ANTIGEN RECEPTORS AND USES THEREOF

Abstract

The present invention generally embraces the treatment of diseases by targeting cells expressing an antigen on the cell surface. In particular the invention relates to recombinant antigen receptors and uses thereof. T cells engineered to express such antigen receptors are useful in the treatment of diseases characterized by expression of one or more antigens bound by the antigen receptors.

Claims

1-81. (canceled)

82-133. (canceled)

134. An antigen receptor, which receptor comprises a first peptide chain and a second peptide chain, wherein the first peptide chain comprises at least an immunoglobulin heavy chain variable domain (VH), an immunoglobulin light chain variable domain (VL) and a human TCR alpha chain constant domain (Cα), and the second peptide chain comprises at least an immunoglobulin heavy chain variable domain (VH), an immunoglobulin light chain variable domain (VL) and a human TCR beta chain constant domain (Cβ); or the first peptide chain comprises at least an immunoglobulin light chain variable domain (VL), an immunoglobulin heavy chain variable domain (VH) and a human TCR alpha chain constant domain (Cα), and the second peptide chain comprises at least an immunoglobulin light chain variable domain (VL), an immunoglobulin heavy chain variable domain (VH) and a human TCR beta chain constant domain (Cβ); or the first peptide chain comprises at least two immunoglobulin light chain variable domains (VL) and a human TCR alpha chain constant domain (Cα), and the second peptide chain comprises at least two immunoglobulin heavy chain variable domains (VH) and a human TCR beta chain constant domain (Cβ); or the first peptide chain comprises at least two immunoglobulin heavy chain variable domains (VH) and a human TCR alpha chain constant domain (Cα); and the second peptide chain comprises at least two immunoglobulin light chain variable domains (VL) and a human TCR beta chain constant domain (Cβ); and wherein a first VH or VL domain from the first peptide chain forms a first antigen binding site together with a first VL or VH domain, respectively, from the second peptide chain, and wherein a second VH or VL domain from the first peptide chain forms a second antigen binding site together with a second VL or VH domain, respectively, from the second peptide chain.

135. The receptor of claim 134, wherein when the first peptide chain comprises at least an immunoglobulin heavy chain variable domain (VH), an immunoglobulin light chain variable domain (VL) and a human TCR alpha chain constant domain (Cα); and the second peptide chain comprises at least an immunoglobulin heavy chain variable domain (VH), an immunoglobulin light chain variable domain (VL) and a human TCR beta chain constant domain (Cβ), either the first peptide chain further comprises a linker between the VH and the VL and/or between the VL and the Cα, or the second peptide chain further comprises a linker between the VH and the VL and/or between the VL and the Cβ, or both the first peptide chain further comprises a linker between the VH and the VL and/or between the VL and the Cα, and the second peptide chain further comprises a linker between the VH and the VL and/or between the VL and the Cβ; or when the first peptide chain comprises at least an immunoglobulin light chain variable domain (VL), an immunoglobulin heavy chain variable domain (VH) and a human TCR alpha chain constant domain (Cα), and the second peptide chain comprises at least an immunoglobulin light chain variable domain (VL), an immunoglobulin heavy chain variable domain (VH) and a human TCR beta chain constant domain (Cβ); either the first peptide chain further comprises a linker between the VL and the VH and/or between the VH and the Cα, or the second peptide chain further comprises a linker between the VL and the VH and/or between the VH and the Cβ, or both the first peptide chain further comprises a linker between the VL and the VH and/or between the VH and the Cα, and the second peptide chain further comprises a linker between the VL and the VH and/or between the VH and the Cβ; or when the first peptide chain comprises at least two immunoglobulin light chain variable domains (VL) and a human TCR alpha chain constant domain (Cα), and the second peptide chain comprises at least two immunoglobulin heavy chain variable domains (VH) and a human TCR beta chain constant domain (Cβ); either the first peptide chain further comprises a linker between the VL domains and/or between the VL and the Cα, or the second peptide chain further comprises a linker between the VH domains and/or between the VH and the Cβ, or both the first peptide chain further comprises a linker between the VL domains and/or between the VL and the Cα, and the second peptide chain further comprises a linker between the VH domains and/or between the VH and the Cβ; or when the first peptide chain comprises at least two immunoglobulin heavy chain variable domains (VH) and a human TCR alpha chain constant domain (Cα); and the second peptide chain comprises at least two immunoglobulin light chain variable domains (VL) and a human TCR beta chain constant domain (Cβ); either the first peptide chain further comprises a linker between the VH domains and/or between the VH and the Cα, or the second peptide chain further comprises a linker between the VL domains and/or between the VL and the Cβ, or both the first peptide chain further comprises a linker between the VH domains and/or between the VH and the Cα, and the second peptide chain further comprises a linker between the VL domains and/or between the VL and the Cβ.

136. The receptor of claim 134, wherein the first and second antigen binding sites bind to the same antigen or different antigens.

137. The receptor of claim 134, wherein the first peptide chain comprises at least two immunoglobulin light chain variable domains (VL) and a human TCR alpha chain constant domain (Cα), and the second peptide chain comprises at least of two immunoglobulin heavy chain variable domains (VH) and a human TCR beta chain constant domain (Cβ); or the first peptide chain comprises at least two immunoglobulin heavy chain variable domains (VH) and a human TCR alpha chain constant domain (Cα); and the second peptide chain comprises at least two immunoglobulin light chain variable domains (VL) and a human TCR beta chain constant domain (Cβ).

138. The receptor of claim 134, wherein an N-terminal domain from the first peptide chain forms together with an N-terminal domain from the second peptide chain an antigen binding site; and a C-terminal domain from the first peptide chain forms together with a C-terminal domain from the second peptide chain an antigen binding site.

139. A peptide chain comprising a first domain and a second domain, wherein the first domain comprises a variable region of a heavy chain of an immunoglobulin (VH) or a variable region of a light chain of an immunoglobulin (VL), and the second domain comprises a variable region of a heavy chain of an immunoglobulin (VH) or a variable region of a light chain of an immunoglobulin (VL), and wherein the peptide chain further comprises a human TCR alpha chain constant domain (Cα) or a human TCR beta chain constant domain (Cβ).

140. A recombinant cell expressing both the first and second peptide chains defined in claim 134.

141. A recombinant cell expressing the peptide chain of claim 139.

142. An ex vivo method for producing a cell expressing an antigen receptor which receptor comprises a first peptide chain and a second peptide chain, the method comprising: (a) providing a human cell; (b) providing a first genetic construct encoding the first peptide chain comprising at least an immunoglobulin heavy chain variable domain (VH), an immunoglobulin light chain variable domain (VL) and a human TCR alpha chain constant domain (Cα); or comprising at least an immunoglobulin light chain variable domain (VL), an immunoglobulin heavy chain variable domain (VH) and a human TCR alpha chain constant domain (Cα); or comprising at least immunoglobulin light chain variable domains (VL) and a human TCR alpha chain constant domain (Cα), or comprising at least two immunoglobulin heavy chain variable domains (VH) and a human TCR alpha chain constant domain (Cα); and (c) providing a second genetic construct encoding the second peptide chain comprising at least of an immunoglobulin heavy chain variable domain (VH), an immunoglobulin light chain variable domain (VL) and a human TCR beta chain constant domain (Cβ); or comprising at least an immunoglobulin light chain variable domain (VL), an immunoglobulin heavy chain variable domain (VH) and a human TCR beta chain constant domain (Cβ); or comprising at least immunoglobulin heavy chain variable domains (VH) and a human TCR beta chain constant domain (Cβ); or comprising at least two immunoglobulin light chain variable domains (VL) and a human TCR beta chain constant domain (Cβ); (d) introducing the first and second genetic constructs into the cell; and (e) allowing the constructs to be expressed in the cell; wherein the cell is a T cell, wherein a first VH or VL domain from the first peptide chain is able to form a first antigen binding site together with a first VL or VH domain, respectively, from the second peptide chain, wherein a second VH or VL domain from the first peptide chain is able to form a second antigen binding site together with a second VL or VH domain, respectively, from the second peptide chain, wherein the first peptide chain and the second peptide chain are provided on a single genetic construct or two separate constructs.

143. The method of claim 142, wherein expression of the antigen receptor is at the cell surface.

144. The method of claim 142, wherein the first peptide chain and the second peptide chain are provided on a single genetic construct.

145. A nucleic acid encoding both the first and second peptide chains defined in claim 134, wherein the nucleic acid is DNA or RNA.

146. A nucleic acid encoding the peptide chain of claim 139, wherein the nucleic acid is DNA or RNA.

147. A pharmaceutical composition comprising (a) the antigen receptor of claim 134, (b) the recombinant cell of claim 140, (c) the nucleic acid of claim 145, or (d) a nucleic acid encoding the first peptide chain and a nucleic acid encoding the second peptide chain, said chains as defined in claim 134; and a pharmaceutically acceptable carrier.

148. A pharmaceutical composition comprising the antigen receptor of claim 134, the recombinant cell of claim 140, or the nucleic acid of claim 145; and a pharmaceutically acceptable carrier for use in the treatment of a disease characterized by expression of at least one antigen which is bound by the antigen receptor, wherein the disease is a cancer disease.

149. The peptide of claim 139, further comprising a linker between the first and second domain and/or between the second domain and the human TCR alpha chain constant domain (Cα) or human TCR beta chain constant domain (Cβ).

150. The receptor of claim 135, wherein the linker comprises (Gly.sub.4Ser).sub.n and n is 2-4.

151. The peptide of claim 149, wherein the linker comprises (Gly.sub.4Ser).sub.n and n is 2-4.

152. A method for the treatment of a disease comprising administering to a subject a therapeutically effective amount of the pharmaceutical composition of claim 147, wherein the disease is characterized by expression of at least one antigen which is bound by the antigen receptor.

153. The method of claim 152, wherein the antigen is a tumor antigen and the disease is cancer.

Description

FIGURES

[0300] FIG. 1 contains a representation of the TCR-CD3 complex. The TCR and the CD3-subunits are composed of ectodomains, a stalk region, a transmembrane domain and cytosolic domains which bear the ITAMs. Expression of the CD4 or CD8 coreceptor determines the commitment to CD4+ or CD8+ T-cell subsets. The intracytoplasmic CD3 immunoreceptor tyrosine-based activation motifs (ITAMs) are indicated as cylinders (adapted from “The T cell receptor facts book”, MP Lefranc, G Lefranc, 2001).

[0301] FIG. 2 shows the design of successive generations of CARs. Schematic representation of the different generations of CARs (1G, first generation, 2G, second generation, 3G, third generation). The first generation contains extracellular scFv’s and the cytoplasmic CD3ζ chain/ZAP70 mediating immediate effector function such as IFNy-secretion or cytotoxicity, the second generation additionally CD28/PI3K promoting proliferation and the third generation furthermore 4-1BB or OX40/TRAF sustaining cell survival (Casucci, M. et al. (2011) 2: 378-382).

[0302] FIG. 3 contains a schematic representation of the different receptor formats for the redirection of T cells against an antigen. Cα/β are the constant domains of a TCR that also operate as a signal transmission domain in relaying the signal of a bound antigen through the cell membrane to the signaling domains of the recruited CD3-complex in the cytoplasma; VH and VL are the variable regions of the heavy and light chains of an immunoglobulin, respectively, and which are representative of the domains on the two peptide chains that form an antigen binding site. Left: a second generation CAR consisting of an antigen-specific scFv fragment, a IgG1-derived spacer domain, a CD28 costimulatory and a CD3ζ signaling domain (Classical single chain CAR); middle: a novel CAR format based on the linkage of the scFv with the constant domain of the murine TCRß chain and coexpression of the constant domain of the murine TCRα chain (monovalent non-combinatory antigen receptor); right: a murine TCR composed of TCR α/β chains (mu, murine). The heterodimer CD3δε and the homodimer CD3ζ ζ is recruited by the Cα-domain while CD3γε is recruited by the Cβ-domain.

[0303] FIGS. 4A and B show the structures of certain of the antigen receptors described herein. Nomenclature of the domains is as described in the legend to FIG. 3. A tandem antigen receptor (tandem AR) comprises a chain comprising 4 variable antibody fragments which form 2 antigen binding sites. An intra/inter-combinatory AR is able of intra- and interchain-binding to antigens. An inter-combinatory AR allows binding of antigens exclusively via interchain-binding. The monovalent non-combinatory AR represents the monovalent prototype of the novel AR-design based on the recruitment of endogenous CD3, the bivalent non-combinatory AR is a CAR, which design has already been suggested by Gross et al., 1992 FASEB J.,(6) 3370-3378 and confines binding to each antigen to a single chain. The monovalent combinatory AR serves as a reference AR or negative control to prove the benefit in function by providing higher valencies by means of scFv-fragments in CAR-design.

[0304] FIG. 5 is a histogram showing the relative expression levels of antigen receptors in T cells. Expression on CAR RNA electroporated T-cells has been validated in flow cytometry analysis by taking advantage of an idiotype antibody directed against the paratope of the Claudin 6-specific antibody IMAB206. The flow cytometry antibody was directly labeled with Dylight-640, expression levels are given in mean fluorescence intensities (MFI).

[0305] FIGS. 6A to 6C are histograms showing the relative induced levels of IFN-γ production, which is an indicator of immune cell activation. FIG. 6A: Detection of IFN-γ production in ELISA for CD8+ T cells electroporated with different antigen receptor expression constructs and controls, and co-cultured with C16 negative and positive immature dendritic cells (iDC). The bivalent antigen receptors capable of inter-chain antigen binding showed good IFN-γ production (inter-combinatory AR 2GS, 3GS, 4GS). Variation of the linker length between the variable domains did not have a significant impact on receptor function, but the 3GS linker appeared to be somewhat better compared to the 2- and 4GS-linker. Truncation of the N-terminal variable domains of the bivalent antigen receptor to obtain a monovalent antigen receptor (monovalent combinatory AR) drastically reduced receptor function, demonstrating the importance of bivalent antigen-binding for superior receptor function. Murine constant domains in the antigen receptor structure lead to slightly improved IFN-γ production (inter-comb. AR Mu 3GS). FIG. 6B: A replicate experiment of that disclosed in FIG. 6A showing similar results but with T cells obtained from a different donor to prove donor-independency of the results. FIG. 6C: Essentially a repeat of the experiments disclosed in FIGS. 6A and 6B. The tandem antigen receptor, and the intra/inter-combinatory antigen receptor (enabling antigen recognition via both intra- and inter-chain VH/VL-domain combination), each lead to a higher induction of IFN-γ expression compared to the monovalent antigen receptor (monovalent non-combinatory AR) and compared to an antigen receptor having only intra-chain antigen binding (bivalent non-combinatory AR).

[0306] FIG. 7 is a histogram showing cytotoxic efficacies of CAR C16-modified T-cells towards ovarian carcinoma cells Sk-ov-3. Tumor cells have been electroporated with increasing amounts of C16 RNA, CD8+ T cells have been electroporated with different antigen receptor expression constructs. T cell effector function as measured by cell lysis decreased with decreasing amounts of C16 expressed in the Sk-ov-3 cells. Cytotoxicity of the T cells recombinantly expressing bivalent antigen receptors (tandem AR and inter-combinatory AR) was less dependent on antigen density on the target cells as compared to the monovalent non-combinatory antigen receptor and classical scCAR.

[0307] FIGS. 8A-8D are histograms showing the results of immune cell activation as determined by cell proliferation against iDCs as APC. FIGS. 8A and 8B: Proliferation of CFSE labeled CD4+ (FIG. 8A) and CD8+ T cells (FIG. 8B), respectively, against C16 negative immature dendritic cells (iDCs). Upon co-stimulation with molecules CD80 and 41BBL in cis (i.e. electroporation into T-cells), the classical single chain chimeric antigen receptor (classical scCAR) revealed background proliferation in the absence of its cognate antigen. This was also observed for T cells incubated without any target cells, indicating that non-specific proliferation is an intrinsic characteristic of T cells expressing the classical scCAR (data not shown). This result was also seen with the C16 negative cell line Sk-Ov-3. FIGS. 8C and 8D: Proliferation of CFSE labeled CD4+ (FIG. 8C) and CD8+ T cells (FIG. 8D), respectively, against C16+ iDCs. CD4+ T cells proliferated against C16 loaded iDCs when electroporated with a construct expressing the inter-combinatory AR. The classical scCAR design led to less proliferation. For CD8+ T cells both, the classical scCAR and inter-combinatory AR showed almost the same good proliferation.

[0308] FIGS. 9A-9B are histograms showing the results of immune cell activation as determined by cell proliferation against tumor cells as APC. Proliferation of CFSE labeled CD4+ (FIG. 9A) and CD8+ (FIG. 9B) T cells against the C16+ ovarian carcinoma cell line Ov-90 electroporated with costimulatory molecules CD80 and 41BBL (co-stimulation in trans). CD4+ T cells were able to proliferate against Ov-90.sup.C16+CD80+41BBL+ when electroporated with the inter-combinatory AR in analogy to iDC.sup.C16+. Both the monovalent and bivalent antigen receptors led to proliferation of CD8+ T cells. The amount of proliferation was comparable with the classical scCAR. Co-stimulation of T cells in cis further enhanced T cell proliferation.

[0309] FIG. 10 is a histogram showing the antigen dose-dependent induction of IFN-γ production. Detection of IFN-γ production in ELISA for CD8+ T cells electroporated with different antigen receptor expression constructs and controls, and co-cultured with immature dendritic cells (iDC) electroporated with relevant full-length C16 RNA in the range of 0.01 .Math.g –1 .Math.g and 1 .Math.g irrelevant gp 100. Bivalent antigen receptors were assessed for cytokine secretion dependent on the hypothesized amount of combinatory chain pairing with the order inter- > inter/intra- > bivalent non-combinatory AR. At high antigen density (1 .Math.g C16) the classical CAR elicited the highest amount of IFNγ-secretion followed by inter- and inter/intra-combinatory AR. Bivalent non-combinatory AR was less functional than the monov. combinatory AR and a monovalent non-combinatory AR. Bivalent non-combinatory AR rely on chain pairing merely between the invariant human C-domains, while the monovalent combinatory AR accomplishes better chain pairing between the human C- and mouse V-domains. The monovalent non-combinatory AR is more functional than the bivalent non-combinatory AR due to the fact that inter-chain pairing is promoted by mouse C-domains instead of weaker human ones. This is also true for any antigen dosis applied in this assay. Notably, when lowering the antigen densities (0.1 .Math.g, 0.01 .Math.g of C16 RNA-electroporated cells) the inter-combinatory AR became increasingly more reactive towards iDCs RNA-pulsed with relevant antigen when compared with the classical CAR. This is in line with a similar trend shown in an antigen dose-dependent cytotoxicity assay (FIG. 7) towards a C16-electroporated tumor cell line Skov-3.

[0310] In FIG. 11A, a comparison of the formats of the classical and the novel combinatory classical CAR is shown. FIG. 11B is a histogram showing the antigen dose-dependent induction of IFN-γ production. Both CARs rely on signaling via the fused CD3ζ-signaling moiety independent of the endogenous CD3 complex in T-cells. FIG. 11B shows the detection of IFN-γ production in ELISA for CD8+ T cells electroporated with the Claudin 6-specific classical CAR and the novel combinatory classical CAR, both carrying CH2-CH3-antibody domains and CD28, and CD3ζ as signal transmission domain. They were co-cultured with immature dendritic cells (iDC) electroporated with relevant full-length C16 RNA in the range of 0.002 .Math.g – 2 .Math.g and 2 .Math.g irrelevant gp100. At higher antigen densities (2 .Math.g – 0.2 .Math.g C16) both CAR formats elicited equal high amounts of IFNγ-secretion. For lower antigen densities the classical CAR seemed to be somewhat more efficient in secreting IFNγ. Importantly, the combinatory classical CAR is functional in the same range of titrated antigen as the classical CAR.

EXAMPLES

[0311] The techniques and methods used herein are described herein or carried out in a manner known per se and as described, for example, in Sambrook et al., Molecular Cloning: A Laboratory Manual, 2.sup.nd Edition (1989) Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y. All methods including the use of kits and reagents are carried out according to the manufacturers’ information unless specifically indicated.

Example 1: Expression of Antigen Receptors in T Cells

[0312] The expression of various antigen receptor constructs was assessed one day after electroporation into CD8+ T cells using a C16 scFv idiotype-specific antibody. The constructs or combination of constructs tested for expression were (i) the constant domain of the murine T cell receptor alpha chain alone (mCα); (ii) VH-VL-mCβ alone (scFv-mCβ); (iii) VH-VL-CH2-CH3-CD28-CD3ζ (Classical scCAR); (iv) mCα and VH-VL-mCβ (monovalent non-combinatory AR); (v) VH-VL-mCα and VH-VL-mCβ (bivalent non-combinatory AR); (vi) VL-VH-mCα and VH-VL-mCβ (Intra/Inter-combinatory AR); (vii) mCα and VH-VL-VH-VL-mCβ (Tandem AR); (viii) VH-hCα and VL-hCβ (monovalent combinatory AR); (ix) VH-(GGGGS)3-VH-hCα and VL-(GGGGS)3-VL-hCβ (Inter-combinatory AR 3GS); and (x) VH-(GGGGS)3-VH-mCα and VL-(GGGGS)3-VL-mCβ (Inter-combinatory AR Mu 3GS) (“m” or “Mu” indicates murine origin and “h” indicates human origin of the constant domain). These different antigen receptor constructs are schematically presented in FIGS. 4A-4B. The data in FIG. 5 represent results from two separate measurements for the bivalent non-combinatory, the intra/inter- combinatory and monovalent combinatory antigen receptors; five measurements for the tandem and inter-combinatory antigen receptors; and up to ten measurements for the classical scCAR and monovalent non-combinatory antigen receptors. The MFI value of each sample was normalized to the corresponding negative control (mCα or scFv-mCβ set to 1).

[0313] As shown in FIG. 5, the classical scCAR design showed the best expression. The monovalent non-combinatory, bivalent non-combinatory, intra/inter-combinatory and the inter-combinatory antigen receptors with human (hC) or murine (mC) constant domains showed similar expression. The monovalent combinatory AR showed reduced surface staining, whereas the tandem antigen receptor showed increased surface staining, as compared to the other constructs, except for the classical scCAR.

Example 2: IFN-γ Secretion Assay

[0314] On day 1 of the experiment, fresh peripheral blood mononuclear cells (“PBMCs”) were isolated from a buffy coat of one healthy donor. From ¾ of PBMCs, CD14+ cells were isolated using MACS sort. MACS flow through and residual PBMCs were then MACS sorted for CD8+ T cells. CD14+ cells were differentiated towards immature dendritic cells (“iDCs”) by administration of IL-4 & GM-CSF (1000 U/ml) on day 1, 3, 6. CD8+ T cells were transferred on OKT3 coated 6 well plates. On day 3, T cells were transferred to new 6 well plates. On day 7, iDCs were electroporated with irrelevant and C16 IVT RNA. OKT3 activated T cells were subsequently electroporated with controls, or antigen receptor constructs as set forth in the individual figures and as described in Example 1. For quality assurance, antigen receptor surface expression on T cells was analyzed using a C16 antigen receptor idiotype-specific antibody labeled with Dylight 650 on day 8. The electroporated T cells and antigen electroporated iDCs were subsequently co-cultured in a 96 well plate for 20 h at an E:T ratio of 10:1 in duplicates. On day 9, culture supernatants were taken and analyzed for the amount of secreted IFN-γ in a sandwich ELISA using the IFN-γ Ready Set Go! kit from eBioscience (#88-7316-88). Absorbance was detected using a Tecan Sunrise ELISA reader.

[0315] As depicted in FIG. 6A, mock (mCα) electroporated T cells in culture with C16 positive or negative iDCs showed no non-specific IFN-y production. Of the antigen receptor positive T cells only the classical scCAR positive cells produced a detectable background of the cytokine when cultured with C16 negative target cells. IFN-γ production from monovalent antigen receptor positive cells could be observed when cultured with C16+ iDCs. Here, the classical scCAR produced high amounts of IFN-γ. When electroporated with a negative control of only one chain of the combinatory antigen receptor, in this case VH-VH-Cα, no IFN-y production could be observed. This control proved that both chains are required for antigen recognition and subsequent activation of the T cells. In contrast, the inter-combinatory AR (3GS) showed a pronounced improvement in IFN-γ production when compared to that produced with the monovalent AR (combinatory and non-combinatory). Different linker lengths, 2 or 4 repeats of the Gly4Ser linker between the variable domains, showed no major impact on function, as compared to the 3 repeats.

[0316] Importantly, a truncation of the N-terminal variable domain in the combinatory antigen receptor structure showed a strong reduction in effector function. This observation clearly proved that bivalent antigen-binding supports T cell activation. The prominent increase in function between the monovalent combinatory AR and the bivalent combinatory AR is also a strong indication that, besides the improvement of chain-pairing mediated by the inter-chain variable domains, the incidence of antigen binding itself across the two chains further stabilizes receptor chain pairing and hence, improves incorporation into the endogenous CD3 complex and subsequent T cell activation/function. From double chain T cell receptors it is well-known that heterodimerisation of the chains is an essential prerequisite for efficient incorporation into the CD3 complex.

[0317] The integration of murine residues into the combinatory AR structure was able to further increase activation, which is reasoned to be due to the known stronger interaction of murine TCR constant Cα/β-domains as compared to human ones. Thus, improved heterodimerisation of the peptide chains of the antigen receptor, either through inter-chain antigen binding to the variable domains or through dimerization of the T cell receptor constant domains on the individual chains, improved integration of the antigen receptor into the endogenous CD3 complex, and thus, improved T cell function. These results were highly reproducible with a different T cell donor (see FIG. 6B).

[0318] Notably, the classical scCAR demonstrated in this experiment a non-specific background of IFN-γ production against C16 negative iDCs. This result is also reproducible and is in line with data published by Long et al. ((2015) Nat. Med., (21) 581-590) discussing tonic signaling by the classical scCAR-CD28-CD3ζ fusion format. Long et al. observed an antigen-independent activation for several classical scCARs of varying antigen specificities. It is assumed that non-specific background activation of classical scCAR positive T cells is not a T cell donor specific effect.

[0319] Further, and as depicted in FIG. 6C, bivalent non-combinatory and intra/inter-combinatory antigen receptors were tested together with the tandem bivalent antigen receptor for IFN-γ production. For comparison, the negative control, the monovalent non-combinatory AR, and as a positive control, the classical scCAR are also shown in the figure. The tandem bivalent antigen receptor showed an improvement compared to the monovalent antigen receptor. The bivalent non-combinatory antigen receptor VH-VL-Cα + VH-VL-Cβ elicited lower IFN-γ production as compared to the related intra/inter-combinatory antigen receptor VL-VH-Cα + VH-VL-Cβ. These observations confirmed the hypothesis that inter-chain interactions conferred by both, V-domain pairing and antigen binding to variable domains on different chains stabilizes the heterodimeric topology and provides for a more robust receptor-dependent T cell signaling response.

Example 3: Cytotoxicity Assay

[0320] On day 1 of the experiment, fresh PBMCs were isolated from two buffy coats from two healthy donors. PBMCs were MACS sorted for CD8+ T cells. CD8+ T cells were transferred on OKT3 coated 6 well plates. They were cultured in medium containing 50 U/ml IL-2. On day 3, T cells were transferred on new 6 well plates and culture medium was changed. On day 7, the ovarian carcinoma cell line Sk-Ov-3 was RNA-electroporated with varying amounts of C16 RNA and 10 .Math.g of luciferase RNA. OKT3 activated T cells were electroporated with an irrelevant classical scCAR, a relevant C16-specific classical scCAR, a monovalent non-combinatory antigen receptor and a tandem antigen receptor as well as an inter-combinatory antigen receptor of the invention as indicated in FIG. 7 and described above in Example 1. For quality assurance, antigen receptor surface expression on T cells was analyzed using a C16 antigen receptor idiotype specific antibody labeled with Dylight 650 on day 8. The antigen receptor electroporated T cells and antigen electroporated Sk-Ov-3 were subsequently co-cultured in a 96 well plate for 3 h at an E:T ratio of 30:1 in triplicates. After 3 h of incubation, luciferin was added to each culture. Specific lysis was detected by a decrease in luciferin-signal due to its turnover by released luciferase in the TECAN reader.

[0321] The Sk-Ov-3 data impressively document that the bivalent antigen receptors (inter-combinatory AR and tandem AR) of the invention showed a marked improvement as compared to the classical scCAR and monovalent non-combinatory AR. Compared to all other antigen receptor constructs, the classical scCAR design showed the best lysis of about 77% against C16 electroporated Sk-Ov-3 cells. However, it should be mentioned that 10 .Math.g of antigen RNA is a non-physiological condition and does not accurately reflect the in vivo situation. At low antigen doses, the monovalent non-combinatory AR was not able to lyse tumor cells in a satisfactory fashion (9.2%). In contrast, the inter-combinatory AR (3GS) still showed good specific lysis (41.3%) compared to the classical scCAR design (48.1%) and conclusively, was less dependent on antigen density for substantial cytotoxic effector function.

Example 4: Proliferation Assay

[0322] On day 1 of the experiment, fresh PBMCs were isolated from a buffy coat of a healthy donor. From ¾ of PBMCs, CD14+ cells were isolated using MACS sort, and residual PBMCs were frozen. CD14+ cells were differentiated to iDCs by administration of IL-4 & GM-CSF (1000 U/ml) on day 1, 3, 6. On day 7 iDCs were electroporated with irrelevant and C16 IVT-RNA. The frozen PBMCs were thawed on the same day and MACS sorted for CD4+ and CD8+ cells. Without any prior activation (OKT3), naive T cells, 6 and 7×10.sup.6 cells, respectively, were subsequently electroporated with controls, classical, monovalent and bivalent antigen receptor constructs as indicated in FIGS. 8A-8D, which constructs were described in Example 1. In an independent set of T cell responders, the same antigen receptor constructs were also electroporated along with the co-stimulatory molecules 41BBL and CD80 to achieve an auto-co-stimulation, or co-stimulation in cis, which was demonstrated to improve effector function.

[0323] For quality assurance, antigen receptor and 41BBL + CD80 expression on T cells was analyzed by FACS staining on day 8. T cells were subsequently labeled with the proliferation marker CFSE. The electroporated T cells and iDCs as well as the ovarian carcinoma cell line OV-90 were subsequently co-cultured in a 96 well plate for 5 days at an E:T ratio of 10:1 in duplicates. On day 5, cultured cells were stained in the 96 well plates with CD4 or CD8 antibodies labeled with APC-Cy7. Proliferation of T cells was detected via FACS by the reduction of CFSE-signal due to dilution in proliferating daughter cells. With minor adjustments, daughter population sizes were assessed using the in Flowjo implemented proliferation tool. The sum of proliferating cells is depicted as daughter generations.

[0324] Background proliferation of T cells was assessed for cells cultured with either the C16 negative cell line Sk-Ov-3 and C16 negative iDCs. Independently of the electroporated antigen receptor construct, neither CD4+ nor CD8+ T cells proliferated against C16 negative cells. Upon co-stimulation in cis, only classical scCAR engineered T cells proliferated non-specifically against C16 negative iDCs and Sk-Ov-3 (Sk-Ov-3 data not shown). CD4+ T cells proliferated overall less efficiently against C16+ cells compared to CD8+ T cells (see FIGS. 8C to 8D). This was in particular true when using C16+ Ov-90 tumor cells as target cells (FIGS. 9A to 9B).

[0325] The results for T cells co-cultured with iDCs proved the overall good function of the inter-combinatory antigen receptor. Of CD4+ T cells without co-stimulation in cis, the inter-combinatory antigen receptor showed superior proliferation and even outperformed the classical scCAR design. This effect was not due to a reduced surface expression of the classical scCAR, as analyzed by CAR idiotype staining. For CD8+ T cells, the inter-combinatory antigen receptor design demonstrated a remarkable proliferation against C16 loaded iDCs (70%). Co-stimulation of the T cells in cis could further enhance T cell responses.

[0326] Interestingly no proliferation of cells could be observed against the C16+ ovarian carcinoma cell line OV-90 (data not shown). This was reasoned to be due to a lack of co-stimulatory molecules on the surface of the tumor cells. In order to compensate this lack of co-stimulus, Ov-90 cells were electroporated with CD80 and 41BBL RNA. In this case, proliferation of CD4+ and CD8+ T cells could be detected (see FIGS. 9A and 9B). Notably, for CD4+ cells this was only the case for the inter-combinatory AR design (15%). For CD8+ cells, the proliferation pattern looked different. The monovalent non-combinatory AR, classical scCAR and inter-combinatory AR responses were comparable with 60% proliferating cells. Upon co-stimulation in cis, proliferation of in particular CD4+ T cells was improved.

[0327] The data clearly indicate that the bivalent antigen receptor construct (Inter-combinatory AR) is capable of proliferating against C16+ tumor cell lines, when co-stimulated. In general, the inter-combinatory AR is by far better than the monovalent non-combinatory AR in proliferating against iDCs loaded with the cognate target antigen. The classical scCAR is prone to non-specific responses when co-stimulated in cis against C16 negative iDCs, indicating a higher susceptibility for antigen-independent T cell signaling.

Example 5: Antigen-titrated IFNγ- Secretion Assay

[0328] On day 1 of the experiment, fresh peripheral blood mononuclear cells (“PBMCs”) were isolated from a buffy coat of one healthy donor. From ¾ of PBMCs, CD14+ cells were isolated using MACS sort. MACS flow through and residual PBMCs were then MACS sorted for CD8+ T cells. CD14+ cells were differentiated towards immature dendritic cells (“iDCs”) by administration of IL-4 & GM-CSF (1000 U/ml) on day 1, 3, 6. CD8+ T cells were transferred on OKT3 coated 6 well plates. On day 3, T cells were transferred to new 6 well plates. On day 7, iDCs were electroporated with irrelevant and C16 IVT RNA dose-dependently. OKT3 activated T cells were subsequently electroporated with controls, or antigen receptor constructs as set forth in the individual figures and as described in Example 1. For quality assurance, antigen receptor surface expression on T cells was analyzed using a C16 antigen receptor idiotype-specific antibody labeled with Dylight 650 on day 8. The electroporated T cells and antigen electroporated iDCs were subsequently co-cultured in a 96 well plate for 20 h at an E:T ratio of 10:1 in duplicates. On day 9, culture supernatants were taken and analyzed for the amount of secreted IFN-γ in a sandwich ELISA using the IFN-γ Ready Set Go! kit from eBioscience (#88-7316-88). Absorbance was detected using a Tecan Sunrise ELISA reader. The result are shown in FIG. 10.

[0329] Bivalent antigen receptors were assessed for the amount of cytokine secretion dependent on the hypothesized propensity to pair in a combinatory fashion which favours stable expression and subsequently, T cell signaling: We speculated to observe exclusive combinatory V-domain inter-chain pairing for the inter-combinatory AR while the inter/intra-combinatory AR may coexist in a proportion of less favourable intra-chain pairing. At high antigen density (1 .Math.g C16) the classical CAR elicited the highest amount of IFNy-secretion followed, as expected, by inter- and inter/intra-combinatory AR. Bivalent non-combinatory AR was less functional than the references of a monovalent combinatory AR and a monovalent non-combinatory AR. Bivalent non-combinatory AR rely on chain pairing merely between the invariant human C-domains, while the monovalent combinatory AR accomplishes better chain pairing between the human C-and mouse V-domains. In line with the expected results, also the monovalent non-combinatory AR is more functional than the bivalent non-combinatory AR because inter-chain pairing, despite the fact that it is also restricted to C-domains here, is promoted by mouse C-domains instead of weaker human ones. This is also true for any antigen dosis applied in this assay.

[0330] Notably, when lowering the antigen densities (0.1 .Math.g, 0.01 .Math.g of C16 RNA-electroporated cells) the inter-combinatory AR became increasingly more reactive towards iDCs pulsed with relevant RNA when compared with the classical CAR. This is in line with a similar trend shown in a antigen dose-dependent cytotoxicity assay (FIG. 7) towards a C16-electroporated tumor cell line Skov-3.

[0331] The classical CAR was also compared with a novel combinatory classical AR in an antigen-titrated IFNy-secretion assay (FIG. 11). Both chimeric antigen receptors comprise CH2-CH3-antibody domains and CD28, and CD3ζ as signal transmission domain. They were co-cultured with immature dendritic cells (iDC) electroporated with relevant full-length C16 RNA in the range of 0.002 .Math.g – 2 .Math.g and 2 .Math.g irrelevant gp100. At higher antigen densities (2 .Math.g – 0.2 .Math.g C16) both CAR formats elicited equal high amounts of IFNy-secretion. For lower antigen densities the classical CAR seemed to be somewhat more efficient in secreting IFNy. Importantly, the combinatory classical CAR is functional in the same range of titrated antigen as the classical CAR down to very low amounts of endogenously expressed Claudin 6. However, this format of a combinatory AR, as opposed to the TCR Cα/Cβ-based combinatory AR design, apparently leads to equal but not better T-cell signaling than the classical CAR presumably due to exogenous CD3ζ-dependent but endogenous CD3-independent signaling. The former mechanism is supposed to be independent of endogenous CD3-recruitment to the TCR or CAR, respectively, which in turn would regulate the amount of T-cell activation.