An optimized single chain TCR scaffold relying on the assembly with the native CD3-complex prevents residual mispairing with endogenous TCRs in human T-cells

Immunotherapy of cancer envisions the adoptive transfer of T-cells genetically engineered with tumor-specific heterodimeric α/β T-cell receptors (TCRα/β). However, potential mispairing of introduced TCRα/β-chains with endogenous β/α-ones may evoke unpredictable autoimmune reactivities. A novel single chain (sc)TCR format relies on the fusion of the Vα-Linker-Vβ-fragment to the TCR Cβ-domain and coexpression of the TCR Cα-domain capable of recruiting the natural CD3-complex for full and hence, native T-cell signaling. Here, we tested whether such a gp100(280-288)- or p53(264-272) tumor antigen-specific scTCR is still prone to mispairing with TCRα. In a human Jurkat-76 T-cell line lacking endogenous TCRs, surface expression and function of a scTCR could be reconstituted by any cointroduced TCRα-chain indicating mispairing to take place on a molecular basis. In contrast, transduction into human TCRα/β-positive T-cells revealed that mispairing is largely reduced. Competition experiments in Jurkat-76 confirmed the preference of dcTCR to selfpair and to spare scTCR. This also allowed for the generation of dc/scTCR-modified cytomegalovirus/tumor antigen-bispecific T-cells to augment T-cell activation in CMV-infected tumor patients. Residual mispairing was prevented by strenghtening the Vα-Li-Vβ-fragment through the design of a novel disulfide bond between a Vα- and a linker-resident residue close to Vβ. Multimer-stainings, and cytotoxicity-, IFNγ-secretion-, and CFSE-proliferation-assays, the latter towards dendritic cells endogenously processing RNA-electroporated gp100 antigen proved the absence of hybrid scTCR/TCRα-formation without impairing avidity of scTCR/Cα in T-cells. Moreover, a fragile cytomegalovirus pp65(495-503)-specific scTCR modified this way acquired enhanced cytotoxicity. Thus, optimized scTCR/Cα inhibits residual TCR mispairing to accomplish safe adoptive immunotherapy for bulk endogenous TCRα/β-positive T-cells.


topology of un/modified TCR constructs used throughout this manuscript. (D)
The top panel depicts the domain arrangements of the wild type human TCRchain of the gp100specificity used as a 'sensor' for mispairing with a 3-domain scTCR of the same antigenspecificity and thus, served as a 'surrogate' for any endogenous TCRchain in mispairing analyses. Additionally, the human wild type or in C-domain murinized (chimerized) double chain TCRs of the gp100-and pp65-specificity, respectively (green or grey). The middle panel depicts the autonomously coexpressed mouse C-domain (grey) along with modified human 3-domain single chain TCRs gp100 murinized in TCR C-domains (green-rimmed).
They were either functionally unresponsive by a silencing mutation S109Q situated on top of the CDR3antigen-recognizing loop, and/or stabilized in V-domain pairing via an artificial disulfide bond bridging VG49C with the C-terminal tail of the Gly/Ser-rich linker at position G17C. The bottom panel illustrates the wild type murine TCRchain also used as a 'sensor' of mispairing and the modified murine single chain TCRs p53 (grey-rimmed) either functionally unresponsive by a silencing mutation D109A on top of CDR3, and/or stabilized in V-domain pairing via an artificial disulfide bond bridging VG51C with the C-terminal tail of the same linker at position G16C. Beside the wild type disulfide bond in TCR Cdomains a routinely used artificial disulfide bond between C T84C / C S79C accomplished stronger C-domain interaction. Enumeration was according to the IMGT database as cited in the main text.

Supplementary Figure 2: A 3-domain scTCR molecularly interacts with TCR in
human Jurkat-76 devoid of endogenous TCRs. (A) Murinization of human TCR Cdomains leads to higher mispairing with a chimerized human scTCR. 5 x 10 6 Jurkat-76 cells were electroporated with 5µg RNA coding for indicated TCR constructs. After 12 hours, TCR expression was analyzed cytofluorometrically by means of v14-staining or antigen recognition (top) as mentioned in Fig. 1. Additionally, the same responder cells were analyzed in IFN-γ spot production (below) in response to gp100(280-288) peptide-pulsed A2.1 + T2 targets at the indicated peptide concentrations at an effector to target cell ratio of 0.3:1. As a control, T2 cells loaded with an A2.1-binding peptide of p53(264-272) were used. Data are shown as mean + SD of duplicates. TCR of the same antigen-specificity operates as a 'sensor' of mispairing and hence, as a 'surrogate' for any (endogenous) TCRMurinization in its C-domain facilitates TCR-mispairing (readout by v14 and multimer) and, if coexpressed with an unrelated TCR pp65, also TCR C-mispairing (readout by multimer).
Mispairing of a murine scTCR p53 with full length mouse or human TCR-chains takes place in Jurkat-76. (B) A 3-domain Wt or functionally unresponsive (silCDR3) scTCR p53 construct was coexpressed with Mu C or diverse antigen (p53, MDM2, gp100, pp65)-and species (mouse, human)-un/related TCR-chains as indicated. Each chain, encoded on a separate plasmid, was retrovirally introduced into J-76 and normalized in gene expression by drug selection, and expanded for at least a week. Expression of the TCR was analyzed by V-and antigen recognition by tetramer p53(264-272)-staining in flow cytometry.

Supplementary Figure 4: Prevention of residual mispairing in human T-cells by a novel artificial disulfide bond designed between V and the C-terminal tail of the linker close
to V for scTCR gp100. 4-10 g of RNA encoding Mu C, or TCRgp100, or different scTCR gp100-constructs as described in Figures 4A, 6A/C were electroporated into MACSpurified human CD8 + T-cells. Unmodified Hu Chim scTCR gp100 or functionally unresponsive Hu Chim scTCR gp100 silCDR3 S109Q were compared with their corresponding scTCR derivatives stabilized in the scTCR-fragment via the novel cystine bridge V-Li(V). Normalized expression was assessed from coelectroporation of Mu C.
TCR-mispairing was assessed from coexpression with TCR gp100. 20 h after coculture with K562-A2 dose-dependently pulsed with the relevant gp100(280-288) or an irrelevant peptide at 10 -6 M at an E:T-ratio of 6:1, responder T-cells were submitted to an IFN-Elispotassay. The incorporation of the scTCR-fragment stabilizing disulfide bond into Hu Chim scTCR gp100 yielded equal IFN-spot production down to 10 -9 M peptide pulse compared with unmodified Hu Chim scTCR gp100 + Mu C and importantly, eliminated residual IFNspot production at 10 -6 M peptide pulse (black box) resulting from TCR-mispairing.