<p>Supplemental Figure 2. The concentration of plasma cfDNA in previous described patients with cancer according to stage. (A) The concentration of cfDNA in patients of all cancer types (9). The concentration of cfDNA in patients with Stage III cancers was significantly higher than that of patients with Stage I cancers (p < 0.01). (B) The concentration of cfDNA in individual cancer types. Stage was classified according to AJCC 7th edition. The concentration of cfDNA in patients with Stage III breast cancer was significantly higher than that of patients with Stage I breast cancer (p < 0.05). For both panels, data were derived from the previously published CancerSEEK study (9).</p>
TP53 is the most commonly mutated cancer driver gene but drugs that target TP53 are not yet available. A peptide derived from the most common p53 mutation R175H (HMTEVVRHC) can be presented by a common human leukocyte antigen (HLA-A*02:01) after proteasomal degradation.1 We aimed to develop T cell receptor (TCR)-mimic antibody targeting this shared neoantigen.
Methods
We constructed a single-chain variable fragment (scFv) phage display library that presents scFvs at an estimated diversity of 3.6e10. Mutant peptide-HLA (pHLA)-specific scFvs were enriched through five rounds of positive and negative selections. Mutant pHLA-specific scFv clones were converted into bispecific T cell retargeting antibodies in the single-chain diabody (scDb) format by linking with the anti-CD3 scFv UCHT1.2 These scDbs were tested in T cell co-culturs in the presence of target pulsed cells or target cells that either overexpress the p53 neoantigen or present the p53 neoantigen at endogenous levels. In vivo efficacy was assessed by administering scDb in NSG xenograft mouse model. The structural basis of the binding specificity was evaluated by X-ray crystallography.
Results
We identified an scFv, termed clone H2, that specifically binds p53 R175H/HLA-A*02:01 pHLA and not its wild-type counterpart at a Kd of 86 nM (figure 1A). H2-scDb induced T cell cytokine release and cytotoxicity in the presence of 1) HLA-A*02:01-expressing cells pulsed with the p53R175H peptide, 2) cells transfected with HLA-A*02:01 and p53 R175H, and 3) cancer cell lines KMS26, KLE, and TYK-nu that express endogenous HLA-A*02:01/p53 R175H (figure 1B-E). T cell activation was diminished when TP53 was knocked out from these cancer cell lines using CRISPR (figure 1E). When administered to NOD scid gamma (NSG) mice systemically engrafted with the KMS-26 cell line, H2-scDb significantly suppressed tumor growth (figure 1 F, G). The structure of p53 R175H/HLA-A2 bound to the H2-Fab fragment shows that four complementarity-determining region loops of the H2 antibody formed a cage-like configuration around the C-terminus of the p53 R175H peptide, trapping the mutant histidine (His175) and the adjacent arginine (Arg174) residues in a stable interaction, which provides the structural basis for the specificity (figure 2).
Conclusions
We have developed a TCR-mimic bispecific T cell engager H2-scDb that recognized the shared neoantigen HLA-A*02:01/p53 R175H pHLA complex with exquisite specificity. It effectively activated T cells and lysed tumor cells both in vitro and in vivo. This approach could in theory be used to target cancers containing mutations that are difficult to target in conventional ways.
Acknowledgements
We thank Surojit Sur, Nicolas Wyhs, Ashley Cook, Marco Dal Molin, Richard L. Blosser, Lilian Dasko-Vincent, Christopher Thoburn, Jianhong Cao, and José Rodríguez Molina for assistance with this study.
Trial Registration
NA
Ethics Approval
NA
Consent
NA
References
Lo W, Parkhurst M, Robbins PF, Tran E, Lu YC, Jia L, Gartner JJ, Pasetto A, Deniger D, Malekzadeh P, Shelton TE, Prickett T, Ray S, Kivitz S, Paria BC, Kriley I, Schrump DS, Rosenberg SA. Immunologic recognition of a shared p53 mutated neoantigen in a patient with metastatic colorectal cancer. Cancer Immunol. Res 2019;7:534–543. Stork R, Campigna E, Robert B, Muller D, Kontermann RE. Biodistribution of a bispecific single-chain diabody and its half-life extended derivatives. J Biol Chem 2009;284:25612–25619
<p>Supplemental Figure 4. Methylation profiles using quadratic programming vs. non-negative least- squares regression using the reference matrix described in Sun et al. (3). Pearson’s correlation coefficient and p values are presented at the bottom of this figure, showing the derived contributions from each of the 12 tissue types that could be assessed.</p>
<p>Supplemental Figure 9. Plasma AST and ALT levels before and ~24 hours after surgery for pancreatic cancer. AST and ALT levels substantially increased in all five patients.</p>
<p>Supplemental Figure 6. Correlation between the cfDNA derived from colon epithelial cells, lung cells, and pancreatic cells, and the fraction of cfDNA derived from neoplastic colon epithelial cells, lung cells, or pancreatic cells.</p>
Abstract Mutation-associated neoantigens (MANAs) are exquisitely cancer-specific therapeutic targets. However, MANAs are present at ultra-low densities on the cancer cell surface (as few as 1-2 copies per cell), leading to the challenge of eliciting a sufficiently robust therapeutic effect. We combined components of both T cell receptors (TCRs) and chimeric antigen receptors (CARs) to create a new receptor with improved potency against an ultra-low-density MANA. From CARs, we utilized the antibody-based antigen recognition domain (i.e. the single chain variable fragment, scFv) and the integrated co-stimulation that amplifies T cell activation. From TCRs, we utilized the multi-chain signaling platform that facilitates high antigen sensitivity. This new receptor, termed a TCR Embedded ScFv for Long-term Activation (TESLA), showed promising characteristics when tested with the H2-scFv which targets the p53 R175H mutation presented on HLA-A*02:01 (R175H/A2). Using CRISPR-based homology directed repair in primary human T cells, we tested 15 configurations of appending the H2-scFv to subunits of the TCR complex to identify a design that maximized T cell cytotoxicity and interferon gamma release in co-cultures with cancer cells expressing endogenous levels of the R175H/A2 antigen. In this system, we showed that the optimal TCR-embedded configuration of the H2-scFv produced similar levels of cytotoxicity and interferon gamma secretion as patient-derived TCRs targeting the same R175H/A2 MANA, while conventional H2-CARs were unable to produce any T-cell activation. We then used a multiple stimulation co-culture system to identify a co-stimulation domain combination (MyD88 and CD40) that improved serial cytotoxicity and proliferation of H2-TESLAs when incorporated on the intracellular side of the TCRbeta chain. Finally, we compared the H2-TESLA receptor to patient-derived TCRs modified with the same MyD88 and CD40 co-stimulation domains. In vivo, H2-TESLAs cured all mice in a tumor model, while co-stimulation-modified TCRs produced only temporary tumor control. Moreover, in vivo, H2-TESLAs elicited 100-fold greater T cell expansion than co-stimulation-modified TCRs. In conclusion, we demonstrated that by combining aspects of both CARs and TCRs, the TESLA receptor improved T cell reactivity against an ultra-low-density neoantigen compared to conventional CARs and patient-derived TCRs. Citation Format: Brian J. Mog, Sarah R. DiNapoli, Michael S. Hwang, Tushar D. Nichakawade, Jacqueline Douglass, Emily Han-Chung Hsiue, Katharine M. Wright, Alexander H. Pearlman, Maximilian F. Konig, Suman Paul, Nicolas Wyhs, Nikita Marcou, Stephanie Glavaris, Jiaxin Ge, Michelle S. Miller, P. Aitana Azurmendi, Evangeline Watson, Drew M. Pardoll, Sandra B. Gabelli, Chetan Bettegowda, Nickolas Papadopoulos, Kenneth W. Kinzler, Bert Vogelstein, Shibin Zhou. Hybrid TCR-CAR design surpasses conventional CARs and patient-derived TCRs in targeting an ultra-low-density neoantigen [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 2 (Clinical Trials and Late-Breaking Research); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(8_Suppl):Abstract nr LB095.
<p>Supplemental Figure 10. In silico mixing experiments (N = 10) of buffy coat bisulfite sequencing data with liver (A), lung (B), colon epithelial cell (C), and left atrium (D) bisulfite sequencing data deconvoluted using the Moss et al. (43) reference matrix and quadratic programming shows excellent agreement between predicted and actual fractional contribution.</p>
