Supplementary Figure Legends and Tables

Abstract IMA101 is an actively personalized, multi-targeted adoptive cell therapy (ACT), whereby autologous T cells are directed against multiple novel defined peptide-HLA (pHLA) cancer targets. HLA-A*02:01-positive patients with relapsed/refractory solid tumors expressing ≥1 of 8 predefined targets underwent leukapheresis. Endogenous T cells specific for up to 4 targets were primed and expanded in vitro. Patients received lymphodepletion (fludarabine, cyclophosphamide), followed by T-cell infusion and low-dose IL2 (Cohort 1). Patients in Cohort 2 received atezolizumab for up to 1 year (NCT02876510). Overall, 214 patients were screened, 15 received lymphodepletion (13 women, 2 men; median age, 44 years), and 14 were treated with T-cell products. IMA101 treatment was feasible and well tolerated. The most common adverse events were cytokine release syndrome (Grade 1, n = 6; Grade 2, n = 4) and expected cytopenias. No patient died during the first 100 days after T-cell therapy. No neurotoxicity was observed. No objective responses were noted. Prolonged disease stabilization was noted in three patients lasting for 13.7, 12.9, and 7.3 months. High frequencies of target-specific T cells (up to 78.7% of CD8+ cells) were detected in the blood of treated patients, persisted for >1 year, and were detectable in posttreatment tumor tissue. Individual T-cell receptors (TCR) contained in T-cell products exhibited broad variation in TCR avidity, with the majority being low avidity. High-avidity TCRs were identified in some patients’ products. This study demonstrates the feasibility and tolerability of an actively personalized ACT directed to multiple defined pHLA cancer targets. Results warrant further evaluation of multi-target ACT approaches using potent high-avidity TCRs. See related Spotlight by Uslu and June, p. 865

Background Several peptide-HLA targets for T cell receptor (TCR)-based immunotherapies are currently being evaluated in the field, however, many are limited by their overall low prevalence, low copy numbers or relevant expression in healthy tissues. A T cell target with nearly ideal properties has high, homogenous and prevalent expression across multiple cancers in the absence of significant safety/toxicity liabilities. Here, we describe the in-depth characterization of an HLA-A*02:01-presented peptide derived from the cancer germline antigen preferentially expressed antigen in melanoma (PRAME) that opens an avenue of new opportunities for patients with solid cancers which we aim to leverage by two distinct TCR-based therapeutic modalities, TCR-engineered T cells (ACTengine® IMA203) and TCR Bispecifics (TCER® IMA402). Methods PRAME target peptide presentation and mRNA expression in tumor and normal tissues was assessed by quantitative mass spectrometry (MS) and transcriptomics. In situ hybridization was used to analyze target homogeneity. We compare target prevalences based on IMADetect® qPCR testing of screening biopsies from clinical trial patients with prevalences based on a PRAME target tailored, MS-based mRNA expression threshold applied to a large RNA sequencing dataset. Results PRAME RNA expression is elevated across many tumor types and only minimal in some normal tissues except testis. As demonstrated by MS, RNA expression does not translate into relevant peptide presentation on normal tissues. Peptide copy numbers range from 100 to 1,000 peptide copies per cell (target density) in the majority of tumor tissues as measured by highly sensitive MS-based AbsQuant® technology. Histologic analysis of PRAME RNA in different solid tumors demonstrates homogenous expression of PRAME with a high frequency of positive tumors cells. PRAME shows a prevalence between 80-100% in uterine and ovarian carcinoma, sarcoma subtypes, cutaneous and uveal melanoma and high prevalence in many other solid cancer types, such as cholangiocarcinoma, lung, kidney, breast, head and neck, esophageal, bladder and hepatocellular carcinoma. Prevalence numbers obtained during patient screening in our clinical trials match predicted prevalences. Interim phase 1a data from the IMA203 TCR-T trial (cut-off Oct 05, 2021) showed clinical responses in head and neck carcinoma, synovial sarcoma, uveal melanoma, and cutaneous melanoma. Conclusions Here, we demonstrate comprehensive target characterization and validation data supporting the nearly ideal target properties of PRAME that can be exploited for the benefit of patients: PRAME is highly cancer-associated, homogenously expressed, presented at high target density, highly prevalent across many solid cancers and clinically validated, underlining its potential to reach a large cancer patient population. Trial Registration NCT03686124 Ethics Approval The study was approved by the institutional review board/ethics committee as required for each participating site.

Knowledge about the peptide repertoire presented by human leukocyte antigens (HLA) holds the key to unlock target-specific cancer immunotherapies such as adoptive cell therapies or bispecific T cell engaging receptors. Therefore, comprehensive and accurate characterization of HLA peptidomes by mass spectrometry (immunopeptidomics) across tissues and disease states is essential. With growing numbers of immunopeptidomics datasets and the scope of peptide identification strategies reaching beyond the canonical proteome, the likelihood for erroneous peptide identification as well as false annotation of HLA-independent peptides as HLA ligands is increasing. Such "fake ligands" can lead to selection of nonexistent targets for immunotherapeutic development and need to be recognized as such as early as possible in the preclinical pipeline. Here we present computational and experimental methods that enable the identification of "fake ligands" that might be introduced at different steps of the immunopeptidomics workflow. The statistics presented herein allow discrimination of true HLA ligands from coisolated HLA-independent proteolytic fragments. In addition, we describe necessary steps to ensure system suitability of the chromatographic system. Furthermore, we illustrate an algorithm for detection of source fragmentation events that are introduced by electrospray ionization during mass spectrometry. For confirmation of peptide sequences, we present an experimental pipeline that enables high-throughput sequence verification through similarity of fragmentation pattern and coelution of synthetic isotope-labeled internal standards. Based on these methods, we show the overall high quality of existing datasets but point out limitations and pitfalls critical for individual peptides and how they can be uncovered in order to identify true ligands.

The additional author support information was erroneously omitted from the Supplementary Information. This has been corrected online.

Patients with glioblastoma currently do not sufficiently benefit from recent breakthroughs in cancer treatment that use checkpoint inhibitors1,2. For treatments using checkpoint inhibitors to be successful, a high mutational load and responses to neoepitopes are thought to be essential3. There is limited intratumoural infiltration of immune cells4 in glioblastoma and these tumours contain only 30–50 non-synonymous mutations5. Exploitation of the full repertoire of tumour antigens—that is, both unmutated antigens and neoepitopes—may offer more effective immunotherapies, especially for tumours with a low mutational load. Here, in the phase I trial GAPVAC-101 of the Glioma Actively Personalized Vaccine Consortium (GAPVAC), we integrated highly individualized vaccinations with both types of tumour antigens into standard care to optimally exploit the limited target space for patients with newly diagnosed glioblastoma. Fifteen patients with glioblastomas positive for human leukocyte antigen (HLA)-A*02:01 or HLA-A*24:02 were treated with a vaccine (APVAC1) derived from a premanufactured library of unmutated antigens followed by treatment with APVAC2, which preferentially targeted neoepitopes. Personalization was based on mutations and analyses of the transcriptomes and immunopeptidomes of the individual tumours. The GAPVAC approach was feasible and vaccines that had poly-ICLC (polyriboinosinic-polyribocytidylic acid-poly-l-lysine carboxymethylcellulose) and granulocyte–macrophage colony-stimulating factor as adjuvants displayed favourable safety and strong immunogenicity. Unmutated APVAC1 antigens elicited sustained responses of central memory CD8+ T cells. APVAC2 induced predominantly CD4+ T cell responses of T helper 1 type against predicted neoepitopes. In a phase I trial, highly individualized peptide vaccines against unmutated tumour antigens and neoepitopes elicited sustained responses in CD8+ and CD4+ T cells, respectively, in patients with newly diagnosed glioblastoma.

Abstract Adoptive cellular therapy (ACT) has dramatically changed the landscape of cancer immunotherapy. ACTolog® and ACTengine® are actively personalized ACT approaches employing T-cell receptor products based on a warehouse of human leucocyte antigen (HLA)-bound peptide targets. Selecting the relevant target candidates requires the establishment of biomarkers predictive for HLA peptide presentation and their development into companion diagnostic devices. Here we describe the development of IMA_Detect, a diagnostic test based on gene expression analysis of primary patient tumor material by qPCR which is predictive for presentation of a target peptide by HLA. To establish mRNA expression levels which indicate actual peptide presentation, data of Immatics' XPRESIDENT® target discovery platform was used by integrating quantitative immunopeptidomics data (label-free LC-MS) with paired transcriptomics data (RNA-Seq). The peptide-specific correlation between peptide presentation and expression of the coding exons was verified for each target individually. The resulting RNA-Seq thresholds were mapped to qPCR thresholds using calibration curves followed by validation of the qPCR assay. The IMA_Detect test is performed in a CLIA/CAP approved setting and is currently being applied in the phase I ACT trials IMA101-101 and IMA201-101 to determine whether a patient's tumor expresses any of the tested targets at levels considered sufficient for potential benefit from the administered T-cell therapy. IMA101-101 uses autologous T-cell products (ACTolog®) for patients with solid cancers while IMA201-101 is based on TCR-engineered T Cells (ACTengine®) in NSCLC and HNSCC patients. We will present first results of the patient screening and the personalized target selection. Citation Format: Jens Fritsche, Arun Satelli, Helen Hörzer, Barbara Rakitsch, Franziska Hoffgaard, Norbert Hilf, Oliver Schoor, Harpreet Singh-Jasuja, Weinschenk Toni. IMA_Detect: Mass spectrometry guided development and clinical application of a companion diagnostic for adoptive cellular therapy against tumor associated HLA peptides [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 662.

Immunotherapy is revolutionizing cancer treatment and has shown success in particular for tumors with a high mutational load. These effects have been linked to neoantigens derived from patient‐specific mutations. To expand efficacious immunotherapy approaches to the vast majority of tumor types and patient populations carrying only a few mutations and maybe not a single presented neoepitope, it is necessary to expand the target space to non‐mutated cancer‐associated antigens. Mass spectrometry enables the direct and unbiased discovery and selection of tumor‐specific human leukocyte antigen (HLA) peptides that can be used to define targets for immunotherapy. Combining these targets into a warehouse allows for multi‐target therapy and accelerated clinical application. For precise personalization aimed at optimally ensuring treatment efficacy and safety, it is necessary to assess the presence of the target on each individual patient's tumor. Here we show how LC‐MS paired with gene expression data was used to define mRNA biomarkers currently being used as diagnostic test IMADETECT™ for patient inclusion and personalized target selection within two clinical trials (NCT02876510, NCT03247309). Thus, we present a way how to translate HLA peptide presentation into gene expression thresholds for companion diagnostics in immunotherapy considering the peptide‐specific correlation to its encoding mRNA.

In addition to genomic mutations, RNA editing is another major mechanism creating sequence variations in proteins by introducing nucleotide changes in mRNA sequences. Deregulated RNA editing contributes to different types of human diseases, including cancers. Here we report that peptides generated as a consequence of RNA editing are indeed naturally presented by human leukocyte antigen (HLA) molecules. We provide evidence that effector CD8(+) T cells specific for edited peptides derived from cyclin I are present in human tumours and attack tumour cells that are presenting these epitopes. We show that subpopulations of cancer patients have increased peptide levels and that levels of edited RNA correlate with peptide copy numbers. These findings demonstrate that RNA editing extends the classes of HLA presented self-antigens and that these antigens can be recognised by the immune system.

A neglect of diatomic differential overlap (NDDO) Hamiltonian has been parametrized as an electronic component of a polarizable force field. Coulomb and exchange potentials derived directly from the NDDO Hamiltonian in principle can be used with classical potentials, thus forming the basis for a new generation of efficiently applicable multipolar polarizable force fields. The new hpCADD Hamiltonian uses force-field-like atom types and reproduces the electrostatic properties (dipole moment, molecular electrostatic potential) and Koopmans' theorem ionization potentials closely, as demonstrated for a large training set and an independent test set of small molecules. The Hamiltonian is not intended to reproduce geometries or total energies well, as these will be controlled by the classical force-field potentials. In order to establish the hpCADD Hamiltonian as an electronic component in force-field-based calculations, we tested its performance in combination with the 3D reference interaction site model (3D RISM) for aqueous solutions. Comparison of the resulting solvation free energies for the training and test sets to atomic charges derived from standard procedures, exact solute-solvent electrostatics based on high-level quantum-chemical reference data, and established semiempirical Hamiltonians demonstrates the advantages of the hpCADD parametrization.

This presentation will describe how structural biology, molecular pharmacology, and medicinal chemistry studies can be combined with molecular modeling and chemoinformatics analyses for a more accurate description and prediction of structural determinants of protein-ligand binding, functional activity, and selectivity.The challenges and possibilities of structural chemogenomics studies will be discussed, including the integration of large volumes of heterogeneous pharmacological and chemical data for different protein targets and the development of structure-based virtual screening and computer-aided drug design approaches to discover novel small molecule ligands with well defined functional activity and protein selectivity profiles.The potential of molecular dynamics simulation methods to complement hybrid structural biology studies will be demonstrated for the investigation the mechanisms of conformational selection and protein-ligand binding kinetics.In the final part of the presentation structural protein-ligand interaction databases will be described that link structure-based protein-ligand interaction maps to protein ligand topology and can be used as structural chemogenomics tools to navigate medicinal chemistry space.

Abstract A major constraint for the broad and safe application of Adoptive Cellular Therapy (ACT) is the limited number of validated tumor targets, especially for solid tumors. For T-cell receptor (TCR)-based approaches, presentation of targeted HLA-peptides on normal tissues can lead to on-target toxicity, such as severe inflammatory colitis reported upon re-directing T cells to an HLA-A*02 restricted carcinoembryonic antigen (CEA) epitope. Independently, off-target cross-reactivity of TCRs occurred in previous ACT trials, e.g. when a MAGEA3-directed TCR cross-recognized an HLA-A*01 restricted epitope from titin expressed on heart, which led to fatal cardiac toxicities. Here we present a novel approach allowing the prediction of severe on- and off-target side effects before entering into clinical trials. We used a target discovery engine (XPRESIDENT) combining highly sensitive, quantitative mass spectrometry (LC-MS/MS), RNA-Seq-based differential transcriptomics, immunology and bioinformatics to characterize the human immunopeptidome directly on shock frozen primary human tissues. Over the last years we have built an according database for > 600 tumor samples from > 20 different tumor types and, importantly, > 300 samples from > 40 different normal tissue types, resulting in hundreds of thousands of unique HLA-peptide sequences. These data allow conclusions on which HLA peptides are actually presented on primary normal tissues in a quantitative manner, taking into account relative differences between normal tissues and tumors as well as absolute peptide copy numbers per cell. In order to assess the off-target risk for a TCR, we predict all theoretical HLA- and TCR-binding peptides in the proteome, ideally based on the binding motif of the TCR, and specifically search for actual peptide presentation by normal tissues. When analyzing the above described CEA case, we were able to detect the CEA-derived peptide IMIGVLVGV on HLA-A*02 positive colorectal cancer samples, but importantly also on normal colorectal samples. In the original study describing the titin case tremendous experimental efforts and sophisticated cell culture models were required to retrospectively identify cross-recognition of the peptide on cardiomyocytes as the cause of toxicity. In contrast, with our approach we easily and directly identified the critical peptide ESDPIVAQY as one of the most abundantly presented peptides on an HLA-A*01 positive primary human heart sample. We show that this approach can lead to noteworthy results also for other pre-clinical and clinical stage TCR candidates. In conclusion our data demonstrate that ultrasensitive LC-MS/MS of primary tissue may represent a fast, straightforward and meaningful complementary method to common in vitro or animal models for the prediction of on- and off-target toxicities in TCR-based immunotherapy approaches. Citation Format: Oliver Schoor, Jens Fritsche, Sarah Kutscher, Andrea Mahr, Lea Stevermann, Annika Sonntag, Franziska Hoffgaard, Dominik Vahrenhorst, Julia Leibold, Valentina Goldfinger, Leonie Alten, Sebastian Bunk, Dominik Maurer, Steffen Walter, Hans-Georg Rammensee, Harpreet Singh-Jasuja, Toni Weinschenk. On- and off target toxicity profiling for adoptive cell therapy by mass spectrometry-based immunopeptidome analysis of primary human normal tissues. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 2291.

Receptor tyrosine kinases represent one of the prime targets in cancer therapy, as the dysregulation of these elementary transducers of extracellular signals, like the epidermal growth factor receptor (EGFR), contributes to the onset of cancer, such as non-small cell lung cancer (NSCLC). Strong efforts were directed to the development of irreversible inhibitors and led to compound CO-1686, which takes advantage of increased residence time at EGFR by alkylating Cys797 and thereby preventing toxic effects. Here, we present a structure-based approach, rationalized by subsequent computational analysis of conformational ligand ensembles in solution, to design novel and irreversible EGFR inhibitors based on a screening hit that was identified in a phenotype screen of 80 NSCLC cell lines against approximately 1500 compounds. Using protein X-ray crystallography, we deciphered the binding mode in engineered cSrc (T338M/S345C), a validated model system for EGFR-T790M, which constituted the basis for further rational design approaches. Chemical synthesis led to further compound collections that revealed increased biochemical potency and, in part, selectivity toward mutated (L858R and L858R/T790M) vs nonmutated EGFR. Further cell-based and kinetic studies were performed to substantiate our initial findings. Utilizing proteolytic digestion and nano-LC-MS/MS analysis, we confirmed the alkylation of Cys797.

The small K+ channel Kcv represents the pore module of complex potassium channels. It was found that its gating can be modified by sensor domains, which are N-terminally coupled to the pore. This implies that the short N-terminus of the channel can transmit conformational changes from upstream sensors to the channel gates. To understand the functional role of the N-terminus in the context of the entire channel protein, we apply combinatorial screening of the mechanical coupling and long-range interactions in the Kcv potassium channel by reduced molecular models. The dynamics and mechanical connections in the channel complex show that the N-terminus is indeed mechanically connected to the pore domain. This includes a long rang coupling to the pore and the inner and outer transmembrane domains. Since the latter domains host the two gates of the channel, the data support the hypothesis that mechanical perturbation of the N-terminus can be transmitted to the channel gates. This effect is solely determined by the topology of the channel; sequence details only have an implicit effect on the coarse-grained dynamics via the fold and not through biochemical details at a smaller scale. This observation has important implications for engineering of synthetic channels on the basis of a K+ channel pore.

The bridge function of a simple liquid is calculated for the repulsive part of the Weeks-Chandler-Andersen (WCA) separation of the Lennard-Jones potential. We employ explicit molecular dynamics simulations of the potential of mean force between constrained dimers in order to extract bridge data near zero separation and illustrate the difference to full Lennard-Jones results. We compare direct, reciprocal space and iterative, real space inversions of the Ornstein-Zernike equation. Bridge functions for various thermodynamic states are analyzed as to their parametric dependence on the renormalized indirect correlation function, which has consequences for the analytic representation of the free energy functional. (C) 2013 Elsevier B.V. All rights reserved.

Modeling solute polarizability is a key ingredient for improving the description of solvation phenomena. In recent years, polarizable molecular mechanics force fields have emerged that circumvent the limitations of classical fixed charge force fields by the ability to adapt their electrostatic potential distribution to a polarizing environment. Solvation phenomena are characterized by the solute's excess chemical potential, which can be computed by expensive fully atomistic free energy simulations. The alternative is to employ an implicit solvent model, which poses a challenge to the formulation of the solute-solvent interaction term within a polarizable framework. Here, we adapt the three-dimensional reference interaction site model (3D RISM) integral equation theory as a solvent model, which analytically yields the chemical potential, to the polarizable AMOEBA force field using an embedding cluster (EC-RISM) strategy. The methodology is analogous to our earlier approach to the coupling of a quantum-chemical solute description with a classical 3D RISM solvent. We describe the conceptual physical and algorithmic basis as well as the performance for several benchmark cases as a proof of principle. The results consistently show reasonable agreement between AMOEBA and quantum-chemical free energies in solution in general and allow for separate assessment of energetic and solvation-related contributions. We find that, depending on the parametrization, AMOEBA reproduces the chemical potential in better agreement with reference quantum-chemical calculations than the intramolecular energies, which suggests possible routes toward systematic improvement of polarizable force fields.

The validity and accuracy of a proposed tertiary structure of a protein can be assessed in several ways. Scoring such a structure by a knowledge‐based potential is a well‐known approach in molecular biophysics, an important task in structure prediction and refinement, and a key step in several experiments on protein structures. Although several parameterizations for such models have been derived over the course of time, improvements in accuracy by explicitly using continuous distance information have not been suggested yet. We close this methodological gap by formulating the parameterization of a protein structure model as a linear program. Optimization of the parameters was performed using amino acid distances calculated for the residues in topology rich 2830 protein structures. We show the capability of our derived model to discriminate between native structures and decoys for a diverse set of proteins. In addition, we discuss the effect of reduced amino acid alphabets on the model. In contrast to studies focusing on binary contact schemes (without considering distance dependencies and proposing five symbols as optimal alphabet size), we find an accurate protein alphabet size to contain at least five symbols, preferably more, to assure a satisfactory fold recognition capability. © 2012 Wiley Periodicals, Inc.

Many papers on parallel random permutation algorithms assume the input size n to be a power of two and imply that these algorithms can be easily generalized to arbitrary n. We show that this simplifying assumption is not necessarily correct since it may result in a bias. Many of these algorithms are, however, consistent, i.e., iterating them ultimately converges against an unbiased permutation. We prove this convergence along with proving exponential convergence speed. Furthermore, we present an analysis of iterating applied to a butterfly permutation network, which works in-place and is well-suited for implementation on many-core systems such as GPUs. We also show a method that improves the convergence speed even further and yields a practical implementation of the permutation network on current GPUs.