METHOD FOR THE CHARACTERIZATION OF PEPTIDE:MHC BINDING POLYPEPTIDES

Abstract

The present invention relates to a method for the characterization of peptide:MHC binding polypeptides, e.g. by mass spectrometry and an analysis of the recognized peptide space, i.e. in order to identify peptides that can be bound in the context of their presentation by MHC, and those who cannot be bound.

Claims

1. A method for identifying the binding of a polypeptide molecule comprising at least one defined peptide binding domain to a peptide of a peptide:MHC complex, comprising a) providing a sample comprising at least one peptide:MHC complex to be analyzed, b) contacting said sample with said polypeptide molecule, wherein said molecule is optionally attached to a matrix material, and allowing said at least one peptide binding domain of said polypeptide molecule to bind, optionally specifically, to said at least one peptide:MHC complex, wherein the amino acid sequence of said peptide binding domain is or is derived from a T-cell receptor (TCR), a T-cell receptor-like polypeptide, and/or an antibody binding domain, and optionally, wherein said polypeptide molecule further comprises at least one attachment site binding to or being attached to said matrix material, c) isolating said at least one peptide:MHC complex bound to said at least one peptide binding domain, and d) identifying said peptide of said at least one peptide:MHC complex as isolated in step c), and thereby identifying the binding of said polypeptide molecule to said peptide of said at least one peptide:MHC complex.

2. The method according to claim 1, wherein said sample is selected from natural or artificial samples comprising at least one peptide:MHC complex, optionally a cellular lysate, or a sample comprising purified or enriched peptide:MHC complexes.

3. The method according to claim 1, wherein said polypeptide molecule comprising said at least one peptide binding domain is selected from bispecific, trispecific, tetraspecific or multispecific molecules.

4. The method according to claim 1, wherein said polypeptide molecules comprising said at least one peptide binding domain are molecules or are derived from molecules selected from a simultaneous multiple interaction T-cell engaging (SMITE) bispecific, a bispecific T-cell engager (BiTE), an scFV, a diabody, a dual-affinity re-targeting antibody (DART), a tandem antibody (TandAb), a soluble TCR, an scTCR, a mutated TCR, for example comprising S-bridges, a truncated TCR, and a bispecific T-cell receptor (TCR)-antibody fusion molecule.

5. The method according to claim 1, wherein said polypeptide molecule comprises at least one second binding domain that is selected from a domain binding to a cell surface molecule known to induce the activation of immune cells, and immune response-related molecules, CD3, such as the CD3, CD3, and CD3 chains, CD4, CD7, CD8, CD10, CD11b, CD11c, CD14, CD16, CD18, CD22, CD25, CD28, CD32a, CD32b, CD33, CD41, CD41b, CD42a, CD42b, CD44, CD45RA, CD49, CD55, CD56, CD61, CD64, CD68, CD94, CD90, CD117, CD123, CD125, CD134, CD137, CD152, CD163, CD193, CD203c, CD235a, CD278, CD279, CD287, Nkp46, NKG2D, GITR, FcRI, TCR/, TCR/, and HLA-DR.

6. The method according to claim 1, wherein said polypeptide molecule comprising said at least one peptide binding domain is a bispecific molecule comprising a peptide binding domain that is derived from a T cell-receptor (TCR).

7. The method according to claim 1, wherein said attachment site is positioned in the at least one binding domain, the at least one second domain or is a separate attachment group, and does not interfere, at least not essentially, with the binding of said molecule.

8. The method according to claim 1, wherein said identifying in step d) comprises a method selected from mass spectrometry and peptide sequencing.

9. The method according to claim 1, further comprising identifying a consensus peptide binding motif for said peptide:MHC binding domain.

10. The method according to claim 9, wherein a specific peptide binding motif is identified, and/or off target binding motifs are identified.

11. The method according to claim 1, further comprising the step of identifying the cross-reactivity for said peptide:MHC binding domain.

12. The method according to claim 1, further comprising the step of identifying the presentation of said peptide motif or peptide motifs on cancerous and/or non-cancerous cells or tissues.

13. The method according to claim 1, wherein said method further comprises adding to said sample at least one peptide having a known sequence and/or at least one defined and/or preselected peptide:MHC complex, optionally in a predetermined amount (spiking)

14. The method according to claim 1, wherein said matrix material is selected from sepharose or agarose.

15. The method according to claim 1, wherein said method further comprises contacting said sample in step c) with other binding domain molecules, such as, for example, broad specific TCRs and/or antibodies.

16. The method according to claim 1, wherein at least one molecule in said method comprises a detectable marker or label.

17. The method according to claim 1, further comprising a computational analysis of said identification and/or off target binding.

18. A kit, comprising materials for performing the method according to claim 1, wherein the materials comprise a) a matrix material, and b) polypeptide molecules comprising said at least one peptide binding domain binding to at least one peptide:MHC complex.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0078] The patent or application file contains at least one drawing executed in color. Copies of this patent or patent application publication with color drawing(s) will be provided by the Office upon request and payment of the necessary fee.

[0079] FIG. 1: Schematic overview of the experimental approach according to the present invention. A sample containing peptide:MHC molecules is provided for example by generating a lysate of peptide:MHC expressing cells derived from a tissue or cell line. Alternatively, the sample can be modified by addition of or constituted by mixtures of artificially produced peptide:MHC molecules. Specific peptide:MHC molecules are isolated from this sample for example by contacting it to a matrix, which has a peptide:MHC binding polypeptide attached to it. Glycine columns which do not contain an attached polypeptide maybe used to deplete the sample from non-specifically isolated peptide:MHC molecules, which interact non-specifically with the matrix. Mass spectrometry can be used to identify the MHC bound peptides, which have been isolated from the sample to identify and sequence the peptide space recognized by the peptide:MHC binding polypeptide. Thereby previously unknown off-target risks can be elucidated, without the need of prediction tools. In a modification of this approach the same peptide:MHC molecule containing sample can be split and in parallel be subjected to a second affinity chromatography with HLA broad specific antibodies or TCRs that bind to these molecules irrespective of the bound peptide species. In the illustrated example the HLA-A*02 pan specific antibody BB7.2 is employed to isolate all peptides presented by HLA-A*02 irrespective of the peptide sequence, herein referred to as the HLA-A*02 immunopeptidome. The abundance of interesting (e.g. off-target) peptides in both isolations can be used to assess the binding affinity of the peptide:MHC binding polypeptide for said peptide in comparison to the employed broad specific antibody or TCR, in this example BB7.2.

[0080] FIG. 2: Cytotoxicity experiments showing killing of a target positive (U2OS) and a target negative (T98G) cell line with two variants of the PRAME-004 peptide:MHC binding polypeptide (black rectangles: original variant, white dots: specificity improved variant following an additional round of maturation of the peptide:MHC specific binder directed against PRAME-004 using the identified off-target peptides as selection determinants). Killing of the target negative cell line T98G is strongly reduced when employing the specificity-improved variant of the peptide:MHC binding polypeptide, whereas killing of the target positive cell line U2OS is only slightly affected.

[0081] FIG. 3: The combination of several cell-lines for sample generation can be used to achieve a high coverage of the HLA-A*02 presented immunopeptidome. Based on XPRESIDENT immunopeptidome data for 60 cell lines it is shown that a combination of already 10 of these cell lines would enable a coverage of more than 60% of the HLA-A*02 immunopeptidome, if peptides are to be considered, for which at least 10 peptide identifications on normal tissues have been previously detected.

[0082] FIG. 4: Cytotoxicity analysis of all peptides identified in the analysis using the PRAME-004 specific peptide:MHC binding polypeptide from Example 1. In brief T2 cell loaded with 10 nM of respective peptides were co-incubated with human CD8+ T-cells in the presence of indicated concentrations of the PRAME-004 specific peptide:MHC binding polypeptide. After 48 h cytotoxicity was quantified by measuring LDH release.

[0083] FIG. 5: Analysis of the expression profile of the peptide encoding source exon of IFT17-003 in different normal as well as tumor tissues. Tumor (black dots) and normal (grey dots) samples are grouped according to organ of origin. Box-and-whisker plots represent median FPKM value, 25th and 75th percentile (box) plus whiskers that extend to the lowest data point still within 1.5 interquartile range (IQR) of the lower quartile and the highest data point still within 1.5 IQR of the upper quartile. Normal organs are ordered alphabetically. FPKM: fragments per kilobase per million mapped reads. Tissues (from left to right): Normal samples: adipose (adipose tissue); adrenal gl (adrenal gland); bile duct; bladder; blood cells; bloodvess (blood vessels); bone marrow; brain; breast; esoph (esophagus); eye; gall bl (gallbladder); head&neck; heart; intest. la (large intestine); intest. sm (small intestine); kidney; liver; lung; lymph node; nerve periph (peripheral nerve); ovary; pancreas; parathyr (parathyroid gland); perit (peritoneum); pituit (pituitary); placenta; pleura; prostate; skel. mus (skeletal muscle); skin; spinal cord; spleen; stomach; testis; thymus; thyroid; trachea; ureter; uterus. Tumor samples: AML (acute myeloid leukemia); BRCA (breast cancer); CCC (cholangiocellular carcinoma); CLL (chronic lymphocytic leukemia); CRC (colorectal cancer); GBC (gallbladder cancer); GBM (glioblastoma); GC (gastric cancer); HCC (hepatocellular carcinoma); HNSCC (head and neck squamous cell carcinoma); MEL (melanoma); NHL (non-hodgkinHodgkin lymphoma); NSCLCadeno (non-small cell lung cancer adenocarcinoma); NSCLCother (NSCLC samples that could not unambiguously be assigned to NSCLCadeno or NSCLCsquam); NSCLCsquam (squamous cell non-small cell lung cancer); OC (ovarian cancer); OSCAR (esophageal cancer); PACA (pancreatic cancer); PRCA (prostate cancer); RCC (renal cell carcinoma); SCLC (small cell lung cancer); UBC (urinary bladder carcinoma); UEC (uterine endometrial cancer).

[0084] FIG. 6: Analysis of the peptide presentation of IFT17-003 on different normal as well as tumor tissues. Upper part: Median MS signal intensities from technical replicate measurements are plotted as dots for single HLA-A*02 positive normal (grey dots, left part of figure) and tumor samples (black dots, right part of figure) on which the peptide was detected. Boxes display median, 25th and 75th percentile of normalized signal intensities, while whiskers extend to the lowest data point still within 1.5 interquartile range (IQR) of the lower quartile, and the highest data point still within 1.5 IQR of the upper quartile. Normal organs are ordered alphabetically. Lower part: The relative peptide detection frequency in every organ is shown as spine plot. Numbers below the panel indicate number of samples on which the peptide was detected out of the total number of samples analyzed for each organ (N=592 for normal samples, N=710 for tumor samples). If the peptide has been detected on a sample but could not be quantified for technical reasons, the sample is included in this representation of detection frequency, but no dot is shown in the upper part of the figure. Tissues (from left to right): Normal samples: adipose (adipose tissue); adrenal gl (adrenal gland); bile duct; bladder; blood cells; bloodvess (blood vessels); bone marrow; brain; breast; esoph (esophagus); eye; gall bl (gallbladder); head&neck; heart; intest. la (large intestine); intest. sm (small intestine); kidney; liver; lung; lymph node; nerve cent (central nerve); nerve periph (peripheral nerve); ovary; pancreas; parathyr (parathyroid gland); perit (peritoneum); pituit (pituitary); placenta; pleura; prostate; skel. mus (skeletal muscle); skin; spinal cord; spleen; stomach; testis; thymus; thyroid; trachea; ureter; uterus. Tumor samples: AML (acute myeloid leukemia); BRCA (breast cancer); CCC (cholangiocellular carcinoma); CLL (chronic lymphocytic leukemia); CRC (colorectal cancer); GBC (gallbladder cancer); GBM (glioblastoma); GC (gastric cancer); GEJC (gastro-esophageal junction cancer); HCC (hepatocellular carcinoma); HNSCC (head and neck squamous cell carcinoma); MEL (melanoma); NHL (non-hodgkinHodgkin lymphoma); NSCLCadeno (non-small cell lung cancer adenocarcinoma); NSCLCother (NSCLC samples that could not unambiguously be assigned to NSCLCadeno or NSCLCsquam); NSCLCsquam (squamous cell non-small cell lung cancer); OC (ovarian cancer); OSCAR (esophageal cancer); PACA (pancreatic cancer); PRCA (prostate cancer); RCC (renal cell carcinoma); SCLC (small cell lung cancer); UBC (urinary bladder carcinoma); UEC (uterine endometrial cancer).

[0085] FIG. 7: Binding motif of the PRAME-004 directed peptide:MHC binding polypeptide determined using the described method. The size of individual amino acids in selected positions reflects their abundance among the identified off-targets. For example, within the identified off-targets histidine (H) in Position 5 of the peptide sequence was more frequent as compared to lysine (K).

[0086] FIG. 8: Identification of the binding motif of the peptide:MHC binding polypeptide directed against PRAME-004 using positional scanning, replacing each amino acid in position 1-9 of the peptide sequence with alanine. The ratio of the KD of the target peptide PRAME-004 to the alanine substituted variant of the peptide sequence is presented for every peptide. A threshold of 50% (dashed line) of the KD ratio is applied to determine positions which are recognized by the binder. KD values were determined by bio-layer interferometry.

[0087] FIG. 9: Complex binding motif determination using amino acid substitutions with all proteinogenic amino acids per position (except cysteine). Ratios of the KD of the target peptide PRAME-004 to the respective positional scanning variant are represented and greyscale-coded (showing low to high values colored from white to dark grey). KD values were determined by bio-layer interferometry.

[0088] FIG. 10: Analysis of the expression profile of the peptide encoding source exon of MAGEA1 in different normal as well as tumor tissues. For a detailed figure description please refer to the legend of FIG. 5.

[0089] FIG. 11: Analysis of the peptide presentation of MAGEA1 on different normal as well as tumor tissues. For a detailed figure description please refer to the legend of FIG. 6.

DETAILED DESCRIPTION

Examples

Example 1

[0090] The targeted MHC peptide used in this example which is presented in the context of HLA-A*02 is derived from Melanoma antigen preferentially expressed in tumors (PRAME) and shows the sequence SLLQHLIGL ((SEQ ID NO: 1), herein also referred to as PRAME-004).

[0091] The peptide:MHC binding polypeptide was exemplified by a modified T-cell receptor molecule which has been engineered to be soluble and showed an enhanced affinity to the PRAME-004 peptide and additionally comprised a CD3-binding antibody moiety.

[0092] As a biological source of peptide:MHC mixtures, the human HLA-A*02 high expressing glioblastoma derived cell line T98G was used. This cell line had been previously tested in cytotoxicity experiments with the described peptide:MHC binding polypeptide directed against PRAME-004 and showed positive killing.

[0093] Five hundred million T98G cells were subjected to lysis in a CHAPS detergent-containing buffer and homogenized assisted by sonification.

[0094] The peptide:MHC binding polypeptide was coupled to a solid sepharose matrix at a pre-determined ratio using chemical coupling after BrCN activation. In parallel the same amount of sepharose was also activated for coupling using the same strategy, yet without addition of the peptide:MHC binding polypeptide. Instead a 0.1 M solution of the amino acid glycine was added to the sepharose, which instead coupled to the chemically activated groups. The T98G lysate containing the mixture of peptide:MHC molecules was then applied to two affinity chromatography columns loaded with 1 ml of the glycine coupled sepharose matrix or 1 ml of the sepharose matrix coupled with the peptide:MHC binding polypeptide. The T98G derived lysate was thereby applied in such a fashion that it would first be run over the glycine coupled sepharose (referred to herein as glycine column) to remove or deplete any peptides, which would bind non-specifically to the column or the sepharose matrix before the isolation of peptides which bind to the peptide:MHC binding polypeptide (FIG. 1). After washing of the affinity columns with PBS and ddH.sub.2O the bound peptide:MHC complexes were eluted from the columns using Trifluoroacetic acid (TFA).

[0095] During this step, MHC bound peptides are also released from the MHC moiety and can be separated from higher molecular weight molecules by ultrafiltration using specified devices with a molecular weight cutoff of less than 10 kDa.

[0096] The isolated peptide mixtures were then finally subjected to liquid chromatography coupled mass spectrometry (LC-MS) using a nanoACQUITY UPLC system (Waters) followed by an Orbitrap Fusion Tribrid mass spectrometer (Thermo Scientific).

[0097] Mass spectrometry instruments were operated in data-dependent mode (DDA) utilizing different fragmentation techniques (in this example, CID and HCD fragmentation) as well as MS/MS spectra readout in two different analyzers (in this example, IonTrap and Orbitrap analyzers).

[0098] Peptide fragment spectra were searched against the human proteome using a modified version of the International protein index (IPI v.378) and the Universal protein resource (UniProt) sequence database with the search engine SEQUEST. All peptides eluted and identified from the glycine column were excluded from further analysis as these represent non-specific binding peptides. Furthermore, known contaminants according to in-house databases and algorithms for their identification were removed from the analysis.

[0099] In total 20 peptides were identified after isolation and processing which are shown in Table 2. For reference, the target peptide PRAME-004 is shown as well, which has however not been identified among the isolated peptides and was not expected to be identified.

[0100] In order to confirm their relevance and analyze the binding strength in comparison to the target peptide, all peptides were subjected to bio-layer interferometry. Measurements were performed on an Octet RED384 system using settings recommended by the manufacturer. Briefly, binding kinetics were measured at 30 C. and 1000 rpm shake speed using PBS, 0.05% Tween-20, 0.1% BSA as buffer. Peptide:MHC complexes were loaded onto biosensors (HIS1 K) prior to analyzing serial dilutions of the peptide:MHC specific binder. The ratio of equilibrium dissociation constants (KD) in comparison to PRAME-004 is presented in the last column of Table 2.

[0101] A selection of these peptides was further tested in cytotoxicity experiments. In brief T2 cell (10,000 cells/well) loaded with 10 nM of respective peptides were co-incubated with human CD8+ T cells (50,000 cells/well) in the presence of indicated concentrations of the PRAME-004 specific peptide:MHC binding polypeptide (FIG. 4). After 48 h cytotoxicity was quantified by measuring LDH release using CytoTox 96 Non-Radioactive Cytotoxicity Assay Kits (PROMEGA). Corresponding EC50 values of tested peptides are also listed in Table 2. The main off-target emanating from this analysis was IFT17-003 which showed similar KD and EC50 values for the peptide:MHC binding polypeptide as compared to the target peptide PRAME-004.

[0102] XPRESIDENT peptide presentation and gene expression data can be used to evaluate the potential safety risk of off-target peptides by differentiating relevant off-targets from less relevant off-targets, which are only presented/expressed in the context of other tumor tissues. In this example, IFT17-003 is considered a highly relevant off-target due to ubiquitous expression (FIG. 5) and presentation (FIG. 6) of the peptide on different normal tissues. Combination of the presented data from this example with additional large-scale peptide presentation or expression data is therefore of additional value for the off-target risk assessment.

[0103] In order to improve the specificity of the peptide:MHC binding polypeptide an additional round of maturation has been performed using the identified peptides as selection determinants. Thereby specificity of the newly generated molecules could be greatly improved shown by a reduced killing of the target negative cell line T98G in FIG. 2. Killing assays were essentially performed as described above. LDH release of target positive or target negative cells (10,000 cells/well) was quantified after co-incubation with human PBMCs (100,000 cells/well) and indicated concentrations of the peptide:MHC binding polypeptide for 48 h. The original peptide:MHC binding polypeptide molecule used for peptide isolation is shown as filled squares, whereas specificity improved variants are shown as open circles. Control peptide:MHC binding polypeptide molecule (square with asterisk) and control without bispecific molecule (circle with asterisk) do not induce target cell killing.

TABLE-US-00003 TABLE2 Overviewoftheidentifiedpeptide:MHCbinding polypeptidespecificpeptides.Indicatedontop isthetargetPRAME-004.TheEC50valuesof cytotoxicityexperimentsusingpeptideloaded T2cellsisspecifiedaswellasthebinding affinitiesdeterminedbybio-layer interferometryusingHIS1Kbiosensors. Binding affinity EC50 fold Peptidecode Peptidesequence [pM] reduction PRAME-004 SLLQHLIGL 1.2 1 (SEQIDNO:1) IFT17-003 FMNPHLISV 1.6 1 (SEQIDNO:2) MCM5-006 MLAKHVITL 16.1 3 (SEQIDNO:3) IFIT1-001 VLLHHQIGL 38.4 8 (SEQIDNO:4) FADS2-001 LLLAHIIAL 83.7 13 (SEQIDNO:5) CTBP1-001 ALMYHTITL 79.31 13 (SEQIDNO:6) ITSN1-001 ILAMHLIDV 1024 36 (SEQIDNO:7) ATP1A1-001 FLPIHLLGL 196 106 (SEQIDNO:8) MCMB-002 YLILHLIST n.a. 127 (SEQIDNO:9) EHD4-001 ALAKHLIKI n.a. 61 (SEQIDNO:10) 5F3B3-005 TLVYHVVGV n.a. 152 (SEQIDNO:11) EHD-001 ALANHLIKV n.a. 159 (SEQIDNO:12) FARSA-001 LTLGHLMGV n.a. 38 (SEQIDNO:13) INT57-002 ILGTHNITV n.a. 57 (SEQIDNO:14) MLXI-001 KLTSHAITL n.a. 12 (SEQIDNO:15) PPP4R1-003 HLMPHLLTL n.a. 16 (SEQIDNO:16) RIF1-004 AIWEKLISL n.a. 156 (SEQIDNO:17) SFXN3-001 SLTKHLPPL n.a. 60 (SEQIDNO:18) TBCK-002 ALSPHNILL n.a. 142 (SEQIDNO:19) TNRC6B-001 SLARHLMTL n.a. 4 (SEQIDNO:20) ZFYVE16-002 ALCPHLKTL n.a. 33 (SEQIDNO:21)

Identification of a Binding Motif

[0104] The identified peptides can be further used to infer a binding motif for the peptide:MHC binding polypeptide, which provides information on which of the amino acids in the peptide sequence are of relevance for the binding of said polypeptide.

[0105] Moreover, additional information on the binding motif can be deduced from the amino acids within the relevant positions. Based on the identified set of peptides only a subset of amino acids is tolerated in positions 1-9 of the amino acid sequence (see FIG. 7 and Table 3).

TABLE-US-00004 TABLE 3 Overview of the tolerated amino acids for each position identified by the presented method. Position Tolerated amino acid residues 1 A, S, F, I, L, H, K, M,T, V, Y 2 L, I, M,T 3 A, L, M, T, C, G, I, N, P, S, V, W 4 P, K, Y, A, E, G, H, I, L, M, N, Q, R, S, T 5 H, K 6 L, N, V, A, I, Q, T 7 I, L, M, K, P, V 8 T, G, S, K, A, D, L, P 9 L, V, I, T

[0106] In contrast to common amino acid scanning approaches, in which amino acids are replaced at individual positions by mutations and subsequently tested in in vitro assays also multiple substitutions with different potentially opposite effects on the overall binding strength can be elucidated. For example, if a substitution in position 6 of the natural amino acid sequence leads to a decrease in the overall binding affinity this might be rescued by a similar substitution in position 8 which can lead to a strong increase in the binding affinity of the peptid:MHC binding polypeptide to the peptide:MHC molecule.

[0107] The thus generated binding motif was used to search different protein sequence databases (e.g. UniProt, IPI) to find additional off-target peptides which reflect and fit to the restrictions imposed by the binding motif (i.e. defined sets of amino acids which are tolerated in relevant positions of the binding motif).

Comparative Example 1

[0108] The following experiments show how currently available methods in the art would not identify the most relevant off-target peptides identified in example 1 and are therefore not able to predict unwanted side effects of peptide:MHC binding polypeptides intended for administration in vivo.

Identification of the Binding Motif Using Positional Scanning:

[0109] Variants of the native PRAME-004 sequence, in which each amino acid is subsequently replaced with the amino acid alanine were tested for their potential to bind to the peptide:MHC binding polypeptide using bio-layer interferometry.

TABLE-US-00005 (SEQIDNO.38) ALLQHLIGL (SEQIDNO.39) SALQHLIGL (SEQIDNO.40) SLAQHLIGL (SEQIDNO.41) SLLAHLIGL (SEQIDNO.42) SLLQALIGL (SEQIDNO.43) SLLQHAIGL (SEQIDNO.44) SLLQHLAGL (SEQIDNO.45) SLLQHLIAL (SEQIDNO.46) SLLQHLIGA

[0110] FIG. 8 shows the results of these experiments. Five of the alanine-substituted peptides lead to a 50% or greater decrease in binding affinity (or 2-fold or greater increase of the KD, respectively) as compared to the wild-type sequence and were therefore considered as essential for binding. Based on these results the binding motif would result in XXXXHLIGL (SEQ ID NO. 22), wherein X represents any amino acid.

[0111] In an extended variant of the positional scanning approach the PRAME-004 sequence was substituted at each position by any of the naturally occurring amino acids in a similar manner as described before. The only proteinogenic amino acid, which was not used for substitution of PRAME-004 was cysteine as this amino acid is known to rapidly undergo several chemical modifications which can lead to false interpretations regarding recognition of peptides during testing. So, in total 9*18=162 peptides were investigated.

[0112] Each peptide was again tested for its binding affinity using bio-layer interferometry (FIG. 9). Peptides which lead to a 50% or greater decrease in binding affinity (or 2-fold or greater increase of the KD, respectively) as compared to the wild-type sequence were considered as not tolerated or detrimental for peptide binding. This resulted in a complex binding motif with a set of different amino acids being tolerated or accepted in position 1-9 of the amino acid sequence:

TABLE-US-00006 X.sub.1X.sub.2X.sub.3X.sub.4HX.sub.5IX.sub.6X.sub.7
wherein X.sub.1 is selected from any of ADEFGHIKLMNPQRSTVWY; X.sub.2 is selected from any of AFGILMQSTVY; X.sub.3 is selected from any of ADGIKLMNQSTVW; X.sub.4 is selected from any of AFGHIKLMNPQRSTVWY; X.sub.5 is selected from any of ILM; X.sub.6 is selected from any of GST; and X.sub.7 is selected from any of EFHIKLMPQTVY.

Similarity Search Based on Ala-Scan Derived Binding Motif:

[0113] An in-house software tool was used to search different protein sequence databases (IPI v. 3.78, Ensembl Version 77 GrCH38 including SNVs, NCBI non-redundant protein database) for human proteins which contain the identified motif sequence (X-X-X-X-X-H-L-I-G-L) (SEQ ID NO. 22), in which X could be constituted by any amino acid. Eight unique peptides were identified: the target itself, PRAME-004 and seven peptides originating from different human proteins: VEZT (Vezatin), HTR2C (5-hydroxytryptamine receptor 2C), HEPHL1 (Hephaestin-like protein 1), COL28A1 (Collagen alpha-1(XXVIII) chain), SLC2A1 (Solute carrier family 2, facilitated glucose transporter member 1), SLC44A3 (Choline transporter-like protein 3), PIEZO2 (Piezo-type mechanosensitive ion channel component 2).

[0114] The amino acid sequences of these peptides are shown below.

TABLE-US-00007 Protein (Uniprot accession number) Sequence PRAME (P78395) S L L Q H L I G L (SEQ ID NO. 1) VEZT (Q9HBM0) H P S Q H L I G L (SEQ ID NO. 23) HTR2C (P28335) S F L V H L I G L (SEQ ID NO. 32) HEPHL1 (Q6MZM0) R V S W H L I G L (SEQ ID NO. 33) COL28A1 (Q2UY09) I N E S H L I G L (SEQ ID NO. 34) SLC2A1 (P11166) R R T L H L I G L (SEQ ID NO. 35) SLC44A3 (Q8N4M1) M W S Y H L I G L (SEQ ID NO. 36) PIEZO2 (Q9H5I5) F T A G H L I G L (SEQ ID NO. 37)

[0115] In-house XPRESIDENT immunopeptidome data from 592 normal tissue samples and 710 tumor tissue samples, all derived from HLA-A*02 typed individuals, showed that none of the 7 predicted off-target peptides, has ever been identified to be presented in the context of HLA-A*02 on any of the analyzed samples. Notably, the VEZT and SLC2A1 derived peptide have been previously identified by XPRESIDENT on tissue samples of non-A*02 positive individuals suggesting that they are presented by different HLA allotypes (HLA-B*07 in case of the VEZT derived peptide and HLA-B*27 in case of the SLC2A1 derived peptide) and are therefore not likely to produce an off-target risk in the context of an HLA-A*02 restricted peptide:MHC binding polypeptide. The positional scanning and prediction approach failed however to identify any of the relevant off-target peptides which could be identified with the superior method described in this application.

Similarity Search Based on the Complex Binding Motif:

[0116] The same in-house software tool was also used to predict peptides derived from the human proteome which fulfill the criteria of the complex binding motif. As cysteine was excluded during substitution, this amino acid was hence additionally allowed for every position in the amino acid sequence resulting in the following motif:

TABLE-US-00008 X.sub.8X.sub.9X.sub.10X.sub.11X.sub.12X.sub.13X.sub.14X.sub.15X.sub.16
wherein X.sub.8 is selected from any of ACDEFGHIKLMNPQRSTVWY; X.sub.9 is selected from any of ACFGILMQSTVY; X.sub.10 is selected from any of ACDGIKLMNQSTVW; X.sub.11 is selected from any of ACFGHIKLMNPQRSTVWY; X.sub.12 is selected from any of CH; X.sub.13 is selected from any of CILM; X.sub.14 is selected from any of CI; X.sub.15 is selected from any of CGST; and X.sub.16 is selected from any of CEFHIKLMPQTVY.

[0117] The search resulted in a total list of 888 different peptides fulfilling the binding motif criteria. Only two peptides (IFT17-003 and ATP1A1-001) were overlapping with the list of relevant off-targets identified by the superior method described in this application in example 1, whereas the rest would not have been identified in the prediction-based approach, even if all 888 peptides would have been tested afterwards in downstream in vitro analyses.

Example 2

[0118] The targeted MHC peptide used in this example which is presented in the context of HLA-A*02 is derived from the melanoma associated antigens A4 and A8 (MAGEA4/A8) and shows the sequence KVLEHVVRV (SEQ ID NO. 24), herein also referred to as MAGEA4/8.

[0119] The peptide:MHC binding polypeptide is constituted by a modified T-cell receptor molecule which has been engineered to be soluble and shows an enhanced affinity to the MAGEA4/A8 derived peptide and additionally comprised a CD3 binding antibody moiety. As a biological source of peptide:MHC mixtures the human HLA-A*02 high expressing and MAGA4/8 positive lung adenocarcinoma derived cell line NCI-H1755 has been employed. Five hundred million cells of this cell line were subjected to lysis in a CHAPS detergent-containing buffer and homogenized assisted by sonification.

[0120] Coupling of the peptide:MHC binding polypeptide and affinity chromatography were carried out as described in Example 1. Before applying the NCI-H1755 lysate containing the mixture of peptide:MHC molecules to the glycine coupled and peptide:MHC binder coupled sepharose the volume was split half and half. The second half of the volume was run in parallel over a different glycine coupled sepharose matrix followed by a sepharose matrix coupled with the HLA-A*02 specific antibody BB7.2. The latter is aimed to isolate the complete spectrum of peptides presented by HLA-A*02 in this cell line (see also FIG. 1).

[0121] Peptides were eluted from all columns and subjected to mass spectrometry analysis as outlined in Example 1. Peptides eluted from glycine columns as well as known contaminants were again excluded from further analysis. In addition, all peptide precursor signals were quantified over all different runs using SuperHirn algorithm (Mueller et al., 2007). Features were extracted and quantified over all mass spectrometry experiments using a fixed retention time window of 3 min, and a mass accuracy of 5 ppm.

[0122] Ratios of the resulting area of individual peptide precursor signals from the MAGA4/8 specific peptide:MHC binding polypeptide to the same precursor signals from the BB7.2 preparation were calculated. These ratios reflect the isolation efficiency of the MAGA4/8 peptide:MHC binding polypeptide in comparison to the HLA-A*02 specific antibody BB7.2. Due to the high affinity of the MAGA4/8 specific peptide:MHC binding polypeptide for their target as well as for potential off-targets peptides bound to the HLA-A*02 molecule, the isolation efficiencies for these peptides are much higher as compared to BB7.2 which has an affinity in the lower nanomolar range towards HLA-A*02 largely independent of the bound peptide species (Parham and Brodsky, 1981). Analysis of the mass spectrometry data identified 10 peptides including the target peptide MAGA4/8 (see Table 4). Ranking of these peptides according to the ratio of areas of the peptide:MHC binding polypeptide and BB7.2 enables a determination of the isolation efficiency of these peptides in comparison to BB7.2 which correlates with the binding affinity employing bio-layer interferometry as described in Example 1. Thereby the risk for off-target toxicities and potential therapeutic windows between target and off-target peptides can be directly deduced from the quantitative data of the mass spectrometry experiments.

[0123] In the presented example in Table 4 the ratio of areas for MAGEA1 is smaller as compared to the target peptide MAGA4/8 (11 as compared to 10.6), which translates into a very small reduction of binding affinity of 4.1. In contrast, for the peptide HEAT5RA the large decrease in the ratio of areas, around 800-fold lower as compared to MAGA4/8, is also reflected in a largely reduced binding affinity of 238 as compared to MAGA4/8.

[0124] Deeper analysis of peptide presentation and gene expression data in XPRESIDENT show that MAGEA1 does not present a relevant off-target risk, as it is exclusively presented on cancer tissues (FIG. 10) and shows a cancer-testis like expression pattern (FIG. 11).

TABLE-US-00009 TABLE4 Overviewoftheidentifiedpeptide:MHCbinding polypeptidespecificpeptides.Indicatedisthe ratioofpeptideselutedfromtheMAGA4/8binding polypeptidetotheHLA-A*02specificbinding peptideBB7.2bymassspectrometry.Inthetop rowthetargetpeptideMAGA4/A8ispresented showingthehighestratioofsignalareas.The PMBECscoreisameasureforpeptidesimilarity tothetargetsequence.Thebindingaffinities weredeterminedbybio-layerinterferometry usingHIS1Kbiosensors. Ratioof signal areas Binding [MAGA4/8 affinity Peptide Peptide binder/ fold code sequence BB7.2] PMBEC reduction MAGEA4/ KVLEHVVRV 11 1.85568 1 A8 (SEQIDNO.24) MAGEA1 KVLEYVIKV 10.615 1.42928 4.1 (SEQIDNO.25) KVLEFLAKV 5.218 1.24828 (SEQIDNO.26) KIIDLLPKV 4.009 0.94598 (SEQIDNO.27) KLQEFLQTL 0.032 0.45866 (SEQIDNO.28) HEAT5RA KVLETLVTV 0.014 1.06691 237.7 (SEQIDNO.29) FAM115A KLGSVPVTV 0.006 0.50865 503.1 (SEQIDNO.30) KIADFGWSV 0.002 0.53985 (SEQIDNO.31)

Abbreviations

[0125] APC Antigen presenting cells
BIRD Blackbody infrared radiative dissociation
BiTE Bispecific T-cell engager
CAR Chimeric antigen receptors
CDR Complementarity determining regions
CID Collision-induced dissociation
DART Dual-affinity re-targeting antibody
DDA Data-dependent acquisition
DIA Data-independent acquisition
DRIP Defective ribosomal particles
ECD Electron-capture dissociation
EDD Electron-detachment dissociation
ETCID Electron-transfer and collision-induced dissociation
ETD Electron-transfer dissociation

ETHCD Electron-Transfer/Higher-Energy Collision Dissociation

[0126] HCD Higher-energy collisional dissociation
IRMPD Infrared multiphoton dissociation
IQR Interquartile range
KD Dissociation constant
NETD Negative electron-transfer dissociation
LDH Lactate dehydrogenase
PBMC Peripheral blood mononuclear cell
SID Surface-induced dissociation
SMITE Simultaneous multiple interaction T-cell engaging
TandAb Tandem antibody
TCR T-cell receptor
TFA Trifluoroacetic acid
TIL Tumor-infiltrating lymphocytes

REFERENCES

[0127] Barnstable C J, Bodmer W F, Brown G, Galfre G, Milstein C, Williams A F, Ziegler A (1978). Production of monoclonal antibodies to group A erythrocytes, HLA and other human cell surface antigens-new tools for genetic analysis. Cell 14, 9-20. [0128] Berger A E, Davis J E, Cresswell P (1982). Monoclonal antibody to HLA-A3. Hybridoma 1, 87-90. [0129] Bijen H M, van der Steen D M, Hagedoorn R S, Wouters A K, Wooldridge L, Falkenburg JHF, Heemskerk MHM (2018). Preclinical Strategies to Identify Off-Target Toxicity of High-Affinity TCRs. Mol Ther 26, 1206-1214. [0130] Birnbaum M E, Mendoza J L, Sethi D K, Dong S, Glanville J, Dobbins J, Ozkan E, Davis M M, Wucherpfennig K W, Garcia K C (2014). Deconstructing the peptide-MHC specificity of T cell recognition. Cell 157, 1073-1087. [0131] Ekeruche-Makinde J, Miles J J, van den Berg H A, Skowera A, Cole D K, Dolton G, Schauenburg A J, Tan M P, Pentier J M, Llewellyn-Lacey S, et al. (2013). Peptide length determines the outcome of TCR/peptide-MHCI engagement. Blood 121, 1112-1123. [0132] Jahn L, van der Steen D M, Hagedoorn R S, Hombrink P, Kester M G, Schoonakker M P, de Ridder D, van Veelen P A, Falkenburg J H, Heemskerk M H (2016). Generation of CD20-specific TCRs for TCR gene therapy of CD20low B-cell malignancies insusceptible to CD20-targeting antibodies. Oncotarget 7, 77021-77037. [0133] Kolstad A, Hansen T, Hannestad K (1987). A human-human hybridoma antibody (TrB12) defining subgroups of HLA-DQw1 and -DQw3. Hum Immunol 20, 219-231. [0134] Lampson L A, Levy R (1980). Two populations of Ia-like molecules on a human B cell line. J Immunol 125, 293-299. [0135] Linette G P, Stadtmauer E A, Maus M V, Rapoport A P, Levine B L, Emery L, Litzky L, Bagg A, Carreno B M, Cimino P J, et al. (2013). Cardiovascular toxicity and titin cross-reactivity of affinity-enhanced T cells in myeloma and melanoma. Blood 122, 863-871 [0136] Maeda H, Hirata R (1984). Separation of four class II molecules from DR2 and DRw6 homozygous B lymphoid cell lines. Immunogenetics 20, 639-647. [0137] Morgan R A, Chinnasamy N, Abate-Daga D, Gros A, Robbins P F, Zheng Z, Dudley M E, Feldman S A, Yang J C, Sherry R M, et al. (2013). Cancer regression and neurological toxicity following anti-MAGE-A3 TCR gene therapy. J Immunother 36, 133-151. [0138] Mueller L N, Rinner O, Schmidt A, Letarte S, Bodenmiller B, Brusniak M Y, Vitek O, Aebersold R, Muller M (2007). SuperHirna novel tool for high resolution L C-M S-based peptide/protein profiling. Proteomics 7, 3470-3480. [0139] Parham P, Brodsky F M (1981). Partial purification and some properties of BB7.2. A cytotoxic monoclonal antibody with specificity for HLA-A2 and a variant of HLA-A28. Hum Immunol 3, 277-299. [0140] Raman M C, Rizkallah P J, Simmons R, Donnellan Z, Dukes J, Bossi G, Le Provost G S, Todorov P, Baston E, Hickman E, et al. (2016). Direct molecular mimicry enables off-target cardiovascular toxicity by an enhanced affinity TCR designed for cancer immunotherapy. Sci Rep 6, 18851. [0141] Rebai N, Mercier P, Kristensen T, Devaux C, Malissen B, Mawas C, Pierres M (1983). Murine H-2Dd-reactive monoclonal antibodies recognize shared antigenic determinant(s) on human HLA-B7 or HLA-B27 molecules or both. Immunogenetics 17, 357-370. [0142] Smith S N, Wang Y, Baylon J L, Singh N K, Baker B M, Tajkhorshid E, Kranz D M (2014). Changing the peptide specificity of a human T-cell receptor by directed evolution. Nat Commun 5, 5223. [0143] Spits H, Keizer G, Borst J, Terhorst C, Hekman A, de Vries J E (1983). Characterization of monoclonal antibodies against cell surface molecules associated with cytotoxic activity of natural and activated killer cells and cloned CTL lines. Hybridoma 2, 423-437. [0144] Stewart-Jones G, Wadle A, Hombach A, Shenderov E, Held G, Fischer E, Kleber S, Nuber N, Stenner-Liewen F, Bauer S, et al. (2009). Rational development of high-affinity T-cell receptor-like antibodies. Proc Natl Acad Sci USA 106, 5784-5788. [0145] Stone J D, Kranz D M (2013). Role of T cell receptor affinity in the efficacy and specificity of adoptive T cell therapies. Front Immunol 4, 244.