CROSS-REFERENCE TO RELATED APPLICATION

[0001] The present application claims benefit of U.S. Provisional Application No. 63/038,816, filed on June 13, 2020, the disclosure of which is hereby incorporated by reference in its entirety.

REFERENCE TO SEQUENCE LISTING SUBMITTED ELECTRONICALLY

[0002] The content of the electronically submitted sequence listing, file name SeqListing- Neoantigen-PCT.txt, size 91,741 bytes, and date of creation June 13, 2021, filed herewith, is incorporated herein by reference in its entirety.

TECHNICAL FIELD OF THE INVENTION

[0003] This present disclosure relates generally to the area of proteomics analysis, and specifically to proteomics analysis by means of liquid chromatography and mass spectrometry (LC-MS), and in more particular to a LC-MS-based method and system for analyzing small peptides, such as neoantigens.

BACKGROUND

[0004] Proteins encoded by genes carrying cancer-related mutations can be processed and mutation-bearing peptides can be presented on the cell surface in the context of human leukocyte antigen (HLA) molecules. Such peptides are called neoantigens. Neoantigens are truly personalized and cancer-specific, making them ideal anti-cancer therapeutic targets(Schumacher & Schreiber, 2015). Neoantigen can be recognized by T cells via their T cell receptor, which is the foundation for cancer immunotherapies(Lauss et ah, 2017; Ott et ah, 2017; Riaz et ah, 2016; Sahin et ah, 2017; Schumacher & Schreiber, 2015). To identify potential drug targets, numerous techniques have been reported recently to reveal the repertoire of neoantigens, including deep profiling of whole immunopeptidome and targeted detection approaches, as well as indirect assays on T cell reactivity (Bassani-Sternberg et ah, 2016; Danilova et ah, 2018; Wang et ah, 2019). We have previously reported a technology termed “MANA-SRM” as a basic research platform for neoantigen detection from approximately 200 million tissue culture tumor cells, which was over 10-fold more sensitive than previously published techniques available for this purpose where over 2-3 billion cells are often required (B as sani- Sternberg et al., 2016; Wang et al., 2019). MANA-SRM is an ideal tool for tissue culture-based neoantigen assays and has enabled the development of the first off-the-shelf cancer vaccine targeting the most frequent mutation hotspot in human tumor suppressor gene TP53 and oncogene K-Ras (Douglass et al., 2021; Hsiue et al., 2021). However, the sensitivity and robustness of MANA-SRM is still not feasible for routine clinical applications given the limited amount of tumor tissue available from biopsy or surgical resection. A more sensitive, rapid and reproducible platform for neoantigen analysis from clinical samples is desperately needed before a neoantigen-based personalized cancer treatment can be established.

[0005] In recent years, mass spectrometry platforms have undergone a series of groundbreaking improvements (Bekker-Jensen et al., 2020; Meier et al., 2018; Zubarev & Makarov, 2013). Meanwhile, proteomic analysis is changing from a multi-hour assay to a short 21-minute run with little compromise on coverage(Bache et al., 2018). These improvements are dramatically revolutionizing mass spectrometry-based proteomics and making it more suitable for clinical applications.

SUMMARY OF THE INVENTION

[0006] This present disclosure provides further improvements in mass spectrometry, which covers several fields including the sample preparation, the hardware configuration and automation, and the mass spectrometry analysis, etc. More specifically, this present disclosure provides an integrated method and system (termed “Valid-NEO pipeline” hereinafter) for the detection and quantification of neoantigens from clinical samples without extensive manual sample processing.

[0007] In a first aspect, a method for a characterization of a target peptide through a detection approach is provided. The method comprises the following steps:

[0008] (1) introducing at least one guard molecule to mix with the target peptide, wherein each of the at least one guard molecule is configured to have similar characteristics as the target peptide, and yet is further configured to be distinguishable from the target peptide by the detection approach; and

[0009] (2) applying the detection approach for the characterization of the target peptide.

[0010] Herein, the detection approach can comprise mass spectrometry analysis, and as such, each of the at least one guard molecule is configured to have an M/z value that is distinguishable from the target peptide by the mass spectrometry analysis. [0011] Besides mass spectrometry analysis, the detection approach covered herein may involve other approaches as well. Examples may include: the use of target peptide sequence specific antibodies, a fluorescent detection method when guard peptide or compound and targt peptide are designed to exhibit different fluorescent signals, the use of polymerase chain reaction (PCR)-based detection methods when the target peptides are associated with PCR- amplifiable signals, the use of nucleic acid sequence-based detection methods when the target peptide are associated with particular sequencing information, etc. Virtually any detection method that can distingui shingly detect the target peptide and the at least one guard compound can be applicable in the method, and is considered to be in the scope of this disclosure.

[0012] As used herein, as well as throughout other part of the disclosure, the term “characterization” of a target peptide may include either or both of detection (i.e. identification, qualification, or alike) and quantification of the target peptide.

[0013] As used herein, the term “guard molecule” refers to a molecule that coexists with the target peptide in one or more steps in the sample preparation and processing pipeline, which serves primarily to protect the target peptide from a loss in the analysis.

[0014] Specifically, if the detection approach involves mass spectrometry analysis, a guard molecule can be configured to have similar characteristics (e.g. similar hydrophobicity and charge status) as the target peptide, and to have a characteristic that can be differentiated from the target peptide by the downstream analytical instrument, such as a different M/z value that is distinguishable from the target peptide when analyzed by the mass spectrometer.

[0015] In another detection approach, such as using peptide-sequence specific antibody for the detection of a target peptide, where the sequence specific antibody can be used to selectively detect the target peptide and does not interact with the guard compound.

[0016] Regardless of the detection approach that is used, the co-presence of the guard molecule with the target peptide can ensure that the guard molecule behave similarly as the target peptides in sample preparation, thereby working as a blocker to prevent nonspecific binding of the target peptide to any surface or any substance that the target peptides may interact during the sample preparation and processing procedures.

[0017] As used herein, the term “similar characteristics” or “similarity” is substantially defined by the separation and fractionation mechanism adopted in a sample preparation and processing procedure, which can typically refer to a situation where two molecules, such as a first molecule (e.g. the target molecule) and a second molecule (e.g. a guard molecule), have substantially same hydrophobicity and charge status at a same pH environment. For example, in certain scenario where a sample preparation pipeline includes only reverse phase columns, a guard molecule that can be used in the method disclosed herein may share the similar hydrophobicity as the target molecule. In another scenario where a size exclusion column is used to separate the analytes, a guard molecule that can be used in the method disclosed herein may have a substantially similar molecular weight as the target molecule to therefore be able to co-exist with the target molecule in the sample preparation and processing procedures. In yet another scenario, both of the above types of columns may be used, thus for the selection of the guard molecule(s), a heavy isotope labeled peptide with the same sequence or peptides with just one amino acid different as target peptides as their guard peptides is the most feasible way to protect the target peptides from loss.

[0018] As used herein, the term “M/z value” is commonly known to people of oridinary skills in the field of mass spectrometry -based peptide analysis, which typically refers to a mass- to-charge ratio of a certain molecule. As is well understood, current mass spectrometry technology can readily separate two molecules that differ by approximately 1/30000 of a proton (i.e. one dalton) of their molecule weights. For example, a spectrometer can separate a first molecule with a molecular weight of 100.0001 from a second molecule with a molecular weight of 100.0002. As such, substitution of an amino acid residue with another amino acid residue having a different molecular weight or a different charge, or heavy isotope labelling of one or more amino acid residues can usually cause at least 1 dalton difference and can cause a target peptide and a guard peptide to be easily separated from one another.

[0019] According to certain embodiments of the method, the at least one guard molecule comprises a guard peptide. Herein, several possible embodiments exist.

[0020] In a first embodiment, the guard peptide may have a same amino acid residue sequence as the target peptide, and at least one amino acid residue in the guard peptide is heavy isotope labeled.

[0021] As used herein, the term “heavy isotope” that is labelled on a certain amino acid residue refers to the fact that in the guard peptide, there is at least one amino acid residue that is a heavy isotoe labeled amino acid residue, such as a Lysine residue with C ¹³ and/or N ¹⁵ labelings (i.e. the standard C ¹² element from the amino acid is replaced by a heavy isotope element of C ¹³ , and N ¹⁴ element can be replaced by a heavy isotpe element of N ¹⁵ ).

[0022] In a second embodiment, only one amino acid residue in the guard peptide may differ from the target peptide. Herein the differing amino acid residue in the guard peptide may, compared with the corresponding amino acid residue in the target peptide, have a same characteristics (i.e. hydrophobicity, charge status at the same pH, etc.) but have a different molecular weight. For example, regarding one target peptide: KRAS Q61H neoantigen sequence ILDTAGHEEY (SEQ ID NO. 496, see Table 1), several possible guard peptides can be used, which may include the following non-limiting examples:

[0023] 1) ILDTAGHDEY (SEQ ID NO. 507), obtained by substituting amino acid residue

E at position 8 of the target peptide with D;

[0024] 2) IVDTAGHEEY (SEQ ID NO. 518), obtained by substituting amino acid residue

L at position 2 of the target peptide with V; and

[0025] 3) ILDSAGHEEY (SEQ ID NO. 529), obtained by substituting amino acid residue

T at position 4 of the target peptide with S.

[0026] By means of these above amino acid residue substitutions (i.e. E/D, L/V, and S/T), each of these guard peptides can behave similarly to that of the target peptides in the sample preparation procedure, but can still be easily differentiated by mass spectrometry.

[0027] In a third embodiment, at least two amino acid residues in the guard peptide may differ from the target peptide. For example, for the above target peptide: KRAS Q61H neoantigen sequence ILDTAGHEEY (SEQ ID NO. 496), one possible guard peptide can be IVDTAGHDEY (SEQ ID NO. 540), which is obtained by substituting amino acid residue L at position 2 with V, and amino acid residue E at position 8 with D.

[0028] In a fourth embodiment, the guard peptide may have a scrambled sequence compared with the target peptide. For example, a possible guard peptide for the above exemplary target sequence ILDTAGHEEY (SEQ ID NO. 496) can be EYILGEDTAH (SEQ ID NO. 541), where they both have the same compositions of amino acid residues but have different sequences therefore will be readily differentiated by an feasible analytical method, such as using a sequence specific antibody or by a mass spectrometer.

[0029] In yet other embodiments, the at least one guard molecule may comprise a non peptide compound. Any compounds that behave similarly (i.e. same hydrophobicity, same charge status at same pH, etc.) as the target peptide in sample preparation can work as a blocker to prevent nonspecific binding of the target peptides to the surfaces target peptides may interact, and also the guard compounds can be differentiated by analytical procedures, such as by sequence-specific antibody or a mass spectrometer by having a different mass and/or M/z. [0030] According to certain embodiments of the method, the target peptide is a neoantigen peptide, and examples of the neoantigen peptide can include, without limitation, KRAS Q61H, KRAS Q61L, KRAS Q61R, IDH2 R140Q, TP53 Y220C, TP53 R248W, TP53 R213L, KRA S_G 12 V_9mer, KRAS_G12V_10mer, KRAS_G12D_9mer, or KRAS_G12D_10mer. These examples will be covered in the Embodiment 1 of the disclosure as set forth below. It is noted that besides characterization of a neoantigen peptide, the method can also be applied to characterize other peptides that are not neoantigens. There is no limitation herein.

[0031] According to some embodiments of the method where the target peptide is a neoantigen peptide, the neoantigen peptide is from a tissue sample obtained from a subject, and the method further comprises a tissue sample preparation step prior to step (1) of introducing at least one guard molecule to mix with the target peptide. The tissue preparation step may comprise the following sub-steps:

[0032] (a) providing the tissue sample, wherein the tissue sample is a frozen tissue sample; [0033] (b) grinding the frozen tissue sample, under an impact of at least 8,000 psi, to thereby obtain a frozen single-cell tissue powder; and

[0034] (c) treating the frozen single-cell tissue powder before obtaining a treated tissue sample.

[0035] Herein, the subject can be a human, but can also be another mammal species, such as a monkey, an ape, a dog, a mouse, a rat, etc., yet can also be a non-mammal species. The tissue sample may be a surgical resection tumor sample, or may be a biopsy sample.

[0036] Herein, in sub-step (a) of the tissue sample preparation step, the tissue sample can be a frozen tissue sample, which preferably can be a tissue sample snap-frozen in liquid nitrogen. The tissue sample can be freshly obtained from a subject by biopsy or by surgical dissection, and can be an FFPE (Formalin Fixed Paraffin Embedded) tissue sample that is processed by liquid nitrogen before the sub-step (b) of grinding. Other forms the tissue sample are also possible and shall be deemed to be covered by the disclosure herein.

[0037] According to certain embodiments, the impact for grinding the frozen tissue sample in sub-step (b) can be approximately 10,000 psi. In other embodiments, the impact force can be approximately 12,000 psi, 15,000 psi, etc.

[0038] According to certain embodiments, the sub-step (c) of treating the frozen single cell tissue powder before obtaining a treated tissue sample comprises: lysis, sonication, and centrifugation, and the treated tissue sample can be from a supernatant after the centrifugation. [0039] In sub-step (c) of the tissue sample preparation step, the frozen single-cell tissue powder can be treated by tissue lysis to lyse the cells within the ground tissue sample, followed by sonication for fragmenting the genomic DNAs within the cells. After centrifuge, the supernatant becomes the treated tissue sample which contains the neoantigen peptide to be characterized. More details for the tissue sample preparation step can be found in Embodiment 1 of the disclosure. [0040] According to certain embodiments, after the sub-step (c) of treating the frozen single-cell tissue powder before obtaining a treated tissue sample, the method can further comprise:

[0041] performing an analysis over genomic DNA obtained from a pellet after the centrifugation.

[0042] According to some embodiments, after the tissue sample preparation step and prior to step (1) of introducing at least one guard molecule to mix with the target peptide, the method further comprises an human leukocyte antigen (HLA) molecule enrichment step, comprising: [0043] passing the treated tissue sample through an HLA enrichment column, wherein the HLA enrichment column comprises a matrix with anti-HLA antibodies immobilized thereon. [0044] According to some embodiments, after the HLA molecules enrichment step, the method further comprises an elution step, comprising:

[0045] applying an elution buffer having a low pH to the HLA enrichment column to thereby obtain an eluate containing the neoantigen peptide.

[0046] Herein, the elution buffer comprises the at least one guard molecule.

[0047] According to some embodiments, after the elution step and prior to step (2) of applying the mass spectrometry analysis for the characterization of the target peptide, the method further comprises an clean-up step, comprising:

[0048] (a) passing the eluate through a trap column for at least one time to thereby trap the neoantigen peptide therewithin, wherein the trap column comprises a matrix capable of binding with the neoantigen peptide but having a lower or no binding affinity to impurities; and [0049] (b) eluting the trap column to thereby obtain a cleaned eluate.

[0050] According to some embodiments, after the clean-up step and prior to step (2) of applying the mass spectrometry analysis for the characterization of the target peptide, the method further comprises a purification step, comprising:

[0051] passing the cleaned eluate through a size exclusion column (SEC column, or SEC), for collecting a neoantigen peptide-containing fraction.

[0052] Herein optionally, before passing the cleaned eluate, the cleaned eluate can be spiked with a peptide ladder, such as NEO-SEC ladder described below, which is purported to define the boundaries for collecting the neoantigens. Specifically, signature chromatography peaks can be monitored to indicate the starting point (e.g. a peak representing 2000 Da in the example of NEO-SEC ladder) and the ending point (e.g. a peak representing 800 Da in the example of NEO-SEC ladder) for the collection. [0053] Regarding the tissue preparation step, the HLA molecules enrichment step, the elution step, the clean-up step, the purification step, and the mass spectrometry analysis step of the method as set forth above, more details can be found in Embodiment 1 of the disclosure. [0054] According to certain embodiments of the method, at least two consecutive steps of the HLA molecules enrichment step, the elution step, the clean-up step and the purification step can be operably connected, thereby substantially realizing an automatic processing with little or no human intervention.

[0055] According to certain embodiments, substantially all steps of the HLA molecules enrichment step, the elution step, the clean-up step and the purification step are operably connected, and are subsantically automatic.

[0056] According to preferred embodiments, the mass spectrometry analysis step whereby the neoantigen peptide-containing sample runs through a mass spectrometer, i.e. step (2) as mentioned above, may be further operably connected with the upstream purification step for automation. In other words, the whole sample processing procedure, including the HLA molecule enrichment step, the elution step, the clean-up step, the purification step, and the mass spectrometry analysis step, can realize an automation with little or no human intervention. [0057] It is noted that each of these steps may not be limited to the steps disclosed herein, and can be realized by an alternative means that is known to people of ordinary skills in the art, yet by applying these steps as disclosed herein, an automation of the sample preparation, HLA molecules enrichment, elution, clean-up, purification, and mass spectrometry analysis can be realized. The integrated system and method, termed “Valid-NEO” pipeline, substantially requires little or no manual intervention, and thereby high sensitivity, reproducibility, and transplantability can be ensured across different diagnostic centers and hospitals.

[0058] In a second aspect, a system capable of implementing the method as set forth above is further provided. The system comprises the following components:

[0059] (1) a tissue sample grinding device, configured to apply an impact of at least 8,000 psi to the frozen tissue sample to thereby obtain the frozen single-cell tissue powder in the tissue sample preparation step;

[0060] (2) an HLA enrichment column, comprising matrix with anti-HLA antibodies immobilized thereon and configured to allow the treated tissue sample obtained in the tissue sample preparation step to pass therethrough so as to enrich the HLA molecules in the HLA molecules enrichment step; [0061] (3) a trap column, comprising a matrix capable of binding with the neoantigen peptide but having a lower or no binding affinity to impurities and configured to trap the neoantigen peptide therewithin in the clean-up step;

[0062] (4) a size exclusion column (SEC), configured to purify the neoantigen peptide in the purification step; and

[0063] (5) a mass spectrometer, configured to implement the mass spectrometry analysis step.

[0064] In Embodiment 1 of the disclosure, one embodiment of the tissue sample grinding device, termed “UniCeller”, is described in more detail.

[0065] According to certain embodiments of the system, at least two neighbors of the HLA enrichment column, the trap column, the size exclusion column (SEC), and the mass spectrometer are sequentially and operably connected with one another to thereby allow an automation.

[0066] Yet according to certain embodiments, all of the HLA enrichment column, the trap column, the size exclusion column (SEC), and the mass spectrometer are sequentially and operably connected with one another to thereby allow an automation.

BRIEF DESCRIPTION OF THE DRAWINGS

[0067] FIG. 1 illustrates a general process for the Valid-NEO pipeline according to certain embodiments of the disclosure;

[0068] FIG. 2A and FIG. 2B together illustrate a scheme of neoantigen isolation and purification in the Valid-NEO pipeline according to certain embodiments of the disclosure; [0069] FIGS. 3A-3F together show a table summarizing the cancer types of the patients, the selected genetic mutation features of their tumors, and the possible neoantigen sequences flanking the highest prevalence mutation site on a cancer driver gene;

[0070] FIG. 4 shows a comparison result of the protein extraction efficiences between UniCeller and three other traditional approaches;

[0071] FIG. 5 shows reproducibility evaluation results of the Valid-NEO pipeline analysis; and

[0072] FIG. 6 shows HPLC chromotograms for neoantigen purification through an SEC column.

DETAILED DESCRIPTION [0073] In order to further describe the neoantigen analysis method and system as provided above, one specific embodiment (i.e. Embodiment 1) is provided below.

[0074] Embodiment 1

[0075] FIG. 1 illustrates a general process for the Valid-NEO pipeline to analyze a neoantigen peptide. Specifically, tumor tissue specimen from a subject is first harvested through biopsy or surgical resection, which is then processed, through tissue lysis, sonication and centrifuge, to obtain HLA molecules (in supernatant) and genomic DNA (in pellet). Patient specific mutations were revealed through genomic sequencing of the tumor DNA from the pellet by DEEPER-Seq pipeline (Wang et al., 2017). Mutation callings are made and potential sequences for neoantigens and neoantigen peptides (called “MaxRec sequences”) were established. Neoantigens are then isolated and purified from tissue lysate superntant through a multi-step procedure which is illustrated in FIGS. 2A and 2B and described below, then neoantigens are directly detected and quantified through mass spectrometry.

[0076] FIGS. 2A and 2B illustrate a scheme of the multi-step process of neoantigen isolation and purification in the Valid-NEO pipeline according to certain embodiments of the disclosure. As illustrated in the two figures, there are mainly 5 steps in Valid-NEO system to obtain neoantigens, Step 1) HLA enrichment by antibody column with repeated loading; Step

2) Elution of HLA molecules and separation of neoantigens followed by trap column loading; Step 3) Neoantigen clean up and removal of HLA molecules; Step 4) Neoantigen elution and loading of SEC column; Step 5) Purification of neoantigens through SEC column.

[0077] In this embodiment of the multi-step process for neoantigen isolation and purification in the Valid-NEO pipeline, it is noted that a series of valves (see “Valves 1-5” in the figure), a series of pumps (see PUMPS 1-5), a series of DAD detectors (see “Detectors 1-

3), and a Fraction Collector, are also included in the system. The schematic configuration and connection for each component in the system is also illustrated in FIGS. 2A and 2B.

[0078] As illustrated, each valve comprises a total of 6 ports (#1-6), each operably and controllably connected to an inlet or an outlet of other devices, such as the “Antibody Column” (i.e. HLA enrichment column), the “Trap Column”, the “SEC Column”, the pumps, DAD detectors, and the Fraction Collector. Each pump is configured to provide a driving force that drives the fluid to flow in the pipeline in a predetermined direction (as shown by the arrows in the figure), and each port is configured to open or close in a controlled manner based on the control signals that it receives. Each of the DAD detectors is configured to detect certain parameter of the fluid that it receives. A processor (not shown) is communicatively connected to each of the above components, and is configured, based on the detection signals transmitted from the DAD detectors, to control the coordinated working of each of the above components in a programed manner. For example, the processor may control the opening/closing status of each port of the valves, and may control the start or stop and flow rate of the each pump. As such, the coordinated working of each component of the system can realize an automatic sample processing, allowing the treated tissue sample (i.e. HLA/neoantigen-containing sample, or the “supernatant” in FIG. 1) to flow through the enrichment column, the trap column, and the SEC column in an automatic and controlled manner before running into the mass spectrometer for characterization.

[0079] Materials and Methods

[0080] Tumor Samples

[0081] Tumor samples from a total of 10 patients were obtained from BioIVT. This study was approved by the Institutional Review Boards for Human Research at Complete Omics Inc. and BioIVT, and complied with Health Insurance Portability and Accountability Act. Cancer types of the patients and selected genetic mutation features of their tumors are listed in the table shown in FIGS. 3A-3F. In the table, the amino acid residues in bold and underlined font represent the mutations of interest in a neoantigen peptide sequence.

[0082] Construction of Valid-NEO

[0083] Valid-NEO is an integrated system composed of five steps essential for neoantigen detection, including 1) Enrichment of HLA molecules, 2) Elution of neoantigens from antibody column, 3) Cleaning of neoantigens, 4) Elution of neoantigens from trap column, 5) Purification of neoantigens through SEC column. This integrated system is composed of a tandem series of HPLC systems, one mass spectrometer, and a set of optimized buffers including the MaxRec system.

[0084] HLA molecule extraction from tissue sample

[0085] Human tumor fresh frozen tissues were obtained from BioIVT (BioIVT, NY). 50 mg frozen tissue were wrapped in aluminum foil such that the tissue chunk was covered by at least four layers of aluminum foil. The wrapped tissue chunk was snap-frozen in liquid nitrogen. UniCeller (Complete Omics Inc, MD), an in-house built device designed to apply strong impact onto frozen tissue packs, was used to produce single-cell level powder from the tissue chunk, and this procedure can be repeated 5 times until the tissue chunk is completely ground into frozen single-cell powder. 1 mL NL buffer (Complete Omics Inc, MD) was added to the tissue powder and the tissue suspension was transferred into a protein lo-bind tube followed by five rounds of sonications through Bioruptor 300 (energy level 4.5, duty step 30 seconds, and delay step 59 seconds). The tissue lysate was incubated on ice for 1 hour, during which the suspension was pipetted up and down 20 times every 10 minutes, and one additional cycle of sonication was performed every 10 minutes. The tissue lysate was centrifuged at 4°C for 30 minutes, and the clear supernatant was transferred to a new protein lo-bind tube. The supernatant containing HLA molecules was diluted with 4 volumes of NC buffer (Complete Omics Inc., MD), after which it was ready for HLA molecule isolation.

[0086] Online enrichment of HLA molecules through antibody -column

[0087] Anti-HLA antibodies (clone W6/32) were immobilized on Protein A agarose beads (ThermoFisher Scientific, MA) through DMP (dimethyl pimelimidate)-based crosslinking reaction. 50 mL beads were then packed into an HLA enrichment column and flushed with 1 L NC buffer (Complete Omics Inc, MD). HLA-neoantigen suspension was filtered through a 0.22 pm filter, diluted with 4 volumes of the NC buffer and injected directly onto the HLA enrichment column. The flow-through was collected into a sample loop and re-injected onto the column. The injection was repeated for 4 more times, for a total of 5 passes of the suspension through the antibody column. During the repeated loadings, HLA molecules were depleted from the mobile phase and captured by the column, while the HLA-suspension was gradually diluted by NC buffer pushed into the system by the pump. The repeated loading ensured an efficient binding of the HLA molecules to the column and the sequential dilution of the sample with the mobile phase facilitates an improved cleaning efficiency and reduced nonspecific binding. The antibody column was then flushed with NC buffer at 1 mL/min for 20 minutes to remove unbound proteins and impurities (including salts and detergents).

[0088] Online elution of neoantigen peptides and antibody column regeneration [0089] Elution of the neoantigen peptides was performed with an increasing gradient (from

0 to 100 % over a period of 5 minutes) of NE buffer (Complete Omics Inc, MD) through the column, followed by a constant flush with 100% NE buffer at 1 mL/min for 2 minutes. The antibody column was then neutralized by running an increasing gradient (from 0 to 100 % over a period of 5 minutes) of NN buffer (Complete Omics Inc, MD), and followed by a 1 hour flushing with NC buffer at 1 mL/min. The eluted HLA molecules and neoantigen peptides were then subjected to further purifications.

[0090] Online isolation and purification of neoantigen peptides

[0091] HLA eluate containing neoantigen peptides was injected to pass through a trap column for a total of 5 times, followed by washing with 10 mL 0.1% formic acid. The cleaned peptides were eluted from the trap column through three cycles of acetonitrile gradients using mobile phase solvent A: 0.1% formic acid in water and mobile phase solvent B: 0.1% formic acid in acetonitrile. The gradient started from 0% solvent B and increased to 60% solvent B over 30 seconds, and then decreased to 0% solvent B over 30 seconds, and this 1-min gradient step was repeated three times at the follow rate of 1 mL/min followed by a high-speed flush at 2 mL/min with 100% solvent B for 1 minute. The follow through was collected 30 seconds after the initial gradient change took place and the collection was stopped 1 minute after the flushing step ended. A total of 4.5 mL of neoantigen peptide suspension was collected with an estimated 30% acetonitrile and 0.1% formic acid. The collected neoantigen suspension was directly loaded onto an SEC column packed with 1.7 pm particles with 125 A pore size (Waters, MA). Before the analysis, NEO-SEC ladder (Complete Omics, MD) was spiked into the system to define the boundaries for collecting the neoantigens. Signature chromatography peaks were monitored to indicate the starting point (a peak representing 2000 Da) and the ending point (a peak representing 800 Da) for the collection. Flow-through containing the isolated neoantigen peptides was collected and subject to lyophilization before mass spectrometry analysis.

[0092] Mass spectrometry method development

[0093] Heavy isotope labeled neoantigen peptides flanking gene mutations in patient cancer genomes were synthesized. Optimization of the detection parameters was performed with a two-step approach. Step 1) All possible ions (first to last) of each peptide were detected with a theoretical collision energy as well as two additional collision energies at 5 eV below and above the theoretical value (three collision energy values in total for each transition). The highest abundance transitions were selected for the next round of optimization. Step 2) High abundance transitions selected from previous step (>20 transitions for each charge status of the peptide target) were subject to a further optimization where for each transition 9 collision energy values were tested including the theoretical collision energy value as well as 4 steps of values below and above the theoretical value with a step-size of 2 eV. After two rounds of optimizations, detection parameters were manually curated to avoid false positive signals from co-detected impurities in the Valid-NEO matrix prepared from a reference human tumor sample, and an average of 8 to 10 transitions were selected as signature transitions for each target. Before and after each batch of analysis, Agilent 6495C Triple Quadrupole mass spectrometer was tuned using manufacturer’s tuning mixture followed by MyProt-SRM Tuning Booster (Complete Omics, MD). Before each assay, to ensure the stable and consistent performance of the mass spectrometer throughout the entire study, MyProt-SRM Performance Standard (Complete Omics, MD), a mixture of standard peptides across a wide range of masses (M/z 100-1400) and a broad range of hydrophobicities, were analyzed. A system performance score was documented before every run.

[0094] Pre-conditioning the system to ensure highest sensitivity

IB [0095] In order to achieve the highest sensitivity for the assay, a strategy is developed to ensure a minimal sample loss by pre-conditioning and co-processing in the system with peptides that are “similar” to the ones being detected. The peptides used to ensure the maximal recovery of the assay are called MaxRec peptides. A MaxRec prediction algorithm was created to generate MaxRec peptide sequences based on the sequences, hydrophobicity and detectability (signal strengths detected in mass spectrometer) of the target peptides desired to be detected from the pipeline. MaxRec peptide sequences used in this study were shown in Table 1, where the amino acid residues in bold and underlined font represent the mutations of interest (i.e. target mutations), and the amino acid residues in italics font represent the altered residues used in the MaxRec peptides. All MaxRec peptides were synthesized at a high purity (>99.9%). A buffer system containing MaxRec peptides at the concentration of 100 femtomole/pL was injected into the Valid-NEO pipeline before each assay. MaxRec peptides passed through the pipeline at much higher concentrations than what would presumably be observed from the target peptides in clinical samples. Before clinical sample injection, the Valid-NEO pipeline was flushed with NC buffer for 30 minutes to deplete excessive unbound MaxRec peptides.

Table 1. MaxRec peptides used in this study [0096] Data deposition

[0097] The data reported in this article have been deposited via ProteomeXchange in [0098] PeptideAtlas SRM Experiment Library (PASSEL) (identifier PASS01588).

[0099] Results

[0100] To maximize the recovery of ELLA molecules from tumor tissue samples, it is critical to homogenize the frozen tissue into single-cell powder rapidly without thawing the sample. For this purpose, an equipment, called the “EiniCeller”, was developed, which is capable of applying a strong impact (-10,000 psi) to frozen tissue chunks. Tissue powder was produced through EiniCeller and was then quickly dissolved in Neoantigen Lysis (NL) buffer (Materials and Methods), followed by repeated pipetting and programmed sonication (Materials and Methods). Through this procedure, it was shown that nearly 100% of the ELLA molecules from the tissue sample was able to be extracted, which represents a greater recovery efficiency than when using traditional approaches including Dounce Homogenizer, Probe Sonicator and Bead Ruptor (see FIG. 4, which shows a comparison results of the recovery efficiences between UniCeller and three other traditional approaches Specifically for the experiment, the same amount (50 mg) of tumor tissue was processed through different approaches, including using Dounce Homogenizer, Probe Sonicator, Bead Ruptor, and UniCeller, for extracting HLA complexes. W6/32 antibody was used for the blot). The higher yield observed when using the UniCeller suggests that HLA molecule, as a protein complex predominantly located on the cell surface, may be vulnerable to temperature change and harsh mechanical force in liquid suspension during extraction. In addition, few to none HLA molecules were left in the pellet from the UniCeller group, indicating that a higher extraction efficiency can be achieved when a strong mechanical impact is applied to rapidly generate single-cell level dry tissue powder, followed immediately by a moderate but repeated HLA extraction in lysis buffer.

[0101] The pellet obtained from the UniCeller tissue lysate was processed to extract genomic DNA (see FIG. 1). Single-stranded exomic regions of a selected panel of cancer driver genes were captured with proprietary dual RNA probes and sequenced through DEEPER-Seq pipeline (Wang et ah, 2017). A mutation calling was made only when the point mutation is observed as a complementary pair of residues on both DNA strands came from the same DNA duplex molecule as previously described (Wang et ah, 2017). Nine tumor samples bearing hotspot mutations in highly frequently mutated cancer driver genes K-Ras and TP53 , as well as the slightly lower frequently mutated driver gene IDH2 were selected for a further evaluation of potential neoantigen presentations. [0102] Antibody-column based affinity chromatography is more efficient and cost- effective than conventional immunoprecipitation and was thus adopted in Valid-NEO pipeline for enriching HLA molecules(Moser & Hage, 2010). To achieve a high enrichment efficiency, an antibody-conjugated column was packed with a 20-fold excess of antibodies (50 mg antibody) relative to the amount needed to enrich HLA molecules from a typical sample (50 - 100 mg wet tissue with > 50% tumor mass), in addition repeated sample loadings were performed to ensure the binding between antibodies and HLA molecules in Neoantigen Capture (NC) buffer (see FIGS. 2A and 2B, Materials and Methods). Through elution from the column and incubation in acidic Neoantigen Elution (NE) buffer (Materials and Methods), HLA complexes were eluted, and neoantigen peptides were dissociated from HLA molecules (see FIGS. 2A and 2B). The HLA enrichment column was then regenerated by Neoantigen Neutralization (NN) buffer. The column can be used for at least 30 enrichment and elution procedure with no significant loss in performance observed for the neoantigens analyzed in this study (see FIG. 5, which shows reproducibility evaluation results of the Valid-NEO pipeline analysis. Specifically for the experiment, 10 tumor samples were processed through the same NeoTrue Valid-NEO pipeline to evaluate their endogenous neoantigen presentations. For each neoantigen, three replicates NeoTrue Valid-NEO assays were performed. There were 9 NeoTrue Valid-NEO assays for different neoantigens between replicate 1 and 2, as well as between replicate 2 and 3 for the detection of each neoantigen).

[0103] HLA molecules and other large proteins were separated from neoantigen peptides by a trap column packed with Cl 8 small pore spherical silica particles (diameter 100 A). Neoantigens (molecular weight around 1.5 kDa) are significantly smaller than HLA molecules (molecular weight around 41 kDa), and will enter the pores therefore be efficiently retained by the Cl 8 matrix that are predominately located inside the pores. The majority of HLA molecules and other large proteins are not efficiently retained by the column. Neoantigens bound to the trap column were then cleaned with 0.1% formic acid to remove HLA molecules and other impurities (see FIGS. 2A and 2B). Neoantigens were then eluted into a suspension composed of 30% acetonitrile 0.1% formic acid. The suspension was spiked with NEO-SEC ladder (Complete Omics Inc, MD), containing two sets of peptides with signature molecular weights at 2,000 and 800 Da (Materials and Methods), and subjected to fractionation through a size exclusion column (short as SEC or SEC column hereinafter) (see FIGS. 2 A and 2B). During elution, absorbance at the wavelength of 280 nm was constantly measured by a diode array detector (i.e. DAD). Neoantigen fraction started to be collected after a signature peak at 2,000 Da was observed, and the collection stopped before an 800 Da signature peak was observed (see FIG. 6, which shows HPLC chromotograms for neoantigen purification through an SEC column. Specifically, neoantigen samples were loaded to SEC column together with NEO-SEC ladders. Two signature peaks were observed at 2000 Da and 800 Da, which marked the boundaries for neoantigen-containing fractions). The flow-through between the two signature peaks was collected and lyophilized. The neoantigen sample was then subjected to Valid-NEO mass spectrometry analysis with pre-optimized conditions (Materials and Methods).

[0104] To further improve the recovery ratio and the sensitivity of the pipeline, a neoantigen recovery system, called “Maximum Recovery (MaxRec) System”, was developed. A key element in MaxRec is a set of peptides with slightly different sequences (varying by only 1 or 2 amino acids) from the target peptides, and they are resuspended in the MaxRec system at a much higher abundance than the endogenous neoantigens. MaxRec peptides are designed to mimic the physical characteristics of the target peptides, so as to saturate the nonspecific binding opportunities in the system, thereby minimizing the loss of the target peptides due to such nonspecific interactions. Though sharing similar physical characteristics as target peptides, MaxRec peptides are chemically different, and can be easily distinguished from neoantigen targets based on the high resolution of modem mass spectrometers (see FIGS. 2A and 2B). The Triple Quadrupole Mass Spectrometer in Valid-NEO system is set up to be completely blind to MaxRec peptides and the presence of MaxRec system will not reduce the sensitivity of the platform, which is a feature of such type of instrument. The Valid-NEO neoantigen isolation system was pre-conditioned and spiked in the samples with the MaxRec peptides (see Table 1 above). Through MaxRec system, a significantly improved performance was able to be achieved in the detection of all neoantigen peptides in this study with an average increase in sensitivity by 27 folds (ranging from 2 to 67 folds) (the results are shown in Table 2 below).

Table 2. Neoantigen quantification through different approaches

[0105] It has been shown that almost all MHC class I associated neoantigens have a length between 8 to 12 amino acids (Sarkizova et al., 2020). For each sample, all potential neoantigen sequences flanking the highest prevalence mutation site on a cancer driver gene (a maximum of 50 possible neoantigen peptides for each missense mutation site) can be directly assayed for in a massively parallel manner without any prediction thus preventing uncertainties (see the table shown in FIGS. 3 A-3F). Using Valid-NEO pipeline, nine fresh frozen tumor samples was analyzed to detect and quantify each patient’s personalized neoantigens and compared their relative performance between MANA-SRM and the current integrated Valid-NEO pipeline (see Table 2). The patients with KRAS mutations at Q61 site have on average 6.6 copies of the neoantigen presented on each tumor cell surface, and the neoantigens flanking G12 site has an average of 32.1 copies presented per tumor cell. The presentation of TP53 neoantigens are low, ranging from 1 to 8 copies per cell, similar to IDH2 mutations with 6.1 copies of neoantigens presented per cell. Each assay was performed for three times, and the reproducibility of the pipeline was thoroughly evaluated (see Table 2, FIG. 5).

[0106] Discussion

[0107] Traditionally, cytotoxic chemotherapies have been the mainstay therapeutic agent for cancers, regardless of a given patient’s individual genetic background of the disease(Bonadonna & Valagussa, 1983; Chan et al., 2012; Savage et al., 2009; Yagoda & Petrylak, 1993). While cytotoxic chemotherapies are still the first line treatment for many cancers, further molecular characterization of cancers has facilitated the development of small molecules or antibody-based agents that can treat a sub-population of the patients who are sharing the same genetic basis of their diseases(Sawyers, 2004; Scaltriti & Baselga, 2006; Sharkey & Goldenberg, 2006). With the development of next-generation sequencing (NGS), it is evident that each individual’s cancer has its own genetic profile with varying degrees of overlaps in cancer driver gene mutations among patients(Bagnyukova et al., 2010; Cancer Genome Atlas Research et al., 2013; Chin et ak, 2011; Vogel stein et al., 2013). In recent years, highly personalized cancer therapeutic approaches have achieved success through targeting a patient’s specific neoantigens, offering hope with regards to the generalizability of such highly personalized treatments(Ott et al., 2017; Sahin et al., 2017). To reveal the neoantigen sequences needed for such personalized cancer therapeutics, algorithm-based or artificial intelligence (Al)-based predictions are often the choice when direct observation is impossible, but such predictions have been proven to be unreliable for clinical applications(Jurtz et al., 2017; Wang et al., 2019). Neoantigens can also be determined through co-culturing tumor cells with autologous T cells, followed by tetramer staining or peptide-pulsing assays, however these functional assays are technically difficult and time consuming, therefore cannot be readily adopted in clinical settings(Danilova et al., 2018; Lu et al., 2014). In Valid-NEO, no prediction is needed, and no cell culture is performed. Additionally, while the neoantigens evaluated in this study are all presented in the context of class I major histocompatibility complexes (MHC I), a similar concept can be readily applied to class II MHC as previously described(Wang et al., 2019).

[0108] Valid-NEO is the only pipeline developed so far to directly validate neoantigens from clinical samples in a sensitive, rapid and reproducible manner, and it helps pave the way for truly personalized cancer therapeutics.

References:

1. Bache, N., Geyer, P. E., Bekker-Jensen, D. B., Hoeming, O., Falkenby, L., Treit, P. V., Doll, S., Paron, L, Muller, J. B., Meier, F., Olsen, J. V., Vorm, O., & Mann, M. (2018, Nov). A Novel LC System Embeds Analytes in Pre-formed Gradients for Rapid, Ultra-robust Proteomics. Mol Cell Proteomics, 77(11), 2284-2296. 2. Bagnyukova, T. V., Serebriiskii, I. G., Zhou, Y., Hopper-Borge, E. A., Golemis, E. A., & Astsaturov, I. (2010, Nov 1). Chemotherapy and signaling: How can targeted therapies supercharge cytotoxic agents? Cancer Biol Ther, 10(9), 839-853.

3. Bassani-Sternberg, M., Braunlein, E., Klar, R., Engleitner, T., Sinitcyn, P., Audehm, S., Straub, M., Weber, J., Slotta-Huspenina, J., Specht, K., Martignoni, M. E., Werner, A., Hein, R., D, H. B., Peschel, C., Rad, R., Cox, J., Mann, M., & Krackhardt, A. M. (2016, Nov 21). Direct identification of clinically relevant neoepitopes presented on native human melanoma tissue by mass spectrometry. Nat Commun, 7, 13404.

4. Bekker-Jensen, D. B., Martinez-Val, A., Steigerwald, S., Ruther, P., Fort, K. L., Arrey, T. N., Harder, A., Makarov, A., & Olsen, J. V. (2020, Apr). A Compact Quadrupole-Orbitrap Mass Spectrometer with FAIMS Interface Improves Proteome Coverage in Short LC Gradients. Mol Cell Proteomics, 19(A), 716-729.

5. Bonadonna, G., & Valagussa, P. (1983, Mar). Chemotherapy of breast cancer: current views and results. Int J Radiat Oncol Biol Phys, 9(3), 279-297.

6. Cancer Genome Atlas Research, N., Weinstein, J. N., Collisson, E. A., Mills, G. B., Shaw, K. R., Ozenberger, B. A., Ellrott, K., Shmulevich, F, Sander, C., & Stuart, J. M. (2013, Oct). The Cancer Genome Atlas Pan-Cancer analysis project. Nat Genet, 45(10), 1113-1120.

7. Chan, D. L., Morris, D. L., Rao, A., & Chua, T. C. (2012). Intraperitoneal chemotherapy in ovarian cancer: a review of tolerance and efficacy. Cancer Manag Res, 4, 413-422.

8. Chin, L., Andersen, J. N., & Futreal, P. A. (2011, Mar). Cancer genomics: from discovery science to personalized medicine. Nat Med, 17(3), 297-303.

9. Danilova, L., Anagnostou, V., Caushi, J. X., Sidhom, J. W., Guo, H., Chan, H. Y., Suri, P., Tam, A., Zhang, J., Asmar, M. E., Marrone, K. A., Naidoo, J., Brahmer, J. R., Forde, P. M., Baras, A. S., Cope, L., Velculescu, V. E., Pardoll, D. M., Housseau, F., & Smith, K. N. (2018, Aug). The Mutation-Associated Neoantigen Functional Expansion of Specific T Cells (MANAFEST) Assay: A Sensitive Platform for Monitoring Antitumor Immunity. Cancer Immunol Res, 6( 8), 888-899.

10. Douglass, J., Hsiue, E. H., Mog, B. J., Hwang, M. S., DiNapoli, S. R., Pearlman, A. H., Miller, M. S., Wright, K. M., Azurmendi, P. A., Wang, Q., Paul, S., Schaefer, A., Skora, A. D., Molin, M. D., Konig, M. F., Liu, Q., Watson, E., Li, Y., Murphy, M. B., Pardoll, D. M., Bettegowda, C., Papadopoulos, N., Gabelli, S. B., Kinzler, K. W., Vogelstein, B., & Zhou, S. (2021, Mar 1). Bispecific antibodies targeting mutant RAS neoantigens. Sci Immunol, 6(51). 11. Hsiue, E. EL, Wright, K. M., Douglass, J., Hwang, M. S., Mog, B. J., Pearlman, A. H., Paul, S., DiNapoli, S. R., Konig, M. F., Wang, Q., Schaefer, A., Miller, M. S., Skora, A. D., Azurmendi, P. A., Murphy, M. B., Liu, Q., Watson, E., Li, Y., Pardoll, D. M., Bettegowda, C., Papadopoulos, N., Kinzler, K. W., Vogelstein, B., Gabelli, S. B., & Zhou, S. (2021, Mar 1). Targeting a neoantigen derived from a common TP53 mutation. Science.

12. Jurtz, V., Paul, S., Andreatta, M., Marcatili, P., Peters, B., & Nielsen, M. (2017, Nov 1). NetMHCpan-4.0: Improved Peptide-MHC Class I Interaction Predictions Integrating Eluted Ligand and Peptide Binding Affinity Data. J Immunol, 199(9), 3360-3368.

13. Lauss, M., Donia, M., Harbst, K., Andersen, R., Mitra, S., Rosengren, F., Salim, M., Vallon-Christersson, J., Torngren, T., Kvist, A., Ringner, M., Svane, I. M., & Jonsson, G. (2017, Nov 23). Mutational and putative neoantigen load predict clinical benefit of adoptive T cell therapy in melanoma. Nat Commun, 5(1), 1738.

14. Lu, Y. C., Yao, X., Crystal, J. S., Li, Y. F., El-Gamil, M., Gross, C., Davis, L., Dudley, M. E., Yang, J. C., Samuels, Y., Rosenberg, S. A., & Robbins, P. F. (2014, Jul 1). Efficient identification of mutated cancer antigens recognized by T cells associated with durable tumor regressions. Clin Cancer Res, 20(13), 3401-3410.

15. Meier, F., Brunner, A. D., Koch, S., Koch, H., Lubeck, M., Krause, M., Goedecke, N., Decker, J., Kosinski, T., Park, M. A., Bache, N., Hoerning, O., Cox, J., Rather, O., & Mann, M. (2018, Dec). Online Parallel Accumulation- Serial Fragmentation (PASEF) with a Novel Trapped Ion Mobility Mass Spectrometer. Mol Cell Proteomics, 77(12), 2534-2545.

16. Moser, A. C., & Hage, D. S. (2010, Apr). Immunoaffmity chromatography: an introduction to applications and recent developments. Bioanalysis, 2(4), 769-790.

17. Ott, P. A., Hu, Z., Keskin, D. B., Shukla, S. A., Sun, J., Bozym, D. J., Zhang, W., Luoma, A., Giobbie-Hurder, A., Peter, L., Chen, C., Olive, O., Carter, T. A., Li, S., Lieb, D. J., Eisenhaure, T., Gjini, E., Stevens, J., Lane, W. J., Javeri, T, Nellaiappan, K., Salazar, A. M., Daley, H., Seaman, M., Buchbinder, E. L, Yoon, C. H., Harden, M., Lennon, N., Gabriel, S., Rodig, S. J., Barouch, D. H., Aster, J. C., Getz, G., Wucherpfennig, K., Neuberg, D., Ritz, J., Lander, E. S., Fritsch, E. F., Hacohen, N., & Wu, C. J. (2017, Jul 13). An immunogenic personal neoantigen vaccine for patients with melanoma. Nature, 547(1662), 217-221.

18. Riaz, N., Morris, L., Havel, J. J., Makarov, V., Desrichard, A., & Chan, T. A. (2016, Aug). The role of neoantigens in response to immune checkpoint blockade. Int Immunol, 25(8), 411-419.

19. Sahin, U., Derhovanessian, E., Miller, M., Kloke, B. P., Simon, P., Lower, M., Bukur, V., Tadmor, A. D., Luxemburger, U., Schrors, B., Omokoko, T., Vormehr, M., Albrecht, C., Paruzynski, A., Kuhn, A. N., Buck, J., Heesch, S., Schreeb, K. H., Muller, F., Ortseifer, L, Vogler, L, Godehardt, E., Attig, S., Rae, R., Breitkreuz, A., Tolliver, C., Suchan, M., Martic, G., Hohberger, A., Sorn, P., Diekmann, J., Ciesla, J., Waksmann, O., Bruck, A. K., Witt, M., Zillgen, M., Rothermel, A., Kasemann, B., Langer, D., Bolte, S., Diken, M., Kreiter, S., Nemecek, R., Gebhardt, C., Grabbe, S., Holler, C., Utikal, J., Huber, C., Loquai, C., & Tureci, O. (2017, Jul 13). Personalized RNA mutanome vaccines mobilize poly-specific therapeutic immunity against cancer. Nature, 577(7662), 222-226.

20. Sarkizova, S., Klaeger, S., Le, P. M., Li, L. W., Oliveira, G., Keshishian, H., Hartigan, C. R., Zhang, W., Braun, D. A., Ligon, K. L., Bachireddy, P., Zervantonakis, I. K., Rosenbluth, J. M., Ouspenskaia, T., Law, T., Justesen, S., Stevens, J., Lane, W. J., Eisenhaure, T., Lan Zhang, G., Clauser, K. R., Hacohen, N., Carr, S. A., Wu, C. J., & Keskin, D. B. (2020, Feb). A large peptidome dataset improves HLA class I epitope prediction across most of the human population . Nat Biotechnol, 38(2), 199-209.

21. Savage, P., Stebbing, J., Bower, M., & Crook, T. (2009, Jan). Why does cytotoxic chemotherapy cure only some cancers? Nat Clin Pract Oncol, 6(1), 43-52.

22. Sawyers, C. (2004, Nov 18). Targeted cancer therapy. Nature, 732(7015), 294-297.

23. Scaltriti, M., & Baselga, J. (2006, Sep 15). The epidermal growth factor receptor pathway: a model for targeted therapy. Clin Cancer Res, 72(18), 5268-5272.

24. Schumacher, T. N., & Schreiber, R. D. (2015, Apr 3). Neoantigens in cancer immunotherapy. Science, 375(6230), 69-74.

25. Sharkey, R. M., & Goldenberg, D. M. (2006, Jul-Aug). Targeted therapy of cancer: new prospects for antibodies and immunoconjugates. CA Cancer J Clin, 56(4), 226-243.

26. Vogelstein, B., Papadopoulos, N., Velculescu, V. E., Zhou, S., Diaz, L. A., Jr., & Kinzler, K. W. (2013, Mar 29). Cancer genome landscapes. Science, 339(6127), 1546-1558.

27. Wang, Q., Douglass, J., Hwang, M. S., Hsiue, E. H., Mog, B. J., Zhang, M., Papadopoulos, N., Kinzler, K. W., Zhou, S., & Vogelstein, B. (2019, Nov). Direct Detection and Quantification of Neoantigens. Cancer Immunol Res, 7(11), 1748-1754.

28. Wang, Q., Wang, X., Tang, P. S., O'Leary G, M., & Zhang, M. (2017, Jun 13). Targeted sequencing of both DNA strands barcoded and captured individually by RNA probes to identify genome-wide ultra-rare mutations. Sci Rep, 7(1), 3356.

29. Yagoda, A., & Petrylak, D. (1993, Feb 1). Cytotoxic chemotherapy for advanced hormone-resistant prostate cancer. Cancer, 71(3 Suppl), 1098-1109.

30. Zubarev, R. A., & Makarov, A. (2013, Jun 4). Orbitrap mass spectrometry. Anal Chem, 55(11), 5288-5296.