CN109682978B - Prediction method for tumor mutant peptide MHC affinity and application thereof - Google Patents

Abstract

The invention provides a prediction method of tumor mutation peptide MHC affinity, which comprises the following steps: firstly, obtaining a tumor specific mutant sequence; (II) for predicting the MHC affinity of the mutant peptide, translating the mutant sequence into an amino acid FASTA sequence with a preset length to obtain a test peptide; respectively evaluating the affinities of the test peptide with MHCI molecules and MHCII molecules, and screening mutant peptides according to a determined grade standard; (III) carrying out proteasome cleavage prediction on the mutant peptide with high affinity screened in the step (II), and screening out the mutant peptide which can be effectively cleaved by proteasome; (IV) Experimental validation Experimental Synthesis procedure (III) validation of the affinity of the mutant peptides to the patient's unique MHC. The invention also provides the application of the prediction method based on the tumor mutation peptide MHC affinity in screening the tumor neoantigen and preparing the personalized tumor vaccine.

Description

Prediction method for tumor mutant peptide MHC affinity and application thereof

Technical Field

The invention relates to the technical field of biology, in particular to an MHC (major histocompatibility complex) affinity prediction method, an application and a personalized tumor vaccine preparation method based on the affinity prediction.

Background

With the development of the NGS sequencing technology, the screening of tumor specific antigens realizes the breakthrough in the technology. In 2013, the Rosenberg team led to the discovery of neoantigens (neoantigens) on tumor cell lines using exon technology and verified their immune response. By using NGS technology and constructing an algorithm model, exon sequencing and transcriptome sequencing can accurately represent DNA and RNA of tumor cells, find out tumor mutation which may cause immune cell recognition, the development of bioinformatics tools improves the screening capability of tumor neoantigens, genome big data and computer algorithm accelerate tumor epitope prediction and MHC (major histocompatibility complex) affinity prediction, and promote the development of cancer personalized immunotherapy (personal vaccenegasident cancer).

Given that cancer prognosis is positively correlated with immune response, specific immune cells are potential targets for immunotherapy. The most potent and dominant antigen presenting cell in the immune system is the Dendritic Cell (DC), and thus such cells are also one of the personalized cancer immunotherapy strategies. Sebastine et al used autologous dendritic cells containing WT1RNA to immunize AML patients and phase II Clinical trial results showed that 43% of AML patients had relapses after chemotherapy and improved overall survival (Blood Clinical Trials and organisms, Aug.23, 2017).

Cancer-personalized immunotherapy another new focus has been to identify tumor-specific muteins using high-throughput sequencing. Recent advances in genome sequencing have shown that there are tens of thousands of different somatic mutations during cancer initiation and progression. Most mutations (passenger mutations) do not confer a tumor growth advantage, but result in instability of the tumor genome. Only a small number of cancer mutations (driver mutations) interfere with normal cellular regulation and contribute to cancer growth and resistance to targeted therapies. To date, about 140 genes have been identified that can drive tumorigenesis. However, both driver and passenger mutations may alter the amino acid coding sequence, collectively referred to as non-synonymous mutations, to form tumor-expressed muteins that are not expressed by normal cells. These abnormal protein sequences are processed intracellularly by proteasome into short peptides, which are then bound by the major histocompatibility complex (MHC, also known as Human Leukocyte Antigen (HLA) in humans) and presented on the cell surface, and thus recognized by T cells as foreign antigens.

In 7 months 2017, two research teams from the American Boston Dana-Farber cancer center (clinical program supported by New Therapeutics) and the American Merzchu Germany (Biopharmaceutical New Technologies, clinical program supported by BioN technology) recently demonstrated a major clinical breakthrough of high throughput sequencing in tumors "personalized vaccines", and published efforts on Nature "entitled" antibiotic personal vaccine for Patients with tumor "(Nature 22991, July 5,2017) and" personalised RNA vaccine mobility-specific therapeutic immunologic immunity accumulator "(Nature 23003, Jyu 5,2017), respectively. In both clinical projects, a tumor tissue sample of a patient is sequenced, unique algorithms are respectively used for predicting which mutations are most likely to cause immune response, and based on prediction results, vaccines based on polypeptide fragments and RNA are respectively developed, and clinical experiments are carried out on middle and late stage recurrent high-risk groups of the late stage melanoma stage 3 or stage 4.

In the Wu doctor led team from the Dana Farbar institute, researchers custom-made polypeptide vaccines containing 13-20 different neoantigens for each patient. Routine clinical data showed that the recurrence rate for this type of advanced tumor was as high as 50%, while the results showed that 4 of 6 patients had no signs of recurrence two years after receiving the vaccine. The other 2 patients showed signs of relapse, but achieved complete remission after receiving PD-1 antibody drug treatment.

Similar results were also found in another set of studies. Two polyepitope RNA vaccines containing neoantigens of 10 different tumors were developed for each patient by the university of minuz team, drawn by professor Ugur Sahin. Of the 13 subjects, 8 had no evidence of recurrence within one year after receiving the vaccine, and 5 other patients had spread the tumor when receiving the vaccine, 2 of which had reduced tumor size after receiving the vaccine, and 1 of which had been completely remitted after receiving the PD-1 antibody drug.

The two studies confirm that the personalized tumor neoantigens predicted by high-throughput sequencing in cancer treatment allow the immune system of a patient to identify the cancer antigens more efficiently and accurately, which provides a very key reference for developing a personalized cancer immunotherapy scheme. Cancer personalized immunotherapy and CAR-T have a consensus on the technical principle: firstly screening individualized cancer specific antigen, and then amplifying the tumor specific antigen in vitro through genetic engineering, thereby activating T lymphocyte and generating active immunity. But the cancer personalized immunotherapy is simpler than the CAR-T technical process, so the clinical application prospect is better.

In general, the prediction of tumor neoantigens is largely divided into two steps:

first, non-synonymous somatic mutations specific to tumor cells are identified, which are potential sources of neoantigens.

Then, screening for neoantigens recognized by T cells based on nonsynonymous somatic mutations is performed, and the screening criteria are mainly based on the expression of the neoantigens, processing, HLA presentation to the surface of tumor cells, and recognition by T cells.

The prediction process of tumor neoantigens is actually a bioinformatics process of multigroup combination and multialgorithm fusion. Based on the result of the sequencing of the full exome of the paired samples, the tumor specific nonsynonymous somatic mutation can be accurately and comprehensively identified, the RNA-seq sequencing with high depth is carried out, and the expressed somatic mutation is screened out. Finally, combining the identified specific HLA of the patient, predicting the affinity between the specific HLA of the patient and the neoantigen corresponding to somatic cell mutation through various algorithms, screening the neoantigen with strong affinity with HLA molecules by adopting a co-censoring strategy, and carrying out target point of targeted tumor specific T cell induction or modification on the patient by the obtained neoantigen through prediction.

For breakthrough of personalized tumor vaccine technology, the key step is to adopt a proper algorithm, and the expression, processing and presentation of the neoantigen and the recognition by MHC and TCR can be predicted according to the sequencing result of a tumor sample. Somatic mutations in tumor cells are random, and the probability of the same neoantigen appearing between different patients with tumors of the same pathological type is less than 1%, and it is not possible to pre-prepare neoantigens. Therefore, each tumor patient must be tested for tumor mutations and analyzed and predicted for neoantigens that may be targets for treatment before they are treated. This is why the study of tumor neoantigens was already open in the first half of the last century, but is limited by detection and analysis techniques and is very slow in clinical application.

Although a certain breakthrough is made in prediction of neoantigens at present, accurate prediction of neoantigens is still the most major challenge in construction of personalized tumor vaccines, the core of the method is an algorithm, each laboratory has a set of own algorithm and flow, the standards are not uniform, and due to the problem of analysis complexity, the industry considers that the accuracy of the current algorithm is estimated to be less than 40%. The method relates to gene data preprocessing, mutation detection, HLA typing identification, expression quantification, protein cleavage, TAP transport, MHC affinity prediction, TCR prediction and the like, wherein the MHC affinity prediction is the most critical loop.

Disclosure of Invention

Aiming at the problems in the identification and screening of the neoantigens, the invention provides an MHC affinity prediction method, which comprises the following steps:

(I) obtaining of tumor-specific mutant sequences

Extracting DNA from the tumor tissue and the normal tissue respectively to perform exon deep sequencing, and analyzing by GMS to obtain a mutant sequence;

(II) prediction of MHC affinity for mutant epitopes

The test peptide is obtained by translating the mutation into a FASTA sequence of 19 amino acids, ideally by substituting 9 amino acids on each side of the amino acid resulting from the mutation. (ii) a

Respectively evaluating the affinity of the test peptide chain with MHCI molecules and MHCII molecules, and screening mutant peptides according to a determined grade standard;

(III) prediction of proteasome cleavage

Performing proteasome cleavage prediction on the mutant peptide chain with high affinity screened in the step (II), and screening out the mutant peptide chain which can be effectively cleaved by a proteasome;

(IV) Experimental verification

And (c) performing MHC competitive binding of the mutant peptide chain screened in the experimental synthesis step (III) and the peptide which is known to be bound to MHC with high affinity.

The method as described above, characterized by: the GMS analysis of the associated mutations comprises the following steps:

1) constructing a 180-and 280-bp library of a normal control and two tumor samples;

2) the library was captured by the agent SureSelect4Human All Exon V6 and duplicates formed by PCR amplification were removed by picard-tools;

3) exon sequence data are generated on an Illumina HiSeq4000 platform, and FastQ data quality control is carried out;

4) BWA (version 0.5.9) was aligned to exonic reads and false positives were filtered in combination with the multiplex settings.

The method as described above, characterized by: the combined multiple-setting filtering false positives comprises the following steps:

1) local alignment re-compares the mismatched regions caused by indels, so as to reduce the error rate of comparison near the indels;

2) stand _ call _ conf in unidifiedGenotyper: a threshold for distinguishing low quality mutation sites from high quality mutation sites during a mutation detection process;

3) VarScan Somatic version 2.2.6 filters false positives.

A method as claimed in any preceding claim, characterized by:

in the step (two), HLA class I and II molecular types of the tumor patients are determined by PCR-rSSO, and further verified by PCR-SSP.

The method as described above, characterized by:

the test peptide is composed of 4 to 10 amino acids each flanking the replacement amino acid resulting from the mutation, preferably 9 amino acids each around.

The method as described above, characterized by: the affinity assessment is the use of NetMHCpan to predict the test peptide chain binding to MHC class i molecules and NetMHCIIpan to predict the test peptide chain binding to MHC class II molecules, respectively. A method as claimed in any preceding claim, characterized by: the ranking criteria include the following five rankings:

(1) new open reading frames (neoORFs) containing predicted binding epitopes;

(2) high affinity somatic nucleotide mutations of less than 150nM due to anchor residue mutations;

(3) a high affinity soma single nucleotide mutation of less than 150nM due to a mutation at other positions of the non-anchor residue;

(4) no new open reading frame predicted to bind the epitope;

(5) a new open reading frame with lower affinity with (2) and (3).

The method as described above, characterized by: after the rank criteria screening is performed, in each ranking group,

firstly, the mutant peptide has strong MHC affinity, and the WT sequence peptide has weak MHC affinity preferentially;

secondly, proto-oncogene mutations are preferred;

again, ordering by difference in mutant peptide affinity;

finally, based on the same affinity, the allele frequency according to the mutation is used as a further ranking

A method as claimed in any preceding claim, characterized by:

the proteasome cleavage prediction is a peptide that predicts a site that can be cleaved by the C-terminus of the proteasome using NetChop C-term 3.0.

The method as described above, characterized by:

the prediction threshold for peptides predicted to be cleaved by the proteasome C-terminus using NetChop C-term 3.0 was 0.5 to distinguish between peptide C-terminus cleavage sites and cleavage sites within peptide epitopes.

The invention also provides the application of the prediction method based on the tumor mutation peptide MHC affinity in screening the tumor neoantigen and preparing the personalized tumor vaccine.

Drawings

FIG. 1 concept diagram of personalized tumor vaccine

FIG. 2 flow chart of personalized tumor vaccine

FIG. 3 distribution of missense mutations identified in tumors at different anatomical locations at different stages of evaluation and identification

FIG. 4 is a photograph of the relevant RNA denaturing PAGE gel electrophoresis and B the northern blot analysis of the relevant RNA.

FIG. 5 wherein A is the results of a 30-lot dendritic cell phenotyping analysis; wherein B is a flow chart of one example of the dendritic cell phenotype.

FIG. 6 is a diagram of RNA expression flow analysis; wherein B is a statistical graph of RNA expression at 10 different voltages; wherein C is a statistical graph of RNA expression of 10 cases at different times.

FIG. 7 is a statistical graph of the expression of 20 different amounts of tumor RNA; wherein B is a statistical chart of 20 different amounts of tumor RNA and quantitative CD40L expression.

FIG. 8 is a statistical chart of IL-12 secretion from 20 cases of dendritic cells into which different amounts of tumor RNA and CD40L were introduced; wherein B is a statistical map of IL-12 secretion for 20 different time periods.

FIG. 9 dendritic cell stimulation-induced tumor neoantigen-specific T cells

Detailed description of the invention

The conceptual diagram from tumor tissue to personalized tumor vaccine shown in fig. 1 includes obtaining tumor tissue, identifying tumor neoantigen, synthesizing RNA in vitro based on the neoantigen and CD40LRNA assisting DC to activate T cells, electrically transferring the synthesized RNA to mature DC cells, and infusing the cells back to the tumor patient, expressing the tumor neoantigen RNA in the DC cells and presenting the RNA to the cell surface, DC homing to lymph nodes, recognizing the presented antigen by T cells, activating T cells, T cells migrating to tumor tissue, and killing tumor cells.

Based on the method of the present invention, based on the above-mentioned conceptual diagram, fig. 2 shows a flow chart of the present invention.

The process of the present invention is further illustrated below with reference to specific examples.

Example 1 tumor neoantigen selection

(one) screening for tumor mutations

Tumor specific coding sequence mutations (tumor specific coding sequence mutations) were obtained using high throughput sequencing, including single, two, and three nucleotide mutations resulting in single amino acid missense mutations and small fragment nucleotide insertions/deletions. Tumor tissues and peripheral blood mononuclear cells (PBMC, normal tissue control) at multiple sites of a patient are obtained, and DNA is extracted for exon sequencing. All tumor tissues are from quick frozen preserved tissues and can also be from paraffin embedded tissues fixed by formaldehyde; normal tissue control PBMCs were cryopreserved as cell pellets after centrifugation. We used our own Genome Modeling System (GMS) to analyze the relevant mutations, and detailed procedures are as follows.

For each patient, using 500ng tumor/PBMC (normal control) genomic DNA samples, one normal and two tumor 180-and 280-bp libraries were constructed, and the antibody SureSelect4Human All Exon V6 capture library, to eliminate DUPLICATES formed during this process due to PCR amplification, we used picard-tools to set REMOVE _ DUPLICATES ═ FALSE to identify these sequences, facilitating the identification of GATK. Generating exon sequence data on an Illumina HiSeq4000 platform, controlling the quality of FastQ data, setting-bqsrBAQGOP 40 in the process, calculating the minimum quality value of two ends of reads to be 2, and if the minimum quality value is smaller than the minimum quality value, not considering the minimum quality value, re-correcting the base quality value of reads in the bam file to enable the quality value of the bases in the reads in the finally output bam file to be closer to the probability of mismatching between a real and a reference genome. BWA (version 0.5.9) was aligned to exonic reads and false positives were filtered in combination with multiple settings: (1) and Local alignment re-aligns the region with mismatch caused by the indel, so that the alignment error rate near the indel is reduced. We realigned these sites with two known reliable indel sites (-knock Mills _ and _1000G _ gold _ standard. indels. hg19. vcf-knock 1000G _ phase1.indels. hg19. vcf). (2) Stand _ call _ conf in unidifiedGenotyper: and (3) a threshold value for distinguishing low-quality variation sites from high-quality variation sites in the variation detection process. Only sites with a quality value above this threshold will be considered high quality. Mutation sites below this quality value will be labeled LowQual in the output. (3) VarScan Somatic version 2.2.6 filters false positives.

(II) prediction of MHC binding peptides

The human major histocompatibility complex (MHC, human is also referred to as HLA) genome is very polymorphic, comprising thousands of alleles, each encoding a different MHC molecule. MHC molecules play an important role in cell-mediated immune responses. The binding of peptides to MHC molecules is a prerequisite for the peptides to be used as immunogens. Peptide bound by MHCII molecule, CD4⁺T cell recognition; MHCI molecule-bound peptide, CD8⁺T cell recognition. HLA class I and class II molecular types in patients were determined by reverse sequence-specific oligonucleotide probes (PCR-rSSO) with further validation by sequence-specific primers (PCR-SSP). The following peptide methods for predicting binding to patient-specific MHC molecules are based on the large and accurate data set derived from the Immune Epitope DataBase (IEDB).

Two independent methods were applied to screen mutant epitopes: 1) NetMHCpan is used to predict patient-specific mutant epitopes that bind to MHC class I molecules; 2) NetMHCIIpan is used to predict patient-specific mutant epitopes that bind to MHC class II molecules. Both methods are pan-specific predictors, encompassing all HLA class molecules with known sequences. The resulting Amino Acid Substitutions (AAS) are each translated into a FASTA sequence of 19 Amino acids, ideally 9 Amino acids on each side of the Amino acid. Each of the 19 amino acid sequences was then evaluated by the two HLA peptide binding algorithms described above, and the epitopes were selected according to the following ranking criteria: (1) new open reading frames (neoORFs) containing predicted binding epitopes; (2) somatic nucleotide mutations with high prediction affinity (<150nM) due to anchor residue mutations; (3) high affinity (<150nM) soma mononucleotide mutations due to mutations at other positions of the non-anchor residue; (4) no new open reading frame predicted to bind the epitope; (5) lower affinity (<150-500nM) to (2) and (3) new open reading frames. The difference in binding affinity of the mutant peptide to the WT sequence peptide was taken into account in each ranking group, and secondly the proto-oncogene mutation became prioritized, otherwise the epitopes were ranked by the difference in mutant peptide affinity, the same affinity epitope, the mutated allele frequency was used as a further differentiation. In addition, various biochemical properties (hydrophobicity, presence of multiple cysteines, etc.) that may affect peptide synthesis or solubility are also considered.

(III) prediction of proteasome cleavage of MHC-binding peptides

The first step of the endogenous antigen processing is the intracellular proteasome cleavage of the C-terminus of the peptide to form a 6-30 amino acid peptide stretch. Thus, native HLA class I peptides are indirect evidence for prediction of proteasome cleavage in vivo. NetChop C-term 3.0 is a database consisting of 1260 MHC class I peptides. Using this database to predict peptides with proteasome C-terminal cleavage sites, a threshold of 0.5 was used to distinguish between peptide C-terminal cleavage sites and cleavage sites within peptide epitopes.

(IV) validation of MHC-bound peptides

The affinity of the specific mutant peptide and HLA is represented by the concentration (IC50) inhibiting the binding of 50% of unlabeled peptide, namely, the high affinity: log (IC50nM) < 3.7; the medium affinity: log (IC50nM) 3.7-4.7; the low affinity: log (IC50nM) 4.7-5.5; and the very low affinity: log (IC50nM) ≥ 6.0.

Based on the above steps, we used exome sequencing to detect tumor-specific missense mutations in tumor samples at different anatomical locations, using three algorithms for union identification. The BRAF allele frequency serves as the upper limit of the tumor variation allele to assess the frequency of other missense mutation-encoding genes. The amino acids corresponding to each missense-encoding mutation were translated into a 19 amino acid sequence and evaluated for candidate peptides capable of binding to HLA and being cleaved by proteasome by relevant HLA software and NetChop C-term 3.0 software. In vitro experiments verify that candidates are further screened based on the predicted difference in HLA binding affinity between the mutant peptide and the WT sequence peptide and/or the predicted high or low HLA binding affinity between the mutant peptide and the WT sequence peptide, and that the HLA-binding peptide can be determined by fluorescence polarization assay. FIG. 3 shows the distribution of missense mutations identified in tumors at different anatomical locations of patient 5 at different stages of evaluation and identification, and patient 5 selected 5 tumor-specific missense mutations as vaccine candidates according to the corresponding criteria.

Example 2 in vitro RNA Synthesis

(1) Synthesis of tumor Neoantigen (Neoantigen) RNA

Synthesizing tumor neoantigen RNA according to the prediction result, and if the mutation is more than or equal to 10, synthesizing the RNA of Tandem gene structures (TMC); if there are few (< 10) mutations, multiple RNAs containing a single mutation are synthesized.

1) Synthesis of RNA for tandem Gene Structure

The tumor neoantigen DNA fragments were synthesized and 7 predicted epitope DNA fragments (each epitope consisting of 19 amino acids with mutations at position 10) joined by non-immunogenic glycine/serine linkers (start linker GGSGGGGSGG, intermediate linker GGSGGGGGGGGG and end linker GGSLGGGGGG) were cloned into a starting vector containing SP and MITD domains (SP, MRVTAPRTLILLLSALTLETWAGS; MITD, IVGIVAGLAVLAVVAVVAVVVVVCRCRKCCRRKSSGGKGGSYSQASSDSAQGSDVSLTA) that optimize HLA class I and class II pathways and backbone sequence elements that improve RNA stability and translation efficiency the DNA was linearized, quantitated by nanodrop ultra-violet spectrophotometer, purified RNA polymerase using T7 under conditions of 7.5mM ATP, CTP, UTP, GTP and 3mM β -S-ACA (D1) cap analogs, magnetic bead electrophoresis and RNA concentration was further assessed by southern blot analysis, pH integrity analysis, endotoxin concentration determination, and further assay.

Optimizing RNA expression and RNA transfer stability into dendritic cells, wherein the RNA expression condition at 7 positions, the influence of 2-7 RNA repeated expressions on specific T cell efficiency and the influence of RNA stability by optimizing amino acid codons of a linker between RNAs are included. MART-1 RNA in different positions and in a repeated tandem of 2-7 was constructed and then electroporated into dendritic cells, and evaluated by flow cytometry for RNA expression and MHC Dextran for specific T cell efficiency effects. Constructing different MART-1 RNAs subjected to amino acid codon optimization, labeling GFP fluorescent labels, then electrically transferring the RNAs into dendritic cells, and evaluating the stability of the RNAs by detecting the percentage of GFP + dendritic cells within 1-24h through a flow cytometer so as to reflect the duration of antigen presentation of the dendritic cells to immune cells. According to the optimization result, RNA expression at 7 positions has no difference, repeated expression has no obvious influence on the efficiency of the specific T cells, and the optimized optimal linker is as follows: the initial linker was GGSGGGGSGG, the intermediate linker was GGSGGGGGGGGG and the terminal linker was GGSLGGGGGG.

2) Synthesis of Single mutant RNA

RNA with only a single mutation, polyA tail containing 70 adenosine residues and I-type cap structure are directly synthesized in vitro, and a fluorescent protein tag is marked.

Based on the experimental procedure described above, 5 RNAs for the 5 tumor neoantigens of the patients were synthesized in vitro according to the prediction and in vitro validation results of example 1.In FIG. 4, A is the electrophoresis chart of the related RNA denaturing PAGE gel, and in FIG. 4, the quality of RNA analyzed by northern blot is shown, and the results show that the quality and size of the RNA synthesized by us are in accordance with the standard.

(2) MART-1 RNA and CD40L RNA Synthesis

MART-1 RNA is used as a standard substance for subsequent evaluation of specific CTL effect induced by dendritic cells; CD40L RNA is transferred into dendritic cells to increase DC IL-12 expression and promote differentiation of specific CTL.

1)MART-1 RNA

Total RNA was extracted from SK-Mel-28 cells, then reverse transcribed using Powerscript reverse transcriptase to produce total cDNA, followed by the use of MART-1 specific primers under the action of PFU enzyme (Stratagene): MART-1 forward: 5'-CCACCATGCCAAGAGAAG-3' and MART-1 reverse: 5'-TTAAGGTG AATAAGGTGG-3', a cDNA encoding MART-1 was obtained. The PCR product was cloned into pGem4Z 64T vector. To obtain a mutation at position 27 of MART-1 open reading frame, alanine was converted to leucine by changing the codon GCC to CTC, site-directed mutagenesis was performed using the QuickChange method (Stratagene). The resulting PCR fragment carried the T7 promoter, CD40L 5' -UTR sequence, HA tag, followed by specific antigen coding sequence, and T64 tail. These PCR fragments serve as templates for in vitro transcription to generate a polyA tail containing specific antigenic RNA comprising CD40L 5' -UTR, HA tag and 64 adenosine residues. In vitro Transcription was performed using the AmpliScribe T7-Flash Transcription Kit from Cellscript according to the manufacturer's instructions. The uncapped RNA was then capped using the CellscriptScriptCap m7G Capping System and ScriptCap2' -O-methyltransferase to generate a type I cap structure.

2)CD40L RNA

Total RNA was extracted from PMA-activated normal volunteer T cells, then reverse transcribed into total cDNA using the Gene Amp Gold kit (Applied Bioscience), followed by CD40L specific primers: CD40L forward: 5'-GCATGATCGAAACATACAACC-3' and CD40L reverse: 5'-GTATTATGAAGACTCCCAGCG-3', 0.8kbCD40L DNA fragment was obtained. The purified PCR fragment was subcloned into pCR2.1 (pCR2.1CD40L. DELTA. XE-MET1) vector. This CD40L cDNA template, used a modified CD40L 5' -UTR for optimal transcription initiation and deleted the first ATG in the open reading frame to generate a single CD40L RNA product for optimal CD40L protein expression. In vitro Transcription was performed using the AmpliScribe T7-Flash Transcription Kit from Cellscript according to the manufacturer's instructions. The uncapped RNA was then capped using the CellscriptScriptCap m7G Capping System and ScriptCap2' -O-methyltransferase to generate a type I cap structure.

Example 3 Dendritic Cell (DC) culture and maturation

1) Monocyte induction generation of immature dendritic cells

Peripheral Blood Mononuclear Cells (PBMC) were isolated from leukocyte concentrates of healthy volunteers by Ficoll-histopaque density centrifugation, washed 4 times with Phosphate Buffered Saline (PBS) at room temperature, and resuspended 2 × 10 in 30mL of AIM-V medium (Invitrogen)⁸Adding PBMC into T150 culture flask, culturing at 37 deg.C and 5% CO2 under RH of 75% or more for 2 hr, removing non-adherent substanceThe cells were rejoined with GM-CSF (R) containing 800U/ml&D) And 800U/mL IL-4 (R)&D) The X-VIVO15 culture medium is cultured for 5 days at 37 ℃ and 5% CO2 under the condition of being equal to or more than 75% RH.

2) Cytokine stimulation of dendritic cell maturation

Day 5, TNF-a (20ng/ml), IFN-r (800U/ml) and PGE were replaced₂(2ug/ml) X-Vivo15 medium, cultured overnight at 37 deg.C under 5% CO2,. gtoreq.75% RH, and harvested on the sixth day.

FIG. 5A shows the results of phenotypic analysis of 30 batches of induced cultured dendritic cells; FIG. 5B shows a diagram of phenotypic flow analysis of one example of dendritic cells, and the results show that we can obtain phenotypically stable mature dendritic cells.

Example 4 introduction of RNA into dendritic cells

RNA was introduced into dendritic cells using electrotransfer the DCs were harvested, washed with PBS, and then washed with 4 × 10⁷The concentration of/mL was resuspended in ice-cold costorsol and placed on ice. DC and mRNA were mixed, placed in a 4mm electroporation cuvette, and electroporated using a Biorad GenePulserXcell Eukarryotic System electroporation apparatus (Voltage 300V, capacitance 200uf, resistance 150ohms, time: 10 ms). The DC after the electrotransformation was immediately treated with a solution containing 800U/ml GM-CSF (R)&D) And 800U/mL IL-4 (R)&D) The culture medium of X-VIVO15 is 1X10⁶At a concentration of/ml, in a 6-well plate or T75 flask, at 37 deg.C, 5% CO2,. gtoreq.75% RH for 4 hours, and then DC is incubated at 3 × 10⁶The/ml concentration jelly was present in 10% dimethyl sulfoxide, 10% glucose (50% solution) and 80% autologous plasma.

Example 5 RNA electrotransfer parameter optimization

The voltage and time for introducing RNA into dendritic cells were optimized and evaluated by cell viability, cell growth status and RNA expression efficiency.

The voltage is an important factor of electrotransfection, and the membrane permeability of cells is increased or pores are formed under the action of an electric field so as to complete the transfection process. Most mammalian cells preferably have an electrical transfer voltage of 400-900V/cm, so that if an electrical transfer cup with an electrode distance of 4mm is used, the voltage range is: 160V-360V. Cell diameter also affects voltage, with smaller diameter cells requiring higher voltage and larger diameter cells requiring lower voltage. Therefore, starting with a voltage of 100V, each 50V increment is optimized.

When the mammalian cells are electrotransfected, square wave pulses are selected, so that the transfection efficiency and the cell survival rate are higher. In square wave pulses, the electrical switching time can be set directly. The electrotransfer time of most mammalian cells is 10-40 msec. In the parameter optimization, the power conversion time should be reduced by referring to the principle that the voltage is increased, and the power conversion time is increased by reducing the voltage. Thus starting with 45msec, optimization is performed with 5msec reduction each time.

We optimized the voltage and time for electrotransfer by introducing CD40L RNA into dendritic cells, and evaluated by cell viability, cell growth status, and RNA expression efficiency. The results showed that cell viability and RNA expression efficiency were best at 300V voltage and 10ms shock time (FIG. 6).

Example 6 optimization of conditions for transfer of multiple RNAs into dendritic cells

Under the condition of optimized electrotransfer, 1 × 10 is introduced into the first time⁶DC RNA amount encoding tumor neoantigen was optimized by cell survival, cell growth status and TCRm⁺/GFP⁺The percentage of cells was evaluated. The results show that in the case of tumor RNA electrotransfer only, the survival rate of the cells is reduced with the increase of the RNA amount, and the 10-8ug of RNA amount is most obvious, which can cause 80% of cell death, and the 7ug of RNA/10 is comprehensively evaluated⁶Cell mass, TCRm⁺/GFP⁺Cell percentage was optimal (fig. 7A); to promote secretion of dendritic cell cytokine, we mix and electroporate a certain amount of CD40L RNA, and as the RNA amount increases, the survival rate of cells decreases, and at the same time, TCRm passes⁺Cells and CD40L⁺Cell percentage analysis showed 7ug RNA/10 electrotransfer⁶The cell, tumor RNA/CD40L was 1:2, which is the optimal tumor RNA electrotransfer condition (FIG. 7B).

Example 7 evaluation of RNA expression and antigen processing

1) Expression of RNA for tumor neoantigens and antigen processing

Fluorescent protein and SVG9 label are added during RNA synthesis, RNA is electrically transferred to dendritic cells, flow cytometry is used for detection, and RNA expression and antigenic peptide processing conditions are analyzed through TCRm + fluorescent protein + cells. TCRm is a T cell receptor mimic monoclonal antibody, and can detect HLA/SVG9 peptide compound expressed on the cell surface, thereby detecting the RNA expression condition, and fluorescent protein can detect the RNA transfer efficiency.

2) CD40L RNA expression

Resuscitating DC cells, adding the DC cells into a flow tube, centrifuging at 1500rpm for 5min, discarding the supernatant, adding 2ul of Dead cell marker, mixing uniformly, irradiating with white light for 10min, adding 2ml of FBS staining buffer, centrifuging, discarding the supernatant, adding 5ul of marker Stop Reagent and 250ul of BD cytotox, mixing uniformly, incubating at 4 ℃ for 10min, adding 2ml of staining buffer, centrifuging, discarding the supernatant, washing the cells twice with 2ml of 1X Perm wash, discarding the supernatant, adding 20ul of APC-CD40L antibody, dyeing in the dark at room temperature for 30min, washing the cells twice with 2ml of 1X Perm wash, centrifuging, discarding the supernatant, resuspending the cells with 300ul of FBS staining buffer, and detecting the expression of CD40L by a flow cytometer.

EXAMPLE 8 evaluation of secretion of IL-12p70 by DC

The expression of IL-12 in dendritic cells can promote the differentiation of specific CTL, so the effect of the dendritic cells is evaluated by detecting the secretion of IL-12p70 by DC, and the resuscitated DC cells and X-VIVO15 culture medium are 1 × 10⁶cell/ml, add 1ml of cells to 24 well plates (1 × 10)⁶cell/well), incubated overnight at 37 ℃ and culture supernatant collected for detection of IL-12p 70. The IL-12p70 secreted by DCs was determined using the Human Soluble Protein Master Buffer Kit (BDbioscience) and Human IL-12p70 Flex Set (BD bioscience). Diluting IL-12 standard and test supernatant samples, adding capture magnetic beads, incubating at room temperature for 1 hour, adding PE labeled detection reagent, incubating at room temperature for 2 hours, adding 1ml of washing buffer, centrifuging at 200g for 5 minutes, resuspending cells in 300ul of buffer, detecting by a BD flow cytometer, and analyzing data by FCAP software.

We evaluated IL-12 secretion by dendritic cells at different levels of RNA and for different time periods when tumor RNA/CD40L was 1: 2. The results showed that 7ug RNA/10 was electrotransferred⁶Cells, 24h dendritic cells, secreted IL-12 in optimal amounts (FIG. 8).

Example 9 Dextran assay to evaluate tumor neoantigen-specific T cells induced by dendritic cell stimulation

(1) Amplification of tumor neoantigen specific CTL

Using CD8⁺T cell isolation kit (stem cell) CD8+ T cells purified from PBMCs. Adding 5x10 into 6-hole plate (corning) according to the ratio of 1:10⁵RNA-introduced DCs and 5X10⁶CD8⁺T cells were co-cultured in 5ml of R-10 medium (containing 0.2U/ml IL-12 and 10ng/ml IL-7) for 3 days, on day 3, R-10 medium containing 20U/ml IL-2 and 10ng/ml IL-7 was replaced, and further cultured for 3 days, DCs into which RNA was introduced were recovered on day 7, co-cultured cells were harvested, and a new round of co-culture was started at a ratio of 1:10 of DCs into which RNA was introduced and CD 8. 3 co-cultures were performed in total, and 3 days after the last co-culture stimulation, tumor neoantigen-specific CTLs and their corresponding functions were detected in the co-cultures.

(2) Flow cytometry detection of tumor neoantigen-specific CTLs in cocultures and their corresponding functions (intracellular cytokine detection)

On day 17 (3 days after 3 rounds of coculture stimulation), coculture cells were harvested, RNA-introduced DCs were revived, DCs and CD8 were cocultured at a 1:10 ratio for 4 hours, 4ul of FITC CD107a antibody, 0.25ul of Brefeldin A and 0.16ul of Monensin (BD Biosciences) were added simultaneously with the incubation for 20 minutes after 1ml of Aquade was added, and after the incubation was completed, surface staining was performed, APC-neoantigen dextramimer was washed, PE-cy7 CD8, BV605CD45RA and PECy5CD28 were added, in order to further determine CTL intracellular cytokine changes, intracellular staining was performed Cytofix buffer (BDsconce) fixed cells, perm/wash and Cytofix buffer (BDsconce) fixed cells, further fixed and buff/bufferfix buffer (BDBiosciences) were added, IFN-CD RA was added, and further IFN-CD 19 + cell counts were analyzed by FACS-fluorescence analysis, CD 21 + CD 19 + CD-CD 11 + CD-CD 19 + CD-CD 11 + PCR-CD 19 + count, CD 11 + PCR + count was analyzed by using a counter.

The results show that we introduce the patient5 dendritic cells of tumor RNA can induce about 7% of tumor-specific CD8⁺T lymphocytes, while these cells highly express TNF-a and IFN-r and have cytotoxic function (FIG. 9).

It should be noted that: although the present invention has been described in detail with reference to the above embodiments, it should be understood by those skilled in the art that: modifications and equivalents may be made thereto without departing from the spirit and scope of the invention and it is intended to cover in the claims the invention any modifications and equivalents.

SEQUENCE LISTING

<110> Dingping

<120> method for predicting tumor mutant peptide MHC affinity and application thereof

<130>2019

<150>CN201711242932.7

<151>2017-11-30

<160>9

<170>PatentIn version 3.3

<210>1

<211>10

<212>PRT

<213> Artificial sequence (artificial sequence)

<400>1

Gly Gly Ser Gly Gly Gly Gly Ser Gly Gly

1 5 10

<210>2

<211>12

<212>PRT

<213> Artificial sequence (artificial sequence)

<400>2

Gly Gly Ser Gly Gly Gly Gly Gly Gly Gly Gly Gly

1 5 10

<210>3

<211>10

<212>PRT

<213> Artificial sequence (artificial sequence)

<400>3

Gly Gly Ser Leu Gly Gly Gly Gly Gly Gly

1 5 10

<210>4

<211>26

<212>PRT

<213> Artificial sequence (artificial sequence)

<400>4

Met Arg Val Thr Ala Pro Arg Thr Leu Ile Leu Leu Leu Ser Gly Ala

1 5 10 15

Leu Ala Leu Thr Glu Thr Trp Ala Gly Ser

20 25

<210>5

<211>55

<212>PRT

<213> Artificial sequence (artificial sequence)

<400>5

Ile Val Gly Ile Val Ala Gly Leu Ala Val Leu Ala Val Val Val Ile

1 5 10 15

Gly Ala Val Val Ala Thr Val Met Cys Arg Arg Lys Ser Ser Gly Gly

20 25 30

Lys Gly Gly Ser Tyr Ser Gln Ala Ala Ser Ser Asp Ser Ala Gln Gly

35 40 45

Ser Asp Val Ser Leu Thr Ala

50 55

<210>6

<211>18

<212>DNA

<213> Artificial sequence (artificial sequence)

<400>6

ccaccatgcc aagagaag 18

<210>7

<211>18

<212>DNA

<213> Artificial sequence (artificial sequence)

<400>7

ttaaggtgaa taaggtgg 18

<210>8

<211>21

<212>DNA

<213> Artificial sequence (artificial sequence)

<400>8

gcatgatcga aacatacaac c 21

<210>9

<211>21

<212>DNA

<213> Artificial sequence (artificial sequence)

<400>9

gtattatgaa gactcccagc g 21

Claims (7)

1. A method for predicting the MHC affinity of tumor mutant peptide includes the following steps:

(I) obtaining of tumor-specific mutant sequences

Extracting DNA from the tumor tissue and the normal tissue respectively to perform exon deep sequencing, and analyzing by GMS to obtain a mutant sequence;

(II) prediction of MHC affinity for mutant peptides

Translating the mutant sequence into a FASTA sequence of 19 amino acids, and replacing 9 amino acids on each side of the amino acids to obtain a test peptide;

respectively evaluating the affinities of the test peptide with MHCI molecules and MHCII molecules, and screening mutant peptides according to a determined grade standard;

wherein said affinity assessment is the prediction of said test peptide chain binding to MHC class I molecules using NetMHCpan and NetMHCIIpan, respectively, to MHC class II molecules;

wherein the ranking criteria include:

(1) new open reading frames (neoORFs) containing predicted binding epitopes; (2) high affinity somatic nucleotide mutations of less than 150nM due to anchor residue mutations;

(3) a high affinity soma single nucleotide mutation of less than 150nM due to a mutation at other positions of the non-anchor residue;

(4) no new open reading frame predicted to bind the epitope;

(5) a new open reading frame with lower affinity than that of (2) and (3);

after the rank criteria screening is performed, in each ranking group,

firstly, the mutant peptide has strong MHC affinity, and the WT sequence peptide has weak MHC affinity preferentially;

secondly, proto-oncogene mutations are preferred;

again, ordering by difference in mutant peptide affinity;

finally, based on the same affinity, the allele frequency according to the mutation is used as a further ranking;

(III) prediction of proteasome cleavage

Performing proteasome cleavage prediction on the mutant peptide with high affinity screened in the step (II), and screening out the mutant peptide which can be effectively cleaved by proteasome;

(IV) Experimental verification

And (3) carrying out MHC competitive combination experiments on the mutant peptide screened in the experimental synthesis step (III) and the peptide which is known to be combined with MHC with high affinity, and verifying the affinity of the mutant peptide and the peculiar MHC of the patient.

2. The method of claim 1, wherein: the GMS analysis of the associated mutations comprises the following steps:

1) constructing a 180-and 280-bp library of a normal control and two tumor samples;

2) aaglient SureSelect4Human All Exon V6 capture libraries and use picard-tools to remove duplicates formed by PCR amplification;

3) exon sequence data are generated on an Illumina HiSeq4000 platform, and FastQ data quality control is carried out;

4) BWA aligned exon reads, combined with multiple settings to filter false positives.

3. The method of claim 2, wherein: the combined multiple-setting filtering false positives comprises the following steps:

1) local alignment re-compares the mismatched regions caused by indels, so as to reduce the error rate of comparison near the indels;

2) stand _ call _ conf in unidifiedGenotyper: a threshold for distinguishing low quality mutation sites from high quality mutation sites during a mutation detection process;

3) VarScan Somatic version 2.2.6 filters false positives.

4. A method according to any of claims 1-3, characterized by:

in the step (two), HLA class I and II molecular types of the tumor patients are determined by PCR-rSSO, and further verified by PCR-SSP.

5. The method of claim 4, wherein:

the proteasome cleavage prediction is a peptide that predicts a site that can be cleaved by the C-terminus of the proteasome using NetChop C-term 3.0.

6. The method of claim 5, wherein:

the prediction threshold for peptides predicted to be cleaved by the proteasome C-terminus using NetChop C-term 3.0 was 0.5 to distinguish between peptide C-terminus cleavage sites and cleavage sites within peptide epitopes.

7. Use of the prediction method according to any one of claims 1 to 6 for screening tumor neoantigens and for preparing personalized tumor vaccines.

Images