Method for accurately identifying antigen miscoordination sites of induced antibody by combining AI and single antigen cytometry and application of method

Abstract

The invention provides a method for precisely identifying antigen mismatch sites of an induced antibody by combining AI and single antigen cytometry and application, and the method comprises the following steps: firstly, finding specific amino acid mismatch sites of donor antigens through sequence comparison of donor antigens and receptor antigens, and then pre-classifying the mismatch sites through AI protein conformation analysis; and finally, verifying the pre-classified mismatched sites according to a binding reaction result of receptor serum and donor single antigen cells without interference antigens. The AI technology and experimental verification are combined, false positive antibody detection caused by mismatch tolerance is reduced, the reliability and practicability of antigen determinant identification are improved, more comprehensive and more reliable rejection antigen classification information can be provided for clinicians, and precise medical decision is supported.

Description

Technical Field

[0001] This invention belongs to the field of biomedical technology and relates to a method and application for accurately identifying antigen mismatch sites that induce antibodies by combining AI and single antigen cell therapy. Background Art

[0002] Transplantation is a medical technique that involves surgically or interventionally implanting cells, tissues, or organs from one individual into the same or other parts of the body, either directly or in another, to replace or enhance the function of the original cells, tissues, or organs. Depending on the type of graft, it is classified as cell transplantation, tissue transplantation, and organ transplantation. The individual providing the graft is called the donor, and the individual receiving the graft is called the recipient.

[0003] Between transplant recipients, mismatches of polymorphic genes can occur, causing donor-specific antigens to induce the production of donor-specific antibodies (DSA) in the recipient. DSA in the recipient can then trigger antibody-mediated rejection (AMR), leading to graft failure and recipient death. Therefore, identifying donors with the fewest mismatched antigens is crucial for ensuring long-term graft survival. Long-term follow-up analysis of large transplant databases of recipient groups with varying numbers of mismatched antigens has also confirmed that fewer antigen mismatches result in a lower incidence of AMR (Transpl Immunol. 2022 Dec;75:101706. doi: 10.1016 / j.trim.2022.101706;Transpl Immunol. 2023 Oct;80:101861. doi: 10.1016 / j.trim.2023.101861.).

[0004] While macro-analysis of the transplant population confirms a correlation between the quantity of mismatched antigens and antigenic determinants between donors and recipients and the incidence of acute rheumatoid arthritis (AMR), when clinicians face a specific transplant recipient and select the best donor, based on the principle of minimizing AMR occurrence, they need to consider not only selecting donors with fewer mismatched antigens, especially antigenic determinants, but also the quality of these mismatched antigens and determinants. They should choose donor mismatched antigens or antigenic determinants that are less likely to induce antibody responses (permissive mismatch) and avoid those that easily induce antibodies (taboomismatch). Relatively speaking, "taboomismatch" mismatches between donors and recipients, especially in terms of antigens and antigenic determinants, are more important because they are the actual target sites for inducing antibody responses. Only "taboo mismatches" truly determine the possibility that a donor mismatch antigen will cause the recipient to produce rejection antibodies, while "permissible mismatches" do not cause antibody reactions under normal circumstances. Their existence interferes with the current analysis of mismatch antigens and antigenic determinants.

[0005] However, it is important to note that in reality, a large number of "natural antibodies" exist in the human body. These antibodies recognize antigens or antigenic determinants that are not normally exposed and are defined as "allowable mismatches." The main reason for this is that humans often ingest animal-based foods with amino acid sequences highly similar to human antigens. These foods contain antigens with high homology to human antigens. When these animal antigens are digested by stomach acid and digestive enzymes, they undergo denaturation, exposing hidden, originally "allowable mismatch" antigenic sites. This leads to the production of antibodies belonging to these "allowable mismatch" antigenic determinants, which are typically defined as natural antibodies. These natural antibodies are not rejection antibodies because their target sites do not exist on the surface of the graft donor's natural antigens.

[0006] Current antibody detection methods can detect both "prohibited mismatch" and "allowable mismatch" antibodies simultaneously, making it impossible to distinguish between them during analysis. DSA antibodies against donor HLA produced by the recipient are the most studied type of antibody in the transplantation field. Currently, the detection of HLA antibodies mainly relies on purified HLA antigens, which are detected by ELISA or Luminex methods (Transplantation. 2003 Jan 15;75(1):43-9. doi: 10.1097 / 00007890-200301150-00008; Ann BiolClin (Paris). 2004 Jan-Feb;62(1):93-8.). However, when using purified HLA to detect its corresponding antibodies, the preparation of the detection reagent involves a series of experimental steps such as purification, elution, concentration, coating, and fixation of HLA antigens. This often leads to significant antigen denaturation when HLA is finally coated onto a solid-phase carrier, resulting in the detection of false positive antibodies due to the presence of denatured HLA antigens. These false positive antibodies refer to antigenic sites that are not exposed on natural antigens and are defined as "tolerable mismatches," and cannot cause AMR (Transplantation 2009;88:226–230. DOI: 10.1097 / TP.0b013e3181ac6198;Chinese Medical Journal, 2022, 102(10): 705-717. DOI: 10.3760 / cma.j.cn112137-20210830-01973;Chinese Journal of Organ Transplantation May 2025; 46(5):344-50. DOI: 10.3760 / cma.j.cn4212032024062600152). This is not conducive to the precise prevention and control of AMR. Summary of the Invention

[0007] To address the aforementioned technical problems, this invention provides a method and application for accurately identifying antigen mismatch sites that induce antibodies by combining AI and single-antigen cell therapy. The identification method of this invention, which first identifies mismatch sites, then performs pre-classification, and finally conducts experimental verification, provides "prohibited mismatch" and "allowable mismatch" results, thereby improving the accuracy of antibody detection.

[0008] The technical solution adopted by this invention to achieve its technical objectives is as follows:

[0009] This invention provides a method for accurately identifying antigen mismatch sites that induce antibodies by combining AI and single antigen cell therapy, comprising: firstly, identifying amino acid mismatch sites unique to donor antigens by sequence alignment of donor and recipient antigens; secondly, pre-classifying the mismatch sites by AI protein conformation analysis; and finally, verifying the pre-classified mismatch sites by the binding reaction results of recipient serum and donor single antigen cells (SACs) with the interfering antigen knocked out.

[0010] Preferably, the method for discovering donor-specific amino acid mismatch sites by comparing donor and recipient antigen sequences includes: inputting high-resolution genotyping data of donor and recipient antigens and their corresponding full-length amino acid sequences, and identifying donor-specific amino acid mismatch sites through a multiple sequence alignment algorithm.

[0011] Preferably, the pre-classification of mismatch sites by AI protein conformation analysis includes: inputting the donor antigen sequence into the AI ​​protein structure prediction model, and pre-classifying the mismatch site as a "forbidden mismatch" or "allowable mismatch" site based on whether the mismatch site is "exposed" or "hidden" in the spatial structure of the corresponding antigen protein.

[0012] More preferably, the AI ​​protein structure prediction model includes AlphaFold2, DI-TASSAR, GEOFlow V2, etc.

[0013] Preferably, the validation of pre-classified mismatch sites by the binding reaction results of recipient serum and donor single antigen cells with the interfering antigen knocked out includes:

[0014] a) Co-incubate donor single antigen cell lines or clones with the interfering antigen knocked out with recipient serum;

[0015] b) Detect whether the corresponding antibody is bound to the cell surface expressing the donor monoantigen by flow cytometry;

[0016] c) Based on the results of the antibody binding reaction, confirm whether the donor-specific amino acid mismatch site is a "forbidden mismatch" or a "permissible mismatch".

[0017] More preferably, if the binding reaction between the recipient's serum and the donor's single antigen cells is positive, the mismatched donor-derived specific site is identified as a "prohibited mismatch"; if the binding reaction between the recipient's serum and the donor's single antigen cells is negative, the mismatched donor-derived specific site is identified as a "permissible mismatch".

[0018] This invention also provides the application of the above method in the preparation of rejection antibody detection and analysis products.

[0019] The innovation of this invention lies in:

[0020] (1) Dual verification mechanism:

[0021] AI conformation prediction (theory) and SAC functionality combined with experiments (empirical) jointly determine mismatch properties, overcoming the shortcomings of purely theoretical derivations in existing technologies.

[0022] (2) Breakthrough in clinical value:

[0023] For the first time, it has achieved a precise distinction between "forbidden mismatch" (targets that actually induce AMR) and "allowable mismatch" (non-pathogenic mismatch), guiding clinicians to avoid high-risk donors.

[0024] (3) Technological synergy:

[0025] Integrating bioinformatics (AI structure prediction), gene editing (SAC construction), and immune detection technology (flow cytometry) to form a closed-loop verification process.

[0026] The beneficial effects of the present invention are:

[0027] (1) Improve the success rate of transplantation:

[0028] By accurately identifying "contraindicated mismatches" and "allowable mismatches," donors less likely to elicit antibody responses can be selected, reducing the incidence of antibody-mediated rejection (AMR) and thus improving the long-term survival rate of grafts.

[0029] (2) Precision prevention and control of AMR:

[0030] a. Provide scientific evidence that is both theoretically grounded and experimentally supported to guide the precise prevention and treatment of AMR.

[0031] b. To help clinicians more accurately assess immune compatibility between donors and recipients when selecting donors.

[0032] (3) Reduce false positive results:

[0033] Current antibody detection methods often result in false positives due to antigen denaturation. This invention, through AI models and SAC validation, reduces false positive antibody detection caused by "tolerable mismatches" and improves detection accuracy.

[0034] (4) Improve the accuracy of immunogenicity assessment:

[0035] Existing methods mainly focus on the "quantity" of mismatches, while this invention also considers the "quality" of mismatches, namely the immunogenicity of the mismatched antigens ("prohibited mismatches" or "permissible mismatches"), providing a more refined means of immunogenicity assessment.

[0036] (5) Combining experimentation and theory:

[0037] This invention combines AI technology with experimental verification, which overcomes the shortcomings of traditional methods that rely solely on theoretical inference, and improves the reliability and practicality of antigenic determinant identification.

[0038] (6) Optimize donor selection:

[0039] This will provide clinicians with a more scientific basis for donor selection, avoid selecting donors that may induce a strong immune response, and thus optimize transplantation treatment plans.

[0040] (7) Reduce interfering factors:

[0041] Validation was performed using single antigen cell lines (SACs) that knock out interfering antigens, thus avoiding interference from non-HLA antibodies and improving the reliability of the validation results.

[0042] (8) Provide comprehensive analysis results:

[0043] By combining AI model predictions and experimental validation results, clinicians can be provided with more comprehensive and reliable information on the classification of rejection antigens, supporting precision medicine decisions. Attached Figure Description

[0044] Figure 1 This is a flowchart of the method for accurately identifying antigen mismatch sites that induce antibodies by combining AI and single-antigen cell therapy according to the present invention.

[0045] Figure 2 This shows the positional characteristics of the two mismatched amino acid sites on their respective antigen protein spatial structures.

[0046] Figure 3 The validation results for mismatch site classification based on single antigen cells (SACs) are shown: A. Binding reaction results of recipient serum and donor HLA-C*03:03 single antigen cell line; B. Binding reaction results of recipient serum and donor HLA-C*08:01 single antigen cell line. Detailed Implementation

[0047] The present invention will now be described in detail with reference to specific embodiments. The following specific embodiments will help those skilled in the art to further understand the present invention, but do not limit the present invention in any way.

[0048] Example

[0049] This embodiment uses two pairs of donors and recipients with HLA-C antigen mismatches to illustrate the methods for discovering, pre-classifying, and experimentally validating antigen-specific amino acid sites for "forbidden mismatches" and "allowable mismatches," respectively. Combined with... Figure 1The process is described in the following steps: identifying amino acid mismatch sites through sequence alignment, pre-classifying mismatch sites through AI conformation analysis, and verifying the pre-classified epitopes through the binding reaction results of recipient serum and donor single antigen cells (SAC).

[0050] For specific preparation methods of single antigen cells (SAC), please refer to the application filed by the applicant on May 28, 2025, with application number 2025107000674, entitled "Transplantation Type Targeted Single Antigen Cell Combination and Its Preparation Method and Application".

[0051] 1. Through sequence alignment, amino acid mismatch sites between the donor and recipient were identified:

[0052] (1) First pair: Donor C*03:03, Recipient C*03:04. Sequence alignment results show that at amino acid position 91, donor antigen C*03:03 is R (arginine, Arg), and recipient antigen C*03:04 is G (glycine, Gly). The 91R of the donor antigen is a potential core mismatched amino acid that can induce a recipient antibody response.

[0053] ① Donor HLA-C*03:03 (91R):

[0054] GSHSMRYFYTAVSRPGRGEPHFIAVGYVDDTQFVRFDSDAASPRGEPRAPWVEQEGPEYWDRETQKYKRQAQTDRVSLRNLRGYYNQSEARSHIIQRMYGCDVGPDGRLLRGYDQYAYDGKDYIALNEDLRSWTAADTAAQITQRKWEAAREAEQLRAYLEGLCVEWLRRY LKNGKETLQRAEHPKTHVTHHPVSDHEATLRCWALGFYPAEITLTWQWDGEDQTQDTELVETRPAGDGTFQKWAAVVVPSGEEQRYTCHVQHEGLPEPLTLRWEPSSQPTIPIVGIVAGLAVLAVLAVLGAVVAVVMCRRKSSGGKGGSCSQAASSNSAQGSDESLIACKA

[0055] ②Recipient HLA-C*03:04 (91G): GSHSMRYFYTAVSRPGRGEPHFIAVGYVDDTQFVRFDSDAASPRGEPRAPWVEQEGPEYWDRETQKYKRQ AQTDRVSLRNLRGYYNQSEAGSHIIQRMYGCDVGPDGRLLRGYDQYAYDGKDYIALNEDLRSWTAADTAAQITQRKWEAAREAEQLRAYLE GLCVEWLRRYLKNGKETLQRAEHPKTHVTHHPVSDHEATLRCWALGFYPAEITLTWQWDGEDQTQDTELVETRPAGDGTFQKWAAVVVPSGEEQRYTCHVQHEGLPEPLTLRWEPSSQPTIPIVGIVAGLAVLAVLAVLGAVVAVVMCRRKSSGGKGGSCSQAASSNSAQGSDESLIACKA

[0056] (2) Second pair: Donor C*08:01, Recipient C*08:03. Sequence alignment results show that at amino acid position 175, donor antigen C*08:01 is G (glycine, Gly), and recipient C*08:03 is R (arginine, Arg). The 175G of the donor antigen is a potential core mismatched amino acid that can induce a recipient antibody response.

[0057] ① Donor HLA-C*08:01 (175g):

[0058] CSHSMRYFYTAVSRPGRGEPRFIAVGYVDDTQFVQFDSDAASPRGEPRAPWVEQEGPEYWDRETQKYKRQAQTDRVSLRNLRGYYNQSEAGSHTLQRMYGCDLGPDGRLLRGYNQFAYDGKDYIALNEDLRSWTAADTAAQITQRKWEAARTAEQLRAYLEGTCVEWLRRY LENGKKTLQRAEHPKTHVTHHPVSDHEATLRCWALGFYPAEITLTWQRDGEDQTQDTELVETRPAGDGTFQKWAAVVVPSGEEQRYTCHVQHEGLPEPLTLRWGPSSQPTIPIVGIVAGLAVLAVLAVLGAVMAVVMCRRKSSGGKGGSCSQAASSNSAQGSDESLIACKA

[0059] ② Recipient HLA-C*08:03 (175R)

[0060] CSHSMRYFYTAVSRPGRGEPRFIAVGYVDDTQFVQFDSDAASPRGEPRAPWVEQEGPEYWDRETQKYKRQAQTDRVSLRNLRGYYNQSEAGSHTLQRMYGCDLGPDGRLLRGYNQFAYDGKDYIALNEDLRSWTAADTAAQITQRKWEAARTAEQLRAYLEGTCVEWLRRY LENRKKTLQRAEHPKTHVTHHPVSDHEATLRCWALGFYPAEITLTWQRDGEDQTQDTELVETRPAGDGTFQKWAAVVVPSGEEQRYTCHVQHEGLPEPLTLRWGPSSQPTIPIVGIVAGLAVLAVLAVLGAVMAVVMCRRKSSGGKGGSCSQAASSNSAQGSDESLIACKA

[0061] 2. AI protein conformation analysis and pre-classification of mismatch sites: "Exposed site - taboo mismatch" or "Hidden site - permissible mismatch". Based on the AI ​​protein structure prediction model for HLA-C antigen proteins, the specific amino acid sites of the two donor-derived antigen proteins, HLA-C*03:03 and HLA-C*08:01, which are mismatched with the corresponding recipient antigens, were identified. Based on the positional characteristics of the two mismatched amino acid sites in their respective antigen protein spatial structures, they were pre-classified as follows (see...). Figure 2 ):

[0062] (1) The first pair: between donor C*03:03 and recipient C*03:04, the 91R amino acid of the mismatched donor antigen is located at the exposed site of the α2 domain of the antigen spatial structure, and is therefore preclassified as a "forbidden mismatch" site.

[0063] (2) The second pair: Between donor C*08:01 and recipient C*08:03, the 175G amino acid of the mismatched donor antigen is located in the hidden part of the α3 domain of the antigen spatial structure, and is therefore preclassified as a "permissible mismatch" site.

[0064] 3. Validation of mismatch site classification based on single antigen cells (SAC) (see...) Figure 3 ):

[0065] A. Binding reaction between recipient serum and donor HLA-C*03:03 single antigen cell line

[0066] a) Negative control: The results of the binding reaction between the recipient's (HLA-C*03:04) serum and cells expressing the recipient's autologous HLA-C*03:04 single antigen were negative.

[0067] b) Positive control: The results of the binding reaction between HLA-C specific monoclonal antibody (CB-1) and donor HLA-C*03:03 single antigen cells showed that the donor single antigen cells had a high level of target antigen expression.

[0068] c) Test: The results of the cell binding reaction between the recipient's (HLA-C*03:04) serum and the donor's HLA-C*03:03 single antigen indicated that the recipient produced antibodies against the donor's HLA-C*03:03 antigen because of the mismatched donor amino acid site 91R between the donor and recipient C antigens, and this site is an exposure site.

[0069] Conclusion: After AI protein conformation analysis, the 91R of the donor HLA-C*03:03 antigen, which was pre-classified as an "exposure site-taboo mismatch", was confirmed by flow cytometry binding reaction to have corresponding antibodies detected in the recipient serum. Therefore, this mismatch was experimentally confirmed as a "taboo mismatch".

[0070] B. Binding reaction between recipient serum and donor HLA-C*08:01 single antigen cell line

[0071] a) Negative control: The results of the binding reaction between the recipient's (HLA-C*08:03) serum and autologous HLA-C*08:03 single antigen cells were negative.

[0072] b) Positive control: The results of the HLA-C specific monoclonal antibody (CB-1) and donor HLA-C*08:01 single antigen cell binding reaction showed that the donor single antigen cells had a high level of target antigen expression.

[0073] c) Test: The results of the cell binding reaction between the recipient's (HLA-C*08:03) serum and the donor's HLA-C*08:01 single antigen indicated that although there was a mismatched donor amino acid site 175G between the donor and recipient C antigens, the recipient did not produce antibodies against the donor's HLA-C*08:01 antigen because this site was a hidden site.

[0074] Conclusion: After AI protein conformation analysis, the donor HLA-C*08:01 antigen 175G, which was pre-classified as "hidden site-permissible mismatch", was not detected in the recipient serum by flow cytometry binding reaction. Therefore, this mismatch was experimentally confirmed as "permissible mismatch".

[0075] It should be noted that, due to the extreme similarity in the amino acid sequences of various HLA antigens and their corresponding subtypes, the three-step method for classifying and identifying mismatch sites provides a more precise dual verification method from discovery to identification for confirming more "prohibited mismatches" and "allowable mismatches." The promotion and application of this invention's method not only facilitates accurate assessment of the clinical risks of mismatched antigens between graft donors and recipients but also facilitates precise diagnosis and treatment of corresponding donor-specific antibodies. The method described in this embodiment is applicable to all patients receiving allogeneic cell or organ therapy, including hematopoietic stem cell transplantation, organ transplantation, platelet transfusion, and universal CAR-T cell therapy in clinical applications.

[0076] Obviously, the above embodiments of the present invention are merely examples to illustrate the present invention more clearly, and are not intended to limit the implementation of the present invention. For those skilled in the art, other variations or modifications can be made based on the above description. It is impossible to exhaustively list all implementation methods here. Any obvious variations or modifications derived from the technical solutions of the present invention are still within the protection scope of the present invention.

Claims

1. A method combining AI and single-antigen cell therapy for precise identification of antigen mismatch sites that induce antibodies, including: First, unique amino acid mismatch sites of donor antigens are identified by sequence alignment of donor and recipient antigens. Then, mismatch sites are pre-classified by AI protein conformation analysis. Finally, the pre-classified mismatch sites are verified by the binding reaction results of recipient serum and donor single antigen cells with the interfering antigen knocked out.

2. The method according to claim 1, characterized in that, Methods for identifying donor-specific amino acid mismatch sites by comparing donor and recipient antigen sequences include: inputting high-resolution genotyping data of donor and recipient antigens and their corresponding full-length amino acid sequences, and identifying donor-specific amino acid mismatch sites through multiple sequence alignment algorithms.

3. The method according to claim 1, characterized in that, Pre-classification of mismatch sites through AI protein conformation analysis includes: inputting the donor antigen sequence into the AI ​​protein structure prediction model, and pre-classifying the mismatch site as a "forbidden mismatch" or "allowable mismatch" site based on whether the mismatch site is "exposed" or "hidden" in the spatial structure of the corresponding antigen protein.

4. The method according to claim 3, characterized in that, The AI ​​protein structure prediction models include AlphaFold2, DI-TASSAR, and GEOFlow V2.

5. The method according to claim 1, characterized in that, Methods for validating pre-classified mismatch sites by analyzing the binding reaction results of recipient serum and donor single antigen cells with the interfering antigen knocked out include: a) Co-incubate donor single antigen cell lines or clones with the interfering antigen knocked out with recipient serum; b) Detect whether the corresponding antibody is bound to the cell surface expressing the donor monoantigen by flow cytometry; c) Based on the results of the antibody binding reaction, confirm whether the donor-specific amino acid mismatch site is a "forbidden mismatch" or a "permissible mismatch".

6. The method according to claim 5, characterized in that, If the binding reaction between the recipient's serum and the donor's monoantigen cells is positive, the mismatched donor-derived specific site is identified as a "prohibited mismatch"; if the binding reaction between the recipient's serum and the donor's monoantigen cells is negative, the mismatched donor-derived specific site is identified as a "permissible mismatch".

7. The application of the method according to any one of claims 1 to 6 in the preparation of rejection antibody detection and analysis products.

Images