Advances in database resources and computational methods for predicting antibody polyreactivity

Abstract

Antibody polyreactivity refers to the ability of a monoclonal antibody to non-specifically bind to a diverse range of antigens. While this property may be an intrinsic mechanism of the immune response, it poses significant challenges in therapeutic antibody development, often leading to off-target effects, poor pharmacokinetics, and potential toxicity. This review compiles the data resources related to polyreactive antibodies and places a particular emphasis on computational models for predicting antibody polyreactivity. The latter includes empirical models based on physicochemical properties, traditional machine learning models, deep learning networks, and protein language models. Through delineating the complexity of antibody polyreactivity, this review emphasizes the critical role and growing potential of computational prediction tools in selecting and engineering antibody drug candidates at early stages, thereby reducing development risks and accelerating the development of safer and better therapeutic antibodies.

Keywords

Antibody polyreactivity / machine learning / protein language models / antibody polyspecificity

Authors contribution

Tang H: Conceptualization, writing-original draft. Zhang Z, Li W: Data curation. Pan X, Zhou Y, Huang J: Writing-review & editing.

Conflicts of interest

The authors declare no conflicts of interest.

Ethical approval

Not applicable.

Consent to participate

Not applicable.

Consent for publication

Not applicable.

Availability of data and materials

Not applicable.

Funding

This work was supported by the National Natural Science Foundation of China (Grant Nos. 62371112 and 62071099) and the Sichuan Science and Technology Program (Grant No. 2024NSFSC0635).

References

[1] Cramer F. Emil fischer’s lock-and-key hypothesis after 100 years-towards a supracellular chemistry. In: Behr JP, editior. Perspectives in supramolecular chemistry: The lock-and-key principle. Chichester: John Wiley & Sons; 1994. p. 1-23.

[2] Jain T, Sun T, Durand S, Hall A, Houston NR, Nett JH, et al. Biophysical properties of the clinical-stage antibody landscape. Proc Natl Acad Sci U S A. 2017; 114(5): 944-949.

[3] Ichiyoshi Y, Casali P . Analysis of the structural correlates for antibody polyreactivity by multiple reassortments of chimeric human immunoglobulin heavy and light chain V segments. J Exp Med. 1994; 180(3): 885-895.

[4] Tsuiji M, Yurasov S, Velinzon K, Thomas S, Nussenzweig MC, Wardemann H, et al. A checkpoint for autoreactivity in human IgM+ memory B cell development. J Exp Med. 2006; 203(2): 393-400.

[5] Dighiero G, Lymberi P, Mazié JC, Rouyre S, Butler-Browne GS, Whalen RG, et al. Murine hybridomas secreting natural monoclonal antibodies reacting with self antigens. J Immunol. 1983; 131(5): 2267-2272.

[6] Pashov AD, Dimitrov JD . Antibody polyreactivity: A challenger of immune paradigms. Immunology. 2025; 176(4): 421-437.

[7] Starr CG, Tessier PM . Selecting and engineering monoclonal antibodies with drug-like specificity. Curr Opin Biotechnol. 2019; 60: 119-127.

[8] Michaelides MC, Eisen HN . The strange cross-reaction of menadione (vitamin K3) and 2,4-dinitrophenyl ligands with a myeloma protein and some conventional antibodies. J Exp Med. 1974; 140(3): 687-702.

[9] Sigounas G, Harindranath N, Donadel G, Notkins AL . Half-life of polyreactive antibodies. J Clin Immunol. 1994; 14(2): 134-140.

[10] Cunningham O, Scott M, Zhou ZS, Finlay WJJ . Polyreactivity and polyspecificity in therapeutic antibody development: Risk factors for failure in preclinical and clinical development campaigns. mAbs. 2021; 13(1): 1999195.

[11] Amash A, Volkers G, Farber P, Griffin D, Davison KS, Goodman A, et al. Developability considerations for bispecific and multispecific antibodies. mAbs. 2024; 16(1): 2394229.

[12] Dyson MR, Masters E, Pazeraitis D, Perera RL, Syrjanen JL, Surade S, et al. Beyond affinity: Selection of antibody variants with optimal biophysical properties and reduced immunogenicity from mammalian display libraries. mAbs. 2020; 12(1): 1829335.

[13] Jang E, Youn J . Contribution of long-lived plasma cells to antibody-mediated allograft rejection. Clin Transplant Res. 2024; 38(4): 341-353.

[14] Geller RL, Bach FH, Turman MA, Casali P, Platt JL . Evidence that polyreactive antibodies are deposited in rejected discordant xenografts. Transplantation. 1993; 55(1): 168-172.

[15] Zhang J, Jacobi AM, Wang T, Berlin R, Volpe BT, Diamond B, et al. Polyreactive autoantibodies in systemic lupus erythematosus have pathogenic potential. J Autoimmun. 2009; 33(3-4): 270-274.

[16] Amara K, Steen J, Murray F, Morbach H, Fernandez-Rodriguez BM, Joshua V, et al. Monoclonal IgG antibodies generated from joint-derived B cells of RA patients have a strong bias toward citrullinated autoantigen recognition. J Exp Med. 2013; 210(3): 445-455.

[17] Mietzner B, Tsuiji M, Scheid J, Velinzon K, Tiller T, Abraham K, et al. Autoreactive IgG memory antibodies in patients with systemic lupus erythematosus arise from nonreactive and polyreactive precursors. Proc Natl Acad Sci U S A. 2008; 105(28): 9727-9732.

[18] Gonzalez S, González-Rodríguez AP, Suárez-Álvarez B, López-Soto A, Huergo-Zapico L, Lopez-Larrea C . Conceptual aspects of self and nonself discrimination. Self Nonself. 2011; 2(1): 19-25.

[19] Carnero LAR, Bedinger D, Cocklin S, Li J, Erasmus MF, D’Angelo S, et al. Identification of polyreactive antibodies by high throughput enzyme-linked immunosorbent assay and surface Plasmon resonance. J Immunol Methods. 2025; 539: 113855.

[20] Mouquet H, Nussenzweig MC . Polyreactive antibodies in adaptive immune responses to viruses. Cell Mol Life Sci. 2012; 69(9): 1435-1445.

[21] Mouquet H, Scheid JF, Zoller MJ, Krogsgaard M, Ott RG, Shukair S, et al. Polyreactivity increases the apparent affinity of anti-HIV antibodies by heteroligation. Nature. 2010; 467(7315): 591-595.

[22] Bostrom J, Yu SF, Kan D, Appleton BA, Lee CV, Billeci K, et al. Variants of the antibody herceptin that interact with HER2 and VEGF at the antigen binding site. Science. 2009; 323(5921): 1610-1614.

[23] Broug E, Bland-Ward PA, Powell J, Johnson KS . Fab-arm exchange. Nat Biotechnol. 2010; 28(2): 123-125.

[24] Bajaj G, Shchelokov D, Demin O Jr, Adams HC, Feng Y, Gibiansky L, et al. Dose selection for DuoBody®-CD40x4-1BB (GEN1042/BNT312) using a mPBPK/RO model leveraging preclinical and clinical data. Clin Pharma Ther. 2025; 118(2): 418-427.

[25] Surowka M, Schaefer W, Klein C . Ten years in the making: Application of CrossMab technology for the development of therapeutic bispecific antibodies and antibody fusion proteins. mAbs. 2021; 13(1): 1967714.

[26] Mitropoulou AN, Bowen H, Dodev TS, Davies AM, Bax HJ, Beavil RL, et al. Structure of a patient-derived antibody in complex with allergen reveals simultaneous conventional and superantigen-like recognition. Proc Natl Acad Sci U S A. 2018; 115(37): E8707-E8716.

[27] James LC, Roversi P, Tawfik DS . Antibody multispecificity mediated by conformational diversity. Science. 2003; 299(5611): 1362-1367.

[28] Seidler CA, Liedl KR . CDR L3 loop rearrangement switches multispecific SPE-7 IgE antibody from hapten to protein binding. J Mol Recognit. 2024; 37(6): e3107.

[29] Beckmann R, Jensen K, Fenn S, Speck J, Krause K, Meier A, et al. DutaFabs are engineered therapeutic Fab fragments that can bind two targets simultaneously. Nat Commun. 2021; 12(1): 708.

[30] Harwardt J, Geyer FK, Schoenfeld K, Baumstark D, Molkenthin V, Kolmar H, et al. Balancing the affinity and tumor cell binding of a two-in-one antibody simultaneously targeting EGFR and PD-L1. Antibodies. 2024; 13(2): 36.

[31] Wymann S, Morales RAV, Toh WH, Remlinger J, Guse K, Ghai R, et al. CSL305: A dual functional therapeutic antibody targeting complement C2 and FcRn. Int J Mol Sci. 2026; 27(5): 2383.

[32] Borowska MT, Boughter CT, Bunker JJ, Guthmiller JJ, Wilson PC, Roux B, et al. Biochemical and biophysical characterization of natural polyreactivity in antibodies. Cell Rep. 2023; 42(10): 113190.

[33] Chen HT, Zhang Y, Huang J, Sawant M, Smith MD, Rajagopal N, et al. Human antibody polyreactivity is governed primarily by the heavy-chain complementarity-determining regions. Cell Rep. 2024; 43(10): 114801.

[34] Torres M, Casadevall A . The immunoglobulin constant region contributes to affinity and specificity. Trends Immunol. 2008; 29(2): 91-97.

[35] Éliás S, Wrzodek C, Deane CM, Tissot AC, Klostermann S, Ros F . Prediction of polyspecificity from antibody sequence data by machine learning. Front Bioinform. 2023; 3: 1286883.

[36] Xu Y, Roach W, Sun T, Jain T, Prinz B, Yu TY, et al. Addressing polyspecificity of antibodies selected from an in vitro yeast presentation system: A FACS-based, high-throughput selection and analytical tool. Protein Eng Des Sel. 2013; 26(10): 663-670.

[37] Almagro JC, Pedraza-Escalona M, Arrieta HI, Pérez-Tapia SM . Phage display libraries for antibody therapeutic discovery and development. Antibodies. 2019; 8(3): 44.

[38] Boughter CT, Borowska MT, Guthmiller JJ, Bendelac A, Wilson PC, Roux B, et al. Biochemical patterns of antibody polyreactivity revealed through a bioinformatics-based analysis of CDR loops. Elife. 2020; 9: e61393.

[39] Frese K, Eisenmann M, Ostendorp R, Brocks B, Pabst S . An automated immunoassay for early specificity profiling of antibodies. MAbs. 2013; 5(2): 279-287.

[40] Shehata L, Maurer DP, Wec AZ, Lilov A, Champney E, Sun T, et al. Affinity maturation enhances antibody specificity but compromises conformational stability. Cell Rep. 2019; 28(13): 3300-3308.

[41] Bailly M, Mieczkowski C, Juan V, Metwally E, Tomazela D, Baker J, et al. Predicting antibody developability profiles through early stage discovery screening. mAbs. 2020; 12(1): 1743053.

[42] Zhou Y, Huang Z, Li W, Wei J, Jiang Q, Yang W, et al. Deep learning in preclinical antibody drug discovery and development. Methods. 2023; 218: 57-71.

[43] Arakawa T, Akuta T . Mechanistic insight into poly-reactivity of immune antibodies upon acid denaturation or arginine mutation in antigen-binding regions. Antibodies. 2023; 12(4): 64.

[44] Arsiwala A, Bhatt R, van Niekerk L, Quintero-Cadena P, Ao X, Rosenbaum A, et al. A high-throughput platform for biophysical antibody developability assessment to enable AI/ML model training. mAbs. 2025; 17(1): 2593055.

[45] Jain T, Prinz B, Marker A, Michel A, Reichel K, Czepczor V, et al. Assessment and incorporation of in vitro correlates to pharmacokinetic outcomes in antibody developability workflows. mAbs. 2024; 16(1): 2384104.

[46] Harvey EP, Shin JE, Skiba MA, Nemeth GR, Hurley JD, Wellner A, et al. An in silico method to assess antibody fragment polyreactivity. Nat Commun. 2022; 13(1): 7554.

[47] Li W, Lin H, Huang Z, Xie S, Zhou Y, Gong R, et al. DOTAD: A database of therapeutic antibody developability. Interdiscip Sci. 2024; 16(3): 623-634.

[48] Gai SA, Wittrup KD . Yeast surface display for protein engineering and characterization. Curr Opin Struct Biol. 2007; 17(4): 467-473.

[49] Xu A, Kim HS, Estee S, ViaJar S, Galush WJ, Gill A, et al. Susceptibility of antibody CDR residues to chemical modifications can be revealed prior to antibody humanization and aid in the lead selection process. Mol Pharm. 2018; 15(10): 4529-4537.

[50] Rabia LA, Zhang Y, Ludwig SD, Julian MC, Tessier PM . Net charge of antibody complementarity-determining regions is a key predictor of specificity. Protein Eng Des Sel. 2018; 31(11): 409-418.

[51] Prigent J, Lorin V, Kök A, Hieu T, Bourgeau S, Mouquet H . Scarcity of autoreactive human blood IgA+ memory B cells. Eur J Immunol. 2016; 46(10): 2340-2351.

[52] Bumbaca Yadav D, Sharma VK, Boswell CA, Hotzel I, Tesar D, Shang Y, et al. Evaluating the use of antibody variable region (Fv) charge as a risk assessment tool for predicting typical cynomolgus monkey pharmacokinetics. J Biol Chem. 2015; 290(50): 29732-29741.

[53] Marks C, Deane CM . How repertoire data are changing antibody science. J Biol Chem. 2020; 295(29): 9823-9837.

[54] Mason DM, Friedensohn S, Weber CR, Jordi C, Wagner B, Meng SM, et al. Optimization of therapeutic antibodies by predicting antigen specificity from antibody sequence via deep learning. Nat Biomed Eng. 2021; 5(6): 600-612.

[55] Jain T, Boland T, Lilov A, Burnina I, Brown M, Xu Y, et al. Prediction of delayed retention of antibodies in hydrophobic interaction chromatography from sequence using machine learning. Bioinformatics. 2017; 33(23): 3758-3766.

[56] Lin Z, Akin H, Rao R, Hie B, Zhu Z, Lu W, et al. Evolutionary-scale prediction of atomic-level protein structure with a language model. Science. 2023; 379(6637): 1123-1130.

[57] Lim H, No KT . Prediction of polyreactive and nonspecific single-chain fragment variables through structural biochemical features and protein language-based descriptors. BMC Bioinformatics. 2022; 23(1): 520.

[58] Yu X, Vangjeli K, Prakash A, Chhaya M, Stanley SJ, Cohen N, et al. Protein language models enable prediction of polyreactivity of monospecific, bispecific, and heavy-chain-only antibodies. Antib Ther. 2024; 7(3): 199-208.

[59] Zhou Y, Tang H, Wu C, Zhang Z, Wei J, Gong R, et al. Enhancing polyreactivity prediction of preclinical antibodies through fine-tuned protein language models. J Pharm Anal. 2025; 15(12): 101448.

[60] Lecerf M, Lacombe RV, Dimitrov JD . Polyreactivity of antibodies from different B-cell subpopulations is determined by distinct sequence patterns of variable region. Front Immunol. 2023; 14: 1266668.

[61] Yang X, Wu C, Liu W, Fu K, Tian Y, Wei X, et al. A clinical-information-free method for early diagnosis of lung cancer from the patients with pulmonary nodules based on backpropagation neural network model. Comput Struct Biotechnol J. 2024; 24: 404-411.

[62] Zhou Y, Liu W, Luo C, Huang Z, Samarappuli Mudiyanselage Savini G, Zhao L, et al. Ab-Amy 2.0: Predicting light chain amyloidogenic risk of therapeutic antibodies based on antibody language model. Methods. 2025; 233: 11-18.

[63] LeCun Y, Bengio Y, Hinton G . Deep learning. Nature. 2015; 521(7553): 436-444.

[64] Wilman W, Wróbel S, Bielska W, Deszynski P, Dudzic P, Jaszczyszyn I, et al. Machine-designed biotherapeutics: Opportunities, feasibility and advantages of deep learning in computational antibody discovery. Brief Bioinform. 2022; 23(4): bbac267.

[65] Alley EC, Khimulya G, Biswas S, AlQuraishi M, Church GM . Unified rational protein engineering with sequence-based deep representation learning. Nat Methods. 2019; 16(12): 1315-1322.

[66] Khurana S, Rawi R, Kunji K, Chuang GY, Bensmail H, Mall R, et al. DeepSol: A deep learning framework for sequence-based protein solubility prediction. Bioinformatics. 2018; 34(15): 2605-2613.

[67] Ruffolo JA, Chu LS, Mahajan SP, Gray JJ . Fast, accurate antibody structure prediction from deep learning on massive set of natural antibodies. Nat Commun. 2023; 14(1): 2389.

[68] Rives A, Meier J, Sercu T, Goyal S, Lin Z, Liu J, et al. Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences. Proc Natl Acad Sci U S A. 2021; 118(15): e2016239118.

[69] Zhang Z, Xu M, Jamasb A, Chenthamarakshan V, Lozano A, Das P, et al. Protein representation learning by geometric structure pretraining. arXiv:2203.06125v5 [Preprint]. 2022.

[70] Ruffolo JA, Sulam J, Gray JJ . Antibody structure prediction using interpretable deep learning. Patterns. 2022; 3(2): 100406.

[71] Ofer D, Brandes N, Linial M . The language of proteins: NLP, machine learning & protein sequences. Comput Struct Biotechnol J. 2021; 19: 1750-1758.

[72] Elnaggar A, Heinzinger M, Dallago C, Rehawi G, Wang Y, Jones L, et al. ProtTrans: Toward understanding the language of life through self-supervised learning. IEEE Trans Pattern Anal Mach Intell. 2022; 44(10): 7112-7127.

[73] Okuno T, Ono N, Altaf-Ul-Amin M, Kanaya S . Self-supervised domain adaptation of protein language model based solely on positive enzyme-reaction pairs. Comput Struct Biotechnol J. 2025; 27: 5441-5449.

[74] Rao R, Bhattacharya N, Thomas N, Duan Y, Chen P, Canny J, et al. Evaluating protein transfer learning with TAPE. Adv Neural Inf Process Syst. 2019; 32: 9689-9701.

[75] Meier J, Rao R, Verkuil R, Liu J, Sercu T, Rives A, et al. Language models enable zero-shot prediction of the effects of mutations on protein function. bioRxiv [Preprint]. 2021.

[76] Schmirler R, Heinzinger M, Rost B . Fine-tuning protein language models boosts predictions across diverse tasks. Nat Commun. 2024; 15(1): 7407.

[77] Robin G, Sato Y, Desplancq D, Rochel N, Weiss E, Martineau P, et al. Erratum to “restricted diversity of antigen binding residues of antibodies revealed by computational alanine scanning of 227 antibody-antigen complexes”. J Mol Biol. 2014; 426(22): 3729-3743.

[78] Kelly RL, Le D, Zhao J, Wittrup KD . Reduction of nonspecificity motifs in synthetic antibody libraries. J Mol Biol. 2018; 430(1): 119-130.

[79] Sakhnini LI, Beltrame L, Fulle S, Sormanni P, Henriksen A, Lorenzen N, et al. Prediction of antibody non-specificity using protein language models and biophysical parameters. mAbs. 2026; 18(1): 2678000.

[80] Gujral O, Bafna M, Alm E, Berger B . Sparse autoencoders uncover biologically interpretable features in protein language model representations. Proc Natl Acad Sci U S A. 2025; 122(34): e2506316122.

[81] Fernández-Quintero ML, Loeffler JR, Bacher LM, Waibl F, Seidler CA, Liedl KR, et al. Local and global rigidification upon antibody affinity maturation. Front Mol Biosci. 2020; 7: 182.