Antibody design using LSTM based deep generative model from phage display library for affinity maturation.
Ontology highlight
ABSTRACT:
SUBMITTER: Saka K
PROVIDER: S-EPMC7955064 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:Antibody leads must fulfill multiple desirable properties to be clinical candidates. Primarily due to the low throughput in the experimental procedure, the need for such multi-property optimization causes the bottleneck in preclinical antibody discovery and development, because addressing one issue usually causes another. We developed a reinforcement learning (RL) method, named AB-Gen, for antibody library design using a generative pre-trained transformer (GPT) as the policy network of the RL agent. We showed that this model can learn the antibody space of heavy chain complementarity determining region 3 (CDRH3) and generate sequences with similar property distributions. Besides, when using human epidermal growth factor receptor-2 (HER2) as the target, the agent model of AB-Gen was able to generate novel CDRH3 sequences that fulfill multi-property constraints. Totally, 509 generated sequences were able to pass all property filters, and three highly conserved residues were identified. The importance of these residues was further demonstrated by molecular dynamics simulations, consolidating that the agent model was capable of grasping important information in this complex optimization task. Overall, the AB-Gen method is able to design novel antibody sequences with an improved success rate than the traditional propose-then-filter approach. It has the potential to be used in practical antibody design, thus empowering the antibody discovery and development process. The source code of AB-Gen is freely available at Zenodo (https://doi.org/10.5281/zenodo.7657016) and BioCode (https://ngdc.cncb.ac.cn/biocode/tools/BT007341).
| S-EPMC10928431 | biostudies-literature
Project description:Phage display libraries of human single-chain variable fragments (scFvs) are a reliable source of fully human antibodies for scientific and clinical applications. Frequently, scFvs form the basis of larger, bivalent formats to increase valency and avidity. A small and versatile bivalent antibody fragment is the diabody, a cross-paired scFv dimer (?55 kDa). However, generation of diabodies from selected scFvs requires decreasing the length of the interdomain scFv linker, typically by overlap PCR. To simplify this process, we designed two scFv linkers with integrated restriction sites for easy linker length reduction (17-residue to 7-residue or 18-residue to 5-residue, respectively) and generated two fully human scFv phage display libraries. The larger library (9 × 10(9) functional members) was employed for selection against a model antigen, human N-cadherin, yielding novel scFv clones with low nanomolar monovalent affinities. ScFv clones from both libraries were reformatted into diabodies by restriction enzyme digestion and re-ligation. Size-exclusion chromatography analysis confirmed the proper dimerization of most of the diabodies. In conclusion, these specially designed scFv phage display libraries allow us to rapidly reformat the selected scFvs into diabodies, which can greatly accelerate early stage antibody development when bivalent fragments are needed for candidate screening.
| S-EPMC4607741 | biostudies-literature
Project description:Heparan sulfate (HS) is a linear polysaccharide with high structural diversity. Different HS epitopes have been detected and localized using single chain variable fragment (scFv) antibodies from a 'single pot' phage display library containing a randomized complementarity determining region of the heavy chain (CDR3). In this study, we created a new library containing anti-HS scFvs that all harbor a dp-38 heavy chain segment where the CDR3 region was engineered to contain the XBBXBX heparin binding consensus site (X?=?any amino acid, B?=?R, K or H). The library contained ~1.73?×?106 unique antibodies and was biopanned against HS from several sources. The selected antibodies were sequenced and chemically/immunohistologically characterized. A number of 67 anti-HS scFv antibodies were selected, of which 31 contained a XBBXBX CDR3 sequence. There was a clear preference for glycine at the first and proline at the fourth position of the CDR3. The sequence GZZP(R/K)X (Z?=?R, K or H, but may also contain N, S, or Q) was unusually overrepresented. Selected antibodies reacted with HS/heparin, but not with other glycosaminoglycans. Antibodies reacted differentially with respect to N-, 2-O, or 6-O-desulfated heparin preparations, and showed distinct topologies of HS epitopes in rat kidney sections. The library may be instrumental in the selection of a large pool of HS epitope-specific antibodies, and - since all antibodies differ only in their 6 amino acid CDR region - may be a tool for a rational design of antibodies recognizing specific HS sulfation patterns.
| S-EPMC7329785 | biostudies-literature
Project description:ObjectiveTo design and construct a new synthetic nanobody library using a structure-based approach that seeks to maintain high protein stability and increase the number of functional variants within the combinatorial space of mutations.ResultsSynthetic nanobody (Nb) libraries are emerging as an attractive alternative to animal immunization for the selection of stable, high affinity Nbs. Two key features define a synthetic Nb library: framework selection and CDR design. We selected the universal VHH framework from the cAbBCII10 Nb. CDR1 and CDR2 were designed with the same fixed length as in cAbBCII10, while for CDR3 we chose a 14-long loop, which creates a convex binding site topology. Based on the analysis of the cAbBCII10 crystal structure, we carefully selected the positions to be randomized and tailored the codon usage at each position, keeping at particular places amino acids that guarantee stability, favoring properties like polarity at solvent-exposed positions and avoiding destabilizing amino acids. Gene synthesis and library construction were carried out by GenScript, using our own phagemid vector. The constructed library has an estimated size of 1.75 × 108. NGS showed that the amino acid diversity and frequency at each randomized position are the expected from the codon usage.
| S-EPMC8966178 | biostudies-literature
Project description:Monoclonal antibodies are important tools in research and since the 1990s have been an important therapeutic class targeting a wide variety of diseases. Earlier methods of mAb production relied exclusively on the lengthy process of making hybridomas. The advent of phage display technology introduced an alternative approach for mAb production. A potential concern with this approach is its complete dependence on an in vitro selection process, which may result in selection of V(H)-V(L) pairs normally eliminated during the in vivo selection process. The diversity of V(H)-V(L) pairs selected from phage display libraries relative to an endogenous response is unknown. To address these questions, we constructed a panel of hybridomas and a phage display library using the spleen of a single tetanus toxoid-immunized mouse and compared the diversity of the immune response generated using each technique. Surprisingly, the tetanus toxoid-specific antibodies produced by the hybridoma library exhibited a higher degree of V(H)-V(L) genetic diversity than their phage display-derived counterparts. Furthermore, the overlap among the V-genes from each library was very limited. Consistent with the notion that accumulation of many small DNA changes lead to increased antigen specificity and affinity, the phage clones displayed substantial micro-heterogeneity. Contrary to previous reports, we found that antigen specificity against tetanus toxoid is encoded by both V(κ) and V(H) genes. Finally, the phage-derived tetanus-specific clones had a lower binding affinity than the hybridomas, a phenomenon thought to be the result of random pairing of the V-genes.
| S-EPMC4182348 | biostudies-literature
Project description:Phage display technology is an established technology particularly useful for the generation of monoclonal antibodies (mAbs). The isolation of phagemid-encoded mAb fragments depends on several features of a phage preparation. The aims of this study were to optimize phage display vectors, and to ascertain if different virion features can be optimized independently of each other. Comparisons were made between phagemid virions assembled by g3p-deficient helper phage, Hyperphage, Ex-phage or Phaberge, or corresponding g3p-sufficient helper phage, M13K07. All g3p-deficient helper phage provided a similar level of antibody display, significantly higher than that of M13K07. Hyperphage packaged virions at least 100-fold more efficiently than did Ex-phage or Phaberge. Phaberge's packaging efficiency improved by using a SupE strain. Different phagemids were also compared. Removal of a 56 base pair fragment from the promoter region resulted in increased display level and increased virion production. This critical fragment encodes a lacZ'-like peptide and is also present in other commonly used phagemids. Increasing display level did not show statistical correlation with phage production, phage infectivity or bacterial growth rate. However, phage production was positively correlated to phage infectivity. In summary, this study demonstrates simultaneously optimization of multiple and independent features of importance for phage selection.
| S-EPMC1866265 | biostudies-literature
Project description:AIM:To identify the scFv antibody fragments specific for hepatocellular carcinoma by biopanning from a large human naive scFv phage display library. METHODS:A large human naive scFv phage library was used to search for the specific targets by biopanning with the hepatocellular carcinoma cell line HepG2 for the positive-selecting and the normal liver cell line L02 for the counter-selecting. After three rounds of biopanning, individual scFv phages binding selectively to HepG2 cells were picked out. PCR was carried out for identification of the clones containing scFv gene sequence. The specific scFv phages were selected by ELISA and flow cytometry. DNA sequences of positive clones were analyzed by using Applied Biosystem Automated DNA sequencers 3 730. The expression proteins of the specific scFv antibody fragments in E.coli HB2151 were purified by the affinity chromatography and detected by SDS-PAGE, Western blot and ELISA. The biological effect of the soluble antibody fragments on the HepG2 cells was investigated by observing the cell proliferation. RESULTS:Two different positive clones were obtained and the functional variable sequences were identified. Their DNA sequences of the scFv antibody fragments were submitted to GenBank (accession nos: AY686498 and AY686499). The soluble scFv antibody fragments were successfully expressed in E.coli HB2151. The relative molecular mass of the expression products was about 36 ku, according to its predicted M(r) value. The two soluble scFv antibody fragments also had specific binding activity and obvious growth inhibition properties to HepG2 cells. CONCLUSION:The phage library biopanning permits identification of specific antibody fragments for hepatocellular carcinoma and affords experiment evidence for its immunotherapy study.
| S-EPMC4502091 | biostudies-literature
Project description:Several synthetic antibody phage display libraries have been created and used for the isolation of human monoclonal antibodies. The performance of antibody libraries, which is usually measured in terms of their ability to yield high-affinity binding specificities against target proteins of interest, depends both on technical aspects (such as library size and quality of cloning) and on design features (which influence the percentage of functional clones in the library and their ability to be used for practical applications). Here, we describe the design, construction and characterization of a combinatorial phage display library, comprising over 40 billion human antibody clones in single-chain fragment variable (scFv) format. The library was designed with the aim to obtain highly stable antibody clones, which can be affinity-purified on protein A supports, even when used in scFv format. The library was found to be highly functional, as >90% of randomly selected clones expressed the corresponding antibody. When selected against more than 15 antigens from various sources, the library always yielded specific and potent binders, at a higher frequency compared to previous antibody libraries. To demonstrate library performance in practical biomedical research projects, we isolated the human antibody G5, which reacts both against human and murine forms of the alternatively spliced BCD segment of tenascin-C, an extracellular matrix component frequently over-expressed in cancer and in chronic inflammation. The new library represents a useful source of binding specificities, both for academic research and for the development of antibody-based therapeutics.
| S-EPMC4065035 | biostudies-literature
Project description:Tetanus neurotoxin (TeNT) produced by Clostridium tetani is one of the most poisonous protein substances. Neutralizing antibodies against TeNT can effectively prevent and cure toxicosis. Using purified Hc fragments of TeNT (TeNT-Hc) as an antigen, three specific neutralizing antibody clones recognizing different epitopes were selected from a human immune scFv antibody phage display library. The three antibodies (2-7G, 2-2D, and S-4-7H) can effectively inhibit the binding between TeNT-Hc and differentiated PC-12 cells in vitro. Moreover, 2-7G inhibited TeNT-Hc binding to the receptor via carbohydrate-binding sites of the W pocket while 2-2D and S-4-7H inhibited binding of the R pocket. Although no single mAb completely protected mice from the toxin, they could both prolong survival when challenged with 20 LD50s (50% of the lethal dose) of TeNT. When used together, the mAbs completely neutralized 1000 LD50s/mg Ab, indicating their high neutralizing potency in vivo. Antibodies recognizing different carbohydrate-binding pockets could have higher synergistic toxin neutralization activities than those that recognize the same pockets. These results could lead to further production of neutralizing antibody drugs against TeNT and indicate that using TeNT-Hc as an antigen for screening human antibodies for TeNT intoxication therapy from human immune antibody library was convenient and effective.
| S-EPMC5037492 | biostudies-literature
Project description:Neutralizing monoclonal antibodies have achieved great efficacy and safety for the treatment of numerous infectious diseases. However, their neutralization potency is often rapidly lost when the target antigen mutates. Instead of isolating new antibodies each time a pathogen variant arises, it can be attractive to adapt existing antibodies, making them active against the new variant. Potential benefits of this approach include reduced development time, cost, and regulatory burden. Here a methodology is described to rapidly evolve neutralizing antibodies of proven activity, improving their function against new pathogen variants without losing efficacy against previous ones. The reported procedure is based on structure-guided affinity maturation using combinatorial mutagenesis and phage display technology. Its use against the novel severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is demonstrated, but it is suitable for any other pathogen. As proof of concept, the method is applied to CoV-X2, a human bispecific antibody that binds with high affinity to the early SARS-CoV-2 variants but lost neutralization potency against Delta. Antibodies emerging from the affinity maturation selection exhibit significantly improved neutralization potency against Delta and no loss of efficacy against the other viral sequences tested. These results illustrate the potential application of structure-guided affinity maturation in facilitating the rapid adaptation of neutralizing antibodies to pathogen variants.
| S-EPMC10566804 | biostudies-literature