Targeted cd47 antibody mutants and uses thereof

Abstract

The application belongs to the field of antibodies, and particularly relates to a CD47-targeting antibody mutant and application thereof, wherein the amino acid sequence of the heavy chain variable region of the CD47-targeting antibody mutant is shown as SEQ ID NO: 3, and the amino acid sequence of the light chain variable region of the CD47-targeting antibody mutant is shown as SEQ ID NO: 4; the application implements multidimensional developability optimization on the light and heavy chain variable regions of the anti-human CD47 antibody through computer-aided molecular design combined with experimental verification, and comprehensively considers parameters such as folding free energy, surface charge distribution, hydrophobic exposure area and local stability, and an optimized mutant is screened; the optimized mutant obtained by the application realizes isoelectric point reduction, hydrophobicity weakening and charge symmetry and hydrophilic exposure improvement under the premise of not reducing affinity, can be used for preparing a drug for treating tumors related to high expression of CD47, and provides a general technical approach for AI-based antibody developability and stability modification.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of antibodies, and particularly relates to a mutant of CD47-targeting antibody and application thereof. BACKGROUND

[0002] CD47 is a glycosylated transmembrane protein widely expressed on the surface of various cells. CD47 interacts with Signal Regulatory Protein alpha (SIRPα) on the surface of myeloid cells to transmit a "don't eat me" signal, inhibiting the phagocytosis and antigen presentation function of macrophages, thereby weakening the innate immunity and subsequent T cell response. Tumor cells often upregulate CD47 to evade immune clearance, so blocking the CD47-SIRPα axis has clear significance for tumor immunotherapy.

[0003] Using antibodies to block the CD47-SIRPα axis can restore the recognition and phagocytosis of macrophages to tumor cells. According to the published data, there are more than 100 CD47 projects in development worldwide, including preclinical projects, among which antibody drugs occupy a dominant position. However, CD47 is widely expressed on the surface of various normal cells, and the development of its targeted drugs needs to take into account the effectiveness of the mechanism and the clinical safety. Specifically, it is necessary to fully consider the problems caused by the antigen sink effect, such as the decrease in drug exposure, the increase in dosage requirement, and the narrowing of the therapeutic window. At the engineering level, it is also necessary to deal with the challenges of high concentration aggregation tendency, increased solution viscosity, increased heterogeneity, and difficult process characterization caused by high isoelectric point, local high charge density, or large hydrophobic patch. Therefore, when optimizing the stability and developability of CD47 antibodies, how to systematically improve the structural stability and developability of antibodies while maintaining the antigen binding activity has become an important research direction in the field of antibody engineering.

[0004] Current strategies for antibody stabilization mainly include CDR grafting and framework humanization, structure-based rational design, consensus mutation or stabilizer introduction, and directed evolution. Although these methods have achieved certain success in specific cases, they still have the following limitations: they are mostly focused on single index optimization, making it difficult to achieve multi-objective trade-off; the efficiency of candidate molecule screening is limited; and the systematic protection of key epitope regions is lacking, which can affect antigen binding. The computability of antibody structure provides a basis for the application of computer-aided drug design (CADD) in its molecular modification. CADD is essentially the integrated use of molecular simulation techniques in drug development, and several successful cases have shown that it can effectively improve antibody stability. CD47, as an immune escape target widely expressed on the surface of various tumor cells, its related antibody drugs are particularly critical for stability and formulation performance. Current modification strategies for CD47 antibodies are mainly focused on affinity or Fc function optimization, and there is a lack of a systematic stability and developability optimization method based on epitope protection, combining multi-source data analysis and artificial intelligence algorithms. SUMMARY

[0005] Invention purpose: The present invention aims to address the developability issues of CD47 target antibodies (using parent D0604 as an example) in the previously granted patent (202211709826.6, Invention name: An antibody targeting CD47 and its application), such as high isoelectric point, local charge enrichment, and high hydrophobicity. Using artificial intelligence technology combined with computer simulation drug design technology, the physicochemical properties of D0604 are evaluated and optimized, and a rational engineering modification strategy under the premise of epitope retention is proposed. The optimized and modified anti-CD47 antibody mutant with better isoelectric point, charge distribution, and hydrophobicity is used to prepare a drug for treating CD47 high expression related tumors.

[0006] Technical solution: A CD47-targeting antibody mutant, the amino acid sequence of its heavy chain variable region is shown in SEQ ID NO: 3; the amino acid sequence of its light chain variable region is shown in SEQ ID NO: 4.

[0007] The CD47-targeting antibody mutant has an IgG1 heavy chain type and a kappa light chain type.

[0008] The CD47-targeting antibody mutant is used in the preparation of a tumor treatment drug.

[0009] The tumor is a CD47 high expression related tumor.

[0010] The mutant is based on the modification of anti-CD47 antibody molecule D0604, and the D0604 is disclosed in the full text of CN115894692B, application date December 29, 2022, and the invention name is: an antibody targeting CD47 and its application, wherein the D0604 heavy chain variable region is as shown in the amino acid sequence of SEQ ID No. 1, and the D0604 light chain variable region is as shown in the amino acid sequence of SEQ ID No. 2.

[0011] The isoelectric point and hydrophobicity of the mutant produced by SEQ ID NO. 3 and SEQ ID NO. 4 are the lowest, and the in vitro stability is also high.

[0012] The application provides a CD47-targeting antibody mutant with high stability, wherein the amino acid sequence of the light chain variable region is selected from any one or more single-point mutated sequences shown in SEQ ID No. 2, and the number of the multiple is less than 4; and the amino acid sequence of the heavy chain variable region is selected from any one or more single-point mutated sequences shown in SEQ ID No. 1, and the number of the multiple is less than 4.

[0013] In the application, the amino acid sequences of the heavy chain variable region and the light chain variable region of the parent antibody D0604 are shown in SEQ ID No. 1 and SEQ ID No. 2 respectively. The directional mutation of multiple single-point amino acids on the light chain can improve the physicochemical properties of the antibody, and the directional mutation includes that Q30 is mutated to Y and K42 is mutated to E. The directional mutation of multiple single-point amino acids on the heavy chain can also improve the physicochemical properties of the antibody, and the directional mutation includes that M47 is mutated to E and G26 is mutated to L.

[0014] An expression vector comprising the above-mentioned CD47-targeting antibody mutant.

[0015] A recombinant cell comprising the above-mentioned CD47-targeting antibody mutant.

[0016] Beneficial effects: In view of the above technical bottlenecks, the present application proposes a computer-aided system methodology for optimizing the structure of CD47 antibody and an optimized antibody mutant obtained therefrom. The method is based on epitope integrity constraints, combined with molecular dynamics simulation (MD), machine learning model (ML) and deep generative network (ESM / ThermoMPNN) for multi-objective mutation screening and energy evaluation, to realize the synergistic optimization of isoelectric point, hydrophobicity, charge symmetry and folding free energy of the antibody, and verify the physical and chemical and functional improvement effect through in vitro expression, to solve the problem that stability improvement and affinity maintenance are difficult to balance in the prior art, that is, the CD47 antibody mutant M1 has lower hydrophobicity, lower isoelectric point and higher apparent solubility compared with the parent D0604; at the same time, the CD47 antibody preferred variant M1 has high binding activity with CD47, maintains high purity and activity under accelerated conditions at 37 DEG C and 4 DEG C, and provides a general reference path for the engineering and industrialization of antibody drugs. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 The binding interface and key amino acid residues of D0604 and CD47 antigen;

[0018] Figure 2 The heavy chain aggregation score of D0604;

[0019] Figure 3 The light chain aggregation score of D0604;

[0020] Figure 4 The M1 binding epitope;

[0021] Figure 5 The M2 binding epitope;

[0022] Figure 6 The M3 binding epitope;

[0023] Figure 7 The RMSD curve of molecular dynamics;

[0024] Figure 8 The Rg curve of molecular dynamics;

[0025] Figure 9 The SASA curve of molecular dynamics;

[0026] Figure 10 The number of hydrogen bonds (H-Bond) curve of molecular dynamics;

[0027] Figure 11Non-reducing SDS-PAGE identification of results, wherein: lanes 1-3 are M1 (VL: Y30Q, K42E), M2 (VL: Y30Q, K42E; VH: M47E), M3 (VL: Y30Q, K42E; VH: M47E, G26L), respectively;

[0028] Figure 12 ELISA to detect the binding ability of the mutant to the target antigen human CD47 at the protein level;

[0029] Figure 13 Non-reducing SDS-PAGE identification of antibody D0604 storage stability;

[0030] Figure 14 Non-reducing SDS-PAGE identification of mutant M1 storage stability;

[0031] Figure 15 ELISA to identify activity after storage. DETAILED DESCRIPTION

[0032] Example 1, structural prediction and computational analysis of the physicochemical properties of anti-CD47 antibody D0604:

[0033] 1. Antibody sequence acquisition and structure modeling:

[0034] Based on the existing anti-CD47 monoclonal antibody D0604 in the laboratory, the variable region sequences of its heavy and light chains (see SEQ ID NO. 1 and SEQ ID NO. 2, respectively) were extracted. The IMGT numbering annotation was performed using the Antibody module in the Molecular Operating Environment (MOE, version 2023) software, and the three-dimensional structure model was constructed through the Antibody Modeler module. Energy minimization processing was performed under the Amber:EHT force field to obtain the optimized D0604 antibody model.

[0035] The Protein Property tool of MOE was used to predict the physicochemical properties of the model, including isoelectric point (PI), hydrophobic patch, and positive / negative charge distribution area (Positive / Negative Patch).

[0036] Results: As shown in Table 1, by analyzing it was found that the predicted isoelectric point of most antibody sequences was less than 7, accounting for 81%, the highest predicted isoelectric point of the recently marketed trastuzumab was 7.21, and the predicted isoelectric point of D0604 was 7.95, which indicated that D0604 had a high risk of isoelectric point; the maximum hydrophobic domain of the marketed antibody sequence was between 80-130, and the D0604 was 150, which had a high risk of easy aggregation and precipitation; other positive and negative charge regions were within the reasonable range.

[0037] Table 1, the isoelectric point and hydrophobic surface area prediction results of antibody D0604

[0038] .

[0039] 2. Structure dynamics and spatial feature prediction:

[0040] To further evaluate the spatial conformation stability and flow characteristics of D0604, MOE was used to calculate the radius of gyration, hydrophilic surface area, protein mobility and solvent accessible surface area (SASA) of D0604.

[0041] Results: As shown in Table 2, the radius of gyration and hydrophilic surface area of D0604 were similar to the average values of the marketed antibodies, indicating that the overall spatial volume and folding form were reasonable; but its protein mobility value was 5.7, which was significantly higher than the average value of the marketed antibody sequences (-0.05), suggesting that D0604 had faster conformation fluctuation and higher flexibility in solution. Higher molecular mobility may lead to local structural instability and denaturation risk during long-term storage, so the reduction of this property should be considered in the subsequent optimization design.

[0042] Table 2, the calculation results of the radius of gyration, hydrophilic surface area, protein mobility and solvent accessible surface area of D0604 using MOE

[0043] .

[0044] 3. Machine learning assisted charge distribution and antibody Fv region structure symmetry prediction:

[0045] Machine learning model was used to predict and analyze the charge characteristics of D0604 antibody light and heavy chains and the structure symmetry of antibody Fv region, and compared with the benchmark data of 11 marketed antibodies.

[0046] Results: As shown in Table 3, the average charge of the heavy chain (VH) and light chain (VL) of the marketed antibody is 2.6004 and 2.2416, respectively, the average viscosity is 24.0141, and the Fv symmetry index is 6.3003. In contrast, the VH and VL charges of D0604 are higher than the average, and the Fv symmetry index is significantly higher, indicating the presence of high local positive or negative charge clusters within the variable region. This asymmetric charge distribution may be related to its high isoelectric point feature and is a potential cause of the increased solution viscosity and decreased conformational stability of D0604.

[0047] In summary, this embodiment reveals the problems of high isoelectric point, excessive hydrophobic patch, high conformational mobility, and asymmetric Fv charge of anti-CD47 antibody D0604 in terms of physicochemical properties through MOE simulation and machine learning prediction, providing clear computational basis and engineering direction for subsequent sequence optimization to reduce isoelectric point, improve surface charge distribution, and enhance structural stability.

[0048] Table 3, Charge, Viscosity, and Mobility of Antibody D0604

[0049] .

[0050] Example 2, Prediction of the Binding Interface and Key Amino Acid Residues of Anti-CD47 Antibody D0604 and CD47 Antigen:

[0051] To determine the binding interface of anti-CD47 antibody D0604 and its target CD47 antigen, and to avoid disrupting key interaction sites in subsequent mutation design, this embodiment uses AlphaFold3 to construct and analyze the complex.

[0052] 1. Complex structure prediction:

[0053] The heavy and light chain variable region amino acid sequences of D0604 antibody (see SEQ ID NO. 1 and SEQ ID NO. 2, respectively) and the extracellular domain sequence of CD47 antigen are input into the AlphaFold3 system for complex modeling. Default parameters are used during the prediction process, and the structure template search and iterative optimization functions are enabled to ensure the reliability of the spatial orientation of the binding interface. The resulting model is displayed in the structure visualization software PyMOL with color coding, where:

[0054] The CD47 antigen part is represented in green;

[0055] The heavy chain (VH) of D0604 antibody is represented in blue;

[0056] The light chain (VL) of D0604 antibody is represented in red.

[0057] 2. Binding interface and hydrogen bond interaction analysis:

[0058] Results: As shown in Figure 1 , by analyzing the interface residues of the predicted complex structure, it was found that eight residues in the CD47 antigen, including 61K, 63R, 82S, 93K, 101D, 103S, 104D, and 107S, formed a significant hydrogen bond network with the D0604 antibody. These residues are distributed in different secondary structure fragments of CD47, showing a non-continuous spatial arrangement.

[0059] Correspondingly, the heavy chain variable region (VH) of the antibody D0604 forms the main binding in the interface, with 51V, 52T, 54T, 55S, and 101N as the key interaction sites, forming stable hydrogen bonds and van der Waals forces with the above-mentioned residues of the CD47 antigen. In contrast, the light chain (VL) is involved in less contact area, mainly playing a supporting role in stability.

[0060] Based on the above results, in the subsequent rational design and mutation screening of antibodies, the above-mentioned 51V, 52T, 54T, 55S, and 101N sites are defined as protected regions, and amino acid substitutions are not implemented at these positions to ensure the binding ability and epitope recognition integrity with the CD47 antigen.

[0061] In summary, AlphaFold3 prediction clearly identifies the key amino acid residues and interface characteristics of the binding of D0604 antibody and CD47 antigen, confirming that D0604 and CD47 form multiple point hydrogen bond interactions in a spatial epitope manner. This information provides precise guidance for subsequent sequence optimization and mutation design at the structural level.

[0062] Example 3, Prediction of Aggregation Tendency and Evolutionary Conservation of Anti-CD47 Antibody D0604 Sequence:

[0063] To identify potential insoluble aggregation sites and non-conserved regions in the sequence of anti-CD47 antibody D0604, this example uses Aggrescan3D 2.0 and ESMscan deep language models for computational analysis at the structural and sequence levels.

[0064] 1. Aggregation hotspot site analysis:

[0065] By using Aggrescan3D to score the aggregation tendency of the three-dimensional structure model of D0604, several residues with high hydrophobic aggregation index were identified.

[0066] Results: As shown in Figure 2 and Figure 3 :

[0067] In the heavy chain (VH), L11, Y31, I97 were determined as the main aggregation amino acid sites;

[0068] In the light chain (VL), V15, Y30, Y92, I106 showed strong aggregation trend.

[0069] These residues are mostly located in the CDR region or its adjacent framework region, exposed to the surface of the molecule, and may promote intermolecular interaction, trigger aggregation or precipitation under high concentration conditions.

[0070] 2. Evolutionary conservation prediction:

[0071] Using the ESMscan deep language model to score the evolutionary conservation of the amino acid sequence of D0604, combined with database alignment results, non-conservative site regions that can be rationally replaced are identified.

[0072] The site score of the results shows that:

[0073] Q1, R30-R35, R45, A46, M47, R50-S56, R96-Y106, S116, S117 in the VH chain are low-conservative regions, and different amino acid type substitutions can be considered to adjust the physicochemical properties;

[0074] E118, T119, E144-A151, N167-E173, Q206-T215 in the VL chain are low-conservative regions, and mutations can be implemented in subsequent design.

[0075] Example 4, virtual saturation mutation scanning and affinity / stability comprehensive prediction:

[0076] After determining the non-conservative sites of the variable region, the virtual mutation module of MOE software was used for saturation mutation scanning. Based on the Amber:EHT force field for energy calculation, 20 kinds of amino acids were introduced into each replaceable site, and the changes in binding free energy (ΔG_bind) and folding stability (ΔG_stab) were calculated.

[0077] The results show that: part of the mutation can reduce the folding energy and improve the binding energy, showing a potential double optimization effect. Through comprehensive sorting and screening, a group of candidate mutation sites with simultaneous improvement of stability and affinity are obtained, and the final mutation scheme is determined.

[0078] Table 4, final mutation scheme of antibody

[0079] .

[0080] Example 5, analysis of changes in the binding epitope of antibody mutants and CD47 antigen:

[0081] To evaluate the changes in the binding mode of the mutated antibodies, AlphaFold3 was used to predict and analyze the interface of the complexes of each mutant (M1, M2, M3) with the CD47 antigen.

[0082] Results: As shown in Figure 4 , Figure 5 , Figure 6 ,

[0083] The key binding residues of the D0604 heavy chain are H51V, H52T, H54T, H55S, H101N, which form stable hydrogen bond interactions with the 61K, 63R, 82S, 93K, 101D, 103S, 104D, 107S of the CD47 antigen.

[0084] The main binding residues of the M1 antibody are H35R, H37Y, H50R, H96R, H98Q, H100I, H103E, H105D, which correspond to the antigen binding sites 53E, 57K, 59K, 64D, 76T, 115E, 117T, 119L, 122E.

[0085] The main binding residues of the M2 antibody are H31Y, H33D, H50R, H56S, H58Y, which form limited contacts with the 79T, 82S, 101D, 103S, 104D of the antigen, with a smaller interface.

[0086] The main binding residues of the M3 antibody are H35R, H50R, H54T, H58Y, L50N, 93G, which interact with the 74K, 75S, 79T, 80D, 82S, 85K, 101D of the antigen.

[0087] In summary, the binding interface of M2 is significantly smaller than that of D0604, while the binding interfaces of M1 and M3 are expanded, especially M1, which significantly increases the number of antibody sites and antigen contact residues, indicating that M1 may have stronger binding stability and affinity.

[0088] Example 6, Gromacs molecular dynamics simulation analysis of antibody stability:

[0089] To verify the influence of mutations on conformational stability, Gromacs 2023 software was used to perform 100 ns molecular dynamics simulation on D0604 and its mutants M1, M2, M3, and dynamic characteristics were compared according to RMSD, Rg, SASA, and the number of hydrogen bonds.

[0090] 1. RMSD analysis:

[0091] The RMSD curve is shown in Figure 7As shown, the stability ranking is: M1 > D0604 > M3 > M2. The M1 curve is stable and has the least fluctuation, indicating that the overall conformational stability is the highest and is closest to the dynamic characteristics of marketed antibodies.

[0092] 2. Rg analysis:

[0093] The trend of radius (Rg) change is as follows Figure 8 As shown, consistent with RMSD, the order is M1>D0604>M3>M2. The Rg value of M1 remains in the lower range, indicating that its molecular compactness is better.

[0094] 3. SASA analysis:

[0095] SASA results are as follows Figure 9 As shown, the soluble exposure area of ​​the parent D0604 was consistently lower than that of the mutants, indicating that its surface hydrophobicity was higher; M3 had the highest SASA value, which was consistent with its presence of more hydrophilic amino acid mutations, suggesting increased hydrophilic exposure.

[0096] 4. Hydrogen bond number analysis:

[0097] After comparison with marketed antibodies enrofloxacin and goxatocillin, the results were as follows: Figure 10 As shown, among the optimized mutants, M2 forms the most hydrogen bonds with the solvent, while D0604, M1, and M3 form similar numbers of hydrogen bonds with the solvent. The overall trend indicates that the mutation did not significantly disrupt solvent interactions, and M1 maintained a reasonable hydrogen bond network while preserving structural stability.

[0098] In summary, the Gromacs molecular dynamics results confirm that M1 exhibits optimal stability across all parameters, which corroborates the calculated and predicted folding energies and epitope analysis results.

[0099] Example 7: Construction and eukaryotic expression of light and heavy chain expression vectors for CD47 monoclonal antibody mutant molecules:

[0100] 1) Design upstream and downstream primers. Using the variable regions of the light and heavy chains of the D0604 antibody as templates, respectively, high-fidelity PCR was used to amplify the variable region sequence of the anti-CD47 antibody and the constant region fragment of the IgG1 antibody. Point mutations were introduced into the original D0604 sequence using overlap PCR to construct different mutation combinations, including Y30Q and K42E in the light chain and M47E and G26L in the heavy chain. Subsequently, the obtained mutant light and heavy chain full-length fragments were ligated into a mammalian cell expression vector using a one-step cloning method after double digestion with EcoRI and NotI. Sequencing verification confirmed that the results were consistent with the design and contained no base errors. Positive clones with correct sequencing were cultured in a culture medium, and high-purity plasmids were extracted for later use.

[0101] 2) The plasmids encoding mutant light and heavy chains were co-transfected into HEK293 cells at a ratio of 2:1. After 5 days of culture at 37 °C and 5% CO2, the culture supernatant was collected, centrifuged at 4 °C to remove cell debris, and purified using a Protein A affinity chromatography column.

[0102] Results: Figure 11 As shown in the SDS-PAGE analysis of the purified products, lanes 1-3 were M1 (VL:Y30Q, K42E), M2 (VL:Y30Q, K42E; VH:M47E), and M3 (VL:Y30Q, K42E; VH:M47E, G26L), respectively. All three showed a single clear band at about 150 kDa, consistent with the theoretical molecular weight of IgG, with no obvious degradation or aggregation bands, indicating that each mutant was correctly expressed in the eukaryotic system and successfully purified.

[0103] Example 8: Gel filtration high-performance liquid chromatography identification of antibodies:

[0104] To evaluate the purity and aggregation of the M series of mutants, gel filtration high-performance liquid chromatography (SEC-HPLC) was used for detection. The Waters Alliance HPLC system was used, with a TSKgel G3000SWXL (7.8 mm x 300 mm) column, a PBS buffer (pH 7.0) mobile phase, a flow rate of 0.5 mL / min, a detection wavelength of 280 nm, and a sample injection volume of 20 μL.

[0105] Results: As shown in Table 5, the main peak of the parent antibody D0604 had a retention time of about 8.05 min, with a monomer peak purity of 97.93% and an aggregation peak of about 2.06%. The main peak of M1 had a retention time of about 7.97 min, with a monomer peak purity of 93.5%, an aggregation peak of 4.69%, and a low molecular peak of 1.81%. The main peak of M2 had a retention time of about 7.81 min, with a monomer peak purity of 92.48%, an aggregation peak of 3.77%, and a low molecular peak of 2.32%. The main peak of M3 had a retention time of about 7.89 min, with a monomer peak purity of 91.85%, an aggregation peak of 4.0%, and a low molecular peak of 2.19%.

[0106] Overall, all four antibodies showed a single main peak, with a purity of more than 90%, indicating that the expression products were stable and had low aggregation content. Compared with D0604, the purity of mutant M1 decreased slightly but still remained at a high level, and the aggregation peak was low, indicating good structural stability. The aggregation ratio of M2 and M3 increased slightly, which may be related to the increased local hydrophobic exposure caused by the mutation sites.

[0107] In summary, the results of gel filtration high performance liquid chromatography analysis showed that the D0604-M series mutants all had good molecular uniformity and preparation stability. Among them, M1 had the best comprehensive performance with the optimal purity and monomer ratio and the least aggregates.

[0108] Table 5, gel filtration high performance liquid chromatography detection results of mutants

[0109] .

[0110] Example 9, hydrophobic chromatography high performance liquid chromatography identification of antibody hydrophobicity:

[0111] Hydrophobic chromatography high performance liquid chromatography (HIC-HPLC) was used to detect the hydrophobic properties of the antibodies. Through the differences in the hydrophobicity of the antibodies, the specific experimental conditions were as follows: chromatograph model was Fuli LC5190, antibody concentration was 2 mg / mL, injection amount was 20 μL, flow rate was 0.8 mL / min, gradient elution was used, detector wavelength was 280 nm, column temperature was 30°C, and chromatograph column model was TSKgel Butyl-NPR.

[0112] Results: As shown in Table 6, the retention times of mutants M1 and M2 were shorter than that of the parent antibody D0604 after base mutation, indicating that the overall hydrophobicity was reduced, the hydrophilicity of the molecular surface was enhanced, and the hydrophobic patch area converged. In contrast, the retention time of M3 was prolonged, suggesting that its hydrophobicity slightly increased, which might be related to the combined mutation of heavy chain G26L and M47E leading to the exposure of local hydrophobic residues.

[0113] Table 6, hydrophobic chromatography high performance liquid chromatography detection results of mutants

[0114] .

[0115] Example 10, ELISA detection of the binding ability of mutants to target antigen human CD47 at the protein level:

[0116] ELISA was used to detect the binding of mutants to recombinant human CD47 antigen. The mFc-tagged hCD47 recombinant protein was diluted to 1 μg / ml with antigen coating solution, and 50 μl / well was added to the enzyme-labeled strip. The duplicate wells were coated at 4°C overnight. After washing with PBST for 3 times, 5% skim milk was added for blocking at 37°C for 2 h. After washing with PBST for 3 times, gradient-diluted M1, M2, M3 and isotype control antibodies (the highest concentration was 200 nM, 4-fold gradient dilution, a total of 12 concentration gradients) were added and incubated at 37°C for 2 h. After washing with PBST for 3 times, HRP-labeled goat anti-human IgG secondary antibody was added and incubated at 37°C for 1 h. After washing with PBST for 6 times, TMB color developing solution was added, and the color developing reaction was incubated in the dark for 15 min. The color developing reaction was terminated. The absorbance values at 450 nm and 630 nm were read on the enzyme-labeled instrument, and the EC 50 ;

[0117] Results: As shown in Table 7 and Figure 2, the mutants can effectively bind to hCD47 protein at the protein level, and the binding ability of mutant M1 is nearly doubled compared with the precursor antibody D0604. Figure 12

[0118] Table 7, EC values of mutants binding to hCD47 protein at the protein level 50

[0119] .

[0120] Example 11, Cation exchange high performance liquid chromatography for identifying changes in isoelectric point of mutants:

[0121] To further verify the influence of antibody mutation on the overall charge property and isoelectric point (pi), CIEX-HPLC was used to analyze D0604 and its mutant M1, and trastuzumab was used as a reference. The experiment used a MabPac SCX-10 column (4.0 mm x 250 mm, Thermo Fisher), mobile phase A was 25 mM MES buffer (pH 6.0), mobile phase B was 25 mM MES buffer + 250 mM NaCl (pH 6.0), gradient elution time was 30 min, flow rate was 0.5 mL / min, and detection wavelength was 280 nm.

[0122] ​​Results: As shown in Table 8. The main peak of trastuzumab appeared at about 18.52 min, corresponding to the predicted isoelectric point value of 7.21. The main peak of the parent antibody D0604 appeared at about 21.0 min, significantly lagging behind trastuzumab, indicating that its overall isoelectric point was higher. The main peak of mutant M1 (VL: Y30Q, K42E) appeared at about 18.4 min, close to that of trastuzumab, indicating that its isoelectric point value had decreased to the same level as the reference antibody (about 7.21), and the charge distribution was significantly improved.

[0123] Comprehensive analysis shows that mutant M1, by replacing the basic residues of the light chain with neutral or acidic amino acids, reduces the overall isoelectric point to the desired range, and the charge distribution tends to be balanced, similar to the charge behavior of trastuzumab standard antibody, consistent with the calculated prediction target.

[0124] Table 8, results of cation exchange high performance liquid chromatography detection of mutants

[0125] .

[0126] Example 12, accelerated stability verification of mutant molecules:

[0127] To verify the structural stability and storage performance of mutant molecules, the stability of parent antibody D0604 and mutant M1 under different temperature and time conditions was accelerated. The purified antibody samples were placed at 4 ℃ and 37 ℃ for 1, 2, 3, and 5 days, respectively, and a 1-month long-term low-temperature stability test was also conducted. At each time point, the samples were taken and analyzed for structural integrity and monomer proportion change by SDS-PAGE and gel filtration high performance liquid chromatography.

[0128] Results: As shown in Table 8. The main peak of trastuzumab appeared at about 18.52 min, corresponding to the predicted isoelectric point value of 7.21. The main peak of the parent antibody D0604 appeared at about 21.0 min, significantly lagging behind trastuzumab, indicating that its overall isoelectric point was higher. The main peak of mutant M1 (VL: Y30Q, K42E) appeared at about 18.4 min, close to that of trastuzumab, indicating that its isoelectric point value had decreased to the same level as the reference antibody (about 7.21), and the charge distribution was significantly improved. Figure 13 and Figure 14 As shown in Table 8. The main peak of trastuzumab appeared at about 18.52 min, corresponding to the predicted isoelectric point value of 7.21. The main peak of the parent antibody D0604 appeared at about 21.0 min, significantly lagging behind trastuzumab, indicating that its overall isoelectric point was higher. The main peak of mutant M1 (VL: Y30Q, K42E) appeared at about 18.4 min, close to that of trastuzumab, indicating that its isoelectric point value had decreased to the same level as the reference antibody (about 7.21), and the charge distribution was significantly improved.

[0129] Comprehensive analysis showed that mutant M1 exhibited better resistance to thermal denaturation and solution stability in accelerated stability test, and could maintain higher monomer proportion and purity under normal storage conditions, and had better conformational robustness and long-term stability compared with the original antibody D0604, which met the quality requirements of therapeutic antibodies.

[0130] Example 13, apparent solubility identification of mutant molecules:

[0131] To evaluate the solubility characteristics of mutant molecules under high concentration conditions, the apparent solubility of parent antibody D0604 and mutant M1 was analyzed by polyethylene glycol (PEG) induced precipitation method, and trastuzumab was used as a reference. The experiment was set up in PBS buffer system with different concentrations of PEG6000 (0-30%, w / v), and the sample concentration was 1 mg / mL. The optical density (OD 350 ) at 350 nm was measured after incubation at 25 °C for 2 h to reflect the turbidity change of antibody solution.

[0132] The results are shown in Table 9. With the increase of PEG concentration, the OD 350 values of the three antibodies gradually increased, but the growth rates were significantly different. Trastuzumab (control) showed a significant aggregation inflection point (EC 50 ) at about 14.15% PEG, the EC 50 of D0604 was 18.60%, and the EC 50 of mutant M1 further increased to 19.75%, indicating that it required a higher PEG concentration to precipitate, indicating that M1 had better solubility.

[0133] Table 9, apparent solubility identification of mutant molecules

[0134] .

[0135] The above results showed that the apparent solubility of mutant M1 was significantly improved, which was consistent with the trend of reduced hydrophobicity in the foregoing hydrophobic chromatography high performance liquid chromatography detection. The mutation effectively improved the solution behavior and solubility characteristics of the antibody molecule by introducing hydrophilic amino acid substitution of the light chain surface hydrophobic residues, verifying the rationality and effectiveness of rational mutation design. In summary, M1 was superior to the parent antibody D0604 in terms of structural stability, solution behavior and storage performance, and had better development potential.

[0136] Example 14, affinity stability identification of mutant:

[0137] To evaluate the antigen binding ability and stability of mutant molecules under accelerated conditions, ELISA binding activity analysis was performed for parent antibody D0604 and its mutant M1. Human CD47 protein (hCD47) was used as the coating antigen in the experiment, and the 96-well enzyme-labeled plate was coated with hCD47 at a concentration of 2 μg / mL at 4 ℃ overnight. After washing the plate three times with PBST, 1% BSA was used for blocking for 2 h, and then different concentrations of antibody samples (0.00001-1000 nM) were added for incubation at 37 ℃ for 2 h. After color development with HRP-labeled secondary antibody, the OD value was read at 450 nm wavelength, and the half maximal effective concentration (EC 50 ) value of each antibody and hCD47 was calculated.

[0138] Results: As shown in Figure 15 and Table 10. The EC50 of parent antibody D0604 binding to hCD47 was 0.16 nM; mutant M1 (VL: Y30Q, K42E) showed stronger binding ability, and the EC 50 decreased to 0.08 nM, indicating that its affinity was about 2 times higher than that of the prototype antibody. The isotype control did not show specific binding signal.

[0139] Table 10, accelerated stability verification of mutant molecules

[0140] .

[0141] Under the stability condition of 37 ℃ for 5 days, D0604 and M1 still maintained a high level of antigen binding ability. The results showed that mutant M1 still had high antigen binding activity and thermal stability under accelerated stress conditions, and the binding ability was better than that of the parent antibody D0604, which was consistent with the results of the above structural stability and physicochemical characterization. The results verified the effectiveness of the strategy of introducing light chain hydrophilic mutations by rational design to significantly improve the stability of antibody molecules without compromising the affinity.

Claims

1. A mutant of a CD47 antibody, characterized in that, The amino acid sequence of the heavy chain variable region thereof is shown as SEQ ID NO: 3, and the amino acid sequence of the light chain variable region thereof is shown as SEQ ID NO:

4.

2. The mutant according to claim 1, wherein The heavy chain type thereof is IgG1. The light chain type thereof is kappa.

Images