High efficiency CHO cell display-based antibody maturation.

Previously, we developed a CHO cell display-based antibody maturation procedure in which an antibody (or other protein) gene of interest was induced to mutate by activation-induced cytidine deaminase (AID) and then form a library by simply proliferating the CHO cells in culture. In this study, we further improved the efficiency of this maturation system by reengineering AID, and optimizing the nucleic acid sequence of the target antibody gene and AID gene as well as the protocol for AID gene transfection. These changes have increased both the mutation rate and the number of mutation type of antibody genes by more than 10 fold, and greatly improved the maturation efficiency of antibody/other proteins.

. The changes made on mammalian AID for improving its activity. AID proteins include nuclear localization signal (NLS), nuclear export signal (NES), catalytic domain, and APOBEC protein-like domains. (A) The wild type AID structure (named AID); (B) The AID without NES (AID-del); (C) The AID-del containing K10E, T82I and E156G point mutations (AID-plus). The mouse and human AID's (mAID and hAID) have the same basic structure. The numbers below the molecules are amino acid sequence numbers, and the numbers above the AID-plus molecule are the amino acid sequence numbers with point mutation. The GFP reporter gene was used to detect mutation efficiency of different AID mutants. The mAID data was used as control, and the hAID, mAID-del, mAID-plus, hAID-del, and hAID-plus data were normalized to mAID to quantitatively compare the mutation efficiencies of the above AID's. both the point mutations (K10E, T82I and E156G), and the deletion of mAID's NES contributed to the improvement of mAID activity (Fig. 2B).
We also constructed hAID-del (human AID without NES) and hAID-plus (hAID-del with the point mutations K10E, T82Iand E156G) and tested their mutation efficiencies (Fig. 2B). Generally speaking, hAID and their mutants had lower activities than their mouse counterparts in CHO cells. The mutations K10E, T82I and E156G on hAID increased its activity, while in contrast to mAID, the NES deletion of hAID did not enhance its activity. These data suggest that mAID-plus has the highest mutating activity, and should be used for antibody affinity maturation in the following experiments.
The contributions of the base optimization of target antibody gene and the engineered AID to mutation efficiency. In the previous section, the engineered AID (AID-plus) demonstrated a superior activity for converting a stop codon into an amino acid and forming a functional GFP gene. A previous study from Honjo's research group found that the AID-induced mutation sites were predisposed to divide into "hot spots" and "cold spots" in B cells 17 . To convert an antibody gene sequence of interest into the one containing as many "hot spots" as possible without changing its amino acid residue sequence, we developed a computer algorithm and converted the variable regions of an anti-TNFα single chain antibody (scFv) (described in Materials and Methods) using this algorithm (S1). However, the converted antibody gene (hsAb) could hardly be displayed (Fig. 3A). Western blot analysis demonstrated that the hsAb did not express while the wtAb expressed normally in host cells (Fig. S1). We inferred that the converted nucleic acids impaired the gene's transcription and/or translation. The lack of expression is not due to rare codons since we intentionally removed all rare codons in the sequences generated from the computer algorithm we developed. Therefore, we had the mutability optimized antibody gene processed using a computer program "OptimumGene" [20][21][22] of the Genscript Biotech company to achieve a maximal expression of the gene (S1). Although this gene (eoAb) was highly expressed and displayed (S1 and Fig. 3A), we found that the computer program generated an identical antibody gene sequence whether it processed the original wild type gene or mutability optimized gene. That is, it erased all the base changes derived from our computer algorithm (S1), thus these two programs are incompatible.
We went ahead to check if the expression-optimization of the antibody gene could lead to a higher mutation efficiency. We paired mAID or mAID-plus with the wtAb or eoAb to investigate the contributions of the modifications of mAID and the sequence-optimization to inducing mutation on this scFv. A round of evolution is composed of (1) transfecting an mAID plasmid into CHO cells that displayed the antibody, (2) then proliferating the CHO cells in medium containing antibiotic to maintain the plasmid in cells, and (3) at last enriching the cells displaying the antibody at high levels as well as strongly binding to TNFα-GFP by flow cytometry sorting. After one round of evolution, the antibody genes isolated from the sorted CHO cells were cloned and sequenced. Fifty clones were sequenced for each sorted cell sample and at least 42 valid sequences were obtained from each of the four samples. The results showed that both the expression optimization and mAID modifications significantly increased mutation efficiency and the number of point mutation type (Table S1 and Fig. 3B). Notably, the types of point mutations from the four samples were hardly overlapping; among 36 different point mutations, only two point mutations (C525G, G726A) occurred in different samples. A single round of evolution had already enriched two point mutations from the two sequenced optimized samples (14 C525G mutations induced by mAID, 7 G371A mutations induced by mAID-plus, Table S1).
Obviously, the above expression-optimization mediated the enhancement in mutation rate and number of mutation type is not due to the introduction of new hot spots. The mutation efficiencies induced by AID was closely proportional to the transcription level of targeted genes 23 . We then checked if the expression-optimization program from Genscript Biotech increased antibody gene transcription. Indeed, the program increased transcription level by 14 fold as revealed by RT-PCR assay (Fig. 3C). This transcription enhancement contributed significantly more to the increase in both mutation rate and number of mutation type than that of mAID-plus. The ratios of mutation rate and number of mutation type of eoAb/mAID to wtAb/mAID are 12.5 and 6, respectively, while those of wtAb/mAID-plus to wtAb/mAID are only 4 and 3, respectively (calculated from Fig. 3B).
Effect of the optimization of AID sequence for transcription on mutation efficiency of antibody gene. In the previous section, we showed that the expression-optimization of the scFv gene sequence for transcription enhanced its mutation rate. We were curious whether the expression-optimization of the AID gene sequence for transcription using the Genscript Biotech computer program would also enhance the mutation rate. We separately transfected with mAID and expression-optimized mAID (eomAID) into antibody-displayed cells and carried out 3 rounds of evolution. Fifty clones of each round of TNFα antibody gene were sequenced and point mutations identified (Table S2). In the cells derived from the first two rounds of evolution (R1, R2), the mutation frequency induced by the optimized mAID was about double that of the wild-type mAID; after the third round of evolution (R3), the mutation frequency induced by the optimized mAID was also significantly higher than that of the wild-type mAID (Fig. 4). The mutation frequency in the cells after the R2 round of evolution induced by the optimized mAID was nearly as high as that induced by the wild-type mAID after the R3 round of evolution.
The effect of the transfection strategy of AID plasmids on the mutation efficiency of antibody gene. We demonstrated that the increased AID expression by AID expression-optimization enhanced the mutations of the targeted antibody gene in the above section. In the past, episomal vectors were used to express both antibody and AID genes for the reason that the episomal vector-bearing genes are not integrated into a chromosome and can then highly express the genes without the influence of a chromosomal location effect [9][10][11] . The high levels of AID and antibody gene transcription confer a high efficiency of antibody gene mutation 23 . Episomal vectors are thought to stay in the cytoplasm and are not integrated into chromosomes 24 . Thus, to conduct multiple rounds of antibody evolution, episomal vectors containing AID gene were transfected into mammalian cells displaying antibodies for each round of evolution followed by culturing the cells in medium containing antibiotic www.nature.com/scientificreports www.nature.com/scientificreports/ to keep the episomal vectors in cells 10,11 . To our surprise, however, the cells derived from 3 months' culture without antibiotic after one round of evolution were still resistant to the antibiotic neomycin; the proliferation rate and morphology of these cells looked the same with or without neomycin in the medium. This indicates that the antibiotic-resistant gene Neo had already been integrated into chromosomes. Therefore, the newly transfected episomal vectors bearing Neo and AID genes after the first round of evolution cannot be kept in cells in medium containing neomycin, and the AID enzyme level should be lower than expected. In this case, the mutation rate of the antibody gene would be lower, resulting in inefficient antibody maturation.
To avoid the undesired situation described above, we constructed a new episomal vector bearing the AID gene as well as the blasticidin-resistant gene (Bsd) (instead of Neo). We compared the AID mRNA levels among the 3 groups of cells: (A) the cells cultured 15 days without neomycin after one round of evolution (15 days after sorting); (B) the cells which were cultured 10 days without neomycin after sorting, then transfected with the episomal vector bearing the AID gene and Neo, and cultured 5 more days in the presence of neomycin; (C) the cells which were cultured 10 days without neomycin after sorting, then transfected with the episomal vector bearing the AID gene and Bsd, and cultured 5 more days in the presence of blasticidin (Fig. 5A). We measured the AID mRNA levels on the 5th, 12th (right before 1st sorting), 22nd (right before 2nd transfection; no 2nd transfection for the 1st group of cells) and 27th day (5 days after 2nd transfection) of the three groups of cells (Fig. 5B). In spite of the unstable experimental system, the data showed that transfection with the vector bearing Bsd yielded a much higher AID mRNA level than that without a 2nd transfection and that with a 2nd transfection with vector bearing Neo, and the AID mRNA level in the B group of cells was slightly higher than that of the A group. These results indicate that the repeated transfection with the vector containing the same antibiotic gene only confers an AID level slightly higher than that without the repeated transfection, and that the repeated transfection with a vector bearing a different antibiotic gene yields an AID level much higher than that of the repeated transfection with the vector containing the same antibiotic gene.
We wanted to know how the antibodies with the three transfection strategies evolved. We detected the display levels and antigen-binding abilities of the antibodies on the cells after two rounds of evolution as shown in Fig. 5C. The C group had more cells with the antigen-binding abilities above the wild type antibody than the A and B groups, and the cell numbers with the highest antigen-binding abilities of the groups A and B were about equal (Fig. 5C). Fifty clones of each group of TNFα antibody gene were sequenced and point mutations identified (Table S3). In the cells derived from the two rounds of evolution (R1, R2), the mutation frequency induced by group C was higher than that of the group A or B.
Antibody affinity maturation. We demonstrated that the reengineered AID, the expression-optimized antibody gene, the expression-optimized AID gene, and an improved AID transfection strategy can enhance the mutations of the targeted antibody gene in the above section. In the past, to mature antibody affinity, the wild type AID was introduced once and the cells displaying antibody were put through multiple rounds of cell proliferation-flow sorting 9 , or the wild type AID was introduced into the cells displaying antibody once after each round of proliferation-flow sorting 11 . In this study, we compared these two maturation procedures with an evolution process in which AID-plus, expression-optimized antibody gene and the improved strategy were used. Specifically, the following three procedures were compared in their abilities to mature antibody affinity: (A) the cells displayed wtAb transfected with the episomal vector bearing the AID gene and Neo in the first round and enriched in the next two rounds; (B) the cells displayed wtAb transfescted with the episomal vector bearing the AID gene and Neo in the two rounds and enriched in the 3rd round; (C) the cells displayed eoAb transfescted with the episomal vector bearing the AID-plus gene and Neo in the first round, then transfected with the episomal vector bearing the AID gene and Bsd in the second round, and enriched in the 3rd round. Fifty clones of TNFα antibody gene derived from the cells after each round of evolution were sequenced and point mutations identified from the successfully sequenced genes (Table S4). We then choose six clones with the highest mutation frequency after the 3rd round of evolution (C344T, G371A, C720G, G721A, C74T + C720A, G81T + G721A, highlighted with colors in Table S4), and examined the affinities of these antibodies in the scFv form. All the mutant clones except G371A have higher affinities than the wild type. The G371A mutant antibody clone failed to express for unknown reason. The other five clones were analyzed by Octet biomolecular interaction technology for their dissociation coefficients (K D ), confirming their improved affinities (Fig. S1, Table 1). We found that the two clones derived from the (C) procedure (C720G and C74T + C720A) had significantly higher affinities than those of the other three mutant clones derived from the (A) and (B) procedures.

Discussion
Compared to the widely used phage display and yeast display, CHO cell display possesses advantages in codon usage, posttranslational modification and ease in library construction 25 , but also has a disadvantage in the limited diversity of a target gene due to a low AID-induced mutation rate. In this study, we reengineered AID, optimized the antibody gene and AID sequences, and formatted a new AID transfection strategy. These changes led to enhancements of both the mutation rate and diversity by more than 10 fold (Fig. 3B), and to a highly efficient maturation procedure (Fig. 5C and Table 1).
AID activity is an important factor in inducing mutations in a gene of interest. We found that mouse AID has higher activity than human AID (Fig. 2B). We thought that this might be because mouse is closer to Chinese hamster than Human in systemic evolution. However, CHO AID has a lower activity than mouse AID (our unpublished data). Interestingly, the deletion of the nuclear export signal of human AID did not increase the gene mutation efficiency as demonstrated for mouse AID (Fig. 2B) although the point mutations of K10E, T82I and E156G in both mouse and human AIDs confer comparable activity enhancements; this phenomenon is beyond our understanding at present. Based on the two unique features of mouse AID, mAID-plus is the best choice for CHO cell display-based maturation. A recent study by Al-Qaisi et al. 26  www.nature.com/scientificreports www.nature.com/scientificreports/ of wild type and its mutant (m7.3) AID's, finding that the mutant induced significantly more mutations in the targeted genes, consistent with our results in this study in which the mAID-plus was actually an m7.3 with the deletion of the AID nuclear export signal (NES). This study presented data showing that the level of the mutations on the RFP gene reached a plateau 10 days after transient transfection of wild type or m7.3 AID into host cells. www.nature.com/scientificreports www.nature.com/scientificreports/ These data are also compatible with our results in which the AID mRNA level at the 5th day after AID transfection into host cells was much higher than that at 12th day (5 times higher, Fig. 5B).
The transcription level of a targeted gene is positively proportional to the rate of AID-induced mutation on this gene 23 . This was confirmed by optimizing the TNFα antibody gene sequence for the best expression using the computer program of the Genscript Biotech Company. The increase in expression is partially due to an improvement in transcription or translation (Fig. 3C).
An episomal vector was thought not to be integrated into cellular chromosomes and used to highly express the gene on the vector due to the lack of location-effects on the gene expression 9,25 . However, our tests showed that the gene on the vector including the antibiotic-resistant gene was integrated in chromosomes after just one round of evolution. The procedures in which an episomal vector bearing an identical antibiotic-resistant gene is repeatedly transfected into the same cells for antibody maturation cannot be kept in the cells by the corresponding antibiotic. Our data demonstrated that transfecting the cells with the vectors containing different antibiotic resistant genes besides AID and culturing cells with their corresponding antibiotics in medium for different rounds of evolution maintained high levels of AID enzyme in cells, and increased the mutation efficiency of the gene of interest. A recent study by Liu LD et al. 27 screened engineered deaminases and CRISPR-deaminase coupling approaches and built diversifying base editors to generate SHM. At the same time, Devilder MC et al. 28 use the CRISPR-Cas 9 to target AID to antibody genes. However, these systems rely on the mammalian cells endogenous DNA repair system, which potentially limits its efficiency or usage.
The types of point mutations from the four samples were barely overlapping; among 36 different point mutations, only two point mutations occurred in different samples (Table S1). This rarity in overlapping point mutations is not only due to the difference in target gene sequences (original versus optimized), but also due to different AID enzymes (mAID versus mAID-plus); only one mutation took place in both samples among 29 different point mutations in the cells that possessed the optimized antibody gene sequence and were mutated respectively by mAID and mAID-plus. These results indicate that the preferred target base sequences of mAID-plus is very different from those of mAID. Notably, the combination of mAID and the optimized antibody sequence yielded a significantly higher mutation rate and diversity than those derived from the combinations of the wild type sequence and mAID or mAID-plus (Table S1), although its levels of mutation rate and diversity were not as high as those resulting from the combination of the optimized sequence and mAID-plus (Fig. 3B). Based on this analysis, it is preferable to use mAID and mAID-plus in different rounds of evolution to increase the opportunity to enrich the desired combinations of point mutations.
The aforementioned changes have made the CHO cell-based evolution procedure highly efficient. In the past, Bowers et al. used a human cell-display based evolution procedure to mature antibody affinity in which the wild type AID was introduced once and the cells displaying antibody were put through multiple rounds of cell proliferation-flow sorting 9 . Our laboratory used a similar procedure, but in which the wild type AID was introduced into the CHO cells displaying antibody once after each round of proliferation-flow sorting 11 . To show the values in antibody affinity maturation of the elements examined in this study (modified AID, expression-optimized AID as well as antibody gene sequences, and transfection strategy), we formatted an affinity maturation procedure in which an example set of changes on these elements were made, and compared the efficiency of this procedure with those of the two above-mentioned previous maturation procedures. Three rounds of evolution using the newly formatted procedure generated two affinity-improved antibodies (N240K and T25I + N240K), and 3 rounds of evolution using the two old procedures produced three affinity-improved antibodies (A115V, E241K and K27N + E241K). The affinities of the two antibodies derived from the new procedure are 3.5 to 8.6 higher than those of the three antibody clones generated from the two old methods (calculated from the data in Table 1), suggesting the superiority of the new procedure to the old ones.
Based on the results in this study, the ideal procedure should be: (1) transfection with a plasmid containing Neo and mAID-plus before the first flow sorting, and (2) transfection with another plasmid bearing Bsd and mAID before the second flow sorting. Of course, all the 4 genes can be paired differently such as Neo and mAID as well as Bsd and mAID-plus, and transfection sequences can be rearranged. This procedure is most likely to produce the highest mutation rate and the highest mutation diversity, and acquire the best mutation combinations conferring the most desired antibody features. We expect that this procedure will be more frequently used in maturing antibodies and other proteins in the future.

Materials and Methods
Plasmid construction and expression. TNFα-GFP is His-tagged on its N-terminus and was prepared by following the procedure by Chen et al. 19 . Briefly, His-TNFα-GFP was synthesized in E. coli containing the PET28a (+)-TNFα-GFP plasmid, was purified with a Ni column. The concentration of the prepared TNFα-GFP was 2 mg/mL.
Anti-human TNFα wild type scFv (αTNFα) is from Jie Tang (Institute of Biophysics, Chinese Academy of Sciences, Beijing, China).The plasmids pF2AC for co-expression of Flpo and iCre recombinases, pFAbL for the replacement of the puro with the antibody (αTNFα antibody) gene in a predetermined genomic locus of CHO-puro cells were described in detail previously 11 .
The exchange vector pFRT-αTNFα-loxP was generated previously in our laboratory 11 , and for the current study pFRT-eo-αTNFα-loxP was created by replacing SP-HA-anti-TNFα single chain antibody gene-TM (SP-HA: signal peptide-HA tag) with SP-HA-eo-anti-TNFα single chain antibody gene-TM (the gene sequence was optimized for the best expression by the Genscript Biotech Company (Nanjing, China)) between BamHI and BglII in pFRT-αTNFα-loxP.
The pCDNA3.1(+)-GFP* plasmids were constructed by Chen et al. 19 . Eight pCEP4-Ig-Ek plasmids expressing various AIDs were used in this study. The AIDs include mouse AID (mAID), mAID with its C-terminal nucleic export signal (NES) deleted (mAID-del), mAID-del bearing K10E, T82I and E156G (mAID-plus), and mAID as well as mAID-plus with their sequences optimized for the best expression (eomAID as well as eomAID-plus). The human AIDs are hAID, hAID-del and hAID-plus. The plasmid expressing mAID was described previously 11 . The DNA sequences of mAID-plus, eomAID, eomAID-plus, hAID and hAID-plus were synthesized by the Genscript Biotech Company. The DNA sequences of mAID-del and hAID-del were synthesized by performing PCRs using mAID and hAID as templates. The PCR primer pairs for the 7 AIDs contained HindIII and XhoI restriction sites, and the PCR products were cut with synthesized by HindIII and XhoI. The purified fragments were inserted between the HindIII and XhoI restriction sites in pCEP4-Ig-Ek plasmid.
Two pCEP4-Ig-Ek-mAID plasmids expressing neomycin and blasticidin resistant gene (Bsd and Neo) respectively were used in this study. The plasmid expressing Neo (pCEP4-Ig-Ek-Neo-mAID) was described previously 11 . pCEP4-Ig-Ek-Bsd-mAID was created by replacing Neo with Bsd in pCEP4-Ig-Ek-Neo-mAID between AgeI and NotI. Bsd was synthesized by the Genscript Biotech Company.
Cell culture. CHO/dhFr cells (12200036) were purchased from The Cell Bank of the Chinese Academy of Sciences, Shanghai, China, and cell lines derived from CHO/dhFr − were propagated in IMDM medium (HyClone, USA) containing 10% fetal bovine serum (HyClone, USA), 0.1 mM hypoxanthin, and 0.016 mM thymidine (Gibco, USA), at 37 °C in a 5% CO 2 incubator. Subcultures were carried out every 2-3 days. CHO-puro cells (The CHO/dhFr − cells with a single copy of retargetable high-level expression cassette) were established previously and described in detail 11 . Transfection and antibody affinity maturation. To test the mutant ability of the eight pCEP4-Ig-Ek plasmids expressing mAID, mAID-del, mAID-plus, hAID, hAID-del and hAID-plus, CHO-puro cells were seeded into each well of a six-well plate. Cells were transfected with a mixture of 1 μg pCDNA3.1(+)-GFP*, 1 μg pCEP4-Ig-Ek plasmids expressing various AIDs, and 5 μL Lipofectamine TM 2000 (Invitrogen, USA) for 6 h, separately. After 48 h, cells were analyzed using a FACSCalibur (BD) flow cytometer.
Generally, to replace the antibody gene integrated in chromosome in CHO cells, CHO-puro cells were seeded into each well of a six-well plate. Cells were transfected with a mixture of 0.5 μg exchange vector, 2 μg pF2AC, and 5μL Lipofectamine TM 2000 (Invitrogen, USA) for 6 h. After transfection, the cells were transferred into a 10 cm dish containing IMDM medium with 10% FCS. Afterwards, the cells were collected and incubated with PE-conjugated anti-HA antibody (Abcam,1:250 in cold opti-MEM medium [Invitrogen, USA], to detect antibody display levels.) for 30 min at 4 °C, washed with cold opti-MEM once, resuspended in cold opti-MEM, and sorted for cells that expressed high displaying antibodies using a FACSAriaIII (BD) flow cytometer.
To mature antibody affinity, CHO-puro cells which displayed anti-TNFα antibodies were seeded into a six-well plate. The cells were transfected with 2 μg of pCEP4-Neo-AID and 5 μL of Lipofectamine TM 2000 for 6 h, washed and maintained in IMDM containing 10% FCS and HT for one day, then the cells were expanded in IMDM with 10% FCS, HT, 1 mg/mL neomycin for 11 days and incubated with PE-conjugated anti-HA antibody and TNFα-GFP (1:10000 in cold opti-MEM medium [Invitrogen, USA], to detect antigen binding ability.) for 30 min at 4 °C, and sorted for cells that expressed high affinity antibodies using a FACSAriaIII (BD) flow cytometer.
Detection of antibody mutations. The genomic DNA of sorted RMCE-Ab cells was purified with a genomic DNA purification system (Promega). The antibody gene fragments were amplified from the genomic DNA and cloned into pMD-19 T-vector. Primers used for antibody gene PCR were as follows: CMV-F: CGCAAATGGGCGGTAGGCGTG; TM-R: CTGCGTGTCCTGGCCCACAGC.
Quantitative real-time PCR analysis. Total RNA was isolated with RNeasyMini Kit (Qiagen) following the manufacturer's protocol. For reverse transcription-PCR (RT-PCR), 2 μg of total RNA was reverse transcribed in a reaction volume of 20 μl to form cDNA using the SuperScript First-Strand Synthesis System (Invitrogen). Real-time PCR was performed using the StepOnePlus system (ABI) with SYBR Green I (Takara) to label amplified DNA. A standard curve method of quantification was used to calculate the expression of target genes relative to the housekeeping gene β-actin. Experiments were performed three times. The following primer pairs were used for the PCR reactions: www.nature.com/scientificreports www.nature.com/scientificreports/ mAID-F1:GGCCACCTTCGCAACAAG; mAID-R1:AGTCTCCGCAGCCCCTCA; β-actin-Chinese hamster -F2:GGCCAACCGTGAAAAGATGA; β-actin-Chinese hamster -R2:CGACCAGAGGCATACAGGGAC. PCR procedures for these genes were template denaturation at 95 °C for 10 min, then 40 cycles of 95 °C for 10 sec, 60 °C for 30 sec, and a final extension at 72 °C for 3 min.
Antibody affinity measurement. The affinities of various antibodies were measured by the Octet biomolecular interaction technology (ForteBio Octet, Menlo Park, CA, USA). TNFα-GFP protein were conjugated with biotin according to the manufacturer (ForteBio Octet), and concentrated to 50 μg/mL. The concentrated antigen was added and fixed to Streptavidin (SA) biosensors. The preparation of the antibody proteins were described above. The detection conditions used were (1) baseline 240 s; (2) loading 600 s; (3) baseline 180 s; (4) association 120 s with a series of concentrations (8000 nM, 4000 nM, 2000 nM, 1000 nM, 500 nM, 250 nM for wild type antibody; 2000 nM, 1000 nM, 500 nM, 250 nM, 125 nM, 62.5 nM for C344T, G721A and G81T + G721A antibody; 250 nM, 125 nM, 62.5 nM, 31.25 nM, 15.62 nM, 7.81 nM for C720G and C74T + C720A antibody.) of αTNF-α scFv; (5) dissociation 180 s. The K on and K off rates were measured by Octet software and K D was calculated for each antibody mutation by the K off /K on ratio.

Data availability
The data sets generated during and/or analyzed during the current study are available from the corresponding author upon reasonable request.