ABSTRACT Monoclonal antibodies (mAbs) are widely utilized as therapeutic drugs for various diseases, such as cancer, autoimmune diseases, and infectious diseases. Using the avian-derived B

cell line DT40, we previously developed an antibody display technology, namely, the ADLib system, which rapidly generates antigen-specific mAbs. Here, we report the development of a human

version of the ADLib system and showcase the streamlined generation and optimization of functional human mAbs. Tailored libraries were first constructed by replacing endogenous

immunoglobulin genes with designed human counterparts. From these libraries, clones producing full-length human IgGs against distinct antigens can be isolated, as exemplified by the

serve as an integrative platform with unique diversity for rapid de novo generation and optimization of therapeutic or diagnostic antibody leads. Furthermore, our results suggest that

libraries can be constructed by introducing exogenous genes into DT40 cells, indicating that the ADLib system has the potential to be applied for the rapid and effective directed evolution

and optimization of proteins in various fields beyond biomedicine. SIMILAR CONTENT BEING VIEWED BY OTHERS DEVELOPMENT OF A NEW AFFINITY MATURATION PROTOCOL FOR THE CONSTRUCTION OF AN

INTERNALIZING ANTI-NUCLEOLIN ANTIBODY LIBRARY Article Open access 08 May 2024 A SINGLE-DOMAIN ANTIBODY LIBRARY BASED ON A STABILITY-ENGINEERED HUMAN VH3 SCAFFOLD Article Open access 31 July

2024 A SINGLE DONOR IS SUFFICIENT TO PRODUCE A HIGHLY FUNCTIONAL IN VITRO ANTIBODY LIBRARY Article Open access 19 March 2021 INTRODUCTION In the past three decades, multiple approaches have

been taken to reduce the immunogenicity of mAbs for therapeutic usage.1,2 Antibody humanization, in which the complementarity-determining regions (CDRs) of non-human-derived mAbs are grafted

into human scaffolds,3 has been widely employed. Alternatively, in vitro display technologies relying on phage, yeast, or mammalian cells to display antibody fragments, such as

antigen-binding fragments (Fabs) and single-chain variable fragments (scFvs), or full-length IgGs containing human variable (V) regions are powerful tools to identify lead candidate

sequences of human origin.4,5,6,7,8,9,10,11,12,13 In contrast to conventional animal immunization, hybridomas created from the spleen cells of genetically engineered mice carrying human

immunoglobulin loci1,14,15,16,17 can be applied for de novo generation of human antibodies.2 Different transgenic animals, such as chickens harboring human antibody variable (V) region

genes, have also been developed.18,19 Despite the invention of these technologies to generate low-immunogenicity therapeutic mAbs, bottlenecks still remain to be resolved.2 For example,

antibody humanization is a laborious process typically performed on a case-by-case basis and often results in unpredictable changes in specificities or affinities for antigens.20 Due to the

limitation of Fabs or scFvs being expressed, further genetic engineering is required to reformat isolated antibody fragments displayed by antigen-specific phages into complete IgG molecules.

For affinity maturation of the antibodies, these conventional methods often need additional rounds of diversification-selection cycles,5,21 prolonging the overall discovery timeline. For de

novo antibody generation platforms, it usually takes months to establish stable hybridomas, and most importantly, there are antigens that are difficult to raise antibodies against due to

immune tolerance. On the other hand, in the case of chickens harboring human V regions, the IgM, IgY, and IgA formats of immunoglobulin heavy chain constant regions are produced after

immunization, thereby necessitating the cloning of each V region gene into the expression vector followed by introduction into culture cells to obtain human IgGs, which is time consuming and

may alter the activities of mAbs.18,19 We previously reported a rapid method to generate mAbs using an avian-derived B cell line.22,23,24 Avian immunoglobulin loci contain only a single

functional variable region for each of the heavy chain (HC) and light chain (LC), whereas clusters of pseudogenes, which show homology to the functional V region, are present in the

near-upstream region. To generate diversity, copies of pseudogenes are transferred to the functional V region via unidirectional gene conversion (GC) (a type of homologous recombination)

(Fig. 1a).25 DT40 is a chicken-derived B cell line in which GC continuously occurs at the immunoglobulin loci.26 We discovered that this GC frequency is markedly enhanced when DT40 cells are

treated with the histone deacetylase inhibitor trichostatin A (TSA). Using TSA-stimulated GC, an autonomously diversifying library (ADLib) was generated, and clones expressing

antigen-specific mAbs can be enriched within ~10 days27,28,29,30 (Fig. 1b). However, as the original ADLib system expresses chicken IgMs, humanization is still required for subsequent

development.29 To overcome this drawback, we aimed to develop a human version of the ADLib system that expresses full-length human IgGs for antibody selection. The exons of the HC and LC

loci of DT40 cells were replaced by human counterparts, and a variety of designed human pseudogenes were inserted upstream of each functional V region (Fig. 2a, b). We confirmed that by TSA

treatment, GC occurred at the humanized immunoglobulin loci of these cells, yielding diversified libraries of human IgGs. From these libraries, clones specific to target antigens, including

vascular endothelial growth factor-A (VEGF-A) and tumor necrosis factor α (TNFα), were enriched, with several possessing neutralizing activities. Moreover, mAbs with improved affinities were

readily obtained through secondary libraries originating from just 1 or 2 weeks of extended culturing of the parental clones, and their enhanced neutralizing activities correlated well with

observed gene conversion events. These results demonstrate that the human ADLib system is a rapid and effective technology platform for the de novo generation and optimization of potential

therapeutic human mAb leads for various diseases, including cancer, autoimmune diseases and emerging infectious diseases. Our data also show that molecular libraries can be constructed by

introducing exogenous genes into specific loci in DT40 cells, suggesting that the ADLib system can potentially be applied for the molecular evolution of non-antibody proteins for various

purposes, including next-generation therapeutics and diagnostics. RESULTS DESIGN OF PSEUDOGENES Since GC is a type of homologous recombination, homology of nucleotide sequences between

donors (pseudogenes) and recipients (functional V region) is required. In avian immunoglobulin genes, four framework regions (FRs) (FR1–4) are highly homologous among pseudogenes and

functional V genes, contributing poorly to the immunoglobulin diversity. Conversely, three CDRs (CDR1–3) are highly diverse and effectively expand the repertoire of immunoglobulin genes (the

sequences of the functional V genes and pseudogenes are shown in Fig. 3 and Supplementary Table 1, respectively). We incorporated these structural features into the design of human

pseudogenes by inserting CDR1–3 sequences derived from the V region cDNA and germline sequences between conserved FRs (see “METHODS” for details). CONSTRUCTION OF THE HUMAN ADLIB SYSTEM The

cells constituting the human ADLib system were prepared by replacing the exons of functional antibody genes of DT40 with their human counterparts as well as inserting a designed human

version of the pseudogenes. We attempted two approaches (Fig. 2c), with the first being “single clone-derived libraries (SCLs)”, in which genetically engineered single clones were cultured

independently in the presence of TSA as a proof-of-concept experiment to enhance GC and acquire diversity. The other approach was “polyclone-derived libraries (PCLs)”, in which diverse HC V

regions derived from human peripheral blood mononuclear cells (PBMCs) were replaced with those of DT40 cells. These cells were further treated with TSA, yielding a mixed state for further

diversification to achieve increased coverage of sequence space (Fig. 2c). To construct SCLs, we genetically engineered DT40 cells by gene targeting31 (Supplementary Figs. 1 and 2). With

regard to the functional HC locus, the HC constant region (CH1, CH2, CH3, M1, and M2) of human IgG1 (hIgG) was introduced into the endogenous constant region of DT40 with simultaneous

deletion of the first exon (Supplementary Fig. 1a, c, and d). Since the ADLib system utilizes membrane-bound and secreted isoforms of antibodies for selection and screening, respectively,

cDNA of the secreted version of hIgG’s constant region was fused with the intron and exon containing the hIgG1 transmembrane domain for alternative splicing and expression of both isoforms.

The endogenous VH, DH and JH regions of DT40 were replaced by the human counterparts derived from human germlines VH3–23, D5–12, and JH4 (three segments were linked directly without P/N

nucleotides) (Fig. 3a, Supplementary Fig. 1b, c, and d). Similarly, regarding the functional LC locus, the chicken VL, JL, and CL regions were also replaced by those of human λ counterparts

derived from human Burkitt’s lymphoma Ramos cells (Fig. 3b, Supplementary Fig. 2a, e). The HC pseudogenes were designed so that the FRs were identical to those of the functional VH region

(generated by linking VH3–23, D5–12, and JH4; Fig. 3a), and the human HC CDRs derived from databases were inserted between the FRs. For the LC pseudogenes (Supplementary Table 1), the FRs

were identical to those of the Ramos V genes (Fig. 3b), and the human LC λ CDRs were derived from the database (Supplementary Table 1). For each of the LC and HC pseudogene loci, 15

pseudogenes (~7.5 kb; all the designed pseudogenes were placed in the same orientation as the functional V regions) were introduced initially as a proof-of-concept experiment. The chicken LC

pseudogene locus was deleted to limit the templates of GC to only the designed human pseudogenes (Supplementary Fig. 2b, e), and the array of 15 designed human LC pseudogenes was

subsequently introduced (Supplementary Fig. 2c, e). Meanwhile, the chicken HC pseudo V locus was not deleted because of its size (~80 kb). The array of 15 designed human HC pseudogenes was

placed upstream of the VH (Supplementary Fig. 1c, d). To enable the ultimate replacement of the pseudogenes via recombination-mediated cassette exchange (RMCE),24,32,33 Cre recombinase

recognition sites (loxm3 and loxm7RE) were inserted at the 5′ flanking region of the designed pseudogene cluster (Supplementary Fig. 1c). In each step, the selection marker was eliminated by

transient expression of Cre recombinase. The resulting clone was referred to as L15H15. We analyzed the L15H15 cells by flow cytometry and confirmed the expression of membrane-bound hIgG

(Supplementary Fig. 3a, b). Immunoblot analysis of L15H15 culture supernatants also showed the presence of secreted hIgGs (Supplementary Fig. 3c). These results suggest that L15H15 cells can

readily be used to generate libraries for mAb selection. PREPARATION OF SCLS USING L15H15 CELLS To construct the SCLs, L15H15 cells were cultured in the presence or absence of TSA for 90

days. On days 21, 42, and 90, the sequence diversity of the HC and LC V regions was analyzed by 454GS Junior (Roche)-based deep sequencing. A total of 239 independent sequences were sampled,

and the number of unique sequences was counted. The number of unique sequences of the VH and VL regions (hereafter, VHDHJH and VLJL are collectively referred to as the “VH region” and “VL

region”, respectively) was markedly increased in the libraries cultured with TSA, whereas those cultured without TSA exhibited little diversification (Fig. 4a). The sequence diversity of the

VH region after 42 days of culture was observed mostly in CDRs (Fig. 4b), whereas FRs had few sequence variations. Importantly, the sequence changes in CDRs were attributable to GC events

templating one of the designed pseudogenes. We also identified single-nucleotide substitutions that had no corresponding sequence in the pseudogenes, possibly due to somatic hypermutation

(SHM). We investigated the usage of pseudogenes after 42 days of culture and found that various pseudogenes were utilized as templates for GC to the VH region (Table 1); on the other hand,

pseudogene usage for the VL region was rather limited compared to that for the VH region (Fig. 5a and Table 2). ENHANCED DIVERSITIES OF LIBRARIES BY ADDITION OF PSEUDOGENE COPY NUMBERS

Arakawa et al. reported that the reduced copy number of pseudogenes lowers the GC frequency and enhances the frequency of SHM, suggesting that the copy number of pseudogenes is an important

determinant of frequent GC.34 To test whether the addition of pseudogene copy numbers could improve the diversity of SCLs, we therefore added 15 LC pseudogenes by target integration upstream

of the existing LC pseudogenes of L15H15 cells (referred to as clone “L30H15”) (Supplementary Fig. 2d, f). We also introduced 30 HC pseudogenes in the forward orientation upstream of the HC

pseudogenes of L30H15 by two rounds of RMCE (15 pseudogenes per round), affording L30H45(fff) (Supplementary Fig. 4a, d). Using these clones, we examined the effects of the copy number of

pseudogenes on sequence diversity. After 21 days of TSA treatment, the VL of L30H45(fff) became more diversified than that of the L15H15 libraries, with the effect persisting into day 42,

although the increase was not strictly proportional to the copy number of pseudogenes (Fig. 4c). In contrast, although the increased diversity of the VH of L30H45(fff) compared to that of

L15H15 showed a significant difference after 21 days of TSA treatment, the values were statistically indistinguishable after 42 days of TSA treatment (Fig. 4c). According to a previous

report, pseudogenes in the reverse orientation relative to the functional V region are preferably utilized as GC templates over those in the forward orientation.35 We thus constructed

L30H45(rrf) (30 LC and 15 HC pseudogenes were inserted in forward orientation, whereas 30 HC pseudogenes were inserted in reverse) in hope of further boosting the diversity (Supplementary

Fig. 4a, b). Contrary to our expectation, the number of unique sequences was comparable between L30H45(fff) and L30H45(rrf) (Fig. 5b). ISOLATION OF ANTIGEN-SPECIFIC MABS FROM HUMANIZED DT40

CELLS We then sought to generate mAbs from SCLs using histidine-tagged human semaphorin 3A (hSema3A) fused with alkaline phosphatase (hSema3A-his-AP) as a model antigen. hSema3A is a factor

involved in the regulation of neuronal axon guidance, the immune system, angiogenesis and osteoprotection.36 We previously succeeded in generating hSema3A-neutralizing mAbs by the original

ADLib system.29,30 L30H45(fff) cells treated with TSA for 86 days were subjected to mAb selection using hSema3A-his-AP-conjugated magnetic beads, followed by limiting dilution cloning. Three

clones specifically reacted with the target (Supplementary Fig. 6a, b), and sequence analysis revealed that these clones had identical HC and LC sequences. In comparison to the original V

region, we detected sequence changes in the VH region, possibly due to the transfer of the pseudogenes pVH09 and pVH04 by GC in HC CDR1 (CDR-H1) (Supplementary Fig. 6c). In CDR-H2, partial

transfer of pVH04 and pVH08 was observed. The LC sequence was identical to that of the original Ramos sequence. The observation of multiple mutated sequences attributed to GC suggests that

the antigen specificity was largely established by GC. CONSTRUCTION OF PCLS The positive results obtained using SCLs led us to explore the feasibility of the second approach, PCLs (Fig. 2c).

Diverse V regions were sourced from human cDNAs of healthy donor PBMCs and cloned into an hVH targeting construct using arm sequences identical to those of the VH knock-in vector. The

resulting targeting plasmid libraries (hVH libraries) were knocked into the VH region of the L30H45(fff) cells (Fig. 6a). The knocked-in clones were treated with TSA to enhance their

diversity in a mixed state. The diversity was later evaluated by comparing hVH sequences between the cDNA libraries and the knocked-in libraries using deep sequencing. The population of

cells that underwent targeted integration of exogenous human VH showed sequence diversity in their HC CDR-H3 length (Fig. 6b). The amino acid usage in the 14-amino-acid-long CDR-H3 of the

knocked-in cells showed a divergent distribution that was well correlated with the hVH plasmid library (Fig. 6c). To further expand the diversity of both SCLs and PCLs, an additional 15 VH

pseudogenes were integrated by RMCE into L30H45(fff) cells in forward orientation to generate L30H60(ffff) cells (Supplementary Fig. 4c). Moreover, as all the abovementioned humanized DT40

cells contained λ LCs, a κ version of the corresponding LC knock-in construct (consisting of the human κ constant region, κ VL region, and 27 designed human κ V pseudogenes) was established

based on L30H60(ffff) to increase the diversity of κ LC (referred to as K27H60(ffff)) (Supplementary Fig. 5a, b). The functional Vκ region was generated by linking the human FRs

(corresponding to those of human germline Vκ 1–39 and Jκ2) and the human CDRs derived from the database (Fig. 3c). For the pseudogenes, κ LC CDRs from the database were inserted between the

FRs (Supplementary Table 1) identical to those of the functional Vκ. The expression of LC was examined by flow cytometry and immunoblot analysis to confirm the expression of κ LCs

(Supplementary Fig. 3a–c). Finally, the hVH plasmid libraries were knocked into L30H45(fff), L30H60(ffff) and K27H60(ffff) cells to create PCLs for subsequent mAb selection. SELECTION OF

FUNCTIONAL MABS AGAINST VEGF PCLs produced by the knocked-in cells (L30H45(fff), L30H60(ffff) and K27H60(ffff)) treated with TSA were combined with SCLs (L30H45(fff), L30H60(ffff) and

K27H60(ffff)), and the whole mixtures were subjected to mAb selection against human VEGF-A (hVEGF-A). Anti-hVEGF-A clones were enriched by magnetic sorting followed by single-cell sorting.

The specificities of the isolated clones were examined by ELISA (Supplementary Fig. 7a). Interestingly, using purified proteins, some of the obtained clones were shown to inhibit the binding

of VEGF-A to VEGFR2 (Supplementary Fig. 7b). However, some clones did not show inhibitory activity in a concentration-dependent manner. This suggests that the affinity of the isolated

anti-VEGF-A clones might be insufficient for functional activity. Thus, in pursuit of improved affinities of the anti-hVEGF-A clones, we moderately diversified the immunoglobulin genes of

anti-hVEGF-A clones (C018 and A033) by prolonged culturing in the absence of TSA. The anti-hVEGF-A clones cultured for 8 days contained a small number of cells reacting with hVEGF-A better

than the parental clones under similar surface IgG expression levels. These cells were isolated by a cell sorter and shown to have relatively high reactivity to VEGF-A (Fig. 7a).

Correspondingly, C018-AM20 exhibited notable concentration-dependent inhibition of VEGF-A binding to VEGFR2 compared to its parental clone (Supplementary Fig. 7b). Next, we carried out a

cell-based assay37,38 to monitor the inhibitory effect of isolated mAbs on the intracellular signal transduction cascades of VEGF. Using human umbilical vein endothelial cells (HUVECs), we

showed that several clones possess inhibitory activity against p44/42 MAPK phosphorylation (Supplementary Fig. 7c). Moreover, these mAbs were shown to inhibit the ability of hVEGF-A to

stimulate the proliferation of HUVECs (Fig. 7b). SPR experiments revealed that all of the isolated anti-VEGF-A clones had nM-order KD values (Fig. 7c, d). Remarkably, the clones obtained

after affinity maturation (A033AM-19 and C018AM-20) displayed ~100-fold higher affinities (Fig. 7c, d: KD = 4.91 × 10−11 M and KD = 1.49 × 10−11 M, respectively) than the corresponding

parental clones (KD = 2.14 × 10−9 M and KD = 1.61 × 10−9 M, respectively) and exhibited improved antagonistic activity (Fig. 7b, Supplementary Fig. 7b, c), suggesting that the streamlined

affinity maturation process contributes to the improvement of the functionalities of mAbs. Strikingly, the increased affinities resulted from a drastic decrease in the dissociation rate

constants (Fig. 7d). Sequence analysis of the HCs revealed that B015, B021, and C018 contain VH3-23-derived sequences (Fig. 7e). Among them, B021 contains a D3-3-derived sequence. Since D3-3

was significantly different from the knocked-in original synthetic VDJ sequence (VH3-23, D5-12, and JH4) designed for the SCL, it was speculated that the sequences were derived from PCLs.

The D regions of B015 and C018 could not be identified due to sequence alteration, possibly by GC events, and their J regions were more similar to JH4b than the synthetic VDJ sequence. For

the LC, B015, B021, and C018 were κ versions, and various GC tracts were observed (Supplementary Fig. 7d). Interestingly, comparison of the C018 and C018AM-20 HCs revealed that they are

different in terms of a pseudogene-derived sequence (pVH04) and a point mutation at CDR2 (Fig. 7e). As their LCs are identical, the GC and SHM likely confer improved affinities. A033 HC

harbors VH1-69 (Supplementary Fig. 7e), indicating that it originated from the PCL. The LC of A033 was of the λ version, and corresponding sequences to two pseudogenes were observed

(Supplementary Fig. 7f). The LC sequence of A033AM-19 was identical to that of A033, implying that the transfer of pVH09 to A033 CDR-H1 by GC is responsible for the improved affinity of

A033AM-19. SELECTION OF FUNCTIONAL MABS AGAINST TNFΑ Next, anti-human TNFα (hTNFα) clones were selected from the mixed library by hTNFα-conjugated magnetic beads. The specificities of the

candidate clones were analyzed by ELISA (Supplementary Fig. 8a). The clone with the strongest ELISA signal (#314) was further cultured without TSA for 11 days, and a subclone reacting to

hTNFα better than #314 was obtained (#314-AM043) (Supplementary Fig. 8b). #314-AM043 was further diversified for an additional 14 days and subjected to a second round of affinity maturation

to obtain a clone with even better reactivity (#314AM-043-01) (Fig. 8a). The neutralizing activity of these clones was examined by NF-κB activation.39 The reporter construct containing a

fluorescent protein (ZsGreen) and an NF-κB response element was stably integrated into BaF3 cells. Binding of TNFα to TNFR1 activates the NF-κB signaling pathway, resulting in upregulated

ZsGreen expression (Fig. 8b, c). The #314 clone was shown to have neutralizing activity (Fig. 8c), and undoubtedly, the clone obtained after two rounds of affinity maturation (#314AM-043-01)

showed improved functionalities, which was correlated with a one-order improvement in affinity as determined by SPR (Fig. 8d, e). Sequence analysis of the HC of #314 revealed that it

contained VH3-23 (Fig. 8f). Although the original D region could not be identified because of alteration by two GC events, this region might have been derived from PCLs since the J regions

are more similar to JH4b than the VDJ sequence designed for SCLs. The sequence corresponding to pseudogene pVH53 was found in the CDR-H2 of #314AM-043-01, indicating that the improved

affinity is conferred by GC. The LC of #314 was of the κ version, and moreover, a sequence corresponding to a pseudogene was found in CDR-L1, which could also be attributed to GC

(Supplementary Fig. 8c). DISCUSSION In this paper, we describe the development of the human version of the ADLib system by replacing chicken immunoglobulin genes with the corresponding human

counterparts and, most importantly, by introducing human pseudogenes. DT40 cells with designed human immunoglobulin loci capable of building unique immunoglobulin repertoires were

diversified by TSA treatment to generate cell-based display libraries for the isolation of mAbs with functional activities against various antigens, as exemplified by neutralizing

anti-hVEGF-A and anti-hTNFα clones. Furthermore, clones with improved affinities and functionalities were seamlessly obtained after being first isolated by a simple culturing process to

moderately diversify the cell population. Taking full advantage of this special property of DT40 cells, the affinity maturation process requires just one to two weeks of culturing followed

by cell sorting. Copious cell-based human mAb display technologies have been reported to date.7,8,9,10,11,12,13 In most cases, mAbs are generated from predetermined libraries constructed by

exogenous introduction of diverse human antibody genes. The ABELmAb library, however, is one of the exceptions.10 In this method, antigen-specific clones are generated from fixed libraries,

and the subsequent overexpression of activation-induced deaminase (an essential factor of SHM40) in these clones introduces SHMs into the immunoglobulin genes, allowing affinity maturation

of the isolated clones. Meanwhile, the human ADLib system is unique in that the libraries are actively diversified by GC in addition to SHM, creating an increased capacity of variation. It

has been reported that GC events can be induced among exogenous sequences incorporated into DT40 cells. For example, Kanayama et al. knocked in blue fluorescent protein (BFP) and green

fluorescent protein (GFP) genes into the immunoglobulin locus of DT40 and observed GC events between BFP and GFP.41 In other reports, Leighton et al. and Schusser et al. replaced the

immunoglobulin genes of DT40 with human counterparts, inserted designed human pseudogenes, and observed that GC occurred moderately among V regions and pseudogenes,42,43 as a prerequisite

proof-of-concept experiment for the establishment of transgenic chickens (OmniChicken) that generate human mAbs in vivo based on GC events among human sequences.18 Therefore, the human ADLib

system is the first demonstration of the use of GC-based technology for the coupled in vitro generation and affinity maturation of human mAbs. Both the ADLib system and OmniChicken

transgenic animal technology rely on the antibody repertoire exhibited by avian B cells in vitro and in vivo, respectively, for the discovery of mAbs with desired functionalities. Although

it is plausible that the DT40 cells constructed by Schusser et al. could provide an in vitro diversified antibody repertoire if treated with TSA, the sequence diversity of the hypothetical

libraries might be lower than that of our system given the presence of considerably fewer copy numbers of pseudogenes (<17 copies for HC and <16 copies for LC) being introduced than

those in the DT40 cells constituting the human ADLib system.42,43 Homologous recombination between the exogenously introduced human V region and pseudogenes has been verified in the B cells

of OmniChicken in vivo, yet higher diversity could, in theory, be achieved if a larger number of pseudogenes are integrated. Considering the differences in the number and the adopted

sequences of pseudogenes, the human ADLib system and OmniChicken could provide different mAb repertoires, suggesting that the two avian-based systems can be used complementarily. Transgenic

animals, including OmniChicken, have been reported to work well as in vivo methods of mAb generation1,14,15,16,17,18,19; however, difficulties in establishing stable chicken hybridomas

necessitate special cloning techniques to retrieve the V region genes from B cells expressing antigen-specific mAbs. Moreover, the HC constant regions of OmniChickens as well as the

engineered DT40 cells are in chicken format (IgM, IgY, and IgA), thereby requiring additional genetic engineering processes to obtain mAbs in human IgG format, not to mention the extra work

of expression and purification of recombinant antibodies for downstream evaluation. In contrast, activities of mAbs identified by our human ADLib system can be directly evaluated in human

IgG format using culture supernatants of DT40. By harnessing the rapid proliferation property of DT40 cells, up to hundreds of micrograms of human IgG can also be easily obtained from the

supernatants by affinity purification for analytical characterization or for other assays requiring relatively low protein amounts. The key features of the human ADLib system for effective

antibody generation are (i) the unique diversity generated by GC, (ii) seamless and efficient affinity maturation after initial isolation of a prototype mAb, and (iii) rapidness and low

number of manipulation steps required for the isolation and affinity maturation of mAbs. Each of these attributes is expected to greatly facilitate the discovery and optimization of lead

therapeutic and diagnostic antibodies for multiple diseases, such as cancer, autoimmune diseases and emerging infectious diseases. Moreover, the ADLib system’s distinctive mechanism for

active building of the antibody library provides an alternative to conventional antibody generation methods, enabling the creation of human mAbs with novel activities and potentially better

developability. Clearly, our results also indicate that artificial libraries can be constructed in DT40 cells by introducing exogenous genes and their corresponding homologs into

immunoglobulin loci, thereby suggesting the possibility of exploiting the ADLib system as an in vitro evolution platform for nonimmunoglobulin proteins. Indeed, in addition to a previous

report,41 we also succeeded in inducing GC events between two homologous genes encoding fluorescent proteins in DT40 cells and showed that novel chimeras of existing fluorescent proteins can

be obtained.44 By leveraging the power of GC activity in DT40 cells as the “engine” of diversification in a controlled manner, the ADLib system could hence potentially be applied for

directed protein evolution and optimization in various fields beyond biomedical applications, such as in material and agricultural sciences. METHODS CELL CULTURE DT40 cells were cultured at

39.5 °C in 5% CO2 in IMDM containing 10% fetal bovine serum (Access), 1% penicillin/streptomycin, 0.5 μM monothioglycerol and 1% chicken serum (Thermo Fisher) as described.22,23,26 For TSA

treatment, cells were passaged every day at a concentration of 3 × 105 cells/mL in medium containing 2.5 ng/mL TSA (Fujifilm). HUVECs were cultured in formulated EGMTM-2 endothelial growth

medium (Lonza) at 37 °C in 5% CO2. BaF3 cells were cultured at 37 °C in RPMI 1640 medium containing 1 ng/mL recombinant mouse IL-3 (R&D System), 1% penicillin/streptomycin, and 10% fetal

bovine serum (Access). DESIGN OF THE PSEUDOGENES A pseudogene unit consisted of FR1, CDR1, FR2, CDR2, FR3, CDR3, and FR4, in that order. For the Igλ and Igκ chains, the FRs of all the

pseudogenes were based on those of hVL, consisting of IGLV2-14*01 and IGLJ2*01, and hVL, consisting of Vκ 1-39 and Jκ2, respectively. For hVH, FRs obtained by linking human germline VH3-23,

D5-12, and JH1 sequences were used. A variety of pairs of CDR1 and CDR2 sequences from Igλ and HC were extracted from the germline V genes found in V-BASE

(http://www2.mrc-lmb.cam.ac.uk/vbase/) and human immunoglobulin V genes found in GenBank (http://www.ncbi.nlm.nih.gov/genbank/) (30 pairs for Igλ and 59 pairs for HC). After the sectioning

of each sequence according to the Kabat numbering scheme, CDR1 and CDR2 sequences were obtained. Regarding the design of CDR3 sequences, approximately 1000 nonredundant CDR3 sequences were

obtained from V-BASE and GenBank. In addition, the hypothetical CDR3 sequences were designed to reflect all combinations of V, D, and J regions. Next, a phylogenetic tree was constructed by

the neighbor-joining method using the CDR3 candidate sequences. The 60 individual CDR sequences were selected for HC pseudogene construction. With regard to Igλ, 30 individual CDR sequences

were selected for Igλ pseudogene construction. These sequences represent major clusters of the phylogenetic tree. Amino acid lengths also vary. For Igκ pseudogene construction, the κ V genes

were obtained from GenBank and sectioned for each CDR. Twenty-seven individual CDR sequences with a Levenshtein distance of 2–7 against the functional κ V gene were selected for each CDR.

The median amino acid lengths were 12, 11, and 9 for HC, Igλ, and Igκ, respectively. Levenshtein distance was calculated by R with the Biostrings package (Supplementary Table 1).

CONSTRUCTION OF THE VL AND CONSTANT REGION KNOCK-IN VECTOR cDNA of Ramos was prepared using SuperScript® III Reverse Transcriptase (Thermo Fisher Scientific), and Ramos λ V region (hVL) VL6

(GenBank Z74694.1) was amplified with the primers F1 and R1　(primers are described in Supplementary Table 3). The Ramos Igλ constant region (hCλ) was amplified with the primers F2 and R2.

The noncoding 1.5-kb genomic region of DT40 adjacent to and upstream of hVL was amplified with the primers F3 and R3 and referred to as the left arm. The 1.7-kb center arm fragment

corresponding to the genomic locus between hVL and hCλ was synthesized. The noncoding 2.0-kb DT40 genomic region adjacent to and downstream of hCλ was synthesized and referred to as the

right arm. The hVL and hCλ regions along with the drug resistance genes and flanking lox sequences were placed between homology arms. All the fragments (left arm, hVL, center arm, hCλ and

right arm) were assembled into pBluescript SK (-) by the In-Fusion system (Takara Bio). For the κ chain, functional VL was designed based on Vκ1-39 and Jκ2 and synthesized along with the hCκ

sequence. The left arm (amplified by F4 and R4), Vκ, blasticidin resistance genes (F5 and R5) and Cκ (F6 and R6) were amplified and fused by assembly PCR. The resulting fragment was

digested with XhoI and NotI and cloned into the XhoI and NotI sites of the Igλ knock-in vector. CONSTRUCTION OF THE CHICKEN LC PSEUDOGENE KNOCK-OUT VECTOR A 2.8-kb DT40 genomic region

adjacent to and upstream of the chicken LC pseudogenes was amplified by the primers F7 and R7 and used as the left arm. A 1.9-kb genomic region adjacent to and downstream of the LC

pseudogenes was amplified by the primers F8 and R8 and used as the right arm. The left and right arms were assembled into pUC19 by In-Fusion along with the neomycin resistance marker.

CONSTRUCTION OF THE LC PSEUDOGENE KNOCK-IN VECTOR Each pseudogene unit was linked by a 125-bp linker sequence in which a 95-bp sequence was derived from the intervening region between

chicken endogenous pseudogenes ψVL1 and ψVL2 (GCCTGACACACACAGCCCTCCCCAGACTGTGGGATGAGGCCCTGTGCCCGCAGTCACATGTGGAATATCAAGACACACACATCTATGACAATCAC) and a 30-bp index sequence. Arrays of 15 or 30

pseudogenes linked by the linkers were synthesized (Genewiz). Each array was cloned into AgeI and ClaI sites located between the downstream lox sequence and the left arm in the LC chicken

pseudogene knock-out construct. For the κ chain, pseudogenes were linked by sequences derived from the chicken genomic region between LC pseudogenes. Linker sequences are shown in

Supplementary Table 2. CONSTRUCTION OF THE HC CONSTANT REGION KNOCK-IN VECTOR The human HC constant region sequence was isolated from the hIgG1 genomic sequence (NW_001838121.1) consisting

of CH1, hinge, CH2, CH3, and transmembrane domains (hIgG1 M1 and M2). The intron between each constant region was derived from the mouse IgG2a intron. The 4.3-kb DT40 genomic region adjacent

to and downstream of the chicken V region was amplified with the primers F9 and R9 and used as a 5′ homology arm (the center arm). The 0.9-kb DT40 genomic region containing chicken Cμ2 was

amplified with the primers F10 and R10 and used as a 3′ homology arm (the right arm). The targeting plasmid was constructed by assembling the left arm, the hIgG1 constant region, the center

arm, the neomycin resistance gene under the control of IRES and the right arm in pBluescript SK(-). CONSTRUCTION OF THE VH KNOCK-IN VECTOR Human hVH was designed by linking human germline

VH3-23 and synthetic VDJ sequences consisting of D5-12 and JH4. The 1.4-kb DT40 genomic region adjacent to and upstream of the chicken signal sequence of VH was amplified with the primers

F11 and R11 and used as the left arm. The 4.4-kb DT40 genomic region adjacent to and downstream of chicken VH was amplified with　the primers F12 and R12 and used as the right arm. Along with

the blasticidin resistance gene, the left arm, hVH and the right arm were cloned into pBluescript SK(-) using the In-Fusion system to construct the hVH knock-in vector. CONSTRUCTION OF THE

HC PSEUDOGENE KNOCK-IN VECTOR A stretch of 15 pseudogenes (pVH01 to pVH15) was linked to a 125-bp linker sequence as described in the LC pseudogene design. For the knock-in vector of the

first 15 pseudogene arrays, the 1.8-kb DT40 genomic region immediately upstream of the left arm of the VH knock-in vector described above was used as the 5′ homology arm (left arm). The

right arm was amplified with the primers F11 and R11. The SV40 promoter, along with the neomycin resistance gene in reverse orientation flanked by lox511 sequences, was placed upstream of

the 15 pseudogene array. Downstream, the SV40 promoter along with duplicated terminators was placed for the screening of targeted clones with sequential RMCE. The loxm3 sequence

(ATAACTTCGTATATGGTATTATATACGAAGTTAT) upstream of the terminators and the loxm7RE sequence (ATAACTTCGTATATTCTATCTTATACGAACGGTA) downstream of the terminators were incorporated for RMCE. The

targeting vector was constructed in pUC19 in the following order: the left arm, lox511RE sequence, SV40-neomycin resistance gene in reverse orientation, lox511LE sequence, SV40 promoter,

loxm3 reverse sequence, duplicated terminators, loxm7RE sequence, array of 15 pseudogenes, right arm and human HC V region. CONSTRUCTION OF THE RMCE DONOR VECTOR For introduction of the

pseudogene array by the first and third rounds of RMCE, the donor vector was designed in the following order: the loxm3 reverse sequence, neomycin resistance gene without the SV40 promoter,

duplicated terminators, loxPLE sequence (ATAACTTCGTATAATGTATGCTATACGAACGGTA), array of 15 pseudogenes, and loxm7LE sequence (TACCGTTCGTATATTCTATCTTATACGAAGTTAT). The donor vector for the

second round of RMCE was assembled in the following order: the loxm3 reverse sequence, blasticidin resistance gene without the SV40 promoter, duplicated terminators, loxm7RE sequence

(ATAACTTCGTATATTCTATCTTATACGAACGGTA), 15 pseudogene array, and loxPRE sequence (TACCGTTCGTATAATGTATGCTATACGAAGTTAT). All the components were cloned into pUC57. The 15 pseudogenes (pVH16 to

pVH30) for RMCE were linked to a 125-bp linker sequence as described in the LC pseudogene design. The 15 pseudogenes (pVH31 to pVH45) for the 2nd round of RMCE were linked with 100-bp

sequences derived from the chicken genomic region between HC pseudogenes. The 15 pseudogenes (pVH46 to pVH60) for the 3rd round of RMCE were linked to linker sequences as described in the LC

pseudogene design except that 30-bp index sequences were used only for pVH48, 50, 52, 53, 55, 57, and 59. Each linker sequence is shown in Supplementary Table 2. CONSTRUCTION OF THE

KNOCK-IN LIBRARY OF THE HUMAN VH REGION AND THE GENERATION OF THE PCL The germline-derived divergent hVH harboring CDR3 was prepared from normal human peripheral blood CD19 + B-cell cDNA

(ALLCELLS) using degenerate primers that annealed to FR3 (F13) and FR4 (R13). The PCR fragment harboring CDRs 1 and 2 was prepared using another primer set that annealed to FR1 (F14) and FR3

(R14). These PCR fragments were assembled by PCR using primers (F15) and (R15) and then cloned into EcoRV and NdeI sites in the hVH targeting vector with the CAG promoter-blasticidin

resistance gene flanked by Vlox sequences (VloxM1; TCAATTTCCGAGAATGACAGTTCTCGGAAATTGA). The resultant construct was transformed into _E. coli_, the colonies were harvested, and plasmids were

purified. The pooled library plasmids were linearized by PciI and introduced into the cell by electroporation. The hIgG1-negative cells were purified by a MACS column (Miltenyi), and the

recovered cells were expanded. The VCre recombinase expression vectors were transiently transfected, and the resultant cells were subjected to MACS purification to concentrate the

hIgG1-positive cells. The obtained cells were diversified by TSA treatment for 21 or 42 days. DNA TRANSFECTION The knock-in vectors were transfected by GenePulser (Bio-Rad Laboratories) as

previously described.24 Vectors were linearized by NotI, except that SalI and ScaI were used for HC constant region knock-in and the hVH knock-in vector, respectively. RMCE donor vectors

were transfected using Nucleofector 2b (Lonza) and Cell Line Nucleofector Kit T (Lonza). A total of 1 × 107 cells were collected and transfected with 8 μg of DNA mixture (7 μg of RMCE

construct and 1 μg of Cre recombinase expression vector) with the Nucleofector 2b (Lonza) optimized transfection program B-023. GENOTYPING PCR Genetic engineering by knock-in and RMCE was

examined by PCR using PrimeSTAR® GXL DNA Polymerase (Takara Bio). The primer pairs are shown in Supplementary Table 3. FLOW CYTOMETRY Cells (5 × 105) were washed twice with FCM buffer (PBS

containing 0.05% bovine (BSA) serum albumin and 2 mM EDTA) and stained with antibodies for 30 min at 4 °C. The antibodies used were R-PE conjugated goat anti-human IgG (gamma chain specific)

(SouthernBiotech), R-PE conjugated goat anti-human Igλ (SouthernBiotech), and FITC conjugated goat anti-chicken IgM (Bethyl Laboratories). After washing with FCM buffer three times, cells

were suspended in FCM buffer and analyzed by FACSCanto II (Becton, Dickinson and Company). The data were analyzed by FlowJo software (Becton, Dickinson and Company). NEXT-GENERATION SEQUENCE

Genomic DNA was isolated from 1 × 106 cells after TSA treatment. The VL regions were amplified by PCR using sense primer (CGTATCGCCTCCCTCGCGCCATCAGNNNNNNNNNNCAGGTTCCCTGGTGCAGGC, where N

indicates index sequences) and antisense primer (CTATGCGCCTTGCCAGCCCGCTCAGGCTTGGTCCCTCCGCCGAA), and VH regions were amplified with sense primer (CTATGCGCCTTGCCAGCCCGCTCAGTCCGTCAGCGCTCTCT)

and antisense primer (CGTATCGCCTCCCTCGCGCCATCAGNNNNNNNNNNTGGGGGGGGTTCATATGAAG, where N indicates index sequences). PCR products were purified and analyzed using a 454 GS-junior system (Roche

Diagnostics). The sequence reads with lengths less than 250 bp between the first read and the sequence position with quality values <15 were filtered out from further analysis, and each

sequence was sectioned into the immunoglobulin framework and CDR according to the Kabat numbering scheme. To evaluate the number of unique sequences (NUS), we repeated 10 attempts of the

following cycles: (i) a certain number of read extractions, (ii) determination of the FR/CDR, and (iii) counting of unique sequence reads. The average of the 10 attempts was determined as

NUS. ANTIGEN PREPARATION The Flag-tagged proteins (FLAG-hTNFα, FLAG-hHer2, and FLAG-hVEGF-A) were transiently expressed in FreeStyle293-F cells (Thermo Fisher Scientific) and purified by

anti-FLAG M2 affinity chromatography (Sigma) followed by gel filtration chromatography (GE Healthcare). hSema3A-his-AP was stably expressed in HEK293 cells (a kind gift from Dr. Yoshio

Goshima) and purified by a HisTrap Excel column (GE Healthcare) followed by gel filtration chromatography. ISOLATION OF ANTIGEN-SPECIFIC CELLS FROM THE ADLIB LIBRARIES The antigen-specific

clones were isolated as described with some modifications.22,23 The purified proteins were immobilized onto Dynabeads M-270 Epoxy, Dynabeads M-270 Carboxylic acid and Dynabeads Isolation and

Pulldown (Thermo Fisher Scientific). The library cells were seeded at a concentration of 1.5 × 107 cells/40 mL and cultured for 24 h. Cells were harvested, washed, and resuspended in 950 μL

of selection buffer (PBS supplemented with 1% BSA), and antigen-immobilized beads were added and incubated at 4 °C for 30 min. The beads were washed out, and the isolated cells were seeded

onto 96-well plates and incubated. After 1 week of culturing, the antigen-specific clones were screened by ELISA and subcloned by limiting dilution. Alternatively, cells were isolated by

single-cell sorting using a FACS Aria Fusion (Becton, Dickinson and Company). The cloned cells were subjected to a second screening by ELISA. IMMUNOBLOTTING Immunoblotting was carried out as

previously described24 with some modifications. The cells were harvested, lysed by sample buffer containing 2-mercaptoethanol and separated by SDS-PAGE using a 4–20% gradient gel. Blotting

was carried out using an iBlot 2 Gel Transfer Device and iBind Western Device (Thermo Fisher Scientific). Human immunoglobulin HC, Igλ, Igκ, and chicken IgM were detected by HRP-conjugated

goat anti-hIgG-Fc (Bethyl), anti-human Igλ (Southern Biotech), anti-human Igκ (Southern Biotech) and anti-chicken IgM (Bethyl), respectively. ECL western blotting detection reagents (GE

Healthcare) and LAS-4000 Mini (Fujifilm) were used for signal detection. ELISA ELISA was carried out as previously described24 with some modifications. Antigens were immobilized onto

MaxiSorp 384-well plates at 62.5 ng/well at 4 °C overnight and then blocked by blocking buffer (PBS containing 1% BSA). After washing, the culture supernatants were added to the wells and

incubated for 1 h at room temperature. Mouse anti-hIgG-Fc HRP conjugated (Southern Biotech) was added to the wells after washing and incubated for 45 min at room temperature. The assay was

developed with TMB (Dako), the reaction was stopped with 1 N sulfuric acid, and the absorbance was read at 450 nm. AFFINITY MATURATION The antigen-specific clones were cultured in medium

without TSA for one week or two weeks. For negative staining, 5.0 × 106 cells were first stained with 100 nM biotinylated Her2 in FCM buffer and further stained with AlexaFluor 488-labeled

streptavidin (Thermo Fisher). The cells were harvested and subsequently incubated with 10 nM biotinylated target antigens followed by staining with AlexaFluor 647-labeled streptavidin

(Thermo Fisher) and PE-conjugated anti-hIgG (Southern Biotech). The staining conditions were identical to those of FCM. The cells showing high specificities for the antigens at given IgG

expression levels were sorted by FACS Aria Fusion (Becton, Dickinson and Company) into 96-well plates. HVEGF COMPETITION ASSAY The competition assay was carried out by ELISA with some

modifications. Recombinant human VEGF R2/KDR Fc chimera protein (R&D Systems) was immobilized onto MaxiSorp 384-well plates at 62.5 ng/well at 4 °C overnight. The plate was washed with

PBS containing 0.05% Tween 20 and soaked with blocking buffer at room temperature for 1 h. During the incubation, the competition reaction mixture of human VEGFR2 (final concentration 20 

ng/mL) and the anti-VEGF-A-IgG1 antibody clone (final concentration 0.1, 1, or 10 μg/mL) was preincubated for 30 min at room temperature. Following the removal of blocking buffer and

washing, the competition reaction mixture was added into 96-well plates and incubated for 30 min at room temperature. After washing, signals were detected by an HRP-conjugated anti-FLAG M2

antibody (Sigma). PHOSPHORYLATION INHIBITION ASSAY The inhibition of MAPK phosphorylation by anti-VEGF-A mAbs was examined as previously described37 using the PathScan Phospho-p44/ 42 MAPK

(Thr202/ Tyr204) Sandwich ELISA Kit (Cell Signaling). Pooled 2.5 × 105 HUVECs were seeded into six-well plates and cultured overnight. The medium was removed and washed with 2 mL of Ham’s

F-12K containing 1% BSA. The cells were incubated for 6 h for starvation. An aliquot of 110 μL of recombinant hVEGF165 (Pepro Tech) diluted with medium (25 nM) and an equal volume of various

anti-VEGF mAbs were mixed and incubated for 30 min at room temperature. Two hundred microliters of the mixture was added to the culture and incubated at 37 °C for 5 min. The plates were

then incubated on ice, and the medium was removed. Finally, the cells were washed with ice-cold PBS and lysed, followed by analysis by sandwich ELISA. CELL PROLIFERATION ASSAY The inhibition

of cell proliferation by anti-VEGF-A mAbs was examined as previously described.38 HUVECs were seeded at 2.0 × 103 cells/well in 96-well plates and incubated at 37 °C overnight. On the next

day, recombinant human VEGF165 was reacted with serially diluted anti-hVEGF-A-hIgG1 mAbs in medium (VEGF165: 20 ng/mL and anti-hVEGF-A-hIgG1 mAbs: 10, 1, or 0.1 μg/mL) at 37 °C for 1 h

before being added to HUVEC culture plates, and then, the cells were further cultured for an additional 2 days in the medium. The viability of the cells was examined with a CellTiter-Glo

Luminescent Cell Viability Assay (Promega). NF-ΚB REPORTER GENE ASSAY The inhibitory activity of anti-TNFα mAbs was examined by a reporter assay essentially as previously described.39 An

aliquot of 24 pM hTNFα was mixed with serially diluted purified anti-hTNFα mAbs in the medium and incubated for 30 min at room temperature. The mixtures were added to 1 × 104 BaF3 cells

carrying the NF-κB reporter construct (Fig. 8b) in 96-well plates and cultured for 4 h. The expression level of ZsGreen was examined by FACSCanto II. SPR ANALYSIS SPR analysis was performed

as previously described24 with some modifications using a Biacore T200 (GE Healthcare). Anti-hIgG capture antibodies (GE Healthcare) were immobilized onto the surface of a CM5 sensor chip 

via amine coupling. The purified antigens serially diluted with HBS-EP + (100, 33.3, 11.1, 3.7, 1.23, and 0.412 nM) were injected for 240 s at a flow rate of 30 µL/min followed by 300 s of

dissociation. Data were fitted with Biacore T200 evaluation software using a 1:1 Langmuir binding model with mass transfer. STATISTICAL ANALYSIS Statistical calculations were performed with

Microsoft Excel and are described in the relevant figure legends. Differences were compared using a one-tailed _t_-test. REFERENCES * Lonberg, N. Human antibodies from transgenic animals.

_Nat. Biotechnol._ 23, 1117–1125 (2005). CAS  PubMed  Google Scholar  * Nelson, A. L. & Reichert, J. M. Development trends for therapeutic antibody fragments. _Nat. Biotechnol._ 27,

331–337 (2009). CAS  PubMed  Google Scholar  * Winter, G., Jones, P. T., Dear, P. H., Neuberger, M. S. & Foote, J. Replacing the complementarity-determining regions in a human antibody

with those from a mouse. _Nature_ 321, 522–525 (2004). Google Scholar  * Jespers, L. S., Roberts, A., Mahler, S. M., Winter, G. & Hoogenboom, H. R. Guiding the selection of human

antibodies from phage display repertoires to a single epitope of an antigen. _Nat. Biotechnol._ 12, 899–903 (1994). CAS  Google Scholar  * Knappik, A., Ge, L., Honegger, A. & Pack, P.

Fully synthetic human combinatorial antibody libraries (HuCAL) based on modular consensus frameworks and CDR randomized with trinucleotides. _J. Mol. Biol._ 296, 57–86 (2000). CAS  PubMed 

Google Scholar  * Frenzel, A., Schirrmann, T. & Hust, M. Phage display-derived human antibodies in clinical development and therapy. _MAbs_ 8, 1177–1194 (2016). CAS  PubMed  PubMed

Central  Google Scholar  * Feldhaus, M. J. et al. Flow-cytometric isolation of human antibodies from a nonimmune Saccharomyces cerevisiae surface display library. _Nat. Biotechnol._ 21,

163–170 (2003). CAS  PubMed  Google Scholar  * Mazor, Y. et al. Isolation of engineered, full-length antibodies from libraries expressed in Escherichia coli. _Nat. Biotechnol._ 25, 563–565

(2007). CAS  PubMed  Google Scholar  * Zhou, C., Jacobsen, F. W., Cai, L., Chen, Q. & Shen, W. D. Development of a novel mammalian cell surface antibody display platform. _MAbs_ 2,

508–518 (2010). PubMed  PubMed Central  Google Scholar  * Bowers, P. M. et al. Coupling mammalian cell surface display with somatic hypermutation for the discovery and maturation of human

antibodies. _Proc. Natl Acad. Sci. USA_ 108, 20455–20460 (2011). CAS  PubMed  Google Scholar  * Breous-Nystrom, E. et al. Retrocyte Display® technology: generation and screening of a high

diversity cellular antibody library. _Methods_ 65, 57–67 (2014). CAS  PubMed  Google Scholar  * Parthiban, K. et al. A comprehensive search of functional sequence space using large mammalian

display libraries created by gene editing. _MAbs_ 11, 884–898 (2019). CAS  PubMed  PubMed Central  Google Scholar  * Achim, D., Laura, R., Stefan, Z. & Harald, K. Therapeutic antibody

engineering by high efficiency cell screening. _FEBS Lett._ 588, 278–287 (2013). Google Scholar  * Green, L. L. et al. Antigen–specific human monoclonal antibodies from mice engineered with

human Ig heavy and light chain YACs. _Nat. Genet._ 7, 13–21 (1994). CAS  PubMed  Google Scholar  * Fishwild, D. M. et al. High-avidity human IgGκ monoclonal antibodies from a novel strain of

minilocus transgenic mice. _Nat. Biotechnol._ 14, 925–928 (1996). Google Scholar  * Ishida, I. et al. Production of human monoclonal and polyclonal antibodies in TransChromo animals.

_Cloning Stem Cells_ 4, 91–102 (2002). CAS  PubMed  Google Scholar  * Murphy, A. J. et al. Mice with megabase humanization of their immunoglobulin genes generate antibodies as efficiently as

normal mice. _Proc. Natl Acad. Sci. USA_ 111, 5153–5158 (2014). CAS  PubMed  Google Scholar  * Ching, K. H. et al. Chickens with humanized immunoglobulin genes generate antibodies with high

affinity and broad epitope coverage to conserved targets. _MAbs_ 10, 71–80 (2018). CAS  PubMed  Google Scholar  * Ching, K. H. et al. Expression of human lambda expands the repertoire of

OmniChickens. _PLoS ONE_ 15, 1–19 (2020). Google Scholar  * Kuramochi, T., Igawa, T., Tsunoda, H. & Hattori, K. Humanization and simultaneous optimization of monoclonal antibody.

_Methods Mol. Biol._ 1904, 213–230 (2019). CAS  PubMed  Google Scholar  * King, D. J., Bowers, P. M., Kehry, M. R. & Horlick, R. A. Mammalian cell display and somatic hypermutation in

vitro for human antibody discovery. _Curr. Drug Discov. Technol._ 11, 56–64 (2014). CAS  PubMed  Google Scholar  * Seo, H. et al. Rapid generation of specific antibodies by enhanced

homologous recombination. _Nat. Biotechnol._ 23, 731–735 (2005). CAS  PubMed  Google Scholar  * Seo, H. et al. An ex vivo method for rapid generation of monoclonal antibodies (ADLib system).

_Nat. Protoc._ 1, 1502–1506 (2006). CAS  PubMed  Google Scholar  * Hashimoto, K., Kurosawa, K., Murayama, A., Seo, H. & Ohta, K. B cell-based seamless engineering of antibody Fc

domains. _PLoS ONE_ 11, 1–22 (2016). Google Scholar  * Reynaud, C. A., Anquez, V., Grimal, H. & Weill, J. C. A hyperconversion mechanism generates the chicken light chain preimmune

repertoire. _Cell_ 48, 379–388 (1987). CAS  PubMed  Google Scholar  * Buerstedde, J. M. et al. Light chain gene conversion continues at high rate in an ALV-induced cell line. _EMBO J._ 9,

921–927 (1990). CAS  PubMed  PubMed Central  Google Scholar  * Omi, K. et al. Noncompetitive immunoassay detection system for haptens on the basis of antimetatype antibodies. _Clin. Chem._

61, 627–635 (2015). CAS  PubMed  Google Scholar  * Guy, A. T. et al. Glycerophospholipid regulation of modality-specific sensory axon guidance in the spinal cord. _Science_ 349, 974–977

(2015). CAS  PubMed  Google Scholar  * Yamashita, N. et al. Anti-semaphorin 3A neutralization monoclonal antibody prevents sepsis development in lipopolysaccharide-treated mice. _Int.

Immunol._ 27, 459–466 (2015). CAS  PubMed  Google Scholar  * Kanemaru, H., Yamada, Y., Ohazama, A., Maeda, T. & Seo, K. Semaphorin 3A inhibits nerve regeneration during early stage after

inferior alveolar nerve transection. _Sci. Rep._ 9, 1–7 (2019). CAS  Google Scholar  * Buerstedde, J. M. & Takeda, S. Increased ratio of targeted to random integration after

transfection of chicken B cell lines. _Cell_ 67, 179–188 (1991). CAS  PubMed  Google Scholar  * Langer, S. J., Ghafoori, A. P., Byrd, M. & Leinwand, L. A genetic screen identifies novel

non-compatible loxP sites. _Nucleic Acids Res._ 30, 3067–3077 (2002). CAS  PubMed  PubMed Central  Google Scholar  * Minorikawa, S. & Nakayama, M. Recombinase-mediated cassette exchange

(RMCE) and BAC gineering via VCre/vloxP and SCre/SloxP systems. _Biotechniques_ 50, 235–246 (2011). CAS  PubMed  Google Scholar  * Arakawa, H., Saribasak, H. & Buerstedde, J.-M.

Activation-induced cytidine deaminase initiates immunoglobulin gene conversion and hypermutation by a common intermediate. _PLoS Biol._ 2, E179 (2004). PubMed  PubMed Central  Google Scholar

  * McCormack, W. T. & Thompson, C. B. Chicken IgL variable region gene conversions display pseudogene donor preference and 5’ to 3’ polarity. _Genes Dev._ 4, 548–558 (1990). CAS  PubMed

  Google Scholar  * Hu, S. & Zhu, L. Semaphorins and their receptors: from axonal guidance to atherosclerosis. _Front. Physiol._ 9, 1–11 (2018). Google Scholar  * Homer-Vanniasinkam, S.

et al. VEGF-A isoforms program differential VEGFR2 signal transduction, trafficking and proteolysis. _Biol. Open_ 5, 571–583 (2016). PubMed  PubMed Central  Google Scholar  * Popkov, M. et

al. Human/mouse cross-reactive anti-VEGF receptor 2 recombinant antibodies selected from an immune b9 allotype rabbit antibody library. _J. Immunol. Methods_ 288, 149–164 (2004). CAS  PubMed

  Google Scholar  * Tamai, R. et al. Cell activation by monosaccharide lipid A analogues utilizing Toll-like receptor 4. _Immunology_ 110, 66–72 (2003). CAS  PubMed  PubMed Central  Google

Scholar  * Yousif, A. S., Stanlie, A., Begum, N. A. & Honjo, T. Opinion: uracil DNA glycosylase (UNG) plays distinct and non-canonical roles in somatic hypermutation and class switch

recombination. _Int. Immunol._ 26, 575–578 (2014). CAS  PubMed  PubMed Central  Google Scholar  * Kanayama, N., Todo, K., Takahashi, S., Magari, M. & Ohmori, H. Genetic manipulation of

an exogenous non-immunoglobulin protein by gene conversion machinery in a chicken B cell line. _Nucleic Acids Res._ 34, e10 (2006). PubMed  PubMed Central  Google Scholar  * Leighton, P. A.,

Schusser, B., Yi, H., Glanville, J. & Harriman, W. A diverse repertoire of human immunoglobulin variable genes in a chicken B cell line is generated by both gene conversion and somatic

hypermutation. _Front. Immunol._ 6, 1–8 (2015). CAS  Google Scholar  * Schusser, B. et al. Harnessing gene conversion in chicken B cells to create a human antibody sequence repertoire. _PLoS

ONE_ 8, e80108 (2013). PubMed  PubMed Central  Google Scholar  * Ohta, K., Seo, H. & Shibata, T. Method of inducing homologous recombination of somatic cell. U.S. Patent No. 9,587,251

(2017). Download references ACKNOWLEDGEMENTS This paper is dedicated to the memory of Dr. Hitoshi Yamatani, who contributed to the basic studies on the ADLib system. We thank Dr. Yoshiro

Goshima (Department of Molecular Pharmacology and Neurobiology, Yokohama City University Graduate School of Medicine) for supplying us with the hSema3A-his-AP expression vector. Financial

supports were provided in part by the New Energy and Industrial Technology Development Organization (NEDO) and by a Core Research for Evolutional Science and Technology (CREST) grant from

the Japan Science and Technology Corporation (JPMJCR18S3). AUTHOR INFORMATION AUTHORS AND AFFILIATIONS * Department of Life Sciences, Graduate School of Arts and Sciences, The University of

Tokyo, Tokyo, Japan Hidetaka Seo & Kunihiro Ohta * Research Laboratories, Chiome Bioscience Inc, Tokyo, Japan Hitomi Masuda, Kenjiro Asagoshi, Tomoaki Uchiki, Shigehisa Kawata, Goh

Sasaki, Takashi Yabuki, Shunsuke Miyai, Naoki Takahashi, Shu-ichi Hashimoto, Atsushi Sawada, Aki Takaiwa, Chika Koyama, Kanako Tamai, Kohei Kurosawa, Ke-Yi Lin & Yukoh Nakazaki Authors *

Hidetaka Seo View author publications You can also search for this author inPubMed Google Scholar * Hitomi Masuda View author publications You can also search for this author inPubMed 

Google Scholar * Kenjiro Asagoshi View author publications You can also search for this author inPubMed Google Scholar * Tomoaki Uchiki View author publications You can also search for this

author inPubMed Google Scholar * Shigehisa Kawata View author publications You can also search for this author inPubMed Google Scholar * Goh Sasaki View author publications You can also

search for this author inPubMed Google Scholar * Takashi Yabuki View author publications You can also search for this author inPubMed Google Scholar * Shunsuke Miyai View author publications

You can also search for this author inPubMed Google Scholar * Naoki Takahashi View author publications You can also search for this author inPubMed Google Scholar * Shu-ichi Hashimoto View

author publications You can also search for this author inPubMed Google Scholar * Atsushi Sawada View author publications You can also search for this author inPubMed Google Scholar * Aki

Takaiwa View author publications You can also search for this author inPubMed Google Scholar * Chika Koyama View author publications You can also search for this author inPubMed Google

Scholar * Kanako Tamai View author publications You can also search for this author inPubMed Google Scholar * Kohei Kurosawa View author publications You can also search for this author

inPubMed Google Scholar * Ke-Yi Lin View author publications You can also search for this author inPubMed Google Scholar * Kunihiro Ohta View author publications You can also search for this

author inPubMed Google Scholar * Yukoh Nakazaki View author publications You can also search for this author inPubMed Google Scholar CONTRIBUTIONS H.S. and S.H. conceived the study. H.S.,

S.H. and Y.N. designed the overall project. H.M., K.A., T.U., T.Y., S.K., S.M., N.T. and A.S. designed and constructed the knock-in vectors and transfected the vectors into the cells. H.M.,

K.A., T.U., T.Y., S.K., S.M. and N.T. constructed the SCLs, and C.K. and A.S. constructed the PCLs. Sequence analysis of the libraries was performed by G.S., K.A. and A.T. H.M. and K.T.

designed and performed the functional assay for the isolated antibodies. The data obtained from the functional assays were analyzed by H.M. and K.T. H.M. and K-Y.L. purified the antigen and

antibody proteins and performed the SPR analysis. Optimization of the affinity maturation process was performed by K.T. H.S., H.M., K.A., S.H., K.K., K-Y.L., K.O. and Y.N. wrote the paper.

CORRESPONDING AUTHORS Correspondence to Hidetaka Seo or Yukoh Nakazaki. ETHICS DECLARATIONS COMPETING INTERESTS This work was supported by Chiome Bioscience Inc. (Japan). H.M., K.A., S.H.,

A.S., A.T., C.K., K.T., K.K., K-Y.L. and Y.N. are employees of Chiome Bioscience Inc., respectively. H.S., T.U., S.K., G.S., T.Y., S.M. and N.T. are former employees of Chiome Bioscience

Inc., respectively. K.O. is a former board member of Chiome Bioscience Inc.. H.S. and K.O. are stockholders of Chiome Bioscience Inc. H.M., K.A., S.H., A.S., A.T., C.K., K.T., K.K., K-Y.L.

and Y.N. are stock option holders of Chiome Bioscience, respectively. Patent applications have been filed (US20170058029A1) for the technology described in this publication. H.M., K.A.,

S.H., A.S., A.T., T.U, T.Y., S.K., S.M. and N.T. are named as the inventors of these patents. SUPPLEMENTARY INFORMATION 41423_2020_440_MOESM1_ESM.PDF Supplementary Fig.1, Supplementary

Fig.2, Supplementary Fig.3, Supplementary Fig.4, Supplementary Fig. 5, Supplementary Fig.6, Supplementary Fig. 7 and Supplementary Fig.8 SUPPLEMENTARY TABLE 1 SUPPLEMENTARY TABLE 2

SUPPLEMENTARY TABLE 3 RIGHTS AND PERMISSIONS OPEN ACCESS This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation,

distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and

indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to

the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will

need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. Reprints and permissions ABOUT THIS ARTICLE

CITE THIS ARTICLE Seo, H., Masuda, H., Asagoshi, K. _et al._ Streamlined human antibody generation and optimization by exploiting designed immunoglobulin loci in a B cell line. _Cell Mol

Immunol_ 18, 1545–1561 (2021). https://doi.org/10.1038/s41423-020-0440-9 Download citation * Received: 13 February 2020 * Revised: 06 April 2020 * Accepted: 07 April 2020 * Published: 26 May

2020 * Issue Date: June 2021 * DOI: https://doi.org/10.1038/s41423-020-0440-9 SHARE THIS ARTICLE Anyone you share the following link with will be able to read this content: Get shareable

link Sorry, a shareable link is not currently available for this article. Copy to clipboard Provided by the Springer Nature SharedIt content-sharing initiative KEYWORDS * Immunoglobulin

rearrangements * homologous recombination * antibody therapeutics * antibody engineering * drug discovery