Coupling Recombinase-Mediated Cassette Exchange with Somatic Hypermutation for Antibody Afﬁnity Maturation in CHO Cells

ARTICLE

Coupling Recombinase-Mediated Cassette Exchange With Somatic Hypermutation for Antibody Afﬁnity Maturation in CHO Cells

Chuan Chen,1,2 Nan Li,1,2 Yun Zhao,1 Haiying Hang1 1 Key Laboratory for Protein and Peptide Pharmaceuticals, National Laboratory of Biomacromolecules, Institute of Biophysics, Chinese Academy of Sciences, Beijing 100101, China; telephone: þ86-10-64888473; fax: þ86-10-64888473; e-mail: [email protected] 2 University of Chinese Academy of Sciences, Beijing, China

Introduction ABSTRACT: Heterologous expression of activation-induced cytidine deaminase (AID) can induce somatic hypermutation Technologies such as phage (de Bruin et al., 1999; Huse et al., 1992; (SHM) for genes of interest in various cells, and several research Smith, 1985; Winter et al., 1994), yeast (Boder and Wittrup, 1997; groups (including ours) have successfully improved antibody affinity in mammalian or chicken cells using AID-induced SHM. These Feldhaus et al., 2003), and bacterial displays (Francisco and affinity maturation systems are time-consuming and inefficient. In Georgiou, 1994; Mazor et al., 2009; Qiu et al., 2010) have been used this study, we developed an antibody affinity maturation platform in for antibody affinity maturation in vitro. In recent years, Chinese hamster ovary (CHO) cells by coupling recombinase- mammalian cell surface display has also been developed (Akamatsu mediated cassette exchange (RMCE) with SHM. Stable CHO cell et al., 2007; Higuchi et al., 1997; Ho et al., 2006; Wolkowicz et al., clones containing a single copy puromycin resistance gene (PuroR) expression cassette flanked by recombination target sequences (FRT 2005); it has advantages in protein folding, post-translational and loxP) being able to highly express a gene of interest placed in the modification, and code usage (Ho et al., 2006). Several groups have cassette were developed. The PuroR gene was replaced with an successfully coupled cell surface display with SHM to improve antibody gene by RMCE, and the antibody was displayed on the cell antibody affinity in either B cell lines (Akamatsu et al., 2007; surface. Cells displaying antibodies on their membrane were Arakawa et al., 2008; Cumbers et al., 2002; Ho et al., 2006; Seo et al., transfected with the AID gene, and mutations of the antibody gene were accumulated by AID-mediated hypermutation during cell 2005) or non-B cell lines (Bowers et al., 2011; Chen et al., 2012; proliferation followed by flow cytometric cell sorting for cells bearing McConnell et al., 2012). SHM is induced by activation-induced antibody mutantswith improved affinity.Affinity improvementswere cytidine deaminase (AID), which converts cytosine to uracil to detected after only one round of cell sorting and proliferation, mutant introduce mutations and amino acid substitutions during antibody fi fi clones with15-fold af nity improvement were isolated within ve affinity maturation (Martin and Scharff, 2002; Maul and Gearhart, rounds of maturation (within 2 months). CHO cells are fast growing, stress-resistant and produce antibody with glycosylations suitable for 2010; Shivarov et al., 2008). B cell lines with endogenous AID therapy. Our antibody-evolution platform based on CHO cells makes expression can generate SHM themselves (Seo et al., 2006; Todo antibody-affinity maturation more efficient and is especially et al., 2006), while other cell lines without endogenous AID convenient for therapeutic antibody affinity improvement. expression were transfected with the AID gene to induce SHM Biotechnol. Bioeng. 2016;113: 39–51. (Bowers et al., 2011; Chen et al., 2012; Yoshikawa et al., 2002). ß 2015 Wiley Periodicals, Inc. Episomal and non-episomal vectors were used to display KEYWORDS: activation-induced cytidine deaminase (AID); antibodies (Bowers et al., 2011; Ho et al., 2006). Episomal vectors fi somatic hypermutation (SHM); af nity maturation; recombinase- cannot be integrated in chromosomes, and it takes a longtime and mediated cassette exchange (RMCE); flow cytometric cell sorting multiple flow cytometric sortings to enrich a cell population that stably displays antibodies at high levels. High-level antibody displaying cell populations or cell clones can be generated by random integration of an antibody-expression vector, but it also takes a longtime and multiple flow cytometric sortings to reach this goal. Even if cells with an antibody highly displayed on the surface were generated, multiple copies of antibody genes often coexist in a

Correspondence to: Dr. Hai-Ying Hang and Dr. Yun Zhao cell (Baer et al., 2000; Horlick et al., 2000). The mixed antibody Received 3 July 2014; Revision received 7 October 2014; Accepted 17 November 2014 display makes antibody affinity maturation inefficient, and more Accepted manuscript online 15 January 2015; time is needed to identify the best antibody gene. Article first published 31 July 2015 online in Wiley Online Library (http://onlinelibrary.wiley.com/doi/10.1002/bit.25541/abstract). CHO cell lines are widely used in antibody expression as they are DOI 10.1002/bit.25541 fast growing, stress-resistant and have identical or similar post-

ß 2015 Wiley Periodicals, Inc. Biotechnology and Bioengineering, Vol. 113, No. 1, January, 2016 39 translational modifications to humans (Schmidt, 2004; Walsh and plasmid pCEP-Neo-mAID was created in two steps: first, the Jefferis, 2006), and it is convenient to generate cell lines with only hygromycin resistance gene between NruI and SalI in pCEP4 one copy of retargetable high-level expression cassette by RMCE in (Invitrogen, USA) was replaced with a neomycin resistance gene CHO cells (Kim and Lee, 2008; Qiao et al., 2009). A DNA fragment of from pCDNA3.1(þ) to form the pCEP-Neo plasmid, and then the interest can be efficiently and unidirectionally inserted into a mAID (mouse activation-induced deaminase (Chen et al., 2012) preferred pre-determined genomic locus flanked with FRTand loxP gene was inserted into pCEP-Neo between HindIII and XhoI. In sequences by simultaneously expressing Flp and Cre recombinases order to compare the stabilities of different vectors to display anti- (dual RMCE) (Anderson et al., 2012)). In this study, we generated a TNF-a antibody, the SP-HA-Ab-TM cassette was PCR amplified and CHO cell clone bearing such a retargetable expression cassette in a inserted into pCDNA3.1-hygro(þ) and pCEP4 vectors between genome locus that maintained high expression of puromycin HindIII and XhoI to generate pCDNA3.1-Ab and pCEP4-Ab. resistance protein; the puromycin resistance gene (PuroR) was then The seven chosen antibody mutants (Table II) were generated as replaced with an antibody gene. Cells highly displaying the antibody hybrid scFv (mouse scFv-human Fc, scFv-hFc) for affinity analysis. were obtained by flow sorting only after 2 days of culture. The cells The scFv-hFc expression cassettes were inserted into the pCEP4 were transfected with AID and a mutant antibody with 15-fold of vectors between HindIII and XhoI to generate pCEP4-scFv-hFC affinity improvement was generated by five rounds of flow sorting- expression vectors. cell proliferation within 2 months. The primers used in constructing the above plasmids are listed in Table I. All the constructed plasmids were confirmed by sequencing. Materials and Methods Cell Culture Vectors CHO/dhFr- cells (12200036, Cell Bank of the Chinese Academy of To establish cell clones in which the FRT-PuroR-loxP cassette is Sciences, Shanghai, China) and cell lines derived from CHO/dhFr- integrated in a chromosome site, which can then be replaced by an were propagated in IMDM medium (HyClone) containing 10% fetal antibody gene of interest, a series of plasmids were created as bovine serum (FCS, HyClone), 0.1 mM hypoxanthin, and 0.016 mM described below. To generate the plasmid pCDNA3.1/hygro(þ)- thymidine (HT, Gibco, USA), at 37 C in a 5% CO2 incubator. DHFR (named as pCD), a SV40 promoter-DHFR-SV40 polyA Subcultures were carried out every 2–3 days. CHO-S cells expression cassette (Huang et al., 2007) was inserted into the (Invitrogen, USA) were maintained in FreeStyleTM CHO Expression pCDNA3.1/hygro(þ) plasmid at the BglII site. The plasmid Medium (Invitrogen, USA) in a round bottle in an ES-W orbital þ pCDNA3.1/hygro( )-FRT-GFP-loxP-DHFR (named as pCDFGL) shaker (BIOTOP, Shanghai, China) at 130 rpm, 37 C, and 8% CO2. was created by inserting the FRT-EGFP-loxP cassette into pCD Flp-In CHO cells (Invitrogen, USA) were maintained in F12 medium between HindIII and XhoI. The FRTand loxP sequences were added (HyClone) with 10% fetal bovine serum (FCS, HyClone), at 37Cina on the two sides of the EGFP gene by PCR using two primers 5% CO2 incubator. containing FRTand loxP sequences on their 50 ends, respectively. To generate the dual recombinase expression plasmid pCI-Flp-2A-Cre Transfection, Stable Cell Line Establishment, and (named as pF2AC), we used the framework of the F2AC containing Antibody Affinity Maturation Flp-2A-Cre cassette (Anderson et al., 2012), but replaced Flpe and Cre with Flpo and iCre by overlapping PCR. The resulting cassette The plasmid pCDFGL was linearized by BglII digestion prior to was then inserted into the pCI plasmid (Promega, Madison, WI) electroporation. CHO/dhFr- cells (1E6) in log phase were collected between NheI and XhoI. Flpo is a variant of Flp, and it is more active by trypsinization and resuspended in 0.1 mlLopti-MEM medium than the wild type Flp (Raymond and Soriano, 2007); iCre is a (Invitrogen, USA) together with 2.5 mg linearized pCDFGL DNA. variant of Cre, and it is more active than the wild type Cre The mixture was then pulsed at 160 V,7.5 ms in a 0.2 cm gap cuvette (Shimshek et al., 2002); the 2Aused here is the TaV variant of the 2A (BIO-RAD Gene Pulser XcellTM). The electroporated cells were peptides, and it is more self-cleaving efficient than the other 2A plated into a 10 cm dish containing 10 mL IMDM medium with 10% peptides (Donnelly et al., 2001). The exchange vectors pFRT-RFP- FCS, HT, and 250 mg/mL hygromycin and grown for 7 days, and loxP (named as pFRL), pFRT-PuroR-loxP (named as pFPL), and then expanded without hygromycin for 8 more days. Cells with the pFRT-Ab-loxP (named as pFAbL) were created by inserting mRFP highest GFP fluorescence intensity were sorted by a FACSAriaIII (Campbell et al., 2002), PuroR (puromycin resistance gene), and SP- (BD) flow cytometer with a sorting window as shown in Figure 2A, HA-Ab-TM (signal peptide-HA tag-antibody-transmembrane the sorted CHO-FRT-GFP-loxP (CHO-FGL) cells were expanded in domain fusion gene (Chen et al., 2012) cassettes into the pCI IMDM medium. plasmid between BamHI and BglII, respectively. The CMV promoter Generally, to replace the gene integrated in chromosome(s) in between BamHI and BglII sites in pCI was deleted as a result. CHO/dhFr- cells, 4E5 cells were seeded into each well of a six-well To generate the plasmid PET28a(þ)-TNF-a-GFP, the open plate. Cells were transfected with a mixture of 0.5 mgexchange reading frame of the EGFP gene was inserted into PET28a(þ) vector, 2 mgpF2AC,and7.5mL LipofectamineTM 2000 (Invi- between SacI and XhoI, then the open reading frame of the human trogen, USA) for 5 h. Afer transfection, the cells were transferred TNF-a gene was inserted into the NcoI and EcoRI sites. The cassette into a 10 cm dish containing IMDM medium with 10% FCS and can express a TNF-a-GFP fusion protein, which can be used for HT before flow cytometric sorting of cells containing the gene of monitoring the surface-displayed TNF-a-binding antibodies. The interest as described specifically below. CHO-FGL cells were co-

40 Biotechnology and Bioengineering, Vol. 113, No. 1, January, 2016 Table I. Primers used for generating vectors

Primers Vectors HindIII-DHFR-P1: 50-GACTGAAGCTTATGGTTCGACCGCTGAACTGCAT-30 pCD XhoI-DHFR-P2: 50-CTGACCTCGAGTTAGCCTTTCTTCTCATAGACTTCAAATTTATACTTGAT-30 Overlap-SV40*-DHFR-P1: 50-CCTAGGCTTTTGCATGGTTCGACCGCTGAACTGCATC-30 Overlap-SV40pA-DHFR-P2: 50-CCAGAGTCCCGCTTAGCCTTTCTTCTCATAGACTTCAAATTTATACTTGAT-30 BamHI-SV40*-P1: 50-TGACTGGATCCGGTGGTTAAAGTCCCCAGGCTCC-30 Overlap-DHFR-SV40*-P2: 50-CGGTCGAACCATGCAAAAGCCTAGGCCTCCAAAAAAGC-30 Overlap-DHFR-SV40pA-P1: 50-GAGAAGAAAGGCTAAGCGGGACTCTGGGGTTCGAAAT-30 BglII-SV40 polyA-P2: 50-GACTGAGATCTGGTATACAGACATGATAAGATACATTGAT-30 HindIII-FRT-BamHI-GFP-P1: pCDFGL 50-GACTGAAGCTTGAAGTTCCTATTCTCTAGAAAGTATAGGAACTTCGGATCCATGGTGAGCAAGGGCGAGGAGCT-30 XhoI-Loxp-EcoRV-GFP-P2: 50-GACTGCTCGAGATAACTTCGTATAGCATACATTATACGAAGTTATGATATCCTACTTGTACAGCTCGTCCATGCCGAG-30 Flpo-NheI-P1: 50-GTCTGGCTAGCATGAGCCAATTTGATATATTATGTAAAACACCAC-30 pF2AC Flpe-overlap-2A-P2: 50-CACGTCCCCGCATGTTAGAAGACTTCCCCTGCCCTCGGCTCTTATGCGTCTATTTATGTAGGATGAAAGG-30 2A-overlap-cre-P1: 50-GAAGTCTTCTAACATGCGGGGACGTGGAGGAAAATCCCGGGCCCATGTCCAATTTACTGACCGTACACC-30 Flag-Cre-XhoI-P2: 50-CAGTCCTCGAGCTACTTGTCGTCATCGTCTTTGTAGTCATCGCCATCTTCCAGCAGGCG-30 BamHI-FRT-HindIII-RFP-P1: pFRL, pFPL, pFAbL, 50-GCACTGGATCCGAAGTTCCTATTCTCTAGAAAGTATAGGAACTTCAAGCTTATGGCCTCCTCCGAGGACGTCAT-30 pCDNA3.1-Ab and pCEP4-Ab BamHI-Loxp-XhoI-RFP-P2: 50-GCACTGGATCCATAACTTCGTATAGCATACATTATACGAAGTTATCTCGAGCTAATAAGATCTGGCGCCGGTGGAG-30 Puro-HindIII-P1: 50-TCACTAAGCTTATGACCGAGTACAAGCCCACGGT-30 Overlap-puro-TK-P1: 50-CCGGTGCC-30 Overlap-puro-TK-P2: 50-GTACGAAGCCATGGCACCGGGCTTGCGGGT-30 TK-XhoI-P2: 50-TGTCACTCGAGTCAGTTAGCCTCCCCCATCTCCC-30 SP-3f8-HindIII-P1: 50-TCACTAAGCTTATGACCCGGCTGACCGTGCT-30 TM-XhoI-P2: 50-GATCACTCGAGCTAACGTGGCTTCTTCTGCCAAAGC-30 GFP-SacI-P1: 50-TACTGGAGCTCATGGTGAGCAAGGGCGAGGAGCT-30 PET28a(þ)-TNFa-GFP GFP-XhoI-A-P2: 50-GACTACTCGAGTCTTGTACAGCTCGTCCATGCCGAG-30 TNFa-NcoI-P1: 50-TGAGTCCATGGTCAGATCATCTTCTCGAACCCCG-30 TNFa-EcoRI-P2: 50-CTGTCGAATTCCAGGGCAATGATCCCAAAGTAGACC-30 pCEP-NEO-NruI-P1: 50-GACTGCTCGCGAAATTCTGTGGAATGTGTGTCAGTTAGGGT-30 pCEP-Neo-mAID pCEP-NEO-SalI-P2: 50-GTCAGGTCGACGGTATACAGACATGATAAGATACATTG-30 CMV-SalI-P1: 50-GTAGCGTCGACCAATTCTCATGTTTGACAGCTTATCATCGC-30 CMV-SalI-P2: 50-TCATCGTCGACCTCATGGCTGCGCCCCGACAC-30 mAID-HindIII-P1: 50-GCACGAAGCTTGCCACCATGGACAGCCTTCTGATGAAGCAAAAG-30 mAID-XhoI-P2: 50-CATCGCTCGAGTCAAAATCCCAACATACGAAATGCATCT-30 3f8-SCFV-EcoRI-P2: 50-GTCACGAATTCCCGTTTTATTTCCAACTTTGTCCCC-30 pCEP4-scFv-hFC hFC-EcoRI-P1: 50-CTGTCGAATTCGAGCCCAAATCTTGTGACAAAACTCAC-30 hFC-XhoI-P2: 50-TGACTCTCGAGTTATTATTTACCCGGAGACAGGGAGAGG-30 transfected with pFRL and pF2AC, and grown without any 10 mg/mL puromycin for 10 days to obtain cells in which RFP was antibiotic for 7 days. The cells expressing RFP but not GFP were replaced by PuroR in the RFP-integrated chromosome site and to sorted and expanded. Then the cells were subjected to two more kill those in which PuroR was not integrated into a chromosome. rounds of sorting (the top 10% of RFP expressing cells) and After puromycin selection, two populations were detected (non- expansion, and the resulting cell population was named as CHO- fluorescent and RFP fluorescent cells) as was shown in Figure 3A FRL-S2. The CHO-FRL-S2 cells were co-transfected with pFPL (RMCE 2). We sorted individual non-fluorescent cells (RMCE 2, (containing the puromycin-resistant gene PurR)andpF2AC gate L) into each well of 96-well plates (FACSAriIII single cell (containing the two recombinases) to mediate the replacement of sorting module) to generate single cell clones in which RFP was RFP with PuroR, and the cells were maintained in IMDM with replaced by PuroR. To obtain antibody expressing cells (named as

Chen et al.: Antibody Affinity Maturation in CHO Cells 41

Biotechnology and Bioengineering RMCE-Ab) CHO-FPL cells were co-transfected with pFAbL and The FreeStyleTM MAX CHO Expression System (Invitrogen, USA) pF2AC and grown for 2 days. Afterwards, the cells were collected was used to generate parent and mutant anti-TNF-a antibodies- and incubated with PE-conjugated anti-HA antibody (Abcam, manufacturer’s protocol. The affinities of various antibodies were 1:300 in cold opti-MEM medium [Invitrogen, USA]) and GFP- measured by the Octet biomolecular interaction technology by hTNF-a (0.2 mg/mL in cold opti-MEM medium [Invitrogen, transient transfection of the pCEP4-scFv-hFC expression vectors USA]) for 25 min at 4C, washed with cold opti-MEM once, according to the platform (ForteBio Octet, Menlo Park, CA, USA). resuspended in cold opti-MEM, and sorted with a sorting window The scFv-hFc fusion protein supernatants were concentrated to as shown in Figure 3C. 100 mg/mL, then detected by seven Anti-hFc kinetic grade To mature antibody affinity, 5E6 RMCE-Ab cells which displayed biosensors (ForteBio:18–5060). The detection conditions used Anti-TNF-a antibodies were seeded into a 10 cm plate. The cells were (I) baseline 240 s; (II) loading 240 s; (III) baseline 180 s; (IV) were transfected with 16 mg of pCEP-Neo-AID and 48 mLof association 120 s with a series of concentrations (1600 nM, 800 nM, LipofectamineTM 2000 for 5 h, washed and maintained in IMDM 400 nM, 200 nM, 100 nM, 50 nM, 25 nM) of GFP-TNF-a antigen; containing 10% FCS and HT for one day, then the cells were (V) dissociation 180 s. The Kon and Koff rates were measured by expanded in IMDM with 10% FCS, HT, 1 mg/mL G418 for 7 days Octet software and Kd was calculated for each antibody mutation by and sorted for cells that expressed high affinity antibodies. the Koff/Kon ratio.

Flow Cytometry Results FACSAriaIII (BD) and FACSCalibur (BD) flow cytometers were used Generation of CHO Cells for High-Level Antibody for cell sorting and analysis. Expression The method to sort cells that express a gene of interest was described in Transfection, stable cell line establishment and In order to generate a highly gene-expressing, stable, and antibody affinity maturation. retargetable cell line for antibody evolution, Flp and Cre dual To analyze the display level and TNF-a binding ability of the RMCE technology was chosen in this study (Anderson et al., 2012), antibody on the cell surface, antibody displaying cells were and several vectors were constructed for RMCE (Fig. 1A–C). An incubated with PE-conjugated anti-HA antibody (Abcam, 1:300 in overview of the RMCE procedure is demonstrated in Figure 1D. cold opti-MEM medium [Invitrogen, USA]) and GFP-hTNF-a A GFP-expressing pCDFGL plasmid (Fig. 1A) was tansfected into (0.2 mg/mL in cold opti-MEM medium [Invitrogen, USA]) for CHO/dhFr- cells by electroporation. The transfectants were 25 min at 4C, cells were then washed and resuspended with cold submitted to hygromycin selection for 7 days and expanded opti-MEM, and measured by a FACSCalibur flow cytometer. without hygromycin for 8 more days. Afterwards, the top 10% of GFP expressing cells (CHO-FGL) were flow-sorted (Fig. 2A) and expanded for RMCE 1. The expansion without hygrommycin Southern Blot followed by flow sorting was intentionally performed to enrich cells Genomic DNA used for Southern blotting was purified with a that highly and stably express GFP without antibiotic (Southern and genomic DNA purification system (HF206, YuanPingHao Bio, Berg, 1982). A similar practice was carried out in the following Beijing, China) according to the manufacturer’s protocol. The DNA procedure to obtain a CHO cell clone to stably express and display a (25 mg) was digested overnight with HindIII and separated by high level of antibody. electrophoresis on a 0.8% agarose gel in 1 TAE buffer. The gel was To further enrich the cells that can stably and highly express the treated with 0.25 M HCl, followed by alkaline denaturation and gene of interest in an integrated reusable cassette, we carried out the neutralization. DNA was transferred to a positively charged nylon following experiments. The expanded CHO-FGL cells were co- membrane (Pharmacia, Amersham, England) in 20 SSC and transfected with the RFP exchange vector pFRL (Fig. 1C) and the immobilized by UV cross-linking. Southern blotting was carried out recombinase-expressing vector pF2AC (Fig. 1B). The Flpo with DIG High Prime DNA Labeling and Detection Starter Kit II (Raymond and Soriano, 2007) and iCre (Shimshek et al., 2002) (Roche, Germany) with a probe against the hygromycin gene. recombinases were expressed and released with a 2A self-cleaving Primers used for PCR probe production were as follows: peptide in the cells, mediating cassette exchange to replace the GFP Hygro-prob-P1: 50-ATGAAAAAGCCTGAACTCACCGCG-30 gene integrated in chromosome(s) with the RFP cassette on the Hygro-prob-P2: 50-GAACCCGCTCGTCTGGCTAAGATC-30 pFRL vector (Fig. 1D, RMCE 1). The transfectants were cultured for 7 days, cells with both RFP and GFP fluorescence should have multiple expression cassettes, and cells with only RFP fluorescence Antibody Gene Sequencing and Antibody Affinity Analysis were most likely those in which GFP was completely replaced by The genomic DNA of sorted RMCE-Ab cells were purified with a the RFP gene; these cells contained a gene-replaceable expressing genomic DNA purification system (Promega). The antibody gene cassette. The RFP fluorescence only cells (CHO-FRL) were collected fragments were amplified from the genomic DNA and cloned into by flow-sorting (Fig. 2B) and processed for two more rounds of pCDNA3.1(þ) between EcoRI and XhoI for sequencing. Primers expanding-sorting to enrich for the top 10% of RFP-expressing used for antibody gene PCR were as follows: cells. The fluorescence intensity order was CHO-FRL-S2 > CHO- T7-P1: 50-TAATACGACTCACTATAGGGAGACCCAAGC-30 FRL-S1 > CHO-FRL (Fig. 2C), indicating a significant enrichment TM-P2: 50-CTAACGTGGCTTCTTCTGCCAAAGC-30 of the two rounds of sorting.

42 Biotechnology and Bioengineering, Vol. 113, No. 1, January, 2016 Figure 1. Vectors used in RMCE and an overview of RMCE. (A) In pCDFGL, a GFP gene (flanked by recombination targets FRT and loxP) are expressed from the CMV promoter, and a dihydrofolate reductase (DHFR) expression cassette is inserted in front of the promoter. (B) In pF2AC, the coding sequences of Flp and Cre are separated by the 2A self- cleaving peptide TaV (refer to Materials and Methods), and expressed from the CMV promoter. (C) In pFRL, pFPL, and pFAbL exchange vectors, RFP, Puro, and anti-TNF-Ab genes are flanked by FRT and loxP sequences for gene exchange and without promoters in front of these genes. (D) An overview of three rounds of RMCE to acquire antibody displaying cells. In RMCE 1, the GFP gene integrated in chromosome(s) were first replaced by RFP cassette, and cells with only RFP fluorescence were sorted and collected; in RMCE 2, the RFP gene were replaced by Puro gene, and cell clones with a single copy Puro expression cassette were generated; in RMCE 3, cells with anti-TNF-a antibody displayed on the surface were generated by exchanging Puro with an anti-TNF antibody gene.

The CHO-FRL-S2 cells were expanded and co-transfected with PuroR-14 cells have two recombination recognition sequences (FRT pFPL (Fig. 1C) and pF2AC vectors to replace the RFP gene with the and LoxP) while Flp-In CHO cells have only one recombination PuroR gene as shown in Figure 1D (RMCE 2). The transfectants recognition sequence (FRT). In order to comprehensively compare were maintained in medium containing 10 mg/mL puromycin for our system with the Flp-In system, two comparisons have been 10 days. Cells that failed to replace the RFP gene with the PuroR made. gene (without puromycin resistance protein expression) were killed We first made the following comparison: PuroR-14 cells were by puromycin and cells that survived and were non-fluorescent were transfected with pFGL (Fig. S1A) and pF2AC to generate FGL-14 those in which RPF was replaced with PuroR gene (Fig. 3A). Sorting cells, the Flp-In cells were transfected with pFGFP (Fig. S1A) and of individual non-fluorescent cells into each well of 96-well plates, the recombinase-expressing vector pOG44. The transfected Flp-In individual cells were then expanded into cell clones. Two clones cells were then maintained in medium with hygromycin to generate (PuroR-12 and PuroR-14) were chosen to detect the copy number of Flp-In-GFP cells. On the eighth day after transfection, the average the integrating plasmid by Southern blotting. The parent CHO/ GFP fluorescence intensity (gate K) of Flp-In-GFP cells was 31% dhFr- cells were chosen for analysis as a negative control. Both higher than FGL-14 cells (Fig. S1B). However, our cells do not need PuroR-12 and PuroR-14 cell clones had one hybridizing DNA band to be cultured in medium containing antibiotics to get rid of the corresponding to a single copy of the integrated plasmid. The sizes cells in which the antibiotic resistant gene is not integrated in of the two DNA bands are the same suggesting that the two cell chromosome(s). Our cells just need 2 days after transfection before clones are most likely derived from a single cell (Fig. 3B). affinity maturation, and Flp-In cells need a much longer time (for The PuroR-14 clone was chosen to compare with Life example, 8 days) before affinity maturation. In addition, Flp-In- Technologies’s Flp-In CHO cell line and also for further studies. GFP cells can also include cells in which multiple expression The Flp-In system is also a convenient system to use to introduce a cassettes are integrated in different chromosome locations since the gene of interest into a designated chromosome site by Flp-In plasmid pFGFP used for recombination contains its own recombination for potential antibody affinity maturation. Our promoter for GFP transcription (Fig. S1A) even without system and the Flp-In system are different not only in the recombination into a specific site. Our cassette vector does not engineered cells, but also in the would-be transfected vectors. contain a promoter for GFP (Fig. S1A), so, the fluorescent FGL-14

Chen et al.: Antibody Affinity Maturation in CHO Cells 43

Biotechnology and Bioengineering Figure 2. Generate high-level RFP expression cells. (A) Enrichment of GFP expression cells. pCDFGL transfected CHO/DHFR- cells were cultured 15 days for plasmid randomly inserted into the genomes and 10% of the cells with highest GFP fluorescence intensity (CHO-FGL) were sorted by a FACSAriaIII flow cytometer. (B) RMCE 1: Cells with RFP expression were detected and the cells (0.3%) with only RFP fluorescence (CHO-FRL) were sorted and collected. (C) Two more rounds of cell sorting to generate the highest RFP expression cells. The parent CHO/dhFr- cells were used as a negative control (CHO-CTL). The CHO-FRL cells were generated at the stage RMCE 1 (refer to Fig. 1D). CHO-FRL-S1 and CHO-FRL-S2 cells were generated from the first and the second rounds of cell sorting enrichment, respectively.

cells contain only one expression cassette. The efficiency of affinity PE-positive cells should be those in which the PuroR gene was maturation using cells bearing multiple antibody genes of interest replaced by anti-TNF-Ab (gate H). will be lower than cells containing a single antibody gene of interest. To make a direct comparison, we created a vector (pFRpA, Quick Establishment of a CHO Cell Population Displaying Fig. S2A) similar to our non-promoter vector (pFRL) for the Flp-In High-Level Antibody system with no promoter in front of the RFP gene. The introduced RFP gene is most unlikely to express if the gene is integrated in It is important to establish a cell population in which the other nonspecific chromosome locations. The pFRL plasmid was antibody gene of interest is highly transcripted as the rate of AID- transfected into our PuroR-14 cells and pFRpA into the Flp-In cells. induced mutation of the antibody gene is positively proportional The average RFP fluorescence (gate P) in our cells (FRL-14) was to the transcription level (Bachl et al., 2001). Considering that the more than 4 times that in their cells (Flp-In-RFP) after 2 days of integrated site of PuroR-14 was derived from multiple rounds of culture in medium containing no antibiotic (data not shown). As sorting without antibiotic pressure and acquired stable the culture time was prolonged, the RFP fluorescence in FRL-14 expression of a gene inserted in the cassette, we used a strategy cells became even higher than that in Flp-In-RFP cells (over 10 to mature antibody affinity without using antibiotics to maintain times on the 5th day after transfection [Fig. S2B]). Therefore, our antibody display. Vectors pCDNA3.1 and episomal vector pCEP4 system is superior to the Flp-In system for antibody affinity have been used to display antibodies by several research groups maturation. (Bowers et al., 2011; Chen et al., 2012). Here, we cloned the anti- þ To replace the PuroR gene with the anti-TNF-Ab gene, PuroR- TNF-Ab gene into the pCDNA3.1-hygro (pCDNA3.1-Ab) and 14 cells were co-transfected with the antibody-expressing pFAbL pCEP4 (pCEP4-Ab) vectors (refer to Materials and Methods) to plasmid (Fig. 1C) and plasmid pF2AC. The anti-TNF-Ab gene in compare their antibody display stabilities with that from our the vector possessed a signal peptide (SP), an anti-TNF single pFAbL vector. pFAbL was transfected together with pF2AC into chain antibody with a HA tag and a transmembrane domain PuroR-14 cells, and pCDNA3.1-Ab and pCEP4-Ab were separately (TM) so the antibody could be displayed on the cell surface (Chen transfected into CHO/dhFr- cells with pF2AC. After 2 days of et al., 2012), and monitored by anti-HA antibody as well as GFP- culture, cells that displayed the highest levels of the antibody TNF-a. The transfected cells were labeled with PE-conjugated were sorted by a flow cytometer (Fig. 4A, gate M) and then anti-HA antibody and assessed by flow cytometry (Fig. 3C). The cultured in medium for 7 days without antibiotic. The cells

44 Biotechnology and Bioengineering, Vol. 113, No. 1, January, 2016 Figure 3. RMCE 2 and RMCE 3. (A) Cells derived from RMCE 2. A significant population of non-fluorescent cells was detected by flow cytometry. The successfully targeted non- fluorescent cells (gate L) were sorted and expanded into cell clones. (B) Southern blot analysis of puromycin resistant cell clones. The Puro-12, Puro-14 clones were derived from RMCE2, and the parent CHO/dhFr- cells were used as a negative control. (C) RMCE 3. Transfection of the Puro-14 cells led to the replacement of puromycin-resistant gene with anti- TNA-a antibody gene. And the antibody was successfully displayed on the cell surface (gate H).

containing pFAbL, pCDNA3.1-Ab, and pCEP4-Ab were named as Affinity Maturation of Anti-TNF-a Antibody RMCE-Ab, Random-Ab, and Episomal-Ab. On day 7, 98.2% RMCE-Ab cells remained positive for antibody display, and the Antibody affinity maturation was carried out with five rounds of display level was nearly unchanged compared to that immediately cell proliferation-flow sorting. For each round of maturation, after flow cytometric sorting, while Random-Ab and Episomal- RMCE-Ab cells were transfected with the AID-expressing plasmid Ab cells were only 1.2% and 2.1%, respectively (Fig. 4B). pCEP-Neo-AID, proliferated (antibody gene mutations were Therefore, without antibiotic pressure, we were able to quickly accumulated during the same period) in neomycin-containing establish a stable, high-level antibody displaying cell population medium (to keep the AID-expressing plasmid in cells) for 7 days, by RMCE, while Random-Ab and Episomal-Ab cells lost almost labeled with PE-conjugated anti-HA antibody (to monitor all the antibody-positive cells during the same period. In order to antibody display levels on cells), and GFP-TNF-a antigen (to obtain cells stably displaying antibodies on the cell surface for monitor the TNF-a-binding ability displayed on the cell surface), affinity maturation, more than 20 days were needed for and the labeled cells with the highest GFP/PE fluorescence ratio pCDNA3.1-Ab vectors to be randomly integrated into the cell (the top 0.02%, around 20,000 cells) were collected by flow genome or pCEP4-Ab vectors to be stably maintained in cells sorting (Fig. 5A). The collected cells were grown for 3 days, and (our unpublished data). the cell number reached around 200,000. The cells were then

Chen et al.: Antibody Affinity Maturation in CHO Cells 45

Biotechnology and Bioengineering Figure 4. Quick establishement of antibody displayed cells. (A) Antibody display levels of cells bearing various vectors were detected 2 days after transfection, and the antibody displayed cells were collected by cell sorting (gate M). (B) Antibody display stabilities of the cells bearing three different vectors, respectively. Antibody display levels were detected by ﬂow cytometry after 7 days of expansion.

transfected with the AID-expressing plasmid for the next round located in the positive display zones (upper left and upper right; of maturation. After each round of sorting, the antibody genes Fig. 5B and C). This feature should contribute to a high mutation were cloned, and 35 clones were sequenced. Overall, a total of 22 rate of the antibody gene, thus, a high efficiency of affinity unique mutations or mutation combinations were obtained from maturation. the five rounds of maturation (Table II). All the antibody genes Three mutations N33K, A145V, and M211I were most frequently containing unique mutations or mutation combinations were detected, and appeared 73, 109, and 14 times, respectively, in all the analyzed with flow cytometry for their abilities to bind GFP- sequenced clones (Table II). N33K and A145V were first detected in TNF-a. Four mutant antibody clones failed to display for an round 2, while the M211I mutation was first detected in round 4. unknown reason. The other 18 antibody clones were successfully After five rounds of maturation, all the clones contained N33K and displayed, and all these mutant antibodies demonstrated stronger A145V mutations, and 11 out of 34 clones possessed all the three capabilities to bind GFP-TNF-a than that of the wild type mutations (Table II and Fig. 6). Two single mutants N33K and antibody. That is, the GFP/PE fluorescence ratios of the cells A145Vappeared multiple times after two rounds of maturation; the displaying these antibodies were larger than that of the parent mutation frequencies were 3/33 and 21/33, respectively, right after antibody (Fig. 5D and data not shown). Seven most frequently the second round of sorting. Interestingly, the higher frequency of appeared clones in Figure 5D were further analyzed by SPR for A145V corresponds to the smaller Kd value (2.12E-09), while the their dissociation coefficients (Kd), confirming their improved lower frequency of N33K corresponds to the larger Kd value affinities (Table III). (4.34E-09). The maturation progress was accompanied with the In the last section, we described that a high-level antibody- replacement of the single mutation clones with the clones displaying cell population was able to be quickly established using containing two or three mutations that conferred even smaller a puromycin resistance cell line by using the RMCE method. Here, Kd values (Table III). These data indicate that the maturation we find that the high level of antibody display was very stable procedure is sensitive enough to distinguish a small difference in during the five rounds of maturation; the majority of cells were antibody affinity and is efficient to accumulate clones with higher

46 Biotechnology and Bioengineering, Vol. 113, No. 1, January, 2016 Figure 5. Affinity maturation. (A) The cells with the highest antigen binding ability (the top 0.02%) were collected by sorting. Only the first round of sorting is shown here. (B) The antibody display levels and their antigen-binding abilities of the sorted cells. The sorted cells from (A) were expanded for 8 days and then stored. The flow cytometric measurements were performed after re-inoculation and 2 days’ culture of these clones stored in a tank. (C) The percentage of cells positively displaying antibody. (D) Analyses of the antibody display levels and their antigen-binding abilities. The assays were made on cells transiently transfected with designated mutant antibody genes, respectively.

Chen et al.: Antibody Affinity Maturation in CHO Cells 47

Biotechnology and Bioengineering Table II. Mutations observed during the afﬁnity maturation procedure

Rounds of evolution Rounds of evolution

Mutation R1 R2 R3 R4 R5 Mutation R1 R2 R3 R4 R5 N33K 3/33 3/34 5/34 N33K þ H212N 1/34 A130T 1/34 A135T þA145V 1/33 A145V 21/33 20/34 6/34 A145V þ M211I 2/34 P229T 1/34 A145V þA262V 1/33 A273T 1/34 A145L þ A229V 3/34 1/34 A145L 1/33 N33K þG136R þ D248G 1/34 N33K þ L49V 1/33 N33KA145V þA199V 1/34 N33K þ A130V 1/34 N33K þ A145V þ Y209D 2/34 N33K þG136R 1/34 N33K þ A145V þ M211I 11/34 N33K þ A145V 2/33 2/34 15/34 23/34 L49V þA62T þA145V 1/34 N33K þC197R 1/34 L49V þA145V þ M211I 1/34

Table III. Afﬁnities of different mutations

Clone Mutation KD (M) Koff (1/s) Kon (1/MS) WT No 7.82E-09 5.29E-04 6.76E þ 04 S1–26 A229T 5.69E-09 4.07E-04 7.15E þ 04 S2–17 N33K 4.34E-09 2.58E-04 5.96E þ 04 S3–19 A145V 2.12E-09 1.57E-04 7.40E þ 04 S2–23 N33K þ A145V 1.25E-09 9.78E-05 7.81E þ 04 S3–18 A145V þ M211I 1.32E-09 1.04E-04 7.87E þ 04 S4–10 N33K þ H212N 2.88E-09 1.66E-04 5.77E þ 04 S5–8 N33K þ A145V þ M211I 5.20E-10 3.94E-05 7.84E þ 04 affinities. Ten days were used for each round of maturation, and Discussion fi only 2 months were needed for ve rounds of maturation with 15- fi fold affinity improvement (between the clone bearing N33K þ AID-induced SHM has been used for antibody af nity maturation A145V þ M211I and the parent clone). in several cell lines such as H1299 (Chen et al., 2012) and 293T cells (Bowers et al., 2011; McConnell et al., 2012). However, they are time-consuming and inefficient. In this study, we combined recombinase-mediated cassette exchange with somatic hypermutation for antibody affinity maturation in CHO cells. This system is simple to perform, highly efficient, and successfully matured the affinity of an anti-TNF-a antibody by 15-fold within 2 months. CHO cells are fast-growing (a doubling time of 16 h), stress- resistant and capable of high-level antibody expression (Lai et al., 2013; Qiao et al., 2009; Ye et al., 2009). Very efficient transcription of an antibody gene is a prerequisite for large scale production and purificationofanantibody(Kimetal.,2012;Laietal.,2013), CHO cells can express an antibody at a level of several gram/L. The rate of transcription is positively proportional to the mutation efficiency of the gene of interest (Bachl et al., 2001), thus, a high-level of antibody gene transcription in CHO cells can lead to a high-level mutation rate of an antibody gene and a fast affinity maturation process as shown in this study (Table III). One of the reasons that CHO cells are the most frequently used cells Figure 6. Accumulation of three most frequently detected mutations through for therapeutic antibody production in industry is because CHO affinity maturation. After each round of affinity maturation, antibody genes were cloned cells produce antibodies with glycosylations identical or similar to and detected. Antibody clones containing single mutation (N33K, A145V, or M211I) or those of human antibodies which are non-antigenic or mildly mutation combinations (N33K þ 145V or N33K þ A145V þ M211I) were plotted against the number of rounds of affinity maturation. antigenic when injected into patients (Ghaderi et al., 2012; Lai et al., 2013). Therefore, this CHO affinity maturation system will

48 Biotechnology and Bioengineering, Vol. 113, No. 1, January, 2016 be very beneficial for the affinity maturation of therapeutic et al., 2012; Ho et al., 2006) or episomal vectors (Bowers et al., antibodies. 2011; McConnell et al., 2012) to generate antibody-displayed cells To establish a cell clone such as PuroR-14 in this study to be for affinity maturation, thus, several antibody genes may express conveniently used for affinity maturation (only a simple in a single cell (Baer et al., 2000; Horlick et al., 2000). Different transfection to replace the PuroR with the antibody gene of mutant and wild type antibodies may be displayed on the surface interest before affinity maturation), we used a three-step of a single cell, thus, affecting the efficiency to sort and collect the strategy:(1) establishment of a CHO cell population bearing a best antibody mutants. In our platform, only one copy of an GFP-expressing cassette(s) in chromosome(s); (2) derivation of antibody gene was exchanged into the expression cassette a much smaller CHO cell population (from the first population) (Fig. 3B), and only one kind of antibody mutation was displayed containing chromosome-integrated cassette(s) that are able to on the surface of a single cell for sorting and isolating the best stably express RFP at high levels; and (3) derivation of a CHO antibody mutants efficiently. Actually, after performing the cell population mostly containing a single PuroR-expressing transfection of the antibody-expressing plasmid pFAbL into the cassette from the second population. At the first step, we parent cell CHO/dhFr- of PuroR-14, no cells that positively intended to collect cells bearing chromosome-integrated GFP- expressed antibody were detected (our data unpublished), expressing cassette(s) without antibiotic pressure by sorting the indicating that all the antibody expression in our cells (RMCE- top 10% of GFP-expressing cells (Fig. 2A). Some of these cells Ab) was from the single chromosome site. Second, our platform is contained multiple integrated GFP-expressing cassettes, some very convenient to generate stable antibody-displayed cells contained crippled GFP-expressing cassette(s) which were (2 days), while other platforms need a longer time to generate unable to be replaced by an antibody gene with recombination, stable cell lines before the antibody affinity maturation can be and some others were less stable and could lose the cassette(s) started (Fig. 4). Third, during the affinity maturation process, or express GFP at low levels after a long-period of culture. To very few cells in our system lose the antibody-display ability solve or mitigate these problems, we took the following step (the (Fig. 5B and C) while previously reported random-integrated or second step): we replaced the GFP gene with the RFP gene by episomal systems have a significant portion of cells with negative RMCE (Fig. 2B), then carried out three consecutive rounds of or reduced antibody-display through the affinity maturation culturing-sorting without antibiotic pressure (Fig. 2B and C). In process (Bowers et al., 2011; Chen et al., 2012). Fourth, CHO cells the last two sortings, we collected only the brightest top 10% of are better suited, because they are fast-growing (doubling time: RFP-expressing cells, the cells stably expressing RFP at high 16 h) and are in general more stress-resistant. The commonly levels (Fig. 2C). Two rounds of RMCE and multiple rounds of used human cells such as 293 and H1299 grow significantly culturing-sorting have also been used by other groups to create slower (doubling time: 21–23 h) and are more sensitive to stable cell lines that are able to express a gene of interest (GOI) transfection, thus, they have a lower survival rate after at high levels (Qiao et al., 2009). Since we only sorted and transfection than CHO. The fast-proliferative and stress-resistant collected the cells that were simultaneously RFP-positive and features allow for quick expansion in cell number with a minimal GFP-negative, most of these cells, if not all, contained the RFP cell loss from the stresses of flow sorting, transfection, and high- gene that was replaceable with an antibody gene of interest. At level antibody display. Our system just needs 10 days to conduct the third step, we intended to replace the RFP gene with a non- one round of affinity maturation (sorting and proliferation) and fluorescent protein gene (Fig. 3A). RFP would interfere with less than 2 months to perform five rounds of affinity maturation dual-color flow cytometric sortings during the affinity matura- and to enhance the affinity of an antibody by 15-fold (Table III). tion, so the RFP gene was replaced at this stage. Also, the Five rounds of affinity maturation using 293T or H1299 human addition of more rounds of replacement-sorting would also cells will take 3.5 months. The above described merits of our reduce the cells containing more than one integrated cassette system were demonstrated by the fact that affinity improvement and enrich those that express a gene of interest stably and at was detected even with only one round of affinity maturation and high levels. In fact, we randomly picked and tested two PuroR- accumulated with each additional round of affinity maturation expressing clones (PuroR-12 and PuroR-14) and found that they (Figs. 5B and 6). each contained a single expression cassette (Fig. 3B). In contrast CHO cells are most widely used in therapeutic antibody to the above described lengthy procedure in creating the cell production because the post-translational modifications on clones PuroR-12 and PuroR-14, the cell clones are very antibodies in CHO cells are the same or similar to those of convenient to use; it only takes a simple transfection and antibodies in the human body, and CHO cells can express 2 days of culture for antibody affinity maturation. In this study, antibody at very high levels (Schmidt, 2004; Walsh and Jefferis, we did not try to generate a retargetable CHO cell line that can 2006). Our affinity maturation system based on CHO cells is display an antibody at the cell surface at an even higher level. It preferable because the mutated genes with high affinities will be worthwhile to find out how high the display level can generated in CHO cells are most likely to express efficiently in reach and how much more efficient the affinity maturation in CHO cells for industrial production. Different cell systems prefer CHO cells will be. to express antibody genes bearing their corresponding mutations Our system has several advantages over the currently used (Persic et al., 1997; Vendel et al., 2012), thus, the antibody genes mammalian cell display systems (Akamatsu et al., 2007; Bowers generated through our maturation system is most likely to meet et al., 2011; Chen et al., 2012; Higuchi et al., 1997; Ho et al., the requirement for the high level production in CHO cells at the 2006). First, other systems usually use random integration (Chen later stage of development.

Chen et al.: Antibody Affinity Maturation in CHO Cells 49

Biotechnology and Bioengineering The PuroR-14 cell line developed in this study can confer Francisco JA, Georgiou G. 1994. The expression of recombinant proteins on the convenient RMCE for any antibody gene of interest for affinity external surface of Escherichia coli. Biotechnological applications. Ann N Y – maturation, and the system developed in this study is highly Acad Sci 745:372 382. Ghaderi D, Zhang M, Hurtado-Ziola N, Varki A. 2012. Production platforms for sensitive to distinguish a small difference in antibody affinity. For fi biotherapeutic glycoproteins. Occurrence, impact, and challenges of non- instance, the af nity difference of N33K over A145V is about human sialylation. Biotechnol Genet Eng Rev 28(1):147–175. twofold, but the mutation accumulation rate of A145V is 7 times Higuchi K, Araki T, Matsuzaki O, Sato A, Kanno K, Kitaguchi N, Ito H. 1997. Cell that of N33K during the first two rounds of maturation display library for gene cloning of variable regions of human antibodies to hepatitis B surface antigen. J Immunol Methods 202(2):193–204. (Table III). The highly distinguishing ability in combination with fi fi Ho M, Nagata S, Pastan I. 2006. Isolation of anti-CD22 Fv with high af nitybyFv the high mutation ef ciency of this system led to the fast display on human cells. Proc Natl Acad Sci U S A 103(25):9637–9642. accumulation of different mutations beneficial for affinity Horlick RA, Schilling AE, Samama P, Swanson RN, Fitzpatrick VD, Robbins AK, improvement (Table III). Damaj B. 2000. Combinatorial gene expression using multiple episomal vectors. Gene 243(1–2):187–194. fi This work was supported by National Basic Research Program of China Huang Y, Li Y, Wang YG, Gu X, Wang Y, Shen BF. 2007. An ef cient and targeted gene (2011CBA00906 and 2014CB910402), National Special Projects for Key integration system for high-level antibody expression. J Immunol Methods – – Scientific Instruments and Equipment Development of Ministry of Science 322(1 2):28 39. and Technology (2011YQ03013404), and National Science and Technology Huse WD, Stinchcombe TJ, Glaser SM, Starr L, MacLean M, Hellstrom KE, Hellstrom fi Major Project (2013ZX10004611-002 and 2013ZX10003006-001-001). It was I, Yelton DE. 1992. Application of a lamentous phage pVIII fusion protein fi also funded by State Key Laboratory of Microbial Resources, Institute of system suitable for ef cient production, screening, and mutagenesis of F(ab) – Microbiology, Chinese Academy of Sciences (SKLMR-20130605). We are antibody fragments. J Immunol 149(12):3914 3920. grateful to Kevin Hopkins from Columbia University at New York for revising Kim JY, Kim YG, Lee GM. 2012. CHO cells in biotechnology for production of this manuscript. recombinant proteins: Current state and further potential. Appl Microbiol Biotechnol 93(3):917–930. Kim MS, Lee GM. 2008. Use of Flp-mediated cassette exchange in the development of References a CHO cell line stably producing erythropoietin. J Microbiol Biotechnol 18(7):1342–1351. Akamatsu Y,Pakabunto K, Xu ZH, Zhang Y, Tsurushita N. 2007. Whole IgG surface Lai T,YangY,Ng SK. 2013. Advances in mammalian cell line development technologies display on mammalian cells: Application to isolation of neutralizing chicken for recombinant protein production. Pharmaceuticals (Basel) 6(5):579–603. monoclonal anti-IL-12 antibodies. J Immunol Methods 327(1–2):40–52. Martin A, Scharff MD. 2002. Somatic hypermutation of the AID transgene in B and Anderson RP, Voziyanova E, Voziyanov Y. 2012. Flp and Cre expressed from Flp-2A- non-B cells. Proc Natl Acad Sci U S A 99(19):12304–12308. Cre and Flp-IRES-Cre transcription units mediate the highest level of dual Maul RW, Gearhart PJ. 2010. AID and Somatic Hypermutation. Advances in recombinase-mediated cassette exchange. Nucleic Acids Res 40(8). Immunology 105(105):159–191. Arakawa H, Kudo H, Batrak V, Caldwell RB, Rieger MA, Ellwart JW, Buerstedde JM. Mazor Y, Van Blarcom T, Iverson BL, Georgiou G. 2009. Isolation of full-length IgG 2008. Protein evolution by hypermutation and selection in the B cell line DT40. antibodies from combinatorial libraries expressed in Escherichia coli. Methods Nucleic Acids Res 36(1). Mol Biol 525:217–239, xiv. Bachl J, Carlson C, Gray-Schopfer V, Dessing M, Olsson C. 2001. Increased McConnellAD,DoM,NebenTY,SpasojevicV,MacLarenJ,ChenAP,AltobellL, transcription levels induce higher mutation rates in a hypermutating cell line. Macomber JL, Berkebile AD, Horlick RA and others. 2012. High affinity J Immunol 166(8):5051–5057. humanized antibodies without making hybridomas; immunization paired Baer A, Schubeler D, Bode J. 2000. Transcriptional properties of genomic transgene with mammalian cell display and in vitro somatic hypermutation. PLoS integration sites marked by electroporation or retroviral infection. Biochemistry ONE 7(11). 39(24):7041–7049. Persic L, Righi M, Roberts A, Hoogenboom HR, Cattaneo A, Bradbury A. 1997. Boder ET, Wittrup KD. 1997. Yeast surface display for screening combinatorial Targeting vectors for intracellular immunisation. Gene 187(1):1–8. polypeptide libraries. Nat Biotechnol 15(6):553–557. Qiao JH, Oumard A, Wegloehner W, Bode J. 2009. Novel tag-and-exchange (RMCE) Bowers PM, Horlick RA, Neben TY, ToobianRM, Tomlinson GL, Dalton JL, Jones HA, strategies generate master cell clones with predictable and stable transgene Chen A, Altobell L, Zhang X and others. 2011. Coupling mammalian cell surface expression properties. J Mol Biol 390(4):579–594. display with somatic hypermutation for the discovery and maturation of human Qiu JK, Jung ST, Georgiou G, Hang HY. 2010. Enrichment of Escherichia coli antibodies. Proc Natl Acad Sci U S A 108(51):20455–20460. spheroplasts displaying scFv antibodies specific for antigens expressed on the Campbell RE, Tour O, Palmer AE, Steinbach PA, Baird GS, Zacharias DA, Tsien RY. human cell surface. Appl Microbiol Biotechnol 88(6):1385–1391. 2002. A monomeric red fluorescent protein. Proc Natl Acad Sci U S A Raymond CS, Soriano P. 2007. High-efficiency FLP and Phi C31 site-specific 99(12):7877–7882. recombination in mammalian cells. PLoS ONE 2(1). Chen SP,Qiu JK, Chen C, Liu CC, Liu YH, An LL, Jia JY, Tang J, Wu LJ, Hang HY. 2012. Schmidt FR. 2004. Recombinant expression systems in the pharmaceutical industry. Affinity maturation of anti-TNF-alpha scFv with somatic hypermutation in Appl Microbiol Biotechnol 65(4):363–372. non-B cells. Protein Cell 3(6):460–469. Seo H, Hashimoto SI, Tsuchiya K, Lin W, Shibata T, Ohta K. 2006. An ex vivo method Cumbers SJ, Williams GT, Davies SL, Grenfell RL, Takeda S, Batista FD, Sale JE, for rapid generation of monoclonal antibodies (ADLib system). Nat Protoc Neuberger MS. 2002. Generation and iterative affinity maturation of 1(3):1502–1506. antibodies in vitro using hypermutating B-cell lines. Nat Biotechnol Seo H, Masuoka M, Murofushi H, Takeda S, Shibata T, Ohta K. 2005. Rapid 20(11):1129–1134. generation of specific antibodies by enhanced homologous recombination. Nat de Bruin R, Spelt K, Mol J, Koes R, Quattrocchio F. 1999. Selection of high-affinity Biotechnol 23(6):731–735. phage antibodies from phage display libraries. Nat Biotechnol 17(4):397–399. Shimshek DR, Kim J, Hubner MR, Spergel DJ, Buchholz F, Casanova E, Stewart AF, Donnelly MLL, Hughes LE, Luke G, Mendoza H, ten Dam, Gani E, Ryan D. 2001. The Seeburg PH, Sprengel R. 2002. Codon-improved Cre recombinase (iCre) ’cleavage’ activities of foot-and-mouth disease virus 2A site-directed mutants expression in the mouse. Genesis 32(1):19–26. and naturally occurring ’2A-like’ sequences. J Gen Virol 82:1027–1041. Shivarov V, Shinkura R, Honjo T. 2008. Dissociation of in vitro DNA deamination Feldhaus MJ, Siegel RW, Opresko LK, Coleman JR, Feldhaus JM, Yeung YA, Cochran activity and physiological functions of AID mutants. Proc Natl Acad Sci U S A JR, Heinzelman P, Colby D, Swers J and others. 2003. Flow-cytometric isolation 105(41):15866–15871. of human antibodies from a nonimmune Saccharomyces cerevisiae surface Smith GP. 1985. Filamentous fusion phage: Novel expression vectors that display display library. Nat Biotechnol 21(2):163–170. cloned antigens on the virion surface. Science 228(4705):1315–1317.

50 Biotechnology and Bioengineering, Vol. 113, No. 1, January, 2016 Southern PJ, Berg P. 1982. Transformation of mammalian cells to antibiotic Wolkowicz R, Jager GC, Nolan GP. 2005. A random peptide library fused to CCR5 for resistance with a bacterial gene under control of the SV40 early region promoter. selection of mimetopes expressed on the mammalian cell surface via retroviral J Mol Appl Genet 1(4):327–341. vectors. J Biol Chem 280(15):15195–15201. Todo K, Miyake K, Magari M, Kanayama N, Ohmori H. 2006. Novel in vitro screening YeJX,KoberV,TellersM,NajiZ,SalmonP,MarkusenJF.2009.High-levelprotein system for monoclonal antibodies using hypermutating chicken B cell library. J expression in scalable CHO transient transfection. Biotech Bioeng 103(3):542–551. Biosci Bioeng 102(5):478–481. Yoshikawa K, Okazaki IM, Eto T, Kinoshita K, Muramatsu M, Nagaoka H, Honjo T. Vendel MC, Favis M, Snyder WB, Huang F, Capili AD, Dong JY, Glaser SM, Miller BR, 2002. AID enzyme-induced hypermutation in an actively transcribed gene in Demarest SJ. 2012. Secretion from bacterial versus mammalian cells yields a fibroblasts. Science 296(5575):2033–2036. recombinant scFv with variable folding properties. Arch Biochem Biophys 526(2):188–193. Supporting Information Walsh G, Jefferis R. 2006. Post-translational modifications in the context of therapeutic proteins. Nat Biotechnol 24(10):1241–1252. Winter G, Griffiths AD, Hawkins RE, Hoogenboom HR. 1994. Making antibodies by Additional supporting information may be found in the online phage display technology. Annu Rev Immunol 12:433–455. version of this article at the publisher’s web-site.

Chen et al.: Antibody Affinity Maturation in CHO Cells 51

Biotechnology and Bioengineering