PEDS Advance Access published September 14, 2014 Protein Engineering, Design & Selection pp. 1 –7, 2014 doi:10.1093/protein/gzu039
Antibody-membrane switch (AMS) technology for facile cell line development Bo Yu1,2,3, John M.Wages2 and James W.Larrick1,2 1
Larix Bioscience, LLC, 1230 Bordeaux Drive, Sunnyvale, CA 94089, USA and 2Panorama Research, Inc., Sunnyvale, CA 94089, USA
3
To whom correspondence should be addressed. E-mail: [email protected] Received June 12, 2014; revised August 3, 2014; accepted August 12, 2014 Edited by James S. Huston
Introduction Since the first regulatory approval in 1986, therapeutic monoclonal antibodies (mAbs) have become the fastest growing class of pharmaceutical drugs (Scott, 2012). Development of novel classes of antibody derivatives (Fc-modified mAb, bispecific mAb and antibody –drug conjugates) promises even more activity and specificity in third-generation therapeutic mAbs. These magic bullets will revolutionize pharmacotherapy of cancer and multiple other disease areas. Development of a therapeutic antibody is a lengthy, expensive process, often taking 10 – 15 years including antibody discovery/engineering, production cell line development, manufacturing process development and clinical studies
# The Author 2014. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: [email protected]
Page 1 of 7
Downloaded from http://peds.oxfordjournals.org/ at University of Otago on October 4, 2014
Generation of high-productivity cell lines remains a major bottleneck in therapeutic antibody development. Conventional cell line development often depends on gene amplification methodologies using dihydrofolate reductase or glutamine synthetase. Higher productivity is associated with an increased gene copy number. However, lack of selection pressure under the conditions of large-scale manufacturing leads to clonal instability. We have developed a novel method for cell line development, antibodymembrane switch (AMS) technology, that does not rely on gene amplification. This fluorescence-activated cell sorting (FACS)-based, high-throughput method is facilitated by cell-surface antibody expression to rapidly and efficiently isolate high-producing cells. The switch between membrane expression and secretion is achieved by alternative splicing and specific DNA recombination. The antibody of interest is initially displayed on the cell surface to facilitate FACS. Isolated high-producing cells are then seamlessly transformed into production cells after removing the membrane-anchoring domain sequence with a DNA recombinase. AMS technology has been applied in a number of antibody cell line development projects, which typically last 2 – 3 months. The top production cell lines exhibit very high specific productivity of 40– 60 pg/cell/day resulting in production titers of 2 – 4 g/l in 10-day batch culture. Keywords: antibody engineering/FACS/monoclonal antibodies/protein engineering
(Agrawal and Bal, 2012). Among these tasks, antibody discovery may take 6 – 18 months, and development of production cell lines may require an additional 10 – 12 months (Jayapal et al., 2007; Li et al., 2010). Production of material for Phase I/II trials is a critical, limiting step in bringing a therapeutic protein to market. Production cell lines must exhibit stable, high-level expression of the antibody that exhibits all of the desired product quality attributes, including protein sequence homogeneity, glycosylation profile, charge variants and aggregation levels. CHO cells are the most popular mammalian cells for production of therapeutic proteins, although other mammalian cells like NS0 or SP2/0 cells have also been used to produce biological therapeutics (Jayapal et al., 2007; Li et al., 2010). Presently, commercial scale manufacturing target yields exceed 2 – 3 g/l. To attain this goal, a cell line producing .20 pg/cell/day is required prior to process development and bioreactor optimization. Because it is rare to have the gene of interest (GOI) integrated into a highly active transcriptional site in the genome of the production cell line, conventional cell line development methodologies often require gene amplification. Historically, dihydrofolate reductase (DHFR) or glutamine synthetase (GS) has been employed to multiply the integrated GOI to enhance expression levels (Agrawal and Bal, 2012). Addition of increasingly high concentrations of methotrexate (MTX) in the DHFR system or methionine sulfoximine (MSX) in the GS system applies selection pressure to amplify antibody genes adjacent to the DHFR or GS. One of the downsides of gene amplification is that cells with multiple copies of the GOI are often unstable, resulting in decreased production yields in the absence of selection pressure. Therefore, multiple production cell lines have to be carried into late-stage process development to ensure at least one stable cell line for largescale manufacturing. The other downside of gene amplification is low throughput resulting from conventional limiting dilution cloning methodology. In a typical cell line screening campaign of 6 – 8 months, 1000 GOI-positive cells from 20– 30 96-well plates are screened for high expression levels, which leads to only dozens of high-productivity clones. In addition, multiple rounds of subcloning are required to achieve clonality. In recent years, several epigenetic approaches have been used to facilitate cell line development: matrix-attachment region or Selexis Genetic Elements (SGE, Selexis), ubiquitous chromatin opening element (UCOE, Millipore) and antirepressor elements (STAR, Crucell). These DNA elements can be incorporated into antibody expression vectors to enhance transcriptional activities, particularly when the expression vector is integrated in an unfavorable locus. Although these methods have been reported to significantly improve production yield for stable pools, their success in screening for top production cell lines has been limited (Browne and Al-Rubeai, 2007; Agrawal and Bal, 2012).
B.Yu et al.
Materials and methods
Cell culture and transfection 293F and CHOS cells (Invitrogen) were maintained in Freestyle 293 expression media and CD FortiCHO media (Invitrogen), respectively.
Cloning of antibody in the AMS expression vector Antibody heavy chain variable sequences were cloned between restriction sites XbaI and NheI in an antibody expression vector containing the constant region, which was under the control of a human EF1a promoter. The vector carries a puromycin resistance gene for stable cell selection and an ampicillin resistance gene for Escherichia coli propagation. Antibody light chain sequences were cloned into a separate mammalian expression vector between restriction sites XbaI and BamHI, under the control of a human EF1a promoter. The light chain expression cassette was then subcloned into the heavy chain expression vector between the restriction sites EcoRV and AscI.
Transfection of 293F cells and CHOS cells Transfection of 293F cells or CHOS cells was carried out with either the Freestyle Max Transfection Reagent or electroporation using the Neon Transfection System (Invitrogen) according to the manufacturer’s suggested protocol. Stable selection of 293F cells or CHOS cells was with 1 or 10 mg/ml of puromycin, respectively.
Determination of antibody production titer Antibody levels in the culture media were determined by dilution enzyme-linked immunosorbent assay (ELISA) in which the antibody was captured with goat anti-human IgG Fc (100 ng/well, Bethyl) and detected with goat anti-human Kappa antibody HRP conjugate (1 : 10 000 dilution, Bethyl). Human IgG antibody (2 mg/ml, Sigma) was used as the standard for IgG quantitation.
Fig. 1. AMS technology. Membrane display of antibodies is achieved by alternative splicing of a modified CH3 exon fused with a MAD. High-expressing cells are sorted by FACS, and then transformed into production cells by execution of a molecular switch to remove the CH3-MAD exon.
Page 2 of 7
Downloaded from http://peds.oxfordjournals.org/ at University of Otago on October 4, 2014
Because only dozens of high-producing clones are generated in conventional cell line development, productivity is usually the only criterion used for selection until a very late stage, when a handful of clones are assessed for other quality attributes critical for large-scale manufacturing. Thus, activity, glycosylation, aggregation, oxidation, deamination and charge are often not assessed until later stages of development. This screening delay increases project risk and raises complex issues regarding downstream development. We have developed a novel approach based on transient cell-surface expression of the target antibody combined with fluorescence-activated cell sorting (FACS) to address all the shortcomings of conventional cell line development technology. Although several FACS-based screening approaches have been described (Carroll and Al-Rubeai, 2004; Sleiman et al., 2008), none of these has come into widespread use. Antibodymembrane switch (AMS) technology utilizes alternative splicing of the immunoglobulin heavy chain. A membraneanchoring domain (MAD) is introduced in the CH3 exon to generate some of the expressed antibodies on the cell surface (Fig. 1). The antibody of interest is secreted, as well as being anchored on the cell surface. Thus, cells can be readily detected and sorted by FACS. The added CH3-MAD exon resides in the intron region situated between the CH2 and CH3 heavy chain domains, and is flanked by a site-specific DNA recombinase recognition sequence (DRRS) such as a LoxP sequence. The MAD can be a GPI-anchored signal sequence (GASS) or a protein transmembrane domain (TM). After selection for high expression levels of membrane-anchored antibodies by FACS using fluorescence-labeled antigen or secondary antibody, the cells are then switched into production mode by transient expression of an appropriate site-specific DNA recombinase (Turen and Bode, 2011; Wang et al., 2011) such as Cre to remove the MAD exon, resulting in a cell line making only secreted antibodies. AMS streamlines and greatly accelerates the development of productive cell lines for antibody production.
Antibody engineering using antibody-membrane switch
Characterization of cell-surface antibody expression The cells were incubated with goat anti-human Fc antibody fluorescein isothiocyanate (FITC) conjugate (1 : 1000, Bethyl) for 30 min. After washing once with phosphate-buffered saline, cells were subjected to flow cytometric analysis.
Switching off the membrane-anchored antibody using Cre recombinase
Screening of highly productive cell lines Antibody sequences were cloned into the AMS expression vector. The plasmid was linearized by digestion of AscI and was used to transfect 0.1–1 l of CHOS cells (1 106 cells/ml). Stable cells were selected with 10 mg/ml of puromycin for 2 weeks. Stable cells (1 – 10 107) were labeled with goat anti-human Fc antibody FITC conjugate (1 : 1000, Bethyl) before FACS sorting. The 0.01% of the cells with the highest fluorescence signal intensity were sorted into a pool. After 2 days, cells were exposed to Cre recombinase either by transfection or addition of the recombinant protein. After 1 week of culturing, cells were labeled with goat anti-human Fc antibody FITC conjugate (1 : 1000, Bethyl). Cells lacking surface antibody were sorted into 10– 20 96-well plates as single cells. Approximately 200– 500 colonies grew out after 2 – 3 weeks. Culture media were screened by ELISA for expression of the antibody, and top clones were expanded and cryopreserved. Results
LoxP site in the middle of antibody H chain intron 3 does not change expression of rituximab (anti-CD20) Antibody expression vectors for development of production cell lines often utilize the genomic DNA for the IgG heavy chain region consisting of four exons (CH1, hinge, CH2 and CH3) and three introns (Fig. 2A). The rituximab heavy chain variable region was cloned into the antibody expression vectors with the wild-type genomic heavy chain or a modified genomic heavy chain with a LoxP site inserted in the middle of intron 3 (Fig. 2B). Two days after co-transfection of 293F cells with the rituximab light chain expression vector using Freestyle Max, antibody expression levels were measured by ELISA. The cells transfected with the LoxP-modified heavy chain (Fig. 3A-b) produced a similar amount of antibodies as the cells transfected with the wild-type heavy chain (Fig. 3A-a), suggesting that the LoxP sequence inserted in the middle of the intron between CH2 and CH3 did not affect the levels of antibody expression. The heavy chain constant region genes in both transfected cells were amplified by reverse
Antibody secretion by cells transfected with AMS vectors The rituximab heavy chain genomic region was modified by insertion of an extra CH3 exon fused with either a transmembrane domain (TM) or a GPI-linked membrane-anchoring sequence (GASS) flanked by LoxP sites in the middle of intron 3. The TM sequence was taken from the human PDGFRb sequence. The GASS was taken from the human decay accelerating factor sequence. Alternative splicing facilitated by one splice donor at the end of CH2 exon and two possible splice acceptors for the CH3 exons (Fig. 2C) permits simultaneous expression of secreted antibody and membrane-anchored antibody. The LoxP sites permit removal of the extra CH3 exon in intron 3 facilitated by specific DNA recombinase Cre after isolation of the high producer cells (Fig. 2D). After co-transfection (Freestyle Max) with rituximab light chain expression vector in 293F cells, the antibody expression levels in the culture media were determined by ELISA. The heavy chain with the extra CH3-TM exon (Fig. 3A-d) produced 50% as much antibody in the culture media as the wild-type control (Fig. 3A-a), whereas the heavy chain with the extra CH3GASS exon (Fig. 3A-c) secreted 30% more antibody than the wild-type control. This is unexpected, albeit consistent with the fact that GPI-linked membrane anchorage is not 100%. This result has been reproduced in three independent experiments.
Membrane antibody expression of the cells transfected with AMS vectors Although 293 cells often exhibit higher transient transfection efficiency (70–90%) than CHO cells (20–40%) using lipidbased transfection reagent, CHO cells are the industry standard for antibody production. The rituximab expression vectors were transfected into CHOS cells (Freestyle Max). Cell-surface antibody was characterized by cytometric analysis after staining with anti-human Fc antibody FITC conjugate. Cells transfected with the wild-type heavy chain vector (Fig. 3B-a) or LoxP-modified CH3 exon vector (Fig. 3B-b) showed only background levels of cell-surface antibody expression (1–2%), whereas cells transfected with alternatively spliced CH3-GASS vector (Fig. 3B-c) or CH3-TM vector (Fig. 3B-d) exhibited cellsurface antibodies in 27.9 or 20.1% of the cells, respectively. The percentage of cells that expressed membrane antibody was consistent with the transfection efficiency. The alternative splicing (Fig. 2C) was confirmed by RT-PCR and sequencing. The ratio of RT-PCR product incorporating the CH3-TM exon and the wild-type CH3 exon was estimated to be 1 : 1 by the occurrence of either exon during sequencing, representing the ratio between membrane-anchored antibody and secreted antibody.
Switching off the membrane-anchored antibody using Cre recombinase Cre recombinase was presented to the cells by either transient transfection of a Cre expression vector or by addition of membrane-permeable recombinant Cre into the culture medium. A stable pool of CHO cells expressing membrane antibody was isolated by FACS after stable transfection with the AMS antibody expression vector. Most of the cells (80%) exhibited strong expression of cell-surface antibody Page 3 of 7
Downloaded from http://peds.oxfordjournals.org/ at University of Otago on October 4, 2014
A Cre recombinase expression vector was constructed. The Cre cDNA was human codon-optimized and fused with a peptide containing a nuclear localization signal (MPKKKRK) at the N-terminus. Expression of Cre was driven by a human EF1a promoter. Cells with stable expression of membraneanchored antibody were transfected with the Cre expression vector and cultured an additional week before characterization of the cell-surface antibody. Alternatively, cells with stable expression of membrane-anchored antibody were treated with 1 mM of recombinant Cre fused with TAT-NLS for nuclear localization (Excellgen) by adding into the culture media for 2 h before replacing with fresh media. Cell-surface antibody was measured after 1 week of culture.
transcription-polymerase chain reaction (RT-PCR) and sequenced. RNA splicing was found to be identical with or without a LoxP site inserted inside the gamma chain intron.
B.Yu et al.
Downloaded from http://peds.oxfordjournals.org/ at University of Otago on October 4, 2014
Fig. 2. Schematic drawing of the antibody heavy chain constant region of the AMS vectors. (A) Schematic of wild-type IgG1 heavy chain constant region gene. (B) A LoxP site is inserted in the middle of intron 3. (C) Alternative splicing created by an extra membrane-anchored CH3 exon inserted in intron 3. TM refers to a transmembrane domain or a GIP-linked membrane-anchoring sequence. (D) The extra CH3-TM exon is removed by Cre recombinase.
(Fig. 4A). After transient transfection by electroporation with a Cre expression vector, 75% of the cells with membraneanchored antibody lost cell-surface antibody after 1 week in culture (Fig. 4A), consistent with the improved transfection efficiency compared with the lipid-based transfection reagent. A stable CHO cell line isolated by FACS after stable transfection with the AMS antibody expression vector expressed antibody on the cell surface (Fig. 4B). To provide the cells with Cre recombinase, 1 mM of cell membrane-permeable Page 4 of 7
recombinant Cre was added to the culture media for 2 h before switching to fresh media. After culturing for one more week, 78% of the cells lost membrane-anchored antibody (Fig. 4B). The cells lacking membrane antibody were sorted into 96-well plates as single cells and cultured for additional 2 weeks. Clones were confirmed to be completely void of cell-surface antibody (Fig. 4C). The removal of the membrane-anchoring CH3 exon from the chromosome was confirmed by DNA sequencing.
Antibody engineering using antibody-membrane switch
Cell line development using AMS technology
Discussion
Fig. 3. Rituximab expression by AMS vector after transient transfection in 293 cells. (A) Secreted rituximab expression. Antibody concentrations in media 2 days after transient transfection were determined by ELISA. LB-K designates the light chain expression vector, whereas LB-H, LB-1, LB-2 and LB-3 designate the heavy chain expression vectors. (a) LB-H: wild-type heavy chain; (b) LB1: a LoxP site inserted in intron 3; (c) LB2: an extra CH3-GASS exon inserted in intron 3; (d) LB4: an extra CH3-TM exon inserted in intron 3. (B) Cell-surface rituximab expression. 293 cells were stained with anti-human Fc antibody FITC conjugate 2 days after transfection with rituximab expression vectors and subjected to cytometric analysis. (a) wild-type heavy chain; (b) a LoxP site inserted in intron 3; (c) an extra CH3-GASS exon inserted in intron 3; (d) an extra CH3-TM exon inserted in intron 3.
Membrane antibody expression correlates with secreted antibody levels Adalimumab sequences were cloned into the AMS expression vector. After stable transfection of CHO cells, six cell lines with different expression levels were isolated. Transfectants expressed both membrane-anchored and secreted antibodies. Membrane antibody expression was measured by staining with anti-human Fc antibody FITC conjugate and cytometric analysis, and was quantified as the geometric mean of total fluorescence intensity (MFI). Secreted antibody expression was characterized by measuring the production titer in culture media after 7-day batch culture. Membrane antibody expression levels correlate strongly with the levels of antibody secretion (Fig. 5).
We have developed AMS technology for rapid development of production cell lines capable of yielding very high specific productivity of 40– 60 pg/cell/day within 2 – 3 months. The essence of AMS technology is a switch mechanism that turns off expression of a reporter (e.g. a cell-surface antibody), which is a surrogate for protein of interest (POI) expression, after isolation of a subpopulation of cells providing for optimal expression of the POI. The simultaneous expression of secreted antibody and cell-surface antibody is achieved by alternative splicing utilizing an extra CH3 exon fused with an MAD, which is flanked by specific DRRSs. The antibody is cell surface expressed to facilitate high-throughput FACS screening to rapidly isolate the cells producing the highest level of antibody. After screening, the top producing cells are seamlessly transformed into production cells by removing the antibody MAD from the chromosome by targeted DNA recombination. AMS technology relies on high-throughput screening rather than gene amplification to isolate the cells with high productivity. FACS is the most efficient high-throughput screening method at more than 100 million cells per day. Thus, AMS technology begins with transfection of more than 100– 1000 times the number of cells utilized in conventional cell line development. Enrichment of the high-producing cells by FACS ensures capture of the rare clones bearing the GOI integrated into a highly transcriptionally active locus. The isolated cell lines likely do not possess many tandem repeats of the integrated DNA sequences characteristic of gene amplification, consistent with the finding of a much higher ratio of stable cell lines (.80%), compared with the cell lines isolated through gene amplification methods (,50%). A significant value of AMS technology results from the generation of more than hundreds to even thousands of highly productive cell lines. These clones produce up to tens of micrograms of antibody in the 96-well plate stage, which can Page 5 of 7
Downloaded from http://peds.oxfordjournals.org/ at University of Otago on October 4, 2014
AMS technology has been applied to production cell line development of multiple antibody projects. Usually 108 – 109 CHO cells are transfected with the linearized AMS antibody expression vector. After 2 weeks of stable selection using 10 mg/ml of puromycin, the stable cells are stained for membrane-anchored antibody and subjected to FACS. The top 0.01% of cells with the highest expression levels of the membrane antibody are isolated as a pool that is subsequently provided with Cre recombinase to switch off membrane antibody expression. After culturing for 1 week, the cells are again stained for the membrane antibody, and those lacking cellsurface antibody are sorted into 96-well plates as single cells. The clones grow out in 2 weeks. The antibody in the culture media is measured by ELISA, and the top clones are expanded and cryopreserved. The whole process takes 2 – 3 months. Production titers of the top clones in commercial media are typically 0.6– 1 g/l in 7-day batch culture or 2 – 4 g/l in 10-day fed-batch culture. The specific productivity is typically 40– 60 pg/cell/day. More than 80% of the isolated clones are found to produce antibody stably (within 70% of the original productivity) after 2 months of continuous culturing without antibiotic selection.
B.Yu et al.
Fig. 5. Antibody secretion correlates with membrane expression. Adalimumab sequences were cloned into the AMS expression vector. After stable transfection in CHO cells, six cell lines with different expression levels were isolated. These cells expressed both membrane-anchored and secreted antibodies. The membrane antibody expression was characterized by staining with anti-human Fc antibody FITC conjugate and cytometric analysis, and was quantified as MFI (geometric mean of fluorescence intensity). The secreted antibody expression was characterized by measuring the production titer in the culture medium after 7-day batch culture.
Page 6 of 7
be purified and subjected to a suite of analytic tools to assess critical product quality attributes at the earliest stage of development (Fig. 6). Recently, more emphasis has been placed on rigorous characterization of product quality attributes of therapeutic biologics. Subtle differences of protein therapeutics have often been observed among various production cell lines, such as glycosylation profile, aggregation levels, oxidation, deamination and charge heterogeneities. Some of these attributes have shown profound clinical effects. Importantly, AMS can provide significant benefit to assess critical quality attributes at an early stage of cell line development to reduce project risk and complexity in the downstream development process. Another interesting application of AMS technology is library screening. The current therapeutic antibody development paradigm is a two-stage process. The first stage is antibody discovery and engineering to identify the therapeutic candidate. The second stage is cell line development and process development for production. One of the limitations of
Downloaded from http://peds.oxfordjournals.org/ at University of Otago on October 4, 2014
Fig. 4. Efficient switching off of the cell-surface antibody by Cre. (A) Transient transfection of Cre expression vector. Stable CHO cells with surface antibody expression generated by the AMS vector were transfected with a Cre expression vector. After a week, the cells were stained with anti-human Fc antibody FITC conjugate and subjected to cytometric analysis. (B) Recombinant Cre. A CHO cell line with surface antibody generated by the AMS vector was treated with 1 mM of recombinant Cre by adding into the culture media for 2 h before switching to fresh media. The cells were cultured for a week before staining with anti-human Fc antibody FITC conjugate and subsequent cytometric analysis. (C) After Cre treatment, the cells were cultured for a week and sorted out as single cells for lack of the surface antibody. After colonies grew out, cells were confirmed to be void of surface antibody after staining with anti-human Fc antibody FITC conjugate and cytometric analysis.
Antibody engineering using antibody-membrane switch
the two-stage process is that the lead molecules out of the discovery stage are often not in the final format of the therapeutic candidate, which is often full-length IgG. Thus, reformatting is often necessary prior to production cell line development, which may lead to downstream complexity. This risk is particularly substantial for the lead molecules from synthetic library screening. It may be highly beneficial to interrogate an antibody library in the same format as the final therapeutic candidate during the discovery stage. It would be even better if the final product were identical to the ones during the discovery stage, i.e. they are produced from the same cell. To achieve this goal, it is necessary to streamline antibody discovery and cell line development into a single process. AMS technology makes this possible. If an antibody library is constructed with the AMS expression vector and integrated site-specifically into a highly transcriptionally active chromosome locus in CHO cells, the cell-surface antibodies can be used to facilitate isolation of antigen binding cells by FACS. The isolated cells can be subsequently transformed seamlessly into production cells by switching to membrane-anchored antibody. Success of a mammalian cell-displayed antibody library depends on the library diversity and a uniform expression level of each member. Although relatively small antibody libraries (104) have been suggested to be sufficient to identify low nanomolar binders (Mao et al., 2010), experience in yeast display and cell-surface display has shown that .108 antibody diversity may be necessary for de novo antibody screening (Beerli et al., 2008; Pepper et al., 2008). We have initiated construction of an antibody library based on AMS technology that utilizes recombinase-mediated cassette exchange to insert a 108 antibody library into a highly active transcriptional locus. The goal is to achieve up to 109 diversity in a liter of cells (106 cells/ml).
Browne,S.M. and Al-Rubeai,M. (2007) Trends Biotechnol., 25, 425–432. Carroll,S. and Al-Rubeai,M. (2004) Expert Opin. Biol. Ther., 4, 1821–1829. Jayapal,K.R., Wlaschin,K.F., Hu,W.S. and Yap,M.G.S. (2007) Chem. Eng. Prog., 103, 40– 47. Li,F., Vijayasankaran,N., Shen,A., Kiss,R. and Amanullah,A. (2010) MAbs, 2, 466–477. Mao,H., Graziano,J.J., Chase,T.M., Bentley,C.A., Bazirgan,O.A., Reddy,N.P., Song,B.D. and Smider,V.V. (2010) Nat. Biotechnol., 28, 1195– 1202. Pepper,L.R., Cho,Y.K., Boder,E.T. and Shusta,E.V. (2008) Comb. Chem. High Throughput Screen., 11, 127– 134. Scott,C. (2012) BioProcess Int., 10, S72– S78. Sleiman,R.J., Gray,P.P., McCall,M.N., Codamo,J. and Sunstrom,N.A. (2008) Biotechnol. Bioeng., 99, 578–587. Turen,S. and Bode,J. (2011) FASEB J., 25, 1 –20. Wang,Y., Yau,Y.Y., Perkins-Balding,D. and Thomson,J.G. (2011) Plant Cell Rep., 30, 267–285.
References Agrawal,V. and Bal,M. (2012) BioProcess Int., 10, 32– 48. Beerli,R.R., Bauer,M., Buser,R.B., Gwerder,M., Muntwiler,S., Maurer,P., Saudan,P. and Bachmann,M.F. (2008) PNAS, 105, 14336–14341.
Page 7 of 7
Downloaded from http://peds.oxfordjournals.org/ at University of Otago on October 4, 2014
Fig. 6. Product quality attribute-based cell line development. After stable transfection, high-producing cells are isolated by FACS as a pool. The membrane antibody is then switched off by Cre. Single cell cloning (SCC) is performed by sorting cells lacking membrane antibody into 96-well plates by FACS. After colonies grow up, up to tens of micrograms of antibodies may be purified from each well and subjected to various analytic assays for product quality attributes.
Antibody-membrane switch (AMS) technology for facile cell line development Bo Yu1,2,3, John M.Wages2 and James W.Larrick1,2 1
Larix Bioscience, LLC, 1230 Bordeaux Drive, Sunnyvale, CA 94089, USA and 2Panorama Research, Inc., Sunnyvale, CA 94089, USA
3
To whom correspondence should be addressed. E-mail: [email protected] Received June 12, 2014; revised August 3, 2014; accepted August 12, 2014 Edited by James S. Huston
Introduction Since the first regulatory approval in 1986, therapeutic monoclonal antibodies (mAbs) have become the fastest growing class of pharmaceutical drugs (Scott, 2012). Development of novel classes of antibody derivatives (Fc-modified mAb, bispecific mAb and antibody –drug conjugates) promises even more activity and specificity in third-generation therapeutic mAbs. These magic bullets will revolutionize pharmacotherapy of cancer and multiple other disease areas. Development of a therapeutic antibody is a lengthy, expensive process, often taking 10 – 15 years including antibody discovery/engineering, production cell line development, manufacturing process development and clinical studies
# The Author 2014. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: [email protected]
Page 1 of 7
Downloaded from http://peds.oxfordjournals.org/ at University of Otago on October 4, 2014
Generation of high-productivity cell lines remains a major bottleneck in therapeutic antibody development. Conventional cell line development often depends on gene amplification methodologies using dihydrofolate reductase or glutamine synthetase. Higher productivity is associated with an increased gene copy number. However, lack of selection pressure under the conditions of large-scale manufacturing leads to clonal instability. We have developed a novel method for cell line development, antibodymembrane switch (AMS) technology, that does not rely on gene amplification. This fluorescence-activated cell sorting (FACS)-based, high-throughput method is facilitated by cell-surface antibody expression to rapidly and efficiently isolate high-producing cells. The switch between membrane expression and secretion is achieved by alternative splicing and specific DNA recombination. The antibody of interest is initially displayed on the cell surface to facilitate FACS. Isolated high-producing cells are then seamlessly transformed into production cells after removing the membrane-anchoring domain sequence with a DNA recombinase. AMS technology has been applied in a number of antibody cell line development projects, which typically last 2 – 3 months. The top production cell lines exhibit very high specific productivity of 40– 60 pg/cell/day resulting in production titers of 2 – 4 g/l in 10-day batch culture. Keywords: antibody engineering/FACS/monoclonal antibodies/protein engineering
(Agrawal and Bal, 2012). Among these tasks, antibody discovery may take 6 – 18 months, and development of production cell lines may require an additional 10 – 12 months (Jayapal et al., 2007; Li et al., 2010). Production of material for Phase I/II trials is a critical, limiting step in bringing a therapeutic protein to market. Production cell lines must exhibit stable, high-level expression of the antibody that exhibits all of the desired product quality attributes, including protein sequence homogeneity, glycosylation profile, charge variants and aggregation levels. CHO cells are the most popular mammalian cells for production of therapeutic proteins, although other mammalian cells like NS0 or SP2/0 cells have also been used to produce biological therapeutics (Jayapal et al., 2007; Li et al., 2010). Presently, commercial scale manufacturing target yields exceed 2 – 3 g/l. To attain this goal, a cell line producing .20 pg/cell/day is required prior to process development and bioreactor optimization. Because it is rare to have the gene of interest (GOI) integrated into a highly active transcriptional site in the genome of the production cell line, conventional cell line development methodologies often require gene amplification. Historically, dihydrofolate reductase (DHFR) or glutamine synthetase (GS) has been employed to multiply the integrated GOI to enhance expression levels (Agrawal and Bal, 2012). Addition of increasingly high concentrations of methotrexate (MTX) in the DHFR system or methionine sulfoximine (MSX) in the GS system applies selection pressure to amplify antibody genes adjacent to the DHFR or GS. One of the downsides of gene amplification is that cells with multiple copies of the GOI are often unstable, resulting in decreased production yields in the absence of selection pressure. Therefore, multiple production cell lines have to be carried into late-stage process development to ensure at least one stable cell line for largescale manufacturing. The other downside of gene amplification is low throughput resulting from conventional limiting dilution cloning methodology. In a typical cell line screening campaign of 6 – 8 months, 1000 GOI-positive cells from 20– 30 96-well plates are screened for high expression levels, which leads to only dozens of high-productivity clones. In addition, multiple rounds of subcloning are required to achieve clonality. In recent years, several epigenetic approaches have been used to facilitate cell line development: matrix-attachment region or Selexis Genetic Elements (SGE, Selexis), ubiquitous chromatin opening element (UCOE, Millipore) and antirepressor elements (STAR, Crucell). These DNA elements can be incorporated into antibody expression vectors to enhance transcriptional activities, particularly when the expression vector is integrated in an unfavorable locus. Although these methods have been reported to significantly improve production yield for stable pools, their success in screening for top production cell lines has been limited (Browne and Al-Rubeai, 2007; Agrawal and Bal, 2012).
B.Yu et al.
Materials and methods
Cell culture and transfection 293F and CHOS cells (Invitrogen) were maintained in Freestyle 293 expression media and CD FortiCHO media (Invitrogen), respectively.
Cloning of antibody in the AMS expression vector Antibody heavy chain variable sequences were cloned between restriction sites XbaI and NheI in an antibody expression vector containing the constant region, which was under the control of a human EF1a promoter. The vector carries a puromycin resistance gene for stable cell selection and an ampicillin resistance gene for Escherichia coli propagation. Antibody light chain sequences were cloned into a separate mammalian expression vector between restriction sites XbaI and BamHI, under the control of a human EF1a promoter. The light chain expression cassette was then subcloned into the heavy chain expression vector between the restriction sites EcoRV and AscI.
Transfection of 293F cells and CHOS cells Transfection of 293F cells or CHOS cells was carried out with either the Freestyle Max Transfection Reagent or electroporation using the Neon Transfection System (Invitrogen) according to the manufacturer’s suggested protocol. Stable selection of 293F cells or CHOS cells was with 1 or 10 mg/ml of puromycin, respectively.
Determination of antibody production titer Antibody levels in the culture media were determined by dilution enzyme-linked immunosorbent assay (ELISA) in which the antibody was captured with goat anti-human IgG Fc (100 ng/well, Bethyl) and detected with goat anti-human Kappa antibody HRP conjugate (1 : 10 000 dilution, Bethyl). Human IgG antibody (2 mg/ml, Sigma) was used as the standard for IgG quantitation.
Fig. 1. AMS technology. Membrane display of antibodies is achieved by alternative splicing of a modified CH3 exon fused with a MAD. High-expressing cells are sorted by FACS, and then transformed into production cells by execution of a molecular switch to remove the CH3-MAD exon.
Page 2 of 7
Downloaded from http://peds.oxfordjournals.org/ at University of Otago on October 4, 2014
Because only dozens of high-producing clones are generated in conventional cell line development, productivity is usually the only criterion used for selection until a very late stage, when a handful of clones are assessed for other quality attributes critical for large-scale manufacturing. Thus, activity, glycosylation, aggregation, oxidation, deamination and charge are often not assessed until later stages of development. This screening delay increases project risk and raises complex issues regarding downstream development. We have developed a novel approach based on transient cell-surface expression of the target antibody combined with fluorescence-activated cell sorting (FACS) to address all the shortcomings of conventional cell line development technology. Although several FACS-based screening approaches have been described (Carroll and Al-Rubeai, 2004; Sleiman et al., 2008), none of these has come into widespread use. Antibodymembrane switch (AMS) technology utilizes alternative splicing of the immunoglobulin heavy chain. A membraneanchoring domain (MAD) is introduced in the CH3 exon to generate some of the expressed antibodies on the cell surface (Fig. 1). The antibody of interest is secreted, as well as being anchored on the cell surface. Thus, cells can be readily detected and sorted by FACS. The added CH3-MAD exon resides in the intron region situated between the CH2 and CH3 heavy chain domains, and is flanked by a site-specific DNA recombinase recognition sequence (DRRS) such as a LoxP sequence. The MAD can be a GPI-anchored signal sequence (GASS) or a protein transmembrane domain (TM). After selection for high expression levels of membrane-anchored antibodies by FACS using fluorescence-labeled antigen or secondary antibody, the cells are then switched into production mode by transient expression of an appropriate site-specific DNA recombinase (Turen and Bode, 2011; Wang et al., 2011) such as Cre to remove the MAD exon, resulting in a cell line making only secreted antibodies. AMS streamlines and greatly accelerates the development of productive cell lines for antibody production.
Antibody engineering using antibody-membrane switch
Characterization of cell-surface antibody expression The cells were incubated with goat anti-human Fc antibody fluorescein isothiocyanate (FITC) conjugate (1 : 1000, Bethyl) for 30 min. After washing once with phosphate-buffered saline, cells were subjected to flow cytometric analysis.
Switching off the membrane-anchored antibody using Cre recombinase
Screening of highly productive cell lines Antibody sequences were cloned into the AMS expression vector. The plasmid was linearized by digestion of AscI and was used to transfect 0.1–1 l of CHOS cells (1 106 cells/ml). Stable cells were selected with 10 mg/ml of puromycin for 2 weeks. Stable cells (1 – 10 107) were labeled with goat anti-human Fc antibody FITC conjugate (1 : 1000, Bethyl) before FACS sorting. The 0.01% of the cells with the highest fluorescence signal intensity were sorted into a pool. After 2 days, cells were exposed to Cre recombinase either by transfection or addition of the recombinant protein. After 1 week of culturing, cells were labeled with goat anti-human Fc antibody FITC conjugate (1 : 1000, Bethyl). Cells lacking surface antibody were sorted into 10– 20 96-well plates as single cells. Approximately 200– 500 colonies grew out after 2 – 3 weeks. Culture media were screened by ELISA for expression of the antibody, and top clones were expanded and cryopreserved. Results
LoxP site in the middle of antibody H chain intron 3 does not change expression of rituximab (anti-CD20) Antibody expression vectors for development of production cell lines often utilize the genomic DNA for the IgG heavy chain region consisting of four exons (CH1, hinge, CH2 and CH3) and three introns (Fig. 2A). The rituximab heavy chain variable region was cloned into the antibody expression vectors with the wild-type genomic heavy chain or a modified genomic heavy chain with a LoxP site inserted in the middle of intron 3 (Fig. 2B). Two days after co-transfection of 293F cells with the rituximab light chain expression vector using Freestyle Max, antibody expression levels were measured by ELISA. The cells transfected with the LoxP-modified heavy chain (Fig. 3A-b) produced a similar amount of antibodies as the cells transfected with the wild-type heavy chain (Fig. 3A-a), suggesting that the LoxP sequence inserted in the middle of the intron between CH2 and CH3 did not affect the levels of antibody expression. The heavy chain constant region genes in both transfected cells were amplified by reverse
Antibody secretion by cells transfected with AMS vectors The rituximab heavy chain genomic region was modified by insertion of an extra CH3 exon fused with either a transmembrane domain (TM) or a GPI-linked membrane-anchoring sequence (GASS) flanked by LoxP sites in the middle of intron 3. The TM sequence was taken from the human PDGFRb sequence. The GASS was taken from the human decay accelerating factor sequence. Alternative splicing facilitated by one splice donor at the end of CH2 exon and two possible splice acceptors for the CH3 exons (Fig. 2C) permits simultaneous expression of secreted antibody and membrane-anchored antibody. The LoxP sites permit removal of the extra CH3 exon in intron 3 facilitated by specific DNA recombinase Cre after isolation of the high producer cells (Fig. 2D). After co-transfection (Freestyle Max) with rituximab light chain expression vector in 293F cells, the antibody expression levels in the culture media were determined by ELISA. The heavy chain with the extra CH3-TM exon (Fig. 3A-d) produced 50% as much antibody in the culture media as the wild-type control (Fig. 3A-a), whereas the heavy chain with the extra CH3GASS exon (Fig. 3A-c) secreted 30% more antibody than the wild-type control. This is unexpected, albeit consistent with the fact that GPI-linked membrane anchorage is not 100%. This result has been reproduced in three independent experiments.
Membrane antibody expression of the cells transfected with AMS vectors Although 293 cells often exhibit higher transient transfection efficiency (70–90%) than CHO cells (20–40%) using lipidbased transfection reagent, CHO cells are the industry standard for antibody production. The rituximab expression vectors were transfected into CHOS cells (Freestyle Max). Cell-surface antibody was characterized by cytometric analysis after staining with anti-human Fc antibody FITC conjugate. Cells transfected with the wild-type heavy chain vector (Fig. 3B-a) or LoxP-modified CH3 exon vector (Fig. 3B-b) showed only background levels of cell-surface antibody expression (1–2%), whereas cells transfected with alternatively spliced CH3-GASS vector (Fig. 3B-c) or CH3-TM vector (Fig. 3B-d) exhibited cellsurface antibodies in 27.9 or 20.1% of the cells, respectively. The percentage of cells that expressed membrane antibody was consistent with the transfection efficiency. The alternative splicing (Fig. 2C) was confirmed by RT-PCR and sequencing. The ratio of RT-PCR product incorporating the CH3-TM exon and the wild-type CH3 exon was estimated to be 1 : 1 by the occurrence of either exon during sequencing, representing the ratio between membrane-anchored antibody and secreted antibody.
Switching off the membrane-anchored antibody using Cre recombinase Cre recombinase was presented to the cells by either transient transfection of a Cre expression vector or by addition of membrane-permeable recombinant Cre into the culture medium. A stable pool of CHO cells expressing membrane antibody was isolated by FACS after stable transfection with the AMS antibody expression vector. Most of the cells (80%) exhibited strong expression of cell-surface antibody Page 3 of 7
Downloaded from http://peds.oxfordjournals.org/ at University of Otago on October 4, 2014
A Cre recombinase expression vector was constructed. The Cre cDNA was human codon-optimized and fused with a peptide containing a nuclear localization signal (MPKKKRK) at the N-terminus. Expression of Cre was driven by a human EF1a promoter. Cells with stable expression of membraneanchored antibody were transfected with the Cre expression vector and cultured an additional week before characterization of the cell-surface antibody. Alternatively, cells with stable expression of membrane-anchored antibody were treated with 1 mM of recombinant Cre fused with TAT-NLS for nuclear localization (Excellgen) by adding into the culture media for 2 h before replacing with fresh media. Cell-surface antibody was measured after 1 week of culture.
transcription-polymerase chain reaction (RT-PCR) and sequenced. RNA splicing was found to be identical with or without a LoxP site inserted inside the gamma chain intron.
B.Yu et al.
Downloaded from http://peds.oxfordjournals.org/ at University of Otago on October 4, 2014
Fig. 2. Schematic drawing of the antibody heavy chain constant region of the AMS vectors. (A) Schematic of wild-type IgG1 heavy chain constant region gene. (B) A LoxP site is inserted in the middle of intron 3. (C) Alternative splicing created by an extra membrane-anchored CH3 exon inserted in intron 3. TM refers to a transmembrane domain or a GIP-linked membrane-anchoring sequence. (D) The extra CH3-TM exon is removed by Cre recombinase.
(Fig. 4A). After transient transfection by electroporation with a Cre expression vector, 75% of the cells with membraneanchored antibody lost cell-surface antibody after 1 week in culture (Fig. 4A), consistent with the improved transfection efficiency compared with the lipid-based transfection reagent. A stable CHO cell line isolated by FACS after stable transfection with the AMS antibody expression vector expressed antibody on the cell surface (Fig. 4B). To provide the cells with Cre recombinase, 1 mM of cell membrane-permeable Page 4 of 7
recombinant Cre was added to the culture media for 2 h before switching to fresh media. After culturing for one more week, 78% of the cells lost membrane-anchored antibody (Fig. 4B). The cells lacking membrane antibody were sorted into 96-well plates as single cells and cultured for additional 2 weeks. Clones were confirmed to be completely void of cell-surface antibody (Fig. 4C). The removal of the membrane-anchoring CH3 exon from the chromosome was confirmed by DNA sequencing.
Antibody engineering using antibody-membrane switch
Cell line development using AMS technology
Discussion
Fig. 3. Rituximab expression by AMS vector after transient transfection in 293 cells. (A) Secreted rituximab expression. Antibody concentrations in media 2 days after transient transfection were determined by ELISA. LB-K designates the light chain expression vector, whereas LB-H, LB-1, LB-2 and LB-3 designate the heavy chain expression vectors. (a) LB-H: wild-type heavy chain; (b) LB1: a LoxP site inserted in intron 3; (c) LB2: an extra CH3-GASS exon inserted in intron 3; (d) LB4: an extra CH3-TM exon inserted in intron 3. (B) Cell-surface rituximab expression. 293 cells were stained with anti-human Fc antibody FITC conjugate 2 days after transfection with rituximab expression vectors and subjected to cytometric analysis. (a) wild-type heavy chain; (b) a LoxP site inserted in intron 3; (c) an extra CH3-GASS exon inserted in intron 3; (d) an extra CH3-TM exon inserted in intron 3.
Membrane antibody expression correlates with secreted antibody levels Adalimumab sequences were cloned into the AMS expression vector. After stable transfection of CHO cells, six cell lines with different expression levels were isolated. Transfectants expressed both membrane-anchored and secreted antibodies. Membrane antibody expression was measured by staining with anti-human Fc antibody FITC conjugate and cytometric analysis, and was quantified as the geometric mean of total fluorescence intensity (MFI). Secreted antibody expression was characterized by measuring the production titer in culture media after 7-day batch culture. Membrane antibody expression levels correlate strongly with the levels of antibody secretion (Fig. 5).
We have developed AMS technology for rapid development of production cell lines capable of yielding very high specific productivity of 40– 60 pg/cell/day within 2 – 3 months. The essence of AMS technology is a switch mechanism that turns off expression of a reporter (e.g. a cell-surface antibody), which is a surrogate for protein of interest (POI) expression, after isolation of a subpopulation of cells providing for optimal expression of the POI. The simultaneous expression of secreted antibody and cell-surface antibody is achieved by alternative splicing utilizing an extra CH3 exon fused with an MAD, which is flanked by specific DRRSs. The antibody is cell surface expressed to facilitate high-throughput FACS screening to rapidly isolate the cells producing the highest level of antibody. After screening, the top producing cells are seamlessly transformed into production cells by removing the antibody MAD from the chromosome by targeted DNA recombination. AMS technology relies on high-throughput screening rather than gene amplification to isolate the cells with high productivity. FACS is the most efficient high-throughput screening method at more than 100 million cells per day. Thus, AMS technology begins with transfection of more than 100– 1000 times the number of cells utilized in conventional cell line development. Enrichment of the high-producing cells by FACS ensures capture of the rare clones bearing the GOI integrated into a highly transcriptionally active locus. The isolated cell lines likely do not possess many tandem repeats of the integrated DNA sequences characteristic of gene amplification, consistent with the finding of a much higher ratio of stable cell lines (.80%), compared with the cell lines isolated through gene amplification methods (,50%). A significant value of AMS technology results from the generation of more than hundreds to even thousands of highly productive cell lines. These clones produce up to tens of micrograms of antibody in the 96-well plate stage, which can Page 5 of 7
Downloaded from http://peds.oxfordjournals.org/ at University of Otago on October 4, 2014
AMS technology has been applied to production cell line development of multiple antibody projects. Usually 108 – 109 CHO cells are transfected with the linearized AMS antibody expression vector. After 2 weeks of stable selection using 10 mg/ml of puromycin, the stable cells are stained for membrane-anchored antibody and subjected to FACS. The top 0.01% of cells with the highest expression levels of the membrane antibody are isolated as a pool that is subsequently provided with Cre recombinase to switch off membrane antibody expression. After culturing for 1 week, the cells are again stained for the membrane antibody, and those lacking cellsurface antibody are sorted into 96-well plates as single cells. The clones grow out in 2 weeks. The antibody in the culture media is measured by ELISA, and the top clones are expanded and cryopreserved. The whole process takes 2 – 3 months. Production titers of the top clones in commercial media are typically 0.6– 1 g/l in 7-day batch culture or 2 – 4 g/l in 10-day fed-batch culture. The specific productivity is typically 40– 60 pg/cell/day. More than 80% of the isolated clones are found to produce antibody stably (within 70% of the original productivity) after 2 months of continuous culturing without antibiotic selection.
B.Yu et al.
Fig. 5. Antibody secretion correlates with membrane expression. Adalimumab sequences were cloned into the AMS expression vector. After stable transfection in CHO cells, six cell lines with different expression levels were isolated. These cells expressed both membrane-anchored and secreted antibodies. The membrane antibody expression was characterized by staining with anti-human Fc antibody FITC conjugate and cytometric analysis, and was quantified as MFI (geometric mean of fluorescence intensity). The secreted antibody expression was characterized by measuring the production titer in the culture medium after 7-day batch culture.
Page 6 of 7
be purified and subjected to a suite of analytic tools to assess critical product quality attributes at the earliest stage of development (Fig. 6). Recently, more emphasis has been placed on rigorous characterization of product quality attributes of therapeutic biologics. Subtle differences of protein therapeutics have often been observed among various production cell lines, such as glycosylation profile, aggregation levels, oxidation, deamination and charge heterogeneities. Some of these attributes have shown profound clinical effects. Importantly, AMS can provide significant benefit to assess critical quality attributes at an early stage of cell line development to reduce project risk and complexity in the downstream development process. Another interesting application of AMS technology is library screening. The current therapeutic antibody development paradigm is a two-stage process. The first stage is antibody discovery and engineering to identify the therapeutic candidate. The second stage is cell line development and process development for production. One of the limitations of
Downloaded from http://peds.oxfordjournals.org/ at University of Otago on October 4, 2014
Fig. 4. Efficient switching off of the cell-surface antibody by Cre. (A) Transient transfection of Cre expression vector. Stable CHO cells with surface antibody expression generated by the AMS vector were transfected with a Cre expression vector. After a week, the cells were stained with anti-human Fc antibody FITC conjugate and subjected to cytometric analysis. (B) Recombinant Cre. A CHO cell line with surface antibody generated by the AMS vector was treated with 1 mM of recombinant Cre by adding into the culture media for 2 h before switching to fresh media. The cells were cultured for a week before staining with anti-human Fc antibody FITC conjugate and subsequent cytometric analysis. (C) After Cre treatment, the cells were cultured for a week and sorted out as single cells for lack of the surface antibody. After colonies grew out, cells were confirmed to be void of surface antibody after staining with anti-human Fc antibody FITC conjugate and cytometric analysis.
Antibody engineering using antibody-membrane switch
the two-stage process is that the lead molecules out of the discovery stage are often not in the final format of the therapeutic candidate, which is often full-length IgG. Thus, reformatting is often necessary prior to production cell line development, which may lead to downstream complexity. This risk is particularly substantial for the lead molecules from synthetic library screening. It may be highly beneficial to interrogate an antibody library in the same format as the final therapeutic candidate during the discovery stage. It would be even better if the final product were identical to the ones during the discovery stage, i.e. they are produced from the same cell. To achieve this goal, it is necessary to streamline antibody discovery and cell line development into a single process. AMS technology makes this possible. If an antibody library is constructed with the AMS expression vector and integrated site-specifically into a highly transcriptionally active chromosome locus in CHO cells, the cell-surface antibodies can be used to facilitate isolation of antigen binding cells by FACS. The isolated cells can be subsequently transformed seamlessly into production cells by switching to membrane-anchored antibody. Success of a mammalian cell-displayed antibody library depends on the library diversity and a uniform expression level of each member. Although relatively small antibody libraries (104) have been suggested to be sufficient to identify low nanomolar binders (Mao et al., 2010), experience in yeast display and cell-surface display has shown that .108 antibody diversity may be necessary for de novo antibody screening (Beerli et al., 2008; Pepper et al., 2008). We have initiated construction of an antibody library based on AMS technology that utilizes recombinase-mediated cassette exchange to insert a 108 antibody library into a highly active transcriptional locus. The goal is to achieve up to 109 diversity in a liter of cells (106 cells/ml).
Browne,S.M. and Al-Rubeai,M. (2007) Trends Biotechnol., 25, 425–432. Carroll,S. and Al-Rubeai,M. (2004) Expert Opin. Biol. Ther., 4, 1821–1829. Jayapal,K.R., Wlaschin,K.F., Hu,W.S. and Yap,M.G.S. (2007) Chem. Eng. Prog., 103, 40– 47. Li,F., Vijayasankaran,N., Shen,A., Kiss,R. and Amanullah,A. (2010) MAbs, 2, 466–477. Mao,H., Graziano,J.J., Chase,T.M., Bentley,C.A., Bazirgan,O.A., Reddy,N.P., Song,B.D. and Smider,V.V. (2010) Nat. Biotechnol., 28, 1195– 1202. Pepper,L.R., Cho,Y.K., Boder,E.T. and Shusta,E.V. (2008) Comb. Chem. High Throughput Screen., 11, 127– 134. Scott,C. (2012) BioProcess Int., 10, S72– S78. Sleiman,R.J., Gray,P.P., McCall,M.N., Codamo,J. and Sunstrom,N.A. (2008) Biotechnol. Bioeng., 99, 578–587. Turen,S. and Bode,J. (2011) FASEB J., 25, 1 –20. Wang,Y., Yau,Y.Y., Perkins-Balding,D. and Thomson,J.G. (2011) Plant Cell Rep., 30, 267–285.
References Agrawal,V. and Bal,M. (2012) BioProcess Int., 10, 32– 48. Beerli,R.R., Bauer,M., Buser,R.B., Gwerder,M., Muntwiler,S., Maurer,P., Saudan,P. and Bachmann,M.F. (2008) PNAS, 105, 14336–14341.
Page 7 of 7
Downloaded from http://peds.oxfordjournals.org/ at University of Otago on October 4, 2014
Fig. 6. Product quality attribute-based cell line development. After stable transfection, high-producing cells are isolated by FACS as a pool. The membrane antibody is then switched off by Cre. Single cell cloning (SCC) is performed by sorting cells lacking membrane antibody into 96-well plates by FACS. After colonies grow up, up to tens of micrograms of antibodies may be purified from each well and subjected to various analytic assays for product quality attributes.
