High-throughput reformatting of phage-displayed antibody fragments to IgGs by one-step emulsion PCR

High-throughput reformatting of phage-displayed antibody fragments to IgGs by one-step emulsion PCR Abstract Single-chain variable fragment (scFv) is the most common format for phage display antibody library. The isolated scFvs need to be reformatted to full-length IgGs for further characterization. High throughput reformatting of scFv to IgG without disrupting VH–VL pairing is of great demanding for exhaustive screening of all antibodies in IgG format. Herein, we developed a strategy based on the overlap extension PCR in emulsion to reformat scFv to IgG while maintain the accuracy and complexity of variable region pairing. Using CD40 as an example target, we reformatted phage display derived CD40 binding scFv library to IgG mammalian display library and isolated high affinity CD40 binding IgGs. This robust and reliable antibody reformatting approach could be integrated into any phage display based antibody drug discovery. Introduction Antibody phage display is an essential technique of antibody-based drug industry and a crucial step to generate antibodies for a variety of clinical applications, such as treatment of metastatic cancer, autoimmune diseases and detoxification (Huse et al., 1989; McCafferty et al., 1990; Barbas et al., 1991). To date, the majority of therapeutic antibodies developed are full-length IgG molecules produced in mammalian cells. In this respect, phage display is limited to antibody fragments such as scFv (Huston et al., 1988). Thus, the scFv selected by phage display had to be converted to full-length IgG format. Many scFv format antibodies lost their activity when transformed into full-length IgGs (Steinwand et al., 2014). It is of great relevance to screen the antibody library in a full-length IgG format. The reformatting from scFv library to full-length IgG library without disrupting VH–VL pairing is a challenging task. Traditional conversion from scFv to IgG was performed on an individual basis (Jostock et al., 2004; Kotera and Nagai, 2008; Chen et al., 2014; Sanmark et al., 2015) which is a time consuming, labor intensive process and the number of antibodies assessed as full-length IgG is only a small fraction of the scFv repertoire. Chao-Guang Chen et al. reported a cloning method, referred to as insert-tagged (InTag) positive selection, for rapid reformatting of phage display antibody fragments to IgGs based on ligation-independent cloning (LIC) method (Chen et al., 2014). Hanna Sanmark et al. reported a modified version of method fully automatic single-tube recombination (FASTR), which enabled VH, VL, CH and CL DNA recombination in the correct order in a single step (Sanmark et al., 2015). However, these methods cannot reformat a library of scFv to IgG as they cannot avoid wrong VH–VL pairing (Sun et al., 2017). Xiaodong Xiao et al. first reported an inverse PCR based approach to reformat scFv library to IgG library without disrupting the variable region pairing (Xiao et al., 2017). Although inverse PCR can ensure the linkage between VH–VL pairing throughout the conversion process, a more straightforward approach to avoid wrong VH–VL pairing is to compartmentalize the reformatting reaction of each individual antibody. Emulsion PCR were developed and used in complex gene library amplification for its high-throughput and accuracy, so we adopted the technology to our strategy (Williams et al., 2006; Turchaninova et al., 2013). Emulsion PCR is based on the compartmentalization of genes into millions of aqueous droplets in a water-in-oil (w/o) emulsion. Templates are segregated in the minute aqueous droplets of the emulsion and amplified by PCR in isolation. Furthermore, emulsion PCR can alleviate the generation of chimeric PCR by-product arising from recombination between homologous regions of DNA. Here we described a straightforward and reliable approach to reformat scFv library to full-length IgG library in one step. The variable region of heavy chain and variable light chain of a single scFv molecule was amplified and full-length IgG was assembled by overlap extension PCR (OE-PCR) in aqueous droplet of water-in-oil emulsion. Coexpression of the assembled heavy chain and light chain was achieved by either T2A or Internal Ribosome Entry Site (IRES). Mammalian display is widely acknowledged as the best way to screen full-length antibodies (Zhou et al., 2010; Zhou and Shen, 2012; Tomimatsu et al., 2013). The full-length antibodies were displayed on mammalian cell surface and sorted by FACS. As a result, several high-affinity CD40 binding IgGs were isolated. To date, the technology presented here allows for more effective mining of phage display scFv libraries for the discovery of full-length IgG of high performance. Material and methods Phage display Human CD40 ectodomain protein (human IgG1 Fc tag, avi-Tag) was biotinylated using biotin protein ligase (GeneCopoeia, BI001). A human naïve scFv library made from PBMC of 30 healthy donors as previously described (Vaughan et al., 1996). The phage display scFv library was used for selections of antibodies against CD40-Fc fusion protein. Briefly, biotinylated CD40-Fc fusion protein was incubated with phage library for 2 h at RT and the phage-CD40 complex were captured by M-280 Streptavidin magnetic beads (Thermofisher, 11 205D). After washes the bound phages were eluted by 0.2 M Glycin-HCl (pH 2.2) for 10 min at RT and neutralized with 1 M Tris pH 8.0 to adjust the pH to 7.5. Then the eluted phages were amplified for next round of panning. Phagemid DNA was purified from the third round of phage display selection for the reformatting study. One step overlap extension emulsion PCR VL, bridge fragment and VH are assembled by the overlap extension PCR. The bridge fragment (CL-T2A/IRES) contains human CL fragment, T2A or IRES and IL-2 signal sequence. Emulsion PCR was performed with an Emulsion Kit (EURx Ltd. E3600). The water phase of the emulsion PCR system contains: optimized amount of template (2 ng, about 5 × 108 molecules of phagemid), 200 ng CL-T2A or 350 ng CL-IRES, 20 nM outer side primers (each), 400 nM inner side primers (each). The inner side primers are JH-RV and VL-FW. The outer side primers are VH-FW and JL-RV as indicated in Fig. 1. The inner side primers are more than the outer side primers to favor the amplification of overlap product. The sequences of primers are listed in Supplemental Table S1, Hot-Start Q5 DNA polymerase (New England Biolabs. M0493), 200 μM dNTP and 0.01 mg/mL acetylated BSA, then a 50 μL PCR system was mixed with 300 μL oil surfactant mixture prepared as described in the kit manual, and the emulsion was aliquoted into four suitable tubes for a thermal cycler. One step extension emulsion PCR program was performed as follows: 94°C 5 min for denaturation, 5×Cycle 1#(denaturation at 94°C for 30 s, annealing at 55°C for 30 s, extension at 72°C for 20 s) for amplifying variable regions, 25×Cycle 2#(denaturation at 94°C for 30 s, annealing at 55°C for 30 s, extension at 72°C for 55 s) for VL-bridge fragment(CL-T2A/IRES)-VH assembling and extension PCR, 72°C 15 min for the last long extension. The emulsion was broken with butanol and DNA was extracted from the opened emulsion by the column. Fig. 1 View largeDownload slide The schematic representation of emulsion PCR based reformatting of scFv to IgG. Individual scFv from a complex phage display library are compartmentalized in the aqueous droplets of a w/o emulsion PCR system such that each droplet contains a single scFv DNA molecule as PCR template. (A) VH chain and VL chain were amplified separately. (B and C) The VL fragment, CL-T2A/IRES-IL2 signal sequence fragment and VH fragment were assembled by overlap PCR. The PCR products were then cloned into the mammalian display vector by standard restriction digestion and ligation. Mammalian display antibody expression vectors consist of IL-2 signal sequence, the constant domain of IgG, and a PDGFR transmembrane domain. T2A or IRES was used for cocurrent expression of light chain and heavy chain of antibody. Fig. 1 View largeDownload slide The schematic representation of emulsion PCR based reformatting of scFv to IgG. Individual scFv from a complex phage display library are compartmentalized in the aqueous droplets of a w/o emulsion PCR system such that each droplet contains a single scFv DNA molecule as PCR template. (A) VH chain and VL chain were amplified separately. (B and C) The VL fragment, CL-T2A/IRES-IL2 signal sequence fragment and VH fragment were assembled by overlap PCR. The PCR products were then cloned into the mammalian display vector by standard restriction digestion and ligation. Mammalian display antibody expression vectors consist of IL-2 signal sequence, the constant domain of IgG, and a PDGFR transmembrane domain. T2A or IRES was used for cocurrent expression of light chain and heavy chain of antibody. Vector and library construction Lentivirus transfer vector pCDH was modified to be a mammalian display vector (Fig. 1). IL-2 signal sequence was added in front of the restriction enzyme site NotI. A NheI site was introduced into the N-terminal of human IgG CH1 which was synonymous mutation of Ala-Ser. The IgG was followed by a 218 linker and PDGFR transmembrane domain to facilitate the IgG displaying on cell surface. Products from one step overlap extension emulsion PCR were cloned into the plasmid by digestion with restriction enzymes NotI and NheI and ligation. NGS sample preparation and data analysis For next generation sequencing, VL and VH of the phagemid scFv library and reformatted IgG library were amplified by emulsion PCR. Individual barcode sequences were added to each sample during library preparation and samples were pooled and sequenced simultaneously during a single run by MiSeq PE250 (Illumina). The 250 bp forward read and the 250 bp reverse read data were processed with CLC Genomics Workbench and merged to assemble an entire heavy or light chain and sorted by the barcodes, then translated to protein sequences. CDR3 regions of heavy chains or light chains were extracted by MiXCR (Bolotin et al., 2015). Mammalian display and FACS sorting Mammalian display IgG library plasmid was cotransfected with plasmids encoding packaging and envelop proteins into HEK293T to generate replication-incompetent lentivirus library. After 48 h, lentiviruses were precipitated by LS buffer from kit (Biomiga, V2001) and titer was determined by Lenti-X p24 Rapid Titer Kit (Clontech, 632 200). HEK293T were infected at a low multiplicity of infection (MOI) to ascertain the expression of a single antibody per infected cell. Puromycin was added to kill cells that were not infected by lentivirus. After one week of puromycin selection cells were detached by accutase (Invitrogen, A1110501) and incubated with 20 nM biotin-CD40 at room temperature for 50 min. Then the cells were stained by 4 μg/mL SA-PE (Thermofisher, 21 627) and 2 μg/mL AF647-conjugated Goat anti Human IgG(H + L) (Thermofisher, A21445) at room temperature for 40 min and subjected for flow cytometric analysis and sorting. Isolation of CD40 binding antibody from sorted cell After three rounds of selection, single positive cells were sorted into 96-well plate and cultured in DMEM complemented with 20% FBS and 2× PS. When the single cell clone grew up to sufficient amount for analysis cells were detached by accutase (Invitrogen, A1110501) and stained for flow cytometry analysis using plate analysis method of BD LSRFortessa. The positive clones were lysed by Cell Lysis Buffer (50 mM KCl, 10 mM Tris-HCl, 0.1% Triton-100, Proteinase K at 16 units/mL, in ddH2O) and incubated at 60°C for 1 h, followed by 10 min inactivation of enzyme at 94°C. The antibody genes were amplified by PCR and cloned into linearized pFuse vector by In-Fusion Cloning Kit (Vazyme, C112) for antibody secretion and purification. IgG secretion and purification The full-length IgG expression vector was transfected into HEK293F cells using PEI (polyscience, 24885) and cultured in freestyle 293 expression medium (Gibco. 12338026) for overexpressing the full-length IgG. The supernatant was harvested after 5 days. IgG was captured by Protein A resin (Genscript, L00210), eluted by sodium citrate (pH 3.5) and neutralized with Tris-HCl (pH 8.0) immediately. The eluted protein was buffer exchanged into PBS with 30 kDa Ultracel Amicon (Merck Millipore, UFC803096). ELISA assay The wells of a microtiter plate were coated with CD40 or BSA as control at 4°C overnight. The coated wells were blocked by 5% milk in PBS at 37°C for 1 h. Then blocking buffer was removed and the antibodies were added and incubated in 37°C for 1 h. After 8 times washes, goat anti-human IgG-HRP (SouthernBiotech, 2048-05) was added and incubated at 37°C for 30 min. Wells were washed for 8 times and ABTS substrate solution (Thermofisher, 2024) were added and plate was incubated at room temperature for 20 min. The OD was evaluated at 405 nm with a plate reader. Cell binding assay and fluorescence statistical analysis The hCD40 overexpressing cells were incubated with a serial dilutes of full-length antibodies at room temperature for 50 min. Then the cells were stained with AF647-conjugated Goat anti Human IgG (H + L) (Thermofisher, A21445) at room temperature for 40 min and analyzed by flow cytometry. Flow cytometry results were analyzed with software Flowjo X version 10.0.7. The MFI of cells was plotted against the antibody concentrations using software GraphPad Prism version 7.0.0. Surface plasmon resonance analysis The affinity of IgG antibody was determined by BIAcore S200 (GE Healthcare). Running buffer (HBS-EP+, GE Healthcare, BR-1006-69) diluted IgG antibodies were immobilized by a Series S Protein A sensor chip (GE Healthcare, 29-1275-55) at 10 μL/min for 30 s. Serial diluted antigen CD40 was injected onto the sensor chip loading at 30 μL/min for 120 s, then dissociated at 30 μL/min for 240 s. Sensor chip was regenerated by Glycine 1.5 (GE Healthcare, BR-1003-54), at 30 μL/min for 20 s. Results were simulated and calculated by BIAevaluation software S200. CD40/NF-κB Reporter cell line The NF-κB GFP reporter construct is stably integrated into the genome of Jurkat cell line. The NF-κB Reporter cell line was then infected with lentivirus expressing full-length human CD40. The GFP gene is controlled by four tandem copies of NF-κB response element located upstream of the TATA promoter. Following activation by human CD40 ligand or agonist antibody, NF-κB transcription factor binds to the DNA response elements to induce transcription of GFP. The reporter cell line was incubated with antibody in presence of an anti-human Fc crosslinker antibody (SouthernBiotech, 2048-01) overnight. GFP expression was detected with flow cytometry. Flow cytometry results were analyzed with software Flowjo X version 10.0.7. The MFI of cells was plotted against the antibody concentrations and EC50 was calculated using software GraphPad Prism version 7.0.0. Results Reformatting of individual antibody using one step emulsion PCR During the progress of reformatting from scFv library to full-length IgG library, maintaining the original pairing of VH and VL is the most critical requirement. Here, taken a TpoR antibody as proof of principle study, we developed one step strategy to reformat from scFv to IgG in one complex emulsion PCR reaction, to avoid intrinsic bias of genetic manipulation by multiple steps. As shown in Fig. 1, pool of scFv DNA molecules are compartmentalized in the aqueous droplets of water-in-oil (w/o) emulsion such that each droplet contains a single template DNA molecule, preventing the formation of wrong pairing of VH–VL by PCR amplification. scFv was reformatted to IgG, consisting of (from 5’ to 3’) part of the IL2 leader sequence, VL, CL, T2A or IRES, IL2 leader sequence, VH, and N-terminus of CH sequence. At first, variable light and heavy chains from the same scFv were amplified separately with primer sets specific to VH and VL frameworks. Secondly, the amplified variable light and heavy chain and bridging fragment consisting of CL, T2A or IRES and IL2 leader sequence were assembled; thirdly, the assembled intermediate product was used as template for PCR with VL FR1 and VH FR4-specific primer sets to facilitate cloning into mammalian display vector as shown in Fig. 1. To display antibody on mammalian cell surface, a transmembrane domain (TM) of PDGFR has been fused in frame to the C-terminus of IgG constant region. Moreover, we introduced self-cleaving T2A peptide or IRES to co-express light chain and heavy chain. IRES is widely used in mammalian display of full-length IgG (Zhou et al., 2010; Zhou and Shen, 2012; Tomimatsu et al., 2013). We proposed that self-cleaving 2A peptide could be a good candidate to replace IRES because of its small size and stoichiometric expression of multiple proteins flanking the 2A peptide (Fang et al., 2005; Chng et al., 2015). We tested the one-step emulsion PCR based reformatting method with a TpoR antibody named 3D9 (Zhang et al., 2013). Both T2A and IRES mediated full-length IgG format of 3D9 were successfully transformed from scFv template, 3D9, which were confirmed by Sanger sequencing. Binding of cell surface IgG to TpoR protein were verified by flow cytometry. T2A and IRES mediated full-length IgG can be displayed at similar level (Fig. 2A). Displayed IgG mediated by T2A and IRES retained specific binding to TpoR protein with Kd of 28.13 nM and 22.39 nM, respectively (Fig. 2B). Considering the short length of T2A peptide, we used T2A thereafter. Fig. 2 View largeDownload slide Mammalian display of T2A or IRES mediated IgG. TpoR antibody 3D9 was reformatted from scFv to T2A or IRES mediated IgG format. (A) HEK293T were infected with lentivirus and selected by puromycin to display T2A (upper) or IRES (lower) mediated IgG format of 3D9. Cells infected with control lentivirus pCDH were used as negative control. The cells were incubated with eight different concentrations of biotinylated TpoR and SA-PE to determine antigen binding ability and anti-human IgG Fc-AF 647 to indicate antibody display level. The representative flow cytometry graphs of cell staining were shown. (B) The plot of TpoR concentration versus the mean fluorescence intensity (MFI) of biotin-TpoR/SA-PE staining. The Kd values of displayed IgGs were calculated. Fig. 2 View largeDownload slide Mammalian display of T2A or IRES mediated IgG. TpoR antibody 3D9 was reformatted from scFv to T2A or IRES mediated IgG format. (A) HEK293T were infected with lentivirus and selected by puromycin to display T2A (upper) or IRES (lower) mediated IgG format of 3D9. Cells infected with control lentivirus pCDH were used as negative control. The cells were incubated with eight different concentrations of biotinylated TpoR and SA-PE to determine antigen binding ability and anti-human IgG Fc-AF 647 to indicate antibody display level. The representative flow cytometry graphs of cell staining were shown. (B) The plot of TpoR concentration versus the mean fluorescence intensity (MFI) of biotin-TpoR/SA-PE staining. The Kd values of displayed IgGs were calculated. Reformatting of scFv mini-library to IgG To maximize the reaction specificity, it should be tightly controlled to limit no more than one template molecule per droplet. We first determined the amount of template by performing PCR with a dilution series of template DNA. The maximum amplification yield was observed with 1010 to 1012 templates indicating that each droplet host at least one template DNA molecules and some droplets host more than one template (Supplemental Fig. S1). Less than one-third of the maximum DNA yield was obtained with 109 template molecules. Based on Poisson distribution, nearly each droplet hosts 0 or 1 copies of template DNA, thus this amount of template in w/o emulsion PCR could prevent wrong pairing of VH and VL from different templates and the formation of chimeric amplification by-products. To evaluate whether the correct VH–VL pairing can be maintained, a ‘mini-library’ comprising 16 individual scFvs of known sequences was reformatted to IgG library and IgG sequences of 50 clones from either emulsion PCR or conventional PCR were analyzed by Sanger sequencing (Table I). In emulsion PCR, 88% of the IgG sequences had correct VH–VL pairing while the remaining IgG representing antibodies with mismatched VH and VL. In conventional PCR only 6% of converted IgG maintained the correct VH–VL pairing. 86% of IgG had promiscuous VH–VL pairing and the remaining 8% are chimeric product of different templates. 3 and 15 original scFv sequences were represented in the 3 and 44 correctly pairing clones in the conventional PCR and emulsion PCR, respectively. Among the 15 original sequences in the emulsion PCR, 3 sequences appeared once, 4 sequences appeared twice, 3 sequences appeared 3 times, 2 appeared 4 times, 2 appeared 5 times and one appeared 6 times (Supplemental Table S2). Table I. Maintenance of VH and VL pairing during batch reformatting of a mini-library of 16 scFv to IgG by conventional PCR or emulsion PCR Correct pairing Wrong pairing Recombination product Conventional PCR 3/50 43/50 4/50 Emulsion PCR 44/50 6/50 0/50 Correct pairing Wrong pairing Recombination product Conventional PCR 3/50 43/50 4/50 Emulsion PCR 44/50 6/50 0/50 50 single clones of reformatted IgG using either conventional PCR or emulsion PCR were sequenced by Sanger sequencing and number of correct pairing of VH and VL, wrong pairing of VH and VL and formation of chimeric product was listed in table. Table I. Maintenance of VH and VL pairing during batch reformatting of a mini-library of 16 scFv to IgG by conventional PCR or emulsion PCR Correct pairing Wrong pairing Recombination product Conventional PCR 3/50 43/50 4/50 Emulsion PCR 44/50 6/50 0/50 Correct pairing Wrong pairing Recombination product Conventional PCR 3/50 43/50 4/50 Emulsion PCR 44/50 6/50 0/50 50 single clones of reformatted IgG using either conventional PCR or emulsion PCR were sequenced by Sanger sequencing and number of correct pairing of VH and VL, wrong pairing of VH and VL and formation of chimeric product was listed in table. Reformatting phage display derived scFv repertoire into IgG library As the one-step emulsion PCR based reformatting strategy was validated using the mini-library of 16 scFvs, we next reformatted a scFv repertoire of phage display selection to IgG library. CD40 is a promising target for cancer immunotherapy and several drugs that target the CD40 pathway have undergone phase I and II clinical trial (Moran et al., 2013; Hassan et al., 2014). Here, we performed phage display screening of naïve scFv library against CD40 ectodomain Fc fusion protein and obtained 1200-fold enrichment after three rounds of biopanning (Supplemental Table S3). Five out of twenty-nine phage clones from the third round output bound to CD40 ectodomain by phage ELISA, among which there were two unique sequences (Supplemental Fig. S2). We thus reformatted the scFv repertoire of the third round of phage display to a full-length IgG library with capacity of 2 × 104. We reformatted the IgG library after the third round of biopanning for two reasons. Firstly, there was excess of nonspecific antibody at earlier rounds as indicated by the phage ELISA result. Secondly, the size of the mammalian display library was often small. The total number of cells that could be screened by fluorescence activating cell sorter (FACS) was also limited. On the premise of bigger mammalian display library can be conveniently constructed and handled, it would be better to reformat the IgG library at earlier rounds. The technology of next-generation sequencing (NGS) can provide millions of sequences of libraries and provide deep insight into the original scFv library and the reformatted full-length IgG library. NGS libraries were constructed via emulsion PCR to avoid the PCR bias during the amplification (van Dijk et al., 2014). The diversity of the third round phage library output and the reformatted library were listed in Supplemental Table S4. The reformatted IgG library can cover about 75% of the output clones in the third round phage output. We compared the ranks of the top 20 HCDR3 and LCDR3 sequences in original scFv library to the reformatted library as well and found the good correlation between two libraries (Fig. 3A, the Pearson correlation value >0.85 and the P value < =0.001). We also observed similar distributions of CDR3 lengths (Fig. 3B) and germline families (Supplemental Fig. S3) between the scFv phagemid library and IgG library. The NGS analysis indicated that the diversity and relative abundance of antibody in the original scFv library were preserved in the reformatted full-length IgG library. Fig. 3 View largeDownload slide NGS analysis of original scFv library and reformatted full-length IgG library. (A) The correlation of the ranks of top 20 CDR3 sequences of original library compared to that of the reformatted full-length IgG library. (B) The distributions and frequencies of total sequences of the original scFv library and reformatted full-length IgG library, according to the length of the CDR3 of heavy chain(H-CDR3) or light chain(L-CDR3). Fig. 3 View largeDownload slide NGS analysis of original scFv library and reformatted full-length IgG library. (A) The correlation of the ranks of top 20 CDR3 sequences of original library compared to that of the reformatted full-length IgG library. (B) The distributions and frequencies of total sequences of the original scFv library and reformatted full-length IgG library, according to the length of the CDR3 of heavy chain(H-CDR3) or light chain(L-CDR3). Screening CD40 binding IgG from reformatted mammalian display IgG library We cloned the full-length IgG library into our modified mammalian cell display vector derived from the third generation lentivirus-based plasmid. The mammalian display IgG library construct was cotransfected with plasmids encoding packaging and envelop proteins to generate replication-incompetent lentivirus library. HEK293T were infected at a low multiplicity of infection (MOI) to ascertain the expression of a single antibody per infected cell. The cells were incubated with biotinylated CD40 antigen and then stained with Streptavidin-PE to indicate the affinity of antibodies (x-axis in Fig. 4) and goat anti-human IgG Fc-Alexa Fluor 488 or 647 to indicate the expression of antibodies (y-axis in Fig. 4). Approximately 1–2% cells were stained positively for biotin-CD40 in the first round of sorting. The cells located upper right of the graph were sorted and expanded for the next round of selection. Fig. 4 View largeDownload slide Isolation of CD40 binding antibody from reformatted IgG library. HEK293T cells were infected at a low MOI with T2A mediated IgG expression lentivirus library reformatted from a phage display scFv library. The infected cells were incubated with biotinylated CD40(lower) or irrelevant protein as negative control(upper) and then stained with SA-PE and anti-Human Fc-AF 647 or anti-Human Fc-AF 488 in alternative rounds. Gated cells were sorted and cultured for the next round of screening. The process was iterated for three rounds. Single cells of the third round were sorted into 96-well plates for further analysis. Gates for sorting are indicated by the quadrangle in the upper right corner of graph. Fig. 4 View largeDownload slide Isolation of CD40 binding antibody from reformatted IgG library. HEK293T cells were infected at a low MOI with T2A mediated IgG expression lentivirus library reformatted from a phage display scFv library. The infected cells were incubated with biotinylated CD40(lower) or irrelevant protein as negative control(upper) and then stained with SA-PE and anti-Human Fc-AF 647 or anti-Human Fc-AF 488 in alternative rounds. Gated cells were sorted and cultured for the next round of screening. The process was iterated for three rounds. Single cells of the third round were sorted into 96-well plates for further analysis. Gates for sorting are indicated by the quadrangle in the upper right corner of graph. Seventeen percent of scFvs could bind to CD40 after three rounds of biopanning (Supplemental Fig. S2). However, upon reformatting, a portion of antibodies may lose the affinity due to the conformational alteration from scFv to IgG. Thus, the reformatting is a crucial step to eliminate the false positive binders in the scFv form. The concentration of biotin-CD40 used in FACS sorting was low to favor the selection of antibody with high affinity, which also account for the low percentage of positive cells in the first round of sorting. Roundwise enrichment of positively staining cell was obtained during three rounds of selection (Fig. 4). Following the last round of selection, the cells with the highest antigen binding affinity were sorted into 96 well plates by single cell sorting for further analysis. Identification and analysis of single clones After 2–3 weeks the sorted individual clones of cells expanded to 70–90% confluency in wells for high throughput flow cytometry analysis of binding to CD40. 27 out of 30 clones exhibited strong binding to CD40. The antibody genes from the positive staining cells were amplified and subjected to Sanger sequencing. Among the sorted and positive IgG clones, there were 14 clones of Ab01, 11 clones of Ab02 and 2 clones of Ab03 (Fig. 5A). They were ranked the third, fourth and eighth in the reformatted libraries according to NGS result. Fig. 5 View largeDownload slide Identification of IgGs obtained via mammalian display. (A) Flow cytometry analysis of IgG of sorted clones on cell surface. Cells were incubated with biotinylated CD40, stained with SA-PE and anti-Human Fc-AF 647 and analyzed by flow cytometry. FACS analysis results of three representative single clones were shown here. (B) ELISA binding assay using purified antibody Ab01, Ab02, Ab03 and an irrelevant antibody HEL. Binding activity of different antibodies was compared by the ratio of the absorbance at 405 nm of CD40 to BSA. (C) The direct binding activity determination of Ab01, Ab02, Ab03 and HEL to hCD40. The binding activity to hCD40 was determined by the mean fluorescence intensity (MFI) quantified via flow cytometry analysis. (D) The affinity of Ab01, Ab02, Ab03 to CD40 measured using Biacore. Different concentration of CD40 were injected onto the antibody immobilized Protein A sensor chip and data presented as an overlay plot aligned at the start of the injection. Fig. 5 View largeDownload slide Identification of IgGs obtained via mammalian display. (A) Flow cytometry analysis of IgG of sorted clones on cell surface. Cells were incubated with biotinylated CD40, stained with SA-PE and anti-Human Fc-AF 647 and analyzed by flow cytometry. FACS analysis results of three representative single clones were shown here. (B) ELISA binding assay using purified antibody Ab01, Ab02, Ab03 and an irrelevant antibody HEL. Binding activity of different antibodies was compared by the ratio of the absorbance at 405 nm of CD40 to BSA. (C) The direct binding activity determination of Ab01, Ab02, Ab03 and HEL to hCD40. The binding activity to hCD40 was determined by the mean fluorescence intensity (MFI) quantified via flow cytometry analysis. (D) The affinity of Ab01, Ab02, Ab03 to CD40 measured using Biacore. Different concentration of CD40 were injected onto the antibody immobilized Protein A sensor chip and data presented as an overlay plot aligned at the start of the injection. Soluble antibodies were expressed and purified using a protein A affinity column and their binding to CD40 was first evaluated by ELISA. In comparison to the binding to BSA as a control, two antibodies Ab01 and Ab03 exhibited strong and specific binding to CD40 and Ab02 showed moderate binding activity to CD40 (Fig. 5B). Measuring antibody and soluble hCD40 ectodomain interaction in solution, away from receptor’s cellular environment, may not provide an affinity value applicable in vivo. Cell surface receptor hCD40 were stained with different concentrations of full-length IgGs and fluorescence secondary antibody for detection. It is demonstrated antigen binding ability of cell surface displayed antibody is positively correlated with its soluble counterpart (Fig. 5C). For binding to either plate immobilized or cell surface CD40, the apparent affinity is affected by the avidity effect. To avoid complications caused by the combined affinity resulting from multivalent binding, the bivalent IgG was immobilized on chip and then the monovalent CD40 protein was added when we measured the binding using Biacore. The affinity of Ab01, Ab02 and Ab03 were 439 nM, 3.89 μM and 287 nM, respectively (Fig. 5D, Table II). Table II. The kinetic profiles of the IgGs against CD40 Antibody KD(M) Ka(1/Ms) Kd(1/s) Ab01 4.39e–07 9.14e+04 4.01e–02 Ab02 3.89e–06 4.01e+02 1.56e–03 Ab03 2.87e–07 3.59e+04 1.03e–02 Antibody KD(M) Ka(1/Ms) Kd(1/s) Ab01 4.39e–07 9.14e+04 4.01e–02 Ab02 3.89e–06 4.01e+02 1.56e–03 Ab03 2.87e–07 3.59e+04 1.03e–02 Table II. The kinetic profiles of the IgGs against CD40 Antibody KD(M) Ka(1/Ms) Kd(1/s) Ab01 4.39e–07 9.14e+04 4.01e–02 Ab02 3.89e–06 4.01e+02 1.56e–03 Ab03 2.87e–07 3.59e+04 1.03e–02 Antibody KD(M) Ka(1/Ms) Kd(1/s) Ab01 4.39e–07 9.14e+04 4.01e–02 Ab02 3.89e–06 4.01e+02 1.56e–03 Ab03 2.87e–07 3.59e+04 1.03e–02 We cannot exclude the possibility of the presence of binders with higher affinities than the identified ones. The reformatted library covered about 75% of the diversity of the phage display output. In addition, only 30 clones of the sorted cells were characterized to demonstrate the feasibility of the method. It is likely that rare, high-affinity binders could be found when more clones of sorted cells were sampled. The activity of these antibodies was further tested by the CD40/NF-κB reporter cell line. Ab01 (EC50 = 3.978 nM) and Ab03 (EC50 = 11.34 nM) exhibited potent agonist activity while Ab02 had no agonist activity measured using the reporter assay (Fig. 6). The agonist activity describes the functional feature of the IgG and is dependent on the epitope of the target that IgG binds to. Fig. 6 View largeDownload slide The biological activity determination of Ab01, Ab02, Ab03 and HEL. hCD40 reporter cell lines were stimulated with antibodies and anti-human-IgG-Fc to crosslink the antibodies, and the value of MFI was used to indicate the activation. Fig. 6 View largeDownload slide The biological activity determination of Ab01, Ab02, Ab03 and HEL. hCD40 reporter cell lines were stimulated with antibodies and anti-human-IgG-Fc to crosslink the antibodies, and the value of MFI was used to indicate the activation. Discussion Phage display is a powerful and versatile technology in the directed evolution of new proteins, particularly for the production of antibody therapeutics (Huse et al., 1989; McCafferty et al., 1990; Barbas et al., 1991; Winter et al., 1994; Smith and Petrenko, 1997). Phage display has led to discovery of antibodies that can neutralize toxins, counteract autoimmune diseases and cure cancer. scFv is well suited for phage display and bacterial expression. However, loss of affinity during conversion from scFv into IgG, which is in most case the final therapeutic format, are frequently observed (Qin and Li, 2014; Steinwand et al., 2014). scFv format often has an enhanced avidity than its Fab counterpart due to formation of dimer, i.e, scFv with short linker between VH and VL can form diabody. A thorough understanding of structural difference between the scFv fragment and its Fab counterpart would resolved by crystallography. Reformatting of phage scFv to IgG allows earlier selection of effective IgGs. The traditional method is to convert scFv to full-length IgG on an individual basis, however, it is not practical for conversion of large number of scFv in a library. Xiao et al. resolved the problem by using inverse PCR to maintain the VH and VL pairing during the population reformatting of phage display libraries (Xiao et al., 2017). Here we proposed an even more straightforward and flexible one-step method by compartmentalization of the scFv-to-IgG conversion OE-PCR in millions of droplets in water-in-oil emulsion. The essential and unique aspects of our approach is that different parts of IgG are spliced in one-pot reaction of OE-PCR and each droplet host only one template in emulsion PCR to ensure correct VH–VL pairing is maintained with high fidelity. In addition to quantitative capability of flow cytometry to discriminate strong and weak binders, mammalian cell display provides several advantages with the ability to co-select antibody for high expression level in mammalian cells, native folding and biophysical properties appropriate for drug development (Beerli et al., 2008; Forsyth et al., 2013; Qin and Li, 2014). Limitation of mammalian cell-based systems, however, has been the smaller library complexity compared to that of phage or yeast display approaches. Thus, phage display and mammalian cell display are complementary technologies. Our one step emulsion PCR based method enable the seamless integration of two technologies into one platform. The antibody isolated by mammalian display need to be expressed as soluble protein for further characterization such as ELISA and function bioassay. To obtain antibodies in soluble form, the genes that encode the antibodies need to be recovered from the mammalian display vector and re-cloned into a new expression vector which contains no transmembrane domain. Bo et al. developed antibody membrane switch (AMS) technology to avoid this step (Yu et al., 2014). The switch between membrane anchored and secretion form is achieved by alternative splicing and specific DNA recombination to facilitate the transformation of the antibody from membrane to secretion format. We proposed that the LoxP sites can be placed flanking a PDGFR-TM domain so that we can knockout PDGFR-TM domain by inducing the expression of Cre recombinase. Yang et al. observed that significant potency loss happened when this scFv-Fc was transformed to full-length IgG form (Yang et al., 2018). Therefore, it is desirable to select antibody of the final format at early stage of drug discovery. Notably, the emulsion PCR based reformatting strategy is flexible to the choice of final antibody format and allows the introduction of various mutations to the Fc of the IgG vector through appropriate design of primers and bridging fragments. For example, human IgG2 imparts FcγR-independent agonistic activity to immune-stimulatory mAbs such as anti-CD40 agonist (Li and Ravetch, 2011; White et al., 2011). The phage display derived scFv library can be easily reformatted to IgG2 library using this approach for the selection of superagonistic immunostimulatory anticancer antibodies. In summary, we developed a method to reformat scFv to full-length IgG based on one-step emulsion PCR and full-length IgG of high affinity can be efficiently isolated when the method is coupled with mammalian display (Fig. 7). The advantages of the method are: (1) it enables the high throughput reformatting of scFv to full-length IgG and maintenance of the diversity and variable region pairing; (2) it is a simplified one-step procedure; (3) it is applicable to any format of antibody. This method provides a user-friendly platform for more effective mining of phage display scFv libraries for the functional antibody discovery. Fig. 7 View largeDownload slide The workflow of our strategy to reformat scFv library to IgG library for mammalian display. ScFv library derived from phage display selection performed against target were reformatted to full length IgG library using our one-step emulsion PCR approach and cloned into lentivirus-based mammalian surface display vector. Lentivirus library was prepared and infection of HEK293T cells at a low MOI generates a pool of infected cells, each expressing at their surface one specific antibody. Cells were stained with antigen and then positively stained cells were sorted by flow cytometry. Antibody genes were extracted from positive cell clones and soluble antibody was expressed and purified for further characterization. Fig. 7 View largeDownload slide The workflow of our strategy to reformat scFv library to IgG library for mammalian display. ScFv library derived from phage display selection performed against target were reformatted to full length IgG library using our one-step emulsion PCR approach and cloned into lentivirus-based mammalian surface display vector. Lentivirus library was prepared and infection of HEK293T cells at a low MOI generates a pool of infected cells, each expressing at their surface one specific antibody. Cells were stained with antigen and then positively stained cells were sorted by flow cytometry. Antibody genes were extracted from positive cell clones and soluble antibody was expressed and purified for further characterization. Funding This work was supported by the National Key R&D Program of China under Grant number 2017YFA0504801; National Natural Science Foundation of China under Grant number 81661148051, 81672010; National Natural Science Foundation of China under Grant number 81872787, 81603010; National Natural Science Foundation of China under Grant number 31500632, 31700807. Conflict of interest No potential conflicts of interest were disclosed. References Barbas , C.F. , Kang , A.S. , Lerner , R.A. and Benkovic , S.J. ( 1991 ) Proc. Natl Acad. Sci. U S A , 88 , 7978 – 7982 . doi:10.1073/pnas.88.18.7978 . Crossref Search ADS PubMed Beerli , R.R. , Bauer , M. , Buser , R.B. , Gwerder , M. , Muntwiler , S. , Maurer , P. , Saudan , P. and Bachmann , M.F. ( 2008 ) Proc. Natl. Acad. Sci. U S A , 105 , 14336 – 14341 . doi:10.1073/pnas.0805942105 . Crossref Search ADS PubMed Bolotin , D.A. , Poslavsky , S. , Mitrophanov , I. , Shugay , M. , Mamedov , I.Z. , Putintseva , E.V. and Chudakov , D.M. ( 2015 ) Nat. Methods , 12 , 380 – 381 . doi:10.1038/nmeth.3364 . Crossref Search ADS PubMed Chen , C.G. , Fabri , L.J. , Wilson , M.J. and Panousis , C. ( 2014 ) Nucleic Acids Res. , 42 , e26 . doi:10.1093/nar/gkt1142 . Crossref Search ADS PubMed Chng , J. , Wang , T. , Nian , R. , Lau , A. , Hoi , K.M. , Ho , S.C. , Gagnon , P. , Bi , X. and Yang , Y. ( 2015 ) mAbs , 7 , 403 – 412 . doi:10.1080/19420862.2015.1008351 . Crossref Search ADS PubMed Fang , J. , Qian , J.J. , Yi , S. , Harding , T.C. , Tu , G.H. , VanRoey , M. and Jooss , K. ( 2005 ) Nat. Biotechnol. , 23 , 584 – 590 . doi:10.1038/nbt1087 . Crossref Search ADS PubMed Forsyth , C.M. , Juan , V. , Akamatsu , Y. , DuBridge , R.B. , Doan , M. , Ivanov , A.V. , Ma , Z. , Polakoff , D. , Razo , J. , Wilson , K. et al. . ( 2013 ) MAbs , 5 , 523 – 532 . doi:10.4161/mabs.24979 . Crossref Search ADS PubMed Hassan , S.B. , Sorensen , J.F. , Olsen , B.N. and Pedersen , A.E. ( 2014 ) Immunopharmacol. Immunotoxicol. , 36 , 96 – 104 . doi:10.3109/08923973.2014.890626 . Crossref Search ADS PubMed Huse , W.D. , Sastry , L. , Iverson , S.A. , Kang , A.S. , Alting-Mees , M. , Burton , D.R. , Benkovic , S.J. and Lerner , R.A. ( 1989 ) Science , 246 , 1275 – 1281 . doi:10.1126/science.2531466 . Crossref Search ADS PubMed Huston , J.S. , Levinson , D. , Mudgetthunter , M. , Tai , M.S. , Novotný , J. , Margolies , M.N. , Ridge , R.J. , Bruccoleri , R.E. , Haber , E. and Crea , R. ( 1988 ) Proc. Natl. Acad. Sci. , 85 , 5879 – 5883 . doi:10.1073/pnas.85.16.5879 . Crossref Search ADS Jostock , T. , Vanhove , M. , Brepoels , E. , Van Gool , R. , Daukandt , M. , Wehnert , A. , Van Hegelsom , R. , Dransfield , D. , Sexton , D. , Devlin , M. et al. . ( 2004 ) J. Immunol. Methods , 289 , 65 – 80 . doi:10.1016/j.jim.2004.03.014 . Crossref Search ADS PubMed Kotera , I. and Nagai , T. ( 2008 ) J. Biotechnol. , 137 , 1 – 7 . doi:10.1016/j.jbiotec.2008.07.1816 . Crossref Search ADS PubMed Li , F. and Ravetch , J.V. ( 2011 ) Science , 333 , 1030 – 1034 . doi:10.1126/science.1206954 . Crossref Search ADS PubMed McCafferty , J. , Griffiths , A.D. , Winter , G. and Chiswell , D.J. ( 1990 ) Nature , 348 , 552 – 554 . doi:10.1038/348552a0 . Crossref Search ADS PubMed Moran , A.E. , Kovacsovics-Bankowski , M. and Weinberg , A.D. ( 2013 ) Curr. Opin. Immunol. , 25 , 230 – 237 . doi:10.1016/j.coi.2013.01.004 . Crossref Search ADS PubMed Qin , C.F. and Li , G.C. ( 2014 ) Int. Immunopharmacol. , 23 , 380 – 386 . doi:10.1016/j.intimp.2014.09.017 . Crossref Search ADS PubMed Sanmark , H. , Huovinen , T. , Matikka , T. , Pettersson , T. , Lahti , M. and Lamminmaki , U. ( 2015 ) J. Immunol. Methods , 426 , 134 – 139 . doi:10.1016/j.jim.2015.08.005 . Crossref Search ADS PubMed Smith , G.P. and Petrenko , V.A. ( 1997 ) Chem. Rev. , 97 , 391 – 410 . doi:10.1021/cr960065d . Crossref Search ADS PubMed Steinwand , M. , Droste , P. , Frenzel , A. , Hust , M. , Dubel , S. and Schirrmann , T. ( 2014 ) MAbs , 6 , 204 – 218 . doi:10.4161/mabs.27227 . Crossref Search ADS PubMed Sun , Z. , Lu , S. , Yang , Z. , Li , J. and Zhang , M.Y. ( 2017 ) Virus Res. , 238 , 156 – 163 . doi:10.1016/j.virusres.2017.06.018 . Crossref Search ADS PubMed Tomimatsu , K. , Matsumoto , S.E. , Tanaka , H. , Yamashita , M. , Nakanishi , H. , Teruya , K. , Kazuno , S. , Kinjo , T. , Hamasaki , T. , Kusumoto , K. et al. . ( 2013 ) Biochem. Biophys. Res. Commun. , 441 , 59 – 64 . doi:10.1016/j.bbrc.2013.10.007 . Crossref Search ADS PubMed Turchaninova , M.A. , Britanova , O.V. , Bolotin , D.A. , Shugay , M. , Putintseva , E.V. , Staroverov , D.B. , Sharonov , G. , Shcherbo , D. , Zvyagin , I.V. , Mamedov , I.Z. et al. . ( 2013 ) Eur. J. Immunol. , 43 , 2507 – 2515 . doi:10.1002/eji.201343453 . Crossref Search ADS PubMed van Dijk , E.L. , Jaszczyszyn , Y. and Thermes , C. ( 2014 ) Exp. Cell Res. , 322 , 12 – 20 . doi:10.1016/j.yexcr.2014.01.008 . Crossref Search ADS PubMed Vaughan , T.J. , Williams , A.J. , Pritchard , K. , Osbourn , J.K. , Pope , A.R. , Earnshaw , J.C. , McCafferty , J. , Hodits , R.A. , Wilton , J. and Johnson , K.S.J.N.b ( 1996 ) Nat. Biotechnol. , 14 , 309 – 314 . Crossref Search ADS PubMed White , A.L. , Chan , H.T. , Roghanian , A. , French , R.R. , Mockridge , C.I. , Tutt , A.L. , Dixon , S.V. , Ajona , D. , Verbeek , J.S. , Al-Shamkhani , A. et al. . ( 2011 ) J. Immunol. , 187 , 1754 – 1763 . doi:10.4049/jimmunol.1101135 . Crossref Search ADS PubMed Williams , R. , Peisajovich , S.G. , Miller , O.J. , Magdassi , S. , Tawfik , D.S. and Griffiths , A.D. ( 2006 ) Nat. Methods , 3 , 545 – 550 . doi:10.1038/nmeth896 . Crossref Search ADS PubMed Winter , G. , Griffiths , A.D. , Hawkins , R.E. and Hoogenboom , H.R. ( 1994 ) Annu. Rev. Immunol. , 12 , 433 – 455 . doi:10.1146/annurev.iy.12.040194.002245 . Crossref Search ADS PubMed Xiao , X. , Douthwaite , J.A. , Chen , Y. , Kemp , B. , Kidd , S. , Percival-Alwyn , J. , Smith , A. , Goode , K. , Swerdlow , B. , Lowe , D. et al. . ( 2017 ) mAbs , 9 , 996 – 1006 . doi:10.1080/19420862.2017.1337617 . Crossref Search ADS PubMed Yang , Z. , Du , M. , Wang , W. , Xin , X. , Ma , P. , Zhang , H. and Lerner , R.A. ( 2018 ) Protein Eng. Des. Sel . doi:10.1093/protein/gzy002 . Yu , B. , Wages , J.M. and Larrick , J.W. ( 2014 ) Protein Eng. Des. Sel. , 27 , 309 – 315 . doi:10.1093/protein/gzu039 . Crossref Search ADS PubMed Zhang , H. , Yea , K. , Xie , J. , Ruiz , D. , Wilson , I.A. and Lerner , R.A. ( 2013 ) Chem. Biol. , 20 , 734 – 741 . doi:10.1016/j.chembiol.2013.04.012 . Crossref Search ADS PubMed Zhou , C. , Jacobsen , F.W. , Cai , L. , Chen , Q. and Shen , W.D. ( 2010 ) mAbs , 2 , 508 – 518 . doi:10.4161/mabs.2.5.12970 . Crossref Search ADS PubMed Zhou , C. and Shen , W.D. ( 2012 ) Methods Mol. Biol. , 907 , 293 – 302 . doi:10.1007/978-1-61779-974-7_17 . Crossref Search ADS PubMed © The Author(s) 2019. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Protein Engineering, Design and Selection Oxford University Press

High-throughput reformatting of phage-displayed antibody fragments to IgGs by one-step emulsion PCR

Loading next page...
 
/lp/oxford-university-press/high-throughput-reformatting-of-phage-displayed-antibody-fragments-to-0kykd04JGM
Publisher
Oxford University Press
Copyright
© The Author(s) 2019. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com
ISSN
1741-0126
eISSN
1741-0134
D.O.I.
10.1093/protein/gzz004
Publisher site
See Article on Publisher Site

Abstract

Abstract Single-chain variable fragment (scFv) is the most common format for phage display antibody library. The isolated scFvs need to be reformatted to full-length IgGs for further characterization. High throughput reformatting of scFv to IgG without disrupting VH–VL pairing is of great demanding for exhaustive screening of all antibodies in IgG format. Herein, we developed a strategy based on the overlap extension PCR in emulsion to reformat scFv to IgG while maintain the accuracy and complexity of variable region pairing. Using CD40 as an example target, we reformatted phage display derived CD40 binding scFv library to IgG mammalian display library and isolated high affinity CD40 binding IgGs. This robust and reliable antibody reformatting approach could be integrated into any phage display based antibody drug discovery. Introduction Antibody phage display is an essential technique of antibody-based drug industry and a crucial step to generate antibodies for a variety of clinical applications, such as treatment of metastatic cancer, autoimmune diseases and detoxification (Huse et al., 1989; McCafferty et al., 1990; Barbas et al., 1991). To date, the majority of therapeutic antibodies developed are full-length IgG molecules produced in mammalian cells. In this respect, phage display is limited to antibody fragments such as scFv (Huston et al., 1988). Thus, the scFv selected by phage display had to be converted to full-length IgG format. Many scFv format antibodies lost their activity when transformed into full-length IgGs (Steinwand et al., 2014). It is of great relevance to screen the antibody library in a full-length IgG format. The reformatting from scFv library to full-length IgG library without disrupting VH–VL pairing is a challenging task. Traditional conversion from scFv to IgG was performed on an individual basis (Jostock et al., 2004; Kotera and Nagai, 2008; Chen et al., 2014; Sanmark et al., 2015) which is a time consuming, labor intensive process and the number of antibodies assessed as full-length IgG is only a small fraction of the scFv repertoire. Chao-Guang Chen et al. reported a cloning method, referred to as insert-tagged (InTag) positive selection, for rapid reformatting of phage display antibody fragments to IgGs based on ligation-independent cloning (LIC) method (Chen et al., 2014). Hanna Sanmark et al. reported a modified version of method fully automatic single-tube recombination (FASTR), which enabled VH, VL, CH and CL DNA recombination in the correct order in a single step (Sanmark et al., 2015). However, these methods cannot reformat a library of scFv to IgG as they cannot avoid wrong VH–VL pairing (Sun et al., 2017). Xiaodong Xiao et al. first reported an inverse PCR based approach to reformat scFv library to IgG library without disrupting the variable region pairing (Xiao et al., 2017). Although inverse PCR can ensure the linkage between VH–VL pairing throughout the conversion process, a more straightforward approach to avoid wrong VH–VL pairing is to compartmentalize the reformatting reaction of each individual antibody. Emulsion PCR were developed and used in complex gene library amplification for its high-throughput and accuracy, so we adopted the technology to our strategy (Williams et al., 2006; Turchaninova et al., 2013). Emulsion PCR is based on the compartmentalization of genes into millions of aqueous droplets in a water-in-oil (w/o) emulsion. Templates are segregated in the minute aqueous droplets of the emulsion and amplified by PCR in isolation. Furthermore, emulsion PCR can alleviate the generation of chimeric PCR by-product arising from recombination between homologous regions of DNA. Here we described a straightforward and reliable approach to reformat scFv library to full-length IgG library in one step. The variable region of heavy chain and variable light chain of a single scFv molecule was amplified and full-length IgG was assembled by overlap extension PCR (OE-PCR) in aqueous droplet of water-in-oil emulsion. Coexpression of the assembled heavy chain and light chain was achieved by either T2A or Internal Ribosome Entry Site (IRES). Mammalian display is widely acknowledged as the best way to screen full-length antibodies (Zhou et al., 2010; Zhou and Shen, 2012; Tomimatsu et al., 2013). The full-length antibodies were displayed on mammalian cell surface and sorted by FACS. As a result, several high-affinity CD40 binding IgGs were isolated. To date, the technology presented here allows for more effective mining of phage display scFv libraries for the discovery of full-length IgG of high performance. Material and methods Phage display Human CD40 ectodomain protein (human IgG1 Fc tag, avi-Tag) was biotinylated using biotin protein ligase (GeneCopoeia, BI001). A human naïve scFv library made from PBMC of 30 healthy donors as previously described (Vaughan et al., 1996). The phage display scFv library was used for selections of antibodies against CD40-Fc fusion protein. Briefly, biotinylated CD40-Fc fusion protein was incubated with phage library for 2 h at RT and the phage-CD40 complex were captured by M-280 Streptavidin magnetic beads (Thermofisher, 11 205D). After washes the bound phages were eluted by 0.2 M Glycin-HCl (pH 2.2) for 10 min at RT and neutralized with 1 M Tris pH 8.0 to adjust the pH to 7.5. Then the eluted phages were amplified for next round of panning. Phagemid DNA was purified from the third round of phage display selection for the reformatting study. One step overlap extension emulsion PCR VL, bridge fragment and VH are assembled by the overlap extension PCR. The bridge fragment (CL-T2A/IRES) contains human CL fragment, T2A or IRES and IL-2 signal sequence. Emulsion PCR was performed with an Emulsion Kit (EURx Ltd. E3600). The water phase of the emulsion PCR system contains: optimized amount of template (2 ng, about 5 × 108 molecules of phagemid), 200 ng CL-T2A or 350 ng CL-IRES, 20 nM outer side primers (each), 400 nM inner side primers (each). The inner side primers are JH-RV and VL-FW. The outer side primers are VH-FW and JL-RV as indicated in Fig. 1. The inner side primers are more than the outer side primers to favor the amplification of overlap product. The sequences of primers are listed in Supplemental Table S1, Hot-Start Q5 DNA polymerase (New England Biolabs. M0493), 200 μM dNTP and 0.01 mg/mL acetylated BSA, then a 50 μL PCR system was mixed with 300 μL oil surfactant mixture prepared as described in the kit manual, and the emulsion was aliquoted into four suitable tubes for a thermal cycler. One step extension emulsion PCR program was performed as follows: 94°C 5 min for denaturation, 5×Cycle 1#(denaturation at 94°C for 30 s, annealing at 55°C for 30 s, extension at 72°C for 20 s) for amplifying variable regions, 25×Cycle 2#(denaturation at 94°C for 30 s, annealing at 55°C for 30 s, extension at 72°C for 55 s) for VL-bridge fragment(CL-T2A/IRES)-VH assembling and extension PCR, 72°C 15 min for the last long extension. The emulsion was broken with butanol and DNA was extracted from the opened emulsion by the column. Fig. 1 View largeDownload slide The schematic representation of emulsion PCR based reformatting of scFv to IgG. Individual scFv from a complex phage display library are compartmentalized in the aqueous droplets of a w/o emulsion PCR system such that each droplet contains a single scFv DNA molecule as PCR template. (A) VH chain and VL chain were amplified separately. (B and C) The VL fragment, CL-T2A/IRES-IL2 signal sequence fragment and VH fragment were assembled by overlap PCR. The PCR products were then cloned into the mammalian display vector by standard restriction digestion and ligation. Mammalian display antibody expression vectors consist of IL-2 signal sequence, the constant domain of IgG, and a PDGFR transmembrane domain. T2A or IRES was used for cocurrent expression of light chain and heavy chain of antibody. Fig. 1 View largeDownload slide The schematic representation of emulsion PCR based reformatting of scFv to IgG. Individual scFv from a complex phage display library are compartmentalized in the aqueous droplets of a w/o emulsion PCR system such that each droplet contains a single scFv DNA molecule as PCR template. (A) VH chain and VL chain were amplified separately. (B and C) The VL fragment, CL-T2A/IRES-IL2 signal sequence fragment and VH fragment were assembled by overlap PCR. The PCR products were then cloned into the mammalian display vector by standard restriction digestion and ligation. Mammalian display antibody expression vectors consist of IL-2 signal sequence, the constant domain of IgG, and a PDGFR transmembrane domain. T2A or IRES was used for cocurrent expression of light chain and heavy chain of antibody. Vector and library construction Lentivirus transfer vector pCDH was modified to be a mammalian display vector (Fig. 1). IL-2 signal sequence was added in front of the restriction enzyme site NotI. A NheI site was introduced into the N-terminal of human IgG CH1 which was synonymous mutation of Ala-Ser. The IgG was followed by a 218 linker and PDGFR transmembrane domain to facilitate the IgG displaying on cell surface. Products from one step overlap extension emulsion PCR were cloned into the plasmid by digestion with restriction enzymes NotI and NheI and ligation. NGS sample preparation and data analysis For next generation sequencing, VL and VH of the phagemid scFv library and reformatted IgG library were amplified by emulsion PCR. Individual barcode sequences were added to each sample during library preparation and samples were pooled and sequenced simultaneously during a single run by MiSeq PE250 (Illumina). The 250 bp forward read and the 250 bp reverse read data were processed with CLC Genomics Workbench and merged to assemble an entire heavy or light chain and sorted by the barcodes, then translated to protein sequences. CDR3 regions of heavy chains or light chains were extracted by MiXCR (Bolotin et al., 2015). Mammalian display and FACS sorting Mammalian display IgG library plasmid was cotransfected with plasmids encoding packaging and envelop proteins into HEK293T to generate replication-incompetent lentivirus library. After 48 h, lentiviruses were precipitated by LS buffer from kit (Biomiga, V2001) and titer was determined by Lenti-X p24 Rapid Titer Kit (Clontech, 632 200). HEK293T were infected at a low multiplicity of infection (MOI) to ascertain the expression of a single antibody per infected cell. Puromycin was added to kill cells that were not infected by lentivirus. After one week of puromycin selection cells were detached by accutase (Invitrogen, A1110501) and incubated with 20 nM biotin-CD40 at room temperature for 50 min. Then the cells were stained by 4 μg/mL SA-PE (Thermofisher, 21 627) and 2 μg/mL AF647-conjugated Goat anti Human IgG(H + L) (Thermofisher, A21445) at room temperature for 40 min and subjected for flow cytometric analysis and sorting. Isolation of CD40 binding antibody from sorted cell After three rounds of selection, single positive cells were sorted into 96-well plate and cultured in DMEM complemented with 20% FBS and 2× PS. When the single cell clone grew up to sufficient amount for analysis cells were detached by accutase (Invitrogen, A1110501) and stained for flow cytometry analysis using plate analysis method of BD LSRFortessa. The positive clones were lysed by Cell Lysis Buffer (50 mM KCl, 10 mM Tris-HCl, 0.1% Triton-100, Proteinase K at 16 units/mL, in ddH2O) and incubated at 60°C for 1 h, followed by 10 min inactivation of enzyme at 94°C. The antibody genes were amplified by PCR and cloned into linearized pFuse vector by In-Fusion Cloning Kit (Vazyme, C112) for antibody secretion and purification. IgG secretion and purification The full-length IgG expression vector was transfected into HEK293F cells using PEI (polyscience, 24885) and cultured in freestyle 293 expression medium (Gibco. 12338026) for overexpressing the full-length IgG. The supernatant was harvested after 5 days. IgG was captured by Protein A resin (Genscript, L00210), eluted by sodium citrate (pH 3.5) and neutralized with Tris-HCl (pH 8.0) immediately. The eluted protein was buffer exchanged into PBS with 30 kDa Ultracel Amicon (Merck Millipore, UFC803096). ELISA assay The wells of a microtiter plate were coated with CD40 or BSA as control at 4°C overnight. The coated wells were blocked by 5% milk in PBS at 37°C for 1 h. Then blocking buffer was removed and the antibodies were added and incubated in 37°C for 1 h. After 8 times washes, goat anti-human IgG-HRP (SouthernBiotech, 2048-05) was added and incubated at 37°C for 30 min. Wells were washed for 8 times and ABTS substrate solution (Thermofisher, 2024) were added and plate was incubated at room temperature for 20 min. The OD was evaluated at 405 nm with a plate reader. Cell binding assay and fluorescence statistical analysis The hCD40 overexpressing cells were incubated with a serial dilutes of full-length antibodies at room temperature for 50 min. Then the cells were stained with AF647-conjugated Goat anti Human IgG (H + L) (Thermofisher, A21445) at room temperature for 40 min and analyzed by flow cytometry. Flow cytometry results were analyzed with software Flowjo X version 10.0.7. The MFI of cells was plotted against the antibody concentrations using software GraphPad Prism version 7.0.0. Surface plasmon resonance analysis The affinity of IgG antibody was determined by BIAcore S200 (GE Healthcare). Running buffer (HBS-EP+, GE Healthcare, BR-1006-69) diluted IgG antibodies were immobilized by a Series S Protein A sensor chip (GE Healthcare, 29-1275-55) at 10 μL/min for 30 s. Serial diluted antigen CD40 was injected onto the sensor chip loading at 30 μL/min for 120 s, then dissociated at 30 μL/min for 240 s. Sensor chip was regenerated by Glycine 1.5 (GE Healthcare, BR-1003-54), at 30 μL/min for 20 s. Results were simulated and calculated by BIAevaluation software S200. CD40/NF-κB Reporter cell line The NF-κB GFP reporter construct is stably integrated into the genome of Jurkat cell line. The NF-κB Reporter cell line was then infected with lentivirus expressing full-length human CD40. The GFP gene is controlled by four tandem copies of NF-κB response element located upstream of the TATA promoter. Following activation by human CD40 ligand or agonist antibody, NF-κB transcription factor binds to the DNA response elements to induce transcription of GFP. The reporter cell line was incubated with antibody in presence of an anti-human Fc crosslinker antibody (SouthernBiotech, 2048-01) overnight. GFP expression was detected with flow cytometry. Flow cytometry results were analyzed with software Flowjo X version 10.0.7. The MFI of cells was plotted against the antibody concentrations and EC50 was calculated using software GraphPad Prism version 7.0.0. Results Reformatting of individual antibody using one step emulsion PCR During the progress of reformatting from scFv library to full-length IgG library, maintaining the original pairing of VH and VL is the most critical requirement. Here, taken a TpoR antibody as proof of principle study, we developed one step strategy to reformat from scFv to IgG in one complex emulsion PCR reaction, to avoid intrinsic bias of genetic manipulation by multiple steps. As shown in Fig. 1, pool of scFv DNA molecules are compartmentalized in the aqueous droplets of water-in-oil (w/o) emulsion such that each droplet contains a single template DNA molecule, preventing the formation of wrong pairing of VH–VL by PCR amplification. scFv was reformatted to IgG, consisting of (from 5’ to 3’) part of the IL2 leader sequence, VL, CL, T2A or IRES, IL2 leader sequence, VH, and N-terminus of CH sequence. At first, variable light and heavy chains from the same scFv were amplified separately with primer sets specific to VH and VL frameworks. Secondly, the amplified variable light and heavy chain and bridging fragment consisting of CL, T2A or IRES and IL2 leader sequence were assembled; thirdly, the assembled intermediate product was used as template for PCR with VL FR1 and VH FR4-specific primer sets to facilitate cloning into mammalian display vector as shown in Fig. 1. To display antibody on mammalian cell surface, a transmembrane domain (TM) of PDGFR has been fused in frame to the C-terminus of IgG constant region. Moreover, we introduced self-cleaving T2A peptide or IRES to co-express light chain and heavy chain. IRES is widely used in mammalian display of full-length IgG (Zhou et al., 2010; Zhou and Shen, 2012; Tomimatsu et al., 2013). We proposed that self-cleaving 2A peptide could be a good candidate to replace IRES because of its small size and stoichiometric expression of multiple proteins flanking the 2A peptide (Fang et al., 2005; Chng et al., 2015). We tested the one-step emulsion PCR based reformatting method with a TpoR antibody named 3D9 (Zhang et al., 2013). Both T2A and IRES mediated full-length IgG format of 3D9 were successfully transformed from scFv template, 3D9, which were confirmed by Sanger sequencing. Binding of cell surface IgG to TpoR protein were verified by flow cytometry. T2A and IRES mediated full-length IgG can be displayed at similar level (Fig. 2A). Displayed IgG mediated by T2A and IRES retained specific binding to TpoR protein with Kd of 28.13 nM and 22.39 nM, respectively (Fig. 2B). Considering the short length of T2A peptide, we used T2A thereafter. Fig. 2 View largeDownload slide Mammalian display of T2A or IRES mediated IgG. TpoR antibody 3D9 was reformatted from scFv to T2A or IRES mediated IgG format. (A) HEK293T were infected with lentivirus and selected by puromycin to display T2A (upper) or IRES (lower) mediated IgG format of 3D9. Cells infected with control lentivirus pCDH were used as negative control. The cells were incubated with eight different concentrations of biotinylated TpoR and SA-PE to determine antigen binding ability and anti-human IgG Fc-AF 647 to indicate antibody display level. The representative flow cytometry graphs of cell staining were shown. (B) The plot of TpoR concentration versus the mean fluorescence intensity (MFI) of biotin-TpoR/SA-PE staining. The Kd values of displayed IgGs were calculated. Fig. 2 View largeDownload slide Mammalian display of T2A or IRES mediated IgG. TpoR antibody 3D9 was reformatted from scFv to T2A or IRES mediated IgG format. (A) HEK293T were infected with lentivirus and selected by puromycin to display T2A (upper) or IRES (lower) mediated IgG format of 3D9. Cells infected with control lentivirus pCDH were used as negative control. The cells were incubated with eight different concentrations of biotinylated TpoR and SA-PE to determine antigen binding ability and anti-human IgG Fc-AF 647 to indicate antibody display level. The representative flow cytometry graphs of cell staining were shown. (B) The plot of TpoR concentration versus the mean fluorescence intensity (MFI) of biotin-TpoR/SA-PE staining. The Kd values of displayed IgGs were calculated. Reformatting of scFv mini-library to IgG To maximize the reaction specificity, it should be tightly controlled to limit no more than one template molecule per droplet. We first determined the amount of template by performing PCR with a dilution series of template DNA. The maximum amplification yield was observed with 1010 to 1012 templates indicating that each droplet host at least one template DNA molecules and some droplets host more than one template (Supplemental Fig. S1). Less than one-third of the maximum DNA yield was obtained with 109 template molecules. Based on Poisson distribution, nearly each droplet hosts 0 or 1 copies of template DNA, thus this amount of template in w/o emulsion PCR could prevent wrong pairing of VH and VL from different templates and the formation of chimeric amplification by-products. To evaluate whether the correct VH–VL pairing can be maintained, a ‘mini-library’ comprising 16 individual scFvs of known sequences was reformatted to IgG library and IgG sequences of 50 clones from either emulsion PCR or conventional PCR were analyzed by Sanger sequencing (Table I). In emulsion PCR, 88% of the IgG sequences had correct VH–VL pairing while the remaining IgG representing antibodies with mismatched VH and VL. In conventional PCR only 6% of converted IgG maintained the correct VH–VL pairing. 86% of IgG had promiscuous VH–VL pairing and the remaining 8% are chimeric product of different templates. 3 and 15 original scFv sequences were represented in the 3 and 44 correctly pairing clones in the conventional PCR and emulsion PCR, respectively. Among the 15 original sequences in the emulsion PCR, 3 sequences appeared once, 4 sequences appeared twice, 3 sequences appeared 3 times, 2 appeared 4 times, 2 appeared 5 times and one appeared 6 times (Supplemental Table S2). Table I. Maintenance of VH and VL pairing during batch reformatting of a mini-library of 16 scFv to IgG by conventional PCR or emulsion PCR Correct pairing Wrong pairing Recombination product Conventional PCR 3/50 43/50 4/50 Emulsion PCR 44/50 6/50 0/50 Correct pairing Wrong pairing Recombination product Conventional PCR 3/50 43/50 4/50 Emulsion PCR 44/50 6/50 0/50 50 single clones of reformatted IgG using either conventional PCR or emulsion PCR were sequenced by Sanger sequencing and number of correct pairing of VH and VL, wrong pairing of VH and VL and formation of chimeric product was listed in table. Table I. Maintenance of VH and VL pairing during batch reformatting of a mini-library of 16 scFv to IgG by conventional PCR or emulsion PCR Correct pairing Wrong pairing Recombination product Conventional PCR 3/50 43/50 4/50 Emulsion PCR 44/50 6/50 0/50 Correct pairing Wrong pairing Recombination product Conventional PCR 3/50 43/50 4/50 Emulsion PCR 44/50 6/50 0/50 50 single clones of reformatted IgG using either conventional PCR or emulsion PCR were sequenced by Sanger sequencing and number of correct pairing of VH and VL, wrong pairing of VH and VL and formation of chimeric product was listed in table. Reformatting phage display derived scFv repertoire into IgG library As the one-step emulsion PCR based reformatting strategy was validated using the mini-library of 16 scFvs, we next reformatted a scFv repertoire of phage display selection to IgG library. CD40 is a promising target for cancer immunotherapy and several drugs that target the CD40 pathway have undergone phase I and II clinical trial (Moran et al., 2013; Hassan et al., 2014). Here, we performed phage display screening of naïve scFv library against CD40 ectodomain Fc fusion protein and obtained 1200-fold enrichment after three rounds of biopanning (Supplemental Table S3). Five out of twenty-nine phage clones from the third round output bound to CD40 ectodomain by phage ELISA, among which there were two unique sequences (Supplemental Fig. S2). We thus reformatted the scFv repertoire of the third round of phage display to a full-length IgG library with capacity of 2 × 104. We reformatted the IgG library after the third round of biopanning for two reasons. Firstly, there was excess of nonspecific antibody at earlier rounds as indicated by the phage ELISA result. Secondly, the size of the mammalian display library was often small. The total number of cells that could be screened by fluorescence activating cell sorter (FACS) was also limited. On the premise of bigger mammalian display library can be conveniently constructed and handled, it would be better to reformat the IgG library at earlier rounds. The technology of next-generation sequencing (NGS) can provide millions of sequences of libraries and provide deep insight into the original scFv library and the reformatted full-length IgG library. NGS libraries were constructed via emulsion PCR to avoid the PCR bias during the amplification (van Dijk et al., 2014). The diversity of the third round phage library output and the reformatted library were listed in Supplemental Table S4. The reformatted IgG library can cover about 75% of the output clones in the third round phage output. We compared the ranks of the top 20 HCDR3 and LCDR3 sequences in original scFv library to the reformatted library as well and found the good correlation between two libraries (Fig. 3A, the Pearson correlation value >0.85 and the P value < =0.001). We also observed similar distributions of CDR3 lengths (Fig. 3B) and germline families (Supplemental Fig. S3) between the scFv phagemid library and IgG library. The NGS analysis indicated that the diversity and relative abundance of antibody in the original scFv library were preserved in the reformatted full-length IgG library. Fig. 3 View largeDownload slide NGS analysis of original scFv library and reformatted full-length IgG library. (A) The correlation of the ranks of top 20 CDR3 sequences of original library compared to that of the reformatted full-length IgG library. (B) The distributions and frequencies of total sequences of the original scFv library and reformatted full-length IgG library, according to the length of the CDR3 of heavy chain(H-CDR3) or light chain(L-CDR3). Fig. 3 View largeDownload slide NGS analysis of original scFv library and reformatted full-length IgG library. (A) The correlation of the ranks of top 20 CDR3 sequences of original library compared to that of the reformatted full-length IgG library. (B) The distributions and frequencies of total sequences of the original scFv library and reformatted full-length IgG library, according to the length of the CDR3 of heavy chain(H-CDR3) or light chain(L-CDR3). Screening CD40 binding IgG from reformatted mammalian display IgG library We cloned the full-length IgG library into our modified mammalian cell display vector derived from the third generation lentivirus-based plasmid. The mammalian display IgG library construct was cotransfected with plasmids encoding packaging and envelop proteins to generate replication-incompetent lentivirus library. HEK293T were infected at a low multiplicity of infection (MOI) to ascertain the expression of a single antibody per infected cell. The cells were incubated with biotinylated CD40 antigen and then stained with Streptavidin-PE to indicate the affinity of antibodies (x-axis in Fig. 4) and goat anti-human IgG Fc-Alexa Fluor 488 or 647 to indicate the expression of antibodies (y-axis in Fig. 4). Approximately 1–2% cells were stained positively for biotin-CD40 in the first round of sorting. The cells located upper right of the graph were sorted and expanded for the next round of selection. Fig. 4 View largeDownload slide Isolation of CD40 binding antibody from reformatted IgG library. HEK293T cells were infected at a low MOI with T2A mediated IgG expression lentivirus library reformatted from a phage display scFv library. The infected cells were incubated with biotinylated CD40(lower) or irrelevant protein as negative control(upper) and then stained with SA-PE and anti-Human Fc-AF 647 or anti-Human Fc-AF 488 in alternative rounds. Gated cells were sorted and cultured for the next round of screening. The process was iterated for three rounds. Single cells of the third round were sorted into 96-well plates for further analysis. Gates for sorting are indicated by the quadrangle in the upper right corner of graph. Fig. 4 View largeDownload slide Isolation of CD40 binding antibody from reformatted IgG library. HEK293T cells were infected at a low MOI with T2A mediated IgG expression lentivirus library reformatted from a phage display scFv library. The infected cells were incubated with biotinylated CD40(lower) or irrelevant protein as negative control(upper) and then stained with SA-PE and anti-Human Fc-AF 647 or anti-Human Fc-AF 488 in alternative rounds. Gated cells were sorted and cultured for the next round of screening. The process was iterated for three rounds. Single cells of the third round were sorted into 96-well plates for further analysis. Gates for sorting are indicated by the quadrangle in the upper right corner of graph. Seventeen percent of scFvs could bind to CD40 after three rounds of biopanning (Supplemental Fig. S2). However, upon reformatting, a portion of antibodies may lose the affinity due to the conformational alteration from scFv to IgG. Thus, the reformatting is a crucial step to eliminate the false positive binders in the scFv form. The concentration of biotin-CD40 used in FACS sorting was low to favor the selection of antibody with high affinity, which also account for the low percentage of positive cells in the first round of sorting. Roundwise enrichment of positively staining cell was obtained during three rounds of selection (Fig. 4). Following the last round of selection, the cells with the highest antigen binding affinity were sorted into 96 well plates by single cell sorting for further analysis. Identification and analysis of single clones After 2–3 weeks the sorted individual clones of cells expanded to 70–90% confluency in wells for high throughput flow cytometry analysis of binding to CD40. 27 out of 30 clones exhibited strong binding to CD40. The antibody genes from the positive staining cells were amplified and subjected to Sanger sequencing. Among the sorted and positive IgG clones, there were 14 clones of Ab01, 11 clones of Ab02 and 2 clones of Ab03 (Fig. 5A). They were ranked the third, fourth and eighth in the reformatted libraries according to NGS result. Fig. 5 View largeDownload slide Identification of IgGs obtained via mammalian display. (A) Flow cytometry analysis of IgG of sorted clones on cell surface. Cells were incubated with biotinylated CD40, stained with SA-PE and anti-Human Fc-AF 647 and analyzed by flow cytometry. FACS analysis results of three representative single clones were shown here. (B) ELISA binding assay using purified antibody Ab01, Ab02, Ab03 and an irrelevant antibody HEL. Binding activity of different antibodies was compared by the ratio of the absorbance at 405 nm of CD40 to BSA. (C) The direct binding activity determination of Ab01, Ab02, Ab03 and HEL to hCD40. The binding activity to hCD40 was determined by the mean fluorescence intensity (MFI) quantified via flow cytometry analysis. (D) The affinity of Ab01, Ab02, Ab03 to CD40 measured using Biacore. Different concentration of CD40 were injected onto the antibody immobilized Protein A sensor chip and data presented as an overlay plot aligned at the start of the injection. Fig. 5 View largeDownload slide Identification of IgGs obtained via mammalian display. (A) Flow cytometry analysis of IgG of sorted clones on cell surface. Cells were incubated with biotinylated CD40, stained with SA-PE and anti-Human Fc-AF 647 and analyzed by flow cytometry. FACS analysis results of three representative single clones were shown here. (B) ELISA binding assay using purified antibody Ab01, Ab02, Ab03 and an irrelevant antibody HEL. Binding activity of different antibodies was compared by the ratio of the absorbance at 405 nm of CD40 to BSA. (C) The direct binding activity determination of Ab01, Ab02, Ab03 and HEL to hCD40. The binding activity to hCD40 was determined by the mean fluorescence intensity (MFI) quantified via flow cytometry analysis. (D) The affinity of Ab01, Ab02, Ab03 to CD40 measured using Biacore. Different concentration of CD40 were injected onto the antibody immobilized Protein A sensor chip and data presented as an overlay plot aligned at the start of the injection. Soluble antibodies were expressed and purified using a protein A affinity column and their binding to CD40 was first evaluated by ELISA. In comparison to the binding to BSA as a control, two antibodies Ab01 and Ab03 exhibited strong and specific binding to CD40 and Ab02 showed moderate binding activity to CD40 (Fig. 5B). Measuring antibody and soluble hCD40 ectodomain interaction in solution, away from receptor’s cellular environment, may not provide an affinity value applicable in vivo. Cell surface receptor hCD40 were stained with different concentrations of full-length IgGs and fluorescence secondary antibody for detection. It is demonstrated antigen binding ability of cell surface displayed antibody is positively correlated with its soluble counterpart (Fig. 5C). For binding to either plate immobilized or cell surface CD40, the apparent affinity is affected by the avidity effect. To avoid complications caused by the combined affinity resulting from multivalent binding, the bivalent IgG was immobilized on chip and then the monovalent CD40 protein was added when we measured the binding using Biacore. The affinity of Ab01, Ab02 and Ab03 were 439 nM, 3.89 μM and 287 nM, respectively (Fig. 5D, Table II). Table II. The kinetic profiles of the IgGs against CD40 Antibody KD(M) Ka(1/Ms) Kd(1/s) Ab01 4.39e–07 9.14e+04 4.01e–02 Ab02 3.89e–06 4.01e+02 1.56e–03 Ab03 2.87e–07 3.59e+04 1.03e–02 Antibody KD(M) Ka(1/Ms) Kd(1/s) Ab01 4.39e–07 9.14e+04 4.01e–02 Ab02 3.89e–06 4.01e+02 1.56e–03 Ab03 2.87e–07 3.59e+04 1.03e–02 Table II. The kinetic profiles of the IgGs against CD40 Antibody KD(M) Ka(1/Ms) Kd(1/s) Ab01 4.39e–07 9.14e+04 4.01e–02 Ab02 3.89e–06 4.01e+02 1.56e–03 Ab03 2.87e–07 3.59e+04 1.03e–02 Antibody KD(M) Ka(1/Ms) Kd(1/s) Ab01 4.39e–07 9.14e+04 4.01e–02 Ab02 3.89e–06 4.01e+02 1.56e–03 Ab03 2.87e–07 3.59e+04 1.03e–02 We cannot exclude the possibility of the presence of binders with higher affinities than the identified ones. The reformatted library covered about 75% of the diversity of the phage display output. In addition, only 30 clones of the sorted cells were characterized to demonstrate the feasibility of the method. It is likely that rare, high-affinity binders could be found when more clones of sorted cells were sampled. The activity of these antibodies was further tested by the CD40/NF-κB reporter cell line. Ab01 (EC50 = 3.978 nM) and Ab03 (EC50 = 11.34 nM) exhibited potent agonist activity while Ab02 had no agonist activity measured using the reporter assay (Fig. 6). The agonist activity describes the functional feature of the IgG and is dependent on the epitope of the target that IgG binds to. Fig. 6 View largeDownload slide The biological activity determination of Ab01, Ab02, Ab03 and HEL. hCD40 reporter cell lines were stimulated with antibodies and anti-human-IgG-Fc to crosslink the antibodies, and the value of MFI was used to indicate the activation. Fig. 6 View largeDownload slide The biological activity determination of Ab01, Ab02, Ab03 and HEL. hCD40 reporter cell lines were stimulated with antibodies and anti-human-IgG-Fc to crosslink the antibodies, and the value of MFI was used to indicate the activation. Discussion Phage display is a powerful and versatile technology in the directed evolution of new proteins, particularly for the production of antibody therapeutics (Huse et al., 1989; McCafferty et al., 1990; Barbas et al., 1991; Winter et al., 1994; Smith and Petrenko, 1997). Phage display has led to discovery of antibodies that can neutralize toxins, counteract autoimmune diseases and cure cancer. scFv is well suited for phage display and bacterial expression. However, loss of affinity during conversion from scFv into IgG, which is in most case the final therapeutic format, are frequently observed (Qin and Li, 2014; Steinwand et al., 2014). scFv format often has an enhanced avidity than its Fab counterpart due to formation of dimer, i.e, scFv with short linker between VH and VL can form diabody. A thorough understanding of structural difference between the scFv fragment and its Fab counterpart would resolved by crystallography. Reformatting of phage scFv to IgG allows earlier selection of effective IgGs. The traditional method is to convert scFv to full-length IgG on an individual basis, however, it is not practical for conversion of large number of scFv in a library. Xiao et al. resolved the problem by using inverse PCR to maintain the VH and VL pairing during the population reformatting of phage display libraries (Xiao et al., 2017). Here we proposed an even more straightforward and flexible one-step method by compartmentalization of the scFv-to-IgG conversion OE-PCR in millions of droplets in water-in-oil emulsion. The essential and unique aspects of our approach is that different parts of IgG are spliced in one-pot reaction of OE-PCR and each droplet host only one template in emulsion PCR to ensure correct VH–VL pairing is maintained with high fidelity. In addition to quantitative capability of flow cytometry to discriminate strong and weak binders, mammalian cell display provides several advantages with the ability to co-select antibody for high expression level in mammalian cells, native folding and biophysical properties appropriate for drug development (Beerli et al., 2008; Forsyth et al., 2013; Qin and Li, 2014). Limitation of mammalian cell-based systems, however, has been the smaller library complexity compared to that of phage or yeast display approaches. Thus, phage display and mammalian cell display are complementary technologies. Our one step emulsion PCR based method enable the seamless integration of two technologies into one platform. The antibody isolated by mammalian display need to be expressed as soluble protein for further characterization such as ELISA and function bioassay. To obtain antibodies in soluble form, the genes that encode the antibodies need to be recovered from the mammalian display vector and re-cloned into a new expression vector which contains no transmembrane domain. Bo et al. developed antibody membrane switch (AMS) technology to avoid this step (Yu et al., 2014). The switch between membrane anchored and secretion form is achieved by alternative splicing and specific DNA recombination to facilitate the transformation of the antibody from membrane to secretion format. We proposed that the LoxP sites can be placed flanking a PDGFR-TM domain so that we can knockout PDGFR-TM domain by inducing the expression of Cre recombinase. Yang et al. observed that significant potency loss happened when this scFv-Fc was transformed to full-length IgG form (Yang et al., 2018). Therefore, it is desirable to select antibody of the final format at early stage of drug discovery. Notably, the emulsion PCR based reformatting strategy is flexible to the choice of final antibody format and allows the introduction of various mutations to the Fc of the IgG vector through appropriate design of primers and bridging fragments. For example, human IgG2 imparts FcγR-independent agonistic activity to immune-stimulatory mAbs such as anti-CD40 agonist (Li and Ravetch, 2011; White et al., 2011). The phage display derived scFv library can be easily reformatted to IgG2 library using this approach for the selection of superagonistic immunostimulatory anticancer antibodies. In summary, we developed a method to reformat scFv to full-length IgG based on one-step emulsion PCR and full-length IgG of high affinity can be efficiently isolated when the method is coupled with mammalian display (Fig. 7). The advantages of the method are: (1) it enables the high throughput reformatting of scFv to full-length IgG and maintenance of the diversity and variable region pairing; (2) it is a simplified one-step procedure; (3) it is applicable to any format of antibody. This method provides a user-friendly platform for more effective mining of phage display scFv libraries for the functional antibody discovery. Fig. 7 View largeDownload slide The workflow of our strategy to reformat scFv library to IgG library for mammalian display. ScFv library derived from phage display selection performed against target were reformatted to full length IgG library using our one-step emulsion PCR approach and cloned into lentivirus-based mammalian surface display vector. Lentivirus library was prepared and infection of HEK293T cells at a low MOI generates a pool of infected cells, each expressing at their surface one specific antibody. Cells were stained with antigen and then positively stained cells were sorted by flow cytometry. Antibody genes were extracted from positive cell clones and soluble antibody was expressed and purified for further characterization. Fig. 7 View largeDownload slide The workflow of our strategy to reformat scFv library to IgG library for mammalian display. ScFv library derived from phage display selection performed against target were reformatted to full length IgG library using our one-step emulsion PCR approach and cloned into lentivirus-based mammalian surface display vector. Lentivirus library was prepared and infection of HEK293T cells at a low MOI generates a pool of infected cells, each expressing at their surface one specific antibody. Cells were stained with antigen and then positively stained cells were sorted by flow cytometry. Antibody genes were extracted from positive cell clones and soluble antibody was expressed and purified for further characterization. Funding This work was supported by the National Key R&D Program of China under Grant number 2017YFA0504801; National Natural Science Foundation of China under Grant number 81661148051, 81672010; National Natural Science Foundation of China under Grant number 81872787, 81603010; National Natural Science Foundation of China under Grant number 31500632, 31700807. Conflict of interest No potential conflicts of interest were disclosed. References Barbas , C.F. , Kang , A.S. , Lerner , R.A. and Benkovic , S.J. ( 1991 ) Proc. Natl Acad. Sci. U S A , 88 , 7978 – 7982 . doi:10.1073/pnas.88.18.7978 . Crossref Search ADS PubMed Beerli , R.R. , Bauer , M. , Buser , R.B. , Gwerder , M. , Muntwiler , S. , Maurer , P. , Saudan , P. and Bachmann , M.F. ( 2008 ) Proc. Natl. Acad. Sci. U S A , 105 , 14336 – 14341 . doi:10.1073/pnas.0805942105 . Crossref Search ADS PubMed Bolotin , D.A. , Poslavsky , S. , Mitrophanov , I. , Shugay , M. , Mamedov , I.Z. , Putintseva , E.V. and Chudakov , D.M. ( 2015 ) Nat. Methods , 12 , 380 – 381 . doi:10.1038/nmeth.3364 . Crossref Search ADS PubMed Chen , C.G. , Fabri , L.J. , Wilson , M.J. and Panousis , C. ( 2014 ) Nucleic Acids Res. , 42 , e26 . doi:10.1093/nar/gkt1142 . Crossref Search ADS PubMed Chng , J. , Wang , T. , Nian , R. , Lau , A. , Hoi , K.M. , Ho , S.C. , Gagnon , P. , Bi , X. and Yang , Y. ( 2015 ) mAbs , 7 , 403 – 412 . doi:10.1080/19420862.2015.1008351 . Crossref Search ADS PubMed Fang , J. , Qian , J.J. , Yi , S. , Harding , T.C. , Tu , G.H. , VanRoey , M. and Jooss , K. ( 2005 ) Nat. Biotechnol. , 23 , 584 – 590 . doi:10.1038/nbt1087 . Crossref Search ADS PubMed Forsyth , C.M. , Juan , V. , Akamatsu , Y. , DuBridge , R.B. , Doan , M. , Ivanov , A.V. , Ma , Z. , Polakoff , D. , Razo , J. , Wilson , K. et al. . ( 2013 ) MAbs , 5 , 523 – 532 . doi:10.4161/mabs.24979 . Crossref Search ADS PubMed Hassan , S.B. , Sorensen , J.F. , Olsen , B.N. and Pedersen , A.E. ( 2014 ) Immunopharmacol. Immunotoxicol. , 36 , 96 – 104 . doi:10.3109/08923973.2014.890626 . Crossref Search ADS PubMed Huse , W.D. , Sastry , L. , Iverson , S.A. , Kang , A.S. , Alting-Mees , M. , Burton , D.R. , Benkovic , S.J. and Lerner , R.A. ( 1989 ) Science , 246 , 1275 – 1281 . doi:10.1126/science.2531466 . Crossref Search ADS PubMed Huston , J.S. , Levinson , D. , Mudgetthunter , M. , Tai , M.S. , Novotný , J. , Margolies , M.N. , Ridge , R.J. , Bruccoleri , R.E. , Haber , E. and Crea , R. ( 1988 ) Proc. Natl. Acad. Sci. , 85 , 5879 – 5883 . doi:10.1073/pnas.85.16.5879 . Crossref Search ADS Jostock , T. , Vanhove , M. , Brepoels , E. , Van Gool , R. , Daukandt , M. , Wehnert , A. , Van Hegelsom , R. , Dransfield , D. , Sexton , D. , Devlin , M. et al. . ( 2004 ) J. Immunol. Methods , 289 , 65 – 80 . doi:10.1016/j.jim.2004.03.014 . Crossref Search ADS PubMed Kotera , I. and Nagai , T. ( 2008 ) J. Biotechnol. , 137 , 1 – 7 . doi:10.1016/j.jbiotec.2008.07.1816 . Crossref Search ADS PubMed Li , F. and Ravetch , J.V. ( 2011 ) Science , 333 , 1030 – 1034 . doi:10.1126/science.1206954 . Crossref Search ADS PubMed McCafferty , J. , Griffiths , A.D. , Winter , G. and Chiswell , D.J. ( 1990 ) Nature , 348 , 552 – 554 . doi:10.1038/348552a0 . Crossref Search ADS PubMed Moran , A.E. , Kovacsovics-Bankowski , M. and Weinberg , A.D. ( 2013 ) Curr. Opin. Immunol. , 25 , 230 – 237 . doi:10.1016/j.coi.2013.01.004 . Crossref Search ADS PubMed Qin , C.F. and Li , G.C. ( 2014 ) Int. Immunopharmacol. , 23 , 380 – 386 . doi:10.1016/j.intimp.2014.09.017 . Crossref Search ADS PubMed Sanmark , H. , Huovinen , T. , Matikka , T. , Pettersson , T. , Lahti , M. and Lamminmaki , U. ( 2015 ) J. Immunol. Methods , 426 , 134 – 139 . doi:10.1016/j.jim.2015.08.005 . Crossref Search ADS PubMed Smith , G.P. and Petrenko , V.A. ( 1997 ) Chem. Rev. , 97 , 391 – 410 . doi:10.1021/cr960065d . Crossref Search ADS PubMed Steinwand , M. , Droste , P. , Frenzel , A. , Hust , M. , Dubel , S. and Schirrmann , T. ( 2014 ) MAbs , 6 , 204 – 218 . doi:10.4161/mabs.27227 . Crossref Search ADS PubMed Sun , Z. , Lu , S. , Yang , Z. , Li , J. and Zhang , M.Y. ( 2017 ) Virus Res. , 238 , 156 – 163 . doi:10.1016/j.virusres.2017.06.018 . Crossref Search ADS PubMed Tomimatsu , K. , Matsumoto , S.E. , Tanaka , H. , Yamashita , M. , Nakanishi , H. , Teruya , K. , Kazuno , S. , Kinjo , T. , Hamasaki , T. , Kusumoto , K. et al. . ( 2013 ) Biochem. Biophys. Res. Commun. , 441 , 59 – 64 . doi:10.1016/j.bbrc.2013.10.007 . Crossref Search ADS PubMed Turchaninova , M.A. , Britanova , O.V. , Bolotin , D.A. , Shugay , M. , Putintseva , E.V. , Staroverov , D.B. , Sharonov , G. , Shcherbo , D. , Zvyagin , I.V. , Mamedov , I.Z. et al. . ( 2013 ) Eur. J. Immunol. , 43 , 2507 – 2515 . doi:10.1002/eji.201343453 . Crossref Search ADS PubMed van Dijk , E.L. , Jaszczyszyn , Y. and Thermes , C. ( 2014 ) Exp. Cell Res. , 322 , 12 – 20 . doi:10.1016/j.yexcr.2014.01.008 . Crossref Search ADS PubMed Vaughan , T.J. , Williams , A.J. , Pritchard , K. , Osbourn , J.K. , Pope , A.R. , Earnshaw , J.C. , McCafferty , J. , Hodits , R.A. , Wilton , J. and Johnson , K.S.J.N.b ( 1996 ) Nat. Biotechnol. , 14 , 309 – 314 . Crossref Search ADS PubMed White , A.L. , Chan , H.T. , Roghanian , A. , French , R.R. , Mockridge , C.I. , Tutt , A.L. , Dixon , S.V. , Ajona , D. , Verbeek , J.S. , Al-Shamkhani , A. et al. . ( 2011 ) J. Immunol. , 187 , 1754 – 1763 . doi:10.4049/jimmunol.1101135 . Crossref Search ADS PubMed Williams , R. , Peisajovich , S.G. , Miller , O.J. , Magdassi , S. , Tawfik , D.S. and Griffiths , A.D. ( 2006 ) Nat. Methods , 3 , 545 – 550 . doi:10.1038/nmeth896 . Crossref Search ADS PubMed Winter , G. , Griffiths , A.D. , Hawkins , R.E. and Hoogenboom , H.R. ( 1994 ) Annu. Rev. Immunol. , 12 , 433 – 455 . doi:10.1146/annurev.iy.12.040194.002245 . Crossref Search ADS PubMed Xiao , X. , Douthwaite , J.A. , Chen , Y. , Kemp , B. , Kidd , S. , Percival-Alwyn , J. , Smith , A. , Goode , K. , Swerdlow , B. , Lowe , D. et al. . ( 2017 ) mAbs , 9 , 996 – 1006 . doi:10.1080/19420862.2017.1337617 . Crossref Search ADS PubMed Yang , Z. , Du , M. , Wang , W. , Xin , X. , Ma , P. , Zhang , H. and Lerner , R.A. ( 2018 ) Protein Eng. Des. Sel . doi:10.1093/protein/gzy002 . Yu , B. , Wages , J.M. and Larrick , J.W. ( 2014 ) Protein Eng. Des. Sel. , 27 , 309 – 315 . doi:10.1093/protein/gzu039 . Crossref Search ADS PubMed Zhang , H. , Yea , K. , Xie , J. , Ruiz , D. , Wilson , I.A. and Lerner , R.A. ( 2013 ) Chem. Biol. , 20 , 734 – 741 . doi:10.1016/j.chembiol.2013.04.012 . Crossref Search ADS PubMed Zhou , C. , Jacobsen , F.W. , Cai , L. , Chen , Q. and Shen , W.D. ( 2010 ) mAbs , 2 , 508 – 518 . doi:10.4161/mabs.2.5.12970 . Crossref Search ADS PubMed Zhou , C. and Shen , W.D. ( 2012 ) Methods Mol. Biol. , 907 , 293 – 302 . doi:10.1007/978-1-61779-974-7_17 . Crossref Search ADS PubMed © The Author(s) 2019. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Journal

Protein Engineering, Design and SelectionOxford University Press

Published: Nov 1, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create folders to
organize your research

Export folders, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off