Validation of glypican-3-specific scFv isolated from paired display/secretory yeast display library
© Li et al; licensee BioMed Central Ltd. 2012
Received: 21 November 2011
Accepted: 7 May 2012
Published: 7 May 2012
Skip to main content
© Li et al; licensee BioMed Central Ltd. 2012
Received: 21 November 2011
Accepted: 7 May 2012
Published: 7 May 2012
Glypican-3 (GPC3) is a heparan-sulfate proteoglycan frequently expressed on the cell membrane of malignant hepatocytes in hepatocellular carcinoma. The capacity for screening potential antibodies in vitro using human hepatocellular lines is critical to ensure binding to this highly post-translationally modified glycophosphatidylinositiol-linked protein. We hypothesized that we could utilize a recently described paired display/secretory yeast library to isolate human-derived scFv against glypican-3 for potential diagnostic and/or therapeutic application.
Using two different biotinylated antigen targets, a synthesized 29mer fragment GPC3550-558 and a truncated GPC3368-548 fused with glutathione S-transferase (GST) we enriched the yeast display library to greater than 30% target-specific yeast with both positive selection and depletion of streptavidin- and GST-specific clones. After cloning of scFv cDNA from the enriched sub-library, scFv specificity was validated by ELISA for binding to recombinant protein from prokaryotic and eukaryotic sources and ultimately naturally presented human protein on the cell membrane of human hepatocellular cell lines. Specificity was confirmed using non-expressing cell lines and shRNA knockdown. Ultimately, five unique scFv with affinity EC50 ranging from 5.0-110.9nM were identified.
Using a paired display/secretory yeast library, five novel and unique scFvs for potential humoral or chimeric therapeutic development in human hepatocellular carcinoma were isolated and characterized.
Hepatocellular carcinoma (HCC) is the fifth most common cancer and the third most common cause of cancer-related death worldwide . During transformation from dysplastic regenerating hepatocytes to malignant hepatoma cells, several tumor-associated proteins are expressed that potentially could allow immune discrimination of malignant hepatocytes from surrounding non-tumor cells. Glypican-3 (GPC3), an oncofetal antigen re-expressed in a high frequency of neoplastic hepatocytes [2–5] has emerged as a useful immunohistochemical diagnostic test [6–8] and potential biomarker [3, 9, 10] for hepatocellular carcinoma. Glypican-3 appears critical for the association of growth factors such as insulin-like growth factor-2, bone morphogenic protein-7 and fibroblast growth factor-2 with growth factor receptors [11, 12] but also may play an immunomodulatory role . Inhibition of glypican-3 function via knockdown [14, 15] or competition [12, 16] has a profound negative effect on HCC cell line proliferation. Unlike any other tumor antigen associated with hepatocellular carcinoma, GPC3 is a glycophosphatidylinositiol-linked membrane-associated protein with a large extracellular domain attractive for antibody-directed therapy. An anti-glypican-3 murine IgG antibody that induces antibody-dependent cytotoxicity has been shown to have anti-tumor effect in a xenograft animal model of hepatocellular carcinoma  but required partial humanization before entering human clinical trials . Thus, while there is a strong rationale for targeting glypican-3 for humoral and potentially chimeric immunotherapy for HCC, an scFv of human origin might be less immunogenic and more flexible for incorporation into downstream applications.
A paired yeast display/secretory scFv library derived from immunoglobulin heavy and light chains originally derived from the B-cells of a human patient with thrombotic thrombocytopenic purpura  has been shown to be a powerful tool for the identification of human scFv against surface-expressed human tumor antigens . Key advantages of this approach include a large repertoire of potential human heavy and light chain pairings, efficient flow cytometric enrichment, eukaryotic-type post-translational modifications, absence of potential xenoreactive sequences and efficient conversion to soluble secreted scFv for validation .
In this study, we report our development and validation of multiple human glypican-3-specific scFv. The high throughput methodology identified human-derived scFv with EC50 ranging from 5.0 – 110.9nM. These scFv bound specifically to glypican-3-expressing cell lines. scFv binding was significantly reduced by shRNA knockdown of glypican-3. We believe these scFv are optimal for development for diagnostic and in vivo therapeutic applications.
Therapeutic options for hepatocellular carcinoma (HCC) remain limited particularly in advanced stages. Immunotherapy with NK- or T-cell augmenting therapies to date has yielded some early promising results [24–27] but the low affinity of endogenous tumor-specific T-cell receptors and the immunosuppressive milieu of the tumor microenvironment represent barriers to effectively harnessing the power of the endogenous immune system to control cancer. Yeast-derived scFv offer many advantageous properties for the development of anti-tumor biologics. scFv are inexpensive to produce, easily modifiable e.g. biotinylation , and facile for subsequent cloning in cis with diagnostic or effector domains.
Identification of an appropriate tumor–associated antigen is an obviously essential requirement for scFv development. Glypican-3 (GPC3), a heparan-sulfate proteoglycan, has recently been identified as a highly specific, membrane-associated tumor antigen found in 49-100% of HCC [2–5]. GPC3 is not expressed (or is expressed very focally ) in non-tumorous cirrhotic liver tissue [7, 30] and expression of GPC3 in other normal tissues appears limited . GPC3 modulates the effect of growth factors such as IGF-2, BMP-7 and FGF-2 on hepatoma cells [11, 12] and may recruit M2 tumor-promoting macrophages to the HCC microenvironment . Emerging evidence also suggests that inhibition of glypican-3 function via knockdown [14, 15] or competition [12, 16] has a profound negative effect on HCC proliferation. Expression on the cell surface makes GPC3 an attractive target for antibody-directed therapy. Another group has shown that a murine anti-hGPC3 antibody induces antibody-dependent cytotoxicity that manifests an anti-tumor effect in a xenograft animal model of hepatocellular carcinoma ; this antibody has subsequently been humanized  and is entering human clinical trials. Thus, available evidence suggests that glypican-3 is a rational target for humoral and potentially chimeric immunotherapy for HCC.
In this study, we utilized the paired display/secretion yeast system to isolate five candidate scFv with affinity in the range from 5.0 – 110.9 nM that each demonstrate specificity for binding the surface of glypican-3-expressing cell lines. scFv binding was significantly reduced after specific knockdown of glypican-3. The paired yeast display/secretion system minimizes post-translational and conformational changes in the conversion from displayed to soluble scFvs, a property that allows for consistency during the high throughput screening and validation process . scFv specificity to the naturally processed glypican-3 protein at physiological conditions was critical given complex post-translational modifications of glypican-3. We utilized increasingly physiological screening criteria to select scFv candidates for further evaluation. Dramatic differences of scFv binding between wild-type and glypican-3-knockdown HepG2 in cell culture conditions confirmed not only the specificity of scFv binding but also the capacity to bind to naturally-processed cell surface glypican-3 in situ. In separate work, we are currently validating a chimeric antigen receptor to redirect T-cells against glypican-3-expressing targets using our 3E11 scFv.
Not surprisingly, scFv had no direct positive or negative impact on cellular proliferation unlike that demonstrated by soluble glypican-3 . The relatively small size of scFv (27 kD) makes competitive inhibition of growth factor binding unlikely. We did not include agonism or antagonism in our screening strategy, and thus lack of agonist and antagonistic effect is not unexpected. Conjugation of our scFv to cellular cytotoxins will be explored as a potential therapeutic application of the scFv technology.
Glypican-3 is a rational target in hepatocellular carcinoma for antibody-based therapy. Utilizing a recently described paired display/secretory yeast library to isolate human-derived scFv against glypican-3, five unique scFv with affinity ranging from 5.0-110.9nM were identified. Each scFv in vitro demonstrated strong surface binding to glypican-3-expressing cell lines that was attenuated by shRNA knockdown, and did not bind glypican-3-nonexpressing cell lines. Ongoing work is characterizing the in vitro and in vivo application of these scFv in chimeric antigen receptor technology for hepatocellular carcinoma.
Cell lines of 293T (American Type Culture Collection, Manassas VA), HepG2 (ATCC), Hep3B (obtained from the Penn Center for Molecular Studies in Digestive and Liver Disease) and GP2-293 cells (Clontech, Mountain View, CA) were maintained in Dulbecco’s modified essential medium DMEM (Invitrogen, Carlsbad, CA) with 10% fetal bovine serum (Sigma, St. Louis MO). HepG2.tdTomato were generated via stable transfection of parental HepG2 with a lentiviral vector encoding tdTomatoRed (a gift from Dr. Carl June) and purified by flow cytometry. 293T.GPC3 were generated by cloning full-length human GPC3 cDNA (NM_004484) into pDisplay (Invitrogen) XmaI and SacII sites using the following forward (5′ CCCGGGGCCACCTGTCACCAAGTCCG 3′) and reverse primer (5′ CCGCGG GTGCACCAGGAAGAAGAAGCAC 3′).
The full-length cDNA of human glypican-3 (NM_004484) was amplified from a human cDNA library using the following forward primer (5′ ATGGCCGGGACCGTGCGCACC 3′) and reverse primer (5′ TCAGTGCACCAGGAAGAAGAAGCA 3′). A 594 bp DNA fragment corresponding to the region of nt1277-1871, which translates a truncated fragment of hGPC3 (aa 368–548) between the CRD cleavage site and putative transmembrane domain, was cloned into the prokaryotic expression vector pGEX-4T using SalI and EcoRI restriction sites (forward primer: 5′CCG GAA TTC GAC AAG AAA GTA TTA AAA GTT GCT CA 3′ and reverse primer: 5’ ACG CGT CGA CGG TGC TTA TCT CGT TGT CCT TC-3′) to generate a plasmid encoding a truncated hGPC3-GST recombinant fusion protein under the control of an IPTG-inducible tac promoter. The plasmid was transformed into E. coli BL21-CodonPlus (DE3)-RIPL (Stratagene, Santa Clara CA), grown in fresh 2YT medium, and induced by 1 mM IPTG at 25c for 6 h. Bacterial cells were collected by centrifugation and lysed by sonication in presence of 1% sarkosyl and 2% Triton X-100. The lysate was incubated with Glutathione Sepharose 4B beads (GE healthcare, Piscataway NJ) at 4c for 4 h, washed, and then eluted 50 mM Tris–HCl buffer (pH 7.4) containing 20 mM reduced glutathione. The recovery of the GST and GPC3-GST fusion protein was confirmed by Coomassie Blue staining. The trhGPC3-GST, GST, and a commercially custom synthesized 29mer GPC3 peptide (aa 530–558, Proimmune Oxford UK) were biotinylated using NHS Biotinylation kit (Pierce, Rockford IL).
The paired yeast-display/secretory scFv library has previously been described , and was screened using existing methodology with minor modifications. Briefly the yeast display library was grown in SD-CAA (2% raffinose, 0.67% yeast nitrogen base, and 0.5% casamino acids) at 30c to an Å600 of ~5. Surface display of scFv was induced by re-inoculating yeast at an Å600 of 0.5 in SGRD-CAA (SD-CAA + 2% galactose) and grown at 20c for 16-36 h. scFv expression by yeast was confirmed by flow cytometry using anti-c-myc mouse mAb (9E10, Santa Cruz biotechnology) and goat anti-mouse Fab Alexa Fluor 488 (AF488, Invitrogen, Carlsbad CA). Two rounds of magnetic bead-based selection were performed as follows: 1x109 induced yeast-display scFv in 500 ul PBE buffer (PBS + 0.5% EDTA) were incubated with biotinylated 29mer GPC3 peptide (100nM) or biotinylated rhGPC3-GST (100 ng/ml) at 25c for 30 min then on ice for 10 min. The rhGPC3-reactive yeast-display scFv were enriched by magnetically sorting over an LS column (Miltenyi Biotec, Auburn, CA). When screening with rhGPC3-GST protein, GST-reactive yeast-display scFv were depleted over an LS column after incubation of yeast-display scFv with biotinylated GST and streptavidin microbeads. Three rounds of flow cytometry-based sorting were performed with gradually decreasing concentration of target antigen as follows: yeast cells were stained with mouse anti-c-myc mAb (1:200), anti-mouse IgG1 AF488, biotinylated antigen (rhGPC3 protein at 40 ng/ml in 1st round, 20 ng/ml in 2nd round, and 10 ng/ml in 3rd round), and either streptavidin-PE (1st and 2nd round, Invitrogen) or neutravidin-PE (3rd round, Invitrogen). AF488+ and PE + double positive yeast were selected and recovered in 96 well plates containing SD-CAA. In all FACS sorts, conditions without antigen, with an irrelevant biotinylated antigen, or with biotinylated GST were included.
scFv cDNA were extracted from the enriched yeast population after the 3rd round of flow sorting, amplified by PCR (forward primer: 5′-GGT TCTGGTGGTGGAGGTTCTGGTGGTGGTGGATCTG-3; reverse 5′-GAGACCGAGGAG AGGGTTAGGGATAGGCTTACCGTCGACCAAGTCTTCTTCAGAATA AGCTT-3′), purified using MiniElute kit (Qiagen, Valencia CA), and then co-transformed with 100 ng of linearized p416-BCCP vector into YVH10 cells. Transformed yeast were plated on Trp + SD-CAA dishes, from which approximately six hundred colonies were transferred to growth medium in deep 96-well plates (Fisher Scientific) and induced by 2% galactose to secrete scFv for up to 72 h. High throughput purification of scFv was performed as previously described by Bergan et al. .
For screening of scFv affinity, Nunc Maxisorb plates were pre-coated target antigen or control at the indicated concentration in carbonate-bicarbonate buffer on overnight at 4c. Target antigens included: 1) GPC3550-558 peptide with media control; 2) rhGPC3.GST expressed in E. coli with GST as control; and 3) rhGPC31-559. His expressed in murine myeloma cell line (R&D Systems, Minneapolis MN) with media control After three washing steps with PBS/0.1% Tween-20 (PBST), 300 ul per well of blocking solution (2% milk in PBS pH 7) was added for 2 h at room temperature then washed three times with PBST. Candidate scFv starting at 100 ug/ml were added, incubated for 1 h at room temperature followed by three washing steps with PBST. scFv binding was detected by adding anti-V5 HRP (Invitrogen), washing x 4 with PBST, washing x 1 with PBS, then adding 50 ul/well of TMB peroxidase substrate (KPL, Gaithersburg MD) plus peroxidase substrate solution B at 1:1 ratio, then stopping the reaction with 50 ul 0.5 M H2SO4. OD450 was measured using a BioRad 680 microplate reader. For determination of functional affinity, ELISA was performed as above with plates coated with rhGPC3.GST at two dilutions and with scFv added at serial dilutions starting at 110 ug/ml. Half-maximal binding concentration (EC50) was calculated with non-linear regression curve fit algorithm the software program PRISM (GraphPad Software, San Diego, CA). rhGPC3 expressed in murine myeloma cell line was commercially obtained (R&D Systems, Minneapolis MN).
Detection of scFv binding to cell lines was detected with anti-V5 mAb (AbD Serotec, Raleigh, NC). Anti-hGPC3 mAb (1G12, Biomosaics Inc., Burlington, VT) was utilized as a positive control. scFv were premixed with anti-V5 APC mAb (AbD Serotec) at a molar ratio of 1:1 for 30 min at RT. scFv-anti-V5 complexes were then incubated with target cell lines for 30 min at 37c. Cells were acquired on a FACSCanto (Becton Dickinson, San Jose CA) and analyzed using FlowJo (Treestar, Ashland, OR).
Target cell lines cultured on 0.2 μm coverslips (Nunc, Rochester, NY) were fixed and stained with indicated scFv-V5 APC complex. Image acquisition was performed on a Fluoview 10 confocal laser microscope (Olympus).
Cell lysates were separated by SDS-PAGE gel and transferred to polyvinylidene difluoride membranes. In dot blot, purified protein (10 ng) was spotted on PVDF membrane. Membranes were blotted with primary Abs followed by incubation with infrared dye IR680-labeled secondary antibodies and quantified with LI-COR Odyssey software.
hGPC3-specific short hairpin RNAs (shRNAs) were prepared in the pSIREN-retroQ-zsGreen retroviral vector by using knockout RNAi systems according to the manufacturer’s instruction (Clontech). Three pairs of 21 nucleotide oligonucleotides, named sh56, sh57, and sh58 as well as a LacZ (negative control), were predicted according to Ambion Silencer Select software, annealed and sub-cloned into pSIREN-retroQ-zsGreen at the BamHI and EcoRI sites. The RNA targeting sequence of these three shRNAs are (sh56: 5′-GCCAAATTATTCTCC TATGTT-3′; sh57: 5′-GCCAATATAGATCTGCTTATT -3′; sh58: 5′- GCTCAAGAAAGA TGGAAGAAA-3′). For testing hairpin silencing, myc-tagged hGPC3(AA 368–551) was cloned into the pDisplay plasmid. Plasmids expressing shRNA and hGPC3.myc plasmids were co-transfected into HEK 293 cells (3:1 ratio, hairpin to target), and cells were lysed after 48 h. hGPC3.myc levels were quantified by Western blot using anti-c-myc mAb. Pseudotyped retrovirus encoding shRNA were then produced. Briefly, GP2-293 cells were seeded in 10 cm cell culture dishes 12 h prior to transfection. At 50% density, cells were transfected with 10 ug pSIREN-shRNA plasmid and 5 ug pVSV-G (Clontech) for pseudotyping using the calcium phosphate transfection method. On day 2 and day 3 after transfection, media containing retroviral particles were collected. Particles were concentrated 100-fold by ultracentrifugation. To infect cells, 10 ul of concentrated virus stock were added into 1 × 106 HepG2 cells in presence of polybrene (4 ug/ml). Transduced cells were isolated by FACS sorting of eGFP + cells and maintained as stable cell lines.
We performed a standard MTT assay using the CellTiter 96® Non-Radioactive Cell Proliferation Assay (Promega Corporation, Madison WI) according to manufacturer’s instructions. HepG2 and HepG2.sh57 cells were plated at a density of 5 × 103 cells/well in triplicate in a 96-well plate and incubated for 2–4 days as indicated. Optical density was measured at 570 nm. Trypisinized cells were manually counted by hematocytometer in validation experiments.
Complementarity determining region
Flow cytometry-assisted cell sorting
Enhanced green fluorescent protein
Magnet-assisted cell sorting
Single-chain fragment variable
short hairpin RNA.
This work was supported by R21 CA149908 (DEK), McCabe Family Research Foundation, and Research Career Development Award from the Veterans Health Administration (DEK), academic development funds from the University of Pennsylvania (DEK). The authors would like to thank Aizhi Zhao for his technical assistance. The content of this article does not reflect the views of the VA or of the US Government.