PEDS Advance Access originally published online on January 22, 2007
Protein Engineering Design and Selection 2007 20(2):81-90; doi:10.1093/protein/gzl057
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Antigen selection from an HIV-1 immune antibody library displayed on yeast yields many novel antibodies compared to selection from the same library displayed on phage
1 Departments of Immunology and Molecular Biology, The Scripps Research Institute, 10550 North Torrey Pines Rd, IMM-2, La Jolla, CA 92037, USA
3 To whom correspondence should be addressed. E-mail: burton{at}scripps.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Phage display of antibody libraries has been widely used for over a decade to generate monoclonal antibodies. Yeast display has been developed more recently. Here the two approaches were directly compared using the same HIV-1 immune scFv cDNA library expressed in phage and yeast display vectors and using the same selecting antigen (HIV-1 gp120). Yeast display was shown to sample the immune antibody repertoire considerably more fully than phage display, selecting all the scFv identified by phage display and twice as many novel antibodies. Positive phage display selection appeared to largely reflect those antibodies that as phage-scFv gave the highest signal in phage ELISAs assessing antigen binding. This signal is thought to reflect the efficiency of expression of folded scFv at the phage surface. Increased access to immune repertoires may increase the rescue of novel antibodies of therapeutic or analytical value that often form a minor part of a typical antibody response.
Keywords: HIV-1 antibody response/phage display/scFv/yeast display
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Phage display of antibodies (Burton and Barbas, 1994
; Winter et al., 1994; Bradbury and Marks, 2004; Hoogenboom, 2002) has contributed to many areas of biomedical science. This technology allows the generation of libraries for the rapid selection of monoclonal antibodies (mAbs) against a desired antigen through a panning process. Of particular note is the ability to generate human antibodies, which has been relatively difficult using hybridoma-based technologies. Phage display has been used to retrieve human mAbs against both foreign and self antigens.
One limitation of phage display is that it requires prokaryotic expression of antibody fragments. It is well known that there is an unpredictable expression bias against some eukaryotic proteins expressed from Escherichia coli because the organism lacks foldases and chaperones present in the endoplasmic reticulum of eukaryotic cells that are necessary for efficient folding of secreted proteins such as antibody fragments. Even minor sequence changes such as single point mutations in the complementarity determining regions (CDRs) of Fab fragments can completely eliminate antibody expression in E. coli (Ulrich et al., 1995), and a random sampling of an scFv library showed that half of the library had no detectable level of scFv in the ulture supernatant (Vaughan et al., 1996). Yeast has emerged as an alternative technology for the display and selection of antibody fragments, both scFv (Boder and Wittrup, 1997; Feldhaus et al., 2003) and Fab (van den Beucken et al., 2003; Weaver-Feldhaus et al., 2004). Because the protein folding and secretory pathways of yeast more closely approximate those of mammalian cells, it has been hypothesized that yeast display could provide access to more or different antibodies than phage display (Boder and Wittrup, 1997).
To compare yeast display to phage display directly, we have subcloned an immune scFv library from the phage display vector pComb3X into a modified yeast display vector derived from pPNL6. As a model system we used a library generated from an asymptomatic long-term HIV-1 infected individual, FDA2, whose serum has been shown to contain anti-viral antibodies that can block the infectivity of a number of HIV-1 strains in vitro (Vujcic and Quinnan, 1995; Fenyo et al., 1996; Moore et al., 1996; Parren et al., 1998). Phage display panning of this library was completed previously and a number of antibody specificities have been characterized and published (Moulard et al., 2002; Zwick et al., 2001, 2003). These pannings focused on antibodies to the HIV-1 envelope spike glycoproteins, gp120 and gp41. This study was undertaken to compare mAbs isolated from both phage and yeast display using recombinant gp120 as selecting antigen. After 3–4 rounds of selection, sequence analysis of individual clones revealed many common antibodies isolated by both techniques, but also revealed many novel antibodies from yeast display selection that had not previously been described from phage display. A representative set of novel and previously identified scFv were characterized on the surface of yeast for their affinity for gp120 and the gp120 epitopes recognized. To our knowledge, this is the first direct comparison of yeast and phage display systems using the same immune antibody library and screening antigen.
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Cell lines and media
Yeast strain EBY100 (GAL1-AGA1::URA3 ura3-52 trp1 leu2
1 his3200 pep4::HIS2 prb11.6R can1 GAL) was maintained in YPD broth (Difco). Transfection of EBY100 with the vector pPNLS was completed using the lithium acetate method and maintained in SD-CAA medium (6.7 g/l yeast nitrogen base, 5 g/l casamino acids, and 20 g/l dextrose, 9.67 g/l NaH2PO4·2H2O and 10.19 g/l Na2HPO4·7H2O) and on SD-CAA plates (SD-CAA + 17 g/l agar). Yeast surface expression of scFv was induced by transferring to SG/R-CAA medium (same as SD-CAA, replacing dextrose with 20 g/l each galactose and raffinose and 1 g/l dextrose). E. coli XL1-Blue was used for cloning, preparation of plasmid DNA and phage production with VCSM13 helper phage. Soluble expression of scFv used E. coli BL21(DE3). Bacterial strains were grown in SB media (30 g/l bactotryptone, 20 g/l yeast extract, and 10 g/l MOPS) supplemented with 20 mM glucose for plasmid preparation and phage production, or 20 mM MgCl2 for soluble scFv production. Dulbecco's modified Eagle's media (Gibco) supplemented with penicillin, L-glutamine and 10% fetal bovine serum was used for growth of 293T cells.
Monomeric gp120JR-FL and soluble CD4 were purchased from Progenics (Tarrytown, NY) and gp120JR-CSF was procured under contract from Advanced Products Enterprises (MD). The human anti-gp120 mAbs used in this study are IgGs b12 (Burton and Barbas 1994), 2G12 (Trkola et al., 1996) (provided by Gabriela Stiegler and Hermann Katinger), C11 (Moore et al., 1994) and F2A3 (provided by James Robinson). Polyclonal human IgG against HIV-1 (HIVIG) was obtained from the AIDS Research Reference Reagent Program, provided by John Mascola. Antibodies were biotinylated using EZ-Link Sulfo-NHS-Biotinylation Kit from Pierce. The polyclonal sheep Ab D7324 was purchased from Cliniqa (Fallbrook, CA). Mouse mAbs anti-HA (12CA5), anti-cmyc (9E10) and rat mAb anti-HA-HRP (3F10) were obtained from Roche. Fluorescent reagents goat-anti-mouse-Alexa 488 (GaM-488), GaM-Alexa 633, GaM-Alexa 546, Strepavidin–phycoerythin (SA–PE) and SA-Alexa 633 were obtained from Molecular Probes.
Variable loop-deleted gp140 mutants (Sanders et al., 2000) (plasmids were generously provided by James Binley) were generated by transient transfection (calcium phosphate precipitation) of 293T cells. At 16 h post-transfection the supernatants of the cell cultures were replaced with fresh media and incubated for an additional 72 h. The supernatants were collected, concentrated 
10-fold and stored at –20°C until use.
Vector modifications and library construction
The yeast display vector pPNL6 was received from M. Feldhaus, Pacific Northwest National Laboratory. Two SfiI restriction sites were inserted before and after the scFv, matching the SfiI sites of the pComb3X phage display vector, using the QuikChangeII site directed mutagenesis kit (Stratragene) with the following oligonucleotides: 1229Sfi: 5'-GGTGGTTCTGCTAGGGCCCAGGCGGCCTGCGGTGG CGG-3', 1229Sfi_AS: 5'-CCGCCACCGCAGGCCGCCTGGGCCCTAGCAGAACCACC-3', 1419Sfi: 5'-CAGGTCGACTGCGGCCAGGCCGGCCAAGGGGGCGGATCC-3' and 1419Sfi_AS: 5'-GGATCCGCCCCCTTGGCCGGCCTGGCCGCAGTCGACCTG-3'. The sequence of the modified vector, pPNLS, was verified by DNA sequencing. The FDA2 scFv kappa and lambda libraries for phage display were completed previously (Barbas et al., 2001; Zwick et al., 2001, 2003). Briefly, RNA was isolated from the bone marrow of patient FDA2, an HIV-1-seropositive individual with broad HIV-1 neutralizing Ab titers, and used to prepare the scFv libraries in pComb3X. The size of each library was estimated at 107 members. These libraries were excised from pComb3X by digestion with SfiI, gel purified and extracted from the gel with the Qiaquick gel extraction kit (Qiagen). The libraries were ligated into SfiI digested and purified pPNLS vector; the ligation reaction was purified using Qiaquick PCR purification kit (Qiagen) and transformed into XL1-Blue electroporation-competent cells (Stratagene). Dilution plates indicated the size of the libraries to be 109, which exceeds the library diversity by 2 orders of magnitude. Inserts of the correct size were found in 100% of tested vectors. The pPNLS-scFv libraries were then transformed into EBY100 using 10 ‘2x’ reactions of the high-efficiency lithium acetate transformation (Gietz and Woods, 2002); the reactions were pooled and grown in SD-CAA at 30°C to saturation (about 40 h). The size of each library in yeast was estimated at 1.5 x 107 and 2.7 x 107 members for the kappa and lamba libraries, respectively.
The yeast libraries were grown as previously described (Feldhaus et al., 2003). Typically yeast were grown in SD-CAA 18–22 h at 30°C and then transferred to SG/R-CAA for 16–18 h at 20°C in culture volumes appropriate for the size of the library. For the first two selection rounds at least 2 x 108 yeast cells were stained in 500 µl volumes; 1 x 107 yeast cells in 100 µl volumes were stained for subsequent sort rounds. Yeast cells were incubated with 100 nM gp120, 200 nM biotinylated anti-gp120 antibody (2G12 or C11), and 5 µg/ml anti-HA (12CA5) antibody for 30 min at room temperature in FACS wash buffer (PBS/0.5% BSA/2 mM EDTA), then washed three times in ice cold wash buffer. The cells were probed by incubation with 5 µg/ml each SA–PE and GaM-488 for 30 min on ice in the dark, then washed three times again and resuspended in 6 or 3 ml FACS wash buffer (depending on number of cells) for sorting by flow cytometry. Selections were performed using a BD Bioscience FACS Vantage DiVa set for purifying selection, and sort gates were determined to select the desired double positive cells. Collected cells were plated on SD-CAA plates with Penn/Strep and grown at 30°C for 2 days. Cell were then resuspended and amplified for the next round, or individual colonies were picked after the final selection round.
Analysis of single yeast clones was performed by first isolating the plasmid from the yeast cells using the Zymoprep yeast plasmid miniprep kit from Zymo Research. The plasmids rescued from yeast were then transformed into electrocompentent XL1-blue E. coli for amplification of plasmid DNA and purified using the QIAprep spin miniprep kit from Qiagen and the scFv insert was sequenced. A representative clone for each sequence was used for subsequent analysis. Individual yeast clones were grown, induced and stained for analysis by flow cytometry. Typically 5 x 105 cells were stained in 30 µl volumes with 30 min incubations and washed twice with 200 µl FACS wash buffer in a V-well 96-well plate. To determine approximate binding constants, 10 concentrations of gp120 (0–200 nM) were used in the presence and absence of 40 nM sCD4 (pre-complexed overnight at 4°C), cells were washed and then incubated on ice with biotinylated-HIVIG and anti-HA. After further washing, the cells were incubated on ice with fluorescent reagents, washed again and resuspended in 150 µl FACS wash buffer. Similarly loop-deleted mutants were titered from 0 to 50 nM (five concentrations); cells were washed then incubated on ice with biotinylated-HIVIG and anti-HA, washed again then incubated on ice with fluorescent reagents, washed and resuspended in l150 µl. For antibody competition, 100 nM gp120 and 100 nM biotinylated (competition) antibody (pre-complexed overnight at 4°C) and anti-HA were incubated with the yeast cells. The cells were washed and then incubated on ice with fluorescent reagents, washed again and resuspended in 150 µl of FACS wash buffer. A BD Bioscience FACSArray plate reader was used for all individual analysis and FlowJo software used for data interpretation.
Phage display, E. coli expression and ELISA binding assays
Individual scFv clones were excised from pPNLS by digestion with SfiI, gel purified and extracted from the gel with QIAquick gel extraction kit and ligated into similarly SfiI digested and purified pComb3x vector. The ligation reaction was transformed into XL1-Blue cells, which were grown for plasmid isolation. Incorporation of the correct scFv was verified by SfiI digest for insert size and by DNA sequencing.
Each scFv clone was displayed on phage by growth of phagemid-bearing XL1-blue for 5 h at 37°C (100 µl overnight culture in 2 ml media), addition of VCSM13 helper phage (109 pfu) and further growth for 2 h at 37°C, followed by addition of kanamycin (70 µg/ml). Following an overnight incubation at 30°C cells were pelleted and the phage-containing supernatant was used for enzyme-linked immunosorbant assay, ELISA. Each clone was also grown for soluble expression from BL21(DE3) cells by growth at 37°C to OD600 = 1, addition of 1 mM IPTG and overnight growth at 30°C. Cells were pelleted, resuspended in PBS, lyzed by freeze-thaw and cell debris pelleted. The scFv-containing supernatant (crude lysate) was utilized for ELISA.
For ELISA, plate wells were coated overnight at 4°C with 50 µl PBS containing gp120JR-FL, gp120JR-CSF or ovalbumin at 2 µg/ml, or, for capture ELISA, sheep Ab D7324 at 4 µg/ml. Wells were washed twice with ELISA wash buffer (PBS/0.5% Tween 20) then blocked with 3% BSA in PBS at 37°C for 1 h. For capture ELISA, 50 µl of gp120 (2 µg/ml) in the presence or absence of sCD4 (40 nM) (pre-complexed overnight at 4°C in ELISA buffer, PBS/1% BSA/0.02% Tween 20) was added and incubated at 37°C for 2 h, then wells were washed four times. Next, for both direct and capture ELISAs, 50 µl of either phage supernatant (undiluted) or crude lysate (titrated in ELISA buffer) was added and incubated for 1 h at room temperature, then the wells were washed four times. Anti-hemagglutinin peroxidase high-affinity 3F10 (Roche) diluted 1:1000 was incubated at room temperature for 30 min, and wells were washed four times. Fifty microliters of TMB (3,3',5,5' -tetramethylbenzidine) solution was prepared according to the manufacturer's instructions (Pierce) and was added to the wells; after about 10 min the reaction was stopped with 50 µl of 2 M H2SO4 and the OD450 was measured on a microplate reader (Molecular Devices).
|
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Generation of the yeast surface display vector pPNLS and subcloning of the immune scFv library
The first reported non-immune scFv library for yeast display utilized the vector pPNL6 (modified from pYD1) (Feldhaus et al., 2003). We modified pPNL6 to include two SfiI restriction enzyme sites that matched the cloning sites in the phage display vector pComb3X. This allowed scFv fragments to be shuttled between the yeast vector, designated pPNLS, and pComb3X. The first SfiI site was inserted directly after the HA affinity tag and (G4S)3 linker sequence, and the second site was inserted directly before the c-myc affinity tag ensuring that both tags would be present on yeast-displayed scFvs.
Preparation of scFv and Fab libraries from donor FDA2 has been previously completed and described (Zwick et al., 2001, 2003). Three mAb specificities from this library have been described in detail: Fab Z13 (Zwick et al., 2001), which targets the membrane proximal external region of HIV-1 gp41; scFv 4KG.5 (Zwick et al., 2003), which targets a unique HIV-1 gp120 epitope that distinguishes the mAb b12 from other CD4bs antibodies; and Fab X5 (Moulard et al., 2002; Labrijn et al., 2003), which targets a CD4i epitope on gp120. The antibody X5 has also been expressed as an scFv and for this study was used as a positive control for gp120 binding. Several additional antibody specificities were identified from the scFv and Fab phage libraries but have not been published.
There are two differences between a previously described non-immune scFv yeast library (Feldhaus et al., 2003) and the current FDA2 immune scFv yeast library: the order of the variable heavy (VH) and variable light (VL) domains are reversed in the FDA2 library so that the VL is first (Fig. 1), and the linker of the FDA2 library is (G4S)3RSS instead of (G4S)3. To ensure that these differences had no effect and that gp120 could be used as an antigen for yeast display, we first subcloned scFv X5 into pPNLS and verified binding to gp120 via FACS. Unlike many other selection protocols for yeast display, the antigen gp120 was not directly tagged, since we did not want to obscure any epitopes or alter gp120's conformation by biotinylation. Instead a mAb to a non-competitive epitope was biotinylated and used to sandwich gp120, which was then visualized with SE–phycoerythin (SA–PE). The binding affinity of scFv X5 for monomeric gp120 was measured by titering the amount of gp120 in the presence and absence of sCD4 and measuring the mean fluorescence intensity (MFI) of antigen binding; binding constants were 1.1 and 14.5 nM, respectively, in agreement with previously published results as estimated from ELISA for Fab X5 (2 and 10 nM) (Moulard et al., 2002).
|
Two scFv phage display libraries were originally generated from the FDA2 donor; both used the same heavy chain PCR pool and overlap PCR was used to combine with kappa and lambda light chains in separate libraries. Here each library of scFv-encoding cassettes was excised as a single SfiI fragment from pComb3X, ligated into similarly digested pPNLS and transformed into electrocompetent XL1-Blue E. coli. The number of independent colonies following transformation was
109, which is two orders of magnitude larger than the estimated diversity of the original libraries. These libraries were transfected into EBY100 yeast cells with an estimated 2 x 107 transformants. Since the original libraries in pComb3X were 107 in size, sequence composition of the libraries in both display formats is expected to be similar. Selection from yeast surface displayed immune scFv library using FACS
ScFv FDA2 yeast-displayed libraries were subjected to multiple rounds of affinity selection sorting for gp120 recognition, and the progress of the sort was monitored by the percentage of induced cells binding to gp120 (Fig. 2). After the first round, secondary-only controls (SE only and capture antibody) were used to determine the appropriate sort gate setting so only gp120-binding yeast were collected. The biotinylated antibodies for gp120 visualization were C11 and 2G12, which were used alternately to minimize selection of non-specific clones. These two antibodies were chosen because their epitopes show no or limited overlap with most of the known epitopes on gp120 (Moore and Sodroski, 1996). C11 binds to the C1–C5 region of gp120, which is buried in the native HIV-1 envelope trimer spike (Moore et al., 1994). 2G12 has a unique domain-swapped antibody structure and binds to the glycosylated ‘silent face’ of gp120 (Trkola et al., 1996; Scanlan et al., 2002; Calarese et al., 2003). Typically, only three–four rounds of selection were necessary to achieve 100% enrichment for specific gp120-binding clones. Following the final round, individual clones were picked and grown for characterization and plasmid isolation. Separate selection rounds were completed for gp120JR-FL and gp120JR-CSF, although most selected clones bound both (data not shown), and subsequent analysis utilized the selecting gp120 for each clone. (The original phage pannings utilized both gp120JR-FL and gp120JR-CSF, as well as several other antigens.). In an attempt to isolate scFv that recognized the CD4 binding site on gp120, an additional round of selection was performed on the gp120 binders in which gp120 was complexed to sCD4 and the non-binders isolated (Supplementary Fig. 1).
|
The set of antibody sequences obtained by yeast display were compared to those obtained from previous phage display selections. The third complementarity determining region of the heavy chain (CDR H3) was used for the comparison. Sequences that were very similar and appeared to be somatic variants of one another were grouped together. Some clones were observed that had the same heavy chain sequences but different light chain usage. Variant clones whose sequences closely matched those of previously identified specificities were counted as part of the same group, whereas novel sequences were grouped separately. Based on these grouping criteria, yeast sorting of the FDA2 scFv library identified six unique anti-gp120 clones, all of which had been previously described from phage display selection. Importantly, twelve additional sequence groups (10 heavy chain sequences) were identified using yeast display. Although the selection of non-binders to the gp120-sCD4 complex did not isolate any novel anti-gp120 clones, the negative selection was successful in that only CD4 binding site clones were observed (save for one non-gp120 binding scFv, sequence not shown). Table I summarizes the CDR loop sequences of the scFv, germline gene usage and number of isolated clones identified for each group; the sequences are arranged according to their heavy chain gene usage. The frequency with which a given clone was observed varied widely, from 1–43 times. In general, clones selected by phage display were observed more frequently than novel clones. The notable exception is the novel clone C18-2 which was observed 20 times.
|
Single scFv analysis using yeast surface display
The first step in analysis of the scFv clones was to determine the equilibrium binding constant (KD). For anti-gp120 antibodies, the KD is usually determined in the presence and absence of sCD4 because complexation with CD4 causes large conformational changes in gp120 and therefore can greatly affect antibody binding. Yeast displaying scFv were incubated with varying amounts of gp120, from 1 to 200 nM, in the presence or absence of 40 nM sCD4. Binding to gp120 was visualized during FACS analysis using biotinylated HIVIG (a polyclonal Ig preparation from seropositive individuals) and SA–PE. By measuring the MFI of antigen binding as a function of gp120 concentration the KD of each scFv could be determined (Table II). All isolated scFv had low nanomolar affinity for either gp120 or the gp120-sCD4 complex as would be expected for antibodies isolated from an immune library.
|
Many epitopes for antibody recognition have been defined on gp120. They include the CD4 binding site (CD4bs), the CD4 induced (CD4i) epitopes, the variable loops (e.g. V1, V2 and V3), and the N- and C-termini. On the basis of the KDs for gp120 with and without sCD4, the CD4bs and CD4i scFv were identified. To further characterize scFv specificities, we used antibodies of known specificities in competition experiments. This approach was preferred to peptide mapping since most antibodies to gp120 recognize conformational epitopes and competition with peptides can be inconclusive. Various biotinylated mAbs were pre-complexed with gp120 and then incubated with the scFv-bearing yeast cells (bio-mAbs had the same level of biotinylation, data not shown). If the scFv clones did not compete with the bio-mAb for gp120 binding, the antigen binding fluorescence was positive by FACS. As an example, Fig. 3 shows three yeast-displayed scFv (C02-7, 4KG.5 and C02-34) with the binding to gp120 visualized by three different biotinylated mAbs (b12, F2A3 and C11). When gp120 is complexed with the anti-V3 loop antibody F2A3 it can no longer bind to cells displaying the anti-V3 loop scFv C02-34. Similarly CD4 binding site antibodies and CD4-induced antibodies have overlapping epitopes and cannot bind to gp120 simultaneously, as shown with cells expressing C02-7 and mAb b12. The scFv 4KG.5 has previously been shown to bind to a unique epitope on gp120 that is dependent on all three variable loops and is enhanced in the presence of b12 (Zwick et al., 2003
). This result is clearly reproduced for 4KG.5 expressed on the surface of yeast (Fig. 3).
|
Additionally, to determine the dependence of our antibodies on the presence of the variable loops, we measured scFv binding to a select panel of HIV-1 envelope variants with truncations in either V1, V2 or V3. Rather than using purified proteins as is typical for FACS analysis, we found that crude supernatant was equally effective for determining antigen binding of yeast-displayed scFv. Wild-type gp140JR-FL and single variable loop truncated mutants (
V1, V2 and V3) were expressed from 293T cells and the supernatant collected, condensed and gp140 concentration estimated. The gp140 supernatants were titrated with each yeast-displayed scFv clone starting at roughly 50 nM and visualized with biotinylated HIVIG. Shown in Fig. 4 are binding curves for a few clones. As expected, clone 4KG.5 did not bind to any of the V mutants.
|
As summarized in Table II, the apparent equilibrium binding constants were rapidly obtained for all of the scFvs and the apparent gp120 binding epitopes were identified with the exception of D02-3. To obtain the fine specificity for D02-3 binding to gp120 further investigation is necessary, e.g. saturation alanine mutagenesis. Although we observed scFv that bound to most of the common epitopes, there was a larger than expected proportion of CD4i antibodies.
Origins of differences between selection from yeast and phage display investigated by transfer of yeast-selected scFvs to phage
Many new scFv were observed using yeast display selection that had not been previously selected by phage display. To gain a better understanding of why these scFvs had not been selected from phage panning, each yeast-derived scFv clone was transferred into the pComb3X vector for display on phage. Individual scFv were displayed on phage and binding to gp120 measured by ELISA with gp120 captured onto the plate by Ab in the presence or absence of sCD4 (Supplementary Fig. 2). Table III summarizes the results as the OD450 observed over background for each scFv. Boxes are colored based on the relationship to saturation with red indicating binding above half maximal. The OD450 signal represents a measure of the phage-scFv antigen interaction that is proportional to the affinity (and possibly avidity) for gp120 and to the efficiency of expression and folding of the scFv on phage. Clearly, there is a correlation between this measurement and selection by phage display: five of the six clones selected by phage display had strong OD450 signals either for gp120 or gp120 with sCD4 and one showed an intermediate OD450 signal (between one-quarter and half maximal). For the 12 scFv clones selected only by yeast display, two had strong signals and three had intermediate signals.
|
In addition to phage display of individual scFv, each scFv was expressed as a soluble fragment from E. coli using non-amber suppressing BL21(DE3) cells. Cells were resuspended in PBS and lysed using freeze-thaw. The supernatant (crude lysate) contains the unpurified scFv with other proteins. The crude lysate for each scFv was titrated for binding to gp120 immobilized on the ELISA plate, or captured in the presence or absence of sCD4 (Supplementary Fig. 3). For many of the soluble scFv it was not possible to obtain a saturating titration curve, so the reciprocal dilution for half-maximal OD450 was estimated (see Supplementary Table I). All but one of the scFv crude lysates showed some level of binding to gp120 or gp120 complexed with sCD4 suggesting that some level of E. coli expression of soluble scFv occurs even for those scFvs not selected by phage display. However, the scFv crude lysates that showed the best binding (affinity and expression) also had the best binding when expressed on phage and had been selected previously by phage display.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Here, we report the direct comparison of yeast and phage antibody display technologies. To our knowledge this is the first such comparison utilizing the same immune antibody library and selection antigen. It appears that for the FDA2 library yeast display allowed the isolation of many scFv that were not selected using phage display. In fact, yeast display selection identified all those clones previously identified by phage display and then twice as many clones again. The novel clones identified by yeast were not simply lower affinity clones. The affinities observed by display on yeast typically correlate well with measurements of purified scFv (Razai et al., 2005
), and all the observed scFv KD values were similar (low nanomolar) and reasonable for an immune library. The clones previously selected by phage display were largely those that gave the highest signals in ELISA measurements of phage-scFv binding to antigen. A number of factors may contribute to the strength of these ELISA signals and it is very difficult to separate relative contributions. However, it appears that the level of expression of correctly folded scFv on the phage surface is most important. This is corroborated by the observation that scFvs positively selected by phage display tend to be those that are well expressed as soluble proteins in E. coli. Further, when displayed on yeast all the scFv appeared to have comparable levels of expression. Therefore, we conclude that eukaryotic processing of the scFv is the greatest reason for the difference in selected clones from yeast and phage display. We also noted other advantages of yeast display in our selection procedures. The ability to visualize both the targeted and non-targeted populations during the selection was a major advantage of yeast display. For example, with FACS sorting we were able to readily isolate a population of specific yeast-scFv that lost binding to gp120 once complexed with sCD4. The stringency of selection is also easily varied using FACS sorting unlike typical phage selection protocols. With FACS, the sort gate can be varied to select a wide range of cells or only select the highest fluorescent cells allowing a variety of antibodies to be selected with the former or steering selection toward a single antibody with the latter. Typical phage selection techniques yield a small number of antibodies at the conclusion of panning which in many cases is the desired result. However, for some purposes, such as the rescue of relatively rare antibodies, it may be desirable to sample a greater range of the antibody response.
We also found yeast display selection to be somewhat less labor intensive than phage display selection, requiring fewer manipulations. Characterization of individual scFv on the surface of yeast allowed KD measurements to be rapidly obtained. This capability is especially advantageous when working with larger numbers of clones since the large-scale expression and purification of many scFv is very time consuming. We were also pleased to find that unpurified antigen can be reliably used for analysis purposes in yeast display, saving significant time in the characterization of scFv.
Interestingly, 50% of the sequences (55% of clones sequenced) appeared to bind to CD4i epitopes on gp120, even though no CD4 was used during the selection rounds. All of the CD4i antibodies had the same germline gene usage, IGVH 1-69, which has been observed previously for CD4i antibodies from other sources (Choe et al., 2003; Huang et al., 2004). It has been noted that tyrosine sulfation of some CD4i antibodies is important for their affinity for gp120, however since yeast lack the requisite tyrosyl protein sulfotransferases (Moore, 2003) the sulfation of scFv cannot account for the large number of CD4i clones observed. The most likely explanation is stability of these scFv. Although scFv stability is greatly dependent on the individual sequence it has been suggested that scFv from heavy chain germline gene families VH1, VH3 and VH5 have the most stable framework (Ewert et al., 2003). It is also possible that although monomeric gp120 alone was used for the selection its conformation as a monomer in solution more easily selects for CD4i antibodies over antibodies to other epitopes. Or the donor from whom the library is derived has a high preponderance of CD4i antibodies. Selections with Fab libraries from this donor and other donors could help clarify these observations.
In conclusion, we have shown that yeast display is able to sample an immune scFv library repertoire considerably more fully than phage display. It remains to be seen whether this increased sensitivity will allow the more reliable rescue of rare antibodies that have important and desirable functions.
|
Footnotes |
|---|
2 Present address: Genmab, Yalelaan 60, 3584 CM Utrecht, The Netherlands
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
The authors thank the flow cytometry core facility at The Scripps Research Institute for expert advice and flow cytometry cell sorting. D.R. Bowley was supported by a predoctoral fellowship from the National Science Foundation.
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Barbas C.F. III, Burton D.R., Scott J.K., Silverman G.J. (2001) Phage Display: A Laboratory Manual(Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY).
Boder E.T. and Wittrup K.D. (1997) Nat. Biotechnol 15:553–557.[CrossRef][Web of Science][Medline]
Bradbury A.R. and Marks J.D. (2004) J. Immunol. Methods 290:29–49.[CrossRef][Web of Science][Medline]
Burton D.R. and Barbas C.F. III. (1994) Adv. Immunol 57:191–280.[Web of Science][Medline]
Burton D.R., et al. (1994) Science 266:1024–1027.
Calarese D.A., et al. (2003) Science 300:2065–2071.
Choe H., et al. (2003) Cell 114:161–170.[CrossRef][Web of Science][Medline]
Ewert S., Huber T., Honegger A., Pluckthun A. (2003) J. Mol. Biol 325:531–553.[CrossRef][Web of Science][Medline]
Feldhaus M.J., et al. (2003) Nat. Biotechnol 21:163–170.[CrossRef][Web of Science][Medline]
Fenyo E.M., Albert J., McKeating J. (1996) Aids 10:Suppl A, S97–S106.
Gietz R.D. and Woods R.A. (2002) Methods Enzymol 350:87–96.[CrossRef][Web of Science][Medline]
Hoogenboom H.R. (2002) Methods Mol. Biol 178:1–37.[Medline]
Huang C.C., et al. (2004) Proc Natl. Acad. Sci. USA. 101:2706–2711.
Labrijn A.F., et al. (2003) J. Virol 77:10557–10565.
Moore K.L. (2003) J. Biol. Chem 278:24243–24246.
Moore J.P. and Sodroski J. (1996) J Virol 70:1863–1872.[Abstract]
Moore J.P., Cao Y., Leu J., Qin L., Korber B., Ho D.D. (1996) J. Virol 70:427–444.[Abstract]
Moore J.P., Willey R.L., Lewis G.K., Robinson J., Sodroski J. (1994) J Virol 68:6836–6847.
Moulard M., et al. (2002) Proc Natl. Acad. Sci. USA 99:6913–6918.
Parren P.W., Wang M., Trkola A., Binley J.M., Purtscher M., Katinger H., Moore J.P., Burton D.R. (1998) J. Virol 72:10270–10274.
Razai A., et al. (2005) J. Mol. Biol 351:158–169.[CrossRef][Web of Science][Medline]
Sanders R.W., Schiffner L., Master A., Kajumo F., Guo Y., Dragic T., Moore J.P., Binley J.M. (2000) J. Virol 74:5091–5100.
Scanlan C.N., Pantophlet R., Wormald M.R., Ollmann Saphire E., Stanfield R., Wilson I.A., Katinger H., Dwek R.A., Rudd P.M., Burton D.R. (2002) J. Virol 76:7306–7321.
Trkola A., Purtscher M., Muster T., Ballaun C., Buchacher A., Sullivan N., Srinivasan K., Sodroski J., Moore J.P., Katinger H. (1996) J. Virol 70:1100–1108.[Abstract]
Ulrich H.D., Patten P.A., Yang P.L., Romesberg F.E., Schultz P.G. (1995) Proc. Natl. Acad. Sci. USA 92:11907–11911.
van den Beucken T., Pieters H., Steukers M., van der Vaart M., Ladner R.C., Hoogenboom H.R., Hufton S.E. (2003) FEBS Lett 546:288–294.[CrossRef][Web of Science][Medline]
Vaughan T.J., Williams A.J., Pritchard K., Osbourn J.K., Pope A.R., Earnshaw J.C., McCafferty J., Hodits R.A., Wilton J., Johnson K.S. (1996) Nat. Biotechnol 14:309–314.[CrossRef][Web of Science][Medline]
Vujcic L.K. and Quinnan G.V. Jr. (1995) AIDS. Res. Hum. Retroviruses 11:783–787.[Web of Science][Medline]
Weaver-Feldhaus J.M., Lou J., Coleman J.R., Siegel R.W., Marks J.D., Feldhaus M.J. (2004) FEBS Lett 564:24–34.[CrossRef][Web of Science][Medline]
Winter G., Griffiths A.D., Hawkins R.E., Hoogenboom H.R. (1994) Annu. Rev. Immunol 12:33–455.
Zwick M.B., Kelleher R., Jensen R., Labrijn A.F., Wang M., Quinnan G.V.Jr., Parren P.W., Burton D.R. (2003) J. Virol 77:6965–6978.
Zwick M.B., et al. (2001) J. Virol 75:10892–10905.
Received November 29, 2006; accepted December 6, 2006.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  D. R. Bowley, T. M. Jones, D. R. Burton, and R. A. Lerner Libraries against libraries for combinatorial selection of replicating antigen-antibody pairs PNAS, February 3, 2009; 106(5): 1380 - 1385. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. R. Beerli, M. Bauer, R. B. Buser, M. Gwerder, S. Muntwiler, P. Maurer, P. Saudan, and M. F. Bachmann Isolation of human monoclonal antibodies by mammalian cell display PNAS, September 23, 2008; 105(38): 14336 - 14341. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. Saerens, G. H. Ghassabeh, and S. Muyldermans Antibody technology in proteomics Brief Funct Genomic Proteomic, July 1, 2008; 7(4): 275 - 282. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||