Abstract

The endonuclease P2A initiates the DNA replication of the bacteriophage P2 by making a covalent bond with its own phosphate backbone. This enzyme has now been exploited as a new in vitro display tool for antibody fragments. We have constructed genetic fusions of P2A with single-chain antibodies ( scFvs ). Linear DNA of these fusion proteins were processed in an in vitro coupled transcription–translation mixture of Escherichia coli S30 lysate. Complexes of scFv–P2A fusion proteins covalently bound to their own DNA were isolated after panning on immobilized antigen, and the enriched DNAs were recovered by PCR and prepared for the subsequent cycles of panning. We have demonstrated the enrichment of scFvs from spiked libraries and the specific selection of different anti-tetanus toxoid scFvs from a V-gene library with 50 million different members prepared from human lymphocytes. This covalent antibody display technology offers a complete in vitro selection system based exclusively on DNA–protein complexes.

INTRODUCTION

Antibodies are protein ligands with a wide range of biomedical applications. They have been displayed and evolved successfully by different in vivo - or in vitro -based selection methods using, for example, phage particles, yeast or bacteria cells ( 13 ). One limitation of the in vivo selection systems is the library size that could be generated. A large library is considered to be important to obtain high-affinity ligands. However, the efficiency of transfer of DNA into cells often limits the library size to 10 9 –10 10 members ( 36 ). In vitro -based antibody selection methods such as ribosome display or mRNA display have proven to be successful in the construction and selection of libraries with a high diversity and complexities (potentially up to 10 14 different members) ( 59 ). Both technologies are based on an initial selection of stabilized mRNA–protein complexes ( 410 ).

All the present antibody selection systems only indirectly link phenotype and genotype via cell membranes, phage–protein coat, mRNA–ribosomes, protein–puromycin–mRNA complexes or water/oil emulsions ( 13 , 7 , 8 , 11 , 12 ). In nature, however, there is a protein from the bacteriophage P2 that offers a direct DNA–antibody selection system. This phage replicates by attaching its early gene product P2A to its own DNA. The first genetic evidence of the cis -activity in vivo was obtained when the wild-type P2 phage did not complement mutations in P2A ( 13 ). P2A initiates the rolling circle replication of the P2 phage in vivo . Its catalytic tyrosine residue (Y454) makes a single-stranded-specific nick at the viral origin of replication, located at base pair 1860 inside the 2.3 kb P2A gene, and forms a covalent bond with the 5′-phosphate group of the coding strand ( 1416 ) ( Figure 1a ).

A single chain antibody (scFv) can be genetically fused to the P2A protein creating the smallest imaginable antibody selection particle: a protein and its gene ( Figure 1b ). Covalent antibody display (CAD) exploits the demonstrated in vivocis -activity of P2A in an in vitro selection system: a fusion protein of P2A and an scFv antibody binds to the same molecule of DNA from which it has been expressed. Following in vitro coupled transcription and translation, the P2A protein makes a covalent link between scFv genotype and scFv phenotype, by producing a stable protein–DNA complex ( 14–17 ). P2A may thus be exploited to select scFvs from a library by using only in vitro methods. These antibody–DNA complexes can be isolated with standard affinity selection strategies. Specific complexes are enriched, eluted and rescued by PCR amplification ( Figure 1c ).

In the present study, we have demonstrated the suitability of P2A for specific selection of scFvs. Fusion proteins of scFv–P2A were expressed and DNA–antibody complexes were specifically recovered on antigen-coated solid phase. In addition, we have applied this technology to select antibodies from spiked and medium complex libraries. We propose that CAD can be exploited as a complete and independent in vitro antibody display tool for affinity selections.

MATERIALS AND METHODS

PCR cloning and assembly

The scFvs anti-phOx [phOx, 4-ethoxymethylene-2-phenyl-2-oxazoline-5-one ( 18 ); anti-phOx ( 19 )] and anti-DOB [DOB, 2,5-dimethoxy-4-bromo-amphetamine ( 20 ); in house made anti-DOB (unpublished data), specific against DOB] were fused to either the N-terminal or C-terminal position of P2A with standard PCR cloning techniques, attaching a GSGSGS linker containing suitable flanking restriction sites (EcoRI, NotI, XhoI or NcoI) and two stop codons at the 3′ end ( Figure 2 ). A vector tacP2aHa (5926 bp) containing the P2A–HA gene under the control of a tac promoter was supplied by Isogenica Ltd. Pfu Turbo DNA polymerase (Stratagene) was applied for generating GS-linker-scFv products for cloning. The PCR mixture was composed of 200 μM dNTP mixture, 30 ng vector DNA-template, 0.6 μM primers GsDOB3F (aaattaaaa ctcgag ggttctggctccggtt ccatgg cggaagtgcagctggtgc; XhoI and NcoI sites are underlined and GSGSGS sequence is in italics)/EcoRIPhog (ttttttt gaattc tatca gtgatggtgatggtgatg tgagtttagg; EcoRI site is underlined and HHHHHH sequence is in italics) or GsPhOxF (aaattaaaa ctcgag ggttctggctccggtt ccatgg cccaggtgcagctggtgc; XhoI and NcoI sites are underlined and GSGSGS sequence is in italics)/EcoRIPhog in the Pfu reaction buffer. Initial denaturation was at 94°C for 5 min followed by 30 cycles at 94°C (1 min), 63°C (1 min) and 72°C (2 min 30 s) in an Eppendorf Mastercycler Gradient PCR machine. The final elongation step was at 72°C for 5 min. The GS-linker-P2A for cloning scFvs at the 5′ of P2A was produced by Pwo Polymerase (Roche, Norway) at an annealing temperature of 65°C [primers: GsP2af (aaattaaat gcggccgc gggttctggctccggtt ccatgg ccgttaaagcctccggg; NotI and NcoI sites are underlined and GSGSGS sequence is in italics)/Lanrb (gtggataaccgtattaccgcc)]. The Pwo PCR programme was as follows: 94°C for 2 min, with initial 10 cycles at 94°C (30 s), 65°C (1 min) and 72°C (2 min), then 20 cycles at 94°C (30 s), 65°C (1 min) and 72°C (2 min) increasing the extension time with 5 s per cycle, and with a final elongation step at 72°C for 7 min. To clone small scFv libraries (scFv) n into the scFv–GSGSGS–P2A vector via NcoI/NotI sites, it was necessary to delete the second NcoI site at the beginning of P2A ( Figure 2 ). This site was removed with the primers GSP2Afnew (aaattaaat gcggccgc g ggttctggctccggttct atggccgttaaagcctccggg; NotI site is underlined, GSGSGS sequence is in italics) and Lanrb and Pwo polymerase as described previously. Standard methods were applied for heat transformation of plasmid DNA into chemical competent Escherichia coli DH5-α or Origami cells (Novagen). The transformed cells were grown in SOC medium at 37°C for 1 h shaking at 260 r.p.m. and plated on SOC/ampicillin (100 μg/ml) agar dishes for 15–20 h at 37°C.

Input linear DNA of P2A–scFv and scFv–P2A clones for coupled transcription and translation were made by either Pwo or Tgo DNA polymerase (Roche, Norway) with a hot-start technique. Preparations of plasmids (ligation mixtures or minipreps; 0.5 μg) were used as template for PCR, and the primers P2AampF (gcttcagtaagccagatgctac, 30 pmol) and LAMPB (tacaccgaactgagata cctac, 30 pmol) were chosen for producing the ∼4 kb DNA fragment comprising tac-promoter, scFv and P2A ( Figure 2 ). The PCR programme for Tgo DNA polymerase was as follows: 94°C (2 min), cycling 30 times at 94°C (30 s), 65°C (1 min) and 72°C (3 min) followed by a 7 min incubation at 72°C. The linear DNA fragments were separated on agarose gel and purified with Qiaquick gel extraction kit (Qiagen). Extra washing steps were necessary in order to remove all the salt components inhibiting transcription and translation. The products were adjusted to a high-concentration sample (1–3 μg/μl) by alcohol precipitation and applied in coupled transcription and translation.

In vitro transcription-translation

Reaction buffer (2.5×) and E.coli [strain SL-119, ( 21 )] S30 extract ( 22 ) with 1 mM DTT (Sigma) were applied. For a 50 μl reaction in an Eppendorf tube, the following components were assembled on ice: S30 2.5× reaction buffer (20 μl), E.coli S30 extract (15 μl), ∼1 mM l -methionine (Sigma), 1 μM protein disulfide isomerase (PDI; Sigma), 40 U RNasin (Promega), nuclease-free water, 0.1% BSA and DNA (1–2.5 μg). Generally, 5–10 μg of DNA was added for each round of selection. Transcription and translation reaction was started by adding DNA (1–5 μg per 50 μl reaction) and processed for 60 min at 30°C. After 40 min of incubation, 2 mM of oxidized and reduced glutathione were added. The reaction was stopped by diluting the mixture in ice-cold panning buffer and incubated with an immobilized target. The transcription and translation extract from Promega ( E.coli S30 extract for linear templates) was also applied, but it contains an excess of DTT, which inhibits the folding of antibodies.

Affinity selections

Dynabeads M-280 Tosylactivated (Dynal Biotech, Oslo, Norway) were coated with anti-haemagglutinin (anti-HA), or with the antigens phOx–BSA ( 18 ) or DOB–BSA ( 20 ) (6 μg antibody or 20 μg hapten–BSA conjugate per mg beads; 20 mg beads/ml) in 0.1 M sodium carbonate, pH 9.6 at 37°C for 20 h on an Eppendorf thermomixer at 1200 r.p.m. Then, they were washed three times in phosphate-buffered saline (PBS), pH 7.4 with 0.1% BSA at 20°C and blocked for another 20 h at 20°C (1200 r.p.m.) in 0.3 M Tris–HCl, pH 8.8 with 0.1% BSA. Finally, they were washed in PBS, pH 7.4 and stored at a concentration of 20 mg beads per ml. For the E7 scFv, an scFv specific for an outer membrane epitope of Neisseria meningitidis (E7), the outer membrane vesicles (OMV) made from the bacteria, were applied as antigen ( 23 ). Immuno tubes (Maxisorp, Nunc, Denmark) were coated with antigens in PBS (pH 7.4) overnight at 20°C on a blood mixer. The affinity matrices were pre-incubated with panning buffer SuperBlock (Pierce) added ∼0.2 mg/ml herring sperm DNA (Sigma), 0.25% (w/v) heparin and 1% Tween-20 ( 6 , 24 , 25 ) for 1–2 h at 20°C on a blood mixer, and then rinsed three times in PBS before panning mixture was added.

DNA–protein complexes were captured by first diluting the transcription and translation mixture into panning buffer, then incubating on an affinity coated matrix for 2–4 h rotation at 4°C or at 20°C on a blood mixer. The enriched complexes were thoroughly washed in PBS with 0.1% Tween-20, panning buffer (including an incubation step at 20°C for 30 min) and in PBS.

The DNA was eluted using proteinase K (50 μg/ml; Sigma) in TE buffer, pH 8.0 for 45 min at 50°C ( 26 ) and ethanol precipitated in seeDNA (Amersham Pharmacia Biotech).

PCR-rescue, detection of P2A–scFv fusions and processing scFv clones

If the P2A–scFv fusions have been expressed and purified, the enriched DNA of the P2A–scFv construct can be identified and amplified by PCR. The seeDNA precipitated DNA was dissolved in 10 μl of nuclease-free sterile MQ-water and 1–5 μl was used for each PCR-screen. If the input DNA in the transcription and translation mixture was made by PCR, it was not possible to use the same primers P2AampF and LAMPB in the PCR mixture as was done for producing it ( Figure 2 ). This was probably due to the partial degradation of DNA after incubation. However, by using primer combinations NcoDOBF(atggcggaagtgcagctggtg)/NotDOBB(gccgcgaagacagatggtgc), NcoPhOxF(atggcccaggtgcagctggtgc)/NotPhOxB (gccgcacctaggacggtgacc) or scFvseqF (cctgttgacaattaatcatggctcg)/scFvseqB (cctgcggcaaatgctgacgg) ( Figure 2 ), they generated products comprising just the scFv gene and short flanking regions (∼0.8–1 kb). For primer combinations NcoDOBF/NotDOBB and NcoPhOxF/NotPhOxB, an annealing temperature at 68°C was chosen, and 55°C for scFvseqF/scFvseqB. For detection purposes, DynaZyme II DNA polymerase (Finnzymes, Finland) was used, and for amplification of new products for next round of selection, Pwo polymerase (Roche) was chosen.

The recovered and amplified scFv DNA from each round of CAD was either cloned into an assay expression vector pHOG21 (having a lac promoter and c -myc/His 6 -tags) ( 27 ) or assembled with tac promoter and P2A gene using restriction/ligation methods into vector tacP2aHa (NcoI/NotI sites; Figure 2 ). DNA for the next round of transcription and translation was generated by PCR using ligation mixture as mentioned above. For the expression of soluble scFvs in 96 deep-well plates, the pHOG21 expression vector constructs were transformed into XL-1 Blue cells and induced in Luria–Bertani (LB) medium containing 100 μg/ml of ampicillin (A) with 100 μM isopropyl-β- d -thiogalactopyranoside (IPTG) at 30°C and shaken for 16 h. The expressed scFvs were detected using enzyme-linked immunosorbent assay (ELISA), or were assayed as described below.

SDS–PAGE, western blotting and enhanced chemiluminescence (ECL) detection

The samples, added SDS-sample buffer with 2-mercaptoethanol, were separated on a 12.5% SDS–PAGE criterion precasted gel (Bio-Rad), electro blotted onto nitrocellulose membranes, washed (PBS), blocked with Blotto (1% dry milk in PBS), and incubated in anti- c -Myc (diluted 1:5000 in Blotto; Invitrogen) or anti-HA (1:2000 in Blotto; Roche) solutions. The membrane was washed (PBS with 0.1% Tween-20) and incubated with detection antibody [rabbit anti-mouse-horseradish peroxidase (HRP); DAKO, Denmark]. The HRP activity on membrane was visualized in ECL-western blotting reagents RPN 2106, Hyperfilm ECL (both Amersham Pharmacia Biotech) and a film developer.

ELISA and filter screening

Maxisorp 96-well plates (Nunc, Denmark) were coated with antigen, and expressed scFvs were detected via their c -Myc tags using antibodies/enzymes as described above. 3,3′,5,5′-Tetramethylbenzidene (TMB) was used as a substrate. Filter screening of 3040 bacterial colonies was performed as described previously ( 28 ). To prepare target filters (0.45 μm pore size cellulose nitrate; Schleicher & Schuell Protan BA 85), these were cut to 22 × 7 cm 2 and coated with 10 ml of phOx BSA or DOB BSA at 0.1 mg/ml. Generally, clones were high-density gridded with a Qpix robot (Genetix, New Milton, UK) on a BSA-coated filter surfacing a 22 × 22 cm 2 LB agar bioassay plate containing 30 μg/ml of tetracycline (T), 100 μg/ml of ampicillin (A) and 0.1 M glucose (G), and grown for 16 h at 37°C. Then, the BSA filter with clones was lifted on top of a sandwich with an antigen-coated filter in the middle and fresh LB agar plate with 100 μM IPTG and A at the bottom, and induced for 16 h at 30°C. The active scFvs will leak from the periplasm of the colonies to the antigen-coated filter underneath. This filter was developed like a western membrane as described above.

RESULTS AND DISCUSSION

Constructing a basic scFv–P2A model system

The A protein of the P2 phage can be used as a potential antibody display tool ( Figure 1b and c ). To test its suitability, we chose two well-characterized scFvs against the haptens: phOx ( 18 ) and DOB ( 20 ). The scFvs were genetically fused to P2A at two different positions connected by a (GS) 3 -linker and expressed from a tac-promoter. The scFv N-terminal to P2A with an HA-tag: tacPO-scFv-(GS) 3 -P2A-HA and scFv C-terminal to P2A with c -Myc tag: tacPO-P2A-(GS) 3 -scFv-cmyc-His 6 were constructed ( Figure 2 ). The fusion proteins and P2A–HA alone were translated in a standard E.coli S30 extract for linear templates (Promega) and sharp bands corresponding to the ∼90 kDa P2A–HA and the ∼110 kDa P2A–scFv proteins were clearly visible, showing efficient in vitro expression ( Figure 3a ). Estimating the yield of protein, the sensitivity of ECL (10 −11 M) indicated a protein band (∼1 ng) where the input of DNA was 1 μg ( Figure 3a ). This suggests that ∼3% of input DNA produces proteins, which would require around 30–40 copies of each scFv to be present in the library to ensure isolation of any unique scFv members.

Antibodies are normally functionally folded in oxidizing or slightly reducing environments ( 6 , 24 ). To allow RNA polymerase activity, which requires a reducing environment, a specially made transcription and translation mixture was prepared (see Materials and Methods). Immediately after transcription and translation, the protein–DNA complexes were enriched onto target antigen-coated beads, eluted and detected by PCR.

The display of functional scFv was evaluated at both the N- and C-terminus of P2A. Figure 3b shows the recovery of rescued DNA for N- and C-terminal position of anti-phOxscFv relative to P2A . For the N-terminal position ( scFv–P2A ), there was more DNA associated with phOx–BSA than BSA beads (target PCR signal > background PCR signal). This illustrates that the DNA enrichment was well correlated with antibody activity. Similar results were also obtained for the anti-DOB fusion proteins (data not shown). In contrast, the C-terminal fusions ( P2A–scFv ) showed no specific enrichment. It is possible that the linker length was too short to accommodate proper folding and functionality of these fusions, or they were structurally destabilized. Premature termination of the translation may also occur if P2A nicks DNA before scFv is synthesized. This may occur more frequently if the scFv is at the C-terminal end of P2A rather than at the N-terminal end.

Different additives improve the functionality of covalent antibody display

Adjusting the redox potential of the transcription and translation mixture generated active fusion proteins associated with DNA, and adding BSA increased the yield of DNA recovered by PCR and improved the signal-to-noise ratio relative to no addition ( Figure 3c ). BSA may stabilize enzymes and components in the transcription and translation mixture during in vitro synthesis, even though BSA also contributes to a stronger unspecific adsorption of DNA ( Figure 3c ). Furthermore, the samples were treated with RNase after the protein–DNA complexes were formed ( Figure 3c ). A similar pre-treatment has been applied for ribosome display, using DNase to remove DNA ( 29 ). RNase increased the signal-to-noise ratio confirming that mRNA is not involved during selection (as for ribosome display). The enriched active species therefore consist of pure DNA–protein complexes. In addition, the coupled transcription and translation mixture of an anti-phOx-P2A-HA DNA construct ( Figure 2 , bottom) was incubated in 6.3 M urea before it was diluted in panning buffer and enriched on anti-HA coated tubes. Still, this harsh treatment made it possible to specifically enrich its DNA on anti-HA-coated tubes (data not shown), and thus did not release the anti-phOx-P2A-HA protein from its mother DNA. This again confirms that the anti-phOx-P2A-HA protein is covalently bound to DNA as shown previously for P2A–HA proteins ( 1316 ).

Enrichment experiments

To demonstrate the recovery of specific antibody DNA–protein complexes from a complex mixture, specific enrichment experiments were performed. In a population where there is only one copy of a specific P2A-scFv gene among, e.g. 10 7 other different non-active P2A-scFv genes, the cis -activity is related to the probability of recovering the specific P2A gene product binding to its own mother DNA. If the P2A protein only finds it own mother DNA molecule, a direct enrichment of P2A protein–DNA complexes is productive. However, all selection systems contain a certain amount of noise (non-relevant gene products) raising the threshold required to detect rare binders from a large diverse library early in the selection process (round 1–2). If there were no cis -activity, the amount of randomness involved would possibly block the enrichment of the specific gene and gene product, even after 10–20 cycles. On the other hand, an early enrichment of active library members from a very diverse mixture should demonstrate that there is a cis -directed selection.

An scFv antibody library from human lymphocytes (PBL) was constructed by RT–PCR using IGHV-D-J , IGKV-J and IGLV-J genes. The library having a diversity of n = 5.0 × 10 7 different clones was cloned as an N-terminal fusion to P2A, (scFv)n–P2A . DNA of anti-phOx–P2A or anti-DOB–P2A fusions were spiked into the library at a ratio of 1:25 000 (anti-phOx) and 1:1000 and 1:100 000 (anti-DOB) and subjected to a complete covalent display selection cycle ( Figure 1c ). Two spiking strategies were employed. For the selection of anti-DOB scFvs spiked into the (scFv)nP2A library at a ratio of 1:10 3 , 4.8 ng of anti-DOB–P2A DNA (1.1 × 10 9 molecules) and 5 μg (1.1 × 10 12 molecules) of library DNA were used. Lowering the same spiking ratio to 1:10 5 , 48 pg of anti-DOB–P2A (1.1 × 10 7 DNA molecules) was added into 5 μg (1.1 × 10 12 DNA molecules) of library DNA. For the selection of anti-phOx, 400 pg of anti-phOx–P2A DNA was spiked into 10 μg of (scFv)nP2A library (1:25 000). The recovered DNA from each round was analysed using filter screen as shown for anti-phOx in Figure 4a . As expected, there were no positive candidates before the selection of 3040 clones in the unselected library. However, after round 1–3 (R1–R3) of CAD, the number of enriched clones increased from 6 (R1), 49 (R2) to 82 clones (R3). The enrichment after each round is also illustrated in Figure 4b . The clones were MvaI digested ( 30 ) and their restriction fragment length polymorphism (RFLP) patterns confirmed their anti-phOx–P2A identity. Enrichment was also seen after analysing 1:1000 and 1:100 000 spiking of anti-DOB–P2A into the same library, respectively. After R1 of CAD 28 positive anti-DOB–P2A clones (1:10 3 spiking) were identified from a random sample of 3040 clones (14-fold enrichment), and similarly 8 clones (∼300-fold enrichment) for the 1:10 5 spiking (data not shown). On studying the results above, where the ratio of scFv–PLA spiked with other irrelevant scFv–PLA DNA molecules is l:25 000, the likelihood that the correct scFv is randomly chosen, is very low. The probability of selecting 82 positive clones from a random sample of 3040 candidates with a population of 5 × 10 7 and having 1 positive among 25 000 clone is as low as 10 −13 [ P , N (0,1)]. This argues strongly that the one given molecule of P2A should bind to its own molecule of DNA that directed its synthesis.

The translation complex formed among DNA, mRNA, ribosomes and proteins is common in prokaryotes. This implies the production of extra mRNAs and an overexpression of the P2A protein. Liu and Haggård-Ljungquist ( 15 ) describe the inability to complement an amber A mutant by the overexpression of normal A protein on a second plasmid. This indicates high-fidelity cis -activity. The authors suggest either compartmentalization of ‘A’ protein or rapid inactivation of A as the mechanism for controlling cis -activity. It may be that the strong cis -effect is due to the coupling of P2A with single-stranded DNA that is available only during the melting of double-stranded DNA with RNA polymerase, and that other free P2A proteins may be inactive for nicking. P2A cannot bind to double-stranded DNA ( 15 ), and thus, free scFv–P2A molecules should not be able to bind to untranscribed double-stranded DNA. It is also possible to purify P2A without covalently bound DNA, suggesting that P2A can be released from the ribosome without binding to DNA. The experiments, however, demonstrate a specific recovery of protein–DNA complexes. The downward trend during selections ( Figure 4 ) may be explained wherein, for each round of selection, more active scFv–P2A fusion proteins are formed. It is most likely that the majority of these protein complexes were not bound to DNA. However, this does not influence the cis -activity since only single-stranded DNA is nicked by P2A. On the other hand, it affected the panning. The free, uncomplexed scFv–P2A fusion proteins compete with the corresponding DNA–protein complexes. Thus, the more the concentration of scFv–P2A protein fusions is increased, the less the DNA–protein complexes are isolated. Panning a more complex library (e.g. n = 10 14 ) should not however involve this type of problem since the initial concentration of expressed active proteins is very low and should not compete with DNA–protein complexes. This was also seen from the spiking experiments. Starting with a more diverse library (1:10 5 ), the initial enrichment from R0 to R1 was larger (300-fold) compared to less diverse spikings 1:2.5 × 10 3 and 1:10 3 that only had 50- and 14-fold enrichments, respectively. The CAD method has to be developed further to deal with these technical problems, which probably only arise at the final enrichment cycles. It may be possible to first purify all DNA before selection on protein complexes. The initial isolation of DNA could be performed with biotinylated DNA and SA-beads before subjecting it to an ordinary panning against active protein.

Panning tetanus toxoid scFvs from a 5 × 10 7 library

The human antibody library, (scFv)nP2A , was also selected against tetanus toxoid (TTx). The serum antibody titre of the donor against TTx was found to be about 1:25 000, implying that it should be possible to identify anti-TTx scFv antibodies derived from donor lymphocytes. For each round of selection against TTx, 10 μg of DNA of the (scFv)n–P2A library was applied as an input for coupled transcription and translation. Indeed, as shown in Figure 5 , after one round (R1) of CAD several anti-TTx scFvs were identified by ELISA. The total yield of available anti-TTx–P2A DNA after each round of selection was estimated from anti-phOx–P2A DNA standards at known concentrations (pg range). The input DNA was correlated with the corresponding PCR yields of anti-TTx scFv DNA versus anti-phOx DNA standards on an agarose gel. After the first round of selection (R1), the amount of input DNA for PCR (recovered DNA) was ∼0.2 pg, which equals a DNA recovered/input ratio of 1/50 000 000. After two and three rounds of selection (R2 and R3), the estimated ratio increased to 1/2 500 000 and 1/420 000, respectively. The CAD-enriched library was also analysed by filter screen with no scFvs isolated before selection (background, 3040 colonies screened). However, after R1 of CAD, the number of positives increased to 26 (26/3040). Most of these clones bound BSA (TTx-5, Figure 5 ). Nevertheless, some anti-TTx scFvs not cross-reacting with BSA were confirmed by ELISA. After R2 and R3, other scFvs were rescued and TTx-12 isolated after two rounds of CAD was specific for TTx. Clones were also analysed by RFLP. The R1 clones were unique. However, TTx-8 and TTx-12 (both R2 clones) had identical RLFP patterns and DNA sequence ( Table 1 ). The same HCDR3 sequence also appeared for TTx-17 (R3 clone), although there was a difference in their LCDR3 region ( Table 1 ), suggesting a sequence enrichment during R1–R3 of CAD. Furthermore, the HCDR3 sequences in this study were aligned with known human anti-TTx antibody sequences [ Table 1 and ( 31 , 32 )]. The residues at the C-terminal region of HCDR3 of TTx-12/TTx-17 (R2/R3 clones, respectively) were very similar to 6-ATT ( 33 ), suggesting a selection on the same epitope of TTx. Similarly, most of the HCDR3 sequences compared in Table 1 have a positively charged N-terminal (at position 105 or 106) and an Asp residue at position 116. For the light chains, they all contain a Ser in the middle of the loop (position 109 or 110).

CAD as a new in vitro antibody display tool

We have demonstrated the successful selections from two independent spiking experiments (phOx and DOB fusions) and from an immunized scFv library with a complexity of 10 7 . The method should thus be well suited for in vitro scFv selections from libraries. However, in order to fully demonstrate its potential, a more diverse library should be screened. P2A can be applied as a general panning tool for antibody library selections at the N-terminal end of the endonuclease. The advantages for such a selection system are numerous, compared to the existing phage display and in vitro mRNA display methods. CAD is the only antibody selection method providing a direct covalent link between an scFv gene and its protein ( 1416 ), increasing the chemical stability of the panning complex. Table 2 compares the more common methods of in vitro selection, and all methods have their advantages and disadvantages. The new DNA-based cis -display methods of panning are however very promising. Generally, CAD may facilitate a very rapid library generation using PCR methods with increased library complexity size (potentially up to 10 13 scFv members) and a fast cycle time. In a few hours, unique scFvs can be enriched, isolated and directly amplified for the next rounds of selection ( Figure 1c ). The system should be very well suited for scaling up and automation, and may be a future standard for in vitro antibody selections. However, at present, the CAD methods must be made more robust to overcome the product inhibition of uncomplexed fusion proteins without DNA. CAD offers a covalent antibody selection tool as an alternative to the latest non-covalent in vitro display selection method of peptides and proteins with the RepA protein ( 36 ).

Figure 1

( a ) The endonuclease P2A (761 residues, 86.3 kDa) makes a single-stranded site-specific nick at Ori of replication (…CCT CGG, *), located inside its own gene at position 1860, and becomes covalently attached (via Y454) to the 5′ phosphate of its own DNA ( 1416 ). ( b ) CAD is the exploitation of P2A to select antibodies or antibody fragments genetically fused to it. In prokaryotes, the transcription and translation are coupled, and the start of translation normally takes place before the transcription is finished ( 17 ). The figure is a model of the complex formation among DNA, RNA polymerase, ribosomes, mRNA and nascent P2A–scFv fusion proteins (colours match the gene products). The P2A protein part of the P2A–scFv fusion protein indirectly attaches the genotype of the scFv covalently to its phenotype. (L = linker). ( c ) Selection cycle for CAD. An scFv repertoire is assembled with the P2A gene using PCR methods and fresh P2A and tac-promoter (1). Protein–DNA complexes are being produced in an in vitro E.coli S30 coupled transcription–translation mixture (2) and selected on the immobilized target (3 and 4). Retained members are eluted and DNA for the next cycle is prepared using PCR (5). Figures are not drawn to scale.

Figure 1

( a ) The endonuclease P2A (761 residues, 86.3 kDa) makes a single-stranded site-specific nick at Ori of replication (…CCT CGG, *), located inside its own gene at position 1860, and becomes covalently attached (via Y454) to the 5′ phosphate of its own DNA ( 1416 ). ( b ) CAD is the exploitation of P2A to select antibodies or antibody fragments genetically fused to it. In prokaryotes, the transcription and translation are coupled, and the start of translation normally takes place before the transcription is finished ( 17 ). The figure is a model of the complex formation among DNA, RNA polymerase, ribosomes, mRNA and nascent P2A–scFv fusion proteins (colours match the gene products). The P2A protein part of the P2A–scFv fusion protein indirectly attaches the genotype of the scFv covalently to its phenotype. (L = linker). ( c ) Selection cycle for CAD. An scFv repertoire is assembled with the P2A gene using PCR methods and fresh P2A and tac-promoter (1). Protein–DNA complexes are being produced in an in vitro E.coli S30 coupled transcription–translation mixture (2) and selected on the immobilized target (3 and 4). Retained members are eluted and DNA for the next cycle is prepared using PCR (5). Figures are not drawn to scale.

Figure 2

Primer and restriction sites for rescue PCR and assembly of DNA for next rounds of CAD. Top: P2A–HA construct; middle: P2A–scFv ; and bottom: scFv–P2A . HA, haemagglutinin peptide epitope of human influenza virus (YPYDVPDYA); c -Myc, myc oncogene epitope (EQKLISEEDL); His 6 , hexahistidine sequence (HHHHHH).

Figure 2

Primer and restriction sites for rescue PCR and assembly of DNA for next rounds of CAD. Top: P2A–HA construct; middle: P2A–scFv ; and bottom: scFv–P2A . HA, haemagglutinin peptide epitope of human influenza virus (YPYDVPDYA); c -Myc, myc oncogene epitope (EQKLISEEDL); His 6 , hexahistidine sequence (HHHHHH).

Figure 3

( a ) Western/ECL identification of P2A-HA (lane 1) and P2A-anti-DOB3-c-Myc (lane 2) proteins expressed in a coupled transcription-translation mixture (Promega E.coli S30 extract for linear templates). The proteins were separated on 12.5% criterion polyacrylamide gel (Bio-Rad), blotted onto nitrocellulose membrane and detected by anti- c -Myc/anti-HA and rabbit anti-mouse IgG-HRP/ECL. ( b ) Agarose gel of rescue PCR of DNA from anti-phOx-P2A (left panel) or P2A-anti-phOx (right panel) protein–DNA complexes selected onto target T (phOx–BSA-coated Dynabeads) or negative control NC (BSA-coated Dynabeads). The protein–DNA complexes were generated in E.coli S30 lysate added 1 μM PDI and 1 mM DTT. They were eluted with proteinase K digestion, alcohol precipitated and detected by PCR. ( c ) Agarose gel of rescue PCR of DNA from scFv(E7)–P2A protein–DNA complexes selected on N.meningitidis outer membrane vesicles (T) or superblock (Pierce, NC). The effect of RNase (Promega) treating samples immediately after transcription–translation (at 30°C, 10 U) or adding 0.1% BSA during transcription and translation. The DNA bands from no additions/post-treatments are also shown (0).

Figure 3

( a ) Western/ECL identification of P2A-HA (lane 1) and P2A-anti-DOB3-c-Myc (lane 2) proteins expressed in a coupled transcription-translation mixture (Promega E.coli S30 extract for linear templates). The proteins were separated on 12.5% criterion polyacrylamide gel (Bio-Rad), blotted onto nitrocellulose membrane and detected by anti- c -Myc/anti-HA and rabbit anti-mouse IgG-HRP/ECL. ( b ) Agarose gel of rescue PCR of DNA from anti-phOx-P2A (left panel) or P2A-anti-phOx (right panel) protein–DNA complexes selected onto target T (phOx–BSA-coated Dynabeads) or negative control NC (BSA-coated Dynabeads). The protein–DNA complexes were generated in E.coli S30 lysate added 1 μM PDI and 1 mM DTT. They were eluted with proteinase K digestion, alcohol precipitated and detected by PCR. ( c ) Agarose gel of rescue PCR of DNA from scFv(E7)–P2A protein–DNA complexes selected on N.meningitidis outer membrane vesicles (T) or superblock (Pierce, NC). The effect of RNase (Promega) treating samples immediately after transcription–translation (at 30°C, 10 U) or adding 0.1% BSA during transcription and translation. The DNA bands from no additions/post-treatments are also shown (0).

Figure 4

( a ) Filter screen of expressed scFvs derived from rescued DNA cloned from each round of CAD selection. Spiking 1:25 000 of anti-phOx–P2A DNA into (scFv)n–P2A library with diversity of n = 50 million unique scFvs, the fraction of positives from a random selection of 3040 clones is shown. ( b ) Enrichment obtained from each round of selection (total of 800-fold). R0–R1: 50-fold; R1–R2: 8-fold; and R2–R3: 2-fold.

Figure 4

( a ) Filter screen of expressed scFvs derived from rescued DNA cloned from each round of CAD selection. Spiking 1:25 000 of anti-phOx–P2A DNA into (scFv)n–P2A library with diversity of n = 50 million unique scFvs, the fraction of positives from a random selection of 3040 clones is shown. ( b ) Enrichment obtained from each round of selection (total of 800-fold). R0–R1: 50-fold; R1–R2: 8-fold; and R2–R3: 2-fold.

Figure 5

Panning an scFv library. (scFv) n –P2A ( n = 50 million unique clones) against tetanus toxoid. ELISA of R1 of CAD. A total of 92 clones were randomly picked and assayed (ELISA). Only clones with A450 > 0.15 are shown (anti-TTx scFv, closed bar; BSA, open bar).

Figure 5

Panning an scFv library. (scFv) n –P2A ( n = 50 million unique clones) against tetanus toxoid. ELISA of R1 of CAD. A total of 92 clones were randomly picked and assayed (ELISA). Only clones with A450 > 0.15 are shown (anti-TTx scFv, closed bar; BSA, open bar).

Table 1

Comparison of human anti-TTx CDR3 regions

TTx-sequences ( IGHV-D-J, IGKV-J and IGLV-J genes) were classified and aligned using IMGT/V-QUEST ( http://imgt.cines.fr ), and the positions 105–117 for CDR3 sequences are shown ( 31 , 32 ). The gene names are the human germline equivalent. IMGT-scores > 1179 reflects at least 95% nucleotide sequence identity ( 33 ). The 6-ATT clone is from Macaca fascicularis Fab ( 33 ). R = Round of CAD. na, not available. The basic residues are in blue, the acidic residues are in red, and serines and threonines are in green.

Table 2

Comparison of different in vitro display technologies

Property  Phage ( 3 ) a  Ribosomal ( 7 ) a  mRNA ( 7 , 8 ) a  IVC ( 11 , 12 ) a CAD (this study)  CIS ( 36 ) a 
Covalent link No No Yes (synthetic) No Yes (inherent) No 
Nucleic acid selected DNA mRNA mRNA/cDNA DNA DNA DNA 
Antibody display scFv scFv  scFv b No scFv scFv 
Typical e.f.  10 2 –10 4  10 1 –10 3  10 1 –10 3  10 1 –10 3  10 1 –10 2  10 3 
Library size  10 9 –10 11  10 12 –10 13  10 12 –10 13  10 9 –10 10 /ml   10 7  10 12 
Potential size  10 11  >10 13  >10 13  >10 10  >10 13  >10 13 
Making libraries Time consuming Fast Fast Fast Fast Fast 
Stability Stable  Stable in Mg 2+ Covalent Stable Covalent >2 days 
Property  Phage ( 3 ) a  Ribosomal ( 7 ) a  mRNA ( 7 , 8 ) a  IVC ( 11 , 12 ) a CAD (this study)  CIS ( 36 ) a 
Covalent link No No Yes (synthetic) No Yes (inherent) No 
Nucleic acid selected DNA mRNA mRNA/cDNA DNA DNA DNA 
Antibody display scFv scFv  scFv b No scFv scFv 
Typical e.f.  10 2 –10 4  10 1 –10 3  10 1 –10 3  10 1 –10 3  10 1 –10 2  10 3 
Library size  10 9 –10 11  10 12 –10 13  10 12 –10 13  10 9 –10 10 /ml   10 7  10 12 
Potential size  10 11  >10 13  >10 13  >10 10  >10 13  >10 13 
Making libraries Time consuming Fast Fast Fast Fast Fast 
Stability Stable  Stable in Mg 2+ Covalent Stable Covalent >2 days 

Abbreviations : e.f., enrichment factor per round of selection; and IVC, in vitro compartmentalization.

a Represent the reference numbers.

b Via IGHV-D-J , IGKV-J and IGL-V-J genes.

Correspondence may also be addressed to Martin Welschof. Email: m.welschof@affitech.com

We thank Dr David Andrews (McMaster University, Hamilton, Canada) for providing E.coli S30 lysates and Dr Terje Michaelsen (The Norwegian Institute of Public Health, Oslo, Norway) for TTx-antigen. We also thank Professor Björn Lindqvist (Department of Biology, University of Oslo, Norway) for inspiration and initial discussions. Funding to pay the Open Access publication charges for this article was provided by Affitech AS.

REFERENCES

1
Li, M.
2000
Applications of display technology in protein analysis
Nat. Biotechnol.
 
18
1251
–1256
2
Lee, S.Y., Choi, J.H., Xu, Z.
2003
Microbial cell-surface display
Trends Biotechnol.
 
21
45
–52
3
Hoogenboom, H.R., de Bruïne, A.P., Hufton, S.E., Hoet, R.M., Arends, J.-W., Roovers, R.C.
1998
Antibody phage display technology and its applications
Immunotechnology
 
4
1
–20
4
Keefe, A.D. and Szostak, J.W.
2001
Functional proteins from a random-sequence library
Nature
 
410
715
–718
5
Wilson, D.S., Keefe, A.D., Szostak, J.W.
2001
The use of mRNA display to select high-affinity protein-binding peptides
Proc. Natl Acad. Sci. USA
 
98
3750
–3755
6
Schaffitzel, C., Hanes, J., Jermutus, L., Plückthun, A.
1999
Ribosome display: an in vitro method for selection and evolution of antibodies from libraries
J. Immunol. Methods
 
231
119
–135
7
Lipovsek, D. and Plückthun, A.
2004
In-vitro protein evolution by ribosome display and mRNA display
J. Immunol. Methods
 
290
51
–67
8
Takahashi, T.T., Austin, R.J., Roberts, R.W.
2003
mRNA display: ligand discovery, interaction analysis and beyond
Trends Biochem. Sci.
 
28
159
–165
9
Hanes, J. and Plückthun, A.
1997
In vitro selection and evolution of functional proteins by using ribosome display
Proc. Natl Acad. Sci. USA
 
94
4937
–4942
10
Amstutz, P., Forrer, P., Zahnd, C., Plückthun, A.
2001
In vitro display technologies: novel developments and applications
Curr. Opin. Biotechnol.
 
12
400
–405
11
Tawfik, D.S. and Griffiths, A.D.
1998
Man-made cell-like compartments for molecular evolution
Nat. Biotechnol.
 
16
652
–656
12
Yonezawa, M., Doi, N., Kawahashi, Y., Higashinakagawa, T., Yanagawa, H.
2003
DNA display for in vitro selection of diverse peptide libraries
Nucleic Acids Res.
 
31
e118
13
Lindahl, G.
1970
Bacteriophage P2: replication of the chromosome requires a protein which acts only on the genome that coded for it
Virology
 
42
522
–533
14
FitzGerald, K.
2000
In vitro display technologies—new tools for drug discovery
Drug Discov. Today
 
5
253
–258
15
Liu, Y. and Haggård-Ljungquist, E.
1994
Studies of bacteriophage P2 DNA replication: localization of the cleavage site of the A protein
Nucleic Acids Res.
 
22
5204
–5210
16
Odegrip, R. and Haggård-Ljungquist, E.
2001
The two active-site tyrosine residues of the A protein play non-equivalent roles during initiation of rolling circle replication of bacteriophage P2
J. Mol. Biol.
 
308
147
–163
17
Voet, D. and Voet, J.G.
1995
Control of transcription in prokaryotes In Voet, D. and Voet, J.G. (Eds.).
Biochemistry
  2nd edn NY John Wiley & Sons Inc. pp.
930
–932 Section 29-3
18
Marks, J.D., Griffiths, A.D., Malmqvist, M., Clackson, T.P., Bye, J.M., Winter, G.
1992
By-passing immunization: building high affinity human antibodies by chain shuffling
Biotechnology (NY)
 
10
779
–783
19
Braunagel, M. and Little, M.
1997
Construction of a semisynthetic antibody library using trinucleotide oligos
Nucleic Acids Res.
 
25
4690
–4691
20
Cody, J.T.
1990
Cross-reactivity of amphetamine analogues with Roche Abuscreen radioimmunoassay reagents
J. Anal. Toxicol.
 
14
50
–53
21
Lesley, S.A., Brow, M.A., Burgess, R.R.
1991
Use of in vitro protein synthesis from polymerase chain reaction-generated templates to study interaction of Escherichia coli transcription factors with core RNA polymerase and for epitope mapping of monoclonal antibodies
J. Biol. Chem.
 
266
2632
–2638
22
Lesley, S.A.
1995
Preparation and use of E.coli S-30 extracts
Methods Mol. Biol.
 
37
265
–278
23
Fredriksen, J.H., Rosenqvist, E., Wedege, E., Bryn, K., Bjune, G., Frøholm, L.O., Lindbak, A.-K., Møgster, B., Namork, E., Rye, U., et al.
1991
Production, characterization and control of MenB-vaccine ‘Folkehelsa’: an outer membrane vesicle vaccine against group B meningococcal disease
NIPH Annal.
 
14
67
–80
24
Osbourn, J. and Holet, T.
2001
Improvements to ribosome display Patent WO0175097
25
Schatz, P.J., Cull, M.G., Miller, J.F., Stemmer, W.P.C., Gates, C.M.
1998
Peptide library and screening method US Patent 6,156,511
26
(Eds.).
Molecular Cloning
 
2001
3rd edn Cold Spring Harbor, NY Cold Spring Harbor Laboratory Press Vol.
3
pp.
A1.11
–A4
27
Kipriyanov, S.M., Moldenhauer, G., Little, M.
1997
High level production of soluble single chain antibodies in small-scale Escherichia coli cultures
J. Immunol. Methods
 
200
69
–77
28
de Wildt, R.M., Mundy, C.R., Gorick, B.D., Tomlinson, I.M.
2000
Antibody arrays for high-throughput screening of antibody-antigen interactions
Nat. Biotechnol.
 
18
989
29
Coia, G., Pontes-Braz, L., Nuttall, S.D., Hudson, P.J., Irving, R.A.
2001
Panning and selection of proteins using ribosome display
J. Immunol. Methods
 
254
191
–197
30
Huang, Q.R., Morris, D., Manolios, N.
1997
Identification and characterisation of polymorphisms in the promoter region of the human Apo-1/Fas (CD95) gene
Mol. Immunol.
 
34
577
–582
31
Giudicelli, V., Chaume, D., Lefranc, M.-P.
2004
IMGT/V-QUEST, an integrated software program for immunoglobulin and T cell receptor V-J and V-D-J rearrangement analysis
Nucleic Acids Res.
 
32
W435
–W440
32
Lefranc, M.P., Pommie, C., Ruiz, M., Giudicelli, V., Foulquier, E., Truong, L., Thouvenin-Contet, V., Lefranc, G.
2003
IMGT unique numbering for immunoglobulin and T cell receptor variable domains and Ig superfamily V-like domains
Dev. Comp. Immunol.
 
27
55
–77
33
Chassagne, S., Laffly, E., Drouet, E., Hérodin, F., Lefranc, M.-P, Thullier, P.
2004
A high-affinity macaque antibody Fab with human-like framework regions obtained from a small phage display immune library
Mol. Immunol.
 
41
539
–546
34
Chin, J., Sohn, Y., Lee, S.H., Park, Y.I., Choi, M.J.
2003
Production of neutralizing human monoclonal antibody directed to tetanus toxin in CHO cell
Biologicals
 
31
45
–53
35
Faber, C., Shan, L., Fan, Z.-C., Guddat, L.W., Furebring, C., Ohlin, M., Borrebaeck, C.A.K., Edmundson, A.B.
1998
Three-dimensional structure of a human Fab with high affinity for tetanus toxoid
Immunotechnology
 
3
253
–270
36
Odegrip, R., Coomber, D., Eldridge, B., Hederer, R., Kuhlman, P.A., Ullman, C., FitzGerald, K., McGregor, D.
2004
CIS display: in vitro selection of peptides from libraries of protein–DNA complexes
Proc. Natl Acad. Sci. USA
 
101
2806
–2810

Comments

0 Comments