New display vector reduces biological bias for expression of antibodies in E. coli

Gene 391 (2007) 120 – 129 www.elsevier.com/locate/gene

New display vector reduces biological bias for expression of antibodies in E. coli Emiliano Pavoni a , Giorgia Monteriù a , Maurizio Cianfriglia b , Olga Minenkova a,⁎ a

Kenton Labs, c/o Sigma-Tau, via Pontina km 30.400, Pomezia (RM), 00040 Italy b Pharmacogenetics, Drug Resistance and Experimental Therapeutics Section, Department of Drug Research and Evaluation of Italian National Institute of Health (Istituto Superiore di Sanità), Rome, Italy Received 16 June 2006; received in revised form 13 December 2006; accepted 14 December 2006 Available online 22 December 2006 Received by J.A. Engler

Abstract We report the development of a novel phagemid vector, pKM19, for display of recombinant antibodies in single-chain format (scFv) on the surface of filamentous phage. This new vector improves efficacy of selection and reduces the biological bias against antibodies that can be harmful to host bacteria. It is useful for generation of large new antibody libraries, and for the subsequent maturation of antibody fragments. In comparison with commonly used plasmids, this vector is designed to have relatively low expression levels of cloned scFv antibodies due to the amber codon positioned in a sequence encoding for the PhoA leader peptide. Moreover, fusion of antibodies to the carboxy terminal part only of the gene III protein improves display of scFv on bacteriophage surface in this system. Despite the lower antibody expression, the functional test performed with a new scFv library derived from human peripheral blood lymphocytes demonstrates that specific antibodies can be easily isolated from the library, even after the second selection round. The use of the pKM19 vector for maturation of an anti-CEA antibody significantly improves the final results. In our previous work, an analogous selection through the use of a phagemid vector, with antibody expression under the control of a lacP promoter, led to isolation of antiCEA phage antibodies with improved affinities, which were not producible in soluble form. Probably due to the toxicity for E. coli of that particular anti-CEA antibody, 70% of maturated clones contained suppressed stop codons, acquired during various selection/amplification rounds. The pKM19 plasmid facilitates an efficient maturation process, resulting in selection of antibodies with improved affinity without any stop codons. © 2006 Elsevier B.V. All rights reserved. Keywords: Phagemid vector; Phage display; scFv library; Amber codon

1. Introduction Bacterial expression systems are a popular means of expressing recombinant proteins. However, good expression of eukaryotic proteins in a heterologous bacterial system, in vivo, is Abbreviations: aa, amino acid(s); AP, alkaline phosphatase; Ap, ampicillin; CDR, complementarity determining region; CEA, carcinoembryonic antigen; GST, glutathione S-transferase; HC, heavy chain; HRP, horseradish peroxidase; IPTG, isopropyl b-D-thiogalactopyranoside; Km, kanamycin; LC, light chain; PAGE, polyacrilamide gel electrophoresis; PBS, phosphate-buffered saline; O.D., optical density; PEG, polyethylene glycol; PFU, plaque-forming unit(s); RBS, ribosome binding site; scFv, single-chain Fv antibody; SDS, sodium dodecyl sulfate; TU, transducing unit(s); wt, wild type. ⁎ Corresponding author. Tel.: +39 06 91628048; fax: +39 06 91629012. E-mail address: [email protected] (O. Minenkova). 0378-1119/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.gene.2006.12.009

the fortunate exception to an unfortunate rule. In many cases, difficulties are related to intracellular protein instability, insolubility, or cellular toxicity. Expression of recombinant antibodies displayed on the filamentous phage in E. coli is exactly that fortunate exception. In fact, filamentous phage morphogenesis, particularly suitable for correct disulfide bond formation and expression of functional antibody fragments, is based on the assembly of phage coat proteins on the inner bacterial membrane, and on the secretion of phage particles into the oxidizing periplasm. However, any library of recombinant antibodies contains millions of different antibody molecules, characterized by their particular capacity to recognize and bind various biological targets. As a result, all antibodies interacting with bacterial proteins may affect bacterial growth or be toxic for host bacteria.

E. Pavoni et al. / Gene 391 (2007) 120–129

The phagemid vectors, commonly utilized for display of scFv antibodies, allow the cloning of foreign DNA encoding for antibody fragments as fusion at the amino terminus of the gene III protein (pIII) of the filamentous phage (Hoogenboom et al., 1991). Superinfection with helper phage results in packaging of the phagemid into a phage particle displaying a cloned antibody fragment. The utilized vectors contain an amber codon inserted at the start of the gene III sequence, thus permitting antibody expression in soluble form in non-suppressor E. coli strains, without recloning the antibody gene. The various tag sequences at the C-terminus of the antibody, such as myc (Ward et al., 1989) or FLAG peptide (Pini et al., 1998) allow the single-chain fragments to be detected in ELISA with commercially available secondary monoclonal antibodies. Recently, we developed a high-affinity single-chain antiCEA antibody, E8, by maturation of an original anti-CEA MA39 antibody (Pavoni et al., 2006) selected from the ETH-2 scFv synthetic library (Viti et al., 2000). A maturation library was generated by cloning the mutated anti-CEA antibody gene in the pDN332 vector (Pini et al., 1998). Several phage clones with increased reactivity were selected. However, we failed to express the soluble maturated antibody in non-suppressor bacteria. Subsequent sequence analysis of scFv genes revealed the presence of TAG and TGA codons in 70% of phage clones, preventing efficient protein production. Although DH5αF′ is not a UGAsuppressor strain, in wild-type E. coli, a UGA triplet can be decoded at a very low frequency by inserting a tryptophan residue (Parker, 1989). Suppression of (i) the TAG codon by supE mutation in the DH5αF′ strain, where the amber codon is translated as glutamine, and (ii) the TGA codon by natural suppression (suppression of nonsense mutation with native, not mutant tRNA) (Parker, 1989), allowed us to select highly reactive functional phage-displayed antibodies, impossible to produce in soluble form without additional manipulation aimed at substituting the stop codons in antibody gene sequence. The

121

stop codons were localized in the antibody regions outside of the zones, which were mutated in the maturation library. Thus, these codons were the result of accidental mutations giving growth advantages to mutated clones which still displayed functional antibodies on the phage capsid. The accumulation of suppressed stop codons during the selection process that reduced antibody expression levels, indicates a strong biological bias against production of the anti-CEA antibody in bacteria, which may interfere with vital bacterial functions and be toxic for host cells. An elevated frequency of suppressed amber codons in antibody genes complicating the identification of positive antibodies during utilization of the synthetic scFv libraries for biopanning experiments was recently described by Marcus et al. (2006). Expression of toxic proteins is a well-known problem which can be addressed by various means, e.g. by using a regulated promoter or by reduction of the plasmid copy number (Khlebnikov et al., 2002; Bowers et al., 2004). Regarding the phage display of toxic proteins, a library of variants of the potato protease inhibitor, P12, resulted in selection of only mutants carrying deletions or amber stop codons when a canonical phagemid vector was applied (Beekwilder et al., 1999). Instead, a new phagemid vector having a lower plasmid copy number and another vector equipped with a psp promoter, inducible upon helper phage infection (both vectors were developed in this study), allowed selection of trypsin inhibitors. Despite this knowledge, a display of antibody libraries is never considered as a display of proteins that may affect the viability of E. coli. In the present work we describe a novel phagemid vector, pKM19, which allows a low-level expression of recombinant antibodies. As a result, an antibody library generated by applying this vector does not accumulate stop codons during selection. This vector improves viability of bacteria harboring harmful antibodies and thus counterbalances the probability of the antibodies, which are either neutral or harmful to bacterial host, to be selected from the library. Moreover, pKM19 provides

Table 1 Oligonucleotide primers used for construction of plasmids and mutant library for affinity maturation of anti-CEA antibody KM161 KM162 KM163 KM164 KM175 KM176 KM181 KM182 KM183 KM184 KM185 KM186 KM180 KM143 KM144 KM148 KM145 KM157

5′-GAGGAAGCTTCCATTAAACGGGTAAAATAC-3′ 5′-TGCAATGGCGGCCGCTAATATTGTTCTGGATATTACCAGC-3′ 5′-AGCTTCCTCATGTAGGCGGCCGCAGGAGACTACAAAGACGACGACGAC AAACACCACCATCACCACCATTAA-3′ 5′-GGCCTTAATGGTGGTGATGGTGGTGTTTGTCGTCGTCGTCTTTGTAGTC TCCTGCGGCCGCCTACATGAGGA-3′ 5′-AGCTTATAAAGGAGGAAATCCTCATGAAACAGAGCACCATCGCACTGGCACTGTTACCGTTACTGTTCACCCCGGTTACCAAA GCACGTACCATGGTTTCC CTTGC-3′ 5′-GGCCGCAAGGGAAACCATGGTACGTGCTTTGGTAACCGGGGTGAACAGTAACGGTAACAGTGCCAGTGCGATGG TGCTCTGTTTCATGAGGATTTCCTCCT TTATA-3′ 5′-GTGGTGATGGAATTCTTTGTCGTCGTCGTCTTTGTAGTC-3′′ 5′-CACCATTAAGGATCCTAATATTGTTCTGGATATTACCAGC-3′ 5′-TCTATTCTGAATTCGCTGAAACTGTTGAAAGTTGTTTAGC-3′ 5′-GCCAATCGGAATTCCTGCCTCAACCTCCTGTCAATGCT-3′ 5′-GAACTGGGATCCTTAAGACTCCTTATTACGCAGTATG-3′ 5′-ACCCGTAAGCTTATAAAGGAGGAAATCCTCATGAAATAGAGCACCATCGC-3′ 5′-TAGCCCCCTTATTAGCGTTTG-3′ 5′-GTCATCGTCGGAATCGTCATCTGC-3′ 5′-TGTGCGAAAAGTAATGAGTTTCTTTTTGACTACTGGGGC-3′ 5′-CTATTGCCTACGGCAGCCGCTGGA-3′ 5′-TCCGCCGAATACCACATAGGGCAACCACGGATAAGAGGAGTTACAGTAATAGTCAGCC-3′ 5′-TTTCGCACAGTAATATACGG-3′

Underlined sequences indicate restriction sites. In a part of oligonucleotides KM144 and KM145 (in italics), each base was replaced with mixture of G/A/T/C at a frequency of 10%.

122

E. Pavoni et al. / Gene 391 (2007) 120–129

more efficient display of recombinant antibodies as compared with canonical vectors, also improving efficacy of selection. 2. Materials and methods 2.1. Bacterial strains Bacterial strain DH5αF′ (supE44 ΔlacU169 (ϕ80 lacZΔM15) hsdR17 recA1 endA1gyrA96 thi-1 relA1 F′ [traD36 proAB+ lacIqlacZΔM15]) was used for soluble and phage antibody production. 2.2. Construction of plasmids The pC89 plasmid (Felici et al., 1991) was amplified by inverse PCR with the KM161 and KM162 oligonucleotides, containing sites for HindIII and NotI restriction enzymes (underlined), Table 1. In inverse PCR, a mixture of Taq polymerase with Pfu DNA polymerase was used to increase fidelity of DNA synthesis. Twenty-five cycles of amplification (95 °C– 30 s, 55 °C–30 s, 72 °C–20 min) were done. The PCR product was digested with HindIII and NotI restriction enzymes and ligated with a KM163–KM164 oligonucleotide duplex encoding for FLAG peptide and 6His-tail. The cloned DNA duplex contained an internal NotI site, upstream of FLAG peptideencoding sequence, while the NotI site, used for cloning of the duplex, was not restored. The resulting pKM15 plasmid was newly digested with HindIII, NotI enzymes and ligated with KM175–KM176 duplex encoding for the leader sequence and for the first two amino acids of the bacterial alkaline phosphatase (PhoA) containing the NcoI cloning site. The pKM16 plasmid was used for soluble single-chain antibody production (Fig. 1). The pKM16 plasmid was amplified by inverse PCR with the KM181 and KM182 oligonucleotides, containing EcoRI and BamHI restriction sites, respectively. The full-length gene III and the 3′ part of the gene, encoding for the last 197 aa of the pIII, were amplified by using the oligonucleotides KM183– KM185 or KM184–KM185 containing BamHI and EcoRI sites (underlined) and ligated into digested pKM16, giving the new plasmids, pKM17 and pKM18, respectively. A short fragment of the pKM18 plasmid containing the PhoA leader peptide gene was PCR-amplified with KM186 and KM180 primers, introducing an amber codon. Then, it was digested with HindIII and NotI and cloned into digested and agarose gel-purified pKM18, to construct the pKM19 plasmid. 2.3. Soluble antibody production A single colony was inoculated into 50 mL of LB containing 100 μg/mL Ap and 2% glucose. The culture was grown at 37 °C for 4–5 h until O.D.600 = 0.8. The cells were recovered by centrifugation, resuspended in 50 mL of fresh LB with 100 μg/ mL Ap, 1 mM IPTG, 20 mM MgCl2, and incubated overnight at 30–32 °C. Bacterial cells were pelleted and then resuspended in 0.5 mL of PBS. After three cycles of freeze and thaw, cell debris was pelleted and the resulting supernatant was used for ELISA or Western blot.

Fig. 1. Plasmid maps. pKM16–pKM19 plasmids are pUC-based vectors constructed either for soluble scFv antibody production (pKM16) or for display of single-chain antibody fragments on the surface of filamentous phage (pKM17, pKM18 and pKM19). pKM16 was obtained by introducing the leader peptide-encoding sequence, cloning sites (NcoI, NotI) and the FLAG/His-tail sequence in plasmid pc89 (Felici et al., 1991). The plasmids pKM17 and pKM18/19 were obtained by inserting, respectively, the full length or the 3′ part of the gene III after FLAG-encoding sequence.

2.4. Purification of lymphocytes from peripheral blood and cDNA synthesis The lymphocytes were isolated from 10 mL of fresh peripheral blood, obtained from breast cancer patient, EC23, through informed consent, with an anticoagulant using FicollPaque Plus (Amersham Pharmacia Biotech, Sweden) according to the manufacturer's instructions. The mRNA was isolated from lymphocytes by using Dynabeads mRNA DIRECT Kit (Dynal, Oslo, Norway). One μg of the poly(A)+ RNA from the lymphocytes was used to synthesize full-length cDNA by using SMART cDNA Library Construction Kit (Clontech, Palo Alto, CA). 2.5. Construction of scFv library from peripheral blood lymphocytes The antibody gene repertoire was amplified using a set of primers designed for amplification of VH and VL antibody domains, while scFv fragments were assembled in vitro as

E. Pavoni et al. / Gene 391 (2007) 120–129

123

described in Pope et al. (1996). The latter were then amplified by PCR with appropriate extension primers, incorporating NcoI and NotI restriction sites, facilitating the cloning of the scFv genes into the pKM19 vector. The resulting PCR products were purified from 1% low-melting agarose gel (NuSieve 3:1 agarose, Rockland, ME), cut with NcoI/NotI and inserted into digested plasmid. The transformed scFvEC23 library contained 1.77 × 107 independent clones with full-length scFv inserts.

collected by centrifuging and resuspended in 40 mL of LB with ampicillin without glucose. About 6 × 109 PFU of helper M13K07 were added to each mL of cell suspension, incubated for 15 min at 37 °C without agitation and for another 2 h in a shaker. Kanamycin was added to obtain a final concentration of 20 μg/mL, and cells were incubated overnight at 32 °C. Phage was purified according to standard PEG/NaCl precipitation (Sambrook et al., 1989).

2.6. Construction of mutated anti-CEA scFv library

2.8. Affinity selection

The maturation library for the anti-CEA scFv was constructed as described in the work of Pavoni et al. (2006). Briefly, mutated scFv gene fragments were generated by PCR amplification with KM144–KM143 and KM148–KM145 primers, introducing random mutations in CDR3 regions of heavy or light chains with low frequency (Table 1). Missing scFv antibody gene parts were amplified with KM148–KM157 and KM158–KM143 primers for HC and LC, respectively (Fig. 2). In order to reconstruct the entire gene, the corresponding fragments were combined and amplified in a PCR-like process without oligonucleotide primers. The resulting product was utilized to amplify the entire gene with external primers. The final DNA fragment was agarose-purified, digested with restriction enzymes NcoI and NotI, and ligated with the digested pKM19 plasmid. The resulting maturation library contained 2.2 × 106 antibody clones.

CEA and recombinant SP2-GST were biotinylated as described earlier (Harlow and Lane, 1988). About 5 × 1011 TU of freshly amplified scFv antibody libraries were preincubated with 50 μL of AD202 bacterial extract in blocking buffer for 30 min at 37 °C. Twenty μg of a biotinylated protein (CEA or SP2-GST) was added to the reaction mixture and incubated for another hour at 37 °C under gentle agitation. The bound phage was captured by using streptavidin-coated Dynabeads M-280 (112.05, Dynal) according to the manufacturer's instructions, washed five to ten times with PBS/Tween (0.05% Tween-20 in PBS), then eluted with 0.1 M HCl (adjusted to pH 2.2 with glycine) and amplified as above. In the second round of affinity selection from the maturation library of anti-CEA scFv variants, only 100 ng of the biotinylated CEA was used. 2.9. Phage ELISA

2.7. Phage amplification Forty μL of scraped bacterial cells or frozen cell suspensions under glycerol were incubated in 40 mL of LB containing ampicillin and 2% glucose until O.D.600 = 0.2. The bacteria were

Multiwell plates (Immunoplate Maxisorb, Nunc, Roskilde, Denmark) were coated overnight at 4 °C with a protein solution at a concentration of 10 mg/mL in 50 mM NaHCO3, pH 9.6. After discarding coating solution, plates were blocked for 1 h at

Fig. 2. Scheme of mutagenesis in vitro. The maturation library was generated by using the partially degenerated oligonucleotides, KM144 and KM145.

124

E. Pavoni et al. / Gene 391 (2007) 120–129

In the case of competition with filtrates of phage supernatants, 10 μL or 50 μL of the MA39 or pKM19/anti-CEA filtrates in 100 μL of blocking buffer were used as competitors. 3. Results 3.1. Plasmid pKM16

Fig. 3. Soluble scFv production by using pKM16 plasmid. Three independent clones obtained by cloning scFv MA39 anti-CEA gene in pKM16 were tested for soluble scFv production (gel lines 1–3). Periplasmic protein fractions were purified from bacteria by freeze–thaw method. The protein size marker is included (left). Western blot membrane was developed with an anti-FLAG alkaline phosphatase-conjugated secondary antibody. Bands corresponding to soluble scFv antibodies (expected molecular weight 26 kDa) migrate between 24.5 and 35.9 kDa bands.

Fig. 1 shows the genetic map of the pKM16 plasmid, used for production of soluble antibodies in scFv configuration, constructed as described in Materials and methods. This plasmid directs protein expression under the control of the lacP promoter. The unique NcoI and NotI cloning sites facilitate insertion of antibody genes, allowing for expression of singlechain antibodies as fusion to the leader peptide and to the first two amino acids of alkaline phosphatase (a bacterial periplasmic enzyme), at the antibody's amino terminus, and as fusion to FLAG/6His-tail at the antibody's carboxy terminus. In order to confirm the plasmid's practical qualities, we amplified a gene of a single-chain antibody of known specificity, anti-CEA MA39 (Pavoni et al., 2006), by PCR and cloned it into the pKM16 vector. We then analyzed freeze–thaw-purified periplasmic proteins in Western blot with an anti-FLAG secondary antibody

37 °C with ELISA blocking buffer (5% non-fat dry milk, 0.05% Tween-20 in PBS). Plates were washed several times with washing buffer (0.05% Tween-20 in PBS). PEG-purified phage in blocking buffer (1:1) was added to each well and incubated for 1 h at 37 °C. The plates were washed and the bound phage was detected by an anti-M13 HRP-conjugated (27-9421-01, Amersham Biosciences, Uppsala, Sweden), or anti-FLAG HRP-conjugated (A9044, Sigma, St. Louis, MO), or antiFLAG AP-conjugated (A9469, Sigma) secondary antibody. In the case of HRP conjugates, the immunoreaction was developed by incubation with TMB liquid substrate (Sigma) for 15 min and stopped by the addition of 25 μL 2 M H2SO4. The results were expressed as the difference between absorbances at 450 and 620 nm, determined by an automated ELISA reader. The AP-conjugated antibody was detected by incubation with 1 mg/ mL solution of p-nitrophenyl phosphate in substrate buffer (10% diethanolamine buffer, 0.5 mM MgCl2, pH 9.8) for 60 min. The results were expressed as the difference between absorbances at 405 and 620 nm. 2.10. Competition with soluble scFv ELISA plates were coated, blocked and washed as above. Various quantities of anti-CEA soluble antibody MA39 in 100 μL of blocking buffer were added to the wells and incubated for 30 min at 37 °C. Then, 10 μL (4.5 × 109 TU) of MA39 phage supernatant or 5 μL (3 × 108 TU) of anti-CEA/ pKM19 supernatant was added to the wells and incubated for another 1 h at 37 °C. The plates were washed and the bound phage detected by an anti-M13 HRP-conjugated antibody. An irrelevant soluble anti-SP2 scFv at a high concentration (400 ng/ well) was used as negative control. A lower quantity of the antiCEA/pKM19 phage, as compared to MA39, was used to moderate ELISA reactivity of this phage.

Fig. 4. Display efficiency of pKM17, pKM18 and pKM19 plasmids in comparison with a classic phagemid system. Anti-CEA scFv antibodies displayed by using three different plasmids, were assayed by ELISA against CEA protein and compared with MA39 phage (anti-CEA/pDN332). The helper phage, M13K07, that does not display antibody fragments, was included as negative control. Data reported are the average values of assays performed in duplicate. The highest phage concentration, labeled by asterisk, corresponds to the 1011 TU for all phages and 3 × 1010 TU for anti-CEA/pKM17.

E. Pavoni et al. / Gene 391 (2007) 120–129

125

Fig. 6. Competition with soluble scFv. Freshly prepared supernatants of MA39 (10 μL) and anti-CEA/pKM19 (5 μL) phages competed with various amounts of the purified soluble anti-CEA antibody. The data are expressed as percentage of reactivity of the supernatants without competitors. The irrelevant soluble antiSP2 scFv was used as negative control.

Fig. 5. Filtration of phage samples. About 2 × 1011 TU/well of each preparation or the corresponding quantity of filtrate samples were tested in ELISA and developed either with anti-M13 or anti-FLAG secondary antibodies. Data reported are the average values of assays performed in duplicate. The data show reactivity of filtrates as percentage of original reactivity of non-filtrated samples (100%).

(Fig. 3). Single-chain antibody bands migrated as proteins with the expected molecular weight. N-terminal protein sequencing by Edman degradation confirmed the correct processing of the leader peptide.

tained a range between 5 and 1 × 1011 TU/mL for MA39, pKM18 and pKM19, displaying the anti-CEA antibody, while anti-CEA/ pKM17 generated five to ten times lower titers. Phage preparations were tested in ELISA, where developing was performed by using the anti-M13, or alternatively, the antiFLAG secondary antibody. Applying different amounts of the phage per ELISA well, we demonstrated higher display efficiency for pKM18 and pKM19 phages in comparison with pKM17 and much higher as compared with MA39 (Fig. 4). It is interesting that the MA39 clone, which produces a higher level of antibodies than anti-CEA/pKM17, as shown by developing with anti-FLAG antibody, has a weaker signal when ELISA is developed with the anti-M13 secondary antibody. This indicates that free scFvs, produced by the classic phagemid system due to the presence of an amber codon between scFv and pIII genes, leak into the medium and coprecipitate with phage particles,

3.2. Phagemids for display of scFv antibodies A classic phagemid (pDN332) displaying the anti-CEA single-chain antibody, MA39, was compared with pKM17, pKM18 and pKM19 vectors displaying the same antibody, for phage particle production and display efficiency. The pKM17 and pKM18 plasmids (Fig. 1) allow the display of antibody fragments on a phage particle by fusion to, respectively, the entire pIII (1–406 aa) or the carboxy terminal domain only (210–406 aa) of the protein. The pKM19 plasmid, a derivative of pKM18, harbors an amber codon in leader sequence, thus providing lower production of scFv–ΔpIII fusion proteins as compared to pKM18 (in supE bacteria an expected suppression efficiency of this TAG codon, which depends on nucleotide context, is about 10–15% (Miller and Albertini, 1983; Edelmann et al., 1987)). We performed functional tests by cloning the antiCEA single-chain antibody gene into the three novel plasmids and confronting them with the original MA39 clone (anti-CEA in pDN332). Three single colonies for each clone were incubated in 10 mL of media and phage was amplified as described in Section 2.7. After phagemid rescue the supernatants were titered. We ob-

Fig. 7. Competition with phage supernatant filtrates. Freshly prepared supernatants of MA39 (10 μL) and anti-CEA/pKM19 (5 μL) phages were competed with 10 μL or 50 μL of filtrates of the same phage supernatants. The data are expressed as percentage of reactivity of the supernatants without competitors.

126

E. Pavoni et al. / Gene 391 (2007) 120–129

concentration of free antibodies in this phage supernatant is already high. Western blot analysis (Fig. 8) of various PEG-purified phages developed with an anti-FLAG antibody detects: (i) the upper band in each sample corresponding to scFv–pIII fusion in the case of MA39 and anti-CEA/pKM17 phages, and scFv–ΔpIII in the case of anti-CEA/pKM18 or anti-CEA/pKM19; (ii) notable presence of free antibodies in MA39 sample; (iii) presence of degradation products in the phage samples as previously described (Kretzschmar and Geiser, 1995). 3.3. Generation of scFv antibody-displayed library and isolation of binding specificities using novel pKM19 plasmid Fig. 8. Western blot of PEG-purified recombinant phages. Protein extracts from about 5 × 109 PFU of phages MA39, anti-CEA/pKM18 and anti-CEA/pKM19, and 1 × 109 PFU of anti-CEA/pKM17 were fractionated by SDS-PAGE and transferred onto a nitrocellulose membrane. The membrane strips were developed with an anti-FLAG AP-conjugated antibody. The protein size marker is included (last strip). The scFv–pIII (66.1 kDa) and scFv–ΔpIII (45.2 kDa) proteins migrate as higher molecular weight bands because of an anomalous moiety of the pIII protein describer earlier (Goldsmith and Konigsberg, 1977).

consequently competing with phage-displayed antibodies for target binding. In order to verify this hypothesis, we filtered fresh preparations of MA39 and anti-CEA/pKM19 phage by using Microcon 100 Centrifugal Filter Devices (Millipore Corporation, Bedford, MA), able to retain large phage particles and pass through soluble scFvs. The ELISA test of phage preparations, before and after filtration, developed with anti-M13 or antiFLAG antibodies, shows that: (i) both filtrates practically lose antibodies displayed on the phage particles, as expected; (ii) the free antibodies are present in both preparations (Fig. 5). However, the level of free antibodies in the anti-CEA/pKM19 sample is markedly lower. The free antibodies in this sample are the result of antibody shedding, inevitable during phage preparation and which might increase as a result of contact with components of the filtration system; while the free antibodies in MA39 samples are the result of free antibody expression and leakage into medium together with shedding. To test the competitive ability of free antibodies in phage supernatants we had the phage supernatants of the MA39 and anti-CEA/pKM19 phages compete either with the soluble antiCEA antibody of known concentration (Fig. 6) or with different quantities of supernatant filtrates of both phages (Fig. 7). These two experiments show that the free scFvs efficiently compete with the phage antibodies. Ten μL of the MA39 filtrate already competes with 10 μL of its own phage supernatant and 5 μL of anti-CEA/pKM19 supernanant, while the same quantity of antiCEA/pKM19 filtrate has no effect. Marked competition is observed only by a ten-fold excess of anti-CEA/pKM19 filtrate with the same phage supernatant (50 μL to 5 μL) (Fig. 7). The increasing competition effect with the growing quantity of filtrates in the case of MA39 is not as visible as in the case of anti-CEA/pKM19 supernatant, since the absolute values of ELISA signals of the MA39 phage are lower and the starting

We used the pKM19 plasmid, which provides lower levels of antibody production, for generation of an scFv library to study whether reduced expression of fused antibodies allows efficient selection of specific antibodies against a target molecule. An scFv antibody library was constructed from human peripheral blood lymphocytes, as described in Sections 2.4 and 2.5. The library was selected against GST fusion of a 168 aa-long Streptoccocus pneumoniae SP2 polypeptide (Beghetto et al., 2006), which was reactive with the blood sample utilized for the scFv library construction. We designed a selection procedure to create a high concentration of the target protein in a small incubation volume, by using a biotinylated protein for panning and streptavidin-coated Dynabeads to capture the bound phage. After completion of two panning rounds, we tested the reactivity of the phage pools in ELISA (Fig. 9). The phage pool, after the second round of affinity selection, was highly reactive with the fusion protein and negative with irrelevant proteins, such as GST, milk and streptavidin, which presented either as protein carrier or components of the selection system and all were used as negative controls in ELISA. Finally, we isolated and sequenced a number of positive clones to confirm the correct scFv sequence (data not shown). One of the identified scFv genes was cloned in pKM16 for production of soluble anti-SP2

Fig. 9. Selection against SP2-GST protein. Reactivity of the phage pools derived from first and second rounds of panning of scFvEC23 library is shown. GST, milk and streptavidin, presented in the selection system, are included as negative controls. Data reported are the average values of assays performed in duplicate. Phage input was normalized. About 3 × 109 TU per single well of each preparation were tested in ELISA.

E. Pavoni et al. / Gene 391 (2007) 120–129

Fig. 10. Affinity selection of maturated anti-CEA gene from mutated library. In this assay, positive immunoreactions were developed by an anti-FLAG APconjugated secondary antibody, in order to moderate positive signals and make visible the growing reactivity during the selection process. The helper phage, M13K07, that does not display antibody fragments, was included as negative control. The reactivity of the original anti-CEA antibody in pKM19 and single clones from the phage pool after second round of affinity selection is shown. Data reported are the average values of assays performed in duplicate. Phage input was normalized. About 3 × 1010 TU per single well of each preparation were tested in ELISA.

antibody, which was used as an irrelevant antibody control in experiments described in Figs. 6 and 11. 3.4. Maturation of anti-CEA scFv antibody by using pKM19 vector

127

Fig. 11. Specificity of maturated clones. About 250 ng per well of original and maturated antibodies in soluble form were assayed with CEA and various irrelevant proteins. The irrelevant anti-SP2 antibody was included as negative control. Data reported are the average values of assays performed in duplicate.

part of pIII, which is absent in the recombinant protein. Thus, in case of increasing amounts of recombinant protein in phage particle, the ELISA signal would likely decrease. Nevertheless, two antibody genes were cloned in pKM16, and soluble antibodies were produced and compared with the original soluble anti-CEA MA39 (Pavoni et al., 2006). Table 2 shows KD of the novel antibodies, indicating their improved affinity. The specificity test on newly selected scFvs shows their low background reactivity with irrelevant proteins, comparable with that of the original antibody (Fig. 11). 4. Discussion

Affinity selection from a maturation library was carried out as described in the work of Pavoni et al. (2006). Fig. 10 shows that phage reactivity against the CEA protein grows in each successive selection round. Single phage clones with improved reactivity were isolated (Fig. 10). We sequenced 19 random clones from the phage pool after the second round of selection. None of the sequenced clones (0 of 19) from the phage pool with increased affinity had any stop codons within antibody genes, whereas 70% (9 of 13) of clones contained such mutations when we used the classic phagemid system (P = 0.00002). Thus, the use of the pKM19 vector for maturation of an anti-CEA antibody significantly improves selection results. The phage antibodies isolated from the maturation library (clones 1 and 2) probably have higher affinity toward the CEA protein, as compared to the original anti-CEA antibody, than a higher number of recombinant antibody molecules per phage particle. This is because a secondary anti-M13 antibody, utilized for ELISA signal developing, recognizes the amino terminal Table 2 Kinetic value of parental and affinity-matured scFv antibodies scFv

kon (+/− SE)

koff (+/− SE)

KD

MA39 (Pavoni et al., 2006)

2.08E+04 (3.49E+02) 4.06E+05 (1.35E+04) 8.01E+05 (3.04E+04)

3.57E− 03 (8.64E−05) 3.61E− 03 (7.45E−05) 2.42E− 03 (5.63E−05)

1.71E− 07

Clone 1 Clone 2

8.90E− 09 3.02E− 09

This work describes the construction of a novel pKM19 phagemid vector for display of single-chain antibodies on filamentous phage. This vector is characterized by several differences, as compared with the canonical system, for display of antibodies on the filamentous phage capsid. First, the classic phagemids contain an amber codon between the scFv and gpIII genes, thus directing production of free scFvs and scFv–pIII fusion in suppressor bacteria, such as TG1, or DH5αF′, or XL1-Blue, generally used for phage amplification. All these bacterial strains, carrying the supE mutation, are glutamine-inserting suppressors with suppression efficiency dependent on the codon following the TAG (Miller and Albertini, 1983; Edelmann et al., 1987). It is obvious that the unassembled scFv–pIII-fused proteins remain anchored in the bacterial membrane, while a possible excess of free soluble scFv antibodies leaks from the periplasm into the medium. Under the standard phage purification protocol by PEG/NaCl, the free scFv antibodies are coprecipitated with phage particles. As a result, the concentration of free antibodies in phage suspension may be five to ten times higher than the concentration of scFv–pIIIfused proteins assembled in the phage particle. In a subsequent selection, the abundant free antibodies compete with phagedisplayed ones for target binding, thus interfering with panning efficiency and delaying the selection process, especially in later panning rounds, where concentration of specific phage is relatively high or in maturation libraries, containing many relative antibodies with the same specificity.

128

E. Pavoni et al. / Gene 391 (2007) 120–129

Second, the presence of an amber codon positioned in a sequence encoding for a phosphatase alkaline leader peptide in pKM19, leads to a relatively low expression level of recombinant antibodies in the amber-suppressor bacteria harboring this plasmid. It is known that induction of the lac promoter by IPTG, during amplification of antibody-displaying phage, reduces the amount of recovered phage (Kretzschmar and Geiser, 1995). Generally, the leakiness of the lac promoter, in the absence of the catabolic repressor, glucose, is sufficient for an efficacious phage display by using classic phagemid. A further reduction of scFv expression, in the case of pKM19, reduces the toxicity of recombinant antibodies for the bacterial host and has no influence on display efficacy. We observed a similar antibody display rate for pKM18 and pKM19 phagemids. A likely explanation for this effect could be that a level of scFv–ΔpIII expression, reduced by the amber codon, already provides a saturating quantity of fused pIII proteins that can be produced by bacterial cells and assembled into the rescued phagemid particles. In a library created by using pKM19, the antibody production is not abundant enough to influence the phage population to favor rapidly growing clones which carry different mutations, reducing antibody expression levels (i.e., frame-shift mutations, stop codons, deletions) and to result in a loss of the real complexity of the generated antibody library. Therefore, the pKM19 phagemid also facilitates selection of harmful antibodies, which interfere with vital bacterial functions. In this case, such antibodies are consequently less toxic due to their lower expression. We demonstrated (i) that the present level of antibody expression is sufficient to produce highly reactive phage antibodies, giving a similar signal in the ELISA test, as compared with pKM18 phage without amber codon and, (ii) that specific antibodies can be easily isolated, after only two selection rounds, from an scFv library constructed from the peripheral blood lymphocytes of a patient with antibodies against a target protein. Third, the pKM19 vector allows the cloning of scFv fragments as amino terminal fusion of the deleted gene III protein. The pIII is a protein of 406 aa residues, composed of two functional domains (Crissman and Smith, 1984). The N-terminal domain is required for bacteriophage infectivity, while the C-terminal domain is incorporated into the virion and is responsible for generation of unit-length particles. Commonly used phage display vectors for scFv lead to incorporation into the phage particles of the entire pIII fused to the antibody fragment, while in the case of pComb3 plasmid utilized for Fab display (Barbas et al., 1991), the antibody fragment is fused to the carboxy terminal half of the pIII. Infectivity of such recombinant phages is obtained during their propagation, since superinfection with a helper phage provides the native gene III protein. According to the present data, fusion of the singlechain antibody to the C-terminal part of pIII improves the phage production and display efficiency of an antibody in comparison with wt pIII protein fusion, in agreement with Kretzschmar's earlier data (Kretzschmar and Geiser, 1995). Improved display efficiency in combination with elimination of free scFv antibodies from the incubation mixture facilitates affinity selection and results in faster enrichment of the phage

pools for specific clones. This also contributes to reduction of stop codons in selected clones, because a lower number of panning/amplification rounds are necessary to complete selection, thus rapidly growing clones have less chance of being isolated. Finally, in bacteria harboring the pKM19 vector, after synthesis of recombinant protein, the PhoA leader peptide is cleaved off by leader peptidase upon membrane translocation, and scFv–ΔpIII is assembled into the phage particle. In this way, the entire cleavage site of the alkaline phosphatase, a genuine periplasmic protein of E. coli, is preserved to guarantee efficient and correct processing and antibody assembly. As a result, the mature protein contains two additional amino acids at the N-terminus of scFv. In the described system, it is necessary to reclone the antibody gene in the appropriate plasmid for the subsequent production of soluble antibodies. In this stage, the additional amino acids can be conserved or eliminated according to specific requirements. The new improved display system was recently applied for selection of tumor-specific antibodies from libraries derived from tumor-infiltrating B lymphocytes (TIL) (Pavoni et al., submitted for publication). It is known that this kind of antibody repertoire is very limited and belongs to clonal groups (Coronella-Wood and Hersh, 2003), indicating TIL as promising sources of -specific antibodies. In fact, a large panel of TIL-derived antibodies capable of recognizing cultured tumor cells was obtained by producing a tumor-derived phage expression library and direct plaque screening protocols which avoided limitations of the phage display system (Wu et al., 2002). However, in using display libraries, several research groups failed to select either a specific antibody discriminating between tumor and normal cells, or one reactive with cell-surface tumor antigens (Coronella et al., 2001; Hansen et al., 2001; Roovers et al., 2001). In applying the pKM19 vector to TIL-derived libraries, we isolated multiple antibodies that specifically bound cultured tumor cells (Pavoni et al., submitted for publication). In conclusion, the combination of relatively low expression of displayed antibodies by introducing the amber codon before antibody gene with improved display efficiency makes the novel pKM19 phagemid useful both for selection of the recombinant scFv antibodies against desired targets from large libraries, as for their affinity maturation. The plasmid guarantees efficient display and allows reduction of biological bias against “difficult” antibodies in the delicate initial selection step. Moreover, this vector is particularly useful for the affinity maturation of antibodies, since high expression levels may increase avidity of phage particles displaying Ab, leading to selection of antibodies with only modest affinity. Acknowledgments We wish to thank Mr. Luca Bruno for always providing excellent technical assistance. We are grateful to Dr. Fiorella Petronzelli (Sigma-Tau, Pomezia (RM), Italy) for Surface Plasmon Resonance (SPR) study with Biacore. We also thank Ms. Marlene Deutsch for the linguistic revision of the text.

E. Pavoni et al. / Gene 391 (2007) 120–129

References Barbas III, C.F., Kang, A.S., Lerner, R.A., Benkovic, S.J., 1991. Assembly of combinatorial antibody libraries on phage surfaces: the gene III site. Proc. Natl. Acad. Sci. U. S. A. 88 (18), 7978–7982. Beekwilder, J., Rakonjac, J., Jongsma, M., Bosch, D., 1999. A phagemid vector using the E. coli phage shock promoter facilitates phage display of toxic proteins. Gene 228, 23–31. Beghetto, E., et al., 2006. Discovery of novel Streptococcus pneumoniae antigens by screening a whole genome lambda-display library. FEMS Microbiol. Lett. 262 (1), 14–21. Bowers, L.M., Lapoint, K., Anthony, L., Pluciennik, A., Filutowicz, M., 2004. Bacterial expression system with tightly regulated gene expression and plasmid copy number. Gene 340 (1), 11–18. Coronella-Wood, J.A., Hersh, E.M., 2003. Naturally occurring B-cell responses to breast cancer. Cancer Immunol. Immunother. 52 (12), 715–738. Coronella, J.A., Telleman, P., Kingsbury, G.A., Truong, T.D., Hays, S., Junghans, R.P., 2001. Evidence for an antigen-driven humoral immune response in medullary ductal breast cancer. Cancer Res. 61 (21), 7889–7899. Crissman, J.W., Smith, G.P., 1984. Gene-III protein of filamentous phages: evidence for a carboxyl-terminal domain with a role in morphogenesis. Virology 132 (2), 445–455. Edelmann, P., Martin, R., Gallant, J., 1987. Nonsense suppression context effects in Escherichia coli bacteriophage T4. Mol. Gen. Genet. 207 (2–3), 517–518. Felici, F., Castagnoli, L., Musacchio, A., Jappelli, R., Cesareni, G., 1991. Selection of antibody ligands from a large library of oligopeptides expressed on a multivalent exposition vector. J. Mol. Biol. 222 (2), 301–310. Goldsmith, M.E., Konigsberg, W.H., 1977. Adsorption protein of the bacteriophage fd: isolation, molecular properties, and location in the virus. Biochemistry 16 (12), 2686–2694. Hansen, M.H., Nielsen, H., Ditzel, H.J., 2001. The tumor-infiltrating B cell response in medullary breast cancer is oligoclonal and directed against the autoantigen actin exposed on the surface of apoptotic cancer cells. Proc. Natl. Acad. Sci. U. S. A. 98 (22), 12659–12664. Harlow, E., Lane, D., 1988. Antibody: A Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, p. 341. Hoogenboom, H.R., Griffiths, A.D., Johnson, K.S., Chiswell, D.J., Hudson, P., Winter, G., 1991. Multi-subunit proteins on the surface of filamentous phage: methodologies for displaying antibody (Fab) heavy and light chains. Nucleic Acids Res. 19 (15), 4133–4137.

129

Khlebnikov, A., Skaug, T., Keasling, J.D., 2002. Modulation of gene expression from the arabinose-inducible araBAD promoter. J. Ind. Microbiol. Biotech. 29 (1), 34–37. Kretzschmar, T., Geiser, M., 1995. Evaluation of antibodies fused to minor coat protein III and major coat protein VIII of bacteriophage M13. Gene 155 (1), 61–65. Marcus, W.D., Lindsay, S.M., Sierks, M.R., 2006. Identification and repair of positive binding antibodies containing randomly generated amber codons from synthetic phage display libraries. Biotechnol. Prog. 22 (3), 919–922. Miller, J.H., Albertini, A.M., 1983. Effects of surrounding sequence on the suppression of nonsense codons. J. Mol. Biol. 164 (1), 59–71. Parker, J., 1989. Errors and alternatives in reading the universal genetic code. Microbiol. Rev. 53 (3), 273–298. Pavoni, E., et al., 2006. Selection, affinity maturation, and characterization of a human scFv antibody against CEA protein. BMC Cancer 6, 41. Pavoni, E., et al., submitted for publication. Tumor-infiltrating B lymphocytes as efficient source of highly specific immunoglobulins recognizing tumor cells. Pini, A., et al., 1998. Design and use of a phage display library. Human antibodies with subnanomolar affinity against a marker of angiogenesis eluted from a two-dimensional gel. J. Biol. Chem. 273 (34), 21769–21776. Pope, A.R., Embleton, M.J., Mernaugh, R., 1996. Construction and use of antibody gene repertoires. In: McCafferty, J., Hoogenboom, H., Chiswell, D. (Eds.), Antibody Engineering — A Practical Approach. Oxford University Press, p. 325. Roovers, R.C., van der Linden, E., Zijlema, H., de Bruine, A., Arends, J.W., Hoogenboom, H.R., 2001. Evidence for a bias toward intracellular antigens in the local humoral anti-tumor immune response of a colorectal cancer patient revealed by phage display. Int. J. Cancer 93 (6), 832–840. Sambrook, J., Fritsch, E.F., Maniatis, T., 1989. Molecular Cloning: A Laboratory Manual, 2nd edn. Cold Spring Harbor Laboratory Press, Cold Spring Harbor. Viti, F., Nilsson, F., Demartis, S., Huber, A., Neri, D., 2000. Design and use of phage display libraries for the selection of antibodies and enzymes. Methods Enzymol. 326, 480–505. Ward, E.S., Gussow, D., Griffiths, A.D., Jones, P.T., Winter, G., 1989. Binding activities of a repertoire of single immunoglobulin variable domains secreted from Escherichia coli. Nature 341 (6242), 544–546 (Oct 12). Wu, H., et al., 2002. Cloning, isolation and characterization of human tumor in situ monoclonal antibodies. Cancer Immunol. Immunother 51 (2), 79–90.