Oncogene (1998) 17, 2445 ± 2456 ã 1998 Stockton Press All rights reserved 0950 ± 9232/98 $12.00 http://www.stockton-press.co.uk/onc Characterization of a new intrabody directed against the N-terminal region of human p53

Pascale A Cohen1, Jean-Claude Mani2 and David P Lane1

1Cancer Research Campaign Cell Transformation Research Group, Department of Biochemistry, Medical Sciences Institute, University of Dundee, Dundee DD1 4HN, UK; 2CNRS UMR 9921, UFR Pharmacie, 15 Av Flahault, 34060 Montpellier cedex 01, France

Genes encoding the rearranged immunoglobulin heavy transcriptional repressor, and down-regulate promo- and light chain variable regions of DO-1, a monoclonal ters lacking wild-type p53 binding sites (Subler et al., directed against human p53, have been used to 1992). construct a single-chain antibody. DO-1 recognizes an The amino-terminal region of p53 is acidic and N-terminal epitope in the region involved in the proline-rich, and holds the transcriptional transactiva- transactivation function of p53 and the binding of tion function of the (Fields and Jang, 1990; Mdm2. The DO-1 single chain scFv expressed in the Raycroft et al., 1990). This region has been shown to periplasm of E. coli or at the surface of the ®lamentous interact with the components of the basal transcrip- phage M13 retained the immunological speci®city and tional machinery TATA-binding (TBP) (Seto anity of the full length antibody. Furthermore, the DO- et al., 1992), the TBP associated coactivator factors 1 recombinant antibody was able to inhibit the in vitro TAFII40, TAFII60 (Thut et al., 1995), TAFII31 (Lu binding of Hdm2, and was shown to be a powerful and Levine, 1995) and members of the transcriptional protecting agent of p53's DNA binding activity at 378C. coactivators p300/CBP family (Gu et al., 1997; Lill et The DO-1 single-chain antibody has been used to al., 1997). The transactivation domain of p53 has been construct single-chain intracellular (intrabo- mapped to residues 1 ± 40 of the N-terminal region of dies) for expression in the cytoplasm and the nucleus of the protein (Unger et al., 1992), and recently, a second mammalian cells. These anti-p53 intrabodies were independent activation domain of p53 has been additionally modi®ed by addition of a Ck domain to identi®ed, and localized to the amino acid 40 ± 83 increase cytoplasmic and nuclear stability. Here we show region (Candau et al., 1997). p53's transactivation that expression of the DO-1 single-chain antibody in the function is negatively regulated by proteins interacting H1299 cell line results in an inhibition of p53's with the N-terminal region, in particular by the E1B- transactivation function. The DO-1 intrabody is a useful 55 kDa protein from adenovirus type 5 (Braithwaite tool to study those functions of p53 driven by the N- and Jenkins, 1989) and the oncogenic Mdm2 protein terminal region of the protein. (Momand et al., 1992; Chen et al., 1993; Oliner et al., 1993). Mutational analysis has demonstrated that Keywords: p53 tumour suppressor; single-chain anti- residues L22 and W23 of the human p53 sequence body; transactivation play a key role in the transcriptional activity of p53, and are crucial for the binding of E1B, Mdm2, TAFs proteins, and CBP protein (Lin et al., 1994; Thut et al., 1995; Lu and Levine, 1995; Gu et al., 1997). This Introduction suggests that inhibition of p53's transcriptional activity by Mdm2 is probably achieved by concealing the The p53 tumour suppressor has become the centre of transactivation domain of p53 (Chen et al., 1993; extensive studies since it has been found to be Oliner et al., 1993). The Mdm2 (murine double minute inactivated, either by missense mutation or interaction gene 2) protein (Cahilly-Snyder et al., 1987) is with oncogenic or viral proteins, in more than half of tumorigenic when overexpressed in nude mice all human cancers. The p53 protein suppresses (Fakharzadeh et al., 1991), and the Mdm2 human oncogenic transformation (Eliyahu et al., 1984; Finlay homolog (Hdm2) is found to be overexpressed in et al., 1989) and induces cell cycle arrest (Leonardo et human sarcomas where the p53 gene remains wild type al., 1994) or programmed cell death (Clarke et al., (Oliner et al., 1992). Much e€ort has been devoted to 1993). p53 acts as a sequence-speci®c transcription the molecular characterization of the Mdm2-p53 factor that activates a number of genes implicated, at interaction. Using synthetic (Picksley et al., 1994) or least in part, in p53-dependent functions. These include phage-displayed (Bottger et al., 1996) peptides, the gadd45 (Kastan et al., 1992), mdm2 (Barak et al., Hdm2-binding site on human p53 has been identi®ed 1993), p21 (waf1/cip1) (El Deiry et al., 1993), cyclin G as the 18TFSDLWKLL26 sequence. A co-crystal- (Okamoto and Beach, 1994), IGF-BP3 (Buckbinder et ization study (Kussie et al., 1996) shows that the N- al., 1995), bax (Miyashita and Reed, 1995) and PIG3 terminus of p53 forms an amphipathic a helix bound (Polyak et al., 1997). p53 can also act as a into a deep hydrophobic pocket on the Mdm2 molecule, and that residues F19, W23 and L26 of p53 are particularly involved in the interaction. In addition to the transactivation function, the N- Correspondence: DP Lane Received 22 December 1997; revised 3 June 1998; accepted 4 June terminal region of p53 seems to be involved in the 1998 regulation of the rate of turnover of the protein. For Characterization of a new anti-p53 scFv PA Cohen et al 2446 instance, deletion of the residues 13-19 results in a cytoplasm of mammalian cells. In this study we show mutant protein more stable than wild-type p53 that the intracellularly expressed scFv is functional as it (Marston et al., 1994). Other studies suggest that results in an inhibition of p53's transactivation calpain is involved in the degradation pathway of p53 function. in the cell system (Kubbutat and Vousden, 1997; Pariat et al., 1997), and that the site of cleavage of calpain on the p53 molecule is localized in vitro in the 20-25 N- Results terminal region (Kubbutat and Vousden, 1997). Ubiquitin-dependent proteolysis is one of the mechan- Engineering of the DO-1 single chain fragment variable isms by which p53 might be degraded (Maki et al., (scFv) 1996); although factors governing recognition by the ubiquitin system are not very well understood, there is Di€erent strategies have been described to produce evidence that residues at the N-terminus of proteins and construct a scFv from a hybridoma cell line. The play a role (Varhavsky, 1992). Ubiquitin-dependent most common method involves separately amplifying proteolysis of p53 has been shown to be promoted by the variable domains of both the heavy (VH) and light the E6 protein from human papillomavirus (Sche€ner (VL) chains of immunoglobulin (Ig), utilizing a set of et al., 1990) and by Mdm2 (Haupt et al., 1997; oligonucleotides which anneal to the 5' and 3' ends of Kubbutat et al., 1997). Interestingly, other proteins the majority of the V region gene families. The DNA belonging to the Mdm2 family have been found to fragments of the two variable regions are then bind to p53 in a similar manner as Mdm2 (Shvarts et assembled into a single gene using a DNA linker al., 1996), opening the way for a new class of fragment encoding the (Gly4Ser)3 sequence. This endogenous proteins modulating p53's functions via approach was unsuccessful for construction of the the N-terminus. DO-1 scFv, because the VH DNA fragment could not Post-translational events such as reversible phos- be ampli®ed. We decided to use an alternative phorylation may also regulate p53's activities. Threo- strategy, through the construction of a cDNA library nine or serine residues within the N-terminal region are from the DO-1 hybridoma cell line. The library was targets for phosphorylation by various kinases: CK I screened for the DO-1 heavy (IgG2a) and light (k) (Milne et al., 1992), raf kinase (Jamal and Zi€, 1995), chains, using two oligonucleotide probes which anneal, MAP kinase (Milne et al., 1994), and DNA-PK (Lees respectively, on the ®rst constant region of IgG2a Miller et al., 1992; Wang and Eckhart, 1992). The mouse heavy chain or on the constant region of DNA-dependent kinase which phosphorylates S15 and mouse kappa light chains. Several positive clones were S37 of human p53 is an interesting candidate, as it has sequenced and primers were designed to construct, by been shown that mutation on S15 modulates the overlap-extension PCR, the DO-1 scFv (see Material protein's activities and stability (Fiscella et al., 1993), and methods). Sequence comparison using data from and that phosphorylation of S15 and S37 impairs the the Immunogenetics Database (DNA plot), showed ability of Mdm2 to inhibit p53-dependent transactiva- that the VH region of the DO-1 antibody belongs to tion (Shieh et al., 1997). p53's interaction with Mdm2 the mouse VH8 family. The third hypervariable may also be a€ected by modi®cation of Mdm2 itself by complementary-determining region (CDR3) is com- the DNA-dependent kinase (Mayo et al., 1997). posed of the DQ52 gene and the JH2 gene. The VL In order to study the regulation of p53's activities region belongs to the mouse Vk1 group II subgroup involving the N-terminal region of the protein, we and uses the JK5 gene. decided to use an innovative approach based on the intracellular expression of recombinant antibodies. E. coli expressed DO-1 scFv keeps the same speci®city Intracellular antibodies (intrabodies) represent one of and the same anity as the DO-1 the most recent advances in antibody engineering (for review: Marasco, 1995). They are synthesized by the To determine the binding activity of the engineered cell and targeted to a speci®c cellular compartment in antibody fragment, the DO-1 scFv was cloned into the order to manipulate the function of the antigen of prokaryotic phagemid expression vector pCANTAB interest. With this in mind, we decided to engineer a 5E. This vector displays the expressed scFv on the recombinant antibody fragment (single chain fragment surface of the ®lamentous M13 phage by fusion with variable or scFv) of the DO-1 monoclonal antibody the gene III minor coat protein (g3p) that is required (Vojtesek et al., 1992). DO-1 antibody binds to human for host infection (McCa€erty et al., 1990). Following p53 at a N-terminal linear epitope 20SDLWKL25 isopropyl-thiogalactoside (IPTG) induction, soluble (Stephen et al., 1995) which overlaps with the Hdm2 scFv can also be expressed in the periplasmic space binding site (Picksley et al., 1994; Bottger et al., 1997) of amber-suppressor bacteria (HB2151 E. coli strain) and the calpain cleavage site (Kubbutat and Vousden, recognizing an amber mutation interposed between 1997). genes encoding the antibody fragment and the coat In our study, we describe the in vitro characteriza- protein (Hoogenboom et al., 1991). To allow the tion of DO-1 scFv after expression in E. coli as a immunodetection of the soluble scFv, the E-tag peptide soluble molecule or at the surface of the ®lamentous was fused in frame at the C-terminus of the Vk domain phage M13. Interaction between human p53 and (Figure 1a). Preparation of pure DO-1 scFv from E. puri®ed scFv from the periplasmic fraction of E. coli coli periplasm was achieved by immunoanity pur- was investigated by enzyme linked immunosorbent i®cation using the anti-E-tag antibody immobilized assays and electrophoretic mobility shift assays. onto a N-hydroxysuccinimide activated Sepharose Di€erent constructs have been designed to allow column (Pharmacia, Uppsala, Sweden). Analysis by stable expression of the scFv in the nucleus or the SDS ± PAGE (silver staining) revealed a band at the Characterization of a new anti-p53 scFv PA Cohen et al 2447 expected size (30 kDa), with a purity greater than 95% DO-1 epitopes are accessible on the tetrameric form of (data not shown). The average yield of immunopur- p53 (Hupp et al., 1992). When the assay is performed i®cation from the periplasmic fractions was 100 ± in the presence of the scFv, the p53/DNA band is 120 mg of scFv per liter of bacterial culture. supershifted, demonstrating that recombinant antibody Phage-displayed or soluble DO-1 scFv were tested is able to recognize the p53/DNA complex. A titration for binding to human p53 by an enzyme linked of DO-1 scFv leads to the formation of three immunosorbent assay (ELISA). As shown on Figure intermediate complexes (Figure 3, lanes 2 ± 7), demon- 2a, the soluble DO-1 scFv puri®ed from E. coli strating that all four N-terminal DO-1 epitopes are periplasm is able to recognize recombinant human accessible on the tetrameric form of p53. p53 expressed in di€erent systems: E. coli, insect cells, Finally, we characterized the kinetic rates of the or human p53 null tumour cell line H1299. This result DO-1/p53 interaction by surface plasmon resonance. was con®rmed for the phage-displayed scFv (data not Analysis of the association and the dissociation phases shown). The DO-1 epitope has been mapped by of the interaction between the immobilized E. coli Stephen and colleagues (1995) as the 20SDLWKL25 human p53 and DO-1 was performed for the full sequence. Fine speci®city of the DO-1 scFv has been length antibody and the scFv. The scFv's ka value investigated by testing the recognition of biotinylated (6.16105 M71 s71) is slightly higher than the DO-1's peptides containing the human DO-1 epitope and the value (1.36105 M71 s71), revealing a better accessibility corresponding region of murine p53. Figure 2b shows of the DO-1 epitope for the smaller molecule (30 kDa) that the recombinant antibody fragment expressed at than for the full length antibody (150 kDa). Analysis the surface of the M13 phage retains the same speci®city as the DO-1 parental antibody; that is that it is able to recognize human but not murine p53. This was also con®rmed with the soluble DO-1 scFv (data a not shown). In order to test the recognition of human p53 in another format, we performed an electrophoretic mobility shift assay (EMSA) in which insect cell- produced recombinant human p53 speci®cally binds to DNA as a tetrameric complex (Hupp and Lane, 1994). A titration of DO-1 parental antibody gave rise to an intermediate form of the p53/DNA complex (Figure 3, lanes 9 ± 15), corresponding to two DO-1 Ig molecules bound per tetrameric p53, suggesting that at least two

b

Figure 2 Expression of the DO-1 scFv in E. coli.(a) an ELISA Figure 1 DO-1 scFv constructs. (a) scFv: single chain fragment capture assay was used to measure the binding activity of the variable comprising the VH and Vk immunoglobulin chains puri®ed soluble DO-1 scFv (100 ng/ml) against recombinant linked together by the (Gly4Ser)3 sequence linker (L), and fused in human p53 expressed in insect cells (0.4 mg/ml), E. coli (0.6 mg/ frame with the E-Tag sequence to allow immunodetection. This ml), or human tumour H1299 cells (black columns). Controls in construct was expressed in E. coli or in mammalian cells where it the white columns represent background signal in the absence of is speci®cally targeted for cytoplasmic expression. (b) scFv-SV40: p53. (b) ELISA to determine the ®ne speci®city of DO-1 antibody as (a), but with addition of a C-terminal SV40 nuclear (2 mg/ml) (white column) and phage-displayed scFv (black localization signal TPPKKKRKV for speci®c expression in the column) against di€erent biotinylated synthetic peptides contain- nuclear compartment of the cell. (c) scFvCk:as(a) for ing the human DO-1 epitope (SDLWKL sequence, peptides A cytoplasmic expression but with the entire mouse Ck domain and B), or the corresponding region of murine p53 (SGLWKL fused in frame between the Vk gene and the E-Tag sequence. (d) sequence, peptides C and D). Peptide A: SGSGEPPLS- scFvCk-SV40: as (c) but with addition of the SV40 sequence fused QETFSDLWKL. Peptide B: SGSQETFSDLWKLLPENN. Pep- in frame to the last amino acid of the E-Tag peptide for nuclear tide C: SGSGTFSGLWKLLPPEDIL. Peptide D: targeting SGSGLWKLLPPEDILPSPHC Characterization of a new anti-p53 scFv PA Cohen et al 2448

a

Figure 3 Formation of intermediates during DO-1 scFv binding to p53/DNA complexes. Standard reactions were assembled with 90 nM of insect-cell produced p53 and the indicated amount of DO-1 scFv: none (lane 1), 30 nM (lane 2), 60 nM (lane 3), 80 nM (lane 4), 125 nM (lane 5), 250 nM (lane 6), 500 nM (lane 7). In a b similar way, a titration was performed with the DO-1 antibody: none (lane 8), 4 nM (lane 9), 8 nM (lane 10), 16 nM (lane 11), 32 nM (lane 12), 65 nM (lane 13), 125 nM (lane 14), 250 nM (lane 15). The white arrow indicates the migration of insect-cell produced p53/DNA complexes. The black arrow marks the migration of the DO-1 scFv/p53/DNA complexes and the hatched arrow the migration of the DO-1 antibody/p53/DNA complexes

of the kd values also shows a more rapid dissociation for the scFv (2.061074 s71) than for the DO-1 antibody (3.661075 s71). These two kinetics para- meters result in an association constant Ka almost identical for both antibodies: 3.06109 M71 for the scFv and 3.66109 M71 for the DO-1 antibody. Taken together, these data indicate that the Figure 4 Inhibition of the Hdm2-p53 interaction by the DO-1 recombinant DO-1 scFv has the same immunological scFv. (a) Sensorgram (BIAcore 2000 technology): preincubation of properties as the full length DO-1 antibody, in terms of 660 nM of DO-1 scFv (curve 1) or 330 nM of DO-1 antibody ®ne speci®city and anity. (curve 2) with E. coli produced human p53 immobilized on the sensor chip leads to the loss of the p53-binding of GST-Hdm2 (260 nM, illustrated on curve 4). Control experiment is performed with an irrelevant anti-Troponin I antibody (330 nM, illustrated on DO-1 scFv inhibits the p53-Hdm2 interaction curve 3). The response is expressed in resonance units (RU), as 1 RU corresponds to a surface concentration of 1 pg of reactant It has been shown that both the DO-1 antibody and 2 Hdm2 interact with p53 at overlapping binding sites per mm of sensor chip. (b) Gel mobility shift assay: insect-cell produced human p53 (90 nM) was preincubated for 30 min alone (Picksley et al., 1994; Stephen et al., 1995; Bottger et al., (lanes 1 and 3) or with GST-Hdm2 (1300 nM) (lanes 2, 4 ± 9). The 1996). DO-1 binds the 20SDLWKL25 sequence and puri®ed DO-1 scFv was then added and the incubation continued Hdm2 the 18TFSDLWKLL26 sequence. An important for 30 min: 1 nM (lane 4), 2 nM (lane 5), 10 nM (lane 6), 30 nM question to determine prior to iniating the expression of (lane 7), 120 nM (lane 8) and 500 nM (lanes 3 and 9). The white arrow indicates insect cell-produced p53/DNA complexes migra- the scFv in cells was whether the DO-1 recombinant tion, and the black arrow the DO-1 scFv/p53/DNA complexes antibody fragment is able to inhibit p53-Hdm2 position interaction. This was investigated by surface plasmon resonance measurements where Hdm2 binding to immobilized human p53 was examined before and after pre-incubation with DO-1 scFv or DO-1 full DO-1 scFv is a powerful protecting agent of DNA length antibody. Figure 4a illustrates a sensorgram binding activity of p53 at 378C where the presence of the DO-1 scFv (660 nM) prevents the binding of Hdm2 to p53 at least to the same extent It is known that the DNA binding activity of p53 is as the DO-1 antibody (330 nM). Control experiment lost following extended incubations at 378C (Hansen shows that an irrelevant antibody (330 nM) has no e€ect et al., 1996). Further stabilization of p53's DNA on the Hdm2-p53 interaction (Figure 4a, curve 3). binding activity at 378C can be regulated by Bottger and colleagues (1997) have shown that p53 interaction with heat shock proteins or monoclonal retains its sequence-speci®c DNA binding activity after antibodies recognizing N-terminal epitopes on p53 complex formation with Hdm2. Figure 4b illustrates (DO2, DO1, 1801 antibodies). Hansen and colleagues the results of DO-1 scFv's e€ects on the Hdm2-p53 (1996) showed that ecient stabilization only takes interaction as determined by EMSA. A titration of the place when all subunits of the tetrameric p53 are DO-1 scFv (Figure 4b, lanes 4 ± 9) leads to the bound by two bivalent antibodies or four monovalent disappearance of Hdm2/p53 complexes and the Fabs. The mechanism of the stabilization event appearance of scFv/p53 complexes, demonstrating induced by the monoclonal antibodies seems to be that the recombinant antibody is able to displace mediated by a protection of the wild-type conforma- Hdm2 from human p53. tion of p53. Characterization of a new anti-p53 scFv PA Cohen et al 2449 It was of interest to investigate the behaviour of the a DO-1 scFv under the same experimental conditions. As described by Hansen and colleagues (1996), incubation of insect cell-produced p53 at 378C for 60 min leads to the loss of the protein's DNA binding activity (Figure 5a, compare lane 1 and lane 5). When the incubation of p53 at 378C is performed in the presence of the DO-1 antibody, p53 is protected by the N-terminal antibody against heat denaturation and DNA binding activity is observed (Figure 5a, compare lane 8 to lane 5). Figure 5a (compare lanes 6 and 7 to lane 5) demonstrates that p53 is still able to bind DNA when the incubation at 378Cis performed in the presence of the DO-1 scFv. In order to quantitatively analyse the protection achieved by the recombinant antibody fragment and by the DO-1 monoclonal antibody, we performed the assay with two di€erent concentrations of the monovalent scFv (250 or 500 nM) and with 250 nM of bivalent DO-1 antibody. Protection is expressed as the relative percentage of DNA binding after incubation at 378C compared with the binding at 48C (Figure 5b). Interestingly, with the same number of paratopes of DO-1 present in the reaction (500 nM for the scFv, 250 nM for the DO-1 full length antibody), the DO-1 scFv is shown to be a more potent stabilizing agent by protecting up to 55+1.5% the DNA binding activity of p53, compared to a 20+5.8% protection for the DO-1 antibody. When the same molarity of both antibodies is tested (250 nM), the scFv is still slightly more powerful than DO-1 full length antibody, by achieving a protection of 35+5.8%. Hansen and colleagues (1996) found that the binding of four monovalent Fabs of the 1801 antibody is able to protect p53 from heat denaturation, but to a lesser extent than the parental 1801 antibody, suggesting that the crosslink which the bivalent 1801 antibody could achieve between monomers of p53 could be important in this phenomenon. Our results with the DO-1 scFv show that the bivalency of N-terminal antibodies is not required to ensure the protection of p53's DNA binding activities at 378C, as the binding of small scFv molecules on the DO- 1 epitope of wild-type p53 results in a more powerful b protection than the one obtained with the binding of DO-1 full length antibody.

Expression of the DO-1 anti-p53 scFv in the cytoplasm and the nucleus of mammalian cells Foreign proteins expressed in the cytoplasm of eukaryotic cells may be prone to rapid degradation in the reducing environment of the cytoplasm. Several anti-p53 expression vectors were constructed to target the anti-p53 DO-1 scFv either to the cytoplasm or to the nucleus of mammalian cells (Mhashilkar et al., 1995). To ensure a high level of expression, the coding sequence of all DO-1 anti-p53 scFvs were modi®ed to contain a strong Kozak sequence and an initiating methionine immediately 5' to amino acid one of framework one of the heavy chain. A derivative was constructed that contained, fused in frame with the C- Figure 5 Stabilization of p53's DNA binding activity at 378Cby the DO-1 scFv. (a) insect cell-produced p53 (15 nM) was tested alone (lanes 1 and 5), with 500 nM of DO-1 scFv (lanes 2 and 6), with 250 nM of DO-1 scFv (lanes 3 and 7), or with 250 nM of cell-produced human p53 in presence of DO-1 scFv (500 and DO-1 antibody (lanes 4 and 8). The mixtures were incubated for 250 nM) or DO-1 antibody (250 nM). Signals are expressed in 60 min at 48C (lanes 1 ± 4) or at 378C (lanes 5 ± 8) before analysis relation to activity at 48C. The data represent the mean of three in a DNA binding assay. (b) Quantitative analysis of the independent experiments. The error bars represent the standard stabilization of DNA binding activity at 378Cof15nM of insect error of the mean Characterization of a new anti-p53 scFv PA Cohen et al 2450 terminus of the VL domain, the entire mouse kappa terminus of the DO-1 scFv dramatically increases the light chain constant domain (Ck), followed by the E- stability of the recombinant antibody in an eukaryotic tag peptide, to allow immunodetection (Figure 1c). context. This construct was called scFvCk and expressed in E. coli to check its ability to bind to human p53 (data not DO-1 scFv inhibits p53-mediated transactivation in shown). For nuclear targeting, the scFv and scFvCk H1299 cell line were additionally modi®ed to contain the C-terminal SV40 nuclear localization signal TPPKKKRKV which p53 activates the transcription of genes containing the has been shown to direct heterologous proteins into the p53 consensus sequence. It has already been described nucleus (Yoneda et al., 1992). The two nucleus-targeted that Hdm2 binding is able to inhibit p53 transactiva- scFvs were called respectively scFv-SV40 (Figure 1b) tion function, probably by concealing the transactiva- and scFvCk-SV40 (Figure 1d). The four scFvs tion domain of p53 from the TAF proteins (Chen et constructs were cloned into the pcDNA3 vector, al., 1993; Oliner et al., 1993). As DO-1 binds to the under the control of the cytomegalovirus promoter same region as Hdm2, we examined whether cotrans- (CMV). fection with pcDNA3-scFvCk and pcDNA3-scFvCk- The ability of the di€erent DO-1 scFvs to be SV40 plasmids would interfere with p53-mediated expressed in mammalian cells was determined by transcriptional activation. Therefore we transfected transient transfection of the p537/7 H1299 cell line p53 null human lung carcinoma H1299 cells with the or the p53+/+ MCF7 cell line. Transfection with the PG13-PY-Luc reporter construct, containing thirteen plasmids encoding the scFvCk or scFvCk-SV40 RGC sequences upstream of a promoter driving proteins (pcDNA3-scFvCk and pcDNA3-scFvCk- transcription of the ®re¯y luciferase gene (Kern et al., SV40), resulted in strong cytoplasmic and nuclear 1992). As shown on Figure 8a, co-transfection with the staining respectively, showing abundant accumulation plasmid expressing human wild-type p53 (pcDNA3- of the scFv proteins (Figure 6). It was noteworthy that human p53) strongly enhanced the expression of the the ¯uorescence intensity observed after transfection reporter construct (lane 2). As previously described, with the plasmid expressing the scFv-SV40 protein was simultaneous transfection with the pCMV-Neobam- weaker than with the scFvCk-SV40 construct, and that hdm2 plasmid inhibits the transcription stimulation the ¯uorescence for the cytoplasmic-targeted scFv activity of p53 (83+5% of inhibition) (Figure 8a, lane construct was almost non existant. We performed a 3). Transfection with pcDNA3-human p53 in presence Western blot with lysates of H1299 cells transfected of the pcDNA3-scFvCk plasmid leads to a 50.7+8% with the same amount of each of the four scFv of inhibition of transcription (Figure 8a, lane 5), and constructs (10 mg of DNA), and probed with the anti- co-transfection with pcDNA3-scFvCk-SV40 induces a E-Tag antibody. A strongly expressed band was visible less marked but reproducible 23.4+1% of inhibition for the scFvCk-SV40 and scFvCk proteins (*40 KDa) (Figure 8a, lane 4). In the absence of p53, co- (Figure 7, lanes 1 and 2), whereas the expression of the transfection of either the scFv-Ck or scFvCk-SV40 scFv-SV40 protein and in particular the scFv protein constructs with the reporter had no signi®cant e€ect on (*30 KDa) were weaker (Figure 7, lanes 4 and 5), the background signal observed with the reporter alone con®rming the immunostaining observations. All these (data not shown). To support the conclusion that p53 data suggest that the construct lacking the Ck domain inactivation results from binding between the human is rapidly degraded in the cytoplasm of mammalian p53 protein and the DO-1 intrabody protein, we cells, and that addition of the Ck domain at the C- performed the same experiment using a plasmid

Figure 6 Expression of scFvCk and scFvCk-SV40 proteins in eukaryotic cells. (a) Immuno¯uorescent staining (using the anti E- Tag antibody) of transiently transfected H1299 cells (p537/7)(b) Immuno¯uorescent staining (using the anti E-Tag antibody) of transiently transfected MCF7 cells (p53+/+). For (a) and (b) left panels, transfection with pcDNA3 vector; middle panels, transfection with pcDNA3-scFvCk; right panels, transfection with pcDNA3-scFvCk-SV40. The calibration bar in (a) and (b) indicates 30 mm Characterization of a new anti-p53 scFv PA Cohen et al 2451

a

Figure 7 Western blot of transiently transfected H1299 cells with 10 mg of: pcDNA3-scFvCk-SV40 (lane 1), pcDNA3-scFvCk (lane 2), pcDNA3 vector (lane 3), pcDNA3-scFv-SV40 (lane 4), and pcDNA3-scFv (lane 5). Detection was performed with the anti E- Tag antibody. On the left side of the ®gure the molecular weight markers (kDa) are shown b expressing the mouse p53 protein, as this is not recognized by the DO-1 antibody. Our peptide binding studies con®rmed the species speci®city of the DO-1 scFv since it is not able to recognize murine p53 (Figure 2b). No inhibition of mouse p53-mediated transcription was observed in the presence of pcDNA3- scFvCk-SV40 or pcDNA3-scFvCk (Figure 8b, lanes 4 and 5), demonstrating that the observed inhibition of human p53-mediated transcription is the result of a speci®c protein-protein interaction, and is not a consequence of a generic e€ect of the intrabody on the cellular transcriptional machinery. Since transfec- tion with the plasmid expressing the Mdm2 protein (pCOC-mdm2X2) is able to eciently inhibit the transcriptional activity of murine p53 in this system (Figure 8b, lane 3), these results provide striking proof of the speci®city of the DO-1 intrabody system. Figure 8 Speci®c inhibition of p53-mediated transcription activation in the p537/7 H1299 cell line. (a) Transfection with the reporter gene PG13-PY-luciferase alone (1) or in presence of pcDNA3-human p53 (2), pcDNA3-human p53+pCMV-Neobam- Discussion hdm2 (3), pcDNA3-human p53+pcDNA3-scFvCk-SV40 (4), pcDNA3-human p53+pcDNA3-scFvCk (5). (b)as(a), but Antibody engineering allows the production of pcDNA3-human p53 is replaced with pcDNA3-mouse p53, and pCMV-Neobam-hdm2 with pCOC-mdm2X2. The data represent recombinant scFv constructs which combine, in a the mean of at least three independent experiments where each single polypeptide chain, both the variable light and point was performed in duplicate. The error bars represent the heavy chain domains of an antibody, covalently joined standard error of the mean by an engineered polypeptide linker. The most frequently used linker is the (Gly4Ser)3 sequence, as NMR evidence suggests this linker is a ¯exible entity which does not change the structure of the variable primary sequence of the antibody. The bacterially- domains (Freund et al., 1993). The production of expressed scFv-binding site must be presented in such a antibody fragments from E. coli is achieved by the way that it accurately represents the ligand binding site functional expression of soluble, correctly folded, of the intact antibody, both with respect to anity for fragments into the periplasmic space of the bacteria, the ligand and ®ne speci®city of the binding site. In while one of the main diculties is the formation of many cases, scFvs show less anity than the whole aggregated insoluble material. The intramolecular antibody, and these di€erences have been attributed to disul®de bond within antibody domains appears to be alterations in the folding of the recombinant protein crucial for stability, and most antibody domains do not molecule, or to the additive impact on anity have sucient free energy of folding to tolerate measurement of two binding sites versus the presence removal of this stabilizing element. Therefore a of a single site, in antibodies and scFvs, respectively. competition between folding and aggregation occurs, In this study, we describe the expression of a whose determinants are external factors such as functional DO-1 scFv in the periplasm of E. coli or temperature and growth parameters, as well as the at the surface of the ®lamentous phage M13. The Characterization of a new anti-p53 scFv PA Cohen et al 2452 soluble recombinant antibody was shown to possess CBP protein (Lin et al., 1994; Thut et al., 1995; Lu and the immunological properties of the full length DO-1 Levine, 1995; Gu et al., 1997). It has been suggested antibody, namely the same ®ne speci®ty and high that Mdm2 inhibits p53-dependent transcription by anity. Moreover, we showed in EMSA experiments concealing p53's transactivation domain from compo- that the monovalent DO-1 scFv is able to protect the nents of the transcriptional machinery (Oliner et al., DNA binding activity of p53 from heat denaturation at 1993). In this work, we show that both the cytoplasmic 378C. Analysis of the data highlights that this scFvCk version and to a less extent the nuclear protection is better achieved by the binding of a small scFvCk-SV40 version lead to a signi®cant inhibition molecule (scFv) on the N-terminal 20-25 region of p53 of the transcriptional activity of human p53 in the than by the full length DO-1 antibody. H1299 cell line. This observation is the result of a Recent advances in antibody engineering have now speci®c interaction between human p53 and the allowed the genes encoding antibodies to be manipu- recombinant antibody, as no inhibition of mouse lated so that antigen binding molecules can be p53-mediated transcription is observed under the same expressed intracellularly. The speci®c and high-anity experimental conditions. The di€erence observed in the binding properties of antibodies, combined with their inhibitory e€ect of the scFvCk (50.7%) and the ability to be stably expressed in precise intracellular scFvCk-SV40 (23.4%) may re¯ect a di€erence of locations inside mammalian cells, has provided a folding of the scFvs in the di€erent cellular compart- powerful new family of molecules for basic research ments, or a better accessibility of DO-1 towards its or (for review, see Marasco, 1995). These epitope on the p53 protein in the cytoplasm rather than intracellular antibodies (intrabodies) can be used to the nucleus. As no trapping of p53 in the cytoplasm by modulate cellular physiology and metabolism by a the scFvCk construct is observed in H1299 cells by wide variety of mechanisms: blocking, stabilizing or double immunostaining (data not shown), it is likely mimicking protein-protein interaction, modulating that the scFvCk/p53 complex occurs as soon as both enzyme function, or diverting proteins from their proteins are synthesized in the cytoplasm, and p53 may usual compartment. Intrabodies can be directed to then shuttle the scFv to the nucleus. The observed the relevant cellular compartments by modifying the inhibition of transcription, which may occur by intrabody genes with N- or C-terminal polypeptide concealing the transactivation domain of p53 to TAFs extensions that encode classic intracellular-tracking and/or CBP/p300 protein, could then be achieved by signals. This approach has been described for di€erent the pre-bound scFvCk. One can imagine that in the antigens, including HIV proteins (Mhashilkar et al., nucleus, the scFvCk-SV40 construct would have to 1995; Maciejewski et al., 1995), erbB-2 (Beerli et al., compete directly for binding with the components of 1994; Wright et al., 1997), EGFR (Jannot et al., 1996), the cellular transcriptional machinery, and would thus and p21ras (Biocca et al., 1993). In the p53 ®eld, the 421 be less e€ective than the scFvCk construct. Recent scFv, recognizing a C-terminal epitope of the protein, work suggests that repression of p53-mediated tran- has been characterized in vitro, but when expressed in scription by Mdm2 could be mediated by a second the cytoplasm or the nucleus of COS-1 cells, was non mechanism involving a direct inhibition of the basal functional and prone to rapid degradation (Jannot and machinery through an interaction with the TFIIE Hynes, 1997). As discussed, an important determinant factor (Thut et al., 1997). In this case, the di€erence of antibody folding is the formation of the intra-chain in the inhibition levels observed between Hdm2 (up to disul®de bonds in the variable regions, and the 87%) and the DO-1 scFv (up to 59%) might re¯ect the reducing environment of the cytosol may therefore mimicing of only one of the mechanisms by which lead to a decrease of stability of the scFv (Biocca et al., Mdm2 inhibits p53's transactivation. 1995). However, despite their reduced state, some scFv Phosphorylation at the N and C-terminus of p53 have been expressed in the cytoplasm and have may modulate the transactivation function of the demonstrated biological e€ects (Duan et al., 1994; protein and this e€ect may vary between the Mhashilkar et al., 1995), suggesting that other features, promoters of the di€erent p53 responsive genes such as the primary sequence of the antibody or the (Hecker et al., 1996; Lorhum and Scheidtmann, speci®c cellular environment also have to be taken into 1996). As the binding of DO-1 to p53 is sensitive to consideration. phosphorylation (T Hupp and A Bottger, manuscripts In this study we describe the intracellular expression in preperation), it would be of interest to determine if of a scFv directed against an N-terminal epitope of DO-1 recognizes a discrete sub-fraction of p53 in the human p53. The DO-1 scFv has been targeted to the cell, and if the DO-1 scFv's activity towards p53 cytoplasm and the nucleus of mammalian cells, and the transactivation varies with di€erent promoters. addition of the Ck domain to the scFv as a scFvCk In our in vitro data, we show that the DO-1 scFv is fusion protein leads to a dramatic increase in the level able to inhibit the binding of Hdm2 to p53; it needs to of intracellular expression. Di€erent observations be determined if the DO-1 scFv is able to prevent the demonstrated that for some scFv the addition of the formation of Hdm2/p53 complexes in cells. Recent Ck domain increases the stability of the recombinant work suggested that Mdm2 is involved in the rapid antibody in an eukaryotic context (Mhashilkar et al., degradation of p53 through the ubiquitin-dependent 1995), but does not for other scFv (Jannot and Hynes, pathway (Haupt et al., 1997; Kubbutat et al., 1997). 1997), showing once more that each scFv is a Other studies suggest that calpain proteolysis of p53 in particular case. the cell system may occur, and that this proteolysis can DO-1 antibody recognizes an N-terminal epitope on be prevented in vitro by the binding of DO-1 antibody human p53 (amino acids 20 ± 25), within this epitope (Kubbutat and Vousden, 1997; Pariat et al., 1997). It the residues L22 and W23 have been shown to be would be of interest to study how the DO-1 scFv crucial for the binding of Mdm2, TAFs proteins and would interfere intracellularly with these di€erent Characterization of a new anti-p53 scFv PA Cohen et al 2453 mechanisms, in order to elucidate the way in which the After centrifugation, supernatants were removed, and the rate of turnover of p53 is controlled. The DO-1 phage suspensions were ®ltered through a 0.45 mM ®lter intrabody should therefore provide an interesting and andstoredat48C. original tool to study the regulation of p53 functions driven by the N-terminal region of the protein. Bacterial expression of the soluble scFv E. coli HB2151 nonsuppressor strain was transformed by infection with the phage suspension. One positive colony Materials and methods isolated on a SOBAG agar plate 100 mg/ml nalidixic acid was then grown in 26YT-AG medium at 308Cuntilthe Engineering of the DO-1 scFv OD600 reaches 0.5. This fresh culture was induced in 26 YT 100 mg/ml ampicillin 1 mM IPTG and shaken overnight The DO-1 anti-p53 monoclonal antibody (IgG2ak)andits at room temperature. The periplasmic fraction containing characteristics have been described previously (Vojtesek et the soluble scFv was prepared by resuspending the al., 1992; Stephen et al., 1995). mRNA was prepared from bacterial pellet in 0.2 M TrisHClpH8,0.5mM EDTA, the DO-1 hybridoma cells using the Quickprep Micro 0.5 M sucrose, and incubating on ice for 45 min. After mRNA puri®cation kit (Pharmacia Biotech, Uppsala, centrifugation, the supernatant containing the soluble scFv Sweden),andthenusedastemplatetoconstructacDNA was stored at 7208C. library in the pSPORT 1 plasmid, according to the SuperScript Plasmid System procedure (Gibco BRL, Gathersburg, MD, USA). The bacterial colonies were Puri®cation of the soluble scFv replicated on a nylon transfer membrane (Hybond N+, The soluble DO-1 scFv was immunopurifed from the Amersham, Buckinghamshire, UK) and then screened by periplasmic fraction on an anti E-Tag Sepharose column hybridization (Sambrook et al., 1989) using the 5'-GCA (Pharmacia). Brie¯y, after equilibration of the column at TCC TAG AGT CAC CGA-3' oligonucleotide (comple- pH 7 with a 0.2 M Na /NaHPO bu€er, the periplasmic mentary to the ®rst constant domain of murine IgG2a) and 2 4 fraction was loaded onto the column. After extensive washes the 5'-GTT GTT CAA GAA GCA CAC G-3' oligonucleo- with the phosphate bu€er, the DO-1 scFv was eluted with tide (complementary to the constant domain of mouse 1 M glycine pH 3 bu€er, and the collected fractions were kappa light chains) labelled with g-32P-ATP (5'-end immediatly equilibrated to pH 7 with 1 Tris pH 8.2 labelling RPN1 1509 kit, Amersham). The positive clones M bu€er. The di€erent fractions were assayed for the presence were sequenced, and primers were designed to PCR- of the scFv by enzyme linked immunosorbent assay (ELISA) amplify the VH and Vk DNA fragments with the Vent and by silver stained SDS gel (ICN, Irvine, CA, USA). The DNA polymerase (New Englands Biolabs, Beverly, MA, scFv containing-fractions were pooled and dialyzed over- USA). The VHreverse-SfIIprimerwas5'-CAT GCC ATG night against 0.1 KH PO pH 7.4 bu€er at 48C. The ACT CGC GGC CCA GCC GGC CAT GGC CCA GGT M 2 4 puri®ed scFv was then aliquoted and stored at 7708C. TAC TCT GAA AGA GTC T-3' and the VHforward primer was 5'-GCC AGA GCC ACC TCC GCC TGA ACC GCC TCC ACC TGA AGA GAC TGT GAG AGG GGT Human p53 ELISA GCC-3'.TheVkreverse primer was 5'-TCA GGC GGA GGT GGC TCT GGC GGT GGC GGA TCG GAT GTT Anti-E tag antibody (Pharmacia), 50 ml, 5 mg/ml in a TTG ATG ACC CAA ATT CCA CTC TCC-3' and the phosphate-bu€ered saline solution (PBS) was adsorbed Vkforward-NotIprimerwas5'-GAG TCA TTC TGC GGC onto a 96-well PVC assay plate (Falcon) overnight at 48C. CGC CCG TTT CAG CTC CAG CTT GGT CCC AGC- After washing with 0.1% Tween 20 PBS (PBST), the wells 3'.TheVkreverse and the VHforward primers contain were blocked with 100 ml PBST containing 5% w/v low fat overlapping sequences (appear in italics) corresponding to milk powder (PBSMT) for 2 h at room temperature (RT). the internal ®fteen amino acid-linker (Gly4Ser)3 of the After washing with PBST, puri®ed bacterial soluble scFv scFv. The DO-1 scFv was then assembled by overlap (0.2 mg/ml in a volume of 50 ml in PBSMT) was added into extension in a PCR reaction by mixing the VHreverse-S®I each well and incubated for 2 h at 48C. After washes with primer, the Vkforward-NotI primer, the VH DNA PBST, human p53 was incubated (50 ml, diluted in fragment, and the Vk DNA fragment. The ampli®ed scFv PBSMT) fo 2 h at 48C.Afterwashes(PBST),therabbit fragment (*800 bp) was then S®IandNotI-digested and anti-p53 serum CM1 was added (50 ml, 1/2000 in PBSMT) inserted into the pCANTAB 5 E phagemid vector and incubated for 2 h at 48C. After washes, detection was (Pharmacia). To engineer the scFvCk construct performed with a swine horse radish peroxidase(HRP)- (*1200 bp), the Vkforward-NotI primer was replaced by conjugated anti-rabbit serum (Dako, Glostrup, Denmark). the Ckforward-NotIprimer(5'-GAG TCA TTC TGC Wild type human p53 was expressed and puri®ed from E. GGC CGC ACA CTC ATT CCT GTT GAA GCT CTT- coli as described elsewhere (Hupp et al., 1992; Hansen et al., 3') which anneals to the 3'-end of the constant domain of 1996) and tested at a 0.6 mg/ml concentration. Human p53 mouse kappa light chain sequence. produced in insect cells was kindly provided by T Hupp and tested at a 0.4 mg/ml concentration. Lysate of H1299 cells transiently transfected with human p53 (Lin and Green, Bacterial expression of the scFv at the surface of the ®lamentous 1989) was prepared as described by Harlow and Lane (1988). phage M13 One colony of E. coli TG1 bacteria transformed with the pCANTAB5E-scFv was grown at 308Cin26TY medium, ELISA with biotinylated synthetic peptides. 100 mg/ml ampicillin, 2% glucose (26YT-AG medium) The human p53 sequence-based peptides have already been until the the culture reached an OD600 of 0.5. To the described (Stephen et al., 1995). The peptide ELISA was bacterial suspensions, helper phage M13KO7 (Pharmacia) performed by coating a 96-well PVC plate with Streptavi- was added to a ®nal concentration of 109 pfu/ml and din (Vector Labs, Peterborough, UK), 100 ml, 5 mg/ml in infection was allowed to proceed for 2 h at 378Cwith PBS overnight at 378C. After washing with PBST, the wells shaking. After centrifugation, the bacterial pellets were were blocked with PBSMT (200 ml) for 1 h at RT. resuspended in 26YT 100 mg/ml ampicillin 50 mg/ml Biotinylated peptides were then added to the streptavidin- kanamycin and grown overnight at 378C with shaking. coated well (50 ml; 5 mg/ml in PBSMT) for 2 h at RT. After Characterization of a new anti-p53 scFv PA Cohen et al 2454 washes, the phages suspension (1/8 in PBSMT) was added ACG CGG TTC CAG CGG ATC-3')wereused.ThePCR in a 50 ml volume, and incubated for 2 h at 48C. After products encoding for the di€erent constructs were cloned in washes with PBS, detection was performed using the HRP- the pGEM-T Easy vector (Promega, Madison, WI, USA), conjugated anti-M13 serum (Pharmacia) for 2 h at 48C. then digested by XbaIandEcoRI enzymes before cloning in Control was performed by replacing the phage suspension the eukaryotic expression vector pcDNA3 (Invitrogen, by DO-1 antibody (2 mg/ml in PBSMT), and the HRP- Leek, Netherlands) under the CMV promoter. The four conjugated anti-phage by a rabbit HRP-conjugated anti- di€erent constructs were called pcDNA3-scFv, pcDNA3- mouse IgG serum (Dako). scFv-SV40, pcDNA3-scFvCk, and pcDNA3-scFvCk-SV40.

Real-time analysis of the DO-1/p53 interaction Transient transfections for immuno¯uorescence stainings or Interaction between DO-1 antibody or DO-1 scFv were Western blot studies of the eukaryotic expressed scfv constructs investigated using BIAcore 2000 technology (Biacore AB, Human breast tumour (p53+/+) MCF7 cells and p53 null Uppsala, Sweden) (Malmqvist, 1993). E. coli recombinant human lung carcinoma H1299 cells were maintained in human p53 was immobilized at pH 6.2 on a sensor chip Dulbecco's Modi®ed Eagle Medium (DMEM), supplemen- surface CM5 previously activated with 100 mM N-ethyl-N'- ted with 10% foetal calf serum (FCS) at 378C. Cells were (3-dimethylaminopropyl) carbodiimide hydrochloride/ seeded 24 h before transfection to 70 ± 80% con¯uence. 400 mM hydroxysuccinimide (EDC : NHS). The amount of Calcium phosphate mediated transfections were performed immobilized p53 was 600 pg/mm2. Several concentrations as previously described (Lin and Green, 1989). Three mgof of either full length DO-1 antibody or DO-1 scFv were each of the plasmids encoding the di€erent scFv proteins injected at a ¯ow rate of 20 ml/min. Dissociation was (pcDNA3-scFv, pcDNA3-scFv-SV40, pcDNA3-scFvCk,or observed for 400 s in running bu€er only. Regeneration of pcDNA3-scFvCk-SV40) or 3 mg of control vector the sensor chip was performed using 10 mlof2M formic (pcDNA3) were transfected per 2.5 cm dish. Twenty-four acid. The kinetics parameters of the binding reactions were hours (H1299 cells) or 48 h (MCF7 cells) post-transfection determined using BIAevaluation 3.0 software (Biacore AB) cells were ®xed on the plastic plate with 50% acetone-50% using the global ®tting method (Karlsson and Falt, 1997). methanol for 2 min. After ®xation, the cells were washed in The dissociation rate (o€-rate) constant kd and the PBS, then incubated with the mouse anti E-Tag antibody association rate constant (on-rate) ka were determined (1 mg/ml, in DMEM-FCS) for 2 h at RT. After washes simultaneously. The self-consistency of the kinetic data was with PBS, the cells were incubated with a FITC-conjugated checked according to Schuck and Minton (1996). anti-mouse IgG serum (Sigma, St Louis, MO, USA), for For the Hdm2 study, the glutathione S-transferase (GST)- 1 h at RT. After washes with PBS, ¯uorescent labelled cells fusion Hdm2 (aa 1 ± 188) protein was described previously were examined under ¯uorescent microscope. (Bottger et al., 1996). Inhibition of GST-Hdm2 binding to E. To compare the stability of the di€erent scFv constructs in coli human p53 was investigated in the same conditions as H1299 cells, 10 mg of each of the four scFv constructs were above, for 260 nM of GST-Hdm2 protein before and after transfected into a subcon¯uent 10 cm dish culture. Control preincubation of DO-1 scFv (660 nM) or DO-1 antibody was performed by transfection with 10 mg of the pcDNA3 (330 nM). Control experiment was performed by replacing vector. Forty-eight hours after transfection, cell lysates DO-1 scFv or DO-1 antibody with an irrelevant anti- (Harlow and Lane, 1988) were fractionated by 12% SDS ± Troponin I antibody (330 nM) (Sano® Diagnostics Pasteur, PAGE and transferred to a nitrocellulose membrane. Blots Montpellier, France). were stained with the anti-Etag antibody (2 mg/ml) and a HRP-conjugated anti-mouse serum (Dako). Electrophoretic mobility shift assay (EMSA) Human p53 was puri®ed from Sf9 cells as previously Luciferase assay described (Hupp and Lane, 1994). Puri®ed fractions active H1299 cells were seeded at 36105 cells per 6 cm plate 24 h for DNA-binding (Hupp and Lane, 1995) were used in the prior transfection. Cells were then transfected as above following EMSA. Binding conditions are described in detail with a total of 7 mg of plasmid and the calcium phospate elsewhere (Hupp et al., 1992). The DNA binding bu€er precipitate was left for 6 h. Each calcium precipitate contained 20% (v/v) glycerol, 50 mM KCl, 40 mM HEPES, contains 3 mg PG13-PY-luciferase reporter construct, pH 8, 0.05 mM EDTA, 5 mM DTT, 0.1% Triton X-100, 10 ng of pcDNA3-human p53 and 3 mgofpcDNA3- 10 mM MgCl2, 1 mg/ml bovine serum albumin. The speci®c scFvCk or 3 mg of pcDNA3-scFvCk-SV40. Control was p53 consensus sequence (PG) (El Deiry et al., 1992) was performed by replacing the plasmids encoding the scFv end-labelled with g-32P-ATP and used in a 20-fold excess of constructs by 1 mg of hdm2 cloned into the pCMV- supercoiled competitor DNA (pBluescript II SK+, Strata- Neobam vector (Oliner et al., 1992). The pcDNA3 vector gene, Cambridge, UK). Displacement of Hdm2 binding was was used in order to maintain a constant amount of CMV performed by incubating for 30 min GST-Hdm2 (1300 nM) promoter in each transfection mixture. Twenty-four hours and insect cells-produced human p53 (90 nM) and then by post-transfection, cells were washed with PBS and lysates incubating for 30 min with di€erent concentrations of DO-1 made in cell culture lysis bu€er (Luciferase reporter assay scFv. A phosphorimager (Molecular Dynamics) was used system, Promega). After a 15 min incubation, the cell for the quanti®cation of radioactive signals. debris was removed by centrifugation, and luciferase activity was determined in a 20 mlsamplebyadding 100 ml luciferase assay reagent (Luciferase reporter assay DO-1 scFv constructs for eukaryotic expression system, Promega), and measuring the light produced with a To PCR amplify the scFv and scFvCk versions, the luminometer (1450 Microbeta plus counter, Wallac). ExpVHreverse primer (5'-TCC CCG CGG CAC CAT Control experiments were performed by replacing human GCA GGT TAC TCT GAA AGA-3') and the ExpETagfor- p53with10ngofpcDNA3-mousep53andpCMV- ward primer (5'-GCT CTA GAG CTT ATT AGG CAC Neobam-hdm2 with 1 mg of pCOC-mdm2X2 (encoding GCG GTT CCA GCG GAT CCG GAT-3')wereused.For the Mdm2 protein, Haupt et al., 1996). the scFv-SV40 and the scFvCk-SV40 versions, the Ex- The PG13-PY ®re¯y luciferase reporter was kindly pVHreverse primer and the ExpSV40forward primer (5'-G provided by B Volgestein (Kern et al., 1992). The pCMV- CTC TAG AGC GGC CGC TTA CTA AAC CTT ACG Neobam-hdm2 and pCOC-mdm2X2 plasmids were a gift from TTT CTT CTT TGG AGG AGT TCC ACC TCC GGC M Oren, and are both under the control of a CMV promoter. Characterization of a new anti-p53 scFv PA Cohen et al 2455 Note added in proof M Pugniere for Biacore assistance, Dr V Bottger for While this work was under revision, microinjection of DO- search in the Immunogenetics database, Dr A Sparks for 1 antibody in human ®broblasts was shown to block the providing DO-1 Ig antibody, and Dr JC Bourdon for activation of the transcriptional function of p53 (Gire and helpful discussions. PAC was supported by grants from Wynford-Thomas, 1998), con®rming our luciferase assay the ARC (Association pour la Recherche sur le Cancer) data. and the european TMR (Training and Mobility Reseach) program. DPL is a Gibb Fellow of the Cancer Research Acknowledgements Campaign. WeareverygratefultoDrKBall,DrBKuehlandProf B Pau for critical reading of the manuscript. We thank Dr

References

Barak Y, Juven T, Ha€ner R and Oren M. (1993). EMBO J., Haupt Y, Maya R, Kazaz A and Oren M. (1997). Nature, 12, 461 ± 468. 387, 296 ± 299. Beerli RR, Wels W and Hynes NE. (1994). J. Biol. Chem., Hecker D, Page G, Lohrum M, Weiland S and Scheidtmann 269, 23931 ± 23936. KH. (1996). Oncogene, 12, 953 ± 961. Biocca S, Pierandrei-Amaldi P and Cattaneo A. (1993). Hoogenboom HR, Griths AD, Johnson KS, Chiswell DJ, Biochem. Biophys. Res. Commun., 197, 422 ± 427. Hudson P and Winter G. (1991). Nucleic Acids Res., 19, BioccaS,RubertiF,TafaniM,Pierandrei-AmaldiPand 4133 ± 4137. Cattaneo A. (1995). Biotechnology, 13, 110 ± 115. Hupp TR and Lane DP. (1994). Curr. Biol., 4, 865 ± 875. Bottger A, Bottger V, Garcia-Echeverria C, Chene P, Hupp TR and Lane DP. (1995). J. Biol. Chem., 270, 18165 ± Hochkeppel HK, Sampson W, Ang K, Howard SF, 18174. Picksley SM and Lane DP. (1997). J. Mol. Biol., 269, Hupp TR, Meek DW, Midgley CA and Lane DP. (1992). 744 ± 756. Cell, 71, 875 ± 886. Bottger V, Bottger A, Howard SF, Picksley SM, Chene P, Jamal S and Zi€ EB. (1995). Oncogene, 10, 2095 ± 2101. Garcia EC, Hochkeppel HK and Lane DP. (1996). Jannot CB, Beerli RR, Mason S, Gullick WJ and Hynes NE. Oncogene, 13, 2141 ± 2147. (1996). Oncogene, 13, 275 ± 282. Braithwaite AW and Jenkins JR. (1989). J. Virol., 63, 1792 ± Jannot CB and Hynes NE. (1997). Biochem. Biophys. Res. 1799. Commun., 230, 242 ± 246. Buckbinder L, Talbott R, Velasco Miguel S, Takenaka I, Karlsson R and Falt A. (1997). J. Immunol. Methods, 200, Faha B, Seizinger BR and Kley N. (1995). Nature, 377, 121 ± 133. 646 ± 649. Kastan MB, Zhan Q, El-Deiry WS, Carrier F, Jacks T, Cahilly-Snyder L, Yang-Feng T, Francke U and George DL. Walsh WV, Plunkett BS, Vogelstein B and Fornace AJ. (1987). Somat. Cell. Mol. Genet., 13, 235 ± 244. (1992). Cell, 71, 587 ± 597. Candau R, Scolnick DM, Darpino P, Ying CY, Halazonetis Kern SE, Pietenpol JA, Thiagalingam S, Seymour A, Kinzler TD and Berger SL. (1997). Oncogene, 15, 807 ± 816. KW and Vogelstein B. (1992). Science, 256, 827 ± 830. Chen J, Marechal V and Levine AJ. (1993). Mol. Cell Biol., Kubbutat MH, Jones SN and Vousden KH. (1997). Nature, 13, 4107 ± 4114. 387, 299 ± 303. Clarke AR, Purdie CA, Harrison DJ, Morris RG, Bird CC, Kubbutat MH and Vousden KH. (1997). Mol. Cell Biol., 17, Hopper ML and Wyllie AH. (1993). Nature, 362, 849 ± 460 ± 468. 852. Kussie PH, Gorina S, Marechal V, Elenbaas B, Moreau J, Duan L, Zhang H, Oakes JW, Bagasra O and Pomerantz RJ. Levine AJ and Pavletich NP. (1996). Science, 274, 948 ± (1994). Hum. Gene Ther., 5, 1315 ± 1324. 953. El Deiry WS, Kern SE, Pietenpol JA, Kinzler KW and Lees Miller SP, Sakaguchi K, Ullrich SJ, Appella E and Vogelstein B. (1992). Nat. Genet., 1, 45 ± 49. Anderson CW. (1992). Mol. Cell Biol., 12, 5041 ± 5049. El Deiry WS, Tokino T, Velculescu VE, Levy DB, Parsons R, Leonardo AD, Linke SP, Clarkin K and Whal GM. (1994). Trent JM, Lin D, Mercer WE, Kinzler KW and Genes and Dev., 8, 2540 ± 2551. Vogelstein, B. (1993). Cell, 75, 817 ± 825. Lill NL, Grossman SR, Ginsberg D, DeCaprio J and Eliyahu D, Raz A, Gruss P, Givol D and Oren M. (1984). Livingston DM. (1997). Nature, 387, 823 ± 827. Nature, 312, 646 ± 649. Lin J, Chen J, Elenbaas B and Levine AJ. (1994). Genes Dev., Fakharzadeh SS, Trusko SP and George DL. (1991). EMBO 8, 1235 ± 1246. J., 10, 1565 ± 1569. LinYSandGreenMR.(1989).Nature, 340, 656 ± 659. Fields S and Jang SK. (1990). Science, 249, 1046 ± 1049. Lohrum M and Scheidtmann KH. (1996). Oncogene, 13, Finlay CA, Hinds PW and Levine AJ. (1989). Cell, 57, 1083 ± 2527 ± 2539. 1093. Lu H and Levine AJ. (1995). Proc. Natl. Acad. Sci. USA, 92, Fiscella M, Ullrich SJ, Zambrano N, Shields MT, Lin D, 5154 ± 5158. Lees Miller SP, Anderson CW, Mercer WE and Appella E. MaciejewskiJP,WeicholdFF,YoungNS,CaraA,ZellaD, (1993). Oncogene, 8, 1519 ± 1528. Reitz MJ and Gallo RC. (1995). Nat. Med., 1, 667 ± 673. Freund C, Ross A, Guth B, Pluckthun A and Holak TA. Maki CG, Huibregtse JM and Howley PM. (1996). Cancer (1993). Febs Lett., 320, 97 ± 100. Res., 56, 2649 ± 2654. Gire V and Wynford-Thomas D. (1998). Mol. Cell. Biol., 18, Malmqvist M. (1993). Nature, 361, 186 ± 187. 1611 ± 1621. Marasco WA. (1995). Immunotechnology, 1, 1 ± 19. Gu W, Shi XL and Roeder RG. (1997). Nature, 387, 819 ± Marston NJ, Crook T and Vousden KH. (1994). Oncogene, 823. 9, 2707 ± 2716. Hansen S, Hupp TR and Lane DP. (1996). J. Biol. Chem., Mayo LD, Turchi JJ and Berberich SJ. (1997). Cancer Res., 271, 3917 ± 3924. 57, 5013 ± 5016. Harlow E and Lane D. (1988). Antibodies. A Laboratory McCa€erty J, Griths AD, Winter G and Chiswell DJ. Manual. Cold Spring Harbor, New York. (1990). Nature, 348, 552 ± 554. Haupt Y, Barak Y and Oren M. (1996). Embo J., 15, 1596 ± Mhashilkar AM, Bagley J, Chen SY, Szilvay AM, Helland 1606. DG and Marasco WA. (1995). EMBO J., 14, 1542 ± 1551. Characterization of a new anti-p53 scFv PA Cohen et al 2456 Milne DM, Campbell DG, Caudwell FB and Meek DW. Shieh SY, Ikeda M, Taya Y and Prives C. (1997). Cell, 91, (1994). J. Biol. Chem., 269, 9253 ± 9260. 325 ± 334. Milne DM, Palmer RH, Campbell DG and Meek DW. Shvarts A, Steegenga WT, Riteco N, van Laar T, Dekker P, (1992). Oncogene, 7, 1361 ± 1369. Bazuine M, van Ham RCA, van der Houven van Oordt W, Miyashita T and Reed JC. (1995). Cell, 80, 293 ± 299. Hateboer G, van der Eb A and Jochemsen AG. (1996). Momand J, Zambetti GP, Olson DC, George D and Levine EMBO J., 15, 5349 ± 5357. AJ. (1992). Cell, 69, 1237 ± 1245. Stephen CW, Helminen P and Lane DP. (1995). J. Mol. Biol., Okamoto K and Beach D. (1994). EMBO J., 13, 4816 ± 4822. 248, 58 ± 78. Oliner JD, Kinzler KW, Meltzer PS, George DL and Subler MA, Martin DW and Deb S. (1992). J. Virol., 66, Vogelstein B. (1992). Nature, 358, 80 ± 83. 4757 ± 4762. Oliner JD, Pietenpol JA, Thiagalingam S, Gyuris J, Kinzler Thut CJ, Chen JL, Klemm R and Tjian R. (1995). Science, KW and Vogelstein B. (1993). Nature, 362, 857 ± 860. 267, 100 ± 104. Pariat M, Carillo S, Molinari M, Salvat C, Debussche L, Thut CJ, Goodrich JA and Tijan R. (1997). Genes and Dev., Bracco L, Milner J and Piechaczyk M. (1997). Mol. Cell 11, 1974 ± 1986. Biol., 17, 2806 ± 2815. Unger T, Nau MM, Segal S and Minna JD. (1992). EMBO Picksley SM, Vojtesek B, Sparks A and Lane DP. (1994). J., 11, 1383 ± 1390. Oncogene, 9, 2523 ± 2529. Varhavsky A. (1992). Cell, 69, 725 ± 735. Polyak K, Xia Y, Zweler JL, Kinzler KW and Vogelstein B. Vojtesek B, Bartek J, Midgley CA and Lane DP. (1992). J. (1997). Nature, 389, 300 ± 305. Immunol. Methods, 151, 237 ± 244. Raycroft L, Wu HY and Lozano G. (1990). Science, 249, Wang Y and Eckhart W. (1992). Proc.Natl.Acad.Sci.USA, 1049 ± 1051. 89, 4231 ± 4235. Sambrook J, Fritsch EF and Maniatis T. (1989). Molecular Wright M, Grim J, Deshane J, Kim M, Strong TV, Siegal GP cloning: A laboratory manual. Cold Spring Harbor, New and Curiel DT. (1997). Gene Ther., 4, 317 ± 322. York. Yoneda Y, Semba T, Kaneda Y, Nobel RL, Matsuoka Y and Sche€ner M, Werness BA, Huibregtse JM, Levine AJ and Kurihara T. (1992). Exp. Cell Res., 201, 313 ± 320. Howley PM. (1990). Cell, 63, 1129 ± 1136. Schuck P and Minton AP. (1996). TIBS, 21, 458 ± 460. Seto E, Usheva A, Zambetti GP, Momand J, Horikoshi N, Weinmann R, Levine AJ and Shenk T. (1992). Proc. Natl. Acad. Sci. USA, 89, 12028 ± 12032.