Bioconjugate Chem. 2002, 13, 7−15
7
ARTICLES In Vivo Tumor Delivery of a Recombinant Single-Chain Fv::Tumor Necrosis Factor: A Fusion Protein Stephen P. Cooke,* R. Barbara Pedley, R. Boden, Richard H. J. Begent, and Kerry A. Chester CRC Targeting and Imaging Group, Academic Department of Oncology, Royal Free and University College Medical School, University College, London, NW3 2PF, U.K. Received December 20, 2000
Locoregional and intratumoral administration of tumor necrosis factor R (TNFR) has been successful in obtaining inhibition or regression of tumor growth in the clinic. This potent antitumor activity of TNFR has not yet been exploited as a systemic agent in cancer therapy, mainly due to high levels of toxicity to normal tissues before a therapeutic dose of TNFR in the tumor has been achieved. To address this, we have targeted TNFR using antitumor antibodies. We have used a genetic fusion of human recombinant TNFR with MFE-23, a single-chain Fv antibody fragment directed against carcinoembryonic antigen. MFE-23::TNFR fusion protein is isolated in high yields (28 mg/L) from bacterial inclusion bodies and purified to homogeneity by affinity chromatography. It is a 144 kDa trimer in native form and possesses the antigen-binding activity of the sFv and the cytotoxicity to both WEHI 164 and a human adenocarcinoma cell line (LoVo) of rhTNFR. Radiolabeled MFE-23::TNFR binds both human and mouse TNF receptor 1 in vitro and is able to localize effectively in nude (nu/nu) mice bearing human LS174T xenografts; tumor/tissue ratios of 21:1 and 60:1 are achieved 24 and 48 h after intravenous injection. These studies indicate that MFE-23::TNFR will provide an effective means for systemically administered cancer therapy with TNFR.
INTRODUCTION
Tumor necrosis factor R (TNFR) is a pleiotropic cytokine produced by a variety of human cells particularly activated macrophages and monocytes. It has the ability to kill tumor cells by binding to two cell surface receptors, TNFR1 and TNFR2 (p55 and p75), thereby inducing apoptosis by activation of the death pathway (Laster et al., 1988; Itoh and Nagata, 1993). Studies involving regional (Lejeune, 1995; Moritz et al., 1989; Abbruzzese et al., 1989; Mavligit et al., 1992) or intratumoral (Watanabe et al., 1994) injection of TNFR in humans, where high tumor levels of the cytokine can be achieved, have demonstrated tumor regression, especially when administered in combination with interferon-γ. This has been particularly impressive in trials involving isolated limb perfusion (Lejeune, 1995; Abbruzzese et al., 1989) and raises the question of whether these results could be matched using TNFR administered to patients via a simple intravenous injection. The use of TNFR as a systemic agent for cancer therapy is restricted by the short circulatory half-life of the molecule and its severe side effects ranging from headache, chills, and fever to systemic shock and sepsis (Asher et al., 1987; Rosenberg et al., 1988). Thus, normal tissue toxicity is limiting before a therapeutic dose to the tumor can be achieved, and therefore, the potent antitumor activity of TNFR remains unharnessed for systemic therapy. Various targeting strategies have been * To whom correspondence should be addrssed. Tel: +44 020 7794 0500 ext 5495. Fax: +44 020 7794 3341. E-mail: s.cooke@ ucl.ac.uk.
developed to address this with the aim of achieving higher doses of TNFR at the tumor site coupled with a low systemic concentration. Systemic delivery of TNFR by liposomes (Manusama et al., 1998) or dextran and metal coordination to improve tumor penetration and retention of TNFR (Tabata et al., 1999) have been explored, but more specific targeting can be achieved using tumor-selective antibodies. Indirect approaches using bispecific (antitumor/anti-TNFR) antibodies to capture TNFR at the tumor site (Robert et al., 1996) or avidin-antibody complexes to target biotinylated TNFR (Moro et al., 1997; Gasparri et al., 1999), have met with some success, but these systems are complex. More straightforward approaches using antibodies directly linked to TNFR are attractive and have shown efficacy in animal models using chemical conjugates of whole antibody coupled to TNFR (Gillies et al., 1993; Pietersz et al., 1998). However, chemical conjugates have demonstrated poor tumor penetration (possibly due in part to their large size), which results in low amounts of TNFR localizing in the tumor tissue. Attempts to optimize the direct antibody targeting approach by using recombinant fusion proteins of TNFR with single-chain Fv antibody fragments (sFvs) have shown encouraging results such as specific cytotoxicity in vitro (Hoogenboom et al., 1991; Yang et al., 1995) and reduction in tumor volume in mice bearing LeY+ xenografts (Scherf et al., 1996). However, these promising therapies were hampered by poor yield of fusion protein, preventing further study and development of a clinically useful product. We describe the production, purification, characterization, and tumor targeting of MFE-23::TNFR fusion protein, which overcomes this obstacle.
10.1021/bc000178a CCC: $22.00 © 2002 American Chemical Society Published on Web 12/22/2001
8 Bioconjugate Chem., Vol. 13, No. 1, 2002
Figure 1. Schematic of MFE-23::TNFR/pET21(d). MFE-23:: TNFR was inserted into pET21(d) via Nco1 and Not1 restriction sites. The flexible linker between the sFv and the TNFR is also shown.
MFE-23 is a filamentous bacteriophage-derived sFv (Chester et al., 1994) that has been shown to localize safely and effectively to carcinoembryonic antigen (CEA) expressing tumors in clinical trials (Begent et al., 1996; Mayer et al., 2000). CEA was chosen as a target because, with highly specific antibodies, it is only detectable on tumors and on the luminal surface of the gut, which is not accessible to IgG antibodies. CEA is expressed on most gastrointestinal carcinomas and on a proportion of breast, lung, and ovarian carcinomas. MFE-23 is readily expressed in bacteria and was selected to bind with high affinity (KD ) 2 × 10-9 M) for CEA (20). The MFE::TNFR fusion protein aims to reduce sequestration and increase tumor concentrations of systemically administered TNFR. EXPERIMENTAL PROCEDURES
Subcloning of MFE-23::TNFr Plasmid. The expression vector pUC119SNpolymyc containing the MFE23:: TNFR fusion protein (a kind gift from Philipp Holliger, LMB, Cambridge, UK) was purified from the Escherichia coli strain TG1 using standard methods. The MFE-23:: TNFR gene was released from the plasmid by digestion with restriction endonucleases Nco1 and Not1 (5 units/ µg DNA; 4 h; 37 °C) and ligated into the expression vector pET21(d) prior to electroporation into E. coli strain BL21(DE3)pLysS. A schematic of the resulting expression vector pET21(d) MFE-23::TNFR is shown in Figure 1. Expression in Inclusion Bodies. Five individual colonies of BL21 (DE3 pLysS) containing the MFE-23:: TNFR gene and controls containing the “empty” pET 21(d) vector alone were selected and grown in 50 mL culture broth (2TY + 0.1% glucose + 100 mg/mL ampicillin). When an OD600 of 0.9 was reached, the cultures were induced with 2 mM isopropylthiogalactoside (IPTG) for 16 h at 30 °C. Cultures were then centrifuged at 4000 rpm for 15 min, and the pellet was incubated in lysis buffer (0.1 M Tris, 20 mM EDTA pH8.0 containing 200 mg/mL lysozyme) for 1 h at room temperature followed by the addition of NaCl to 0.5 M and Triton-X 100 to a final concentration of 2%. After the pellet was incubated a further 3 h, tubes were centrifuged at 25000g for 1 h to isolate the inclusion body fraction. Inclusion bodies were washed three times in wash buffer (0.1 M Tris, 20 mM EDTA pH 8.0) before freezing at -70 °C overnight. The inclusion body pellet was resuspended in a second lysis buffer (0.1 M tris pH 8.0, 2 mM EDTA, 300 mM DTE, 8 M urea) and incubated overnight at room temperature. Tubes were centrifuged at 30000g, and the
Cooke et al.
supernatant was recovered. Aliquots were analyzed for protein production by SDS-PAGE (stained with Coomassie Blue). The supernatants containing protein of the expected molecular weight were pooled and dialyzed extensively against 8 M urea to remove all DTE prior to refolding. Refolding and Purification by Affinity Chromatography. The refolding protocol was based on that of Scherf et al. (1996). Single aliquots (5-10 mL) of supernatant were added every 30 min to 1 L of refolding buffer (0.1 M Tris pH 8.0, 2 mM EDTA, 1 mM reduced glutathione (GSSG), 1 mM oxidized glutathione (GSG), 0.5 M L-arginine) and allowed to refold at 4 °C for 96 h with constant agitation. The resulting protein solution (1200 mL) was filtered to remove aggregates before passage through an CNBr-activated Sepharose-CEA affinity column. The solution was circulated though the column via a peristaltic pump, and the column was washed with 50 mL of PBS before bound protein was eluted by the addition of 5 mL aliquots of 50 mM diethylamine. Optical density measurements were taken (OD280 nm), and fractions containing protein were pooled and dialyzed into PBS before being sterile filtered, aliquoted, and stored at -70 °C. This affinity chromatography procedure was repeated three times. SDS-PAGE and Western Blot Analysis. For the initial analysis of the crude supernatants and controls obtained after inclusion body lysis, 20 µL samples were loaded into the wells of a 10% polyacrylamide gel under reducing conditions alongside molecular markers covering a range of molecular weights from 250 to 4 kDa. After electrophoresis, gels were stained with Coomassie Blue and destained in water/acetic acid/methanol until bands were visible. After refolding of the fusion protein from the inclusion body supernatant and purification by affinity chromatography, 20 µL samples were loaded into the wells of a 10% polyacrylamide gel under reducing conditions. After electrophoresis, the gels were cut in half and proteins transferred to PVDF membranes. After being blocked in 5% Marvel, blots were incubated with either rabbit antiMFE or rabbit anti-human TNFR polyclonal antibodies diluted 1:500 in 1% Marvel in PBS/0.1% Tween-20 for 1 h. After being washed with PBS/0.1% Tween-20, blots were incubated with goat anti-rabbit IgG conjugated to horseradish peroxidase (HRP) diluted 1:2000 for 1 h prior to visualization with diaminobenzidine/H2O2. Assessment of CEA Binding Activity. ELISA plates were coated with CEA at 2 µg/mL and blocked for 1 h with 5% Marvel in PBS/0.1% Tween-20. Fusion protein was serially diluted in PBS/0.1% Tween-20 and 100 mL added to the wells in duplicate. Plates were washed, and 100 mL rabbit anti-MFE or rabbit anti-human TNFR polyclonal antibodies were added at 1:500 for 1 h followed by 100 µL of goat anti-rabbit IgG conjugated to HRP at 1:2000 for a further 1 h. Plates were developed by addition of 100 µL of o-phenylenediamine (OPD)/H2O2 in citrate buffer pH 5.0. The reaction was stopped by the addition of 50 µL of 4 M HCl and plates read at 490 nm on an ELISA plate reader. Native Molecular Weight Determination. One hundred micrograms of purified fusion protein was loaded onto a Superose 12 size-exclusion column and analyzed by fast-performance liquid chromatography. One milliliter fractions were collected, and absorbance at OD280 nm was used as the readout. The column was first calibrated by addition of 100 µg each of protein of known molecular weight (MFE-23; 28 kDa, IgG; 150 kDa, b-amylase; 205 kDa, apoferritin; 440 kDa and thyroglobu-
Tumor Delivery of a sFv/rhTNFR Fusion Protein
lin; 690 kDa) and a calibration curve of molecular weight versus elution volume constructed. The fraction number of the eluted peak was determined relative to these standards and the molecular weight estimated from the calibration curve. Cytotoxicity Assay for TNFr Activity. The TNFRsensitive cell line WEHI 164 was used in this assay. Cells were plated at 2 × 105 cells/well in 96-well plates and incubated overnight at 37 °C/5%CO2. Cells were treated with actinomycin D at 10 µg/mL before addition of fusion protein or rhTNFR controls in tripling dilutions (adjusted to yield from 1000 pg/mL to 4 pg/mL w/v TNFR), in triplicate in 200 µL culture medium. sFv MFE-23 alone was used as control (also at 1000 pg/mL to 4 pg/mL). Culture medium alone was used as a negative control. Plates were incubated overnight before addition of 10 µL of a 5 mg/mL solution of MTT [3-(4,5-Dimethylthiazol2-yl)-2,5-diphenyltetrazolium bromide]/well. After a further 4 h incubation at 37 °C, cells were lyzed by addition of 100 µL of 10% SDS in 0.01 N HCl. Color development was allowed to proceed overnight at 37 °C, and plates were read at 570 nm on an ELISA plate reader. The reduction in optical density is a measure of decreased cell viability. Results were expressed in terms of percentage cytotoxicity using the formula [(ODmax - ODsample)/ (ODmax -ODmin)] × 100, where ODmax is the optical density of wells containing culture medium alone and ODmin is the optical density at which no viable cells were visible by phase-contrast microscopy. The differences between sets of triplicate wells were tested for significance using a nonpaired, single tail students t-test, and significance in the text refers to a p value of
