Abstract
Purpose: Our aim was to isolate a novel human mini-antibody(scFv) that specifically targets ErbB2-positive cancer cells. ErbB2, a tyrosinekinase receptor, is overexpressed in clinically significant tumors, such as breast, ovary, and lung carcinomas. In normal tissues, it is expressed only in certain epithelial cell types.
Experimental Design: A large phagemid library (Griffin.1 library) of human scFv was used for the isolation of the ErbB2-specific scFv. A very effective strategy was developed for the isolation, consisting in a double subtractive selection, the use of two different combinations of “positive,” i.e., ErbB2-bearing, and “negative” cell lines.
Results: Here we report the isolation of the first human anti-ErbB2 mini-antibody endowed with antitumor action. Both in its soluble and phage format, it binds specifically to ErbB2, inhibits its autophosphorylation, is internalized by target cells, and exerts a strong and specific antiproliferative action on ErbB2-positive target cells. A correlation was found between the extent of this antiproliferative effect and the expression levels of ErbB2 on target cells, with a strong cytotoxicity for hyper-expressing cells, such as SKBR3, in which apoptosis was evidenced.
Conclusions: This scFv is a potentially effective immunoreagent for diagnostics and therapeutics of certain cancers, both as a readily diffused molecule in solid tumors and as an essential asset for the construction of fully human anticancer drugs.
Introduction
The ErbB2 transmembrane tyrosine kinase receptor, homologous to the EGF4 receptor (1 , 2) , is highly expressed in breast, ovary, and lung carcinomas (3 , 4) , as well as in salivary gland and gastric tumor-derived cell lines (5 , 6) . Its overexpression, which occurs most commonly via gene amplification, can reach as many as 2 × 106 molecules/cell. In normal tissues it is expressed only in certain epithelial cell types (7) . ErbB2 plays a central role in tumor progression, because it potentiates and prolongs the signal transduction cascades elicited by ligand activation of other ErbB tyrosine kinase receptors (8 , 9) . Overexpression of ErbB2 may also increase resistance of tumor cells to host defenses through the evasion of immune surveillance exerted by activated macrophages (10) . The accessibility of ErbB2 on cell surface and its implication in the development of malignancy of these tumors make it an attractive target for immunotherapy.
Several groups have isolated mouse and rat mAbs against ErbB2 extracellular domain (11, 12, 13) , which display useful properties for immunotherapy. For instance, these reagents can induce endocytosis, and some of them have antiproliferative activity on tumor cells (11, 12, 13, 14, 15) . However, as a consequence of their nonhuman origin, the use of these mAbs as immunotherapeutic drugs is limited. A clear progress in this area of research consisted of the development of the antibody humanization technology with the production of humanized versions of rodent antibodies (16) with reduced immunogenic potential. A humanized version of an anti-ErbB2 murine antibody (Herceptin) is currently in use for treatment of advanced breast cancer (17) .
Recently, fully human scFvs have been generated with the phage display technology through the expression of large repertoires of antibody variable regions on filamentous phages, after their fusion to a phage coat protein (18 , 19) .
Human scFvs specific for ErbB2 have been produced, using for their selection the isolated recombinant extracellular domain (20 , 21) or more recently breast tumor cells (22) . Given their high affinity for the receptor, these immunoreagents may be considered precious tools as delivery vehicles for specifically directing cytotoxic agents to antigen-bearing tumor cells. However, none of these exhibit antitumor activity.
Herein we describe the successful isolation from a large human scFv phagemid library (19) of a novel scFv directed against the ErbB2 receptor. We used a very effective parallel selection procedure based on intact cells either expressing high levels of the receptor antigen or virtually lacking the antigen. Recently, similar strategies have been successfully used by other authors (23 , 24) . The human anti-ErbB2 single chain antibody was found to be endowed with biological properties not described for other anti-ErbB2 scFvs isolated thus far. This novel scFv is internalized upon specific antigen recognition by ErbB2-expressing target cells; it strongly inhibits receptor autophosphorylation and displays a strong inhibitory activity on the growth of ErbB2-positive cell lines. In addition, a clear cytotoxic effect was evidenced toward ErbB2 hyper-expressing SKBR3 cells in which apoptotic death is induced. Interestingly, these features are present both in soluble scFv and in its phage format.
Materials and Methods
Cell Cultures.
The SKBR3 human breast tumor cell line and the A431 human epidermoid carcinoma cell line (kindly provided by Menarini Research, Pomezia, Italy) were cultured in RPMI 1640 (Life Technologies, Inc., Paisley, United Kingdom). The BT-474 and MDA-MB453 human breast tumor cell lines (a kind gift of H. C. Hurst, Imperial Cancer Research Fund, London, United Kingdom), the SK-OV-3 human ovarian cell line (a kind gift of I. McNeish, Imperial Cancer Research Fund, London, United Kingdom), and the NIH-3T3 murine fibroblasts (American Type Culture Collection, Rockville, MD) were grown in DMEM (Life Technologies, Inc.). The NIH-3T3 fibroblast cell line transfected with human ErbB2, kindly provided by N. E. Hynes, Friedrich Miescher Institute, Basel, Switzerland (25) , were cultured in DMEM containing 1 mg/ml G418 (Life Technologies, Inc.). Medium were supplemented with 10% FCS, 50 units/ml penicillin, and 50 μg/ml streptomycin (all from Life Technologies, Inc.).
Antibodies.
The following antibodies were used in the current study: murine anti-M13 mAb (Amersham Pharmacia Biotech, Little Chalfont, United Kingdom); murine mAb 9E10 directed against the myc tag protein (26) ; murine anti-His tag mAb (Qiagen, West Sussex, United Kingdom); murine anti-ErbB2 MgR6 mAb (gift from Menarini Research; Ref. 27 ); FITC-conjugated rabbit antimouse immunoglobulin antibody, and horseradish peroxidase-conjugated rabbit antimouse immunoglobulins (both from Dako, Cambridgeshire, United Kingdom); murine anti-phosphotyrosine mAb P-Tyr (PY99; Santa Cruz Biotechnology Inc., Santa Cruz, CA); and murine anti-actin mAb (Sigma, St. Louis, MO).
Anti-NIP (28) and anti-thyroglobulin scFv were kindly provided by Dr. G. Winter, Center for Protein Engineering, Cambridge, United Kingdom; gp200-MR6 scFv was isolated as described (29) .
Selections of scFv-Phage on Live Cells.
ErbB2-positive cells were labeled as follows. The human breast tumor SKBR3 cell line, naturally expressing high levels of ErbB2, and the NIH-3T3 fibroblasts, transfected with human ErbB2, grown in 250-ml flasks (Becton Dickinson, Oxford, United Kingdom) to 70–80% confluency, were detached with the cell dissociation solution (Sigma) and washed twice with PBS. Cells were then resuspended in 1 ml of prewarmed PBS containing 15 μm 5(6) -CFDA, SE (5-(and-6)-carboxyfluorescein diacetate, succinimidyl ester mixed isomers; Molecular Probes, Eugene, OR), and incubated for 30 min at 37°C. After three washes with cold PBS, cells were resuspended (1 × 106 cells/ml), and the level of fluorescence analyzed by flow cytometry before each round of phage selection.
Phagemid particles were rescued with M13-K07 from the Griffin library, as described previously (18) . For each round of panning, phages (1013 cfu) were blocked with 5% milk powder (Marvel) in PBS for 15 min. The blocked phages were incubated for 16 h at 4°C with labeled positive cells (1 × 106) in the presence of unlabeled negative cells (9 × 106) by gently rotating in a final volume of 5 ml containing 2% Marvel. Cells were then pelleted by centrifugation at 600 × g for 5 min at 4°C and washed twice in 50 ml of PBS. The positive-labeled cells were sorted by FACS. To elute phages from positive cells, these were incubated with 0.5 ml PBS containing 50 mm citric acid (pH 2.5) for 5 min, and then neutralized with 0.4 ml of 1 m Tris-HCl (pH 7.4). The recovered phages were amplified by infecting Escherichia coli TG1 cells to prepare phage for the next round of selection. Phage screening was carried out by cell ELISA assays as described previously (30) .
Analysis of Clone Diversity.
To determine the number of individual selected clones, DNA fingerprinting analysis was performed with the restriction endonuclease BstNI or BsaJI (New England Biolabs, Hertfordshire, United Kingdom) as described (18) . DNA encoding the variable region of positive clones was amplified by PCR from the pHEN2 plasmid, using as primers 5′-CAGTCTATGCGGCCCCATTCA-3′ (complementary to the sequence located between gene III and the c-myc peptide tag) and 5′-ATGAAATACCTATTGCCTACG-3′ (pel B leader sequence). Reactions were performed with Taq DNA Polymerase (Promega, Southampton, United Kingdom) in a volume of 20 μl for 30 cycles under the following conditions: 30 s denaturing at 94°C, 30 s annealing at 55°C, and 1 min extension at 72°C. The amplified products, analyzed by electrophoresis on 1% agarose gel, were used for DNA fingerprinting and sequence analyses. The nucleotide sequences encoding scFv were determined using the ABI automated sequencer (Perkin-Elmer, Warrington, United Kingdom) and were analyzed with the V-BASE sequence alignment program.5
Cell Lysis, Immunoprecipitation, and Western Blot Analyses.
SKBR3 cell lysates were prepared by resuspending in 0.5 ml of lysis buffer [10 mm Tris-HCl (pH 7.4), 150 mm NaCl, and 0.5% NP40 containing Complete proteases inhibitor; Boehringer Mannheim, Germany] ∼7.5 × 106 cells, detached previously with the cell dissociation solution (Sigma), and washed three times in PBS. After 20 min at 0°C the extracts were clarified by centrifugation at 12,000 rpm for 10 min.
ErbB2 immunoprecipitation was carried out by incubating the anti-ErbB2 MgR6 mAb with the cell lysates for 3 h at 4°C. The immune complex was then collected by adsorption to protein G-Sepharose (Sigma) for 1 h at 4°C. After four washes, the proteins, released by boiling in loading buffer (31) , were fractionated by 7.5% SDS-PAGE and electroblotted onto polyvinylidene difluoride membranes (Millipore Corporation, Bedford, MA). The ErbB2 protein was detected using either anti-ErbB2 mAbs or scFv-phage preparations, as described previously (29) .
Preparation of Monoclonal Phage Antibodies for Functional Assays.
scFv carrying phages were prepared from individual ampicillin-resistant colonies grown in 100 ml of 2xTY medium, purified by polyethylene glycol precipitation (32) and washed with 20 ml of sterile water. After an additional polyethylene glycol precipitation step, phages were resuspended in PBS, centrifuged at 12,000 rpm for 15 min, and stored at 4°C until later use.
Soluble scFv Expression and Purification.
Cultures of E. coli SF110 (33) , infected previously with anti-ErbB2 scFv-phage or with anti-NIP scFv-phage, were grown at 37°C in 2xTY medium containing 100 μg/ml ampicillin and 1% glucose, until an absorbency of 1 at 600 nm was reached. Cells were centrifuged at 6,000 rpm for 15 min and resuspended in glucose-free medium. The expression of soluble scFv was induced by the addition of isopropyl-1-thio-β-d-galactopyranoside (Alexis, Nottingham, United Kingdom) to a final concentration of 1 mm in the cell culture, which was then grown at room temperature overnight. Cells were harvested by centrifugation at 6,000 rpm for 15 min, and a periplasmic extract was obtained by resuspending cells in ice-cold 50 mm Tris-HCl (pH 7.4), 1 mm EDTA, and 20% sucrose. After an incubation of 1 h on ice, the periplasmic extract, obtained by centrifugation at 12,000 rpm for 30 min at 4°C, was dialyzed in PBS. Alternatively, soluble periplasmic proteins were isolated using the B-PER buffer (Pierce, Rockford, IL) according to the manufacturer’s recommendations.
Soluble scFv was purified on immobilized-metal affinity chromatography, by incubating the periplasmic extract with Ni-NTA agarose (Qiagen) overnight at 4°C. After extensive washes with PBS containing 20 mm imidazole, the protein was eluted in 50 mm NaH2PO4 (pH 8.0) containing 0.3 m NaCl and 250 mm imidazole. Additional purification was achieved by gel-filtration on a Superdex 75 Hi-Load 10/30 column (fast protein liquid chromatography; Pharmacia Biotech, Uppsala, Sweden) equilibrated in PBS containing 0.16 m NaCl, carried out at a flow rate of 0.3 ml/min. The purity of the final preparation was evaluated by SDS-PAGE. Protein bands were detected by Coomassie staining. Purified scFv, analyzed by Western blotting, was detected using either mAb 9E10 or anti-His tag mAb, followed by rabbit antimouse horseradish peroxidase-conjugated mAb, as described previously (29) .
Determination of Tyrosine Phosphorylation.
SKBR3 cells were grown for 24 h at 37°C in serum-free RPMI medium, then treated with EGF (Collaborative Research Inc., Waltham, MA) or soluble anti-ErbB2 scFv at a concentration of 100 ng/ml and 12 μg/ml, respectively, in fresh, serum-free medium. At the indicated times, cells were washed with PBS, harvested, and lysed in the presence of 1 mm sodium orthovanadate (Sigma), as described above. Immunoblotting analyses were performed with an antiphosphotyrosine mAb. The signal intensity of reactive bands was quantitated with a phosphorimager (GS-710; Bio-Rad, Hercules, CA).
Internalization of Phage Antibodies and Native scFv.
Cells grown on coverslips to 60% confluency were incubated with either phage scFv (1011 cfu/ml) or soluble scFv (20 μg/ml) for 2 or 16 h at 37°C. Cells were then washed, fixed, and permeabilized as described elsewhere (34) . Intracellular phage scFv or soluble scFv were detected with either anti-M13 mAb or mAb 9E10, respectively, followed by FITC-conjugated rabbit antimouse immunoglobulin. Optical confocal sections were taken using a confocal microscope (Zeiss, Axiovert S100TV).
Flow Cytometry.
Approximately 1 × 106 cells were incubated with 100 μl of phage particles (1012 cfu/ml), which were blocked previously with 25 μl of 10% Marvel/PBS (29) . Bound phage particles were detected using murine anti-M13 mAb, followed by FITC-conjugated rabbit antimouse immunoglobulin (Dako). The anti-ErbB2 mAb MgR6 was used at saturating concentration in PBS containing 2% normal human serum and detected using the FITC-conjugated rabbit antimouse immunoglobulin (Dako). Controls included cells incubated with the appropriate isotype-matched antibodies. For Annexin V immunofluorescence, cells were resuspended in binding buffer [10 mm HEPES (pH 7.4), 140 mm NaCl, and 2.5 mm CaCl2] and then stained with Annexin V-FITC and 7-amino-actinomycin D according to the manufacturer’s recommendations (PharMingen, Oxford, United Kingdom). Labeled cells were analyzed using the FACS Calibur flow cytometer (Becton Dickinson); the data were processed using CellQuest software (Becton Dickinson).
Cell Growth Inhibition Assays.
Cells were seeded in 96-well plates; SKBR3, BT-474, and MDA-MB453 cells at a density of 1.5 × 104/well in 150 μl; A431, NIH-3T3, and NIH-3T3 cells transfected with human ErbB2 at a density of 5 × 103/well. Phage scFv (1010-1011 cfu/ml) or soluble purified scFv (1–32 μg/ml) was added, and at suitable time intervals surviving cells were counted. Cell counts were determined in triplicate by trypan blue exclusion. In parallel experiments cells were pulsed for 8 h with [3H]thymidine (Amersham-Pharmacia Biotech, Little Chalfont, United Kingdom) before harvest, and the incorporated radioactivity was measured.
To test apoptotic death, SKBR3 cells were seeded in six-well plates at a density of 3 × 105/well in the absence or in the presence of the anti-ErbB2 scFv, either expressed as phage scFv (1011 cfu/ml) or as soluble scFv (15 μg/ml). The irrelevant anti-NIP scFv was tested in its phage and soluble format as a control. After 24 h, cells were harvested, washed in PBS, and treated with Annexin V as described above. The apoptotic inducer puromycin (10 μg/ml) was used as a positive control.
Results
Parallel Selection on Different Cell Lines to Isolate a Human, ErbB2-specific scFv.
The strategy devised for the isolation of an anti-ErbB2 scFv from the Griffin.1 library consisted in a double selection with the use of two different combinations of positive, i.e., antigen-bearing, and negative cell lines. In the first combination, cells from a human breast carcinoma (SKBR3 cell line), naturally expressing high levels of ErbB2 receptor, were used as antigen-positive cells, and cells from a human epidermoid carcinoma (A431 cell line), expressing the receptor at minimal levels, as antigen-negative cells. In the second combination, NIH 3T3 cells transfected with human ErbB2 (25) and untransfected NIH 3T3 were used as antigen-positive and antigen-negative cells, respectively. The strategy of using two combinations of positive and negative cell lines was devised to guarantee an efficient selection of the anti-ErbB2 clones through successive selection rounds.
In each selection round, a mixture of positive cells (∼10%), labeled previously with a fluorochrome (see “Materials and Methods”), and unlabeled negative cells (90%) were incubated with the antibody phage display library (1013 cfu/selection). The negative cells were used to deplete the library of phage antibodies that bound to common antigens. After 16 h of incubation at 4°C, cells were washed, and the labeled ones (∼1 × 106 cells) were isolated by FACS. Phages bound to the cell surface were displaced and used to infect E. coli TG1. Initially, two rounds of selection were performed with SKBR3 and A431 cell lines. Positive phages, which selectively bound to the ErbB2-positive cell line, were submitted to two additional rounds of selection, using either the same cell combination (strategy 1) or the NIH 3T3 transfected and untransfected cell lines (strategy 2). Strategy 2 was implemented to verify the possibility of recruiting phages with higher binding affinity by using the NIH 3T3-transfected cells that express lower levels of ErbB2.
From the last selection round of strategies 1 and 2, the titer of phage recovered after elution increased ∼10-fold compared with the previous round. Approximately 25% of the clones from either selection procedures were identified as ErbB2-positive clones using cell ELISA. On the basis of their DNA fingerprinting analyses, one distinct restriction pattern was observed, shared by clones isolated with both selection strategies.
One of these clones, clone A7, was selected for additional analyses. Sequence analysis of clone A7 DNA revealed that the scFv VH belongs to the VH5 family (derived from the VH germ-line gene DP-73), whereas the VL belongs to the VL8 family (derived from the VL germ-line gene DPL-21). The scFv displayed on phages from clone A7 was named ph-A7.
Characterization of Phage-Antibody A7 Specificity.
To determine whether ph-A7 was specific for ErbB2, we tested it by Western blot analyses on SKBR3 cell lysates. As shown in Fig. 1⇓ , ph-A7 recognized a protein of ∼Mr 185,000, the molecular weight expected for the ErbB2 antigen. In the same experiment a protein corresponding to the same molecular size was recognized by the murine anti-ErbB2 MgR6 mAb (27) . No bands were detected when an antithyroglobulin scFv-phage preparation was used as a negative control (see Fig. 1⇓ ). More conclusively, ph-A7 was able to recognize a single protein of the molecular weight expected for ErbB2 when this was isolated previously from SKBR3 cell extracts by immunoprecipitation with the anti-ErbB2 MgR6 mAb.
Western blot analysis of cell extracts prepared from SKBR3 cells. Extracts were probed with: ph-A7 (scFv displayed on phages from clone A7; Lane 1); anti-ErbB2 MgR6 mAb (Lane 2); anti-thyroglobulin scFv displayed on phage (Lane 3). In Lane 4, a cell extract from SKBR3 cells, previously immunoprecipitated with anti-ErbB2 MgR6 mAb, was probed with ph-A7.
The specificity of ph-A7 was additionally investigated by flow cytometry using a panel of cell lines, which was not used in the selection procedure and expresses different levels of ErbB2 (Fig. 2⇓ ; Table 1⇓ ). The phage ph-A7 gave strong labeling of MDA-MB453 and BT-474 cells from breast carcinomas, and of SK-OV-3 cells from ovarian adenocarcinoma, all expressing high levels of ErbB2. On the contrary, no fluorescence was detected when the same cells were incubated with an irrelevant anti-NIP scFv-phage (see Fig. 2⇓ ). More conclusively, ph-A7 failed to stain A431 cells, which express ErbB2 at low levels (35 , 36) , and NIH 3T3 fibroblasts, but did stain ErbB2-transfected NIH 3T3 fibroblasts. Furthermore, an identical pattern of reactivity was observed when these cell lines were tested with the anti-ErbB2 MgR6 mAb. These results, together with those from the Western blotting analyses, clearly demonstrate that ph-A7 specifically recognizes the extracellular domain of ErbB2.
Flow cytometric analysis of ph-A7 binding to ErbB2 expressing cell lines. The following cell lines were tested: MDA-MB453 (A and B); BT-474 (C and D); SK-OV-3 (E and F); and A431 (G and H). Cells were probed with ph-A7 (A, C, E, and G, shaded peaks), or with a control anti-NIP scFv displayed on phages (unshaded peaks). In B, D, F, and H cells were probed with anti-ErbB2 MgR6 mAb (shaded peaks) or with OKT3, a control, unrelated anti-CD3 mAb (unshaded peaks).
Flow cytometric analyses of ph-A7 binding to a panel of cell lines expressing different levels of the ErbB2 receptor
Binding was evaluated as mean fluorescence intensity (MFI). Receptor expression levels were detected with murine MgR6 mAb as a specific anti-ErbB2 antibody. The effects of ph-A7 on cell proliferation after 96 h of treatment, expressed as IC50 values, are tabulated as means of at least three independent determinations.
Expression and Characterization of Soluble scFv.
To prepare human anti-ErbB2 scFv as a soluble molecule, the pHEN2 phagemid vector (a derivative of the pHEN1 vector; Ref. 28 ), containing the DNA encoding A7 scFv, was expressed in the bacterial strain SF110. After induction with isopropyl-1-thio-β-d-galactopyranoside, a periplasmic extract was prepared as described previously (37) .
To verify whether soluble anti-ErbB2 scFv retained the binding properties of the scFv displayed on phages, the periplasmic extract was analyzed by ELISA as well as by flow cytometry using the cell lines tested previously with the anti-ErbB2 scFv in its phage format. The results from both analyses showed that the anti-ErbB2 soluble scFv selectively binds to the antigen-bearing cells (data not shown). The scFv immunoreagent in its soluble format was named s-A7.
Because the DNA encoding the scFv was cloned into the pHEN2 vector fused to a COOH-terminal hexahistidine sequence, recombinant s-A7 was purified by immobilized-metal affinity chromatography by using Ni-NTA agarose followed by fast protein gel-filtration chromatography. In the gel-filtration pattern, s-A7 eluted as a protein of ∼ Mr 27,000 (data not shown). When analyzed by SDS-PAGE electrophoresis (Fig. 3)⇓ , s-A7 migrated as a single band of the expected molecular size (∼Mr 27,000). These results indicate that s-A7 is monomeric, hence that it bears a single antigen-binding site.
SDS-gel electrophoresis and Western blot analyses of s-A7. Lanes 1 and 2, Coomassie staining of s-A7 and control anti-NIP soluble scFv, respectively. In Lanes 3-6 Western blot analyses are shown of s-A7 (Lanes 3 and 5) and anti-NIP scFv (Lanes 4 and 6) probed with anti-myc 9E10 mAb (Lanes 3 and 4) or with anti-His mAb (Lanes 5 and 6).
Purified s-A7 was also analyzed by Western blotting either with an anti-His tag mAb or with anti-myc tag 9E10 mAb (see “Materials and Methods”). By both analyses a band of the expected size, ∼ Mr 27,000, was visualized (Fig. 3)⇓ .
Internalization of A7 by SKBR3 Cells.
Several studies have shown that monoclonals specific to ErbB2 can induce receptor-mediated endocytosis and inhibit tumor cells growth (11, 12, 13, 14, 15) . This could occur by removing ErbB2 from the cell surface, thereby preventing activation of the receptor. We tested the anti-ErbB2 scFv, both in the phage and in the soluble format, for its ability to undergo receptor-mediated endocytosis in SKBR3 cells. To test the immunoreagent in the phage format, cells grown on coverslips were incubated with ph-A7 (1011 cfu/ml) for 16 h at 37°C. Cells were then extensively washed with PBS to remove nonspecific binding followed by four washes with a high salt and low pH, stripping glycine buffer to remove phages specifically bound to the cell surface (34) . Cells were then fixed and permeabilized, and internalized phages were visualized with an anti-M13 mAb, followed by a rabbit antimouse FITC-conjugated antibody. As a control, an anti-NIP scFv-phage preparation (1012 cfu/ml) was used.
By confocal microscopy a strong intracellular staining was observed for ph-A7, whereas no staining was detected with the anti-NIP scFv-phage (see Fig. 4, A and C⇓ ). To determine whether infectious antibody-equipped phage particles could be recovered from within the cells, the experiment was repeated on cells grown in six-well plates, then incubated with the antibody carrying phages for 2 h at 37°C. After the last stripping wash, cells were dissociated from the culture plates by trypsinization, washed three times with PBS, and then lysed with 100 mm triethylamine. Phage particles, recovered in the cell lysates, were titrated by infection of E. coli TG1 strain, as described previously (34) . The titer of ph-A7 in the triethylamine fraction was much higher (at least 1 order of magnitude) than that obtained using an antithyroglobulin scFv-phage as a control (data not shown).
Internalization of ph-A7 and s-A7 in SKBR3 cells as visualized by confocal microscopy. Cells were incubated for 16 h with ph-A7 (A) or with s-A7 (B). Anti-NIP scFv displayed on phages (C) or soluble anti-NIP scFv (D) were used in parallel as controls. Magnification 1:1000.
Identical results were obtained with the scFv in the soluble format. When s-A7 was incubated with SKBR3 cells for 16 h at 37°C, a strong intracellular staining was visualized (Fig. 4B)⇓ by confocal microscopy using anti-myc tag 9E10 mAb followed by rabbit antimouse FITC-conjugated antibody. No staining was detected on incubation for the same time period with an irrelevant anti-NIP soluble scFv (Fig. 4D)⇓ .
Effects on Tumor Cell Proliferation.
To determine whether the anti-ErbB2 scFv could affect tumor cell growth, cells were plated in the presence or in the absence of increasing concentrations of ph-A7. After a suitable time interval the extent of cellular proliferation was measured by cell counts (Fig. 5A)⇓ or DNA synthesis (Fig. 5B)⇓ . When tested on ErbB2-positive cells, such as SKBR3, MDA-MB453, BT-474, and SK-OV-3 cell lines, ph-A7 was found to severely inhibit their proliferation, with a dose-dependent inhibitory effect (Fig. 5A)⇓ . In contrast, when SKBR3 cells (Fig. 5B)⇓ or the other mentioned ErbB2-positive cell lines were incubated with phage preparations either lacking the scFv moiety or displaying an irrelevant scFv, such as anti-NIP or anti-gp200-MR6 (29) , no effects on cell proliferation were detected (data not shown). Moreover, ph-A7 had no effect on the proliferation of ErbB2-negative cells, such as A431 and NIH 3T3 cell lines (see Fig. 5A⇓ ). When instead NIH 3T3 fibroblasts were tested on transfection with human ErbB2, they were found to be severely inhibited in their proliferation by ph-A7 (see Table 1⇓ ). These data clearly indicate the specificity of the antiproliferative activity of the anti-ErbB2 scFv.
The effects of ph-A7 and s-A7 on the proliferation of ErbB2-positive and -negative cell lines. Cells were treated for 96 h with either ph-A7 (A) or s-A7 (C). The cells tested were: A431 (○); NIH 3T3 (□); SK-OV-3 (•); MDA-MB453 (▴); BT474 (♦); SKBR3 (▪). In the dose-response curves, cell survival is expressed as percentage of live cells with respect to untreated cells (∼3 × 104 cells). In B and D, the antiproliferative effects of ph-A7 (2 × 1010 cfu/ml; B) or s-A7 (20 μg/ml; D) on SKBR3 cells are expressed as the percentage of DNA synthesis in treated versus control cells. In control cells 1.2 × 103 cpm of [3H]thymidine were incorporated. Unrelated scFv (anti-NIP, anti-gp200-MR6), in their phage (ph-) or soluble format, and phage lacking the scFv moiety (wt-phage) were tested as controls; bars, ±SD.
The IC50 values, i.e., the phage concentrations capable of inhibiting cellular proliferation by 50%, calculated for the above mentioned cell lines, are tabulated in Table 1⇓ . With the exception of the SKBR3 cells, for which an IC50 value of ∼2 × 1010 cfu/ml was determined (Table 1)⇓ , the IC50 values obtained with ErbB2-positive cell lines (MDA-MB453, BT-474, SK-OV-3, and transfected NIH 3T3 cells) were found to be in the range of 1011 cfu/ml, i.e., 1 order of magnitude higher than the value determined for SKBR3 cells. It should also be noted that the effect of ph-A7 on these four cell lines appeared to be cytostatic, rather than cytotoxic, as the number of surviving cells was never found to be lower than the number of plated cells; on SKBR3 cells instead, a clear cytotoxic effect was evidenced. The higher sensitivity of SKBR3 cells to ph-A7 is additionally supported by the observation that incubation of SKBR3 cells with ph-A7 leads to a dramatic change in cell morphology and the appearance of cell debris; no such changes were seen with the other ErbB2-positive cell lines tested in this study (data not shown).
When soluble s-A7 was tested on the panel of ErbB2-positive cell lines, it was found to retain its ability to severely inhibit cell proliferation and to reduce viable cell number in a dose-dependent manner (see Fig. 5C⇓ ), with a stronger effect on SKBR3 cells with respect to the other receptor-positive cell lines. The IC50 values obtained for the SKBR3, the MDA-MB453, the BT474, and the SK-OV-3 cell lines were 5.4 ± 0.2, 9.0 ± 0.45, 11.0 ± 0.6, and 23.3 ± 0.9 μg/ml, respectively. No effects were detected instead with the control anti-NIP scFv on SKBR3 cells (Fig. 5D)⇓ or on the other positive cell lines (data not shown); likewise, s-A7 was ineffective on A431 and NIH 3T3 cells (Fig. 5C)⇓ but effective on NIH 3T3 ErbB2-transfected cells (data not shown). In this case, given the lower level of ErbB2 with respect to the other receptor-positive cells tested (see Table 1⇓ ), cells were starved before treatment to enhance the immunoreagent effect. Under these conditions, <50% of cells survived, whereas the untransfected cells, tested in parallel, were unaffected by s-A7 (data not shown).
To directly compare the potency of the inhibitory effect of ph-A7 and s-A7 on ErbB2-positive cell growth, the IC50 values calculated for ph-A7 (Table 1)⇓ were expressed also in terms of scFv concentration in the phage preparation, assuming that one molecule of scFv is displayed per phage particle. By comparing these values to those obtained on the same cell lines with s-A7 (see above), the scFv in the phage format appeared to be at least 1000-fold more active as an antitumor agent than the corresponding soluble scFv.
To determine whether the mechanism of cell death occurs through induction of apoptosis, we used annexin V to measure the appearance of phosphatidylserine, a marker of apoptosis, on the outer leaflet of the plasma membrane of SKBR3 cells. Cells treated with A7, either in its phage or soluble format, were found to bind 2-fold more FITC-conjugated annexin V than untreated cells or cells treated with a control anti-NIP scFv, either in its phage or its soluble format (Table 2)⇓ .
Extent of apoptosis in SKBR3 cells treated for 24 h with ph-A7, s-A7, or irrelevant immunoreagents
Effects on Tyrosine Phosphorylation of ErbB2.
Because tyrosine kinase receptors are activated by ligand binding with an increase in their phosphotyrosine content, we tested the effects of s-A7 on ErbB2 phosphorylation. Cell extracts of starved SKBR3 cells, treated at 37°C with s-A7 (12 μg/ml), were analyzed by parallel Western blottings using either P-Tyr mAb specific for phosphotyrosine or anti-ErbB2 MgR6 mAb. Both analyses were performed in the presence of an antiactin mAb to directly compare the levels of ErbB2 receptor with those of tyrosine phosphorylation. The signal intensity of positive bands was estimated by phosphorimaging. In Fig. 6⇓ , the effects of s-A7 on ErbB2 phosphorylation is shown. A strong inhibitory effect was already detectable after a 1-h treatment, and after 7 h the inhibition reached 74% in comparison with untreated cells. In the same experiment, EGF was found to effectively stimulate phosphorylation of ErbB2 (see inset of Fig. 6⇓ ), but this stimulatory effect was significantly reduced when cells were incubated previously with s-A7 (see inset of Fig. 6⇓ ).
The effects of s-A7 on ErbB2 phosphorylation. The levels of ErbB2 phosphorylation in lysates from SKBR3 cells, treated for the indicated times with s-A7 (12 μg/ml), are reported as percentages of the phosphorylation level detected in untreated cells. In the inset, the effects are shown of: a 2-h treatment with s-A7 (12 μg/ml); a 15-min treatment with EGF (100 ng/ml); the same treatment with EGF on cells treated previously for 2 h with s-A7. Control, untreated cells.
Discussion
Here we report the successful isolation of a novel human anti-ErbB2 scFv from a large synthetic human antibody phage display library. This mini-antibody exhibits unique functional properties, because it not only induces internalization, but also displays an intrinsic, strong and specific antiproliferative activity toward ErbB2-overexpressing cells. Thus, this is the first report, to our knowledge, of a human anti-ErbB2 scFv endowed with antitumor activity.
The majority of antibodies thus far isolated from combinatorial libraries expressed on phages have been selected using purified antigens or peptides immobilized on artificial surfaces. The disadvantage of this approach is that it may lead to the selection of antibodies that do not recognize the antigen in its native state, i.e., in its physiological context (38) . For example, an anti-ErbB2 scFv isolated using the purified extracellular domain of ErbB2 was unable to bind to ErbB2 on the surface of SKBR3 cells (20) .
In contrast, direct panning of a scFv library on live cells has been shown to have considerable potential in the isolation of phage antibodies recognizing cell surface antigens in their native configuration (23 , 24 , 39) . Furthermore, this strategy also allows the identification of novel cell surface antigens, which may be of clinical, diagnostic, or therapeutic use (22 , 39, 40, 41) .
The anti-ErbB2 scFv was selected by panning the phage library on live ErbB2-overexpressing cells, using a subtractive selection strategy based on the use of two combinations of ErbB2-positive and -negative cell lines. Using this approach, clone A7 was isolated. Flow cytometric and Western blotting analyses, together with functional studies, clearly demonstrate that the clone A7 specifically recognizes ErbB2 and displays no cross-reactivity with the structurally related EGF receptor (ErbB1) expressed at high levels on A431 cells (Refs. 42 , 43 ; see Table 1⇓ ). In addition, it should be noted that as ph-A7 binds efficiently to SK-OV-3 and SKBR3 cells, which express high levels of ErbB2 and low levels of ErbB3 and ErbB4, respectively (44) , we can conclude that ph-A7 is also capable of discriminating ErbB2 from all other members of the ErbB family.
When prepared as a soluble pure protein, named s-A7, it was found to retain the binding specificity for the receptor. Furthermore, A7 scFv, both in its phage and soluble format, was rapidly endocytosed into the ErbB2-overexpressing SKBR3 cells, as demonstrated by confocal microscopy, as well as by titration of the internalized phages. Because the phagemid used in this study is engineered to display only a single copy of a monovalent scFv per phage particle, the results reported in this investigation suggest that monovalent antibody-antigen interactions are sufficient to induce endocytosis.
On the basis of previous investigations, the divalent nature of antibodies was considered as essential for the induction of internalization through receptor dimerization (45) . Furthermore, it has been reported that divalent phage antibodies can be generated spontaneously from monovalent scFv, albeit at low frequency (46) . However, Poul et al. (22) have recently isolated a panel of monovalent scFv-carrying phages that can be internalized, including scFv specific for ErbB2 or the transferrin receptor. Thus, our data with ph-A7 and s-A7 are in line with the latter findings on the ability of monovalent scFv-phage to induce internalization.
The anti-ErbB2 scFv reported in the literature to date can only undergo endocytosis (22 , 34) , whereas s-A7 can also strongly inhibit receptor phosphorylation and induce inhibition of tumor cell growth. This suggests that the inhibitory action of s-A7 on tumor cell proliferation is attributable at least in part to its inhibition of receptor phosphorylation. Indeed, the conventional murine anti-ErbB2 4D5 mAb, which is known to have antitumor activity, can also inhibit ErbB2 phosphorylation (11 , 47) . In contrast, immunoreagents that do not alter ErbB2 phosphorylation, do not inhibit cell growth (12 , 22) . Hence, we can surmise that one possible mechanism of action is that s-A7 acts as an antagonist of the putative ligand of ErbB2, possibly by inhibiting receptor dimerization and/or by interfering with the signaling pathway that stimulates mitogenesis. This hypothesis is also supported by the finding that the stimulatory effect of EGF on ErbB2 phosphorylation is substantially inhibited by s-A7. Work is in progress to elucidate the mechanism underlying the cell growth inhibition effects displayed by s-A7.
Interestingly, the antitumor activity of A7 scFv correlates with the level of ErbB2 expression on ErbB2-positive cell lines, whereas no effects on ErbB2-negative cell lines were detected. Moreover, A7 scFv was found to have an antiproliferative effect on all of the ErbB2-positive tumor target cells tested, whereas its effect on SKBR3 cells is cytotoxic and related to induction of apoptosis. A similar effect has been observed previously with an anti-ErbB2 murine mAb (13) . The mechanism underlying this high sensitivity of SKBR3 cells to anti-ErbB2 immunoreagents is not known. However, these cells may be particularly dependent on an autocrine loop between the ErbB2 receptor and its ligand; this loop would be disrupted by the anti-ErbB2 scFv.
To our knowledge, ph-A7 represents the first example of a cytostatic/cytotoxic scFv displayed on phage. In fact, the anti-ErbB2 scFv reported here is more active as an antitumor agent in the phage format than when it is expressed as a soluble scFv. A plausible interpretation of this observation is that when the scFv is displayed on phage, it is more stable and/or it acquires a different conformation with a higher stability, which enhances its biological effects. Alternatively, although the majority of phage will express only a single scFv, it may not be excluded that some phages may express more copies of the scFv fragments. This may increase the antibody valency with enhanced efficacy of the scFv in its phage format.
As the biological properties described for the soluble scFv are retained when the scFv moiety is expressed on filamentous phages, the data reported above indicate that it is possible to develop new phage selection procedures based on the scFv-specific bioactivity.
In conclusion, we have isolated a novel human anti-ErbB2 scFv capable, both in its soluble and phage format, to be effectively internalized by ErbB2-overexpressing target cells and to specifically inhibit their growth. Because of the entirely human origin of the scFv, it is expected to be nonimmunogenic in humans. Therefore, this scFv represents an ideal immunobullet for ErbB2-positive cancer cells. Additionally, ph-A7 and s-A7, because of their capacity to be effectively internalized by target cells, should provide useful vehicles to deliver drugs or toxins into the cytosol of tumor target cells. These chimeric reagents would additionally enhance the antitumor potential of the isolated moieties while reducing the systemic toxicity of most toxins.
Acknowledgments
We thank Dr. G. Winter (Medical Research Council, Cambridge, United Kingdom) for kindly supplying the Griffin.1 library; Drs. A. Mele and D. Parente (Menarini Ricerche, Pomezia, Italy) for providing us with the murine MgR6 mAb; Dr. N. E. Hynes (Friedrich Miescher Institute, Basel, Switzerland) for supplying the ErbB2-transfected NIH 3T3 cell line; Dr. G. Warnes for cell sorting; and Dr. E. Eren for photography.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
↵1 Supported by grants from the Associazione Italiana per la Ricerca sul Cancro, Italy, and from the Medical Research Council, United Kingdom. D. B. P. is a recipient of an MRC Career Development Award.
↵2 These authors contributed equally to this work.
↵3 To whom requests for reprints should be addressed, at Department of Biological Chemistry, University of Naples Federico II, via Mezzocannone 16, 80134 Naples, Italy. Phone: 39-081-2534731; Fax: 39-081-5521217; E-mail: dalessio{at}unina.it
↵4 The abbreviations used are: EGF, epidermal growth factor; scFv, single-chain variable fragment; ph-A7, phages from clone A7 bearing the human anti-ErbB2 scFv; s-A7, the human anti-ErbB2 scFv in the soluble format; FACS, fluorescence-activated cell sorting; NIP, 4-hydroxy-3-nitro-5-iodophenyl-acetyl; VH, heavy chain variable region; mAb, monoclonal antibody; VL, light chain variable region; cfu, colony-forming unit.
↵5 Tomlinson, I. M., Williams, S. C., Corbett, S. J., Cox, J. P. L., and Winter, G. The V BASE Directory of Human Variable Gene Sequences. MRC Centre for Protein Engineering, Hills Road, Cambridge, CB2 2QH, United Kingdom. Internet address: http://www.mrc-cpe.cam.ac.uk/imt-doc/vbase-home-page.html, 1996.
- Received November 5, 2001.
- Revision received February 20, 2002.
- Accepted March 8, 2002.
References
Volume 8, Issue 6
