J Cancer Res Clin Oncol DOI 10.1007/s00432-015-1975-5

ORIGINAL ARTICLE – CANCER RESEARCH

Novel EGFR‑specific immunotoxins based on panitumumab and cetuximab show in vitro and ex vivo activity against different tumor entities Judith Niesen1 · Christoph Stein1,2 · Hannes Brehm2 · Grit Hehmann‑Titt4 · Rolf Fendel1,2 · Georg Melmer4 · Rainer Fischer1,3 · Stefan Barth1,2 

Received: 12 March 2015 / Accepted: 15 April 2015 © Springer-Verlag Berlin Heidelberg 2015

Abstract  Purpose  The epidermal growth factor receptor (EGFR) is overexpressed in many solid tumors. EGFR-specific monoclonal antibodies (mAbs), such as cetuximab and panitumumab, have been approved for the treatment of colorectal and head and neck cancer. To increase tissue penetration, we constructed single-chain fragment variable (scFv) antibodies derived from these mAbs and evaluated their potential for targeted cancer therapy. The resulting scFv-based EGFR-specific immunotoxins (ITs) combine target specificity of the full-size mAb with the cell-killing activity of a toxic effector domain, a truncated version of Pseudomonas exotoxin A (ETA′). Methods  The ITs and corresponding imaging probes were tested in vitro against four solid tumor entities (rhabdomyosarcoma, breast, prostate and pancreatic cancer). Specific binding and internalization of the ITs scFv2112ETA′ (from cetuximab) and scFv1711-ETA′ (from panitumumab) were demonstrated by flow cytometry and for the Electronic supplementary material  The online version of this article (doi:10.1007/s00432-015-1975-5) contains supplementary material, which is available to authorized users. * Judith Niesen [email protected] 1

Fraunhofer Institute for Molecular Biology and Applied Ecology IME, Forckenbeckstrasse 6, 52074 Aachen, Germany

2

Department of Experimental Medicine and Immunotherapy, Institute of Applied Medical Engineering, RWTH Aachen University Clinic, Aachen, Germany

3

Institute of Molecular Biotechnology (Biology VII), RWTH Aachen University, Aachen, Germany

4

Pharmedartis GmbH, Aachen, Germany







scFv-SNAP-tag imaging probes by live cell imaging. Cytotoxic potential of the ITs was analyzed in cell viability and apoptosis assays. Binding of the ITs was proofed ex vivo on rhabdomyosarcoma, prostate and breast cancer formalin-fixed paraffin-embedded biopsies. Results  Both novel ITs showed significant pro-apoptotic and anti-proliferative effects toward the target cells, achieving IC50 values of 4 pM (high EGFR expression) to 460 pM (moderate EGFR expression). Additionally, rapid internalization and specific in vitro and ex vivo binding on patient tissue were confirmed. Conclusions  These data demonstrate the potent therapeutic activity of two novel EGFR-specific ETA′-based ITs. Both molecules are promising candidates for further development toward clinical use in the treatment of various solid tumors to supplement the existing therapeutic regimes. Keywords  Epidermal growth factor receptor (EGFR) · Immunotoxin (IT) · Single-chain fragment variable (scFv) · Pseudomonas exotoxin A (ETA′) · SNAP-tag · Cancer therapy

Introduction The epidermal growth factor receptor (EGFR/ErbB1) a member of the ErbB family, which also includes HER-2/ ErbB2, HER-3/ErbB3 and HER-4/ErbB4, was the first receptor directly associated with human cancer (de Larco and Todaro 1978; Koefoed et al. 2011). It is one of the bestcharacterized tyrosine kinases and a validated target for cancer therapy (Nicholson et al. 2001). EGFR comprises an extracellular domain, which can be subdivided into four subdomains (DI–DIV), a single transmembrane domain and an intracellular kinase domain. After ligand binding,

13



DI and DIII undergo conformational changes that bring them together to form a ligand-binding pocket (Koefoed et al. 2011; Li et al. 2005; Yewale et al. 2013). Seven mammalian ligands are known to bind to the EGFR: epidermal growth factor (EGF), heparin-binding EGF-like growth factor (HB-EGF), transforming growth factor α (TGFα), amphiregulin, betacellulin, epiregulin and epigen. Ligandbinding promotes receptor heterodimerization or homodimerization causing the autotransphosphorylation of tyrosine residues in the intracellular tyrosine kinase domain and the stimulation of cell proliferation (Sasaki et al. 2013; Schneider and Yarden 2014). The EGFR is internalized in the absence of ligands with a half-life of 30 min and is rapidly recycled back to the cell surface. In tumor cells, the metabolic half-life of EGFR increases up to 20 h, so a single EGFR molecule cycles many times through the endocytic pathway during its lifetime (Yewale et al. 2013). The EGFR signaling pathway becomes more active in many cancers (Pedersen et al. 2010; Yewale et al. 2013). This may reflect an increase in EGFR gene expression, the greater availability of ligands so that the receptor is constantly stimulated in the tumor microenvironment or mutations in the EGFR gene that generate a constitutively active receptor (Li et al. 2005; Pedersen et al. 2010; Pines et al. 2010; Sasaki et al. 2013; Yewale et al. 2013). EGFR overexpression has been confirmed in many types of tumor such as in solid and epithelial tumors; this is associated with accelerated tumor growth, greater risk of metastasis, poorer prognosis and resistance to chemotherapy (Bruell et al. 2003; Mendelsohn 2002; Nicholson et al. 2001). Therapeutic strategies that target EGFR are therefore beneficial, especially for tumors that resist standard chemotherapy. For example, EGFR overexpression has been observed in prostate cancer and EGFR-targeted therapy has proven beneficial for the treatment of metastatic, castration-resistant prostate tumors. A phase II clinical trial involving patients with EGFR+ tumors showed that a combination of the EGFRspecific monoclonal antibody (mAb) cetuximab (Erbitux®, Merck Sereno) and docetaxel chemotherapy can re-induce anti-tumor responses and achieve longer progression-free survival (Cathomas et al. 2012). EGFR overexpression is also observed in half of all cases of triple-negative breast cancer (TNBC), which is characterized by the absence of three typical chemotherapeutic targets: estrogen receptor, progesterone receptor and HER-2/neu. Therefore, new treatment options targeting, e.g., the EGFR are urgently needed. There are EGFR-targeted therapeutic approaches for TNBC published, using mAbs such as cetuximab and panitumumab (Vectibix®, Amgen) (Carey et al. 2012; Nabholtz et al. 2012). EGFR is also commonly expressed in pancreatic cancer, which is highly malignant and difficult to treat (Faller and Burtness 2009). In rhabdomyosarcoma (RMS), EGFR is overexpressed in up to 76 % of all cases

13

J Cancer Res Clin Oncol

of embryonal RMS (ERMS) and in up to 50 % of all cases of the more aggressive alveolar RMS (ARMS) (Armistead et al. 2007; Ganti et al. 2006; Ricci et al. 2000). The tumorspecific overexpression of EGFR and the relatively low expression levels in surrounding healthy tissues have therefore made this receptor a suitable target for cancer diagnosis and therapy (Azemar et al. 2000; Tebbutt et al. 2013). The two most advanced mAbs targeting the EGFR extracellular domain are cetuximab and panitumumab. Cetuximab is a chimeric (mouse/human) mAb indicated for colorectal and head and neck cancer (Barnea et al. 2013; Schlessinger et al. 2001). Panitumumab is a fully human mAb approved for metastatic colorectal cancer with disease progression after prior treatment and non-mutated wild-type KRAS tumors (Jakobovits et al. 2001; Tebbutt et al. 2013). Both mAbs are approved for therapeutic use in humans, along with trastuzumab, which binds the extracellular domain of another ErbB family receptor, HER-2 (Gilabert-Oriol et al. 2013). When these antibodies bind to the extracellular domain of their target receptors, they prevent ligand binding, inhibit cell cycle progression and the corresponding signal transduction pathways and therefore induce the target cells to undergo apoptosis (Koefoed et al. 2011; Yewale et al. 2013). Although several mAbs and antibody–drug conjugates carrying therapeutic agents such as toxins to cancer cells have been approved for therapeutic use in humans, their size (150 kDa) can limit tumor penetration (Bruell et al. 2005; Singh et al. 2007). Single-chain fragment variable (scFvs) antibodies comprise the immunoglobulin variable domains of the parent mAb joined by a flexible polypeptide linker, and these smaller molecules (~30 kDa) achieve better tumor penetration (Asano et al. 2013; Kampmeier et al. 2010). Fusion proteins based on scFvs and truncated toxins derived from plants or bacteria result in recombinant immunotoxins (ITs), these are promising tools for targeted cancer treatment (Pastan et al. 2007). They bind receptors or other antigens on cancer cells and are taken up by endocytosis and subsequently induce apoptosis by triggering the corresponding pathways (Antignani and Fitzgerald 2013). The absence of the Fc region in scFvs as recombinant antibody fragment part in ITs reduces also immunogenicity of scFvs compared to full-length mAbs. Many human anti-mouse mAbs show reactivity with the Fc region and are directed against the Fc domain of therapeutic mAbs; therefore, scFvs are in many applications favorable therapeutic agents (Ahmad et al. 2012; Monnier et al. 2013; Thorpe et al. 2003). Additionally, full-length mAbs are often taken up by Fc receptor expressing cells, which reduces the number of therapeutically active mAbs at the side of the tumor. Furthermore, scFvs are easier to clone and express in a functional form in bacterial and mammalian cells, reducing the costs of production in large quantities (Ahmad et al. 2012).

J Cancer Res Clin Oncol

Some of the earliest recombinant ITs targeted EGFR and used the ligands EGF or TGFα as the targeting component and Pseudomonas exotoxin A (ETA) as the effector domain (Chaudhary et al. 1987; Kreitman 2006). Several scFv-ETA ITs targeting the EGFR family are available, including D2C7-(scdsFv)-PE38KDEL which binds to the extracellular domain of wild-type EGFR and EGFRvIII, a constitutively active receptor tyrosine kinase overexpressed in glioma cells. Preclinical studies suggest that this construct could eventually be used for treatment of brain tumors (Chandramohan et al. 2013; Chandramohan and Bigner 2013). Promising results have also been reported for the EGFR-specific IT 425(scFv)-ETA′, which showed activity against pancreatic and epidermoid cancer cells in vitro and in vivo and also more recently against RMS cells in vitro and ex vivo (Bruell et al. 2003, 2005; Niesen et al. 2014; Pardo et al. 2012). This IT was therefore used as an internal reference in the experiments described herein. Here we report the successful derivation of scFvs from the mAbs cetuximab and panitumumab (based on published sequences) and their genetic fusion to the truncated version of Pseudomonas exotoxin A (ETA′), yielding the ITs scFv2112-ETA′ (derived from cetuximab) and scFv1711-ETA′ (derived from panitumumab). Those were characterized in vitro and ex vivo. We observed target-specific binding and cytotoxic activity against the EGFR+ cancer cell lines A431 (epidermoid carcinoma), MDA-MB-468 (TNBC), C4-2 (prostate cancer), L3.6pl (pancreatic cancer) and RD (ERMS). We choose these cell lines because of the reasons explained above. Both scFvs were also combined with the well-established SNAP-tag technology to generate imaging probes for targeting and for diagnostic validation in vitro and for potential diagnostic use in vivo (Amoury et al. 2013; Kampmeier et al. 2009, 2010). We found that both ITs showed therapeutic potential and should be characterized in more detail for their application in anticancer tumor therapy.

Materials and methods Bacterial strains and plasmids All plasmids were prepared using Escherichia coli (E. coli) OneShot® TOP10 cells (F-mcrA Δ(mrr-hsdRMS-mcrBC) Φ80lacZΔM15 Δ lacX74 recA1 araD139 Δ(araleu)7697 galU galK rpsL (StrR) endA1 nupG). Synthetic oligonucleotides were prepared by MWG (Martinsried, Germany). We used a modified version of the pSecTAg/HygroB vector named pMS (Stocker et al. 2003) for expression in mammalian cells and a modified version of the pET27b vector named pMT (Barth 2002; Matthey et al. 1999; Tur et al. 2003) for bacterial expression.

Cell lines and culture conditions HEK 293T human embryonic kidney cells were obtained from the American Type Culture Collection (ATCC, Manassas, USA) and cultivated in RPMI 1640 medium (Gibco Invitrogen, Carlsbad, USA) supplemented with 10 % (v/v) heatinactivated fetal bovine serum, 2 mM l-glutamine, 100 U/ ml penicillin and 0.1 mg/ml streptomycin (ATCC-No.: CRL11268). HEK 293T cells were transfected with RotiFect (Carl Roth GmbH, Karlsruhe, Germany) using up to 1 µg of plasmid and 3 µl RotiFect, according to the manufacturer’s instructions and were cultivated as described above in medium containing 100 µg/ml Zeocin™ (Invitrogen, San Diego, USA). The epidermoid carcinoma cell line A431 (DSMZNo. ACC91) and the histiocytic lymphoma cell line U937 (DSMZ-No. ACC-5) were obtained from the DSMZ (German Collection of Microorganisms and Cell Cultures, Braunschweig, Germany). RD is a human ERMS cell line (kindly provided by Dr. Annette Paschen, Deutsches Krebsforschungszentrum DKFZ Heidelberg, Germany). The human prostate cancer cell line C4-2 (kindly provided by Prof. Dr. Elsässer-Beile, University Hospital Freiburg, Germany) is derived from the androgen-sensitive human adenocarcinoma prostate cancer cell line LNCaP. The triple-negative breast cancer cell line MDA-MB-468 was obtained from ATCC (No. HTB-132). The pancreatic carcinoma cell line L3.6pl is derived from COLO357 (Bruns et al. 1999). All cell lines and transfected HEK 293T cells were incubated at 37 °C in a humidified 5 % CO2, 95 % air incubator using the recommended media. Construction and generation of the ITs and SNAP‑tag imaging probes The scFv2112 derived from cetuximab was synthesized based on published sequences (US Patent 6,217,866 B1, 2001). The synthetic gene was prepared by Life Technologies™ GeneArt® and optimized for maximum performance in E. coli using the GeneOptimizer® Software. The scFv1711 derived from panitumumab was synthesized based on the published sequences (US Patent 6,235,883 B1, 2001). Missing sequences in the framework regions were replaced with amino acids adapted to germ-line gene sequences. We added the 5′ restriction site SfiI and the 3′ restriction site NotI to facilitate further cloning steps. For bacterial expression, each scFv was directly inserted into the bacterial expression vector pMT using restriction sites SfiI and NotI. This vector contains an isopropyl-β-d-1thiogalactopyranoside (IPTG)-inducible lac operator, a pelB signal peptide, an Enterokinase-cleavable His6-tag and a modified ETA′ sequence (Bruell et al. 2003). For mammalian expression, the scFv-SNAP fusion proteins were constructed by cloning each scFv sequence

13



(standardized by adding SfiI and NotI sites) in the pMS vector system, which contains a modified SNAP-tag sequence as described (Kampmeier et al. 2009, 2010). The vector also includes a cytomegalovirus (CMV) promoter (pCMV), a Zeocin resistance gene (ZeoR) used for the selection of transfected HEK 293T cells, an Igκ-leader (Igκ-L) for secretion, a tandem Myc-His6-tag for detection and purification and an internal ribosomal entry site (IRES), which enables the co-translation of the enhanced green fluorescent protein (EGFP) (Kampmeier et al. 2009; Stocker et al. 2003). Successful cloning was verified by sequencing and control restriction digest. Expression and purification of ITs and SNAP‑tag imaging probes After transformation, the recombinant ITs scFv2112-ETA′ and scFv1711-ETA′ were expressed in the periplasm of E. coli strain BL21 Star (DE3) under stress in the presence of compatible solutes as previously described (Barth 2002). Briefly, when the bacteria reached an OD600 of 1.6, the culture was supplemented with 0.5 M sorbitol, 10 mM betaine and 4 % (w/w) NaCl to induce stress and incubated at 26 °C for 30 min while shaking (170 rpm). Expression was induced by adding 2 mM IPTG, and the cells were harvested after 16-h incubation at 26 °C while shaking. After centrifugation (4000×g, 20 min, 4 °C), the pellet was frozen immediately at −80 °C, stored overnight and then re-suspended at 4 °C in preparation buffer [75 mM Tris– HCl, 300 mM NaCl, 5 mM dithiothreitol (DTT), 10 mM EDTA, 10 % (v/w) glycerol, pH 8.0] containing a complete protease inhibitor cocktail tablet (Roche, Mannheim, Germany) and sonicated six times for 60 s at 200 W. After centrifugation (30,000×g, 30 min, 4 °C), the periplasmic fraction was recovered and EDTA was removed by overnight dialysis against phosphate-buffered saline (PBS, pH 7.4). His6-tagged recombinant ITs were purified by immobilized metal ion affinity chromatography (IMAC) on an Äkta-Purifier System (GE Healthcare, Freiburg, Germany). Pure protein was dialyzed against PBS, and aliquots were stored at −80 °C. After the transfection of HEK 293T cells with the scFvSNAP constructs, SNAP-tag activity was tested by staining 20 µl of cell culture supernatant with the BG-derivative SNAP-Vista® Green (New England BioLabs, Schwalbach, Germany) followed by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and visualization under a UV transilluminator. The protein was purified from cell-free culture supernatant by IMAC to capture the His6-tag using an Äkta-FPLC system with a 5 ml Ni–NTA Superflow column (Qiagen, Hilden, Germany) as previously described (Kampmeier et al. 2009). The proteins were concentrated using a 10-kDa molecular weight cutoff

13

J Cancer Res Clin Oncol

(MWCO) VivaSpin column (Sigma-Aldrich, Taufkirchen, Germany), 1 mM of DTT was added, and aliquots were stored at −80 °C. The scFv-SNAP proteins were labeled with BG-modified dyes as described elsewhere (Kampmeier et al. 2009, 2010). SDS‑PAGE and Western blot analysis SDS-PAGE and Western blots were carried out as previously described (Bruell et al. 2003). Purified recombinant proteins were separated by SDS-PAGE and stained with Coomassie Brilliant Blue. Separated protein was transferred to a nitrocellulose membrane. Protein concentrations were determined by densitometric analysis against bovine serum albumin as a standard and were quantified using AIDA software (Raytest GmbH, Straubenhardt, Germany). The ITs were detected by Western blot using the in-house ETAspecific mouse antibody (TC-1) (diluted 1:5000, 230 ng/ µl) and an alkaline phosphatase (AP)-conjugated goat antimouse IgG (GAMAP) mAb diluted 1:5000 (Dianova, Hamburg, Germany). Final staining was performed using nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate (NBT/BCIP) substrate (Life Technologies, Thermo Fisher Scientific, Waltham, MA, USA), and the protein size was compared to a pre-stained broad-range protein marker (New England BioLabs, Schwalbach, Germany). The coupling efficiency of the scFv-SNAP probes was determined by SDS-PAGE with the Cri Maestro imaging system (PerkinElmer, Waltham, MA, USA) using the appropriate filter sets. Flow cytometry and internalization analysis Specific binding of the fusion proteins to the target cells was analyzed by flow cytometry using a BD FACSVERSE (BD Biosciences, Franklin Lakes, NJ, USA) and the corresponding software. We incubated 4 × 105 cells with 800 ng of the ITs in 100 µl PBS for 30 min on ice, followed by incubation steps with the ETA-specific mouse antibody (TC-1) and a fluorescein isothiocyanate (FITC)-conjugated goat antimouse (GAMFITC) mAb (Dianova, Hamburg, Germany). The corresponding mAbs were added at equimolar concentrations to the cells and detected with an FITC-labeled goat anti-human IgG (H + L) F(ab′)2 (Chemicon, Merck Millipore, Darmstadt, Germany). Both detection antibodies were diluted 1:100 in 50 µl PBS before use, and the cells were washed with PBS between incubation steps. The QIFIKIT® kit (Dako, Hamburg, Germany) was used to detect the number of EGFR antigens on the cell surface by flow cytometry according to the manufacturer’s protocol. A sample of cetuximab was kindly provided by Dr. Agnieszka Weinandy (University Hospital Aachen, Neurosurgery Clinic, Aachen, Germany) and a sample of panitumumab by Martina

J Cancer Res Clin Oncol

Weichbrodt (University Hospital Aachen, Pharmacology, Aachen, Germany). The rate of IT internalization was determined as described by Cizeau et al. (2009). We used scFv2112-ETA′, scFv1711-ETA′ and 425(scFv)-ETA′ in the same concentrations and incubated them with the cells for 30 min at 4 °C. After a washing step, cell culture medium was added and the cells were incubated for 15, 30, 60, 120 and 180 min at 37 °C. We used the same antibodies as described above. Results were presented as the percentage change in geometric mean fluorescence (MFI), assigning the MFI of the 4 °C control a score of 100 %. Confocal microscopy and Opera® high‑content screening To visualize the internalization of the scFv-SNAP probes coupled to SNAP-Surface® Alexa Fluor® 488 (New England BioLabs, Schwalbach, Germany), the proteins were added at a concentration of 1 µg (~100 nM) to the cancer cells (2 × 105 cells/ml) seeded in eight-well chamber slides (Thermo Fisher Scientific, Waltham, MA, USA) for confocal microscopy or 96-well black cell culture microplates (Greiner-Bio-One GmbH, Frickenhausen, Germany) for Opera high-content screening. The cells were incubated at 37 °C for 60–120 min, and the medium was replaced. The nucleus was counterstained with Bisbenzimide Hoechst 33342 (Sigma-Aldrich, Taufkirchen, Germany) for confocal analysis and 4′,6-diamidino-2-phenylindole (DAPI,) for Opera analysis. Internalization was monitored using a LEICA TCS SP8 confocal microscope and the corresponding software (Leica Microsystems GmbH, Wetzlar, Germany) or with the PerkinElmer Opera® automated image acquisition system and subsequent image analysis (A Capella Software). Colorimetric cell proliferation assay The cytotoxic activity of both ITs against the target cells was measured using a cell proliferation assay as previously described (Schiffer et al. 2013). Briefly, 1 × 105 cells/ml were seeded in 96-well plates and incubated with different dilutions of the ITs for 72 h at 37 °C, 5 % CO2 and 100 % humidity. Untreated cells, scFv-SNAP constructs, mock-ETA′ and the parental mAbs were used as controls. Substrate conversion was measured by adding 50 µl XTT/ phenazine methosulfate (Serva and Sigma, Steinheim, Germany) to each well and incubating the plates for 2–4 h before measuring the absorbance at 450- and 630-nm reference wavelength using an Epoch Microplate Spectrophotometer (Biotek, Bad Friedrichshall, Germany). All experiments were carried out at least three times in triplicate or quadruplicate. The concentration required to achieve a

50 % reduction in protein synthesis (IC50), 95 % confidence interval (95 % CI) relative to the untreated control cells was calculated using the Hill equation with GraphPad Prism v5, (GraphPad Software, La Jolla, CA, USA). Apoptosis assay Apoptosis was measured by annexin V/propidium iodide (PI) staining according to Stahnke (Stahnke et al. 2008). We seeded 4 × 105 cells/ml in 300 µl medium and incubated them with 80 nM of each IT in a 24-well plate for 48 h at 37 °C, 5 % CO2 and 100 % humidity. The mAbs, mock-ETA′ and buffer (PBS) were used as controls, as well as the EGFR− cell line U937. After incubation, cells were washed in 1× annexin V binding buffer (10 mM HEPES, 150 mM NaCl, 5 mM KCl and 2 mM CaCl2, pH 7.4) and stained with 450 µl annexin V-EGFP cell culture supernatant and 50 µl 10× annexin V binding buffer for 15 min at room temperature. After washing, the cells were re-suspended in 1× annexin V binding buffer containing 1 µg/ml PI and analyzed by flow cytometry using BD FACSVERSE (BD; Franklin Lakes, NJ, USA). Experiments were carried out independently in duplicates at least four times, and the results were presented as means ± standard errors (SEMs), with significance determined using a two-tailed unpaired Student’s t test (**p ≤ 0.01, ***p ≤ 0.001). Ex vivo binding to human tumor tissue Formalin-fixed, paraffin-embedded (FFPE) tissue sections were treated to remove paraffin as described elsewhere (Niesen et al. 2014). Circles were painted around each section using a Dako-pen (Dako, Hamburg, Germany). Blocking solution (PBS supplemented with 1 % (v/v) goat serum) was added to the dry slides and tapped off after 1-h incubation. The slides were incubated overnight at 4 °C with either the parental mAbs or the recombinant IT scFv2112-ETA′ or scFv1711-ETA′ at an amount of 1 µg. After incubation, the slides were washed three times for 5 min with PBS. Slides previously incubated with the recombinant ITs were incubated for up to 12 h at 16 °C with the primary antibody TC-1, diluted 1:40 in blocking solution followed by a PBS washing step. A secondary antibody GAMAP (Dianova, Hamburg, Germany) for the ITs or an AP-labeled goat antihuman mAb (Sigma-Aldrich, Taufkirchen, Germany) for the full-size mAbs, each diluted 1:50 in blocking solution, was added to the slides and incubated at 4 °C overnight. After washing in PBS, AP activity was detected using naphthol AS-BI phosphate (sodium salt, 50 mg/100 ml; Sigma-Aldrich, Taufkirchen, Germany) as a substrate and New Fuchsin (Merck, Darmstadt, Germany) as a chromogen dissolved in 0.1 M Tris–HCl (pH 8.5). Endogenous AP activity was inhibited by adding 0.35 mg/

13



J Cancer Res Clin Oncol

ml levamisole (Sigma-Aldrich, Taufkirchen, Germany). Slides were counterstained with ready-to-use hematoxylin and eosin (H&E) (Sigma-Aldrich, Taufkirchen, Germany). Images were analyzed using a Leica DMR-HC light microscope and the corresponding Leica QWin software (Leica Microsystems GmbH, Wetzlar, Germany). The tissue samples were obtained during routine clinical practice at the University Hospital Giessen, in accordance with the principles of the Declaration of Helsinki.

Results Expression and purification of the ITs and SNAP‑tag constructs The ITs scFv2112-ETA′ and scFv1711-ETA′ (~72 kDa) were expressed in the periplasm of E. coli strain BL21 Star (DE3). The imaging probes scFv2112-SNAP, scFv1711SNAP and 425(scFv)-SNAP were expressed in HEK 293T cells. All proteins were successfully purified under native conditions by immobilized metal ion affinity chromatography (IMAC) using the His6-tag. The purified ITs were detected by SDS-PAGE and Western blot using the ETA′specific mouse antibody (TC-1) which recognizes both scFv2112-ETA′ (Fig. 1a) and scFv1711-ETA′ (Fig. 1b). The final yields of the ITs were in the range 0.2–0.3 mg/l bacterial culture. The well-characterized IT 425(scFv)ETA′ was used as an internal reference in most experiments, and the expression and purification of this protein are described elsewhere (Bruell et al. 2005; Pardo et al. 2012). The yield of the scFv-SNAP constructs was 2–5 mg/l culture supernatant for scFv2112-SNAP and scFv1711-SNAP and up to 10 mg/l culture supernatant for 425(scFv)-SNAP. Successful benzylguanine (BG)-Alexa Fluor® 488 labeling was confirmed by Maestro™ in vivo fluorescence imaging system after separation by SDSPAGE, as shown for scFv2112-SNAP BG-488 (Fig. 1c) and scFv1711-SNAP BG-488 (Fig. 1d). Binding of the immunotoxins and mAbs to different cancer cell lines The binding characteristics of the ITs were analyzed by flow cytometry. These experiments confirmed the specific binding activity of scFv2112-ETA′ and scFv1711-ETA′ to the EGFR+ cell lines A431 (epidermoid carcinoma), MDAMB-468 (TNBC), C4-2 (prostate adenocarcinoma), L3.6pl (pancreatic cancer) and RD (ERMS). The specific binding of the internal reference 425(scFv)-ETA′ was shown by using A431 cells as representative example. Strong fluorescence signals for the ITs were observed on the target cell lines but not on the EGFR− control cell line U937 (Fig. 2).

13

Fig. 1  Enrichment of the ITs scFv2112-ETA′ (a) and scFv1711ETA′ (b) and the corresponding SNAP-tag probes (c, d) shown by SDS-PAGE and Western blot analysis. The purified ITs are fractionated by SDS-PAGE and stained with Coomassie Brilliant Blue. The Western blot membrane is incubated with the antibody TC-1 and a goat anti-mouse AP-labeled secondary antibody. SNAP-Surface® Alexa Fluor® 488 labeled scFv-SNAP-tag fusion proteins scFv2112SNAP (c) and scFv1711-SNAP (d) are separated by SDS-PAGE, visualized using the CRi Maestro Imaging System with the Maestro software and the blue filter set (500–720 nm) revealing by a green band and afterward stained with Coomassie Brilliant Blue. The anticipated sizes are ~72 kDa for the ITs and ~48 kDa for the SNAP-tag probes

The mean fluorescence intensity (MFI) was ~15–200 times higher on target cell lines incubated with the ITs compared to the control cell line U937 or the background control (cells incubated with the detection antibody alone). The different MFI values depended on the cell line and the EGFR expression level. The parental mAbs cetuximab and panitumumab, which were used as positive binding controls, confirmed these results and neither bound to the EGFR− cell line U937 (Fig. 2). For the parental mAbs, the MFI values were up to ~400 times higher than the U937 or background controls. The abundance of cell-surface EGFR in each cell line was determined using a commercial kit (QIFIKIT®, DAKO). This showed that EGFR was expressed at the highest level on the surface of A431 and MDA-MB-468 cells, followed by C4-2, L 3.6pl and finally RD cells (Table 1). Internalization of EGFR‑specific ITs and imaging probes An efficient IT must be internalized rapidly. We therefore characterized the internalization of scFv2112-ETA′, scFv1711-ETA′ and 425(scFv)-ETA′ by different EGFR+ cell lines using flow cytometry-based internalization assays

J Cancer Res Clin Oncol Fig. 2  The specific cell binding activity of the recombinant ITs is analyzed by flow cytometry. The proteins are detected using antibody TC-1 and GAMFITC for the ITs or a FITC-labeled goat antihuman IgG (H + L) F(ab′)2 for the mAbs (FL-1 fluorescence channel/FITC). The filled black curve shows the background control for the ITs (a). The binding of scFv2112ETA′ is shown as a solid black curve (b), and the binding of scFv1711-ETA′ is shown as a dashed black curve (c). The binding of 425(scFv)-ETA′ (d) is shown on A431 and U937 cells only as representative cell lines. The filled gray curve gives the background control for the mAbs (e), the solid gray curved demonstrates the binding of cetuximab (f), and the dashed gray curve demonstrates the binding of panitumumab (g)

Table 1  IC50 values (pM) of scFv2112-ETA′, scFv1711-ETA′ and 425(scFv)-ETA′ against selected cell lines with different EGFR expression levels Construct

A431 IC50 (pM) [95 % CI]

MDA-MB-468 IC50 (pM) [95 % CI]

C4-2 IC50 (pM) [95 % CI]

L3.6pl IC50 (pM) [95 % CI]

RD IC50 (pM) [95 % CI]

U937 IC50 (pM) [95 % CI]

scFv2112-ETA′

4 [3–7]

11 [9–14]

55 [18–165]

290 [170–497]

460 [266–487]

No effect

scFv1711-ETA′

18 [12–34]

32 [26–39]

192 [143–258]

260 [186–385]

240 [132–427]

No effect

425(scFv)-ETA′ scFv-SNAP/mockETA′/mAbs

2 [1–2]

4 [3–4]

35 [10–121]

80 [46–134]

598 [288–1245]

No effect

No effect (up to highest starting concentration of 80 nM)

EGFR expression level

148,723 (±15,992 SD)

21,290 (±5257 SD)

12,490 (±4220 SD)

54 (±23 SD)

85,891 (±17,973 SD)

22,902 (±2775 SD)

The IC50 values (confidence interval: 95 % CI) indicate the concentrations of the ITs required to achieve a 50 % reduction in protein synthesis relative to untreated control cells, as shown in Fig. 5. The scFv-SNAP, mock-ETA′ and mAb controls did not show any cytotoxicity (Fig. S2). No effect of the ITs could be measured (>80 nM) on the U937 cell line, which is used as EGFR− cell line. The EGFR expression level on the cell lines is determined using the QIFIKIT® kit. These data are presented as mean ± SD, from three independent experiments

13



J Cancer Res Clin Oncol

Fig. 3  The efficiency of internalization for scFv2112-ETA′, scFv1711-ETA′ and 425(scFv)-ETA′ are measured by flow cytometry at five different time points. The geometric mean (percentage MFI) is

shown at different time points. Cells incubated at 4 °C are assigned as 100 % in the evaluation. The ITs are detected with the ETA-specific mouse antibody TC-1 and a GAMFITC secondary antibody

(Fig.  3), and compared the internalization behavior of the imaging probes scFv2112-SNAP, scFv1711-SNAP and 425(scFv)-SNAP by confocal microscopy (Fig. 4). We also used Opera high-content screening of the probes by MDAMB-468 and L3.6pl cells. Both assays showed comparable results. In the flow cytometry assays, cells incubated at 37 °C were assayed for internalization behavior and control cells incubated at 4 °C were used to set MFI at 100 % (representing no uptake of the ITs). As shown in Fig. 3, RD cells internalized almost 90 % of scFv2112-ETA′ and scFv1711ETA′ within 60 min, but only ~50 % of 425(scFv)-ETA′ was taken up in this time. The prostate cancer cell line C4-2 internalized ~30 % of scFv2112-ETA′ and ~20 % of scFv1711-ETA′ after 180 min, and 425(scFv)-ETA′ was again taken up more slowly (Fig. 3). The L3.6pl and MDAMB-468 cells internalized ~50 % of scFv2112-ETA′ and scFv1711-ETA′ after 30 min, and once again 425(scFv)ETA′ was internalized more slowly (Fig. 3). Confocal microscopy showed that the internalization of scFv2112-SNAP (Fig. 4a) and scFv1711-SNAP (Fig. 4b) commenced after ~60 min at 37 °C, whereas the internalization of 425(scFv)-SNAP (Fig. 4c) commenced after ~80 min at the same temperature. As expected, no internalization was observed at 4 °C (Fig. 4c shows the data for RD cells incubated with 425(scFv)-SNAP as an example). As observed for the ITs, the SNAP-tag proteins were taken up most efficiently into RD cells (Fig. 4a, c). None of the SNAP-tag proteins were taken up by the EGFR− cell line

U937, as shown in supplementary Figure S1 for scFv2112SNAP as a representative example. All scFv-SNAP constructs were labeled with BG-Alexa Fluor® 488.

13

Dose‑dependent cytotoxicity toward cell lines from different tumor entities All five EGFR+ cell lines demonstrated the ability to bind and internalize the two novel ITs so next we used XTTbased colorimetric cell viability assays to determine the specific cytotoxicity of the EGFR-specific ITs in vitro. The cells were incubated with decreasing concentrations of the ITs and controls. Proliferation was compared to untreated control cells after incubation for 72 h. As shown in Fig. 5, the viability of the target cells was reduced in the presence of different concentrations of the two novel ITs. The calculated IC50 values for scFv2112-ETA′ ranged from 4 pM [95 % CI 3–7] on A431 cells and 11 pM [95 % CI 9–14] on MDA-MB-468 cells, with the highest EGFR expression levels, to 460 pM [95 % CI 266–487] on RD cells, the cell line with the lowest EGFR expression level (Table 1). The IC50 values for scFv1711-ETA′ ranged from 18 pM [95 % CI 12–34] on A431 cells and 32 pM [95 % CI 26–36] on MDA-MB-468 cells to 240 pM [132–427] on RD cells. The internal reference 425(scFv)-ETA′ showed a similar range of IC50 values. No unspecific cytotoxic effects of the ITs could be measured on the EGFR− cell line U937 with a concentration >80 nM, which was the highest start

J Cancer Res Clin Oncol

Fig. 4  The internalization behavior of scFv2112-SNAP, scFv1711SNAP and 425(scFv)-SNAP labeled with BG-Alexa Fluor® 488 visualized in different EGFR+ cell lines by confocal microscopy and Opera® high-content screening. Nuclei are counterstained with Hoechst 33342 or DAPI. The internalization of scFv2112-SNAP (a) and scFv1711-SNAP (b) is measured after incubation for 60 min at

37 °C; for scFv2112-SNAP, all five EGFR+ cell lines are demonstrated; for scFv1711-SNAP, only four cell lines are shown exemplarily. The internalization of 425(scFv)-SNAP by RD cells as representative example (c) is shown after incubation for 80 min at 37 °C or 4 °C for 30 min (non-internalization control). Each experiment is carried out at least twice. Scale bars 10 µm

concentration of the ITs on the target cell lines. To confirm the cytotoxic specificity of the ITs, scFv2112-SNAP, scFv1711-SNAP, cetuximab, panitumumab and a nonbinding mock-ETA′ construct were used as controls, but as expected, these did not show any cytotoxic effects (see supplementary Figure S2).

population of both apoptotic stages increased after incubation with the ITs, confirming the induction of apoptosis (statistically significant: **p ≤ 0.01, ***p ≤ 0.001). Early and late apoptosis/necrosis are combined in a column diagram for each cell line in Fig. 6b. On A431 and MDA-MB-468, the induction of apoptosis is induced in ~60 % of the cells for the two novel ITs and the internal reference. On C4-2 and L3.6pl, the induction of apoptotic effects reaches around 80 % for scFv2112-ETA′, and on C4-2, 80 % apoptosis induction was also reached for scFv1711-ETA′. The IT scFv1711-ETA′ only triggers apoptosis in ~60 % of L3.6pl cells. On RD cells, all ITs showed a similar effect (~70 % apoptotic cells). As expected, the PBS control, full-size mAbs and mock-ETA′ showed only weak or no apoptotic effects.

Induction of apoptosis by the ITs An annexin V/propidium iodide (PI) assay was used to determine whether the observed inhibitory effect of the ITs was as well caused by the induction of apoptosis. The cells were incubated with the ITs followed by staining with annexin V (AV)-EGFP and PI. Each IT was applied at a concentration of 80 nM to distinguish cells in the early and late apoptotic stages after 48-h incubation. Both ITs significantly reduced the cell number at relatively low concentrations. Figure 6a shows the dot blots for cell line A431, which was chosen as representative dataset. The lower right quadrant shows the population of early apoptotic cells, and the upper right corner shows the population of late apoptotic/necrotic cells. The

Binding of the ITs to human tumor biopsies ex vivo We gained the first clinically relevant information about our novel ITs by investigating the ability to bind to primary cells from formalin-fixed and paraffin-embedded (FFPE) sections of prostate, breast and RMS tumor cells/biopsies.

13



J Cancer Res Clin Oncol

Fig. 5  The cytotoxicity of each IT is determined using an XTT assay. The concentration required to achieve a 50 % reduction in protein synthesis (IC50) relative to untreated control cells is calculated using GraphPad Prism software. These experiments are carried out at least

four times in triplicate or quadruplicate. The cell lines A431, MDAMB-468, C4-2, L3.6pl, RD and U937 are incubated with serial dilutions of the sterile ITs. Data are shown as means ± SEMs

The EGFR-specific binding of both ITs was confirmed by New Fuchsin staining for both ITs, whereas hematoxylin and eosin staining was used to verify the presence of tumor cells. No signal was detected in the negative controls stained with the detection antibodies in the absence of the ITs scFv2112-ETA′ and scFv1711-ETA′ (Fig. 7a control).

The binding of cetuximab and panitumumab to breast cancer tissue is shown as an example in Fig. 7b. In agreement with their specific binding characteristics in vitro, both ITs were able to bind specifically to primary material confirming their ability to recognize primary cells from prostate cancer, breast cancer and RMS biopsies ex vivo.

13

J Cancer Res Clin Oncol

Fig. 6  The apoptotic activity of the ITs is determined by annexin V-EGFP/PI staining. All cell lines are incubated for 48 h with 80 nM of each IT, PBS as a negative control or camptothecin as a positive control. Both full-size mAbs and a mock-ETA′ are included as additional negative controls. Early and late apoptosis/necrosis are combined in a bar chart, which shows the results for all cell lines (b). Dot blots for cell line A431 as a representative example: lower left viable

cells; lower right early apoptotic cells; upper right late apoptotic/ necrotic cells. The number in each quadrant indicates the percentage of cells in each category (a). All experiments are carried out in duplicate at least four times, and data are shown as mean ± SEM. Statistical significance is determined using a two-tailed unpaired Student’s t test, (**p ≤ 0.01; ***p ≤ 0.001)

13



J Cancer Res Clin Oncol

Fig. 7  Tumor tissue biopsies from prostate cancer, breast cancer and RMS patients are shown. Representative images of the different FFPE tumor sections stained with New Fuchsin substrate and H&E (objective ×10, scale bar 100 µm). a The specific binding of scFv2112-ETA′ and scFv1711-ETA′ is indicated by red-stained EGFR+ tumor cells. The antibodies TC-1 and GaMAP are used in

the absence of the IT, as a control (as shown in the third column). b Specific binding of the corresponding mAbs cetuximab and panitumumab on breast cancer tissue as a representative example. The APlabeled goat antihuman antibody is used in the absence of cetuximab or panitumumab as a control, as shown in the third column

Discussion

(Becker and Benhar 2012; Pastan et al. 2007). EGFR is a promising target for directed tumor therapy because this well-characterized receptor is over-expressed in a variety of tumors, whereas relatively low expression levels occur in surrounding healthy tissues (Mendelsohn 2002; Nicholson

The use of ITs or immunoconjugates which trigger the direct killing of tumor cells following internalization has become increasingly important for targeted cancer therapy

13

J Cancer Res Clin Oncol

et al. 2001; Tebbutt et al. 2013; Yewale et al. 2013). We therefore tested the cytotoxicity of two novel EGFR-specific recombinant ITs against different EGFR+ cell lines and cancer types. For proof of principle, we fused the wellcharacterized truncated toxin ETAʹ to two different scFvs derived from mAbs that are currently approved for EGFRspecific antibody-based cancer therapy, the chimeric cetuximab and the fully human panitumumab (Jakobovits et al. 2001; Schlessinger et al. 2001; Sliwkowski and Mellman 2013), with the benefit that scFv1711 (derived from panitumumab) is completely human. Antibody-based therapy is an established therapeutic option for cancer (Scott et al. 2012; Sliwkowski and Mellman 2013; Wilkins and Mayer 2006). More than a dozen mAbs are currently approved by the FDA for different oncology indications with many more undergoing clinical development (Reichert 2014; Schrama et al. 2006; Scott et al. 2012; Wilkins and Mayer 2006). Cetuximab and panitumumab are the most advanced EGFR-specific therapeutic mAbs, and both bind to the EGFR extracellular domain (Tebbutt et al. 2013). However, the administration of a single therapeutic mAb to patients with solid tumors often shows low efficacy, whereas combinations of mAbs and standard therapeutic approaches have a more beneficial impact (Gerber et al. 2013). Antibody–drug conjugates (ADCs) have therefore been developed to combine the targeting specificity of mAbs with the cell-killing activity of cytotoxic drugs or radionuclides (Panowksi et al. 2014). One example is the EGFR-specific immunoliposome formulation, comprising a Fab′ fragment derived from matuzumab or cetuximab covalently linked to stabilized liposomes containing chemotherapeutic drugs (Mamot et al. 2006). A similar preparation has been developed in which panitumumab is covalently conjugated to liposomes containing encapsulated doxorubicin (Lukianova-Hleb et al. 2012). ETA has also been used as approach for ADC development by using it as a model toxin to identify the best antibody candidates targeting HER-2 (de Goeij et al. 2014). The disadvantages of mAbs include their larger size and greater immunogenicity compared to scFvs due to the presence of the Fc region (Ahmad et al. 2012). Therefore, scFvs are more suitable for diagnostic and therapeutic applications, because they retain the binding functionality and specificity of the parental full-length mAb but can penetrate tissues more efficiently. Furthermore, in contrast to ITs, in general, EGFR-directed mAbs do not kill the tumor cells directly but instead inhibit ligand binding, block signal transduction and so inhibit EGFR gene expression in the tumor cells (Bruell et al. 2003; Scott et al. 2012). We therefore developed ITs containing the scFvs derived from cetuximab and panitumumab for the targeted killing of EGFR+ tumor cells. These ITs should achieve direct and specific killing of the tumor cell upon internalization

without affecting surrounding tissues (Allen 2002; Madhumathi and Verma 2012; Weldon and Pastan 2011). Functional scFv2112-ETAʹ and scFv1711-ETAʹ were successfully isolated from the periplasmic space of E. coli under osmotic stress. The yield following purification by IMAC was 0.2–0.3 mg/l bacterial culture, which was comparable to or higher than yields reported for other recombinant scFv-ETA proteins (Schmidt et al. 1997; Schwenkert et al. 2008; Stein et al. 2010; Wolf et al. 2010). Specific binding to the five selected EGFR+ cancer cell lines was confirmed. Concerning the binding sides of the three used scFvs, it seems that they are all different. The internal reference 425(scFv) and the corresponding mAb425 (IgG2a) bind to the EGFR extracellular domain and to glyco- and aglycoreceptor forms, indicating that the epitope is part of the polypeptide chain containing amino acid residues G460/ S461 and is linked closely to the EGF ligand-binding active site. The homodimeric 425(scFv) has the same binding properties as the corresponding mAb (Kamat et al. 2008; Muller et al. 1998; Murthy et al. 1987). Cetuximab recognizes a conformational epitope on extracellular domain III, whereas the exact binding side of panitumumab is not clear (Voigt et al. 2012). Both mAbs may bind adjacent or overlapping epitopes belonging to epitope bin III/B, but definitely the epitopes are not identical (Alvarenga et al. 2012; Koefoed et al. 2011; Voigt et al. 2012). Panitumumab with specificity for the ligand-binding region binds to EGFR with a higher affinity (Kd = 50 pM) than cetuximab (Kd = 390 or 200 pM) (Freeman 2009; Kim and Grothey 2008; Shim 2011). Rapid and efficient internalization by target cells is necessary for the activity of ITs so we investigated the uptake of scFv2112-ETA′, scFv1711-ETA′ and 425(scFv)-ETA′ by flow cytometry. The prostate cancer cell line C4-2, the breast cancer cell line MDA-MB-468 and the pancreatic cancer cell line L3.6pl took ~60 min to internalize 50 % of the available scFv2112-ETA′ and scfv1711-ETA′, whereas the ERMS cell line RD took only ~30 min and almost 90 % of both ITs had been taken up after 180 min. Similar internalization times were described for another ETA-based IT (Cizeau et al. 2009). Surprisingly, we found that our internal reference 425(scFv)-ETA′ was taken more slowly by all the cell lines except the epidermoid carcinoma cell line A431. The rapid internalization on RD cells was also observed by confocal microscopy using the corresponding scFv-SNAP-tag constructs, which were expressed in HEK 293T cells with an average yield of 2–5 mg/l of purified protein, and we could demonstrate specific binding and internalization. We previously developed the scFv-SNAPtag technology because it allows the binding and internalization of scFv-SNAP in real time, without additional antibodies for detection (Amoury et al. 2013; Kampmeier et al. 2009, 2010). The simple and rapid labeling reaction using

13



BG-modified fluorescent dyes with different wavelengths (including near infrared) thus allow imaging probes to be used to determine whether novel scFvs are suitable for the construction of ITs (Amoury et al. 2013). We found that svFv2112-ETA′ and scFv1711-ETA′ showed specific and potent cytotoxicity against all selected EGFR+ cancer cell lines with IC50 values in the pico-molar range starting at 4 pM (Fig. 5), resulting in the induction of significant apoptotic effects (Fig. 6). The IC50 values of the internal reference 425(scFv)-ETA′ were similar or slightly better than those of the new ITs. Lower IC50 values were achieved against the cell lines expressing the highest level of EGFR (Table 1). Similar results have been reported for the EGFR-specific IT SE comprising EGF as the targeting component and saporin-3 as the effector domain when tested against different human cervical carcinoma cell lines (Bachran et al. 2010). They demonstrated a clear correlation between EGFR expression and sensitivity of the IT SE (Bachran et al. 2010). Besides, we want to point out that both novel ITs also significantly reduced the number of EGFR+ cells by inducing >80 % apoptosis at relatively low concentrations. ETA-based ITs are renowned for their pro-apoptotic effects (Kreitman 2006; Stein et al. 2010). Many groups use ETA-based ITs, including some that target EGFR (Azemar et al. 2000; Chandramohan and Bigner 2013; Nachreiner et al. 2008; Schwenkert et al. 2008; Stein et al. 2010; Wolf et al. 2006). We have recently shown that the EGFR-specific IT 425(scFv)-ETAʹ is effective in vitro, ex vivo and in vivo, which is why we chose this construct and its corresponding 425(scFv)-SNAP fusion protein as internal references in our experiments (Bruell et al. 2003, 2005; Hussain et al. 2011; Kampmeier et al. 2009, 2010; Niesen et al. 2014; Pardo et al. 2012). Several recombinant scFv-ETA-based ITs are currently undergoing clinical evaluation, some of which target members of the EGFR family (Becker and Benhar 2012; Kreitman 2006; Schrama et al. 2006). One example is scFv(FRP5)-ETA, which targets ErbB2/HER-2 and demonstrates IC50 values of between ~10 pM (