Efficient growth suppression in pancreatic cancer PDX model by fully human anti-mesothelin CAR-T cells

Pancreatic cancer is a devastating disease ranked as the 4th leading cause of cancer-related deaths in the United States, and its incidence rate is increasing according to the latest statistics. The overall survival rates for patients with pancreatic cancer have not significantly improved over the past thirty years (Siegel et al., 2012; Simard et al., 2012). One of the reasons for the high mortality rates is the high resistance of pancreatic cancer to chemotherapy and radiation. Most patients are diagnosed at late stages of the disease. Approximately 15%–20% of patients diagnosed with pancreatic cancer are eligible for surgical resection, and 85% of these patients eventually experience relapse and ultimately cancer-related death (Siegel et al., 2012). In recent years, increasing evidence indicates that the fibro-inflammatory stroma is a source of cellular and molecular components contributing to tumor progression and metastasis (Feig et al., 2012; Waghray et al., 2013). Importantly, increased levels of stroma are positively related to a poor prognosis (Erkan et al., 2008). Despite the broader understanding of pancreatic cancer biology, gemcitabine, a chemotherapeutic approved for pancreatic cancer treatment approximately twenty years ago, still remains the standard of care (Burris et al., 1997). Thus, the development of novel treatment strategies for this devastating disease is urgently needed. Immunotherapy based on T cells modified with a chimeric antigen receptor (CAR) has been demonstrated to be a promising strategy for cancer treatment. CAR T cells specifically recognize tumor-associated antigens and eliminate tumor cells in a non-major histocompatibility complexrestricted manner. Several pilot clinical trials using CAR T cells have recently been reported to have promising clinical outcomes, even in solid tumors (Brown et al., 2016; Kershaw et al., 2013). Mesothelin (MSLN) is a membrane protein that is overexpressed in many cancer types, including pancreatic cancers, and is expressed only at low levels on normal peritoneal, pleural, and pericardial mesothelial surfaces (Chang and Pastan, 1996). Previously, several types of MSLN-targeted CAR-T cells were developed and have been found to have impressive antitumor activities in mesothelioma and ovarian cancer models (Carpenito et al., 2009; Lanitis et al., 2012). However, there are no reports on the antitumor activities of anti-MSLNCAR-T cells toward pancreatic tumor xenograft models. No study has yet examined the use of CAR T cells in PDX models of pancreatic cancer. Therefore, it is necessary perform a preclinical evaluation of novel CAR T cells as a treatment for pancreatic cancer in PDX models. In this study, we developed a novel fully human anti-mesothelin antibody. To investigate the binding properties of anti-MSLN antibody, we fist established the MSLN-overexpressed cell lines CHO-K1-MSLN and PANC-1-MSLN. The expression of mesothelin in these two established cell lines was confirmed by Western blotting (Fig. 1A). The fully human anti-MSLN antibody was screened from a fully human naïve antibody library by using phage display technology. The binding specificity of the anti-mesothelin antibody was tested on CHO-K1-MSLN and PANC-1-MSLN cells. The scFv proteins of anti-mesothelin antibody were produced transiently in FreeStyleTM 293F cells and purified by protein A affinity chromatography (Fig. S1). The results in Fig. 1C indicated that P1A6E and P3F2 scFv bound specifically to MSLN-expressing cells but not to cells without MSLN expression. Additionally, we compared the two fully human antibodies P1A6E and P3F2 with the SS1 and C10 antibodies. SS1 and C10 have a high binding affinity to mesothelin (Chowdhury and Pastan, 1999), and SS1 has also been found to be safe in patients when administered as a recombinant immunotoxin (Hassan et al., 2007). The results indicated that P1A6E and P3F2 had a significantly higher binding affinity than SS1 and C10 to MSLN-expressing cells (MFI value in PANC-1-MSLN cells: scFvP1A6E: 327.5, scFv-P3F2: 308.8, scFv-SS1: 48.9 and scFvC10: 46.8; MFI value in CHO-K1-MSLN cells: scFv-P1A6E: 452.3, scFv-P3F2: 445.1, scFv-SS1: 65.5 and scFv-C10: 80.2). The mean fluorescence intensity (MFI) of different scFv proteins bound cells as determined by flow cytometric analysis is shown in Fig. 1B. To test the affinity of antibody binding to mesothelin, we used Biacore Surface Plasmon resonance (SPR). The binding sensorgrams were collected at 25°C. The data were double-referenced by using


Selection of anti-mesothelin antibodies from the phage library
The fully human anti-MSLN antibody was screened from a fully human naï ve antibody library by using phage display technology. The recombinant mesothelin was biotinylated with EZ-LinkTM sulfo-NHS-Biotin (Pierce, Rockford, IL) according to the manufacturer's instructions. The biotinylated antigen (final concentration 10 -7 M) was incubated with 10 12 phage particles (a naï ve scFv library) in 1 ml 1% BSA for 40 min on a shaker at room temperature. The phage-antigen complex was captured in avidin-coated or streptavidin-coated Maxisorp wells (Thermo Fisher Scientific, Waltham, MA). The phage-antigen mix was distributed in eight wells and incubated for 20 min on a shaker at RT. The wells were rinsed 10 times with phosphate-buffered saline (PBS) 0.1% Tween-20 and 10 times with PBS. Bound phage was eluted by incubation with 200μl/well 100mM triethylamine for 5-10 min. The eluted phage was used to infect exponentially growing Escherichia coli TG-1.

Expression and purification of P1A6E and P3F2 scFv-Fc fusion protein
To express P1A6E and P3F2 scFv-Fc fusion proteins in FreeStyle™ 293F cells, two expression vectors were constructed separately. Primers V5-P1A6E-F (5'-ACAGTGCTAGCACAGGTACAGCTGGAACAG-3') and V5-P1A6E-R (5-'TTGTCGGATCCACCTAGGACGGTGACC-3') were used to amplify the scFv fragment of P1A6E from a phagemid, and primers V5-P3F2-F (5'-ACAGTGCTAGCACAGATGCAGCTAGTGC-3') and V5-P3F2-F (5'-TTGTCGGATCCACGTTTGATCTCCAGC-3') were used to amplify the scFv fragment from a phagemid. The scFv fragment was digested with NheI and BamHI and cloned into the expression vector pCMV-V5-Fc, which contains the Fc (hinge+CH2+CH3) part of IgG1 (purchased from Shanghai raygene biotechnology). FreeStyle™ 293F cells were transfected with the expression vector according to the manufacturer's instructions. The cells were incubated at 37℃ with 8% CO 2 for 6-7 days. ScFv-Fc fusion proteins were collected and the cell culture supernatants were purified by protein A affinity chromatography. The control antibodies SS1 and C10 (US7081518B1) were expressed and purified through the same method.

Surface plasmon resonance
Kinetic measurements were performed by using surface plasmon resonance with a Biacore T200 TM . The 1×HBS-EP+ buffer was chosen as the running buffer. Approximately 1000 response units (RU) anti-human Fc antibody (GE, #BR100839) were immobilized on a CM5 sensor chip through standard amine coupling. Purified scFv-Fc was injected at predetermined concentrations to achieve 100 RU for the captured scFv-Fc. Association and dissociation rate constants were determined by injection of a concentration range of the recombinant mesothelin at a constant flow rate (10μl/min). The channel was regenerated by injection of 3 M MgCl 2 over 2 min.

Western blot analysis
To confirm the levels of mesothelin expression in cell lines of stablely transfected human mesothelin protein, 3 × 10 6 cells were lysed in 200µl lysis buffer for 60 min on ice. Cell lysate was then removed by centrifugation at 12,000 × g for 10 min. Each sample was denatured under reducing conditions and electrophoresed by 12% SDS-PAGE. The samples were then transferred to a nitrocellulose membrane (Bio-Rad, Hercules, CA) and immunoblotted with a mouse monoclonal antibody mesothelin (K1) (Santa Cruz Biotechnology, Santa Cruz, CA). GAPDH was used as a loading control. The blots were incubated with horseradish peroxidase-conjugated anti-mouse IgG (Kangchen Biotech, Shanghai, China) and detected using the ECL western blot analysis system (Pierce, Thermo Scientific, Rockford, IL) in accordance with the manufacturer's instructions.

Mesothelin-specific CARs construction
The sequence encoding the MSLN-scFv antibody in the VL-VH orientation was obtained by polymerase chain reaction (PCR) amplification from a plasmid encoding the scFv-P3F2-Fc. As shown in Fig. 2A, the MSLN-28Z CAR contained the human CD8α signal peptide, followed by the MSLN -scFv linked in-frame to the hinge domain of the CD8a molecule, the transmembrane region of the human CD28 molecule, and the intracellular signaling domains of the CD28 and CD3ζ molecules. The MSLN-BBZ CAR contained the human CD8α signal peptide, followed by the MSLN-scFv linked in-frame to the hinge domain of the CD8a molecule, the transmembrane region of the human CD8 molecule, and the intracellular signaling domains of the CD137 and CD3ζ molecules. The MSLN-28BBZ CAR contained the human CD8α signal peptide, followed by the MSLN-scFv linked in-frame to the hinge domain of the CD8a molecule, the transmembrane region of the human CD28 molecule, and intracellular signaling domains comprising CD28, CD137, and CD3ζ. The nucleotide sequences of the human CD8α hinge and transmembrane region of the human CD28 molecule and intracellular signaling domains of CD28, 4 -1BB, and CD3ζ were assembled in-frame to produce three different configurations referred to as 28Z, BBZ or 28BBZ. The 28Z, BBZ or 28BBZ fragments were produced by PCR amplification using a plasmid encoding the corresponding fragments (Gao et al., 2014) as a template. The scFv DNA fragments and 28Z, BBZ or 28BBZ fragments were then recombined using PCR. These fragments were designed to have a MluI site at the 5' end and a SalI site at the 3' end. The synthesized fragments were digested with MluI and SalI restriction enzymes (New England Biolabs, USA) and ligated individually into the similarly digested pRRLSIN.cPPT-GFP.WPRE vector plasmid. The sequence integrity of all vectors described in this paper was confirmed by DNA sequencing. The Mock construct was transduced using a pRRLSIN.cPPT-GFP.WPRE lentiviral vector.

Lentivirus production
The 293T cells were seeded at 6 × 10 6 per 10-cm dish before transduction. The next day, 293T-cells were transfected with pRRLSIN.cPPT-GFP.WPRE vector (mock) or the different recombinant expression vectors in addition to the lentiviral packaging plasmid pMDLg/pRRE, pRSV-Rev and envelope-expressing plasmid pCMV-VSV-G (from Addgene) by using a polyethylenimine-based DNA transfection reagent. The viral supernatants were harvested at 48 or 72 h after transfection. Then, the lentiviral particles were concentrated 30-fold by ultracentrifugation (Beckman Optima™ XL-100 K, Beckman, Germany) for 2 h at 28,000 rpm.

Transduction and culture of primary T cells
Peripheral blood mononuclear cells (PBMCs) derived from human donors were provided by the Shanghai Blood Center. PBMC cells were cultured in AIM-V medium (Invitrogen, Carlsbad, CA) with 2% human AB serum (Huayueyang Biotechnology, China) and recombinant human IL2 (Huaxin High Biotech, China). For the transduction of primary T cells, PBMCs were stimulated for 48 h with anti-CD3/anti-CD28 antibodies immobilized on tosyl-activated paramagnetic beads (Invitrogen, Carlsbad, CA) before infection. After stimulation, the T cells were transduced with lentivirus particles in 24-well plates coated with RetroNectin (TaKaRa, Japan). The transduced T cells were cultured at a concentration of 5×10 5 cells/ml in the presence of rhIL2 (300 IU/ml).

Flow cytometric analysis
To detect scFv binding to target cells, 2×10 5 cells were collected by centrifugation and stained with anti-mesothelin scFv-Fc antibodies (10μg/ml) for 1 h at 4℃ followed by FITC conjugated goat anti-human secondary antibody (Kang-Chen Bio-tech, Shanghai, China) in the dark for 45 min at 4℃.
To measure CAR expression, the various genetically modified T cells were detected by using a biotinylated anti-human-F(ab')2 fragment (1:50, Jackson) and incubated at 4℃ for 45 min. After being washed with FACS buffer, cells were then incubated with PE-conjugated streptavidin for 45 min at 4℃ (eBioscience, San Diego, CA). For analysis of the absolute number of human T cells, CD4 + and CD8 + T cells were quantified using TruCount tubes (BD Biosciences, San Jose, CA) as described in the manufacturer's instructions. Fluorescence was assessed using a BD FACSCelesta Flow cytometer and data were analyzed with FlowJo7.6 software.

Cytotoxicity assays in vitro
Two human pancreas cells were co-cultured with the genetically modified T cells at different effector: target ratios of 3:1, 1:1 and 1:3. After 18 h of culture, the level of released LDH in the supernatant was measured using a CytoTox 96 ® non-radioactive cytotoxicity Kit (Promega, Madison,WI) as previously described (Gao et al., 2014). This experiment was repeated three times with consistent results.

Cytokine release assay
Cytokine measurements were performed by co-culturing transduced T cells with cancer cells in 96-well culture plates. After co-culture for 24 h, the IFN-γ, TNF-α and IL-2 cytokines secreted by the genetically modified T cells stimulated by the target cells were measured using an ELISA kit according to the manufacturer's instructions (MultiSciences Biotechnology, Hangzhou, China).

Pancreatic cancer PDX tumor models
The immunocompromised mice engrafted with pancreatic cancer patient-derived xenograft were kindly provided from Crown Bioscience. The 6-8-week-old female, non-obese diabetic/severe combined immunodeficient (NOD/SCID) mice were housed and treated under specific-pathogen-free conditions and used for tumor engraftment. Briefly, the tumors were sliced into 3 mm × 3 mm × 3 mm fragments and inoculated subcutaneously on the right flank. When the tumor burdens were approximately 100 mm 3 , the mice were randomly separated into three groups (n = 6) and injected intravenously (i.v.) with different CAR-T cells (1×10 7 CAR-T cells/mouse) after lymphocyte depletion with cyclophosphamide (100mg/kg). The tumor dimensions were measured twice weekly with calipers, and the tumor volumes were calculated using the formula V =π/6 × (length × width 2 ), where the length is the greatest longitudinal diameter and the width is the greatest transverse diameter (Goldstein et al., 1995). Mice were euthanized when their body weight loss was greater than 20% of the initial weight, when they exhibited inability to ambulate, or when tumor ulceration was observed in the control groups. NOD/SCID mice were housed and treated according to protocols approved by the Shanghai Medical Experimental Animal Care Commission.

Immunohistochemistry
To assess the infiltration of human T cells into xenograft tumors, formalin-fixed, paraffin-embedded tumor tissues were immunostained using an anti-CD3 antibody (Thermo Fisher Scientific, Waltham, MA). A normal rabbit IgG served as an isotype control. The procedures were performed as previously described (Gao et al., 2014).
Briefly, after deparaffinization and rehydration, the sections were exposed to 3% H 2 O 2 in methanol to eliminate endogenous peroxidase activity. Bovine serum albumin (1%) was used to block the sections for 30 min at room temperature (RT). The primary rabbit anti-human CD3 monoclonal antibody was incubated overnight at 4℃. The sections were then washed with PBS and incubated with an HRP-conjugated goat anti-rabbit secondary antibody (Kangchen Biotech, Shanghai, China) for 45 min at RT. The sections were visualized using a diaminobenzidine staining kit (Tiangen Biotech, Beijing, China) and then counterstained with hematoxylin, dehydrated, cleared, mounted and photographed. DAB-immunostained sections were analyzed by bright-field microscopy using an Olympus microscope (OLYMPUS IX71, Japan). CD3 + cells were quantified by measuring the number of stained T-cells in each section from three mice in each group by using Image-Pro Plus software. The mean count of the three areas was taken and expressed as the absolute number of CD3+ cells per 0.95 mm 2 (×200 field).

Statistical analysis
All data are presented as the mean ± standard error of the mean (SEM). Data were analyzed using a two-way ANOVA with a Bonferroni post-test for multi-sample comparisons for multi-sample comparisons or Student's t test for two sample comparisons. GraphPad Prism 5.0 was used for the statistical calculations. A P value less than 0.05 was considered significant.

Figure S1. Expression and purification of scFv-P1A6E-Fc and scFv-P3F2-Fc.
ScFv-Fc proteins were produced transiently in FreeStyle™ 293F cells. After the cells were cultured for 6-7 days, proteins were harvested by centrifugation and the cell culture supernatant was purified by protein A affinity chromatography. The purified protein were analyzed by SDS-PAGE.