We and others have recently shown cisplatin resistance-related protein 9 (CRR9)/Cleft Lip and Palate Transmembrane 1-Like (CLPTM1L) to affect survival and proliferation in lung and pancreatic tumor cells. Our research has indicated that CLPTM1L affects multiple survival signaling pathways in tumor cells under oncogenic, genotoxic, and microenvironmental stress. We have confirmed the association of CLPTM1L with pancreatic cancer by demonstrating overexpression of CLPTM1L in pancreatic tumors and poor survival in patients with high tumor expression of CLPTM1L. Predicting a transmembrane structure, we determined that CLPTM1L could be targeted at the plasma membrane. Herein, we describe the development of mAbs targeting CLPTM1L. Lead antibodies inhibited surface accumulation of CLPTM1L, Akt phosphorylation, anchorage-independent growth, and chemotherapeutic resistance in lung and pancreatic tumor cells. Gemcitabine promoted a physical interaction between CLPTM1L and p110α in pancreatic tumor cells, which was inhibited by anti-CLPTM1L. In vivo treatment with anti-CLPTM1L robustly inhibited the growth of both lung and pancreatic adenocarcinoma xenografts. The efficacy of anti-CLPTM1L correlated with specific epitopes representing important targets in human cancers, particularly those driven by KRas, for which effective targeted therapies have been elusive. This study is the first to report cell-surface exposure of the tumor survival protein CLPTM1L and inhibition of the function of surface CLPTM1L with novel, systematically developed inhibitory mAbs establishing proof of concept of clinically practical agents inhibiting this compelling new tumor survival target in cancer. Mol Cancer Ther; 15(5); 985–97. ©2016 AACR.
The CRR9/Clptm1l gene resides within a locus at chromosome 5p15.33 that is associated with risk of non–small cell lung cancer, pancreatic cancer, and a variety of other cancers (1–11). CLPTM1L is commonly overexpressed in multiple tumor types (refs.12–14; Oncomine Compendia Bioscience). There is increasing evidence for tumor-specific CLPTM1L expression. In fact, increased CLPTM1L transcript expression was found to be a highly specific marker of laryngeal carcinoma using microarray analysis (13). In addition, an in-frame fusion of the 297 amino acid N-terminus of CLPTM1L with the GLIS3 in a rare pediatric case of hepatocellular carcinoma was found to be oncogenic (15). Little has been reported regarding the expression of CLPTM1L in normal tissues. However, we have shown that CLPTM1L expression in normal lung epithelium is low relative to that in tumor tissues (16). CLPTM1L expression was characterized as weak in 30% to 40% of normal pancreatic ductal epithelial tissues, with expression being undetectable in the remainder of normal tissues in one study (14). In addition to tumor-specific expression, CLPTM1L functions in several tumor-specific pathologic processes. In particular, we previously found CLPTM1L to confer resistance to chemotherapy and survival of anoikis in the context of anchorage-independent tumor growth (12). We found that the role of CLPMT1L in chemoresistance was not limited to cisplatin, as resistance to Topoisomerase I inhibition was also dependent on CLPTM1L. Our laboratory was the first to demonstrate a role for CLPTM1L in tumorigenesis. Our previous results demonstrated that CLPTM1L was necessary for Ras-induced Akt phosphorylation in association with an interaction between PI3K and CLPTM1L in mouse fibroblasts. CLPTM1L blockade also caused decreased expression of Bcl-xL, which was independent of Akt status (16). Anchorage-independent growth inhibition by CLPTM1L blockade in these cells was dependent on inhibition of both Akt phosphorylation and Bcl-xL expression.
CLPTM1L is predicted to have 6 hydrophobic helical domains as indicated by Uniprot annotation, indicating membrane localization. It was previously suggested that CLPTM1L may exert its survival function in the mitochondria (17). Glycosylation of CLPTM1L has previously been overlooked, and given its role in the intrinsic apoptosis pathway, a focus on the mitochondria has resulted in a failure to recognize the presence of CLPTM1L on the cell surface by previous studies. However, recent studies in pancreatic tumor cells and the current study provide strong evidence that CLPTM1L resides predominantly in the endoplasmic reticulum and on the cell surface (14). It is unknown whether cell surface-associated CLPTM1L plays a specific and critical role in survival signaling or if it can be targeted therapeutically. We have shown CLPTM1L to be critical for Ras transformation and tumorigenesis, making this signaling pathway a novel potential way to target Ras-driven cancers. Therefore, this study focuses on lung and pancreatic cancers, of which approximately 35% and 90% are driven by K-Ras, respectively.
Here, we present the first evidence for cell-surface exposure of CLPTM1L in lung and pancreatic tumor cells and demonstrate an interaction of CLPTM1L with p110α in tumor cells. We demonstrate the first proof of concept of clinically practical anticancer agents inhibiting the function of surface CLPTM1L with systematically developed inhibitory monoclonal antibodies. These novel agents inhibit p110α–CLPTM1L interaction and Akt phosphorylation, inhibit tumor cell growth and anchorage independence, and sensitize lung and pancreatic tumor cells to killing by cisplatin and gemcitabine, respectively. Furthermore, we demonstrate effective inhibition of the growth of established adenocarcinomas in in vivo models. Our findings provide strong justification for investigation of CLPTM1L-targeting antibodies as therapeutics and chemosensitizers for human cancers.
Materials and Methods
Cell culture and reagents
Panc1, MiaPaCa, A549, H838, HeLa, U251, GBM4, and Beas-2B cells were obtained from ATCC or authenticated by DNA typing of STR and microsatellite loci and comparison with ATCC reference profiles within 6 months of experiments. Primary human pancreatic adenocarcinoma cell lines MCW462 and MCW670 were established at the Surgical Oncology Biorepository at Medical College of Wisconsin (MCW; Milwaukee, WI) and maintained in DMEM/F12 with 6% FBS and supplements. Cell lines were generated from heterotopic murine xenografts established from primary and metastatic human pancreatic cancer (PC) specimens. Cell lines were established after enzymatic digestion of the xenografts. Mouse CD326- MHC Class I+(H-2Kd) cells were eliminated from the cell lines by flow cytometric FACS sorting using human-specific CD326 (EpCAM) and murine-specific MHC Class I (H-2Kd) antibodies (eBioscience). Short tandem repeat (STR) profiling was performed using seventeen STR loci plus the gender determining locus using the commercially available PowerPlex 18 D Genetic Analyzer. Data were analyzed using GeneMapper ID-X v1.2 (Applied Biosystems). Samples did not match any cell line in either the ATCC database. Cell lines were characterized by IHC for epithelial (CK19) and pancreatic (PDX-1) markers, doubling time, colony-forming efficiency, and in vivo tumorigenicity. Mutations in KRAS and TP53 were assessed using Sanger sequencing. Both cell lines harbor KRAS G12A mutations. Human lung adenocarcinoma cell lines (A549 and H838) were cultured in RPMI-1640 plus 10% FBS (Life Technologies). Beas-2B cells were cultured in LHC-8 media plus epinephrine (Life Technologies). Panc1 cells were grown in DMEM/F12 media with 10% FBS. Cisplatin and gemcitabine were purchased from Sigma-Aldrich and prepared immediately before use in 5 and 50 mmol/L aqueous stock solutions, respectively. Polyclonal anti-CLPTM1L (ab155119, Abcam) was used in polyclonal anti-CLPTM1L inhibition studies. Antibody diluent as described by Abcam was used as a vehicle control for polyclonal antibody treatment where indicated to account for any effect of diluent constituents. Normal mouse IgG was used as a nonspecific antibody control for monoclonal antibody treatment where indicated. Rabbit α-HA (Santa Cruz Biotechnology) was used as a nonspecific antibody control for polyclonal antibody treatment where indicated. Mouse α-HA (Cell Signaling Technology) was used as a nonspecific antibody control for experiments with purified monoclonal antibodies, and mouse α-human Von Willibrand Factor (hVWF) ascites was used for experiments with monoclonal ascites. Monoclonal antibody production was contracted to Biomatik Corporation. Polyclonal antibodies provided preliminary results and the use of polyclonal, monoclonal ascites, and purified monoclonal antibody was dictated by appropriate characterization of each antibody and secondarily by availability. The shRNA depletion of CLPTM1L with shCLP, as well as the specificity and efficacy of knockdown with this vector were previously validated and described (12, 16).
IHC staining was performed on the Dako Autostainer Plus using the Dako EnVision FLEX High pH Detection Kit (K8010; Dako). Slides were deparaffinized to DI water. Antigen Retrieval was performed on Dako PT Link water bath. The antigen retrieval was done at 97°C for 20 minutes. The slides were cooled until they reached 65°C. All slides for all antibodies were placed in Tris/EDTA pH 9 (Dako TRS High pH). Slides were washed in Dako wash buffer for 5 minutes. Slides were subjected to a peroxidase Block for 5 minutes. Slides were rinsed twice with wash buffer. Slides were incubated with primary antibody CLPTM1L (rabbit polyclonal, Sigma Aldrich cat# HPA014791, lot A57952) diluted to 1:400 for 30 minutes. Slides were rinsed with wash buffer. Slides were incubated with secondary antibody for 20 minutes and rinsed twice with wash buffer. Slides were incubated with DAB substrate for 10 minutes and rinsed with wash buffer. Slides were stained with hematoxylin for 7 minutes and rinsed with DI water. Slides were dehydrated and coverslipped for viewing. Staining intensity in tumor tissue was scored by three pathologists and averaged. Tumors were scored as negative – 0, weak – 1, intermediate – 2, or strong – 3. Independent scores were averaged. Antibodies for IHC on xenograft tumors were Ki-67 (CRM 325 A,B,C) and cleaved caspase-3 (Asp175, Cell Signaling Technology) 1:300.
Twenty-five pancreatic cancer patients with IHC CLPTM1L expression data on treatment naïve tumors were divided into tertiles according to CLPTM1L staining intensity in tumor tissues, which resulted in division into groups with average staining intensities of <1.15, 1.15–1.75, and >1.75 in low, intermediate, and high expression groups, respectively. Differences in survival and clinicopathologic variables between expression groups were analyzed using STATA software (StataCorp) χ2 tests.
Cell-surface protein isolation
Proteins exposed on the cell surface were biotinylated and isolated using the Pierce Cell Surface Protein Isolation Kit according to manufacturer's instructions (Pierce Biotechnology). Briefly, subconfluent cells in T75 flasks were washed with cold PBS and treated with sulfo-NHS-SS-Biotin in PBS for 30 minutes at 4°C. The reaction was stopped with quenching buffer and cells were scraped into solution in a 50-mL conical tube. Cells were pelleted, resuspended and washed in TBS, and repelleted. Cells were lysed in the presence of protease inhibitors and sonicated. Lysates were incubated on ice with periodic vortexing for 30 minutes and clarified by centrifugation. Biotinylated proteins were captured and washed in a NeutrAvidin Agarose loaded column. Proteins were eluted with 500 μL of SDS-PAGE Sample Buffer (62.5 mmol/L Tris HCl, pH 6.8, 1% SDS, 10% glycerol) with 50 mmol/L DTT for 60 minutes at room temperature. Lysates were analyzed by Western blotting as described. CLPTM1L antibody precipitated Panc1 cell lysate (CLP +Ct) and a whole-cell Panc1 lysate (Lysate +Ct) were used as positive controls. Surface-associated EGFR and nucleus-associated Cyclin D1 were blotted for as loading and surface specificity controls, respectively.
Cells were grown on tissue culture-treated slides, washed with PBS, and fixed with 4% formaldehyde at room temperature for 30 minutes. Cells were gently washed twice with PBS for 2 minutes. Cells were blocked with 10% FBS in PBS for 1 hour at room temperature. Slides were incubated with the indicated primary antibodies (1:100–1:250) in PBS with 10% FBS overnight at 4°C. Slides were washed three times for 5 minutes with PBS with gentle shaking. Slides were incubated with secondary Alexa Fluor 594 goat anti-mouse, Alexa Fluor 488 goat anti-rabbit (Molecular Probes, Inc.), or anti-mouse Cy2 antibodies (1:500) at room temperature for 1 hour protected from light. Slides were washed three times for 5 minutes with PBS. For plasma membrane staining, wheat germ agglutinin (WGA) conjugated with Alexa Fluor 594 (ThermoFisher Scientific) was added to the slides for 10 minutes and slides were washed with PBS per manufacturer's instructions. Thirty microliters of SlowFade Gold with DAPI anti-fading agent (Molecular Probes, Inc.) was added to the slide, over which a coverslip was laid. Signal was observed using fluorescence and confocal microscopy on an Olympus Fluoview FV1000 confocal laser scanning microscope (Olympus Life Science Europa GmbH) with a UPLSAPO 20× (dry, NA: 0.75) objective.
Panc-1 cells were harvested, centrifuged, and resuspended in cold FACS buffer (2% FCS in PBS) at a density of 1 × 106/mL. Samples (200,000 cells) were incubated for 2 hours with 1:50 anti-CLPTM1L (5-2) ascites at 4°C. Cells were washed with 1 mL FACS buffer three times and incubated with 1:100 anti-mouse-PE (R&D Systems) for 1 hour at 4°C. After washing three times with 1 mL FACS buffer, analysis was carried out on FACS Calibur cytofluorimeter using CellQuest software (Beckton Dickinson).
Cells were lysed with 100 μL of 1× NP40 lysis buffer containing proteinase inhibitors, sheared 10 times with a 28-gauge needle, spun at 16,000 × g for 30 minutes, normalized by protein concentration as determined by the Bradford method, and the supernatant boiled for 5 minutes. Twenty microliters of normalized lysate were resolved by SDS-PAGE and immunoblotting analyzed with indicated antibodies. The following antibodies were used: rabbit anti-CLPTM1L (Sigma-Aldrich), mouse anti-Actin (Santa Cruz Biotechnology), anti–Bcl-xL-rabbit Bcl2L1 (Abcam), anti-AKT, anti-pAKT (Thr308), anti- EGFR, and anti-Cyclin D1 (Cell Signaling Technology). Quantitation of Western blot analyses of three independent cultures was done using ImageJ software (NIH, Bethesda, MD).
Subconfluent cells plated in triplicate at equal densities on 12-well tissue culture plates were treated with antibodies and/or cisplatin at the indicated concentrations for 24 or 48 hours. One hundred microliters of CellTiter 96 Aqueous One-Solution Cell Proliferation Assay reagent (MTS, Promega) were added to 0.5 mL of media on the growing cells. The assay was incubated for 3 hours at 37°C and 5% CO2. The media and MTS solution were mixed with a pipette and 200 μL was transferred to a 96-well assay plate for measurement of absorbance at 490 nm.
AminoLink Co-immunoprecipitation Kit (Pierce Biotechnology) was used according to the manufacturer's instructions for coprecipitation of proteins from tumor cell lysates using the indicated antibodies. Polyclonal antibodies α-p110α (Cell Signaling Technology) and α-CLPTM1L (Sigma Aldrich) were covalently linked to resin for immunoprecipitation. Eluted proteins were analyzed by Western blotting as described.
Cells were plate at equal density on 24-well tissue culture plates and treated as indicated. Cells were treated with purified 5 μmol/L monoclonal anti-CLPTM1L (5-2) or control antibody for 24 hours followed by 72 hours of treatment with 200 μmol/L gemcitabine or vehicle control (sterile distilled water). Dead cells were fluorescently stained with Yoyo-1 (Life Technologies) and enumerated on an Incucyte FLR live cell imager (Essen Bioscience). Total cell numbers were then enumerated on the imager by staining with Vybrant Dye Cycle Green (Life Technologies). A killing index was calculated for each well by dividing the number of dead cells by the number of total cells. The killing indices of triplicate groups were averaged.
Soft agar assay
For anchorage-independent growth, cells were suspended in 0.4% agarose in complete growth media and plated over 0.8% bottom agar at 10,000 cells per well of a 6-well tissue culture dish in triplicate. Cells were fed twice a week over 4 weeks in culture, and colonies were stained using cell staining reagent and protocol from Millipore's Cell Transformation Detection Assay. Images of wells were captured and analyzed by ImageJ software to count colonies. Average counts per well were determined. P values were determined using a two-tailed Student t test.
Colony formation assay
Panc1 cells were plated at a density of 1,000 per well on 6-well tissue culture plates in triplicate for each treatment group. Cells were treated with the indicated antibodies (5 μg/mL). Colonies were methanol fixed and stained after 7 days in culture with 0.5% crystal violet and counted using ImageJ software using a particle (colony) size cutoff of 4 square pixels (14, 18). Average counts per well were determined.
Tumor spheroid assay
Cells were counted and plated (2,500 cells/well) on ultra-low attachment round-bottom tissue culture plates (Corning Life Sciences, Inc.). Plates were centrifuged at 1,000 g for 5 minutes and placed in an Incucyte FLR live cell imager (Essen Biosciences) for monitoring of tumor spheroid size using confluence as a measure.
A549 xenograft assay
Tumor cells were cultured, counted, and resuspended in sterile PBS at a concentration of 1 × 107 cells/mL. A volume of 200 μL (2 × 106 cells, respectively) was injected subcutaneously into the flanks of athymic nude mice at an age of 6 weeks. Mice were kept in individual ventilated cages to avoid infection. All mice were fed a commercial sterile diet and water ad libitum and were housed in an animal facility under a 12-hour light/dark cycle at constant temperature and humidity. Five mice per treatment group were treated with 1 mL/kg of the indicated ascites intraperitoneally in 200 μL of sterile PBS on days 0, 5, and 12 after growth of tumors to an average of 30 mm3 and randomization into groups. The health of these mice was monitored three times weekly and tumor sizes were measured periodically using digital calipers until sacrifice. Tumor volumes were calculated by the formula for an ellipsoid, 4/3πrlrwrh. P values were determined by one-way ANOVA with Tukey HSD posttest where K>2. Animal weights were measured throughout the study to monitor toxicity. Significance of differences between groups was determined by two-tailed Student t-test. Mice were euthanized and tumors excised and frozen for further analysis. Protein was isolated from 30 mg of tumor tissue using a Tissuelyser LT (Qiagen) and 300 μL of RIPA buffer with HALT protease/phosphatase inhibitor (Life Technologies).
Panc1/MiaPaCa xenograft assay
Twelve-week-old female NOD CB17-Prkdcscid/NcrCrl mice (Charles River) were kept in individual ventilated cages to avoid infection. All mice were fed a commercial sterile diet and water ad libitum and were housed in an animal facility under a 12-hour light/dark cycle at constant temperature and humidity. For the induction of rodent model of human pancreas tumors, 5 × 106 Panc1 or MiaPaCa human pancreatic ductal adenocarcinoma cells were injected in a 100-mL volume of PBS subcutaneously into the right flank of the mice. Fifteen days after cancer cell inoculation, mice were randomized and treated once weekly intravenously through the tail vein for 6 weeks. Tumors were measured by calipers every 3 days. Two different groups, each containing 9 mice, were studied: (i) ascites control (2.5 mL/kg) and (ii) ascites containing antibody 5-2 (2.5 mL/kg). For MiaPaCa studies, 2.5 mg/kg purified control or indicated antibody was administered intravenously as described. The study was performed according to the Institutional and National Animal Experimentation and Ethics Guidelines under ethical clearance (XXIX./3610/2012). Animal weights were measured throughout the study to monitor toxicity. Significance of differences between groups was determined by two-tailed Student t test. Mice were euthanized and tumors excised and frozen for further analysis. Protein was isolated from 30 mg of tumor tissue using a Tissuelyser LT (Qiagen) and 300 μL of RIPA buffer with HALT protease/phosphatase inhibitor (Life Technologies). Panc1 tumor samples (collected 8 days after last anti-CLPTM1L treatment) were paraffin embedded. HE and IHC staining were carried out.
CLPTM1L is overexpressed in lung and pancreatic tumors and associated with poor clinical outcome
Having previously demonstrated overexpression of CLPTM1L in lung adenocarcinomas compared with that in adjacent normal lung tissue (12), we employed IHC analysis to evaluate its expression in human pancreatic ductal adenocarcinoma and confirm its expression in human lung tumors. Staining was uniformly negative in normal ductal epithelia (n = 16), while positive in 90% (28/31) of tumor cells (r = 0.78, P < 5 × 10−10; Fig. 1A). Staining intensity, while variable between patients, was highly consistent within the tumor tissue of individual patients. Therefore, distribution of staining was invariably 100% and was not a factor in calculating positivity. Negative tissues surrounding the positively stained tumor cells in these sections are tumor stroma. Tumors had an average staining score of 1.46, whereas that of normal ductal epithelium was 0. Tumors expressing low levels of CLPTM1L as defined by an average score of 1.15 or less were associated with increased overall survival, whereas patients with higher expression had shorter median overall survival times. Median overall survival among all patients was 18.5 months. At that time point, the survival rate of patients with low expression in tumor tissues was significantly higher (P = 0.0294) at 86%, compared with 44% and 33% in intermediate- and low-expressing groups. Patients with low CLPTM1L expression had a median overall survival time of 31 months, whereas patients with intermediate and high expression had median survival times of 17 and 14 months, respectively (Supplementary Table S1). χ2 testing verified that clinicopathologic features of age, race, gender, and disease stage did not affect expression. Survival studies excluded 6 patients who had received neoadjuvant therapy. Representative tissues stained with CLPTM1L antibody are shown in Fig. 1B, showing positivity in tumor over normal tissues. Similarly, human lung adenocarcinoma tissues were positive for CLPTM1L expression by IHC (Fig. 1C). Demographic information and clinical characteristics of the patients in this study are detailed in Supplementary Table S1.
Cell-surface localization of CLPTM1L in lung and pancreatic tumor cells
Immunofluorescence staining of nonpermeabilized A549 cells for CLPTM1L revealed a punctate cell-surface staining pattern in both pancreatic adenocarcinoma cells and in lung adenocarcinoma cells (Fig. 1D). To confirm the localization of CLPTM1L at the surface of tumor cells, we isolated cell-surface proteins from A549 human lung tumor cells by biotinylation and extraction with a streptavidin resin. Western analysis of whole-cell and cell-surface lysates demonstrated a prevalence of CLPTM1L at the cell surface (Fig. 1E). EGFR, as a cell surface-localized control protein, was detected at the cell surface. The ratio of signal for cell surface CLPTM1L to that for total protein was greater than that of EGFR, suggesting that a larger fraction of CLPTM1L was localized to the cell surface in these cells. Cyclin D1, used as a nuclear protein control, was not detected at the cell surface. We confirmed surface exposure of CLPTM1L in Panc1 cells by labeling of live cells with purified anti-CLPTM1L mAb and a phycoerythrin-conjugated secondary antibody. Cells labeled with anti-CLPTM1L were positive for fluorescence with mean fluorescence intensity of 341 units compared with 149 in control-labeled cells (Fig. 1F) confirming the presence of surface-exposed CLPTM1L in live pancreatic tumor cells.
Transmembrane structure prediction of CLPTM1L
CLPTM1L is predicted to be a serpentine transmembrane protein with 6 transmembrane domains by analysis using a hidden Markov model (TMHMM - http://www.cbs.dtu.dk/services/TMHMM/; Supplementary Fig. S1A). Protein disorder prediction using IUPred (http://iupred.enzim.hu) supports this model and indicates a disordered region within the large predicted extracellular region near the N-terminus of the protein (Supplementary Fig. S1B). Areas of high order are predicted to contain helical transmembrane regions, and those of the highest disorder are predicted to be without secondary structure. Areas on either side of the disordered region (residues 141-162) are predicted extracellular globular domains. We have designated these domains extracellular region 1 (ECR1) and extracellular region 2 (ECR2). Figure 2A represents the overall structural prediction for CLPTM1L including transmembrane domains, extracellular globular domains, the disordered region, and N-glycosylated residues.
Design and development of anti-CLPTM1L mAbs
To effectively target CLPTM1L, we developed a panel of mAbs against exposed epitopes of the protein. A target region of the CLPTM1L protein was defined as the hydrophilic loop region that is likely to be exposed on the cell surface in the protein's native form (residues 32–264), excludes the identified disordered region (residues 141–162), and excludes N-glycosylation sites (residues 91, 101, and 229). The hydrophilic loop region has been found to be necessary for CLPTM1L function (14). A proprietary algorithm was utilized to select epitopes with minimal similarity to other proteins, hydrophilicity, surface orientation, and flexibility (Supplementary Fig. S2). mAbs were raised in balb/c mice by immunization with peptides representing these epitopes, generation of hybridoma clones, collection of ascites, and Protein A affinity purification. The resulting ten antibodies targeted epitopes designated as epitopes 1 through 5. Epitopes 1 and 2 reside within ECR1, while epitopes 3, 4, and 5 reside within ECR2. These epitopes are represented in Supplementary Table S2. Aliquots of these mAbs were used for in vitro and in vivo screening. The three best performing candidates (1-1, 5-2, and 5-3) in terms of inhibition of survival signaling, cisplatin resistance, anchorage-independent growth, and tumorigenesis were selected for propagation of hybridoma cell lines and purification of antibodies.
Our panel of 10 mAbs targeting the extracellular globular domains of CLPTM1L was tested for detection of CLPTM1L and immunofluorescent labeling of surface CLPTM1L. mAbs labeled plasma membrane-localized CLPTM1L in nonpermeabilized Panc1 cells (Fig. 2B). All antibodies resulted in a cell-surface staining pattern when used for immunofluorescence on nonpermeabilized A549 cells, with antibodies targeting ECR2 exhibiting a more distinct punctate staining in cisplatin-treated A549 cells (Fig. 2C). mAbs strongly detected a single 62-kDa CLPTM1L protein band by Western blot analysis corresponding to the major band detected with polyclonal anti-CLPTM1L (Fig. 2D, left). The antibody clone 1-1 blot included two faint higher molecular weight bands. No additional bands were detected. In addition, the single band detected by mAbs was diminished upon specific depletion of CLPTM1L with shRNA, demonstrating the specificity of the antibodies for CLPTM1L (Fig. 2D, center). Specificity of the polyclonal antibody and lead mAb 5-2 was confirmed by Western blotting in cells expressing endogenous CLPTM1L and those also expressing a FLAG-tagged exogenous CLPTM1L (Fig. 2D, right).
Anti-CLPTM1L inhibits surface accumulation and chemotherapy-induced interaction of CLPTM1L with p110α
With our previous evidence that CLPTM1L interacts with the catalytic subunit of PI3Ks in fibroblasts, we sought to determine whether CLPTM1L interacts with p110α in tumor cells, and whether that interaction is affected by anti-CLPTM1L treatment. Given the role of CLPTM1L in chemoresistance, we also sought to determine whether gemcitabine treatment affected this interaction in human pancreatic adenocarcinoma cells. Coimmunoprecipitation assays were conducted in Panc1 cells. Gemcitabine caused a modest increase in the amount of CLPTM1L that coprecipitated with p110α in these cells, suggesting that the interaction between these proteins may be enhanced upon treatment (Fig. 3A). The increase in coprecipitation of CLPTM1L and p110α was not observed upon treatment with monoclonal α-CLPTM1L (5-2) or (1-1). To determine whether cell-surface accumulation of CLPTM1L was diminished by treatment with anti-CLPTM1L, which occurs through internalization with other antagonistic immunoglobulin drugs, we conducted biotinylation of surface proteins followed by isolation and Western blotting. Cells were treated with anti-CLPTM1L (5-2) ascites or control ascites (1:1,000) for 96 hours. Anti-CLPTM1L treatment resulted in downregulation of surface-associated CLPTM1L (Fig. 3B). As evidenced in Fig. 3A, levels of CLPTM1L in CLPTM1L immunoprecipitated and whole-cell lysates were not appreciably diminished upon anti-CLPTM1L treatment, suggesting that downregulation was cell surface specific. Masking by the treatment mAb is not expected as detection was performed using a polyclonal antibody raised against the larger residues 75–210 of the protein which does not include the monoclonal epitope.
Inhibition of AKT phosphorylation and chemotherapeutic resistance with anti-CLPTM1L antibodies
We previously demonstrated that targeting CLPTM1L in tumor cells modulates Bcl-xL and Akt survival signaling and sensitizes cells to killing with cisplatin (12, 16). In the current study, A549 and H838 human lung tumor cells were treated with monoclonal anti-CLPTM1L ascites at 1:5,000 and 1:1,000 dilutions. After culture for 48 hours with treatment, Western blotting for Bcl-xL, total Akt1, and phospho-Akt1 was conducted. Decreased accumulation of Akt1 with polyclonal anti-CLPTM1L treatment was observed in human lung tumor cell lines (Fig. 4A). To determine the effect of treatment with polyclonal antibody on sensitivity to cisplatin, cells were treated with control antibody or dilutions of anti-CLPTM1L from 0.01 to 1 μg/mL for 24 hours. Cells were then exposed to cisplatin at 0, 10, and 20 μmol/L concentrations for 48 hours. Treatment with anti-CLPTM1L resulted in a decrease in cell viability following cisplatin treatment that was dependent on the dose of antibody (Fig. 4B). These results demonstrate sensitization to cisplatin killing using anti-CLPTM1L. Likewise, treatment of pancreatic adenocarcinoma cells with anti-CLPTM1L 5-2 resulted in dose-dependent inhibition of Akt phosphorylation (Fig. 4C). Akt phosphorylation was also inhibited in MCW670 primary human pancreatic tumor cells upon treatment with anti-CLPTM1L (Supplementary Fig. S3A). Treatment of pancreatic adenocarcinoma cells with anti-CLPTM1L resulted in significantly increased killing by gemcitabine (P = 0.007; Fig. 4D). Anti-CLPTM1L treatment resulted in an approximately 2-fold increase in gemcitabine killing (20% vs. 10% with gemcitabine alone).
Treatment of A549 and H838 human lung tumor cells with our panel of mAbs resulted in depletion of Bcl-xL and Akt phosphorylation with a range of efficacies. Western blot analysis revealed that treatment with antibodies targeting epitopes 1 and 5 resulted in dose-dependent decreases in Bcl-xL and phospho-Akt1 accumulation (Supplementary Fig. S3B and S3C). Likewise, sensitization to cisplatin killing was accomplished by treatment of tumor cells with anti-CLPTM1L mAbs with a range of efficacies (Supplementary Fig. S3D). A summary of these results for each antibody clone is given in Supplementary Table S3.
Inhibition of both KRas and EGF-driven anchorage-independent growth and colony formation by anti-CLPTM1L
With the knowledge that inhibition of CLPTM1L can inhibit K-Ras driven tumor spheroid formation and tumorigenesis (16), we sought to determine whether antibodies targeting of CLPTM1L were inhibitory to the anchorage-independent growth of human lung tumor cells. Treatment of A549 and H838 lung tumor cells with 1 μg/mL anti-CLPTM1L significantly decreased the number of anchorage-independent colonies by 55% and 65%, respectively, compared with solvent control-treated cells (P<0.05; Fig. 5A). Treatments with 1 μg/mL antibody or control IgG were administered one day after plating and again one week later. EGF treatment induced anchorage-independent growth in Beas2B human bronchial epithelial cells, which was inhibited by 76% upon treatment with 1 μg/mL anti-CLPTM1L antibody (Fig. 5B). Colony formation of Panc1 cells in adherent tissue culture was inhibited by anti-CLPTM1L mAbs, particularly those targeting epitope 5 (P<0.005; Fig. 5C). Anti-CLPTM1L (1-1, 5-2, and 5-3) also robustly inhibited colony formation of A549 lung, GBM4, and U-251 glioblastoma, and HeLa cervical human tumor cells (Supplementary Fig. S4). To investigate the role of Akt signaling in colony formation inhibition by anti-CLPTM1L, Panc1 cells were stably transfected with a constitutively active myristoylated Akt (mAkt) or control vector. Expression of mAkt modestly rescued colony formation, with an average of 15 colonies per well with mAkt compared with six with vector control (Fig. 5D).
Inhibition of anchorage-independent growth of A549 cells by mAbs is summarized in Supplementary Table S3. Among all mAbs raised against the extracellular domains of CLPTM1L, only those raised against epitope 5 were effective in inhibiting anchorage-independent tumor spheroid growth of H838 cells (Supplementary Fig. S5; Supplementary Table S3).
Inhibition of the growth of established tumors in vivo by anti-CLPTM1L antibodies
To determine whether the growth of established human lung tumor xenografts could be inhibited using antibodies targeting CLPTM1L, we utilized an athymic nude mouse model of tumorigenesis. Mice bearing established subcutaneous A549 cell tumors were separated into groups of equal average tumor size and variance. mAbs recognizing epitope 5 in ECR2 were evaluated in xenograft models for inhibition of the growth of established human lung tumor xenografts. A549 cell xenografts were established and mice were separated into groups of equal average tumor size and variance. Mice were treated intraperitoneally with 1 mL/kg of the indicated mAb or control ascites on days 0, 5, and 12 postrandomization. Antibodies 5-2 and 5-3 had a robust and significant inhibitory effect on tumor growth (Fig. 6A). Monoclonal clone 5-2 inhibited tumor growth by 88% (P<0.01). In Panc1 human pancreatic xenograft models, lead candidate monoclonal anti-CLPTM1L 5-2 significantly inhibited tumor growth by 75% over 55 days posttreatment initiation (P<0.01; Fig. 6B). The difference in average tumor size was significant from day 21 posttreatment on. Similar results were obtained with purified mAb treatment in MiaPaCa human pancreatic adenocarcinoma xenografts with significantly decreased growth and regression of tumors as early as 7 days posttreatment initiation (Supplementary Fig. S6A). Panc1-xenografted mice were treated with 2.5 mL/kg antibodies on days 0, 7, 14, 21, and 28, postrandomization. Both control and treated Panc1 tumors had the same immunophenotype as observed by hematoxylin and eosin (H&E) stain (blind to pathologist; Fig. 6C). Treated tumor tissue was blindly determined by the pathologist to exhibit more widespread cell death and much less vital tumor tissue. IHC staining of tumor tissues revealed that positive staining for CLPTM1L was diminished in anti-CLPTM1L–treated mice. As determined by pathologist review, heterogeneity in the staining signal for CLPTM1L was due to technical variation within the sections with regard to thickness of the section, proximity to necrotic tissue, and vacuolation. Again, masking by the treatment mAb is not expected because IHC was performed using a polyclonal antibody raised against a different region of the protein. Accumulation of phosphorylated Akt in tumor tissues was ablated by anti-CLPTM1L treatment of both Panc1 (P < 0.005) and A549 (P < 0.05) xenografts (Fig. 6D). There were no significant differences in average animal weight from that of the control group indicating an absence of acute toxicity (Supplementary Fig. S6B).
A growing body of evidence has demonstrated a critical role for the tumor survival protein CLPTM1L in tumorigenesis. Our discovery that CLPTM1L resides on the surface of tumor cells reveals the potential targetability of this protein for preventive or therapeutic purposes. As previous publications have overlooked the presence of CLPTM1L at the cell surface, the use of nonpermeabilized cells for IF, isolation of surface-exposed proteins, or flow cytometry may be necessary methods for the determination of cell line and tissue specificity of surface expression.
To our knowledge, we are the first to report an association of CLPTM1L expression with poor outcome in pancreatic cancer patients. While significant, these findings are limited by the inherently small number of available tumor tissues and will require further validation with larger datasets.
Mutant KRas is the prevalent driver of pancreatic cancer, and KRas and Akt are predominant contributors to chemoresistance in pancreatic tumors (19). K-Ras mutations in lung tumors are associated with poor survival (20). Effective strategies to target K-Ras have been highly sought after. However, little success in targeting this prevalent oncoprotein has been had. Likewise, in lung cancer, targeted therapies exist for EGFR-driven tumors, but resistance remains an issue and no effective targeted therapy exists for KRas-driven tumors. Our results provide a potential approach for targeting Ras-driven cancers as well as cancers driven by other oncogenes. Although our mechanistic findings are not specific to K-Ras, it is apparent that there is a critical interaction between mutant K-Ras signaling, and the CLPTM1L/Akt signaling axis (16). Here, we demonstrate the effectiveness of targeting CLPTM1L with specific antibodies in the inhibition of anchorage-independent growth of both K-Ras–transformed lung and pancreatic tumor cells and EGF-transformed human bronchial epithelial cells. A subset of anti-CLPTM1L mAbs was also effective in the inhibition of H838 tumor spheroid growth. H838 tumor cells do not harbor activating mutations in K-Ras, or H-Ras, or N-Ras, indicating that anti-CLPTM1L antibody treatment is also effective in non-Ras–driven tumor cells as well as those driven by Ras oncogenes.
Here, we provide a description of the design and development of a panel of mAbs targeting specific regions of the CLPTM1L protein, a subset of which affect tumor cell survival signaling and chemotherapy resistance and are highly effective in inhibition of anchorage-independent growth and tumorigenesis. Monoclonal inhibitory anti-CLPTM1L antibodies were effective in inhibiting the growth of established tumors in vivo, particularly those targeting epitope 5. Overall, antibodies targeting this epitope were most inhibitory to CLPTM1L function. Antibodies targeting ECR1 of CLPTM1L (1-1, 1-2, 2-1, and 2-2) were generally less inhibitory and resulted in a more diffuse, less intense staining pattern with immunofluorescence, although clone 1-1 did inhibit Akt phosphorylation. It is possible that antibodies targeting different domains as 1-1 and 5-2 do may bind a differently localized subset of protein, or directly affect localization or staining appearance. Staining appears more punctate and may be more vesicular in Panc1 cells compared with A549 cells. We posit that this may be simply due to differences in cellular morphology. Because of the need to collect endpoint tissues from xenografted mice while treatment remained effective and a finite supply of treatment antibody, the duration of mouse experiments was restricted. This resulted in endpoint tumor volumes that were smaller than is often seen in xenograft experiments that are conducted for longer duration, or with higher cell numbers, but demonstrated striking efficacy against human lung and pancreatic tumors. Clearing of CLPTM1L-positive cells from tumor tissues may cause a decrease in the number of positively staining cells by IHC; however, diminished signal intensity upon anti-CLPTM1L treatment as seen in xenograft tumor samples may also be a result of decreased surface-localized CLPTM1L as demonstrated in Fig. 3.
Our results also demonstrate that targeting CLPTM1L can sensitize tumor cells to killing by genotoxic agents, including cisplatin and gemcitabine. We have demonstrated chemosensitization by anti-CLPTM1L antibodies in pancreatic and lung cancer cells. Akt and Bcl-xL are intimately related to survival of tumor cells harboring oncogenic drivers such as K-Ras (21, 22) and chemoresistance in lung and pancreatic cancers, and are key targets for chemosensitization (22–26). Treatment of human pancreatic adenocarcinoma cells with gemcitabine promoted the interaction of CLPTM1L with PI3K p110α, implicating this interaction and downstream Akt signaling as playing a role in chemoresistance. Furthermore, this promotion of the CLPTM1L–p110α interaction was inhibited by monoclonal anti-CLPTM1L, providing mechanistic insight into the functional inhibition of CLPTM1L by these immunoglobulin agents. The effect on CLPTM1L–p110α interaction may have been due to the fact that anti-CLPTM1L caused a general downregulation of CLPTM1L accumulation at the cell surface, potentially because of internalization of the antibody–antigen complex. This mechanism of downregulation has been previously described for antagonistic antibodies, including cetuximab (27). Expression of a myristoylated Akt partially rescued colony formation, providing further evidence for inhibition of the PI3K/Akt signaling pathway as a mechanism of action. Exogenous activation of this pathway only resulted in a modest increase in colony formation, suggesting that other signaling pathways may be involved. Although we previously showed constitutive activation of Akt to restore anchorage-independent growth, additional CLPTM1L-regulated factors may influence proliferation or survival in adherent culture. Further research, including mechanistic studies and additional in vivo models of combination therapy and chemosensitization will be important for the clinical translation of therapy using CLPTM1L inhibition.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: L.G. Puskás, M.A. James
Development of methodology: I. Man, G. Szebeni, M.A. James
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): L.G. Puskás, G. Szebeni, L. Tiszlavicz, S. Tsai, M.A. James
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): I. Man, G. Szebeni, L. Tiszlavicz, M.A. James
Writing, review, and/or revision of the manuscript: L.G. Puskás, M.A. James
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M.A. James
Study supervision: L.G. Puskás, M.A. James
The authors thank Dr. Marshall Anderson and Elena Kupert for their collaboration in mAb production, Dr. Rebecca Gundry for help with using Protter software, Dr. Brian Volkman for help with transmembrane structure prediction, the MCW Surgical Oncology Biorepository and MCW Clinical and Translational Research Core for their services and guidance, Drs. Craig MacKinnon and Kiyoko Oshima for pathology review of tissues, and the You Laboratory for help with preliminary studies in lung tumor cells.
This work was supported by NIH grant U19CA148127 [Principal investigator (PI): C. Amos], the Batterman Foundation, and the Ronald Burkland Eich Pancreatic Cancer Research Fund, Department of Surgery, Medical College of Wisconsin, the MCW Cancer Center, and Institutional Research Grant # 14-247-29-IRG from the American Cancer Society (PI: M.A. James).
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.