Through a whole-cell panning approach, we previously identified a panel of antibodies that bound to prostate cancer cell surface antigens. One such antigen, CUB domain-containing protein 1 (CDCP1), was recognized by monoclonal antibody 25A11 and is a single transmembrane molecule highly expressed in several metastatic cancers as well as on CD34+CD133+ myeloid leukemic blast cells. We show CDCP1 expression on prostate cancer cell lines by real-time quantitative PCR (RT-qPCR), flow cytometry, and immunohistochemistry and on prostate cancer patient samples by RT-qPCR and immunohistochemical staining. In cell-based assays, antibody 25A11 inhibited prostate cancer cell migration and invasion in vitro. Further characterization showed that CDCP1 is internalized on antibody binding. When 25A11 was coupled to the cytotoxin saporin either directly or via a secondary antibody, both resulted in prostate cancer cell killing in vitro. In vivo targeting studies with an anti-CDCP1 immunotoxin showed significant inhibition of primary tumor growth as well as metastasis in a mouse xenograft model. These data provide support for continued evaluation of anti-CDCP1 therapy for potential use in cancer in primary and metastatic disease. [Cancer Res 2008;68(10):3759–66]
In the United States, prostate cancer is the most frequently diagnosed and second leading cause of cancer death in men (1). Worldwide, prostate cancer ranks third in mortality behind lung and colon cancer, and in developed countries, it is the most common cancer in men (2). Currently, treatment and prognosis is based on clinical stage, biopsy Gleason grade, and serum prostate-specific antigen levels. However, although patients with localized tumor can be effectively treated with ∼100% 5-year relative survival (3), prognostic indicators are not indicative of clinical outcome for those patients with late-stage metastatic disease. Even those patients diagnosed early in their disease course may still have micrometastasis, leading to secondary tumor growth (4). This progression of solid tumors to metastatic disease includes essential steps that involve the detachment of cells from the surrounding extracellular matrix, tumor cell survival in the circulation, and tissue invasion (5, 6).
Through a disease-specific combinatorial antibody library and cell-based panning approach, we identified a panel of antibodies that bind to prostate cancer cell surface antigens (7). One antibody, 25A11, recognizes CUB domain-containing protein 1 (CDCP1). CDCP1 was first identified as an epithelial tumor antigen that is significantly overexpressed in lung cancer cell lines compared with normal lung tissues and also up-regulated in colon adenocarcinomas (8). CDCP1 was independently identified in cancer through subtractive immunization using a highly metastatic human epidermoid carcinoma cell line against a nonmetastatic variant (9). Subsequent studies showed high expression in the metastatic PC-3 prostate cancer line as well as the DLD-1 colon cancer cell line and localization to malignant cells in colon carcinomas. In breast cancer patient samples, elevated CDCP1 expression levels correlated with increased cell proliferation (10). CDCP1 is also expressed on CD34+CD133+ myeloid leukemic blast cells as well as on CD34+CD38− bone marrow stem/progenitor cells (11, 12).
CDCP1 has a single transmembrane predicted structure with three extracellular CUB (initials of the first three identified proteins containing such domains: complement factor C1r/C1s, embryonic sea urchin protein uEGF, and bone morphogenetic protein-1) domains. The 140-kDa molecule is a trypsin-sensitive precursor to the 80-kDa membrane glycoprotein p80. CDCP1 interacts directly with Src, protein kinase Cδ (PKCδ), and Fyn (13–15) in a phosphorylation-dependent pathway known to be involved in cancer cell migration and invasion (reviewed in refs. 16, 17).
In the present study, we show that 25A11 inhibits cancer cell migration and invasion in vitro. In studies to evaluate the potential of anti–CDCP1-targeted therapy, we show that 25A11 is an internalizing antibody. When 25A11 is conjugated to saporin (a ribosomal-inactivating toxin), primary tumor growth as well as size and incidence of lymph node metastases are inhibited in a prostate cancer spontaneous metastatic tumor model. As entry of tumor cells into the circulation is the critical first step in the metastatic process, and the presence of circulating tumor cells in blood is associated with poor overall survival in patients with metastatic prostate cancer (18), antibody targeting that either blocks the function of CDCP1 on these cells or kills circulating tumor cells may have clinical relevance. Together, our studies suggest that CDCP1 is a potential target for treatment of prostate cancer and metastasis.
Materials and Methods
Cell lines. Prostate adenocarcinoma PC-3, prostate carcinoma lymph node metastasis LNCaP, and prostate carcinoma brain metastasis DU145 cell lines were obtained from the American Type Culture Collection and grown as described (7). The PrEC normal prostate epithelial cell line was obtained from Cambrex Bio Science and cultured in prostate epithelial growth medium according to the manufacturer's instructions.
Reagents and antibodies. Phage display antibody production, antibody selection by cell surface panning with stringent negative selection, and antigen identification for 25A11 was described previously (7). Murine IgG 25A11 was made by cloning the 25A11 murine Fab into a murine IgG vector; chimeric antibodies were made by inserting variable regions of the murine Fab by overlap PCR into Fab vectors containing human constant regions and then subcloning into a human IgG vector. Anti-CDCP1 antibodies used in flow cytometry, immunohistochemistry, and internalization assays included chimeric 25A11 (ch25A11) monoclonal antibody (mAb) and murine CUB1 (MBL). Isotype control antibodies used in internalization assays included an in-house chimeric antibody and murine anti-OX7 (Advanced Targeting Systems). Saporin-conjugated goat anti-mouse IgG (mAb-ZAP), goat anti-human IgG (Hum-ZAP), and goat IgG isotype control (Goat IgG-SAP) secondary antibodies as well as the ch25A11-saporin custom direct conjugate (ch25A11-Sap) were purchased from Advanced Targeting Systems. The ratio of toxin to antibody in the ch25A11-Sap conjugate was ∼2:1. CUB1-Zenon labeling was performed using the Zenon Mouse IgG Labeling kit (Invitrogen). Src inhibitor PP2 [4-amino-5-(4-chlorophenyl)-7-(t-butyl)pyrazolo[3,4-d]pyrimidine] was purchased from Calbiochem.
RNA samples. Total RNA for the normal tissue panel was purchased from BD Biosciences and BioChain Institute, Inc. Total RNA from frozen sections of prostate cell lines and severe combined immunodeficient (SCID) xenografts was extracted using RNeasy Mini kits (Qiagen) or Trizol reagent (Invitrogen) according to the manufacturer's instructions. RNA from 13 prostate patient primary tumor and 2 normal samples was purchased from Ardais Corp. RNA from one additional prostate cancer patient sample was obtained from Asterand. All RNA samples were treated with DNaseI, and cDNA was prepared using the High-Capacity cDNA Archive kit (Applied Biosystems).
Real-time quantitative PCR. Relative gene expression levels were determined by real-time quantitative PCR (RT-qPCR) using 18S rRNA for normalization. Assays-on-Demand Taqman probes for CDCP1 and 18S rRNA were used with Taqman Universal PCR master mix (Applied Biosystems) for cDNA amplification. Amplification and analysis were performed using the ABI 7500 sequence detection system.
Flow cytometry. PC-3, DU145, and LNCaP cells were incubated with 200 nmol/L ch25A11 and stained with R-phycoerythrin (R-PE)-conjugated goat anti-human IgG (H+L; Jackson ImmunoResearch, Inc.). For unique epitope determination, PC-3 cells at 1.25 × 106/mL in 100 μL were bound to 160 nmol/L of murine 25A11 Fab (∼80% saturation) and stained with R-PE–conjugated goat anti-mouse IgG (Sigma). After washing with PBS twice, Zenon-labeled CUB1 antibody was added to each reaction at 0.01, 0.3, 1.3, 6.4, 32, or 160 nmol/L and incubated on ice for 30 min. Control reactions of CUB1 binding to PC-3 cells without 25A11 prebinding were set up accordingly. All staining was detected on a Becton Dickinson FACSCalibur flow cytometric analyzer.
Immunohistochemistry. Antibody titrations on fresh-frozen tissues with murine antibody 25A11 IgG1 (for human tissues) or chimeric ch25A11 IgG1 (for mouse tissues) identified 2.5 μg/mL as the concentration that would result in minimal background and maximum detection of signal. Detection of primary antibodies was performed with DAKO Envision peroxidase-labeled polymer with 3,3′-diaminobenzidine as the chromogen, which produces a brown-colored deposit. Tissues were also stained with positive control antibodies (CD31 and vimentin) to ensure that the tissue antigens were preserved and accessible for immunohistochemistry. The negative control consisted of treating adjacent sections similarly but in the absence of primary antibody.
Cell migration and invasion assays. Cell migration and invasion assays were purchased from Chemicon/Millipore International and performed according to the manufacturer's instructions. Briefly, PC-3 cells were treated with 4-fold dilutions of CUB1 and ch25A11 in PBS and tested for inhibition of cell migration and invasion through an 8-μm pore size polycarbonate membrane. Invasion assays used a membrane coated with a uniform layer of basement membrane matrix solution, which serves as a barrier to discriminate invasive cells from noninvasive cells. PP2 was titrated from 25 to 0.02 μmol/L on PC-3 cells, with the optimal concentration for maximum cell migration and the least cell toxicity being 2.5 μmol/L. In both assays, PP2 at 2.5 μmol/L was used as a control for inhibition of migration/invasion. Other controls included a serum-free medium control for assay background and a complete medium control containing 10% fetal bovine serum for the maximum cell migration or invasion readout. For quantitation, the plates were incubated for 24 h at 37°C, and cells that migrated to the lower surface were stained with crystal violet and counted (five fields per well) on an Olympus IX70 microscope.
In vitro internalization assay. PC-3 cells were plated in 96-well microplate wells at 2,500 cells/90 μL. The plates were incubated for 16 h at 37°C in the presence of 5% CO2. Primary antibodies with either mAb-ZAP or Hum-ZAP secondary antibodies, or ch25A11-Sap with PBS vehicle instead of secondary, were added in a volume of 10 μL for a final well volume of 100 μL. Recognition and internalization of the primary antibody results in delivery of the saporin-antibody complex to the cell interior followed by cell killing. Goat IgG-SAP was used as a nontargeted saporin control for the mAb-ZAP and Hum-ZAP secondary antibodies. To avoid binding and internalization of primary antibody before its interaction with toxin-conjugated secondary antibody, the secondary saporin conjugate (mAb-ZAP or Hum-ZAP) was added to the plate first. Plates were incubated for 72 h at 37°C in the presence of 5% CO2. Cell viability was assayed using CellTiter 96 AQueous One Solution Cell Proliferation Assay according to the manufacturer's instructions (Promega), and the plates were read at 490 nm in a Molecular Devices Vmax microplate reader. Percent cell viability was calculated by assigning the average of the readings from secondary antibody Goat IgG-SAP isotype controls with PBS as 100% cell viability. The positive antibody controls for internalization were anti-epidermal growth factor receptor (EGFR) mAb-225 (19) and anti-Her2 mAb-4D5 (20), which were tested on A431 and PC-3 cell lines.
In vivo immunotoxin studies. SCID CB17 mice were used for in vivo studies. For the dose-finding study using ch25A11-Sap, mice were randomized and divided into seven groups of 12 mice each that were i.v. injected as follows: 200 μL PBS on days 0, 2, and 5 (group 1); 0.286 mg/kg ch25A11 (the equivalent antibody dose of the saporin conjugate) on day 0 (group 2) or days 0, 2, and 5 (group 3); 0.4 mg/kg ch25A11-Sap on day 0 (groups 4), days 0 and 2 (group 5), days 0, 2, and 5 (group 6), or days 0, 5, 7, and 9 (group 7). Sera were collected from two mice from each group at day −2 or after injection at 2 min, 30 min, 6 h, 24 h, 3 d, 7 d, 14 d, and 21 d for a pharmacokinetic study of ch25A11 and ch25A11-Sap, whereas 10 mice were used for the dose finding. The amount of ch25A11 in serum was tested by ELISA and compared with the standard curve. The pharmacokinetic variables of ch25A11-Sap could not be obtained because of the masking effect of saporin on the antibody in the conjugate, resulting in unsuccessful capture and/or detection of the antibody in the ELISA assay. The pharmacokinetic variables of ch25A11 were obtained from group 2 by noncompartmental analysis/first-order kinetics.
The spontaneously metastatic PC-3 tumor model was used to evaluate the anticancer activity of the ch25A11-Sap conjugate. Briefly, 3 × 106 cells were s.c. injected into SCID CB17 mice on the lower back on day 0. On day 7, mice were randomized and divided into six groups, 7 to 10 mice per group. Three 200 μL i.v. or s.c. injections were given to each group on days 7, 10, and 17 at the doses specified: group 1, PBS alone, i.v.; group 2, 0.286 mg/kg ch25A11 antibody alone (equivalent antibody dose of the conjugate), i.v.; group 3, 0.014 mg/kg saporin alone (equivalent toxin dose of the conjugate), i.v.; group 4, 0.4 mg/kg ch25A11-Sap, i.v.; group 5, PBS, s.c. (injection into the flank region of each mouse, at least 1 cm away from the tumor site); group 6, 0.4 mg/kg ch25A11-Sap, s.c. Primary tumors were measured twice weekly to assess the antitumor activity of ch25A11 and ch25A11-Sap. On day 23, primary tumors were removed from all mice. To evaluate the anticancer activity of ch25A11 without toxin, a postsurgery treatment with ch25A11 twice weekly for 3 wk was added to group 2. The same postsurgery PBS treatment was added to group 1 as a control. At days 46 and 50, s.c. lymph node metastases were measured with a caliper and analyzed by incidence for all groups. The body weights of mice were measured daily to assess liver toxicity in both studies.
Expression of CDCP1 mRNA in prostate cell lines, SCID xenografts, and prostate patient samples by RT-qPCR. The CDCP1 mRNA expression profile was determined by RT-qPCR for normal human tissues, prostate cancer patient samples, and prostate cancer cell lines and their corresponding xenografts. In normal human tissues, the highest CDCP1 expression was found in colon (∼2.5-fold higher than normal prostate) followed by skin, small intestine, and normal prostate (Fig. 1A). Lower CDCP1 expression (approximately half the level of normal prostate) was found in kidney, lung, pancreas, bladder, placenta, uterus, and stomach. We evaluated 10 prostate cancer primary tumor samples and found the average overall CDCP1 transcript level to be ∼1.9-fold lower than in normal prostate tissue (Fig. 1B). CDCP1 mRNA expression was evaluated in two prostate intraepithelial neoplasia and two benign hyperplasia samples, which were approximately equal to that of nondiseased prostate. CDCP1 transcripts were detected in all three prostate cancer cell lines examined, PC-3, DU145, and LNCaP, as well as the corresponding SCID mouse tumor xenografts (Fig. 1C). Although these data suggest that the cell lines express higher levels of CDCP1 than primary prostate, we cannot make direct comparisons by RT-qPCR as the primary samples are a heterogeneous mixture of cell types compared with a homogeneous population of cultured cells. A more direct comparison may be the level of CDCP1 in PrEC, a cell line of epithelial origin (same origin as PC-3), which is 1.6-fold higher than the level of CDCP1 in PC-3. CDCP1 protein expression on the cell surface of PC-3, DU145, and LNCaP cell lines was confirmed by flow cytometry with ch25A11 (Fig. 1D).
CDCP1 protein expression in prostate cancer and normal human and normal mouse tissues by immunohistochemistry. CDCP1 was found to be present on normal prostate epithelial cells as well as on malignant cells in four of five primary prostate tumor patient samples examined by immunohistochemistry (summarized in Table 1). In this study, 25A11 was evaluated on frozen sections of normal prostate and on samples of prostate cancer with associated benign glands. The most prominent staining was observed in benign glandular epithelium. Malignant glands were also positive, but staining was generally less prevalent and less intense than adjacent benign glands and benign glandular epithelium in normal samples (Fig. 2A). Staining in both benign and malignant glands was found in the luminal layer and predominantly membranous. Other cell types identified in this study were negative, which included inflammatory cells, endothelium, vascular smooth muscle, nerves, and prostatic fibromuscular stroma.
|Patient .||Age .||Gleason score .||Cell type .||Cellular location .||Expression intensity .||Expression frequency (%) .|
|Patient .||Age .||Gleason score .||Cell type .||Cellular location .||Expression intensity .||Expression frequency (%) .|
NOTE: Immunohistochemical staining intensity: negative (−), blush positive (+), faint positive (++), moderate positive (+++), and strong positive (++++). Immunohistochemical staining frequency: rare (0.1–10%), focal (10–30%), frequent (30–75%), most (75–90%), and no modifier (>90%).
Abbreviations: M, membranous staining; C, cytoplasmic staining; NA, not available.
Before further model development and to consider CDCP1 as a target for antibody therapy, we addressed potential toxicity concerns by evaluating CDCP1 protein expression by immunohistochemical staining in several additional critical normal human tissues (Fig. 2B). As CDCP1 is expressed in normal human colon (9), normal human colon was used as a positive control in this study (data not shown). Moderate staining was identified in colonic epithelium, bile ducts, pancreatic ducts, and respiratory epithelium. Less intense staining was identified in a subset of renal tubular epithelium, mucous glands in the bronchi, and pancreatic islets, although many other cell types in pancreas were negative, including endothelium, smooth muscle, fibroblasts, and peripheral nerves. Liver hepatocytes showed slight staining. Heart and spleen tissues were negative. Glomeruli, including parietal epithelial cells, visceral epithelial cells, mesangial cells, and glomerular capillary endothelium, were also negative.
For animal model evaluation, we tested cross-reactivity of 25A11 to murine CDCP1 by immunohistochemical staining of mouse tissues, as gene expression profiling indicates that murine CDCP1 is highly expressed in mouse colon (21). Although mouse and human CDCP1 proteins share 81.3% identity, 25A11 immunohistochemistry was negative on most normal mouse tissues, including colon, with the exception of weak staining in mouse hepatocytes (data not shown).
25A11 binds to a distinct epitope of CDCP1 and blocks cell migration and invasion in vitro. In a series of in vitro studies, we compared 25A11 with a commercially available anti-CDCP1 antibody, CUB1 (12). To determine if 25A11 and CUB1 bind to different epitopes, PC-3 cells were bound with 25A11 Fab at 80% saturation before the binding of CUB1. CUB1 binding to PC-3 cells increases with higher concentrations at the same rate, regardless of whether 25A11 is prebound or not (Fig. 3A), suggesting that 25A11 and CUB1 bind to a different epitopes.
In vitro migration and invasion assays were performed using Boyden chambers in which the effect of 25A11 and CUB1 on the migration or invasion of PC-3 cells through the membrane into the outer chamber was evaluated (described in the Materials and Methods). Addition of the Src family kinase inhibitor PP2 was used as a positive control in this study, as it has been shown to significantly inhibit DU145, PC-3, and LNCaP cell migration in vitro (22). In the cell migration assay, ch25A11 treatment at 0.8 μmol/L resulted in a 78% inhibition of PC-3 cell migration compared with the PBS/complete medium control (Fig. 3B and C). This was superior to the 57% inhibition obtained by treatment with 2.5 μmol/L PP2, the concentration that was experimentally determined to be the optimal concentration for maximum PC-3 cell migration inhibition with the least cell killing. ch25A11 at a concentration of 0.8 μmol/L also effectively inhibited PC-3 cell invasion at levels comparable with 2.5 μmol/L PP2, resulting in ∼45% inhibition of invasion compared with the PBS/complete medium control (Fig. 3C). Although CUB1 had a modest effect of 32% inhibition of cell migration, it was not found to inhibit cell invasion.
CDCP1 immunotoxin-mediated cell cytotoxicity. To more fully characterize antibody binding to CDCP1, we evaluated whether antibody bound to CDCP1 is internalized as an additional mechanism to antibody-mediated therapy. For internalization studies, we evaluated antibody-toxin conjugates, where internalization-dependent toxin delivery mediates cell killing in vitro. PC-3 cells were treated with anti-CDCP1 antibodies, isotype control antibodies, or positive control antibodies (anti-EGFR mAb-225 and anti-Her2 mAb-4D5), which are known to bind, internalize, and kill target cells (19, 20). Murine 25A11 and ch25A11 as well as murine CUB1 primary antibodies are internalized similarly as evidenced by dose-dependent PC-3 cell killing when used in conjunction with the appropriate saporin secondary conjugates (Fig. 4). No cell killing is observed with the isotype control primary antibodies (Fig. 4) or the nonspecific secondary saporin conjugate Goat IgG-SAP (data not shown). The ch25A11-Sap direct immunoconjugate shows dose-dependent PC-3 cell killing similar to the 25A11 and CUB1 antibodies with their appropriate saporin-conjugated secondary antibodies (Fig. 4). The killing curves at these antibody concentrations were similar to those obtained with the positive control antibodies on PC-3 cells (data not shown).
ch25A11-Sap conjugate directly kills PC-3 cells in vivo. To extend the in vitro cell killing data, we evaluated in vivo cell killing using the ch25A11-Sap direct conjugate. To begin these studies, we first performed a dose-finding study that showed that ch25A11 or saporin treatments did not have toxicity in SCID CB17 mice. In contrast, one to four doses of ch25A11-Sap treatment caused immediate body weight loss up to 24%, indicative of acute toxicity (data not shown); however, none of the mice died. The dose-finding study indicated that the optimal regimen for ch25A11-Sap was a three-dose treatment on days 7, 10, and 17. Pharmacokinetic variables indicate that ch25A11 has a long elimination half-life in mice of 8.4 days (Fig. 5A). The volume of distribution (Vd) of ch25A11 is close to the circulation volume of SCID mice at the same age, indicating adequate drug exposure. The small clearance (Cl) value, reasonable antibody concentration at time 0 (C0), and area under the curve (AUC) support the regimen used in the efficacy study.
In the efficacy study, the effects of ch25A11, saporin, or ch25A11-Sap were evaluated for PC-3 tumor growth inhibition and compared with the PBS vehicle control. The ch25A11 and saporin doses were designed to approximate the amounts of each present in the 0.4 mg/kg immunoconjugate dose. Administration of ch25A11 i.v. and saporin i.v. did not affect PC-3 tumor growth at the levels used (Fig. 5B). ch25A11-Sap i.v. inhibited tumor growth approximately 66% at day 18, 67% at day 22, and 63% at day 23, which were significant (P < 0.05) by Mann-Whitney test. ch25A11-Sap s.c. did not inhibit tumor growth, suggesting possible poor bioavailability and low drug exposure at the primary tumor site by the s.c. route. The ch25A11-alone and saporin-alone groups showed slightly larger tumor burdens than the PBS control, but this was not statistically significant. In the 25A11-Sap i.v. treated group, we did note a slight increase in tumor growth weekly after the last dose at day 17, suggesting that, once the drug is cleared, any remaining surviving tumor cells can continue to proliferate.
In addition to effects of the antibody-toxin fusion on primary tumors, this model allows evaluation of the antibody-toxin fusion effect on metastasis. In the PC-3 spontaneous metastatic tumor model, a secondary tumor caused by metastasis is often observed in one of the superficial axillary lymph nodes or inguinal lymph nodes. To investigate the effect of the anti-CDCP1 immunoconjugate on PC-3 metastasis, primary tumors on the lower back were surgically removed on day 23 to differentiate lymph node metastasis from primary tumor growth. S.c. lymph node metastasis in the superficial axillary lymph nodes of all groups was analyzed by size and incidence of metastatic lesions (summarized in Fig. 5C). Both i.v. and s.c. administration of ch25A11-Sap inhibited tumor metastasis, suggesting that although ch25A11-Sap s.c. may not have sufficient bioavailability to inhibit primary tumor growth, the inhibition of metastases may be a direct result of tumor cell killing in the circulation.
Consistent with the dose-finding study, both i.v. and s.c. administration of ch25A11-Sap caused acute toxicity, shown by the acute but reversible body weight loss 1 day after injection (Fig. 5D). However, after primary tumor removal on day 23, mice in both groups treated with ch25A11-Sap regained body weight similar to that of control mice without tumors by day 35 due to inhibition of metastasis. In ch25A11-treated and saporin-treated groups, body weight loss was caused mainly by large tumor burden (days 10–23) and severe lymph node metastasis (after day 30) but was not due to drug-related toxicity.
CDCP1 is a substrate for phosphorylation by Src kinases in vitro, and increased phosphorylation status is associated with anchorage independence (13–15). The tyrosine kinase inhibitor PD173955, which is selective for Src family kinases, inhibits phosphorylation of CDCP1 (14). Interestingly, as Src kinase activity is necessary for the interaction of CDCP1 with Src and PKCδ (13), the Src family kinase inhibitor dasatinib can suppress the proliferation of PC-3 human prostate cancer cells (23) and inhibit the adhesion, migration, and invasion of DU145 human prostate cancer cells (24). Additional Src inhibitors have been reported to reduce prostate cancer growth and metastasis in mouse xenograft studies (reviewed in ref. 17). Consistent with a role in cell adhesion, CDCP1 also interacts with the adhesion proteins N-cadherin and P-cadherin, the matrix proteins syndecans 1 and 4, and the membrane serine protease MT-SP1, and overexpression of CDCP1 results in a loss of adhesion phenotype (14). Recently, it has also been shown that CDCP1 is essential for lung cancer metastasis formation in vivo (15).
In the present study, we show that, similar to Src kinase inhibition, 25A11 antibody blocking of CDCP1 inhibits cancer cell migration and invasion in vitro, possibly through blocking the interaction of CDCP1 with another cell surface molecule necessary for activation of the Src kinase pathway. Although both 25A11 and CUB1 anti-CDCP1 antibodies bound to PC-3 cells by flow cytometry, 25A11 treatment showed increased inhibition of cell migration as well as invasion compared with treatment with CUB1, likely a result of 25A11 binding to a different epitope.
Several antibodies targeting pan-cancer antigens that bind to prostate tumors are in preclinical and clinical development, including antibodies against prostate stem cell antigen (PSCA), which have been shown to positively correlate with disease progression (25–27). Although our data indicate that the tumor-associated antigen CDCP1 is not up-regulated in prostate cancer, the rationale for targeting CDCP1 is based on the presence and proposed physiologic function of the protein, as an antibody that inhibits a cell surface target upstream of the Src family kinases could be clinically relevant. Similar to anti-PSCA antibodies (28), we show that 25A11 is an internalizing antibody and that targeting CDCP1 with an immunotoxin leads to cell killing in vitro. This observation allowed us to evaluate immunotoxin targeting of cancer cells via CDCP1 in a tumor xenograft model. As we examined the effects of the various treatment regimens on primary tumor growth, we found that 25A11-Sap when administered i.v. significantly inhibited primary tumor growth, whereas Sap i.v. or ch25A11 i.v. alone at the dose of the immunoconjugate did not have an effect. After primary tumor removal on day 23, mice in groups treated with ch25A11-Sap through both routes of administration also showed significant inhibition of metastasis, which correlated well with regaining body weight (Fig. 5C and D). Although ch25A11-Sap s.c. does not seem to have sufficient bioavailability to affect primary tumor growth, our data suggest that it can target tumor cells in the circulation to inhibit metastasis formation.
To consider the potential for toxicity of targeted anti-CDCP1 therapies, we performed RT-qPCR and immunohistochemical studies, which indicate that CDCP1 has low to moderate expression on many essential human tissues. When targeting cancer cells using immunotoxins or radiolabeled conjugates, even low expression of the target on normal tissues can be problematic, and standard tissue cross-reactivity and toxicity studies in relevant animals, including nonhuman primates, would be warranted before any consideration of immunotoxin trials in humans.
However, due to our in vitro data and the implied role of CDCP1 in the metastatic process, less toxic treatment options may include an anti-CDCP1 IgG1 antibody that can bind and promote antibody-dependent cellular cytotoxicity–mediated cell killing or an anti-CDCP1 antibody devoid of effector function whose main mechanism of action could be blocking CDCP1 function. Although unconjugated ch25A11 did not show antibody efficacy in vivo when compared with the immunoconjugate, this may be due to the extremely low dose of antibody (0.286 mg/kg) used to control for the amount in the ch25A11-Sap immunoconjugate, which would likely be below the therapeutic range for the inhibition of primary tumors and metastasis. Future experiments should address the potential of higher levels of unconjugated ch25A11 antibodies in inhibiting metastasis formation. Therefore, as CDCP1 is an interesting and potentially promising target for antibody-mediated therapy, additional studies to elucidate the function of CDCP1 in the Src signaling pathway and its role in metastasis may be helpful in going forward with drug development either in man or in animal models.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
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.
We thank Linh Tran for construction of the fully murine 25A11 IgG1 molecule, Kathleen Autote for assistance with antibody purification, and Matthew G. Archer for assistance with assay development.