Tissue factor expression in ovarian cancer: implications for immunotherapy with hI-con1, a factor VII-IgGF(c) chimeric protein targeting tissue factor.

A total of 63 fresh-frozen ovarian tumors including 30 OSPC, 14 CC, 12 END and 7 UND carcinomas were tested by real-time-PCR for TF expression. In addition, all five primary EOC cell lines available to this study, including 3 serous and 2 CC primary tumors ( Table 1 ) were also tested for TF expression by qRT-PCR. We found ovarian carcinomas with CC histology to significantly over-express TF (mean ± SEM = 20.4 ± 6.5, range 0.1–80.5) when compared to serous (1.4 ± 0.2, range 0.08–4.1, P < 0.0001), to END (1.7 ± 0.3, range 0.1–3.8, P = 0.01) or to UND tumors (2.4 ± 0.8, range 0.3–6.9, P = 0.01). Of the 5 primary tumor cell lines tested, four showed a high mRNA copy number for TF (i.e., OSPC-1 ARK-1, OSPC-2 ARK-2, CC-1 ARK-1 and CC-2 ARK-2), ranging from 122.1 to 827.1 with a mean ± standard error of 528.4 ± 160.7 ( Table 2 ), while one (i.e., OSPC-3 ARK-3) showed low TF expression by qRT-PCR ( Table 2 ). Consistent with the results obtained in fresh frozen ovarian cancer tissues, TF expression was found to be significantly higher in CC primary ovarian cancer cell lines when compared to those from serous tumors ( Table 2 , P < 0.03) or when compared to the human NOVA used as controls ( P < 0.01). We performed immunohistochemical analysis of TF protein expression on formalin fixed tumor tissue from 25 paraffin-embedded EOC including 9 OSPC, 10 CC and 6 END ovarian adenocarcinomas. As representatively shown in Fig. 1 , we found high TF expression (i.e., 2+ or above) in 10 out of 10 (100%) of the CC ovarian carcinoma tested (i.e., three 4+, five 3+ and two 2+, respectively), in 6 out of 9 (67%) of the OSPC tumors (i.e., one 4+, two 3+, three 2+ and three 1+, respectively) and in 5 out of 6 (83%) of the END tumors tested (i.e., one 4+, two 3+, two 2+ and one showing no TF expression). With a single exception, all EOC samples tested showed either membrane and/or cytoplasmic immunoreactivity for TF (i.e., 24 out of 25 samples = 96%), while the non-neoplastic ovarian controls were found consistently negative for TF ( Fig. 1 ). Surface TF receptor and mCRP expression were evaluated by FACS analysis on all 5 primary EOC cell lines using hI-con1 and anti-CD46, anti-CD55, anti-CD59 and an anti-human TF control mAbs. As negative controls, several PHA-stimulated PBL established from healthy donors or the same EOC patients from which the tumor cell lines had been established were also studied. In agreement with the qRT-PCR results, high reactivity against TF was found by flow cytometry in OSPC-1 ARK-1, OSPC-2 ARK-2, CC-1 ARK-1 and CC-2 ARK-2 cell lines stained with hI-con1 ( Table 2 ; Fig. 2 ). In contrast, significantly lower TF surface expression was detected in the OSPC-3 ARK-3 cell line ( Table 2 ). Mean fluorescence intensity (MFI) ranged from 65.4 to 124.5 in high TF expressor cell lines versus a MFI of 29 in the OSPC-3 cell line and 15 in PHA-stimulated PBL used as negative controls ( Table 2 ; Fig. 2 , P = 0.03). When mCRP surface expression was evaluated by flow cytometry, we found that with the exception of CC-2 ARK-2 where CD59 was not detected ( Table 3 ), all remaining chemotherapy resistant ovarian cancer cell lines expressed high levels of CD46, CD55 and CD59. All five primary EOC cell lines available were tested for their sensitivity to NK cytotoxicity when challenged with heterologous PBL, collected from several healthy donors, in a standard 5-h 51 Cr-release assay. EOC primary cell lines were consistently found to be resistant to NK-mediated cytotoxicity when combined with PBL at effector:target (E:T) ratios varying from 25:1 to 50:1 (range of cytotoxicity from 0.7 to 12.8% with all E:T ratios). Similarly, EOC cell lines incubated with rituximab (2.5 μg/ml) control antibody displayed no significant cytotoxicity (range from 1.1 to 11.2%). We then investigated the sensitivity of EOC cell lines expressing different levels of TF to heterologous PBL in the presence of hI-con1 (30 μg/ml). While negligible levels of cytotoxicity were detected against PHA stimulated PBL (mean ± SD, 5.8 ± 0.8%, Fig. 3 , P > 0.05), tumor cell lines expressing high levels of TF (i.e., OSPC-1 ARK-1, OSPC-2 ARK-2, CC-1 ARK-1 and CC-2 ARK-2) were found highly sensitive to hI-con1-induced cell death (mean ± SD, 50.2 ± 10.2%, range, 33.0–66.4% for the E:T ratio 50:1, Fig. 3 , P < 0.001). In contrast, the ovarian tumor cell line expressing low levels of TF (i.e., OSPC-3 ARK-3), was not significantly killed (mean ± SD, 6.9 ± 0.07% for the E:T ratio 50:1; Fig. 3 ) when compared to PHA-stimulated PBL control incubated with hI-con1. In order to evaluate the effect of complement on hI-con1-mediated-cell death as well as its potential inhibition by physiological IgG serum concentrations, human plasma diluted 1:2 (with and without heat inactivation) was added during standard 5 h 51 Cr release cytotoxicity assays against two clear cell tumor cell lines expressing high levels of TF ( Table 2 ) but showing differential expression of the membrane-bound complement regulatory protein CD59 (i.e., CC-1 ARK-1 showing high level of CD59 and CC-2 ARK-2 showing low-negligible levels of CD59, Table 3 ). In multiple experiments we consistently noted a significantly decreased hI-con1-dependent cell-killing after incubation of CC-1 ARK-1 with plasma with or without heat inactivation when compared to PBL alone ( Fig. 4a , P = 0.02). These results illustrate the potential inhibitory effect of high concentrations of IgG on hI-con1-mediated cytotoxicity in vitro and the potential lack of CC-1 ARK-1 sensitivity to complement-dependent cell death. Consistent with this view, additional cytotoxicity experiments with CC-1 ARK-1 in the presence of plasma with or without hI-con1 and in the absence of PBL demonstrated negligible levels of complement-dependent-cell death (mean killing = 5%, range 1.7–7%). In contrast, when CC-2 ARK-2 tumor cell line was evaluated in cytotoxicity assays, we found that the addition of plasma to hI-con1 significantly increases IDCC against this ovarian tumor ( Fig. 4b , P = 0.001), while heat-inactivated plasma did not result in a significant decrease in CC-2 ARK-2 cell killing when compared to PBL alone ( Fig. 4b , P = 0.45). These results illustrate the potential high sensitivity to complement of CC-2 ARK-2 and the lack of a significant inhibitory effect of high concentrations of IgG on hI-con1-mediated cytotoxicity against this complement-sensitive tumor in vitro. Consistent with this view, additional cytotoxicity experiments in the presence of plasma with or without hI-con1-and in the absence of PBL confirmed significant levels of complement-dependent-cell death against CC-2 ARK-2 (mean killing = 22.6%, range 21–24%). Finally, as shown in Fig. 4c for CC-1 ARK-1, knockdown experiments of mCRP CD59, demonstrated a significant increase in IDCC in the presence of complement in vitro ( P = 0.01). These results demonstrating the importance of CD59 expression in potentially inhibiting complement attack during hI-con1-dependent-cytotoxicity experiments against aggressive ovarian tumor overexpressing TF. To investigate the effect of low doses of IL-2 in combination with hI-con1 (30 μg/ml) on IDCC against EOC cell lines, PBL from healthy donors were incubated for 5 h in the presence of 50–100 IU/ml of IL-2. As representatively shown in Fig. 5 for CC-1 ARK-1, in both EOC tested showing high TF overexpression (i.e., OSPC-2 ARK-2 and CC-1 ARK-1), IDCC was increased in the presence of low doses of IL-2. Administration of 50–100 IU/ml of IL-2 to the effector PBL at the start of the assay increased the cytotoxic activity against EOC cell lines overexpressing TF compared with the use of hI-con1 alone ( Fig. 5 , P < 0.01) whereas no significant increase in cytotoxicity was detected after IL-2 treatment in the absence of hI-con1 or in the presence of rituximab control mAb ( Fig. 5 ). These results suggest that low levels of IL-2 may enhance hI-con1-mediated-cell-cytotoxicity in EOC cell lines in vitro.

EOC remains the most lethal gynecologic malignancy in the US and Western Europe. Although the majority of women diagnosed with advanced ovarian cancer initially respond to the combination of surgery and chemotherapy, most of these patients inevitably succumb to the onset of chemotherapy-resistant/recurrent disease [ 2 ]. Thus, development of novel, effective targeted therapies against chemotherapy-resistant ovarian carcinoma remains an unmet medical need. Ovarian cancer is morphologically and biologically a heterogeneous disease characterized by multiple histological subtypes including serous, endometrioid, mucinous, undifferentiated and clear-cell. While CC ovarian cancer constitute only 3.7–12.1% of the EOC cases, a large number of relapses and deaths occur in this group of patients [ 20 , 21 ]. The main factor to explain this disproportionate poor outcome is a reported lower sensitivity of CC ovarian cancer to conventional platinum- or taxane-based chemotherapeutic regimens [ 20 – 22 ]. Moreover, an additional typical feature of the clear cell histology is the lower cell proliferation activity that may further contribute to the drug resistance mechanism [ 22 ]. We have evaluated TF expression by IHC and qRT-PCR in multiple formalin-fixed and fresh-frozen ovarian carcinoma samples with multiple histologies. In addition, we have studied TF expression at RNA and protein levels in primary ovarian carcinoma cell lines derived from patients harboring chemotherapy-resistant tumors and subsequently tested the in vitro activity of hI-con1, a previously characterized immuno-conjugate molecule that targets TF [ 9 – 11 ], against these chemotherapy-resistant EOC cell lines in vitro. In this regard, TF is known to be involved in pathological angiogenesis and is abnormally overexpressed in multiple human tumors including ovarian cancer and in tumor vascular endothelial cells but not on normal quiescent vascular endothelial cells [ 4 , 5 ]. Although TF, as a cell surface receptor, is physiologically expressed on extra-vascular cells of many organs and in the adventitial layer of the blood vessel, it is shielded from coagulation factor VII (fVII), a natural ligand for TF, at these sites by the tight endothelial cell–cell junctions of the normal vasculature [ 4 , 5 ]. Thus, pathologically expressed TF may provide a target for the development of novel cancer therapies effective not only against tumor cells but also tumor blood vessels [ 9 – 11 , 23 ]. We observed TF overexpression on the membrane and/ or in the cytoplasm of the majority of ovarian tumor samples tested by IHC (24 out of 25 = 96%) while normal control ovarian cells were found consistently negative for TF expression. Of interest, CC ovarian tumors were found to express significantly higher TF levels than the other histological types of ovarian cancer tested by qRT-PCR. These data suggest that TF expression may be a common and important event in malignant transformation of ovarian cells into biologically aggressive cancers and particularly in CC ovarian tumors. Our results are in agreement with multiple recent reports showing the presence of TF-coagulation factor VII complex on ovarian cancer cells, particularly in tumors with clear cell histology [ 24 , 25 ] and also consistent with the recent discovery that ectopic synthesis of Factor VII may activate ovarian cancer cell migration and invasion [ 26 ]. Of interest, in these studies, high levels of TF expression, specifically in patients harboring CC ovarian tumors, were associated with a high incidence of deep-vein thrombosis and subsequent pulmonary thromboembolism [ 24 , 25 ]. Consistent with the results obtained using formalin-fixed and fresh frozen tumor tissues, four out of five of the chemotherapy-resistant tumor cell lines tested in our study were found to overexpress TF by qRT-PCR and by flow cytometry. Moreover, both CC ovarian cell lines available to this study showed significantly higher TF expression when compared to primary serous cell lines. Importantly, primary OSPC and CC cell lines overexpressing TF, regardless to their resistance to chemotherapy or histology, were found to be highly sensitive to hI-con1-mediated cytotoxicity in vitro. Previous studies have shown that transfection of TF into ovarian carcinoma cell lines increased invasive and meta-static properties in murine models [ 27 ]. In addition, elevated preoperative serum TF levels have been previously reported to represent an independent prognostic factor for early death from disease in ovarian carcinoma patients [ 28 ]. It has been speculated that tumor cell TF plays an important role in the tumor metastatic process, possibly by inducing the coating of the tumor cell with fibrin that would trap the cells in the microvasculature, thereby aiding metastases [ 29 ]. More recently, however, a possible direct role for TF in tumor growth has also been suggested by studies showing dramatically reduced tumor growth in mice where a selective reduction in TF was achieved using small interfering RNA [ 30 ]. Of great interest, in these studies, the reduction of TF expression did not affect growth of the tumor cells in vitro, suggesting that TF-mediated enhancement of tumor growth requires a factor present in vivo that is not present when cells are grown in vitro [ 30 ]. A potential candidate to explain these findings is therefore fVIIa, which would form a TF:fVIIa complex on the surface of tumor cells in vivo leading to activation of PAR2-dependent signaling [ 8 ]. These findings combined with our results suggest that TF overexpression may potentially provide an additional growth advantage to biologically aggressive EOC in vivo. The potential cytotoxic activity of hI-con1 against human melanoma and prostate tumor cells has previously been demonstrated by Hu et al. [ 9 – 11 ]. In a recent study, we extended the results of Hu et al. evaluating the cytotoxic potential of hI-con1 against multiple high grade, biologically aggressive, Type II endometrial cancer cell lines [ 31 ]. In that study we found all endometrial carcinomas tested showing high TF expression, regardless of their high or low HER2/neu expression, to be highly susceptible to IDCC in the presence of effector cells [ 31 ]. In the current report, we expanded our research work with hI-con1 to multiple primary ovarian carcinoma cell lines with serous and CC histology highly resistant to chemotherapy. It is worth noting that although these cell lines were resistant to natural killer cytotoxic activity, IDCC resulted in killing of up to 50% of tumor cells in 5-h 51 Cr release assays. Taken together, these in vitro results strongly suggest that TF may provide a novel target for the treatment of chemotherapy resistant/residual ovarian disease and, potentially, their tumor vasculature, that may result in hI-con1-induced lysis of tumor cells, as well as pathological endothelial cells, in vivo [ 9 – 11 ]. For effective in vivo hI-con1 cytotoxicity, effector cells must be able to interact with the immune-conjugate at the target site, even in the presence of high concentrations of human IgG. Moreover, because complement activation may be an additional effector mechanism triggered by hI-con1, inhibition of complement activation by tumor cells due to overexpression of membrane-bound complement regulatory protein such as CD46, CD55 and CD59 may potentially reduce in vivo hI-con1 therapeutic efficacy by decreasing complement-dependent cellular cytotoxicity [ 15 – 17 ]. Consistent with this view, in this study we have evaluated the expression of mCRP on the surface of multiple chemotherapy resistant ovarian tumor cell lines overexpressing TF as well as the effect of complement and high concentration of irrelevant IgG during hI-con1-mediated killing in vitro. With the exception of CC-2 ARK-2 primary cell line, which was found to lack significant expression of CD59 and to be highly sensitive to complement-depend cytotoxicity, all other primary chemotherapy resistant ovarian tumors were found to express high levels of CD46, CD55 and CD59 mCRPs. Importantly, consistent with previous reports demonstrating CD59 expression as the most effective surface inhibitor to protect cells from complement attack [ 15 , 17 , 32 ], the addition of high concentrations (50%) of human plasma resulted in a significant increase in IDCC against CC-2 ARK-2 cell line. In contrast, CD59 expressing CC-1 ARK-1 tumor cells were found highly resistance to complement attack in the presence or absence of hI-con1. Indeed, in these experiments a significant decrease in hI-con1-mediated cytotoxicity was detected in the presence of plasma with or without heat inactivation. These results are explained by the high resistance of CC-1 ARK-1 to complement and the concomitant inhibition of IDCC mediated by the large number of irrelevant IgG present in human plasma. This interpretation of the results is consistent with our previous findings [ 16 ] as well as those of others [ 17 ] who also identified the inhibitory effect of endogenous non-specific IgG during ADCC assay in vitro in the presence of plasma. Furthermore, knockdown of CD59 mCRP by siRNA in CC-1 ARK-1 resulted in a significant increase in IDCC in the presence of plasma. These results support the importance of CD59 expression in TF overexpressing tumor cells as a major component of resistance to complement attack against ovarian tumors. Moreover, these results also suggest that CD59 silencing may represent a potential target for ovarian cancer gene therapy [ 33 ] that may potentially synergize with IDCC in vitro as well as in vivo. The degree of ADCC in vitro and in vivo is known to be mainly dependent on the number and function of NK cells [ 34 , 35 ]. NK cells are considered the best suited lymphocytes for ADCC because they carry F c γ RIII receptors, which are activator molecules, and not F c γ RIIb receptors, which inhibit ADCC [ 34 – 37 ]. NK cells also carry receptors for IL-2, a cytokine which induces expansion of the NK cell population and enhances their function [ 36 ]. Outpatient low dose IL-2 has, therefore, been previously used to enhance cancer patients’ immune response to monoclonal antibodies with little toxicity [ 34 – 37 ]. These findings are particularly interesting because a modulation of the number and function of NK cells has been associated with tumor progression in both experimental and animal models and pre-treatment of the PBL with IL-2 can increase the cytotoxicity levels in patients with suppressed ADCC to levels similar to those of normal donors [ 36 , 37 ]. Consistent with this data, our in vitro experiments reveal a significant increase of IDCC after the brief incubation of PBL and tumor cells with IL-2 compared to the cytotoxicity induced by hI-con1 in the absence of IL-2. IL-2 seems to therefore enhance the cytotoxic potential of the effector cells. The administration of low doses of IL-2 might therefore be a valid therapeutic option in order to increase IDCC in heavily pretreated ovarian carcinoma patients. The use of the human hI-con molecule in animals harboring xenografts of human ovarian cancer has not been investigated in our study. This is because SCID or Nude mice harboring human xenografts represent suboptimal models for studying the efficacy of hI-con as a new form of immunotherapy against human tumors due to the fact that: (a) human hI-con (unlike murine I-con not available to our study) does not cross react with mouse TF expressed on the host tumor vessels feeding human ovarian xenografts in vivo, (b) these animals are profoundly immunocompromised and, (c) a low level of cross reactivity is present between mouse NK (present in limited amount in Nude mice) and the human hI-con FC receptor. Nevertheless, previous studies have demonstrated the safety and efficacy of using the murine I-con in vivo in mice carrying mouse and human tumors [ 9 – 11 ] or harboring human implants of pathologic endometriosis [ 38 ], as well as in eradicating choroidal neovascularization in mouse and pig models of wet-form macular degeneration [ 39 ]. Thus, current studies in multiple animal models suggest that hI-con1 use is potentially safe in vivo and, therefore, may have great potential as a neovascular targeting agent with broad applications for the immunotherapy of pathological angiogenesis-involved diseases in humans. In conclusion, expression of TF in chemotherapy-resistant ovarian carcinomas and specifically in CC ovarian tumors makes hI-con1 an attractive agent for immunotherapy of chemotherapy-resistant ovarian disease. In this study, we have demonstrated significant hI-con1-mediated killing of high grade, primary chemotherapy-resistant ovarian cancer cell lines for which treatment options are very limited. hI-con1 might therefore represent a fascinating new addition to the treatment of this aggressive disease and, potentially, multiple other human tumors overexpressing TF.

Epithelial ovarian carcinoma (EOC) is the fifth most common malignancy affecting women in the United States with 21,800 new cases and 13,850 deaths estimated for 2010 [ 1 ]. At the time of diagnosis, two-thirds of patients have advanced disease (i.e., Stage III–IV), and those with advanced disease have a response rate of 73–77% after first-line therapy with platinum agents and paclitaxel [ 2 ]. Unfortunately, ovarian cancer will recur in the majority of patients, and an important prognostic factor is whether the recurrence occurs at 6 months (platinum-sensitive disease) after completion of chemotherapy. Patients with platinum-resistant disease have response rates of<10% when re-treated with platinum compounds, and the prognosis for these patients remains dismal. These figures illustrate the dire need for novel approaches to treat chemotherapy-resistant ovarian cancer. Angiogenesis, the formation of new vessels from pre-existing vasculature, is critical to ascites development and metastasis in ovarian cancer [ 3 ]. Tissue factor (TF), a transmembrane receptor for coagulation factor VII/VIIa, is aberrantly expressed in human cancers and on endothelial cells within the tumor vasculature, albeit not on the surface of normal vascular endothelium [ 4 , 5 ]. Importantly, tumor cells characterized by a high production of TF and vascular endothelial growth factor (VEGF), a crucial initiator of angiogenesis, are known to generate solid tumors characterized by intense vascularity and highly aggressive behavior [ 6 ]. Consistent with this view, several studies have shown that VEGF is overexpressed in a variety of human tumors including ovarian carcinoma [ 3 ] and the degree of neovascularization has been previously associated with a poor overall and disease-free survival in patients with advanced ovarian carcinoma [ 7 ]. While a direct regulation of VEGF expression in human tumor cells by the cytoplasmic tail of TF has been previously demonstrated [ 6 ], recent studies indicate that type-2 proteinase activated receptor (PAR-2) is intimately involved in TF-mediated signaling and angiogenesis [ 8 ]. These data suggest a potential direct role for TF in tumor growth [ 8 ]. The hI-con1 ™ molecule is a previously characterized, all-human immuno-conjugate molecule that specifically targets TF [ 9 – 11 ]. It is composed of two identical protein chains consisting of an inactive mutant form of human factor VII (fVII) as the targeting domain fused to human IgG 1 F c as the effector domain; the two chains are held together by the disulfide bonds in the hinge region normally present in IgG. The hI-con1 molecule is designed to bind to TF with far higher affinity and specificity than can be achieved with an anti-TF antibody. Indeed, the hI-con1 has several important advantages over monoclonal antibodies for targeting TF including: (1) the K d for fVII binding to TF is about 10 −12 M [ 12 ], in contrast to the K d of anti-TF antibodies that is typically in the range of 10 −8 to 10 −9 M for TF [ 13 ]; and (2) hI-con1 is produced by recombinant DNA technology, allowing this completely human protein to be conveniently produced for future clinical trials. Because binding of the fVII to TF could induce disseminated intravascular coagulation, a potentially lethal vascular disease, an amino acid substitution was introduced into the fVII domain of hI-con1 (Lys 341 to Ala) which inhibits initiation of the coagulation pathway without reducing its strong affinity for TF [ 9 , 14 ]. The human F c domain of the hI-con1 can thus potentially activate powerful cytolytic responses mediated by ADCC (antibody-dependent cell-mediated cytotoxicity) against both TF-expressing tumor cells and tumor vascular endothelial cells that bind the hI-con1 molecule. The efficacy of cancer immunotherapy with complement-activating antibodies in vivo may be limited by overexpression of one or more membrane-bound complement regulatory protein (mCRPs: CD46, CD55, CD59) on the surface of tumor cells as well as the presence of high concentrations of human IgG in the human plasma [ 15 – 17 ]. Consistent with this view, upregulation of mCRPs has been described for various human tumors including colorectal cancer, cervical cancer, prostate cancer and renal cell carcinoma [reviewed in 15 ]. Moreover, a significant decreased in the level of ADCC against human tumors by endogenous non-specific IgG as the major inhibitory component of human serum has been previoulsy reported by us [ 16 ] as well as others [ 17 ]. In this report we studied the expression of TF at mRNA and protein levels and that of CD46, CD55, CD59 mCRPs in multiple histological types of ovarian cancer, and evaluated for the first time the in vitro potential of hI-con1 as a novel immunotherapeutic agent against biologically aggressive and chemotherapy-resistant primary ovarian cancer cell lines overexpressing TF.

Primary ovarian tumor cell lines from five patients harboring advanced ovarian cancer were obtained from fresh tumor biopsies collected at the time of primary surgery (CC-2 ARK-2) or at the time of tumor recurrence (OSPC-1 ARK-1, OSPC-2 ARK-2, OSPC-3 ARK-3, CC-1 ARK-1), under approval of the Institutional Review Board. Source-patient characteristics of these five cell lines are described in Table 1 . Tumors were staged according to the International Federation of Gynecologists and Obstetricians operative staging system. All patients received a combination of carboplatin and paclitaxel as their primary chemotherapy regimen. All five primary ovarian cancer cell lines used in this study were found highly resistant to multiple chemotherapy drugs in vitro including carboplatin, cisplatin, paclitaxel, doxorubicin, ifofosfamide, gemcitabine and topotecan by Extreme Drug Resistant (EDR) assays (Oncotech, Irvine, CA) [ 16 ]. RNA isolation from 63 fresh-frozen ovarian tumors including 30 serous (OSPC, 1 stage I, 3 stage II, 15 stage III and 11 stage IV), 14 clear cells (CC, 6 stage I, 5 stage II and 3 stage III), 12 endometrioid (END, 1 stage I, 4 stage II, 6 stage III and 1 stage IV) and 7 undifferentiated (UND, 1 stage II, 3 stage III and 3 stage IV) carcinomas and 5 primary ovarian cancer cell lines used in the cytotoxicity experiments ( Table 1 ) were performed using TRIzol Reagent (Invitrogen, Carlsbad, CA) according to the manufacturer’s instructions. Quantitative PCR was done with a 7500 Real Time RT-PCR System (Applied Biosystems, Foster City, CA) to evaluate expression of TF in all samples. Briefly, 5 μg of total RNA from each sample was reverse transcribed using SuperScript III first-strand cDNA synthesis (Invitrogen). Five μl of reverse transcribed RNA samples (from 500 μl of total volume) were amplified by using the Taq Man Universal PCR Master Mix (Applied Biosystems) to produce PCR products specific for TF. The primers and probe for TF were obtained from Applied Biosystems (Assay ID Hs01076032_m1). The comparative threshold cycle ( C T ) method (Applied Biosystems) was used to determine gene expression in each sample relative to the value observed in the lowest nonmalignant ovarian epithelial cell sample (i.e., NOVA), using glyceraldehyde-3-phosphate dehydrogenase (Assay ID Hs99999905_m1) RNA as internal control. Formalin-fixed, paraffin-embedded tissue blocks from a separate set of 25 patients harboring stage I (8 patients), stage II (4 patients), stage III (7 patients) and stage IV (6 patients) EOC were retrieved from the surgical pathology files at the Yale New-Haven Hospital. EOC tissue samples studied by IHC included 9 OSPC, 10 CC and 6 END ovarian adenocarcinomas. Specimens were reviewed by a surgical pathologist (NB). The level of TF expression was then evaluated on the most representative block by standard immunohistochemical staining. For immunohistochemistry, 4 μm sections were cut from the formalin-fixed paraffin-embedded blocks. Following deparaffinization and rehydration, endogenous peroxidase was blocked in 3% H 2 O 2 . Steam and high pH (pH 9) were used for antigen retrieval. The slides were then incubated overnight at 4°C with monoclonal anti-TF antibody (No.4509, 1:10 dilution, American Diagnostica, Stamford, CT). EnVisionTM system (Dako, Carpinteria, CA) was used for secondary detection and the reactions were visualized with diaminobenzidine. Appropriate positive and negative controls were used with each case. Both cytoplasmic and membranous immunoreactivity was considered positive. Immunostaining was assessed using a semi-quantitative scoring system, as follows: 0, negative (0–5% staining); 1+, weakly positive (10–20%); 2+, moderately positive (20–50%); 3+ and 4+, strongly positive (50–75% and over 75%, respectively), as previously described [ 18 ]. Commercially available antibodies against membrane-bound complement regulatory protein CD46 (cat#555948), CD55 (cat#555691) and CD59 (cat#555761) were purchased from BD Pharmingen, San Diego, CA, while clinical grade hI-con1 was produced for Iconic Therapeutics Inc. (Atlanta, GA) by Laureate Pharma (Princeton, NJ) by cultivation of BHK cells transfected with a vector containing the gene sequence originally described by Hu and Garen [ 9 – 11 ]. The protein was purified by a series of chromatographic steps to a purity adequate for clinical use, formulated in 15 mM HEPES, 150 mM NaCl, 5 mM CaCl 2 , 25 mM arginine, 0.01% Tween 80, pH 7.4 buffer and sterile-filtered. Briefly, EOC cell lines and phytohemagglutinin (PHA)-stimulated control PBL were stained with 1 μg/200 μl of anti-CD46, anti-CD55, anti-CD59 or hI-con1 at a concentration of 30 μg/ml for 30 min on ice or with a commercially available FITC-conjugated mouse anti-human TF immunoglobulin (i.e., positive control, BioSource International, Camarillo, CA). After hI-con1 staining, cells were washed twice with the same buffer and secondary mouse-anti-human antibody (IgG1-FITC, catalog # F0767 Sigma-Aldrich, St. Louis, MO) was added for a further 30 min. Analysis was conducted with a FAC-SCalibur instrument using Cell Quest software (Becton-Dickinson, Franklin Lakes, NJ). CD59 specific siRNA oligonucleotides [i.e., 5′ ggaccugugu aacuuuaacuu 3′ (sense) and 3′ uuccuggacacauugaaauug 5′ (antisense)], were purchased from Ambion, Inc. (Austin, TX). Briefly, CC-1 ARK-1 cells were cultured in 6-well plates and transfected with siRNA duplexes using Lipofectamine ™ RNAiMAX (Invitrogen) following the manufacturer’s instructions. 12 nM of anti-CD59 specific siRNA were incubated with 5 μl of Lipofectamine ™ RNAiMAX. Mock transfections and nonspecific siRNA duplexes were used as negative controls. CC-1 ARK-1 cells were treated for 72 h (i.e., the time we found required for maximal down-regulation of CD59, based on qRT-PCR, data not shown), after which they were used in hI-con1-dependent cell-mediated cytotoxicity (IDCC) in the presence or absence of plasma as a source of complement. A standard 5-h chromium ( 51 Cr) release assay was performed to measure the cytotoxic reactivity of Ficoll-Hypaque-separated PBL from several healthy donors in combination with hI-con1 against EOC target cell lines. The release of 51 Cr from the target cells was measured as described [ 19 ] as evidence of tumor cell lysis after exposure of tumor cells to various concentrations of hI-con1 (ranging from 1 to 80 μg/ml) and different target/effector cells ratios. Controls included the incubation of target cells alone or with PBL or hI-con1 separately. The chimeric anti-CD20 mAb rituximab (Rituxan ® ; Genentech) was used as control in all bioassays. IDCC was calculated as the percentage of killing of target cells observed with mAb plus effector cells as compared with 51 Cr release from target cells incubated alone. A standard 5-h 51 Cr release assay identical to those used for IDCC assays was used, except that human plasma in a dilution of 1:2 was added in place of the effector cells. This human plasma was used as a source of complement to test for complement-mediated target cell lysis. To test for the possible inhibition of IDCC against ovarian cancer cell lines by physiological human plasma concentrations of γ -immunoglobulin, heat inactivated (56°C for 60 min) human plasma was diluted 1:2 before being added in the presence or absence of effector PBL. In some hI-con1-mediated cytotoxicity experiments non-heat-inactivated human plasma (diluted 1:2) was added in the presence of effector PBL while ovarian cancer cells knocked-down in CD59 expression by siRNA (i.e., CC-1 ARK-1) were also used as targets. In all experiments, controls included the incubation of target cells alone or with either lymphocytes or mAb separately. Rituximab was used as isotype control mAb. To investigate the effect of IL-2 on IDCC, effector PBL were incubated for 5 h at 37°C at a final concentration of IL-2 (Aldesleukin; Chiron Therapeutics, Emeryville, CA) ranging from 50 to 100 IU/ml in 96-well microtiter plates. Target cells were primary EOC cell lines exposed to hI-con1 (concentrations ranging from 1 to 80 μg/ml), whereas controls included the incubation of target cells alone or with PBL in the presence or absence of IL-2 or mAb, respectively. Rituximab was used as a control mAb. IDCC was calculated as the percentage of killing of target cells observed with hI-con1 plus effector PBL, as compared with target cells incubated alone. Each experiment was performed with PBL collected from multiple normal donors, with results from a representative donor presented. qRT-PCR data were evaluated with unequal-variance t -test for differences between EOC-versus-NOVA and ovarian tumors with different histology. Differences in TF expression by IHC and flow cytometry were analyzed by the unpaired t -test. Kruskal–Wallis test and chi-square analysis were used to evaluate differences in hI-con1-induced cellular cytotoxicity levels in primary tumor cell lines. Statistical analysis was performed using SPSS version 17 (SPSS, Chicago, IL). A P value of <0.05 was considered statistically significant.