Targeting Serous Epithelial Ovarian Cancer with Designer Zinc Finger Transcription Factors*

Background: There is a need for novel targeted therapies for metastatic ovarian cancers. Results: We reactivated the tumor suppressor Maspin in ovarian carcinoma cells by delivering tumor-specific nanoparticles encapsulating a chemically modified ATF-mRNA. Conclusion: LPR nanoparticles encapsulating the ATF mRNA inhibited ovarian cancer tumor growth in a mouse model. Significance: We report the first non-viral delivery of an ATF in vivo and the discovery of novel anti-metastatic targets for ovarian cancer. Ovarian cancer is the leading cause of death among gynecological malignancies. It is detected at late stages when the disease is spread through the abdominal cavity in a condition known as peritoneal carcinomatosis. Thus, there is an urgent need to develop novel therapeutic interventions to target advanced stages of ovarian cancer. Mammary serine protease inhibitor (Maspin) represents an important metastasis suppressor initially identified in breast cancer. Herein we have generated a sequence-specific zinc finger artificial transcription factor (ATF) to up-regulate the Maspin promoter in aggressive ovarian cancer cell lines and to interrogate the therapeutic potential of Maspin in ovarian cancer. We found that although Maspin was expressed in some primary ovarian tumors, the promoter was epigenetically silenced in cell lines derived from ascites. Transduction of the ATF in MOVCAR 5009 cells derived from ascitic cultures of a TgMISIIR-TAg mouse model of ovarian cancer resulted in tumor cell growth inhibition, impaired cell invasion, and severe disruption of actin cytoskeleton. Systemic delivery of lipid-protamine-RNA nanoparticles encapsulating a chemically modified ATF mRNA resulted in inhibition of ovarian cancer cell growth in nude mice accompanied with Maspin re-expression in the treated tumors. Gene expression microarrays of ATF-transduced cells revealed an exceptional specificity for the Maspin promoter. These analyses identified novel targets co-regulated with Maspin in human short-term cultures derived from ascites, such as TSPAN12, that could mediate the anti-metastatic phenotype of the ATF. Our work outlined the first targeted, non-viral delivery of ATFs into tumors with potential clinical applications for metastatic ovarian cancers.

Ovarian cancer is the leading cause of death among gynecological malignancies. It is detected at late stages when the disease is spread through the abdominal cavity in a condition known as peritoneal carcinomatosis. Thus, there is an urgent need to develop novel therapeutic interventions to target advanced stages of ovarian cancer. Mammary serine protease inhibitor (Maspin) represents an important metastasis suppressor initially identified in breast cancer. Herein we have generated a sequence-specific zinc finger artificial transcription factor (ATF) to up-regulate the Maspin promoter in aggressive ovarian cancer cell lines and to interrogate the therapeutic potential of Maspin in ovarian cancer. We found that although Maspin was expressed in some primary ovarian tumors, the promoter was epigenetically silenced in cell lines derived from ascites. Transduction of the ATF in MOVCAR 5009 cells derived from ascitic cultures of a TgMISIIR-TAg mouse model of ovarian cancer resulted in tumor cell growth inhibition, impaired cell invasion, and severe disruption of actin cytoskeleton. Systemic delivery of lipid-protamine-RNA nanoparticles encapsulating a chemically modified ATF mRNA resulted in inhibition of ovarian cancer cell growth in nude mice accompanied with Maspin re-expression in the treated tumors. Gene expression microarrays of ATF-transduced cells revealed an exceptional specificity for the Maspin promoter. These analyses identified novel targets coregulated with Maspin in human short-term cultures derived from ascites, such as TSPAN12, that could mediate the antimetastatic phenotype of the ATF. Our work outlined the first targeted, non-viral delivery of ATFs into tumors with potential clinical applications for metastatic ovarian cancers.
Epithelial ovarian carcinoma (EOC) 3 is the seventh most fatal cancer worldwide and the deadliest malignancy affecting the female reproductive organs (1). Serous ovarian carcinoma (SOC) is the most common form of EOC, comprising 30 -70% of the cases (2). SOC is predominantly associated with p53 mutations, and loss of BRCA1/2 may also predispose to the development of the disease (3,4). This form is mostly detected at advanced stages, when disease is widely spread and metastasized into the abdomen, in a condition known as peritoneal carcinomatosis. Late diagnosis is explained by the absence of alarming symptoms and the lack of effective screening meth-ods. The lethality of this condition is due not only to late diagnosis but also to transient response to available therapies. Thus, despite achieving optimal de-bulking with surgery and obtaining adequate response to adjuvant chemotherapy, the majority of cases will recur, and patients finally die because of resistant metastatic disease (5)(6)(7). Unfortunately, the discovery of biomarkers of metastatic progression and the development of more effective treatments for SOC has been impeded due to our limited understanding of the etiology and progression of the disease.
Genetically engineered mouse models (GEMMs) of epithelial cancers represent powerful model systems as they recapitulate the essential molecular hallmarks of disease development and progression that occur in humans (8). One of those models, the C57BL/6 TgMISIIR-TAg transgenic mouse, develops EOC with metastatic features. Mouse ovarian carcinoma (MOVCAR) cell lines derived from metastatic lesions (ascites) of TgMISIIR-TAg mice recapitulate essential features of SOC, particularly the metastatic potential (9). These cells have the advantage that they can be easily manipulated in vitro to elucidate novel biomarkers of metastatic disease and to establish novel delivery systems for therapeutic intervention (9). Our laboratory has previously described a therapeutic approach to target tumor and metastasis suppressors in cancer cells using arrays of engineered, sequence-specific C 2 H 2 zinc finger (ZF) domains (10). Each ZF is composed of a recognition ␣-helix that binds 3 bp of DNA with high selectivity (11). Six zinc finger (6ZF) arrays read an 18-base pair (18-bp) sequence that is potentially unique in the human genome and provide a high degree of genomic specificity and selectivity (12). Engineering binding specificity is achieved by grafting the ␣-helical domain of each ZF known to interact with the target DNA triplet (13). We have constructed multimodular 6ZF proteins referred as artificial transcription factors (ATFs), recognizing sequences in targeted promoters with dissociation constants in the picomolar range (10,(13)(14)(15)(16)(17)(18). In an ATF, the 6ZF scaffold can be linked to a variety of protein modules to promote transcriptional activation (19 -23), repression (24,25), and more recently, epigenetic editing (26).
We have recently described an ATF (ATF-126) targeting the human mammary serine protease inhibitor (Maspin), a tumor and metastasis suppressor gene initially identified in breast cells (27). Although most epithelial cells express high levels of Maspin, the promoter is down-regulated by epigenetic mechanisms in several cancers, such as breast (28,29) and ovarian cancer (30). Endogenous Maspin reactivation by ATF-126 was associated with decreased tumor growth by enhancement of apoptosis (10), cell invasion (10), and suppression of metastatic colonization in breast (18) and lung (31) cancer cells. Although most studies have been focused on breast, prostate, and lung cancer, the functional role of Maspin as tumor and metastasis suppressor in EOC has not been investigated. Both cytoplasmic and nuclear Maspin expression has been reported in some primary ovarian tumors and cell lines, with nuclear expression being a favorable prognosis factor in ovarian cancer patients (32). Decreased nuclear Maspin expression has been associated with tumor grade and disease progression, suggesting a role of Maspin silencing in advanced stages of ovarian cancer, potentially in metastatic disease (32)(33)(34).
In this manuscript we took advantage of ZF technology to target the endogenous murine Maspin promoter in metastatic MOVCAR cell lines to interrogate the functional role and therapeutic potential of Maspin in EOC. Delivery of engineered ZF proteins has historically been a major limitation for translational applications, most notably in cancer models. This paper reports the first non-viral delivery of ATF mRNA for future therapeutic treatment of serous epithelial ovarian cancer of advanced stage and potentially for metastatic disease. We also describe a panel of novel targets co-regulated with Maspin, which could be used as potential biomarkers for future diagnosis and treatment of metastatic ovarian cancer.

EXPERIMENTAL PROCEDURES
Construction of a Murine-specific ATF Recognizing the Maspin Promoter (mATF)-The murine-specific mATF was designed to recognize an 18-bp duplex 5Ј-GAAGACCTGG-GTGTGGTC-3Ј located at Ϫ127 nucleotides upstream the first ATG triplet (translation start site) in the murine Maspin promoter (Fig. 1A). The specific ZF coding sequences were generated by overlapping PCR as previously described (10), and the specific ␣-helical sequences used for the PCR are shown in supplemental Fig. S1. The 6ZF cassette was cloned into the SfiI sites of the retroviral vector pMX-ss-IRES-GFP (10) for endogenous gene regulation studies.
Retroviral Transduction of mATF in Cancer Cell Lines-Retroviral transduction in the packaging cell line 293TGagPol (ATCC Number CRL-11654) and infection of the host cell lines was performed as described (17) using the retroviral vector pMX-IRES-GFP and the envelope protein plasmid pMDG.1. The cells were infected every 8 -12 h for a total of 3 times and collected or fixed 72 h after the first infection for further experiments. The Retro-X Tet-On expression system (catalog number 632104 CloneTech, Mountain View, CA) was used to generate a MOVCAR 5009 stable cell expressing the human Maspin cDNA (purchased from Origene, catalogue number SC303231, Rockville, MD) as described (18). To induce Maspin cDNA expression, cells were treated for 48 h with doxycycline 100 ng/l and collected for further experiments. The 293TGagPol cells were grown in DMEM (Sigma) supplemented with 10% of fetal bovine serum (FBS) and antibiotic-antimycotic solution (Sigma).

Treatment of the MOVCAR 5009 Cell Line with Epigenetic
Inhibitors-The MOVCAR 5009 cell line was treated with the 5-aza-2ЈdC DNA-methyltransferase inhibitor and two histone deacetylase inhibitors, suberoylanilide hydroxamic acid and trichostatin A. Cells were treated at saturating concentrations of drugs, as determined empirically for the MOVCAR 5009 cell line: 5 M 5-aza-2ЈdC, 10 M suberoylanilide hydroxamic acid, and 100 nM trichostatin A. All drugs were purchased from Sigma. Cells were plated at a density of 3 ϫ 10 5 cells in 100-mm plates (Corning, NY) in DMEM media and treated with the corresponding inhibitors during 48 h. Cells were collected, and RNA was extracted with RNeasy Kit (Qiagen) and processed for quantitative real time-PCR (qRT-PCR).
qRT-PCR Assays-RNA from adherent cell cultures and tumor samples were extracted with the RNeasy kit (Qiagen) and reverse-transcribed to cDNA using the High Capacity cDNA reverse transcription kit (Applied Biosystems, Carlsbad, CA). TaqMan Fast Universal Master Mix (Applied Biosystems) and 150 ng of cDNA were used in the PCR reactions. GAPDH was used as an endogenous control. Primers used for qRT-PCR studies are shown in supplemental Table S4. The 7500 Software Version 2.0.5. (Applied Biosystems) was used to analyze the data. qRT-PCR was performed in duplicate wells and in three independent experiments. Statistical differences were determined by Student's t test considering p Յ 0.05 as significant (*), p Յ 0.01 as highly significant (**), and p Յ 0.001 as extremely significant (***).
Western Blot-Total protein extract was obtained with radioimmune precipitation assay buffer (Sigma R0278). The detection was performed with the Amersham Biosciences ECL detection system (GE Healthcare). A detailed list of the antibodies used in this study, the corresponding suppliers, and the experimental dilutions are shown in supplemental Table S5.
Chromatin Immunoprecipitation (ChIP)-ChIP assay to detect the ATF/polymerase II binding to the murine Maspin promoter was performed as described elsewhere (40,41). Protein-DNA complexes were pulled down with A/G beads (Santa Cruz, sc-2003, Santa Cruz, CA). The beads were washed three times with low salt buffer, three times with high salt buffer, and two times with Tris-EDTA buffer (40,41). The beads were eluted overnight at 65°C in elution buffer (Tris-EDTA buffer, 1% SDS, and 2 l of proteinase K). DNA was isolated by phenolchloroform extraction. The immunoprecipitated DNA was amplified by PCR using the following primers flanking the mATF binding site in the Maspin promoter: 5Ј-CTGGGTGT-GGTCACAGGTGAGC-3Ј and 5Ј-TCCCTTTGCTTACCTT-GAGTTGC-3. The primers against the murine GAPDH promoter were used as controls for the PCR amplifications: 5Ј-TACTCGCGGCTTTACGGG-3Ј and 5Ј-TGGAACAGGGA-GGAGCAGAGAGCA-3Ј (42). The PCR products were loaded in an agarose gel, and the relative -fold enrichment was determined by densitometry (ImageJ, NCBI, Bethesda, MD) and normalized to the input samples.
Immunofluorescence-Cells were seeded in fibronectincoated coverslips and fixed with 10% formalin-PBS solution at room temperature for 10 min. Cells were permeabilized with 0.5% Triton X, PBS for 15 min at 4°C, and 4% BSA, PBS was added for overnight blocking at 4°C. Fibronectin, formalin, and Triton-X were purchased from Sigma. MOVCAR 5009 cells were transduced with mATF, control, or Maspin cDNA and stained for detection of mATF, Maspin, and actin (supplemental Table S5). For immunofluorescence detection of Maspin and tetraspanin 12 (TSPAN12) on epithelial ovarian cancer specimens, paraffin sections (obtained from the UNC Tissue Procurement Core) were deparaffinized, blocked at room temperature with 5% BSA/PBS for 2 h, and incubated with primary antibody overnight at 4°C. The slides were next washed with 0.03% Tween 20, 1% BSA, PBS and stained with secondary antibodies at room temperature for 1 h. Details regarding the antibody sources and dilutions are specified in supplemental Table  S5.
Cell Viability Assay-Cell viability was determined by Cell-Titer Glo assay (Promega, Madison, WI) as described by the manufacturer. Twenty-four hours post-transduction, 1000 cells per well were seeded in 96-well plates, and cell viability was followed every 24 h for a total of 4 days. Statistical differences were determined by Student's t test considering p Յ 0.05 as significant (*), p Յ 0.01 as highly significant (**), and p Յ 0.001 as extremely significant (***).
Soft Agar Assay-Cells were collected 72 h post-transduction. Colony formation was assessed using six-well plates (Corning). Five thousand cells per well were seeded on the top layer of 0.3% agar with a 0.6% agar base layer. Plates were incubated for 20 days in a 37°C incubator with 5% CO 2 . Colonies were visualized and counted after a 2-h treatment with a 0.005% crystal violet solution. Cells were seeded in triplicate, and the experiment was repeated three times. Materials were purchased from Sigma unless otherwise stated.
Matrigel Invasion Assay-Matrigel Invasion Chambers (BD Biosciences) were used to study cell invasion as previously described (10). Briefly, cells were starved 24 h before the assay, and 1 ϫ 10 5 cells were seeded in the chambers. 24 h later cells were fixed and stained with hematoxylin and eosin.
Preparation of Modified mRNA-The mATF coding sequence and enhanced green fluorescent protein (EGFP) were cloned into pcDNA 3.1 vector flanked by an untranslated 5Ј strong Kozak translational initiation signal and a 3Ј ␣-globin sequence for higher translation efficiency and longer half-life. The untranslated region and gene of interest was amplified by PCR with 5Ј primer, CTAGAGAACCCACTGCTTACTGGC-TTATCG, and 3Ј primer, 120 TGCGTCGACACTAGTTCTA-GACCCT. The amplicons were used as templates for in vitro transcription. Modified mRNA was synthesized with Ambion MEGAScript T7 kit (Invitrogen). A mix of 1.6 g of purified amplicons, 6 mM 3Ј-0-Me-m 7 G(5Ј)ppp(5Ј)G (New England BioLabs, Ipswitch, MA), 1.5 mM guanosine triphosphate, 7.5 mM adenosine triphosphate, 7.5 mM 5-methylcytidine triphos-phate, 7.5 mM pseudouridine triphosphate, and 4 l of T7 enzyme were incubated at 37°C for 4ϳ6 h. The mRNA was purified with Ambion MEGAclear kit (Invitrogen).
Preparation of Liposome and Liposome Protamine RNA (LPR)-DOTAP (Avanti Polar Lipids, Alabaster, AL) and cholesterol (Sigma) (1:1 mol/mol) were dissolved in chloroform, and the solvent was removed under reduced pressure. The lipid film was hydrated overnight with distilled water to make the final concentration of 10 mM DOTAP and cholesterol. The liposome was sequentially extruded through 400-, 200-, 100-, and 50-nm polycarbonate membranes (Whatman, Piscataway, NJ) to form 80 ϳ 100-nm unilamellar liposomes. The LPR core was prepared by mixing 146 l of solution A (10 g of mRNA) and 146 l of solution B (3 l of 2 mg/ml protamine, 12 l of 10 mM DOTAP/cholesterol liposome, and 131 l of distilled water). The resulting solution was incubated at room temperature for 10 min. The pegylation was performed by adding 4 l of DSPE-PEG 2000 (Avanti Polar Lipids) (10 mg/ml) and 4 l of DSPE-PEG 2000 -AA (10 mg/ml) to the LPR core and incubating the mix at 50°C for 15 min. The DSPE-PEG-AA was synthesized as previously described (44).
In Vitro Transfection of Nanoparticles in MOVCAR 5009 Cells-Ninety-six well plates were seeded with 1 ϫ 10 4 cells per well. The cells were transfected with either LPR-AA-EGFP or LPR-EGFP nanoparticles equivalent to 0.5 g of modified mRNA encoding EGFP. Nanoparticles were added to each well in the presence of Opti-MEM medium (Invitrogen) and incubated at 37°C in 5% CO 2 . The medium was replaced with complete medium for 4 h post-transfection. The transfection efficiency was determined using a BD FACS Canto Flow Cytometer (BD Biosciences). MOVCAR 5009 cells were transfected with either LPR-AA-mATF or LPR-AA-control. The control refers to the mATF modified mRNA lacking the 5Ј cap nucleotide, which inhibits translation. The HA.11 antibody was applied to determine mATF expression by immunofluorescence 24 h after transfection (supplemental Table S5). Transfected cells were processed by MTT cell proliferation assays 2 days post-transfection as described by the manufacturer's instructions (Roche Applied Science). The absorbance was read at 570 nm to determine cell survival.
Molecular Beacon Design and Transfection-The software Beacon Designer (PREMIER Biosoft International, Palo Alto, CA) was employed to design a molecular beacon (MB) probe to hybridize with the mouse Maspin mRNA. The MB oligonucleotide sequence 5Ј-(Texas Red)(C6 amino)-CGCGATCTGTTT-CTGATTCAGGAAATTTCTTCATGATCGCG-(3ЈBHQ-2)-3Ј was synthesized by The Midland Certified Reagent Co. (Midland, TX). LPR nanoparticles loaded with 0.2 g of MB and 0.8 g of mATF-mRNA were added into 8-well Lab-Tek Chamber Slides. Twenty-four hours post-transfection cells were examined with fluorescence microscopy observation.
Tumor Reduction Study-4 -7-Week-old athymic nude mice (obtained from University of North Carolina animal models core facility) were inoculated with 1 ϫ 10 7 MOVCAR 5009 cells in the flank. Mice received an intravenous injection (via tail vein) of nanoparticles equivalent to 10 g of mATF-mRNA/control as a palpable solid tumor was formed (on day 16). Mice received an injection every other day for five times. The tumor sizes were monitored as an indication of therapeutic effect. Statistical analysis was undertaken using Prism 4.0 GraphPad Software. The tumors were removed and analyzed for expression of the mATF and Maspin reactivation by Western blot.
siRNA Transfection-Cells previously transduced with either control or mATF were collected 72 h post-transduction for the siRNA experiment. The murine Maspin siRNA and a nonspecific control siRNA targeting luciferase were purchased from Thermo Scientific (Walthman, MA). Transfection was performed using the manufacturer's instructions. Briefly, 1 ϫ 10 5 cells were plated in 6-well plates, and DharmaFECT 4 was used to transfect the siRNA with a final concentration of 50 nM. Transfected cells were collected 48 h post-transfection and processed for qRT-PCR analysis of Maspin and Tspan12 expression. Data were normalized to control-transfected samples.
Ovarian Cancer Patient-derived Samples-The normal ovary preparation SR06, ovarian primary tumors, and ascites were obtained from the Universidad Católica de Chile with approved Institutional Review Board protocols. These samples were resuspended in Trizol (Invitrogen) and processed for expression analyses. Human ascites (ASC13 to 17) were obtained from University of North Carolina hospitals with approved Institutional Review Board protocols. These cells were grown in RPMI media with 5% FBS in adherent tissue culture plates as described above. Tumor cells were passaged three times in tissue culture dishes and processed for qRT-PCR or Western blot. The epithelial origin of these cells was verified by Western blotting using anti-Pan-Cytokeratin antibodies (supplemental Fig. S4). For the qRT-PCR experiments the samples were analyzed using the GAPDH gene. Two additional endogenous controls recommended for ovarian tumor specimens (the genes glucuronidase ␤ (GUSB) and peptidylprolyl isomerase A (PPIA) (45,46)) were also used to validate the quantifications.
Gene Expression Microarrays-Total RNA was purified from control and mATF-transduced cells (three biological replicates) as described (18). The mRNA was amplified, labeled, and hybridized as previously described (47) using Agilent mouse 4 ϫ 180 K oligo microarrays (Agilent Technologies). All microarray data have been deposited in the Gene Expression Omnibus (GEO) data base under the accession numbers: GSM891209, GSM891210, GSM891211, GSM891212, GSM891213, and GSM891214. The probes/genes were filtered by requiring the lowest normalized intensity values to be Ͼ10 in both samples and controls. The normalized log2 ratios (Cy5 sample/Cy3 control) of probes mapping to the same gene were averaged to generate independent expression estimates. Statistical analysis of microarray for gene expression data were performed using Significance Analysis of Microarray. Gene ontology (GO) analysis was performed using EASE: the Expression Analysis Systematic Explorer (48).

mATF Binds the Murine Maspin Proximal Promoter and Upregulates Its Expression in a Panel of Cell Lines Derived from
GEMMs-To up-regulate the murine Maspin promoter, we designed a mATF targeting an 18-bp duplex located ϳ127 nucleotides upstream the translation start site (Fig. 1A). This specific site was chosen because an 18-bp site at the same relative position in the human promoter was successfully targeted by an artificial zinc finger factor, ATF-126 (10). Moreover, because the targeted site was not fully conserved between the murine and the human promoter, the 6ZF helices were re-engineered to regulate the murine counterpart (supplemental Fig. S1).
To have an initial assessment of the potency of the mATF in reactivating Maspin, we took advantage of a panel of breast and ovarian cell lines derived from several GEMMs (Fig. 1, B and C). We delivered the mATF and an empty vector (control) using the pMX-IRES-GFP retroviral vector system (17). The mATF was able to significantly up-regulate Maspin in all the murine cancer cell lines examined carrying low or silenced Maspin. The gene expression data were normalized to the parental MMTV-PyMT breast cancer line carrying the highest expression of Maspin (Fig. 1B). As expected, no significant differences in gene expression were observed between the parental and controltransduced cell lines. We found that although Maspin was expressed in some breast cancer cell lines, the gene was silenced in all the ovarian cell lines derived from the TgMISIIR-TAg transgenic mouse. Because of their high invasive potential and their reported phenotypic resemblance to high grade EOC, we focused further experiments using the MOVCAR 5009 cell line. In this background the mATF increased Maspin mRNA expression by more than 70,000-fold relative to the parental line, empty vector control cells, or a control vector carrying no ZFs (VP64-SS sample, Fig. 2A). This regulation required a specific DNA binding as the delivery of a scramble library of ZF proteins linked to VP64 (67) resulted in a significant reduction (more than 3 orders of magnitude) of Maspin activation relative to the mATF, indicating that the specific 6ZF domains used for the construction of the protein provided sequence-selectivity ( Fig. 2A). Interestingly, the reactivation of Maspin by the engineered protein was clearly superior to the reactivation of the endogenous gene observed when the parental cell line MOVCAR 5009 was treated with epigenetic inhibitors, such as the DNA methyltransferase inhibitor 5-aza-2ЈdC (activation of 2564.9-fold over vehicle-control cells) and the histone deacetylase inhibitor trichostatin A (72.96-fold activation over vehicle control-treated cells). Unlike the human Maspin promoter (5), the murine counterpart was not significantly up-regulated upon exposure to the histone deacetylase inhibitor suberoylanilide hydroxamic acid. Nevertheless, the clear reactivation of Maspin with the methyltransferase inhibitor 5-aza-2ЈdC suggested that Maspin was silenced by promoter methylation in the MOVCAR 5009 cell line. Our results are in agreement with previous mapping of methylated CpG islands by sodium bisulfate sequencing in human ovarian cell lines (30) and suggest that the Maspin promoter in metastatic ovarian cell lines is down-regulated by epigenetic mechanisms.
To demonstrate the binding of the mATF to its cognate sequence in the Maspin promoter, ChIP assays were performed in the MOVCAR 5009 cells (Fig. 2B). Chromatin from empty vector (control) and mATF-transduced cells was isolated, sonicated, and processed for immunoprecipitation assays using an anti-HA antibody to detect the tag engineered at the C terminus of the ZF construct (Fig. 2B). The ChIP products were amplified using primers specific for the Maspin promoter sequences flanking the mATF site, and the enrichment of the PCR products was evaluated by densitometry. As shown in Fig.  2B, mATF-transduced samples, but not controls, yielded the expected Maspin-specific 135-bp amplicon, suggesting that the mATF was physically bound to its cognate site in the Maspin promoter. Furthermore, when the ChIP assay was performed using an anti-RNA-polymerase II antibody, the immunoprecipitated DNA was enriched in the mATF samples relative to control. This result suggests that the mATF transduction was increasing the RNA-polymerase II binding in the Maspin promoter, resulting in an enhanced transcriptional activity, as confirmed by qRT-PCR and Western blot (Fig. 1, B and C).
The reactivation of Maspin by mATF in the MOVCAR 5009 cell line was further confirmed by immunofluorescence (Fig.   2C). In mATF-transduced cells, Maspin was found localized in the cytoplasm and in the nucleus of the tumor cells. When a Maspin cDNA-expressing retrovirus was used to exogenously express Maspin, only cytoplasmic Maspin was detected in the cells. The molecular basis of the differential Maspin localization observed between the mATF-and the cDNA-transduced cells is not known. A molecular hallmark of the ATFs is their ability  Student's t test was applied to analyze the difference in tumor cell survival between the control and mATF cells, with p Ͻ 0.001. Three different transductions were performed, each one monitored in triplicate wells. B, mATF decreases anchorageindependent growth of MOVCAR 5009 cells. Soft agar colony formation assays were performed on control and mATF-transduced cells. The arrow in the picture (left) points to a colony amplified on the left corner. Colony quantification was performed 20 days after the seeding. Three independent experiments were carried out, and Student's t test was applied to analyze the data; p ϭ 0.0054. C, mATF-transduced cells abolished tumor formation potential, as evaluated by tumorsphere assay formation. Cells were retrovirally transduced with either mATF or control and seeded in low-attachment culture plates with spheroid media (43). After 10 days, representative pictures of the experiment are shown outlining the abolishment of spheroid formation in mATF transduced cells. (p ϭ 0.0002). D, mATF transduction in MOVCAR 5009 cells inhibits the invasion potential of the cells relative to control, as assessed by Matrigel-invasion assays (10). A representative picture of the fixed cells after 24 h of invasion is indicated. Three independent experiments were performed, and Student's t test was applied to analyze the data; p ϭ 0.0001. E, immunofluorescence staining of actin (green) and Maspin (red) on mATF and control-transduced cells is shown. To detect actin, Alexa 488-phalloidin was applied, and anti-Maspin-Alexa 594 was used as the secondary antibody for Maspin detection. Hoechst 33258 was used to stain the nuclei (blue).
to activate the endogenous promoters (23), resulting in transcription of the physiologically relevant mRNA variants in the proper ratios, which represents a fundamental difference from the exogenous delivery of cDNAs (18).
mATF Expression Decreases Tumor Cell Viability and Inhibits Tumor Cell Invasion in MOVCAR 5009-To assess the effect of the mATF expression on the phenotype of MOVCAR 5009 cells, we first performed cell viability assays using a CellTiter Glo assay (Fig. 3A). The mATF-transduced cells and empty vector control were harvested 72 h post-transduction, and the cell viability was monitored every 24 h for a total of 4 days. We found that the ATF significantly decreased the tumor cell viability over time relative to control (p Ͻ 0.001). A significant decrease in cell viability was also observed when mATF was transduced in other MOVCAR cell lines carrying silenced Maspin, such as MOVCAR 5447 and 5612 (supplemental Fig. S2). Transduction of mATF also suppressed anchorage-independent growth of MOVCAR 5009 cells in soft agar colony formation assays (Fig. 3B). Furthermore, mATF completely suppressed the ability of MOVCAR 5009 cells to form tumorspheres in non-adherent plates, demonstrating that the engineered protein suppressed the tumorigenic properties of the cells in vitro (Fig. 3C). In addition to tumor suppression, Maspin re-expression in breast cancer cells has been associated with decreased motility and invasion (10,18,49). Consistent with this second function of Maspin, we found that the mATF strongly suppressed the ability of MOVCAR 5009 cells to invade the Matrigel (Fig. 3D). Notably, the mATF-expressing cells underwent severe changes in cell shape, outlined by a disruption and loss of actin filaments (Fig. 3E).

LPR Nanoparticles Encapsulating a Chemically Modified mATF mRNA Are Efficiently Internalized into MOVCAR 5009 Cells in Vitro and in Tumor Allografts-We next investigated
whether the mATF could be delivered in the tumor cells using a non-viral method by delivering chemically modified messenger RNA, eliminating the possibility of undesired genomic integration due to viral vector or plasmid DNA delivery. We generated LPR nanoparticles encapsulating the mATF mRNA. LPRs are liposome-based nanocarriers capable of delivering chemically modified RNAs to the targeted tumor site with high efficiency and specificity ( Ref. 76; Fig. 4A). The chemical ligand anisamide (AA) was grafted at distal end PEG polymer on the surface of LPR particles to target the sigma receptors (SRs), which are overexpressed in a large spectrum of human tumors and cell lines (50). In our study we employed LPR-AA nanoparticle technology to deliver the mATF to the MOVCAR 5009 cell line. The expression of SR1 in MOVCAR 5009 cells was confirmed by Western blotting (supplemental Fig. S3). To evaluate the transfection efficiency of the LPR-AA formulation, we encapsulated an mRNA encoding EGFP, and the expression level and percentage of EGFP ϩ cells were evaluated by flow cytometry. As shown in Fig. 4B, MOVCAR 5009 cells were transfected with ϳ85% efficiency and, thus, with a yield comparable with retroviral or lentiviral vectors. In addition, the transfection efficiency was ligand-dependent. AA-LPR resulted in a 3-fold transgene expression reflected by fluorescence intensity and a 2-fold percent of transfected cells compared with non-targeted LPR. This implied that efficient gene delivery with high specificity was achieved as a result of enhanced receptor mediated cellular uptake. Next, we investigated whether the delivery of functional LPR-AA nanoparticles encapsulating the mATF mRNA resulted in efficient translation of the mATF and in nuclear localization of the designer protein. The expression of mATF was detected by immunohistochemistry using an anti-HA antibody to detect the C-terminal epitope tag of the protein. As shown in Fig. 4C, the functional mATF, but not an inactive control ATF mRNA without 5Ј cap structure, was detected in punctate structures in the nucleus of the cells, demonstrating effective translation. To investigate whether the mATF was transcriptionally competent in the cells, we co-transfected either the mATF mRNA or the uncapped ATF mRNA control, with a molecular beacon designed to specifically hybridize with Maspin mRNA. In the presence of untranslated control mRNA, no Maspin mRNA transcript was detected by the molecular beacon in the cells. However, when active mATF mRNA was included in the nanoparticle, an activation of endogenous Maspin mRNA was detected by the molecular beacon probe that allowed the real-time imaging of Maspin mRNA transcripts with a perinuclear localization (Fig. 4D). Overall, these studies demonstrated the delivery of our designer ZF protein into nanoparticles to target the ovarian cancer cells under study and effective mATF translation leading ultimately to Maspin re-expression.
At last, the ability of mATF-loaded nanoparticles to induce phenotypical changes in the cell line was evaluated in vitro (Fig.  5A). We found that LPR-AA-mATF nanoparticles significantly decreased survival of the MOVCAR 5009 cells by ϳ40% relative to control-loaded nanoparticles, as determined by a cell viability MTT assay (Fig. 5A). To assess the efficiency of LPR-AA nanoparticles to target MOVCAR 5009 cells in vivo, LPR-AA-mATF and control nanoparticles were delivered via intravenous injections into mice bearing MOVCAR 5009 subcutaneous allografts. Nanoparticle injections were performed via tail vein every 2 days with a total of 5 treatments. 24 days postinjection, a significant inhibition of tumor growth was detected on LPR-AA-mATF-injected animals relative to control (p ϭ 0.0220) or vehicle-treated animals (p ϭ 0.0253) (Fig. 5B). Tumors samples were processed by Western blot to demonstrate effective expression of the mATF in the tumors and concomitant Maspin regulation (Fig. 5C). In summary, these studies demonstrated that nanoparticle delivery of mATF mRNA led to a significant therapeutic effect inhibiting tumor cell growth of MOVCAR 5009 tumor allografts.
Genome-wide Changes in Gene Expression by mATF Delivery in MOVCAR 5009 Cells-To address the specificity of mATF in the MOVCAR 5009 cells, we performed gene expression microarrays using Agilent mouse 4 ϫ 180 K oligo microarrays. Cells were transduced in three independent biological replicates with either an empty retroviral vector (control) or with a mATF-expressing retrovirus. Significantly differentially expressed targets were analyzed using a two-class, unpaired SAM, with a false discovery rate Ͻ5%. Supplemental Tables S1 and S2 contain the list of genes significantly up-regulated (505 genes) and down-regulated (383 genes), respectively. A gene ontology (GO) analysis of the gene sets revealed gene pathways involved in nuclear signaling, signal transduction by small GTPases, cell motility, and adhesion (supplemental Table S3).

FIGURE 5. Transfection of LPR-AA-mATF nanoparticles decreases the proliferation of MOVCAR 5009 cells in vitro and inhibits tumor growth in vivo.
A, delivery of LPR-AA-mATF nanoparticles decreases survival of MOVCAR 5009 cells relative to control-treated targeted nanoparticles. Cell survival was determined by MTT cell proliferation assay on cells transfected with either LPR-AA-mATF or control. Student's t test was applied to analyze the difference on survival between mATF and control-transfected cells; p ϭ 0.0002. Data represent the averages and S.E. of three independent experiments, each one done in triplicate wells. B, LPR-AA-mATF nanoparticles inhibit the growth of MOVCAR 5009 cells in subcutaneous allografts in nude mice. MOVCAR 5009 cells were inoculated in nude mice to induce subcutaneous tumors, and nanoparticles were injected intravenously at the time points indicated in the graph (arrows). Student's t test was applied to analyze the difference between control (n ϭ 9) and mATF treatment (n ϭ 9; p ϭ 0.0220) and between vehicle and mATF (p ϭ 0.0253). C, mATF and Maspin expression were detected on the treated tumors by Western blot. An anti-HA antibody was used to detect the ATF by Western blot.
Notably, the gene expression microarray datasets revealed only 3 genes, including Maspin, with a relative mRNA up-regulation of 30-fold or higher over empty vector control (Table 1). These findings were validated by qRT-PCR, as indicated in Fig. 6A. The strongest mATF up-regulated target was Maspin, outlining the high degree of potency and specificity of the mATF in MOVCAR 5009 cells. The genes Tetraspanin12 (Tspan12), calcium and integrin binding family member 2 (Cib2), potassium channel tetramerization domain containing 12 (Kctd12), and potassium large conductance calcium-activated channel, subfamily M, ␤ member 4 (Kcnmb4) were also found up-regulated in mATF-transduced samples relative to control-transduced or parental line (p Ͻ 0.0001). As shown in Table 1, the genes upregulated by mATF are membrane-associated proteins involved in signal transduction, cell adhesion and migration, and tumor suppression. The down-regulated targets (Ϸ0.13fold relative to control) included potential oncogenes such as Wingless-related MMTV integration site 10a (Wnt10a) and CCAAT/enhancer binding protein ␦ (Cebpd) and regulators of cell adhesion, such as Nidogen1 (Nid1) (Fig. 6B). In summary, the analysis also revealed important novel targets co-regulated with Maspin, which could mediate the anti-metastatic and antitumorigenic phenotype of the ZF factor in EOC cell lines. Whether these targets represent indeed bona fide direct targets of the ATF or downstream genetic cascades of Maspin is presently not known and will require future molecular analyses, for example by zinc finger mapping in their promoter sites. However, these genome-wide data documented a high degree of potency and selectivity of mATF for its cognate Maspin target in MOVCAR 5009 cells.
Maspin and TSPAN12 Are Down-regulated in Short-term Cultures Derived from Human Ascites-We focused on the target TSPAN12, a member of the tetraspanin family of cell-surface proteins, some of which have been reported to mediate metastasis suppressive functions in several cancer models (51). We interrogated whether TSPAN12 was co-regulated with Maspin in a panel of primary ovarian cancer tumors specimens and cell lines and whether the gene was silenced in ascites preparations from ovarian cancer patients. These samples were analyzed for Maspin and TSPAN12 expression by qRT-PCR (Fig.  7A). Data were normalized to the SR06 sample, a normal total ovary preparation. We found that Maspin was expressed in some ovarian primary tumors specimens and ovarian cell lines but was low or undetectable in most of the ascites samples. TSPAN12 mRNA expression followed a similar pattern and was significantly lower in the vast majority of the ascites analyzed as compared with primary tumors and ovarian cancer cell lines. In one case examined, from which pair samples of tumor and ascites were available, the primary tumor (UC064) had significantly higher Maspin and TSPAN12 mRNA expression than the ascites derived from the same patient (ASCUC064). As shown in Fig. 7D and supplemental Fig. S5, the expression of Maspin and TSPAN12 in primary EOC tumors was confirmed by immunofluorescence in epithelial ovarian cancer tissue specimens, Maspin having a predominantly cytoplasmic distribution, whereas TSPAN12 was localized on the plasma membrane. The down-regulation of Maspin and TSPAN12 in the short-term cultures from ascites fluids was also validated by Western blotting using the OVCAR-3 tumor cell line as reference (Fig. 7B).
The above results suggest that Maspin and TSPAN12 are down-regulated in metastatic cells, opening the possibility that TSPAN12 could be a downstream target of Maspin in ovarian cancer. Alternatively, TSPAN12 could be regulated as a result of mATF-dependent gene cascades, for example, by direct binding of the ATF or by indirect regulation of downstream targets. To address whether the activation of TSPAN12 in mATF-transduced cells was dependent on Maspin, we challenged MOVCAR 5009 cells transduced with mATF with either a Maspin-specific siRNA or a control siRNA. Changes in TSPAN12 were assessed in the transfected cells by qRT-PCR. As shown in Fig. 7C, the Maspin knockdown resulted in a significant down-regulation of TSPAN12 in the MOVCAR 5009 cells, suggesting that Maspin could be controlling TSPAN12 activation in these cells. In addition, we were unable to find putative binding sites of the mATF in the TSPAN12 promoter, suggesting that TSPAN12 is not a direct target of the ATF. Moreover, more analyses need to be done, for example, using genome-wide ZF mapping to validate the exact number and genomic location of all mATF-binding sites in the cancer cell genome. In summary, the gene expression analyses demonstrated that the ZF technology could be used for cancer inves-  AUGUST 24, 2012 • VOLUME 287 • NUMBER 35

Designer Zinc Fingers in Epithelial Ovarian Cancer
tigators to identify novel targets potentially involved in migration and metastatic progression for future diagnosis and treatment of aggressive metastatic disease.

DISCUSSION
Mammary serine protease inhibitor (Maspin) has emerged as an important clinical target possessing tumor and metastasis suppressive functions. Although the function of Maspin has been well examined in breast (52), lung (53), and prostate (54) models, its function as a tumor and metastasis suppressor in ovarian cancer has not been well investigated. In this paper we took advantage of ZF technology to interrogate the role of Maspin in metastatic ovarian cancer. We have described an mATF able to up-regulate endogenous Maspin in metastatic murine ovarian cancer cell lines derived from a TgMISRII-TAg GEMM of serous ovarian cancer (9). The mATF had an exceptionally strong transcriptional activation in multiple murine cancer cell models derived from GEMMs, with levels of expression that were comparable with ectopic cDNA overexpression. The higher transcriptional potency of the mATF relative to the previously reported human-specific ATF (ATF-126) could be related to the configuration of the chromatin and the promoter context of the mouse promoter. Moreover, MOVCAR 5009 cells derived from ascitic fluids from the C57BL/6 TgMISRII-TAg GEMM exhibited an epigenetically silenced Maspin, which could be partially re-activated by DNA methyltransferase (DNMT) and histone deacetylase inhibitors. In this background the mATF was able to bind its target site in the cell and reactivated Maspin mRNA expression by more than 70,000fold relative to control-transduced and parental cells. The much higher potency of the ATF in activating Maspin relative to chromatin remodelers could be explained by the fact that the ZFs confer locus selectivity. In contrast, epigenetic inhibitors, such as DNA methyltransferase (DNMT) and histone deacetylase inhibitors, are believed to target a large spectrum of tumor suppressors. In addition, the Maspin-specific ZFs are linked to the VP64 transactivator domain, which enhances transcription by recruitment of several components of the transcription machinery (55). Consistently, our ChIP analysis demonstrated an enrichment of RNA-polymerase II binding in the Maspin promoter upon mATF overexpression. Interestingly, the transduction of mATF elevated both cytoplasmic and nuclear forms of Maspin, whereas the exogenous cDNA overexpression upregulated only cytosolic Maspin. Recent publications outline the importance of nuclear Maspin as being a favorable prognosis factor in breast (56), lung (57), and ovarian carcinomas (33). This result emphasizes the significance of the endogenous mechanisms of regulation of targeted promoters, which results in activation of the physiologically relevant isoform of the targeted gene. The phenotypic outcomes of Maspin reactivation in metastatic MOVCAR 5009 cells recapitulated the effect of the human ATF-126 in breast cancer models (10,18). Retroviral delivery of mATF resulted in decreased cell viability, tumorigenicity, and strongly suppressed cell invasion. mATF induced profound remodeling in the actin cytoskeleton, which could have a negative impact in cell adhesion, migration, and mitosis. mATF-transduction induced the formation of multiple focal adhesions, indicating the cells were less prone to migration.
Our data are consistent with previous studies showing that treatment with a recombinant Maspin protein, rMaspin, resulted in increased adhesion and inhibition of cell invasion of the breast carcinoma MDA-MB-435 cell line (58). rMaspin led to changes in the cytoskeleton structure and increased the formation of focal adhesions in metastatic breast cancer cells (59). Our results with our designer protein confirm the role of Maspin as metastasis suppressor in metastatic ovarian cancer by modulating cell surface signaling pathways resulting in cytoskeleton remodeling.
To evaluate the specificity of the mATF in MOVCAR 5009 cells and to identify genes mediating the anti-tumor/metastatic phenotype of the protein, we performed genome-wide gene expression microarrays. These analyses revealed that only two  Table S1 and  Table S2) were validated by qRT-PCR. A, genes differentially up-regulated by transduction of mATF in MOVCAR 5009 cells are shown. B, genes down-regulated by mATF are shown. Data were normalized to the parental MOVCAR 5009 cell line and represent the mean and S.D. of at least three independent experiments. Differences in gene expression were calculated with a student t test; ***, p Յ 0.0001. targets were up-regulated together with Maspin, with a -fold overexpression higher than 30 relative to control-transduced cells. The subsequent qRT-PCR validation of these targets demonstrated that Maspin was the most regulated target of the ATF, outlining the high potency and specificity of the engineered factor.
TSPAN12 belongs to the tetraspanin family of signal transduction proteins, many of which act as tumor suppressors and regulators of metastasis (51). Tetraspanins are transmembrane proteins that physically associate with integrins to transduce signals inside the cell (60). CD9 is a tetraspanin involved in regulation of cell adhesion and motility (61). Down-regulation FIGURE 7. Down-regulation of Maspin and TSPAN12 in short term human ascites preparations from ovarian carcinoma patients. A, expression of Maspin and TSPAN12 in a panel of human ovarian primary tumors, ascites, and established cancer cell lines, as assessed by qRT-PCR is shown. Data were normalized to SR06, a preparation of normal ovary. UC064 and ASCUC064 represent tumor sample and ascites preparation, respectively, from the same patient. Tumor specimens, ascites preparations, and ovarian cell lines were homogenized and resuspended in Trizol for mRNA extraction and analyzed by quantitative qRT-PCR. B, shown is an immunoblot of Maspin and TSPAN12 on protein extracts from human ascites, the human cell line OVCAR-3, and parental, control, and mATF-transduced MOVCAR 5009 cells. C, down-regulation of Maspin and Tspan12 after Maspin siRNA knockdown in the mATF-transduced MOVCAR 5009 cell line as assessed by qRT-PCR is shown. Data were normalized to the cell line transfected with a control (anti-luciferase) siRNA (*, p Յ 0.05; **, p Յ 0.01; ***, p Յ 0.001). D, shown is expression of Maspin and TSPAN12 by immunofluorescence in representative sections of epithelial ovarian cancer specimens (see also supplemental Fig. S5).
of tetraspanin CD9 has been associated with metastatic behavior and poor prognosis in lung (62), breast (63), and ovarian cancers (64). CD9 down-regulates the expression of several Wnt-associated proteins in lung adenocarcinoma and fibrosarcoma cell lines (65). Interestingly, mATF delivery in MOVCAR 5009 cells led to an up-regulation of (TSPAN12) and a downregulation of Wnt10a suggesting that modulation of Wnt signaling could be of therapeutic value for treatment of metastatic ovarian cancer. Our results identified a novel target, TSPAN12, that was found down-regulated together with Maspin in human metastatic ovarian cancer cells derived from ascites. In summary, these results demonstrate that ZF could be used to potentially discover novel molecular targets and regulators of metastasis in ovarian cancer.
Delivery of ZF proteins has been an unbearable problem for the transition of these agents into the clinic. In addition to effective delivery in pre-existing tumors, another fundamental problem is the specificity to the tumor tissue. To deliver the mATF in MOVCAR 5009 tumor allografts, we developed LPR-AA nanoparticles. The anisamide (AA) ligand coated at the surface of the particles provides specificity for the sigma receptor (SR1) (44), which is expressed at high levels in a vast spectrum of cancer cell lines (50), including the MOVCAR 5009 cells. These nanoparticles were designed with a condensed core comprising protamine and a chemically modified mATF mRNA. The delivery of modified mRNA circumvents limitations associated with traditional plasmid DNA, such as nuclear delivery of DNA through the nuclear envelope and host immunoresponse (76). In vitro synthesis of chemically modified RNA using ribonucleotide analogues has been shown to suppress the innate immune response by stimulating toll-like receptor and concomitantly extend the half-life of the mRNA molecule (66). Previous work has demonstrated that chemically modified mRNA can be used for delivery of reprogramming transcription factors for inducible pluripotent stem cells in vitro (66). Indeed, we have found that our LPR-AA nanoparticles transfected MOVCAR 5009 cells with efficiencies similar to retroviral delivery (ϳ60%). In addition, the same LPR formulation was able to transfect some breast cancer cell lines with efficiencies close to 100% (data not shown) outlining the potential of the technology to effectively deliver ZF proteins in tumor models. The LPR nanoparticles were formulated such that high density of hydrophilic PEG was coated on the surface of the nanoparticles. The PEG coating stabilized LPR in the presence of serum proteins and reduced nonspecific uptake by reticuloendothelial system, extending the half-life in the blood circulation after systemic administration. This led to the enhanced tumor accumulation due to the enhanced permeability and retention. Notably, we demonstrated that the LPR-mATF-AA nanoparticles resulted in translation of ZF proteins in the tumor cells and in transactivation of the endogenous Maspin. When delivered intravenously in a MOVCAR 5009 allograft model, the LPR-mATF-AA nanoparticles significantly inhibited tumor growth, which was accompanied by Maspin reactivation in the tumors. These results are very encouraging and provide the first non-viral delivery of ZF proteins in vivo for its potential use in clinical trials to treat metastatic disease. Although our results demonstrate proof-of-concept using nude mice, the next step will require the intravenous injection of LPRs in a syngeneic or even in the GEMM C57BL/6 TgMISRII-TAg of ovarian cancer, which develops ovarian cancer in a similar progression as in humans (9). In addition, future research will involve the intraperitoneal injection of LPRs in mice carrying abdominal metastases to assess the efficiency of the mATF to abolish pre-established metastases. Given the explosion of publications using both ZF and transcription activator-like effectors (TALEs) technologies in the past decade (particularly in the domain of gene therapy (67)), our results provide a novel non-viral delivery system of mRNA to be used for a broad spectrum of investigators working in the area of designer ZF factors.