Adenovirus-Mediated Restoration of Expression of the Tumor Suppressor

ORIGINAL ARTICLE Adenovirus-mediated restoration of expression of the tumor suppressor gene DLC1 inhibits the proliferation and tumorigenicity of aggressive, androgen-independent human prostate cancer cell lines: prospects for gene therapy M Guan1,2, V Tripathi2, X Zhou and NC Popescu Laboratory of Experimental Carcinogenesis, Center for Cancer Research, National Cancer Institute, Bethesda, MD, USA

Our recent study showing highly recurrent loss of function of DLC1 (deleted in liver cancer 1), a tumor suppressor gene in primary prostate carcinoma (PCA), implicates this gene in the pathogenesis of this disease. To evaluate the response of PCA to oncosuppressive activity of DLC1, we examined now the effects of adenoviral vector for human DLC1 transduction into the DLC1- deficient, androgen-independent (AI) and aggressive human PCA cell lines PC-3 and C4-2-B2. Adenovirus-mediated restoration of DLC1 expression inhibited the proliferation, invasiveness and anchorage-independent growth of PC-3 and C4-2-B2 cells in vitro as well as the tumorigenicity of PC-3 cells in nude mice. It also induced cell-cycle arrest, inhibited the activation of RhoA and the formation of actin stress fibers. DLC1 induced apoptosis in C4-2-B2 cells, whereas it did not elicit such an effect in PC-3 cells. The abundance of the antiapoptotic protein Bcl-2 was greater in PC-3 cells than in C4-2-B2 cells, and PC-3 cells were rendered sensitive to DLC1-induced apoptosis by treatment with the Bcl-2 inhibitor HA14-1. These results suggest that adenovirus-mediated DLC1 transfer, alone or together with other agents, such as inhibitors of Bcl-2 or histone deacetylase, might prove effective in the treatment of aggressive, AI-PCA. Cancer Gene Therapy (2008) 15, 371–381; doi:10.1038/cgt.2008.13; published online 28 March 2008 Keywords: deleted in liver cancer; prostate cancer; tumor suppressor gene; suppression of tumor cell proliferation; apoptosis; gene therapy

Introduction However, in about one-third of patients who undergo radical prostatectomy, the tumors recur, become androgen Despite progress in early detection and therapy, prostate independent (AI), and adopt an aggressive phenotype 4–6 cancer (PCA) continues to be the second leading cause of characterized by preferential metastasis to bone. The cancer-related deaths in men.1 In contrast to most other existing therapy for metastasis of PCA is unsatisfactory, 6–8 forms of cancer, the risk of developing PCA increases with the condition being invariably fatal. The significantly with age. In developed countries, due to a development of novel therapies, possibly including gene boom in the aging population and western-style diet therapy, for management of PCA is thus a clinical resulted in a marked increase in the incidence of PCA.2,3 priority. Screening based on the level of prostate-specific antigen Chromosomal regions that are frequently deleted in (PSA) in serum has enabled the early detection of PCA in cancer contain tumor suppressor genes whose loss or men with locally confined tumors that are treatable inactivation contributes to deregulation of cell prolifera- with surgery, hormone ablation or radiation therapy. tion. Prostate carcinoma (PCA), like several other types of solid tumor or lymphoma, exhibits recurrent loss of DNA copy number and loss of heterozygosity at Correspondence: Dr NC Popescu, Laboratory of Experimenatal chromosomal region 8p21–22.9,10 This region contains Carcinogenesis, Center for Cancer Research, National Cancer the gene DLC1 (deleted in liver cancer 1), which functions Institute, Building 37, Room 4128, 37 Convent Drive, MSC 4262, as a bona fide tumor suppressor gene in several common Bethesda, MD 20892-4262, USA. cancers and encodes a Rho GTPase-activating protein E-mail: [email protected] 1Current address: Department of Laboratory Medicine, Huashan (Rho-GAP). Reactivation of DLC1 function results in Hospital, Fudan University, Shanghai 200040, China suppression of tumor cell proliferation and induction of 2These authors contributed equally to this work apoptosis in vitro as well as prevention of or a reduction in 11 Received 1 November 2007; revised 5 January 2008; accepted 17 tumorigenicity in vivo. The recent identification of January 2008; published online 28 March 2008 recurrent functional alterations of DLC1 in primary DLC1 inhibits prostate cancer cell proliferation M Guan et al 372 PCA has also implicated this tumor suppressor gene in the TOPO Cloning Kit (Invitrogen). The cDNA insert was pathogenesis of PCA.12 Downregulation or loss of DLC1 then transferred to the pAd/CMV/V5-DEST vector protein expression resulting from methylation of the (Invitrogen) by means of the Gateway system and with DLC1 promoter or the deacetylation of histones asso- the use of LR Clonase (Invitrogen), and the resulting ciated with the gene was also detected in a substantial plasmid was purified and digested with PacI (New proportion of individuals with PCA or benign prostatic England Biolabs, Beverly, MA, USA). The linearized hyperplasia as well as in PCA cell lines. Methylation of plasmid (1–2 mg) was mixed with 3 ml of Lipofectamine DLC1 correlated with the age of patients with PCA and 2000 (Invitrogen) in 250 ml of Opti-MEM medium with the blood level of PSA in those with benign prostatic (Invitrogen), and the mixture was added to subconfluent hyperplasia.12 293A cells cultured in 1 ml of Opti-MEM in six-well Given that loss of function of tumor suppressor genes is plates. The Ad-DLC1 vector produced by the transiently directly related to carcinogenesis, these genes have been transfected 293A cells was amplified by infection of a targeted for the development of anticancer therapeutics. fresh batch of 293A cells. The pAd/CMV/V5-GW/lacZ The introduction of tumor suppressor genes into cancer vector (Invitrogen) was digested with PacI and introduced cells prevents tumor growth and, in many instances, yields into 293A cells by transfection to produce the Ad-LacZ a complete response without toxicity to normal cells.13–15 vector as a control. Both Ad-DLC1 and Ad-LacZ were To assess the feasibility of DLC1-based gene therapy, purified with the use of a Viralbind Adenovirus Purifica- we have now examined the effects of restoration of DLC1 tion Kit (Cell Biolabs, San Diego, CA, USA) and stored expression in human PCA cell lines on cell proliferation in at À80 1C. The viral titer (plaque-forming units (PFU)) vitro and in vivo. Given that adenoviral vectors are was determined with 293A cells. considered appropriate vehicles for gene therapy, allow- ing a high efficiency of gene transfer and a high level of Adenovirus infection in vitro transgene expression16,17 we constructed an adenoviral C4-2-B2 or PC-3 cells were seeded at 70–80% confluency vector harboring human DLC1 cDNA for gene transduc- 1 day before infection. Either Ad-DLC1 or Ad-LacZ was tion in PCA cells. Adenovirus-mediated restoration of added to the wells at a multiplicity of infection (MOI) of DLC1 expression resulted in suppression of cell prolifera- 10 or 50, and the cells were incubated with the viruses for tion and tumorigenesis as well as in induction of 6 or 18 h, respectively. The virus-containing medium was apoptosis or cell-cycle arrest in aggressive, AI-PCA cells. then removed, and the cells were washed once with phosphate-buffered saline (PBS) and cultured in complete RPMI 1640 or T-medium, respectively, for 48 h before experiments. Materials and methods Cell lines and culture conditions Immunoblot analysis The human PCA cell lines LNCaP and PC-3 and the Cells were lysed with NP-40 Lysis Buffer (BioSource, human kidney epithelial cell line 239A were obtained Camarillo, CA, USA) containing a Protease Inhibitor from American Type Culture Collection (Rockville, MD, Cocktail (Sigma Aldrich, St Louis, MO, USA). Lysates USA) and were cultured in RPMI 1640 or Dulbecco’s were centrifuged at 12 000 g for 10 min at 4 1C to remove modified Eagle’s medium (Invitrogen), respectively, each debris, and the protein concentration of the resulting supplemented with 10% heat-inactivated fetal bovine supernatants was determined with the BCA protein assay serum. The AI metastatic PCA cell line C4-2-B2 was (Pierce, Rockford, IL, USA). Equal amounts of super- obtained from ViroMed (Minneapolis, MN) and was natant protein were fractionated by SDS-polyacrylamide cultured in T-medium (Invitrogen, Carlsbad, CA, USA) gel electrophoresis on a 4–12% gradient gel (Invitrogen), supplemented with 10% heat-inactivated fetal bovine transferred to a nitrocellulose membrane (Invitrogen) and serum. All cells were maintained at 37 1C in a humidified probed with antibodies to DLC1, Bcl-2, Bax, caspase-3 incubator containing 5% CO2. and glyceraldehyde 3-phosphate dehydrogenase (Santa Cruz, San Diego, CA, USA). Immune complexes were Real-time polymerase chain reaction analysis detected with horseradish peroxidase-conjugated second- Real-time polymerase chain reaction analysis of DLC1 ary antibodies and a chemiluminescent substrate (Milli- mRNA was performed with an ABI Prism 7900 Sequence pore, Billerica, MA, USA). Detection System (Perkin-Elmer Applied Biosystems, Foster City, CA, USA), as described previously.12 The Assay of cell proliferation and viability amount of DLC1 mRNA was normalized by that of Cells were seeded in 96-well plates and infected with glyceraldehyde-3-phosphate dehydrogenase mRNA. Ad-DLC1 or Ad-LacZ. The cell proliferation assay was preformed with the Cell Proliferation Kit I (MTT), Vector construction and adenovirus production according to manufacturer’s instructions (Roche, An adenovirus harboring DLC1 cDNA was prepared Indianapolis, IN, USA). Absorbance was measured at with a Viral Power Adenovirus Expression System 490 nm with the use of a microplate reader (Molecular (Invitrogen) according to the Gateway method. In brief, Devices, Sunnyvale, CA, USA). For assay of cell viability, the full-length cDNA was subcloned into the pENTR/ cells were seeded in 24-well plates, infected with D-TOPO vector with the use of a pENTR Directional Ad-DLC1 or Ad-LacZ and incubated for various times

Cancer Gene Therapy DLC1 inhibits prostate cancer cell proliferation M Guan et al 373 before exposure to trypsin. The number of viable cells was Newark, DE, USA). The insert was examined with a light determined with the use of a Cellometer (Nexcelom microscope, and five fields were evaluated for invading Bioscience, Lawrence, MA, USA). In some experiments, cells. the cells were incubated in the presence of ethyl 2-amino- 6-bromo- 4-(1-cyano-2-ethoxy-2-oxoethyl)-4H-chromene- In vivo tumorigenesis assay 3-carboxylate (HA14-1) (Biomol International, Plymouth PC-3 or C4-2-B2 cells infected with Ad-DLC1 or Meeting, PA, USA) before determination of cell viability. AD-LacZ were harvested with trypsin and resuspended in sterile PBS at a density of 1 Â 107 mlÀ1, and 200 mlof Detection of apoptosis the cell suspension were injected subcutaneously into the Cells infected with Ad-DLC1 or Ad-LacZ as well as flank of 6-week-old male nude mice. The diameter of the incubated in the absence or presence of HA14-1 were resulting tumors was measured with calipers every 3 days, harvested by exposure to trypsin, washed twice with ice- and the tumor volume was calculated as the volume of a cold PBS and resuspended in a mixture of fluorescein sphere. isothiocyanate–labeled annexin V and propidium iodide (Chemicon, Temecula, CA). Apoptotic (annexin Assay of RhoA activity and staining of cells for F-actin V-positive, propidium iodide-negative) cells were detected PC-3 or C4-2-B2 cells were infected with Ad-DLC1 or by flow cytometry with a FACScan instrument (Becton Ad-LacZ, deprived of serum for 24 h and then stimulated Dickinson, San Jose, CA). with lysophosphatidic acid (LPA) for 20 min. The amount of active RhoA in cell lysates was then measured with the Cell-cycle analysis use of a G-LISA kit (Cytoskeleton, Denver, CO, USA). 6 PC-3 or C4-2-B2 cells (1 Â 10 ) infected with Ad-DLC1 or Alternatively, NIH 3T3 cells similarly infected with Ad-LacZ were harvested by exposure to trypsin, adenoviruses and stimulated with LPA were fixed suspended in PBS and fixed with 70% ethanol overnight with 4% formaldehyde in PBS for 20 min at 37 1C at 4 1C. The fixed cells were isolated by centrifugation and and then stained with rhodamine-conjugated phalloidin 1 then incubated first with RNase A (10 mgmlÀ ) for 20 min (Molecular Probes, Eugene, OR, USA). The cells were at 37 1C and then with propidium iodide (50 mgmlÀ1) for examined with a laser-scanning confocal microscope 30 min at room temperature. The cells were then analyzed (LSM 510, Carl Zeiss, Thornwood, NY, USA). for cell-cycle distribution by flow cytometry with a FACSort instrument (Becton Dickinson) and FlowJo 6.4.1 software (Tree Star, Ashland, OR, USA). Results Assay of caspase-3 activity Adenovirus-mediated restoration of DLC1 expression Caspase-3 activity in lysates of cells infected with To examine the effects of adenovirus-mediated DLC1 Ad-LacZ or Ad-DLC1 was measured with an assay kit transduction in PCA cells, we selected the PC-3 and C4-2- (Chemicon) based on the spectrophotometric detection of B2 cell lines, both of which are aggressive, AI- and the chromophore p-nitroaniline (p-NA) generated by DLC1-deficient. PC-3 cells form tumors in nude mice, cleavage of the labeled substrate DEVD-pNA. whereas C4-2-B2 cells metastasize to bone in these animals and were established from LNCaP, an andro- Assay of anchorage-independent cell growth gen-dependent and nonmetastatic PCA cell line.4 To Colony formation in soft agar was evaluated with the use determine whether the acquisition of metastatic potential of a Cell Transformation Detection Assay Kit (Chemicon was associated with a change in DLC1 expression, we International, Billerica, MA, USA). Cells infected with examined the expression of this gene in LNCaP and C4-2- Ad-LacZ or Ad-DLC1 were suspended in 0.35% B2 cells. Reverse transcription and real-time polymerase low-melting-point agarose and were plated at a density chain reaction analysis showed that the amount of DLC1 of 2 Â 104 per well on the layer of a six-well culture plate mRNA in the nonmetastatic LNCaP cells was about five containing 0.8% agarose, according to the manufacturer’s times that in the metastatic C4-2-B2 cells (Figure 1a). instructions. The number of colonies with a diameter of DLC1 immunoreactivity was not detected in PC-3 or C4- 4100 mm was counted after culture for 3 weeks. 2-B2 cells by immunoblot analysis (Figure 1b). Infection with a recombinant adenovirus encoding human DLC1 Cell invasion assay (Ad-DLC1) at an MOI of 10 or 50 conferred a high level Assay of cell invasion was performed with the use of of DLC1 expression in both cell lines, whereas infection transwell chambers (Becton Dickinson, Franklin Lakes, with the control adenovirus Ad-LacZ had no such effect NJ, USA). In brief, cells infected with Ad-DLC1 or (Figure 1b). Ad-LacZ were plated at a density of 1 Â 106 per well on ECMatrix gel-coated inserts (pore size, 8 mm; Becton Effects of DLC1 on cell growth and viability Dickinson) and cultured for 24 h. The noninvading cells The effects of restoration of DLC1 expression on PCA and the ECMatrix gel were removed from the upper cell growth and viability were evaluated at 24-h intervals surface of the insert with a cotton-tipped swab, and the over a period of 4 days after infection with Ad-DLC1. invading cells on the lower surface of the membrane were Infection with Ad-DLC1 at 50 MOI resulted in marked stained by dipping the insert in DiffQuick (Dade Behring, inhibition of the proliferation of both C4-2-B2 and

Cancer Gene Therapy DLC1 inhibits prostate cancer cell proliferation M Guan et al 374

Figure 1 Restoration of DLC1 expression in PCA cells with an adenoviral vector. (a) Total RNA isolated from LNCaP or C4-2-B2 cells was assayed for DLC1 mRNA by real-time polymerase chain reaction analysis. Data were normalized by the corresponding amount of GAPDH mRNA and are means±s.e.m. from three independent experiments. (b) C4-2-B2 or PC-3 cells were infected (or not) with Ad-DLC1 or Ad-LacZ at 50 MOI. The cells were subsequently lysed and subjected to immunoblot analysis with antibodies to DLC1 or to GAPDH (loading control). DLC1, deleted in liver cancer 1; GAPDH, Glyceraldehyde 3-phosphate dehydrogenase; Figure 2 Effects of DLC1 on the proliferation and viability of C4-2- MOI, multiplicity of infection; PCA, prostate carcinoma. B2 and PC-3 cells. (a) C4-2-B2 or PC-3 cells were infected (or not) with Ad-DLC1 of Ad-LacZ at an MOI of 10 or 50 and were subsequently incubated for the indicated times before evaluation of PC-3 cells compared with that apparent for noninfected cell proliferation with the MTT assay. Data are expressed as cells or cells infected with Ad-LacZ (Figure 2a). Infection absorbance units at 490 nm and are means±s.e.m. from three with Ad-DLC1 at 50 MOI also resulted in an approxi- independent experiments. (b) Cells treated as in (a) were assayed mately 50% decrease in the viability of C4-2-B2 cells over for viability on the basis of trypan blue exclusion. Data are expressed as the surviving percentage of cells relative to control of noninfected 4 days compared with that apparent for noninfected or cells and are means±s.e.m. from three independent experiments. Ad-LacZ-infected cells, whereas it had no substantial DLC1, deleted in liver cancer 1; MOI, multiplicity of infection. effect on the viability of PC-3 cells (Figure 2b).

Induction of apoptosis by DLC1 DLC1 on the amounts of these proteins in PCA cells by Given the effects of DLC1 on cell proliferation and immunoblot analysis. Infection with Ad-DLC1 resulted in viability, we next examined whether Ad-DLC1 infection a MOI-dependent reduction in the amount of Bcl-2 in C4- induced apoptosis in the PCA cells. Cells in the early stage 2-B2 cells, whereas restoration of DLC1 expression had of apoptosis were detected by staining with annexin V and no effect on the abundance of Bcl-2 in PC-3 cells propidium iodide. Flow cytometry revealed that infection (Figure 4a). In contrast, DLC1 had no apparent effect of C4-2-B2 cells with Ad-DLC1 at 50 MOI induced a on the amount of Bax in either cell line (Figure 4a). marked increase in the percentage of apoptotic (annexin Restoration of DLC1 expression in liver cancer cells was V-positive, propidium iodide-negative) cells compared previously shown to result in downregulation of Bcl-2 with that apparent for Ad-LacZ-infected or noninfected without an effect on Bax abundance.19 Immunoblot control cells, whereas restoration of DLC1 expression in analysis also confirmed that DLC1 induced cleavage of PC-3 cells did not have such an effect (Figure 3a). the precursor form of caspase-3 in C4-2-B2 cells, but not Activation of caspases has a central role in apoptosis, and in PC-3 cells (Figure 4b). DLC1-mediated inhibition of liver, ovarian or lung cancer cell proliferation was previously shown to be associated Role of Bcl-2 in DLC1-induced apoptosis with the induction of apoptosis mediated by caspase-3.18–20 Given that the constitutive level of Bcl-2 expression in Consistent with these previous observations, infection PC-3 cells was higher than that in C4-2-B2 cells with Ad-DLC1 induced an approximately 2.5-fold (Figure 4a), we next examined whether this difference increase in caspase-3 activity in C4-2-B2 cells (Figure 3b). might be responsible for the difference in sensitivity of the Given that the balance between the antiapoptotic two cell lines to DLC1-induced apoptosis. We, therefore, protein Bcl-2 and the proapoptotic protein Bax investigated whether the Bcl-2 inhibitor HA14-1 might determines cell survival or death after exposure to an sensitize PC-3 cells to the induction of apoptosis by 21 apoptotic stimulus, we examined the possible effects of DLC1. Treatment with 10 mM HA14-1 for 4 h resulted in a

Cancer Gene Therapy DLC1 inhibits prostate cancer cell proliferation M Guan et al 375

Figure 3 Effects of DLC1 on apoptosis in C4-2-B2 and PC-3 cells. (a) C4-2-B2 or PC-3 cells were infected (or not) with Ad-DLC1 or Ad-LacZ at a 50 MOI and were subsequently stained with FITC-labeled annexin V and propidium iodide (PI) for determination of the proportion of apoptotic cells by flow cytometry. Data are from a representative experiment. (b) C4-2-B2 cells infected as in (a) were lysed and assayed for caspase-3 activity. Data are expressed relative to the activity of noninfected cells and are means±s.e.m. from three independent experiments. DLC1, deleted in liver cancer 1; FITC, fluorescein isothiocyanate; MOI, multiplicity of infection.

marked increase in the incidence of apoptosis in PC-3 inhibited the invasive activity of both C4-2-B2 and PC-3 cells infected with Ad-DLC1, but not in noninfected cells cells by approximately 50% (Figures 6c and d). (Figure 5a). This combined effect of DLC1 and HA14-1 We next tested the effect of Ad-DLC1 infection was also apparent in the cell-survival assay (Figure 5b) on tumorigenicity of PCA cells in vivo. Restoration and was accompanied by cleavage of the precursor form of DLC1 expression resulted in marked inhibition of of caspase-3 (Figure 5c). the growth of tumors formed by PC-3 cells in nude mice. All five mice injected with PC-3 cells infected with Ad-LacZ developed tumors, whereas three of the five Oncosuppressive effect of DLC1 in PCA cells mice injected with PC-3 cells infected with Ad-DLC1 To determine whether restoration of DLC1 expression had not developed tumors after 40 days of observation. exerts antioncogenic activity in PCA cells, as it has been The size of those two tumors developed after infection shown to do in breast, liver and lung cancers,11 we first with Ad-DLC1 were less than a half the size of tumors investigated the effects of Ad-DLC1 infection on C4-2-B2 formed in mice infected with Ad-LacZ (Supplementary and PC-3 cell invasiveness and growth in semisolid information). The anchorage-independent growth of medium. Quantification of anchorage-independent cell C4-2-B2 cells in vitro was previously shown to reflect growth on the basis of the number and size of colonies the in vivo growth and metastatic potential of these formed in soft agar revealed that restoration of DLC1 cells.4 Under our conditions, however, using a different expression inhibited such growth by 84 and 81% in C4-2- strain of mice, we did not detect the formation of tumors B2 and PC-3 cells, respectively (Figures 6a and b). A by C4-2-B2 cells after 10 weeks of observation. This transwell invasion assay also revealed that DLC1 discrepancy may be due to our use of a different strain

Cancer Gene Therapy DLC1 inhibits prostate cancer cell proliferation M Guan et al 376 3T3 cells infected with Ad-LacZ exhibited only a low level of F-actin staining with rhodamine–phalloidin. Treatment of these cells with LPA induced a pronounced increase in the amount of cortical F-actin. The morphology of serum-deprived NIH 3T3 cells infected with Ad-DLC1 did not differ substantially from that of those infected with Ad-LacZ, ectopic expression of DLC1 largely prevented the LPA-induced formation of actin stress fibers (Figure 7c).

Discussion

We have shown that restoration of DLC1 expression inhibited the proliferation of aggressive, AI-PCA (C4-2- B2 and PC-3) cells, thus expanding the spectrum of Figure 4 Effects of DLC1 on the expression of apoptosis-related common forms of cancer that are sensitive to the proteins in PCA cells. C4-2-B2 or PC-3 cells were infected (or not) antioncogenic function of this tumor suppressor gene. with Ad-DLC1 or Ad-LacZ at an MOI of 10 or 50 and were We also found that DLC1 induced apoptosis in the subsequently lysed and subjected to immunoblot analysis with metastatic C4-2-B2 cell line, but not in the nonmetastatic antibodies to Bcl-2 and to Bax (a) or with those to caspase-3 (b). The PC-3 line. Loss of chromosome 8p or downregulation of positions of the precursor (37 kDa) and cleaved (18 kDa) forms of DLC1 expression has been associated with metastatic caspase-3 are indicated. DLC1, deleted in liver cancer 1; MOI, potential in various clones of liver or breast tumor cell multiplicity of infection; PCA, prostate carcinoma. lines as well as with reduced survival both of patients with high-grade gliomas and of hepatitis C virus–positive of mice or to not monitoring the mice for a longer period individuals with hepatocellular carcinoma awaiting liver of time. transplantation.24–29 The C4-2-B2 cell line, which meta- stasizes to bone, was isolated from LNCaP cells, which Effects of DLC1 on cell-cycle distribution, RhoA activity are androgen-dependent, nontumorigenic and nonmeta- and the actin cytoskeleton static and possess one methylated DLC1 allele.4,12 We To gain insight into the mechanism of the oncosuppres- have now shown that the amount of DLC1 mRNA in sive effect of DLC1 in PCA cells, we first examined LNCaP cells is about five times that in C4-2-B2 cells. This cell-cycle distribution. Arrest of cell-cycle progression reduction in DLC1 expression in the metastatic cell line is contributes to the inhibition of tumor cell growth, attributable, at least in part, to loss of copy number of 8p migration or invasion. Flow cytometric analysis of cells as revealed by karyotype analysis.4 Similarly, human stained with propidium iodide revealed that Ad-DLC1 MDA-MB-231 breast cancer cells selected for the ability infection induced arrest of PC-3 cells at the G2–M to metastasize to bone exhibit a lower level of DLC1 transition of the cell cycle, whereas it resulted in G1 expression compared with clones that are only weakly 30 arrest as well as the accumulation of cells with a sub-G1 metastatic or nonmetastatic. These observations with DNA content (corresponding to apoptotic cells) in different types of cancer cells are thus indicative of a link C4-2-B2 cells (Figure 7a). between DLC1 deficiency and metastasis that may have In liver cancer cells, DLC1 was shown to increase the prognostic implications with regard to the metastatic hydrolysis of GTP bound to RhoA and, to a lesser extent, potential of primary tumors. that bound to Cdc42, but it had only a minimal effect on Infection with an adenoviral vector for DLC1 resulted the GTPase activity of Rac1.22,23 We, therefore, examined in a high level of protein expression in both PC-3 and whether restoration of DLC1 expression affected the C4-2-B2 cells. Restoration of DLC1 expression sup- activation status of RhoA in PCA cells. PC-3 and C4-2- pressed cell proliferation, invasion and anchorage-inde- B2 cells were deprived of serum and then stimulated with pendent growth (an in vitro measure of tumorigenicity) in LPA for 20 min, after which the amount of the GTP- both cell lines as well as reduced the tumorigenicity of bound (activated) form of RhoA coupled with a PC-3 cells in vivo. The use of an adenoviral vector as a Rhotekin-binding domain peptide was determined. vehicle for DLC1 gene transfer may therefore hold Infection with Ad-DLC1 resulted in a MOI-dependent promise as a therapeutic approach for aggressive and reduction in the amount of the active form of RhoA metastatic PCA. in both C4-2-B2 and PC-3 cells relative to that in The suppression of cell growth by DLC1 was associated noninfected cells, whereas infection with Ad-LacZ had with cell-cycle arrest, and a reduction in RhoA activity in no such effect (Figure 7b). both C4-2-B2 and PC-3 cells and ectopic expression of Given that RhoA regulates the formation of actin stress DLC1 in NIH 3T3 fibroblasts inhibited the LPA-induced fibers, we examined the possible effect of forced DLC1 formation of actin stress fibers. The Rho-GAP domain of expression on RhoA-mediated cytoskeletal reorganization DLC1 is thought to be essential for its growth-inhibitory and changes in cell morphology. Serum-deprived NIH effect on tumor cells.22,23 Recently, in lung cancer cells,

Cancer Gene Therapy DLC1 inhibits prostate cancer cell proliferation M Guan et al 377

Figure 5 Effect of HA14-1 on apoptosis in PC-3 cells expressing DLC1. (a) PC-3 cells infected (or not) with Ad-DLC1 or Ad-LacZ at 50 MOI were incubated with 10 mM HA14-1 or vehicle (DMSO; final concentration, 0.5%) for 4 h before analysis of apoptosis by staining with FITC- annexin V and propidium iodide and flow cytometry. Data are from a representative experiment. (b) PC-3 cells infected (or not) as in (a) were incubated with HA14-1 at 5 or 10 mM for the indicated times, after which cell viability was evaluated by staining with trypan blue. Data are expressed as the surviving percentage of cells relative to control of noninfected cells are means±s.e.m. from three independent experiments. (c) PC-3 cells infected (or not) with Ad-DLC1 or Ad-LacZ at an MOI of 10 or 50 were incubated for 4 h in the absence or presence of 5.0 mM HA14- 1, after which cell lysates were prepared and subjected to immunoblot analysis with antibodies to caspase-3. The positions of the precursor (37 kDa) and cleaved (18 kDa) forms of caspase-3 are indicated. DLC1, deleted in liver cancer 1; DMSO, dimethyl sulfoxide; FITC, fluorescein isothiocyanate; MOI, multiplicity of infection. the Rho-GAP domain of DLC1 was shown to stimulate cell-cycle progression and the induction of apoptosis. the GTPase activity of RhoA as well as that of RhoB, DLC1 thus induced arrest in G1 and G2–M phases of the RhoC and Cdc42 and inhibit cell proliferation by cell cycle in C4-2-B2 and PC-3 cells, respectively, effects both Rho-GAP domain-dependent and -independent that likely contribute to the inhibition of cell proliferation mechanisms.31 DLC1 was also recently shown to interact in each cell line.35 Furthermore, accumulation of sub- with members of the tensin family of focal adhesion diploid (apoptotic) cells was apparent only for C4-2-B2 proteins32–34 and demonstrated that suppression of the cells infected with Ad-DLC1. migration of human lung cancer cells by DLC1 requires We previously showed that among Du145, LNCaP, cooperation between the Rho-GAP and tensin-binding PC-3 and 22Rv1 PCA cell lines, DLC1 was expressed only domains.34 in Du145 cells and only LNCaP cells exhibited partial Our results have revealed differences between C4-2-B2 methylation of the DLC1 promoter.12 In the absence of and PC-3 cells with regard to the effects of DLC1 on aberrant methylation or genomic deletion of DLC1,

Cancer Gene Therapy DLC1 inhibits prostate cancer cell proliferation M Guan et al 378

Figure 6 Effects of DLC1 on colony formation and invasive activity of PCA cells. (a) C4-2-B2 or PC-3 cells infected with Ad-DLC1 or Ad-LacZ at 50 MOI were assayed for the ability to form colonies during incubation in soft agar for 3 weeks. (b) The number of colonies with a diameter of 4100 mm formed in experiments similar to that shown in (a) was determined. Data are expressed as number of colonies per field and are means±s.e.m. from three independent experiments. (c) Cells infected as in (a) were assayed for invasive activity with a transwell apparatus. Cells that had migrated to the lower surface of the membrane were stained with DiffQuick. (d) The numbers of migrated cells in experiments similar to that shown in (c) were counted in five fields, averaged and expressed as means±s.e.m. from three independent experiments. DLC1, deleted in liver cancer 1; MOI, multiplicity of infection; PCA, prostate carcinoma.

we considered histone deacetylation as a mechanism Consistent with these results, we have now shown that responsible for downregulation of DLC1 expression in adenovirus-mediated restoration of DLC1 expression PCA cell lines and treated the cells with trichostatin A, an induced apoptosis accompanied by the activation of inhibitor of histone deacetylase. This drug restored DLC1 caspase-3 in PC-3 cells only after chemical inhibition of expression in LNCaP and 22Rv1 cells, but was ineffective Bcl-2 with HA14-1. The similar response of PC-3 cells to in PC-3 cells.12 A study examining the possible factors the induction of apoptosis by unrelated antioncogenic contributing to the resistance of certain PCA cell lines to factors, chemotherapeutic agents, radiation and a tumor cell death induced by another histone deacetylase suppressor protein thus indicate that inhibition of Bcl-2 inhibitor suberoylanilide hydroxamic acid (SAHA) found family members is crucial for the induction of apoptosis that Du145 cells were highly sensitive to SAHA-induced by different apoptotic stimulus. death and had no detectable Bcl-2 protein, whereas PC-3 In conclusion, our results suggest that silencing of DLC1 cells were resistant to SAHA-induced death, with SAHA signaling is an important event in the progression and inducing upregulation of Bcl-2 in these cells. Inhibition of metastasis of PCA. The effectiveness of adenovirus- Bcl-2 activity with HA14-1 rendered PC-3 cells sensitive mediated restoration of DLC1 expression in suppressing to SAHA-induced apoptosis in the absence of caspase-3 PCA cell proliferation and tumorigenicity suggests that or caspase-9 activation,7 and an antisense oligonucleotide therapy based on DLC1 transfer alone or combined specific for Bcl-2 mRNA also sensitized PC-3 cells to with inhibitors of the antiapoptotic activity of Bcl-2 or radiation-induced apoptosis and enhanced the radiation- histone deacetylase inhibitors37,38–40 may prove beneficial induced activation of caspase-3 and caspase-7.36 for the treatment of aggressive, AI-PCA. To determine the Furthermore, an antisense oligonucleotide that depleted oncosuppressive effects of DLC1 on primary and/or the mRNAs for several members of the Bcl family, metastatic tumors in various model systems such as including Bcl-2, induced apoptosis in PC-3 cells orthopic transplantation, xenografts and intracardiac and enhanced their chemosensitivity to paclitaxel.37 injection will be used for direct in vivo adenovirus/

Cancer Gene Therapy DLC1 inhibits prostate cancer cell proliferation M Guan et al 379

Figure 7 Effects of DLC1 on cell-cycle distribution, RhoA activity and cell morphology in PCA cells. (a) PC-3 or C4-2-B2 cells were infected (or not) with Ad-DLC1 or Ad-LacZ at 50 MOI and were then examined for cell-cycle distribution by staining with propidium iodide and flow cytometry. The percentages of cells in G1, S and G2–M phases of the cell cycle are indicated. Data are from a representative experiment. (b) C4- 2-B2 and PC-3 cells, respectively, were infected (or not) with Ad-DLC1 or Ad-LacZ at an MOI of 10 or 50, deprived of serum for 24 h, and stimulated with LPA (1.0 mgmlÀ1) for 20 min. Cell lysates were then subjected to the G-LISA assay for the amount of GTP-bound RhoA. Data are means±s.e.m. from three independent experiments (c) NIH 3T3 cells were infected with Ad-DLC1 or Ad-LacZ at 50 MOI, deprived of serum for 24 h and incubated in the absence or presence of LPA (1.0 mgmlÀ1) for 20 min. They were then fixed, stained with rhodamine-phalloidin and examined by confocal microscopy. DLC1, deleted in liver cancer 1; LPA, lysophosphatidic acid; MOI, multiplicity of infection; PCA, prostate carcinoma.

DLC1 transfer into the tumor tissue or combined treatment Acknowledgements with other agents.41 Additional insights into the role of DLC1 in the progression of prostate tumorigenesis This work was supported by The Intramural Research may come from prostate-specific ablation of DLC1 Program of the National Cancer Institute supported this expression in genetically engineered mice predisposed to work, NIH. We thank Dr Paul Marks for most helpful PCA development. suggestions.

Cancer Gene Therapy DLC1 inhibits prostate cancer cell proliferation M Guan et al 380 References 19 Zhou X, Thorgeirsson SS, Popescu NC. Restoration of DLC-1 gene expression induces apoptosis and inhibits both 1 Jemal A, Murray T, Samuels A, Ghafoor A, Ward E, cell growth and tumorigenicity in human hepatocellular Thun MJ. Cancer statistics. CA Cancer J Clin 2003; 53: 5–26. carcinoma cells. Oncogene 2004; 23: 1308–1313. 2 Gronberg H. Prostate cancer epidemiology. Lancet 2003; 20 Yuan BZ, Jefferson AM, Millecchia L, Popescu NC, 361: 859–864. Reynolds SH. Morphological changes and nuclear trans- 3 Hsing AW, Devesa SS. Trends and patterns of prostate location of DLC1 tumor suppressor protein precede cancer: what do they suggest? Epidemiol Rev 2001; 23: 3–13. apoptosis in human non-small cell lung carcinoma cells. 4 Thalmann GN, Sikes RA, Wu TT, Degeorges A, Chang SM, Exp Cell Res 2007; 313: 3868–3880. Ozen M et al. LNCaP progression model of human prostate 21 Oltvai ZN, Milliman CL, Korsmeyer SJ. Bcl-2 heterodi- cancer: androgen-independence and osseous metastasis. merizes in-vivo with a conserved homolog, bax, that Prostate 2000; 44: 91–103. accelerates programmed cell-death. Cell 1993; 74: 609–619. 5 Thalmann GN, Anezinis PE, Chang SM, Zhau HE, Kim EE, 22 Wong CM, Yam JW, Ching YP, Yau TO, Leung TH, Hopwood VL et al. Androgen-independent cancer progres- Jin DY et al. Rho GTPaseactivating protein deleted in liver sion and bone metastasis in the LNCaP model of human cancer suppresses cell proliferation and invasion in hepato- prostate cancer. Cancer Res 1994; 54: 2577–2581. cellular carcinoma. Cancer Res 2005; 65: 8861–8868. 6 Li X, Raikwar SP, Liu YH, Lee SJ, Zhang YP, Zhang S et al. 23 Kim TY, Lee JW, Kim HP, Jong HS, Kim TY, Jung M et al. Combination therapy of androgen-independent prostate DLC-1, a GTPase activating protein for Rho, is associated cancer using a prostate restricted replicative adenovirus with cell proliferation, morphology and migration in human and a replication-defective adenovirus encoding human hepatocellular carcinoma. Biochem Biophys Res Commun endostatin-angiostatin fusion gene. Mol Cancer Ther 2006; 2007; 355: 72–77. 5: 676–684. 24 Qin LX, Tang ZY, Sham JST, Ma ZC, Ye SL, Zhou XD 7 Xu W, Ngo L, Perez G, Dokmanovic M, Marks PA. et al. The association of chromosome 8p deletion and Intrinsic apoptotic and thioredoxin pathways in human tumormetastasis in human hepatocellular carcinoma. Cancer prostate cancer cell response to histone deacetylase inhibitor. Res 1999; 59: 5662–5665. Proc Natl Acad Sci USA 2006; 103: 15540–15545. 25 Qin LX, Tang ZY, Ye SL, Liu YK, Ma ZC, Zhou XD et al. 8 Scher HI, Sawyers CL. Biology of progressive, castration- Chromosome 8p deletion is associated with metastasis of resistant prostate cancer: directed therapies targeting the human hepatocellular carcinoma when high and low androgen-receptor signaling axis. J Clin Oncol 2005; 23: metastatic models are compared. J Cancer Res Clin Oncol 8253–8261. 2001; 127: 482–488. 9 Arbieva ZH, Banerjee K, Kim SY, Edassery SL, 26 Goodison S, Yuan J, Sloan D, Kim R, Li C, Popescu NC Maniatis VS, Horrigan SK et al. High-resolution physical et al. The RhoGAP protein DLC-1 functions as a metastasis map and transcript identification of a prostate cancer suppressor in breast cancer cells. Cancer Res 2005; 65: deletion interval on 8p22. Genome Res 2000; 10: 244–257. 6042–6053. 10 Brothman AR. Cytogenetics and molecular genetics of 27 Song LJ, Ye SL, Wang KF, Weng YQ, Liang CM, Sun RX cancer of the prostate. Am J Med Genet 2002; 115: 150–156. et al. Relationship between DLC-1 expressions and meta- 11 Durkin ME, Yuan BZ, Zhou X, Zimonjic DB, Lowy DR, stasis in hepatocellular carcinoma. Zhonghua Gan Zang Bing Thorgeirsson SS et al. DLC-1: a Rho GTPase-activating Za Zhi 2005; 13: 428–431. protein and tumor suppressor. J Cell Mol Med 2007; 11: 28 Czernicki T, Zegarska J, Paczek L, Cukrowska B, 1185–1207. Grajkowska W, Zajaczkowska A et al. Gene expression 12 Guan M, Zhou X, Soulitzis N, Spandidos DA, Popescu NC. Int J Aberrant methylation and deacetylation of deleted in liver profile as a prognostic factor in high-grade gliomas. cancer-1 gene in prostate cancer: potential clinical applica- Oncol 2007; 30: 55–64. tions. Clin Cancer Res 2006; 12: 1412–1419. 29 Mas VR, Fisher RA, Archer KJ, Yanek KC, Williams B, 13 Fang BL, Roth JA. Tumor-suppressing gene therapy. Cancer Dumur CI et al. Genes associated with progression and Biol Ther 2003; 2: S115–S121. recurrence of hepatocellular carcinoma in hepatitis C 14 Roth JA, Grammer SF. Tumor suppressor gene therapy. In: patients waiting and undergoing liver transplantation: El-Deiry WS (eds). Tumor Supressor Genes: Regulation, preliminary results. Transplantation 2007; 83: 973–981. Function and Medicinal Applications. vol.2. Humana Press: 30 Kang Y, Siegel PM, Shu W, Drobnjak M, Kakonen SM, Totowa, 2003 pp 577–597. Cordo´ n-Cardo C et al. A multigenic program mediating 15 Osborne C, Wilson P, Tripathy D. Oncogenes and tumor breast cancer metastasis to bone. Cancer Cell 2003; 3: suppressor genes in breast cancer: potential diagnostic and 537–549. therapeutic applications. Oncologist 2004; 9: 361–377. 31 Healy KD, Hodgson L, Kim TY, Shutes AT, Maddileti S, 16 MacRae EJ, Giannoudis A, Ryan R, Brown NJ, Hamdy FC, Juliano RL et al. DLC1 suppresses non-small lung Maitland N et al. Gene therapy for prostate cancer: current cancer growth and invasion by RhoGAP-dependent and strategies and new cell-based approaches. Prostate 2006; 66: independent mechanisms. Mol Carcinogenesis 2007 doi: 470–494. 10.1002/mc.20389. 17 Relph KL, Harrington KJ, Pandha H. Adenoviral strategies 32 Yam JW, Ko FC, Chan CY, Jin DY, Ng IO. Interaction of for the gene therapy of cancer. Semin Oncol 2005; 32: deleted in liver cancer 1 with tensin2 in caveolae and impli- 573–582. cations in tumor suppression. Cancer Res 2006; 66: 8367–8372. 18 Syed V, Mukherjee K, Lyons-Weiler J, Lau KM, Mashima T, 33 Liao YC, Si L, deVere White RW, Lo SH. The phospho- Tsuruo T et al. Identification of ATF-3, caveolin-1, tyrosine-independent interaction of DLC-1 and the SH2 DLC-1, and NM23-H2 as putative antitumorigenic, domain of cten regulates focal adhesion localization and progesterone-regulated genes for ovarian cancer cells by gene growth suppression activity of DLC-1. J Cell Biol 2007; 176: profiling. Oncogene 2005; 24: 1774–1787. 43–49.

Cancer Gene Therapy DLC1 inhibits prostate cancer cell proliferation M Guan et al 381 34 Qian X, Li G, Asmussen HK, Asnaghi L, Vass WC, 38 Butler LM, Agus DB, Scher HI, Higgins B, Rose A, Braverman R et al. Oncogenic inhibition by a deleted in Cordon-Cardo C et al. Suberoylanilide hydroxamic acid, liver cancer gene requires cooperation between tensin binding an inhibitor of histone deacetylase, suppresses the growth of and Rho-specific GTPase activating protein activities. Proc prostate cancer cells in vitro and in vivo. Cancer Res 2000; 60: Natl Acad Sci USA 2007; 104: 9012–9017. 5165–5170. 35 Ullmannova V, Popescu NC. Inhibition of cell proliferatio- 39 Kelly WK, O’Connor OA, Krug LM, Chiao JH, Heaney M, n,induction of apoptosis,reactivation of DLC1, and modula- Curley T et al. Phase I study of an oral histone deacetylase tion of other gene expression by dietary flavone in breast inhibitor, suberoylanilide hydroxamic acid, in patients with cancer cell lines. Cancer Detect Prev 2007; 31: 110–118. advanced cancer. J Clin Oncol 2005; 23: 3923–3931. 36 Mu Z, Hachem P, Pollack A. Antisense Bcl-2 sensitizes 40 O’Connor OA, Heaney ML, Schwartz L, Richardson S, prostate cancer cells to radiation. Prostate 2005; 65: 331–340. Willim R, MacGregor-Cortelli B et al. Clinical experience 37 Yamanaka K, Rocchi P, Miyake H, Fazli L, Vessella B, with intravenous and oral formulations of the novel histone Zangemeister-Wittke U et al. A novel antisense oligonucleo- deacetylase inhibitor suberoylanilide hydroxamic acid in tide inhibiting several antiapoptotic Bcl-2 family members patients with advanced hematologic malignancies. J Clin induces apoptosis and enhances chemosensitivity in andro- Oncol 2006; 24: 166–173. gen-independent human prostate cancer PC3 cells. Mol 41 Singh AS, Figg WD. In vivo models of prostate cancer Cancer Ther 2005; 4: 1689–1698. metastasis to bone. J Urol 2005; 174: 820–826.

Supplementary Information accompanies the paper on Cancer Gene Therapy website (http://www.nature.com/cgt)

Cancer Gene Therapy