Darinaparsin Inhibits Prostate Tumor–Initiating Cells and Du145 Xenografts and Is an Inhibitor of Hedgehog Signaling

Arsenic derivatives have been therapeutically used for more than 2,000 years. In the early 20th century, use of arsenic trioxide (ATO) in treating leukemia was first reported, and by the mid-20th century, its effectiveness in patients with relapsed acute promyelocytic leukemia (APL) was demonstrated (1). A randomized clinical trial in United States led to FDA approval of ATO for relapsed or refractory APL (1). Recently, the combination of ATO and all trans-retinoic acid has been recommended as first-line treatment of APL (2). ATO has been investigated in the treatment of other non-APL cancers; however, it was less effective at clinically relevant doses and was highly toxic at higher concentrations (3). Therefore, other arsenicals with antitumor activity and with less toxicity and oral availability have been sought. Darinaparsin is an organic arsenic (S-dimethylarsino-glutathione; Z-101) made by conjugating dimethylarsenic to glutathione (4, 5). Screening of the NCI-60 panel of cells indicated that Ic-50 concentrations with darinaparsin ranged from 0.02 to 7.3 μmol/L. Mouse toxicity studies showed that the LD₅₀ of darinaparsin was approximately 50-fold higher than that of ATO. Phase I studies with darinaparsin in patients with advanced refractory solid tumors showed that 300 mg/m² i.v. for 5 consecutive days every 4 weeks was well tolerated (6). Phase II studies in both hematologic malignancies and solid cancers are currently under way (7, 8). In vitro studies showed that darinaparsin more potently induces growth arrest, apoptosis, and oxidative stress than ATO in several hematologic malignancies (8). Also, unlike ATO, it does not increase bcl-2 protein levels. Importantly, darinaparsin is effective in ATO-resistant leukemic cell lines that overexpress multidrug-resistant protein 1/ATP-binding cassette, subfamily C, member1 (MRP1/ABCC1; refs. 3, 9). Darinaparsin showed increased antiangiogenic activity in both in vitro human umbilical vascular endothelial cell microtubule formation and in vivo Matrigel plug models (10). Recent investigations have also shown that darinaparsin is a multivalent molecule that can induce an incomplete stress response by disrupting microtubules and sonic hedgehog (Shh) signaling (11).

In this study, we examined the cytotoxic effects of darinaparsin in both established prostate cancer cell lines and in fresh prostate cancer cells from patients. Here, we show that darinaparsin is a potent cytotoxic against various human prostate cancer cell lines as well as primary prostate cancer cells and is effective in inhibiting prostate tumor–initiating cells (TIC). Studies with the Du145 cell line also showed synergistic cell kill with taxotere. Du145 tumors propagated in nude mice were also sensitive to darinaparsin. Prompted by studies showing that ATO inhibited Hedgehog signaling (12), we show that darinaparsin also inhibits Hedgehog signaling in prostate cancer by downregulating Gli-2 transcriptional activity.

Collagen I (rat tail collagen) was purchased from BD Biosciences, darinaparsin/Z101 was obtained from Ziopharm Oncology Inc., NOD/SCIDγ mice were purchased from The Jackson Laboratory, Gli-2, GAPDH, and β-tubulin antibodies were purchased from Cell Signaling Technology, α2-FITC, CD44-APC (BD Biosciences), SAG (Smoothened agonist), and ATO were purchased from Sigma-Aldrich. Gli-2 luciferase reporter plasmid was obtained as a kind gift from P.A. Beachy (Stanford University School of Medicine, Stanford, CA).

Du145, LnCap, PC3, and CWR22 cells were purchased from the ATCC in 2010. The ATCC uses isoenzymology method for species determination and short-tandem repeats methods for identity verification of the cell line. The cells were also tested for Mycoplasma (http://www.atcc.org/~/media/PDFs/Technical%20Bulletins/tb08.ashx). Upon receipt, the cells were maintained at low passage numbers in RPMI (Gibco BRL) supplemented with 10% fetal bovine serum and 1% penicillin–streptomycin. The cells were routinely tested for Mycoplasma contamination in the laboratory using the Lonza Mycoplasma Testing Kit. No other method was used to authenticate the cells.

Human primary prostate tumors were obtained from the Robert Wood Johnson Hospital (New Brunswick, NJ) after prostatectomy on an Institutional Review Board (IRB)–approved protocol. To obtain single cells from fresh prostate tumors, the specimen was minced in small pieces and incubated in 200 μg/mL of collagenase I in RPMI for 2 to 4 hours, the longer time for larger size specimens. After 4 hours, the media containing the minced tissue was strained and the supernatant was collected and washed twice in 1× PBS at 250 × g for 30 seconds (this step eliminates the fibroblasts from the epithelial cells). The pellet was suspended in PROSTAlife medium (Life Science Technologies) and plated in a flask. The cells were incubated at 37°C for 3 to 4 days to allow the epithelial organoids to attach. The media was replaced and cells passaged upon 70% confluency. The cells were stained with anti-EPCAM and anti-androgen receptor antibodies and analyzed with flow cytometry to confirm that the isolated cells were prostate epithelial cells.

For enrichment of prostate TICs from early-passage Du145 cells, a collagen I attachment assay was used to harvest 5-minute attached cells (13). These cells, enriched for TICs, were maintained in keratinocyte serum-free medium (KSFM) supplemented with epidermal growth factor (EGF) and bovine pituitary extract (13). Briefly, tissue culture plates were coated with 70 μg collagen I and blocked with 1% BSA. Early-passage cells were allowed to attach on collagen I–coated dishes for 5 minutes. The 5-minute attached cells were collected after trypsinization (0.5% trypsin). The unattached cells after 20 minutes were also collected and used as a control (non-TICs).

For three-dimensional spheroid formation, 5-minute attached tumor cells enriched for TIC's and 20-minute nonattached cells were suspended and plated in 1% agarose-coated 6-well plates. Every 3 days, half of the media was supplemented with fresh KSFM. For drug treatment assays, drugs at specified IC₅₀ concentrations were added to the spheroids. After 48 or 72 hours, spheroids were collected and dissociated and used for colony-forming assays. For colony formation assay from the spheroids, 48 or 72 hours after drug treatment, floating cell spheroids were collected, washed, and dissociated with 0.1% trypsin to form single cells. Equal numbers of single cells (25–300 cells) were then plated in 6-well tissue culture plates. After 14 days, colonies were stained with 0.25% crystal violet in 95% ethanol. Colonies consisting of more than 50 cells were counted.

Du145, PC3, and LnCap prostate cancer cells and primary prostate cells were plated at a density of 3,500 cells/200 μL in 96-well plates. Twenty-four hours after plating, darinaparsin, taxotere and other drugs were added at various concentrations. Seventy-two hours after treatment, the MTS reagent was added (Promega) to the cells. Color change was monitored at 490 nm and data were acquired by SOFT max pro. Cell viability and IC₅₀ concentrations were determined using PRISM software. For synergy experiment, CalcuSyn software was used to determine the combination index (CI) values between taxotere and darinaparsin (14).

Flow cytometry using FACS was used to determine cell-surface markers: α2-FITC and CD44-APC. Data were analyzed using the CellQuest software. For cell-cycle analysis, the cells were collected at 72 hours, washed, and fixed in ethanol overnight. The fixed cells were then washed and stained with PI-RNase (BD Pharmigen) and cells were acquired using flow cytometry. The data were analyzed using MODFIT software.

Du145 cells were plated at 100,000 cells/mL in 24-well plates. Sixteen hours after plating, cells were cotransfected with 0.5 μg pcDNA-Gli2 and 0.5 μg 8xGli-luciferase reporter. After 8 hours of transfection, cells were treated with a Shh agonist (10 nmol/L SAG from Sigma-Aldrich) with or without darinaparsin for 24 hours. Luciferase assays were performed with the Promega Luciferase System. Assays were done in triplicates and the results were normalized to total protein levels.

Du145 cells treated with darinaparsin for 72 hours were harvested and total protein was extracted with RIPA lysis buffer. Fifty μg of total protein was electrophoresed on a SDS-PAGE gel and then transferred to nitrocellulose membrane (Millipore). Membranes were then blocked in 5% milk and probed with anti-Gli-2, β-tubulin, GAPDH, and Gli-1 antibodies overnight.

For the study involving Du145 cells, 10⁶ cells in 100 μL of PBS were injected subcutaneously into the right flank of 4-week-old nude female mice. Once tumors were palpable, the mice were randomized to different groups, each group had 8 mice: control, taxotere, darinaparsin, and combinations of taxotere and darinaparsin. Mice were treated with taxotere (6 mg/kg) i.p. once a week for 3 weeks and with darinaparsin (100 mg/kg) i.p. every other day for 10 days. Treatment with the combination followed the same schedule and doses. For the study involving TICs, 20,000 α2β1^hi/CD44^hi cells (5-minute collagen-attached cells) or 20,000 α2β1^low/CD44^low cells (20-minute unattached cells) were injected subcutaneously in nude mice and the next day treatment was initiated with injections of darinaparsin or taxotere (doses and schedules as above). Tumor size and body weights were measured three times a week and the tumor volume was calculated using the formula width² × (length/2). For the study involving PC3 cells, 2 × 10⁶ cells in 100 μL of PBS were injected subcutaneously into the right flank of NOD/SCIDγ mice. Once tumors were palpable, the mice were randomized into two groups (n = 4), control and darinaparsin. Mice were treated with the same dosage and schedule as described above for Du145 cells. Results are presented as mean ± SEM.

All experiments were performed three times, and each experiment was done in triplicate. Statistical analysis was performed using Prism software (GraphPad). In all cases, ANOVA followed by two-tailed, unpaired Student t tests were performed to analyze statistical differences between groups. P values of <0.05 were considered statistically significant.

Prostate cancer cell lines LnCap, Du145, and PC3 cells were treated with darinaparsin or ATO at various concentrations for 72 hours. After 72 hours, cell viability was measured using the MTS, and as shown in Fig. 1A, all three cell lines were sensitive to both darinaparsin and ATO; IC₅₀ concentrations of Du145, PC3, and LnCap cells ranged from approximately 5 to 10 μmol/L. Du145 and LnCap cells were more sensitive to darinaparsin than ATO, while PC3 cells were equally sensitive to both darinaparsin and ATO. Similar to the prostate cancer cell lines, primary prostate cancer cells isolated from five different patients (Gleason score 8) were equally sensitive to darinaparsin with IC₅₀ concentrations ranging from 2.5 to 20 μmol/L. (Fig. 1B). Primary prostate cancer cells isolated from primary prostate tumors were confirmed for the prostate epithelial cells by testing for EpCam and AR markers (Supplementary Fig. S1A and S1B).

Relapse in most solid tumors after conventional chemotherapy has been attributed to the existence of drug-resistant TICs. We previously developed a novel assay for TIC prostate tumor enrichment based on rapid adherence of TICs to collagen I (13). In brief, cells that attach on collagen I have high expression of CD44 and α2β1 (α2β1hi/CD44^hi) as compared with 20-minute unattached cells (α2β1^low/CD44^low). We also showed that α2β1^hi/CD44^hi cells have increased colony forming, migration, and adhesion ability and can form tumors in nude mice and are the TICs. As shown in Fig. 2A, darinaparsin treatment at an IC₅₀ concentration or 2-fold higher reduced the number of Du145 cells and PC3 cells that attached on collagen I after 5 minutes. Darinaparsin-treated cells as compared with taxotere or ATO-treated cells had fewer α2β1^hi/CD44^hi cells when grown as monolayer (Fig. 2B) or as spheroids (Fig. 2C; IC₅₀ of darinaparsin in spheroids from Du145 was determined by the MTS assay; Supplementary Fig. S2). Dot plots are in Supplementary Figs. S3 and S4. Darinaparsin treatment also reduced secondary colonies formed from dissociating α2β1^hi/CD44^hi spheroids (Fig. 3A), indicating an effect on the TICs. Darinaparsin also decreased levels of CD44^hi and α2β1^hi cells (Fig. 3B and C) from primary prostate cells, further validating the inhibitory effect of darinaparsin on TICs.

Darinaparsin induces an incomplete stress response with disruption of microtubules and Shh signaling (11). As darinaparsin is a potent inhibitor of growth of Du145 cells, we tested the effect of darinaparsin on Shh signaling. Prostate cancer–initiating cells from Du145 cultures (early-passage cells) isolated by the collagen attachment assay showed an increase in Gli-2 levels as compared with the parental or 20 minutes unattached β1^low/CD44^low cells, at both the protein and mRNA level (Fig. 4A, i and ii). Furthermore, treatment of cells with darinaparsin decreased the protein levels of Gli-2 (Fig. 4B). Interestingly, Gli-2 levels were decreased at 1.5 μmol/L concentration but not at 3 μmol/L. This experiment was repeated three times and the same pattern was observed. The mechanism by which Gli2 levels come back may be due to some feedback mechanism. We also investigated whether darinaparsin inhibits Gli-2 activity by measuring its effect on the Gli responsive luciferase promoter pGL38xGli, which contains 8 GLI DNA-binding sites attached to the chicken lens crystalline promoter followed by the luciferase gene (12). Du145 cells were transfected with the luciferase plasmid and cotransfected with pcDNA-Gli-2 plasmids. After 8 hours of transfection, cells were treated with a SAG and cotreated with darinaparsin or ATO or left untreated. Twenty-four hours later, cells were harvested and luciferase activity was assayed. The luciferase levels were normalized with total protein. As shown in Fig. 5A, such as ATO (15), darinaparsin also inhibited Gli transcriptional activity compared with vehicle control. The effect of darinaparsin on the cell cycle was also examined. Darinaparsin blocks the cells in G₂–M phase of cell cycle and but did not induce apoptosis (Fig. 5B). Therefore, the cytotoxic effect of darinaparsin may be attributed to the block in the G₂–M phase as well as inhibition of Gli2.

To show the antitumor effect of darinaparsin and to determine whether the combination of darinaparsin with taxotere produced additive or synergistic antitumor effects in vivo, nude mice bearing the Du145 tumor were injected with darinaparsin every other day for 5 times, taxotere every once a week for three times, or with both taxotere and darinaparsin in the same doses and schedule to determine toxicity. Interestingly, darinaparsin inhibited tumor growth as effectively as taxotere. However, unlike the in vitro study (Supplementary Fig. S5), the combination of taxotere and darinaparsin did not have more of an effect than either of the drugs alone (Fig. 6A). Neither drug, nor the combination caused toxicity, was measured by weight loss (Fig. 6b). We also tested the efficacy of darinaparsin in human PC3 xenografts, also castrate resistant, to determine whether other prostate cancers also were inhibited by darinaparsin. As shown in Fig. 6C, darinaparsin treatment inhibited tumor growth without evidence of toxicity, as measured by a decrease in body weight (Fig. 6D).

Mice were injected with 20,000 α2β1^hi/CD44^hi cells (enriched for TICs) or with 20,000 α2β1^low/CD44^low cells (depleted of TICs) and treated with darinaparsin or with taxotere before tumors were palpable (day 1 after implantation). As shown in Table 1, only 3 of 8 darinaparsin-treated mice developed tumors, whereas 5 of 8 taxotere-treated mice developed tumors. Six of 8 untreated mice formed tumors (Table 1).

In this study, we report that darinaparsin is effective in killing both prostate cancer cells from cell lines as well as cells from fresh prostate tumors at low micromolar concentrations. The IC₅₀ concentration of darinaparsin was lower than ATO in Du145 and LnCap cells, whereas in PC3 cells, IC₅₀s of both darinaparsin and ATO were similar. Importantly, darinaparsin was efficient is reducing the number of α2β1^hi/CD44^hi cells and reducing the number of primary and secondary colonies formed from α2β1^hi/CD44^hi spheroids, indicating that darinaparsin and ATO are cytotoxic to TICs, and darinaparsin is more potent. We previously showed that clinically used chemotherapeutic drugs did not affect TICs from Du145 TICs (12).

ATO antagonizes Shh signaling by reducing Gli2, a transcriptional effector (12). ATO treatment inhibits the ciliary accumulation of Gli2 and hence blocks trafficking required for activation. In addition, ATO caused a reduction of overall levels of Gli2. Moreover, in SMO inhibitor–resistant cells, ATO efficiently inhibited tumor growth either alone or in combination with itraconazole, a compound that also inhibited Shh signaling (16). In this study, we show that darinaparsin also inhibited Shh signaling. We also show that darinaparsin blocks cell in G₂–M. Reports showing that darinaparsin inhibits microtubule polymerization and induces stress granule formation (11) may be the mechanism by which it blocks cells in G₂–M (Fig. 3D).

The mechanism by which darinaparsin kills cells appears to be complex. In hematopoietic cell lines, an increase in oxidative stress was shown to play an important role in darinaparsin action. In solid cancers, gene expression studies comparing darinaparsin to ATO (17) showed that darinaparsin decreased the expression of genes that were upregulated by Myc. In this study, we show that darinaparsin inhibited growth of two human prostate cancer xenografts in nude mice and reduced the number of mice with tumors that were initiated with TICs. However, in contrast to the in vitro results that showed synergistic cell kill between taxotere and darinaparsin (Supplementary Fig. S3), the in vivo study with a combination of taxotere and darinaparsin was not even additive. A possible explanation for lack of additivity or synergy is that darinaparsin and taxotere both block cells in G₂–M. It is also possible that different doses and schedules would cause additive or synergistic effects. Itraconazole, also inhibits Shh signaling by inhibiting downstream gli-2 transcription, such as ATO and darinaparsin, (16). In a phase II study, itraconazole showed only modest antitumor activity, and dose escalation was limited by toxicity (15).

The development of an oral form of darinaparsin and evidence of antitumor activity in phase I and II trials (6, 18), together with the results of this study, provide a rationale for evaluating darinaparsin in patients with castrate-resistant prostate cancer.

J. Lewis has ownership interest (including patents) in Ziopharm Oncology Inc. J.R. Bertino reports receiving a commercial research grant from and is a consultant/advisory board member for Scientific Review Committee of Ziopharm Oncology, Inc. No potential conflicts of interest were disclosed by the other authors.

Conception and design: N. Bansal, L. Wu, J.R. Bertino

Development of methodology: N. Bansal, H. Youssoufian

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): N. Bansal, N.J. Farley, L. Wu

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): L. Wu, J. Lewis, H. Youssoufian, J.R. Bertino

Writing, review, and/or revision of the manuscript: N. Bansal, J. Lewis, H. Youssoufian, J.R. Bertino

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): N. Bansal, J. Lewis, H. Youssoufian

Study supervision: N. Bansal, J.R. Bertino

The authors thank Dr. P.A. Beachy for providing them with 8xGli-luciferase reporter plasmid construct.

This work was supported by a grant from Ziopharm Oncology Inc. to J.R. Bertino.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.