A Novel Transcriptional Inhibitor Induces Apoptosis in Tumor Cells and Exhibits Antiangiogenic Activity

In eukaryotes, mRNA synthesis is mediated by the concerted action of a number of factors, chief among them being the RNA polymerase II (1). The process of RNA polymerase II transcription consists of the preinitiation, initiation, and the elongation stages (2). Several inhibitors of transcription are known, which work by blocking one or more of these stages. For instance, actinomycin D, which is both a transcriptional inhibitor and a DNA damage agent, intercalates within the DNA and thus inhibits the initiation stage of transcription (3). Flavopiridol and 5,6-dichloro-1-β-ribofuranosylbenzimidazole (DRB) target the elongation stage of transcription, by inhibiting positive transcription elongation factor b (P-TEFb, a complex of Cdk9/cyclin T1; ref. 4–6), whose phosphorylation of RNA polymerase II is essential for this stage. α-amanitin, on the other hand, binds directly to RNA polymerase II, which leads to the inhibition of both initiation and elongation stages (7–10).

General transcriptional inhibitors may be useful in cancer therapies, and in some instances, have been shown to work as antiviral agents (11, 12). For example, flavopiridol is a very efficient inducer of apoptosis in malignant cells and it also potentiates the lethal effects of other cytotoxic drugs (11, 13). In addition, it inhibits cell migration and displays potent antiangiogenic activity (14, 15). Specifically, this class of drugs may be useful against tumors that express labile antiapoptotic proteins due to their ability to down-regulate proteins of short half-life (11). Understandably, these drugs may also act synergistically with certain factors such as tumor necrosis factor-α, which is known to transcriptionally induce antiapoptotic proteins (16).

In this study, using a high-throughput cell-based assay system, we identified a novel small molecule that acts as a general transcriptional inhibitor by inhibiting RNA polymerase II phosphorylation. Interestingly, this compound induces apoptosis in transformed and cancer but not normal cells and exhibits potent antiangiogenic activity in vitro.

Cell lines and media. Colon cancer LIM1215, breast cancer MCF-7, liver cancer HepG2, gastric cancer AGS, and prostate cancer LNCaP cells were obtained from American Type Culture Collection (Manassas, VA). Wild-type and SV40-transformed MRC-5 human fetal lung fibroblasts were obtained from Coriell Institute (Camden, NJ). Isogenic colon cancer HCT-116 wild-type and p53^−/− cell lines were a kind gift from Dr. Vogelstein (Howard Hughes Medical Institute and The Sidney Kimmel Comprehensive Cancer Center, Johns Hopkins University, Baltimore, MD). MCF-7 and HepG2 cells with stable knock-down of p53 (MCF-7-p53si and HepG2-p53si, respectively) have been described previously (17). The cells were grown in different media as described (18).

High-throughput screening of chemical library. LIM1215 colon carcinoma cells harboring the lacZ reporter gene under the control of a 2.3 kb p21 promoter (containing two p53-binding sites) were grown in 96-well plates. Diversity set, a chemical compound library containing ∼2,000 compounds (dissolved in DMSO) was obtained from the National Cancer Institute (NCI) and was used at a final concentration of 10 μmol/L on the cells for 24 hours. lacZ encoded β-Gal was detected by X-Gal staining as described previously (19).

Immunoblot analysis. Immunoblotting was done as previously described (20) with antibodies specific for p53 (sc-126 horseradish peroxidase; Santa Cruz Biotechnology, Santa Cruz, CA), hdm2 (sc-965; Santa Cruz), p21 (556431; BD PharMingen, San Diego, CA), Sp1 (sc-59; Santa Cruz), survivin (sc-10811; Santa Cruz), phospho RNA polymerase II (4735; Cell Signaling Technology, Danvers, MA), total RNA polymerase II (8WG16; a gift from Dr. Schlegel, Department of Pathology and The Cancer Center, University of Illinois at Chicago, Chicago, IL), α-tubulin (T9026; Sigma, St. Louis, MO), cleaved caspase-3 (9664; Cell Signaling), caspase-9 (9502; Cell Signaling), and β-actin (A5441; Sigma-Aldrich, St. Louis, MO) antibodies.

Nuclear run-on assay. Run on assays with isolated nuclei were done as described (4). Briefly, cells treated with the indicated agents were lysed with hypotonic buffer [10 mmol/L Tris-HCl (pH 7.4), 10 mmol/L NaCl, and 35 mmol/L MgCl₂] and nuclei collected by centrifugation and resuspended in nuclear storage buffer [50 mmol/L HEPES (pH 8.0), 5 mmol/L MgCl₂, 0.5 mmol/L DTT, 1 mg/mL bovine serum albumin, and 25% (v/v) glycerol] at about 5 × 10⁸ nuclei/mL and stored at −80°C. Transcription reactions were done in 200 μL volume consisting of 1 × 10⁷ nuclei in the presence of 0.12 mol/L KCl, 7 mmol/L Mg(Ac)₂, 25 μCi of [³²P]GTP, 500 μmol/L ATP, UTP, and CTP at 30°C. Samples were collected at various time points (18 μL/time point—0, 5, 10, 20, and 30 minutes) and the reaction terminated by the addition of 57 μL of Sarkosyl solution [1% Sarkosyl, 0.1 mol/L Tris (pH 8.0), 0.1 mol/L NaCl, 10 mmol/L EDTA, and 200 μg/mL tRNA]. The stopped reactions were transferred to Whatman DE81 paper, and washed four times with wash buffer (5% K₂HPO₄ and 0.3% Na₄P₂O₇) for 10 minutes, followed by a 5-minute water wash. Then, the paper was briefly rinsed with 95% ethanol and allowed to dry completely. Radiation was quantitated using a liquid scintillation counter.

Kinase assays. Purified P-TEFb (Cdk9/cyclin T1) protein and glutathione S-transferase-tagged COOH-terminal domain of RNA polymerase II (GST-CTD) expression plasmid were gifts from Dr. Nekhai (Center for Sickle Cell Disease and Department of Biochemistry and Molecular Biology, Howard University, Washington, DC). GST-CTD protein was expressed in BL21 Escherichia coli (Stratagene, La Jolla, CA) and purified using a GST protein purification kit (Amersham Biosciences, Piscataway, NJ). The kinase assay was done at 30°C for 1 hour in the presence of P-TEFb with or without the addition of 4-amino-6-hydrazino-7-β-d-ribofuranosyl-7H-pyrrolo(2,3-d)-pyrimidine-5-carboxamide (ARC). The phosphorylation levels of GST-CTD were detected with anti-phospho-RNA polymerase II antibody (4735; Cell Signaling).

Apoptosis assays. Apoptosis was detected either by 4′,6-diamidino-2-phenylindole (DAPI) staining or by Flow Cytometry after propidium iodide staining. For DAPI staining, all treatments (DMSO or ARC) were done in triplicate in six-well plates and cells were stained with DAPI and visualized by fluorescent microscopy. Four random fields for each sample were photographed and at least 500 cells per field were counted to estimate apoptosis. The data is represented as mean ± SD.

For flow cytometry, cells were fixed in 70% ethanol and stored at −20°C until further analysis. Equal numbers of fixed cells were then stained with propidium iodide solution (0.1% Triton X-100, 0.2 mg/mL DNase-free RNase, and 0.02 mg/mL propidium iodide) made in PBS for 15 minutes at 37°C and analyzed by a flow cytometer.

Angiogenesis assays. Different angiogenesis assays such as cord formation assay, motility assay, and proliferation inhibition assay were done at the developmental therapeutics program (DTP) branch of the NCI as previously described (21). Human umbilical vein endothelial cells (HUVEC) were used for all the assays. Taxol and TNP-470 were used as positive controls for the assays.

A novel compound ARC represses p53 targets but increases p53 levels. Recent studies have shown that blocking the function of p21 enhances apoptosis of cancer cells (22–26), which could be beneficial in cancer therapy. In order to find small-molecule transcriptional inhibitors of p21, we generated LIM1215 colon cancer cells, carrying lacZ under the control of p21 promoter, as a screening system for testing ∼2,000 structurally diverse compounds (diversity set) obtained from the NCI. Out of the five compounds that were able to repress the p21 promoter (judged on their ability to attenuate p21 promoter driven β-Gal expression), we chose the most potent one for detailed characterization. The ability of the compound, which we named ARC (NSC-188491), to inhibit p21 promoter is illustrated in a sample of the screen (Fig. 1A). ARC is a nucleoside analogue (Fig. 1B), and was able to repress p21 levels in a variety of cell lines derived from cancers of different origin (Fig. 1C). In addition, ARC was able to down-regulate hdm2 (an induced target of p53) and survivin (a repressed target of p53), although it increased p53 levels (Fig. 1C). The repression of p21 and hdm2 were reproducible in a PC3 cell line where temperature-sensitive p53 was overexpressed (Supplementary Fig. S1). To determine if these effects mediated by ARC were dose-dependent, we treated HCT-116 colon carcinoma cells with different concentrations of ARC. We found that ARC increased p53 but simultaneously repressed its target genes p21 and hdm2 in a dose-dependent manner, whereas no significant changes were observed in Sp1 and β-actin levels (Fig. 1D).

In order to find out if the ARC-mediated repression of p21, hdm2 and survivin are p53-dependent, we employed a pair of isogenic colon cancer cell lines differing only in their p53 status—HCT-116 wild-type and HCT-116-p53^−/− cells (a gift from Dr. Bert Vogelstein). Comparison of the p21, hdm2, and survivin levels in HCT-116-p53^+/+ and HCT-116-p53^−/− before and after treatment with ARC revealed that these genes are repressed irrespective of their p53 status (Fig. 1E), implying that ARC operates through a p53-independent mechanism.

ARC is a general transcriptional inhibitor that functions by inhibiting RNA polymerase II phosphorylation. Some transcriptional inhibitors such as DRB and flavopiridol repress p21 and hdm2 whereas increasing p53 levels (4, 16). Because ARC displays a similar pattern, it is possible that it acts as a global transcriptional inhibitor. To test this hypothesis, HCT-116 cells were treated with DMSO (control), ARC, or DRB, and nuclei were isolated. These nuclei were assayed for rate of transcription either in the presence or absence of α-amanitin using the nuclear run-on assay. We found that the rates of transcription were lower in nuclei treated with ARC and DRB than the control nuclei (Fig. 2A), clearly demonstrating that ARC could act as an inhibitor at the level of transcription. It is known that α-amanitin inhibits RNA polymerase II, but does not affect RNA polymerase I- and III-based transcription. If ARC inhibits polymerase I and III transcription, the rate of transcription should be substantially lower in the ARC + α-amanitin (curve 5, Fig. 2A) when compared with DMSO + α-amanitin (curve 4, Fig. 2A). However, both of these rates of transcription were similar, implying that ARC might preferentially affect only RNA polymerase II transcription.

In order to find if ARC can bind to and inhibit RNA polymerase II directly (like α-amanitin) or indirectly (like DRB), we isolated nuclei from normally growing untreated HCT-116 cells and treated them with each of these agents, and then assayed for rate of transcription by nuclear run-on assay. The addition of DRB or ARC did not affect the rate of transcription, whereas the addition of α-amanitin dramatically inhibited transcription (Fig. 2B), implying that ARC affects polymerase II transcription in an indirect manner.

It is known that DRB and flavopiridol inhibit transcription by blocking the kinase activity of P-TEFb, which in turn leads to decreased phosphorylation of RNA polymerase II CTD (4). To test if ARC operates by a similar mechanism, we did a kinase assay with purified P-TEFb using GST-CTD as the substrate. The addition of ARC potently decreased the phosphorylation of GST-CTD (Fig. 2C), suggesting that ARC can inhibit P-TEFb kinase activity. To test if this is true in vivo, we treated HCT-116 cells with ARC for 3 hours and analyzed the levels of phospho-RNA polymerase II by immunoblotting. We found that whereas the total amount of RNA polymerase II remained unchanged, phosphorylation of the RNA polymerase II CTD decreased upon ARC treatment (Fig. 2D), consistent with the in vitro kinase assay.

ARC induces apoptosis in transformed cells and cancer cells, but not normal cells. Flavopiridol and DRB have been shown to cause the apoptosis of cancer cells (11, 27). We tested the ability of ARC to induce apoptosis in wild-type and SV40-transformed MRC-5 human fetal lung fibroblasts. Twenty-four hours after treatment with ARC, transformed fibroblasts underwent robust apoptosis (∼50% and 70%, respectively, with 5 and 10 μmol/L of ARC; Fig. 3A and B). However, the wild-type fibroblasts were not susceptible to ARC-mediated cell killing, even at 10 or 20 μmol/L of ARC (Fig. 3A). We also observed the appearance of cleaved caspase-3 upon ARC treatment in transformed but not normal fibroblasts (Fig. 3C), clearly demonstrating that ARC promotes apoptosis specifically in transformed cell types. Interestingly, we were able to observe several-fold higher levels of survivin in transformed fibroblasts when compared with the wild-type, and that these levels were further reduced after ARC treatment in both cell types (Fig. 3C). We speculate that the transformed cell types could be more dependent on the survivin levels for their survival than their wild-type counterparts, and hence, the reduction in these levels could be lethal to these cells.

We further analyzed the effect of 20 μmol/L ARC treatment on wild-type fibroblasts and found that there is little change in the cell cycle profile after 24 hours (data not shown) and that they undergo G2-arrest after 48 hours (Fig. 3D). These results suggest that whereas ARC induces efficient apoptosis in transformed cells, it causes a cell cycle block in normal cells, making it an ideal candidate for anticancer drug development.

In order to further characterize ARC with respect to its apoptotic potential, we used cell lines derived from cancers of different origin. We observed that MCF-7 breast cancer cells underwent apoptosis (∼45% and 55%, respectively, with 10 and 20 μmol/L of ARC) after 24 hours of treatment (Fig. 4A and B). Moreover, we observed cleavage of caspase-9 upon ARC treatment in these cells (Fig. 4C), suggesting that these cells undergo caspase-mediated apoptosis. To test if ARC-mediated apoptosis is p53-dependent, we used a MCF-7 cell line that stably expresses a short hairpin RNA targeting p53 (MCF-7-p53si; ref. 17). We were able to detect cleavage of caspase-9, even in the absence of p53 in MCF-7-p53si cells (Fig. 4D), implying that cell death mediated by ARC is p53-independent. To determine if ARC induces the apoptosis of other cancer cells, we then treated different cell lines (LIM1215 colon cancer, AGS gastric cancer, and HepG2 liver cancer cells) with ARC for 48 hours and detected apoptosis (data not shown) and cleavage of caspase-3 (Fig. 4E). Also, HepG2 cell line with stable knock-down of p53 (HepG2-p53si; ref. 17) showed cleavage of caspase-3 confirming that ARC-mediated apoptosis is p53-independent (Fig. 4E).

In order to evaluate if ARC induces apoptosis comparable to that of the other known nucleoside analogue, DRB, we treated LIM1215 colon cancer cells with either ARC or DRB for 48 hours. Using propidium iodide staining followed by flow cytometry, we found that 5 μmol/L of ARC treatment resulted in 37.8% apoptosis, whereas a 50 μmol/L of DRB treatment only showed 23.4% apoptosis (Fig. 4F), suggesting that ARC is more efficient in eliciting apoptosis than DRB at least in these LIM1215 cells.

ARC is a potent antiangiogenic agent in vitro. Next, we wanted to check if ARC could inhibit angiogenesis (a process of blood vessel formation that is necessary for tumors in vivo). For this purpose, we sent the compound ARC to NCI/DTP where three types of angiogenesis assays (cord formation assay, motility assay, and proliferation inhibition assay) were done. HUVECs which are similar to the endothelial cells that line the blood vessels are used for these assays. Normal endothelial cells in vivo divide very rarely (once about every 3 years on the average), except during tumor development when they divide actively, contributing to the growth of a new network of blood vessels. Therefore, any inhibition of their growth and motility by ARC in vitro would suggest that ARC could be a potential antiangiogenic agent.

For cord formation assay, HUVECs were plated on a three-dimensional layer of Matrigel, where they formed cord-like structures. Later, they were treated with vehicle (DMSO) or different concentrations of ARC for 24 hours, after which cord formation was quantitated as a function of cord junctions and cord length. A dose-dependent decrease was seen in response to ARC treatment in both junction and length (Fig. 5A) with respective IC₅₀ values of 344.87 and 382.3 nmol/L (averages from three experiments). These values are about half of what was observed for TNP-470, an exclusive antiangiogenic agent in clinical trials (28), for which the IC₅₀ is ∼700 nmol/L (29), suggesting that ARC is a more potent angiogenesis inhibitor in this assay. For the motility assay, HUVECs were pretreated for 24 hours with different concentrations of ARC and used in the assay with vascular endothelial growth factor (VEGF) as a chemoattractant. ARC induced a dose-dependent reduction in cell motility (Fig. 5B) with an IC₅₀ value of 830 nmol/L (average from two experiments). This value is comparable to that of TNP-470, which has an IC₅₀ of 600 nmol/L (29). For the proliferation inhibition assay, HUVECs were grown in the absence of growth factors for 24 hours and then stimulated with 10 ng/mL VEGF, and were treated with either DMSO or different concentrations of ARC for 24, 48, or 72 hours. ARC inhibited HUVEC growth in a time- and concentration-dependent manner (Fig. 5C) with an IC₅₀ of 46.9 nmol/L (average from two experiments). Taken together, these data suggest that ARC could be a potent antiangiogenic agent.

In this study, we isolated a novel nucleoside analogue ARC in a screen for inhibitors of p21 transcription. Although ARC reproducibly repressed p21 in several cell lines, indicating that the screen was successful, it also suppressed other genes such as hdm2 and survivin (Fig. 1C-E). Further investigation revealed that ARC is a global transcriptional inhibitor (Fig. 2A). One common feature of the genes repressed by ARC is their short half-life at both the mRNA and protein levels (5). It has been suggested that induction of p53, accompanied by down-regulation of p21 and hdm2, is a hallmark of repression of transcription (16, 27, 30). This increase in p53 could be due to the down-regulation of the short-lived protein, hdm2, which is a well-established negative regulator of p53 (16, 31). In agreement with this notion, we saw a similar effect in several cancer cell lines upon treatment with ARC (Fig. 1C).

In our hands, ARC and another nucleoside analogue DRB, repressed RNA polymerase II transcription to similar levels as measured by nuclear run-on assays (Fig. 2A). One way to repress RNA polymerase II is by direct interaction as seen in the case of α-amanitin (10). However, treatment of nuclei with ARC in vitro failed to inhibit transcription (Fig. 2B), ruling out its direct interaction with RNA polymerase II. We found that ARC employs an indirect mechanism to inhibit transcription, similar to DRB and flavopiridol (4, 32). Phosphorylation of the CTD of RNA polymerase II by P-TEFb kinase is a requisite step for transcriptional elongation (33, 34), and ARC blocks this process as seen by its inhibition of P-TEFb, leading to decreased phosphorylation on RNA polymerase II (Fig. 2C and D). Being an adenosine analogue, it seems conceivable that ARC might compete for the ATP-binding site of the P-TEFb kinase leading to its reduced kinase activity (Fig. 2C).

A noteworthy property of ARC is its ability to induce apoptosis in transformed and various cancer cells (Figs. 3 and 4). Apoptosis in mammalian cells is a multistep process that results in the activation of caspases, a subfamily of cysteine proteases, followed by execution of cell death. Due to the lack of functional caspase-3 expression, MCF-7 breast cancer cells do not undergo apoptosis easily (35). However, our data indicate that ARC is able to induce efficient apoptosis in MCF-7 cells (Fig. 4A and B), suggesting that ARC invokes a caspase-3-independent cell death pathway in these cells. The other cell lines that we tested (SV40-transformed fibroblasts, LIM1215, AGS, and HepG2) have caspase-3 expression and hence showed caspase-3 cleavage as a result of ARC treatment (Figs. 3C and 4E).

A striking feature of ARC-mediated apoptosis is that it is p53-independent. We have shown that MCF-7 and HepG2 cells along with their p53-knocked-down counterparts undergo apoptosis to the same extent (Fig. 4D and E; data not shown). This is in agreement with the observation that p53 is a marker but not a mediator of flavopiridol-induced cytotoxicity (16). However, other transcriptional inhibitors DRB and α-amanitin, were suggested to induce p53-dependent apoptosis (27, 30). It was shown recently that α-amanitin induced apoptosis in the absence of p53-dependent transcription by direct translocation of p53 to mitochondria, a phenomenon which has also been observed by other investigators in different situations (36–38). In our study, however, despite the accumulation of transcriptionally incompetent p53 as seen by its inability to induce its target genes (Fig. 1A and B) in response to ARC treatment, it did not lead to enhanced apoptosis. The reason for this discrepancy is unclear. The differences in the cell lines or experimental approaches used in these studies could be the reason behind the difference in outcome. Also, because α-amanitin represses transcription by a mechanism other than ARC, its mode of inducing apoptosis could be different.

Although we have shown that ARC causes p53-independent, caspase-mediated, apoptosis, the precise step that is targeted by ARC to elicit this response has been elusive. The down-regulation of myeloid cell leukemia 1, a short-lived antiapoptotic protein, was suggested as a possible mechanism of flavopiridol-induced apoptosis in leukemia cells (39). In our case, it is reasonable to speculate that the ability of ARC to down-regulate antiapoptotic proteins including, but not limited to survivin and p21 (Figs. 1C and 4C), may play an important role mechanistically. Survivin, a member of the mammalian inhibitors of apoptosis protein family is overexpressed in a variety of tumors and has been proposed as an attractive target for cancer therapy (40, 41). It was found that survivin expression increased gradually in the transition from normal colorectal mucosas to adenomas with low-grade dysplasia to high-grade dysplasia/carcinomas (42). In agreement with this fact, we saw higher levels of survivin expression in the SV40-transformed fibroblasts when compared with the wild-type fibroblasts (Fig. 3C). Apart from apoptosis inhibition, survivin also plays an important role in mitosis (41). Accordingly, it is possible that the low levels of survivin in wild-type fibroblasts, when further down-regulated by ARC, resulted in G2 block (Fig. 3D), due to the inability of the cells to proceed to mitosis. However, in the SV40-transformed fibroblasts, survivin may primarily be antiapoptotic, thus, leading to cell death when these levels are down-regulated by ARC (Fig. 3A-C).

Similarly, p21 may also act in an antiapoptotic fashion by inducing cell cycle arrest or by inhibiting proapoptotic molecules such as procaspase-3 and apoptosis signal-regulating kinase-1 (43, 44). It has been suggested that inhibition of p21 may be an effective strategy for enhancing apoptosis in cancer cells mediated by anticancer agents (22–24, 45, 46). The fact that ARC represses both survivin and p21 makes a strong case for consideration of ARC as a potential antitumor agent. This is further supported by the observation that ARC was >10-fold more efficient in inducing apoptosis than the other nucleoside analogue, DRB (Fig. 4F). This is especially intriguing because we have shown that ARC and DRB repress global transcription to a similar extent (Fig. 2A). The difference in apoptosis suggests that ARC, apart from transcriptional repression, might employ additional means of inducing cell death. Our results also suggest that ARC is antiangiogenic in nanomolar concentrations in vitro, and its activity is comparable to TNP-470 (29). Although we have not tested ARC in any animal models, preliminary toxicity studies from NCI/DTP indicate that concentrations as high as 200 mg/kg were nontoxic to B6D2F1 mice during a 5-day period, suggesting that ARC could be a promising candidate for in vivo studies.

In summary, the ability of ARC to induce apoptosis in transformed cells and cancer cells, but not normal cells, along with its potent antiangiogenic activity, makes it ideally suited for anticancer and/or antiangiogenic drug development.

Grant support: NIH grant CA91146, the 2006 Penny Severns award from the Illinois Department of Public Health and start-up funds from the Department of Medicine, University of Illinois at Chicago to A.L. Gartel.

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

We thank Dr. Price for helpful suggestions for the nuclear run-on assays, Dr. Nekhai for providing P-TEFb protein and GST-CTD constructs, Dr. Vogelstein for the HCT-116 (wild-type and p53^−/−) cells, and Dr. Schlegel for anti-total RNA polymerase II antibody; NCI/DTP for performing the angiogenesis assays; and Drs. Tyner, Kandel, and Costa for helpful discussions.