Small-molecule targeting hairpin loop of hTERT promoter G-quadruplex induces cancer cell death

SUMMARY

Increased telomerase activity is associated with malignancy and poor prognosis in human cancer, but the development of targeted agents has not yet provided clinical benefit. Here we report that, instead of targeting the telomerase enzyme directly, small molecules that bind to the G-hairpin of the hTERT G-quadruplex-forming sequence kill selectively malignant cells without altering the function of normal cells. RG260 targets the hTERT G-quadruplex stem-loop folding but not tetrad DNAs, leading to downregulation of hTERT expression. To improve physicochemical and pharmacokinetic properties, a small-molecule analog, RG1603, was derived from the parent compound. RG1603 induces mitochondrial defects including PGC1α and NRF2 inhibition and increased oxidative stress, followed by DNA damage and apoptosis. RG1603 injected as a single agent has tolerable toxicity while achieving strong anticancer efficacy in a tumor xenograft mouse model. These results demonstrate a unique approach to inhibiting the hTERT that functions by impairing mitochondrial activity, inducing cell death.

INTRODUCTION

Telomerase is an essential enzyme to cancer cells in maintaining telomere function (Harley et al., 1994; Wellinger et al., 1996). Human cancer cells acquire telomerase activity by activating or upregulating the normally silent human telomerase reverse transcriptase gene (hTERT) (Nugent and Lundblad, 1998). hTERT gene is usually silenced in somatic cells but expressed at high levels in ~90% of human cancers (Blasco and Hahn, 2003; Kim et al., 1994). Recent studies have shown that key mutations in hTERT promoter region elevate hTERT expression, telomerase activity, and telomere length in various cancer types, including melanoma, glioblastoma, and bladder cancer (Borah et al., 2015; Horn et al., 2013; Killela et al., 2013). In these studies, patients with tumors expressing high levels of hTERT showed significantly worse overall survival rates than those with lower expression of hTERT, directly suggesting the role of the promoter in controlling patient outcomes.

Telomeric DNA is capable of folding into four-stranded guanine quadruplex (G4) structures (Bochman et al., 2012; Palumbo et al., 2009). Starting from single-stranded DNA, the hTERT G4 at the 3′-end nearest the transcription start site presents an unusual 26-base hairpin loop, which is proposed to act as the initiation point for cooperative folding of the G4 (Kang et al., 2016; Yu et al., 2012). This sequence is critical to the inhibition of hTERT transcription because somatic mutations found naturally in the same hairpin of the G4 resulted in disruption of G4 folding, leading to overexpression of hTERT (Kang et al., 2016). Targeting the major G4 in the hTERT promoter with the guanidine-substituted acridine compound GTC365 resulted in inhibition of hTERT activity. GTC365 acted as a pharmacological chaperone by re-steering the folding process such that the effect of misfolding of the silencer G4 caused by somatic mutations could be reversed. The generation of G4 drugs that target the unique heteroduplex hairpin loop without targeting the G4 tetrad would be predicted to increase selectivity because tetrads are the conserved features of G4s found in many other genes, such as BCL2, MYC, KRAS, and PDGFR-β (Bugaut et al., 2010; Burger et al., 2005; Zhang et al., 2017a). Sequence variability and associated structural variability in the loops provide the opportunity for both better selectivity and drug-like properties.

We set out to see if we could identify a different set of molecules that lacked the acridine moiety but still possessed the chaperone properties of GTC365. We find that the folding pattern of the final G-quadruplex formed from binding of compounds that lack the tetrad targeting moiety leads to a different folded product than with GTC365 that can be rationalized based on the initial drug binding site and subsequent folding steps that most probably mimic the natural folding process. This singular mechanism of action permits reduced telomerase production and telomere shortening. However, the early onset of anticancer effects, for example, inhibition of cell proliferation and induction of apoptosis, are driven by telomere-independent mechanisms. In vivo studies demonstrate that systemic administration of small molecules stabilizing hTERT G4s significantly delays tumor growth. These studies provide data supporting hTERT as an important cancer drug target.

RESULTS

Discovery of a small molecule that selectively targets the hairpin stem-loop of the WT hTERT G4 to downregulate hTERT expression

Single-stranded oligomer for the full-length (FL)-G4 in the hTERT promoter element can form a tandem G-quadruplex containing an unusually large 26-base hairpin stem-loop (Figure S1A) (Bochman et al., 2012; Palumbo et al., 2009). We carried out a FRET assay–based screening of ~1500 compounds in the NCI diversity set (Kang et al., 2016) and identified two sets of structurally distinct compounds that quenched the fluorescence in the FRET assay (Figure 1A). The subsequent DMS footprinting experiments on hTERT G4 in presence of GTC365 were performed with 5 mM KCl buffer rather than the 140 mM KCl buffer. The results of these experiments show that the footprinting on the GTC365-bound WT G4 in 5 mM KCl is similar to the footprinting obtained on the hTERT WT sequence in 140 mM KCl. This indicates that GTC365 interacts with the same final folding pattern for the tandem G4s with the 26-base hairpin contained in the 5′-G4 (Figure S1A). GTC365 consists of an acridine-derived moiety similar to BRACO19 that binds to G-quadruplexes through the recognition of the G-tetrad and a guanidino group that binds to the lower part of the 26-base hairpin loop (Kang et al., 2016).

Figure 1. Identification and characterization of RG260 as an hTERT downregulator by stabilizing the hTERT promoter G4.

(A) Principle of compound screening by FRET assay. The ends of the 5−12 G-tract were labeled with FAM at the 5′-end and TAMRA at the 3′-end, and the folding of the oligomer brings each end closer. Subsequently, FAM fluorescence is quenched by TAMRA. A representative result of screening of the NCI Diversity Set III is shown.

(B) Chemical structures of GTC365 and RG260 (NSC654260).

(C, D) DMS footprinting showing the different folding pattern of the 1–12 full-length G-quadruplex by RG260 (20 equiv) and GTC365 (20 equiv) with 5 mM KCl (C) or 100 mM CaCl₂ (D).

(E) Dose-dependent decrease of hTERT core promoter activity by RG260. MCF7 cells were transfected with pGL3 constructs and pRL-TK for 6 h and then treated with RG260 for 24 h. A dual luciferase assay was performed to obtain luciferase activity of firefly and renilla for normalization. Data are shown as mean ± SD of four replicates.

(F) Dose-dependent inhibition of telomerase activity by RG260. MCF7 cells were treated with RG260 for 72 h before being subjected to TRAP assay.

From this same FRET screening assay we identified a second set of structurally distinct compounds (Figure 1A), typified by RG260 (NSC654260). These compounds are the subject of this contribution, and the characterization of their effect on the folding pattern of the hTERT G4 was determined by DMS footprinting in 5 mM KCl, as was the WT G4 complex with GTC365. RG260 contains a benzoylphenylurea (BPU) moiety but lacks the tetrad-binding acridine moiety found in GTC365 (Figure 1B). We initially speculated that the BPU moiety might mimic the guanidino moiety of GTC365, which has been shown by chemical footprinting to bind to the base of the 26-base hairpin stem-loop rather than the 3′-tetrad in the G4. Unexpectedly, RG260 showed no significant change of conformation of the G4 as measured by CD at a pH of 7.5 (Figure S1B). To gain some insight into how RG260 interacts with the 5–12 G4, we first checked if changes in pH would affect the protonation status of cytosine and adenine, but not guanine, thereby inducing a change of Watson–Crick base-pairings in the stem-loop but not G-tetrads. The maximum CD signal at around 262 nm at pH 6.8 was significantly decreased with a little increase in a shoulder at around 290 nm compared to the spectra at pH 7.5 (Figure S1C). This change in CD is indicative of a conformational change just in the hairpin stem-loop structure. When RG260 was added, the T_m was increased by 2 °C (Figure S1D), which confirms our hypothesis that at a lower pH, conformational change of the hairpin stem-loop is specifically induced by addition of the drug so that the entire structure is stabilized. This result is in line with the role of the hairpin stem-loop of the 5–12 G4 as a nucleation site for folding of hTERT G4, as proposed previously (Kang et al., 2016).

In the next series of experiments, we compared the DMS footprinting pattern of the FL-G4 in the presence of GTC365 and RG260 using the 5 mM KCl buffer that we had used in the previous experiments with GTC365 (Figure 1C). We were surprised to see that the GTC365- and RG260-bound DMS footprints were quite distinct and that the WT footprint in 5 mM KCl buffer was different to that found in the initial DMS footprinting previously carried out in 140 mM KCl buffer. This clearly shows that the KCl buffer concentration was a determinant of the final folding pattern, which is most likely determined by the initial folding intermediates. In reflection, this is not too surprising because the higher concentration of KCl (140 mM) used in the earlier experiments is more likely to initially favor formation of the G4s, whereas the lower concentration of KCl (5 mM) used in the later experiments is more likely to favor initial formation of the hairpin loop. The similar footprinting pattern in the DMSO control and with RG260 suggests that either RG260 was not binding to the hTERT G4 or that the drug was binding without significant change in the DMS footprinting pattern. If the latter was the case, then this suggests that the binding of the two different drugs results in two different G4 folded products, and only in the case of GTC365 is the WT folding pattern changed in 5 mM KCl. At this point it became important to determine whether RG260 was binding to the hTERT G4, especially since the DMS footprinting did not confirm this.

T_m and CD results (Figure S1B and S1C) support the idea that RG260 is bound to the hTERT G4 and that the structure retains two G4s, but the arrangement of the intervening G-runs (5–10) was different in the two drug-bound species. Indeed G-runs 5–10 have distinctly different DMS footprinting patterns in the presence of either GTC365 or RG260 (Figure 1D). This leads us to suspect that G-runs 5–8 might form a similar heteroduplex in the control and in the presence of RG260 but that in the presence of GTC365 G-runs 7–10 were utilized to form the heteroduplex. Furthermore, this suggests that while the origin of the 5′-G4 is the same for both GTC365 and RG260 (G-runs 1–4), the 3′-G4 is formed from G-runs 9–12 for RG260 or from G-runs 6–7 and 11–12 for GTC365. A critical structural difference between GTC365 and RG260 is that GTC365 contains an acridine ring, the G-tetrad binding moiety, while RG260 lacks this recognition feature, which could be instrumental in defining the early intermediates in the different folding pathways. Since the DMSO footprinting in 5 mM KCl buffer did not reveal any significant difference between the control and the RG260-bound species, we explored the effect of different buffer conditions. In this regard we were particularly interested in the effect of calcium ion on the DMS footprinting pattern of the FL-G4 in the presence of GTC365 and RG260, because calcium ion is known to stabilize the G-triplex (Jiang et al., 2015), an intermediate that could form in an early step during the formation of the stem loop involving G-runs 5–8. Indeed, with a 100 mM CaCl₂ buffer we found that G-runs 6–8 showed a somewhat enhanced DMS cleavage in the presence of RG260, whereas there was no significant change of DMS footprinting pattern by GTC365 compared to the DMSO (Figure 1D). These results show that different buffers (KCl versus CaCl₂) result in different folding patterns, and in the 100 mM CaCl₂ buffer it is clearly demonstrated that RG260 indeed binds to the hTERT G4. The enhanced DMS cleavage in G-runs 6–8 in the presence of RG260 in 100 mM CaCl₂ buffer suggests this might be the binding site for the BPU moiety of the RG260 compound in contrast to G-runs 7–10 for the guanidino moiety of GTC365.

To evaluate whether this structural change of the hTERT G4 induced by RG260 affects hTERT promoter activity, a luciferase assay was conducted in human breast cancer MCF7 cells. RG260 reduced the luciferase activity in a dose-dependent manner by 25 % at 2 μM, indicating that RG260 inhibited hTERT promoter activity (Figure 1E). The effect on the telomerase activity by RG260 was also evaluated by a modified TRAP assay. As shown in Figure 1F, 72 h treatment of RG260 decreased the telomerase activity in a dose-dependent manner by 70 % at 60 nM.

Determination of the binding site for RG260 series compounds on the hTERT G4.

To support the role of the BPU structure of RG260 for conformational change in the hairpin stem-loop structure, we set out to design and synthesize new analogs (Figure 2A). Then, we determined the comparative effects of RG260 and other similar analogs on hTERT mRNA expression and hTERT promoter luciferase activity (Figure 2B and 2C). While RG1533 was inactive in both assays, RG1534, RG1601 and RG1603 were more potent than RG260. Likewise, in a cytotoxicity experiment using human prostate cancer (PCa) PC3-LN4 cells, RG1533 was also inactive (Figure 2D). We next designed a series of three different size oligomers for FRET experiments with the objective to determine the drug binding site for the RG260 series compounds. In descending order of size they represented the heteroduplex plus the 3′G4 (42-mer), the full heteroduplex (31-mer), and what we predicted would be the first intermediate in the folding pathway (18-mer) (Figure 2E). The FRET measurements on RG260 and this series of analogs (RG1533, RG1534, RG1601, and RG1603) are shown for each oligomer. The FRET measurement is the average of the full set of equilibrating species in each experiment. The results demonstrate that the 18-mer overall shows the most significant reduction in FRET measurement upon addition of the RG260 analogs, while the 31-mer intermediate and the 42-mer have the least reduction in FRET. The most biologically active RG260 analogs (i.e., RG1603) show the most potent effect on FRET with the shorter oligomers (31-mer and 18-mer). This set of data suggests that the drug-binding site for the RG260 series is within the 18-mer, most likely in proximity to the set of four G–G base pairs and the adjacent G–C base pair. To further define the binding site, we mutated G-run 6 in the 18-mer, which is positioned at the top of the hairpin. As expected, if this G-run is not involved directly in the binding of RG1603, drug binding to the mutant sequence still caused significant reduction of the FRET signal upon addition of RG1603, which was almost as much as the parent WT sequence (Figure 2F). Furthermore, using the WT 18-mer sequence, we demonstrated a dose dependency in reduction of FRET signal upon addition of RG1603 (Figure 2G). To further support the 18-mer as the drug-binding site for RG1603, we carried out DMS footprinting. In a comparison of the gel lanes of the treated versus non-treated samples, it was G-runs 5 and 7 that show the most change in extent of DMS cleavage, confirming that the G–G base pairing along with the adjacent G–C base pair is the site for drug occupancy (Figure 2H). Differential binding sites and pathways for folding of the G-quadruplexes for RG1603 and GTC365 are illustrated in the schematic diagram in Figure. S2.

Figure 2. Site-specific binding of small molecules with hTERT G4.

(A) A series of compounds based on the initial hit RG260.

(B, C) Relative hTERT mRNA level change (B) and promoter activity change (C) by RG260 analogs. MCF7 cells were treated with 250 nM of RG260 analogs for 48 h and then subjected to qRT-PCR. For luciferase assay, MCF7 cells were transfected with pGL3-hTERT construct and pRL-TK for 6 h and then treated with 500 nM of RG260 analogs for 18 h.

(D) Dose-response effects of PC3-LN4 cells to RG260, RG1534, RG1601, RG1603, and RG1533 (72 h). Data are shown as mean ± SD of four replicates.

(E) Identification of G-runs 5–7 with 18-mer as a binding site of RG260 analogs by FRET assay. FRET probes with FAM and TAMRA at each end of oligomers were annealed with compounds at an indicated concentration in a buffer containing 10 mM Na-cacodylate (pH 7.5) and 100 mM CaCl₂ before the measurement of fluorescence intensity of FAM. The relative fluorescence intensity was obtained compared to the DMSO sample.

(F, G) Determination of G-hairpin as a binding target of RG1603 (left) and dose-dependent FRET change (F). WT or a mutant with G-to-A of G-run 6 was annealed with indicated concentration of RG1603 in a buffer containing 10 mM Na-cacodylate (pH 7.5) and 100 mM CaCl₂ before the measurement of fluorescence intensity of FAM (G).

(H) Annealed mixture of FAM-labeled 18-mer and RG1603 in a buffer including 100 mM CaCl₂ was subjected to DMS footprinting. Data are shown as mean ± SD of four replicates. P-values (**P < 0.01, ***P < 0.001, ****P < 0.0001) were obtained by two-tailed t-test.

Overall, these results from DMS footprinting on the 18-mer, together with the FRET studies on the RG260 analogs, strongly suggest that RG260 only binds to the hairpin stem-loop, and this is sufficient to induce a folding of the 5–12 G4 to inhibit the transcription of hTERT. However, in the absence of direct evidence for the binding site, this conclusion must remain tentative. Although attempts were made by NMR to directly determine the binding site in the 18-mer hairpin, these were unsuccessful due to the weak binding of these compounds.

Treatment of prostate cancer cells with RG260 results in selective toxicity to cells expressing hTERT

Having identified RG260 as a downregulator of hTERT that works by a similar mechanism to GTC365, we sought to evaluate anticancer efficacy of RG260 focusing on PCa. The critical oncogenic function of TERT in PCa during tumorigenesis has been demonstrated by others using genetically engineered mouse knock-in models inducing expression of TERT in normal prostate epithelium (Hu et al., 2012). The RG260 activity was characterized in culture system using human PCa lines including DU145, PC-3 and LNCaP, and mouse prostate epithelial cancer (mPrEC) cells. Because the unique G4 DNA is present in the core promoter region of human TERT but not in mouse TERT, we hypothesized that RG260 possessed a selectivity toward human cancers. After exposure of RG260, morphological examination and MTT assay showed cell death across all human PCa lines tested, but not the mPrEC cells. These later cells were resistant even after prolonged incubation (Figure 3A). Similarly, the RWPE1 cell line which is immortalized but does not express elevated hTERT, was not growth inhibited by this compound (Figure 3B and 3C). Previously reported telomerase-targeting compounds, including BRACO-19 and BIBR1532 (Kleideiter et al., 2007), exerting their effects through telomeric G4s or the RNA component of the telomerase holoenzyme showed acute toxicity to hTERT-negative RWPE1 cells.

Figure 3. Selective cytotoxicity of RG260 to hTERT-expressing prostate cancer cells.

(A) Representative dose-response effects of human PCa cells (DU145, PC-3, and LNCaP) and mPrEC cells to RG260 (72 h treatment). Data are shown as mean ± SD of four replicates.

(B) The hTERT mRNA level in RWPE-1 cells and PC3-LN4 cells. qRT-PCR was conducted to obtain the relative mRNA level of hTERT to β-ACTIN. Additionally, HK2, COXII, and MYC mRNA levels were compared.

(C) Cell viability of RWPE1- and PC3-LN4 cells after 72 h treatment with RG260. Data are shown as mean ± SD of four replicates.

(D) Immunoblot analysis of cleavages of caspase-9 (Casp-9), capase-3 (Casp-3), and PARP-1 in PCa cells following treatment with RG260.

(E) Relative telomere length changes induced by RG260. Human PCa cells were treated with 0.1 μM of RG260 for 72 h. Genomic DNA was extracted and then subjected to qPCR. Ct values of telomeres were normalized to that of 36B4 (single copy gene) to obtain ΔCt (triplicates/group).

(F) Relative mRNA expression changes induced by RG260 was assessed by qPCR analysis. MCF7 cells were treated with 0, 0.13, 0.25, or 0.5 μM of RG260 for 24 h.

To characterize the anti-neoplastic effects of RG260 in both DU145 and LNCaP cells, RG260 treatment in a dose dependent manner caused the cleavage of caspase-9 (Asp330) producing a p37 fragment, that is known to amplify the apoptotic response (Figure 3D). Cleaved caspase-9 led to caspase-3 cleavage to trigger PARP-1 cleavage. Next, we measured telomere length in cells treated with RG260 using genomic DNA extracts by PCR amplification with oligonucleotide primers designed to hybridize to the telomere repeats (Cawthon, 2002). RG260 treatment resulted in significant reduction in telomere length in both DU145 and LNCaP (Figure 3E). This change, followed by hTERT downregulation, was not medicated by altering expression of c-MYC that binds to the E-box of the hTERT promoter and contains G4 tetrads (Figure 3F). These results indicate that RG260 binding to the heteroduplex loop not the conserved tetrads, induces selectivity toward cancer cells resulting in downregulation of hTERT expression and apoptotic cell death.

Synthetically derived analogs of RG260 interacting with the hairpin stem-loop in the hTERT G4 element induce apoptotic cell death selectively in cancer cells without harming normal epithelial cells.

To improve the molecular properties of RG260 and achieve a more bioavailable small molecule drug, RG260 analogs were synthesized with the following considerations in molecular weight (<500), cLogP (<4.0), PSA (<140 A2), and a number of rotatable bonds (<8). The design of analogs with varying groups at R1, R2, and R3 resulted in compounds RG1534, RG1601, and RG1603, and these analogs showed improved potency to inhibit hTERT promoter activity (Figure 2A–C). Because the MTT assay after 72 h exposure to these analogs indicated that RG1534 has superior effects on growth inhibition when compared to RG1601 and RG260 (Figure 2D) and caused a striking downregulation of hTERT expression and robust cleavage of caspase-9 and PARP-1, apoptotic markers, in PCa cells (Figure S3A), pharmacokinetics studies were carried out showing that both peak plasma concentration (C_max) and area under the plasma concentration curve (AUC) were proportional to dose and increased in a roughly linear fashion (Figure S3B). Using a subcutaneous mouse DU145 model, when RG1534 was dosed i.p. every other day at 2 and 5 mg/kg, RG1534 induced moderate tumor growth inhibition in a dose dependent manner (Figure S3C) without significant weight loss (<5% weight loss). Treatment with RG1534 at 10 mg/kg (i.p., QOD) induced complete response in all RG1534-treated animals. However, weight loss was higher than 10 % in all animals that received this dose (Figure S3D).

Since RG1534 produced weight loss in animals and RG1603 showed a dose-dependent anticancer activity in a broad range of PCa lines without killing normal prostate epithelial cells (Figure 4A), we next set out to evaluate RG1603 as an alternative lead candidate. Regardless of p53 status in PCa cells, RG1603 increased p21 expression, a cell cycle inhibitor, followed by downregulation of hTERT expression (Figure 4B). Since metastatic PCa harbor frequent mutations in p53 and loss of p53 causes genomic instability and chemoresistance, RG1603 might be an alternative option for treating metastatic PCa. Real-time qPCR analysis showed downregulation of hTERT mRNA expression by RG1603 in PCa cells (Figure 4C). To show whether the strong growth inhibitory effects of RG1603 was mediated through induction of cell death, cells were stained with ANNEXIN V/7-AAD and then subjected to flow cytometry analysis. In these PCa cells, cells positive for ANNEXIN V/7-AAD staining were increased with RG1603 treatment in a time dependent manner (Figure 4D). After RG1603 treatment, cleavage of both caspase-3 and PARP-1 were also detected by immunoblotting (Figure 4E). Consistent with this result, examination of cytosolic fraction from the cells treated with RG1603 showed an increased cytochrome c in cytosol (Figure 4F), a known component of the electron transfer chain in mitochondria. Once cytochrome c gets released to cytosol, it binds to a partner protein Apaf-1 and procaspase-9 to trigger apoptotic cell death through caspase cleavages (Li et al., 1997; Song et al., 2003). Collectively, our data indicated that RG1603 triggers apoptotic cell death selectively in cancer cells while sparing normal cells.

Figure 4. Selective cancer cell death induced by RG1603.

(A) Dose-response effects on proliferation of RG1603 after 72 h treatment of PCa cell lines and normal prostate epithelial RWPE-1 cells. Data are shown as mean ± SD of four replicates.

(B) Immunoblot analysis of hTERT, p53, and p21 in PCa cells. Cells were treated with RG1534 for 24 h and total lysates were subjected to immunoblotting.

(C) hTERT mRNA levels change in response to RG1603 doses in DU145 and RWPE-1 cells. qRT-PCR was conducted to obtain the relative mRNA of hTERT to β-ACTIN.

(D) Time-dependent increase of Annexin V binding by RG1603. Human prostate cancer DU145 and PC3-LN4 cells were treated with RG1603 for 24 and 48 h before Annexin V binding assay using flow cytometry.

(E) Immunoblot analysis of cleavages of caspase-3 (Casp-3) and PARP-1 in DU145 and LNCaP cells. Cells were treated with RG1603 for 24 h and total lysates were subjected to immunoblotting.

(F) Immunoblot analysis of cytochrome c release in cytosolic fraction of cells from the treatment of (E).

Molecular mechanism(s) by which RG1603 inhibits the survival and growth of prostate cancer cells.

In addition to dysfunction and shortening of telomeres and DNA damage, TERT knockout in mice causes mitochondrial dysfunction and elevated oxidative stress (Sahin et al., 2011). This result suggests that the action of this hTERT promoter inhibitor could also involve these specific pathways. RG1603 treatment markedly downregulated peroxisome proliferator-activated receptor-gamma coactivator alpha (PGC1α) expression, a transcription factor important for mitochondrial biogenesis, and expression of NQO1 and HMOX1 (Figure 5A). The latter two proteins are targets of the nuclear factor erythroid 2-related factor 2 (NRF2)-mediated antioxidant/electrophile-responsive element (ARE/ERE) signaling. This result together with increased cytochrome c release in RG1603-treated PCa cells suggests that RG1603 triggers catastrophic damage to mitochondria.

Figure 5. Mechanism of action responsible for RG1603-induced cell death.

(A, B) Immunoblot analysis of PGC1α, NQO1 and HMOX1 expression in Taxotere naïve LNCaP cells and Taxotere-resistant (Tx-R). Cells were either treated with RG1603 or 100 nM Taxotere for 24 h and total lysates were subjected to immunoblotting.

(C) DU145 cells were transfected with scramble control (si-C) or hTERT siRNAs. After 72 h, total lysates were subjected to immunoblot analysis of hTERT, NRF2, NQO1, and HMOX1.

(D) Representative images of crystal violet-stained Tx-R cells expressing either shRNA scramble control or shRNA NRF2. Cells were infected with lentiviruses containing scramble control or shNRF2 for 48 h and then further treated with or without 1 μM RG1603 for 72 h.

(E) An immunoblot with the oxidative stress defense Western blot cocktail. Tx-R cells were treated with DMSO or 1 μM RG1603 for 24 h and the resulting total lysates were subjected to immunoblotting. Relative changes of Catalase (60 kDa) Smooth Muscle Actin (42 kDa) Superoxide Dismutase 1 (16 kDa) Thioredoxin (12 kDa) are shown on the one blot. Morphological cell death induced by RG1603 is depicted in phase contrast images.

(F) Detection of superoxide anion and reactive oxygen species (ROS). DU145 cells were treated with DMSO or 0.3 μM RG1603 for 24 h and then stained with MitoSox Red or H2DCF-DA for 15 min prior to flow cytometry analysis.

(G) Representative fluorescence images of DCF (ROS) and nuclear (Hoechst 33342) staining in Taxotere naïve and resistant (Tx-R) cells at 0, 6, or 16 h treatment with 0.3 μM RG1603. Relative fluorescent intensity (RFI) was counted using a fluorescent microplate reader (excitation wavelength at 485 nm and emission wavelength at 535nm). Data are shown as mean ± SD of triplicates.

(H) (I) Immunoblot analysis of γ-H2AX (S139) and DFF45 expression in DU145, LNCaP, and Tx-R cells. Cells were treated with RG1603 as indicated for 24 h.

Taxotere is the only chemotherapeutic agent used to treat prostate cancer that extends life. However, every patient treated with this therapy eventually becomes resistant. To explore whether the ability of RG1603 to damage mitochondria and induce apoptosis can overcome specific chemo-resistance, LNCaP cells induced to become Taxotere resistant (Tx-R) by culturing in increased doses of this chemotherapeutic agent were studied (Domingo-Domenech et al., 2012; Vidal et al., 2015). Real-time qPCR analysis confirmed that expression of hTERT mRNA was increased in Tx-R cells (Figure S4) and these cells had elevated expression of CK44 and CK9, previously identified markers of Tx-R (Domingo-Domenech et al., 2012). Tx-R LNCaP cells have both HMOX1 and PGC1α expression increases (Figure 5B) (Vidal et al., 2015) when compared to the wild type control. As shown in Figure 5A, RG1603 treatment reversed the PGC1α increase found in Tx-R cells. Similarly, NQO1 and HMOX1 expression, two targets of the NRF2 transcription factor, were also decreased by RG1603 treatment. To determine whether NRF2 signaling is regulated by hTERT, hTERT expression in PCa cells was depleted by the transduction of small interference RNA (siRNA). Depletion of hTERT expression led to downregulation of NRF2 as well as targets of NRF2, NQO1 and HMOX1 (Figure 5C). NRF2 depletion inhibited both the survival and growth of Tx-R cells, suggesting the importance of this transcription factor to growth of these cells (Figure 5D). To examine whether killing of Tx-R cells by RG1603 is mediated by inhibition of cellular redox mechanisms, the levels of proteins controlling cellular redox, catalase, SOD1, and thioredoxin were examined by immunoblotting. Expression of all of these ROS scavengers was reduced in cells treated with RG1603 (Figure 5E). In a similar fashion to the results seen with the knock-down of hTERT, marked decreases in expression of ROS scavengers was induced by RG1603 treatment. This change would permit accumulation of oxidative stress in the RG1603 treated cells (Sahin et al., 2011). Consistent with the ability of this agent to decrease ROS scavenger proteins, after RG1603 treatment, MitoSox Red staining demonstrates that RG1603-mediated hTERT downregulation increases mitochondrial superoxide anion production (Figure 5F). Similarly, after RG1603 treatment H₂DCF-DA staining followed by flow analysis demonstrates elevated levels of total cellular ROS (Figure 5G and F). The marked induction of γ-H2AX expression by RG1603 treatment (Figure 5H), and the detection of cleavage products of DNA fragmentation factor 45 (DFF45/ICAD) (Figure 5I) implies that RG1603 is inducing cell death through inhibiting hTERT synthesis which in turn blocks mitochondrial function and leads to oxidative DNA damage.

In vivo evaluation of anticancer efficacy and safety of RG1603 in a tumor xenograft mouse model.

Following a single dose of RG1603 (25 mg/kg) by intraperitoneal administration, plasma samples were collected to measure pharmacokinetics. The plasma concentration of RG1603 was maintained at higher levels for 4 h and then gradually decreased by 12 h (Figure 6A). Pharmacokinetics studies demonstrated that administration of 25 mg/kg RG1603 to mice gave peak plasma drug concentrations of RG1603 of ~8.4 μM corresponding to the doses that effectively killed PCa cells in culture. When engrafted DU145 tumors were grown on the flank of SCID mice and reached size at ~150 mm³, mice were randomly assigned to two groups: vehicle control and RG1603. Mice were then treated with either vehicle or 25 mg/kg RG1603 (i.p., QOD) and tumor burden was evaluated for 21 days. RG1603 treatment caused significant growth inhibition of established PCa tumors in vivo xenografts (Figure 6B–D). Blood chemistry analyses including alkaline phosphatase, alanine aminotransferase, blood urea nitrogen, glucose and total protein not significantly different between vehicle, RG1603 treatment groups, and normal mice (Figure S5). Hematological profiles disclosed that the RG1603-treated group had a mild hypochromia (two mice) or polychromasia (three mice). To verify whether tumor suppression was the result of hTERT inhibition, at the end of treatment, tumor tissues were dissected from xenograft mice and then used for preparation of tumor extracts and immunohistochemistry (IHC) analysis. Immunoblot analysis of tumor extracts revealed that the striking decrease of hTERT expression in tumors of RG1603-treated mice (Figure 6E). Compared to the tumors of control mice, reduced NRF2 signaling was apparent in the tumors of mice treated with RG1603 as mirrored evidenced by lower expression of HMOX1 and SOD2 ROS scavengers. In treated animals, IHC analysis demonstrated increased levels of 8-hydroxydeoxyguanosine (8-OH-dG), a marker of oxidative damage, in mice receiving RG1603 treatment when compared with controls (Figure 6F). In tumors from RG1603-treated mice, IHC analysis with an antibody specific for cleaved caspase-3 demonstrated increased staining while decreased staining for tumor proliferation antigen Ki-67 was evident, consistent with the induction of apoptosis. Collectively, our results show that the hTERT inhibitor RG1603 is capable of inducing oxidative stress, DNA damage, and tumor apoptosis.

Figure 6. Pharmacokinetic analysis and efficacy of RG1603 in prostate tumor xenograft model.

(A) Pharmacokinetic profile in mouse plasma after single dose of 25 mg/kg RG1603 i.p.

(B) SCID mice implanted subcutaneously with DU145 cells were treated every other day with either vehicle or 25 mg/kg RG1603 by i.p. for 21 days to measure the effect on tumor growth. Data are shown as mean ± SEM. ***Significant difference in tumor volume at days 21 between RG1603-treated or vehicle-treated mice. Student t test P < 0.001.

(C) (D) Differences in tumor size and weight at the endpoint of (B) treatments (vehicle vs. RG1603). ***Significant difference in tumor weight at days 21 between RG1603-treated or vehicle-treated mice. Student t test P < 0.001.

(E) Immunoblot analysis of DU145 tumors from mice treated with 25 mg/kg RG1603 at the times shown.

(F) Representative immunohistochemical staining for H&E, Ki67, 8-ox-dG and cleaved caspase-3 of tumor tissues of (C).

DISCUSSION

Telomerase activation is a hallmark of cancer and a target for therapeutic intervention. Despite extensive studies and clinical trials, only moderate clinical success has been reported. Telomerase inhibition to date requires proliferating cells over multiple population doublings to induce tumor suppressive effects. However, long term telomerase inhibition may have undesirable effects on normal stem cells and progenitor cells which need telomerase for survival (Hiyama and Hiyama, 2007).

hTERT, the catalytic subunit of telomerase, synthesizes the new telomeric DNA from the RNA template hTR (Lai et al., 2001; Lingner et al., 1997). While hTR is ubiquitously expressed across cancer and normal tissues (Yan et al., 2001), hTERT expression is upregulated in the vast majority of cancers including cervical carcinomas, hepatocellular carcinoma, lung, breast carcinomas, and neuroblastomas but not in normal cells, suggesting that there would be minimal side effect on normal cells. Moreover, recent studies in glioblastoma, melanoma, and bladder cancer indicated that promoter mutations or unusual epigenetic changes can promote constitutive activation of hTERT (Borah et al., 2015; Horn et al., 2013; Killela et al., 2013). Somatic mutations such as cytosine to thymine or guanine to adenine mutations were shown to increase hTERT expression and telomerase activity (Horn et al., 2013). These mutations all occur in a G-rich region of the hTERT promoter which has previously been shown to form quadruplex DNA. These mutations abrogate G4 folding process and result in aberrant activation of hTERT (Kang et al., 2016), suggesting a negative function of G4 structure on hTERT promoter activity. In order to avoid the toxicity associated with binding in a nonspecific manner with the external tetrads in G4s, which are widespread through the human genome, we attempted to identify compounds that lack the tetrad-binding moiety but still retain the hairpin stem-loop binding properties. These compounds, exemplified by RG1603, still retain the ability to facilitate the cooperative folding process leading to the silencer element in the hTERT promoter. Thus, this approach differs radically from those previously used to target G4s because it does not rely on binding to the core G4 structure.

RG260 and the related series of compounds described in this contribution did not produce the same G4 folding pattern as GTC365 in 5 mM KCl buffer. In contrast to GTC365, this new series of compounds appears to produce a folding pattern that mimics that found naturally in the WT sequence. However, this depends upon the buffer conditions. In 5 mM KCl buffer DMS footprinting reveals that RG260 forms a similar folded form as in the control, while GTC365 forms a different folded structure. In contrast, in CaCl₂ buffer the GTC365 and control form the same folded structure, but RG260 forms a different folded species. The origin of the differently induced folding patterns produced by the GTC365 and RG260 series compounds is most probably due to their initial binding interactions with the G4-forming units or the hairpin loop, which contains multiple G-runs that can form intermediate species such as G-triplexes (Jiang et al., 2015). In the 5 mM KCl buffer, while RG260 binds to the top of the hairpin loop, which is the earliest intermediate in the folding process and thus appears to mimic the natural folding pathway, GTC365 may bind at an artificially induced step, which may involve preferential stabilization of the G4s due to the presence of the acridine moiety of GTC365. In CaCl₂ buffer the intermediates in the folding pathway, such as G-triplexes, are stabilized in the loop region. RG260 more favorably binds to these early intermediates in the loop region, whereas the presence of GTC365 favors the early folding of the 3′-G4 and its stabilization by the acridine moiety. The difference between the two series of compounds is the presence of an acridine G-tetrad binding moiety present solely in GTC365. The presence of this moiety may facilitate the formation of the second G4 using G-runs 5 and 6 alongside 11 and 12 prior to allowing the hairpin to form from G-runs 7–10. The transcription activation of hTERT produced by the somatic mutations at –124 and –146 (G to A) can be explained by the generation of new binding sites for GABP at the duplex level (Mancini et al., 2018). Furthermore, both these mutations and those at −124/125 and −138/139 (GG to AA) would also be expected to compromise the naturally occurring folding pathway involving such intermediates as shown in Figure 2E, which would also lead to overexpression of hTERT. It is intriguing and perhaps not coincidental that the pairs of mutations at –124/125 and –138/139 are directly opposite each other in the hairpin, in contrast to the less obvious mutation pattern found in the hTERT G4 folding pattern induced by GTC365. We believe that we have uncovered a DNA folding process that mimics RNA riboswitches, with the RGT365 and RG260 series compounds acting as small-molecule chaperones at the initial stages of the folding process (Dethoff et al., 2012). In this study we found that RG1603 decreased hTERT mRNA expression and induced apoptotic cell death prior to telomere shortening, which is consistent with observation by others that hTERT loss triggers dysfunctional DNA capping. In contrast to robust cell death in PCa cells, normal human prostate epithelial RWPE-1 cells deficient with hTERT expression (Huang et al., 2008) were insensitive to RG1603. Moreover, no growth inhibitory activity of RG1603 was observed in mouse tumor cells which have a lack of G4 stem-loop in mouse TERT promoter. In contrast to previous generations of G4 molecules targeting the telomeric G4s, which exhibited acute toxicities regardless of cell type, the improved specificity of RG1603 on the target, while sparing normal cells, is highly desirable to the development of clinically useful anticancer therapeutics.

Mechanistically, it is not well understood how enhancing the hTERT G4 DNA cooperative folding process triggers apoptotic cell death. RNAi-mediated hTERT knockdown resulted in suppression of tumor cell growth and induced a variable degree of programmed cell death prior to telomere shortening (Oliver et al., 2017; Shen et al., 2008). These results suggest that the hTERT enzyme has telomerase independent activity. hTERT was previously shown to regulate pro-survival Bcl-2 expression preventing mitochondrial apoptosis (Del Bufalo et al., 2005). Mitochondrial localization of hTERT has also been associated with oxidative stress (Singhapol et al., 2013). To understand how hTERT promoter control regulates mitochondrial function we examined two transcription factors, NRF2 and PGC1α both important to mitochondrial function and implicated in oxidative stress regulation (Dinkova-Kostova and Abramov, 2015; Valle et al., 2005). We showed that RG1603 decreased NRF2 expression and the expression of its transcription target HMOX1. siRNA-mediated depletion of hTERT expression also resulted in inhibition of NRF2 and HMOX1 expression. In our study, DNA damage as shown by γ-H2AX induction and DNA fragmentation by RG1603 coincided with PGC1α inhibition. Decreases in the level of NRF2 and PGC1α by RG1603 resulted in generation of ROS leading to mitochondrial damage and cell death. Similarly, a previous study found that hTERT-induced cisplatin-resistance in osteosarcoma cells was associated with an increased glutathione antioxidant defenses that functioned to reduce ROS level (Indran et al., 2011; Zhang et al., 2017b).

In conclusion, small molecule targeting of the cooperative folding process in the hTERT G4 promoter element produces telomerase independent events which impairs mitochondrial processes resulting in oxidative stress and inhibition of prostate cancer cell growth.

STAR METHODS

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Cell lines

The human prostate cancer cell lines LNCaP, PC-3, DU145 (ATCC), and PC-3LN4 were cultured as previously described (Pettaway et al., 1996; Song and Kraft, 2012; Song et al., 2016). RWPE1 cells (ATCC) were maintained in keratinocyte medium supplemented with 5 ng/mL human recombinant epidermal growth factor and 0.05 mg/mL bovine pituitary extract (Invitrogen). Mouse prostate epithelial cells were grown in DMEM medium supplemented with 2.5% charcoal stripped FCS, 5 μg/mL of insulin/transferring/selenium, 10 μg/mL of bovine pituitary extract, 10 μg/mL of epidermal growth factor, 1 μg/mL of cholera toxin as described previously (Sun et al., 2011). HEK293T cells were cultured in 10% FBS-containing DMEM. MCF7 cells were cultured in RPMI-1640 with 10% FBS and 1% penicillin/streptomycin. All cell lines were maintained at 37 °C in 5 % CO₂ and were authenticated by short tandem repeat DNA profiling performed by the University of Arizona Genetics Core Facility. The cell lines were routinely tested for mycoplasma and used for fewer than 50 passages.

Animal studies

All in vivo studies were approved by and conducted in accordance with the guidelines of the Institutional Animal Care and Use Committees at the University of Arizona Cancer Center. Male SCID immune-deficient mice (C.B-Igh-1^b/IcrTac-Prkdc^scid) were purchased from Taconic Laboratories and maintained by the Experimental Mouse Shared Resource in an ABSL-2 immunodeficient animal housing facility at University of Arizona. Mice of 8 weeks old were randomly assigned to treatment groups. For the pharmacokinetics study, male FVB/NJ mice were purchased from The Jackson Laboratory (stock number #001800).

METHOD DETAILS

FRET assay

FRET assay for compound screening and FRET probe of the 5−12 G4 were reported previously (Kang et al., 2016). FRET probes with FAM (Ex. 490 nm/Em. 520 nm) and TAMRA (Ex. 560 nm/Em. 580 nm) at each end were synthesized and HPLC purified by MGW Operon Inc (sequences are provided in Key Resources Table). The WT probe (50 nM) was annealed in a buffer containing 10 mM Tris-HCl (pH 7.5) and 5 mM KCl by heating at 95 °C for 5 min and slowly cooling to room temperature. Compounds (50 μM) and probe were incubated for 1 h at room temperature. The same volume of DMSO served as a control. For K_d value determination, the WT, G124/125A, and G146A probes were annealed by heating for 5 min at 95 °C and subsequent slow cooling to room temperature, and then several concentrations of the compound were treated for 1 h at room temperature. Dose-dependent fluorescence intensity at 520 nm was measured by a microplate reader (BioTek Synergy HT). The data were corrected with the blank signal of buffer and compound. The relative fluorescence intensity compared to DMSO was used for binding curve fitting to determine the K_d value using GraphPad Prism software.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
See Table S1.
Chemicals, Peptides, and Recombinant Proteins
MitoSox Red	Invitrogen	Cat# M36008
CM-H2DCF-DA	Invitrogen	Cat# C6827
Thiazolyl blue tetrazolium bromide	Sigma-Aldrich	Cat# M5655
RG260 (NSC654260)	NCI	https://wiki.nci.nih.gov/display/NCIDTPdata/Compound+Sets
RG1534	This paper	N/A
RG1603	This paper	N/A
RG1601	This paper	N/A
RG1533	This paper	N/A
Experimental Models: Cell Lines
Human Prostate cancer PC-3	ATCC	Cat# ATCC CRL-1435
Human Prostate cancer DU145	ATCC	Cat# ATCC HTB81
Human Prostate cancer LNCaP	ATCC	Cat# ATCC CRL-1740
Human Prostate cancer PC3-LN4	The University of Texas M. D. Anderson Cancer Center (Pettaway et al., 1996)	N/A
Human prostate epithelial RWPE1	ATCC	Cat# ATCC-CRL-11609
Human breast cancer MCF7	ATCC	Cat# ATCC-HTB-22
HEK293T	Dharmacon	Cat# HCL4517
Mouse prostate epithelial cancer cells	Roswell Park Cancer Center	N/A
Experimental Models: Organisms/Strains
Tx-R LNCaP cells	This paper	N/A
hTERT luciferase expressing MCF7 cells	(Kang et al., 2016)	N/A
Oligonucleotides
Oligonucleotides used in DMS footprinting and FRET (see Table S2)
Primers for real time PCR (see Table S3)
ON-TARGETplus Human TERT (7015) siRNA	Dharmacon	Cat# L-003547-00-0005
ON-TARGETplus Non-targeting Pool siRNA	Dharmacon	Cat# D-001810-10-05
Recombinant DNA
pLKO human shRNA NFE2L2	Dharmacon	Cat# RHS3979-201739828
TRC Lentiviral Non-targeting shRNA Control	Dharmacon	Cat. #RHS6848
VSV-G	Addgene	Cat# 8454
psPAX2	Addgene	Cat# 12260
Software and Algorithms
Microsoft Office 365 Excel	Microsoft	https://www.office.com/
GraphPad Prism	GraphPad Software	https://www.graphpad.com/scientific-software/prism/

DMS footprinting

FAM-labeled oligomers with full-length and 18-nt were synthesized by MGW Operon Inc. and PAGE purified: FRET 5–12 G4, [FAM]-AGGGGGCTGGGCCGGGGACCCGGGAGGGGTCGGGACGGGGCGGGGT-[TAMRA]; FRET WT 31nt (G-runs 5–8), [FAM]-GCCCGGAGGGGGCTGGGCCGGGGACCCGGGA-[TAMRA]; FRET WT 18nt (G-runs 5–7), [FAM]-AGGGGGCTGGGCCGGGGA-[TAMRA]; FRET G131–133A 18nt, [FAM]-AGGGGGCTAAACCGGGGA-[TAMRA]; FAM-Full length, [6-FAM] AGGGGAGGGGCTGGGAGGGCCCGGAGGGGGCTGGGCCGGGGACCCGGGAGGGGTCGGGACGGGGCGGGGTTTTTTTT; FAM-18nt, [FAM]-AGGGGGCTGGGCCGGGGATTTTTTTT. These oligomers (250 nM) and 5 μmol/L of compounds were dissolved in a buffer containing 10 mmol/L Na-cacodylate (pH 7.5), and 5 mmol/L KCl or 100 mmol/L CaCl₂. The mixtures were annealed by heating at 95 °C for 30 s and then cooled down for 1–2 min at room temperature. For the DMS reaction, annealed samples were incubated with 2 μg of salmon sperm DNA (Sigma, D1626) and 5% DMS in 50% ethanol for 8 min for full-length and 5 min for 18-nt, respectively, at room temperature. The reaction was stopped by 10% of β-mercaptoethanol and then subjected to ethanol precipitation and cleavage by 10% piperidine with incubation at 93 °C for 12 min. Cleaved product was washed twice by water and then separated by 15% denaturing PAGE with 7 M urea. Fluorescent bands were scanned by Bio-Rad PharosFX™ Plus.

Luciferase assay

The pGL3 luciferase constructs used in a previous publication (Kang et al., 2016) were used for this study. MCF7 cells in a 24-well plate were incubated with serum-free RPMI-1640 and then transfected with 200 ng of pGL3 construct and 5 ng of pRL-TK by FuGENE® HD or Lipofectamine 3000 and then incubated for 6 h. After replacing media by RPMI-1640 with 10% FBS, cells were treated with 0.1% DMSO or 500 nmol/L of compounds. After 18 h of incubation, cells were lysed by passive lysis buffer and subjected to dual-luciferase assay (Promega, #E1910) using an FB12 luminometer or GloMax® 20/20 Luminometer (Promega). Relative ratio of firefly to renilla luciferase activity compared to the DMSO was obtained. Each experiment was performed in quadruplicate.

Circular Dichroism (CD) analysis

The WT 5–12 G4-forming oligomer was synthesized and HPSF-purified by MWG Operon, Inc. The concentration of oligomers dissolved in deionized water was determined by the Lambert–Beer law using molecular extinction coefficient, 463,850 L (mol*cm). For CD analysis, the oligomers (5 μM) in a buffer containing 10 mM Tris-HCl (pH 6.8 or pH 7.5) with 5 mM KCl were heated at 95 °C for 5 min and then cooled at room temperature. After 1–2 h of incubation of annealed oligomers with DMSO or RG260, samples were subjected to CD spectra and melting analysis using a Jasco 810 spectropolarimeter with parameters of 2 nm band width and a rate of 1.6 °C/min. The spectra of oligomers were corrected by subtraction of the spectra of buffer with DMSO or RG260. T_m was determined by sigmoidal curve fitting of a melting curve using GraphPad Prism software.

Gene transduction

Cells were transfected with SMARTpool: ON-TARGETplus TERT siRNA (Dharmacon) using lipofectamine 3000 (Thermo Fisher Scientific) as previously described (Song et al., 2015; Song et al., 2016). To knock down expression of NRF2, short hairpin RNAs against human NRF2 (TRC gene set 7555) and TRC lentiviral non-targeting shRNA control were purchased from Dharmacon. Lentiviruses were produced in 293T cells coexpressing the packaging vectors (pPAX2 and VSVG), concentrated by ultracentrifugation (20,000 g for 2 h at 4 °C), and resuspended with culture media for 48 h infection.

Cell proliferation and apoptosis assays

Cells were seeded in 96-well plates and treated with drugs as indicated. At the end of treatment, the medium was removed gently and the cells were washed once with 200 μl of PBS, then 100 μl of medium containing MTT assay solution (1 mg/mL thiazolyl blue tetrazolium bromide, Sigma Cat# M5655) was added to each well. The plates were placed in a water-jacketed incubator at 37°C for 2 h and then MTT containing medium was removed. The color development was monitored by addition of 100 μl of DMSO to each well prior to absorbance reading at 490 nm using the microplate reader (Spectra Max 190, Molecular Devices). Each drug treatment was performed in quadruplicate and verified in three independent experiments. Apoptotic cells were quantitated by flow cytometry using PE Annexin V Apoptosis Detection kit 1 (BD Pharmingen, Catalog # 559763).

Immunoblotting

Cells were lysed in RIPA buffer (50 mmol/L Tris-HCl pH 7.4, 150 mmol/L NaCl, 0.5 % sodium deoxycholate, 0.1 % SDS, 1% NP-40, 5 mmol/L EDTA) supplemented with complete protease/phosphatase inhibitor cocktail (Cell Signaling Cat# 5872S). After centrifugation (15,000 g for 15 min at 4 °C), aliquots of total lystates were subjected to SDS-PAGE as described previously (Song et al., 2015). The antibodies were listed in Table S1.

Immunofluorescence and immunohistochemistry

Staining with MitoSox Red and H₂DCF-DA followed by flow cytometry was as previously described (Song et al., 2015). Phase contrast and fluorescence images were taken with EVOS FL Auto (Bio-Rad). Immunohistochemical staining of tumors tissues were performed with anti-8-ox-dG (Trevigen, Cat # 4354-MC-050), cleaved caapse-3 (Cell Signaling, Cat # 9661) and Ki67 (Leica, Cat # NCL-L-Ki67-MM1) antibodies. Hematoxylin 560 MX (Leica Biosystems, Cat # 3801575) and Eosin Phloxine 515 (Leica Biosystems, Cat# 3801606) were used for H&E staining.

Measurements of telomere length and telomerase enzyme activity

To measure telomere length in cells, genomic DNA (gDNA) was extracted with DNA extraction kit (Qiagen). A pair of Tel1 (GGTTTTTGAGGGTGAGGGTGAGGGTGAGGGTGAGGGT) and Tel2 (TCCCGACTATCCCTATCCCTATCCCTATCCCTATCCCTA) primers was used for amplification of the telomere region and a pair of 36B4F (CAGCAAGTGGGAAGGTGTAATCC) and 36B4R (CCCATTCTATCATCAACGGGTACAA) primers for amplification of a single-copy gene to normalize data. The PCR was initiated at 95 °C for 3 min and then 27 cycles at 95 °C for 3 s and 60 °C for 2 min. The fluorescence signal at 60 °C was acquired. Triplicate data were averaged and normalized to 36B4 to obtain ΔCt. The relative telomere length was determined compared to the DMSO. Telomerase enzyme activity was assessed by TRAP assay as previously reported (Kang et al., 2016).

Real-time PCR analysis

Total RNA was extracted from cells by using Qiagen QIAshredder and RNeasy Mini kit. Equal amounts of total RNA (1 μg RNA) was subjected to first-strand cDNA synthesis using iScript cDNA synthesis kit according to the manufacturer’s protocol. Real-time PCR reactions with Bio-Rad SsoAdvanced Universal SYBR Green Supermix were performed using a Bio-Rad CFX96 Touch System. Samples were assayed in triplicate and the data were normalized to 18S mRNA levels. The primer sets for TERT, NRF2, HK2, MYC, COX2, and β-ACTIN genes were purchased from Sigma-Aldrich to measure gene expression (see Table S3 for primer sequences).

Synthesis of RG260 analogs

RG260 is a benzoylphenyl urea derivative in which two aromatic rings are connected by the urea functionality. RG1534, RG1601, and RG1603 are structural analogs of RG260 (Supplemental Scheme 1). Oxalyl chloride (0.5 g) was added to 2-nitrobenzamide (0.2 g) in toluene at 0 °C. The solution was warmed to room temperature, then refluxed with stirring for 24 h. The solvent was evaporated and product 2-nitrobenzoyl isocyanate was used directly without further purification for the next reaction. A solution of 2,5-dichloropyrimidine (382 mg), 4-amino-2-methylphenol (319 mg), and K₂CO₃ (718 mg) in dry DMSO (20 mL) was stirred at 120 °C for 2.5 h. After cooling to room temperature, the reaction mixture was poured into water and extracted with ethyl acetate. The organic layer was washed with water, saturated brine and then dried. The organic solvent was evaporated to give a residue that was purified using silica gel column chromatography (ethyl acetate-hexane 1:3) to give product 190 mg of 4-((5-chloropyrimidin-2-yl)oxy)-3-methylaniline. ¹H NMR (400 MHz, Chloroform-d) δ 8.46 (s, 2H), 6.87 (d, J = 8.4 Hz, 1H), 6.60 – 6.49 (m, 2H), 3.66 (s, 1H), 2.08 (s, 3H). A solution of 2-nitrobenzoyl isocyanate (115 mg) in 15 mL dry dioxane was added dropwise to a solution of 4-((5-chloropyrimidin-2-yl)oxy)-3-methylaniline (141 mg) in dry 1,4-dioxane with stirring at room temperature. The reaction mixture was stirred for 18 h and then diluted with water. The precipitated solid was collected by filtration and washed with water. The solid was dissolved in ethyl acetate, and the organic layer was washed with (3 × 30 mL) water, dried and concentrated to give 180 mg of N-((4-((5-chloropyrimidin-2-yl)oxy)-3-methylphenyl)carbamoyl)-2-nitrobenzamide. ¹H NMR (400 MHz, DMSO-d6) δ 11.29 (s, 1H), 10.22 (s, 1H), 8.80 (s, 2H), 8.22 (ddd, J = 8.1, 1.2, 0.6 Hz, 1H), 7.94 – 7.88 (m, 1H), 7.83 – 7.76 (m, 2H), 7.47 (d, J = 19.8 Hz, 2H), 7.13 (d, J = 8.6 Hz, 1H), 2.08 (s, 3H). Iron powder (160 mg) was added in portions to a mixture of N-((4-((5-chloropyrimidin-2-yl)oxy)-3-methylphenyl)carbamoyl)-2-nitrobenzamide (256 mg) and ammonium chloride (335 mg) in ethanol at 80 °C. The reaction mixture was refluxed for 30 min and then cooled to room temperature and diluted with water. The precipitated solid was collected by filtration. The solid was dissolved in excess ethyl acetate and filtered. The filtrate was dried and concentrated to give a residue that was purified by column chromatography (ethyl acetate: hexane 2:3) to give 110 mg compound RG1534, 2-amino-N-((4-((5-chloropyrimidin-2-yl)oxy)-3-methylphenyl)carbamoyl)benzamide. Nuclear magnetic resonance (NMR) data: ¹H NMR (400 MHz, DMSO-d6) δ 10.77 (s, 1H), 10.57 (s, 1H), 8.75 (s, 2H), 7.72 (dd, J = 8.1, 1.5 Hz, 1H), 7.53 – 7.45 (m, 2H), 7.26 (dt, J = 8.4, 1.5 Hz, 1H), 7.16 – 7.08 (m, 1H), 6.79 (dd, J = 8.4, 1.2 Hz, 1H), 6.66 – 6.50 (m, 3H), 2.09 (s, 3H). ¹³C NMR (101 MHz, DMSO) δ 170.98, 163.48, 158.71, 151.80, 151.34, 147.42, 135.72, 134.18, 130.93, 129.84, 125.16, 122.87, 122.74, 119.21, 117.25, 115.17, 112.24, 16.43. High-Performance Liquid Chromatography (HPLC) Purity = 97.246 % (retention time: 1.790 min). Column: Zorbax SB C18, 4.6 ×150 mm, 3.5 u, Mobile phase: 35/65/0.25, Acetonitrile: water: Acetic acid, Flow rate: 0.5 mL/min. High Resolution Mass Spectra (HRMS): Found = 398.1014 (MH⁺) (Theoretically = 398.1020).

In vivo studies

To investigate the pharmacokinetics of RG1603 after a single oral dose of 25 mg/kg given by intraperitoneally (for RG1534, 10 and 50 mg/kg i.p.), at each time point the plasma concentrations were measured in three FVB mice using triplicate determinations. Blood samples were collected from the mice via the cheek pouch or retro-orbital bleeding. Total plasma concentrations of RG1603 were determined by high performance liquid chromatography/tandem mass spectrometry method. RG1603 was extracted from plasma by protein precipitation with acetonitrile. After centrifugation, the supernatant was injected onto the HPLC-MS system for analysis. HPLC separation was achieved using a C-18 column and a gradient mobile phase of 10 mM ammonium acetate and acetonitrile. Detection was performed on a TSQ Quantum Ultra Triple Quadrupole system operating in negative polarity utilizing atmospheric pressure chemical ionization. The instrument is operated in multiple reaction monitoring mode, monitoring the fragment mass of 117.01 from the parent mass of 395.9 at a collision energy of 44eV with argon as the collision gas. For xenograft studies, DU145 cells (5 × 10⁶) were implanted with Matrigel subcutaneously into the left flank of mice (male SCID, 8 weeks). When the tumor size reached ~150 to 200 mm³, mice were randomly assigned and treated every other day with vehicle (0.5% hydroxypropyl methylcellulose), RG1603 by intraperitoneal injection. Tumor volume was measured twice per week with calipers and calculated as tumor volume = (length × width²) × 0.5. The unpaired Student t test was used to evaluate the significance of differences observed between groups, accepting P < 0.05 as a threshold of significance. All animal studies were approved by the Institutional Animal Care and Use Committee at the University of Arizona.

Quantification and statistical analysis

Each drug treatment was performed in quadruplicate and verified in three independent experiments unless stated otherwise. Graphs were generated using Microsoft Excel and GraphPad Prism. The unpaired Student t test was used to evaluate the significance of differences observed between groups, accepting P < 0.05 as a threshold of significance. Data analyses were performed using the Prism software (GraphPad).

SIGNIFICANCE.

We have identified small molecules targeting the unique hTERT G4 stem-loop but not tetrad DNAs, which increases the selectivity toward hTERT DNA. These drugs act as pharmacological chaperones facilitating the cooperative folding of the G4 silencer element rather than acting by stabilizing the core G4 structure. The ability of RG1603 to selectively killing human cancer cells while sparing normal epithelial cells demonstrates for the first time in vivo anticancer efficacy of an hTERT G4–targeting small molecule. Importantly, using these drugs, we uncovered the role of hTERT in NRF2 and PGC1α regulation of redox signaling and DNA damage response. The potential use of hTERT inhibitory following Tx-R of prostate cancer may provide an alternate therapeutic option to overcome tumor resistance.

Abbreviations used:

G4

guanine quadruplex (G-quadruplex)

HMOX1

heme oxygenase 1

hTERT

human telomerase reverse transcriptase

hTR

telomerase RNA component

i.p.

intraperitoneal

mPrEC

mouse prostate epithelial cancer

NQO1

NAD(P)H:quinone oxidoreductase 1

NRF2

nuclear factor erythroid 2-related factor 2

PCa

prostate cancer

PG

PicoGreen

PGC1α

peroxisome proliferator-activated receptor gamma coactivator 1 alpha

QOD

every other day

ROS

reactive oxygen species

SOD

superoxide dismutase

Tx-R

Taxotere resistance/taxotere-resistant