Targeting RNA Polymerase I with Hernandonine Inhibits Ribosomal RNA Synthesis and Tumor Cell Growth

Ribosome biogenesis is the complex and profoundly organized cell process that prompts the creation of ribosomes and is strictly reliant on the activity of the RNA polymerase I (RNA Pol. I) transcriptional machinery (1). RNA Pol. I-dependent transcription of 47S rDNA progressively reacts to growth signaling and cellular stresses to build up a plenitude of ribosomal RNAs and directly manages cellular protein translational limit and along these lines proliferative growth rate (2). rRNA is a major component of the ribosome, and, therefore, its synthesis is a requirement for carcinogenesis (2–4). rDNA transcription happens in nucleoli, which are formed around actively transcribed rDNA repeats in early G₁ before being dismantled in mitosis when rDNA transcription is stopped. Accordingly, raised rDNA transcription by RNA Pol. I is a typical phenomenon of cancer (2, 5), and the enlarged nucleoli, which are generated from a morphologic consequence known as nucleolar hypertrophy and are characteristic of hyperactivated rDNA transcription, have been used as markers of aggressive tumors (6).

The nucleolus is isolated into particular compartments, to be specific, the fibrillar center, dense fibrillar component and granular component with characterized capacities that help RNA Pol. I transcription, rRNA processing and maturation of the rRNAs, and ribosome assembly, respectively (7, 8). Alteration of ribosome biosynthesis can occur at various levels and incorporate changed spatial dispersion of nucleolar proteins and downregulation of RNA Pol. I transcription, precursor rRNA (pre-rRNA) processing and transport, or changes in ribosome structure (8). Stresses that lead to blocked RNA Pol. I transcription cause fast and dynamic revamping of the nucleolar structures and proteins (9, 10). For instance, proteins of the fibrillar center and the dense fibrillar component relocalize to nucleolar periphery forming nucleolar cap structures, while granular component proteins commonly translocate to the nucleoplasm (11). Specifically, nucleolar localization controls the function of key oncogenes and tumor suppressors, for example, ARF and MDM2, the two of which are basic for the regulation of p53 (TP53; ref. 12). Accordingly, RNA Pol. I–dependent transcription and nucleolar integrity are significant determinants for the various processes required for the expansion of malignancy cells and give convincing thinking to target of RNA Pol. I transcription as a therapeutic strategy.

Most chemotherapeutic agents initiate cytotoxicity in cancer cells by damaging DNA, inhibiting DNA synthesis, and disturbing the mitotic process, even as the cytotoxic effects influence the multiplying cells of normal tissues, in this way building up strict pharmacologic limit on the utilization of these anticancer medications (13, 14). More than 3,000 traditional Chinese plants have been reported to treat cancer; therefore, traditional Chinese medicine is a valuable and alternative resource for identifying novel anticancer agents (14, 15). In this study, we found that hernandonine (C₁₈H₉NO₅), an oxoaporphine-related compound isolated from Hernandia nymphaeifolia (16, 17), represses RNA Pol. I transcription, initiates the degradation of RNA Pol. I catalytic subunit RPA194, and shows intense anticancer action. We further demonstrate that the molecule acts in a p53-dependent or p53-independent manner and that RNA Pol. I inhibition happens upstream of the activated p53. This investigation characterizes new and significant molecules for focusing on RNA Pol. I transcription.

Human colorectal carcinoma cells HCT116-p53^+/+ and HCT116-p53^−/− were grown in McCoy 5a medium modified supplemented with 10% FBS. Human cervical carcinoma cells (HeLa) and human lung adenocarcinoma cells (A549) (ATCC) were grown in DMEM supplemented with 10% FBS and in RPMI1640 medium supplemented with 10% FBS, respectively. Human normal lung fibroblast (IMR-90) were grown in modified Eagle medium (MEM) supplemented with 10% FBS. Human immortalized bronchial epithelial cells (Beas-2B) were grown in RPMI1640 medium supplemented with 10% FBS. These cell lines shown above were purchased from ATCC. Human oral squamous carcinoma cells (SAS, OECM-1 cells) and human normal oral keratinocyte (NOK) were provided by Dr. Kuo-Wei Chang at the Institute of Oral Biology, School of Dentistry, National Yang-Ming University (Taipei, Taiwan), and were authenticated by short tandem repeat analysis (18). SAS cells were grown in DMEM supplemented with 10% FBS. OECM-1 cells were grown in RPMI1640 medium supplemented with 10% FBS. Human normal oral keratinocyte (NOK) were grown in keratinocyte-SFM (KFSM, 1×, Invitrogen) medium supplemented with human recombinant 0.2 ng/mL EGF 1–53 and 25 μg/mL bovine pituitary extract (Invitrogen).

The cytotoxicity of hernandonine were determined using a modified 3-(4,5- dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium (MTT; Sigma) assay and a lactate dehydrogenase (LDH) leakage assay, as described previously (19). Briefly, for MTT assay, cells (5 × 10³/well) seeded in 96-well plates overnight were treated with hernandonine (0–50 μg/mL, 24 hours). The resulting formazan dissolved with DMSO were measured at 570 nm and results were presented as the percentage of the control values. For LDH leakage assay, cells (2.5 × 10⁴/well) seeded in 24-well plates overnight were treated with hernandonine (0–50 μg/mL, 24 hours). LDH activities were measured using the LDH Detection Kit (Sigma) as described in the manufacturer's protocol. The extent of cellular damage was determined dependent on the percent LDH activity in the supernatants in respect to that in the cell lysates. All of these experiments were performed in triplicate and were repeated independently at least three times.

The procedure of immunofluorescence assay was described previously (19, 20). Briefly, cells were seeded, fixed, and permeabilization with 0.5% (v/v) Triton X-100. After blocking in 0.3% (w/v) BSA, and 0.05% (v/v) Tween-20 in PBS at room temperature for 1 hour, cells were incubated with nucleolin (1:500, Cell Signaling Technology, #14574), B23 (1:250, Zymed), UBF (1:100, Santa Cruz Biotechnology, sc-9131), or γH2AX (Ser 139, 1:100, Millipore, #299783) at 4°C for overnight, followed by incubation of 80 ng/mL DAPI at room temperature for 1 hour. The suitable fluorophore-conjugated secondary antibodies (1:200, FITC or Rhodamine; Molecular Probes) were utilized and immunofluorescent images were observed with a fluorescence laser-scanning confocal microscope (Olympus FV10i).

The procedure of immunoprecipitation (IP) assay was followed according to manufacturer's instructions (Dynabeads protein G, Invitrogen), as described previously (19). Briefly, cells were washed and scraped in PBS, then lysed for IP. Briefly, MDM2 (2 μg, ab16895, Abcam) or IgG (Santa Cruz Biotechnology Inc) antibody was mixed with dynabeads protein G (100 μL) on rotating platform for 2 hours at 4°C. After removing antibody solution using the Dynal magnet system, protein samples (1 mg in 300 μL of IP lysis buffer) were added and incubated overnight on rotating platform at 4°C. Beads were then washed and antibodies/protein complexes were eluted with 1× SDS sample buffer followed by immunoblotting analysis.

The procedure of immunoblotting analysis was done as described previously (21). Briefly, protein samples (30 μg) were run on 8–10% SDS-PAGE and then transferred onto a polyvinylidene difluoride (Bio-Rad) at 90 V for 120 minutes. Blots were probed with primary antibodies overnight at 4°C. Primary antibodies included: B23 (1:250, Zymed), caspase-3 (1:1,000, Cell Signaling Technology, # 9662), caspase-9 (1:1,000, Cell Signaling Technology, # 9502), E2F-1 (1:1,000, Cell Signaling Technology, #3742), cyclin B1 (1:1,000, Santa Cruz Biotechnology, #sc-752), p53 (1:1,000, Calbiochem), Cdc25C (1:1,000, Cell Signaling Technology, #4688), PARP (1:1,000, Cell Signaling Technology, #9542), nucleolin (1:500, Cell Signaling Technology, #14574), RNA Pol. I (RPA194, 1:100, Santa Cruz Biotechnology, sc-28714) UBF (1:100, Santa Cruz Biotechnology, sc-9131). After primary antibody incubation, the membrane was washed and incubated with a horseradish peroxidase–conjugated secondary IgG (1:3,000; Millipore) for 1 hour at room temperature. Immunoreative bands were detected using Amersham Enhanced Chemiluminescence (Amersham Pharmacia Biotech).

Immunodetection of nascent rRNA was performed by incorporation of 5-fluorouridine (5-FU), according to the method described previously (19). Briefly, hernandonine-treated cells were incubated with 2 mmol/L 5-FU (Sigma) for 15 minutes, at that point washed with cold PBS and fixed in 4% paraformaldehyde and 1% Triton X-100 in PBS for 10 minutes. Along these lines, the cells were immunofluorescently stained with a particular mAb for halogenated uridine [1:400, Sigma (BU-33)]. Quantification of incorporation of 5-FU into rRNA were performed using Olympus cellSens Dimension software (Olympus Life Science). Histograms showed the values (mean ± SD) of three independent experiments.

Total RNA was isolated from harvested cells using TRIzol reagent (Thermo Fisher Scientific), as described previously (19). Reverse transcription was RevertAid Reverse Transcriptase (Thermo Fisher Scientific) according to the manufacturers' instructions. Subsequent real-time RT-PCR analysis of cDNA will be performed in triplicate using the SYBR Green dye on the StepOnePlus Real-Time PCR System (Applied Biosystems). The primers (5′–3′) are CTCCGTTATGGTAGCGCTGC and GCGGAACCCTCGCTTCTC for 45S; CGACGACCCATTCGAACGTCT and CTCTCCGGAATCGAACCCTGA for 18S; GGTGAAGCCAAAGGCAGATGT and TATGATGCGGCTTCTGGCAGGT for RNA Pol. I; CCTGGAGGGTCCTGTACAATCTC and GCAGGCACCTAATTGGGCTC for PUMA; GTGGCCTTCTTTGAGTTCGGT and GTGCCGGTTCAGGTACTGAGT for BCL2; R: CCGTCTAGAAAAACCTGCC and GCCAAATTCGTTGTCATACC for GAPDH. For the relative RNA expression, GAPDH were used as an internal control for all qRT-PCR reactions and compared with control groups. Histograms showed the values (mean ± SD) of three independent experiments.

The procedure of cell-cycle analysis using flow cytometry was described previously (19). Briefly, cells (5 × 10⁵/6-well plate) treated with hernandonine (0–10 μg/mL, 24 hours; 5 or 7.5 μg/mL for 3–24 hours) were harvested and fixed in ice-cold 70% ethanol for at least 30 minutes at 4°C. Then, cells were digested with DNase-free RNase A (50 U/mL) at 37°C for 30 minutes followed by DNA staining with 500 μL propidium iodide (PI, 10 μg/mL; Sigma). PI signals were measured to indicate cell-cycle status using a Becton Dickinson FACScan instrument and Cell Quest software. Histograms showed the values (mean ± SD) of three independent experiments.

FITC Annexin V Apoptosis Detection Kit I (BD Biosciences Canada) was used as described previously (19). Briefly, cells (1 × 10⁶/10-cm dish) treated with hernandonine (0–10 μg/mL, 24 hours) were washed and resuspended in 1 mL of binding buffer (10 mmol/L HEPES/NaOH, pH 7.5, 140 mmol/L NaCl, and 2.5 mmol/L CaCl₂). Cell suspension (500 μL) was incubated with 5 μL of Annexin V-FITC and 10 μL of PI for 10 minutes at room temperature in the dark, and then analyzed using a Becton Dickinson FACScan instrument and Cell Quest software. The results were described as the proportion of live cells, apoptotic cells (early- and late-stage apoptosis), and necrotic cells. Histograms showed the values (mean ± SD) of three independent experiments.

The global protein synthesis was assessed using Cayman Protein Synthesis Assay Kit (Cayman Chemical, # 601100), as described previously (19). Briefly, the protein synthesis rate was assessed by incorporation of puromycin analogue o-propargyl-puromycin (OPP) added to the hernandonine-treated cells resuspended in complete medium. These truncated C-terminal alkyne-labeled proteins were then subsequently measured via copper-catalyzed click chemistry utilizing 5 FAM-Azide using a Becton Dickinson FACScan instrument and Cell Quest software. All of these experiments were performed in duplicate and were repeated independently at least three times.

Student t tests were utilized to decide statistical significance and two-tailed P values are shown. At least three independent replicate experiments were performed to justify the use of statistical tests.

Hernandonine, an aporphine compound isolated from H. nymphaeifolia, has been shown to have cytotoxic effects (refs. 16, 17; Fig. 1A). To analyze its cytotoxic mechanisms and explore its potential as a chemotherapeutic agent, we tested the cytotoxic effects of hernandonine in various cancer cell lines, including human cervical cancer cells (HeLa), human colon cancer cells (HCT116-p53^+/+ and HCT116-p53^−/−), human oral squamous cancer cells (SAS and OECM-1), human lung adenocarcinoma (A549), and three normal human cell lines: normal human lung fibroblasts (IMR90), human normal oral keratinocytes (NOK) cells, and immortalized bronchial epithelial cells (Beas-2B). The results show that hernandonine induces higher cytotoxicity in human cancer cells, namely, HeLa, SAS, OECM-1, HCT116-p53^+/+, HCT116-p53^−/−, and A549 (IC₅₀ = 5, 7.5, 8, 7.5, 7.5, and 15 μg/mL, respectively), compared with normal human cells (IC₅₀ > 50 μg/mL for the IMR90, NOK, and Beas-2B cells; Fig. 1B). To understand the different sensitivities toward hernandonine-induced cytotoxicity of cancer and normal cells, we chose two sensitive cancer cell lines, HeLa and SAS, and compared them to two normal cell lines, NOK and Beas-2B. First, we confirmed that hernandonine induces higher cytotoxicity in the HeLa and SAS cells compared with the NOK and Beas-2B cells using a lactate dehydrogenase (LDH) release assay (Supplementary Fig. S1A). Furthermore, we found that hernandonine induced an increase in the number of in HeLa and SAS cells in the sub-G₁ phase and in early/late apoptosis in a dose- and time-dependent manner (0–10 μg/mL, 24 hours; 5 or 7.5 μg/mL for 3–8 hours) using cell-cycle analysis and Annexin V analysis, respectively (Fig. 1C–F; Supplementary Fig. S1B and S1C). On the other hand, we found that hernandonine induces an increase in the population of NOK and Beas-2B cells in G₂–M, indicating that hernandonine induces G₂–M arrest in normal cells (Fig. 1G–J).

RNA polymerase (Pol.) I, the multiprotein complex that synthesizes rRNA, is activated widely in cancer (22). We further investigated the effect of hernandonine on RNA Pol. I in HeLa, SAS, NOK, and Beas-2B cells compared it with the actinomycin D (Act. D), which has been suggested to inhibit RNA Pol. I (23). In results that differed from those for Act. D, hernandonine decreased the protein expression of RPA194, the large catalytic subunit protein of the RNA Pol. I holocomplex in a dose- and time-dependent manner in both the cancer cells (HeLa, SAS, and OECM-1) and the normal cells (NOK and Beas-2B; Fig. 2A–D; Supplementary Fig. S2A). However, neither hernandonine nor Act. D decreased the expression of RPA194 mRNA in these cells (Fig. 2E and F). Interestingly, we found that hernandonine increased the expression of RPA194 mRNA in HeLa and Beas-2B cells. This increase is likely due to intracellular negative feedback from the degradation of RNA Pol. I that had been induced by hernandonine to maintain a stable RNA Pol. I protein level. However, the details of this mechanism require further investigation. In addition, pretreatment with MG132, a proteasome degradation inhibitor, restored the hernandonine-induced reduction in RPA194 expression in a dose- and time-dependent manner in both the cancer cells (HeLa and SAS) and the normal cells (NOK and Beas-2B; 0–10 μg/mL for 24 hours; 5 or 7.5 μg/mL for 3–8 hours; Fig. 2G–L). These results suggest that hernandonine induces proteasome degradation of RNA Pol. I in both the cancer cells (HeLa and SAS) and the normal cells (NOK and Beas-2B).

RNA Pol. I is crucial for ribosomal biogenesis (1); therefore, we further investigated the effect of hernandonine on rRNA transcription and protein synthesis. Using an immunofluorescence assay, we found that hernandonine decreased nucleolar 5-FU incorporation into nascent rRNA, showing that hernandonine represses rRNA synthesis in both the cancer cells (HeLa, SAS, and OECM-1) and the normal cells (NOK and Beas-2B), which is a similar result to that observed for Act. D (Fig. 3A–E; Supplementary Fig. S2B and S2C). This finding is predictable with the outcomes from the investigation of quantitative real-time RT-PCR, which demonstrated that hernandonine essentially diminished the expression of 45S pre-rRNA yet not that of 18S rRNA (Fig. 3F–I). Furthermore, hernandonine also reduced global protein synthesis in both the cancer cells (HeLa, SAS, and OECM-1) and the normal cells (NOK and Beas-2B; Fig. 3J; Supplementary Fig. S2D). These results indicate that hernandonine inhibits ribosomal biogenesis, including rRNA transcription and protein translation, in human normal and cancer cells.

It has been demonstrated that rRNA synthesis repression is identified with disintegration of nucleolar structures (24). We decided the impact of hernandonine on nucleolus morphology and molecular rearrangements and contrasted these progressions with those actuated by Act. D, a well-described nucleolar disruptor (10). Similar to the observations after Act. D treatment, hernandonine-induced upstream binding factor (UBF) segregated into caps around the DAPI-sparse nucleoli to such an extent that nucleophosmin (B23) and nucleolin (NCL) were mislocalized over the nucleoplasm in both the cancer cells (HeLa, SAS, and OECM-1) and the normal cells (NOK and Beas-2B; Fig. 4A–H; Supplementary Fig. S2E and S2F). Act. D is an antitumor drug that particularly intercalates into transcriptionally active regions of chromosomal DNA and thereby abrogates RNA synthesis (25). Previous studies have shown that Act. D, at a concentration that inhibits transcription, can induce γH2AX foci formation (0.5 μg/mL; ref. 25). Our results showed that hernandonine induced lower γH2AX foci formation, as determined using Western blot analysis and immunofluorescence analysis in both the cancer cells (HeLa and SAS) and the normal cells (NOK and Beas-2B) treated with hernandonine (0–10 μg/mL) compared with Act. D (IC₅₀ = 0.1 μmol/L for both the HeLa and SAS cells; Fig. 5A–G). These results suggest that hernandonine caused less DNA damage than Act. D. In addition, hernandonine induced less γH2AX formation in the normal cells (NOK and Beas-2B) compared with the cancer cells (HeLa, SAS) within the same dose range (0–10 μg/mL; Fig. 5). These results suggest that hernandonine induces nucleolar reorganization but does not activate DNA damage signaling.

The nucleolus is the site of ribosome biogenesis, which is a fundamental cellular process (1, 7), and debilitation of ribosome biogenesis induces ribosomal stress (otherwise called nucleolar stress; refs. 7, 10, 12). Both p53-dependent and p53-independent nucleolar stress pathways have been shown to induce cell-cycle arrest or apoptosis (26, 27). For this study, we investigated the effect of hernandonine on the nucleolar stress pathway and the cellular fate in both the normal and the cancer cells. The results showed that hernandonine induced intracellular apoptotic pathways, such as those leading to cleavage of caspase-9, -3, and PARP, in different types of cancer cells, such as those from the HeLa, SAS, HCT116, and OECM-1 cell line; Fig. 6A and B; Supplementary Fig. S3A–S3C), but not in normal cells, such as NOK and Beas-2B cells (Fig. 6C and D). Taking into account that both p53 and E2F1 are pivotal controllers involving p53-dependent and p53-independent nucleolar stress pathways, respectively (27, 28), we next examined their expression and the subsequent signaling they induce. The results showed that hernandonine increased expression of a p53 downstream proapoptotic gene, PUMA (29), and induced downregulation of E2F1 and decreased expression of a downstream antiapoptotic gene, BCL-2 (30), in HeLa and SAS cells (Fig. 6A and B; Supplementary Fig. S4A–S4D). These results suggest that both p53-dependent and p53-independent nucleolar stress pathways contribute to hernandonine-induced apoptosis in cancer cells.

As a result of decreased ribosome biogenesis, a few ribosomal proteins (RP), for example, L5, L11, L23, and S7, are not constantly utilized for ribosome generation, but rather bind to MDM2 to diminish its hindrance on p53, just as its protection toward E2F1 (31–33). Here, we found that hernandonine reduced the amounts of p53 and E2F1 in complex with MDM2 (Fig. 6E and F). On the other hand, the amount of RPL11 related with MDM2 was increased in hernandonine-treated cells (Fig. 6E and F) in HeLa and SAS cells. The results demonstrate that the mechanism of p53 stabilization or E2F1 degradation after inhibition of ribosome biogenesis may be the result of changes in the function and physical communications of these proteins with MDM2. On the other hand, we found that hernandonine induces cell-cycle arrest at G₂–M but not apoptosis in normal cell lines, namely, the NOK and Beas-2B cells (Figs. 1G–J and 6C and D). Previous studies have shown that the gene encoding the mitosis-promoting phosphatase Cdc25C is transcriptionally repressed by p53, contributing to the p53-driven G₂ arrest during exposure to different types of stresses (34–36). Here, we found that hernandonine induced stabilization of p53 and phosphorylation of p53 followed by a decrease in Cdc25C expression in these normal cells, although an increase in cyclin B1 was observed (Fig. 6G and H). These results indicate that downregulation of Cdc25C through activation of p53 results in G₂–M arrest in normal cells treated with hernandonine.

Deregulated ribosomal RNA synthesis is related with uncontrolled malignancy cell proliferation. RNA polymerase (RNA Pol. I), the multiprotein complex that synthesizes rRNA, is broadly enacted in cancer, and its action provides compelling reasons to target RNA Pol. I transcription as a therapeutic strategy (22). In this study, we found that an oxoaporphine-related compound, hernandonine, causes nucleolar stress that is represented by relocalization of nucleolar proteins, hindrance of RNA Pol. I transcription, and loss of the RNA the Pol. I catalytic subunit RPA194, in this manner exhibiting its anticancer activity. We further demonstrated that this molecule acts in a p53-dependent or p53-independent manner and that RNA Pol. I repression happens upstream of p53 activation. This study identified new and significant molecules for targeting RNA Pol. I transcription.

Hernandia nymphaeifolia (Presl) Kubitzki (Hernandiaceae) is an evergreen tree that is distributed in the tropical island shores of the Indian Ocean and western Pacific Ocean, and its seed is used as a cathartic drug (37). Aporphine-related compounds isolated from H. nymphaeifolia have been shown to display cytotoxic effects (16, 17). We tested different aporphine-related compounds isolated from H. nymphaeifolia for nucleolar stress, but only hernandonine-induced nucleolar disintegration (Supplementary Fig. S5). In addition, we found that hernandonine induced proteasome-dependent degradation of RPA194, the large catalytic subunit protein of the RNA Pol. I holocomplex in both the cancer cells (HeLa and SAS) and normal cells (NOK and Beas-2B) in a dose- and time-dependent manner (Fig. 2). Previous studies have shown that small-molecule BMH compounds destabilize RPA194 in a proteasome-dependent manner, resulting in the inhibition of nascent rRNA synthesis and expression of the 45S rRNA precursor (38, 39). However, the structures of these BMH compounds are quite distinct from each other. Within these compounds, pyridoquinazolinecarboxamide BMH-21 has been suggested to bind GC-rich sequences, which are present at high frequency in ribosomal DNA genes and rapidly repress RNA Pol. I transcription (38). Previous studies have also shown that CX-5461, which appears to inhibit the formation of the RNA Pol. I preinitiation complex, is structurally and mechanistically distinct from the BMH compounds (38–40). The current study identifies another structurally new molecule, oxoaporphine, for RNA Pol. I degradation. However, structure and activity analyses of aporphine-related compounds against RNA Pol. I are required to further understand the mechanism in detail.

Previous studies have demonstrated that the inhibition of rRNA synthesis has deleterious effects only on cancer cells due to loss of checkpoint mechanisms and upregulated ribosome biogenesis in the cancer cells but does not affect normal tissues (4, 40–42). In this study, we found that hernandonine inhibits RNA Pol. I transcription in both the cancer and the normal cells (Fig. 3). However, hernandonine induces cellular apoptosis in the cancer cells, while hernandonine induces cell-cycle arrest in the normal cells (Figs. 1 and 6). We further found that normal cells (NOK and Beas-2B), but not cancer cells (HeLa and SAS), were able to resume the normal cell cycle and survive after the hernandonine was removed (Supplementary Fig. S6). A high dose of hernandonine (>10 μg/mL) causes cellular apoptosis in normal cells (Supplementary Fig. S7). Therefore, the results suggest that the rRNA synthesis inhibition induced by a lower dose of hernandonine (< 10 μg/mL) in normal cells triggers cell-cycle arrest and that these cells could survive after hernandonine is removed. However, rRNA inhibition induced by hernandonine to a great extent causes cellular apoptosis in normal cells. This finding explains the higher hernandonine-induced cytotoxicity in human cancer cells (HeLa, SAS, OECM-1, HCT116, and A549 cells with IC₅₀ = 5, 7.5, 8, 7.5, and 15 μg/mL, respectively) compared with normal human cells (IC₅₀ >50 μg/mL for the IMR90, NOK, and Beas-2B cells; Fig. 1B).

A widely accepted mechanism of p53 checkpoint activation is based on alteration of ribosome biogenesis (31–33). To further investigate the mechanism by which hernandonine induced the different fates of normal cells and cancer cells, we analyzed the p53-dependent nucleolar stress pathway. The results showed that hernandonine-induced cytotoxicity and cellular apoptosis in the cancer cell lines (Fig. 1B–F) and that hernandonine-induced p53 stabilization in the HeLa, SAS, and HCT116-p53^+/+ cells but not in the HCT116-p53^−/− cells (Fig. 6A and B; Supplementary Fig. S3). Therefore, this result suggests that p53 is likely not required for the nucleolar stress responses by hernandonine. It was found that a transcriptional factor, E2F1 or c-Myc, which is involved in rRNA transcription, plays an important role in the p53-independent mechanism that links nucleolar stress to cellular apoptosis (26, 27, 43). Here, we found that degradation of E2F1, followed by a decrease in BCL-2, may be involved in the p53-independent nucleolar stress pathway in the hernandonine-treated HeLa and SAS cancer cells (Fig. 6A and B; Supplementary Fig. S4). Previous studies have demonstrated that MDM2 adversely regulates p53 activity in two different methods: by interfering with its transactivation activity and by facilitating p53 proteasomal degradation (44). Conversely, it has been demonstrated that MDM2 binds to the E2F1 protein and shields it from proteasome-mediated degradation (45). Our results showed that the amounts of p53 and E2F1 in complex with MDM2 were diminished, while the amount of RPL11 related with MDM2 was increased, in hernandonine-treated cancer cells (Fig. 6E and F). We decipher these results as showing that hernandonine-induced rRNA synthesis repression causes breaking down of the polysome and the ribosomal structure (Fig. 3 and 4). Thus, ribosomal proteins, including RPL11, are discharged from the ribosomal structure and RPL11 is then ready to bind to MDM2. The RPL11-bound MDM2 loses its capacity to intervene p53 degradation and can never again ensure E2F1 against proteasomal degradation (Fig. 6A and B). Taken together, these findings show that hernandonine induces cellular apoptosis through a p53-dependent or p53-independent process in solid tumor cell lines.

Conversely, we found that hernandonine induced G₂–M arrest in normal cells. Progression through G₂–M is driven by a complex of cyclin B1 and cdc2. Cyclin B1 and cdc2 accumulate and associate in the cytoplasm during G₂. The complex is kept inactive by phosphorylation of cdc2, and entry into mitosis is triggered by translocation of the complex to the nucleus where it is activated by dephosphorylation by Cdc25 phosphatases (46). Cdc25C is a dual-specificity phosphatase that promotes entry into mitosis by removing the inhibitory phosphates on the cyclin-dependent kinases, cdc2 (34, 47). Previous work recognized the Cdc25C gene as a target for direct transcriptional repression by p53, and Cdc25C downregulation was demonstrated to be required for maintenance of G₂ arrest (34–36, 48). Here, we found that hernandonine activated p53, increased cyclin B1 expression, and reduced Cdc25C expression, resulting in G₂–M arrest in the NOK and Beas-2B cells (Fig. 6G and H). These results suggest that Cdc25C plays a crucial role in hernandonine-induced G₂–M arrest and that an increase in cyclin B1 is most likely the result of G₂ arrest. Interestingly, we found that hernandonine also induced S-phase arrest in NOK cells, although the effect was not as significant as that in hernandonine-induced G₂–M arrest (Fig. 1G and H). However, this phenomenon was not observed in Beas-2B cell (Fig. 1I and J). This is probably due to cell-specific effect. The detailed mechanism needs further investigation. To sum up, the cell-cycle arrest observed in hernandonine-treated NOK and Beas-2B cells is because of the p53 checkpoint pathway activation (49), which is reversible. After recuperation of ribosome biogenesis, cell-cycle progression resumes, and cells accomplish a ribosomal complement adequate for giving rise to normal daughter cells (Supplementary Fig. S6C and S6D). Notwithstanding, the result was not the equivalent for cancer cells in which rRNA repression did not modify the cell-cycle progression and cells underwent progressive ribosome depletion that prompted apoptotic death (Supplementary Fig. S6A and S6B).

Most chemotherapeutic agents exert their toxic action on cancer cells by damaging DNA, hindering DNA synthesis, and disrupting the mitotic process (13). The absence of a DNA damage response may be a preferable mechanism of action because it potentially lowers the toxicity, including genotoxicity, in normal cells (22). Act. D is a well-characterized nucleolar disruptor (10) that is induced through inhibition of RNA Pol. I (23). However, it has been shown that Act. D intercalates into specific transcriptionally active regions of chromosomal DNA, resulting in γH2AX foci formation and thereby abrogating RNA synthesis (0.5 μg/mL; ref. 25). Here, we found that different doses of hernandonine (0–10 μg/mL) induced less γH2AX formation compared with Act. D (IC₅₀ = 0.1 μmol/L for the HeLa and SAS cells) in both the cancer cells and the normal cells (Fig. 5). The dose of hernandonine to induce cytotoxicity in cancer cell lines was far higher than that for Act. D, but less DNA damage was produced by hernandonine in these cells. In addition, hernandonine induced less γH2AX formation in the normal cells (NOK and Beas-2B) than in the cancer cells (HeLa, SAS) (Fig. 5). These results suggest that hernandonine has potential antitumorigenic activity, but does not activate DNA damage signaling. This also explains the reason that hernandonine causes less cytotoxicity in normal human cells (IC₅₀ >50 μg/mL for the IMR90, NOK, and Beas-2B cells; Fig. 1B). However, previous studies have shown that oxoaporphine alkaloid derivatives have intercalating DNA properties and inhibited topoisomerase I activity, which results in cytotoxicity (50). In addition, numerous cancer therapeutics, particularly the topoisomerase I and II poisons, hinder RNA Pol. I transcription by causing torsional stress of the rDNA (24). These actions are applicable to consider and screen for their potential therapeutic advantage. In this study, we provide evidence showing that hernandonine inhibits RNA Pol. I transcription and activates degradation of the RNA Pol. I catalytic subunit RPA194. Therefore, these findings do not rule out the possibility that hernandonine can bind to DNA or inhibit topoisomerase activity or that such activity is related to hernandonine-induced cellular responses.

In conclusion, we have studied the anticancer potential of a small molecule, hernandonine, which is isolated from H. nymphaeifolia for use as a traditional Chinese medicine, and provided evidence of its mode of action. Our results show that hernandonine destabilizes RPA194 in a proteasome-dependent manner and inhibits nascent rRNA synthesis and expression of the 45S rRNA precursor. Finally, hernandonine induces cellular apoptosis through a p53-dependent or p53-independent process in solid tumor cell lines. Overall, these results define hernandonine as a new and important molecule for targeting RNA PoI. I transcription and provide strong support that RNA Pol. I is a rational and tractable target for cancer therapeutics.

No potential conflicts of interest were disclosed.

Conception and design: H.-T. Wang

Development of methodology: Y.-T. Chen

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y.-T. Chen, H.-T. Wang

Writing, review, and/or revision of the manuscript: Y.-T. Chen, H.-T. Wang

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J.-J. Chen, H.-T. Wang

This work was supported by the Ministry of Science and Technology, Taiwan under grant nos. 106-3114-B-010001 and 107-2320-B-010-018 (to H.-T. Wang) and the Yin Yen-Liang Foundation Development and Construction Plan of the School of Medicine, National Yang-Ming University.

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.