We have defined the mechanism of action of lurbinectedin, a marine-derived drug exhibiting a potent antitumor activity across several cancer cell lines and tumor xenografts. This drug, currently undergoing clinical evaluation in ovarian, breast, and small cell lung cancer patients, inhibits the transcription process through (i) its binding to CG-rich sequences, mainly located around promoters of protein-coding genes; (ii) the irreversible stalling of elongating RNA polymerase II (Pol II) on the DNA template and its specific degradation by the ubiquitin/proteasome machinery; and (iii) the generation of DNA breaks and subsequent apoptosis. The finding that inhibition of Pol II phosphorylation prevents its degradation and the formation of DNA breaks after drug treatment underscores the connection between transcription elongation and DNA repair. Our results not only help to better understand the high specificity of this drug in cancer therapy but also improve our understanding of an important transcription regulation mechanism. Mol Cancer Ther; 15(10); 2399–412. ©2016 AACR.
Cancer cells aberrantly deregulate specific gene expression programs with critical functions in cell differentiation, proliferation, and survival (1). Differently from noncancer cells, those altered gene programs in cancer cells have a striking dependence on continuous active transcription. For example, small cell lung cancer (SCLC) cells are addicted to lineage-specific and proto-oncogenic transcription factors that support their growth (2–7). Similarly, triple-negative breast cancer (TNBC) is highly dependent on uninterrupted transcription of a specific key set of genes (8, 9). Pharmacologic modulation of transcription of protein-coding genes may thus provide an approach to identify and treat tumor types that are dependent on deregulated transcription for maintenance of their oncogenic state.
Targeting DNA in tumor cells happened to be the most explored therapeutic strategy to block DNA processing enzymes such as those involved in transcription (e.g., cisplatin and derivatives, anthracyclines, etc.; ref. 10). Currently, several laboratories are developing inhibitors of cyclin-dependent kinases (CDK) that have a critical role in regulating transcription initiation, pause release, and elongation (e.g., CDK7, CDK8, or CDK9), the three main steps involved in RNA synthesis (11, 12). Other approaches are inhibition of DNA repair mechanisms (e.g., irinotecan, topotecan, olaparib; ref. 13) or chromatin remodeling (HDAC inhibitors or demethylating agents; refs. 14, 15). Although these compounds have already entered clinical trials, the mechanisms by which they disturb transcription as well as those driving to cancer cell death are far from being understood.
Here, we describe the inhibition of transcription by lurbinectedin (PM01183; Fig. 1A), an anticancer agent that is being evaluated in late-stage (phases II and III) clinical trials. Lurbinectedin is structurally related to trabectedin, containing the same pentacyclic skeleton of the fused tetrahydroisoquinoline rings, but differing by the presence of a tetrahydro-B-carboline replacing the additional tetrahydroisoquinoline of trabectedin. The pentacyclic skeleton is mostly responsible for DNA minor groove recognition and binding. Lurbinectedin reacts with the exocyclic amino group of guanines in the minor groove of DNA forming a covalent bond. The resulting adduct is additionally stabilized through the establishment of van der Waals interactions and one or more hydrogen bonds with neighboring nucleotides in the opposite strand of the DNA double helix (16). The additional tetrahydro β-carboline moiety protrudes from the DNA minor groove and could be interacting directly with specific factors involved in DNA repair and transcription pathways. Indeed, it is possible that this part of the molecule interacts directly with TC-NER factors and could interfere with the repair mechanism. In this sense, lurbinectedin is able to attenuate the repair of specific nucleotide excision repair (NER) substrates (17, 18). In addition to its activity in tumor cells, it was recently shown that lurbinectedin affects the inflammatory microenvironment, with a selective apoptotic-inducing effect on mononuclear phagocytes and a specific inhibition of production of inflammatory cytokines (19, 20). In this work, we show that, following its specific target on CG-rich sequences located at promoters of protein-coding genes, lurbinectedin induces the specific degradation of elongating (phosphorylated) RNA polymerase II (Pol II) by the ubiquitin-proteasome machinery. This process occurs specifically on activated genes and is associated with the formation of DNA breaks that drive tumor cells to apoptosis. Inhibition of Pol II phosphorylation prevents its degradation and the formation of DNA breaks. These investigations not only show how lurbinectedin causes a cascade of events on the transcription process that can explain its antiproliferative activity on tumor cells, but also improve our understanding of the fate of Pol II when it encounters a lesion on the DNA.
Materials and Methods
Lurbinectedin was produced by PharmaMar through a semisynthetic method. Z-Leu-Leu-Leu-al (MG-132), 5,6-dichlorobenzimidazole-1-α-D-ribofuranoside (DRB), and flavopiridol were purchased from Sigma. The following antibodies were used for Western blotting: POLR2A (RPB1, Pol II) (clones N-20 and H-224), POLR1A (RNA Pol I), POLRMT (B-1), CCNH (Cyclin H) (B-1), CDK9 (C-20), ERCC2 (p80-TFIIH) (H-150), TP53 (FL-393) from Santa Cruz Biotechnology; TBP, GTF2H1 (TFIIH), POLR2D (RPB4, RNA Pol II), POLR2B (RPB2, RNA Pol II), CRCP (RNA Pol III) from Abcam; CDK7 from Cell Signaling Technology; and anti–phospho-Ser2Pol II (clone H5) and anti–phospho-Ser5 Pol II (clone H14) from Covance. The following antibodies were used for chromatin immunoprecipitation (ChIP) and immunoprecipitation (IP) experiments: Antibodies against phospho-Ser2 Pol II (clone 3E8) and phospho-Ser2 Pol II (clone 3E10) were from Active Motif. Polyclonal antibodies against POLR2A (H-224), CDK7 (C-19), CDK9 (H-169), UBB (Ubiquitin clone FL-76 or clone A-5), Biotin(33), and ERCC4 (XPF, clone H-300) were from Santa Cruz Biotechnology. Antibody against γ-H2AX (ab2893) was obtained from Abcam. POLR2A (1PB-7C2), PSMC5 (Sug-1, clone 3SCO), and GTF2F2 (TFIIF β) antibodies were from IGBMC.
The following cell lines were obtained from the ATCC in 2007: A549 (lung adenocarcinoma; CCL-185), A673 (human muscle Ewing's sarcoma; CRL-1598), HCT-116 (human colorectal carcinoma; CCL-247), HT-29 (human colorectal carcinoma; HTB-38), MCF7 (breast adenocarcinoma; HTB-22), MDA-MB-231 (breast adenocarcinoma; HTB-26), and PSN-1 (pancreatic adenocarcinoma; CRM-CRL-3211). HeLa and HeLa siXPF are cervical carcinoma cells and were obtained from Tebu-Bio. Dr. M. D'Incalci (Istituto Mario Negri) generously provided IGROV1 and IGROV-ET ovarian cancer cell lines. All cell lines were cultured in the medium recommended by the supplier supplemented with 10% FBS, 2 mmol/L l-glutamine, and penicillin–streptomycin mix (Sigma). None of the cell lines used in the article have been authenticated in the laboratory in the last 6 months.
Cell proliferation was studied by [3-(4,5-dimethythiazol-2-yl)-2,5-diphenyl] tetrazolium bromide (MTT) assays that were performed following the manufacturer's instructions (MTT Cell Proliferation Kit I; Roche Diagnostics). Briefly, cells were seeded in 96-well trays. Serial dilutions of lurbinectedin, PM030779, or PM120306 were added to the medium. Exposure to the drugs was maintained during 72 hours. Determination of IC50 values was performed by iterative nonlinear curve fitting using the Prism 5.0 statistical software (GraphPad). The data presented are the average of three independent experiments performed in triplicate.
DNA electrophoretic mobility shift assay
The binding assay was performed with a 250 pb PCR product from the human adiponectin gene. After incubation with appropriate concentrations of the compounds at 25°C during 1 hour, the DNA was subjected to electrophoresis in a 2% (w/v) agarose/TAE gel, stained with ethidium bromide (Sigma) and photographed.
DNase I footprinting assays
Radiolabeled AS/CGG and OS/CCG were bound to magnetic beads (Dynabeads) and incubated for 30 minutes at room temperature with the indicated drug concentrations (21). After extensive washings, DNase I digestion was performed for 45 seconds at room temperature. Purified nucleic acids were resolved on an 8% denaturing Urea-polyacrylamide gel.
In vitro transcription assays
Run-off transcription assays were performed using recombinant TFIIB, TFIIE, TFIIF, TBP, TFIIH, and RNA pol II, as previously described (22).
RNA synthesis quantification in tumor cells.
A549 (3.5 × 104 cells/well), A673 (2.6 × 105), HCT116 (1.8 × 105), HeLa (1.5 × 105), and MDA-MB-231 (2 × 105) were seeded in 24-well plates and incubated with lurbinectedin or vehicle (DMSO) for 30, 45, 60, and 90 minutes and pulsed with 5 μCi [3H] uridine (Perkin Elmer) for 30 minutes in medium supplemented with 10% FCS. Then, cells were rinsed twice in PBS and fixed with chilled 10% trichloroacetic acid for 10 minutes, and the monolayers were washed with ethanol and air-dried at room temperature. The precipitated macromolecules were dissolved in 0.5 N NaOH and 0.1% SDS and diluted in Ultima-Flo M (Perkin Elmer). The radioactivity was measured using a β-counter Hidex 300SL.
Western blotting analysis.
Cell protein extracts were prepared following standard procedures in RIPA buffer in the presence of protease inhibitors (Complete, Roche Diagnostics) and phosphatase inhibitors (PhosStop, Roche Diagnostics). After quantification with the Micro-BCA Protein Assay Kit (Thermo Scientific), 15 to 25 μg of protein were separated by SDS-PAGE and transferred to PVDF membranes (Immobilon-P; Millipore). After using appropriated primary and secondary antibodies, blots were developed by a peroxidase reaction using the ECL detection system (Amersham-G.E. Healthcare).
Pol II half-life measurement
Pulse-chase analyses were carried out in cells that were pretreated during 2 hours in DMEM cys−/met− medium and then metabolically labeled with 100 μCi/mL 35S-methionine for 1 hour in DMEM cys−/met− medium. After washing steps, cells have been treated with lurbinectedin. Whole-cell extracts were prepared using RIPA buffer (0.01 mol/L Tris–HCl, pH 8.0, 0.14 mol/L NaCl, 1% Triton X-100, 0.1% Na Deoxycholate, 0.1% SDS), and Pol II was immunoprecipitated using a specific antibody and resolved by SDS–PAGE. 35S-Pol II bands were quantified by phosphorimager analysis using ImageJ software (open source).
A549 cells were treated with the appropriate concentration of lurbinectedin for 4 hours in the absence or presence of the transcription inhibitor 20 μmol/L DRB (preincubation of 1 hour), washed, fixed (4% paraformaldehyde), permeabilized (0.5% Triton X-100), and incubated with the primary anti-Pol II monoclonal antibody for 1 hour at 37°C. Thereafter, the cells were washed and incubated with the AlexaFluor 594 secondary goat anti-mouse IgG (Invitrogen) for 30 minutes at 37°C. Finally, the slides were incubated with Hoechst 33342 (Sigma) and mounted with Mowiol mounting medium. Pictures were taken with a Leica DM IRM fluorescence microscope equipped with a 100x oil immersion objective and a DFC 340 FX digital camera (Leica).
A549 whole-cell extracts were made using RIPA buffer supplemented with PIC (Roche) and phosphatase inhibitors (Active Motif). Antibodies were incubated with magnetic beads and lately protein whole-cell extract was added. Sequential washes were done, and the resulting sample was loaded on SDS-PAGE for Western blotting. When Tandem Ubiquitin Binding Entity (TUBES 2) technology (Lifesensors) was used, equilibrated slurry was initially incubated with protein whole cell extract at 4°C. After extensive washing steps, the resulting sample was loaded on SDS-PAGE for Western blotting.
Subcellular protein fractionation
A549 cells were incubated with the different compounds at different time points, and subcellular protein fractionation was performed using the Subcellular Protein Fractionation Kit for Cultured Cells (Thermo Scientific Waltham) following the manufacturer's instructions.
Reverse transcription and quantitative PCR
Total RNA was isolated using the RNeasy Mini Kit (QIAGEN) and reverse transcribed with SuperScript II reverse transcriptase (Invitrogen). The quantitative PCR was done using the Lightcycler 480 SYBR Green and the Lightcycler 480 (Roche Diagnostics). The primers used in the real-time PCR experiments were as follows: for RARβ2: fw 5′-CCAGCAAGCCTCACATGTTTCCAA-3′ and rv 5′-TACACGCTCTGCACCTTTAGCACT-3′; for Glyceraldehyde 3-phosphate dehydrogenase (GAPDH): fw 5′-TCGACAGTCAGCCGCATCTTCTTT-3′ and rv 5′-ACCAAATCCGTTGACTCCGACCTT-3′. RARβ2 mRNA levels represent the ratio between values obtained from treated and untreated cells normalized against the housekeeping GAPDH mRNA.
Cells were cross-linked at room temperature for 10 minutes with 1% formaldehyde. Chromatin was isolated and sonicated (23). Samples were immunoprecipitated (IP) with antibodies at 4°C overnight, and protein G-Sepharose beads (Upstate) were added, incubated 3 hours at 4°C, and sequentially washed. DNA fragments were purified using the QIAquick PCR purification Kit (QIAGEN) and analyzed by real-time PCR using sets of primers targeting RARβ2 gene promoter: fw 5′-TGGTGATGTCAGACTAGTTGGGTC-3′ and rev 5′-GCTCACTTCCTACTACTTCTGTCAC-3′; and RARβ2 exon 4: fw 5′-TCCAGCTGTCAGGAATGACAGGAA-3′ and rev 5′-TGAGATCGTCCAACTCAGCTGTCA-3′. Biotin-ChIP assays have been performed as previously described (23).
OS/CCG DNA template (containing a single lurbinectedin adduct; ref. 21) has been bound to magnetic beads (Dynabeads) and incubated with the indicated drug concentrations. Dual incision assays were next carried out after addition of XPG, XPF/ERCC1, XPC/hHR23B, RPA, XPA, and TFIIH. The reactions were conducted as previously described (24).
After lurbinectedin or PM030779 treatment, single-cell gel electrophoresis assay has been used (CometAssay; Trevigen) following the manufacturer's instructions. Experiments and pictures analyses have been performed as previously described (25).
Antitumor activity in xenograft murine models
All animal protocols were reviewed and approved according to regional Institutional Animal Care and Use Committees. Mice used in the following experiments were always female 4 to 6 weeks of age, 16 to 25 gr, athymic nude-Foxn1 nu/nu obtained from Envigo (Italy). Mice were housed in individually ventilated cages on a 12-hour light–dark cycle at 21 to 23ºC and 40% to 60% humidity. Mice were allowed free access to an irradiated diet and sterilized water. Design, randomization, and monitoring of experiments (including body weights and tumor measurements) were performed using NewLab Software v2.25.06.00 (NewLab Oncology). All mice were s.c. xenografted with A549 cancer cells into their right flank with c. 3 × 106 cells in 0.2 mL of a mixture (50:50; v:v) of Matrigel basement membrane matrix (Becton Dickinson) and serum-free medium. When tumors reached approximately 150 mm3, mice were randomly assigned to treatment or control groups. Lurbinectedin was intravenously administered in three consecutive weekly doses (0.18 mg/kg/day), whereas the control animals received an equal volume of vehicle with the same schedule. Caliper measurements of the tumor diameters were made twice weekly, and tumor volumes were calculated according to the following formula: (a x b)2/2, where a and b were the longest and shortest diameters, respectively. Animals were humanely killed when their tumors reached 3,000 mm3 or if significant toxicity (e.g., severe body weight reduction) was observed. Differences in tumor volumes between treated and control groups were evaluated using the Mann–Whitney U test. Statistical analyses were performed by LabCat v8.0 SP1 (Innovative Programming Associates, Inc.).
A549 cells treated with 300 nmol/L of PM120306 during 4 hours were cross-linked with 1% formaldehyde at room temperature for 10 minutes. After sonication, the soluble chromatin was incubated with 80 μL of Streptavidin beads (Life technologies) overnight. After wash the captured DNA was eluted, de-cross-linked and purified by QiAquick Spin columns (Qiagen). The DNA was used to construct a library and sequenced on an Illumina HiSeq 2500 system. The sequencing reads were aligned to the hg19 assembly by using Bowtie 1.0. The HT-seq data were visualized by preparing custom tracks for the UCSC (University of California Santa Cruz) genome browser (https://genome.ucsc.edu). Annotations were performed by using both MACS (http://liulab.dfci.harvard.edu/MACS/; ref. 26) and Ensembl database Release 75. CG motifs search was performed by using Hypergeometic Optimization of Motif Enrichment (HOMER; http://homer.salk.edu; ref. 27). The pattern search of the motif CGG in the summit (100 bp) of all the identified peaks was done using a home-made Java program.
Lurbinectedin targets DNA to arrest cancer cell growth
We analyzed the effect of lurbinectedin on several human cancer cell lines including lung (A549), Ewing sarcoma (A673), colon (HCT-116, HT-29), breast (MCF7, MDA-MB-231), cervix (HeLa), and pancreas (PSN-1), over a 72-hour period (Fig. 1B; Supplementary Table S1). Lurbinectedin showed a potent antiproliferative activity with IC50 values in the low nanomolar range on all the cancer cell lines tested so far. We then performed xenograft studies to check whether the antiproliferative effect of lurbinectedin translated into in vivo antitumor activity. A549 cells were xenografted into the right flank of athymic nu/nu mice. Once the tumors reached c.150 mm3, mice were randomized into groups (n = 10) and either vehicle or lurbinectedin (0.18 mg/kg/day) was intravenously administered in three consecutive weekly doses. In those conditions, lurbinectedin presented antitumor activity with a statistically significant inhibition of tumor growth (Fig. 1C). In control mice, mean tumor volumes at 0, 7, and 14 days were 165 mm3, 582 mm3, and 1,575 mm3, respectively, whereas in lurbinectedin-treated mice, mean tumor volumes at the same time intervals were 170 mm3, 168 mm3, and 278 mm3, respectively (P = 0.7, P = 0.004, and P = 0.002, respectively). Lurbinectedin also induced an improvement of mice survival (mean survival for control mice: 18 days vs. mean survival in lurbinectedin-treated mice: 47 days; P = 0.001; Fig. 1D). No significant toxicity or body weight loss was observed in the treated animals (data not shown).
Interestingly, IGROV-ET ovarian cancer cells, overexpressing P-glycoprotein and previously shown to be resistant to doxorubicin, etoposide, and trabectedin (28), were less sensitive to lurbinectedin (Fig. 1E). These data indicated that lurbinectedin needed to accumulate in the cell to exert its antiproliferative activity. Band shift assays next demonstrated that the drug bound to DNA (Fig. 1F). Indeed, we observed a delay in electrophoretic migration of a 250 bp DNA fragment treated with lurbinectedin, while when treated with PM030779, a structural analogue lacking the hydroxyl group at position 21 that is involved in the covalent binding to guanines (ref. 29; Supplementary Fig. S1A), DNA migrated as the control-untreated fragment. We then analyzed the effect of either lurbinectedin or PM030779 on A549 lung cancer cells over a 72-hour period (Fig. 1G). Lurbinectedin showed a potent antiproliferative activity with IC50 values in the low nanomolar range, whereas PM030779 was inactive.
Lurbinectedin specifically targets DNA CG-rich regions
We next searched for the target site of the drug using a DNase I footprinting assay. We designed a DNA template that contained a unique, high-affinity adduct-forming site (CGG triplet) in just one strand. We observed that lurbinectedin was, in fact, bound to both strands (Fig. 2A). Lurbinectedin-bound DNA strand (AS) was protected from DNase I digestion from nucleotides T-4 to T+6, being G0 the guanine to which lurbinectedin was covalently bound (Fig. 2A, lanes 4–6). We also observed that the drug interacted with the opposite strand (OS) through hydrogen bonds and van der Waals forces (Fig. 2A, lanes 11–12).
To identify genomic binding sites of lurbinectedin in A549 cells, we next performed a chemical affinity capture (Chem-Seq) which measures the incorporation of the biotin-linked lurbinectedin analogue (PM120306; Supplementary Fig. S1B) within DNA followed by DNA sequencing. This analogue that binds guanine through its hydroxyl group (as lurbinectedin) was shown to exhibit antiproliferative properties similar to lurbinectedin (Supplementary Fig. S1C and S1D, see below). Sonicated chromatin from cells treated by PM120306 was incubated with magnetic streptavidin beads to isolate the biotinylated DNA fragments followed by massively parallel DNA sequencing. These sequences were used to reveal DNA regions enriched (peaks) in PM120306-bound sites genome wide. We identified approximately 1,000 peaks in the sequenced DNA. Interestingly, we detected a high density of peaks in the vicinity (+/- 10 Kb) of promoter-transcription start sites (TSS; Fig. 2B). We also observed that CpG-rich regions around TSS (Fig. 2C) overlapped with the peaks previously identified (Fig. 2B). Knowing that the optimal binding site of lurbinectedin is the CGG triplet, we did a pattern search of this motif in all the identified peaks: 50% of them included at least one CGG triplet (data not shown).
Altogether, our data show that lurbinectedin specifically targets CG-rich regions that are largely located in areas surrounding promoter regions.
Lurbinectedin inhibits RNA synthesis and induces Pol II degradation
The inhibition of RNA synthesis by lurbinectedin was investigated in A549 lung cancer cells (Fig. 3A) as well as in Ewing sarcoma (A673), colon (HCT-116), cervix (HeLa), and breast (MDA-MB-231) cancer cell lines (Supplementary Fig. S2A). All cell lines were treated with lurbinectedin (30 nmol/L) over time before being pulsed with [3H] uridine. Total RNA synthesis was inhibited by around 40% after 30 minutes and almost 80% after 2-hour treatment in all the cell lines analyzed. We also investigated whether lurbinectedin would affect RNA synthesis when added to a well-defined in vitro transcription assay containing the adenovirus major late promoter (AdMLP) as a template, in addition to TFIIB, TBP, TFIIE, TFIIF, TFIIH basal transcription factors, and RNA pol II (Pol II; ref. 21). The exposure of the DNA template to increasing amounts of lurbinectedin before adding the transcription machinery led to a progressive inhibition of RNA synthesis (Fig. 3B).
Having observed that the drug induced RNA synthesis inhibition, we next examined the fate of components involved in the transcription process overtime after lurbinectedin treatment. In A549 whole-cell extracts, we observed a rapid disappearance of both the hypo- (IIa) and hyper- (IIo) phosphorylated forms of Rpb1, the largest subunit of Pol II, in a time-dependent manner (Fig. 3C); Western blot also showed that the carboxy-terminal domain (CTD contains 52 Serine and threonine rich hepta-repeats) of Rpb1 was phosphorylated on both Serine 5 and Serine 2 (Fig. 3D), indicating that the drug treatment did not prevent the Pol II phosphorylation. Pol II disappearance was also observed in all the other human tumor cells so far tested (Supplementary Fig. S2B). We also noticed the disappearance of Rpb2, the second largest subunit of Pol II, after 15 hours, whereas the amounts of other subunits such as Rpb4 remained unchanged along the time course (Fig. 3D). Remarkably, other RNA polymerases, such as Pol I (Rpa194 subunit), Pol III (Rpc9 subunit), or mitochondrial RNA Pol, remained present in the cell extracts several hours after drug treatment (Fig. 3D). Western blot analyses showed that the drug did not affect other transcription factors, such as TFIIH (as visualized by the presence of CDK7 kinase and p62 subunits), TFIIF (β subunit), P-TEFb (CDK9 subunit) as well as TBP (a member of TFIID), all of them being involved in various steps of the RNA synthesis process (Fig. 3E).
We next studied the turnover of Pol II following lurbinectedin treatment by conventional pulse chase. A549 cells were metabolically labeled with [35S]-methionine for 1 hour and then treated with the drug. Cells were collected at different times, and Pol II was immunoprecipitated before being resolved by SDS–PAGE and autoradiographed (Fig. 3F). Newly synthesized [35S]-Pol II was detected up to 6 hours in untreated cells, whereas in drug-treated cells, Pol II labeling was almost absent after the first hour.
Phosphorylation of Pol II is essential for its degradation
To further investigate the potential connection between the phosphorylation and the degradation of Pol II after lurbinectedin treatment, we pretreated A549 cells with 20 μmol/L of DRB. This agent inhibits the CDK7 kinase subunit of TFIIH that phosphorylates serine 5 of CTD of Pol II (Fig. 4A, lanes 1 and 3), thus preventing Pol II elongation and, consequently, RNA synthesis. Pretreatment of cells with DRB prevented the Rpb1 degradation induced by lurbinectedin (Fig. 4A, lanes 2 and 4). Similar effects have been observed when A549 cells were pretreated with flavopiridol (Flv; Fig. 4A, lanes 6 and 7), a flavonoid that inhibits several cyclin-dependent kinases, including CDK9, that phosphorylate serine 2 of CTD. In parallel, confocal microscopy revealed that the amount of Pol II (in green) strongly decreased 4 hours after lurbinectedin treatment (Fig. 4B, panels control and Lur), in clear contrast to what occurred upon DRB pretreatment (Fig. 4B, panels DRB and DRB+Lur). Altogether, lurbinectedin activity required Pol II phosphorylation.
We then investigated the mechanism of lurbinectedin-induced Pol II degradation. A549 cells were treated with lurbinectedin, and, using a Pol II antibody, we performed an immunoprecipitation from whole-cell extracts over time. The immune-precipitated Pol II fraction appeared to be part of an ubiquitinated complex, as evidenced by the smear observable from 0.5 hour of treatment (Fig. 5A, lanes 2–4). Converse experiments showed that the immunoprecipitated fraction by TUBES (tandem ubiquitin binding entity) contained phosphorylated Pol II that was no more present after 2 hours of treatment (Fig. 5B, lanes 1–4).
To further investigate whether Pol II degradation induced by lurbinectedin was dependent on the ubiquitin-proteasome system, A549 cells were pretreated with PYR-41, a chemical inhibitor of UBA1, an ubiquitin-activating enzyme (30) before adding the drug. In those conditions, the phosphorylated Pol II was accumulated (Fig. 5C, lanes 2 and 4), suggesting that mono-ubiquitination, one of the first steps of the ubiquitin/proteasome degradation process, was needed for lurbinectedin activity. Pretreatment of cells with the proteasome inhibitor MG-132 similarly prevented drug-mediated degradation of Pol II (Fig. 5D, lanes 2–3 and 5–6).
The ubiquitin proteasome degradation process occurs at activated genes
The above data prompted us to investigate if the degradation of phosphorylated (elongating) Pol II that started few hours after lurbinectedin treatment was occurring in transcriptionally active genes. Western blot analysis revealed that Pol II was mainly located in the chromatin-bound fraction and not in the soluble extract several hours after drug treatment (Fig. 6A, top and bottom panels). This suggested that the degradation process might have been initiated at the gene level. The effect of the drug on gene expression in A549 cells was thus tested by monitoring the transcription of RARβ2, a retinoic acid receptor responsive gene that contains CGG triplets that are highly favorable for lurbinectedin bonding and whose expression was induced by trans-retinoic acid (t-RA). While in untreated cells, RARβ2 expression was peaking around 3 hours after t-RA treatment (Fig. 6B), in drug-treated cells, RARβ2 expression was hardly initiated and completely abolished few hours after treatment.
We next monitored the recruitment of the transcription machinery on RARβ2 by using ChIP coupled to real-time PCR. In untreated cells, ChIP assays showed that phosphorylated Pol II as well as TFIIH (as indicated by the presence of its CDK7 subunit) was abundantly recruited at the RARβ2 promoter 3 hours after t-RA treatment (Fig. 6C1); the time course paralleled the peak of RNA synthesis (Fig. 6B). On the contrary, in drug-treated cells, we detected low amounts of Pol II and TFIIH (Fig. 6D1). Interestingly, P-TEFb (detected by the presence of CDK9), which was visible at the promoter and exon 4 of RARβ2 in untreated cells, accumulated much more at exon 4 in lurbinectedin-treated cells (Fig. 6C2, D2 and C4, D4). In this case, CDK9 as well as phosphorylated Pol II (S2-P), which were present at 2 hours, disappeared at 3 hours; in contrast, they were still detected in untreated cells (Fig. 6C4), which was indicative of active RNA synthesis at that time point (Fig. 6B). The differences in the ratio between total Pol II and S2-P observed at exon 4 in untreated versus treated cells could be explained by the regulation of Pol II activity by some phospho/dephosphorylation processes (31). In lurbinectedin-treated cells, ChIP-Ubi/reChIP-Pol II showed that an ubiquitination process was already engaged on the RARβ2 as opposed to untreated cells (Fig. 6C3 and 6D3); we also noticed the presence of SUG1 (a subunit of the proteasome) at the promoter.
The above data strongly suggest that Pol II degradation by the ubiquitin-proteasome machinery is initiated in actively transcribing genes upon treatment with lurbinectedin, and only the phosphorylated elongating Pol II is degraded in the process.
DNA breaks are dependent on Pol II phosphorylation
We also examined whether the NER pathway could eliminate the lurbinectedin covalently bound to DNA triplet. A labeled DNA template (containing a single lurbinectedin adduct) was incubated with XPC/HR23B, TFIIH, XPA, RPA, XPG, and as indicated XPF/ERCC1; we did not observe the removal of the damaged oligonucleotide, but rather unexpectedly, we detected several DNA breaks on the opposite DNA strand (Fig. 7A) originated only when the XPF endonuclease was present (21). However, it was worthwhile to notice that phosphorylated histone H2AX (γ-H2AX), a hallmark of DNA breaks, was detected in lurbinectedin-treated cell extracts that coincided with the disappearance of phosphorylated Pol II (S2-P; Fig. 7B, lanes 1–7). Furthermore, pretreatment of cells with MG-132 resulted in the accumulation of S2-P, a decrease in the presence of γ-H2AX, and accumulation of XPF (Fig. 7B, lanes 8–9).
DNA breaks were next evaluated in A549 cells using the Comet assay, which is based on the neutral or alkaline lysis of labile DNA at sites of damage. DNA from cells that have accumulated DNA breaks appeared as fluorescent comets with tails of DNA fragmentation or unwinding, whereas normal, undamaged DNA did not migrate far from the origin. In lurbinectedin-treated cells, we observed a clear increase in DNA strand breaks at 4 hours when compared with untreated cells that were almost abolished in cells pretreated with either DRB or Flv (Figures 7C and D). It should be noticed that Comet assays allow the detection of both single-stranded (SSD) and double-stranded (DSB) DNA breaks. Additional assays then showed that the absence of DNA breaks was not due to DRB itself because increasing concentrations of DRB did not inhibit in vitro NER (Supplementary Fig. S3). We should notice that Comet assays also demonstrated that the level of DNA breaks induced by the drug was significantly lower in XPF-deficient cells compared with XPF-proficient cells (Fig. 7E and F). Interestingly XPF as well as γ-H2AX was significantly present after drug treatment at exon 4 of RARβ2 in lurbinectedin-treated cells compared with the untreated ones (Fig. 6C5 and 6D5). In addition, a bioChIP assay that measured the incorporation of biotinylated dUTP within broken DNA (32) also showed the presence of DNA breaks at exon 4 (Fig. 6C5 and 6D5).
Altogether, the above data strongly underline the connection between Pol II elongation (following its phosphorylation), Pol II degradation, and the generation of DNA breaks in cells treated with lurbinectedin.
Current research efforts in cancer therapy are aimed to disturb transcription, one of the fundamental processes involved in the maintenance of the oncogenic state. Here, we have studied the mechanism of action of lurbinectedin, an anticancer agent under clinical investigation with very promising results, and defined the fate of Pol II upon genotoxic attack. We first demonstrated the antiproliferative activity of lurbinectedin in several cancer cell lines with IC50 values in low nanomolar range (Fig. 1B); this potent activity was confirmed in human tumors xenografted in nude mice where lurbinectedin induced a significant inhibition of tumor growth and improvement of mice survival (Fig. 1C and D). Molecular biology approaches next showed that lurbinectedin binds specifically to CG-rich sequences located in the vicinity of the promoters of protein-coding genes (Fig. 2B). Further investigations demonstrated that lurbinectedin induced in tumor cells, a specific and irreversible degradation of Pol II that paralleled the formation of DNA breaks (Figs. 3C–E, 6D1 and D5, 7B–F). Based on these results, we propose lurbinectedin drive cancer cells to apoptosis by the mechanism of action described in Supplementary Fig. S4.
Some of the CG-rich sequences targeted by lurbinectedin (Fig. 2C) can be highly related to the regulation of gene expression by constituting CpG islands that can be methylated to silence the corresponding gene. The GC-dependent (specific) binding preference of lurbinectedin, to areas surrounding the protein-coding gene promoters, does not exclude the possibility that the drug binds also to other CG-rich sites of the genome targeted by transcription factors, such as SP1 (21, 33), or involved in rDNA transcription (34). At any rate, in tumor cells, Pol II is very active and could be arrested by the drug already bound to the DNA. Indeed, Pol II degradation occurred once the CTD of its Rpb1 subunit was phosphorylated (Fig. 3C–E). Certainly, Pol II phosphorylation by TFIIH (CDK7) and P-TEFb (CDK9) allows elongation and therefore RNA synthesis (35). Neither Pol II transcription factor partners nor other RNA polymerases were degraded (Fig. 3B–C), underlining a second level of specificity of the drug. At the same time, Comet and Biochip experiments evidenced the formation of DNA breaks in lurbinectedin-treated cells (Figs. 6D5 and 7C–F) and the presence of γ-H2AX (a DNA break marker) on the activated RARβ2 gene (Figs. 6D5 and 7B). Surprisingly, inhibition of Pol II phosphorylation with either DRB (an inhibitor of the TFIIH transcription initiation factor) or flavopiridol (an inhibitor of the P-TEFb elongation factor) prevented its degradation (Fig. 4A) and the formation of DNA breaks (Fig. 7C). Thus, it seems that, when lurbinectedin is bound to the DNA, the elongating (phosphorylated) Pol II is stalled in front of the lurbinectedin-DNA adduct, and although DNA breaks appeared on the genome (Fig. 6D5), it fails to promote the elimination of the lesion. Indeed, due to the specific interaction of lurbinectedin to both DNA strands (16), DNA breaks could have resulted from an incomplete DNA repair activity. It is likely that DNA repair factors that have been either carried by the elongating Pol II (36) or recruited following the conventional transcription coupled repair (TCR) mechanism (i.e., recruitment of NER factors by the stalled Pol II in front of a lesion; refs. 37, 38) fail to remove the lesion (Fig. 7A). Indeed, the endonuclease XPF, involved in the repair of DNA inter and intra-crosslinks (39), was found colocalizing with γ-H2AX at the RARβ2 (Fig. 6D5). Moreover, XPF-deficient cells presented much lower amounts of DNA breaks compared with their proficient counterparts (Fig. 7E and F). However, this does not exclude the possible involvement of other factors such as Topo IIα that was shown to be essential for AR nuclear receptor transactivation (40). Finally, and in agreement with our proposed model, the structural analogue of lurbinectedin that failed to bind DNA (Fig. 1F) did not produce any degradation of Pol II as well as DNA damage (Supplementary Fig. S5A), demonstrating that the DNA binding of the drug to the CG-rich regions is responsible for the disturbance of the transcription process. Similarly, in P-glycoprotein–overexpressing IGROV-ET ovarian cancer cells (28), the lower activity of lurbinectedin was correlated to the absence of RNA synthesis inhibition and Pol II degradation and to lower amounts of DNA breaks induced by the drug (Supplementary Fig. S5B).
Our results can also be used to explain the regulation of transcription mechanisms after genotoxic insults. Indeed, we can infer from them that, in the absence of DNA damage removal and subsequent restart of RNA synthesis, it is likely that there is a mechanism that restricts the amount of time that any stalled Pol II can reside on an activated gene. However, a failure of TCR could not by itself explain the Pol II degradation that also occurred following treatment of cells with α-amanitin, which binds with high affinity to the Rpb1 subunit of Pol II. It thus seems that in both cases (binding of either lurbinectedin to DNA or α-amanitin to Rpb1), degradation occurs because elongating Pol II is irreversibly stalled on the DNA template. As a consequence, the ubiquitin-proteasome degradation process will be engaged (41, 42). In the case of lurbinectedin, this is characterized by the ubiquitination of Pol II (Fig. 5A and B) and the recruitment of Sug1 (Fig. 6D3) in the vicinity of the adduct site.
Taking together, our results show how lurbinectedin, by inhibiting transcription, causes a cascade of events that can explain its potent effects on tumor cells. The relevance of Pol II degradation and formation of DNA breaks was finally confirmed in vivo (Supplementary Fig. S6). In these xenografted models, the significant inhibition of tumor growth observed after lurbinectedin treatment was correlated to both Pol II degradation and induction of DNA damage. Therefore, lurbinectedin may help in the treatment of tumors with transcription addiction. In this regard, it was recently reported that, in combination with doxorubicin, lurbinectedin has compelling activity as a second-line treatment in patients with SCLC (43), a tumor type that could be most sensitive to transcription-targeting drugs (3–5, 7, 44, 45). Lurbinectedin has also showed impressive activity in platinum-resistant ovarian cancer (46), a subtype of ovarian cancer that has been related to gene expression alterations affecting different oncogenic pathways related to drug resistance and tumor microenvironment (47–49). Targeting transcription addiction with lurbinectedin might be similarly beneficial in the treatment of other tumor types that are known to be dependent on transcription dysregulation (50, 51).
In summary, here we have described the mode of action of lurbinectedin for cancer targeting. This drug inhibits the transcription process by (1) its binding to CG-rich sequences, mainly located around promoters of protein-coding genes; (2) the irreversible stalling of elongating RNA Pol II on the DNA template and its specific degradation by the ubiquitin/proteasome machinery; and (3) the generation of DNA breaks and subsequent apoptosis. These results also improved our understanding of an important transcription regulation mechanism.
Disclosure of Potential Conflicts of Interest
C.M. Galmarini and P. Aviles are Senior Managers and have ownership (including patents) in PharmaMar. G. Santamaría Nuñez and J.F. Martínez-Leal are PharmaMar employees. J.-M. Egly reports receiving commercial research grant from PharmaMar. No potential conflicts of interest were disclosed by the other authors.
Conception and design: G. Santamaría Nuñez, C. Giraudon, P. Aviles, C.M. Galmarini, J.-M. Egly
Development of methodology: G. Santamaría Nuñez, C.M. Genes Robles, E. Compe, J.-M. Egly
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): G. Santamaría Nuñez, C.M. Genes Robles
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): G. Santamaría Nuñez, C.M. Genes Robles, C. Giraudon, J.F. Martínez-Leal, E. Compe, F. Coin, P. Aviles, C.M. Galmarini, J.-M. Egly
Writing, review, and/or revision of the manuscript: G. Santamaría Nuñez, C.M. Genes Robles, C. Giraudon, J.F. Martínez-Leal, E. Compe, P. Aviles, C.M. Galmarini, J.-M. Egly
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C. Giraudon, J.F. Martínez-Leal
Study supervision: C. Giraudon, J.F. Martínez-Leal, P. Aviles, C.M. Galmarini, J.M. Egly
J.M. Egly has been awarded with the following grants: ERC Advanced grant, l′Agence Nationale de la Recherche (N#ANR1 08MIEN-022-03), l′Association de la Recherche contre le Cancer (SL22013060782), the Institut National du Cancer (INCA-2008-041), and Ligue Nationale contre le Cancer.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank María José Guillen-Navarro, María José Muñoz, Jose Manuel Molina-Guijarro, Cathy Braun, and the IGBMC facilities for their expertise and technical support. We also are grateful to Carmen Cuevas Marchante, José María Fernández Sousa Faro, and Nicolas Le May for fruitful discussions. Sequencing and bioinformatics analysis (with the expertise of Tao Ye and Baptiste Bidon) were performed by the IGBMC Microarray and Sequencing platform, a member of the “France Génomique” consortium (ANR-10-INBS-0009).
Note: Supplementary data for this article are available at Molecular Cancer Therapeutics Online (http://mct.aacrjournals.org/).
G. Santamaría Nuñez and C.M. Genes Robles are first coauthors.
- Received March 28, 2016.
- Revision received May 23, 2016.
- Accepted June 15, 2016.
- ©2016 American Association for Cancer Research.