The chemotherapeutic agent CX-5461 irreversibly blocks RNA polymerase I initiation and promoter release to cause nucleolar disruption, DNA damage and cell inviability

Abstract

In the search for drugs to effectively treat cancer, the last 10 years have seen a resurgence of interest in targeting ribosome biogenesis. CX-5461 is a potential inhibitor of ribosomal RNA synthesis that is now showing promise in phase I trials as a chemotherapeutic agent for a range of malignancies. Here, we show that CX-5461 irreversibly inhibits ribosomal RNA transcription by arresting RNA polymerase I (RPI/Pol1/PolR1) in a transcription initiation complex. CX-5461 does not achieve this by preventing formation of the pre-initiation complex nor does it affect the promoter recruitment of the SL1 TBP complex or the HMGB-box upstream binding factor (UBF/UBTF). CX-5461 also does not prevent the subsequent recruitment of the initiation-competent RPI-Rrn3 complex. Rather, CX-5461 blocks promoter release of RPI-Rrn3, which remains irreversibly locked in the pre-initiation complex even after extensive drug removal. Unexpectedly, this results in an unproductive mode of RPI recruitment that correlates with the onset of nucleolar stress, inhibition of DNA replication, genome-wide DNA damage and cellular senescence. Our data demonstrate that the cytotoxicity of CX-5461 is at least in part the result of an irreversible inhibition of RPI transcription initiation and hence are of direct relevance to the design of improved strategies of chemotherapy.

Figure 1.

CX-5461 inhibits de novo rRNA synthesis and disrupts subnucleolar organization. (A) 30-min metabolic RNA labelling in NIH3T3 MEFs after 15-min incubation of cells with the indicated concentrations of CX-5461 as analysed by electrophoresis. (B) Quantitation of uridine incorporation into 47S rRNA derived from two biological replicates. RNA gel fractionations as in (A) and normalized to bulk 28S levels. The standard error of the mean (SEM) is indicated. (C) Comparative analysis of 47S synthesis in NIH 3T3 and E14 mESCs treated with 10 μM CX-5461, 40 nM Act-D or mock treated and analysed as in (B) except that [³H]-labelled uridine was present from the start of the drug treatments and RNA was recovered 15 or 30 min later. Data derived from two biological replicates per cell type and the SEM is indicated. (D) Examples of nuclear distributions of fibrillarin, a 47S processing factor (grey) and UBF, indicating the transcriptionally active rDNA (red), before and after treatment with 10 μM CX-5461 as compared with their distributions in MEFs depleted for the essential RPI-associated factor Rrn3 (29). Single confocal planes from indirect IF microscopy are shown and areas enclosed by a rectangle are shown at higher resolution at the top right of each pane. Nuclear DNA (DAPI) is shown superimposed (blue). See Supplementary Figure S1 for a time course of CX-5461-induced changes in nuclear/nucleolar fibrillarin and UBF distributions.

Figure 2.

CX-5461 inhibits initiation but not recruitment of RPI–Rrn3. (A) Graphical representation of qPCR amplicons within the mouse rDNA repeat unit. (B, C) ChIP-qPCR analysis, respectively, of RPI and Rrn3 interactions across the rDNA unit of NIH3T3 cells after 15 min exposure to either 2 or 10 μM CX-5461 or mock treated by the addition of vehicle (NaH₂PO₄). Standard errors were <0.1 and <0.03 for RPI and Rrn3, respectively, and so have been omitted. (D) DChIP-seq profiles of RPI and Rrn3 interactions across the NIH3T3 rDNA unit after treatment with 10 μM CX-5461, 40 nM ActD or mock treatment for 15 min. The lower panel shows an enlargement of the region boxed in the upper panel. The key elements of the rDNA repeat are indicated graphically below both panels. The scale of ChIP enrichment in (D) is indicated at the upper right of each lane. In (A) to (D), SpPr refers to the Spacer promoter [yellow in (D)]; Tsp, Spacer promoter adjacent terminator [red in (D)]; T₀, 47S promoter proximal terminator; 47SPr, 47S pre-rRNA promoter; 47S 5′, amplicon covering 5′ base pairs 159–320 of the 47S pre-rRNA coding region; ETS, external transcribed sequence; 18S, 5.8S and 28S indicate structural rRNA coding regions; T_1–10, 47S termination elements; and T_1–3, the qPCR amplicon for this region.

Figure 3.

De novo RPI–Rrn3 initiation is arrested by CX-5461 but ongoing transcription elongation is not affected. (A) Time course of DChIP-seq rDNA interaction profiles for RPI and Rrn3 during the first 15 min of inhibition by 10 μM CX-5461 in NIH3T3 cells. The scale of ChIP enrichment is indicated at the upper left of each lane. The key elements of the rDNA repeat are indicated and labelled as in Figure 2D. Supplementary Figure S3A shows the same data but across the full rDNA unit and includes a 30 min time point. The elongation rate in the presence of CX-5461 was determined from the positions of the RPI boundaries formed after 2 and 5 min of drug treatment. Detailed treatments are given in Supplementary Figure S3D and E, but briefly the boundary positions were defined by sigmoidal curve fits to the RPI ChIP-seq datasets giving initial elongation rates in the presence of CX-5461 of 13.6 and 25.6 nucleotides/s, respectively, at 2 and 5 min. These initial rates did not account of the delay in drug uptake, but by assuming the elongation rate should remain constant during the first 5 min of drug treatment, this could be estimated as 71 s, giving a true elongation rate of 33.5 nucleotides/s. Similar estimates were made using linear curve fits to RPI levels in independent ChIP-qPCR experiments. This approach gave an independent estimate for elongation as 39 nucleotides/s and a drug uptake delay of 56 s. Taking the two independent estimates gave a mean elongation rate of 37 ± 3 nucleotides/s and a drug uptake delay of 64 ± 8 s. (B) High-resolution profiles of RPI and Rrn3 across the 47S promoter after mock treatment or 15-min treatment with 10 μM CX-5461. Gaussian curve fits (dotted lines) were used to estimate the indicated central positions of the factors relative to the 47S initiation site. It should be noted that for the ‘mock treatment’ the distribution of RPI and Rrn3 at the 47S promoter continues into the adjacent 47S coding region; hence, the curve fit was only to the 5′ rise in RPI and Rrn3 loading profile (solid line). The positions of the two functional 47S promoter sequence elements, the upstream element (UPE) and core (yellow), and the adjacent T₀ termination site (red) are indicated below each panel. Supplementary Figure S3C shows these data side by side with mapping of UBF and TAF1B.

Figure 4.

CX-5461 inhibits RPI transcription initiation but not elongation in vitro. Cell-free transcription assays were carried out using a template containing the mouse rDNA 47S promoter fragment from −168 to 292 bp relative to the 47S initiation site in which bases +1 to +34 bp lacked G residues on the non-coding (top) strand. The template was linearized within the adjacent pUC9 vector sequence to give an RNA run-off product of 320 bases. An initial reaction was performed in the absence of GTP during which transcription was arrested at the first G-residue, giving a 34-base product. Subsequent addition of GTP to the reaction (chase reaction) allowed transcripts to elongate to the end of the template, and in the absence of CX-5461 (CX) also allowed multiple rounds of initiation/elongation that could be prevented by addition of heparin (compare lane 5 with lanes 6–8). Addition of 10 μM CX-5461 during the initial reaction eliminated synthesis of both 34-base and full-length transcripts but its addition during the GTP chase reaction had no effect on the efficiency of elongation of 34-base transcripts to full length. The asterisk in lanes 3 and 4 indicates a degree of readthrough at the end of the G-less cassette, bands corresponding to the positions of subsequent G-residues.

Figure 5.

CX-5461 irreversibly disrupts rDNA transcription and nucleolar organization. (A) Timeline of ‘CX-pulse’ experiments. (B) NIH3T3 cells were subjected to a CX-pulse (5 min, 10 μM CX-5461) or mock treated and then incubated for varying times in the absence of drug and finally subjected to RNA metabolic labelling with [³H]-uridine for 30 min. The left panel shows electrophoretic analysis of labelled RNAs and the right panel shows the quantitation of [³H]-uridine incorporation into 47S rRNA normalized to bulk 28S levels (EtBr 28S). The error bars represent the SEM for two biological duplicates. (C) ChIP-qPCR analysis of RPI and Rrn3 interactions across the rDNA unit of NIH3T3 cells at various times after a CX-pulse or mock as in (A). The amplicons sampled are those shown graphically in Figure 2A and labelled as in the previous figures. To aid interpretation and avoid clutter, only data from a subset of time points are shown and error bars have been omitted. The data are the mean of three biological replicates and are shown in full in Supplementary Figure S5A along with the SEM for each data point. (D) A comparative histogram of RPI and Rrn3 interactions over the 47S 5′ amplicon [shaded in (C)] at different times after the CX-pulse treatment. The data are taken from the interaction curves in (C) and Supplementary Figure S5A and the SEM is indicated by vertical bars. (E) Left panel: epifluorescence imaging of de novo RNA synthesis (EU 30-min labelling) and of RPI in NIH3T3 at 90 min after a 5-min CX-pulse or mock treatment. Arrows indicate residual EU incorporation into RNA corresponding with RPI foci after the CX-pulse. Right panel: quantitation of EU incorporation within the nucleoplasmic and the nucleolar volumes. An average of 50 nuclei per time point from 3D image stacks were subjected to volumetric analysis using Volocity software (Quorum Technologies) and the SEM for each dataset indicated (see the ‘Materials and Methods’ section). A complete time course of imaging is shown in Supplementary Figure S5.

Figure 6.

Arrest of 47S pre-RNA synthesis by CX-5461 leads to DNA damage and arrest of replication. (A) Epifluorescence imaging of NIH3T3 cells treated with a 5-min 10 μM CX-pulse and probed for γH2A.X and 53BP1 DNA damage foci and for fibrillarin at time points after CX-5461 removal. Arrows indicate colocalization of 53BP1 foci and fibrillarin (see Supplementary Figure S6 for a more complete image set). (B) The left panel shows the fraction of γH2A.X positive nuclei and the right panel shows mean number of dominant 53BP1 foci per nucleus at different time points after the CX-pulse. Wide-field 3D image stacks from three biological replicas were subjected to volumetric analysis using Volocity software (Quorum Technologies) (see the ‘Materials and Methods’ section). Staining intensity >250% above the Otsu intensity threshold (41) was considered positive. Dominant 53BP1 foci were also defined as having volumes between 1 and 10 μm³. More than 150 nuclei were analysed per experimental time point and the SEM for each is indicated. (C) Relative nuclear EdU incorporation at time points after the CX-pulse determined from staining intensities in wide-field images, examples of which are shown in Supplementary Figure S7. Quantitation of EdU incorporation was again obtained using the Volocity software (Quorum Technologies) to analyse ≥5 image fields, and an average of 40 nuclei per time point and treatment; the SEM is shown. (D) Western blot analysis of NIH3T3 total cellular protein for the expression of p53, S15-phospho p53 and p21 at indicated time points after the CX-pulse. Tubulin was used as an internal loading standard.