Trapping of PARP1 and PARP2 by Clinical PARP Inhibitors

Abstract

Small-molecule inhibitors of PARP are thought to mediate their antitumor effects as catalytic inhibitors that block repair of DNA single-strand breaks (SSB). However, the mechanism of action of PARP inhibitors with regard to their effects in cancer cells is not fully understood. In this study, we show that PARP inhibitors trap the PARP1 and PARP2 enzymes at damaged DNA. Trapped PARP-DNA complexes were more cytotoxic than unrepaired SSBs caused by PARP inactivation, arguing that PARP inhibitors act in part as poisons that trap PARP enzyme on DNA. Moreover, the potency in trapping PARP differed markedly among inhibitors with niraparib (MK-4827) > olaparib (AZD-2281) >> veliparib (ABT-888), a pattern not correlated with the catalytic inhibitory properties for each drug. We also analyzed repair pathways for PARP-DNA complexes using 30 genetically altered avian DT40 cell lines with preestablished deletions in specific DNA repair genes. This analysis revealed that, in addition to homologous recombination, postreplication repair, the Fanconi anemia pathway, polymerase β, and FEN1 are critical for repairing trapped PARP-DNA complexes. In summary, our study provides a new mechanistic foundation for the rational application of PARP inhibitors in cancer therapy.

Conflict of interest statement

Conflict of interest: The authors declare that they have no conflict of interest.

©2012 AACR.

Figures

Figure 1. PARP1 is required for cell killing by olaparib

(A) Western blot of whole cell lysates prepared from wild-type and PARP1−/− DT40 cells. Blots were probed with the indicated antibodies. (B) Nuclear localization of PARP1 in wild-type DT40 cells and loss of signal in PARP1−/− DT40 cells. Immunofluorescence microscopy was performed with anti-PARP1 antibody and nuclei were stained with DAPI. (C) PAR level in wild-type and PARP1−/− DT40 cells measured by ELISA. Cells were treated with the indicated olaparib concentrations for 2 hours. (D) Survival curves after continuous olaparib treatment for 72 hours. Cellular ATP activity was used to measure cell viability. The survival of untreated cells was set as 100%. Error bars represent standard deviation (SD) (n=3). (E) Cell cycle analyses after olaparib treatment. The indicated cell lines were treated with 10 μM olaparib for 24 hours; after which, cells were analyzed by flow cytometry. The left and right peaks indicate G1 and late S and G2 populations, respectively. (F) DNA damage measured by immunofluorescence microscopy with anti-γH2AX antibody after 2 hours treatments with or without 10 μM olaparib. Left panels show representative images and right plot shows a quantitation of the number ofγH2AX foci per individual cells. The number of cells analyzed (N) and the average number ofγH2AX foci (Ave.) are shown above the plot.

Figure 2. Olaparib stabilizes PARP1-DNA complexes that are more toxic than unrepaired SSBs

(A) Western blot analysis of nuclear soluble and chromatin bound fractions prepared from wild-type DT40 cells. Cells were treated for 30 min without drug (lane 1 and 7) or as indicated. Blots were probed with indicated antibodies. Histone H3 and topoisomerase I were used as positive markers for chromatin and nuclear soluble fractions, respectively. The asterisk indicates a non-specific band. (B) Quantification of PARP1 levels in chromatin binding fractions 30 min after the indicated drug treatments. Signal intensity was quantified using Image J software (NIH) from 4–7 independent Western blot analyses (mean values ± SD), and normalized to untreated cells set as one. (C) Survival curves of wild-type (left) and PARP1−/− (right) cells treated with MMS alone or with the indicated olaparib concentrations (indicated beside each curve in micromolar unit). The survival of treated cells without MMS was set as 100%. Data are mean ± SD (n=3). (D) Cell cycle analysis of wild-type and PARP1−/− DT40 cells after 2 and/or 4 hour continuous treatments. Cells were pulse-labeled with BrdU for 10 min following the drug treatment to measure DNA synthesis.

Figure 3. Stabilization of toxic PARP1- and PARP2-DNA complexes by olaparib

(A) Survival curves of wild-type, PARP1−/−, PARP1−/− complemented with human PARP1 (PARP1−/−;hPARP1) and PARP2 (PARP1−/−;hPARP2) DT40 cell lines treated with olaparib. Cells were treated with olaparib as in Figure 1D. Error bars represent SD (n=3). (B–E) Western blotting of nuclear soluble and chromatin bound fractions prepared from human DU145 cells. Blots were probed with the indicated antibodies. (B) Cells were treated for 4 hours without drug (lane 1 and 7), or as indicated. (C) Rapid reduction of PARP1 and PARP2 binding to chromatin after washing out olaparib. Cells were treated for 4 hours without drug or with 10 μM olaparib in the presence of 0.01% MMS. Drugs were removed by washing cells with cold PBS three times, followed by incubation in pre-warmed medium for the indicated times. (D) Control siRNA (siCtrl), PARR1 siRNA (siP1) and PARP2 siRNA (siP2) were transfected into DU145 cells. Three days later, cells were treated for 4 hours with10 μM olaparib in the presence of 0.01% MMS. (E) Reduced cytotoxicity of olaparib in DU145 cells downregulated for PARP1. Cells were transfected with the indicated siRNAs for 60 hours before adding olaparib for an additional 72 hours. Sensitivity was determined as Figure 1D. Error bars represent SD (n=3).

Figure 4. Three clinical PARP inhibitors differ in their potency to poison PARP1 and PARP2 irrespective of their potency to inhibit PARP catalytic activity

(A) Drug-induced PARylation inhibition. Western blotting of PAR levels in whole cell lysates from wild-type DT40 cells treated as indicated for 30 min. Blots were probed with anti-PAR antibody. Whole cell lysates from untreated PARP1−/− cells were used as a control. Asterisks indicate non-specific bands. (B) Representative PAR ELISA assays in DT40 cells treated with olaparib, veliparib or MK-4827 for 120 min. PAR level of untreated cells was set as 100%. IC50 (inhibitory concentration 50%) of olaparib, veliparib and MK-4827 were 1.2 nM, 10.5 nM and 50.5 nM, respectively. (C–F) Differential cytotoxicity of the three PARP inhibitors. Viability was measured as in Figure 1D. Error bars represent SD (n=3). Survival curves of the wild-type cells (C), PARP1−/− cells (D), wild-type cells treated with a subtoxic MMS concentration (0.00025%; see Fig 2C) (E), and BRCA2-deficient DT40 cells (F). Sensitivity was determined as in Figure 1D. Error bars represent SD (n=3). The viability of PARP1−/− cell to 0.00025% MMS (22%) was shown with a horizontal dashed line with annotation (E). (G) Survival curves of DU145 treated with the indicated PARP inhibitors for 72 hours. (H) Survival curves of DU145 treated with MMS alone or MMS plus 1 μM olaparib or veliparib or MK-4827 for 72 hours.

Figure 5. Differential cellular trapping of PARP1 and PARP2 by clinical PARP inhibitors

(A) Western blotting of nuclear soluble and chromatin bound fractions prepared from wild-type DT40 cells (upper) and DU145 cells (lower) with or without drug treatment as indicated. Blots were probed with indicated antibodies. (B,C) Fractionation of PARP-DNA complexes stabilized by PARP inhibitors. DU145 cells were treated for 4 hours without drug (lanes 1–5) or as indicated. (B) Scheme for the preparation of indicated fractions (P1, A–D) using different stringency buffers (Supplementary methods). (C) The eluted proteins in each fraction were blotted with indicated antibodies. Black arrows indicate monoubiqutinated PCNA.

Figure 6. Differential biochemical trapping of PARP1 by clinical PARP inhibitors

(A) Scheme of the fluorescence anisotropy (FA) binding assay. The star indicates the site labeled on the DNA substrate with Alexa Fluor488. Unbound nicked DNA substrate rotates fast and gives low FA. PARP1 binding to the substrate slows the rotation and gives high FA. Addition of NAD+ leads to PARP1 dissociation from DNA due to autoPARylation. (B) Concentration-dependent PARP1-DNA association using the three different PARP inhibitors. FA was measured 60 min after adding NAD+. (C) Time-course of PARP1-DNA dissociation in the presence of the three different PARP inhibitors (1 μM each). The absence of PARP inhibitors immediately reduces PARP1-DNA complexes (DMSO control). In the absence of NAD+, PARP1-DNA complexes remained stable for at least 120 min (no NAD+).

Figure 7. DNA repair pathways for PARP inhibitors

(A) Sensitivity profiles of olaparib, veliparib and MK-4827 in a panel of thirty DT40 isogenic repair mutant DT40 cell lines (see Table S1 for additional information on the cell lines). The sensitivity in each mutant cell line was determined relative to wild-type cells (see Fig. 1D). Negative and positive scores indicate that a given cell line is hypersensitive or resistant to the drug, respectively. Each bar length reflects the degree of sensitivity or resistance to the drug (log2 units). Each bar is colored according to the category of DNA repair function (see Table S1): black: non-homologous end joining; pink: checkpoint; dark blue: homologous recombination (HR); brown: post replication repair; light blue: PARP1 and PARP1/RAD18; light green; removal of Top1 or Top2 cleavage complexes; magenta; Fanconi anemia (FA) pathway; orange: DNA polymerases; purple; FEN1; red; nucleotide excision repair; green: RecQ helicases. The IC90 for olaparib and MK-4827 and IC50 for veliparib are shown at the bottom (mean ± SD, n = 13). (B and C) Schematic illustration of the dual inhibitory mechanism of PARP inhibitors (poisoning and catalytic inhibition). (B) Molecular interaction map showing the regulatory pathways for the formation and dissociation of PARP-DNA complexes. Symbols are: Binding of PARP and DNA with the node corresponding to the PARP-DNA complex; Activation; Inhibition (50). PARP inhibitors enhance the PARP-DNA complexes by two mechanisms: 1: inhibition of NAD+ binding, and 2: binding of PARP inhibitor in the NAD site activates (allosterically) the binding of PARP to DNA. See discussion for further details. (C) Dual cytotoxic mechanisms of PARP inhibitors. 1: Catalytic inhibition (upper pathway) interferes with the repair of SSBs, leading to replication fork damage that requires homologous recombination (HR) repair. 2: Trapping of PARP-DNA complexes also leads to replication fork damage but utilizes additional repair pathways including Fanconi pathway (FA), template switching (TS), ATM, FEN1 (replicative flap endonuclease) and polymerase β.