ZNF432 stimulates PARylation and inhibits DNA resection to balance PARPi sensitivity and resistance

Abstract

Zinc finger (ZNF) motifs are some of the most frequently occurring domains in the human genome. It was only recently that ZNF proteins emerged as key regulators of genome integrity in mammalian cells. In this study, we report a new role for the Krüppel-type ZNF-containing protein ZNF432 as a novel poly(ADP-ribose) (PAR) reader that regulates the DNA damage response. We show that ZNF432 is recruited to DNA lesions via DNA- and PAR-dependent mechanisms. Remarkably, ZNF432 stimulates PARP-1 activity in vitro and in cellulo. Knockdown of ZNF432 inhibits phospho-DNA-PKcs and increases RAD51 foci formation following irradiation. Moreover, purified ZNF432 preferentially binds single-stranded DNA and impairs EXO1-mediated DNA resection. Consequently, the loss of ZNF432 in a cellular system leads to resistance to PARP inhibitors while its overexpression results in sensitivity. Taken together, our results support the emerging concept that ZNF-containing proteins can modulate PARylation, which can be embodied by the pivotal role of ZNF432 to finely balance the outcome of PARPi response by regulating homologous recombination.

Graphical Abstract

Figure 1.

Identification of potential resection regulators in a subset of ZFPs identified by proteome-wide microarray analysis of PAR readers. HeLa cells transfected with control or gene-specific siRNAs against 7 selected ZFPs were irradiated (5 Gy) and screened for RAD51 (A) and BrdU (B) foci formation 3 h post-irradiation. siRNA-mediated PARP-1 knockdown and inhibition with 1 μM BMN673 for 16 h were taken as positive controls. The data presented in the graphs are the median of three independent experiments. The effects of ZNF432 depletion on RAD51 (C), BrdU foci formation (D), pRPA (E) and γ-H2AX (F), are shown. S-phase cells were identified by selecting Geminin for RAD51 and γ-H2AX foci analysis or EdU Click-iT labeling of EdU incorporated into cells for pRPA foci analysis. The error bars represent ± s.d. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001 (Mann–Whitney U-test).

Figure 2.

Loss of ZNF432 decreases NHEJ. The effect of ZNF432 knockdown in HeLa cells following IR-induced damage on (A) phospho-DNA-PKcs, (B) 53BP1 and (C) RIF1 foci formation. The cells were selected for S-phase by using EdU Click-iT (A) or S-G2-phase cells with Geminin staining (B, C). The error bars indicate ± s.d. of three independent experiments. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001 (Mann–Whitney U-test). (D) The attenuation of ZNF432 decreases NHEJ. EJ5-GFP HEK293T cells were transfected with pCAG.I-Sce1 and the percentage of GFP-positive cells was evaluated 48 h later by fluorescence microscopy. The siControl and DMSO conditions were normalized to 1.0. The error bars represent s.d. from three independent experiments. **P ≤ 0.01, *** P ≤ 0.001 (Student's t-test).

Figure 3.

Recruitment of ZNF432 to DNA damage depends on PARP-1 and PAR synthesis. (A) ZNF432 protein structure. (B) Cellular localization of ZNF432. Green—GFP-ZNF432 full length, blue—DNA (Hoechst). (C) Graph and representative images showing the recruitment of full-length GFP-ZNF432 and its subdomains GFP-ZNFs and GFP-KRAB in HEK293T cells following the damage induced by laser-microirradiation. (D) Effect of inhibition of PAR (1 μM BMN673 for 16 h) and loss of PARP-1 (PARP-1 KO) on recruitment of ZNF432 (GFP-ZNF432) to DNA lesions in HEK293T cells, following the damage induced by laser-microirradiation. The error bars represent ± SEM of three independent experiments.

Figure 4.

ZNF432 interacts with PAR and PARP-1. (A) Sodium dodecyl sulfate-polyacrylamide gel electrophoresis of purified human ZNF432 protein. (B) Overlay assay showing the affinity of ZNF432 for PAR in vitro. Five pmoles of recombinant human ZNF432 were slot-blotted on a nitrocellulose membrane and probed with 100 nM protein-free DHBB-purified PAR. Histone H1 and DNAse I were respectively used as positive and negative PAR binding controls. The presence of PAR bound to proteins was revealed by western blot (WB) with the anti-PAR antibody clone 10H. SyproRuby protein blot staining was used as a loading control. (C) GFP-ZNF432 interacts with endogenous PARP-1. (D) Experimental design for the identification of ZNF432 protein interaction networks by affinity-purification coupled to mass spectrometry (AP-MS). (E) Venn diagram of overlapped and unique ZNF432-associated proteins identified by AP-MS. The number of proteins in the overlapping and non overlapping area are shown. Whole cell extracts of wild-type (WT) and PARP-1 depleted HEK 293T cells (PARP-1 KO) were prepared following exposure to hydrogen peroxide (H₂O₂) or left untreated. Refer to Supplementary Table S6 for a complete listing of proteins in each area. (F) ZNF432 resides in a core protein complex in which TRIM28/KAP1 acts as a central hub. The interaction network (derived from the STRING-DB website) shows some of the most abundant ZNF432 interaction partners. (G) ZNF432 protein interaction network generated from 139 proteins assigned to DNA damage response and chromatin remodeling pathways. Nodes with high connectivity in the protein interaction network were isolated. The edges indicate both functional and physical protein associations. Line thickness indicates the strength of data support.

Figure 5.

ZNF432 promotes PARP-1 assembly and activation. (A) Left: Knockdown of ZNF432 reduces PARylation at localized DSBs. i265 U2OS cells were transfected with control or ZNF432 siRNAs for 7 days before being analyzed at 1 h after DSB induction. The PAR signal is shown in green, while the DSB sites were marked by mCherry-FokI in red. Nuclei were stained by Hoechst 33342 in blue. Right: A violin plot displaying the quantification of relative PAR intensity in siControl or siZNF432 cells in the experimental setting of Figure 5A. The dotted line indicates the mean. The slide line below the mean labels the 1st quartile and the slide line above the mean demarcate the 3rd quartile, respectively. P-values were calculated by unpaired two-tailed Student's t-test as shown. (B) i265 U2OS cells transfected with siRNAs (control siRNA and ZNF432 smart pool) were processed at 1 h post-DSB induction for immunofluorescence. The violin plot represents the quantification of the relative intensity of PAR, PARP-1 and γ-H2AX foci on the array. (C) Overexpression of GFP-ZNF432 increase PARylation as measured with a PBZ-mRuby PAR sensor. (D) Monitoring of the mRuby-PAR relative fluorescence intensity over time. (E) ZNF432 stimulates PARP-1 activity. Overexpression of GFP-ZNF432 increases intracellular PAR levels under basal conditions and in IR-treated cells. Whole cell extracts of ZNF432-expressing cells were analyzed for the presence of PAR by western blot. (F) Western blot showing the in vitro automodification pattern of PARP-1 in the presence of full length human ZNF432 (1, 2 and 5 pmoles). PARP-1 was incubated with ZNF432, calf-thymus activated DNA and NAD+. Reaction products were resolved by SDS-PAGE and PAR polymers were identified by western blot. A polyclonal antibody against PAR (96–10) was used in both blots E and F to reveal the presence of PAR polymers. (G) The ZNFs region of ZNF432 stimulates PARP-1 activity. The Western blot shows the in vitro auto modification pattern of PARP-1 in the presence of bovine serum albumin (BSA, control), full length ZNF432 (1–652), the N-terminal fragment containing the KRAB domain (1–205) and the C-terminal domain containing the array of 16 ZNFs repeats (205–652). PAR polymers were revealed by Western blot analysis using a polyclonal antibody against MAR/PAR (E6F6A). (H) ZNF432 and HPF1 produce an additive stimulatory effect on PARP-1 activity. PARP-1 was auto modified in vitro in the presence of ZNF432, HPF1 and both proteins. Reaction products were subjected to 1M hydroxylamine (HA) hydrolysis to cleave Asp/Glu-linked ADP-ribosylation or PARG treatment to erase PAR polymers. PAR polymers were revealed by Western blot analysis using a polyclonal antibody against PAR (96–10).

Figure 6.

ZNF432 preferentially binds to single-stranded DNA. Purified ZNF432 DNA binding capabilities were monitored on the following labeled DNA substrates: (A) double-strand DNA, (B) single-strand DNA. (C) Competition bandshift assays using the single-strand DNA probe in competition with the double-strand DNA probe. All assays were done at 37ºC with glutaraldehyde fixation before running on 1×-TBE–acrylamide gels. The error bars indicate ± SEM. of three independent experiments and images are representative. (D) K_d and B_max values of full-length ZNF432 bound to single-strand DNA.

Figure 7.

ZNF432 is an inhibitor of EXO1-mediated DNA resection. In vitro resection assay using 2.7 Kb double-strand [α-³²P] radiolabeled DNA probe and EXO1 (6.5 nM) with (A) increasing concentrations of ZNF432 (0, 10, 20, 30 and 40 nM) and (B) with ZNF432 (20 nM) in presence or absence of 250 nM purified PAR. All images are representative of three independent biological repeats. The error bars indicate ± SEM. *P = 0.05. (C) Recruitment kinetics for GFP-EXO1 WT in U2OS cells transfected with siCTRL or siZNF432 to laser-induced DSBs. Mean curves ± SEM (n = 145) are shown. (D) ER-AsiS1 U2OS cells were knocked down for ZNF432 for 72 h, followed by induction with 300 nM 4-hydroxytamoxifen for 4 h and quantification of ssDNA. The error bars indicate ± s.d. from four independent experiments. *P ≤ 0.05, *** P ≤ 0.001 (Student's t-test).

Figure 8.

Loss of ZNF432 leads to resistance to BMN673. (A) Effect of BMN673 (0–6.4 μM) on survival of U2OS-Control and U2OS-ZNF432-KD cells. Effect of BMN673 (0–6.4 μM) on survival of cells overexpressing GFP or GFP-ZNF432: (B) U2OS-Control, (C) U2OS-ZNF432-KD cells. The error bar indicates ± s.d. of three independent experiments. The P-value was calculated by the unpaired t-test. (D) Expression of ZNF432 in ovarian normal (n = 110) cancer (n = 1516) tissue. The P-value was calculated by an unpaired t-test. (E) Each population (normal/endometroid) were separated based on their mean ZNF432 expression level. Comparison of the healthy population with high ZNF432 expression to the endometroid population with relatively high ZNF432 expression (left) and analysis for the lower expressing subpopulations in both groups (right). The error bar indicates ± s.d. The P-value was calculated by an ordinary one-way ANOVA test. The expression data was obtained from http://gent2.appex.kr/gent2/ (accessed on 24 January 2022). *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001.

Figure 9.

Modulation of ZNF432 sensitizes the PARPi resistant ovarian cancer cell line to PARP inhibition. Effect of BMN673 on viability following ZNF432 overexpression (A, E) or ZNF432 KD (C, G) in COV362 and COV362 Olaparib resistant (COV362-R) cells respectively. Effect of AZD2281 survival following ZNF432 overexpression (B, F) or ZNF432 KD (D, H) in COV362 and COV362-R cells. The error bars indicate ± s.d. from three independent experiments. The P-value was calculated by the unpaired t-test. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001. (I) Knockdown of ZNF432 in COV362 53BP1^−/− cells leads to additive resistance to Olaparib. The error bars indicate ± s.d. from three independent experiments.

Figure 10.

Model. In wild-type cells, PARP-1 is recruited and activated at DNA damage sites, which then engage ZNF432 to stimulate PARylation. ZNF432 promotes NHEJ and inhibits DNA end resection by binding ssDNA, although to a certain extent, so the cells can perform DNA repair. If ZNF432 is depleted, ZNF432 is no longer recruited to DSBs, it cannot activate PARP-1 and protect ssDNA, a loss of 53BP1 and RIF1 foci occurs, leading to DSB hyper-resection. This enhances PARPi resistance. Conversely, overexpression of ZNF432 downregulates DNA end resection and therefore sensitizes cells to PARPi.