PARP1-TRIM44-MRN loop dictates the response to PARP inhibitors

Abstract

PARP inhibitors (PARPi) show selective efficacy in tumors with homologous recombination repair (HRR)-defects but the activation mechanism of HRR pathway in PARPi-treated cells remains enigmatic. To unveil it, we searched for the mediator bridging PARP1 to ATM pathways by screening 211 human ubiquitin-related proteins. We discovered TRIM44 as a crucial mediator that recruits the MRN complex to damaged chromatin, independent of PARP1 activity. TRIM44 binds PARP1 and regulates the ubiquitination-PARylation balance of PARP1, which facilitates timely recruitment of the MRN complex for DSB repair. Upon exposure to PARPi, TRIM44 shifts its binding from PARP1 to the MRN complex via its ZnF UBP domain. Knockdown of TRIM44 in cells significantly enhances the sensitivity to olaparib and overcomes the resistance to olaparib induced by 53BP1 deficiency. These observations emphasize the central role of TRIM44 in tethering PARP1 to the ATM-mediated repair pathway. Suppression of TRIM44 may enhance PARPi effectiveness and broaden their use even to HR-proficient tumors.

Graphical Abstract

Graphical Abstract.

Introduction

Failure in DNA repair results in genomic instability and contributes to the development of cancer (1). DNA double-strand breaks (DSBs) are considered as the most deleterious DNA damage (2). Poly (ADP-ribose) polymerase 1 (PARP1) is an initial responder that detects DNA strand breaks. PARP1 at damaged chromatin synthesizes multibranched poly (ADP-ribose) (PAR) chains, primarily targeting itself (3). The resulting PAR chains serve as a platform for the recruitment of downstream repair factors, instigating the base excision repair (BER) at DNA single-strand breaks (SSBs) (4). The dynamic turnover of PAR is executed by the primary PAR-degrading enzyme, poly (ADP-ribose) glycohydrolase (PARG) (5). PARP1 is also rapidly activated by DSBs. The MRE11–RAD50–NBS1 (MRN) complex detects DSBs and activates ATM kinase(6). ATM phosphorylates MDC1 and H2AX, which generates phosphorylated H2AX (γH2AX)(7). This signal amplifies and propagates the repair signal, leading to the recruitment of various repair proteins including the ubiquitin E3 ligases, BRCA1 and 53BP1 (8). These lead to error-prone non-homologous end-joining (NHEJ) or error-free homologous recombination repair (HRR), depending on the phase of cell cycle. PARP1 is actively involved in the ATM-mediated DNA damage response and repair pathways. It facilitates the rapid recruitment of MRE11 and NBS1 to DSB sites (9) and its shortage leads to impaired ATM kinase activity (10,11). PARP1 also plays a role in the prompt recruitment of BRCA1 to DSBs, which is subject to the HRR(12). Meanwhile, PARP1 complex with DNA-PK and KU70/80 blocks DNA end resection, triggering NHEJ (13,14).

PARP inhibitors (PARPi) are used in clinical settings for treating patients with BRCA1 or BRCA2 mutations (15,16). PARPi not only suppresses catalytic activity of PARP1 but also triggers PARP1 trapping, which results in replication fork collapse and the creation of DSBs during DNA replication (17). In HR-proficient cells, DSBs caused by PARPi are repaired through HRR, enabling cell survival; in HR-deficient cells with BRCA1/2 mutants, DSBs are repaired by NHEJ or Pol theta-mediated Alt-EJ, leading to accumulation of genomic instability and cell death (18). Despite the success of PARPi in patients with HR-deficient tumors, multiple mechanisms of resistance have emerged in patients with advanced cancer. Resistance to PARPi can arise from loss of PARG(19) or from acquisition of revertant mutations in BRCA1 or BRCA2 that restore sufficient HRR function (20). In addition, loss of 53BP1 partially restores HR in BRCA1-deficient cells (21,22). Similarly, Loss of REV7/MAD2L2 leads to HR restoration and PARP inhibitor resistance in BRCA1-deficient cells (23). Additionally, several lines of evidence suggest the importance of ATM/ATR activity in the efficacy of PARP inhibitors. ATM-deficient cells exhibit greater sensitivity PARPi (24) or to the combined treatment of ATR inhibitor with PARPi (25,26). It is unclear how the ATM-mediated DNA damage response (DDR) and repair is activated when PARP activity is inhibited by PARPi.

The DDR orchestrated by PARP1 and/or ATM kinase regulates a downstream cascade linked to ubiquitin signaling. Typically, histones ubiquitination by RNF8 and RNF168 E3 ligases plays an active role in recruiting BRCA1 and 53BP1 to DSB sites (27,28). PARP1 is subject to ubiquitination by different E3 ubiquitin ligases, which modulates its degradation or activity (29,30). A temporal and spatial regulation of ubiquitin signaling at DNA lesions is essential for proper DDR and repair.

Here, we explored the relationship between PARP1 and the ATM machinery at DSB sites, focusing on how these affect the sensitivity to PARP inhibitors. From screening 211 human ubiquitin-related proteins, we identified TRIM44 as a key link between PARP1 and the MRN complex. TRIM44 ensures the MRN complex's recruitment to DNA lesion, independent of PARP1 activity. Hence, suppression of TRIM44 expression led to increased sensitivity to both phleomycin and olaparib. TRIM44 could be a potential therapeutic target to enhance PARPi effectiveness in HR-proficient tumors.

Materials and methods

Cell lines

Human U2OS, HEK293T, HEK293 and HEK293TN cells were cultured in Dulbecco's modified Eagle's medium (DMEM, Welgene, cat#LM001-05) containing 10% fetal bovine serum (FBS, Gibco, cat#12483020). U2OS 2–6–3 (31) and 2–6–5 (32) cells were kindly gifted from Dr Roger A. Greenberg (University of Pennsylvania) and cultured in DMEM containing 10% FBS and puromycin (1 μg/ml). The U2OS-based HR reporter cell (DR-GFP U2OS) (33) was kindly provided by Dr Jeremy M. Stark (Beckman Research Institute of the City of Hope) and maintained in DMEM high glucose without sodium pyruvate (Gibco, cat#11965092) supplemented with 10% FBS and puromycin (1 μg/ml). AsiSI-ER-U2OS (DIvA) cell was kindly provided by Dr Gaëlle Legube (34) and cultured in DMEM (Gibco, Cat# 11995065) containing 10% FBS and 1% penicillin/streptomycin (Gibco, Cat# 15140122). All cells were incubated in a temperature-controlled chamber (37ºC, 5% CO₂).

Establishment of TRIM44 knockout U2OS cells line

To generate TRIM44 knockout (KO) cell line, single guide RNA (sgRNA) targeting to TRIM44 was designed by an online CRISPR design tool (http://chopchop.cbu.uib.no/). The designed sgRNA was cloned into lentiCRISPRv2 (Addgene, #52961) according to the protocol provided by Zhang lab (35) and transfected into HEK293TN cells for production of lentivirus. After 48 hr, virus-containing supernatant was harvested. Viruses were infected into U2OS cells to generate TRIM44 KO cells with treatment of puromycin (1 μg/ml). Selected cells were seeded onto 96 well plate and cultured for 10 days to make single clones. Over 10 clones were tested whether it has TRIM44 or not using an immunoblotting with TRIM44 antibody.

Ubiquitin-related protein library cloning and plasmids

To generate pENTR^TM 211 vectors of human ubiquitin-related proteins (URPs) PCR amplified with individual primer by human liver cDNA libraries. And then all of URPs cloned into pENTR^TM were sequenced and transferred pDEST53 (N-terminal GFP) vector using a Gateway LR Cloning system (Invitrogen). Each truncated mutant of TRIM44 was constructed by a classical PCR method, cloned into pCR8/GW/TOPO vector for entry clones and then subcloned into SFB vector (gifted from Dr Hongtae Kim, Ulsan National Institute of Science and Technology). All deletion regions were validated by DNA sequencing analysis, and each primer set used in this study is described. All detail information for other plasmids and chemicals used in this study are summarized in Supplementary Table S3.

RNA interference and plasmid transfection

The siRNAs for indicate genes purchased from Bioneer were transfected with Lipofectamine RNAiMAX for 72 h. Plasmids were transfected with polyethylenimine (PEI, Polysciences) for 48–72 h.

Laser microirradiation and immunofluorescence

For laser microirradiation, U2OS TRIM44 WT and U2OS TRIM44 KO cells were seeded onto 35 mm round glass bottom dishes (SPL, Korea). Cells were transfected with siRNA and DNA for 48−72 h and incubated with 10 μM 5-bromo-2 deoxyuridine (BrdU) for 24 h before laser-induced DNA damage. Single- and double-strand DNA breaks were induced by laser microirradiation using 405 nm laser in Nikon A1R confocal microscope (Nikon). Cells were subjected to laser-induced DNA damage with two different conditions [3 s/ 32 line for live cell imaging and 1 s/32 line for fixed cell imaging] using 60× oil objective. For fixed cell imaging, cells were fixed with 2% (v/v) solution formaldehyde solution (Sigma) in PBS at room temperature after damaged for indicated times. Cells were blocked for 30 min at room temperature in blocking solution (1% BSA in PBS). Cells were incubated with primary antibodies for overnight at 4ºC. Next day, secondary antibodies (Invitrogen, Carlsbad, CA, USA) were incubated for 1 h at room temperature and mounted with VECTASHIELD® with DAPI (Vector Laboratories).

Immunofluorescence for DNA damage-induced foci formation

For analysis of damage-induced RAD51 foci formation, U2OS cells were plated on 35 mm glass bottom dishes (SPL, Korea). Cells were transfected with siRNAs using RNAiMAX according to manufacturer's instruction for 48 h and then treated with 200 ng/ml Neocarzinostatin (NCS, Sigma-Aldrich) prior to 2 h fixation. Cells were fixed with 4% PFA in PBS for 15 min at room temperature (RT) following NCS treatment. Fixed cells were washed three times with PBS and permeabilized with 0.5% TritonX-100 in PBS for 15 min at RT. Non-specific signals were blocked using blocking solution (PBS containing 1% BSA and 0.5% TritonX-100) at RT for 1 h. After then, primary antibodies diluted in the blocking solution were overnight incubated at 4°C. Following day, Alexa Fluor-conjugated secondary antibodies were incubated for 2 h at RT and washed three times with PBS. Cells were counterstained with 1 μg/ml DAPI (Sigma-Aldrich) diluted in PBS at RT for 10 min, washed with PBS, and then mounted onto glass using VECTASHIELD mounting solution (Vector laboratories). Detailed information of the antibodies used in this study are described in Supplementary Table S3.

For quantification of foci formation, all images were obtained by Nikon A1R confocal microscope using 60× oil immersion objective at the same experimental setting. Images were used for manually foci counting using NIS-element AR software. The cells having ≥5 or 10 RAD51 foci per nucleus were counted by manually and the percentage of cells was calculated. The statistical analysis was performed using GraphPad Prism 8.0 software (GraphPad).

Homologous recombination (HR) repair assay

The repair efficiency of Homologous recombination (HR) was estimated using the DR-GFP U2OS reporter cell line. To evaluate HR efficiency, the DR-GFP U2OS cells were plated in 12-well dishes and transfected with siRNAs using Lipofectamine RNAiMAX (Invitrogen) according to manufacturer's instructions. After 24 h, I-SceI endonuclease was delivered to cells by Lipofectamine 2000 (Invitrogen). After 72 h, cells were trypsinized and centrifuged at 1500 × g for 5 min. Subsequently, cells were washed with PBS and resuspended in the same buffer. The cell suspension was then transferred to 5 ml Round Bottom Polystyrene Tube (Falcon) for FACS analysis. GFP-positive cells were quantified using the FACSAria III (BD biosciences).

DNA end resection assay

For measurement the efficiency of DSB end resection, we used the method as previously described(36). AsiSI-ER-U2OS (DIvA) cells seeded onto 100 mm dishes (1.2 × 10⁶ cells per dish) were transfected with siRNAs. After 2 days, cells were treated or not 300 nM 4-OHT (Sigma-Aldrich) for 4 h to induce DNA double strand breaks. Subsequently, the cells were collected and genomic DNA from the cells was isolated using the DNeasy Blood & Tissue Kit (Qiagen) with RNase A treatment during lysis according to the manufacturer's instructions. After that, 1000 ng of the extracted gDNA was mock-digested or digested with 20 units of BsrGI-HF (NEB, #R3575) restriction enzyme at 37°C for 20 h and the enzyme was heat-inactivated at 65°C for 20 min. 50 ng of each digested gDNA was used as a template for quantitative PCR using SYBR Green Supermix (Bio-Rad) to measure resection site at loci 335 nt (Chr1: 89231183). For each sample, a ΔCt was calculated by subtracting Ct value of mock-digested sample from Ct of digested sample. The percentage of ssDNA (ssDNA %) for each sample was calculated with the following equation: ssDNA % = 1/(2(△Ct − 1) + 0.5) × 100. The used primers are listed in Supplementary Table S2.

FokI assay

For FokI-screening, indicated plasmids were transfected into U2OS-2–6–3 reporter cells integrated to Lac operator repeats (×256) along with mCherry-LacI-FokI endonuclease plasmids using Lipofectamine2000 (Invitrogen). After 48 h transfection, cells were fixed with solution containing 4% formaldehyde in PBS for 10 min at RT and positive signal was analyzed and visualized by Nikon A1R confocal microscope using 60× oil objective. U2OS 2–6–5 cells, which stably express mCherry-LacI-FokI, were seeded on glass bottom dish (SPL) and transfected with GFP-tagged plasmid using Lipofectamine 2000 (Invitrogen). After 48 h incubation, DSBs were induced by treatment with 4-OHT (1 μM) and Shield1 (1 μM) for 4 hr. Cells were fixed with 2% formaldehyde (Sigma) and mounted with VECTASHIELD® with DAPI (Vector Laboratories).

Western blotting

For western blotting assay, samples were boiled with 1× sample buffer and resolved by SDS-PAGE using gradient gel (4−20% acrylamide gel). The separated proteins were transferred onto nitrocellulose membrane and blocked for 1 h with 5% skim milk or 1% BSA in TBST. The membrane was incubated with indicated primary antibodies for overnight at 4ºC and secondary antibodies (Bio-Rad) for 1 h at room temperature. The immunoblotted proteins were detected with ECL reagents (GE Healthcare).

Pull-down assay

HEK 293T cells were lysed with NETN buffer (50 mM Tris, pH 8.0, 150 mM NaCl, 1% Nonidet P-40, 50 U/ml MNase, 50 U/ml Benzonase) including proteases and phosphatase inhibitor cocktail (Roche) and incubated for 20 min at room temperature. Cell lysate was centrifuged at 13 000 rpm for 15 min at 4ºC. For pull down assay of SFB-tagged proteins, the supernatant was incubated with Streptavidin Sepharose High Performance affinity resin (GE Healthcare) for 2 h at 4ºC and washed four times with NETN buffer. The washed precipitates were boiled with 2× sample buffer and used for western blotting assay.

Whole-cell extraction

U2OS cells were harvested with 1× sample buffer and sonicated using EpiShear Probe Sonicator (Active motif). Cell lysates were boiled and used for western blotting assay.

Immunoprecipitation

U2OS cells were lysed with NETN buffer (50 mM Tris, pH 8.0, 150 mM NaCl, 1% Nonidet P-40 and 5 mM EDTA) including proteases and phosphatase inhibitor cocktail (Roche). Cell lysate was sonicated using EpiShear Probe Sonicator (Active motif) and centrifuged at 13 000 rpm for 15 min at 4ºC. For immunoprecipitation, the supernatant was incubated with primary antibody for overnight at 4ºC. Next day, the immunoprecipitates were captured by incubation with Protein A Sepharose Fast-Flow (GE Healthcare) for 1 h and the beads were washed four times with NETN buffer. The washed precipitates were boiled with 2× sample buffer and used for western blotting assay.

Nuclear fraction

U2OS cells were lysed with lysis buffer A (10 mM HEPES, pH 7.4, 10 mM KCl, 1.5 mM MgCl₂, 0.34 M sucrose, 10% glycerol, 0.1% Tritons X-100, 1 mM DTT) including proteases and phosphatase inhibitor cocktails (Roche) and incubated for 5 min at 4ºC. The lysate was centrifuged at 1300 rcf for 4 min at 4ºC. For cytosol fraction, soluble was centrifuged again at 13 000 rpm for 15 min at 4ºC. Nuclear fractions were isolated with lysis buffer B (20 mM Tris, pH 8.0, 0.4 M NaCl, 15% glycerol, 1.5% Triton X-100) including proteases and phosphatase inhibitor cocktails (Roche) and incubated for 30 min at 4ºC. The lysate was centrifuged at 5000 rcf for 5 min at 4ºC.

Chromatin fractionation

U2OS TRIM44 WT and TRIM44 KO cells were transfected with indicated siRNAs and DNAs After 72 h, cells were first lysed with NETN buffer (50 mM Tris, pH 8.0, 150 mM NaCl, 0.5% Nonidet P-40) including proteases and phosphatase inhibitor cocktails (Roche) and incubated for 20 min at 4ºC. The lysate was centrifuged at 12 000 rpm for 10 min at 4ºC. Chromatin fractions were isolated with chromatin buffer (50 mM Tris, pH 8.0, 150 mM NaCl, 1% Nonidet P-40, 25 U/ml MNase, 25 U/ml Benzonase) including proteases and phosphatase inhibitor cocktails (Roche) and incubated for 20 min in 37ºC shaking incubator. Soluble fractions isolated from whole-cell lysates and chromatin fractions were used for western blotting assay. Each experiment was repeated at least three times.

Alkaline comet assay

U2OS cells were treated with 1 mM of H2O2 for 30 min (Release 0 h) and released after indicated time points (Release 3, 6 and 12 h), and Alkaline Comet Assay was performed using reagents from Trevigen in accordance with the manufacturer's protocol. Imaging was performed with a fluorescence microscope and Tail Moment determined using the OpenComet analysis software.

Clonogenic survival assay

U2OS cells were transfected with indicated siRNAs. Next day, cells were seeded (3000 cells/each well) on 6-well plate (TPP). Next day, cells were treated with phleomycin or olaparib for indicated concentrations. Cells were incubated in a temperature-controlled chamber (37ºC, 5% CO₂) for 10–14 days. Cells were fixed with purified methanol (Sigma) for 10 min and stained with 0.01% crystal violet for 20min. Colony areas were measured and analyzed by ImageJ.

Image quantification and statistical analysis

All values in this study are reported as mean ± standard error for the mean (SEM). Representative of an average of at least three independent experiments is shown. Student's t-test or one-way ANOVA followed by Dunnett's multiple comparisons test was used for statistical analysis. P value < 0.05 was considered statistically significant. Significance is indicated by asterisk. *P< 0.05, **P < 0.01, ***P < 0.001, **** P < 0.0001, n.s., not significant.

Results

Identification and functional dissection of novel ubiquitin-related proteins at damaged chromatin

While ubiquitin-mediated signaling is vital in the ATM-mediated DDR, its significance is confined in PARP1 signaling. We cloned 211 GFP-fused ubiquitin-related proteins (URPs) including ubiquitin E3 ligases derived from human liver cDNA library and examined whether these proteins translocate to DSB sites that were generated by the mCherry-fused FokI endonuclease reporter system of the U2OS 2–6–3 cell line(37). Among the 211 URPs examined (Figure 1A; Supplementary Table S1), 27 (12.8%) exhibited co-localization with mCherry-FokI at DSBs (Figure 1B; Supplementary Figure S1). We identified 16 novel URPs subgroups (11 of the 211 URPs, 7.6%) harbouring evolutionally intact motifs, such as tripartite, interesting new gene (RING), homologous to the E6-AP carboxyl terminus (HECT) and Di19 (Figure 1C). We also found several URPs (11 of the 211 URPs, 5.2%) that have well known functions in DDR: CHFR, PELI1, RNF2, RNF8, RNF20, RNF40, RNF138, RNF146, RNF166, RNF168 and RNF206 (27,38–42) (Figure 1B,C; Supplementary Figure S1). To investigate the physiological dissection of potential DDR-associated ubiquitin-related E3s, we first examined whether ATM inhibitor, KU55933, or PARP inhibitor, PJ34, alter their translocation to laser microirradiation (mIR)-induced DNA lesions. Of the 24 DDR-E3s tested, 11 E3 ligases were found at DNA lesions in the presence of KU55993, but not upon treatment with PJ34. Meanwhile, 11 other E3 ligases were clearly enriched at mIR-induced DNA lesions in the presence of PJ34, whereas they were not upon treatment with KU55993. Two other proteins, FBXL and PELI1, moved to DNA lesions regardless of the presence of KU55993 or PJ34 (Figure 1D; Supplementary Figure S2A, B, C). Consequently, we have categorized 24 DDR-E3s into three groups: ATM-dependent (11), PARP1-dependent (11) and ATM- and PARP1-independent (2). As 11 E3 ligases that moved to DNA lesions in a PARP1-dependent manner were not affected by the ATM inhibitor and other E3 ligases recruited to DNA lesions in an ATM-dependent manner were not affected by the PARP inhibitor, we concluded that these proteins in the PARP1 and ATM pathways seem to be apart in their functions. We were particularly interested in PARP-dependent ubiquitin signalling and here focused on TRIM (tripartite motif) family proteins. The TRIM family encompasses approximately 90 proteins that play diverse roles in cellular processes such as signalling, development, protein quality control and carcinogenesis (43). Recently, some TRIM family proteins in the DDR have been identified (44–47). We observed that TRIM31, TRIM36(v), TRIM44 and TRIM61(v) migrated to DSB sites (Figure 1B, E; Supplementary Figure S1). Among these proteins, TRIM31 and TRIM36(v) moved to DNA lesions in an ATM-dependent fashion while TRIM44 and TRIM61(v) translocated DNA lesions in a PARP1-dependent manner. (Figure 1F, G). TRIM31, TRIM36(v) and TRIM61(v) are members of the RBCC-motif subfamily, which contains a RING domain that generally functions as a ubiquitin E3 ligase. In contrast, TRIM44 belongs to the BCC-motif subfamily, which does not contain a RING domain. Initially classified as an E3 ligase due to its TRIM family membership, TRIM44 has been found to function as a ubiquitin-specific protease (USP) via its ZnF-UBP domain (48,49).

Figure 1.

Screening of ubiquitin-related proteins (URPs) at damaged chromatin. (A) Schematic illustrating the FokI-screening process for 211 GFP-tagged human URPs. (B) Co-localization of GFP-tagged URPs and mCherry-FokI nuclease at DSB sites. Representative images indicate mCherry-FokI-positive human URPs associated with DDR (left panel). URPs are categorized as either previously uncharacterized (red; 16 out of 211) or known (black; 11 out of 211) in their association with DDR (right panel). Nuclei were counterstained with DAPI. (C) Identification of DDR-related URPs possessing evolutionarily conserved motifs. (D) Classification of the 24 URPs based on regulation by either PARP1 or ATM activity. (E–G) Visualization of TRIMs' accumulation at DNA break sites post-laser microirradiation. U2OS cells, transfected with GFP-TRIMs underwent laser microirradiation. Pre-treatments included control DMSO (E), PARP1 inhibitor PJ34 at 5 μM (F), and ATM inhibitor KU55933 at 10 μM (G) for 1 h before microirradiation. Cells were fixed using 4% paraformaldehyde 10 min post-microirradiation. The co-localization frequency of GFP-TRIMs with DNA damage marker γH2AX at the laser-irradiated regions is depicted in each panel. Scale bars, 5 μm.

The glu-rich domain of TRIM44 is necessary for PARP1 binding

TRIM44 moves to DNA lesions in a PARP1-dependent manner and it contains a zinc finger ubiquitin-binding domain (ZnF UBP) that binds to the C-terminal diglycine motif of ubiquitin (50). We hypothesized that PARP1-dependent TRIM44 may regulate the DDR by recognizing ubiquitin chains on the damaged chromatin. We verified that endogenous TRIM44 and GFP-TRIM44 moved to the single DSB site generated by FokI endonuclease (Figure 2A). Consistent with the findings in Figure 1F, G, TRIM44 recruitment to DSB was not affected by ATM inhibitor whereas its movement was hindered by treatment with the PARP inhibitor, AG14361 (Figure 2B), which inhibits PARP1’s catalytic activity of PARP1 but does not affect its movement to DNA lesions. The pull-down assay was performed using cell lysates expressing SFB-tagged TRIM44, indicating the strong interaction between TRIM44 and PARP1. These interactions display a slight enhancement when cells were exposed to zeocin, a radiomimetic agent, for 2 h (Figure 2C). TRIM44 was observed to have a weak interaction with ATM. To address whether PARP1 loss abolishes TRIM44 interaction with chromatin, we conducted experiments using siRNA to deplete PARP1 (siPARP1). Our findings indicate that TRIM44’s recruitment to DNA damage sites is highly dependent on the presence and activity of PARP1. The diminished interaction between TRIM44 and H2AX in siPARP1-depleted cells suggests that TRIM44 does not directly bind to H2AX but rather relies on PARP1 for its localization to damaged chromatin (Supplementary Figure S3A). The inhibition of TRIM44 translocation by PARPi further supports the notion that PARP1’s catalytic activity is crucial for TRIM44’s function in the DNA damage response. Therefore, these data underscore the importance of PARP1 in mediating TRIM44’s role at DNA damage sites and highlight the specific mechanism by which TRIM44 is recruited through interaction with PARP1. We then addressed which domains of TRIM44 are important for PARP1 binding. TRIM44 comprises of four domains: ZnF UBP located at the N-terminus, glutamate-rich (GR) domain, B-box (BB) and coiled-coil domain (CC). It is worth noting that the last two domains are well-conserved among TRIM family proteins(43). Four TRIM44 deletion mutants were generated with each domain removed (ΔUBP, ΔBB, ΔCC and ΔGR) (Figure 2D). These mutants were then expressed in 293T cells and treated with zeocin. A pull-down assay revealed that the ΔUBP, ΔBB and ΔCC mutants bound PARP1 as efficiently as TRIM44 WT (Figure 2E). They also retained their binding to not only ATM but also to RAD50 and MRE11 of the MRN complex as TRIM44 WT. In contrast, the ΔGR mutant was unable to bind to PARP1, indicating that GR domain of TRIM44 is critical for the interaction with PARP1. It is worth noting that the ΔGR mutant showed heightened binding to ATM as well as RAD50 and MRE11 in comparison to the other three mutants. These findings suggest a potential inverse correlation between the binding ability of TRIM44 to PARP1 and its binding to the MRN complex. The significance of the GR domain of TRIM44 in its recruitment to DSB was confirmed in the FokI system, where it was observed that the recruitment of TRIM44 ΔGR to the DSB site was measurably lower than that of WT TRIM44 (Figure 2F). Likewise, all deletion mutants of TRIM44, except for ΔGR, moved to the damaged DNA strips created by microirradiation (Supplementary Figure S3B). Impediment of TRIM44 ΔGR moving to DSB lesion was not attributed to different subcellular localization because WT and all deletion mutants of TRIM44 were observed in both nucleus and cytosol in immunocytochemistry and cellular fractionation assay (Supplementary Figure S3C, D). Together, our data clarify that TRIM44 translocates to DSBs in a PARP1-dependent manner. Moreover, the stable localization of TRIM44 at DNA damage sites relies on its physical interaction with PARP1 through the GR domain.

Figure 2.

Domain analysis of TRIM44 binding to PARP1. (A) U2OS 2–6–5 cells were transfected with GFP-TRIM44 and treated with 4-OHT (1 μM) and shield1 (1 μM) for 4 h to induce DNA damage. After induction, cells were fixed stained with TRIM44. (B) U2OS 2–6–5 cells were transfected with GFP-TRIM44, pretreated with DMSO, ATM inhibitor or PARP inhibitor for 1 h and treated with 4-OHT (1 μM) and shield1 (1 μM) for 4 h to induce DNA damage. After induction, cells were fixed. (C) 293T cells were transfected empty vector or SFB-TRIM44. Cells were treated with zeocin (100 μg/ml) for 1 h to induce DSBs. Cells were lysed, subjected to Streptavidin beads pulldown assay, and analyzed by immunoblotting with the indicated antibodies. (D) Schematic illustration of domain structure of human TRIM44. (E) 293T cells were transfected with the indicated combinations of SFB-tagged TRIM44 truncated mutants or SFB vector. Cells were treated with zeocin (300 μg/ml) for 1 h to induce DSBs. Cells were lysed, subjected to Streptavidin beads pull-down assay, and analyzed by immunoblotting with the indicated antibodies. (F) U2OS 2–6–5 cells were transfected with GFP-TRIM44 WT and GFP-TRIM44 ΔGR and treated with 4-OHT (1 μM) and shield1 (1 μM) for 4 h to induce DNA damage. After DSBs induction, cells were fixed. Fixed cells image (upper) and graph (lower). Scale bar, 10 μm. Student's t-test was used for statistical analysis.

TRIM44 prevents the PARP1 hyperactivation

We next addressed whether TRIM44 has an impact on PARP1 activity under DNA damage. To achieve this, we analyzed the product of PARP1-mediated ADP ribosylation activity on PARP1 itself and histones, two major targets of PARP1(3,51). Hydrogen peroxide (H₂O₂) induces both DNA SSBs and DSBs(52) and activates PARP1(53). When U2OS cells were treated with 1 mM of H₂O₂ for 5–30 min, subsequently probed with anti-poly (ADP-ribose) antibodies, a marked auto-PARylation of PARP1 occurred along with a correlated rise in histone ADP-ribosylation within 5–10 min of H₂O₂ treatment in control cells (Figure 3A). This effect vanished in cells cotreated with PARP inhibitor, AG14361. Of note that hyper-PARylation on PARP1 as well as elevation of histone ADP-ribosylation were observed in cells depleted of TRIM44, suggesting that TRIM44 regulates the PARP1 activity. Similarly, a marked increase in PAR levels was also observed in TRIM44 depleted cells treated with varying doses of H₂O₂ (Supplementary Figure S4A). When cells were exposed to methyl methanesulfonate (MMS) that induces SSBs, a moderate increase of PAR levels were observed by TRIM44 siRNA (Supplementary Figure S4A). Elevated PARylation in TRIM44 knockdown cells was verified by laser microirradiation experiment. Immunofluorescence staining with anti-PAR antibodies revealed rapid accumulation of PAR chains at the DNA stripe within 1 min, which almost entirely disappeared within 10 min in control cells (Figure 3B). However, in cells treated with TRIM44 siRNA, the initial accumulation of PAR chains was stronger, and a significant portion of the chains remained even after 10 min. Thus, depletion of TRIM44 resulted in a more intense initial production of PAR chains and a significant delay at subsequent time-points. We verified that hyper-PARylation of PARP1 in TRIM44 depleted cells was not attributable to impediment of PARG recruitment to the DNA lesion (Supplementary Figure S4C) because GFP-PARG in TRIM44 depleted cells recruited to microirradiated DNA strips as efficiently as that in control cells. Given that the binding of TRIM44 to PARP1 is important for the regulation of PARP1 activity, the GR domain of TRIM44 that mediates the PARP1 binding would be central to regulate the PARP1 activity. Thus, we reconstituted each deletion mutant of TRIM44 in TRIM44 knockout (KO) cells and compared the auto-PARylation levels of PARP1 10 and 15 min after H₂O₂ treatment. In TRIM44 KO cells, auto-PARylation of PARP1 was sustained for 15 min whereas reconstitution of wild-type of TRIM44 substantially reduced the PAR levels in 15 min (Figure 3C; Supplementary Figure S4B). Similar decreases in PAR levels were observed in cells introduced with either ΔBB or ΔCC construct. In contrast, the PAR levels at 15 min time-point were sustained in cells expressing ΔGR as TRIM44 KO cells. Of note that accumulation of PAR chains was much stronger in cells introduced with ΔUBP mutant. Interestingly, the data here suggest that not only GR domain but also ZnF UBP domain of TRIM44 plays a key role in the upregulation of PARP1 activity. So, the next question is how the ZnF UBP of TRIM44 regulates PARP1 activity. It is shown that the ZnF UBP in TRIM44 is conserved in some deubiquitinating enzymes (50) and removes the K-48 linked poly-ubiquitin chains, increasing protein stability (48,54). Thus, we hypothesized that TRIM44 KD or TRIM44 ΔUBP may lead to accumulation of PARP1 ubiquitination, which in turn may affect the status of hyper-PARylation of PARP1. In vivo ubiquitylation assay of PARP1 immunoprecipitates using whole cell lysates introduced with HA-ub construct showed that the PARP1 ubiquitination substantially increased after H₂O₂ treatment. Surprisingly, however, knockdown of TRIM44 resulted in decreased PARP1 ubiquitination under DNA damage (Figure 3D). These findings were observed consistently across cells at different time-points (1–15 min) following treatment with H₂O₂. PARP1 ubiquitination increased in a time-dependent manner up to 10 min, while TRIM44 KD cells had constant lower levels of PARP1 ubiquitination. In consistent notion in Figure 3A-C, a significant build-up of PAR chains on PARP1 was observed in TRIM44 KD cells, showing an inverse correlation with its ubiquitination. To determine if these results were due to TRIM44 affecting the total levels of ubiquitylated proteins, we examined the total ubiquitylation levels in the input lysates. Our findings indicate that TRIM44 deficiency does not impact the overall levels of total ubiquitylated proteins (Figure 3D). Additionally, we assessed the basal levels of PARP1 ubiquitylation before H₂O₂ treatment. The results demonstrate that the basal ubiquitylation of PARP1 is comparable between siCtrl and siTRIM44 cells, indicating that TRIM44 silencing does not affect PARP1 ubiquitylation under unstressed conditions. Upon H₂O₂ treatment, differential effects on PARP1 ubiquitylation were observed, underscoring the role of TRIM44 in the DNA damage response. To clarify which domain of TRIM44 affects PARP1 ubiquitination, experiments were conducted using TRIM44 WT, ΔUBP, or ΔGR mutants under overexpression conditions. As previously observed, the restoration of PARP1 ubiquitination was seen in cells overexpressing TRIM44 WT, but not in those expressing ΔUBP or ΔGR mutants (Figure 3E). These findings suggest that TRIM44 binds to PARP1 through its GR domain and subsequently preserves PARP1 ubiquitination through its ZnF UBP domain, thereby limiting PARP1 hyperactivity. Given these results, we next sought to explore the role of HPF1 in PARP1 ubiquitination, given that HPF1 is known to promote serine PARylation of PARP1 after DNA damage (55). To investigate this, we used three distinct HPF1-targeting siRNAs (Supplementary Figure S4D). We observed the accumulation of PAR chains in cells treated with siRNAs for TRIM44 or HPF1. Furthermore, the increase in auto-PARylation of PARP1 was more pronounced in cells with both TRIM44 and HPF1 knocked down(51,53), showing an additive effect (Supplementary Figure S4E). These findings suggest that TRIM44 and HPF1 target different amino acids for PARylation. We also analyzed PARP1 ubiquitination patterns in these cells and found that HPF1 depletion resulted in decreased PARP1 ubiquitination, similar to the effect of TRIM44 depletion (Supplementary Figure S4F). Auto-PARylation of PARP1 occurs at serine (S), aspartic acid (D), glutamic acid (E), and lysine (K) residues within its BRCT and WGR domains. This suggests that there may be competition between ubiquitination and PARylation on the lysine residues of PARP1 in TRIM44-depleted cells. HPF1 has been shown to facilitate serine PARylation of PARP1, indicating that the inverse relationship between PARylation and ubiquitination observed in HPF1-depleted cells may operate through different mechanisms. These results highlight the distinct regulatory roles of HPF1 and TRIM44 in modulating PARP1 activity and stability during the DNA damage response. The relationship between auto-PARylation of PARP1 and its ubiquitination was further examined by PARG KD by siRNA (Figure 3F) or by overexpression of GFP-PARG (Supplementary Figure S4G). PARG KD led to accumulation of PAR chains, accompanied with reduced PARP1 ubiquitination whereas overexpression of PARG significantly increased the PARP1 ubiquitination. Taken together, the data suggests that TRIM44 sustains the poly-ubiquitin chain of PARP1 to constrain PARP1 hyperactivation.

Figure 3.

Correlation between PARP1 hyperactivation and its ubiquitination. (A) U2OS cells were transfected with the indicated siRNAs and pretreated with DMSO and PARP inhibitor (AG14361). Cells were treated with H2O2 (1 mM) for indicated time-points to induce DNA damage and whole-cell extracts were analyzed by immunoblotting with the indicated antibodies. (B) U2OS cells were transfected with indicated siRNAs and subjected to laser microirradiation. After damaged, cells were fixed and co-stained with PAR and γH2AX. Scale bar, 10 μm. Student's t-test was used for statistical analyses. (C) U2OS TRIM44 KO cells were transfected with the indicated combinations of SFB-tagged TRIM44 truncated mutants. Cells were treated with H₂O₂ (1 mM) for indicated time-points to induce DNA damage and whole-cell extracts were analyzed by immunoblotting with the indicated antibodies. (D) U2OS cells were transfected with the indicated siRNAs and HA-tagged ubiquitin. Cells were treated with H₂O₂ (1 mM) for indicated time-points to induce DNA damage and immunoprecipitated (IP) with a PARP1 antibody. Cells were analyzed by immunoblotting with the indicated antibodies. (E) U2OS cells were transfected with the indicated combinations of GFP-tagged TRIM44 truncated mutants and HA-tagged ubiquitin. Cells were treated with H₂O₂ (1 mM) for 5min to induce DNA damage and immunoprecipitated (IP) with a PARP1 antibody. Cells were analyzed by immunoblotting with the indicated antibodies. (F) U2OS cells were transfected with the indicated siRNAs and HA-tagged ubiquitin. Cells were treated with H₂O₂ (1 mM) for indicated time to induce DNA damage and immunoprecipitated (IP) with a PARP1 antibody. Cells were analyzed by immunoblotting with the indicated antibodies.

Hyperactivation of PARP1 inhibits accumulation of the MRN complex to DNA lesions

PARP1 activity is shown to contribute to the early recruitment of MRN complex components, MRE11 and NBS1 to DSB sites, and the activity of ATM is also affected by PARP1 activity(9,11). We noticed that TRIM44 ΔGR showed stronger binding to the ATM/MRN complex than the other three deletion mutants (Figure 2E). These findings indicate that the TRIM44 regulates not only PARP1 activity, but also the ATM-mediated DDR signaling. Indeed, we found that the p-ATM level was reduced by TRIM44 siRNA under DNA damage and reconstitution of GFP-TRIM44 in TRIM44 KD cells restored the phosphorylated ATM (p-ATM) as well as γH2AX levels (Figure 4A). Since ATM relies on the MRN complex to recognize and bind to DNA lesions(6), we compared the amount of the MRN complex and ATM present at the chromatin. In TRIM44 WT cells, the induction of double-strand breaks (DSBs) by zeocin resulted in increased accumulation of ATM and the MRN complex on the chromatin. However, this accumulation was not observed in TRIM44 KO cells (Figure 4B). To determine if these results were due to TRIM44 affecting the total levels of the ATM and the MRN complexes, we examined the total levels of these proteins. Our findings indicate that TRIM44 deficiency does not impact the overall levels of ATM and the MRN complex (Supplementary Figure S5A). Importantly, we also found that the levels of MRE11 and NBS1 were lower in TRIM44 KO cells transfected with ΔUBP than those in cells transfected with WT (Figure 4C), suggesting that hyperactivation of PARP1 may affect the recruitment of the MRN complex to the damage chromatin. Interestingly, a decrease in chromatin binding of NBS1 and MRE11 in PARG knockdown cells was observed whereas an increase in chromatin binding was found in PARG overexpression cells (Supplementary Figure S5B), supporting that hyperactivation of PARP1 hampers the recruitment of the MRN complex to the damaged chromatin. To further validate our findings, we carried out more sensitive microirradiation approaches. At the damaged DNA strips, the accumulation of MRE11 was significantly reduced in TRIM44 KD cells at various time-points of 5–30 min (Figure 4D), and NBS1 accumulation in control cells was also higher than those in TRIM44 KD cells (Figure 4E). Together, TRIM44 limits PARP1 hyperactivation, which allows the recruitment of the MRN complex and activation of ATM at the damaged chromatin. Thus, we propose that TRIM44 is a dual regulator of PARP1 activity and the ATM-mediated DDR.

Figure 4.

Effect of TRIM44 on the accumulation of the MRN complex to DNA lesions. (A) U2OS cells were co-transfected with indicated siRNAs and pcDNA or GFP-TRIM44 WT. Cells were treated with zeocin (300 μg/ml) and subjected to whole-cell extraction. (B) U2OS TRIM44 WT or KO cells were treated with zeocin (300 μg/ml) for 1 h to induce DSBs and subjected to chromatin fraction. (C) U2OS TRIM44 KO (KO) cells were transfected with the indicated combinations of SFB-tagged TRIM44 truncated mutants and SFB empty vector. Cells were treated with zeocin (300 μg/ml) for 1 h to induce DSBs. Cells were lysed, subjected to chromatin fraction, and analyzed by immunoblotting with the indicated antibodies. (D and E) U2OS cells were transfected with indicated siRNAs and subjected to laser microirradiation. After damaged, cells were fixed and co-stained with MRE11 and γH2AX. Scale bar, 10 μm.

TRIM44 is necessary for the DSB repair

Because TRIM44 is important for recruitment of MRN complex to DNA lesion, we next determined the effect of TRIM44 on the repair efficiency of damaged chromatin using Comet assay in which damaged tail image is automatically analyzed by the OpenComet software. When U2OS cells were exposed to H₂O₂ for 30 min, the index of tail moment was increased but 3 hr after wash-off (R3) that index was reduced and returned to the control level at 6 h, suggesting that damaged chromatin was fully repaired (Figure 5A). By contrast, the damaged chromatin in cells with TRIM44 siRNA was only slightly repaired at 3−6 h after wash-off, and still retained the tail. To further confirm the role of TRIM44 in the HR pathway, we carried out homologous recombination (HR) repair assays, DNA end resection assays, and RAD51 foci formation assays (Figure 5B; Supplementary Figure S6A–C). Consistent with the findings in Figure 4, all of these processes were impaired in TRIM44 KD cells. Similarly, in clonogenic survival assay cells with TRIM44 siRNA showed a much lower survival upon phleomycin treatment (Figure 5C). Approximately 30% of cells survived upon treatment of 1 μg/ml) of phleomycin and 10% of cells survived with 2 μg/ml of phleomycin treatment. Knockdown of PARP1 by siRNA did not affect the survival rate, suggesting PARP1 is not actively involved in the repair pathway to phleomycin. On the other hand, either TRIM44 siRNA or BRCA1 siRNA resulted in almost the same reduction rate in cell survival in response to 0.5−2 μg/ml of phleomycin treatment. Approximately 10% of cells survived after 1 μg/ml of phleomycin treatment in cells with either TRIM44 or BRCA1 siRNA. Thus, the data suggest that TRIM44 is as crucial as BRCA1 in DSB repair. Together, TRIM44 is necessary for the DSB repair. In cells that have functional PARP1, TRIM44 plays a role in inhibiting the PARP1 hyperactivation, which leads to proper DDR and repair.

Figure 5.

Effect of TRIM44 on DSB repair. (A) U2OS cells were transfected with indicated siRNAs and treated with H₂O₂ for 1h to induce DNA damage. After induction, cells were release with fresh medium for 0 hr, 3 hr, 6 hr and 12 hr and proceeded to comet assay (upper) and western blotting (lower). (B) Analysis of homologous recombination repair efficiency. The effect of TRIM44 depletion was assessed in DR-GFP-U2OS cells. Statistical significance was determined using one-way ANOVA, followed by Dunnett's multiple comparisons test. Data represent mean ± s.e.m. of six independent experiments. **** P < 0.0001. (C) U2OS cells were transfected with indicated siRNAs and treated with various dose of phleomycin (0, 0.5, 1 and 2 μg/ml) for clonogenic survival assay. Graph (left) and colony (right). Student's t-test was used for statistical analysis.

TRIM44 shifted its binding to the MRN complex in the presence of PARPi

PARPi represents the first approved drugs for clinical use to utilize the concept of synthetic lethality in patients carrying germline mutations in BRCA1(16,17,56). The induction of DSBs by PARPi is resolved via an ATM-dependent HR repair. We looked into whether TRIM44 is also involved in the recruitment of the MRN complex in cells treated with PARP inhibitor. U2OS cells were subjected to 1 h pretreatment with AG14361 prior to the addition of zeocin before cell harvest. In the chromatin fractionation assay, we observed a decrease in the accumulation of MRE11 and NBS1 at the damaged chromatin in TRIM44 KD cells, irrespective of the presence of PARPi (Figure 6A), implying a consistent role of TRIM44 in the recruitment of the MRN complex to the damaged chromatin. This was confirmed by a microirradiation experiment followed by immunostaining. Control and TRIM44 KD U2OS cells were treated with AG14361 for 1 h before microirradiation. Indeed, the accumulation of both MRE11 and NBS1 at the DNA strips was significantly reduced in TRIM44 KD cells at 10 and 20 min after laser microirradiation (Figures 6B–E). As the TRIM44 ΔGR mutant, which loses its binding to PARP1, showed stronger binding to RAD50, MRE11 and ATM compared to three other PARP1-interacting deletion mutants (Figure 2E), it is possible that the interaction of TRIM44 with the MRN complex might change when PARP1 activity was destroyed. In the pull-down assay, a reduction in the interaction between TRIM44 and PARP1 was observed following AG14361 treatment. In contrast, TRIM44 showed a stronger interaction with ATM as well as with MRE11 and NBS1 (Figure 6F). Thus, the data showed that TRIM44 switches its primary binding partner to the MRN complex in the presence of PARPi. The domain of TRIM44 crucial for its interaction with the MRN complex in these conditions was investigated. In the pull-down assay, we found that TRIM44 ΔUBP almost completely lost its ability to bind ATM and MRN complex whereas WT and the other three deletion mutants of TRIM44 retained their ability to bind them (Figure 6G). In similar to the notion in Figure 2E, the ΔGR mutant showed a complete loss in the binding to PARP1 but it retained its binding to the MRN complex, suggesting that TRIM44 may directly interact with the MRN complex in the presence of PARPi. Together, the data indicate that TRIM44 acts as a functional mediator of the MRN recruitment to DSB sites where the ZnF UBP domain plays a critical in binding to the MRN complex in the presence of PARP inhibitor.

Figure 6.

Interaction of TRIM44 with the MRN complex in the presence of PARPi. (A) U2OS cells were transfected with the indicated combinations of SFB-tagged TRIM44 truncated mutants and SFB empty vector (EV). Cells were pretreated with DMSO or PARP inhibitor (AG14361) for 1hr and treated with zeocin (300 μg/ml) for 1 h to induce DSBs. Cells were lysed, subjected to chromatin fraction, and analyzed by immunoblotting with the indicated antibodies. (B and C) U2OS cells were transfected with indicated siRNAs. Cells were pretreated with DMSO or PARP inhibitor (AG14361) for 1hr and subjected to laser microirradiation. After damaged, cells were fixed and co-stained with MRE11, NBS1 and γH2AX. Scale bar, 10 μm. (D) Quantification of (B). Student's t-test was used for statistical analysis. (E) Quantification of (C). Student's t-test was used for statistical analysis. (F) 293T cells were transfected with SFB empty vector or SFB-tagged TRIM44 WT and pretreated with DMSO or PARP inhibitor (AG14361) for 1 hr. Cells were treated with zeocin (300 μg/ml) for 1 h to induce DSBs. Cells were lysed, subjected to Streptavidin beads pull-down, and analyzed by immunoblotting with the indicated antibodies. (G) 293T cells were transfected with SFB empty vector or SFB-tagged TRIM44 truncated mutants and pretreated with PARP inhibitor (AG14361) for 1 h. Cells were treated with zeocin (300 μg/ml) for 1 h to induce DSBs. Cells were lysed, subjected to streptavidin beads pull-down, and analyzed by immunoblotting with the indicated antibodies.

TRIM44 is a key upstream factor in determining the sensitivity to PARP inhibitors

PARPi represents the first approved drugs for clinical use to utilize the concept of synthetic lethality in patients carrying germline mutations in BRCA1(16,17,56). The induction of DSBs by PARPi is resolved via an ATM-dependent HR repair. Given that TRIM44 is the molecule responsible for recruiting the MRN complex to DSBs in cells challenged with PARPi, it would be a critical factor to determine the sensitivity to PARPi. In clonogenic survival assays, approximately 30 percent of U2OS cells were found to survive by treatment of 2 μM of olaparib, showing a mild effect. PARP1 KD by siRNA restored the cell survival rates to 70–90% (Figure 7A). On the other hand, the survival rates in cells with TRIM44 or BRCA1 siRNA were significantly diminished, up to 5% (Figure 7A). Thus, the efficiency of TRIM44 KD was comparable to that of BRCA1 KD. Double-KD of TRIM44 and BRCA1 did not display any additive or synergistic effects on cell death (Supplementary Figure S7A, B), suggesting that TRIM44 and BRCA1 share the same repair pathway. Thus, our data demonstrated that suppression of TRIM44 expression sensitizes cells to both phleomycin (Figure 5C) and olaparib. In order to explore the potential therapeutic effect of TRIM44 in cancers we searched for the TRIM44 expression in various cancers. In TCGA RNA-seq dataset(57), the levels of TRIM44 mRNA expression were noticeably lower among the patients diagnosed with renal cell carcinoma (Figure 7C). The Kaplan–Meier analysis showed that patients with low expression levels of TRIM44 exhibited lower overall survival rates when compared to those in the high TRIM44 expression group (Figure 7D). Like that observed in U2OS cells (Figure 7A), knockdown of TRIM44 by siRNA in HEK293 cells, a kidney cancer cell line, resulted in a significant reduction in survival rate, even with lower doses of olaparib treatment (0.25–1 μM) (Figure 7E; Supplementary Figure S7D, E). Thus, the data indicate that patients with KIRC displaying low TRIM44 expression could be suitable for olaparib treatment. Previous studies have shown that resistance to PARPi treatment is due to secondary mutations that restore BRCA1 function or BRCA2-induced HR due to dysfunction of the 53BP1-RIF1-shieldin complex(58). Since TRIM44 plays a role in the most upstream step of DSB repair, we determined if TRIM44 affects the overcoming of PARPi resistance caused by 53BP1-deficeint. We found that 53BP1 KD by siRNA increased the cell survival rate at the concentrations of 1–2 μM of olaparib and double-KD of TRIM44 and 53BP1 significantly rescued PARPi sensitivity (Figures 7B; Supplementary Figure S7C). Specifically, as shown in the Figure 7B, the survival rates of cells treated with TRIM44 siRNA were significantly reduced, reaching up to 20% following the administration of 2 μM of Olaparib. In contrast, 53BP1 knockdown (KD) by siRNA resulted in an increase in cell survival at concentrations of 1–2 μM of Olaparib. This suggests that TRIM44 and 53BP1 do not operate within the same pathway in response to PARP inhibitors. Notably, the simultaneous knockdown of TRIM44 and 53BP1 resulted in a reduction of the enhanced cell survival observed following 53BP1 KD. These data indicate that TRIM44 plays a role in overcoming PARP inhibitor resistance caused by 53BP1 deficiency. Taken together, these data suggest that TRIM44 is a key upstream determinant of PARP inhibitor sensitivity. Once PARP1 activity turns off, TRIM44 recruits the MRN complex to the damaged chromatin, initiating the ATM-mediated DDR and repair.

Figure 7.

Effect of TRIM44 on the sensitivity to PARP inhibitors. (A and B) U2OS cells were transfected with indicated siRNAs and treated with various dose of olaparib (0, 0.5, 1 and 2 μM) for clonogenic survival assay. Graph (left) and colony (right). Student's t-test was used for statistical analysis. (C) Expression level of TRIM44 in kidney renal clear cell carcinoma (KIRC). (D) Kaplan-Meier plot for TRIM44 in KIRC (http://www.oncodb.org/). (E) HEK293 cells were transfected with indicated siRNAs and treated with various dose of olaparib or phleomycin (0, 0.5, 1 and 2 μM) for clonogenic survival assay. Student's t-test was used for statistical analysis. (F) Functional mechanism of TRIM44 at the DNA lesions in the presence of PARPi.

Discussion

Here, we unveil TRIM44 as a previously unidentified mediator of PARP1 to ATM-mediated DDR. At DSBs, TRIM44, in tandem with PARP1, optimizes the MRN complex binding to damaged chromatin, thereby restraining excessive PARP1 activity. With PARPi exposure, TRIM44 switches its binding to ATM/MRN complexes and facilitates their accumulation at damaged chromatin (Figure 6). This makes TRIM44-deficient cells notably more vulnerable to zeocin or PARPi. We propose that TRIM44 directly connects of PARP1 to the ATM-mediated DDR, independent of the PARP1 activity.

In DDR, we showed that TRIM44 has a strong correlation with PARP1. Its prompt relocation to DSBs relies on PARP1 activity (Figure 1F). TRIM44 also directly interacts with PARP1 via its GR domain (Figure 2E) on DNA lesions. The GR domain is a disordered region in which approximately 50% of the domain contains glutamic acid. Deletion of the GR domain in TRIM44 resulted in the abolition of its binding to PARP1, whereas the deletion of the ZnF UBP, BB and CC domains did not have the same effect (Figure 2E). Given that TRIM44 moves to damaged chromatin in a PARP1-dependent manner (Figure 1F), it can be postulated that the TRIM44 GR domain plays a role in maintaining its stable binding to PARP1 at the chromatin following its translocation to DNA lesions. Furthermore, we observed that TRIM44 WT retained its binding to PARP1 in cells treated with PARPi, despite its initial translocation to DNA lesions being PARP1 activity-dependent (Figure 1F). The TRIM44 ΔGR mutant was observed to lose its binding to PARP1 in cells treated with PARPi (Figure 6G). These data indicate that the GR domain in TRIM44 is responsible for the binding of trapped PARP1 in these cells. Moreover, the TRIM44 ZnF UBP domain plays a crucial role in regulating PARP1 activity. Hyper-PARylation on PARP1 itself as well as histones was shown in cells expressed TRIM44 ΔUBP (Figure 3C), suggesting that TRIM44 UBP is involved in the regulation of PARP1 activity. Meanwhile, ubiquitination pattern on PARP1 was reduced in TRIM44 KD cells (Figure 3D), showing an inverse correlation between ubiquitination and auto-PARylation of PARP1. Because the ZnF UBP domain is capable of binding ubiquitin chains, we assumed that TRIM44 ZnF UBP domain binds ubiquitin chains on PARP1, which protects them from accessibility of other deubiquitinating enzymes (DUBs) or other modifying enzymes. This is different from the previously known function of TRIM44. TRIM44’s ZnF UBP domain, observed to be consistent with specific deubiquitinating enzymes(50), enhances the stability of proteins like VISA, p62, and HIF-1α by removing their poly-ubiquitin chains (48,54). The question then arises as to how ubiquitination on PARP1 impacts its auto-PARylation or activity. Since both ubiquitination and ADP-ribosylation target the lysine residue, they might be in competitive processes (59–61). Our experiments, showing altered ubiquitination and ADP-ribosylation of PARP1 upon PARG manipulation (Figure 3F; Supplementary Figure S4G), support this notion. Several E3 ubiquitin ligases, including TRIP12 and MDM2, mediate PARP1 ubiquitination, leading to its degradation (29,40). Notably, SMURF2 also affects PARP1’s activity via ubiquitination, irrespective of degradation (30). Moreover, DUB inhibitors enhance PARP1 ubiquitination and subsequent PARylation (62), suggesting ubiquitination's role in PARP1 activation. These reports support our data here that protection of PARP1 ubiquitination by TRIM44 restricts the hyperactivation of PARP1.

At DSBs, TRIM44 acts as a direct mediator of PARP1 to the ATM-mediated DDR and repair. It directly influences p-ATM and γH2AX levels, and is crucial for the MRN complex's movement to DNA lesions (Figure 4A, B). Our findings support previous claims that PARP1 rapidly recruits MRE11 and NBS1 to DSB sites (9) and its deficiency results in the defective ATM kinase activity(10,11). We suggest that TRIM44 serves as an essential link between PARP1 and the MRN complex. When TRIM44 is absent at DNA lesions, there's a notable decline in repair efficiency (Figure 5A, B). This is in line with studies noting increased cisplatin sensitivity in lung adenocarcinoma when TRIM44 is depleted (63,64). While accumulated PAR chains on damaged chromatin are typically beneficial for recruiting repair factors (4), our study indicates that excessive PARylation can hinder MRN complex recruitment (Figure 4B). In TRIM44 KO cells transfected with ΔUBP, there's a decrease in MRE11 and NBS1 at the chromatin compared to WT-transfected cells. This pattern is consistent in cells with manipulated PARG levels. While overexpressing PARG marginally increases ATM machinery on the chromatin, depleting it does the opposite, particularly with significant PAR chain build-up (Supplementary Figure S5B). We noticed that γH2AX levels were not consistent in TRIM44 depleted cells under DNA damage. Unlike the reduction of γH2AX levels in immunoblotting in cells exposed to phleomycin or zeocin (Figure 4A), γH2AX levels remained stable in TRIM44 KD cells when cells were exposed to H₂O₂ (Figure 3A) or under laser microirradiation (Figure 3B). It is shown that H₂O₂ induces both DNA single- and doble-strands breaks(52) and ATM and ATR mediate production of γH2AX(7). Likewise, under laser microirradiation not only DSBs but also SSBs and pyrimidine dimers occur(65).

Importantly, we demonstrated that TRIM44 would be a valuable target for evaluating the PARPi sensitivity. TRIM44 exhibited a robust interaction with ATM as well as with the MRN complex in the presence of PARPi (Figure 6F). These interactions are mediated by the ZnF UBP domain of TRIM44 as TRIM44 ΔUBP almost completely lost its ability to bind ATM and MRN complex (Figure 6G). It is worth noting that ubiquitination of MRE11 is regulated by the presence of UBQLN4, a proteasomal shuttling factor, and RNF126 E3 ligase(66,67). And the accumulation of ubiquitinated MRE11 at damaged sites facilitates HR repair. In addition, we previously reported that Pellino1 mediates the NBS1 ubiquitination, which promotes the ATM activation at DSBs(41). Thus, ubiquitinated MRE11 and/or NBS1 could be suitable targets for TRIM44 to bind and facilitate the HRR process. It was also observed that significantly lower survival rates were displayed in TRIM44-depleted cells that received olaparib treatment (Figure 7A). While BRCA1/2 mutated cancers are primary PARPi targets, some HR-proficient cancers resist these inhibitors. Our data suggest that by targeting TRIM44, the efficacy of PARPi therapy could extend beyond HR status, given TRIM44’s role in recruiting MRN to damaged chromatin and initiating ATM-mediated DDR. Supporting this, we found that TRIM44 depletion sensitizes 53BP1-deficient cells to PARPi (Figure 7B). TRIM44’s overexpression in numerous cancers links it to tumor progression, poor prognosis, and chemoresistance (48,63,68). Investigating whether higher DNA repair activity is related to these cancers with high TRIM44 expression would be an interesting topic for future research. Recent reports also showed that TRIM44 raised the expression levels of BRCA1 and 53BP1, promoting DNA repair (64). This suggests that TRIM44 expression might play a role in regulating repair activity at DSB sites. In summary, TRIM44 is a vital upstream element in DSB repair. Targeting TRIM44 offers solutions not only in overcoming PARPi resistance but also in widening its usage to tumors with HR-proficient.

Data availability

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Ho Chul Kang (hckang@ajou.ac.kr)

This study did not generate new unique reagents. Requests for cell lines and plasmids generated in this study should be directed to the lead contact.

This paper does not report the original code. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Supplementary data

Supplementary Data are available at NAR Online.

Funding

National Research Foundation of Korea (NRF) grants funded by the Korean government [2020R1A2C3011423, 2023R1A2C1004074, 2022R1I1A1A01064018]; Korea Health Technology R&D project of the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare, Republic of Korea [HR16C0001, HR22C1734]. Funding for the open access charge: KHIDI [HR22C1734].

Conflict of interest statement. None declared.