When BRCA-deficient cells become PARPi resistant, their defects in HR, fork protection, and gap protection are often partially or fully reverted, suggesting that these restored activities may contribute to the resistance. Additionally, BRCA-deficient but PARPi-resistant cells may be increasingly dependent on alternative DSB repair pathways for survival. These possibilities have fueled the development of strategies to overcome PARPi resistance by combining PARPi with other DNA damage response (DDR) drugs.

Combinations of PARPi with ATR/Chk1 inhibitors. The ataxia telangiectasia and Rad3-related (ATR) kinase, a master regulator of the replication stress response, plays a crucial role in stabilizing replication forks (64, 91). In addition to inducing fork instability, ATR inhibition also blocks HR at DSBs by preventing RAD51 recruitment (92). Thus, ATR inhibitors (ATRis) may be the ideal drugs to increase replication fork collapse and simultaneously prevent the repair of resulting DSBs.

In a panel of PARPi-resistant cell lines derived from a BRCA1-deficient cell line, ATRi effectively suppressed the restored HR and fork protection activities, resensitizing the resistant cells to PARPi (55). In addition, BRCA1-deficient, PARPi-resistant cells also displayed restored abilities to prevent gap-initiated nascent DNA degradation and to activate the ATR checkpoint upon PARPi-induced fork collapse (15, 64). Again, these restored activities in the resistant cells are suppressed by ATRis. These results raise the possibility that a common underlying mechanism may be responsible for the various phenotypes associated with PARPi resistance and that this mechanism is disrupted by ATR inhibition. The exact mechanism that drives PARPi resistance in BRCA-deficient cells remains unclear, but restored RAD51 loading to ssDNA gaps and collapsed forks would alleviate PARPi-induced DNA damage. Indeed, RAD51 foci were observed in the cancer cells from BRCA1/2 mutant patients that acquired PARPi resistance (93). Importantly, ATR inhibition abolishes the restored RAD51 foci and the loading of RAD51 to replication forks in BRCA1-deficient, PARPi-resistant cancer cells (55), supporting the idea that ATR inhibition disrupts a common RAD51-mediated mechanism driving PARPi resistance (Figure 3A).

Combination therapies to overcome PARPi resistance. (A) ATR and Chk1 inhibitors (ATRi, Chk1i) overcome PARPi resistance by disrupting restored HR, fork-protection, and gap-protection activities. (B) WEE1 and PTMYT1 inhibitors (WEE1i, PKMYT1i) may overcome PARPi resistance by increasing replication and overriding the G2/M checkpoint. (C) POLQi overcomes PARPi resistance by blocking alt-EJ and/or ssDNA gap repair. REV1 and USP1 inhibitors (REV1i, USP1i) may also overcome PARPi resistance by blocking ssDNA gap repair. (D) Inhibition of DNPH, MTHFD2, and ALC1 (DNPHi, MTHFD2i, ALC1i) may overcome PARPi resistance by increasing PARP trapping. More studies are needed to confirm whether MTHFD2i and ALC1 can overcome resistance.

The ability of ATR inhibition to overcome the PARPi resistance of BRCA1/2-deficient cells is seen not only in cell lines, but also in patient-derived xenografts (PDXs) and organoids of BRCA-deficient tumors (88, 94, 95). Inhibitors of Chk1, the effector kinase of ATR, have similar effects in PARPi-resistant cell lines and PDXs (Figure 3A) (94, 96). In a clinical trial, the combination of ATRi and PARPi showed efficacy in HR-deficient, PARPi-resistant, high-grade serous ovarian cancer (HGSOC) patients (97). Several ATRis are being tested in clinical trials (98), providing opportunities to overcome PARPi resistance in the near future.

Combinations of PARPi with WEE1/PKMYT1 inhibitors. The tyrosine kinase WEE1 inhibits both CDK1 and CDK2 by phosphorylating the tyrosine 15 of these two kinases (99). WEE1 inhibitor (WEE1i) induces hyperactivation of CDK1/2 and overrides the G2/M DNA damage checkpoint, leading to excessive replication origin firing, replication catastrophe in S phase, and mitotic catastrophe in mitosis (99, 100). The combination of WEE1i and PARPi overcomes PARPi resistance in breast and ovarian cancer models (101). WEE1i and PARPi also display a synergy in preclinical models of BRCA wild-type triple-negative breast cancer (TNBC) by activating antitumor immune responses (102). WEE1 inhibition also increases PARPi sensitivity in BRCA wild-type pancreatic cancer cells (103) and HGSOC cells (104), suggesting a utility of WEE1i as a PARPi sensitizer. A recent study suggested that PKMYT1, another member of the WEE kinase family regulating the G2/M transition, is a promising therapeutic target in cancer cells overexpressing cyclin E1, which are under high replication stress (105). PKMYT1 inhibitor (PKMYT1i) induces unscheduled activation of CDK1 in S phase, driving cells into mitosis before completion of DNA replication. It is interesting to note that PARPi-resistant, BRCA1-deficient cells regain the abilities to suppress ssDNA and activate the checkpoint response (15, 59, 64) and that the abilities of WEE1i and PTMYT1i to increase replication and override the checkpoint may enable them to revert the changes in PARPi-resistant cells and overcome PARPi resistance (Figure 3B).

Combinations of PARPi with inhibitors of gap repair. DNA polymerase θ (POLθ or POLQ) plays a crucial role in repairing DSBs through the alternative end-joining (alt-EJ) pathway (also known as the microhomology-mediated end-joining or MMEJ pathway). The expression of POLQ is upregulated in HR-defective epithelial ovarian cancers, suggesting that alt-EJ functions as a backup DSB-repair pathway to compensate for the loss of HR (106). POLQ is also shown to fill in ssDNA gaps in BRCA-deficient cancer cells (107). The antibiotic novobiocin (NVB) inhibits POLQ by binding to its ATPase domain and overcomes the PARPi-resistance of BRCA-deficient cells and PDXs (108). ART558, which inhibits the polymerase activity of POLQ, also overcomes the PARPi resistance of BRCA-deficient tumors (109). Notably, POLQ inhibitors (POLQis) induce high levels of ssDNA in PARPi-resistant, BRCA-deficient cells, suggesting that they prevent the repair of resected DSB ends and/or ssDNA gaps. Like POLQ, the translesion synthesis (TLS) pathway is also implicated in the repair of ssDNA gaps in BRCA1-deficient cells (14). TLS is initiated by PCNA monoubiquitylation at stalled forks or gaps, which allows the recruitment of REV1 and several TLS DNA polymerases to bypass various DNA lesions. JH-RE-06, a TLS inhibitor that disrupts the interaction between REV1 and POLζ, preferentially kills BRCA1 mutant cells and overcomes their PARPi resistance (14). Inhibition of USP1, a deubiquitylase of PCNA, leads to persistent PCNA mono- and polyubiquitylation and fork instability, preferentially killing BRCA-deficient cells (110). Recent studies showed that USP1 inhibitors (USP1is) increase ssDNA gaps in BRCA1-deficient cells, synergize with PARPi in killing BRCA-deficient cells, and overcome PARPi resistance in BRCA1-deficient cells and PDXs (111, 112). Together, these results suggest that combining PARPi with inhibitors of ssDNA gap repair is a promising strategy for overcoming the PARPi resistance of BRCA-deficient cells (Figure 3C).

Combinations of PARPi with drugs increasing PARP trapping. DNPH1 (2′-deoxynucleoside 5′-monophosphate N-glycosidase, also known as RCL) is an enzyme that eliminates the cytotoxic hydro5-hydroxymethyl-deoxyuridine (hmdU) monophosphate. Inhibition of DNPH leads to increased hmdU misincorporation, PARP trapping, and fork collapse in BRCA1-deficient cells (113). Consequently, DNPH inhibitor (DNPHi) preferentially kills BRCA1-deficient cancer cells and overcomes their PARPi resistance. Notably, the effects of DNPH1 are dependent on the SMUG glycosylase, suggesting that DNA nicks or gaps are involved. Somewhat analogously, inhibition of the folate metabolism enzyme methylenetetrahydrofolate dehydrogenase/cyclohydrolase (MTHFD2) leads to an imbalanced dUTP:dTTP pool, increased replication stress, and preferential killing of acute myeloid leukemia (AML) cells (114). It would be interesting to determine whether MTHFD2 inhibitor synergizes with PARPi in BRCA-deficient cells to overcome their PARPi resistance. The trapping of PARP by PARPi is also stimulated by the loss of PAR-binding chromatin remodeling factor ALC1 (CHD1L) (115–117), likely due to the decrease of chromatin accessibility at DNA damage sites and reduced recruitment of repair proteins. Loss of ALC1 drastically increases the PARPi sensitivity of BRCA-deficient cells and overcomes their PARPi resistance, making ALC1 an attractive therapeutic target (Figure 3D). The studies on ALC1 suggest that nucleosome remodeling, by influencing the repair of DNA nicks or gaps, is a key determinant of PARP trapping and the PARPi sensitivity of HR-deficient cells.

It is worth noting that recent CRISPR/Cas9 loss-of-function screens have served as a powerful and unbiased tool to explore synthetic lethal interactions with PARPis in BRCA-proficient and -deficient cells (118). In addition to ALC1 (115), RNase H2 was identified as a strong synthetic lethal screen hit with olaparib (119). Loss of RNase H2 renders cells hypersensitive to PARPi and also selectively kills BRCA1/2-deficient cells. CRISPR screens in prostate cancer cells revealed that loss of MMS22L drastically increases PARPi sensitivity (120). On the other hand, CRISPR screens also revealed mechanisms of PARPi resistance, such as point mutations in PARP1 (39), loss of CHK2 (120), and loss of ARH3 (121).

It is also important to note that, while combinations of PARPis with other targeted drugs may overcome PARPi resistance, these drug combinations may also increase general cellular toxicity and side effects, including hematological toxicity. Optimization of drug scheduling and dosing is likely important for achieving the maximal efficacy of these combination therapies. It remains unclear whether the selectivity of these combinations toward BRCA-deficient cells are reduced or enhanced compared with PARPis, which is an important question to address in clinical trials.