SLFN11 enhances the sensitivity of BRCA1/2-deficient cells to PARPis.
To examine the reciprocal effects of BRCA1/2 and SLFN11, we employed the SLFN11-expressing ovarian endometrioid adenocarcinoma TOV-112D and medulloblastoma DAOY cell lines in their SLFN11-knockout (SLFN11-KO) counterparts, which we recently shown to be hypersensitive to camptothecin (CPT) and cisplatin [32, 33], Deleterious mutations in BRCA1 and BRCA2 are absent in either cell line according to the Cancer Cell Line Encyclopedia database (https://discover.nci.nih.gov/rsconnect/cellminercdb/) [39].
After knocking down (KD) BRCA2 expression using a mixture of siRNAs in parent and SLFN11-KO cells in each cell line (Figs. 1A and S1A), we compared four conditions, presented in the following order in Figs. 1 and S1: SLFN11-KO/control siRNA (siCON), SLFN11-KO/BRCA2 siRNA (siBRCA2), parent (SLFN11-proficient)/siCON, and parent/siBRCA2. In cellular viability analyses with 48 h continuous 100 nM CPT treatment (Fig. 1B, left), TOV-112D SLFN11-KO/siCON cells were the most resistant. SLFN11-KO/siBRCA2 cells or parent/siCON cells were more sensitive than SLFN11-KO/siCON cells, and parent/siBRCA2 cells were the most sensitive. Under 25 nM CPT (Fig. 1B, right), SLFN11-KO/siBRCA2 cells were more sensitive than SLFN11-KO/siCON cells, but parent/siCON cells were as resistant as SLFN11-KO/siCON cells. Parent/siBRCA2 cells were as sensitive as SLFN11-KO/siBRCA2 cells. These results demonstrate that SLFN11 expression and BRCA2 deficiency additively enhance CPT sensitivity.
We next examined the impact of BRCA2 and SLFN11 on the activity of the four clinical PARPis: talazoparib, niraparib, olaparib, and veliparib. At relatively high concentrations (125 nM talazoparib, 1.2 µM niraparib, and 10 µM olaparib), we observed additive effects of BRCA2-KD and SLFN11 expression in cell viability assays (Fig. 1C). By contrast, at relatively low drug concentrations (8 nM talazoparib, 312 nM niraparib, 2.5 µM olaparib, and 25 µM veliparib), where the effects of SLFN11 expression were marginal in parent/siCON cells compared with SLFN11-KO/siCON cells, the effects of SLFN11 expression were significant in parent/siBRCA2 cells compared with SLFN11-KO/siBRCA2 cells (Fig. 1C). We obtained comparable results with the DAOY cell set (Figure S1B, C). Additionally, flow cytometry revealed an 11% increase in dead cells in SLFN11-KO/siBRCA2 cells, a 9% increase in parent/siCON cells, and a 32% increase in parent/siBRCA2 cells compared to SLFN11-KO/siCON cells (Fig. 1D). These results establish that SLFN11 further enhances the activity of PARPis in the cells lacking functional BRCA2.
To generalize these results, we performed parallel experiments by knocking down BRCA1 and examining the combinational effects of BRCA1 inactivation and SLFN11 expression (Fig. 1E). However, the reduced proliferation of BRCA1-KD cells compared to their WT counterparts (Fig. 1F) rendered it difficult to assess a potential difference in cell viability 48 h after olaparib treatment. Yet at 72 h, we observed additive effects of BRCA1-KD and SLFN11 expression (Fig. 1G). These results collectively demonstrate that SLFN11 enhances PARPis sensitivity in BRCA1- and BRCA2-deficient cells. To circumvent any issues arising from the slow proliferation of BRCA1-KD cells, we primarily focused on analyzing the effect of BRCA2 in the rest of the study.
SLFN11 expression and BRCA2 deficiency increase chromatin-bound RPA2 in cells treated with PARPis.
Generation of single-strand DNA (ssDNA) gaps has been correlated with cell killing by PARPis [17]. To examine whether these previous findings were consistent with our results, we measured chromatin-bound replication protein A2 (RPA2), which reflects the amount of ssDNA in genomic DNA. Immunoblotting demonstrated that chromatin-bound RPA2 increased in TOV-112D SLFN11- KO/siBRCA2, parent/siCON, and furthermore increased in parent/siBRCA2 cells in the presence of 10 µM olaparib (Fig. 2A, B). Comparable results were obtained with niraparib (Figure S2). Thus, we conclude that chromatin-bound RPA2, which reflects an accumulation of ssDNA was well correlated with impaired cell viability in response to PARPis (Fig. 1C, olaparib 10 µM). By contrast, under 100 nM CPT treatment, chromatin-bound RPA2 was increased similarly in SLFN11-KO/siCON and SLFN11-KO/siBRCA2 cells but was relatively restricted in parent/siCON and parent/siBRCA2 cells (Fig. 2C). Hence, chromatin-bound RPA2 was not correlated with cell killing in the case of CPT (Fig. 1B, CPT 100 nM). The results with CPT are consistent with our previous findings demonstrating that SLFN11 restricts the extension of uncoupled replication forks carrying ssDNA gaps [26]. The results of immunoblotting with olaparib treatment (Fig. 2A, B) matched well with the immunofluorescence with pre-extraction techniques that detect only chromatin-bound RPA2 (Fig. 2D, E). Together, these experiments demonstrate that PARPis generate ssDNA gaps that are enhanced by SLFN11 expression and BRCA2 deficiency, and even further in cells both proficient for SLFN11 and deficient for BRCA2.
SLFN11 expression and BRCA1/2 deficiency increase ssDNA gaps in the presence of PARPis.
To further study the formation of ssDNA gaps induced by PARPis, we performed alkaline BrdU comet assays [18]. TOV-112D cells were treated with or without PARPis for 6 h, labeled with BrdU for 30 min during that time, and incubated in BrdU-free medium for the last 90 min (Fig. 3A). Tail moments of BrdU-labeled cells were measured to score the amount of ssDNA gaps [18]. Without drug treatment, tail moments were minimal and were not significantly affected by BRCA2 deficiency or SLFN11 expression (Figs. 3B, C). Olaparib (10 µM) increased the tail moments in SLFN11-KO/siCON cells and further increased in SLFN11-KO/siBRCA2 cells (Figs. 3B, C). Notably, parent/siCON cells exhibited significantly higher tail moments than SLFN11-KO/siCON cells, indicating that SLFN11 impairs OFP under olaparib treatment as BRCA2 deficiency does. Notably, parent/siBRCA2 cells exhibited the highest tail moments. Niraparib showed comparable results to olaparib (Figure S3). We also obtained comparable results among BRCA1-deficient and/or SLFN11-expressing conditions (Fig. 3D). Together, these results demonstrate that SLFN11 expression increases ssDNA gaps in cells treated with PARPis and that this effect is independent of BRCA1/2 but enhanced in BRCA1/2-deficient cells.
We next examined whether catalytic inhibition or PARP-trapping was more important for OFP impairment. To this end, we first checked the catalytic inhibitory concentration for PARylation by olaparib. Because under normal conditions, PARylation was not detectable in TOV-112D cells, we used the alkylating agent methanesulfonate (MMS) to induce PARylation and titrate the olaparib concentration required for inhibiting PARylation in the TOV-112D cells (Fig. 3E). As olaparib suppressed PARylation even at 10 nM, we used 100 nM olaparib to sufficiently suppress PARylation. Under these conditions, the tail moments were not increased under all cellular conditions (Fig. 3F), suggesting that catalytic inhibition of PARP1/2 alone is insufficient to impair OFP. By contrast, 100 nM CPT increased the tail moment but at similar levels in all cellular conditions (Fig. 3G), suggesting that CPT induces ssDNA gaps regardless of SLFN11 expression and BRCA2 deficiency. According to the known mechanism of CPT action, these single-stranded DNA gaps are caused by the formation of TOP1-cleavage complexes (TOP1-trapping) [40]. Together, these results reveal the unique effects of PARPis in increasing ssDNA gaps in a SLFN11-dependent and BRCA1/2 deficiency manner likely through PARP-trapping at Okazaki fragments [16, 18].
BRCA2 deficiency enhances the chromatin recruitment of SLFN11 under PARPi treatment.
Given the ability of SLFN11 to bind RPA complex-coated ssDNA as well as ssDNA [25, 26] and our current finding that ssDNA gaps increase in SLFN11 expressing cells treated with PARPis, we hypothesized that SLFN11 could be recruited to the PARPi-induced ssDNA gaps. To test this possibility, we collected chromatin-bound fractions from TOV-112D parent/siCON or parent/siBRCA2 cells after olaparib treatment. We observed that chromatin-bound PARP1 was increased in parent/siBRCA2 cells, indicating that more PARP-trapping lesions are generated under BRCA2-deficient conditions (Fig. 4A, B). In parallel, chromatin-bound SLFN11 was increased in parent/siCON cells and was further increased in parent/siBRCA2 cells. The kinetics of SLFN11 recruitment to chromatin coincided with the kinetics of RPA2 signals (Fig. 4A, B). These results demonstrate that SLFN11, RPA2 and PARP1 associate with chromatin with similar kinetics in cells treated with olaparib.
We next analyzed chromatin-bound SLFN11 using immunofluorescence with pre-extraction. As the chromatin binding of endogenous SLFN11 was hard to detect in TOV-112D parent cells, we generated SLFN11-overexpressing (OE) TOV-112D parent cells (Fig. 4C). Knocking down BRCA2 by siRNA was as effective as in the original parent cells (compare Figs. 1A and 4C). More importantly, in these parent/siBRCA2 cells treated with olaparib, chromatin-bound SLFN11 was increased to a greater extent than in the parent/siCON cells (Fig. 4D, E). In addition, line plots showed that the chromatin-bound SLFN11 was colocalized with chromatin-bound RPA2 (Fig. 4D). From these results, we conclude that SLFN11 binds RPA complex-coated ssDNA gaps under olaparib treatment and that BRCA2 deficiency generates more PARP-trapping and ssDNA gaps behind replication forks, where SLFN11 recruitment is enhanced.
SLFN11 recruitment at ssDNA gaps behind replication forks does not block replication.
To elucidate the potential differences between SLFN11 recruitment behind replication forks in response to PARP inhibition and SLFN11 recruitment at replication forks in response to CPT, we first performed cell cycle analyses. In response to CPT, SLFN11-KO/siCON cells resumed replication at 24 h (Fig. 5A, top) and SLFN11-KO/siBRCA2 cells showed a faster replication state than SLFN11-KO/siCON cells at 6, 12, and 24 h (Fig. 5A, second from top). This is likely because the S-phase checkpoint was less activated in SLFN11-KO/siBRCA2 cells than in SLFN11-KO/siCON cells, as inferred from a recent report [41] and the reduced phosphorylation of Checkpoint Kinase 1 (CHK1) at S345 (Figure S4A). SLFN11-expressing cells (parent/siCON and parent/siBRCA2 cells) treated with CPT showed a persistent replication block at 12 and 24 h (Fig. 5A, second from bottom and bottom). Yet, the S-phase checkpoint was less activated in BRCA2-deficient than in parent/siCON cells (Figure S4A), indicating the dominant replication blocking effects of SLFN11 [26].
We next performed immunofluorescence for RPA2 and SLFN11 with pre-extraction in cells treated with 5-ethynyl-2’-deoxyuridine (EdU) 1 h before harvesting to determine the location of replication foci. After CPT treatment for 12 h, SLFN11 formed foci at the nuclear periphery and the inner nucleus, where RPA2 colocalized (Fig. 5B, second from top and bottom) [26]. Cells with SLFN11 foci were mostly negative for EdU signals indicative of replication block in these cells and EdU-positive cells were significantly decreased (Fig. 5C, left). These results confirmed our previous findings that SLFN11 at replication forks blocks replication in response to CPT [26].
Parallel cell cycle analyses in TOV-112D cells treated with 10 µM olaparib showed a lesser impact on cell cycle progression at 6 h and 12 h compared to CPT in the four subsets. Additionally, cells in each subset kept progressing toward G2-phase until 24 h (Fig. 5D). BRCA2-deficient cells (SLFN11-KO/siBRCA2 and parent/siBRCA2) exhibited slightly faster cell cycle progression than each counterpart (SLFN11-KO/siCON and parent/siCON) at 24 h with less S-phase checkpoint activation (Fig. 5D and S4B). These results indicate that, by contrast to CPT, SLFN11 has little impact on replication in cells treated with olaparib at concentrations that trap PARP1. Consistently, the EdU-positive population was not affected by olaparib (Fig. 5C, right), and chromatin-bound SLFN11 was mostly detected in EdU-positive cells with colocalization with RPA2 under olaparib treatment (Fig. 5E). The patterns of SLFN11 foci and RPA2 foci in the presence of olaparib were clearly different from those in the presence of CPT, where, as discussed above, the SLFN11 foci induced by CPT tend to be most visible at the nuclear periphery (Fig. 5B, E). Nevertheless, we detected a small number of cells with SLFN11 foci at the nuclear periphery and inner nucleus under olaparib treatment; however, these cells were negative for EdU overall (Fig. 5E, bottom). Comparable results were obtained with the DAOY cell set (Figure S4C-F, S5). Together, these results suggest that olaparib induces ssDNA gaps both behind replication forks and at replication forks, both of which recruit SLFN11 to chromatin, and the recruitment of SLFN11 behind replication forks does not block replication.
Resection by MRE11 is required for SLFN11 recruitment to chromatin.
Given that ssDNA gaps were increased by SLFN11 in cells lacking BRCA1/2, we tested the role of the 3’ to 5’ meiotic recombination 11 (MRE11) exonuclease in the generation of the ssDNA gaps observed under olaparib treatment. We treated the four subsets of TOV-112D cells with olaparib and with or without mirin, an established MRE11 inhibitor. In SLFN11-KO/siCON and SLFN11-KO/siBRCA2 cells in which olaparib treatment increased chromatin-bound RPA2, the addition of mirin decreased the chromatin-bound RPA2 almost to control levels (Fig. 6A, B). In the SLFN11-expressing cells (parent/siCON and parent/siBRCA2) treated with olaparib, chromatin-bound RPA2 and SLFN11 were also decreased to control levels by the addition of mirin (Fig. 6C, D). Hence, MRE11-mediated resection is likely critical for generating ssDNA gaps in response to PARPis, regardless of BRCA2 or SLFN11.
SLFN11 expression is a biomarker of favorable response to olaparib in ovarian cancers.
Given that SLFN11 expression, both alone and in combination with BRCA deficiency, leads to increased ssDNA gap formation and hypersensitivity to PARPis, we conducted a retrospective analysis of SLFN11 expression using immunohistochemistry in a cohort of 73 ovarian cancer patients who received olaparib as maintenance therapy at Keio University Hospital (Fig. 7A, B). The median duration of olaparib treatment was 11.3 months, with a range of 1.6 to 47.6 months. In our examination of progression-free survival (PFS), we categorized patients into two groups: super-responders, defined as having more than 2 years without disease progression under olaparib treatment, and short-responders, who experienced disease progression within 6 months during olaparib treatment. A significantly higher number of SLFN11-positive cases was observed among the 11 super-responders (median 32.2 months; 24.0-47.6 months) compared to the 11 short-responders (median 4.0 months; 2.1-6.0 months) (Fig. 7C). Germline BRCA status was available in 9 out of 11 super-responders, and 6 out of 9 patients had BRCA mutations. The 6 patients with BRCA mutation were all positive for SLFN11 (Fig. 7C). Although the analysis was based on a limited number of samples as responses to PARP inhibitors beyond two years are rare, these findings suggest that SLFN11 expression may serve as a predictive biomarker for identifying patients who are likely to respond favorably to olaparib maintenance therapy.