Exosome component 1 cleaves single-stranded DNA and sensitizes kidney renal clear cell carcinoma cells to poly(ADP-ribose) polymerase inhibitor

Targeting DNA repair pathway offers an important therapeutic strategy for cancers. However, the failure of DNA repair inhibitors to markedly benefit patients necessitates the development of new strategies. Here, we show that exosome component 1 (EXOSC1) promotes DNA damages and sensitizes kidney renal clear cell carcinoma (KIRC) cells to DNA repair inhibitor. Considering that endogenous source of mutation (ESM) constantly assaults genomic DNA and likely sensitize cancer cells to the inhibitor, we first analyzed the statistical relationship between the expression of individual genes and the mutations for KIRC. Among the candidates, EXOSC1 most notably promoted DNA damages and subsequent mutations via preferentially cleaving C site(s) in single-stranded DNA. Consistently, EXOSC1 was more significantly correlated with C>A transversions in coding strands than these in template strands in KIRC. Notably, KIRC patients with high EXOSC1 showed a poor prognosis, and EXOSC1 sensitized cancer cells to poly(ADP-ribose) polymerase inhibitor. These results show that EXOSC1 acts as an ESM in KIRC, and targeting EXOSC1 might be a potential therapeutic strategy.

In this study, we show that EXOSC1 acts as an ESM and sensitizes cancer cells to 97 PARPi in KIRC. Due to the role of exosome in maintaining genomic stability, these 98 results also indicate that a unit of multiprotein complex can play a role opposite to 99 that of the complex. 100

Identification of candidate ESMs in KIRC 103
Because that ESMs constantly assaults genomic DNA, we hypothesized that ESMs 104 likely sensitized cancer cells to the inhibitors of DNA repair pathways. Considering 105 that substitution is the most abundant mutation in all cancers, we initiated this study to 106 identify candidate ESMs responsible for substitution mutations. To identify the 107 candidate ESMs other than deamination, we focused on KIRC for three reasons: (1) 108 KIRC shows the lowest proportion of mutations at CG in major cancer types (Burns   Spearman's rank analysis was first performed to assess the correlation between 120 each c-substitution type and the genome-wide expression of individual genes. 121 Resultant p and r values were used for the further analyses. For example, GAPDH 122 showed a p = 0.0011 and r = 0.16 correlation with C>A/G>T c-substitution, indicating 123 that GAPDH expression displayed a positive correlation with C>A/G>T ( Figure 1E). 124 Similarly, CRB3 (p = 0.0003, r =0.17) was positively correlated with A>T/T>A 125 ( Figure 1E). Although the p values of multiple genes were lower than 0.05 ( Figure  126 1F), only top ranked 200 genes (approximately 1% of the genome-wide genes) were 127 taken as the candidates for each c-substitution type (Supplementary Table S1). 128 Student's t-test analysis was then used to determine whether the expression 129 difference of individual gene between the high and low c-substitution groups is 130 significant. The expression of individual genes in each patient was normalized by a 131 house keeping gene, TATA-binding protein (TBP) as previously described (Burns et 132 al., 2013a;Burns et al., 2013b). According to each c-substitution, 532 KIRC patients 133 were groups into 3 groups (high, medium and low). The difference of individual gene 134 between high and low c-substitution mutation groups was then analyzed by student's 135 t-test. Resultant p and fold of change (FC) values were used for the further analyses. 136 For example, ASNS with p = 0.0005 and FC = 1.39, indicating that ASNS was 137 increased in the high group ( Figure 1G). Although the p values of multiple genes were 138 lower than 0.05 ( Figure 1H), only top 200 genes with high FC and p < 0.05 were 139 taken as the candidates (Supplementary Table S2). Notably, none of APOBEC family 140 members were identified as candidate by correlation or student's t-test analyses, 141 supporting that deamination contributes less to the mutations in KIRC. 142 Next, we performed meta-analyses to determine which of the 6 c-substitution types 143 to be focused on. Mutation frequencies of c-substitution types were first analyzed in 144 five major cancers: breast adenocarcinoma (BRCA), glioblastoma multiforme (GBM), 145 bladder urothelial carcinoma (BLCA), acute myeloid leukemia (AML) and KIRC, 146 which potentially suffer less from the EOSMs. As shown in Figure 1I To evaluate the capability of the candidate gene to promote mutation, 168 rifampicin-resistant assay in E. coli was performed as previously described 169   Figure 2B). Notably, EXOSC1 more significantly increased 179 mutations than AID did (p = 4.08 X 10 -5 ) ( Figure 2B). We then evaluated the 180 capabilities of EXOSC1 homologs (EXOSC2-EXOSC9) to promote mutations 181  Considering that the exosome is well known to degrade RNA, we speculated that 208 suggesting that EXOSC1 preferred to cleave C sites in ssDNA. Since EXOSC1 was 244 correlated with the C>A/G>T c-substitution type, it was likely that EXOSC1 cleaved 245 C sites in ssDNA and subsequently resulted in C>A mutations through -A‖ rule DNA 246 repair. 247 To evaluate the above hypothesis, we then determined whether EXOSC1-promoted 248 mutations displayed strand asymmetries. Considering that EXOSC1 cleaved C sites in 249 ssDNA but not DNA-RNA hybrid, we speculated that -transcribed‖ temple strands 250 likely bound by RNA were less cleaved by EXOSC1. As shown in Figure 4F  C>A from 0% to 69% (p = 2.67 X 10 -7 ), while EXOSC1 only enhanced G>T from 6% 260 to 17% even without a significance (p = 0.27) (Figure 4-figure supplement 1E and 261 F). Next, we evaluated C>A strand asymmetry in KIRC using spearman's rank and 262 student's t-test analyses. Spearman's rank analyses indicated that EXOSC1 showed 263 the highest correlation (r) with C>A, and the correlation between EXOSC1 and G>T 264 was even lower than that between EXOSC1 and C>A/G>T ( Figure 4G and Figure  265 4-figure supplement 1G). To evaluate the impact of group number on the further 266 student's t-test analyses, the KIRC patients were grouped into 2, 3, 4 or 5 groups 267 according to the mutation types C>A, G>T, C>A/G>T and total (12 substitution 268 types). As expected, the EXOSC1 differences between the low and high C>A/G>T 269 groups were more significant than those between the low and high groups of total (12 270      Figure 6A and B). The median DFS in the 320 low EXOSC1 group was 32.0-month longer than that in the high group (p = 9.78 X 321 10 -8 , log-rank test) ( Figure 6A). Consistently, the median OS in the low EXOSC1 322 group was 36.9-month longer than that in high group (p = 2.2 X 10 -8 ) ( Figure 6B). As 323 expected, the best-separation KM analysis also indicated that high EXOSC1 group 324 significantly showed poor OS (p = 2.6 X 10 -12 ) ( Figure 6-figure supplement 1B). Considering that EXOSC1 increases DNA damage, we speculated that EXOSC1 332 potentially sensitizes KIRC cells to the inhibitor of poly(ADP-ribose) polymerase 333 (PARP), which treats cancers via blocking DNA repair. As previously described (Li et  Caki-2 KIRC cells to the inhibitor (Figure 6-figure supplement 1D). Next, we 342 determined whether EXOSC1 could sensitize KIRC cells to niraparib in xenograft 343 mouse models. The control and EXOSC1-OE cells were subcutaneously injected. 344 Resultant tumor-bearing mice were grouped and treated by vehicle or niraparib. 345 Consistent with the ex vivo results, niraparib more notably inhibited the tumor with 346 enhanced EXOSC1 ( Figure 6G and I), indicating that EXOSC1 sensitized KIRC 347 xenografts to the inhibitor. No significant weight loss was observed throughout the 348 study, suggesting that the niraparib treatment was well tolerated (Figure 6H and J). The potential pathological significance of EXOSC1 is supported by its association 379 with poor DFS and OS in KIRC. Due to the capability of EXOSC1 to cleave DNA 380 and promote mutations, EXOSC1 might enhance mutations and consequently provide 381 genetic fuel for cancer development, metastasis, and even therapy resistance. 382 Therefore, EXOSC1 might represent not only a KIRC marker but also a target to 383 decrease the rate of KIRC evolution and stabilize the targets of existing therapeutics. However, several limitations of this study should be noted. First, we observed a 396 notable variation in terms of the correlation with a different c-substitution type for a 397 given gene, implying the need for further studies. Second, although we showed that 398 EXOSC1 could cleave ssDNA and act as an ESM, we did not directly identify the 399 mechanism responsible for turning the DNA cleavages into mutations. The role of 400 XRCC1 in EXOSC1-promoted mutations was only briefly evaluated. Hence, we 401 cannot exclude the possibility that other proteins might contribute to this process. 402 Third, because that DNA cleavage by EXOSC1 should be independent of the cancer 403 type, we propose that EXOSC1 likely contributes to the mutations in the cancers other 404 than KIRC. And more work is still needed. Despite these limitations, our results still 405 indicate that EXOSC1 acts as an ESM in KIRC. 406

Sample preparation 409
Samples of 532 KIRC patients from TCGA used for expression and mutation analyses  Limited, Hornsby Westfield, NSW, Australia) to construct the pET-28a-Gene-6XHis E. 446 coli expression plasmids. PCR primers for the amplification of above genes are 447 described in Supplementary Table S5. 448 449

Immunoblotting and immunofluorescence 450
Immunoblotting and immunofluorescence were carried out as described in our 451 previous study (Song et al., 2018;Wang et al., 2020). primers. PCR products were then analyzed by gel electrophoresis, cloned into 530 pMD20-T vector, and sequenced. 531 532

Colony-forming assay 533
The colony-forming assays were performed as described in our previous study (Li et 534 al., 2020). 535 536

Subcutaneous xenograft tumor growth in vivo 537
The following animal-handling procedures were approved by the Animal Care and 538 Use Committee of Dalian Medical University. Xenograft models were carried out as 539  Briefly, 2 × 10 6 stable control/EXOSC1-OE 769-P and Caki-2 cells were suspended and injected subcutaneously into the flank of 6-week-old nude mice. After 7 days, 542 these tumor-bearing mice were randomized into 4 groups (6 mice per group) and 543 treated by oral gavage twice a day with vehicle or niraparib (4 mg/kg). The mice were 544 observed daily and weighed once per week. Tumor size was measured using a caliper, 545 and the tumor volume was calculated using the following formula: 0.52 × L × W 2 , 546 where L is the longest diameter and W is the shortest diameter. Mice were euthanized 547 when the tumors reached 1500 mm 3 or showed necrosis.