APE1 and NPM1 protect cancer cells from platinum compounds cytotoxicity and their expression pattern has a prognostic value in TNBC

Breast cancer (BC) is a heterogeneous disease comprising several biologically different subtypes characterized by distinctive prognosis and potential therapeutic targets [1, 2]. Although gene expression profiling technologies are capable to finely describe four main intrinsic subtypes, the logistic issues in deploying this methodology in common clinical practice has resulted in the widespread adoption of their immunohistochemistry (IHC)- defined approximation, which is still capable of predicting clinical outcome, pattern of metastatization and benefit from distinct types of therapy [3‐7]. Among BC subtypes, triple negative breast cancer (TNBC) represents a challenge because of its heterogeneity and the absence of a well-defined druggable target, restricting clinical decision making to chemotherapy [8‐10]. Different approaches have been explored to better dissect the TNBC profile in order to optimize therapeutic choices and exploit new strategies, such as genotoxic agents and interference with different DNA repair pathways [11, 12]. Among the therapeutic options already employed in the clinical practice of several tumors including TNBC, platinum (Pt)-based compounds emerge for their ability to interfere with several cellular processes, including DNA replication and transcription [13‐15]. The most common Pt-salts include cisplatin (CDDP), carboplatin (CBDCA) and oxaliplatin [16]. Of note, the activity of the single agent CBDCA was investigated through the randomized phase III trial TNT. After a median follow-up of 11 months, no significant improvements were observed in the total population in terms of overall response rate (ORR) with CBDCA versus docetaxel. However, in patients with a germline BRCA1/2 mutation, ORR was more than two-fold higher when treated with CBDCA, with a longer progression-free survival (PFS) [17]. Regarding the molecular effects of Pt-salts, although several mechanisms of action were hypothesized and are still under investigation, the most accredited is the generation of mono-adducts, intra-strand and inter-strand (ICL) DNA cross links, able to distort the DNA backbone causing the formation of toxic single (SSB) and double (DSB) strand breaks [16]. If not efficiently repaired, DNA damages induced by Pt-salts may cause an extensive blockage of the physiological cellular processes, including DNA replication and transcription, resulting, to an ultimate extent, in cell cycle arrest and apoptosis. The preferable DNA repair pathways acting in repairing bulky lesions induced by Pt-salts are the nucleotide excision repair (NER), the homologous recombination (HR) and the Fanconi anemia (FA) pathways [18, 19]. Just recently, a role of base excision repair (BER) pathway has been also hypothesized [19‐21]. Typically, BER is active in repairing non-bulky DNA lesions induced by oxidative, alkylating or methylation stressors [22]. Upon the recognition of the damaged base by specific glycosylases, the apurinic/apyrimidinic endonuclease 1 (APE1, also known as REF-1) cleaves the newly generated abasic site allowing the accomplishment of the DNA repair [23]. The pivotal role of APE1 is demonstrated by its function in cellular viability and embryonic development [24], due to its role in DNA repair activity and other recently characterized non-canonical functions. In fact, APE1 also plays an important role as redox effector on many transcriptional factors, such as NF-κB, HIF-1α, STAT-3, PAX8, AP-1 and p53 [23‐27], regulating important genes involved in tumor progression. Moreover, new interesting molecular functions involved in RNA metabolism were recently discovered in our laboratory [23], including processing of damaged RNA [28], miRNAs [29] and abasic and oxidized ribonucleotides embedded in DNA [30]. The different functions of APE1 are finely modulated by expression, localization, posttranslational modifications (PTMs) [31‐36] and by its protein-protein interactome, as well. Indeed, APE1 localizes into the nucleus with a peculiar nucleolar accumulation, which depends on active rDNA transcription [37‐41], but the precise significance of this subnuclear distribution is currently non-completely clear [42]. Interestingly, a proper shuttling from nucleoli to nucleoplasm is essential for an efficient response to genotoxic damage [43]. Up to now, it is well accepted that APE1 subcellular localization changes are associated with several cellular functions, as well as to cancer onset and progression [28, 31, 41]. Interestingly, one of the major determinants of APE1 accumulation within nucleoli is its interaction with nucleophosmin (NPM1), which is modulated by APE1 acetylation [41, 44]. NPM1 is a phosphoprotein, which canonically acts as a central factor in rRNA gene processing and as a chaperone in ribosome biogenesis involved in cell proliferation [45, 46]. Depending on the cellular context, NPM1 may act both as a proto-oncogene and as a tumor suppressor and its perturbations are often involved in tumorigenesis and cancer progression [47, 48]. The NPM1 gene is also involved in several chromosomal translocation characterizing several tumors and involving genes such as ALK, RAR and MLF1 [49]. In addition, an aberrant overexpression of the NPM1 protein is another causing factor of several tumors including colon and ovarian cancers [48, 50, 51]. Notably, its localization has an impact on tumorigenesis. Indeed, NPM1 prevalently localizes within the nucleoli, but it constantly shuttles between the nucleus and the cytoplasm [45, 46, 52, 53]. We have already demonstrated that NPM1, and its localization, have an impact on BER activity. In fact, NPM1 is an important functional regulator of BER factors, specifically controlling levels and localization of BER proteins, including APE1 [43]. Moreover, in acute myeloid leukemia (AML)- associated mutations, the mutated NPM1 gene determines the formation of an aberrant NPM1 protein (NPM1c+) which re-localizes in the cytoplasm. This mis-localization hampers canonical functions of NPM1 [54‐56] and affects APE1 nuclear BER function in cancer cells, through relocalization of APE1 itself in the cytoplasm [41]. Finally, it has been demonstrated that higher levels of APE1, often detected in several cancers, confer acquired resistance to chemotherapeutic agents [57] and that hyperacetylation of APE1 is associated with the TNBC phenotype [31]. For these reasons, APE1 is an emerging promising therapeutic target for cancer treatment [58]. To this aim, research has been recently focused on the interference of APE1 functions, including the AP-endonuclease function (e.g. Compound #3) and the redox function (e.g. APX3330) [59, 60] (Codrich et al., submitted), and on efficiently disrupting the APE1/NPM1 interaction, such as SB206553, Fiduxosin and Spiclomazine [61]. One of our purposes was testing whether the treatment with BER inhibitors could sensitize cancer cells to genotoxic agents [61]. Although partially investigated, the relationship between BER and Pt-salts needs to be further explored [20, 21, 62‐68]. Based on the above mentioned evidences, we deemed fundamental to investigate the cytotoxicity induced by the combined treatment of Pt-compounds and APE1- inhibitors, which may have synergistic therapeutic effects in the treatment of cancers such as TNBC [69, 70].

For this reason, starting from the emerging importance of Pt-salts for the treatment of TNBC patients and, in parallel, from the continuously evolving knowledge on APE1 functions, the purpose of this study was to understand the role of APE1, and of its interactor NPM1, in TNBC cell lines treated with Pt-compounds, including CDDP and CBDCA. Specifically, by using different cancer cell lines and specific NPM1- or APE1- gene knockout cell models, we explored: i) the protective role of APE1 and NPM1 in CDDP cytotoxicity and ii) whether the APE1 and NPM1 proteins were modulated in terms of level and subcellular localization upon Pt-compounds treatment in TNBC cancer cells. Moreover, we investigated whether targeting APE1 endonuclease activity or its interaction with NPM1 may sensitize TNBC cancer cells to Pt-compounds treatment. To corroborate our in vitro data, we also considered APE1 and NPM1 levels in a real-world cohort of patients affected by TNBC and explored their potential prognostic impact for further hypothesis-generation and potential clinical utility. Finally, we analyzed the TCGA-BRCA dataset (n = 1105), focusing in particular on TNBC patients (n = 180), and identified several gene signatures whose expression levels correlated in a positive or negative way with APE1 and NPM1 expression. Notably, we examined the differences in patients that had an overall survival above/below 5 years, or that underwent disease recurrence after at least 1 year from diagnosis, highlighting the genes and the underlying molecular processes that may be involved in this different outcome.

Pt-salts represent a common chemotherapeutic option for the treatment of several tumors, including breast cancer. Based on the evidences regarding the role of BER factors in regulating Pt-salts cytotoxicity, and in order to shed more light on the molecular basis of BER protective function to Pt-compounds treatment, the present study was aimed at characterizing the functional relationship between APE1 and NPM1, as well as their interaction with Pt-compounds. We focused specifically on TNBC, a particularly aggressive tumor subtype, characterized by poor prognosis and lack of novel therapeutic targets [11, 84], in which, currently, genotoxic agents are gaining momentum as an important option [15]. The genotoxic effect of Pt-salts, generating bulky lesions on DNA, can be efficiently opposed by different DNA repair pathways, including NER, HR and FA. Only recently, BER has emerged as a putative DNA repair pathway involved in cancer cell resistance to Pt-salts treatments. However, the way in which BER is involved still remains object of study. It has been demonstrated that Pol β [85] could be considered a prognostic factor in colorectal and esophageal cancer, for its ability to affect Pt-salts susceptibility [63, 86, 87]. Similarly, APE1, which is overexpressed in lung cancer, is directly involved in CDDP resistance and its inhibition sensitizes cancer cells to the treatment [68]. Conflicting observations were reported by Kothandapani et al., which have hypothesized that BER could play a role in mediating the CDDP cytotoxicity, showing how a BER blockage, through Pol β or UNG1 knockout or treatment with Methoxyamine, an indirect inhibitor of APE1 endonuclease activity, may result in an increased resistance to CDDP treatment [20, 21]. What is actually unclear is why BER could be directly involved in the CDDP- induced DNA damage, since it is not able to process any bulky DNA distortion produced by Pt-induced adducts. Interestingly, several authors have hypothesized an indirect involvement of BER in mediating cisplatin cytotoxicity [64, 65]. Again, Kothandapani et al. showed that CDDP-treatment induces abasic sites accumulation close to ICLs [20]. It also seems that CDDP and, to a lesser extent, CBDCA and oxaliplatin may cause the generation of reactive oxygen species (ROS), significantly increasing oxidative stress levels. Moreover, interesting findings have recently shown how Pt-salts may increment ROS and DNA oxidation, leading to an increased cell death [64], and have suggested a possible use of APE1 redox inhibitors for the treatment/prevention of secondary effects like Pt-induced sensory neuropathy [88]. A further hypothesis regards the emerging discoveries about the non-canonical roles of BER proteins such as those associated with RNA metabolism [23], that could also explain the role of BER in Pt-salts induced effects. In fact, a strong but not completely characterized association between Pt-salts and RNA-damage, which inhibits ribosomal RNA synthesis in vivo [89] and mRNA translation in vitro [90], has been found.

In this study, by using APE1- or NPM1-defective isogenic cell lines, we first demonstrated a leading role of APE1 and NPM1 expression, as well as their functional interaction, accounting for cell resistance to CDDP cytotoxicity. Then, by using two TNBC cell lines (HCC70 and HCC1937 cells) characterized by a different p53- and BRCA1-status, we initially categorized their response to Pt-salts (CDDP and CBDCA) in vitro. Specifically, we observed that HCC70 cells, having lower levels of both APE1 and NPM1 proteins, were more sensitive to Pt-salts treatments compared to HCC1937 cells; in particular, we found that this occurred in a NPM1-dependent manner, as demonstrated by the protective effect due to the overexpression of NPM1 in the HCC70 cell line. In order to evaluate whether APE1 was also involved in the response to Pt-salts in TNBC cells, we tested the effect of the combination of CDDP and CBDCA with different APE1 inhibitors targeting the endonuclease activity of the protein (e.g. Compound #3) and its interaction with NPM1 (Fiduxosin, Spiclomazine and SB206553) [59, 61]. After evaluating the response of both TNBC cell lines to these drugs as single agents, which demonstrated that the HCC70 cell line was more sensitive to APE1-inibitors, in agreement with the lower level of the protein with respect to HCC1937, we analyzed the response of the combined treatment with Pt-salts. Interestingly, we observed that only HCC70 cells, expressing lower levels of both APE1 and NPM1, were sensitized by the combination of Compound #3 or Fiduxosin with both crosslinking agents. We previously reported that APE1 accumulates within the nucleoli through the interaction with NPM1 [28]. Notably, in that paper we observed that CDDP induced a re-localization of both NPM1 and APE1 to the nucleoplasm in HeLa cells and that impairment of the nucleolar-nucleoplasmic shuttling kinetics of APE1 was associated with a higher sensitivity to CDDP induced cytotoxicity [43]. Our present data in TNBC cell lines, showing nucleolar emptying upon Pt-salts treatment but having different kinetics between HCC70 and HCC1937 cell lines, allow to conclude that APE1 and NPM1 proteins re-localization from nucleoli to nucleoplasm after treatment with Pt-compounds is a general phenomenon of cancer cells, even though with different kinetics. The molecular mechanisms at the basis of this phenomenon should be better clarified, as well as their biological significance and relevance for cancer therapeutics, and work is actually ongoing in our laboratory along these lines. In order to explore the translational and clinical relevance of our results, we explored the mutual interactions of APE1 and NPM1 and their associations with clinico-pathological features in a real-world cohort of TNBC patients. Both APE1 and NPM1 biomarkers were analyzed using a healthcare-grade analytical pipeline, without any ad-hoc adaptations, rendering this approach scalable and integrable with the current clinical workflow. Moreover, expression level correlations and localization associations were consistent with those observed in our preclinical models. Notably, APE1 was not associated with any common clinical characteristics, while NPM1 was associated with worse disease onset and interesting results were observed in the exploratory outcome analysis, in terms of both protein levels and localization. On the other hand, NPM1 failed to show a prognostic role for the total population; however, the subgroup analysis highlighted a potential role in terms of EFI among patients not treated with anthracyclines in the neoadjuvant or adjuvant setting, suggesting that anthracycline naïve patients with low NPM1 have a significantly higher probability of having a local or distant relapse. Since NPM1 has a pivotal role in controlling ribosomal biogenesis and genome stability [91], these data are an intriguing hint in refining patients’ stratification according to potential response to genotoxic agents. As the anthracycline-based regimen seems to bridge the gap between the NPM1 low and high groups, these data suggest that there is a subgroup of TNBC patients that not only is potentially prone to respond, but that could be burdened by a worse outcome if different therapeutic strategies were put in place. These results are of particular clinical interest since DNA-damaging strategies are an emerging opportunity for the treatment of TNBC, but a solid predictive marker is currently needed [12, 17, 92]. The inability to translate this difference on OS could be due to the small sample size, post-relapse treatments or underlying differences in terms of tumor biology. In order to further elucidate this aspect, we investigated the overall impact of APE1 and NPM1 levels through TCGA-derived RNA-seq data. To find promising companion proteins, based on the hints derived by our clinical results, we then compared patients with an OS lower or higher than 5 years, and those with a recurrence after at least 1 year from diagnosis and we identified subsets of genes (Additional file 7: Figure S6) that positively or negatively correlated with APE1 and NPM1 levels. Interestingly, the functional terms associated with these signatures highlighted biological processes that could represent new hubs for personalized anti-cancer treatments. We found of particular interest the alterations concerning cell-cycle dependent proteins observed among patients with an OS < 5 years. Targeting cell cycle is a new emerging strategy that was particularly effective in luminal-like breast cancer in reverting endocrine resistance [93], but interesting data suggest a possible role of CDK4–6 inhibitors also as chemo-companions, since CDK6 may interfere with Pt-induced cell death through FOXO3 [94]. The present study, therefore, supports the rationale of finding optimal DNA-repair - focused companions, capable to both synergize with chemotherapy and to revert potential resistance mechanisms that could defeat the DNA - damaging potential of Pt-salts [12]. Synthetic lethality involving the inhibition of the APE1 endonuclease activity is only recently emerging in literature [95, 96]. The differential response of the TNBC cell lines here tested is certainly interesting, as it could be helpful in understanding TNBC heterogeneity and consequently tailor the right therapeutic associations and sequences on the single patient. This possibility is currently under investigations in our laboratory, but it is likely that the different mutational status of p53 could have a role in the observed differences. As reported in the Materials and Methods section, the mutational status of TP53 is different among the two TNBC cell lines used in the present study. The R²⁴⁸Q mutation, occurring in HCC70 cells, represents a “hot spot” mutation in the DNA binding domain (DBD) of the transcription factor (amino acids 109–288), often found in several types of tumors. This mutation belongs to the class of mutagenic p53 that affects the DNA binding activity without destabilizing the DBD [97, 98]. Generally, R²⁴⁸Q is a functional mutation of p53 that causes an aberrant overexpression of the p53 protein [99]. Just recently, Olszewski et al. have demonstrated that the R²⁴⁸ mutational status may affect cancer cell biology, in a cell-type dependent manner, demonstrating how in breast cancer the expression of R²⁴⁸Q p53 decreases the motility and invasiveness [100], differently from what has been published in other tumors types [101, 102]. Contrarily to the HCC70 cell line, the HCC1937 cells have an acquired non-sense mutation (C³⁰⁶T) occurring on the TP53 gene and giving rise to a conversion of a codon encoding for R³⁰⁶ to a STOP codon, which occurs close to the tetramerization domain (amino acids 324–359) and thus causes a negative expression of p53 [103]. Data not shown from our laboratory (Codrich et al., submitted) clearly demonstrated that the inhibition of the APE1 endonuclease activity by Compound #3, in different cancer cell lines including HCC70 and HCC1937, causes a reduction of cell viability in a p53-dependent manner, possibly involving nucleolar stress and impairment of ribosome biogenesis [43]. Based on these premises, the lack of sensitization response of HCC1937 cells to the combined-treatment with Pt-compounds and Compound #3 could indeed be linked to a potential differential effect mediated by the p53 functional status.

In the case of BRCA1 gene, which is important for correct functional activity of HR in repairing DSBs formation generated by Pt-compounds, it was somehow surprising that HCC1937 cells (BRCA1-deficient), see Materials and Methods section for details, were more resistant to cisplatin than HCC70 cells (BRCA1-proficient). We believe that, in the time-frame and doses we considered in our experimental set up, the effect on cell viability was not only due to a difference in the DNA repair activity by HR, by means of BRCA1 functional protein, but also to damages associated to RNA processing mechanisms, which are correlated with APE1/NPM1 functional activities. Moreover, a major role of NHEJ in repairing DSBs induced by Pt-compounds treatment by HCC1937 cell line, cannot be excluded at present. Thus, further studies are needed to drive definitive conclusions on these hypotheses.

The present study analyzed the relevance of APE1, NPM1 and their interaction towards Pt-salts cell cytotoxicity and explored their role as clinically transferable biomarkers for patients’ selection and hypothesis generation. APE1 and NPM1 protect cancer cells from Pt-compounds cytotoxicity, suggesting a possible improvement of the activity of Pt-based therapy for TNBC, using the NPM1 and APE1 proteins as second therapeutic targets. We propose several TNBC gene signatures having positive or negative correlation with APE1 and NPM1. Of particular interest, APE1/NPM1 gene expression levels are associated with the alterations concerning cell-cycle dependent proteins observed among patients with an OS < 5 years.