APE1 has been found not only to be required for animal viability, as deletion of both alleles of the APE1 gene in mice leads to embryonic lethality, but also for cell viability in culture [12,13]. Elevated levels of APE1 have been found in medulloblastoma and primitive neuroectodermal tumors, prostate cancers, headand-neck cancers, non-small cell lung carcinomas, gliomas, and osteosarcomas [4]. Over-expression of APE1 has been correlated with increased cellular resistance to chemotherapeutic agents. Moreover, APE1-deficient cells exhibit hypersensitivity to methyl methanesulfonate (MMS), hydrogen peroxide, bleomycin, temozolomide, gemcitabine, 1,3-bis (2-chloroethyl)-1-nitrosourea (a.k.a. Carmustine), and the nucleoside analogue ?L-dioxolane-cytidine (a.k.a. Troxacitabine) [4,14]. Furthermore, the expression of a dominant-negative APE1 protein (termed ED), which binds with high affinity to substrate DNA and blocks subsequent repair steps [15], augments the cell killing effect of 5-fluorouracil and 5fluorodeoxyuridine, implicating BER in the cellular response to such anti-metabolites [16]. These data indicate that APE1 is an attractive and rational target in the effort to improve therapeutic efficacy of clinical DNA-interactive drugs through the inactivation of the critical BER pathway. A significant limitation of anti-cancer cytotoxins is their harmful side-effects on normal tissue.
While combinatorial treatment strategies are still of interest, researchers and clinicians have been pursuing the idea of synthetic lethality to reduce potential offtarget toxicities. In this scenario, inhibition of two independent processes separately has little cellular consequence, whereas inactivation of both pathways simultaneously leads to cell death. This model has been exploited in the case of cancers deficient in the breast cancer-related homologous repair proteins, BRCA1 and BRCA2. Here, inhibitors against the poly(ADP-ribose) polymerase protein, PARP-1, which operates in strand break responses, including the single-strand break repair sub-pathway of BER, have been shown to induce selective cell killing of BRCA-deficient cells [8,17], presumably due to replication fork collapse and increased genetic instability. Relevant to the effort within, inhibitors against APE1 have been found to be synthetically lethal to cells deficient in BRCA1 or BRCA2, or the checkpoint signaling protein ATM, inducing accumulation of DNA double-strand breaks as well as G2/M cell cycle arrest [18]. A number of chemical libraries have been screened to identify small molecule inhibitors of APE1 endonuclease activity [19,20,21,22,23]. Several molecules were identified from these efforts, including 7-nitro-indole-2-carboxylic acid identified from a screen of a 5000-compound collection [19]; several arylstibonic acid derivatives identified from a screen of the National Cancer Institute Diversity Set [20]; Reactive Blue 2, 6-hydroxy-DLDOPA, and myricetin, reported as prioritized hits from a screen of the LOPAC1280 collection of bioactive compounds [21]; and 2,4,9trimethylbenzo [b] [1,8] naphthyridin-5-amine (dubbed AR03), selected as the top hit from a 60,000-member library screen [23]. Additionally, an in-silico screen based on a pharmacophore approach has led to the identification of several APE1 inhibitors sharing a hydrophobic middle segment to which at least two carboxyl substituents (or other negatively charged groups) are attached via a range of linkers [22]; however, APE1 inhibition has not been demonstrated for these compounds in cell-based models. At present, none of the above compounds has been shown to have clinical utility and, with very few exceptions, the inhibitors reported to date are not readily amenable to further optimization by medicinal chemistry due to multiple liabilities stemming from their chemical structure [24]. We describe herein the first small molecule inhibitors of human APE1 identified by quantitative high-throughput screening (qHTS) [25] of a large public compound collection, the Molecular Libraries Small Molecule Repository (MLSMR) of .300,000 compounds, as well as additional public libraries of the NIH Chemical Genomics Center. Prioritized hits were further characterized by a panel of biochemical assays and in MMS cell toxicity potentiation models. Select compounds were also tested in an AP site measurement assay designed to ascertain the inhibitors’ effect on APE1 within a cellular context.
Materials and Methods Reagents
Dimethyl sulfoxide (DMSO, certified ACS grade) was purchased from Thermo Fisher Scientific. Tris-HCl, Thiazole Orange (ThO), Tween-20, NaCl, and MgCl2 were obtained from Sigma, while the arylstibonic control inhibitor NSC-13755 was supplied by the National Cancer Institute Developmental Therapeutics Program Repository. The enzymes and substrates employed in this study (human APE1 and E. coli Endo IV enzymes, fluorogenic and radioassay substrates) were prepared and characterized as described previously [21]. The fluorescent probe employed in the fluorescence polarization (FP) displacement assay was of the same composition as the fluorogenic HTS substrate with the exception of the Black Hole Quencher-2 moiety. The double-stranded 17-mer DNA fragment used in the ThO DNA binding assay was of the same sequence as the fluorogenic HTS substrate but carried no dye labels.
Compound Library
The 352,498-member library comprised two main subsets: NIH Molecular Libraries Small Molecule Repository (MLSMR), prepared as 10 mM stock solutions in 384-well plates and delivered by Biofocus DPI (South San Francisco, CA, http:// mli.nih.gov/mli/compound-repository/mlsmrcompounds/), and NCGC internal exploratory collection, which consisted of several commercially available libraries, as well as collections from academic compound libraries. The compound library was sourced as DMSO solutions at initial concentrations ranging between 2 and 10 mM and was further serially diluted for qHTS in 1536-well format as described in detail elsewhere [26,27]. For follow-up testing of primary screen actives, screening hits and their analogues were sourced as powders from the respective original suppliers (Sigma-Aldrich, NCI, Asinex, ChemBridge, Tocris, Ambinter, and ChemDiv), dissolved in DMSO to produce 10 mM initial stock solutions, and serially diluted in twofold steps for a total of 12 concentrations in duplicate.