Development and evaluation of human AP endonuclease inhibitors in melanoma and glioma cell lines

Aims: Modulation of DNA base excision repair (BER) has the potential to enhance response to chemotherapy and improve outcomes in tumours such as melanoma and glioma. APE1, a critical protein in BER that processes potentially cytotoxic abasic sites (AP sites), is a promising new target in cancer. In the current study, we aimed to develop small molecule inhibitors of APE1 for cancer therapy. Methods: An industry-standard high throughput virtual screening strategy was adopted. The Sybyl8.0 (Tripos, St Louis, MO, USA) molecular modelling software suite was used to build inhibitor templates. Similarity searching strategies were then applied using ROCS 2.3 (Open Eye Scientific, Santa Fe, NM, USA) to extract pharmacophorically related subsets of compounds from a chemically diverse database of 2.6 million compounds. The compounds in these subsets were subjected to docking against the active site of the APE1 model, using the genetic algorithm-based programme GOLD2.7 (CCDC, Cambridge, UK). Predicted ligand poses were ranked on the basis of several scoring functions. The top virtual hits with promising pharmaceutical properties underwent detailed in vitro analyses using fluorescence-based APE1 cleavage assays and counter screened using endonuclease IV cleavage assays, fluorescence quenching assays and radiolabelled oligonucleotide assays. Biochemical APE1 inhibitors were then subjected to detailed cytotoxicity analyses. Results: Several specific APE1 inhibitors were isolated by this approach. The IC50 for APE1 inhibition ranged between 30 nM and 50 μM. We demonstrated that APE1 inhibitors lead to accumulation of AP sites in genomic DNA and potentiated the cytotoxicity of alkylating agents in melanoma and glioma cell lines. Conclusions: Our study provides evidence that APE1 is an emerging drug target and could have therapeutic application in patients with melanoma and glioma.

Monofunctional alkylating agents are routinely used for the treatment of patients with advanced melanoma and glioma. However, the response rate to chemotherapy is modest and the overall prognosis is poor. The cytotoxicity of alkylating agents is directly related to their propensity to induce genomic DNA damage. However, the ability of cancer cells to recognize this damage and initiate DNA repair is an important mechanism for therapeutic resistance that negatively impacts upon therapeutic efficacy. Pharmacological inhibition of DNA repair, therefore, has the potential to enhance the cytotoxicity of alkylating agents and improve patient outcomes Madhusudan and Middleton, 2005).
The DNA base excision repair (BER) pathway is critically involved in the repair of bases that have been damaged by alkylating agents such as temozolomide and dacarbazine (Hoeijmakers, 2001). Although there is more than one sub-pathway of BER, in most cases base excision is initiated by a DNA glycosylase, which recognizes a damaged base and cleaves the N-glycosidic bond, leaving a potentially cytotoxic apurinic/apyrimidinic (AP) site intermediate (Hickson et al, 2000). This product is a target for the human AP endonuclease (APE1). The DNA repair domain of APE1 cleaves the phosphodiester backbone on the 5 0 side of the AP site resulting in a single-strand break, which is further processed by proteins of the BER pathway. AP endonuclease 1 accounts for over 95% of the total AP endonuclease activity in human cell lines (Demple et al, 1991). In addition to its DNA repair activity, APE1 also performs functions such as redox regulation (mediated through a separate redox domain) and transcriptional regulation (Xanthoudakis et al, 1992;Okazaki et al, 1994;Bhakat et al, 2003). AP endonuclease 1 is a member of the highly conserved exonuclease III family of AP endonucleases, named after the E. coli homologue of APE1 . The endonuclease IV family of AP endonucleases, the prototypical member of which is E. coli endonuclease IV (Ramotar, 1997), is structurally unrelated to APE1, despite being able to carry out the comparable AP site incision reaction Gorman et al, 1997;Hosfield et al, 1999). Using either antisense oligonucleotides or RNA interference approaches, several groups have reported that depletion of intracellular APE1 sensitizes mammalian cells to a variety of DNA damaging agents (Chen and Olkowski, 1994;Walker et al, 1994;Silber et al, 2002). In melanoma cell lines, APE1 downregulation led to increased apoptosis, whereas APE1 overexpression conferred protection from chemotherapy-or hydrogen peroxide-induced apoptosis. (Yang et al, 2005). Antisense oligonucleotides directed APE1 depletion in SNB19, a human glioma cell line lacking O(6)-methylguanine-DNA-methyltransferase, lead to potentiation of MMS and temozolomide cytotoxicity (Silber et al, 2002).
In patient tumours, APE1 expression may have prognostic and/or predictive significance. We have recently shown that APE1 expression has prognostic significance in ovarian, gastrooesophageal and pancreatico-biliary cancers (Al-Attar et al, 2010). AP endonuclease 1 is also aberrantly expressed in other human tumours and strong nuclear expression has consistently been observed in these studies (reviewed in ). In head and neck cancer, nuclear localisation of APE1 was associated with resistance to chemoradiotherapy and poor outcome (Koukourakis et al, 2001), and in cervical cancer, an inverse relationship between intrinsic radiosensitivity and levels of APE1 has been demonstrated (Herring et al, 1998).
Preclinical and clinical studies suggest that APE1 is a viable anticancer drug target. We recently initiated a drug discovery programme to identify small molecule inhibitor-lead compounds of APE1 . Fluorescence-based high throughput screening of a chemical library, as well as biochemical and cellular investigations were undertaken. We reported the identification and characterisation of CRT0044876 (7-nitro-1Hindole-2-carboxylic acid), the first small molecule inhibitor of APE1 that potentiated the cytotoxicity of alkylating agents such as temozolomide . The ability of CRT0044876 to block BER has also been demonstrated independently by other investigators (Guikema et al, 2007;Koll et al, 2008). In a recent study, BER inhibition using CRT0044876 was shown to confer selectively enhanced cytotoxicity in an acidic tumour microenvironment (Seo and Kinsella, 2009). However, the ability of CRT0044876 to block BER has not been consistently demonstrated by other groups (Fishel and Kelley, 2007) implying that further work needs to be done before a genuine lead inhibitor could emerge.
Here, we report on a new structure-based drug design strategy to identify APE1 inhibitors. This approach has allowed us to identify several novel APE1 inhibitors that potentiate the cytotoxicity of alkylating agents and that have potential as lead compounds for further optimisation and development. We also present preclinical data that support APE1 modulation as a particularly promising new strategy in melanoma and glioma where alkylating agents remain an important treatment modality.

Enzymes, oligonucleotides and chemicals
Human APE1, uracil-DNA glycosylase and E. coli endonuclease IV were obtained from New England Biolabs (Ipswich, MA, USA).

Virtual screening strategy
Virtual screening was done against the high resolution crystal structure of APE1 (PDB accession code 1BIX). Sybyl8.0 was used to build inhibitor templates based on the previously reported APE1 inhibitor  and three new pharmacophore templates designed in silico (M1, M2 and M3) based on the structural features of the APE1 active site (see results and discussion). Using these templates, ROCS 2.3 (Open Eye Scientific, Santa Fe, NM, USA) (Hawkins et al, 2007) was used to extract pharmacophorically-related (Tanimoto cut-off between 0.6 and 0.75) subsets of compounds from the ZINC database (http:// zinc.docking.org/; 2008 version with ca. 2.6 million drug-like compounds) (Irwin and Shoichet, 2005). The 1679 filtered ligands were docked into the APE active site pocket using GOLD2.7 (Hartshorn et al, 2007). Predicted ligand poses were ranked on the basis of two fitness scoring functions: GOLDScore (Jones et al, 1997)and ChemScore (Verdonk et al, 2003). A total of 100 docking runs were performed for each ligand.

Fluorescence-based AP site cleavage assay
A fluorescence-based AP site cleavage assay was performed as described previously with slight modifications . Briefly, APE1 (50 nM) (New England Biolabs) was incubated in a buffer system consisting of 50 mM Tris-HCl, pH 8.0, 1 mM MgCl 2 , 50 mM NaCl, 2 mM DTT at 371C for 10 min. 5 0 -F-GCCCC CXGGGGACGTACGATATCCCGCTCC-3 0 and its complementary Q-labelled oligonucleotide (see above) were annealed in a buffer containing 100 mM Tris-HCl, 50 mM NaCl and 1 mM EDTA. AP-site cleavage was initiated by addition of the annealed substrate (25 nM) to the reaction mix. Fluorescence readings were taken at 5 min intervals during 30 min incubation at 371C using an Envision Multilabel reader from Perkins Elmer (Cambridge, UK) with a 495 nM excitation and a 512 nM emission filter. If the DNA is cleaved at the abasic site at position 7 from the 5 0 -end by APE1, the 6-mer fluorescein-containing product will dissociate from its complement by thermal melting. As a result, the quenching effect of the 3 0 dabcyl (which absorbs fluorescein fluorescence when in close proximity) is lost, and APE1 activity is measured indirectly as an increase in fluorescence signal ( Figure 2A). Similar assays were developed for monitoring the AP endonuclease activity of endonuclease IV using a buffering system containing 10 mM HEPES-KOH, pH 7.4, 100 mM KCl and 60 ng of endonuclease IV (Trevigen, Abingdon, UK). The final DMSO concentration was maintained at 1.2% in all assays. APE1 wild-type and D148E polymorph was quantified using NanoDrop 2000c spectrophotometer (Thermo Scientific, Wilmington, NC, USA), and 50 nM of protein was used in all assays. D148E polymorph was generated as described previously (Hadi et al, 2000). Experiments were repeated at least five times.
Screening of virtual APE1 inhibitor candidates APE1 was incubated with the candidate inhibitors at 100 mM (final DMSO concentration, 1.2%) before initiating the AP site cleavage assay described in the previous section. Those candidates that showed 490% inhibition of APE1 activity were subjected to serial dilution experiments for IC 50 calculations. In addition, screening of potential inhibitors for their specificity (at 100 mM concentration) was performed using endonuclease IV cleavage assays.

IC 50 value estimations
To estimate IC 50 for APE1 inhibition, the ability of the compounds to inhibit APE1 at a range of concentrations (10nM-100 mM) were evaluated in black 384-well plates. The reactions were set up as before and fluorescence intensity was measured every 5 min for 30 min following reaction initiation. Using the initial rate values from the assay, percent activity was calculated for each sample relative to a negative DMSO only control. The data was fitted to a sigmoidal dose-response model using Graphpad Prism 3.0 (GraphPad Software, La Jolla, CA, USA) and IC 50 values were determined using the formula: % Activity ¼ 100/(1 þ 10 (log[I]Àlog IC 50) ).

Fluorescence quenching assay
To investigate the possibility that compounds might possess intrinsic quenching activity, fluorescence quenching assays were performed. Briefly, the oligonucleotides 5 0 -F-oligonucleotide (see above) and 3-CGGGGGCCCCCTGCATGCTATAGGGCGAGG-5 0 were annealed as described previously. The double stranded oligonucleotide (5 nM) was incubated with 100 mM of potential APE1 inhibitor in a buffer consisting of 50 mM Tris-HCl, pH 8.0, 1 mM MgCl 2 , 50 mM NaCl and 2 mM DTT at 371C for 30 min. Fluorescence intensity was measured every 5 min. Any hits that showed a decrease of more than 50% in the fluorescence intensity were considered as quenchers and discarded from further analyses.

Radiolabelled oligonucleotide-based APE1 cleavage assay
This basic assay was performed as described previously . Briefly, a radiolabelled uracil-containing oligonucleotide (5 0 -CTCGCAAGUGGGTACCGA-3 0 ) was annealed to a complementary oligonucleotide. To generate AP sites, the annealed DNA substrate was pretreated with uracil-DNA glycosylase and the resulting AP site was chemically reduced by the addition of sodium borohydride. AP site cleavage reaction consisted of 50 nM APE1 and 0.75 ng reduced AP site double-stranded oligonucleotide incubated at 371C for 1 h. The sample was resolved on a 15% TBE Criterion Pre Cast Gel (Bio-Rad, Hemel Hempstead, Herts, UK) and the radiolabelled substrate and reaction products were visualised using a phosphorImager (Molecular Dynamics, Buckinghamshire, UK).

Whole-cell extract AP-site cleavage assay
HeLa cells -maintained in DMEM with 10% fetal bovine serum and 1% penicillin -Streptomycin -were harvested, washed with 1 Â PBS, and the pellet was resuspended in cold 222 mM KCl plus protease inhibitors (0.5 mM PMSF, 1 mg ml À1 each of Leupepetin and Pepstatin A), incubated on ice for 30 min, and clarified by centrifugation at 12 000 Â g for 15 min at 41C (Simeonov et al, 2009). The supernatant WCE was retained, the protein concentration determined using the Bio-Rad Bradford reagent, and aliquots were stored at À801C. AP endonuclease activity assays using 18-mer radiolabelled oligonucleotide substrates (see above) were performed. In brief, all potential APE1 inhibitors were incubated at 100 mM concentrations with 30 ng of HeLa WCE at room temperature for 15 min in incision buffer consisting of 50 mM Tris-HCl, pH 8, 1 mM MgCl 2 , 50 mM NaCl and 2 mM DTT. After incubation, 0.5 pmol 32 P-radiolabeled THF-containing 18-mer double-stranded DNA substrate was added. Incision reactions were then carried out immediately at 371C for 5 min in a final volume of 10 ml after which the reaction was terminated by the addition of an equal volume of stop buffer (0.05% bromophenol blue and xylene cynol, 20 mM EDTA, 95% formamide), followed by denaturation of samples at 951C for 10 min. The radiolabeled substrate and product were separated on a standard polyacrylamidedenaturing gel and quantified by phosphorimager analysis.

Kinetics analysis
APE1 protein (80 ng) was incubated at room temperature for 30 min without or with APE1 inhibitor (5, 10 and 20 mM). Fluorescent DNA substrate was then added to a final concentration of 100, 200 and 500 nM (in 40 ml final volume), and enzyme activity was allowed to proceed for 30 min at 371C. The percentage APE1 cleavage activity was plotted. Lineweaver -Burk plots and kinetic parameters (k cat and K M ) were determined from eight independent data points.

Quantification of AP sites in genomic DNA
AP sites were quantified as described previously . Genomic DNA was extracted from a pellet of 1 Â 10 6 cells using the guanidine/detergent lysis method. Briefly, 0.5 ml APE1 inhibitors for melanoma and glioma MZ Mohammed et al of DNAzol (Helena Biosciences, Gateshead, UK) was added to the pellet and the cell lysate was gently passed several times through a pipette. The resultant viscous solution was centrifuged at 10 000 g for 10 min at 251C. DNA was precipitated from the supernatant using 0.25 ml of 100% ethanol by gently inverting the tube 5 -8 times at room temperature for 1 -3 min. The DNA was washed twice in 0.4 ml of 75% ethanol. The DNA was then solubilized in TE buffer (pH 8.0), and the final concentration was adjusted to 100 mg ml À1 (using a Gene Quant pro spectrophotometer). AP-site determinations were performed on the genomic DNA using an aldehyde reactive probe assay kit using the protocol provided by the manufacturer (BioVision Research Products, Mountainview, CA, USA). Untreated cells were compared with cells exposed to either MMS alone, APE1 inhibitor alone or combination of MMS and APE1 inhibitor. DNA was extracted at 90 min and AP site quantified as described previously. All experiments were performed in triplicate.

AQ ueous non-radioactive cell proliferation assay (MTS assay)
To evaluate intrinsic cytotoxicity and to evaluate the potentiation of toxicity of cytotoxic agents by APE1 inhibitors, MTS assays were performed as per the manufacturer's recommendation (Promega, Southampton, UK). Briefly, 2000 cells per well (in 200 ml of medium) were seeded into a 96-well plate. For HUVEC cells, 5 ml of 2% type 2 gelatine (Sigma) was added to the wells and the plates were preincubated for 20 min at 371C before seeding of cells. For intrinsic cytotoxicity assessments, cells were incubated with varying concentrations of APE1 inhibitors and the MTS assay was performed on day 5. For potentiation experiments, cells were preincubated with a relatively nontoxic concentration of APE1 inhibitor for 24 h and then exposed to MMS, temozolomide or doxorubicin. Non-radioactive cell proliferation assay was conducted as described previously.

Virtual screening
The virtual screening process requires the precise definition of the ligand-binding site in the target protein. The DNA repair domain active site was localised on the basis of the previously reported 10 critical amino acid residues that are essential for the AP endonuclease activity of APE1 (D70, D90, E96, Y171, D210, N212, D219, D283, D308 and H309) Rothwell and Hickson, 1996;Erzberger and Wilson, 1999;Fritz et al, 2003;Mundle et al, 2004). The active site is a welldefined deep V-shaped cleft, with a Mg 2 þ ion at its 'elbow' ( Figure 1A).
Our virtual screening strategy was to take a known 'first generation' APE1 inhibitor, plus prototypical molecular scaffolds designed on the basis of the shape of the ligand-binding site, and perform a rapid structure-based similarity search of a large virtual library of drug-like molecules. 'Hits' from this search were then subjected to the more computationally costly process of dockingbased evaluation. We used Sybyl8.0 to build molecular models for the previously reported APE1 inhibitor, CRT0044876 (Figures 1B), and to build models for three prototypical scaffolds (M1, M2 and M3) (Figures 1B) that were predicted to fit well into the APE1 binding site cleft and interact with key residues. Template M1 features a central tetrahedral centre bearing a potential Mg 2 þ -interacting carboxylate group plus two heteroaromatic branches that have dimensions and relative orientations designed to fit snugly into the active site groove. Template M2 bears the same key features, but the heteroaromatic substituents are extended to interact with more of the groove . Template M3 bears an additional heteroaromatic sidechain that can access a subsidiary cleft in one branch of the ligand-binding groove (Figures 1B).
Using these templates, a shape-based similarity searching strategy using ROCS 2.3 (OpenEye Scientific)(Hawkins et al,  . The conformations of these compounds were then energy minimised and subjected to docking against the active site of the APE1 model. A consensus score plot was constructed for each virtual hit by adding the GOLDScore and ChemScore (Figure 2A). The top ranking 25% of the compounds were shortlisted from the consensus plot and subjected to detailed biochemical analyses.

Inhibitory activity of compounds against the D148E polymorphic variant of APE1
The D148E polymorphic variant of APE1 has been implicated in cancer predisposition including melanoma (Li et al, 2006;Farkasova et al, 2008;Gu et al, 2009). In addition, the D148E polymorph may also alter ionising radiation sensitivity (Hu et al, 2001). We tested if our isolated inhibitors would have differential  Figure 2 (A) Consensus score plot was constructed by plotting Gold Score (x-axis) and Chem Score (y-axis) for the 1679 virtual APE1 inhibitor candidates. The top ranking 25% of the compounds were shortlisted from the consensus plot and subjected to detailed biochemical analyses. (B) Primary screening. Fluorescence-based APE1 cleavage assay is shown here. If the DNA is cleaved at the abasic site at position 7 from the 5 0 end by APE1, the 6-mer fluorescein-containing product will dissociate from its complement by thermal melting. As a result, the quenching effect of the 3 0 dabcyl (which absorbs fluorescein fluorescence when in close proximity) is lost, and APE1 activity is measured indirectly as an increase in fluorescence signal. For detailed protocol see Materials and methods section. (C) APE1 inhibition by CRT0044876 is shown here. Control ¼ no APE1 in reaction. (D) APE1 inhibition by compound 4 is shown here (IC 50 ¼ 11 mM).  activity against the variant compared with the wild-type protein.
Although the AP-site cleavage activity of D148E variant was similar to that of the wild type ( Figure 4C), consistent with a previous report (Hadi et al, 2000), Figure 4D demonstrates that for compound 4, the IC 50 for APE1 inhibition was significantly reduced by 50.5% for the D148E protein (5.56 mM) compared with the wild type (11 mM). The preferential inhibitory activity of compound 4 towards the D148E protein was also confirmed in radiolabelled oligonucleotide assays (data not shown). We were not able to demonstrate preferential inhibitory activity of other compounds either in fluorescence or radiolabelled assays.

Kinetics analyses
To evaluate potential mechanism of action of APE1 inhibitor, kinetic analysis was performed ( Figure 5). As compound 4 had the strongest inhibitory activity (490% inhibition) in whole-cell extracts, we selected this compound for kinetic analysis. Lineweaver -Burk plots and kinetic parameters was determined from eight independent data points. K M and k cat decreased at each inhibitor concentration (compared with no inhibitor) and the k cat /K M decreased at increasing inhibitor concentration. The data is consistent with uncompetitive inhibition.
However, we cannot exclude the possibility that compound 4 operates as a weak uncompetitive inhibitor (meaning it binds the protein -DNA substrate complex), as we observed a reproducibly lower K M in the presence of the compound, though this is unlikely.

Genomic AP site accumulation in cells
In order to test the biological activity of APE1 inhibitors under physiological conditions, analysis was then undertaken in melanoma cell lines (MeWo, SKMel and MM418) and glioma cell lines To evaluate potential mechanism of action of APE1 inhibitor, kinetic analysis was performed. Lineweaver -Burk plots and kinetic parameters determined from eight independent data points (note: error bars are in some cases too small to see) for compound 4 is shown here. The APE1 inhibitor was tested at three dose levels (5, 10 and 20 mM) and oligonucleotide substrate was evaluated at three different concentrations (100, 200 and 500 nM). The reaction was performed as described in methods. K M and k cat decreased at each inhibitor concentration (compared with no inhibitor) and the k cat /K M decreased at increasing inhibitor concentration. The data is consistent with uncompetitive inhibition.
(U89MG and SNB-19). We initially tested if these cell lines expressed APE1 protein. Robust APE1 expression was seen in these cell lines using western blot analyses ( Figures 6A and 7A). In order to confirm that the isolated inhibitors block APE1 function in living cells, the aldehyde reactive probe assay that allows quantification of genomic AP sites was utilised in this study. Figure 6B shows that compared with untreated cells, glioma cells exposed to compound 4 accumulated AP sites confirming target  Figure 7D). We took the survival fraction as 100%. The percentage survival for those cells exposed to both inhibitor and temozolomide was plotted as a relative survival to cells exposed to the inhibitor alone. Potentiation of cytotoxicity of MMS by compound 4 (10 mM) in U89 MG cell line is shown here.
(D) Potentiation of temozolomide by compound 4 (10 mM) in U89MG cell line is shown here.  Figure 7D). We took the survival fraction as 100%. The percentage survival for those cells exposed to both inhibitor and temozolomide was plotted as a relative survival to cells exposed to the inhibitor alone. Compound 4 was relatively nontoxic to HUVEC cells.
inhibition in vivo. As AP sites are obligatory intermediates during the repair of MMS-induced base damage, accumulation of AP sites were also demonstrated in cells exposed to MMS alone. Moreover, AP-site accumulation in cells exposed to a combination of APE1 inhibitor and MMS was more than the cells exposed to either agent alone. Similar accumulation of AP sites was also demonstrated in melanoma cells.
Cytotoxicity analysis in melanoma, glioma and HUVEC endothelial cell lines Potentiation of cytotoxicity was also demonstrated with other APE1 inhibitors that showed moderate to strong WCE AP-site cleavage inhibition (compound 2, 5 and 6) but not with mild WCE AP-site cleavage inhibition (compound 1). Compound 7, which was a non-specific inhibitor (i.e blocked both APE1 and endonuclease IV), did not show any potentiation of cytotoxicity and Compound 3, which was a specific APE1 inhibitor but had no activity in WCE assay, also did not shown any potentiation of cytotoxicity (data not shown).
To exclude non-specific activity and potentiation, we performed toxic studies using doxorubicin. Compound 4 did not potentiate the cytotoxicity of doxorubicin in melanoma (SK-Mel30) and glioma cell line (U89MG) (Figures 8A and B). Similar results were seen for MeWo, MM418 and SNB-19 cells.
In order to investigate whether APE1 inhibitor was toxic to noncancer cells, we conducted toxicity analysis in HUVEC endothelial cells. Figure 7D shows that compound 4 was relatively non-toxic to HUVECs compared with melanoma (SK-Mel30) and glioma (U89MG) cell lines. Similar results were seen for MeWo, MM418 and SNB-19 cells.

DISCUSSION
The overall prognosis of advanced melanoma and glioma remains poor and strategies to improve tumour response to chemotherapy remain a high priority. Blocking DNA repair may enhance cell kill in cancer and improve outcomes Madhusudan and Middleton, 2005). APE1, a critical protein in BER, is involved in the pathogenesis of glioma and melanoma. Elevated AP endonuclease activity is frequently seen in human glioma tumours (Bobola et al, 2001). Moreover, in preclinical studies, antisense oligonucleotides directed APE1 depletion in SNB19, a human glioma cell line lacking O(6)-methylguanine-DNA-methyltransferase, lead to potentiation of MMS and temozolomide cytotoxicity, implying that pharmacological modulation of APE1 is a promising strategy in glioma (Silber et al, 2002). A recent study has demonstrated that microphthalmia-associated transcription factor (MiTF), a key transcription factor for melanocyte lineage survival, regulates APE1 expression. Microphthalmia-associated transcription factor-positive melanoma cell lines accumulated high levels of APE1 (Liu et al, 2009). In a separate study, downregulation of APE1 using antisense constructs promoted apoptosis in melanoma cell lines (Yang et al, 2005). Interestingly, the APE1 genetic polymorphism D148E may also alter melanoma predisposition (Li et al, 2006). These studies therefore suggest that APE1 is also a novel target in melanoma. In this investigation, we have focussed on the development of novel APE1 small molecule inhibitors and have provided the first evidence that blocking APE1 is a promising strategy in melanoma and glioma cells.
Our previous study provided the first evidence that small molecule inhibition of APE1 is a viable anticancer strategy . In order to develop novel drug-like chemotypes, we recently adopted a virtual screening approach. The architecture of the active site of APE1 in the absence and presence of bound AP-DNA indicates that there is little or no remodelling of the active site upon substrate binding, a feature that is suitable for a virtual screen Gorman et al, 1997). We have exploited the structural features of APE1 to develop an enhanced virtual screening strategy and identified several novel small molecule inhibitors for further drug development. Three new pharmacophore templates were designed in silico (M1, M2 and M3) and a total of 1679 virtual hits with similarities to the templates were identified (CRT template ¼ 359, M1 template ¼ 373, M2 template ¼ 459 and M3 template ¼ 488). Detailed biochemical screening showed that majority of the compounds conform to the M3 template, which bears an additional heteroaromatic sidechain that can access a subsidiary cleft in one branch of the ligand-binding groove ( Figures 1B). Although the structural details of M3 template binding to APE1 active site is unknown, cocrytallization trials may provide structural insight to guide a rational drug-design strategy.
In this study, we also provide evidence for the first time that certain APE1 inhibitors may be more effective in blocking the endonuclease activity of the D148E polymorph (a common polymorph associated with cancer predisposition) compared with the wild type. The inability of six of the seven compounds examined to inhibit the activity of endonuclease IV provides presumptive evidence that the compounds indeed act by interaction with APE1 rather than by obscuring the abasic site on the DNA substrate. Moreover, the kinetics analysis has provided insight into the mechanism of action of the inhibitor. We have shown that compound 4 decreased K M , k cat (compared with no inhibitor) and decreased the k cat /K M implying uncompetitive inhibition. Future cocrytallization experiments in the presence of DNA are likely to provide further information regarding the exact mechanism of action of this compound. To assess potency and specificity of our compounds, we screened their ability to block AP-site cleavage activity using WCE. This is a good system to screen for compounds that may have non-specific binding to other cellular proteins. Compound 4 exhibited more than 90% inhibition in the WCE assays, implying strong potency and specificity. Although compound 3 blocked APE1-directed AP-site cleavage activity in purified APE1-based assay, it had no effect in the WCE assay. This implies that the compound has 'off target' non-specific protein-binding effect and suggests that it is unlikely to be a good development candidate.
In order to provide further preclinical evidence that blocking the repair domain of APE1 is a potential treatment strategy, we conducted studies in glioma and melanoma cell lines. We confirmed APE1 expression in these cancer cell lines. We then confirmed accumulation of AP sites in vivo in cells exposed to inhibitor, providing direct evidence of target inhibition in vivo. Intrinsic cytotoxicity for several of the inhibitors was demonstrated in glioma and melanoma cell lines, a finding consistent with the observation that APE1 downregulation in melanoma cell lines promotes apoptosis, although non-specific toxicity at higher doses of the compound cannot be excluded in our study (Yang et al, 2005). Interestingly, the inhibitors were relatively non-toxic to HUVEC cells implying selectivity to cancer cells. In a recent study, BER inhibition using CRT0044876 was shown to confer selectively enhanced cytotoxicity in an acidic tumour microenvironment (Seo and Kinsella, 2009), suggesting a further novel opportunity to target tumours. We then showed potentiation of MMS and temozolomide cytotoxicity in melanoma and glioma cell lines. We did not observe potentiation of doxorubicin toxicity in these cell lines implying that APE1 inhibitor potentiates chemotherapy that induce base damage and repaired through BER. Moreover, potentiation of cytotoxicity was not demonstrated in HUVEC cells, again implying selectivity to cancer cells. These studies indicate that APE1 inhibitors, either alone or in combination with chemotherapy, may be a promising strategy in cancer.
Following our initial report, other investigators have identified various APE1 inhibitors for potential pharmaceutical application (Seiple et al, 2008;Simeonov et al, 2009;Zawahir et al, 2009). In conclusion, these studies and our two reports (including this one), confirm the validity of APE1 as an emerging anti-cancer drug target.