Nedd8-Activating Enzyme Inhibitor MLN4924 Provides Synergy with Mitomycin C Through Interactions with ATR, BRCA1/BRCA2, and Chromatin Dynamics Pathways

Published OnlineFirst March 26, 2014; DOI: 10.1158/1535-7163.MCT-13-0634

Molecular Cancer Cancer Biology and Signal Transduction Therapeutics

Nedd8-Activating Enzyme Inhibitor MLN4924 Provides Synergy with Mitomycin C through Interactions with ATR, BRCA1/BRCA2, and Chromatin Dynamics Pathways

Khristofer Garcia1, Jonathan L. Blank1, David C. Bouck1, Xiaozhen J. Liu1, Darshan S. Sappal1, Greg Hather2, Katherine Cosmopoulos1, Michael P. Thomas1, Mike Kuranda1, Michael D. Pickard2, Ray Liu2, Syamala Bandi3, Peter G. Smith1, and Eric S. Lightcap1

Abstract MLN4924 is an investigational small-molecule inhibitor of the Nedd8-activating enzyme currently in phase I clinical trials. MLN4924 induces DNA damage via rereplication in most cell lines. This distinct mechanism of DNA damage may affect its ability to combine with standard-of-care agents and may affect the clinical development of MLN4924. As such, we studied its interaction with other DNA-damaging agents. Mitomycin C, cisplatin, cytarabine, UV radiation, SN-38, and gemcitabine demonstrated synergy in combination with MLN4924 in vitro. The combination of mitomycin C and MLN4924 was shown to be synergistic in a mouse xenograft model. Importantly, depletion of genes within the ataxia telangiectasia and Rad3 related (ATR) and BRCA1/BRCA2 pathways, chromatin modiﬁcation, and transcription-coupled repair reduced the synergy between mitomycin C and MLN4924. In addition, comet assay demonstrated increased DNA strand breaks with the combination of MLN4924 and mitomycin C. Our data suggest that mitomycin C causes stalled replication forks, which when combined with rereplication induced by MLN4924 results in frequent replication fork collisions, leading to cell death. This study provides a straightforward approach to understand the mechanism of synergy, which may provide useful information for the clinical development of these combinations. Mol Cancer Ther; 13(6); 1625–35. 2014 AACR.

Introduction and regulation of p21, p27, p53, Cdc25A, Wee1, claspin, and MLN4924 is being developed as an inhibitor of the FANCM (6–10). Many standards of care in cancer chemo- Nedd8-activating enzyme (1) and is currently in phase I therapy are DNA-damaging agents and mechanistic inter- clinical trials. Nedd8 is a small ubiquitin-like protein in actions between MLN4924 and these agents may influence which the best characterized role is the activation of a class the clinical development of MLN4924. of E3 ubiquitin ligases known as cullin RING ligases (CRL; Although the mechanism engaged by MLN4924 to ref. 2). Multiple studies have demonstrated the import- induce DNA damage seemingly differs from other ance of the induction of rereplication, resulting in DNA DNA-damaging agents, the pathways involved in the damage, to the mechanism of cell death by MLN4924 (3–5). repair of the DNA damage may overlap. A previous In addition, CRLs are known to be involved in the regu- siRNA screen that evaluated the genetics of sensitivity of lation of multiple DNA replication and repair pathways, A375 and HCT-116 cell lines to MLN4924 demonstrated including nucleotide excision repair, histone modification, an engagement of a p21-dependent intra–S-phase checkpoint in A375 cells and an Emi1-dependent G2–M checkpoint in HCT-116 cells (5). In addition, roles for p53, Authors' Affiliations: Departments of 1Discovery, 2Clinical Biostatistics, BRCA1/BRCA2, and nucleotide excision repair were and 3Information Technology, Takeda Pharmaceuticals International Co., Cambridge, Massachusetts identified. Oncogene-induced replication stress seems to play a central role in tumor progression (11, 12), suggest- Note: Supplementary data for this article are available at Molecular Cancer Therapeutics Online (http://mct.aacrjournals.org/). ing that rereplication may directly challenge tumor vul- nerabilities. However, it is not clear that rereplication will Current address for K. Cosmopoulos: Agios Pharmaceuticals, Cambridge, be the dominant mechanism of DNA damage in combi- Massachusetts; and current address for M.P. Thomas and P.G. Smith, H3 Biomedicine, Inc., Cambridge, Massachusetts. nation with other DNA-damaging agents and, thus, it is difficult to predict the best combination partners. Corresponding Author: Eric S. Lightcap, Discovery Oncology Biology, Takeda Pharmaceuticals International Co., 40 Landsdowne Street, Cam- We have conducted a systematic evaluation of the bridge, MA 02139. Phone: 617-551-3717; Fax: 617-551-8906; E-mail: effectiveness of combining MLN4924 with DNA-damag- [email protected] ing agents across four different cancer cell types in vitro doi: 10.1158/1535-7163.MCT-13-0634 and demonstrate synergy with mitomycin C, cytara- 2014 American Association for Cancer Research. bine, cisplatin, UV, SN-38, melphalan, etoposide, and

www.aacrjournals.org 1625

Garcia et al.

gemcitabine. Mitomycin C gave the most consistent syn- CCL-185, HCT-116 ATCC CCL-247, and U-2 OS ATCC ergistic interaction with MLN4924 across cell lines. Much HTB-96). Consistent regions of amplification and dele- work has been done to understand the mechanism of tions were found between the cell lines of the same origin, DNA repair following mitomycin C treatment (13, 14), and correlation on genotypes was greater than 0.995. Cell providing a background to understand these results. lines were passaged for less than 8 weeks following Mitomycin C alkylates DNA with about 10% of DNA resuscitation. modifications resulting in interstrand crosslinks, the remainder being monoadduct and intrastrand crosslinks Evaluation of cell line sensitivity to DNA-damaging (14). The mechanism by which the cell repairs this DNA agents and evaluation of combination effects damage is determined by how the adduct is detected, Cells were grown in their respective growth media, which is significantly influenced by the stage of the cell supplemented with 10% FBS: A375, Dulbecco’s Modified cycle in which the damage occurs. During G1, adduct can Eagle Medium; A549, Ham’s F-12K (Kaighn’s) medium; be detected by components of either global genome nucle- HCT-116 and U-2 OS, McCoy’s 5A medium with 1% otide excision repair or by transcription-coupled nucleo- glutamine. tide excision repair (TC-NER) if the interstrand crosslink A375 melanoma (800 cells/well), HCT-116 colorectal blocks RNA polymerase. During S-phase, encounter of carcinoma (800 cells/well), A549 lung carcinoma (1,000 the interstrand crosslink by the replication fork becomes a cells/well), and U-2 OS osteosarcoma (800 cells/well) significant mechanism for detection of the crosslink and were seeded on 384-well poly-D-lysine (PDL)-coated the repair mechanism is believed to be substantially black, clear-bottom plates (BD BioCoat) and allowed to different from that experienced during G1 (14). Notably, adhere for 24 hours at 37 C, 6% CO2. Cells were then during S and G2 phases of the cell cycle, sister chromatids treated with compounds, either alone or in combination are available for exchange by homologous recombination, with MLN4924, at various doses for 48 hours (HCT-116 providing a high-fidelity mechanism of repair. and A375, doubling times of 18 and 16 hours, respectively) To gain a more complete understanding of the synergy or 72 hours (A549 and U-2 OS, doubling times of approx- between MLN4924 and mitomycin C, we used an siRNA imately 24 hours). Viability was assessed with CellTiter- screen to evaluate the effect of gene depletion on this Glo Cell Viability reagent according to the manufacturer’s synergy. Many of the genes identified in the siRNA screen instructions (Promega). Luminescence was measured underwent posttranslational modifications to their pro- using a LEADseeker imaging system (GE Healthcare). tein products upon treatment with the combination of MLN4924 will be made available to qualified researchers MLN4924 and mitomycin C, confirming their likely once a standard Material Transfer Agreement has been involvement. Comet assay experiments demonstrated executed. early induction of DNA strand breaks by the combination of MLN4924 and mitomycin C. Our results suggest that Calculation of combination metrics mitomycin C increases the frequency of replication fork The relationship between the normalized viability collision induced by MLN4924 and that the BRCA1–clas- and drug concentrations was fit with a nine parameter pin–ATR–Chk1 pathway plays an important role in the response surface model (15). To quantify the synergy, response to these collisions. These studies may provide a the combination index (16) or nonlinear blending (17) way of identifying promising drug combinations and was computed. Additional details can be found in understanding the mechanistic details of their synergy. Supplementary Materials and Methods. Such knowledge may more efficiently guide clinical development of chemotherapeutic combinations. The effect of depletion of DNA damage response genes on the synergy between MLN4924 and mitomycin C Materials and Methods A375 cells were cultured and reverse transfected with Genomic characterization and authentication of cell siRNA SMARTpools or duplexes targeting DNA damage lines response genes using DharmaFect4 transfection lipid A375, A549, HCT-116, and U-2 OS were received from (Dharmacon, GE Healthcare), as previously described the American Type culture collection (ATCC) in July 2006, (5). After 48 hours knockdown, cells were treated with February 2004, June 2009, and July 2006, respectively, and DMSO (dimethyl sulfoxide; negative control), 330 nmol/L passaged twice before freezing. Genomic DNA was iso- MLN4924, 1.6 or 1 mmol/L mitomycin C, or the combi- lated from 2 million cells under vendor-specified proto- nation of 330 nmol/L MLN4924, and 1.6 or 1 mmol/L cols. Of note, 500 mg DNA per sample was then amplified, mitomycin C, in the continued presence of siRNA oligos, labeled, and processed on the Affymetrix SNP-6 whole and incubated for a further 48 hours, after which viability genome array platform on November 7, 2011, and eval- was assessed with ATPlite reagent according to the man- uated using Partek Genomics Suite for copy-number ufacturer’s instructions (PerkinElmer). HCT-116 cells variation analysis. Spearman correlation was determined were also transfected as previously described (5). After for each cell line against an ATCC-authenticated stock vial 48 hours knockdown, cells were treated with DMSO received at Takeda (A375 ATCC CRL-1619, A549 ATCC (negative control), 110 nmol/L MLN4924, 450 nmol/L

1626 Mol Cancer Ther; 13(6) June 2014 Molecular Cancer Therapeutics

MLN4924 DNA-Damaging Agent Combinations

mitomycin C, or 110 nmol/L MLN4924, and 450 nmol/L Odyssey Infrared Imager (LI-COR Biosciences) for visu- mitomycin C for an additional 48 hours. alization and quantitation. Antibodies and their condi- The interaction (BI score) between MLN4924 and mito- tions are given in Supplementary Table S1A. Full-length mycin C was calculated from log2-transformed data as Western blots are included in Supplementary Fig. S1. ¼ V þV V V V BI x4m x x4 xm, where x4m is the viability in the presence of siRNA, MLN4924, and mitomycin C, Vx is Alkaline comet assay V 6 the viability in the presence of siRNA alone, x4 is the A375 cells were plated (0.2 10 cells/well) and incu- viability in the presence of siRNA and MLN4924, and Vxm bated overnight in 6-well tissue culture dishes. Cells were is the viability in the presence of siRNA and mitomycin C. then treated with compounds, either as single agent or in combination treatments using mitomycin C and Evaluation of synergy in the A375 xenograft model MLN4924 for 6 hours. A 6-hour treatment using etoposide Female Balb/c Nude mice (Shanghai Laboratory Ani- was used as a positive control for induction of DNA strand mal Center, Shanghai, China) were inoculated with 5 breaks. 106 A375 tumor cells (0.1 mL in 50% Matrigel) in the right Preparation and execution of the alkaline comet assay flank. After tumors grew to an average size 150 mm3, mice was done according to the Trevigen CometAssay protocol were randomly assigned into treatment groups of 10 mice for single-cell gel electrophoresis provided with the each (vehicle, MLN4924, mitomycin C, or the combination reagent kit apart from the electrophoresis protocol (cat- of MLN4924 and mitomycin C). Vehicle (20% HPbCD) alog# 4250–050-K). Additional details can be found in was dosed subcutaneously on days 1, 4, 8, 11, 15, and 18 Supplementary Materials and Methods. (twice weekly 3). MLN4924 was formulated in 20% HPbCD and dosed subcutaneously twice weekly 3at Results 180 mg/kg. Mitomycin C (Intas Pharmaceuticals, Ltd.) was formulated in 0.9% saline and dosed intravenously at MLN4924 synergizes with multiple DNA-damaging 2.5 mg/kg on days 1, 8, and 15 (once a week for 3 weeks). agents Body weight and tumor growth were monitored twice a We profiled the cell lines A375 (melanoma), A549 (non– week. All mice had access to food and water ad libitum and small cell lung cancer), HCT-116 (colon), and U-2 OS were housed and handled in the AAALAC-certified facil- (osteosarcoma) to evaluate their sensitivity to 14 agents ity at the Medicilon Preclinical Research, LLC, in accor- (Supplementary Fig. S2), selected across a breadth of dance with the Medicilion Institutional Animal Care and mechanism, including cross-linking agents, topoisomer- Use Committee Guidelines. ase inhibitors, nucleoside analogs, and the PARP inhibitor ABT-888 for evaluation in combination with MLN4924 in Cell culture and reverse transfection for Western the four cancer-derived cell lines. All four cells lines were > m blot analysis insensitive to the PARP inhibitor (LC50 25 mol/L), as A375 melanoma cells were maintained as previously expected (18–20). Paclitaxel and bortezomib were includ- described (5). For Western blot analysis of protein knocked as inhibitors that do not directly cause DNA damage, down, A375 cells (57,000 cells/plate) were reverse trans- although bortezomib is likely to affect DNA damage fected in 10-cm diameter PDL-coated cell culture plates repair (21). (CELLCOAT; Greiner Bio-One) with DharmaFECT 4 Compound combinations were evaluated as a 10 10 reagent (DH4; Dharmacon) using a final concentration of matrix of drug concentrations in duplicate with simulta- 15 nmol/L siRNA (siGENOME SMARTpool; Dharma- neous addition to cells. Data were fitted to a dose– con) and incubated for 48 hours at 37C to allow for response surface (15) and isobolograms were derived and protein knockdown. Cells were then treated with 330 plotted (Fig. 1 and Supplementary Fig. S3). Combination nmol/L MLN4924 and 1 mmol/L mitomycin C or vehicle index (16) and nonlinear blending (17) values were control (0.07% DMSO) for a further 24 hours before lysis. calculated for each experiment, based upon the LC50 isobologram (Fig. 2 and Supplementary Fig. S4). Both Sample preparation and Western blot analysis calculations use Loewe additivity as the null model Cell extracts were prepared by resuspension of the cell (22). As expected, MLN4924 combined with itself was pellet in ice-cold RIPA buffer with 0.5 U/mL benzonase clearly additive (Fig. 2 and Supplementary Fig. S3 and S4). nuclease HC (Novagen). Protein concentrations were Mitomycin C, cisplatin, cytarabine, and gemcitabine determined with the BCA Kit (Pierce) according to the gave significant synergy (Fig. 1, concave curves). Gener- manufacturer’s instructions using bovine serum albumin ally, responses to the various compounds correlated as standard. Samples (5–10 mg protein) were run on between cell lines (Supplementary Fig. S3). Gemcitabine NuPAGE 4% to 12% Bis–Tris gels in MES/SDS buffer or was the exception, with the observation of both significant on NuPAGE 3% to 8% Tris–Acetate gels in Tris–Acetate/ synergy and antagonism (convex curves), depending on SDS buffer (Invitrogen). Proteins were transferred to the cell line (Fig. 1D). Intriguingly, synergy with gemci- immobilon-FL polyvinylidene difluoride membranes tabine occurred in those cell lines with more extensive using the semidry method and probed with the indicated rereplication induction by MLN4924 [>4N DNA ¼ 30% antisera using tubulin as a normalization control and an (U-2 OS) and 31% (HCT-116)], whereas antagonism

www.aacrjournals.org Mol Cancer Ther; 13(6) June 2014 1627

Garcia et al.

A B Mitomycin C Cisplatin

A375 A549 A375 A549 MLN4924 (nmol/L) MLN4924 Mitomycin C (nmol/L) Cisplatin (nmol/L) Cisplatin (nmol/L) 200 400 600 800 200 400 600 800 5,000 15,000 25,000 0 0 0 0 20,000 60,000 0 500 1,500 2,500 3,500 0 500 1,000 2,000 0 200 600 1,000 1,400 0 2,000 4,000 6,000 Mitomycin C (nmol/L) MLN4924 (nmol/L) MLN4924 (nmol/L) MLN4924 (nmol/L)

HCT116 U2OS HCT116 U2OS 10,000 60,000 Cisplatin (nmol/L) Cisplatin (nmol/L) Mitomycin C (nmol/L) Mitomycin C (nmol/L) 20,000 2,000 6,000 10,000 10,000 30,000 50,000 0 0 2,000 6,000 0 0 0 100 200 300 400 0 500 1,500 2,500 0 100 200 300 400 500 600 0 500 1,000 1,500 MLN4924 (nmol/L) MLN4924 (nmol/L) MLN4924 (nmol/L) MLN4924 (nmol/L) C D Cytarabine Gemcitabine

A375 A549 A375 A549 200 300 400 500 Cytarabine (nmol/L) Cytarabine (nmol/L) Gemcitabine (nmol/L) Gemcitabine (nmol/L) Gemcitabine 100 300 500 700 0 100 010203040 0 02 468101214 00100 200 300 400 500 500 1,000 1,500 0 100 200 300 400 500 0 500 1,000 1,500 2,000 MLN4924 (nmol/L) MLN4924 (nmol/L) MLN4924 (nmol/L) MLN4924 (nmol/L)

HCT116 U2OS HCT116 U2OS 1,500 Cytarabine (nmol/L) Cytarabine (nmol/L) 4,000 8,000 12,000 Gemcitabine (nmol/L) Gemcitabine (nmol/L) 0 010203040506070 0 20 40 60 80 100 0 500 1,000 0 100 200 300 400 500 0 500 1,000 1,500 2,000 0 100 200 300 400 500 0 500 1,0001,500 2,000 MLN4924 (nmol/L) MLN4924 (nmol/L) MLN4924 (nmol/L) MLN4924 (nmol/L)

Figure 1. Isobolograms of select DNA-damaging agents interacting with MLN4924 across A375 (melanoma), A549 (lung), HCT-116 (colon), and U-2 OS (osteosarcoma) cancer cell lines. A, evaluation of the combination of mitomycin C with MLN4924. Lines, a contour plot of the response surface from a10 10 matrix, in which each line represents a decade of viability (LC10,LC20, etc.). Concave curves suggest synergy, linear curves suggest additivity, and convex curves suggest Loewe antagonism. B, isobolograms for the combination of cisplatin with MLN4924. C, isobolograms for the combination of cytarabine and MLN4924. D, isobolograms for the combination of gemcitabine with MLN4924. Isobolograms for additional combinations with MLN4924 are shown in Supplementary Fig. S3.

occurred in the cell lines that exhibited less extensive four agents in A549, four agents in U-2 OS, and seven rereplication [>4N DNA ¼ 8% (A375) and 11% (A549); agents in HCT-116, demonstrating a sufﬁcient breadth of ref. 5]. genetic background differences for our studies. Mitomy- As judged by both combination index and nonlinear cin C, cisplatin, and UV demonstrated synergy with blending values, MLN4924 demonstrated some degree of MLN4924 in three of four cell lines. All three agents are synergy with only 1 agent in A375 (mitomycin C) but with capable of cross-linking DNA. Cytarabine, SN-38 (the

1628 Mol Cancer Ther; 13(6) June 2014 Molecular Cancer Therapeutics

MLN4924 DNA-Damaging Agent Combinations

Figure 2. Interactions between MLN4924 and DNA damage–associated agents. A, a plot of the combination index along the LC50 isobologram calculated for the indicated combinations with MLN4924 from the four cell lines, two to three repeats are given for each combination. The long vertical line represents Loewe additivity. The long dashed vertical lines represent combination indexes of 0.8 or 1.2. The short vertical line represents the median value across the cell lines for each agent. A similar plot with nonlinear blending values instead of the combination index is included in Supplementary Fig. S4. B, antitumor effect of MLN4924 combined with mitomycin C was examined in female Balb/c nude mice bearing A375 human malignant melanoma xenografts. Animals were treated with a subcutaneous dose of MLN4924 twice weekly and an intravenous dose of mitomycin C (MMC) once weekly. Treatment was stopped at day 18 (solid vertical line). Tumor growth was compared with both vehicle-treated and single-agent treated arms (n ¼ 10; data, mean SEM). Average starting tumor volume for mice in the combination set is indicated by the dashed horizontal line. The combination proved synergistic when evaluated over the 21 days (P < 0.001). active metabolite of irinotecan), and gemcitabine demon- regrowth was detectable 1 week after stopping treat- strated synergy with MLN4924 in two of four cell lines, ment. Importantly, the combination of MLN4924 and whereas melphalan and etoposide reliably demonstrated mitomycin C in A375 xenograft was synergistic when synergy in one of the four cell lines. Melphalan was assessed through day 21 (P < 0.001; ref. 23). essentially inactive in HCT-116 and U-2 OS and the addition of MLN4924 did not alter this insensitivity (data RNAi against synergistic combination of MLN4924 not shown). Loewe antagonism was seen in at least one and mitomycin C demonstrates involvement of ATR cell line with gemcitabine, paclitaxel, daunorubicin, tra- and BRCA1 pathways bectedin, and bortezomib. We sought to understand the role of DNA damage genes in the response of A375 cells to the combination of MLN4924 and mitomycin C are synergistic in the MLN4924 and mitomycin C. Specifically, we knocked A375 xenograft mouse model down 320 genes using SMARTpool siRNAs targeting It is often easier to achieve synergy in vitro than it is in DNA damage responses and treated A375 cells with vivo m , especially because the combinations are often 330 nmol/L MLN4924 (LC36) or 1.6 mol/L mitomycin compromised by dose reductions due to increased toxi- C (LC58) or the combination of both treatments. The cities and drug–drug interactions. Therefore, we observed viability for the combination with control siRNA wanted to test whether our most significantly synergis- (GL2) was 6.7%, 4-fold lower than that predicted by Bliss tic example from the A375 cell line data was tolerated independence [(1 0.36) (1 0.58) ¼ 0.27), giving a BI and synergistic in vivo. The combination of mitomycin score of 2.00 0.07 (ref. 24)]. We looked for genes that C dosed at 2.5 mg/kg intravenously once weekly and when depleted would enhance or reduce the synergy, MLN4924 dosed at 180 mg/kg subcutaneously twice although genes enhancing synergy are expected to be weekly was well tolerated for 3 weeks. Although treat- more difficult to find because the viability is already low ment with either single agent alone resulted in only (Supplementary Table S1B). We then evaluated the four moderate growth inhibition, the combination kept the individual oligos against each gene that had demonstrat- tumor from growing for the entire time of treatment ed particularly large effects, to show the likelihood that (Fig. 2B), with day 21 treated versus control tumor these were on-target effects (Supplementary Table S1C, volume ratios of 0.7, 0.5, and 0.17 for mitomycin C, requiring three of four oligos to give the same phenotype MLN4924, and the combination, respectively. Tumor to qualify as on-target), including an additional 28 genes

www.aacrjournals.org Mol Cancer Ther; 13(6) June 2014 1629

Garcia et al.

Figure 3. Effect on synergy between MLN4924 and mitomycin C by depletion of genes involved in DNA damage response. A, of note, 348 genes with known roles in DNA damage response and repair were knocked down and evaluated for their impact on MLN4924–mitomycin C synergy in A375. Only knockdown of CDKN1A (encoding p21) significantly increased synergy. Knockdown of 40 genes gave the most significant reduction of synergy. These genes have been classified according to their known functions. Data, mean value for the third greatest effect size of four oligos (n ¼ 3; error bars, SEM) from a representative run, except GL2, in which only one oligo was evaluated and n ¼ 48. Annotation of genes in which knockdown had synthetic lethal (L), epistatic (E), or suppressor (S) interactions with either MLN4924 (4) or mitomycin C (M) is given parenthetically. Data supporting Fig. 3A are included in Supplementary Table S1B and S1C. B, the same experiment as Fig. 3A, except repeated in HCT-116. Knockdown of 22 genes gave the most significant reduction of synergy. Data supporting Fig. 3B are included in Supplementary Table S1D. C, depletion of CUL4A and CUL4B or depletion of DDB1 results in increased p21 levels, both with vehicle and 330 nmol/L MLN4924 treatment.

from the MLN4924 screen (5). In the deconvoluted exper- also affected A375. Greater transfection efficiency in A375 iment, we adjusted the combination to 330 nmol/L cells likely allowed for more genes to be discovered in that m MLN4924 (LC33) and 1.0 mol/L mitomycin C (LC47). setting (Supplementary Table S1C and S1D). The observed viability of the combination with control In particular, DDB1 and EP400 depletion resulted in siRNA was 14%, giving a BI score of 1.26 0.09. All complete loss of synergy between MLN4924 and mito- reported genetic interactions were seen in at least two of mycin C in A375. Ddb1 is a core component of the Cul4A four runs (Fig. 3A). and Cul4B E3 ubiquitin ligases, so loss of DDB1 should The only gene whose depletion substantially increased mimic the effect of MLN4924 on Cul4A/B activity and the synergy was CDKN1A (encoding p21), suggesting that the addition of MLN4924 would not enhance the synergy p21-dependent intra–S-phase checkpoint helps to block already experienced by depletion of DDB1. Depletion of some degree of synergy between MLN4924 and mitomy- DDB1 or CUL4A and CUL4B in A375 resulted in an cin C in A375. The 40 genes with the strongest reduction of increase in p21 levels (Fig. 3C), suggesting that this effect synergy are classified according to their known biology, may account for the loss of synergy. EP400 encodes p400, with key genes represented in the ATR, BRCA1/BRCA2, which is important for remodeling of the chromatin to chromatin dynamics, and TC-NER pathways. A parallel allow for the recruitment of DNA damage repair proteins, experiment in HCT-116 gave similar results (Fig. 3B; including BRCA1 (25) and Rad51 (26). Finally, CDC7, Supplementary Table S1D), with about 50% of the genes MCM5, and PCNA depletion significantly reduced the reducing synergy in A375 also affecting HCT-116. Con- synergy, suggesting that replication licensing is important versely, 82% of the genes reducing synergy in HCT-116 for this synergy.

1630 Mol Cancer Ther; 13(6) June 2014 Molecular Cancer Therapeutics

MLN4924 DNA-Damaging Agent Combinations

A Vehicle 4924 MMC Combination Vehicle 4924 MMC Combination Vehicle 4924 MMC Combination BARD1 Claspin RAD51 (pT309) BRCA1 FANCD2 Rpa70 p21 FANCI Rpa32 Chk1 H2Ax (pS139) Rpa32 (pT21) Chk1 (pS296) PCNA UBE2L3 Chk1 (pS317) Chk1 (pS345) RAD17 (pS656) Tubulin

B Vehicle MLN4924 + MMC

siRNA RFC4 CLSPN RPA2 BARD1 BRCA1 CHEK1 FANCI RPA2 GL2 ATR CLSPN RFC4 ATR BRCA1 CHEK1 GL2 BARD1 FANCI Nedd8-Cullins BRCA1 BARD1

H2Ax (pS139)

RFC4 Claspin Rpa32 (pT21) * Chk1 (pS345) FANCI Tubulin

Figure 4. The combination of MLN4924 and mitomycin C results in a BRCA1-dependent activation of ATR following induction of DNA strand breaks. A, A375 cells were treated with vehicle, 330 nmol/L MLN4924, 1 mmol/L mitomycin C, or MLN4924 and mitomycin C for 24 hours and prepared for Western blot analysis against the protein products for genes identiﬁed within Fig. 3A. Protein products with a greater than 2-fold change are shown, except for RAD51 (pT309), which did not change. Arrows indicate bands consistent with monoubiquitination of the protein. Protein products with a less than 2-fold change are shown in Supplementary Fig. S5. B, A375 cells were treated with siRNA SMARTpools against the indicated genes for 48 hours before treatment with vehicle or the combination of 330 nmol/L MLN4924 and 1 mmol/L mitomycin C for 24 hours and then prepared for Western blot analysis. The effect of gene depletion on relevant signaling events is shown. All effects were seen in at least two independent experiments. The slower migrating bands for FANCD2, FANCI, p-H2Ax, and PCNA represent monoubiqutination. The multiple bands within p-RAD17 likely represent different spliceforms. The slower migrating band for UBE2L3 may represent thiol-stable monoubiquitination; , a nonspeciﬁc band.

Protein changes induced by the combination of with generally greater effects by MLN4924 alone. Only MLN4924 and mitomycin C changes in the slower migrating band of Ube2L3, the Posttranslational changes to the proteins involved in fastest migrating band of Rad17, and p-Ub-H2Ax were the synergy of MLN4924 and mitomycin C could provide up more than 2-fold with the combination treatment clues about the mechanism by which the combination relative to either of the single agents. promotes cell death. In addition, posttranslational Chk1 has been shown to promote homologous recom- changes speciﬁc to the combination might provide phar- bination by phosphorylation of Rad51 (27, 28), although macodynamic biomarkers for the combination. As such, we do not see any changes in the phosphorylation status of we have evaluated 33 of the hits and associated proteins Rad51 (Fig. 4A). Conversely, BRCA1 has a role in pro- from Fig. 3A for protein changes following treatment with moting ATR activation (29). Thus, it is possible that the MLN4924, mitomycin C, or the combination. Those pro- ATR-associated genes and the BRCA1-associated genes teins that changed more than 2-fold are shown in Fig. 4A, work within a single pathway, although which effect is those without such changes in Supplementary Fig. S5, primary is not clear. although Rad51 (pT309) does not change. All of the To understand whether these two gene sets were reg- proteins that were affected by mitomycin C as a single ulating the same signal transduction pathway, we deplet- agent were also affected by MLN4924 as a single agent ed ATR-associated genes (ATR, CHEK1, RFC4, and RPA2),

www.aacrjournals.org Mol Cancer Ther; 13(6) June 2014 1631

Garcia et al.

BRCA1-associated genes (BARD1, BRCA1, and FANCI), was a signiﬁcant increase in comet tails with the combi- and CLSPN, which is associated with both ATR and nation (P ¼ 1.7 10 3 for 0.3 mmol/L MLN4924 þ 1.0 BRCA1, and treated A375 cells with vehicle or the com- mmol/L mitomycin C; P ¼ 7.2 10 10 for 1.0 mmol/L bination of MLN4924 and mitomycin C. Chk1 phosphor- MLN4924 þ 1.0 mmol/L mitomycin C). This increase ylation by the combination treatment is reduced by RNAi compares favorably with 0.5 mmol/L etoposide (P ¼ 3.1 targeting ATR, CHEK1, CLSPN, and RPA2 (Fig. 4B), as 10 14). In addition, phosphorylation of H2Ax Ser139 is expected, but also by RNAi targeting BARD1 and BRCA1. unchanged with the single agents within 6 hours, in agree- Phosphorylation of threonine-21 on Rpa32 (encoded by ment with prior observations (5), but is increased 3- and 10- RPA2) is reduced by a similar set of genes; however, ATR fold with the combination of 0.3 or 1 mmol/L MLN4924 and depletion does not reduce Rpa32 phosphorylation and, 1 mmol/L mitomycin C, respectively (Fig. 5C and Supple- thus, ATR is not its relevant kinase. mentary Fig. S6). The increased DNA strand breaks likely BRCA1 is both stabilized and hyperphosphorylated lead to the synergy between these two agents. in response to MLN4924 and mitomycin C (Fig. 4A), resulting in slower gel migration. Although the stabilization of BRCA1 was reduced by RNAi targeting ATR, Discussion CHEK1, CLSPN,orRPA2, its hyperphosphorylation was In this study, we explored how MLN4924 might com- unaffected. These results demonstrate that BRCA1 is bine with cancer chemotherapeutic standards of care, required for the activation of the ATR–Chk1 pathway, many of which are DNA-damaging agents. Furthermore, but that ATR and Chk1 are not responsible for the DNA damage is an important contributor to the mecha- phosphorylation of BRCA1. Importantly, these results nism of MLN4924, because MLN4924 results in the sta- suggest that MLN4924 and mitomycin C interact at least bilization of Cdt1 and the induction of rereplication (2–5). in part through the regulation of the BRCA1–ATR–Chk1 Thus, we focused on how other DNA-damaging agents pathway. might interact with MLN4924. The four cell lines that we have used are generally Combination of MLN4924 and mitomycin C results sensitive to a broad array of DNA-damaging agents (Sup- in increased DNA strand breaks plementary Fig. S2). This provides us with a platform to Comet assay analysis of the combination of MLN4924 explore the interaction of MLN4924 with DNA-damaging and mitomycin C demonstrated increased DNA strand agents across multiple genetic backgrounds. Consistent breaks at 6 hours compared with the single agents (Fig. 5A with MLN4924 inducing DNA damage on its own, and Supplementary Fig. S6). Within 6 hours, it is unlikely MLN4924 frequently demonstrated some level of inter- that all cells are actively replicating their DNA. Figure 5B action with other DNA-damaging agents in at least one demonstrates that the single-agent treatments do not sig- cell line (Fig. 2), and interestingly, this interaction could be niﬁcantly increase comet tails over vehicle, whereas there either synergy or antagonism.

Figure 5. The combination of A DMSO MMC MLN4924 MLN4924 Control MLN4924 and mitomycin C leads (1.0 μmol/L) (0.3 μmol/L) (1.0 μmol/L) etoposide to increased DNA strand breaks. A, alkaline comet assays were μ performed with the treatment of 0.5 mol/L A375 with 1.0 mmol/L mitomycin C, Etoposide 0.3 or 1.0 mmol/L MLN4924, or the combination for 6 hours. Etoposide (0.5 or 5.0 mmol/L) is included as a + 1.0 μmol/L 5.0 μmol/L DNA-damaging control. B, the ratio MMC Etoposide of the comet tail to the head from Fig. 5A was calculated for each cell (minimum 189 cells) and box-and-whisker plots made. Box B C plots represent 25th, 50th, and 75th percentile, whereas whiskers represent 5th and 95th percentile. The t test against DMSO: , P < 0.01; , P < 0.0001. Dashed line, the median tail/head ratio for MMC1.0 DMSO MLN4924-0.3 MLN4924-1.0 4924-0.3 + MMC1.0 Etoposide-0.5 Etoposide-5.0 4924-1.0 + MMC1.0 DMSO. C, Western blots demonstrate phosphorylation of p-H2Ax H2Ax on Serine-139 within 6 hours of treatment by the combination of Tubulin MLN4924 and mitomycin C. The arrow indicates a band consistent with monoubiquitination of the protein.

1632 Mol Cancer Ther; 13(6) June 2014 Molecular Cancer Therapeutics

MLN4924 DNA-Damaging Agent Combinations

Stabilized stalled replication fork colliding with rereplication fork

Mre11

Rad50 5′ BARD1 3′ Nbs1 PALB2 BRCA1 BRCA2 CLSPN FANCM ATRIP Polδ Chk1 ATR MMC–DNA Crosslink M C M Polε

CDC7 DBF4

Figure 6. Schematic for proposed mechanism of synergy between mitomycin C and MLN4924. The following model explains the data in Figs. 3 and 5. Mitomycin C (MMC) induces DNA damage at least in part via interstrand DNA crosslinks. MLN4924 should inhibit nucleotide excision repair processes during G1 (not shown) and thereby promote detection of mitomycin C crosslinks in S-phase. Collision of the replication fork with the mitomycin C crosslink will result in stalling of the fork and recruitment of FANCM. FANCM promotes the formation of a "chicken-foot" form of the DNA, which is then stabilized by binding the BRCA1–BRCA2–Rad51 complex, which also inhibits Mre11 nuclease activity. MLN4924 induces additional origin of replication ﬁring through its stabilization of Cdt1, leading to rereplication and increasing the likelihood that a second replication fork will collide with the stalled replication fork. The DNA fragments resulting from this head-to-tail replication fork collision then results in cell death. Blue and yellow ovals, hits from Fig. 3A. Triangles, phosphorylation sites. Yellow, changes seen within Fig. 4A. Small green circles, RPA (replication protein A); whereas small red circles, Rad51.

Paclitaxel blocks mitotic progression, whereas MLN4924 to DNA damage combinations, our results (Fig. 3) suggest requires S-phase to induce DNA damage. Therefore, simul- that depletion of ATR, BRCA1/2, and TC-NER pathways taneous addition of these agents may affect their activity actually reduced the synergy between MLN4924 and through these cell-cycle effects. Our results in the four cells mitomycin C. The ability of gene knockdown to reduce lines were consistent with this (Supplementary Fig. S4). synergy may be due to an equivalence of function between Bortezomib may block the ability of MLN4924 to induce gene depletion and drug effect. For example, both DDB1 rereplication by its stabilization of geminin, an APC/C depletion and MLN4924 treatment result in p21 stabili- substrate and inhibitor of Cdt1 (30). This may explain the zation, increasing engagement of the intra–S-phase check- antagonism observed between MLN4924 and bortezomib point (Fig. 3C). Likewise, both ERCC1 depletion and (Supplementary Fig. S4). Agents that result in the accumu- MLN4924 treatment might be expected to inhibit TC-NER lation of cells in S-phase may demonstrate synergy with (32). Adding MLN4924 to ERCC1-depleted cells might MLN4924. Of note, mitomycin C was found to be the agent result in reduced synergy with mitomycin C because TC- with the most consistent synergy with MLN4924 in this NER would already be inhibited. However, loss of syn- study. ergy is not necessarily equivalent to less cell death overall, Previous work on mitomycin C suggests that adducts are as the viability seen with the treatment of ERCC1-depleted repaired by different mechanisms depending upon how cells by mitomycin C may already be as low as that seen they are detected (13, 14). MLN4924 could substantially with control cells treated with the combination of mito- shift their mechanism of detection, because MLN4924 mycin C and MLN4924. Finally, cells lacking a particular causes an accumulation of cells actively replicating their DNA damage repair pathway may ultimately select a DNA(3,4).Inotherwords,replicationforkcollisionwith pathway that results in a different terminal outcome. the mitomycin C–induced interstrand crosslink would In contrast with TC-NER, the role of Nedd8 in ATR and become the major mechanism of damage detection in the BRCA1 regulation is not well characterized. Understand- presence of MLN4924 (Fig. 6). Comet assay results support ing the effect of gene depletion on the viability effect of the this hypothesis as the combination is leading to a substan- individual agents can help us better understand the effect tial increase in DNA strand breaks within 6 hours. Because of gene depletion on the combination. Knockdown of the mitomycin C as a cross-linking agent inhibits migration of ATR pathway does not seem to have a signiﬁcant impact DNA into the comet tail (31), these results support an on cell death induced by MLN4924 alone, whereas knock- increase in MLN4924-driven DNA strand breaks. down of the BRCA1 pathway did (Supplementary Table Although depletion of DNA damage repair mechan- S1C; ref. 5). Recent work has demonstrated a role for isms would intuitively be expected to increase sensitivity FANCD2, BRCA1, BRCA2, and Rad51 in protecting

www.aacrjournals.org Mol Cancer Ther; 13(6) June 2014 1633

Garcia et al.

stalled replication forks from degradation by the nuclease cisplatin was synergistic with MLN4924 in their setting activity of Mre11 (33–36). Although this BRCA complex is (40). preparing the DNA for repair, the stabilization of stalled Similarly, we do not see the described effects on forks would increase the likelihood of fork collision under FANCD2 or Chk1 in A375 (Fig. 4A and Supplementary the rereplication stress induced by Cdt1 stabilization (37), Fig. S5), even though mitomycin C is clearly synergistic here promoted by MLN4924. Importantly, siRNA oligos with MLN4924 in this background. In addition, although targeting RAD50, a subunit of the Mre11–Rad50–Nbs1 all of the known Fanconi anemia genes were included in (MRN) complex, were synthetic lethal with MLN4924, our experiments, only knockdown of FANCM, BRCA2 whereas siRNA oligos targeting FANCM and BARD1 (FANCD1), and PALB2 (FANCN) affected MLN4924– were suppressor (5), consistent with the BRCA1/BRCA2 mitomycin C synergy. Finally, efficient knockdown of complex protecting stalled forks from MRN nuclease FANCI did not block Chk1 phosphorylation (Fig. 4B). activity and increasing the frequency of strand breaks Therefore, the mechanism of synergy between MLN4924 from replication fork collision. Mre11 destabilizes the and mitomycin C seems to be independent of FANCD2/ stalled replication forks, allowing for an alternative mech- FANCI monoubiquitination. Changes to FANCD2, anism to resolve stalled forks and decreasing the likeli- FANCM, and Chk1 do not seem to be associated with hood of replication fork collision. synergy between MLN4924 and cross-linking agents On the other hand, the BRCA1 pathway did not seem to across cell lines. have a significant role in sensitivity to mitomycin C alone In conclusion, we have studied the interaction of in this setting (Supplementary Table S1C), but the ATR MLN4924 with a broad array of DNA-damaging agents. pathway did. Importantly, depletion of ERCC6 (CSB) was We were able to develop an RNAi-based approach to synthetic lethal with mitomycin C in A375, consistent with study the mechanism of synergy between MLN4924 and single-agent mitomycin C–induced damage being mitomycin C. These results implicated the role of BRCA1 repaired by TC-NER in G1 (14). CSB is degraded by promoting the ATR–Chk1 pathway and of the BRCA2– Cul4A/B (32) whose function is inhibited by MLN4924 Rad51 complex stabilizing replication forks as driving the (4), thereby promoting the accumulation of mitomycin C synergistic interaction. We expect that an improved mech- cross-linked DNA at stalled replication forks in S-phase. anistic understanding from studies such as detailed here Previous studies (29) have characterized a role for may allow for a more informed selection of combinations BRCA1 as a decision point following replication fork– for clinical development. blocking UV-induced DNA damage, promoting ATR activation, ERCC1/XPF recruitment, and inhibiting trans- Disclosure of Potential Conflicts of Interest lesion DNA polymerases. What other proteins might be No potential conflicts of interest were disclosed. associated with BRCA1 in this decision point role is not currently known. Our data suggest that this role of Authors' Contributions BRCA1 is important for MLN4924–mitomycin C synergy. Conception and design: D.S. Sappal, K. Cosmopoulos, M. Kuranda, Intriguingly, we do see PCNA monoubiquitination fol- R. Liu, P.G. Smith, E.S. Lightcap Development of methodology: K. Garcia, J.L. Blank, D.C. Bouck, X.J. Liu, lowing MLN4924 treatment (Fig. 4A), suggesting that D.S. Sappal, M. Kuranda translesion bypass (TLS) may be activated. The loss of Acquisition of data (provided animals, acquired and managed patients, BRCA1 may promote TLS as a DNA damage tolerance provided facilities, etc.): K. Garcia, J.L. Blank, D.C. Bouck, X.J. Liu, D.S. Sappal, K. Cosmopoulos, M.P. Thomas, M. Kuranda, P.G. Smith mechanism. Although RNF111-dependent neddylation Analysis and interpretation of data (e.g., statistical analysis, biostatis- has been suggested to be involved in BRCA1 regulation tics, computational analysis): D.C. Bouck, D.S. Sappal, G. Hather, M.D. Pickard, R. Liu, P.G. Smith, E.S. Lightcap (38), knockdown of RNF111, RNF8, or RNF168 did not Writing, review, and/or revision of the manuscript: J.L. Blank, affect the synergy between MLN4924 and mitomycin C D.C. Bouck, D.S. Sappal, G. Hather, M.D. Pickard, E.S. Lightcap (Supplementary Table S1C and S1D). Administrative, technical, or material support (i.e., reporting or orga- nizing data, constructing databases): X.J. Liu, M.P. Thomas, S. Bandi, MLN4924 has been studied for interaction with inter- E.S. Lightcap strand cross-linking agents (39). Kee and colleagues dem- Study supervision: E.S. Lightcap onstrated that MLN4924 blocked UV-induced FANCD2 monoubiquitination and Chk1 phosphorylation in HCT- Acknowledgments 116, HeLa, MCF7, U-2 OS, and the Fanconi anemia-defi- The authors thank Allison Berger, Doug Bowman, Ben Amidon, Mar- cient ovarian cancer cell line 2008, as well as following garet Quinlan, and John Newcomb for assistance with the article. cisplatin treatment in HeLa, in which MLN4924 seemed to stabilize FANCM. They observed that addition of Grant Support MLN4924 increased the cytotoxicity of mitomycin C in This work was financially supported by Takeda Pharmaceuticals Inter- national Co. HCT-116 and of cisplatin in the Fanconi anemia-deficient The costs of publication of this article were defrayed in part by the ovarian cancer cell line 2008 and its FANCF complemen- payment of page charges. This article must therefore be hereby marked ted counterpart 2008 þ F. They suggest that the effects of advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. MLN4924 on FANCD2 and Chk1 account for the observed sensitization. However, Jazaeri and colleagues did not Received August 9, 2013; revised March 13, 2014; accepted March 17, observe similar effects in SKOV3 or ES2 cells, even though 2014; published OnlineFirst March 26, 2014.

1634 Mol Cancer Ther; 13(6) June 2014 Molecular Cancer Therapeutics

MLN4924 DNA-Damaging Agent Combinations

References 1. Soucy TA, Smith PG, Milhollen MA, Berger AJ, Gavin JM, Adhikari S, 23. Shi JQ, Lasky K, Shinde V, Stringer B, Qian MG, Liao D, et al. MLN0905, et al. An inhibitor of NEDD8-activating enzyme as a new approach to a small-molecule PLK1 inhibitor, induces antitumor responses in treat cancer. Nature 2009;458:732–6. human models of diffuse large B-cell lymphoma. Mol Cancer Ther 2. Sarikas A, Hartmann T, Pan ZQ. The cullin protein family. Genome Biol 2012;11:2045–53. 2011;12:220. 24. Chen S, Blank JL, Peters T, Liu XJ, Rappoli DM, Pickard MD, et al. 3. Lin JJ, Milhollen MA, Smith PG, Narayanan U, Dutta A. NEDD8- Genome-wide siRNA screen for modulators of cell death induc- targeting drug MLN4924 elicits DNA rereplication by stabilized by proteasome inhibitor bortezomib. Cancer Res 2010;70: ing Cdt1 in S phase, triggering checkpoint activation, apopto- 4318–26. sis, and senescence in cancer cells. Cancer Res 2010;70: 25. Xu Y, Sun Y, Jiang X, Ayrapetov MK, Moskwa P, Yang S, et al. The p400 10310–20. ATPase regulates nucleosome stability and chromatin ubiquitination 4. Milhollen MA, Narayanan U, Soucy TA, Veiby PO, Smith PG, Amidon B. during DNA repair. J Cell Biol 2010;191:31–43. Inhibition of NEDD8-activating enzyme induces rereplication and 26. Courilleau C, Chailleux C, Jauneau A, Grimal F, Briois S, Boutet- apoptosis in human tumor cells consistent with deregulating CDT1 Robinet E, et al. The chromatin remodeler p400 ATPase facilitates turnover. Cancer Res 2011;71:3042–51. Rad51-mediated repair of DNA double-strand breaks. J Cell Biol 2012; 5. Blank JL, Liu XJ, Cosmopoulos K, Bouck DC, Garcia K, Bernard H, 199:1067–81. et al. Novel DNA damage checkpoints mediating cell death induced by 27. Sørensen CS, Hansen LT, Dziegielewski J, Syljuasen RG, Lundin C, the NEDD8-activating enzyme inhibitor MLN4924. Cancer Res 2013; Bartek J, et al. The cell-cycle checkpoint kinase Chk1 is required for 73:225–34. mammalian homologous recombination repair. Nat Cell Biol 2005; 6. Li JM, Jin J. CRL ubiquitin ligases and DNA damage response. Front 7:195–201. Oncol 2012;2:29. 28. Bahassi EM, Ovesen JL, Riesenberg AL, Bernstein WZ, Hasty PE, 7. Hannah J, Zhou P. Regulation of DNA damage response pathways by Stambrook PJ. The checkpoint kinases Chk1 and Chk2 regulate the the cullin-RING ubiquitin ligases. DNA Repair 2009;8:536–43. functional associations between hBRCA2 and Rad51 in response to 8. Hannb R, Dubiel W. COP9 signalosome function in the DDR. FEBS Lett DNA damage. Oncogene 2008;27:3977–85. 2011;585:2845–52. 29. Pathania S, Nguyen J, Hill SJ, Scully R, Adelmant GO, Marto JA, et al. 9. Abbas T, Dutta A. CRL4Cdt2: master coordinator of cell-cycle pro- BRCA1 is required for postreplication repair after UV-induced DNA gression and genome stability. Cell Cycle 2011;10:241–9. damage. Mol Cell 2011;44:235–51. 10. Kee Y, Kim JM, D'Andrea AD. Regulated degradation of FANCM in 30. Saxena S, Dutta A. Geminin-Cdt1 balance is critical for genetic sta- the Fanconi anemia pathway during mitosis. Genes Dev 2009; bility. Mutat Res 2005;569:111–21. 23:555–60. 31. McKenna DJ, Gallus M, McKeown SR, Downes CS, McKelvey- 11. Halazonetis TD, Gorgoulis VG, Bartek J. An oncogene-induced DNA Martin VJ. Modification of the alkaline Comet assay to allow simul- damage model for cancer development. Science 2008;319:1352–5. taneous evaluation of mitomycin C-induced DNA cross-link dam- 12. Evertts A, Coller HA. Back to the origin: reconsidering replication, age and repair of specific DNA sequences in RT4 cells. DNA Repair transcription, epigenetics, and cell-cycle control. Genes Cancer 2012; 2003;2:879–90. 3:678–96. 32. Groisman R, Kuraoka I, Chevallier O, Gaye N, Magnaldo T, Tanaka K, 13. Deans AJ, West SC. DNA interstrand crosslink repair and cancer. Nat et al. CSA-dependent degradation of CSB by the ubiquitin-protea- Rev Cancer 2011;11:467–80. some pathway establishes a link between complementation factors of 14. Muniandy PA, Liu J, Majumdar A, Liu ST, Seidman MM. DNA inter- the Cockayne syndrome. Genes Dev 2006;20:1429–34. strand crosslink repair in mammalian cells: step by step. Crit Rev 33. Schlacher K, Wu H, Jasin M. A distinct replication fork protection Biochem Mol Biol 2009;45:23–49. pathway connects Fanconi anemia tumor suppressors to RAD51- 15. Minto CF, Schnider TW, Short TG, Gregg KM, Gentilini A, Shafer SL. BRCA1/2. Cancer Cell 2012;22:106–16. Response surface model for anesthetic drug interactions. Anesthesi- 34. Ying S, Hamdy FC, Helleday T. Mre11-dependent degradation of ology 2000;92:1603–16. stalled DNA replication forks is prevented by BRCA2 and PARP1. 16. Chou TC. Drug combination studies and their synergy quantification Cancer Res 2012;72:2814–21. using the Chou–Talalay method. Cancer Res 2010;70:440–6. 35. Schlacher K, Christ N, Siaud N, Egashira A, Wu H, Jasin M. Double- 17. Peterson JJ, Novick SJ. Nonlinear blending: a useful general concept strand break repair-independent role for BRCA2 in blocking stalled for the assessment of combination drug synergy. J Recept Signal replication fork degradation by MRE11. Cell 2011;145:529–42. Transduct Res 2007;27:125–46. 36. Hashimoto Y, Ray Chaudhuri A, Lopes M, Costanzo V. Rad51 protects 18. Donawho CK, Luo Y, Luo Y, Penning TD, Bauch JL, Bouska JJ, et al. nascent DNA from Mre11-dependent degradation and promotes con- ABT-888, an orally active poly(ADP-ribose) polymerase inhibitor that tinuous DNA synthesis. Nat Struct Mol Biol 2010;17:1305–11. potentiates DNA-damaging agents in preclinical tumor models. Clin 37. Davidson IF, Li A, Blow JJ. Deregulated replication licensing causes Cancer Res 2007;13:2728–37. DNA fragmentation consistent with head-to-tail fork collision. Mol Cell 19. Bryant HE, Schultz N, Thomas HD, Parker KM, Flower D, Lopez E, et al. 2006;24:433–43. Specific killing of BRCA2-deficient tumours with inhibitors of poly 38. Ma T, Chen Y, Zhang F, Yang CY, Wang S, Yu X. RNF111-dependent (ADP-ribose) polymerase. Nature 2005;434:913–7. neddylation activates DNA damage-induced ubiquitination. Mol Cell 20. Farmer H, McCabe N, Lord CJ, Tutt AN, Johnson DA, Richardson TB, 2013;49:897–907. et al. Targeting the DNA repair defect in BRCA mutant cells as a 39. Kee Y, Huang M, Chang S, Moreau LA, Park E, Smith PG, et al. therapeutic strategy. Nature 2005;434:917–21. Inhibition of the Nedd8 system sensitizes cells to DNA interstrand 21. Motegi A, Murakawa Y, Takeda S. The vital link between the ubiquitin– cross-linking agents. Mol Cancer Res 2012;10:369–77. proteasome pathway and DNA repair: impact on cancer therapy. 40. Jazaeri AA, Shibata E, Park J, Bryant JL, Conaway MR, Modesitt SC, Cancer Lett 2009;283:1–9. et al. Overcoming platinum resistance in preclinical models of ovarian 22. Loewe S. The problem of synergism and antagonism of combined cancer using the neddylation inhibitor MLN4924. Mol Cancer Ther drugs. Arzneimittelforschung 1953;3:285–90. 2013;12:1958–67.

www.aacrjournals.org Mol Cancer Ther; 13(6) June 2014 1635

Nedd8-Activating Enzyme Inhibitor MLN4924 Provides Synergy with Mitomycin C through Interactions with ATR, BRCA1/BRCA2, and Chromatin Dynamics Pathways

Khristofer Garcia, Jonathan L. Blank, David C. Bouck, et al.

Mol Cancer Ther 2014;13:1625-1635. Published OnlineFirst March 26, 2014.

Updated version Access the most recent version of this article at: doi:10.1158/1535-7163.MCT-13-0634

Supplementary Access the most recent supplemental material at: Material http://mct.aacrjournals.org/content/suppl/2014/03/27/1535-7163.MCT-13-0634.DC1

Cited articles This article cites 40 articles, 15 of which you can access for free at: http://mct.aacrjournals.org/content/13/6/1625.full#ref-list-1

Citing articles This article has been cited by 8 HighWire-hosted articles. Access the articles at: http://mct.aacrjournals.org/content/13/6/1625.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://mct.aacrjournals.org/content/13/6/1625. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.