The Mismatch Repair System Modulates Curcumin Sensitivity through Induction of DNA Strand Breaks and Activation of G₂-M Checkpoint Free

Curcumin (diferuloylmethane), a dietary pigment derived from the rhizome Curcuma longa, is a promising anticancer drug that is currently in phase II clinical trial (1, 2). The anticancer property of curcumin is attributed to its selective cell death–inducing ability in tumor cells (2–4). In normal and primary cells, curcumin is either inactive or inhibits proliferation but does not elicit a cytotoxic response. For example, in primary untransformed mouse embryonic fibroblast line C3H/10T1/2, rat embryonic fibroblasts, and human foreskin fibroblasts, curcumin failed to initiate cell death although it inhibited proliferation of both normal and transformed cells in a nonselective manner (2–4).

The principal mode of cell death induced by curcumin is apoptosis (5). Failure to undergo growth arrest before or during mitosis in response to stress triggers aberrant chromosome segregation, which ultimately culminates in the activation of apoptosis through a mechanism termed mitotic catastrophe. Indeed, a recent study showed that curcumin disrupts mitotic spindle structure and induces micronucleation (6). It has been proposed that mitotic catastrophe induced by some anticancer agents such as paclitaxel stems from cellular damage in combination with dysregulated cell cycle checkpoint activation resulting in spindle and chromosome segregation abnormalities during mitosis (7–9). More specifically, inactivation or dysfunction of the chromosomal passenger complex, which is necessary for coordinating chromosomal and cytoskeletal events during mitosis, often leads to mitotic cell death (10). To that end, a recent report documented that in BCR-Abl–transformed cells, curcumin induces mitotic catastrophe and cell death through the disruption of chromosomal passenger complex and downregulation of survivin, a modulator of cell division and apoptosis (11). In the apoptosis-resistant HL-60 subline HCW-2, curcumin induces mitotic catastrophe by inhibiting the expression of survivin (12), and in the breast cancer line MCF7, curcumin-induced apoptosis was associated with the assembly of aberrant, monopolar mitotic spindles that are impaired in their ability to segregate chromosomes (6, 13). Other studies documented that curcumin-induced apoptosis and cell cycle arrest in melanoma cells was associated with the downregulation of NFκB activation; decreased iNOS and DNA-PK catalytic subunit expression; and upregulation of p53, p21(Waf1/Cip1), p27(Kip1), and Chk2 (14, 15). Curcumin-mediated inhibition of NF-κB through the NIK/IKK signaling complex results in the excessive generation of reactive oxygen species (ROS), which ultimately trigger apoptosis through the activation of c-Abl>c-Jun-NH₂-kinase signaling (16, 17). In sum, available evidence indicates that curcumin targets multiple signaling pathways to induce cell cycle arrest and/or cell death.

The mismatch repair (MMR) system is an evolutionarily conserved DNA repair mechanism primarily responsible for resolving postreplicative mismatches in DNA (18, 19). Expectedly, deficiencies in MMR increase the rate of genomic mutations and susceptibility to multiple forms of cancer, including colorectal tumors (18, 19). In addition to its capacity as a repair mechanism, MMR is essential for the activation of signaling cascades activated in response to certain genotoxic insults (20, 21). For example, MMR-deficient cells fail to trigger G₂-M cell cycle arrest following treatment with S_N1-methylating agents such as N-methyl-N'-nitro-N-nitrosoguanidine, procarbazine, temozolomide, and the antimetabolite 6-thioguanine (18–21). MMR-deficient cell lines are also resistant to the cytotoxic effects of these agents, a phenotypic effect termed alkylation tolerance (20, 21). MMR complexes also recognize other types of damage such as cisplatin adducts (22) and oxidized bases such as 8-oxo-dG (23).

Recently, defects in G₂-M checkpoint activation observed in MMR-deficient cells following S_N1 methylator exposure have been attributed to the dysregulation of MMR-dependent activation of the cell cycle checkpoint kinases Chk1 and Chk2 (24, 25). Biochemical studies revealed that components of the MMR system interact with the damage-responsive kinases ATM/ATR, potentially facilitating phosphorylation and activation of Chk1/Chk2 kinases (24–27). Taken together, the absence of MMR confers resistance to S_N1 methylators due to failure to recognize and respond to DNA lesions that normally activate a robust cellular response.

In this report, we show that cells compromised or deficient in MMR function (Msh2 and Mlh1) exhibit increased sensitivity to curcumin. Elucidation of the mechanism behind this enhanced sensitivity revealed that the levels of oxidative DNA damage induced by curcumin were independent of the MMR status. However, the activation of Chk1 and Chk2 kinases and the induction of G₂-M arrest triggered by curcumin required MMR function. Compared with MMR-proficient cells, MMR-deficient counterparts displayed reduced DNA double-strand breaks (DSB) and ATM activation, suggesting that DSB formation induced by curcumin is predominantly a MMR-dependent process. Together, the findings indicate that the MMR system protects cells from curcumin cytotoxicity, in part, by activating the G₂-M checkpoint. In clear contrast, MMR-deficient cells fail to activate this response and likely transit into mitosis with a damaged genome, which subsequently activates a robust apoptotic response.

Curcumin, NAC, and glutathione was obtained from Sigma-Aldrich and dissolved in DMSO before use. 7-Hydroxystaurosporine (UCN-01) was obtained from the National Cancer Institute. 2-[4-(4-Chlorophenoxy) phenyl]-1H-benzimidazole-5-carboxamide (Chk2 inhibitor II) was purchased from Calbiochem. Antibodies specific for phospho-Chk1 (S345), phospho-Chk2 (T68), and total Chk1 and Chk2 were obtained from Cell Signaling. Antibodies specific for γH2AX, MLH1, and MSH2 were obtained from Santa Cruz Biotechnology. Horseradish peroxidase–conjugated secondary antibodies were purchased from Novus Biologicals. Cleaved poly ADP ribose polymerase (PARP) and caspae-3 antibodies were obtained from AbCam.

The MMR-proficient colorectal tumor line HCT116+ch3 was created by the stable transfer of a portion of human chromosome 3, bearing a wild-type copy of the hMlh1 gene, into MLH1-deficient line HCT116 (28). The control HCT116+ch2 cell line is an MLH1-deficient derivative of HCT116 that has a portion of human chromosome 2 introduced by microcell fusion. HCT116+ch2 and HCT116+ch3 cells were maintained in DMEM containing 10% fetal bovine serum supplemented with 400 μg/mL of geneticin (G418) as described (28). The MSH2-proficient derivative of the human endometrial adenocarcinoma cell line Hec59 (Hec59+ch2) was maintained in RPMI containing 400 μg/mL geneticin (29). The colorectal tumor line RKO was cultured in DMEM containing 5-azacytidine (5 μmol/L) for 5 d to restore the expression of the epigenetically silenced hMlh1 gene. The resultant RKO/Mlh1+ and the parental RKO cells were cultured as described in ref. (30).

MTS assay was done using the CellTiter 96 Aqueous One Solution Proliferation Assay System (Promega). This assay measures the bioreduction by intracellular dehydrogenases of the tetrazolium compound MTS in the presence of the electron-coupling reagent phenazine methosulfate. MTS and phenazine methosulfate were added to the culture wells, and the mixture was incubated at 37°C for 3 h. Absorbance was measured at 490 nm using a microplate reader and is directly proportional to the number of viable cells in the cultures. The relative toxicity was calculated by comparing with nontreated cells.

Overlapping synthetic oligonucleotides corresponding to sequences specific for the human Chk1 (5′-GAAGCAGTCGCAGTGAAGAT-3′), Chk2 (5′-GAACCTGAGG-ACCAAGAACC-3′), MSH2 (5′-GTTCGTCAGTATAG AGTTGAA-3′), and MLH1 (5′-GGTTCACTACTAGTAAACTG-3′) transcripts were hybridized and cloned into pSIREN-RETRO-Q (Clontech). The recombinant pSiren plasmid was cotransfected with pCL-ampho plasmid encoding the packaging viral DNA into 293T cells using Lipofectamine 2000 (Invitrogen). The supernatant containing the viral DNA was collected, filtered, and used to infect HCT116+ch3 cells. Cells were selected by incubation with puromycin (1 μg/mL) for 4 d and downregulation of target gene expression was confirmed by immunoblotting.

Cells were harvested by scraping, washed with ice-cold PBS, and lysed in cold 1× lysis buffer containing 10 mmol/L Tris-HCl (pH 8.0), 240 mmol/L NaCl, 5 mmol/L EDTA, 1 mmol/L DTT, 0.1 mmol/L phenylmethylsulfonyl fluoride, 1% Triton X-100, 1 mmol/L sodium vanadate, and 1 μg/mL of leupeptin, pepstatin, and aprotinin by incubation on ice for 20 min. Lysates were cleared by centrifugation and protein concentration was determined using the Bradford assay. For immunoblotting, proteins were resolved on 4% to 12% SDS-PAGE and electrotransferred onto Immobilon-P (Millipore) membrane. The membrane was then probed with indicated primary antibody, subsequently with horseradish peroxidase–conjugated secondary antibody, and developed using chemiluminescence. Where indicated, the membrane was stripped by incubation in stripping buffer containing 65 mmol/L Tris-HCl (pH 6.7), 100 mmol/L β-mercaptoethanol, and 2% SDS for 30 min at 40°C. The membrane was then reprobed with the indicated antibody.

Cells were grown on presterilized glass coverslips and treated with DMSO (mock) or curcumin (30 μmol/L) in the presence or absence of NAC (5 mmol/L) for 1 h. Cells were then washed thrice with HBSS containing 10 mmol/L HEPES, 2 mmol/L CaCl₂, and 4 mg/mL bovine serum albumin. Cells were fixed in 4% paraformaldehyde in HBBS for 5 min and then permeabilized in 100% methanol for 5 min. Cells were then stained for γH2AX foci by incubation with the anti-H2AX antibody for 1 h at 37°C. Cells were subsequently stained with Texas red–conjugated secondary antibody. DNA was counterstained with 4′,6-diamidino-2-phenylindole. γH2AX foci in each nucleus were counted using a Nikon fluorescence microscope (TE S2000) equipped with a charge-coupled device camera. At least 100 cells were scored for each time point.

Mock (DMSO) and curcumin-treated cells were washed twice with 1× PBS and fixed in ice-cold 70% ethanol for 30 min on ice and stored at 4°C before analysis. For staining, cells were incubated in PBS containing 1 mg/mL RNase A and 40 μg/mL propidium iodide for 30 min in the dark at 37°C and then analyzed by flow cytometry. Approximately 3 × 10⁴ cells were evaluated in each sample and DNA histograms were analyzed using the ModFit (Verity Software) software. All flow cytometry experiments were done in triplicate and Student's t test was conducted to determine statistical significance.

8-Oxo-guanine (8-oxo-G) was measured using the 8-OxyDNA assay kit from Calbiochem. Following treatment with DMSO (mock) or curcumin, cells were fixed, permeabilized, and the FITC-antibody conjugate, which binds 8-oxo-G moiety, is added. The presence of oxidized DNA is indicated by a green/yellow fluorescence and measured using flow cytometry.

Logarithmically growing cells were plated at a density of 3 × 10³ cells per well in a 10-cm dish and treated the next day with varying concentrations of curcumin (0, 10, 20, 30, 40, and 50 μmol/L) for 5 h. Cells were washed and replaced with fresh media and grown for 14 d. Colonies were stained using 0.25% crystal violet and 10% formalin in 80% methanol for 30 min, washed with water, and counted.

In addition to resolving postreplicative bp errors, the MMR system plays a critical role in the activation of key cellular responses induced by genotoxins such as S_N1 methylators (21). To evaluate a potential role for MMR in curcumin-induced cellular responses, we exposed matched MMR-proficient (HCT116+ch3, HEC59+ch2, and RKO/MLH+) and MMR-deficient (HCT116+ch2, HEC59, and RKO) cells to 30 μmol/L curcumin and then assessed cytotoxic response by trypan blue exclusion (Fig. 1A). Curcumin was mildly cytotoxic in the MMR-proficient cell lines at 48 hours after treatment, with each line showing <10% trypan blue staining. In contrast, MMR-deficient lines displayed >30% trypan blue–positive cells in response to curcumin at this time point. Statistical analysis showed significant difference (P < 0.01) in trypan blue–positive cells between matched MMR-proficient/deficient cell lines.

As an independent approach, we used MTS assay to examine the effect of MMR on curcumin sensitivity. In this set of experiments, we not only scored the cytotoxic effects of curcumin in our MMR-proficient/deficient cells but also examined the dose-dependent nature of this response. In close agreement with the results obtained using trypan blue exclusion, we observed a modest cytotoxic effect in MMR-proficient cells at each curcumin dose tested (5, 10, and 20 μmol/L; Fig. 1B). Again, we observed a heightened cytotoxic response in MMR-deficient cells and measured a statistically significant (P < 0.01) difference between matched MMR-proficient and MMR-deficient cells treated with 10 and 20 μmol/L curcumin.

Next, we examined the effect of MMR status on cell survival in response to curcumin using clonogenic assays on our panel of matched MMR-proficient/deficient cells. Colony-forming ability in each cell line was measured in response to 10, 20, 30, 40, and 50 μmol/L curcumin. This analysis revealed that survival in the MSH2-deficient HEC59 line was significantly reduced (P < 0.05, Student's t test) compared with matched MSH2-proficient HEC59+ch2 cells treated with 40 and 50 μmol/L curcumin (Fig. 1C). Consistent with this result, we measured statistically significant (P < 0.01) reduced survival at curcumin doses of >40 μmol/L in MLH1-deficient HCT116+ch2 cells compared with matched MLH1-proficient HCT116+ch3 cells. MLH1-deficient RKO showed significantly reduced (P < 0.05) survival following treatment with >30 μmol/L curcumin when compared with MLH1-proficient RKO/MLH1+ cells.

To further evaluate the role of MMR in activating the cell death response to curcumin, we knocked down MSH2 or MLH1 expression in MMR-proficient cells (HCT116+ch3) by RNA interference, specifically using retroviruses encoding gene-specific short hairpin RNA (shRNA) sequences. Immunoblotting of the lysates transduced with MLH1 and MSH2 shRNA–encoding retroviruses confirmed the marked suppression of MLH1 and MSH2 expression compared with cells transduced with control luciferase shRNA (shLuc)–encoding virus (Fig. 2A). This panel of cell lines was treated with 30 μmol/L curcumin and the sub-G₁ cell population was measured 48 hours posttreatment by flow cytometry following staining with propidium iodide (Fig. 2B). Results showed notable accumulation of sub-G₁ population in curcumin-treated cells compared with mock-treated cells. However, a comparison of cell death induction within the treated panel of cell lines revealed notably less cell death in shLuc cells (8.6% sub-G₁) than in shMSH2 or shMLH1 cells (27.2% and 17.9% sub-G₁, respectively). When the results of three independent experiments were combined, shMLH1 and shMSH2 cells showed a statistically significant (P < 0.01, P < 0.05, respectively) increase in sub-G₁ than in shLuc cells in response to curcumin treatment (Fig. 2C). Immunoblotting of lysates formed from curcumin-treated cells with anti–caspase-3 and anti-PARP antibody showed increased cleavage in treated shMSH2 and shMLH1 cells than in treated shLuc cells (Fig. 2D) consistent with the activation of the intrinsic apoptotic pathway. Parallel findings were also observed with matched MMR-proficient/deficient cells (Supplementary Figs. S1 and S2A and B). Here, again, at 24 and 48 hours after curcumin treatment, MMR-negative cells (H2; Hec59) displayed increased cell death compared with matched MMR-proficient cells. At the 24-hour time point, MMR-positive cells (H3; Hec59+Ch2) showed accumulation of cells with 4N DNA content (G₂-M arrest), but at a later time point (48 hours), a decrease in the G₂-M population and an apparent increase in the G₁ population was observed. These results indicate that the cell cycle arrest induced by curcumin in MMR+ cells is transient in nature, presumably allowing these cells enhanced survival against curcumin. Immunoblotting of lysates obtained from these cells with anti–caspase-3 and anti-PARP antibodies confirmed reduced apoptotic response in MMR+ cells compared with MMR cells (Supplementary Figs. S1 and S2C). Together, these findings reinforced that loss of either MSH2 or MLH1 enhances curcumin sensitivity.

Following our observation that MMR-deficient cells display enhanced sensitivity to curcumin, we determined if curcumin-induced cell stress is dependent on a functional MMR system. Because curcumin has been characterized as raising cellular levels of ROS (17, 18), we measured ROS production using 2',7'-dichlorofluorescein diacetate. Result showed no significant difference in DCF fluorescence in the matched sets of MMR-proficient/deficient cells (see Supplementary Fig. S3A and B). We also measured 8-oxo-G, a prominent mutagenic lesion formed by increased levels of ROS, in shLuc, shMSH2, and shMLH1 cells. Again, no significant difference in curcumin-induced accumulation of 8-oxo-G was observed in these lines (see Supplementary Fig. S3C and D). Clearly, the disparity in curcumin sensitivity observed in MMR-deficient and MMR-proficient cells is not attributable to quantitative differences in the stress itself.

Recently, our group and others determined that the dysregulated checkpoint response exhibited by MMR-deficient cells in response to S_N1 methylators stems, in part, from defective Chk1 and Chk2 activation (24, 25, 31). In addition to the increased cell death observed in MMR-deficient cells, we also observed a decreased G₂-M arrest when compared with MMR-proficient cells similarly treated with curcumin (Fig. 2C). Based on this rationale, we examined MMR⁺ and MMR⁻ cells for the activation of Chk1 and Chk2 by immunoblotting with phospho-specific Chk1 (phospho-S317) and Chk2 (phospho-T68) antibodies. Whereas MMR-positive (HCT116+ch3; H3) cells showed clearly detectable levels of phosphorylated Chk1 and Chk2, matched MMR-negative (HCT116+ch2; H2) cells showed a blunted response at 2, 4, and 6 h after curcumin treatment (Fig. 3A). Similarly, curcumin-induced phosphorylation of Chk1 and Chk2 was observed in treated MSH2-proficient (Hec59+ch2) cells but not in MSH2-deficient (Hec59) cells (Fig. 3B). Blunted Chk1 and Chk2 phosphorylation were also observed in shMLH1 and shMSH2 cells in response to curcumin treatment (Fig. 3C). Together, the results showed that curcumin activates Chk1/Chk2 kinase through a mechanism that is dependent on the intact MMR system.

To determine whether the MMR-dependent phosphorylation of Chk1 and Chk2 mediates G₂-M arrest and protects cells from curcumin-induced cytotoxicity, we inhibited Chk1 activity with the drug UCN-01 (32). As expected, the treatment of MMR-proficient (HCT116+ch3) cells with 30 μmol/L curcumin induced robust accumulation of cells in the G₂-M phase of the cell cycle (44.1%) accompanied by minimal (5.5%) levels of cell death (sub-G₁ cells; Fig. 4A). However, incubation of HCT116+ch3 cells with 0.3 μmol/L UCN-01 resulted in 24.3% of the cells displaying sub-G₁ DNA content representing a statistically significant (P < 0.05) increase in curcumin-induced cell death compared with control cells. Chk2 inhibitor II, a compound that specifically inhibits this kinase caused a modest increase in the percentage of apoptotic cells in response to curcumin treatment (Fig. 4A and B).

To independently assess the roles of Chk1 and Chk2 in curcumin-induced cytotoxicity, we suppressed Chk1 and Chk2 expression in HCT116+ch3 (H3) using gene-specific shRNAs. After confirming reduced Chk1 and Chk2 expression in shChk1 and shChk2 cells by immunoblotting (Fig. 4C), these cells, along with control shLuc cells, were treated with 30 μmol/L curcumin and analyzed at 36 hours by flow cytometry. As anticipated, in response to curcumin treatment, shChk1 cells displayed a significant increase (P < 0.01) in the sub-G₁ population in contrast to shLuc cells (20.3% versus 6.3%, respectively; Fig. 4D). Conversely, curcumin-induced G₂-M arrest was decreased significantly (P < 0.01) in shChk1 cells compared with curcumin-treated shLuc cells (8.4% versus 51.3%; Fig. 4E). Curcumin-treated shChk2 cells displayed a modest decrease in G₂-M cells than treated shLuc cells. In sum, the data showed that in these colorectal cancer cells, Chk1 confers protection against curcumin toxicity by activating G₂-M checkpoint response.

The antiproliferative effects of curcumin are associated with elevated levels of intracellular ROS. Thus, we sought to test if curcumin-generated ROS activates Chk1 and Chk2. To address this issue, we examined Chk1 and Chk2 activation by curcumin in HCT116+ch3 (H3) cells in the presence of the radical scavenger, N-acetyl-l-cysteine (NAC) or glutathione. Results showed that both NAC and glutathione efficiently abrogated Chk1 and Chk2 phosphorylation induced by curcumin (Fig. 5A). Additionally, flow cytometric analysis of these cells indicated a significant decrease (P < 0.01) in G₂-M population when HCT116+ch3 cells were pretreated with NAC before the addition of curcumin (Fig. 5B). These results established that curcumin-induced Chk1/Chk2 activation and G₂-M checkpoint induction is mediated by ROS.

Elevated intracellular levels of ROS oxidize nucleotides, resulting in the replication-dependent induction of DSBs (33). Thus, we examined the formation of γH2AX foci, a well-known marker for DSB, in curcumin-treated shLuc, shMSH2, and shMLH1 cells. Staining of cells with phospho-specific (S139) H2AX antibody revealed the γH2AX foci within 60 minutes after curcumin (30 μmol/L) treatment, whereas very few mock-treated cells stained positive for γH2AX foci (Fig. 6A). Immunoblotting of the lysates prepared at various time points (1, 2, and 4 hours) after treatment showed a time-dependent increase in H2AX phosphorylation (Fig. 6B). Consistent with the role of ROS in DSB formation, we also observed that H2AX phosphorylation was blocked by the treatment of cells with NAC (Fig. 6B). To address the MMR dependency of DSB formation following curcumin exposure, we examined H2AX phosphorylation in curcumin-treated shMLH1 and shMSH2 cells by immunoblotting. Results indicated substantial reduction in H2AX foci and phosphorylation in HCT116+ch3 with reduced expression of either MSH2 or MLH1 when compared with curcumin-treated control (shLuc) cells (Fig. 6C and D). Collectively, these results indicated that curcumin-generated ROS induces DSB largely through a MMR-dependent mechanism.

Numerous groups have investigated the effects of curcumin on cultures of human tumor cell lines and determined that curcumin exhibits strong antiproliferative activity, activates the G₂-M checkpoint and, depending on the cell type under investigation, triggers apoptosis (3, 5, 6, 12). In addition, it has been determined that the antiproliferative effect of curcumin is attributable to the sharp increase in superoxide anion concentrations in treated cells (16, 17). Similarly, we observed that curcumin exposure results in increased ROS levels and the oxidized purine adduct, 8-oxo-G. Furthermore, we observed that the increased levels of ROS stemming from curcumin exposure are responsible for the cell cycle arrest and cell death response triggered by curcumin. In sum, curcumin exposure drives ROS accumulation, resulting in cellular stress in the form of genotoxic damage. Thus, curcumin functions as a genotoxic agent.

Consistent with a general response to genotoxic stress, we found that curcumin exposure activates the kinases Chk1 and Chk2. Among these molecules, Chk1 is the principal signal-transducing kinase responsible for the establishment of the G₂-M checkpoint in response to genotoxic stress (34–37). In support of the importance of this kinase in triggering the checkpoint response to curcumin exposure, we observed that RNAi-induced knockdown and pharmacologic inhibition of Chk1 strongly blunted G₂-M arrest. Inhibition of Chk2 showed a more modest effect on the activation of this checkpoint. This outcome parallels findings observed with other types of genotoxic agents (38).

We also observed that activation of Chk1 and Chk2 in response to curcumin are MMR-dependent events. We and others have observed MMR-dependent activation of Chk1 and Chk2 and other kinases in response to S_N1 methylating agents (24–26, 39). It is widely viewed that the mutagenic O⁶MeG adducts resulting from S_N1 methylator exposure is the lesion that triggers MMR-dependent responses (21, 23).

Although it is undetermined which lesion(s) result in the engagement of the MMR system in response to curcumin, this response most likely stems from base oxidation due to increased ROS because radical scavengers blunted the Chk1/Chk2 response following curcumin administration. Given that curcumin generates 8-oxo-G lesions, it is reasonable to assume that this lesion, and perhaps and other such mutagenic lesions, is recognized by MMR, and processing of this damage is responsible for the generation of the DSB observed in curcumin-treated MMR-positive cells. When taken together, our observations lead us to propose a model elucidating how cells respond to curcumin (Fig. 6D). In this model, we propose that at the doses used, curcumin-induced DNA damage engages the MMR system, most likely through the recognition of oxidative damage to the genome. MMR-dependent processing of these lesions subsequently leads to the development of DSBs within the genome. The cell consequently responds to these DSBs by activating the G₂-M cell cycle checkpoint through the Chk1/Chk2 signal transduction pathway. Proper activation of this checkpoint response significantly blunts the cytotoxic effects of curcumin. Higher doses or longer incubation times with curcumin, however, results in the triggering of cell death in MMR+ cells through necrotic cell death (data not shown). Although the MMR system has no effect on the levels of ROS generated or consequential genome damage in response to curcumin, in the absence of this repair mechanism, cells fail to develop DSBs. Owing to reduced DSBs, MMR-deficient cells do not activate the G₂-M checkpoint and, as a result, likely enter mitosis and subsequently trigger apoptosis through mitotic catastrophe (6, 12).

Our model proposes that MMR plays a significant role in eliciting cellular response to oxidative stress induced by curcumin. Of note, others have observed that MMR-deficient cells exhibit increased sensitivity to oxidative stress. Specifically, Chang et al. (40) observed that MMR-deficient cells are more sensitive to oxidative damage induced by H₂O₂ than MMR-proficient cells. This result is similar to what we have documented in response to curcumin. Interestingly, unlike what we have observed using curcumin, H₂O₂ treatment resulted in activation of G₂-M arrest through a MMR-independent mechanism. Oxyradicals generate a broad myriad of DNA damage (33); consequently, there are potential differences in the nature of lesions generated by curcumin versus those generated by other sources of oxidative stress such as H₂O₂. Nonetheless, our results show that curcumin-induced G₂-M arrest clearly proceeds in a MMR-dependent manner and its abrogation by UCN-01 enhanced curcumin sensitivity. Although failure to activate cell cycle arrest probably contributes to curcumin sensitivity in MMR-deficient cells, it should be borne in mind that curcumin activates multiple signaling pathways. For example, curcumin treatment is associated with the downregulation of NF-κB activation (14), decreased iNOS and DNA-PK catalytic subunit expression, and upregulation of p53, p21, p27(Kip1), and Chk2 (15). Curcumin activates c-Abl and c-Jun-NH₂-kinase kinases that are required for apoptosis induced by this anticancer agent (17). Moreover, a recent study showed that curcumin modulates the radiosensitivity of colorectal cancer cells by suppressing constitutive and inducible NF-κB activity (41). When taken together, these observations suggest that in addition to abrogated checkpoint signaling, the increased curcumin sensitivity observed in MMR-deficient cells may be attributable to the dysregulation of other signaling mechanisms. In addition to activating apoptosis, recent studies showed that in normal human cells, and malignant glioma cells, curcumin induced nonapoptotic, autophagic cell death (42). Clearly, the involvement of autophagy, if any, in conferring curcumin sensitivity in MMR-deficient cells remains unknown.

The results presented here have clear potential therapeutic value. Microsatellite instability–positive phenotype is commonly observed in several tumor types including colorectal, endometrial, gastric, and ovarian cancer (43). This feature is associated with the inactivation of the MMR system (44). Currently, curcumin is being evaluated clinically for the treatment of colorectal tumors (1, 45). Our findings clearly suggest that this anticancer agent may be especially useful in the treatment of microsatellite instability–positive tumors. Additionally, our findings suggest that combining curcumin with Chk1 inhibitors such as UCN-01 may potentially sensitize microsatellite instability–negative (MMR-proficient) tumor cells consequently enhancing cell death. Although curcumin bioavailability is suboptimal, liposomal curcumin, curcumin nanoparticles, and curcumin phospholipid complexes are being tested as alternate approaches for enhanced bioavailability (46).

No potential conflicts of interest were disclosed.

Grant Support: NIH grants GM 60945 (R. Baskaran), CA90787 (J.C. Yalowich), and CA102289 (K.D. Brown).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.