An MLH1 Mutation Links BACH1/FANCJ to Colon Cancer, Signaling, and Insight toward Directed Therapy Free

Read the Perspective on this article by Lynch et al., p. 1371

In the absence of DNA repair proteins, cell cycle checkpoints or DNA damage repair pathways are not properly activated, propelling tumorigenesis. Moreover, in the absence of DNA repair proteins, cancer cells can develop resistance to DNA-damaging agents used as chemotherapy. For example, loss of the mismatch repair (MMR) pathway is associated with hereditary nonpolyposis colon cancer (HNPCC; ref. 1) and with chemoresistance (2, 3). MMR-deficient cells not only resist DNA damage–induced arrest and evade apoptosis but also have a greatly enhanced mutation frequency. Thus, MMR-deficient cells often have a mutator phenotype associated with microsatellite instability (4).

MMR proteins are required to activate apoptosis in response to certain types of DNA lesions. For example, methyl nitrosourea (MNU) generates methylation at O⁶ in guanine of DNA to form O⁶-methylguanine (O⁶-meG). This lesion is sensed by MMR through the heterodimer MutSα (MSH2 and MSH6), which recruits the heterodimer MutLα (MLH1 and PMS2) to initiate MMR signaling and repair. However, the contribution of these responses to apoptosis is not fully understood (1). In one model, MMR proteins are hypothesized to facilitate misguided attempts to repair DNA methylation, ultimately leading to more severe secondary lesions, such as double-strand breaks (5). In an alternative model, MMR proteins are proposed to function directly in activating checkpoint and apoptosis independent of a repair function (6). Consistent with the latter model, in response to O⁶-meG lesions, the MutSα and MutLα complexes are required to recruit and activate the checkpoint kinase ATR (7). In either model, the initiation of a MMR response and apoptosis can be minimized if damage is reversed by the enzyme methylguanine methyltransferase (MGMT). MGMT can transfer a methyl group onto itself, thus repairing an O⁶-meG lesion in a single step.

Separation-of-function mutations in MMR genes show that both repair and checkpoint functions are critical for tumor suppression. For example, Msh2-null mice develop tumors with faster onset than mice that carry a missense mutation, Msh2^G674A, which disrupts repair but not checkpoint function (8). Moreover, Mlh1-null mice develop a full range of tumors, whereas mice that carry a missense mutation that disrupts repair, Mlh1^G67R, present fewer intestinal tumors (9). A separation-of-function mutant that disrupts checkpoint but not repair remains to be identified. Conceivably, such a mutant could exist among HNPCC sequence variants that are characterized by unknown pathogenicity and/or a microsatellite stable (MSS) phenotype.

Loss of DNA repair and checkpoint functions as well as cancer is also a characteristic associated with defects in the BRCA1-associated helicase FANCJ (also known as BACH1/BRIP1). FANCJ mutations were identified in breast cancer (10, 11) and also in the cancer-prone disease Fanconi anemia (12–14). Treatment of FANCJ-null FA-J cells with DNA cross-linking agents, such as mitomycin C (MMC), generates cellular sensitivity and a prolonged checkpoint response. These outcomes are corrected by complementation with wild-type FANCJ, but not with an MLH1-interaction defective mutant FANCJ^K141/142A (15). Given this finding, we considered that loss of the FANCJ/MLH1 interaction could be associated with cancer.

In this study, we uncover that loss of the FANCJ/MLH1 interaction is associated with HNPCC. Specifically, we identified an MLH1 clinical mutation, MLH1^L607H, which ablates MLH1 binding to FANCJ and alters the DNA damage response. We identify that expression of MLH1^L607H in MLH1-null cells generates sensitivity to MMC but resistance to MNU. The resistance to MNU is dependent on MGMT activity. Likewise, we find that FANCJ-null and FANCJ-depleted cells are resistant to MNU when MGMT is active. The data indicate that the MGMT dependence of our findings is due to reduced MMR function, which allows greater time for MGMT lesion reversal. In particular, we find that FANCJ deficiency delays MMR checkpoint and apoptotic responses. We suggest that this delay could explain some HNPCC familial cancers, such as MLH^L607H, characterized as MSS and with intact repair. The link between FANCJ, MMR signaling, and colon cancer suppression also provides insight toward directed therapy because loss of the FANCJ/MLH1 interaction sensitizes cells to DNA cross-linking agents.

MCF7 and 293T cells were grown in DMEM supplemented with 10% fetal bovine serum and penicillin/streptomycin (100 units/mL each). HCT116 cells were grown in McCoy's 5A medium supplemented with 10% fetal bovine serum and penicillin/streptomycin (100 units/mL each). FA-D2, FA-D2+FANCD2, FA-C, and FA-G cell lines were obtained from Fanconi Anemia Research Fund (FARF) and grown in DMEM supplemented with 15% fetal bovine serum and penicillin/streptomycin (100 units/mL each). FA-J and fibroblast cell lines were cultured as previously described (12). FA-J cells were infected with the POZ retroviral vector (16) containing no insert, FANCJ^WT, FANCJ^K52R, and FANCJ^K141/142A and pCDNA3 vectors were described previously (15).

Cells were untreated or treated with 1 mmol/L hydroxyurea (continuous treatment) or MNU (1 hour, serum-free), harvested, lysed, and processed for Western blot analysis as described previously (15). The antibodies used for immunoprecipitation and Western blot assays included FANCJ, monocolonal pool (1A3, 2G7, and IG5; ref. 10) and polyclonal, E67 (17), β-actin (Sigma), MLH1 (BD Biosciences), PMS2 (BD Biosciences), MSH6 (BD Biosciences), MSH2 (Calbiochem), Chk1 (Bethyl), pChk1317 (Bethyl), and MGMT (NeoMarkers). To prepare chromatin extracts, MCF7 cell lines were prepared as described (18). Briefly, cells were left untreated or treated with 1 mmol/L MNU for 1 hour, washed, and incubated for 24 hours. The cells were washed extensively with hypotonic buffer containing digitonin, collected, and sonicated.

FA-J complemented cells, stable MCF7 cells, or transiently transfected HCT116 cells were seeded onto six-well plates at 1,000 per well; incubated overnight; and treated with increasing doses of MNU (1 hour, serum-free), methyl methanesulfonate (MMS) (1 hour, serum-free), N-methyl-N-nitro-N-nitrosoguanidine (1 hour, serum-free), or N-(2-chloroethyl)-N′-cyclohexyl-N-nitrosourea (1 hour, serum-free). To suppress MGMT, 2 hours before exposure to DNA damage agents, cells were incubated with 20 μmol/L O⁶-benzylguanine (O⁶-BZG). In addition, O⁶-BZG was also included during and after treatment. DNA-damaged cells were counted after 5 to 8 days using a hemocytomerter or measured photometrically in a model 3550 microplate reader (Perkin-Elmer) as the relative growth (in luciferase units) using the CellTiter-Glo viability assay (Promega). Colonies were stained with Giemsa (LabChem) and counted. Percent growth or colony survival was calculated as (treated cells/untreated cells) × 100.

Stable MCF7 cells were plated onto 60-mm plates, incubated overnight, and treated with 1 mmol/L MNU (1 hour, serum-free). O⁶-BZG was used to suppress MGMT. Cells were washed 72 and 96 hours after treatment, and 1 × 10⁶ cells were stained with Annexin V-FITC (Sigma) and propidium iodide (Sigma). Stained cells were analyzed by fluorescence-activated cell sorting.

Previously, we found that the FANCJ/MLH1 interaction was required for FANCJ function in DNA cross-link repair. Unlike wild-type FANCJ, the MLH1-interaction–defective mutant FANCJ ^K141/142A failed to restore MMC resistance to FA-J cells (15). To perform the reciprocal experiment, we sought to identify a FANCJ-interaction–defective mutant of MLH1. Deletion of MLH1 amino acids 478 to 744 disrupted not only FANCJ but also PMS2 binding (ref. 15; Fig. 1A). To selectively disrupt FANCJ binding, we considered whether an MLH1 missense mutation associated with HNPCC could be useful. Thus, MLH1 clinical mutants targeting the region of amino acids 478 to 744 and proficient for PMS2 binding were screened for FANCJ binding. Wild-type or mutant MLH1 constructs were cotransfected with PMS2 into MutLα–deficient 293T or HCT116 cells. MLH1 antibody precipitation revealed that most of the MLH1 mutants precipitated PMS2 and FANCJ similar to wild-type MLH1. Notably, mutation of MLH1 at amino acid 607, from leucine to histidine (MLH1^L607H), disrupted FANCJ, but not PMS2, binding (Fig. 1B and C). In the reciprocal immunoprecipitation with FANCJ antibody, FANCJ precipitated wild-type MLH1, but not the MLH1^L607H mutant (Fig. 1C).

Given that cells lacking the FANCJ/MLH1 interaction were sensitive to MMC (15), we considered whether expression of MLH1^L607H, as compared with wild-type MLH1, differentially affected the MMC sensitivity of HCT116 cells. To address this possibility, vector, wild-type, or MLH1^L607H were cotransfected with PMS2; cells were plated at equal numbers; and cell proliferation was assessed in response to increasing dose of MMC. Expression of wild-type MLH1 as compared with vector control did not measurably alter the sensitivity of HCT116 cells to MMC. In contrast, expression of MLH1^L607H enhanced MMC sensitivity (Fig. 1D). Thus, the clinical MLH1^L607H mutant resembled the FANCJ^K141/142A mutant in sensitivity to MMC, further indicating that the FANCJ/MLH1 interaction is important for processing DNA cross-links.

MLH1 functions in the MMR pathway, which is required for cellular sensitivity to DNA methylating agents such as MNU (19). To address whether this sensitivity requires an intact FANCJ/MLH1 interaction, transfected HCT116 cells were treated with increasing doses of MNU. Introduction of wild-type MLH1 restored MNU sensitivity to HCT116 cells. As expected, the MNU-induced sensitivity was most dramatic when DNA methylation reversal by MGMT was inhibited with O⁶-BZG (Fig. 1E). Similar to wild-type MLH1, introduction of MLH1^L607H restored MNU sensitivity when MGMT was inhibited (Fig. 1E). Surprisingly, when MGMT was active, in contrast to wild-type MLH1, introduction of MLH1^L607H did not enhance MNU sensitivity. Instead, cells expressing MLH1^L607H were resistant to MNU, similar to vector-transfected cells (Fig. 1E). In summary, a defective FANCJ/MLH1 interaction enhanced MMC sensitivity but reduced MNU sensitivity when MGMT was active.

If FANCJ expression was required for a robust MNU-induced sensitivity, we reasoned that FANCJ-null FA-J fibroblast cells should be more resistant to MNU than FA-J cells complemented with wild-type FANCJ. To address this idea, we first confirmed that wild-type FANCJ was functionally expressed in FA-J cells by immunoblot analysis and restored resistance to MMC (Fig. 2A; Supplementary Fig. S1). Next, we scored resistance to MNU with MGMT active or inhibited in proliferation assays. Remarkably, wild-type FANCJ restored MNU sensitivity when MGMT was active but had no effect when MGMT was inhibited (Fig. 2B). In contrast, wild-type FANCD2 complementation in FANCD2-null FA-D2 patient cells had no effect on MNU resistance with active MGMT (Fig. 2C). Furthermore, in comparison with other Fanconi anemia patient cell lines, FA-C (PD331) and FA-G (PD352), or normal fibroblasts, FA-J cell lines (2833 and EUFA30F) were more resistant to MNU (Fig. 2D).

The MGMT dependence of our findings could reflect a role for FANCJ in modulating MGMT protein levels or activity. However, FANCJ expression in FA-J cells did not dramatically change the expression of MGMT in response to MNU (Fig. 3A). Greater MGMT activity should reduce the toxicity of N-(2-chloroethyl)-N′-cyclohexyl-N-nitrosourea, which generates DNA methylation processed by MGMT, but not by MMR (20). Instead, toxicity was greater in FANCJ-deficient as compared with wild-type FANCJ–expressing FA-J cells (Fig. 3B). Similar to MNU, however, FANCJ deficiency enhanced resistance to N-methyl-N-nitro-N-nitrosoguanidine (Fig. 3B), a DNA methylating agent that, similar to MNU, generates O⁶-meG lesions processed by MMR. Taken together, the data indicated that FANCJ deficiency altered MMR function, but not MGMT function.

Given these findings, we sought to test the prediction that FANCJ promoted MMR function through a direct FANCJ/MLH1 interaction. Thus, FA-J cells were complemented with wild-type FANCJ or the MLH1-interactive defective mutant FANCJ^K141/142A, as well as the enzyme inactive mutant FANCJ^K52R (15, 17). Whereas protein expression was similar (Fig. 3C), unlike wild-type FANCJ, FANCJ^K52R and FANCJ^K141/142A failed to restore MNU sensitivity. Instead, MNU resistance in the mutant-expressing lines was similar to that in FA-J cells complemented with vector (Fig. 3D), indicating that both the MLH1 binding and enzyme activity of FANCJ were required to restore MNU sensitivity.

Conceivably, the MNU resistance in FANCJ-deficient cells could reflect reduced processing by MMR, which in turn enables a greater number of lesions to be reversed by MGMT. To test this idea, we sought to directly compare how FANCJ or MLH1 deficiency affected the MNU-induced checkpoint and apoptotic response. Western blot confirmed that shRNA reagents targeted FANCJ and MLH1 proteins in MCF7 breast cancer cells (Fig. 4A). FANCJ deficiency did not alter the expression of the MutLα complex (Fig. 4A). In contrast, shRNA targeting MLH1 suppressed both MLH1 and PMS2, confirming reports that MLH1 stabilizes PMS2 (ref. 21; Fig. 4A). As found in FA-J cells, FANCJ-deficient MCF7 cells were resistant to MNU when MGMT was active (Fig. 4C). In fact, colony survival data in MLH1- or FANCJ-depleted MCF7 cells were similar; both had ∼2-fold enhanced colony survival as compared with nonsilencing controls (Fig. 4C). When MGMT was inhibited, FANCJ depletion had no effect on MNU-induced colony survival as compared with MLH1 depletion, which dramatically enhanced the number of surviving colonies (Fig. 4D). Likewise, MLH1 depletion reduced apoptosis following MNU treatment irrespective of MGMT activity (Fig. 4B). Instead, FANCJ depletion only had a measurable effect on apoptosis in MGMT-active cells (∼2-fold) as compared with nonsilencing controls (Fig. 4B).

MMR is required for Chk1 activation following MNU treatment (7). Consistent with this point, depletion of MLH1 reduced MNU-induced, but not hydroxyurea-induced, Chk1 phosphorylation (Fig. 4E and F). As expected, in comparison with MGMT-inhibited cells, MGMT-active cells required a higher dose of MNU (∼1.5 mmol/L) to induce Chk1 phosphorylation (data not shown and Fig. 4E and F). Depletion of MLH1 reduced MNU-induced Chk1 phosphorylation by ∼1.7- and ∼5.5-fold at 1.5 and 2 mmol/L MNU, respectively (Fig. 4E and F). Interestingly, as compared with nonsilencing control, FANCJ depletion also decreased Chk1 phosphorylation induced with 1.5 and 2 mmol/L MNU by ∼1.7- and ∼2.2-fold, respectively. Furthermore, depletion of FANCJ with two distinct siRNAs also reduced MNU-induced Chk1 phosphorylation by ∼1.7 - and ∼1.8-fold at 1.5 mmol/L MNU (Supplementary Fig. S2). Thus, FANCJ is required for maximal MNU-induced checkpoint signaling, sensitivity, and apoptosis when MGMT is active.

The complex role of MMR proteins in the DNA damage response has complicated efforts to understand which function(s) is required for tumor suppression. Furthermore, MMR-associated HNPCC tumors have heterogeneous defects and varying levels of MSI. Here, we uncover the potential pathogenic defect of an HNPCC-associated MLH1 mutant, MLH1^L607H, to be lack of FANCJ binding. Loss of FANCJ binding is unlikely to affect repair given that the MLH1^L607H mutant was shown previously to be intact for MMR, have normal expression, and generate tumors that are MSS (22–24). Instead, our data indicate that loss of FANCJ binding reduces, but does not inactivate, MMR signaling. Most notably, in the absence of FANCJ, cells can initiate processing, checkpoint activation, and undergo cell death. However, the timing of these events seems to be delayed. With this delay, the MMR-independent methylation reversal by MGMT likely has time to enhance DNA methylation resistance (see model in Fig. 5). An intriguing question is whether this delay in MMR signaling could be a mechanism linked to cancer and/or chemoresistance.

Our findings further show that a functional relationship exists between FANCJ and MLH1 in the DNA damage response. Similar to the response to DNA cross-links, the response to DNA methylation requires FANCJ helicase and MLH1 binding activities. Why does loss of the FANCJ/MLH1 interaction enhance sensitivity to agents that induce DNA cross-links but reduce sensitivity to agents that induce O⁶-meG? In response to DNA cross-links, the delay in MMR signaling likely interferes with repair and/or recovery. In contrast, in response to O⁶-meG lesions, the delay in MMR signaling is prosurvival because MGMT reversal is available. If MGMT is inactivated, however, O⁶-meG lesions will inevitably be processed by MMR and induce apoptosis. Because MGMT is typically inhibited in most studies analyzing MMR contribution to DNA methylation signaling, it is unclear whether other MLH1 or HNPCC-associated mutations will generate resistance to MNU when MGMT is active. If so, tumors derived from such patients may also be selectively sensitive to MMC.

Our data show that FANCJ contributes to MMR signaling. Whether FANCJ facilitates loading, translocation, or stabilization of MMR complexes on chromatin is not clear. Loss of FANCJ or the FANCJ/MLH1 interaction may not overtly limit MMR activity, given that patients with the MLH1^L607H mutation are characterized as MSS (25) and MMR activity for this mutant is similar to wild-type MLH1 (24). It is possible, however, that delayed MMR signaling in MLH1^L607H patient cells might provide a selective advantage in the initial stages of tumorigenesis, or under high mutation stress, enhance the mutation frequency. Conceivably, more DNA damage could be tolerated without activating the MMR-induced checkpoint. This lowered checkpoint barrier could enhance genomic instability and cancer. Mouse model studies will be useful to estimate the risk associated with the MLH1^L607H mutant allele.

No potential conflicts of interest were disclosed.

Grant support: Grant RO1 CA129514-01A1 (S.B. Cantor). We thank Mr. and Mrs. Edward T. Vitone, Jr. for charitable contributions.

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.