Mismatch repair system decreases cell survival by stabilizing the tetraploid G1 arrest in response to SN-38

Abstract

The role of the mismatch repair (MMR) system in correcting base–base mismatches is well established; its involvement in the response to DNA double strand breaks, however, is less clear. We investigated the influence of the essential component of MMR, the hMLH1 protein, on the cellular response to DNA-double strand breaks induced by treatment with SN-38, the active metabolite of topoisomerase I inhibitor irinotecan, in a strictly isogenic cell system (p53^wt, hMLH1⁺/p53^wt, hMLH1⁻). By using hMLH1 expressing clones or cells transduced with the hMLH1-expressing adenovirus as well as siRNA technology, we show that in response to SN-38-induced DNA damage the MMR proficient (MMR⁺) cells make: (i) a stronger G2/M arrest, (ii) a subsequent longer tetraploid G1 arrest, (iii) a stronger activation of Chk1 and Chk2 kinases than the MMR deficient (MMR⁻) counterparts. Both Cdk2 and Cdk4 kinases contribute to the basal tetraploid G1 arrest in MMR⁺ and MMR⁻ cells. Although the Chk1 kinase is involved in the G2/M arrest, neither Chk1 nor Chk2 are involved in the enhancement of the tetraploid G1 arrest. The long-lasting tetraploid G1 arrest of MMR⁺ cells is associated with their lower clonogenic survival after SN-38 treatment, the abrogation of the tetraploid G1 arrest resulted in their better clonogenic survival. These data show that the stabilization of the tetraploid G1 arrest in response to double strand breaks is a novel function of the MMR system that contributes to the lesser survival of MMR⁺ cells.

The DNA mismatch repair (MMR) system recognizes and repairs erroneous insertions, deletions and mis-incorporations of bases that can arise during DNA replication and recombination.1 It also recognizes and repairs bulky lesions created in the DNA by certain chemical agents, for example, MNU, MNNG, 6-TG or cisplatin. If the load of the lesions is in excess of the repair capacity, MMR triggers apoptosis thus contributing to a lesser survival of MMR proficient cells.2

MMR proficient cells have also been reported to be more sensitive to irinotecan-induced double strand breaks than the semi-isogenic hMLH1 deficient counterparts,3, 4 a reaction which can not be explained by the repair function of the MMR system. The situation was complicated by the fact that the observations were made in semi-isogenic cell systems, with their known limitations.5

Why the MMR proficient cells are more sensitive to double strand breaks has not been clarified. In fact, the role of MMR in the reaction to double strand breaks is debated.6 Double strand breaks caused by topoisomerase I inhibitors, like irinotecan, are predominantly repaired by homologous recombination,7 which may be affected by the MMR system.8 It has been hypothesized that after irradiation double strand breaks can lead to oxidative products like 8-oxoguanosines, which could be recognized in the nascent DNA strand by the MMR recognition complex9; this is less likely after irinotecan treatment.

We and others showed previously, that the main cellular effect of irinotecan or its metabolite SN-38 on p53^wt colon carcinoma cells is the triggering of a short p53-independent G2/M arrest followed by a p53-dependent G1 arrest of tetraploid cells.10, 11 Tetraploid G1 arrest is a durable arrest following adaptation of the G2- or mitotic arrest.12 The maintenance of this arrest determines the long-term growth of tumor cells after irinotecan treatment.13

We also reported that the maintenance of the p53-dependent long-term tetraploid G1 arrest after treatment with irinotecan is longer in MMR-proficient than in MMR-deficient cells.14

The nature of the signal triggering this tetraploid G1 arrest is not known (for review, see Ref. 15). It has been shown to be dependent on p21 protein, which induces a G1 arrest through inhibition of Cdk2 and Cdk4 kinase activity thus preventing phosphorylation of the Rb protein16 and leading to E2F1 recruitment and cell arrest. The contribution of Cdk1 inhibition to tetraploid G1 arrest in irradiated HCT116 cells has also been demonstrated.17 The relationship between the MMR system and the tetraploid G1 arrest have not been investigated.

Evidence has recently accumulated that MMR not only repairs DNA, but also, independently, stimulates the G2 checkpoint induced by MMR-activating agents. The creation of “separation of function” mutants allowed to demonstrate that the function of MMR in response to DNA damage is independent from its function in the checkpoint.18, 19 Another way of separating the repair function from the checkpoint stimulating function, applied in the present work, is the infliction of DNA damage not causing mismatches and following the contribution of MMR to transduction of the checkpoint signals and survival.

To precisely address the questions of the influence of MMR on the cellular response of p53^wt cells to DNA-damage, we generated cell systems isogenic for hMLH1 by selecting clones of the HCT116 cell line, stably transfected either with pcDNA vector (mock transfectants) or with pcDNA-hMLH1 (hMLH1 transfectants). Additionally, HCT116 transiently transduced with an adenovirus expressing functional hMLH1 (AdV-hMLH1) were used as a second isogenic model cell system. The treatment with SN-38 triggered in all clones a short G2/M arrest, followed by a long-term tetraploid G1 arrest. The presented data indicate that in the MMR proficient cells both arrest steps are enhanced and the latter one suppresses long-term cell survival. Some aspects of the enhancement of the tetraploid G1 arrest in MMR proficient cells were investigated in detail.

Material and Methods

Cell lines

HCT116 (MMR⁻), HCT116+chr2 (MMR⁻), HCT116+chr3 (MMR⁺) cells were a kind gift from Dr. C.R. Boland (Baylor University Medical Center, Dallas, USA). The HCT116Chk2^−/− cell line was kindly donated by Dr. Bert Vogelstein. HCT116 cell lines were cultured in DMEM with 10% FCS. The culture medium for HCT116+chr3 cells contained 0.4 mg/ml G418. The ovarian carcinoma cell lines A2780 (hMLH1⁺) and A2780/cp70 (hMLH1⁻) were a kind gift from Dr. J. Plumb (CRUK Centre for Oncology and Applied Pharmacology, Beatson Laboratories, University of Glasgow, UK). They were cultured in RPMI with 10% FCS (all media and FCS were obtained from Biochrom, Berlin, Germany).

Transfections

Parental HCT116 cells were stably transfected with 6 μg pcDNA3.1-myc-hMLH1 (kindly provided by Dr. Trojan, Medizinische Klinik I, Klinikum der Johann Wolfgang Goethe-Universitat, Frankfurt a.M., Germany) or with pcDNA3.1-myc using Fugene 6 transfection reagent (Roche Molecular Biochemicals, Mannheim, Germany) as per the manufacturer's recommendations. Forty-eight hours after transfection, the cells were trypsinized and seeded in medium (DMEM+10% FCS) containing G418 (1 mg/ml). Colonies were individually picked and expanded. The clones were characterized for the expression of hMLH1 as well as for the functionality of the MMR system by performing a clonogenic assay and cell cycle distribution analysis in the presence of MNU. Only those clones that showed both hMLH1 expression and sensitivity to MNU were used for further investigation.

For transient transfections, HCT116 cells were seeded and transfected 24 h later with the following plasmids: pEGFP-C1-Chk1 WT, pEGFP-C1-Chk1 MUT (D130A), kindly donated by Dr. Piwnica-Worms,20 pEGFP-C2-Chk2, kindly donated by Dr. Giacomo Buscemi21 and pEGFP-C2 (empty plasmid) using Lipofectamine 2000 (Invitrogen, Karlsruhe, Germany) according to the manufacturer's instructions. In short, 24 h after seeding FCS containing medium was removed, cells were washed once with serum-free medium and transfection was done in the absence of serum.

Five hours later, transfection solution was removed and medium with 10% FCS was added. After 24 h, cells were trypsinized and replated. Twenty-four hours later, cells were treated with SN-38 (10 nM) for 2 days and then harvested for lysates or in parallel fixed in 70% ethanol for flow cytometry. Cells that were harvested at day 4 were washed with PBS 2 days after treatment and kept in drug-free medium for another 2 days.

Treatment protocols

For characterization of the MMR function, the cells were treated with up to 3 mM MNU (N-Nitroso-N-Methylurea, Sigma-Aldrich, Taufkirchen, Germany) for 2 h (clonogenic assay) or daily with 1 mM for up to 4 days (cell cycle analysis). For the analysis of SN-38-induced responses, the cell lines were treated with 10 nM SN-38 for 2 days, then washed, cultured in fresh drug-free medium and harvested. The different time points after start of treatment are given in the legends.

Suppression of gene expression

hMLH1-expressing clones were transfected by electroporation at 290 V and 1050 μF capacitance in a GenePulser (BioRad Laboratories, München, Germany) with 20 μg pSUPER-EGFP (a kind gift from Dr. M. Truss, Charité Campus Virchow, Humboldt University, Berlin, Germany) or with pSUPER-hMLH1 as described.22 Cells were then seeded in 2 ml prewarmed culture medium and incubated at 37°C for 24 h. Cells were washed and cultured in medium containing SN-38 and were harvested after 2 and 5 days after start of treatment.

Alternatively, siRNA directed against hMLH1, Cdk2, Cdk4, Cdk6, Chk1 (all from Qiagen, Hilden, Germany) was used for gene suppression. As a control, siRNAGFP or Stealth siRNA P1 (Invitrogen) was taken. Cells were seeded and 24 h later transfection was done with HiPerFect transfection reagent (Qiagen). For hMLH1 and Chk1, transfection was done a second time after 8 h. On the next day, cells were trypsinized, replated and after 24 h treated with SN-38 (10 nM) for 2 days. Cells were either harvested on day 2 or washed and further cultured in drug-free medium until the indicated time points.

Western blotting

At different time points after treatment, adherent cells were trypsinized, washed with PBS, and cell pellets were lysed in lysis buffer (0.2%NP-40, 137 mM NaCl, 20 mM Tris-HCl pH 7.5, 1 mM sodium orthovanadate, 12 mM β-glycerophosphate, 1 mM EDTA, 1.5 mM MgCl₂, 50 mM NaF, 10% glycerol, 1 mM PMSF and 10 μg/ml aprotinin). SDS-PAGE and Western blotting were performed as described recently.11 The following antibodies against human proteins were employed: mouse anti-p53 (DO-7) (Dako, Hamburg, Germany), rabbit anti-p21 (C-19), rabbit anti-cyclin B1 (H-433; sc-752), rabbit anti-Cdk1 (H-297; sc747), rabbit anti-Chk1 (FL476), rabbit anti-Chk2 (H-300) (all from Santa Cruz); rabbit anti-phospho-Cdk1 (Tyr-15), rabbit anti-phospho-Chk1 (Ser-317, -345), rabbit anti-phospho-Chk2 (Thr-68), mouse anti-Cdk2, rabbit anti-phospho-Rb antibodies against Ser807/811 (all from Cell Signaling, Frankfurt, Germany); mouse anti p27/Kip1, mouse anti-hMLH1, mouse anti-Cdk4, (all from BD Biosciences, Heidelberg, Germany), mouse anti-β-actin (clone AC-15) and mouse anti-γ-tubulin (Sigma-Aldrich Chemie, Taufkirchen, Germany); anti-cyclin D1 (Ab-3; Neomarkers, Fremont, CA). Two peroxidase-conjugated second antibodies were used: goat anti-mouse (IgG+IgM) or goat anti-rabbit IgG (Dianova, Hamburg, Germany). Detection was carried out using SuperSignal West Pico chemiluminescent substrate (Pierce, Rockford, IL). All Western blots were performed 3–5 times.

Clonogenic assays

About 200–300 cells were seeded as a single cell suspension in 60-mm diameter tissue culture dishes and were allowed to adhere for 24 h. The cells were treated for 2 h (MNU) or for 2 days (SN-38, in the presence or absence of UCN-01, as mentioned in the figure legend), followed by thorough washing with PBS and further culturing in fresh drug-free medium for 14 days. The colonies were fixed and stained with 0.2% crystal violet and counted.

Selection and investigation of SN-38-resistant cells

Cells of the hMLH1 transfectant clone 43 were seeded for clonogenic assay, treated with 6 nM SN-38 for 48 h and the surviving colonies were used for a second and a third identical clonogenic assay. The cells surviving the third clonogenic assay were collected after 2 weeks and further investigated for cell cycle distribution.

Caffeine treatment

Cells of mock transfectants and hMLH1-clone 43 were seeded and 24 h later pretreated with caffeine (final concentration for mock 0.5 mM, for clone 43 2 mM) for 24 h. Then SN-38 (10 nM) was added for 2 days. At day 2, cells were harvested and fixed in 70% ethanol as described for cell cycle analysis. Dishes for 6-day time point were washed with PBS and left in drug-free medium for another 4 days.

Flow cytometry

For cell cycle analysis, adherent cells were trypsinized and pooled with the floating cells. 10⁶ cells were fixed in 1 ml ice-cold 70% ethanol for 2 h at −20°C, washed in PBS and stained in PBS containing 0.1% Triton-X 100, 200 μg/ml RNaseA and 20 μg/ml propidium iodide (all from Sigma) for 30 min. For experiments with pEGFP plasmids, transfected cells were fixed in 70% ethanol for 30 min at 4°C, washed with 1% BSA in PBS and stained with propidium iodide as for cell cycle analysis. Analysis of DNA content was done only in GFP-positive cells as described previously.20 Data were analyzed on a FACS Calibur using CellQuest (BD Biosciences). Mean ± SD of 3 independent experiments was calculated for each time point.

Immunoprecipitation of CDK2, CDK4, CDK6

Immunoprecipitation was performed from 500 μg of whole cell lysates. The lysates were precleared with 30 μl of either mouse- or rabbit IgG-agarose (Sigma) for 1 h at 4°C by gently shaking the samples. The samples were centrifuged at high speed for 2 min, the precleared supernatant was collected and 3 μg of the primary antibody was added. The samples were gently shaken overnight at 4°C followed by the addition of 50 μl of anti-mouse- or anti-rabbit-IgG-agarose (Sigma) for 2 h at 4°C. The agarose-immunoprecipitates were washed 3 times with lysis buffer (0.2% NP-40, 137 mM NaCl, 20 mM Tris-HCl pH 7.5, 1 mM sodium orthovanadate, 12 mM β-glycerophosphate, 1 mM EDTA, 1.5 mM MgCl₂, 50 mM NaF, 10% glycerol, 1 mM PMSF and 10 μg/ml aprotinin). Finally, 2 × SDS sample buffer was added and the samples were boiled for 5 min. The supernatant was subjected to SDS-PAGE and detected for the presence of various proteins as mentioned in the figure.

Kinase assay

For Cdk4 or CDK6 kinase activity, immunoprecipitation was done with 3 μg rabbit anti-Cyclin D1(M-20) (Santa Cruz) or 3 μg rabbit anti-CDK6 antibody (Santa Cruz), as described above. For Cdk2 kinase activity, rabbit anti-Cdk2-agarose (M-2) (Santa Cruz) was used. Agarose-immunoprecipitates were washed 2 times with washing buffer [50 mM HEPES, pH 7.5, 0.1% NP-40. 150 mM NaCl, 10 mM ß-Glycerophosphate, 2.5 mM sodium pyrophosphate, 1 mM sodium orthovanadate, 5 mM NaF, 1 mM PMSF, 1 mM EDTA, 1 mM EGTA, Leupeptin, Aprotinin, Pepstatin (10 μg/ml each)] and 3 times with kinase buffer (20 mM HEPES, pH 7.5, 5 mM MgCl₂, 2.5 mM MnCl₂, 1 mM DTT). The kinase reaction was carried out in a 20 μl volume of kinase buffer containing 1 μg recombinant Rb (aminoacids 769-921, Santa Cruz) and 50 μM ATP (Sigma) for 30 min at 37°C. The phosphorylation reactions were stopped by adding an equal volume of 2 × SDS sample buffer. Samples were boiled for 5 min, centrifuged; the supernatant was recovered and loaded onto SDS-PAGE. The phosphorylation of Rb (Ser807/811) was detected in Western blot with the specific antibody. Specific enzymatic activity was determined by densitometry as the ratio of the phospho Rb Ser807/811 signal and the kinase protein signal. For detection of phospho Rb Ser807/811 and Cdk4 signals in the kinase assay samples the rabbit-TrueBlot System from eBioscience (NatuTec, Frankfurt, Germany) was used, according to the manufacturer's recommendations.

Overexpression of hMLH1 protein using adenovirus expression system

HCT116 cells were seeded, and after 24 h they were incubated with the adenovirus (AdV-hMLH1 or AdV-LacZ as a control) at the desired multiplicity of infection (MOI) for 90 min at 37°C in medium without serum. Then, medium was added with a final FCS concentration of 10%. The cells were harvested for preparation of lysates at indicated time points.

Results

Establishment and characterization of stable hMLH1-transfectant clones and the hMLH1-expressing adenovirus

The HCT116 cell line was stably transfected with the hMLH1 plasmid or the empty vector (mock transfection) and the transfectants were grown in selective medium. Several transfectant clones were expressing hMLH1, detectable in Western blot and were more sensitive to MNU in the clonogenic assay than the mock transfectants or the parental cell line. The clones 26 and 43 were investigated in detail.

Whole cell lysates of mock- and hMLH1-transfectant clones as well as those of cell lines HCT116 and HCT116+chr3, A2780 and A2780cp70 were tested for the expression of hMLH1 by Western blotting. The mock-transfected clone as well as the HCT116 and A2780cp70 cell lines showed no detectable hMLH1 protein as expected, whereas the hMLH1-transfected clones and the cell lines HCT116+chr3 and A2780 showed comparable hMLH1 expression (Fig. 1a).

Figure 1. Transfection of HCT116 cells with the hMLH1-expressing plasmid or transduction with AdV-hMLH1 restores the signaling function of hMLH1. (a) The clones 26 and 43 as well as the cell lines HCT116+ch3 and A2780 express hMLH1 detectable in Western blot, while mock-transfectants, A2780/cp70 and HCT116 do not; (b) Two days after treatment with MNU (1 mM) the clones 26 and 43 show a G2/M arrest while the mock clone is not affected; (c) Clones 26 and 43 are more sensitive to MNU treatment (1–3 mM, 2 h) in clonogenic assay than mock-transfectants; (d) Transduction of HCT116 cells with the hMLH1-expressing adenovirus (AdV-hMLH1 at 1 and 10 MOI)) decreases their clonogenic survival in a dose-dependent manner after MNU treatment, mean values of triplicates. Inset: expression of hMLH1 protein after transduction with 1 and 10 MOI of AdV-hMLH1. The expression at 1 MOI corresponded to that in clone 43 (not shown); (e) Transduction with the AdV-hMLH1 enhances in a dose-dependent manner the MNU-induced G2/M arrest after 1 day of treatment. Means of 3 experiments ± SD. Transduction with Adv-hMLH1 alone does not induce any G2/M arrest (not shown).

To determine the functional status of the expressed hMLH1 protein, the short- and the long-term response to N-methyl-N-nitrosourea (MNU), an alkylating agent known to activate the MMR was analyzed. Two days after treatment, the hMLH1 transfectants but not the mock transfectants exhibited a strong G2/M arrest (Fig. 1b). The clonogenic assay showed that the hMLH1-expressing clones were more sensitive to MNU than the mock clone (Fig. 1c).

These data indicated that the obtained clones 26 and 43 expressed a functionally competent hMLH1 protein. These clones were used as a model system to investigate the role of hMLH1 in response to topoisomerase I inhibitor SN-38.

Transduction of HCT116 cells with the hMLH1-expressing adenovirus (AdV-hMLH1) decreased the clonogenic survival after MNU treatment for 2 days (Fig. 1d) and triggerred a G2/M arrest 1 day after treatment (Fig. 1e) in a dose-dependent manner. At MOI of about 10 the extent of the arrest was 60%, that is, it was similar to the arrest of the transfectant clones. This indicated that the adenovirus expressed hMLH1 protein, which at MOI of 10 exerted the same function as the endogenous hMLH1. Since the transduction with 1 MOI yielded hMLH1 expression comparable to that in clone 43, it was evident that the adenovirally coded protein was less active than the physiologically expressed one (Fig. 1d, inset).

MMR enhances the G2/M arrest and stabilizes the subsequent tetraploid G1 arrest after SN-38 treatment

We showed previously13 that several p53^wt colon carcinoma cell lines treated for 2 days with SN-38 enter a G2/M arrest, which after removal of the drug is followed by a long-term tetraploid G1 arrest.

We asked the question how this p53-dependent response is affected by the MMR status.

The effect of hMLH1 protein expression on the time course of this response was investigated in the transfectant clones for 9 days after start of treatment. There was negligible apoptosis in the clones 26 and 43 after SN-38 treatment; the mock transfectants showed a slight apoptosis on day 4 and 6 after start of treatment (not shown). Both, the G2/M arrest, measured at day 2 and the tetraploid G1 arrest, measured 3–9 days after start of treatment, were stronger in hMLH1 clones than in the mock transfectant (Fig. 2a). A higher tetraploid G1 arrest in MMR proficient cells was also observed 6 days after start of treatment in the semi-isogenic cell pair HCT116+chr2/HCT116+chr3 (34% vs. 65%, Supporting Information Fig. S1A), thus independently confirming the results obtained with the stably transfected clones.

Figure 2. hMLH1 expression enhances G2/M and as well as the tetraploid G1 arrest after SN-38 treatment. (a) The G2/M arrest is stronger and the release is slower in clones 26 and 43 than in mock transfectants. The curves represent 2 different tetraploid states: G2/M (Day 1–2) and tetraploid G1 (Day 3–9, see Fig. 3); Bar indicates the duration of treatment. Mean values ± SD of at least 3 experiments. (b) Transduction of mock transfectants with AdV-hMLH1 (1 MOI) enhances the tetraploid arrest on day 4 by 13% as compared to cells transduced with the vector expressing β-galactosidase. Representative of 2 experiments. Transduction with the virus alone does not affect the cell cycle (not shown). (c) Suppression of hMLH1 with pSuper-hMLH1 in clone 43 leads to (d) A decrease of the G2/M arrest at day 2 (p = 0.01) and of the tetraploid G1 arrest at day 5 (p = 0.036). Mean values of 3 experiments ± SD. Similar results were obtained with clone 26 (not shown).

When hMLH1 signaling function was restored in the mock transfectant by transduction with amounts of AdV-hMLH1 expressing physiological levels of the hMLH1 protein, the tetraploid arrest at day 6 after SN-38 treatment was increased (Fig. 2b).

Conversely, the suppression of hMLH1 by pSuper-hMLH1 in clone 43 (as well as in clone 26) prior to treatment with SN-38 leads to suppression of the G2/M arrest at day 2 and of the tetraploid G1 arrest at day 5 (Fig. 2d).

The time course of the transition from the G2/M arrest to the tetraploid G1 arrest after SN-38 treatment was investigated in detail in mock transfectants and in both clones (Fig. 3). The phosphorylation of Cdk1 (P-Cdk1) and the expression of cyclin B1 persisted 72 h after start of treatment in hMLH1 clones, versus 48 h in mock transfectants, indicating that the latter left the G2/M arrest earlier. This result is consistent with our previous findings14 and the findings of others.9 Consequently, the tetraploid G1 arrest begun earlier in the mock-transfected cells, indicated by the earlier expression of cyclin D1 (Fig. 3a).

Figure 3. hMLH1 delays the exit from the G2/M arrest and, independently, enhances the tetraploid G1 arrest. (a) Markers of the G2/M arrest (P-Cdk1 and cyclin B1) at day 1–2 are stronger and they persist longer while the marker of the tetraploid G1 arrest on days 3–6 (cyclin D1 expression) appears later in clone 43 than in mock transfectants. (b) The difference between mock and clone 43 in the tetraploid G1 arrest at day 6 is retained also after the G2/M arrest was adjusted to the same level by addition of caffeine. Mean values of 3 independent experiments ± SD. p calculated in t-test.

Conversely, when hMLH1 was suppressed by siRNA, the exit from the G2/M arrest was accelerated, visible by decreased expression of cyclin B1 and increased expression of cyclin D1 at day 2 (Supporting Information Fig. S2).

The different tetraploid G1 arrest could be dependent on the extent of the preceding G2/M arrest in the cell lines. Since the G2/M arrest is dependent on the initial ATM signal, we adjusted the G2/M arrest to the same level by addition of the ATM inhibitor caffeine, at a concentration of 0.5 mM to mock and 2 mM to clone 43 one day before start of treatment. Two days after start of treatment, both caffeine and SN-38 were removed and the tetraploid arrest was followed in the regular, nonsupplemented medium. It was evident that the tetraploid G1 arrest is higher in clone 43 than in mock even when the starting G2/M arrest was the same (Fig. 3b), supporting the notion that the G2/M arrest and the tetraploid G1 arrest are regulated independently.

Thus, the intact hMLH1 protein enhanced the G2/M arrest, delayed the exit from G2/M by about 1 day and strongly enhanced the tetraploid G1 arrest. The long maintenance of the tetraploid G1 arrest in hMLH1 clones did not appear to be a mere result of the delayed exit from the G2/M arrest.

Differences in molecular response of mock transfectants and hMLH1-clones to DNA damage

The expression and the phosphorylation of proteins potentially controlling the triggering or maintenance of the tetraploid arrest were monitored by Western blotting for up to 6 days after start of SN-38 treatment in mock transfectants and in hMLH1-expressing clones. The time course and the extent of activation of p53 and of the downstream targets p21 and p27 were similar in mock- and the transfectant clones (Fig. 4a).

Figure 4. Time course of the alteration of markers of the G2/M arrest and of the tetraploid G1 arrest after SN-38 treatment. (a) Activation of the kinase Chk1 at day 2 or Chk2 at day 4 is higher, and the phosphorylation of the Rb protein (at day 4–6) is lower in clones 43 and 26 than in mock transfectants after SN-38 treatment. β-actin was detected to control equal protein loading. Representative for 4 experiments in either clone; (b) Rb protein is less phosphorylated at Ser807/811 at day 6 in clone 26 and in clone 43 than in the mock transfectant.

It has been reported previously that DNA damaging agents, including SN-38, activate both Chk1 and Chk2 kinases.23 Here we observe in addition, that both Chk1 and, to a lesser extent, Chk2 are more phosphorylated, that is, activated in hMLH1-expressing clones than in the mock transfectants. The MMR status-dependent stronger Chk1 activation on day 2 and of Chk2 activation on days 2–6 was observed after SN-38 treatment also in the cell line pair HCT116+chr2/HCT116+chr3 (Supporting Information Fig. S3). The Rb protein was less phosphorylated on day 6 in both clones than in the mock transfected cells. The difference of the overall phosphorylation of Rb (Fig. 4a) between mock and clones corresponded to the differences in the phosphorylation detected at Ser807/811 (Fig. 4b). It was also evident that the phosphorylation of histone H3, indicating more proliferation, was higher in the mock than in the hMLH1 transfectant clones 6 days after start of treatment.

In sum, the hMLH1-expressing clones showed 2 days after start of treatment a stronger activation of the G2/M regulators Chk1 and Chk2 and 6 days after start of treatment a stronger activation of the G1-S regulator Rb.

Chk1 phosphorylation enhances the G2/M arrest

To analyze the potential role of Chk1 and Chk2 activity in the G2/M and in the tetraploid G1 arrest, we followed in Western blot the time course of their phosphorylation and plotted the total amount of phosphorylated, that is, active kinase per cellular tubulin. Although the amount of activated/phosphorylated Chk1 exhibited a sharp peak extending only for the time of treatment (Figs. 4a and 5a), the amount of phosphorylated Chk2 started to increase at day 2 and reached the maximum on day 5 (Figs. 4a and 5b).

Figure 5. Stronger Chk1 activation enhances the G2/M arrest in clone 43. (a) Active Chk1 disappears at day 3, data representative of 3 experiments; (b) Active Chk2 reaches its peak at day 5, representative of 4 experiments; The densitometric values were always related to the value of the γ-tubulin band in each lane. Clone 26 gave similar results; (c) Suppression of Chk1 protein in clone 43 decreases the G2/M arrest: Values of 4 N DNA are means of 5 experiments ± SD, p = 0.005; (d) Lack of Chk2 in HCT116Chk2^−/− cells does not affect G2/M nor tetraploid G1 arrest, representative of 3 experiments; (e) Transient overexpression of Chk1 protein enhances the G2/M arrest, transient overexpression of Chk2 does not affect G2/M nor tetraploid G1 arrest. Mean values ± SD, determined by FACS only in the EGFP-positive cells in 3 experiments. p calculated in paired t-test.

The relationship between the activation of Chk1 and the G2/M arrest was further investigated after selective suppression of the Chk1 protein. Chk1 was visible only on the second day of treatment; at day 6 it was hardly detectable in Western blot (Figs. 4a and 5c). The suppression of Chk1 was, expectedly, concomitant with a consistent decrease of G2/M arrest (from 82 to 64%).

We used HCT116Chk2^−/− cells to directly investigate the potential role of Chk2 kinase in the tetraploid G1 arrest. The Chk2 knock-out cells showed a similar tetraploid time course and the extent of the tetraploid arrest as the parental HCT116 cells (Fig. 5d). This confirmed previously published data indicating that Chk2 is not involved in G2/M arrest regulation24 and showed further that it does not affect the tetraploid G1 arrest.

These data were supported by the overexpression of a functional Chk1 (Supporting Information Fig. S4), which showed an increase in the G2/M arrest at day 2 as compared to the transfectants with mutated Chk1 (Fig. 5e, Day 2 and Day 4 shown), while neither G2/M- nor tetraploid G1 arrest was affected by overexpression of Chk2 (Fig. 5e, Supporting Information Fig. S4).

The contribution of Chk1 activity to the DNA-damage-triggered G2/M arrest was reported previously. The present data reveal a stronger activation of Chk1 in MMR proficient cells and its causal effect on the enhanced G2/M arrest in these cells. By contrast, Chk2 activity did not affect G2/M- nor tetraploid G1 arrest in the present experimental setup.

The basal tetraploid arrest is due to CDK2 and CDK4 activity but the difference between mock transfectants and hMLH1 clones is neither due to CDK2 nor CDK4

To test if the cyclin-dependent kinases contribute to the tetraploid G1 arrest in the present cell system, Cdk2, Cdk4 or Cdk6 were suppressed in mock transfectants with the corresponding siRNA; on the next day the cells were treated with SN-38 (2 days, 10 nM), and the tetraploid arrest was monitored on day 6 after start of treatment (Fig. 6).

Figure 6. The suppression of Cdk2 or of Cdk4 increases the tetraploid G1 arrest induced by SN-38 treatment. The suppression of Cdk2 or of Cdk4 in mock transfectants increases the tetraploid G1 arrest at day 6. Representative for 3 experiments. The suppression of Cdk6 had no detectable effect on the tetraploid arrest (not shown).

The suppression of either Cdk2 or Cdk4 increased the tetraploid G1 arrest 6 days after start of treatment by 14–19% (Fig. 6), while the suppression of Cdk6 had no effect on tetraploid G1 arrest and was not further investigated (data not shown).

We tested if the activity of any of the 2 kinases Cdk2 and Cdk4 was different between hMLH1-transfectants and mock-transfectans: Cdk2 or Cdk4 were immunoprecipitated and the ability of each kinase to phosphorylate the Rb protein in vitro at Ser807/811 was assayed by using the recombinant Rb protein fragment as the substrate (Supporting Information Figs. S5A–D). As expected, the specific activity, expressed as the ratio of the Rb protein phosphorylation on Ser 807/811 to precipitated Cdk2 or Cdk4, dramatically decreased after SN-38 treatment. This was explained by the higher content of p21 in the precipitates after SN-38 treatment (Supporting Information Fig. S5E). The composition of the Cdk4 complexes (Supporting Information Fig. S5E) and of the Cdk2 complexes (not shown) in mock and clones (clone 26 shown) and the specific activity of the kinases (Supporting Information Figs. S5C,D) was, however, similar.

These data indicated that while the basal tetraploid arrest is likely to be due to Cdk2 and Cdk4 activity, the difference in tetraploid G1 arrest between the mock and hMLH1-clones is due to other factors.

Long tetraploid G1 arrest is associated with a lesser clonogenic survival

Both hMLH1-transfected clones exhibited a higher sensitivity to SN-38 in the clonogenic survival assay than the mock clone (Fig. 7a). Conversely, when the hMLH1 signal was reconstituted in the mock transfectants by transduction with AdV-hMLH1 (1.5–6 MOI), the clonogenic survival of the transduced cells decreased in a dose-dependent fashion (Fig. 7b).

Figure 7. hMLH1 expression decreases the clonogenic survival after SN-38 treatment. (a) Clonogenic survival of mock transfectants and clones 26 and 43 after treatment with SN-28 for 2 days; (b) The clonogenic survival of mock transfectants decreases after transduction with 1.5 or 6 MOI of AdV-hMLH1; (c) In nontreated cells of clone 43, 25% are in the tetraploid state, the treatment (10 nM SN-38, 48 hr) leads to 74% tetraploid arrest in clone 43 at day 4; the selectants react to treatment with a lower tetraploid arrest of 43% and a better clonogenic survival (Supporting Information Fig. S6); (d) Four days after start of treatment, the selectants express less hMLH1 than the original clone 43.

We observed lower clonogenic survival of MMR-proficient than of MMR-deficient cell lines after SN-38 treatment also in the cell line pair A2780/A2780cp70 (not shown) and also, in agreement with previous reports,3, 4 in the semi-isogenic pairHCT116+chr2/HCT116+chr3 (Supporting Information Fig. S1B).

Neither the suppression of hMLH1 by pSuper-hMLH1 nor siRNA-hMLH1 was, however, sufficient to increase the clonogenic survival (not shown), although a small but significant difference in the tetraploid G1 arrest (Fig. 2d) and the different duration of G2/M arrest (Fig. 3a) were evident. Higher amounts of pSuper-hMLH1 were detrimental to the cells and were not used (data not shown).

If the clonogenic survival is related to the long-term tetraploid arrest, then the survivors should exhibit a diminished ability to arrest in tetraploid G1. To test this hypothesis, we investigated the properties of the colonies of clone 43 surviving 3 consecutive clonogenic assays (lasting 12–15 days each) after SN-38 treatment: the ability of these SN-38 resistant selectants to carry out the long-term tetraploid arrest at day 4 after start of SN-38 treatment diminished from 74 to 43% (Figs. 7c,d) and their clonogenic survival after treatment with SN-38 was better than that of the parental cells (Supporting Information Fig. S6). The selectants also expressed less hMLH1 after treatment than the original clone 43 (Fig. 7d). The lesser tendency of the selectants to arrest as single cells and their greater propensity to form colonies was evident upon microscopic inspection of the dishes (Supporting Information Fig. S6).

These data confirmed the association between the extended tetraploid G1 arrest and the lower survival of hMLH1-expressing cells.

UCN-01 abrogates the tetraploid G1 arrest and increases the clonogenic survival of hMLH1-expressing cells

UCN-01 is a broad-range kinase inhibitor, which was recently shown to inhibit Chk1, Chk2 as well as a number of other essential kinases regulating the cell cycle.25

We observed that simultaneous 2-day treatment of cells with SN-38 and UCN-01 led not only to a partial abrogation of the G2/M arrest at day 2 as reported previously,26, 27 but also to a significant decrease of the percentage of 4 N DNA in hMLH1 transfectants at day 6 (by 23% at 125 nM, Fig. 8a, and by 34% at 500 nM, not shown). The decrease of the tetraploid G1 arrest was concomitant with an improved survival in clonogenic assay (Fig. 8b).

Figure 8. Abrogation of the long-term tetraploid G1 arrest is associated with a higher clonogenic survival after SN-38 treatment. (a) Simultaneous treatment with SN-38 (10 nm, 48 h) and with UCN-01 (125 nM, 48 h) leads to a decrease of the G2/M arrest (in mock and clone 43) and of the tetraploid G1 arrest (in clone 43). Mean values ± SD of 3 experiments. Bar indicates the duration of treatment; (b) Improved clonogenic survival of clone 43 but not of mock transfectants after SN-38 treatment (10 nM for 2 days) in the presence of 125 nM UCN-01; Mean values ± SD of 3 experiments. (c) Colonies of mock transfectants and of (d) Clone 43 and of (e) Clone 26 after treatment with SN-38; (f) The addition of 500 nM UCN-01 completely eliminates the long-term arrested single cells in clones 26 and 43 (clone 43 shown here). Bar = 1 mm. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

By contrast, the effect on the tetraploid G1 arrest and on survival of mock transfectants was minor (Figs. 8a and 8b). The inspection of the colonies under the microscope revealed in the transfectants 26 and 43, but not in the mock transfectants (Fig. 8c), numerous single cells, arrested after SN-38 treatment (Figs. 8d and 8e). These single cells did not exhibit any signs of apotosis. Upon addition of UCN-01 no single cells were detectable (Fig. 8f), but only colonies, whose relative number was higher than in the group without UCN-01 (Fig. 8b). Similar effect of simultaneous treatment with SN-38 and UCN-01 on tetraploid G1 arrest and an improved clonogenic survival were observed in HCT116+chr3 cells (Supporting Information Fig. S7).

Collectively, these data show that UCN-01 suppresses the tetraploid G1 arrest in hMLH1-expressing cells treated with SN-38. In combination with the data presented in Figure 2, they support the notion that the extension of the tetraploid G1 arrest is the main cause of the decreased survival of MMR proficient cells.

Discussion

Several authors reported that MMR-proficient but not the -deficient cells treated with MMR activating agents like MNNG, MNU or 6-TG undergo a G2/M arrest and then apoptosis. These consistent observations explained the poor survival of MMR proficient cells after treatment with MMR activating agents. There is, however, a controversy about the role of MMR in the response to DNA double strand breaks induced by irradiation or by chemotherapeutic agents.

To circumvent the ambiguities associated with the hitherto used semi-isogenic systems, in the present work we generated genetically matched hMLH1⁺/hMLH1⁻ colon carcinoma cells by stable transfection of the HCT116 cell line with hMLH1-expressing plasmid and subsequent cloning. In 2 hMLH1-expressig clones 26 and 43, whose behaviour corresponded to that of cells with a functioning MMR system, we investigated the role of MMR status in the response to DNA double strand breaks.

In this isogenic cell system, we show that the DNA-damage caused by SN-38 treatment leads to a stronger G2/M arrest and a longer tetraploid G1 arrest (Fig. 2) in the presence of the intact MMR, clearly indicating that it modulates the reaction to SN-38-inflicted damage. The stronger tetraploid G1 arrest was retained in the MMR proficient cells also when the G2/M arrest was equalized, suggesting an independent regulation of each process (Fig. 3b).

SN-38 treatment has previously been shown to activate Chk1 and, to a lesser extent, Chk2.23 We show for the first time, that the activation of both kinases after double strand breaks is enhanced by intact MMR system. The MMR-dependent enhancement of kinase activation was also observed in the HCT116+chr2/HCT116+chr3 pair (Supporting Information Fig. S3), suggesting that this was a general phenomenon and not a peculiarity of the clones.

The stronger G2/M arrest at day 2 after start of treatment was dependent on the stronger Chk1 activation. This represents the mechanistic link between MMR and the enhanced G2/M arrest induced by double strand breaks.

Although the stronger tetraploid G1 arrest coincided with a stronger Chk2 activation, the Chk2-suppression (not shown) and -overexpression (Fig. 5e) experiments as well as the time course of the arrest of the HCT116Chk2 knock-out cell line (Fig. 5d) did not support the hypothesis that Chk2 affects the tetraploid G1 arrest.

By contrast, the suppression of Cdk2 and Cdk4 clearly showed that the basal tetraploid arrest was regulated by both kinases (Fig. 6). The immunoprecipitation of Cdk4 or Cdk2 revealed that the amounts of cyclin D1 as well as of p21 proteins bound to the kinases were similar in mock- and hMLH1-transfectants (Supporting Information Fig. S5E and not shown). Furthermore, the capacity of both kinases to phosphorylate Rb in vitro on Ser807/811 (Supporting Information Figs. S5A,B) was decreased at day 6 after start of treatment to the same extent (Supporting Information Figs. S5C,D). These data clearly indicate that the observed difference in phosphorylation of cellular Rb at day 6 (Fig. 4b) must be due to another, hMLH1-dependent, kinase.

Together, our data support the notion17 that p21 is essential for the tetraploid G1 arrest. They do not explain the origin of the difference in tetraploid G1 arrest (Fig. 2a) nor in Rb phosphorylation at day 6 (Fig. 4b) between MMR-proficient and -deficient cells.

We have recently demonstrated in different established colon carcinoma cell lines that in p53^wt cells the low clonogenic survival after SN-38 treatment is associated with a long-term cell arrest.13 In the present work, we show that the MMR-dependent stronger G2/M arrest and a longer tetraploid G1 arrest (Fig. 2a) are associated with a lower clonogenic survival (Fig. 7a).

G2/M arrest has been previously shown to have little effect on clonogenic survival,9 an observation consistent with the present data. Indeed, the SN-38 resistant selectants obtained after 3 rounds of SN-38 treatment exhibited a better clonogenic survival (Supporting Information Fig. S6) and 31% less tetraploid G1 arrest (Fig. 7c), but only 14% less G2/M arrest (not shown). They also showed less single arrested cells (Supporting Information Figs. S6B,C).

By contrast, the abrogation of the SN-38 induced tetraploid G1 arrest by simultaneous treatment with UCN-01 (Fig. 8a) increased the clonogenic survival (Fig. 8b). The long-term arrested cells evident in the clonogenic assay of hMLH1-expressing clones (Figs. 8d and 8e) were virtually absent in the mock clone (Fig. 8c) and in the clones simultaneously treated with UCN-01 (Fig. 8f). The numerous potential molecular targets that could be responsible for the observed UCN-01 effect25 have not been investigated; Chk2 could be excluded as a target since the extent of tetraploid G1 arrest in HCT116Chk2 knock-out cells was the same as that in HCT116 cells (Fig. 5d).

It is of importance that in HT-29 cells (p53^mut, MMR⁺), UCN-01 treatment after DNA damage decreases cell survival (data not shown and Ref. 28). Similarly, the suppression of G2/M checkpoint by Chk1 suppression strongly diminished clonogenic survival of SN-38 treated HeLa cells,23 which behave like a p53^mut cell line and after treatment undergo mitotic catastrophe and cell death.

Thus in the present experimental setup, simultaneous addition of UCN-01 increased clonogenic cell survival of p53^wt cells, which respond to SN-38 with an indefinite arrest and not with apoptosis. By contrast, in p53^mut cells, which usually respond to SN-38 with a short G2/M arrest followed by a mitotic catastrophe, the abrogation of the (protective) G2/M arrest is expected to promote cell death.

The present data establish a connection between MMR and the response to SN-38-induced DNA damage and point to a new function of the MMR system, not related to the MMR per se. Our data show that lesser survival of MMR-proficient than MMR-deficient cells is due to a higher and more sustained tetraploid G1 cell arrest in the proficient cells after SN-38 treatment. Thus the role of MMR in the response to MMR activating agents is qualitatively different from that in the response to double strand breaks: in both cases, however, the survival of the MMR proficient cells is diminished (Fig. 9).

Figure 9. p53^wt, MMR-proficient cells respond to mismatch-inducing agents with mismatch repair and apotosis. They respond to DNA double strand breaks with an enhancement of the tetraploid G1 arrest.

The hMLH1-expressing clones used in this study as well as the semi-isogenic cell pair HCT116+chr2/HCT116+chr3 originated from the cell line HCT116. The observed alterations could thus be due to a unique property of the parental cells, for example, to a defect of homologous recombination repair (HRR). HRR, however, would not be reconstituted by the transfection of hMLH1.8 Furthermore, the results obtained after AdV-hMLH1 transduction as well as the difference in survival seen in cell pairs A2780/A2780cp70 and LoVo/LoVo+chr24 argue against this hypothesis. Clearly, more data are necessary to establish the generality and the mechanistic connection between the MMR function and the enhancement of the tetraploid G1 arrest after DNA damage.

The present data indicate that an analysis of the role of MMR in patients undergoing chemotherapy will be more valid within a group that is homogeneous in respect to the p53 status. They further reveal that it is essential to know the type of the individual response to DNA damage in the individual patient: if the response of the tumor is due mainly to the long-term arrest, the application of modulators abrogating this arrest during the therapy may be counterproductive.

Acknowledgements

The authors thank Mr. Donnie Owens, Pfizer Inc. for a kind gift of SN-38, Dr. Helen Piwnica-Worms, Dr. Giacomo Buscemi and Dr. Jörg Trojan for the donation of plasmids and Dr. Bert Vogelstein for HCT116 knock-out cell lines and Dr. Boland for HCT116+chr3 cell line.