Main

Minor groove binders (MGBs) represent an interesting class of anticancer agents, which have been shown to be highly effective in in vitro and in vivo preclinical tumour models unresponsive to other antineoplastic agents (Martin et al, 1981; Li et al, 1982, 1992; Hartley et al, 1988; D'Alessio et al, 1994; D'Incalci, 1994; Colella et al, 1999; Marchini et al, 1999; Geroni et al, 2002). The main representatives of this class, which reached the clinic, are the antitumour agents derived from CC-1065, that is, adozelesin, carzelesin, and bizelesin, and the distamycin A derivative tallimustine. These ‘classical’ MGBs have been shown to be highly DNA sequence-specific (Lee et al, 1993; D'Incalci, 1994) and to exert their cytotoxic effect through the ability to per se directly alkylate DNA mainly at the N3 position of adenines exposed in (TA)-rich sequences in the DNA minor groove (Hurley et al, 1984; Reynolds et al, 1985; Broggini et al, 1995, 1991; Sun and Hurley, 1992; D'Incalci, 1994; Marchini et al, 1998), without the requirement to be activated by other pathways (e.g., enzymatic activation of the drug). The absence of significant antitumour activity for nonalkylating MGBs (Marchini et al, 1998) indicates that the N3 alkylation activity of these compounds is a prerequisite for their cytotoxicity. MGBs activity, however, has previously been reported (Colella et al, 1999) to be associated with reduced susceptibility to the cytotoxic effect in tumour cells with defects in DNA mismatch repair (MMR), similar to certain chemicals, including MNNG, which alkylates O6 of guanines, and anticancer agents such as doxorubicin and cisplatin (Branch et al, 1995; Drummond et al, 1996).

MMR proteins recognise mismatched base pairs in the DNA, arising either spontaneously during DNA metabolism or from modified nucleotides provoked by physical and chemical agents, and are thought to link DNA damage recognition to an apoptotic pathway, thereby preventing mutagenesis, tumorigenesis, and tumour progression (Modrich, 1991; Fink et al, 1998). Tumours resulting from MMR-deficiency include the hereditary nonpolyposis colon cancer (HNPCC) and some sporadic carcinomas such as mammary, ovarian, or endometrial cancers (Peltomaki, 2001). The development of novel MGBs able to overcome the involvement of MMR assumes great clinical importance with respect to the treatment of tumours deficient in MMR. A novel α-bromoacryloyl derivative of distamycin A, PNU-151807, which exhibits no alkylating activity per se, has been identified (Marchini et al, 1999). The cytotoxic effect has been shown to not depend on MLH1 in some tumour cells (Colella et al, 1999) and has been attributed to the α-bromoacrylic moiety of the compound, which seems to interfere with cell cycle progression via yet unknown pathways (Cozzi, 2000; Geroni et al, 2002).

Recently, brostallicin (PNU-166196), a synthetic α-bromoacrylic MGB structurally related to PNU-151807, has been selected for clinical development. Brostallicin has shown very promising activity in experimental tumour models; its in vitro and in vivo activity is increased in tumour cells with higher glutathione (GSH) and/or glutathione-S-transferase (GST) levels (Geroni et al, 2002). The α-bromoacrylic moiety of brostallicin was found to react with GSH, in a reaction catalysed by GST, with the possible formation of a highly reactive GSH-complex able to bind covalently to DNA (Geroni et al, 2002; Cozzi, 2003).

The present study was aimed at investigating the effect of loss of MMR on the sensitivity to brostallicin compared to the structurally related tallimustine, using cell lines deficient or proficient in MLH1, MSH2, or PMS2, respectively. A putative involvement of two members of the PI3-like kinase family, ATM and DNA-PK, which link DNA damage and cell cycle response in drug-induced cytotoxicity, was also investigated. We report that MMR-deficient cells retain sensitivity to brostallicin, thereby extending the list of potential anticancer agents for use in the treatment of MMR-deficient tumours, and that brostallicin-induced cytotoxicity may not require ATM and DNA-PK.

Materials and methods

Cell lines

The MLH1-deficient human colorectal adenocarcinoma cell line HCT116 was obtained from the American Type Culture Collection (ATCC CCL 247), and a subline complemented with chromosome 3 carrying the wild-type gene for hMLH1 (clone HCT116/3–6, identified here as HCT116+ch3) was obtained from Dr M Koi (Koi et al, 1994) as were the MSH2-deficient human endometrial adenocarcinoma cell line HEC59 (Umar et al, 1997) and a subline complemented with chromosome 2 carrying the wild-type gene for hMSH2 (clone HEC59/2–4, identified here as HEC59+ch2). HCT116 cells contain a hemizygous mutation in MLH1 resulting in a truncated, nonfunctional protein (Boyer et al, 1995). Similarly, the HEC59 cells are mutated at different loci on both alleles of MSH2 and are deficient in repair activity (Umar et al, 1997). The chromosome-complemented sublines HCT116+ch3 and HEC59+ch2 are competent in DNA MMR. HCT116 and HEC59 cell lines were maintained in Iscove's modified Dulbecco's medium (Life Technologies, Basel, Switzerland) supplemented with 2 mM L-glutamine and 10% heat-inactivated foetal bovine serum (Oxoid, Basel, Switzerland). The chromosome-complemented lines were maintained in a medium supplemented with geneticin (400 μg ml−1 for HCT116+ch3, and 600 μg ml−1 for HEC59+ch2) (Life Technologies). Although the extent of possible effects of the introduction of an extra chromosome are not fully clear, it is generally acknowledged that it does not spoil the effects of loss of MMR on drug sensitivity. PMS2−/−/p53−/− and PMS2+/+/p53−/− cell lines, established from E1A/Ha-Ras-transformed knockout mice primary fibroblasts, were generously provided by Dr P Glazer (Zeng et al, 2000). Cells are maintained in culture for a limited number of passages and are routinely tested for the expression of MMR proteins. The ATM+/+/p53−/− and ATM−/−/p53−/−, and the DNA-PK+/+/p53−/− and DNA-PK−/−/p53−/− mouse embryonic fibroblasts, were generously provided by Dr P Leder (Westphal et al, 1997) and Dr EH Goodwin (Bailey et al, 1999), respectively. The cells were maintained in DMEM medium supplemented with 2 mM L-glutamine (Life Technologies), 10% heat-inactivated foetal calf serum (Oxoid) and penicillin/streptomycin (100 U ml−1/100 μg ml−1, Life Technologies). All cell lines were tested negative for contamination with Mycoplasma spp. and maintained in a controlled environment of 5% CO2 and 95% relative humidity at 37°C. Except for the ATM+/+/p53−/− and ATM−/−/p53−/− mouse cells, which grow as a monolayer and do not form colonies, all other cell lines used in these experiments form well-defined individual colonies when seeded sparsely on standard tissue culture plates.

Reagents

Distamycin A and its derivatives brostallicin (PNU-166196) and tallimustine (PNU-152241) were synthesised by Pharmacia Italy (Nerviano, Italy). The chemical structures of the derivatives are presented in Figure 1. Brostallicin was dissolved in methanol, tallimustine in DMSO, and distamycin A in water. Stock solutions were stored at −20°C. The final concentration of DMSO or methanol in the cultures was <0.1% at all drug concentrations and in controls. Previous experiments (data not shown) have shown that neither 0.1% DMSO nor 0.1% methanol affects the viability or growth of these cell lines.

Figure 1
figure 1

Chemical structures of brostallicin and tallimustine. Both molecules share the distamycin A backbone.

Full size image

MPE footprinting analysis

The MPE footprinting method has been previously described in detail (Hertzberg and Dervan, 1984). The 4492- and 751-bp fragments of SV40-labelled plasmid previously described (Marchini et al, 1999) were incubated with distamycin A, tallimustine, and brostalicin (50 μ M) for 1 h at room temperature and treated for 30 min at room temperature with a solution of MPE-(NH4)2-Fe(SO4)2 (synthesised by Pharmacia, Italy, according to the published method; Hertzberg and Dervan, 1984). After precipitation, DNA was resuspended in loading buffer and electrophoresed on an 8% polyacrylamide 7 M urea gel and autoradiographed.

Taq polymerase stop assay

Studies with the Taq stop assay were based on a previously reported method (Ponti et al, 1991). Prior to drug-DNA incubation, plasmid pBSSK-TOPO II was linearised with a PstI restriction enzyme (NEB) to provide a stop for the Taq polymerase, downstream from the primer. After drug treatment, the DNA was precipitated and washed as described (Ponti et al, 1991). The primer was 5′-end labelled with T4 polynucleotide kinase (NEB) and [γ-32P] ATP (5000 Ci mmol−1, Amersham). The synthetic primer sequence and the linear PCR amplification conditions were performed as described (Marchini et al, 1998). Samples were then purified by extraction with an equal volume of chloroform–isoamyl alcohol (24 : 1), and then precipitated and washed following the standard protocol. Dried samples were resuspended in loading buffer and denatured at 90°C for 2 min before loading onto an 8% polyacrylamide denatured gel. After the run, the gel was dried and autoradiographed.

Clonogenic survival and MTT proliferation assays

Clonogenic survival in response to drug treatment was performed by plating 250 cells in 60 mm cell culture dishes. After 24 h, the drug was added, followed by incubation in a drug-containing medium for 2 h or 24 h and then in a drug-free medium for another 6–8 days at 37°C in a humidified atmosphere containing 5% carbon dioxide. Cells were then fixed with 25% acetic acid in ethanol and stained with Giemsa. Colonies of at least 50 cells were scored visually. Each experiment was performed a minimum of three times using triplicate cultures for each drug concentration. The logarithm of relative colony formation was plotted against the concentration of the drug. The IC50 was estimated by linear interpolation of the logarithmic transformed relative plating efficiencies.

For ATM+/+/p53−/− and ATM−/−/p53−/− mouse cells that do not form distinct colonies, the drug sensitivity was determined by the MTT assay (Mosmann, 1983). MTT (3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazoliumbromide) measures the mitochondrial dehydrogenase of surviving cells. Cells growing in the log phase were harvested by brief trypsinisation. A total of 1000 cells were plated (96 well plates) 24 h prior to 2 h drug treatment. Cells were then grown in a drug-free medium for another 4 days at 37°C in a humidified atmosphere containing 5% carbon dioxide. A volume of 20 μl MTT in PBS to a final concentration of 0.5 mg ml−1 was added, followed by incubation at 37°C for 4 h, aspiration of the medium, and addition of 200 μl DMSO. The optical density was measured by the Emax microplate reader E9336 (Molecular Devices, Clearwater, MN, USA) at 540 nm, setting the value of the cell lines in the medium to 1.0 (control) and the value of the no cells blank to zero. Differences in drug sensitivity of the respective cell lines were determined from at least four independent experiments and are reported as the concentration required to suppress proliferation by 50% (IC50).

Statistical analysis

The mean±s.d. values were calculated for all data sets. The two-sided paired t-test was used to compare the effects on drug sensitivity. P<0.05 was considered to be statistically significant.

Results

Brostallicin does not alkylate DNA per se but through the interaction with GSH/GST

Noncovalent interactions of brostallicin and tallimustine (TAM) with DNA were compared to those of distamycin A (DISTA). The data reported in Figure 2 show an autoradiograph of a classical ladder of an MPE-footprinting experiment tested on the 751-bp (panel A) and 4492-bp fragments (panel B) of the SV40 DNA plasmid. Each band corresponds to a DNA fragment differing in size by a single nucleotide. When a DNA region is protected by the presence of a molecule, chemical digestion is blocked and a ‘gap’ is present on the autoradiograph. In the control (CTR, Figure 2), all fragments are present with broadly the same signal intensity on the gel, while in all the treated sample lanes a typical ‘gap’ is common in AT-rich regions. The brackets highlight these regions. The distamycin A backbone present in the brostallicin chemical structure drives the DNA interaction towards AT-rich regions in the same way as previously shown for tallimustine. In fact, brostallicin shows a noncovalent DNA interaction effect superimposable to that of tallimustine and distamycin A (internal positive control). These regions are highlighted by brackets. The differences in band intensities were due to differences in gel loading.

Figure 2
figure 2

MPE footprinting analysis of the SV40 DNA plasmid (751-bp, panel A; and 4492-bp, panel B) treated with distamycin A (DISTA), tallimustine (TAM), and brostallicin as described in Materials and Methods section. CTR=untreated control DNA. Brackets on the right-hand side of the picture refer to AT-rich regions determined by DNA sequencing.

Full size image

On the basis of the previously reported data showing that brostallicin is able to covalently interact with DNA upon in vitro reduction by the GSH/GST system (Geroni et al, 2002), we further tested this hypothesis by incubating the drug-DNA solution with and without GSH and GST in an in vitro system. The sequence-specific, covalent DNA interaction of brostallicin in comparison with tallimustine was analysed by the Taq polymerase stop assay. This assay is a linear amplification method employing the properties of DNA polymerase to investigate the sequence selectivity of the interaction between DNA-damaging agents and the DNA. As expected, tallimustine retained its high sequence specificity in alkylating DNA at the target hexamer (5′-TTTTGA), while brostallicin per se was completely unable to produce any alkylation in the selected DNA region (Figure 3). Brostallicin alone did not alkylate DNA, while a band was observed when GST/GSH was added to the reaction mixture. In the absence of brostallicin, GSH/GST did not induce any alteration able to block Taq polymerase. It is important to underline the fact that, although the interaction of brostallicin with DNA involves AT-rich regions, the compound binds to a sequence (AAAG) different from those previously reported for tallimustine. Studies are still in progress to better define the sequence of the alkylated regions.

Figure 3
figure 3

Autoradiography of a typical Taq Stop Assay on topoisomerase II cDNA after treatment with brostallicin and tallimustine (TAM). The experiment was performed as described in Materials and Methods section. CTR=control untreated sample. Arrows indicate the alkylated regions.

Full size image

Loss of MLH1 or MSH2 does not alter sensitivity to brostallicin

The question was addressed as to whether loss of either MLH1 or MSH2 affects the sensitivity to brostallicin using the clonogenic assay. The data presented in Table 1 show that MLH1-deficient HCT116 cells are nearly as sensitive as MLH1-proficient HCT116+ch3 cells to this drug (P=0.36). This indicates that MLH1 is not involved in brostallicin-mediated cytotoxicity. Furthermore, MSH2-deficient HEC59 cells are as sensitive to brostallicin as MSH2-proficient HEC59+ch2 cells (P=0.41), indicating that brostallicin-mediated cytotoxicity does not require functional MSH2. Brostallicin cytotoxiciy has been compared to tallimustine. The results show that MLH1-deficient and MSH2-deficient cells are three-fold (P<0.01) and 1.8-fold (P=0.03), respectively, less sensitive to tallimustine than their respective proficient counterparts.

Table 1 IC50 concentrations for clonogenic survival of MMR-proficient or -deficient cells in response to treatment with brostallicin or tallimustine
Full size table

Sensitivity to brostallicin, but not to tallimustine, is retained after loss of PMS2

Although less frequently mutated than MLH1 or MSH2 in human cancers, PMS2 may nevertheless be relevant in this respect since it forms a heterodimer with MLH1 and the lack of one or the other partner affects MMR activity. Based on the model that cytotoxicity of tallimustine, but not the α-bromoacrylic derivatives, is dependent on functional MMR, it is anticipated that loss of PMS2 negatively affects sensitivity to tallimustine, but not to brostallicin. The effect of loss of PMS2 on drug sensitivity was investigated in p53-deficient cells derived from knockout mice. Table 2 shows that the clonogenic survival after treatment with brostallicin in PMS2-deficient cells was not different from that in PMS2-proficient cells (P=0.79). In contrast, PMS2-deficient cells were 1.6-fold less sensitive to tallimustine than PMS2-proficient cells (P=0.02).

Table 2 IC50 concentrations for clonogenic survival of PMS2-proficient or -deficient mouse cells in a p53-deficient genetic setting in response to drug treatment
Full size table

Thus, PMS2-deficient p53-null mouse fibroblasts retain sensitivity to brostallicin. The 1.6-fold resistance to tallimustine in PMS2-deficient cells indicates a role for PMS2 in sensitivity to this compound.

Loss of ATM or DNA-PK does not affect sensitivity to brostallicin

It has previously been proposed that the cytotoxic effect of the α-bromoacrylic derivative PNU-151807 interferes with the cell cycle checkpoint control (Marchini et al, 1999). Although yet unknown, a possible pathway may include ATM or DNA-PK, members of the PI3-like kinase family, which are important kinases for connecting DNA damage monitoring and cellular responses such as cell cycle checkpoint activation and apoptosis. The question was addressed as to whether the sensitivity to brostallicin is affected by loss of ATM or DNA-PK in a p53-deficient genetic background. We used embryonic fibroblasts from knockout mice. The data presented in Figure 4 show that ATM-deficient cells (0.8±0.3 μ M) were as sensitive to brostallicin as ATM-proficient cells (0.9±0.2 μ M) in a p53-deficient genetic setting (P=0.60). Likewise, DNA-PK-deficient cells (17.5±0.7 nM) were as sensitive to brostallicin as DNA-PK-proficient cells (21.0±1.4 nM) in a p53-null background (P=0.13).

Figure 4
figure 4

Sensitivity to brostallicin of ATM−/−/p53−/− (□) and ATM+/+/p53−/− (▪) mouse cells determined by the MTT assay (A) and of DNA-PK−/−/p53−/− (○) and DNA-PK+/+/p53−/− (•) mouse cells determined by the clonogenic assay (B), expressed as the percentage of the untreated controls. Each point represents the mean±s.d. of at least four independent experiments.

Full size image

Thus, neither ATM nor DNA-PK seems to be involved in the sensitivity to brostallicin in p53-deficient mouse cells.

Discussion

The present study demonstrates that brostallicin, a novel α-bromoacrylic, second-generation DNA MGB structurally related to distamycin A, maintains its cytotoxic effect in cells deficient for the MMR proteins MLH1, MSH2, or PMS2. The data permit drawing several conclusions. First, brostallicin, the lead compound of a novel class of MGBs in clinical trials, exerts its cytotoxic effect regardless of the MMR status, suggesting that further clinical testing of brostallicin in tumours deficient in MMR is to be recommended. Second, brostallicin-induced cytotoxicity can occur in the absence of functional ATM or DNA-PK in p53-deficient cells, indicating that brostallicin-induced cytotoxicity in this setting is independent of PI3-like kinase family status. Third, brostallicin is the first MGB unable to per se covalently interact with DNA. It requires the GSH/GST system to alkylate DNA with a sequence specificity different from that reported for previously tested alkylating molecules.

MMR plays an important role in the correction of spontaneously occurring errors during DNA processing that have escaped the DNA polymerase proof-reading activity, thereby preserving the integrity of the genome by preventing the occurrence of gene mutations and tumorigenesis (Modrich, 1991). Spontaneous tumours arising from MMR deficiency include the hereditary nonpolyposis colon cancer as well as some sporadic carcinomas such as mammary, ovarian, or endometrial cancers (Peltomaki, 2001). MMR monitors specific types of DNA damage introduced by DNA-damaging agents, and subsequently triggers an apoptotic response (Fink et al, 1998). Loss of MMR hence results in resistance to a variety of widely used anticancer drugs, including the topoisomerase I poisons camptothecin and topotecan, the topoisomerase II poisons doxorubicin, epirubicin, mitoxantrone and etoposide, and some platinum compounds such as cisplatin and carboplatin, as well as some alkylating agents including MNNG and busulphan (Branch et al, 1995; Drummond et al, 1996; Fink et al, 1998, 1996; Fedier et al, 2001).

Interestingly, the MMR status also affects the activity of several MGBs such as CC-1065 analogues and the distamycin-derivative tallimustine, but not that of the α-bromoacryoyl derivative of distamycin A (PNU-151807) (Colella et al, 1999). The present study expands on this previous finding by demonstrating that brostallicin, a novel second-generation DNA MGB structurally related to PNU-151807, exerts its cytotoxic effect regardless of the MMR status. Brostallicin as well as the class of the taxanes (Fedier et al, 2001) and photodynamic therapy (Schwarz et al, 2002) may thus represent valuable options for the treatment of tumours disabled in MMR.

The α-bromoacryoyl moiety has been proposed to be important since it reacts with GSH, and reactive drug-GSH intermediates may then modify the DNA by mechanisms not yet fully understood (Geroni et al, 2002; Cozzi, 2003). DNA interaction data reported in the present study suggest that the distamycin A backbone drives the brostallicin molecule towards the AT regions present in the minor groove of the DNA. In addition, brostallicin binds covalently to DNA through interaction with the GSH/GST system. Brostallicin covalently binds to DNA with a completely different sequence specificity than tallimustine. One hypothesis for the different behaviour of brostallicin against MMR status is that this covalent interaction is not substrate for MMR, whereas the alkylated DNA by tallimustine is recognised by MMR. It should be noted that no direct interaction between MMR and the GSH/GST system is known, and that the GSH/GST status of the cell lines under study does not matter for the experiments because the cell lines are quasi-isogenic, that is, they differ only in their MMR status and the extra chromosomes.

Moreover, as reported for PNU-151807, the bromoacryloyl moiety seems to be relevant for cell cycle checkpoint control (Marchini et al, 1999). The identity of mediators for signalling between DNA damage and downstream effectors is not clear. One possibility is that the DNA damage is recognised by one or several members of BASC (BRCA1-associated genome surveillance complex), a multiprotein complex including BRCA1, ATM, MMR proteins, and other proteins implicated in DNA repair (Wang et al, 2000). Our data, however, show that deficiency in ATM or DNA-PK did not affect brostallicin sensitivity in p53-deficient cells, arguing against a role of these kinases in these cells. Since these kinases are activated upon DNA double-strand breaks introduced by radiation or radiomimetic drugs (Jackson, 1997; Smith et al, 1999), α-bromoacryoyl derivatives seem unlikely to produce this type of lesion. Although the cytotoxic effect of tallimustine and PNU-151807 has been shown not to be dependent on the p53 status (Marchini et al, 1999), the data for these kinases obtained in p53-deficient cells may not be conclusive for p53-proficient cells. There is an apparent higher sensitivity to brostallicin of the DNA-PK data set compared to the ATM data set, but this is likely due to the use of two assays that differ in their sensitivities.

Mutations in the p53 tumour suppressor gene are found in a large fraction of human cancers (Hollstein et al, 1991) and this may be the genetic basis underlying failure to respond to chemotherapy (Ferreira et al, 1999). PNU-151807 has recently been reported to retain sensitivity against cells disabled in p53 function (Marchini et al, 1999), indicating that PNU-151807-mediated cytotoxicity does not require functional p53. We have recently shown that additional loss of PMS2 in p53-deficient cells increases cytotoxicity to a variety of anticancer agents (Fedier et al, 2002). This hypersensitising effect, however, was not observed in response to treatment with brostallicin. For tallimustine, even an opposite effect was observed in PMS2-deficient cells, suggesting that tallimustine-induced DNA damage is a substrate for MMR in p53-deficient cells. Consistent with this, tallimustine-induced DNA damage has already been shown to be a substrate for MMR in p53-proficient cells (Colella et al, 1999).

We also observed that tallimustine is less toxic than brostallicin in p53-deficient cells and that this effect is much greater than the difference in sensitivity to tallimustine between MMR-deficient and -proficient cells. This marked effect was not observed in p53-proficient cells. As the status of p53 has been reported not to markedly affect the sensitivity of human tumour cells to either tallimustine or PNU-151807 (Marchini et al, 1998), this effect in p53-deficient cells may be ascribed to the mouse origin and/or to the fibroblast cell type.

In summary, the present study demonstrates that brostallicin-mediated cytotoxicity does not depend on the MMR status of tumour cells, and that, at least in p53-deficient mouse cells, functional ATM or DNA-PK is not required. Brostallicin potentially offers the advantage of having efficacy on MMR-defective tumours that are refractory to several anticancer agents. Since the responsiveness to cisplatin treatment is affected by both MMR status and GSH/GST level/expression, brostallicin is a good candidate for clinical protocols.