Action of a Novel Anticancer Agent, CHS 828, on Mouse Fibroblasts: Increased Sensitivity of Cells Lacking Poly (ADP-Ribose) Polymerase-1¹

One important chemotherapeutic strategy for human malignancies is not only to inhibit cell proliferation, but also to eliminate tumor cells primarily by the induction of apoptosis. Therefore, the systematic investigations and detection of properties of the molecules sensitizing cells to drug-induced cytotoxicity is of great importance.

CHS 828 is a novel anticancer drug, representing a new class of drugs for cancer treatment, with an unknown primary mechanism of action. The drug is highly potent in a panel of human tumor cell lines, as well as in different in vivo models (1, 2). CHS 828 shows low correlation of activity pattern with standard cytotoxic drugs (2). U937 lymphoma cells exposed to CHS 828 show a loss of membrane integrity without displaying the typical signs of apoptosis (3). One target of interest in the search for the mechanism of action have been the mitochondria. It has been shown that cell treatment by CHS 828 induces an increase in extracellular acidification rate. This has been interpreted as an inhibition of the mitochondrial respiration and a subsequent compensatory increase in the glycolysis. However, this does not seem to be necessary nor sufficient to fully explain the cytotoxic activity of CHS 828 in tumor cell lines (4). It has been shown that combining CHS 828 with 3-aminobenzamide, an inhibitor of PARP-1,is able to increase the IC₅₀ of CHS 828 and to shift the mode of cell death from necrosis to apoptosis (5). The high efficacy of the CHS 828 attested in biological models is very promising, and the drug is currently under clinical investigations in Phase II trials in the therapy of hematological malignancies.

PARP-1 is a nuclear enzyme that is abundantly expressed in different cell and tissue types (for review, see Refs. 6 and 7). Appearance of single-strand breaks in DNA, induced by cytotoxic drugs or other genotoxic stress stimuli, results in a strong activation of PARP-1, which after cleavage of the binding between nicotinamide and ADP-ribose moiety, transfers the latter to various acceptor proteins. Excessive activation of PARP-1 generates multiple oligo- or poly(ADP-ribosyl)ated proteins and leads to a substantial depletion of intracellular NAD⁺ and consequently to a fall in intracellular ATP levels (8). Moreover, nicotinamide released by PARP-1 from NAD can be recycled to NAD in a reaction that needs ATP. The rapid activation of PARP-1 resulting in the dramatic exhaustion of ATP seems to be involved in the balance between survival and cell death. In addition, several studies have implicated PARP-1 as a key molecule important for the decision to induce apoptosis or necrosis (9). In an effort to elucidate the mechanism of action, and to consider this aspect of PARP-1 activation, we decided to investigate whether PARP-1 status would play any role in the induction of cellular response to treatment by the novel anticancer drug CHS 828. We compared the susceptibility of normal mouse fibroblasts and PARP-1 KO cells to the drug. Surprisingly, mouse cells in which the PARP-1 gene was inactivated showed enhanced sensitivity to cytotoxic action of CHS 828 as compared with the normal counterparts. On the other hand, CHS 828 induced, in a dose-dependent manner, p53 response in normal but not in mutant mouse cells. Our results indicate that inactivation of PARP-1 sensitizes cells to therapy by CHS 828.

Mice lacking PARP-1 were generated by homologous recombination (10). Immortalized MEFs were obtained from PARP-1 +/+ (A-19) and from PARP-1 −/− (A-11 and A-12) mice. Cells were grown in DMEM supplemented with 10% FCS in an atmosphere of 8% CO₂. PARP-1 −/− cells (clone A-11) were reconstituted with human PARP-1 (9). An eukaryotic expression construct containing full-length human cDNA under the control of SV40 promoter was used to generate stable cell lines expressing exogenous PARP-1. Cells were positively selected with hygromycin and resistant clones (A-11/wt2 and A-11/wt3) were isolated and characterized by Southern blot analysis (9).

Different anti-p53 antibodies recognizing distinct epitopes were used. Monoclonal anti-p53 antibodies PAb421 (Ab-1) directed against an epitope within the COOH terminus of the mouse protein, DO-1 antibodies specific for an amino-terminal epitope of the human antigen, monoclonal antibodies against MCM7 (clone DCS141.2) and against PCNA (clone PC10) were obtained from Oncogene Research Products (Cambridge, MA.). Monoclonal anti-PARP-1-antibodies (C-2-10) were from Dr. G. Poirier (Laval University, Quebec, Canada). Monoclonal anti-p27^kip1 antibodies (clone G173-524) were from BD PharMingen (San Diego, CA) and anti-cyclin D₁ were from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Anti-actin antibodies (clone C4) were from ICN (Aurora, OH). Appropriate secondary antibodies linked to horseradish peroxidase, Cy-2, or Cy-3 were from Amersham International (Little Chalfont, Buckinghamshire, United Kingdom).

MDR modulators (probenecid and verapamil hydrochloride) and LMB, inhibitor of protein export (11), were from Sigma Co. (St. Louis, MI). Proteasome inhibitors were from Calbiochem-Novabiochem Corp. (La Jolla, CA). CHS 828 was supplied by LEO Pharmaceutical Products (Ballerup, Denmark). Stock solution was prepared in DMSO and further diluted with culture medium to desired concentrations.

Cells were treated for indicated periods of time with CHS 828 at concentrations ranging between 0.01 μm to 30 μm. In some cases, it was combined with modulators of MDR (10 μm verapamil or 1 mg/ml probenecid). Protein export was inhibited by 20 nm LMB for 6 h or 24 h. To inactivate proteasome-mediated protein degradation, cells were incubated for 24 h with proteasome inhibitors (lactacystin, omuralide) at a final concentration of 5 μm. PARP-1-deficient cells were treated by a combination of proteasome inhibitors with probenecid.

Sensitivity of cells to the drug was determined by proliferation assay to measure the effect of the drug on the cells ability to divide and by a dye-exclusion test to assess the direct cytotoxicity. The cell proliferation was assessed using a microtiter plate colorimetric assay, which measures the ability of viable cells to cleave the tetrazolium salt (WTS-1; Roche Diagnostics GmbH, Vienna, Austria) to a water-soluble purple formazan, and by an alternative assay CellTiter-Glo Luminescent Cell Viability assay (Promega Corporation, Madison, WI). The latter is a homogenous method for determining the number of viable, metabolically active cells in a culture based on quantification of the ATP concentration. The assay involves adding a reagent, which in a single step, generates a luminescent signal proportional to the amount of ATP present in cells.

Cells were plated at the appropriate density (10,000 cells/well) in a 24-well multiwell plate. Twenty-four hours after plating, cells were exposed to various concentrations of CHS 828 for 24 h. DMSO in PBS was used as a solvent control. After treatment, the medium was removed and replaced with 200 μl of drug-free medium, and cells were incubated further for 48 h. At this time, 20 μl of WTS-1 was added to each well. After a 4-h incubation period at 37°C, plates were shaken for 5 min. Absorbance was measured at 460 nm on a Labsystem Multiskan microtiterplate reader (12). Alternatively, for detection of the luminescent signal, CellTiter-Glo reagent was added, and light was measured in the Wallac 1420 Victor, a multilabel, multitask plate counter. Each point represents the mean ± SD (bars) of four values from one representative experiment. At least three independent experiments were performed by each method. To assess the direct cytotoxic effect of the drug, cells were treated by CHS 828 at a final concentration of 300 nm, 1 μm, and 5 μm for 24 and 48 h. The phenotype and the adherence of cells were evaluated under phase-contrast microscopy (inverted microscope Eclipse TE300; Nikon Corporation, Tokyo), and then cells were washed once in PBS. Trypan blue dye diluted 1:1 with PBS was added, and after 10 min the accumulation of the dye was evaluated. Cells in parallel Petri dishes were fixed in 3.7% paraformaldehyde and stained by Hoechst 33258 (8 μg/ml) for 20 min. Hoechst-stained specimens were used for detection of morphological changes that are characteristic for apoptosis.

To assess the effect of drugs on cells progressing through S-phase, incorporation of ³H-thymidine was determined. Cells were plated at the appropriate density (10,000 cells/well) in a 24-well multiwell plate. Twenty-four hours after plating, cells were exposed for 24 h to various concentrations of CHS 828. Cells were pulse labeled with ³H-thymidine for 3 h before harvesting. Cells were washed twice with PBS, once with PBS containing 100 μm cold thymidine, lysed, and measured by liquid scintillation counting. The DNA labeling (dpm) in treated cells was expressed as a percentage of values measured for corresponding untreated controls. Each column represents the mean of four values. For each cell line, at least 3 independent experiments were performed.

Proteins dissolved in reduced SDS-sample buffer were separated on SDS-polyacrylamide gels and electrophoretically transferred onto polyvinylidene difluoride membrane (Amersham International, Little Chalfont, Buckinghamshire, United Kingdom). Protein loading and electroblotting were checked by Ponceau S staining. Blots were incubated with specific primary antibodies at the appropriate dilution, and the immune complexes were detected autoradiographically using appropriate peroxidase-conjugated secondary antibodies and enhanced chemiluminescent detection reagent ECL+ (Amersham International) as described previously (12, 13). In some cases, blots were stripped and used for several sequential incubations. Equal protein loading was confirmed by immunoblotting with anti-actin antibodies. WCL from human Ewing sarcoma cells overexpressing mutant p53 were used as positive p53 controls (14).

In the first step, we determined the effect of CHS 828 on the proliferation of mouse fibroblasts. We measured cell numbers by two alternative methods to exclude the possibility that the drug impaired the enzymatic activity converting substrate for colorimetric assay. Both cell-proliferation assays revealed different sensitivity of mouse cells to CHS 828 depending on the PARP-1 status (Fig. 1). Whereas proliferation of normal mouse cells remained almost unaffected by CHS 828 in concentrations ranging from 10 nm to 500 nm, the growth rate of PARP-1 −/− cells was strongly inhibited by the drug. The reconstitution of mutant cells with human PARP-1 rendered them more resistant to the action of CHS 828. These results clearly show that cells lacking the PARP-1 gene are more sensitive to the action of the drug than the normal or PARP-1-reconstituted counterparts.

Exposure of mouse cells to CHS 828 for 24 h even at high concentrations did not result in any visible phenotype changes such as cell detachment or cell shrinking, which usually are indicative for ongoing apoptosis and increased cytotoxicity (not shown). The treatment for 24 h up to a final concentration of 5 μm did not result in enhanced accumulation of trypan blue, indicating that the drug did not impair the integrity of the plasma membrane of treated mouse fibroblasts. However, prolonged exposure of mouse cells to CHS 828 for an additional 24 h revealed again its increased cytotoxic action on cells in which PARP-1 was inactivated. A 48-h exposure to CHS 828 at a final concentration of 300 nm damaged the cell membrane, as detected by increased trypan blue accumulation in mouse cells lacking PARP-1 (not shown). Surprisingly, cells accumulating the dye remained adherent and only a few cells were released into the medium. This direct cytotoxicity of CHS 828 was dose dependent. At higher CHS 828 concentrations, the majority of cells showed high accumulation of trypan blue. The introduction of human PARP-1 (clones A-11/wt2 and A-11/wt3) gave mouse cells the resistance to cytotoxic action of CHS 828 (not shown). Thus, the extent of cytotoxicity exerted by CHS 828 seems to depend on the PARP-1 status of the exposed cells. Wild-type counterparts or PARP-1-reconstituted cells were negligibly affected by prolonged CHS 828 exposure even at higher concentrations. Interestingly, untreated wild-type fibroblasts (A-19) after cultivation for 48 h became confluent and several cells released from the monolayer accumulated trypan blue. This spontaneous process was not observed in PARP-1-deficient or -reconstituted cells.

The increased sensitivity of PARP-1 KO cells to CHS 828, a novel anticancer drug, is intriguing because our previous studies on these cell lines have shown that inactivation of PARP-1 renders the cells resistant to standard cytostatic drugs like doxorubicin (15) or to proteasome inhibitors that are increasingly used as novel anticancer agents (12). The observed resistance of PARP-1-deficient cells to conventional chemotherapy was reversed by verapamil and has been shown to be attributable to an up-regulation of the MDR gene product P-glycoprotein in PARP-1-deficient cells (15). The up-regulation of the P-glycoprotein was a consequence of the reduced basal level of wild-type p53 in PARP-1 KO cells (13).

To examine which pathway is affected by CHS 828, we studied its effect on DNA synthesis and induction of apoptosis. CHS 828 significantly reduced the incorporation of ³H-thymidine into the DNA in mouse fibroblasts. Interestingly, DNA synthesis in PARP-1 −/− cells was inhibited by approximately 70 to 80% at 300 nm CHS 828. The 10-fold concentration of the drug was necessary to reduce the incorporation of ³H-thymidine by ∼50% in normal mouse fibroblasts, indicating that the latter are more resistant to the anticancer drug (Fig. 2). To assess the effect of the drug on the cell phenotype, unexposed controls and CHS 828 exposed cells were paraformaldehyde fixed, and chromatin was visualized by Hoechst 33258. The examination of the stained nuclei in specimens treated by CHS 828 for 24 h and 48 h did not reveal any significant changes characteristic of apoptosis (not shown). Thus, it is apparent that CHS 828 stronger affected the proliferative potential and viability of PARP-1-deficient cells. However, the exposure of mouse cells to the drug even for 48 h did not induce changes characteristic of apoptotic cell death.

To further characterize the biological activity of CHS 828, we examined the ability of CHS 828 to induce p53 response. It is generally known that a variety of anticancer drugs activate the p53 protein, a crucial sensor of DNA damage and a regulator of cell-cycle progression, resulting in the stimulation of p53 downstream mediators (16, 17). Monitoring the level of p53 by immunoblotting revealed that this novel anticancer agent is able to induce p53 response in normal but not in PARP-1-deficient cells. The induction of p53 in wild-type mouse fibroblasts was dose dependent. The highest increase of p53 level was observed at 1 μm concentration (Fig. 3 and Fig. 4). However, CHS 828 failed to induce p53 protein in mouse cells lacking PARP-1 even after prolonged exposure at the highest concentrations. The lack of the induction of p53 response in PARP-1 −/− MEFs was reminiscent of our previous observation that PARP-1 −/− cells failed to activate p53 protein after treatment by doxorubicin (15). This was attributable to the described up-regulation of the MDR gene product of P-glycoprotein in PARP-1-deficient cells (15). After combined treatment of mutant cells by doxorubicin and verapamil, p53 was strongly increased. To exclude the possibility that the lack of induction of p53 response in PARP-1-deficient cells was a result of enhanced activity of the P-glycoprotein, we additionally tested the action of CHS 828 after inhibition of the membrane pump by two distinct MDR modulators, verapamil and probenecid. As shown in Fig. 4, combined CHS 828 exposure with verapamil or probenecid did not affect the level of p53 protein in PARP-1 KO cells. This observation is in accordance with previous data showing that CHS 828 is not a substrate of membrane-bound MDR proteins (2). To eliminate the possibility that p53 response was impaired in mouse cells lacking PARP-1, we examined p53 expression after tumor necrosis factor-α alone, or combined with actinomycin D (not shown), after action of proteasome inhibitors or after inhibition of protein export by LMB (11). These treatments are known to induce a p53 response in these cell lines (12, 15). As shown in Fig. 4, inhibition of protein export by LMB led to an accumulation of p53 protein in PARP-1-deficient cells in a time-dependent manner, thereby evidencing that p53 response was inducible in PARP-1-deficient cells. Interestingly, combined treatment by lactacystin and probenecid was less efficient regarding the p53 increase. This result also implicates that the activation of p53 protein response differs between PARP-1 KO cells and wild-type counterparts and seems to be drug specific.

To characterize the action of CHS 828 on the progression of the cell cycle, we examined the expression, activity, and intracellular distribution of some selected markers for cell proliferation and progression through the cell cycle. The total cellular level of PCNA and cyclin D₁ did not change in wild-type and PARP-1 KO mouse cells after 24 h of drug administration (not shown). Previously partial chromosomal gains encompassing, among others, the tumor suppressor gene Rb were detected in PARP-1 KO cells by comparative genomic hybridization (18, 19, 20). Therefore, the restriction point R, regulated by hyperphosphorylation of retinoblastoma protein, is impaired in PARP-1 KO cells. Considering this fact, we examined neither the activity of CDKs catalyzing the stepwise phosphorylation of Rb nor the complex formation between Rb and E2F1 or DP-1 proteins. However, we examined the expression and intracellular distribution of MCM7, a member of a family of DNA licensing factors, and p27^kip1, an inhibitor of cyclin E-cdk2 kinase (for review, see Ref. 21). MCM2–7 proteins are essential components of prereplicative complexes that accumulate cooperatively in the nucleus during the G₁ phase and are required to establish the S-phase checkpoint. The nuclear exclusion of the MCM2–7 complex is one of the inhibitory pathways preventing the re-replication of the DNA within every cell cycle. The treatment of mouse cells by CHS 828 did not affect the total level of MCM7 protein (Fig. 3). However, its nuclear exclusion accompanied by cytoplasmic accumulation was observed posttreatment in normal and PARP-1 KO cells (Fig. 3). This effect seems to be dose dependent. The effect of CHS 828 on p27^kip1, another regulator of the cell cycle, also became evident. As depicted in Fig. 3, CHS 828 negatively regulated the level of p27 protein. Although p27^kip1 can be transcriptionally activated by some factors, the main regulatory mechanism for p27^kip1 is posttranslational. p27^kip1 levels fluctuates during the cell cycle. This is mostly attributable to cell-cycle dependent proteolysis. CDK-dependent phosphorylation of p27^kip1 on threonine 187 is essential for p27^kip1 ubiquitination and subsequent degradation by 26S proteasome. Interestingly, PARP-1 KO cells express much a higher basal level of p27 as in contrast to their normal counterpart. CHS 828-mediated down-regulation of p27^kip1 mimics its reduction occurring in the S and G₂ phases. Thus, our preliminary results show that CHS 828 affects the expression and localization of some key regulators of the cell cycle progression.

What is the mechanism of the increased susceptibility of PARP-1-deficient cells to the novel anticancer drug? One could speculate that CHS 828 exerts a stronger direct cytotoxic effect on cells lacking PARP-1. However, this assumption seems not to be supported by the results of the trypan blue test performed after a 24-h exposure of the cells to the drug. The viability of PARP-1 KO cells was not significantly affected after 24-h treatment. Only a few cells accumulated the dye. One possible explanation for the proliferation blocking effect of the drug on PARP-1-deficient cells could be that they possess a diminished capability to repair damaged DNA. Indeed, PARP-1 is an essential component of the base-excision repair complex (22), and its activity is necessary for the formation of a functional complex with XRCC1 and for fine regulation of its activity. PARP-1 is also involved in the stabilization of wild-type p53 protein (13, 23). The genomic instability of PARP-1 animals and cells seems to be a consequence of the reduced stability of wild-type p53 protein (18, 19, 20). Furthermore, the lack of induction of p53 response in PARP-1 −/− cells exposed to CHS 838 could also contribute to the reduced DNA repair, because activated wild-type p53 protein is known to induce distinct genes directly involved in the DNA repair such as p53R2 ribonucleotide reductase (24) and GADD-45 (25). The assumption, that reduced capability of mutant cells to repair damaged DNA render them more sensitive to the action of CHS 828, is at least partially supported by our results. We observed that the incorporation of ³H-thymidine into DNA in CHS 828 treated PARP-1-deficient cells was markedly lower than in their normal counterparts. Because we did not use any specific inhibitors of distinct DNA polymerases in our assay, we were not able to discriminate between labeling during replicative and unscheduled DNA synthesis.

However, previous studies of the mechanism of action of CHS 828 on human tumor cell lines do not show any signs of PARP-1 activation (3, 5) or any other signs of DNA damage (5). There are no previous data reporting on the role of p53 in CHS 828-induced cytotoxicity.

This study shows for the first time that CHS 828 is able to increase p53 level. We observed a dose-dependent induction of p53 protein in normal, but not in PARP-1-deficient MEFs, that are more sensitive to the cytotoxicity of CHS 828. In most cell systems, an induction of p53 would lead to growth arrest or induction of apoptosis. These intriguing results require further investigation to identify what is the function of CHS 828-induced p53 protein, whether increased p53 is active as a transcription factor, and which p53-dependent target genes are primarily induced in cells exposed to CHS 828.

We thank Mrs. Maria Eisenbauer for the cultivation of cells and Mr. Paul Breit for preparations of photomicrographs.