Variations in Mre11/Rad50/Nbs1 status and DNA damage-induced S-phase arrest in the cell lines of the NCI60 panel

NCI60 cancer cell lines were obtained from the Developmental Therapeutics Program at the National Cancer Institute and were cultured in RPMI 1640 media (Hyclone) supplemented with 10% heat-inactivated fetal bovine serum (Hyclone) and Antibiotic-antimycotic (Gibco). Cells were cultured at 37°C with 5% CO₂.

Elevated expression of Mre11 was achieved by cloning Mre11 from pDB20 (obtained from John Petrini, Memorial Sloan-Kettering Cancer Center) into pCR3.1 (Invitrogen) [18]. This plasmid was transfected into HCT116 cells using lipofectamine (Invitrogen), according to the manufacturer's protocol. Cells were selected in G418 (500 μg/ml) and multiple colonies were screened for maximum expression.

SN38, the active metabolite of irinotecan, was obtained from Pfizer, and 2-(4-amino-3-methylphenyl)-5-fluorobenzothiazole (5F 203, NSC 703786) was obtained from the Developmental Therapeutics Program at the National Cancer Institute (Bethesda, MD). Both drugs were dissolved in DMSO and stored at -20°C.

For cell cycle analysis, cells were rinsed with phosphate-buffered saline (PBS), harvested with 0.05% trypsin, and fixed in 70% ethanol overnight. Cells were stained with 100 μg/ml propidium iodide in 1 mg/ml ribonuclease A and analyzed on a FACScan flow cytometer (Becton Dickinson).

Following treatment, cells were rinsed with phosphate buffered saline, lysed by direct addition of Laemmli buffer to the wells, boiled for 5 min, and stored at -80°C. Protein levels were quantified using EZQ Protein Quantitation Kit (Molecular Probes). 15 μg of each lysate was separated by SDS-PAGE, and transferred to polyvinylidene difluoride membranes. Membranes were blocked and incubated overnight at 4°C with primary antibodies specific for Nbs1 (Cell Signaling), Rad50 (Novus Biologicals), Mre11 or actin (Calbiochem). Membranes were washed and incubated for 30 min at room temperature with anti-rabbit or anti-mouse secondary antibody conjugated to horseradish peroxidase. Proteins were visualized with enhanced chemiluminescence. Densitometry scans were quantified using Scion Image software (Scion Corporation) and normalized to a serial dilution of the same protein from MDA-MB-231 present on each blot. Several exposures of each western blot were scanned to avoid over-exposure and to ensure that the samples were within the range of the standards.

Cells were seeded at low density (500-1000 cells) in 96-well plates and then incubated with drug for 24 h. Following treatments, cells were washed and grown in fresh media for 5-7 days at 37°C. Prior to attaining confluence, cells were washed, lysed, and stained with Hoechst 33258, as previously described [19]. Fluorescence was read on a microplate spectrofluorometer (Spectramax M2). Within each experiment, each treatment was repeated in at least eight wells. Sensitivity to the "NSC" compounds was determined by the Developmental Therapeutics program at NCI and the data were obtained from their web site (http://​dtp.​nci.​nih.​gov). In their screen, all cell lines were incubated with a range of drug concentrations, usually for 48 h, and then fixed and stained with sulforhodamine B.

DNA damage, such as that caused by the topoisomerase I inhibitor SN38, triggers checkpoints that arrest cell cycle progression to allow for DNA repair. We have previously compared seven cell lines for their ability to arrest in S phase [20]. In general, most cell lines arrest in late, middle or early S phase as the SN38 concentration increases. However, we found that HCT116 cells were notably different in that they showed primarily G2 arrest at high concentrations of SN38. Other studies have demonstrated that HCT116 cells have low levels of the MRN complex [9]. We confirmed that these cells have very low expression of Mre11 protein compared to MDA-MB-231 cells, as measured by western blotting (Figure 1). Rad50 and Nbs1 levels were also low in HCT116. Since Mre11 has been shown to be involved in the S-phase checkpoint, we hypothesized that the Mre11 defect was responsible for the lack of S-phase arrest in HCT116 cells. To test this hypothesis, we transfected HCT116 cells with a vector expressing Mre11. We selected individual colonies that expressed higher levels of Mre11 protein over control or vector-transfected cells (Figure 1). HCT116+Mre11 cells not only had increased levels of Mre11, but also had higher levels of Rad50 and Nbs1 proteins, suggesting that the protein stability of Rad50 and Nbs1 depends upon complex formation with Mre11. Similar results were also obtained in an earlier study that transfected Mre11 into HCT116 cells [21]. Importantly, we also show that expression of Mre11 rescued SN38-induced S-phase arrest in HCT116 cells (Figure 1). We conclude that the failure of HCT116 cells to arrest in S phase on SN38 is due to their very low level of MRN.

The National Cancer Institute has screened more than 100,000 compounds in a panel of 60 cell lines and this data is publicly available. The database also has information on the gene expression patterns in each cell line. However, our data in Figure 1 suggests that MRN proteins are primarily regulated by stability so their levels are unlikely to be reflected by mRNA levels. In order to determine the frequency of defects in the MRN complex we made lysates from all of the cell lines. We assessed expression of Mre11, Rad50 and Nbs1 by western blot, and quantified the intensity of the protein bands (Figure 2A, B and Table S1 in Additional File 1). Expression of each protein was assessed relative to the levels in MDA-MB-231 which was set at a value of 1. There was a large range of expression with relative protein levels ranging from 0.06 to 2.4 for Mre11. There was also more than 10-fold difference in protein levels of Rad50 and Nbs1 (0.23 to 3.1 for Rad50, and 0.19 to 2.4 for Nbs1). There was a strong correlation between protein levels when we compared Mre11/Rad50, Mre11/Nbs1 and Rad50/Nbs1 (Figure 3A-C). This is consistent with our finding that transfection with Mre11 also increased Rad50 and Nbs1 levels. In addition to HCT116, a number of other cell lines expressed low levels of Mre11, in particular U251 had levels as low as HCT116. The next 5 lowest cell lines were HL60, MALME-3M, SK-MEL-2, IGROV1 and TK10. Each of these cell lines also had relatively low levels of Rad50 and Nbs1, consistent with the intact complex regulating the stability of each protein.

While this analysis was designed to identify cell lines with low levels of the MRN proteins, we also identified several lines in which the ratio of these proteins far exceeded the expected values. For example, SW620 cells had about 10 fold more Mre11 than Rad50, whereas Nbs1 was intermediate between these two levels. On the other hand, HCC2998, K562 and EKVX had about 2 fold more Rad50 than Mre11. For these latter cell lines, Mre11 and Nbs1 were relatively similar suggesting that it is the elevated level of Rad50 that maybe abnormal. Whether these cell lines contain functionally defective MRN remains to be determined.

To extend this analysis further, we compared the protein levels to mRNA levels obtained from the NCI database (Figure 2C and Table S1 in Additional File 1). The correlation coefficient for Mre11 mRNA and protein was 0.55, but only 0.24 and 0.15 for Rad50 and Nbs1, respectively (Figure 3D-F). This observation suggests that Mre11 protein levels are more highly regulated by mRNA levels whereas the Rad50 and Nbs1 protein levels are relatively independent of mRNA levels and likely depend more on protein stability mediated via formation of the MRN complex. It is also important to note that most of the cells we identified as having low levels of Mre11 protein had relatively normal levels of MRN mRNA suggesting that, like HCT116, they may contain a defective protein rather than the level of protein being regulated at the transcriptional level. The one notable exception was IGROV1 which has a very low level of Mre11 mRNA; surprisingly, these cells also exhibited a very low level of mRNA for Rad50. Therefore in this case, it appears that the reduced levels of the MRN complex are a consequence of transcription.

Considering that defective MRN can cause chemo- and radio-sensitivity, we hypothesized that other drugs in the NCI database would correlate with the low levels of MRN observed. Accordingly, we used the NCI COMPARE program to find correlations between MRN proteins and compound activity. We set the correlation coefficient to 0.5 to restrict the number of compounds identified. The search recovers both positive and negative correlations, that is, a positive correlation represents high protein level correlating with sensitivity whereas a negative correlation represents a cell line with low protein but high sensitivity to a compound. Mre11 and Nbs1 protein levels correlated with the activity of only a few compounds, and approximately equally distributed between positive and negative correlations (Table 1). In contrast, Rad50 protein levels correlated with many compounds, most of them positive. Three compounds were identified twice in the search, once each for Mre11 and Rad50 proteins. NSC 734406 correlated with low levels of Mre11 and Rad50 and therefore warrants validation and further study (Figure 4). The other two compounds, NSC359452 and NSC740487, correlated with high Mre11 and Rad50, but further examination of the dose response curves revealed the compounds were only marginally active in HCC2998 and HT29 cells and not in the other cell lines even at the highest concentration tested (10^-4 M). Hence, sensitivity to these latter compounds does not correlate with expression of the MRN proteins. This reflects a limitation of the COMPARE program that is discussed below.

While the MRN complex has a critical role in the S-phase checkpoint, there are many other proteins that are also involved. In search of other possible S phase checkpoint defects, we screened all of the cell lines in the NCI60 panel for their ability to S-phase arrest in response to the topoisomerase I inhibitor SN38 (Figure 5). Several of the cell lines exhibited multiple peaks in untreated cells which is due to a mix of diploid and tetraploid cells. A mid S-phase arrest is observed in most of the cell lines when incubated with 3 - 10 ng/ml SN38. Several cell lines exhibited much more marked S-phase arrest at lower concentrations of SN38, for example T47D. One cell line, HL60, showed predominantly sub-G1 DNA content above 3 ng/ml indicative of apoptosis rather than arrest. Of the 59 cells lines tested, there were four that primarily arrested in G2 phase rather than S at the maximum concentration tested: HCT116, IGROV1, TK-10 and HCC2998 (these cell lines are designated "100" in Figure 6). Interestingly, the first three of these cell lines also had low MRN levels. For these cell lines, it is possible that the low MRN can explain their S-phase defect. In contrast, HCC2998 had one of the highest levels of Mre11 but lacked S-phase arrest. One possibility as discussed above is that the high Mre11 might reflect a functional defect in Mre11 and this would then explain their S-phase defect. However, several other cell lines, including U251 and HL-60, had low levels of MRN, but were still proficient for S-phase arrest, hence the level of MRN does not by itself predict whether a cell will arrest.

Using the COMPARE program to analyze the values in Figure 6, we identified 157 compounds that correlated with the lack of S arrest, but only 11 compounds that had a negative correlation (top 10 hits shown in Table 2). The correlation coefficients for the top compounds were much higher than those observed for correlations with MRN expression shown in Table 1. The top 30 compounds were analyzed individually. It was noted that many compounds exhibited poor correlation with the cell lines that failed to S-phase arrest, rather the relatively high correlation coefficients resulted from strong correlation with the other 55 cell lines rather than the 4 that failed to arrest. However, six of the top 30 correlations all derived from a single cluster (k15.21) of the 3dMind Map, a self organizing map which clusters similar cytotoxicity data into a two dimensional visual translation of related compounds (Figure 7). These clusters generally reflect compounds with a similar mechanism of action, although cluster k15.21 is unrelated to any common anticancer mechanism. About half of the compounds in this cluster, have closely related structures and are known as benzothiazoles, although other structures such as NSC 680467 also showed high correlation with cells that failed to S phase arrest. The correlation with S phase arrest for two of the benzothiazoles is shown in Figure 8. In general, most of the compounds in cluster k15.21 inhibit growth of 5 cell lines, HCC2998, IGROV-1, TK10, MCF7 and T47D (although the latter is not evident for the compounds shown in Figure 6B), but only the first three of these failed to arrest in S phase when incubated with SN38. To validate these results, we obtained the benzothiazole NSC 703786 (5F 203; the only one available from NCI) and assessed growth inhibition in various cell lines including those that lack S-phase arrest (Figure 9). Our results were consistent with those from the NCI database.

The sensitivity of cells to benzothiazoles is reported to be due to the ability of cells to metabolically activate these compounds which then go on to damage DNA [22, 23]. MCF7 cells exhibit the highest metabolism of benzothiazoles and we therefore anticipated that they should sensitize non-responsive cells in co-culture. We used transwell plates in a 96 well format and plated ~50,000 MCF7 cells in the upper well. Other cell lines, including HCT116, were plated at low density in the lower wells. NSC703783 (1 μM) was added for 24 h then removed, in an experiment similar to that in Figure 9. The MCF7 cells had no impact on the growth of six cell lines tested. We conclude that differences in metabolism do not adequately explain the differences in sensitivity of cells to the benzothiazoles.