Pharmacogenomic Profiling and Pathway Analyses Identify MAPK-Dependent Migration as an Acute Response to SN38 in p53 Null and p53-Mutant Colorectal Cancer Cells

In advanced colorectal cancer, response rates remain disappointingly low, at 40% to 50% for combined 5-FU–based regimes (1, 2). The main reason for the lack of response is acute or acquired drug resistance. The identification and manipulation of drug resistance mechanisms in colorectal cancer could significantly enhance patient response to therapy.

The p53 tumor suppressor gene is the most frequently mutated gene in all human cancers and has been shown to be mutated in at least 50% of colorectal cancer tumors (3). A number of studies have reported that p53 is a predictive marker of sensitivity to chemotherapy (4–8); however, other studies have reported conflicting findings (9, 10). Previously, we reported that colorectal cancer cells display similar levels of sensitivity to irinotecan irrespective of p53 status; this is not the case for either 5-FU or oxaliplatin both of which have significantly reduced cytotoxic effects in p53 null cells (11). Ravi and colleagues showed that the combination of irinotecan and TRAIL eliminates hepatic metastasis in both p53 wild-type and null colorectal cancer cells in vivo. They further showed that the synergy displayed between irinotecan and TRAIL was mediated via a p53-independent mechanism that involved the inhibition of JAK–STAT3/5 signaling (8). Indeed, work carried out in our laboratory showed that irinotecan, but not 5-FU or oxaliplatin, upregulated Fas cell-surface expression via a novel p53-independent mechanism that involved the activation of STAT1 followed by enhanced Fas cell-surface trafficking (11). Bhonde and colleagues used expression profiling to identify the genes that were associated with induction of apoptosis in p53-mutant cells following SN38 treatment. They identified a significant number of mitotic genes that were differentially regulated between p53 wild-type and p53-mutant cells and further showed that suppression of the mitotic gene, hMps1, reduced the apoptosis induced following SN38 treatment in p53-mutant cells (12).

The aim of this study was to carry out transcriptional profiling of isogenic p53 wild-type and p53 null colorectal cancer cells following treatment with SN38 (the active metabolite of irinotecan) to identify the signaling pathways activated by this agent in these genetic backgrounds. Bioinformatics analyses revealed that, in response to SN38, a number of pathways involved in promoting cell migration were activated in the p53 null setting, but not in the p53 wild-type setting. Importantly, cell migration assays revealed that SN38 treatment increased the migratory potential of p53 null and p53-mutant colorectal cancer cells, but not p53 wild-type cells. Of note, inhibitors of mitogen-activated protein kinase (MAPK) signaling, such as the mitogen-activated protein/extracellular signal–regulated kinase (MEK) inhibitor AZD6244, attenuated SN38-induced migration in p53 null and p53-mutant colorectal cancer cells. These results suggested that in the absence of wild-type p53, SN38 promotes migration of colorectal cancer cells. Moreover, inhibiting MAPK blocks this potentially prometastatic adaptive response in p53-mutant/null cells.

All colorectal cancer cells were grown as previously described (13). Following receipt, cells were grown up and as soon as surplus cells became available, they were frozen as a seed stock. All cells were passaged for a maximum of 2 months, after which new seed stocks were thawed for experimental use. All cell lines were tested for Mycoplasma contamination at least every month. RKO cells were obtained from the American Type Culture Collection [ATCC; authentication by short-tandem repeat (STR) profiling/karyotyping/isoenzyme analysis] and maintained in Dulbecco's Modified Eagle's Medium (DMEM) 2001. The p53 wild-type HCT116 human colorectal adenocarcinoma cell line and a matched isogenic p53 null cell line (kindly provided by B. Vogelstein, Johns Hopkins University School of Medicine, Baltimore, MD) were maintained in McCoy's medium (GIBCO; Invitrogen). LoVo cells were obtained from the European Collection of Cell Cultures, maintained in DMEM in 2004. HCC2998 cells were obtained from the National Cancer Institute–Frederick Cancer DCT Tumor repository [October 2008; authentication: SNP arrays, oligonucleotide base HLA typing, karyotyping and STR (May 2007)] and maintained in RPMI-1640. DLD-1 cells were kindly provided by Senji Shirasawa (Department of Cell Biology, Faculty of Medicine, Fukoka University, Fukuoka, Japan) in August 2008 and were maintained in DMEM. HCT116, HT29, LoVo, DLD-1, and RKO cell lines were validated by STR profiling by LGC Standards (14) in May 2011. The p53 wild-type and p53 null SN38-resistant HCT116 sublines were generated in our laboratory as previously described (6). All medium was supplemented with 10% dialyzed fetal calf serum, 50 μg/mL penicillin–streptomycin, 2 mmol/L l-glutamine, and 1 mmol/L sodium pyruvate (all medium and supplements from Invitrogen Life Technologies Corp.). All cell lines were maintained at 37°C in a humidified atmosphere containing 5% CO₂.

The following reagents were used SN38 (Abatra Technology Co., Ltd; Fig. 1A), U0126 (Cell Signalling Technology, Inc.), AZD6244 (Astra Zeneca; ref. 15).

HCT116 p53wild-type and p53 null cells were treated with 5 nmol/L SN38 for 0, 6, 12, and 24 hours. Total RNA was isolated from 3 independent experiments using the RNA STAT-60 Total RNA isolation reagent (Tel-Test, Inc.) according to the manufacturer's instructions. Five micrograms of total RNA was sent to Almac Diagnostics (Craigavon) for cDNA synthesis, cRNA synthesis, fragmentation, and hybridization onto Affymetrix HG U133 Plus2.0 microarrays. Detailed experimental protocols and raw expression data are available at http://www.ebi.ac.uk/arrayexpress/ (Acession numbers E-MEXP-1171 and E-MEXP-1194).

Microarray data analysis was carried out using GeneSpring v7.3 (Agilent Technologies UK Limited) as previously described (16). First, the HCT116 p53 wild-type and p53 null arrays were assessed as a single experiment. Following standard normalization and filtering, a 4,800 drug-inducible gene list was created using a 1.5-fold cut-off for each gene relative to their zero-hour control, with genes meeting this criterion in at least one time point retained. This gene list was used as the input for principal component analysis (PCA) and Hierarchical clustering analysis. For PCA, the gene list was used for all samples, with mean centering and scaling. For hierarchical clustering, the gene list was analyzed using Pearson correlation as the similarity measure and the average linkage clustering algorithm. Second, constitutive gene expression in p53 wild-type and p53 null cells was examined. Normalizations and filtering were carried out, and those genes that were altered 1.5-fold or more were retained as the basal gene list and used in pathway analysis. Finally, the drug-inducible data from each cell line were analyzed separately to identify SN38-inducible gene lists for p53 wild-type and null cells. A Venn diagram was used to identify genes that were common and unique to each experimental setting; these common and unique gene lists were retained for pathway analysis.

Pathway analysis was conducted using KEGG and GenMAPP pathways. Pathway analysis was carried out on the basal and drug-inducible gene lists described above. Pathway statistical significance was assessed using hypergeometric statistics. For both basal and SN38-inducible analyses, pathways were retained that contained 10 or more genes per pathway. Pathway analysis from the p53 wild-type experiments were compared with the p53 null experiments, and common pathways were retained and further analyzed for differential expression either basally or following SN38 treatment.

Total RNA was isolated as described above. Reverse transcription was carried out as previously described (16) using 2 μg of RNA using a Moloney murine leukemia virus–based reverse transcriptase kit (Invitrogen) according to the manufacturer's instructions. All amplifications were primed by pairs of chemically synthesized 18- to 22-mer oligonucleotides designed using freely available primer design software (17).

Western blots were carried out as previously described (11). Anti–phospho-extracellular signal-regulated kinase 1/2 (pErk1/2; Thr²⁰²/Tyr²⁰⁴; Cell Signalling) mouse monoclonal antibody was used in conjunction with a horseradish peroxidase–conjugated anti-mouse secondary antibody (Amersham), and anti-Erk1/2 (K-23; Santa Cruz) rabbit polyclonal antibody was used in conjunction with a horseradish peroxidase-conjugated anti-rabbit secondary antibody (Amersham).

Cell migration was assessed using the CIM-plate 16 and the Xcelligence system (Roche Applied Sciences), according to the manufacturer's instructions.

Pearson product moment correlation (r) and the coefficient of determination (r²) were calculated using MS excel. The significance of the correlation was determined using a 2-tailed test of significance (18). The unpaired 2-tailed t test was used to determine statistically significant differences between treatment effects. Significance was defined as P < 0.05.

Transcriptional profiling was carried out to compare the acute changes in mRNA expression induced by SN38 in isogenic p53 wild-type and p53 null HCT116 colorectal cancer cell lines. The results were combined into a single data set that contained 4,800 genes that changed by 1.5-fold or greater in at least one treatment condition. These 4,800 genes were used as the starting point for PCA and hierarchical clustering. The PCA revealed that the greatest variability in gene expression was because of p53 status, with the p53 wild-type and p53 null treatment groups clustering separately (Fig. 1B). Secondly, within these groups, the drug-treated samples were more closely grouped than the nontreated control samples. Hierarchical clustering also showed 2 main groups: one branch that contained all of the p53 wild-type samples and one that contained all the p53 null samples (Fig. 1C). Within each branch, the 12-hour and 24-hour samples were more closely related than the 6-hour sample, and the least closely correlated sample was the control sample. Thus, both PCA and hierarchical clustering analyses segregated the samples according to p53 status and drug treatment; therefore the signaling pathways activated by SN38 in p53 null cells are significantly different from those activated in p53 wild-type cells.

To further compare the mechanism of action of SN38 in p53 wild-type and p53 null cells, we analyzed each data set independently and identified the unique and common genes in the 2 lists. Following normalization and filtering, it was found that 3,296 genes changed 1.5-fold or more in at least one time point in p53 wild-type cells, whereas 2,738 genes changed 1.5-fold or more in at least one time point in p53 null cells. Of these genes, only 752 were common to both p53 wild-type and null samples (Fig. 1D and Supplementary Table S1), whereas 2,544 genes (77.2%) were unique to p53 wild type (Fig. 1D and Supplementary Table S2) and 1,986 (72.5%) genes were unique to p53 null cells (Fig. 1D and Supplementary Table S3). From this analysis, it was again clear that the downstream effects of SN38 are significantly different in p53 wild-type and null cells. The constitutive differences in gene expression between the p53 wild-type and null cells were also assessed, with a significant number of genes (2,102) found to be constitutively altered 1.5-fold or more between the 2 cell lines (Supplementary Table S4).

Q-PCR analysis was carried out to validate the microarray datasets. Validation of the p53 wild-type dataset following SN38 treatment has previously been reported (16). For the comparison of constitutive gene expression in the p53 wild-type and p53 null cells, the results for selected genes are displayed in Supplementary Table S5. Overall, when the Q-PCR data was compared with the microarray data, the Pearson correlation product moment (r) was 0.789 (r² = 0.6225 and P = 0.0002; Fig. 2A). The results of the Q-PCR analyses for the genes induced by SN38 treatment in the p53 null cells are presented in Supplementary Table S6. For the selected genes, Pearson correlation showed an r value of 0.73, with r² = 0.5383 (P = 0.000015; Fig. 2B). Overall, statistically significant correlations (P < 0.05) were observed for 24 of the 26 (∼92%) genes (data not shown). In the analysis of the p53 wild-type cells following SN38 treatment, 25 genes were analyzed by Q-PCR and showed excellent correlation with the microarray results, with an r value of 0.80 and r² = 0.61 (P = 0.000001; ref. 16). Altogether, these results indicated a strong overall concordance of gene expression trends between the microarray analysis and real-time reverse transcriptase PCR (RT-PCR) analysis and showed the robustness of the microarray datasets for further analysis.

Pathway analysis was carried out to identify those signaling pathways constitutively altered between p53 wild-type and p53 null cells and those altered following SN38 treatment in the p53 wild-type and p53 null settings. Eleven pathways containing 10 or more genes were found to be differentially active in p53 wild-type and p53 null cells (P < 0.05; Supplementary Table S7). The majority of the identified pathways were downregulated in the p53 null cells compared with the p53 wild-type cells. Further analyses showed that a total of 32 pathways were altered following SN38 treatment in either p53 wild-type, p53 null, or both settings (Table 1). Of the 32 pathways, 16 contained genes that were altered in both the p53 wild-type and null settings (Table 1, highlighted in bold). Of these 16 pathways, 5 were differentially regulated in p53 wild-type and p53 null cells following SN38 treatment (Table 2). The pathways that displayed differential regulation were adherens junction, focal adhesion, JAK-STAT signaling, MAPK signaling, and regulation of the actin cytoskeleton, and in all cases, the pathways were upregulated in p53 null samples following SN38 treatment compared with p53 wild-type samples. In addition, all of these pathways were constitutively lower in the p53 null samples compared with p53 wild-type samples (Supplementary Table S7).

To validate the pathway analysis results, we selected 15 members of the focal adhesion pathway and analyzed their mRNA expression in p53 wild-type and p53 null cells following SN38 treatment. The focal adhesion pathway was selected, as this pathway included differentially expressed genes that were also members of the actin signaling (ITGB4, ITGA6, ITGA3, MYLK, ROCK1, PIK3R1, PIK3CD, ACTN1, CALM3, MAPK1, ACT, FAK, B-Raf, PAK1, and PAK2) and MAPK signaling (MAPK1, GRB2, PRKCA, B-Raf, RAP1A, MAPK8, c-Jun, PAK1, and PAK2) pathways. Genes from various points in the pathway were chosen, including receptors (INTGA6 and INTGA3), members of cell motility signaling (Actn1, Actn4, ZYX, VASP, CPN, and Actb) and cell proliferation genes (CCND3, MAPK1, DUSP5, DUSP6, PRKCA, MAPK8, and Jun). Analysis of these genes showed that ZYX Actb, CCND3, MAPK1 (ERK2), MAPK8, Jun, INTGA3, and DUSP5 were downregulated in p53 wild-type cells, but were highly upregulated in p53 null cells following SN38 treatment (Fig. 2C). INTGA6, Actn1, Actn4, VASP, CPN, PRKCA, and DUSP6 were upregulated in both p53 wild-type and p53 null cells following SN38 treatment; however, the level of upregulation was significantly greater in the p53 null setting. Overall, these results showed differential activation of the focal adhesion pathway in p53 wild-type and null cells and thereby confirmed the results of the pathway analysis.

The results of the pathway analyses indicated that adherens junctions, ECM receptor interaction, focal adhesion, actin cytoskeleton, MAPK signaling, and JAK-STAT signaling were all constitutively downregulated in p53 null cells compared with their p53 wild-type counterparts, which suggested that the p53 null cells may migrate more slowly than the p53 wild-type cells. In addition, it was noted that following SN38 treatment, all of these pathways were upregulated in p53 null cells compared with p53 wild-type cells, suggesting that SN38 may increase cell migration in the p53 null setting. To test these hypotheses, we measured migration of p53 wild-type and null isogenic cells in the presence and absence of SN38. Both cell lines started to migrate after approximately 8 hours, and active migration continued to approximately 28 hours, therefore all measurements of cell migration, cell index, and slope of the curve were measured between 8 and 28 hours. As predicted by the pathway analyses, we found that constitutively, the p53 null cells migrated more slowly than the p53 wild-type cells, as the p53 null cells displayed a reduced slope of the curve (0.115/h) compared with the p53 wild-type cells (0.278/h; Fig. 3A, P < 0.001). Importantly, following treatment with SN38, the migration of p53 null cells significantly increased (Fig. 3B, P < 0.001). In contrast, the p53 wild-type cells did not display any increase in cell migration following SN38 treatment (Fig. 3B). To assess whether these findings were unique to SN38, we carried out migration experiments following treatment with the 2 other chemotherapies commonly used in the treatment of colorectal cancer, 5-FU and oxaliplatin. Treatment with 5-FU induced cell migration in both p53 wild-type and null cells to a similar extent, whereas treatment with oxaliplatin significantly reduced migration of p53 wild-type cells, but did not significantly alter migration of p53 null cells (data not shown). These results suggested that in this panel of clinically relevant chemotherapeutic agents, SN38 is unique in its ability to differentially induce migration based on p53 status.

We also compared the migratory potential of p53 wild-type and p53 null SN38-resistant daughter cell lines (6). The p53 wild-type SN38-resistant daughter cell line displayed highly reduced migratory ability compared with its parental SN38-sensitive cell line (Fig. 3C, P < 0.001). Most notably however, the p53 null SN38-resistant daughter cell line migrated at a significantly faster rate than the p53 null SN38-sensitive parental line (Fig. 3C, P < 0.001). These results further strengthened the hypothesis that increased migratory potential is an important adaptive response to SN38 treatment in p53 null colorectal cancer cells.

As the pathway analyses of the microarray data (Tables 1 and 2) and the molecular analyses (Fig. 2) indicated differential MAPK signaling in p53 wild-type and p53 null cells following SN38 treatment and given the role of this pathway in cell migration (19), we next assessed whether MAPK signaling was involved in promoting the migration of p53 null cells following SN38 treatment. Western blot analyses showed that ERK1/2 activation was inhibited following treatment with the MEK inhibitor, AZD6244 (Fig. 4A). Importantly, it was found that the increased migration of HCT116 p53 null cells following SN38 treatment was significantly reduced following cotreatment with the MEK inhibitors U0126 and AZD6244 (Fig. 4A, P < 0.001 and data not shown). To determine whether these observations were cell line dependent, we assessed migration in 4 other colorectal cancer cell lines: DLD1 and HCC2998 (both p53 mutant), and LoVo and RKO (both p53 wild-type). These analyses showed that cell migration was significantly induced following SN38 treatment in p53-mutant cells (DLD1, P < 0.001 and HCC2998, P < 0.001) and that this was inhibited following cotreatment with AZD6244 (Fig. 4B). In agreement with results from the HCT116 model, SN38 treatment did not induce cell migration in the p53 wild-type LoVo and RKO cell lines (Fig. 4C). These results provided further evidence that in the absence of functional p53, SN38 enhances MAPK-dependent migration of colorectal cancer cells.

p53 is one of the most commonly mutated genes in cancer and at least 50% of colorectal cancer tumors have dysfunctional p53 (3, 20, 21). Several studies have evaluated the role p53 plays in determining sensitivity to irinotecan therapy, with some suggesting a predictive capability and others reporting no correlation (6–10). Indeed, studies from our own laboratory have shown that there was no significant difference in the sensitivity of p53 wild-type and p53 null colorectal cancer cells to irinotecan (6); this was not the case for either 5-FU or oxaliplatin, the other 2 chemotherapeutic agents used to treat colorectal cancer, both of which had reduced activity in the p53 null setting. Previous studies have shown that p53 wild-type colorectal cancer cells induce a prolonged G₂/M arrest in response to irinotecan, whereas p53 null colorectal cancer cells undergo a more short-term arrest followed by apoptosis (22). We previously reported that although p53 wild-type and p53 null colorectal cancer cells are similarly sensitive to irinotecan, the molecular signaling pathways activated in the 2 settings can be significantly different (11). Therefore the aim of this study was to understand p53-independent signaling in response to irinotecan in colorectal cancer.

Following transcriptional profiling of SN38-treated p53 wild-type and null HCT116 cells, PCA and hierarchical clustering analysis showed that there was a clear separation between the p53 wild-type and p53 null samples. Furthermore, it was found that only approximately 25% of the genes significantly altered by SN38 treatment in each cell line were similarly altered in the other cell line. These findings indicate that distinct genes/pathways are activated in response to SN38 in the different p53 backgrounds. The microarray results were validated by Q-PCR, and the level of correlation was found to be highly significant and similar to those of other previously published studies (16, 23–25).

Using KEGG and GenMAPP pathway analysis programs, we identified adherens junction, focal adhesion, JAK-STAT signaling, MAPK signaling, and actin cytoskeleton as the key pathways that were differentially regulated following SN38 treatment in p53 wild-type and p53 null cells. Previous work from our laboratory investigated the role p53 and STAT1 play in regulating Fas-mediated apoptosis in response to chemotherapy in colorectal cancer cells (11). In that study, we found that irinotecan treatment resulted in STAT1 activation in p53 null cells, but not p53 wild-type cells, and this resulted in upregulation of Fas receptor expression at the cell membrane. The microarray data and pathway analyses presented in this study also indicated that the JAK–STAT pathway was activated in p53 null cells following exposure to SN38. Further work is ongoing to investigate the roles of JAK1/2 and STAT3 in colorectal cancer in regulating the response of colorectal cancer cells to chemotherapy in clinically relevant genetic contexts (manuscript submitted).

In this study, we focused on the role of the focal adhesion pathway in regulating the response of colorectal cancer cells to SN38. This pathway contains components of MAPK signaling and regulators of the actin cytoskeleton, which pathway analyses also identified as being differentially regulated between p53 wild-type and null cells following SN38 treatment. Several studies have shown a link between focal adhesion kinase (FAK) and p53; specifically, p53 binds to the FAK promoter and regulates its activity. In addition, several studies have shown that FAK is upregulated in breast and colorectal cancer tumors, which contain p53 mutations compared with those expressing wild-type p53 (26–28). Real-time RT-PCR results showed that following SN38 treatment, a number of key components of the focal adhesion pathway were differentially regulated in p53 wild-type and p53 null cells. Some of the most marked differences between p53 wild-type and p53 null cells were observed for those focal adhesion pathway genes that are also components of the MAPK signaling pathway.

As several pathways that regulate cell migration were identified as being differentially regulated in p53 wild-type and p53 null cells, we carried out a number of cell migration experiments. Constitutively, p53 null cells displayed a lower level of cell migration compared with the p53 wild-type cells, which is in agreement with the results of the pathway analyses. Importantly, following SN38 treatment, the p53 null cells displayed a significantly faster rate of migration compared with the p53 wild-type cells, which again is in agreement with the transcriptional profiling and pathway analysis results. In addition, we also found that a stably SN38-resistant p53 null daughter cell line migrated at a faster rate than the parental p53 null cell line; in contrast, the corresponding p53 wild-type SN38-resistant cell line failed to migrate at all. We speculated that the differences between the migratory potential of the p53 wild-type and null HCT116 isogenic cell lines was likely due to the differences we observed in the focal adhesion signaling pathway, which was found to be constitutively lower in the p53 null setting but which, following SN38 treatment, was upregulated in p53 null cells, but not their p53 wild-type counterparts. In support of this, the increased migratory potential observed in p53 null cells following SN38 treatment was completely abrogated following cotreatment with MEK inhibitors. This suggests that following SN38 treatment in the p53 null setting, activation of the MAPK signaling pathway leads to enhanced migratory potential, which can be reversed by MEK inhibition. One of the limitations of this study is that all of the initial findings have been carried out in one isogenic cell line model system; it would be of benefit to confirm these findings in additional p53 wt and null isogenic cell line pairs. In confirmation of our results in the isogenic HCT116 cell line models, we also found that SN38 treatment increased the migratory potential of the p53-mutant DLD1 and HCC2998 cell lines in a MEK-dependent manner, whereas SN38 had no effect or actually decreased the migration of the p53 wild-type LoVo and RKO cell lines, respectively. A recent study by Muller and colleagues showed that mutant p53 can promote cell migration in vitro and in vivo through recycling of integrins to the membrane (29). We also found differential expression of integrins following SN38 treatment in the isogenic HCT116 cell line models, with a higher expression of INTGA6 and INTGA3 in p53 null compared with p53 wild-type cells. In terms of SN38 resistance, MAPK signaling pathway plays a key role in inflammatory response, differentiation, cell-cycle arrest, apoptosis, senescence, and survival (30), with many of the substrates of ERKs involved in promoting cell proliferation and survival. Previously, a number of studies have highlighted the importance of p38 MAPK signaling in regulating resistance to irinotecan therapy. Specifically, p38 has been shown to be activated in SN38-resistant colorectal cancer cells (31). In addition, pharmacologic inhibition of p38α and β was found to sensitize SN38-resistant colorectal cancer cells to treatment with SN38 (32).

In conclusion, we have used microarray profiling to identify a number of functionally relevant pathways that are differentially regulated following SN38 treatment in p53 wild-type and p53 null cells. Notably, following SN38 treatment, the focal adhesion pathway is activated in p53 null cells but downregulated in p53 wild-type cells. Furthermore, we report that a component of the focal adhesion pathway, MAPK, is involved in mediating the enhanced migratory potential of p53 null cells following SN38 treatment. We have also shown that there are differential migration rates in parental and SN38-resistant daughter cells that differ depending on p53 status. In addition, we have shown that the increased migratory potential observed in p53null/mutant cells is a unique feature of SN38 treatment as the other 2 treatments used in the management of advanced colorectal cancer, 5-FU and oxaliplatin, do not differentially effect colorectal cancer cell migration based on p53 status. Thus, targeting MAPK signaling may prove a viable strategy for enhancing the therapeutic potential of irinotecan in p53 null or p53-mutant colorectal cancer by decreasing drug-induced cell migration. Finally, in terms of personalized therapy, p53 status would represent an important prognostic biomarker for patients who receive irinotecan therapy. This will require further retrospective and prospective clinical testing. Moreover, further analyses are needed to assess the effectiveness of irinotecan-based therapy in patients with p53null/mutant tumors.

P.G. Johnston is employed by and has an ownership interest in Almac Diagnostics and Fusion Antibodies. He has received a research grant from Invest NI/McClay Foundation/QUB. He is a consultant/advisor for Chugai pharmaceuticals, Sanofi-Aventis, and on the board of the Society for Translational Oncology. He has honoraria from Chugai Pharmaceuticals, Sanofi-Aventis, and Precision therapeutics.

Conception and design: W.L. Allen, S.V. Schaeybroeck, D.B. Longley, P.G. Johnston

Development of methodology: W.L. Allen, L. Stevenson, G. Carson, S.V. Schaeybroeck, P.G. Johnston

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): W.L. Allen, R.C. Turkington, V.M. Coyle, S. Hector, P.G. Johnston

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): W.L. Allen, R.C. Turkington, V.M. Coyle, S.V. Schaeybroeck, D.B. Longley, P.G. Johnston

Writing, review, and/or revision of the manuscript: W.L. Allen, R.C. Turkington, L. Stevenson, S. Hector, P. Dunne, S.V. Schaeybroeck, D.B. Longley, P.G. Johnston

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): W.L. Allen, L. Stevenson, G. Carson, S.V. Schaeybroeck

Study supervision: W.L. Allen, S.V. Schaeybroeck, D.B. Longley, P.G. Johnston

The authors thank Cathy Fenning for technical assistance.

This work was supported by Cancer Research UK (C212/A7402, P.G. Johnston).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.