The p53-dependent G2/M checkpoint plays a key role in the maintenance of genomic integrity, thereby protecting cells from neoplastic progression. Reprimo, a gene involved in the p53-induced G2 cell cycle arrest, has been recently identified as a novel target for aberrant methylation in pancreatic and other cancers. The biological and clinical relevance of Reprimo methylation in pancreatic cancer was investigated.
The methylation status of Reprimo CpG island was analyzed by methylation-specific polymerase chain reaction in a large series of pancreatic cancers and was correlated with p53 mutation status, genetic instability (as measured by the fractional allelic loss), and clinicopathologic features.
Aberrant methylation of Reprimo was identified in 60% (75 of 125) of pancreatic cancer xenografts and primary pancreatic adenocarcinomas. Reprimo methylation was also detectable in 30% (19 of 63) of pancreatic intraepithelial neoplasias (PanIN), known precursors to infiltrating carcinoma. Reprimo methylation was unrelated to the p53 mutation status and associated with the increased degree of genetic instability (P = .04). Furthermore, we found that patients with Reprimo methylation in their primary pancreatic cancers have significantly worse prognosis than those without Reprimo methylation (P = .007). In contrast, other methylation targets in pancreatic cancers (SPARC and CXCR4) did not correlate with prognosis.
Impaired G2/M checkpoint machinery through genetic and/or epigenetic alterations in the p53 tumor suppressor gene as well as its downstream target genes (such as 14-3-3 sigma) plays a central role in the development of genetic instability and the pathogenesis of human cancers.1–4Reprimo is a novel member of genes involved in the p53-induced G2 cell cycle arrest.5 Ectopic expression of Reprimo results in cell cycle arrest at the G2 phase through regulating Cdc2/cyclin B1 activity.5 Loss of Reprimo expression was associated with uterine sarcoma developed in p53 transgenic mice (val135/wt) exposed to a colon carcinogen, 1,2-dimethylhydrazine.6 Furthermore, Reprimo is mapped to the chromosome 2q23, a locus frequently deleted in several cancers and endocrine tumors.7–10 These findings collectively suggest that Reprimo is a candidate tumor-suppressor gene. Using a microarray-based approach, we recently identified Reprimo as 1 of the target genes for aberrant methylation in pancreatic cancer.11 Studies from other groups have shown that aberrant methylation of Reprimo is also detectable in most types of solid as well as hematological cancers including lung, gastric, colorectal, esophageal, gallbladder, breast cancers, lymphoma, and leukemia.12, 13 Importantly, methylation, but not mutation, of Reprimo was shown to be tightly associated with loss of expression in these cancers.12, 13 These findings suggest that epigenetic inactivation of Reprimo is a common and important event in human cancers. To investigate the role of Reprimo methylation in pancreatic neoplastic progression, we analyzed a large series of pancreatic cancers for a possible association between Reprimo methylation and p53 mutation status, genetic instability, and clinicopathologic variables.
MATERIALS AND METHODS
Cells, Xenografts, and Tissue Samples
An hTERT-immortalized cell line derived from normal human pancreatic ductal epithelium (HPNE)14 was generously provided by Dr. M.M. Ouellette (University of Nebraska Medical Center, Omaha, NE). Pancreatic cancer cell lines were obtained from the ATCC (American Type Culture Collection, Rockville, MD) as previously described.11 A series of pancreatic cancer xenografts were established from surgically resected primary pancreatic carcinomas,15 and 38 xenografts were randomly selected for this analysis. The p53 gene mutation status in these xenografts was reported in previous publications.16 Archival tissues of primary pancreatic adenocarcinoma were obtained from surgical specimens resected at the Johns Hopkins Medical Institutions between 1998 and 2003. DNA extraction was done from 10-μm serial sections of formalin-fixed and paraffin-embedded tissue using the DNeasy tissue kit (Qiagen, Veronica, CA). Pancreatic intraepithelial neoplasias (PanIN) were microdissected and subjected to DNA extraction as described previously.17 The 63 PanINs analyzed in this study included 17 PanIN-1A, 19 PanIN-1B, 15 PanIN-2, and 12 PanIN-3.
The RT-PCR reaction for Reprimo was performed under the following conditions: 95°C for 5 minutes; then 32 cycles of 95°C for 20 seconds, 63°C for 20 seconds, and 72°C for 30 seconds; and a final extension of 4 minutes at 72°C. Primer sequences were 5′-GGA AGA GTG ACC TGT TAA AAG-3′ (forward) and 5′-AGA AAA AGA CTT ATC AAA CAT TTG-3′ (reverse). To check the integrity of mRNA, glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was also amplified using the same amount of cDNA.
Methylation-Specific PCR (MSP)
DNA samples were treated with sodium bisulfite (Sigma, St. Louis, MO) and purified with the Wizard DNA clean-up system (Promega, Madison, WI). One μL of the bisulfite-treated DNA (resuspended in 50 μL TE buffer) was amplified using primers specific for either the unmethylated or for the methylated DNA. Primer sequences for Reprimo MSP were TTG TGA GTG AGT GTT TAG TTT G (forward) and TAA TTA CCT AAA ACC AAA TTC ATC (reverse) for unmethylated PCR and GCG AGT GAG CGT TTA GTT C (forward) and TAC CTA AAA CCG AAT TCA TCG (reverse) for methylated PCR. PCR conditions were as follows: 95°C for 5 minutes; then 40 cycles of 95°C for 20 seconds, 60°C for 20 seconds, and 72°C for 30 seconds; and a final extension of 4 minutes at 72°C.
Fractional Allelic Loss
The degree of genetic instability was measured in pancreatic cancer xenografts by the fractional allelic loss (FAL, the percent of markers deleted/markers analyzed) using 386 microsatellite markers, as described previously.18
Differences in frequencies of Reprimo methylation or p53 gene mutation between groups were examined using Fisher exact probability test. Difference in FAL was assessed by 2-tailed Student t-test. Survival probability was analyzed according to the Kaplan–Meier method, and the difference in their distribution was evaluated by the log rank test. Cox proportional hazards model was also used to determine the prognostic value for various factors in univariate and multivariate analysis.
Methylation Analysis of Reprimo in Pancreatic Cancer and Its Precursor Lesions
By screening for genes that are up-regulated by pharmacologic modifications of epigenetic status in pancreatic cancer cells, we identified Reprimo as 1 of the genes aberrantly methylated in pancreatic cancer.11
We first examined if aberrant methylation of Reprimo is associated with its transcriptional repression in pancreatic cancer. Using RT-PCR, we found that Reprimo mRNA expression was present in HPNE (a cell line derived from normal pancreatic duct) and 2 pancreatic cancer cell lines (CFPAC1 and Su8686) with unmethylated Reprimo, whereas the expression was weak or absent in 9 pancreatic cancer cell lines with methylated Reprimo (Fig. 1A). Reprimo expression in BxPC3 was robustly increased upon treatment with a methyltransferase inhibitor (5-aza-2′-deoxycytidine) and histone deacetylase inhibitor (trichostatin A) (Fig. 1A), suggesting that Reprimo transcription is regulated at least in part through epigenetic mechanisms.
To determine the extent of abnormal methylation of Reprimo in pancreatic cancer, we used MSP to examine the methylation status of Reprimo in a large panel of normal pancreatic tissues (n = 10), pancreatic cancer xenografts (n = 38), and in surgically resected primary pancreatic adenocarcinomas (n = 87). The Reprimo CpG island was unmethylated in all the normal pancreatic tissues examined, whereas aberrant methylation was detected in 25 (66%) of 38 xenografts and in 50 (57%) of 87 primary pancreatic cancers (Fig. 1b) (Table 1).
Table 1. Methylation Profile of Reprimo in Pancreatic Cancers and PanINs
For comparison, we also examined the promoter methylation of another G2/M (mitotic) checkpoint gene, CHFR, which has been demonstrated to undergo methylation-associated silencing in various cancers.19 Aberrant hypermethylation of CHFR was found in only 1 (4%) of 23 pancreatic cancer xenografts examined (data not shown), which is in striking contrast to the high prevalence of Reprimo methylation in these cancers.
To determine the timing of aberrant Reprimo methylation during the development of pancreatic cancer, we then examined the Reprimo methylation in different grades of PanINs, the precursor to infiltrating pancreatic ductal adenocarcinoma. Overall, we were able to detect aberrant methylation of Reprimo in 19 (30%) of 63 independent PanIN lesions selectively microdissected from resected pancreata (Fig. 1c). Reprimo methylation was present even in the earliest PanIN lesions (PanIN-1A), although they showed weaker methylation signals than the advanced stages of PanINs (PanIN-3) (Fig. 1c). The prevalence of Reprimo methylation increased significantly with increasing PanIN grade; Reprimo methylation was present in 11 (22%) of 51 early (low-grade) PanIN lesions (PanIN-1/2) and in 8 (67%) of 12 advanced (high-grade) PanINs (PanIN-3) (P = .004) (Table 1). Thus, these findings suggest that aberrant methylation of Reprimo can occur only in a minority of early PanINs but becomes a common feature in PanIN-3.
Relation between Reprimo Methylation and p53 Mutation Status or Genetic Instability in Pancreatic Cancer
Because Reprimo is a transcriptional target of the p53 tumor suppressor, we first looked at the relation between aberrant methylation of Reprimo and p53 genetic alteration in pancreatic cancer. The p53 mutation status was available from 23 pancreatic cancers (including 7 pancreatic cancer cell lines [AsPC1, BxPC3, Capan1, Capan2, CFPAC1, MiaPaCa2, and Panc1] and 16 pancreatic xenografts) with known Reprimo methylation status.16, 20 Overall, 17 (74%) of 23 pancreatic cancers harbor p53 mutations, but there was no correlation between Reprimo methylation and p53 gene status in these cancers; Reprimo methylation was found in 14 (82%) of 17 cancers containing p53 mutation and in 3 (50%) of 6 cancers with wildtype p53 (difference not significant, P = .3). Thus, although the number of cases analyzed is small, it appears that Reprimo methylation occurs independently of the p53 gene alterations.
Loss of checkpoint gene function through either genetic or epigenetic mechanisms has been considered 1 of the mechanisms responsible for the development of genetic instability in cancer. We therefore examined the correlation between Reprimo methylation and genetic instability in a total of 38 pancreatic cancer xenografts. Genetic instability in each xenograft was measured by FAL by using 386 microsatellite markers, which represents the proportion of genomic deletions across the whole genome.18 The FAL was significantly higher in 25 xenografts with methylated Reprimo than in 13 xenografts with unmethylated Reprimo (17.9% vs. 13.3%, P = .04) (Fig. 2). Thus, Reprimo methylation is associated with an increased magnitude of genetic instability in pancreatic cancer.
Relation between Reprimo Methylation and Clinicopathologic Variables in Pancreatic Cancer
We then analyzed a large series of 87 primary pancreatic cancers (of which 50 [57%] were methylated for Reprimo, as described above) for the correlation between Reprimo methylation and clinicopathologic variables, including patient age, gender, tumor size, tumor differentiation (poorly differentiated or well/moderately differentiated), the presence of lymph node metastasis, positive or negative surgical margins, and overall prognosis after surgical resection. There was no difference in age, gender, tumor size, tumor differentiation, positive lymph nodes, or positive surgical margins between patients with Reprimo methylation and those without Reprimo methylation (Table 2); however, the overall prognosis in patients with methylated Reprimo in their cancers was significantly poorer than in those without methylation (P = .007, log rank test) (Fig. 3A).
Table 2. Characteristics of Patients with Pancreatic Cancer with or without Reprimo Methylation
Reprimo methylation (−)
Reprimo methylation (+)
64.6 ± 11.1
68.6 ± 11.3
66.9 ± 11.3
Tumor size, cm
2.7 ± 0.8
3.0 ± 1.1
2.8 ± 1.0
Well or moderately
Lymph node status
To further determine the prognostic value of Reprimo methylation for pancreatic cancer, clinicopathologic variables as well as Reprimo methylation status were analyzed using Cox proportional hazards model. The clinicopathologic variables analyzed were age, tumor size, tumor differentiation (poor vs. well to moderate), positive lymph node metastasis, and positive surgical margins. In a univariate model, factors significantly associated with poor prognosis included Reprimo methylation (hazard ratio, 2.2; P = .009), tumor size of 3 cm or greater (hazard ratio, 2.8; P = .0004), and poor differentiation (hazard ratio, 3.0; P = .0001). In a multivariate analysis, tumor size (hazard ratio, 2.4; P = .004) and poor differentiation (hazard ratio, 2.4; P = .003) were found to be independent prognostic factors, and patients whose cancers had Reprimo methylation tended to have a poorer prognosis (hazard ratio, 1.8; P = .07).
To test whether this prognostic significance was specific for Reprimo, we also analyzed the same series of primary pancreatic cancers for methylation status of 2 genes (SPARC and CXCR4) that we previously identified as aberrantly methylated in pancreatic cancer.21, 22 Using MSP, aberrant methylation was detectable in 62% (54 of 87) for SPARC and 62% (54 of 87) for CXCR4; however, we found no significant correlation between their methylation status and prognosis in this series of patients. For example, although CXCR4 methylation was detected at a frequency (∼60%) similar to Reprimo methylation, there was no difference in overall survival between patients with CXCR4 methylation and those without CXCR4 methylation (P = .6, log rank test) (Fig. 3B).
In the present study we investigated the biological and clinical relevance of aberrant hypermethylation of Reprimo in pancreatic cancer. The major findings in this study were 1) aberrant methylation of Reprimo was frequently (∼60%) found in primary pancreatic adenocarcinomas; 2) Reprimo methylation was also detectable commonly in late PanINs; 3) Reprimo methylation was associated with increased degree of genetic instability; and 4) Reprimo methylation correlated with poor prognosis in a large series of resected pancreatic cancers. These results raise the possibility that aberrant methylation of Reprimo is an epigenetic event in pancreatic cancer that has a mechanistic role in neoplastic progression.
The mechanism of how Reprimo methylation occurs during pancreatic carcinogenesis is unknown. It has been proposed that methylation change in cancer can be a secondary event that may occur as a consequence of genetic or other events, such as loss of transcription factor(s), that alter the transcriptional activities of affected promoters. For example, an oncogenic transcription factor promyelocytic leukemia (PML) retinoic acid receptor (RAR) fusion protein has been shown to induce hypermethylation and silencing of its target gene, RAR 2, suggesting that genetic event can lead to DNA methylation.23 In addition, a recent study has shown that loss of estrogen receptor (ER) expression by RNA interference results in silencing of downstream target genes, followed by a progressive accumulation of DNA methylation in their promoter CpG islands.24 It is possible, therefore, that loss of p53 function could induce silencing and subsequent methylation of its transcriptional target gene Reprimo in pancreatic cancer. However, this is unlikely because we found no significant correlation between Reprimo methylation and p53 gene mutation status in pancreatic cancers. In agreement with our finding, a previous study showed no correlation between Reprimo methylation and p53 mutation in 20 lung cancer cell lines.12 We also demonstrate that Reprimo methylation is present, albeit at low levels, even in the earliest precursors (PanIN-1A) in which p53 mutations have not been identified.25 Thus, aberrant methylation of Reprimo may occur in relatively early stages of pancreatic carcinogenesis, irrespective of the p53 gene inactivation, most likely as a part of a genome-wide process of CpG island hypermethylation.
The G2/M cell cycle checkpoint is tightly and cooperatively regulated by transcriptional targets of p53, including p21, 14-3-3 sigma, Gadd45, and Reprimo.2, 26 Because the G2 checkpoint is critical to maintain genomic stability, inactivation of these checkpoint mediators could lead to or accelerate genetic instability. For example, embryo fibroblasts from Gadd45a-null mice exhibited genetic instability characterized by aneuploidy and centrosome amplification.27 Similarly, cells lacking 14-3-3 sigma expression showed increased genomic instability associated with loss of telomeric repeat sequences, chromosome end-to-end fusions, and nonreciprocal translocations.28 These findings raise the possibility that epigenetic silencing of Reprimo could contribute to the development of chromosomal instability in cancer. In fact, we demonstrate that abnormal Reprimo methylation is significantly associated with increased levels of genetic instability (FAL) in pancreatic cancer xenografts. Interestingly, 14-3-3 sigma has been shown to be silenced by hypermethylation in several types of cancers such as breast and prostate cancers,4 but was found to be overexpressed and hypomethylated in the majority of pancreatic cancers.29, 30 We also demonstrate that a well-characterized checkpoint gene, CHFR, is rarely (<5%) methylated in pancreatic cancer. These observations suggest cancer-type specific alterations in DNA methylation targeting these G2/M checkpoint genes, and epigenetic silencing of Reprimo could be an important pathway leading to defects in the cell cycle control and genetic instability in pancreatic cancer.
In the present study, we demonstrate that aberrant methylation of Reprimo, but not of other genes, is a predictor of poor outcome in patients undergoing resection for pancreatic cancer. Several previous studies have found that methylation of individual genes had prognostic significance, such as DAPK in lung cancer,31CDH1 (E-cadherin) in diffuse gastric cancer,32 and PGP9.5/UCHL1 in esophageal cancer.33 As with these genes, the correlation between Reprimo methylation and poor prognosis does not prove that Reprimo methylation directly contributes to the poor prognosis. It is still possible that methylation of Reprimo reflects a biologically distinct group of cancers (for example, CpG island methylator phenotype or CIMP) that is not necessarily related to the function of this gene. Nevertheless, our present results suggest a potential clinical use of Reprimo methylation as a prognostic marker in patients with pancreatic cancer.
In summary, we demonstrate that Reprimo methylation is a frequent event during pancreatic ductal carcinogenesis, and might contribute to genetic instability and tumor progression.
We thank Dr. Scott E. Kern (Johns Hopkins Medical Institutions, Baltimore, MD) for providing DNA samples from pancreatic cancer xenografts. Dr. Goggins and Dr. Sato have entered into licensing agreements with Oncomethylome Sciences, who wish to develop commercial products related to the Reprimo methylation assay used in this study.