Pancreatic cancer ranks fifth among cancer related deaths in Japan and fourth in the United States.[1, 2] Recent works have led to important advances in the understanding of pancreatic cancer biology. Pancreatic cancers develop from ductal cells, acinar cells, and islet cells, and the ductal origin is the most frequent and poorest prognosis among these; we focus on tumors of pancreatic ductal origin in this review and briefly summarize recent research findings on genetics and epigenetics of pancreatic cancer, precursor lesions, and familial cases. In addition we also describe the most common genetically engineered mouse models.
By genomic and epigenomic screening techniques, substantial progress has been made in our understanding of pancreatic cancer. The comprehensive studies of the pancreatic cancer genome have revealed that most genetic alterations are identified to be associated with specific core signaling pathways including high-frequency mutated genes such as KRAS, CDKN2A, TP53, and SMAD4 along with several low-frequency mutated genes. Three types of histological precursors of pancreatic cancer: pancreatic intraepithelial neoplasia, mucinous cystic neoplasm, and intraductal papillary mucinous neoplasm, had been recognized by morphological studies and the recent genomic screening techniques revealed that each of these precursor lesions were associated with specific molecular alterations. In the familial pancreatic cancer cases, several responsible genes were discovered. Epigenetic changes also play an important role in the progression of pancreatic cancer. Several tumor suppressor genes were silenced due to aberrant promoter CpG island hypermethylation. Several genetically engineered mouse models, based on the Kras mutation, were created, and provided reliable tools to identify the key molecules responsible for the development or progression of pancreatic cancer.
High Frequency of Alterations in Driver Genes
The most frequent mutations of pancreatic ductal adenocarcinoma (PDAC) are KRAS, CDKN2A, TP53, and SMAD4.3 The first gene, KRAS, is mutated in the most pancreatic cancers, more than 95%, and is activated by point mutations mostly at codon 12.[4, 5] The KRAS gene is located at chromosome 12p12.1 and encodes a membrane-bound guanosine triphosphate (GTP) binding protein. Activated mutations of KRAS abolish the regulated GTPase activity of the KRAS protein, which results in constitutive activation of MAPK signaling cascade (Fig. 1a). In the downstream from KRAS in MAPK cascade, RAF, particularly BRAF, is activated by missense mutation of V600E, and was observed in 5% of pancreatic cancers that do not possess KRAS mutation.
The CDKN2A gene is located on chromosome 9q21.3 and encodes two tumor suppressor proteins, P16INK4A and P14ARF by sharing exons 2 and 3 with distinct reading frames. The CDKN2A gene is also inactivated in most of the PDAC cases, caused by several different mechanisms such as homozygous deletion, inactivating mutation in one allele accompanied by loss of the other allele, or promoter hypermethylation, in more than 95%.[7-10] The P16INK4A protein regulates cell cycle at the G1/S checkpoint by inhibiting cyclin D/CDK4/6 complex, which in turn inhibits RB phosphorylation. P14ARF inhibits mdm2, thus promoting TP53 functions. Because P16INK4A and P14ARF share two of three exons of CDKN2A, most of the mutations in CDKN2A inactivate both proteins; mounting evidence suggests that P14ARF also possesses several functions as a suppressor of PDAC progression.[11, 12]
The TP53 gene on chromosome 17p13.1 is inactivated in about 75% or more of pancreatic cancer cases.[3, 7] TP53 is a stress-inducible transcription factor that exerts this protective effect through the induction of either cell cycle arrest or apoptosis in damaged cells. Functional loss of the TP53 protein enables cellular survival and division in the presence of DNA damage; this facilitates the accumulation of further genetic abnormalities. Unlike most tumor suppressor genes, which are predominantly inactivated as a result of deletion or truncation-causing mutations such as nonsense mutation, frameshift mutation and splice-site mutation, inactivation of the TP53 gene typically occurs through missense mutations of one allele, accompanied with loss of the other allele. The great majority of these missense mutations are clustered in hot spot residues, mainly within the DNA binding domain. While wild-type TP53 under unstressed conditions is a very short-lived protein, these missense mutations lead to the production of full-length non-functioning TP53 protein with a prolonged half-life and result in accumulation in the nucleus, easily detected immunohisitochemically.
SMAD4 on chromosome 18q21.2 is inactivated in more than half of pancreatic cancers, about 55%, mainly by homozygous deletion.[15, 16] SMAD4 is a key mediator for TGF-β signals; binding of the TGF-β to its receptor triggers phosphorylation of SMAD2/3 to form a complex with SMAD4 and other factors to stimulate transcription of genes that normally regulate cellular growth. Loss of SMAD4 function abolishes the SMAD4-dependent TGF-β pathway and gives rise to unregulated cellular proliferation.
Jones and colleagues sequenced 20 661 protein cording genes in 24 ductal adenocarcinomas and found genetic alterations including low frequency driver gene; the most frequent mutations are represented 4 mountains involving KRAS, CDKN2A, SMAD4, and TP53, with numerous hills involving SMARC4A, CDH1, EPHA3, FBXW7, EGFR, IDH1, and NF1. They also proposed 12 core signaling pathways: K-Ras, TGFβ, JNK, Integrin, Wnt/Notch, Hedgehog, small GTPase pathways, control of G1/S phase transition, apoptosis, DNA damage control, invasion and hemophilic cell adhesion. These component genes were frequently altered in most pancreatic cancers. Biankin and colleagues analyzed 99 informative ductal adenocarcinomas and newly identified frequent and diverse somatic aberrations in axon guidance, particularly SLIT/ROBO signaling, which was also evident in murine somatic mutagenesis models of pancreatic cancer.
Three precursor lesions of ductal pancreatic cancer have been recognized (Fig. 2): pancreatic intraepithelial neoplasia (PanIN), intraductal papillary-mucinous neoplasm (IPMN) and mucinous cystic neoplasm (MCN). Conventional PDAC mainly originates from PanIN, and some investigators denote PDAC only for conventional type. However, there exist PDACs derived from sources other than PanIN, such as IPMN or MCN. Recently, atypical flat lesions derived from centroacinar-compartment are suggested to be a site at which pancreatic cancer develops in patients with a strong family history.[20, 21]
PanINs are microscopic papillary or flat noninvasive epithelial neoplasms arising in pancreatic ducts characterized by mucin-containing cuboidal to columnar cells, and grade on the basis of histological and cytological criteria. In 1976, Cubilla and Fitzgerald found that papillary proliferative lesions were more common in the pancreas with cancer than they were in the pancreas without cancer and that papillary proliferations with significant atypia were seen in the pancreas with cancer. These observations have also been confirmed by several groups.[23, 24] Although it was felt that these lesions in the pancreatic ducts were the precursors to developing infiltrating cancer, there was no direct evidence that they actually progressed to infiltrating cancer at that point. If the ductal lesions are indeed the precursors to invasive cancer, these lesions should harbor the same mutations found in their associated infiltrating cancer. In 1993, frequent KRAS mutation was reported by Yanagisawa and colleagues in mucous cell hyperplasia in patients with chronic pancreatitis; this evidence suggests the precancerous nature of these lesions. The term PanIN was introduced by David Klimstra and Daniel Longnecker in 1994. This concept suggests that all the ductal changes are thought to be precursor of PDAC. Briefly, PanIN-1 lesions have a flat (PanIN-1A) or papillary (PanIN-1B) mucinous epithelium without cellular atypia, whereas PanIN-2 lesions show increasing signs of cellular atypia and prevalence or papillary architecture. Finally PanIN-3 lesions correspond to carcinoma in situ[27-29] (Fig. 3).
To examine whether the morphologically suggested progression from PanIN-1 to PanIN-2 and -3 lesions, KRAS mutations were focused on because of their high incidence of involvement in invasive PDAC. Using a combination of microdissection and genetic studies, approximately 45% of PanIN-1 lesions were found to harbor KRAS mutation.[9, 30-32] These results suggest that one of the earliest genetic events in pancreatic intraepithelial neoplasia (PanIN) is activation of KRAS. Telomere shortening was also found in approximately 90% of PanIN-1 lesions. Loss of heterozygosity (LOH) analyses of CDKN2A, TP53, and SMAD4 revealed a rising incidence with increasing PanIN grade.[34, 35] Inactivating mutations of CDKN2A occur in PanIN-2 lesions, after KRAS activation. CDKN2A was also shown to undergo promoter hypermethylation and silencing. Immunohistochemically, Wilentz and colleagues showed that loss of expression of CDKN2A was observed in 30% of flat duct lesions without significant atypia, 55% of papillary duct lesions without significant atypia, and 71% of papillary duct lesions with significant atypia. These results suggest that functional abrogation of CDKN2A is one of the early events in pancreatic ductal carcinogenesis. Inactivation of TP53 and SMAD4 are generally associated with PanIN-3. In PanIN-3 lesions, almost as many LOHs have accumulated as the corresponding invasive carcinomas.[36, 39] These findings provided a basis for a progression model of PanINs for the development of PDAC.
A natural history of PanINs is unknown so far. Sipos and colleagues. suggested that PanIN-1 is probably of low risk for invasive PDAC and the lesions with a clearly increased risk are PanIN-2 in which some genetic changes have already been accumulated. There is a long phase between the first occurrence of a low-grade PanIN lesion and its final outcome.
Intraductal papillary mucinous neoplasms (IPMNs), demonstrated in Fig. 4, are mucin-producing epithelial neoplasms, usually with papillary architecture; they arise from the main pancreatic duct or branch ducts. IPMNs are histologically classified into four variants, including gastric, intestinal, pancreatobiliary, and oncocytic types. Activating point mutations of KRAS occur in approximately 50% of IPMNs with low-grade dysplasia, and the prevalence of KRAS mutations increases with the degree of dysplasia. Inactivation of CDKN2A and TP53 are found in IPMNs with high grade dysplasia. Loss of SMAD4 is observed only in a small subset (3%) of IPMN. Recently, GNAS mutations are reported as a frequently observed early genetic aberration in IPMNs.[44, 45] G-protein alpha-subunit (Gsα) encoded by GNAS on chromosome 20q13.32 forms a heterotrimer with β and γ G-protein subunits, which then couples with a membrane-bound receptor, GPCR. When GPCR is activated by ligand-binding, the GTP bound Gsα dissociates from the receptor and the βγ subunits, and proceeds to activate specific effector molecules including adenylyl cyclase, which produces cAMP that can act as a second messenger. Common mutations in GNAS observed in IPMNs are R201C and R201H, the same mutation as observed in endocrine neoplasms causing disruption of the intrinsic hydrolytic activity of Gsα followed by functionally constitutive activation (Fig. 1b). By unique ligation assay using oligonucleotide probes complementary to either the wild type or the mutant sequences, Wu et al. analyzed 132 IPMNs and found that 66% of IPMNs harbored GNAS mutations, 81% harbored KRAS mutations, and 51% harbored both GNAS and KRAS mutations. By direct sequencing, Furukawa et al. found that 41% of the 118 IPMNs harbored GNAS mutations, 48% harbored KRAS mutations, and 25% harbored both GNAS and KRAS mutations. The differences of frequency of this mutation might have been caused by different detection methods In invasive adenocarcinomas without association with IPMN, GNAS mutations were not found. GNAS mutations were observed both in low-grade and high-grade tumors as well as in invasive tumors; GNAS mutation probably plays an important role in initiation, rather than progression, of the pathogenesis of IPMN.
MCNs (mucinous cystic neoplasms) occur predominantly in women and associated with ovarian type stroma. In contrast to IPMNs, MCNs do not generally communicate with the pancreatic duct. The genetic alterations of MCNs have not been well defined. Studies of MCNs have indicated the prevalence of KRAS mutations and aberrant nuclear TP53 accumulation with increasing degrees of dysplasia.[48-50] SMAD4 mutation and loss of nuclear expression are not observed in most of the noninvasive MCNs, but SMAD4 expression is frequently lost when infiltrating cancers arise from MCNs; inactivation of SMAD4 is suggested to occur as the late event of neoplastic progression from MCNs. Recently, RNF43, a gene coding for a protein with intrinsic E3 ubiquitin ligase activity, was suggested to be a tumor suppressor for both IPMNs and MCNs. Wu and colleagues performed the exomic sequences, and found six of eight IPMNs and three of the eight MCNs harbored mutations of RNF43.
Hereditary Pancreatic Cancer
It has been suggested that nearly 10% of pancreatic cancer patients have family history of the disease.[54, 55] While some of the aggregation of pancreatic cancer in families is due to chance of shared environmental exposure, much of this accumulation is caused by genetic alteration. According to a consensus conference, the term familial pancreatic cancer applies to families with two or more first-degree relatives with pancreatic cancer that do not fulfil the criteria of any other inherited tumor syndromes with increased risks of pancreatic cancer. Several of the genes responsible for the familial pancreatic cancer have been discovered. BRCA2, located on chromosome 13q13.1, is probably the best characterized of all of the familial pancreatic cancer genes. The protein product of the BRCA2 gene functions to repair DNA cross-linking damage. Germline mutations in the BRCA2 gene cause familial breast cancer, and increase the risk of pancreatic cancer approximately 5.9-fold. FANCC and FANCG, which function in the same DNA repair pathway as BRCA2, have also been linked to the familial clustering of pancreatic cancer.[59, 60] The PALB2 gene, discovered by Jones and colleagues, encodes a BRCA2 associating protein and helps to localize BRCA2 to the nucleus, and aberration was found in about 3% of familial pancreatic cancer.
Roberts NJ and colleagues recently identified heterozygous ATM (ataxia telangiectasia mutated) gene mutations in two kindreds with familial pancreatic cancer, and tumor analysis demonstrated LOH of the wild-type allele. They analyzed 166 familial pancreatic cancer probands and identified four additional patients with deleterious mutations in the ATM gene.
Germline mutations in the PRSS1/TRY1 and SPINK1/PSTI genes have both been shown to cause familial pancreatitis.[63, 64] PRSS1 located on 7q34 encodes cationic trypsinogen, and mutations in this gene cause an autosomal dominant form of hereditary pancreatitis, while mutations of the serine protease inhibitor SPINK1 gene located on 5q32 cause autosomal recessive form of hereditary pancreatitis. Patients with familial pancreatitis have a 26–53-fold increased risk for developing pancreatic cancer.[65, 66] This evidence may imply that chronic inflammation sets up a cancer-prone space.
Some familial cancer syndromes are known to be associated with an increased risk of pancreatic cancer. Most of Peutz-Jergers syndrome (>80%) cases are caused by inherited mutations in the STK11/LKB1 tumor suppressor gene, on chromosome 19p13, which encodes for a serine-threonine kinase. Individuals with Peutz-Jergers syndrome have been shown to be at an increased risk of developing cancers in organs such as esophagus, stomach, small intestine, colon, lung, breast, uterus, ovary and pancreas.[67-69] Individuals with Peutz-Jeghers syndrome harbor a 132-fold increased risk of pancreatic cancer. Recent studies have suggested that pancreatic cancers in patients with Peutz-Jeghers syndrome may arise via the IPMN.[71, 72]
Germline mutations in the CDKN2A gene cause of the familial atypical multiple mole melanoma (FAMMM) syndrome. Patients with FAMMM due to CDKN2A gene mutations also have a 9 to 47-fold increased risk of developing pancreatic cancer.[73-76] The majority of mutations are located in exon 2, shared coding exon for both P16INK4 and P14ARF.
Pancreatic cancer may occur in association with other disease, such as familial adenomatous polyposis, Li-Fraumeni syndrome or hereditary non-polyposis colorectal carcinoma syndrome.
Epigenetics of Pancreatic Cancer
Research into molecular mechanisms of pancreatic cancer has revealed that the disease is due not only to genetic change but also to epigenetic change. The main epigenetic mechanisms that may affect gene expression include DNA methylation and histone modification. DNA methylation is the covalent binding of a methyl group to position 5 of cystosine residue at the CpG dinucleotide residues, catalyzed and maintained by a family of enzymes DNA methyltransferases, namely DNMT1, DNMT3A, and DNMT3B. DNMT1 is involved in preserving parental methylation patterns and transferring these patterns to offspring. DNMT3A and DNMT3B are involved in de novo methylation. A major pattern of DNA methylation occurs in CpG islands, frequently located near the transcriptional start sites of genes. Some CpG islands are methylated in tissue-specific manner. Aberrant hypermethylation of promoter CpG islands is tightly associated with gene silencing and may be associated with loss of tumor suppressor function in cancer.
Several tumor suppressor genes were shown to undergo promoter hypermethylation and silencing in pancreatic cancer such as CDKN2A. DUSP6/MKP-3 at 12q21 encodes a dual specificity phosphatase that inactivates MAPK1/ERK2. The expression of DUSP6 is often upregulated in intraductal lesions but reduced or abolished in invasive ductal carcinoma of the pancreas; this evidence suggests its role in the progression from noninvasive to invasive state.[80, 81] The loss of DUSP6 induces constitutive active MAPK1/ERK2 activity (Fig 1a), and is associated with hypermethylation of the CpG islands in intron 1 of this gene. One of the downstream targets of MAPK1/ERK2 is AURKA, and is frequently upregulated in several cancers including pancreatic cancer; if AURKA is successfully downregulated, cell growth has suppressed and chemosensitivity to taxane has increased.
MLH1 at 3p22.2, one of the DNA mismatch repair genes, and CDH1 encoding E-cadherin show aberrant methylation in a subset of pancreatic cancers. TP53 and SMAD4, and STK11, have not been shown to undergo epigenetic silencing by DNA methylation.
In addition to hypermethylation of genes functioning as tumor suppressors, hypomethylation-induced activation of oncogenic driver genes is also a commonly observed event in pancreatic adenocarcinomas. Examples of such events involve S100 genes such as S100A4 at 1q21.3 that encodes one of the subfamilies of the EF-hand Ca2+-binding protein is suggested to associate with cell proliferation, invasion, and metastasis. Hypomethylation at specific CpG sites in S100A4 gene is associated with protein overexpression, and controlling of this protein altered cell growth in association with apoptosis, cell motility, and invasion,[84, 85] and positive association with neural invasion was particularly observed. Other frequently hypomethylated genes, were also reported such as CLDN4 (claudin-4), LCN2 (lipocalin-2), SFN/14-3-3σ, TFF2 (trefoil factor 2), MSLN (emsothelin), and PSCA (prostate stem cell antigen).
Genetically Engineered Mouse Models
By using animal models, we can introduce particular genetic alteration of one of the key molecules for human diseases, and can provide manageable reliable evidences that the molecule play a role on development of the disease. Since an activating mutation of the KRAS oncogene is the most frequent genetic alteration of pancreatic cancer, activation of this pathway might be requisite. Thus most of the genetically engineered mouse models are based on the Kras mutation, as summarized in Table 1.
|Genotype||Time of expression||Time to tumor development (months)||Phenotype||Survival (months)||Reference|
|PDX-1-Cre; LSL-KrasG12D||E8.5||6||PanIN, PDAC (long latency)||16||Hingorani, Cancer Cell 2003;4:437-450|
|P48±Cre; LSL-KrasG12D||E9.5||8||PanIN, PDAC (long latency)||16||Hingorani, Cancer Cell 2003;4:437-450|
|P48±Cre; LSL-KrasG12D; LSL-Trp53R172H/-||E8.5||2–3||PanIN, PDAC||5–6||Hingorani, Cancer Cell 2005;7:469-483|
|PDX-1-Cre; LSL-KrasG12D; Smad4flox/flox||E8.5||2–3||IPMN, PanIN,||2–6||Bardeesy, Genes Dev 2006;20:3130-3146|
|PDX-1-Cre; LSL-KrasG12D; Ink4a/Arfflox/flox||E8.5||2||PanIN, pooly differentiated PDAC||2–3||Aguirre AJ, Genes Dev 2003;17:3112-3126|
Hingorani and colleagues developed a mouse model expressing physiological levels of KrasG12D by crossing mice with Cre-activated KrasG12D allele and mice expressing Cre recombinase under Pdx1 promoter or Cre knockin mouse at the Ptf1a locus (see Fig. 5a). Pdx1 and Ptf1a are critical transcription factors in the development of the pancreas (see Fig. 5b). Cre-activated KrasG12D allele was first constructed by Jackson and colleagues. Briefly, genetic inhibitory element flanked by loxP site was inserted into the mouse genomic Kras locus upstream of locus 1 to contain G-A transition at the second nucleotide of codon 12 (G12D). Mutant mice (PDX1-Cre; LSL- KrasG12D, or P48+/-Cre; LSL- KrasG12D) develop ductal lesions similar to all three stages of human PanINs. At low frequency, these lesions progress spontaneously to invasive and metastatic ductal adenocarcinoma within a year. These results suggest that Kras activation plays an important role in the triggering of initiation of PDAC carcinogenesis.
Several mouse models of invasive pancreatic cancer have been developed by combination of Kras activation and inactivation of certain tumor suppressor genes. Hingorani and colleagues generated a conditionally expressed Trp53R172H mutant, the ortholog of Li-Fraumeni syndrome. Activation of both the KrasG12D and the Trp53R172H alleles occurs in pancreatic progenitor cells through crossing with PDX-1-Cre mice, and such mice with both mutant KrasG12D and Trp53R172H develop spontaneous pancreatic ductal adenocarcinoma and die of cancer before 12 months of age.
Mice with constitutive deletion of both or either of the partially-shared P16Ink4a/P19Arf locus do not develop spontaneous pancreatic cancer. However, Aguirre generated mutant mice expressing the KrasG12D and the pancreas-specific homozygous deletion of P16Ink4a/P19Arf show highly invasive pancreatic tumor along with frequent exhibition of the epithelial-to-mesenchymal transition (EMT).
Bardeesy et al. generated a conditional knockout allele of Smad4lox and crossed with homozygous mice of either PDX1-Cre or Ptf1a-Cre transgene. These mice showed no evidence of any gross pathological abnormalities, but the combination of KrasG12D and Smad4 deficiency result in acceleration of the progression of PanIN and the rapid development of tumors resembling human intraductal papillary mucinous neoplasia (IPMN) of the gastric type. Smad4 deficiency also accelerated PDAC development of KrasG12D and P16Ink4a/P19Arf heterozygous mice. These results provide genetic confirmation that SMAD4 is a PDAC tumor suppressor, functioning to block the progression of KRasG12D -initiated neoplasms.
Genetically engineered mouse models also exert great power to identify new genes with a significant role in PDAC progression. Pérez-Mancera and colleagues used Sleeping Beauty transposon-mediated mutagenesis in a Pdx1-Cre induced KrasG12D mouse model, and identified genes that cooperated with KrasG12D to accelerate tumorgenesis and promote progression. The most commonly mutated gene was the X-linked deubiquitinase Usp9x, which is inactivated in 50% of the tumors. In humans, low USP9X expression levels correlate with poor survival, the USP9X gene may play a major role as a tumor suppressor.
Recent advances in the research on pancreatic cancer encompass three significant implementations to accelerate further development of therapy and prevention of pancreatic cancer. One of the advances is the understanding of the earliest histological and genetic changes that occur in precursor of the pancreas. The identification of molecular signatures at early stage during carcinogenesis including KRAS or GNAS mutations should enhance the trend to identify novel molecular signatures for early diagnosis during carcinogenesis. Secondly, a discovery of the pivotal genes associated with familial pancreatic cancer can provide a significant impact in modification of natural history of the disease allowing a tailor-made treatment for those at high risk through a genetic screening and counseling. Lastly, evolution of genetically engineered mouse models of pancreatic cancer carries a tremendous potential to further elucidation of the mechanisms of carcinogenesis in this lethal disease. Persistent investigation activities utilizing these models can reveal physical functions of newly identified candidate genes to disclose certain clues for molecular-targeted therapy for this life-threatening entity.
We are grateful to B. L. S. Pierce (University of Maryland University College) for editorial work in the preparation of this manuscript, and to T. Furukawa (Tokyo Women's Medical University) and F. Motoi, S. Egawa, and M. Unno (Tohoku University School of Medicine) for a fruitful discussion. This work was supported by Grants-in Aid for Scientific Research from the Ministries of Education, Culture, Sports, Science and Technology, and Health, Labour, and Welfare of Japan, Pancreas Research Foundation of Japan, and Gonryo Medical Foundation.