Mist1-Kras^G12D Knock-In Mice Develop Mixed Differentiation Metastatic Exocrine Pancreatic Carcinoma and Hepatocellular Carcinoma

As pancreatic ductal adenocarcinoma (PDA) is among the most lethal human malignancies, the identification of molecular and cellular pathways important in PDA development is essential for the formulation of rational approaches to prevention, diagnosis, and therapy. Accordingly, mutant mouse strains that recapitulate certain aspects of PDA have been generated to assess the contribution of various common genetic lesions identified in patients, including Kras, p53, and p16Ink4a. Indeed, mice that express the endogenous Kras^G12D allele throughout the developing and adult pancreas develop preinvasive pancreatic intraepithelial neoplasias (PanIN) and eventually succumb from invasive and metastatic PDA (1). This process is substantially accelerated by the biallelic deletion of Ink4a/ARF (2) or by coexpression of an endogenous Trp53^R172H allele (3). Although these approaches show that PanIN and PDA can be modeled in mice, the contribution of tissue progenitor cells as opposed to mature cell types is uncertain. To determine which specific cellular compartments may initiate the development of PanIN and PDA in response to oncogenic Kras, transgenic mice have been generated that ectopically express oncogenic Kras from either mature ductal or acinar promoters. Interestingly, the expression of Kras^G12V from the ductal-specific cytokeratin 19 promoter induced a periductal lymphocytic infiltrate in the absence of obvious ductal hyperplasia or dysplasia (4). Additionally, the directed expression of Kras^G12D from the rat elastase promoter produced nonlethal pancreatic acinar hyperplasia and dysplasia accompanied by infrequent preinvasive acinar adenomas and cystic neoplasms (5). These findings show that ectopic oncogenic Kras expression is unable to promote advanced pancreatic cancer when expressed in mature ductal or mature acinar epithelial cells. However, these models do not preclude the importance of less differentiated cell types of either lineage in mediating this process.

To further investigate whether the ectopic expression of oncogenic Kras could be used to identify a cellular compartment capable of initiating PanIN and PDA, we used homologous recombination to target the expression of Kras^G12D to the Mist1 locus, a gene known to be expressed at earlier stages of pancreatic exocrine development. Mist1 is a basic helix-loop-helix transcription factor that is expressed at low levels in the embryonic pancreas at day 10.5 as determined by a β-galactosidase knock-in reporter into the Mist1 locus (6–8). In the adult, Mist1 protein is restricted to mature pancreatic acinar cells with no measurable expression in mature ductal or islet cells (6, 9, 10). Mist1 is required for proper acinar architecture as Mist1^−/− mice develop progressive acinar dysplasia and pancreatic fibrosis, whereas Mist1^+/− mice have no discernable phenotype (6). We describe here that Mist1^KrasG12D/+ mice develop lethal metastatic pancreatic cancer with mixed histologic features and find that Trp53 mutation exacerbates the clincopathologic manifestations of this disease. Therefore, Mist1-expressing cells may participate in the genesis of pancreatic cancer.

Generation of the Mist1^KrasG12D/+ mice. The Kras4B^G12D cDNA (11) was cloned in-frame downstream of the initiating methionine in the first coding exon (exon 2) of a Mist1 genomic clone. A similar strategy was previously used to express β-galactosidase from the Mist1 promoter when the Mist1 gene was disrupted (6). TC1 embryonic stem cells (12) were electroporated with 40 μg of targeting vector previously linearized with NotI. Southern blot analysis of BamH1-digested embryonic stem cell DNA and probed with a 3′ external probe revealed that 12 of 33 embryonic stem cell clones contained a properly integrated construct. Two independent embryonic stem cell clones were used for the production of chimeric mice and the germ line phenotypes of each mouse were identical. Trp53^+/− mice have been previously described (13). All mice analyzed were F₁ generation on a 129Sv/C57Bl/6 background. Genotyping of Mist1^KrasG12D/+ mice was accomplished by amplifying tail genomic DNA with primers specific to the Kras4B cDNA, using forward primer 5′-AgggAATAAgTgTgATTTgCC-OH and reverse primer 5′-AAACTgCAgTCACATAACTgTACACCTTgTC-OH and 35 cycles of amplification at 94°C/58°C/72°C to yield the expected 400 bp fragment.

Statistical analyses. Kaplan-Meier curves were computed using the survival time for each mouse. The log-rank test was used to test for significant differences between the four groups of mice (14). Based on Bonferroni's correction for multiple comparisons, individual P values for three comparisons (e.g., the Mist1^KrasG12D/+ survival curve with each of the other three survival curves) of <0.0167 were considered statistically significant.

Histologic and pathologic analyses. Standard histologic analyses were done with sections stained with H&E or Alcian blue. Pathologic nomenclature used followed the “Pathology of genetically engineered mouse models of pancreatic cancer: consensus report and recommendations.”⁷

Immunohistochemical analyses. Frozen and paraffin-embedded pancreatic sections were processed for immunohistochemistry by standard procedures. For paraffin immunohistochemistry, sections were deparaffinized and rehydrated, and antigens were retrieved using the 2100-Retriever (PickCell Laboratories, Amsterdam, the Netherlands) and Antigen Unmasking Solution (Vector Laboratories, Burlingame, CA). For cytokeratin 19 immunolabeling, antigen retrieval was done by digesting sections with 250 μg/mL proteinase K in 2.5 mmol/L CaCl₂ and 10 mmol/L Tris-HCl (pH 7.5) for 6 minutes at room temperature. Samples were washed with PBS and blocked using the MOM blocking reagent. Primary antibodies were incubated for 1 hour at room temperature or 4°C overnight. Primary antibodies included rabbit amylase (1:100, Calbiochem, San Diego, CA), rabbit Mist1 (1:2,000), rabbit connexin32 (1:200, Zymed Laboratories, San Francisco, CA), mouse Ki67 (1:100, Novocastra, Newcastle Upon Tyne, United Kingdom), and rat cytokeratin 19 (TROMA 3, 1:100, gift of Rolf Kemler, Department of Molecular Embryology, Max-Planck Institute, Freiburg, Germany). Following the addition of primary antibody, sections were incubated with biotinylated (1:200, Vector Laboratories) or Oregon green–conjugated (1:200, Molecular Probes, Eugene, OR) secondary antibodies for 20 minutes at room temperature followed by incubation with Alexa 594–conjugated tertiary antibodies (1:400, Molecular Probes) and 4′,6-diamidino-2-phenylindole (DAPI, Sigma, St. Louis, MO) for 5 minutes.

Protein immunoblot assays. Tissue protein extraction, protein electrophoresis, and immunoblotting were done as described previously (6, 9). For immunoblot analysis, 50 μg of whole cell protein extracts were electrophoresed on acrylamide gels, transferred to polyvinylidene difluoride membranes, and incubated with primary antibodies against rabbit Mist1 (1:1000), mouse β-gal (1:500, Developmental Studies Hybridoma Bank, Iowa City, IA), goat amylase (1:2,000, Santa Cruz, Santa Cruz, CA), rabbit Akt, phospho-Akt, p44/42 mitogen-activated protein kinase (MAPK), phospho-p44/42 MAPK (1:1,000, Cell Signaling, Beverly, MA), mouse pan-Ras (1:40, Calbiochem), goat clusterin (1:1,000, Santa Cruz), and cytokeratin 19. Following secondary antibody incubation, the immunoblots were developed using an ECL kit (Pierce, Rockford, IL) as per instructions of the manufacturer.

RNA expression analysis. Total RNA (1 μg) was reverse transcribed using the iScript cDNA synthesis kit (Bio-Rad, Hercules, CA). cDNA reactions were amplified using Taq polymerase and gene specific primers for Hes1 (5′-TCTACACCAGCAACAGTG-3′, 5′-TCAAACATCTTTGGCATCAC-3′), Hey1 (5′-GCGGACGAGAATGGAAACTTG-3′, 5′-GCTCAGATAACGGGCAACTTCG-3′), and Hey2 (5′-TGAGCATTGGATTCCGAGAGTG-3′, 5′-ATACCGACAAGGGTGGGCTGATTG). Target sequences were amplified within linear ranges using 95°C/40 seconds, 55°C/40 seconds, and 72°C/55 seconds conditions.

Mist1^KrasG12D/+ mice have reduced life spans. Properly targeted embryonic stem cells were used to generate Mist1^KrasG12D/+ mice (Fig. 1B and C). Mist1^KrasG12D/+ mice were generated from high-contribution male chimera donors, and germ line animals showed expected Mendelian inheritance. Although there was no obvious defect in early development, carriers (n = 44) were found to have a diminished median survival of 10.8 months compared with littermate control wild-type animals with a median survival of 24.2 months (P < 0.001; Fig. 1D). A smaller cohort of Mist1^KrasG12D/+; Trp53^+/− mice (n = 12) was produced to determine the effects of Trp53 deficiency on this background, and the 6.5-month median survival of such mice was substantially reduced relative to the 10.8- and 11.7-month median survival for each genetic trait separately.

Mist1^KrasG12D/+ mice develop acinar-ductal metaplasia. Histologic analysis of pancreata obtained from asymptomatic newborn or adult Mist1^KrasG12D/+ mice showed diffuse acinar hyperplasia (Fig. 2B) with an increased proliferative index (not shown). Acinar adenomas of solid or cystic nature were obvious in 2-month-old mice (Fig. 2B and C). Metaplastic ductal structures with mucinous cytoplasm that resembled murine PanIN-1a were found scattered throughout the pancreas in close association with metaplastic acini (Fig. 2D). These metaplastic ducts were characterized by the presence of the ductal protein cytokeratin 19 and acidic mucin staining with Alcian blue (not shown). Based on the recent Pancreatic Pathology Nomenclature Consensus Report, such lesions are not designated murine PanIN-1a but rather “acinar-ductal metaplasia” due to the diffuse background of metaplastic acini.

Mist1^KrasG12D/+ mice die of advanced pancreatic exocrine carcinoma. Mist1^KrasG12D/+ mice became cachectic and presented frequently with hemorrhagic ascites, starting at 3 months of age. Gross necropsy showed large pancreatic primary tumors with visible metastases studding the visceral organs and the i.p. space. The primary tumors and metastases were solid and cystic in nature and up to several centimeters in diameter, and some likely contributed to the decreased survival of the mutant mice due to bowel strangulation. The predominant histologic subtype of pancreatic cancer noted was invasive acinar cell carcinoma across the spectrum of well-differentiated to poorly differentiated high-grade lesions (Fig. 3A and data not shown). Additionally, cystic neoplasms that were papillary with acinar cell features were commonly noted (Fig. 3B and C). Based on the new nomenclature, these would be designated “cystic papillary neoplasms with acinar differentiation without ovarian stroma.” Classic glandular ductal adenocarcinomas were only rarely observed (Fig. 3D) but several examples of mixed carcinomas with acinar and ductal features were noted (Fig. 3A). The invasive carcinomas and metastases were frequently accompanied by a rich collagenous stroma and accompanying fibroblasts, typical of the desmoplastic reaction present in human PDA.

Mist1^KrasG12D/+ mice develop early and advanced hepatocellular carcinoma. At necropsy, the Mist1^KrasG12D/+ mice oftentimes had hepatomegaly with macroscopic nodules. Histologic analysis showed multifocal hepatocellular adenomas (Fig. 4A and B) and carcinomas (Fig. 4C and D) in the majority of Mist1^KrasG12D/+ mice (n = 25 of 44), whereas only one wild-type control mouse (n = 1 of 16) developed hepatocellular carcinoma (Fisher's exact test, two-sided, P < 0.001). Furthermore, three of six Mist1^KrasG12D/+ mice that succumbed before developing invasive pancreatic carcinoma had a significant burden of hepatocellular carcinoma. The increased incidence of hepatocellular carcinoma may be due to the expression of the Mist1^KrasG12D knock-in gene in the developing foregut endoderm (6), although Mist1 is not known to be expressed in hepatocytes. Alternatively, the local and systemic effects of pancreatic neoplasia in Mist1^KrasG12D/+ mice may induce hepatic injury and initiate hepatic neoplasia. Consistent with this possibility, eosinophilic (Fig. 4B), basophilic, and clear cell hepatocyte foci were evident in the livers of Mist1^KrasG12D/+ mice, and often abutted hepatic adenomas (Fig. 4B) and frank hepatocellular carcinoma.

Trp53^+/− cooperates with Mist1^KrasG12D/+ to promote advanced histologic findings and metastasis. In addition to decreased survival (Fig. 1D), the number of mice that presented with metastatic disease to any site was slightly higher in Mist1^KrasG12D/+; Trp53^+/− mice (9 of 12) compared with Mist1^KrasG12D/+ mice (18 of 44; Fisher's exact test, two-sided, P = 0.051; Supplementary Table S1). An examination of the primary pancreatic tumors and metastases showed that in addition to the diverse histologic findings in Mist1^KrasG12D/+ mice (Figs. 2 and 3), compound mutant Mist1^KrasG12D/+; Trp53^+/− mice showed advanced pathologic features, such as pleomorphic carcinomas with ductal characteristics (Fig. 5A) and undifferentiated carcinomas with tumor giant cells (Fig. 5B). Interestingly, upon close examination, none of the Mist1^KrasG12D/+ mice developed parenchymal liver metastases, whereas 5 of 12 of the Mist1^KrasG12D/+; Trp53^+/− mice showed this finding (Fisher's exact test, two-sided, P < 0.001; Fig. 5C). Furthermore, lung metastases were only noted in the Mist1^KrasG12D/+; Trp53^+/− mice (n = 2 of 12, Fisher's exact test, two sided, P = 0.042). Consistent with loss of heterozygosity at the Trp53 locus, semiquantitative PCR of genomic DNA showed reduction in the intensity of the wild-type Trp53 allele in macrodissected pancreatic tumors obtained from several Mist1^KrasG12D/+; Trp53^+/− mice (Fig. 5D , lanes 4-6 and data not shown), despite the contamination by a significant fraction of stromal cells. Similar observations of promoting hematogenous metastases have been previously reported for a murine model of hepatocellular carcinoma (15). Nonpancreatic neoplasms, particularly hepatocellular carcinoma, were much less common in Mist1^KrasG12D/+; Trp53^+/− mice (n = 1 of 12 mice), possibly reflecting the shortened life spans of the compound mutant mice compared with the parental Mist1^KrasG12D/+ strain.

Biochemical pathway usage in pancreata of Mist1^KrasG12D/+ mice. Pancreatic tissue lysates from Mist1^KrasG12D/+ mice revealed elevated levels of Ras protein and the active phosphorylated forms of the Ras effectors extracellular signal-regulated kinase and AKT (not shown). Additionally, Ras-GTP levels were elevated (not shown) and a well-described Ras effector target gene, cyclin D1, was present at higher levels, consistent with its role in cellular proliferation. Activation of the notch signaling pathway, a prominent feature of human and mouse PDA (16), was also evident by the presence of mRNA representing the Notch target genes Hes1, Hey1, and Hey2 (not shown).

Pronounced reduction in Mist1 protein in Mist1^KrasG12D/+ mice. Pancreatic lysates prepared from wild-type, Mist1^+/−, and Mist1^KrasG12D/+ mice showed unexpectedly decreased levels of Mist1 protein in Mist1^KrasG12D/+ pancreata and a corresponding increase in the immature acinar cell protein clusterin and the duct cell–specific protein cytokeratin 19 (Fig. 6A). A similar reduction in Mist1 nuclear accumulation was also observed by immunohistochemistry (Supplementary Fig. S1A-D), with little Mist1 protein detectable in areas of acinar metaplasia (Supplementary Fig. S1D). To evaluate whether Mist1 function was compromised in the acinar compartment of Mist1^KrasG12D/+ mice, the presence of gap junctions was assessed by performing immunohistochemistry with antibodies against connexin32, the major connexin protein found in pancreatic acinar gap junctions (17, 18). Consistent with the observation that Mist1 levels were markedly depressed in Mist1^KrasG12D/+ pancreata, gap junctions were also absent in Mist1^KrasG12D/+ acini but not in control Mist1^+/− acini (Fig. 6B and C).

Previous work with oncogenic Kras mutant mice has led to the proposal that PanIN/PDA may develop in chiefly two different manners: the dedifferentiation/transdifferentiation of mature pancreatic epithelial cell types into neoplastic cells with ductal features or the emergence of neoplastic PanIN/PDA clones from a tissue progenitor or stem cell compartment. The evidence that supports the former is the ectopic expression of Kras^G12D from the rat elastase promoter, where progressive acinar metaplasia precedes the emergence of preinvasive and nonlethal acinar and cystic adenomas in a minority of mice by 1 year of age (5).⁷ Mouse models that support the progenitor cell hypothesis of pancreatic carcinoma genesis include those in which an endogenous Kras oncogene is expressed in the earliest stages of pancreatic development, and therefore in all cells of the mature organ (1–3). In these models, acinar metaplasia is not a prominent feature and thus acinar-to-ductal transdifferentiation is not required for PanIN/PDA formation.

Here, we have ectopically expressed Kras^G12D from the endogenous Mist1 promoter and find that Mist1^KrasG12D/+ mice develop acinar and ductal hyperplasia and metaplasia with complete penetrance before the appearance of invasive pancreatic carcinoma with mixed histologic features. Therefore, the presence of superimposed acinar metaplasia and mixed lineage carcinoma suggests that both acinar dedifferentiation and progenitor cell proliferation may contribute to pancreatic tumorigenesis in Mist1^KrasG12D/+ mice.

In addition to pancreatic carcinomas, Mist1^KrasG12D/+ mice frequently developed hepatocellular carcinomas. Although the etiology of hepatocellular carcinoma formation in Mist1^KrasG12D/+ mice is currently unknown, it could reflect the cellular plasticity of Mist1^KrasG12D/+-expressing hepatic progenitor cells. Indeed, “hepatopancreatic stem cells” with the capability of multilineage differentiation are hypothesized to exist in adult hepatic and pancreatic tissues (19). Despite the lack of obvious expression of the Mist1 promoter in embryonic or adult hepatic tissue of Mist1^{β-gal/β-gal} mice (6), rare progenitor or stem cells may express Mist1 and could account for hepatocellular carcinoma in Mist1^KrasG12D/+ mice. Alternatively, the use of β-galactosidase activity as an indirect measure of Mist1 promoter function may be insufficient to exclude a low level of Kras^G12D expression in hepatocytes of Mist1^KrasG12D/+ mice that may be biologically relevant and sufficient to initiate tumor formation in the liver. The alternative hypothesis that chronic hepatocyte damage is induced by pancreatic disease in Mist1^KrasG12D/+ mice is equally plausible, given the presence of hepatocyte foci in histologic examination (Fig. 4). Importantly, the hepatocellular carcinoma phenotype is not fully penetrant in Mist1^KrasG12D/+ mice, and even among mice with hepatocellular carcinoma, pancreatic cancer is the most likely cause of death.

Several factors may explain the more severe phenotype of Mist1^KrasG12D/+ mice relative to that of Ela-Kras^G12D mice. First, oncogenic Kras levels profoundly influence the biological properties of primary cells (11, 20), and the Mist1^KrasG12D/+ strain uses a Kras^G12D knock-in allele that may be expressed more uniformly and at different levels than the transgenic Ela-Kras^G12D allele. Second, the Mist1 promoter is expressed earlier in mouse development at embryonic day 10.5 (6, 9, 10) compared with elastase expression which occurs at embryonic day 14 (Fig. 1A; refs. 21–24). By extension, Mist1 may thus be expressed in adult progenitor cells in contrast to the elastase gene.

Additionally, Mist1 haploinsufficiency may play an important role in promoting carcinogenesis in Mist1^KrasG12D/+ mice in two distinct manners. First, previous work has shown that Mist1^{β-gal/ β-gal} acinar cells that lack Mist1 protein become progressively dedifferentiated and metaplastic, with loss of gap junctions (6, 17). Indeed, we have shown that embryonic and adult Mist1^KrasG12D/+ pancreata contain unexpectedly low levels of the Mist1 protein and increased levels of the immature acinar cell protein clusterin and the mature ductal cell protein cytokeratin 19 (Fig. 6A; ref. 18), and that metaplastic regions of Mist1^KrasG12D/+ pancreata phenocopies the Mist1 null state with respect to the loss of expression of the gap junction protein connexin32 (Fig. 6B and C). Second, depressed Mist1 levels in Mist1^KrasG12D/+ mice may also result in the expansion of pancreatic progenitor cells because Mist1^{β-gal/ β-gal} pancreata contain cells that coexpress markers of mature ductal and acinar cells (6). These “double positive” or biphenotypic cells in Mist1^{β-gal/ β-gal} mice pancreata have also been identified in Mist1^KrasG12D/+ pancreata (not shown). We are currently pursuing the molecular basis of depressed Mist1 levels in Mist1^KrasG12D/+ pancreata and the potential importance of biphenotypic cells in promoting carcinogenesis in Mist1^KrasG12D/+ mice. As Trp53 loss promotes carcinogenesis and hematogenous metastasis in Mist1^KrasG12D/+ mice (Fig. 5), it will be of interest to determine whether Mist1 levels are more severely depressed and the biphenotypic cell population is expanded in this genetic background. Finally, we are directly evaluating whether the genetic deficiency of Mist1 promotes the emergence of PanIN/PDA in our models of endogenous pancreatic Kras^G12D expression.

In summary, we describe a new model of Kras^G12D-dependent metastatic exocrine pancreatic cancer and hepatocellular carcinoma that for the first time implicates the Mist1 transcription factor and Mist1-expressing cells in tumorigenesis. It is currently not possible to separate the potential contributions of acinar-to-ductal metaplasia from the expansion of a Mist1⁺ progenitor cell compartment in Mist1^KrasG12D/+ mice in the promotion of pancreatic neoplasia because of the confounding continuous dependence of Kras^G12D gene expression upon Mist1 promoter activity. Indeed, further experiments that use strategies to express Cre recombinase from the Mist1 promoter exclusively in either immature/developing cells or mature acinar cells will be needed to directly evaluate the ability of the endogenous conditional LSL-Kras^G12D allele to initiate pancreatic tumorigenesis in Mist1-expressing progenitor cells compared with mature acinar cells. The Mist1^KrasG12D/+ mice represent an extremely robust single allele model of pancreatic exocrine neoplasia that should afford an opportunity to explore important questions concerning the etiology of pancreatic cancer and various preclinical applications including the evaluation of pharmacologic inhibitors of the Notch and Ras signaling pathways.

Grant support: AACR-PanCAN Career Development Award (D.A. Tuveson); NIH grants R01 CA101973 (D.A. Tuveson), U01 CA084291 (D.A. Tuveson), and R01 DK55489 (S.F. Konieczny); Department of Defense grant BC043093 (S.F. Konieczny); and the Purdue Cancer Center (S.F. Konieczny).

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.

We thank Catrina King for technical assistance, and Denise Crowley and Drs. Roderick Bronson and Jonathan Glickman for initial pathologic consultation.