Malignant transformation of duct-like cells originating from acini in transforming growth factor α transgenic mice

GASTROENTEROLOGY 1998;115:1254–1262

Malignant Transformation of Duct-like Cells Originating From Acini in Transforming Growth Factor a Transgenic Mice ¨ HRS,* GU ¨ NTHER KLO ¨ PPEL,‡ GUIDO ADLER,* and ROLAND M. SCHMID* MARTIN WAGNER,* HARDI LU *Department of Internal Medicine I, Ulm University, Ulm, and ‡Department of Pathology, University of Kiel, Kiel, Germany

Background & Aims: In transgenic mice overexpressing transforming growth factor (TGF)-a in the exocrine pancreas, progressive pancreatic fibrosis and a transdifferentiation of acinar cells to duct-like cells occurs. The present study was undertaken to analyze this transdifferentiation process. Methods: Pancreatic specimens were characterized using light microscopy and immunohistochemistry. Expression of the epidermal growth factor receptor (EGFR) and TGF-a was evaluated with slot blot and Western analysis. To identify other generic events, K-ras mutations were screened with an enriched polymerase chain reaction approach and p53 expression was detected with immunohistochemistry. Results: Morphological examination revealed an aggregation of interlobular fibroblasts and a decrease in acinar cell height starting at day 14 after birth. In older animals, these acinar cells change to duct-like cells, which form tubular structures and express ductal markers. Evidence for dysplastic changes was found in 12 of 21 TGF-a transgenic mice older than 1 year. We also observed four malignant pancreatic tumors, which were multicentric and originated from dysplastic tubular complexes. They displayed a mixed cystic-papillary phenotype strongly positive for carbonic anhydrase activity. EGFR expression progressively increased in the transition from acinar to duct-like and transformed cells. Activating K-ras mutations could not be detected; however, tubular complexes and tumors displayed increased immunoreactivity for nuclear p53. Conclusions: These data suggest an involvement of the TGF-a/EGFR pathway in conjunction with other yet unknown events in pancreatic tumor development. Furthermore, these observations are in favor of an acinar-ductal carcinoma sequence. Thus, these transgenic animals will be useful to define genetic alterations associated with a transition from acinar cells to a neoplastic ductal phenotype.

lthough a multistep model of carcinogenesis in colon cancer is generally accepted, such a sequence has not been established for pancreatic cancer.1 Hyperplastic changes in ductal epithelium are frequently observed, but their role in the pathogenesis of pancreatic carcinoma is not yet settled.2–4 Papillary ductal hyperplasia without

A

atypia is frequently seen at autopsy in otherwise normal pancreas or in patients with chronic pancreatitis.3,5 Further characterization revealed the presence of K-ras mutations in ductal lesions in chronic pancreatitis.6 In addition, epidemiological studies provide evidence for an association of long-lasting chronic pancreatitis and pancreatic carcinoma.7 The origin of pancreatic cancer from duct cells is based on the phenotypic appearance of the tumor cells.8 However, the possibility remains that mature pancreatic acinar cells have the potential to alter their phenotype towards ductal cells. On the other hand, it has been shown that a small stem cell population can divide and subsequently differentiate into acinar, duct, or islet cells.9–13 Cultured mouse acinar cells lose their typical cytoplasmic ultrastructure and a specific antigen of mature acinar cells, but acquire a more duct cell–like phenotype and express duct cell–specific antigens.14,15 Hall and Lemoine15 reported that within 4 days, acinar cells lost amylase immunoreactivity and gained expression of cytokeratin 19, specific for duct cells in humans. In another study, an increase in the duct-specific messenger RNAs for cystic fibrosis transduction regulator and carbonic anhydrase II has been detected in cultured human pancreas epithelial cells. In addition, the cytokeratin pattern switched to cytokeratin 7– and cytokeratin 19–positive cells.16 Overexpression of transforming growth factor (TGF)-a in the pancreas causes a progressive fibrosis and structural transition from acinar cells to tubular complexes as characterized by Bockman et al. and others.17–19 In the present study, we present evidence that, in TGF-a transgenic mice, acinar cells transdifferentiate to ductlike cells that de novo express a ductal marker and are strongly positive for carbonic anhydrase activity. Furthermore, 12 TGF-a overexpressing mice developed dysplasAbbreviations used in this paper: bp, base pair(s); DAB, 3,38diaminobenzidine; EGFR, epidermal growth factor receptor; mK-ras, mouse K-ras; PCR, polymerase chain reaction; TGF, transforming growth factor. r 1998 by the American Gastroenterological Association 0016-5085/98/$3.00

November 1998

tic tubular complexes and 4 animals showed further malignant progression within these dysplastic foci. Transformed cells developed from duct-like cells within tubular complexes and were highly positive for carbonic anhydrase activity and ductal markers. These observations provide arguments in favor of tubular complexes being precursors of invasive carcinoma and point toward an acinar-ductal-carcinoma sequence in this model. Screening for the p53 status of tubular complexes and tumors revealed an enhanced nuclear immunoreactivity of p53 compared with wild-type pancreas. No evidence for activating point mutations of K-ras was detected using polymerase chain reaction (PCR) analysis. The tubular complexes and the tumor cells expressed high levels of the epidermal growth factor receptor (EGFR) similar to human pancreatic cancer and chronic pancreatitis. These data provide functional evidence for an autocrine loop that may contribute to hyperplastic and malignant changes in the pancreas.20,21

Materials and Methods Animals The transgenic mouse line EL-TGFa-hGH (#2261-3) was kindly provided by Sandgren et al. and has been described earlier.17 Mice were bred to C57BL/6 and kept as heterozygotes. Transgenic animals were identified by Southern blot analysis of mouse tail DNA using a 1.3–base pair (bp) EcoRI fragment of human growth hormone polyA as a probe.17

Light Microscopy For light microscopy, pancreatic tissue was fixed in 2% phosphate-buffered paraformaldehyde for 12 hours, embedded in paraffin, and sectioned (4 µm). Sections were stained with H&E and examined with a Zeiss photomicroscope (Zeiss, Oberkochen, Germany).

Source of Antibodies The following antibodies were used to stain frozen sections: rat acinar-1 and duct-1 monoclonal antibodies provided by R. C. De Lisle (Department of Anatomy and Cell Biology, University of Kansas)14; rabbit antiserum to human amylase purchased from Sigma (Deisenhofen, Germany); rat cytokeratin 8 (TROMA1) and cytokeratin 18 (TROMA2) provided by H. Kemmler (Max Planck Institute for Immunobiology, Freiburg, Germany)22; rabbit antiserum to the human EGFR purchased from Santa Cruz Biotechnology (Santa Cruz, CA); and rabbit polyclonal antiserum to p53, a generous gift from W. Deppert (Heinrich-Pette-Institute for Experimental Virology and Immunology, University of Hamburg, Hamburg, Germany).

Immunohistochemistry Sections of shock frozen pancreas were air-dried, incubated in acetone at 220°C for 5 minutes, and further treated

ACINAR–DUCTAL CARCINOMA SEQUENCE

1255

with blocking solution, primary antibody, biotinylated secondary antibody, and a peroxidase-avidin complex according to the manufacturer’s instructions (ABC Vectorstain Kit; Vector Laboratory, Burlingame, CA). The optimal dilution and incubation time of the primary antibody was established in previous experiments. After these incubations, the bound peroxidase was visualized using 3,38-diaminobenzidine (DAB) as chromogen. For immunohistochemical detection of p53, DAB staining was nickel enhanced to obtain a black nuclear staining. After the DAB step, sections were washed again, counterstained with hematoxylin, dehydrated, embedded, and analyzed with a photomicroscope (Zeiss). The following dilutions of antibodies were used: acinar-1 and duct-1, 1:1; antiamylase, 1:100; TROMA1, 1:20; TROMA2, 1:20; anti-EGFR, 1:250; and anti-p53, 1:2000.

Staining for Carbonic Anhydrase Activity Carbonic anhydrase activity was detected by a modified cobalt/phosphate method.23 Sections were incubated in a solution containing 8.75 mmol/L CoSO4, 5.8 mmol/L KH2PO4, 157 mmol/L NaHCO3, 0.5% Triton X-100, and 53 mmol/L H2SO4 for 2 seconds followed by a drying period of 30 seconds. These two steps were repeated 15 times. Sections were washed in a solution containing 0.67 mmol/L KH2PO4 and 150 mmol/L NaCl (pH 5.9) for 2 minutes followed by 0.5% (NH4)2S for 15 seconds. After a final wash with H2O for 2 minutes, sections were counterstained with nuclear fast red, rinsed in H2O, and mounted with Rotihistol (Roth, Karlsruhe, Germany). As negative controls, sections were incubated in a reaction solution containing 20 µmol/L acetazolamide.

Western Blotting Pancreata of TGF-a transgenic mice and littermate controls were homogenized on ice in lysis buffer (150 mmol/L NaCl, 50 mmol/L Tris-HCl [pH 7.2], 2 mmol/L EDTA, and 1% Nonidet P40) supplemented with a protease inhibitor cocktail (50 µg/mL antipain, 1 µg/mL aprotinin, 40 µg/mL bestatin, 50 µg/mL chymostatin, 5 µg/mL E-64, 0.5 µg/mL leupeptin, 1 mg/mL pefabloc, 0.7 µg/mL pepstatin, and 50 µg/m Na-p-tosyl-L-lysine-chloromethyl ketone; Boehringer Mannheim, Mannheim, Germany). Protein concentrations were measured with Bradford reagent (Bio-Rad, Munich, Germany), and equal amounts of total protein (40 µg) were separated on a 7.5% sodium dodecyl sulfate–polyacrylamide gel electrophoresis and blotted to polyvinylidene difluoride membranes. Membranes were blocked in 2.5% nonfat milk and 0.02% Tween 20 in phosphate-buffered saline (PBS), pH 8.0 (blocking solution), overnight at 4°C, incubated with a rabbit polyclonal EGFR antibody (1:1000) in blocking solution for 2 hours, washed three times for 10 minutes in PBS/0.02% Tween 20, and incubated with a 1:5000 dilution of a horseradish peroxidase– labeled secondary antibody in blocking solution for 1 hour. After three 10-minute washes in PBS/0.02% Tween 20, binding of antibody was visualized with enhanced chemiluminescence reagent (Amersham Life Science, Buckinghamshire, England).

1256

WAGNER ET AL.

GASTROENTEROLOGY Vol. 115, No. 5

Slot Blot Analysis for TGF-a Total RNA was extracted from frozen pancreas of transgenic animals and controls according to the method of Chomczynski et al.24 RNA was denatured at 65°C for 10 minutes in a 1:1 solution of 37% formaldehyde/standard saline citrate 203, blotted to nitrocellulose with a slot blot device (Bio-Rad, Munich, Germany), and UV crosslinked. A random prime-labeled Smal fragment of the rat TGF-a complementary DNA was used as a probe. Filters were washed under high-stringency conditions and visualized by autoradiography.

Enriched PCR for Activating K-ras Mutations Total DNA was extracted with the QIAamp tissue kit (Quiagen, Hilden, Germany) from frozen sections of tumor tissue as judged by phase-contrast microscopy. Enriched PCR for mutated mouse K-ras (mK-ras) at codon 12 was established according to the procedure for human K-ras.25 The intron-exon sequence of mK-ras was previously sequenced (R. Blumenthal, unpublished results). The following oligonucleotide primers were used for amplification of mK-ras (a nucleotide substitution is underlined): mK-ras 58, 58 A C T G A G T A T A A A C T T G T G G T G G T T G G A C C T 38; mK-ras 38 wild type, 58 T A T C T T T T T C A A A G C G G C T G G C T G 38; mK-ras 38 mutated, 58 T A T C T T T T T C A A A G C G C C T G G C T G 38. In the first step, 100 ng DNA was amplified in a 50-µL reaction containing 5 pmol of each primer (mK-ras 58 and mK-ras 38 wild type), 0.2 mmol/L deoxyribonucleotide triphosphates, 10 mmol/L Tris-HCl (pH 8.8), 1.5 mmol/L MgCl2, 75 mmol/L KCl, and 2.5 U of Taq polymerase (GIBCO BRL, Eggenstein, Germany). The first round of PCR consisted of 15 cycles at 94°C for 30 seconds, 62°C for 30 seconds, and 72°C for 30 seconds using a programmable thermocycler (GeneAmp 9700; Perkin Elmer Applied Biosystems, Forster City, CA). A 30-µL aliquot of this amplification was digested with 20 U BstNI (New England Biolabs, Beverly, MA) for 3 hours at 60°C. The second amplification was carried out in a 50-µL reaction with 30 pmol of primers mK-ras 58 and mK-ras 38 mut, 0.2 mmol/L deoxyribonucleotide triphosphates, 10 mmol/L Tris-HCl (pH 9.2), 1.5 mmol/L MgCl2, 25 mmol/L KCl, and 2.5 U of Taq polymerase. After 35 cycles and a second digest with BstNI, products were separated on 4% agarose gels (NuSieve GTG; FCM, Rockland, ME) and stained with

ethidium bromide. A 115-bp fragment indicates wild-type mK-ras. In case of a mutated mK-ras at codon 12, a 162-bp fragment is expected. Genomic DNA extracted from mouse tails and NIH 3T3 were used as negative controls, and a mK-ras genomic fragment mutated at codon 12 was used as a positive control.

Results Development of Fibrosis and Tubular Complexes in TGF-a Transgenic Mice TGF-a transgenic animals show no signs of altered behavior or growth abnormalities compared with littermate controls but develop a drastically enlarged and fibrotic pancreas. The characteristics of TGF-a transgenic mice and littermate controls are summarized in Table 1. An increase in the interlobular fibroblast population was already detected at 14 days of age, which progressed to a massive fibrosis with extensive extracellular matrix accumulation at 180 days (Figure 1). These fibrotic changes were paralleled by an altered acinar cell structure. Cells decreased in height, leaving an enlarged acinar lumen (Figure 1A). At day 28, some of these cells became elongated and flat, indicative of tubular complex formation (Figure 1B). At 180 days, extensive tubular structures were interspersed with areas of normal-appearing acini and acini representing early stages of transdifferentiation (Figure 1C and D). Electron microscopy revealed a subsequent loss of acinar cell characteristics, e.g., zymogen granules and a gain of structures indicative of mature ductal cells including mucin granules, apical microvilli, and basal tight junctions (data not shown). Expression of Acinar and Ductal Markers in TGF-a Transgenic Mice and Littermate Controls At 180 days, acinar cells of control animals showed homogenous immunoreactivity for amylase, which was markedly reduced in cells within tubular complexes of TGF-a transgenic animals. Similar results were ob-

Table 1. Characteristics of Animals Analyzed Age of animals (days)

Littermate controls

TGF-a transgenic mice

Mice without dysplasia

Mice with large cystic changes and intracystic proliferation

Mice with proliferating tubular complexes

Papillary to cystic tumors

0–180 180–360 360–540

82 40 17

94 44 21

94 42 9b

0 2 10

0 0 5

0 2a 2

NOTE. The number of cystic changes and proliferating tubular complexes exceeds the total number of mice because these changes were seen simultaneously in individual mice. aTumors were first observed in 2 animals at 280 days (animals 315 and 316). bIn addition to the pancreatic tumors, we observed a single high-grade T-cell lymphoma involving the spleen, liver, and lung but not the pancreas in a transgenic mouse at 450 days (animal 275).

November 1998

ACINAR–DUCTAL CARCINOMA SEQUENCE

1257

Figure 1. Development of tubular complexes in TGF-a transgenic mice. Tissue sections were stained with H&E. (A ) Early changes at day 14 include mild interlobular fibrosis (arrows) and decreased acinar cell height (asterisks). (B ) At day 28, fibrosis (arrows) and altered acinar morphology (asterisks) is visible. (C ) After 180 days, pancreatic fibrosis and typical tubular complexes (asterisks) can be seen next to normal acinar cells (arrowhead ). At a higher magnification, the transition from acinar toward ductal morphology is evident. The inserts represent pancreas sections of littermate controls at the same age. A, B, and C : original magnification 503; D : 2503.

tained when sections were stained with acinar-1, detecting an antigen present on acini but not on duct cells.11 Some tubular complexes were entirely negative for this marker, whereas others were still positive (Figure 2A). In littermate controls, carbonic anhydrase activity was limited to ducts and the centroacinar region, whereas the loss of acinar markers was associated with a gain of duct-specific markers like carboanhydrase activity and duct-1 in TGF-a transgenic mice. Duct-1 only stained ductal cells in wild-type pancreas (Figure 2B). TROMA1 and TROMA2 have been shown to recognize cytokeratin 8 and cytokeratin 18, respectively.22 In TGF-a transgenic mice, expression of cytokeratin 8 and 18 is not altered in the tubular complexes compared with the acinar cells in wild-type animals (Figure 2C). Thus, tubular complexes keep the cytokeratin expression pattern of epithelial cells, whereas the expression of cell type–specific markers like amylase and carbonic anhydrase changed. To test whether the acinar-specific overexpression of TGF-a and formation of tubular complexes is associated with an autocrine or paracrine stimulation, we examined the distribution and expression of EGFR in transgenic and control pancreas. In wild-type pancreas, staining was

mainly restricted to intra- and interlobular ducts (Figure 3A), whereas in transgenic mice tubular complexes expressed high levels of EGFR (Figure 3B). These findings were confirmed by Western blot analysis (Figure 3C), which revealed an increase of the EGFR expression as early as 28 days. The expression of TGF-a remained constant (Figure 3D). These data suggest a role of this signaling pathway in tubular complex formation. Malignant Tumors Originate From Tubular Complexes Malignant transformation occurred only in animals older than 180 days (Table 1). Although tumors in animals 315 and 316 were observed in routine autopsy due to the age of the animals, animal 191 was found dead in the cage at the age of 520 days. The macroscopic findings indicate major bleedings in large pancreatic cysts as possible cause of death. To get more insight in the tumor development, we killed and analyzed an additional 20 animals older than one year. Together with animal 191, we found dysplastic changes and tumors in 12 of 21 animals older than 1 year (Table 1). Although we saw different types of dysplastic changes, the underlying lesion was the transdifferentiation of acinar cells to

1258

WAGNER ET AL.

tubular complexes in all cases. Two types of dysplastic lesions were found: highly proliferating tubular complexes in 5 animals and large cystic changes in 12 TGF-a transgenic mice (Figure 4). In contrast to the welldifferentiated cells in regular tubular complexes, the

GASTROENTEROLOGY Vol. 115, No. 5

highly proliferating ductal structures were clustered in areas with delicate connective tissue. These proliferating cells displaced and invaded the surrounding parenchyma (Figure 4A). They are characterized by an increase in the nuclear to cytoplasmic ratio and displayed a marked nuclear anaplasia (Figure 4B). Because we did not find signs of vascular or extrapancreatic invasion, we cannot define these highly proliferating ductal structures as invasive pancreatic cancer at the moment. Cysts originating from tubular complexes were lined with flat epithelium without signs of nuclear anaplasia (Figure 4C). However, signs of intracystic papillary proliferations were found in five cases. Cells within these areas showed various degrees of cellular anaplasia defining these lesions as dysplastic (Figure 4D). A progression of these intracystic proliferations to invasive pancreatic tumors was evident in 4 transgenic animals (Figure 4E and F). Tumor

C

D

Figure 2. Expression of acinar and ductal markers in the pancreas of TGF-a transgenic mice. Sections were incubated with the respective antibodies or stained for carbohydrase activity. (A ) Staining for amylase and acinar-1 in 180-day-old transgenic mice (TGF-a) compared with littermate controls (WT). (B ) Carbonic anhydrase activity in tubular complexes of TGF-a and in controls (WT). Duct-1 expression is strongly expressed in tubular complexes. In control mice (WT) it is restricted to the ducts. (C ) Staining for cytokeratin 8 (TROMA1) and 18 (TROMA2) in TGF-a transgenic (TGF-a) animals and littermate controls (WT). Original magnification 503.

Figure 3. Expression of the EGFR and TGF-a in the pancreas of TGF-a transgenic mice and control animals. In wild-type animals, (A ) the EGFR is only expressed in ductal cells, (B ) whereas cells within tubular complexes express high levels of EGFR. (C ) Western analysis indicates an increase of the EGFR at days 28 and 180. (D ) Slot blot analysis with 10 µg of total RNA shows a constant high level of TGF-a expression in the pancreas of 0 to 180-day-old TGF-a transgenic animals. The 185 ribosomal probe served as control.

November 1998

ACINAR–DUCTAL CARCINOMA SEQUENCE

1259

Figure 4. Dysplastic changes and tumors in the pancreas of TGF-a transgenic mice. (A ) Highly proliferating tubular complexes are clustered and surrounded by delicate connective tissue (animal 195; original magnification 503). (B) The higher magnification (animal 190; original magnification 1003) reveals irregular cellular morphology. This includes pleomorphic nuclei with abnormal chromatin patterns and frequent mitotic figures in ductal structures. (C ) Large cysts are lined with flat epithelium (animal 333; original magnification 503). (D ) In some animals we found intracystic proliferations characterized by a multilayer epithelium and intracystic, papillar y projections with a varying degree of nuclear anaplasia (animal 194; original magnification 503). (E ) Tumors display a papillary to cystic phenotype surrounded by dense connective tissue (animal 316; original magnification 12.53). (F ) A higher magnification reveals the origin of the tumors from large cysts. Tumors invade the connective tissue and show a marked nuclear anaplasia (animal 315; original magnification 1003).

formation occurred multifocal and displayed a papillary to cystic phenotype. No obvious signs of metastasis were detected. Tumors Developing in TGF-a Transgenic Mice Express Ductal Markers To confirm the duct-like phenotype of pancreatic tumors, we stained frozen sections of pancreas with tumors with acinar- and duct-specific markers (Figure 5). Similar to tubular complexes, cell clusters within the tumors still stained weakly for amylase immunoreactivity and for acinar-1, suggesting an origin from acinar cells. On the other hand, tumors were strongly positive for carbonic anhydrase activity and stained for duct-1 (Figure 5A). Tumors were also positive for cytokeratin 8 and 18, suggesting a relation to tubular complexes. The EGFR expression was further increased on tumor cells compared

with cells within tubular complexes, pointing toward a selection process for cells with high levels of EGFR expression (Figure 5B). Tubular Complexes and Tumors Show Increased Nuclear Staining for p53 Immunohistochemical staining revealed a markedly increased nuclear staining for p53 in tumors and in tubular complexes compared with wild-type pancreas (Figure 6). To identify other genetic events, we screened for activating K-ras mutations in the observed tumors. Using an enriched PCR approach for murine K-ras with specific primers followed by a BstNI digest,25 we were not able to detect mutations of K-ras at codon 12. The specificity and sensitivity of this approach was confirmed using mK-ras mutated at codon 12 (data not shown).

1260

WAGNER ET AL.

Figure 5. Characterization of tumors with acinar and ductal markers. (A ) Sections were incubated with antibodies as indicated or stained for carbonic anhydrase activity. Compared with normal pancreatic tissue, tumor sections display reduced immunoreactivity for amylase and acinar-1, whereas carbonic anhydrase activity and reactivity for duct-1 was increased. (B ) The expression of cytokeratin 8 and 18 remained unchanged within tubular complexes. Tumors overexpressed the EGFR. Original magnification 503.

Discussion In the present study, we show that transgenic mice overexpressing TGF-a in the pancreas show a differentiation of acinar cells toward duct-like cells. In addition, these mice developed malignant tumors with a duct-like phenotype. At 4 weeks of age, tubular complexes appeared that were intermingled with acini. The acini showed a dilatation of acinar lumen, which was lined by a monolayer epithelium mostly composed of flattened acinar cells and cells displaying an intermediate phenotype of acinar and duct-like cells. In older animals (180 days), the transitional forms between acinar cells and duct-like cells disappeared in favor of ductal cells, which completely lost ultrastructural criteria typical for mature acinar cells. There were no zymogen granules and the endoplasmic reticulum was scarce. Interestingly, these cells were strongly positive for carbonic anhydrase activity starting at day 28 and duct-1, which are both specific for duct cells. The de novo expression of these ductal markers together with

GASTROENTEROLOGY Vol. 115, No. 5

the intermediate filament pattern indicate that tubular complexes formed in TGF-a overexpressing mice represent a transdifferentiation process rather than a state of dedifferentiation. Whether these changes are due to a direct effect of TGF-a on the acinar cell or mediated by an unknown autocrine and/or paracrine mechanism is unclear. The phenotypic switch of acinar cells is somewhat similar to the changes reported in vitro.11–13,26 However, in vitro this phenomenon may be explained by the lack of cell-cell contacts and/or growth factors controlling maintenance of acinar cell differentiation. The plasticity of acinar cells has also been observed in vivo in several experimental models of pancreatitis and in cystic fibrosis.27–32 In acute pancreatitis, the formation of tubular complexes represents most likely acinar cell degeneration. These duct-like cells revealed a loss of pancreatic enzymes, an increased expression of keratin and actin, and a weak staining for carcinoembryonic antigen.33 Further evidence for the plasticity of acinar cells in vivo came from transgenic mice overexpressing interferon gamma under control of an islet-specific promoter.34 Inflammatory destruction of islets was followed by local overexpression of EGF, TGF- a, and the EGFR in neighboring acinar cells and a transdifferentiation to a ductal phenotype and appearance of endocrine cells. In addition to tubular complex formation, we found papillary-cystic neoplasms originating from tubular complexes because these tumors expressed the same ductal markers. The observed pattern is comparable with that occurring in a particular pancreatic tumor called intraductal papillary mucinous tumor. As to the pathogenesis of tubular complexes and papillary-cystic tumors, it is interesting that TGF-a binding to the EGFR activates a signaling pathway involving the activation of Ras, which is known for its close association with the development of pancreatic carcinoma. Therefore, similar signaling cascades might be responsible for the formation of tubular complexes in

Figure 6. Expression of p53 in the pancreas of TGF-a transgenic mice and control animals. Sections were incubated with polyclonal antiserum to p53. Tumors and tubular complexes show enhanced nuclear staining for p53 compared with wild-type pancreas and the remaining normal acinar cells in TGF-a transgenic mice. Original magnification 2503.

November 1998

these models. Furthermore, EGFR expression is very high in transformed cells in TGF-a transgenic mice originating from tubular complexes. The activation of this pathway might therefore be an initial event in the multistep process of carcinogenesis in the pancreas. According to the multistep model of colorectal cancer, the accumulation of a number of mutations rather than the order of mutations is important for tumor development. Therefore, it is very likely that additional mutations occurred in TGF-a transgenic mice. We are not able to define these events at the moment. Using an enriched PCR approach, we did not detect mK-ras mutations at codon 12. However, the activation of K-ras is most likely not important because transgenic mice overexpressing TGF-a in keratinocytes bypassed the need for an additional H-Ras mutation in skin tumorigenesis.35 Therefore, other pathways are more likely to be critical for tumor progression in this transgenic model. Both the tubular complexes and the tumors showed a marked enhanced nuclear staining for p53. Because the used antibody stains both wild-type and mutant p53, we cannot confine the role of p53 in the tumor development in TGF-a transgenic mice. TGF-a transgenic mice are currently intercrossed with p53 mice to investigate whether tumor induction is accelerated in the presence of an inherited mutated p53 allele. In summary, overexpression of TGF-a in the pancreas induced a differentiation program of acinar cells to duct-like cells. In old animals, transformation occurred, originating from these transdifferentiated cells. Duct-like cells and even more tumor cells overexpressed the EGFR underlining the importance of this signaling pathway in malignant transformation. Because tumors developed from acinar cells, these data favor a role for TGF-a in initiating an acinar-ductal carcinoma sequence. Because tumor formation is a rather late event and not obligatory, other genetic events are likely to determine the progression to invasive cancer.

References 1. Vogelstein B, Fearon ER, Hamilton SR, Kern SE, Preisinger AC, Leppert M, Nakamura Y, White R, Smits A, Bos JL. Genetic alterations during colorectal tumor development. N Engl J Med 1988;319:525–532. 2. Cubilla AL, Fitzgerald PJ. Morphological lesions associated with human primary invasive nonendocrine pancreatic cancer. Cancer Res 1975;35:2234–2248. 3. Klo¨ppel G, Bommer G, Ru¨ckert K, Seifert G. Intraductal proliferation in the pancreas and its relationship to human and experimental carcinogenesis. Virchows Arch 1980;387:221–223. 4. Kozuka S, Sassa R, Taki T, Masamoto K, Nagasawa S, Saga S, Hasegawa B, Takeuchi M. Relation of pancreatic duct hyperplasia to carcinoma. Cancer 1979;43:1418–1428. 5. Sommers SC, Murphy SA, Warren S. Pancreatic duct hyperplasia and cancer. Gastroenterology 1954;27:629–640.

ACINAR–DUCTAL CARCINOMA SEQUENCE

1261

6. Yanagisawa A, Ohtake K, Ohashi K, Masaharu H, Kitagawa T, Sugano H, Kato Y. Frequent c-Ki-ras oncogene activation in mucous cell hyperplasia of pancreas suffering from chronic pancreatitis. Cancer Res 1993;53:953–956. 7. Lowenfels AB, Maisonneuve P, Cavallini G, Ammann RW, Lankisch PG, Andersen JR, Dimagno EP, Andren-Sandberg A, Domello¨f L, The International Pancreatitis Study Group. Pancreatitis and the risk of pancreatic cancer. N Engl J Med 1993;328:1433–1437. 8. Klo¨ppel G. The pancreas: biology, pathology, and disease. In: Go VL, DiMagno EP, Gardener JD, Lebenthal E, Reber HA, Scheele GA, eds. Pathology of nonendocrine pancreatic tumors. 2nd ed. New York: Raven, 1993;46:871–898. 9. Githens S. The pancreatic duct cell: proliferative capabilities, specific characteristic, metaplasia, isolation, and culture. J Pediatr Gastroenterol Nutr 1988;7:486–506. 10. Jamieson JD, Ingber DE, Muresan V, Hull BE, Sarras MP, MayliePfenninger MF, Iwanij V. Cell surface properties of normal, differentiating, and neoplastic pancreatic acinar cells. Cancer 1981;47:1516–1525. 11. Wang RN, Klo¨ppel G, Bouwens L. Duct- to islet-cell differentiation and islet growth in the pancreas of duct-ligated adult rats. Diabetologia 1995;38:1405–1411. 12. Wang RN, Rehfeld JF, Nielsen FC, Klo¨ppel G. Expression of gastrin and transforming growth factor a during duct to islet cell differentiation in the pancreas of duct-ligated adult rats. Diabetologia 1997;40:887–893. 13. Bouwens L, Klo¨ppel G. Islet cell neogenesis in the pancreas. Virchows Arch 1996;427:553–560. 14. De Listle BC, Logsdon CD. Pancreatic acinar cells in culture: expression of acinar and ductal antigens in a growth-related manner. Eur J Cell Biol 1990;51:64–75. 15. Hall PA, Lemoine NR. Rapid acinar to ductal transdifferentiation in cultured human exocrine pancreas. J Pathol 1992;166:97–103. 16. Vila MR, Lloreta J, Real FX. Normal human pancreas cultures display functional ductal characteristics. Lab Invest 1994;71:423– 431. 17. Sandgren EP, Luetteke NC, Palmiter RD, Brinster RL, Lee DC. Overexpression of TGF-a in transgenic mice: induction of epithelial hyperplasia, pancreatic metaplasia, and carcinoma on the breast. Cell 1990;61:1121–1135. 18. Jhappan C, Stahle C, Harkins RN, Fausto N, Smith GH, Merlino GT. TGF-a overexpression in transgenic mice induces liver neoplasia and abnormal development of the mammary gland and pancreas. Cell 1990;61:1137–1146. 19. Bockman DE, Merlino G. Cytological changes in the pancreas of transgenic mice overexpressing transforming growth factor a. Gastroenterology 1992;103:1883–1892. 20. Lemoine NR, Hughes CM, Barton CM, Poulsom R, Jeffery RE, Klo¨ppel G, Hall PA, Gullick WJ. The epidermal growth factor receptor in human pancreatic cancer. J Pathol 1992;166:7–12. 21. Barton CM, Hall PA, Hughes CM, Gullick WJ, Lemoine NR. Transforming growth factor alpha and epidermal growth factor in human pancreatic cancer. J Pathol 1991;163:111–116. 22. Boller K, Kemmler R, Baribault H, Doetschman T. Differential distribution of cytokeratins after microinjection of anti-cytokeratin monoclonal antibodies. Eur J Cell Biol 1987;43:459–468. 23. Githens S, Finley JJ, Patke CL, Schexnayder JA, Fallon KB, Ruby JR. Biochemical and histochemical characterization of cultured rat and hamster pancreatic ducts. Pancreas 1987;2:427–438. 24. Chomczynski P, Sacchi N. Single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction. Anal Biochem 1987;162:156–159. 25. Kahn SM, Jiang W, Culbertson TA, Weinstein IB, Williams GM, Tomita N, Ronai Z. Rapid and sensitive nonradioactive detection of mutant K-ras genes via enriched PCR amplification. Oncogene 1991;6:1079–1083.

1262

WAGNER ET AL.

26. Arias AE, Bendayan M. Differentiation of pancreatic acinar cells into duct-like cells in vitro. Lab Invest 1993;69:518–530. 27. Churg A, Ward RR. Early changes in the exocrine pancreas of the dog and rat after ligation of the pancreatic duct. Am J Pathol 1971;63:521–534. 28. Bockman DE, Black O Jr, Mills LR, Webster PD. Origin of tubular complexes during developing induction of pancreatic adenocarcinoma by 7,12-dimethylbenz(a)antracene. Am J Pathol 1978;90: 645–651. 29. Mundlos S, Adler G, Schaar M, Koop I, Arnold R. Exocrine function in oleic acid–induced pancreatic insufficiency in rats. Pancreas 1986;1:29–36. 30. Elsa¨sser HP, Adler G, Kern HF. Time course and cellular source of pancreatic regeneration following acute pancreatitis in the rat. Pancreas 1986;1:421–429. 31. Porta EA, Stein AA, Patterson P. Ultrastructural changes of the pancreas and liver in cystic fibrosis. Am J Clin Pathol 1964;42: 451–465. 32. Willemer S, Elsa¨sser HP, Kern HF, Adler G. Tubular complexes in cerulein and oleic acid induced pancreatitis in rats: glycoconjugate pattern, immunocytochemical and ultrastructural findings. Pancreas 1987;2:669–675. 33. Willemer S, Adler G. Histochemical and ultrastructural characteristics of tubular complexes in human acute pancreatitis. Dig Dis Sci 1989;34:46–55. 34. Arnush M, Gu D, Baugh C, Sawyer SP, Mroczkowski B, Krahl T, Sarvetnick N. Growth factors in the regenerating pancreas of gamma interferon transgenic mice. Lab Invest 1996;74:985– 990.

GASTROENTEROLOGY Vol. 115, No. 5

35. Vassar R, Hutton ME, Fuchs E. Transgenic overexpression of transforming growth factor alpha bypasses the need for c-Ha-ras mutations in mouse skin tumorigenesis. Mol Cell Biol 1992;12: 4643–4653.

Received October 17, 1997. Accepted July 29, 1998. Address requests for reprints to: Roland M. Schmid, M.D., Department of Internal Medicine, University of Ulm, Robert-Koch-Strasse 8, 89081 Ulm, Germany. Fax: (49) 731-502-4302. Supported by a grant from Bausteinfo ¨rderung P176 and the Bundesministerium fu ¨r Bildung und Forschung (to R.M.S.). The authors thank E. Sandgren (Laboratory of Reproductive Physiology, University of Pennsylvania, Philadelphia) for the generous gift of the transgenic mouse line EL-TGFa-hGH (#2261.3); R. D. Palmiter (Howard Hughes Medical Institute, University of Washington, Seattle) for the hGH polyA; H. Kemmler (Max Planck Institute for Immunobiology, Freiburg), for the antibodies TROMA1 and TROMA2; R. C. De Lisle (Department of Anatomy and Cell Biology, University of Kansas) for the acinar-1 and duct-1; W. Deppert (Heinrich-Pette-Institute for Experimental Virology and Immunology, University of Hamburg), for the p53 antiserum; H. F. Kern, Institute for Cytobiology and Cytopathology (PhilippsUniversity, Marburg) for helpful discussion; T. Binder for classifying the high-grade lymphoma of animal #225; E. Wolff-Hieber and E. Schacher for excellent technical assistance; and S. Aigner for assistance with manuscript preparation.