Advertisement
Research Article

Inactivation of TIF1γ Cooperates with KrasG12D to Induce Cystic Tumors of the Pancreas

  • David F. Vincent equal contributor,

    equal contributor Contributed equally to this work with: David F. Vincent, Kai-Ping Yan

    Affiliations: INSERM, U590, IFR62, Lyon, France, Univ Lyon, Lyon, France, INSERM “Avenir” group, Lyon, France, Centre Léon Bérard, Lyon, France

    X
  • Kai-Ping Yan equal contributor,

    equal contributor Contributed equally to this work with: David F. Vincent, Kai-Ping Yan

    Affiliation: IGBMC, Strasbourg, Illkirch-Cedex, France

    Current address: The Wistar Institute, Philadelphia, Pennsylvania, United States of America

    X
  • Isabelle Treilleux,

    Affiliation: Centre Léon Bérard, Lyon, France

    X
  • Fabien Gay,

    Affiliation: INSERM, U865, Faculté Laennec, Lyon, France

    X
  • Vanessa Arfi,

    Affiliations: INSERM, U590, IFR62, Lyon, France, Univ Lyon, Lyon, France, INSERM “Avenir” group, Lyon, France, Centre Léon Bérard, Lyon, France

    X
  • Bastien Kaniewsky,

    Affiliations: INSERM, U590, IFR62, Lyon, France, Univ Lyon, Lyon, France, INSERM “Avenir” group, Lyon, France, Centre Léon Bérard, Lyon, France

    X
  • Julien C. Marie,

    Affiliation: INSERM, U758, Lyon, France

    X
  • Florian Lepinasse,

    Affiliations: Univ Lyon, Lyon, France, INSERM, U865, Faculté Laennec, Lyon, France, Hospices Civils de Lyon, Hôpital Edouard Herriot, Lyon, France

    X
  • Sylvie Martel,

    Affiliations: INSERM, U590, IFR62, Lyon, France, INSERM “Avenir” group, Lyon, France, Centre Léon Bérard, Lyon, France

    X
  • Sophie Goddard-Leon,

    Affiliation: Centre Léon Bérard, Lyon, France

    X
  • Juan L. Iovanna,

    Affiliation: INSERM, U624, Marseille, France

    X
  • Pierre Dubus,

    Affiliation: EA 2406, Univ Bordeaux, France

    X
  • Stéphane Garcia,

    Affiliation: INSERM, U624, Marseille, France

    X
  • Alain Puisieux,

    Affiliations: INSERM, U590, IFR62, Lyon, France, Univ Lyon, Lyon, France, Centre Léon Bérard, Lyon, France

    X
  • Ruth Rimokh,

    Affiliations: INSERM, U590, IFR62, Lyon, France, Univ Lyon, Lyon, France

    X
  • Nabeel Bardeesy,

    Affiliation: Massachusetts General Hospital Cancer Center, Harvard Medical School, Boston, Massachusetts, United States of America

    X
  • Jean-Yves Scoazec,

    Affiliations: Univ Lyon, Lyon, France, INSERM, U865, Faculté Laennec, Lyon, France, Hospices Civils de Lyon, Hôpital Edouard Herriot, Lyon, France

    X
  • Régine Losson,

    Affiliation: IGBMC, Strasbourg, Illkirch-Cedex, France

    X
  • Laurent Bartholin mail

    bartholi@lyon.fnclcc.fr

    Affiliations: INSERM, U590, IFR62, Lyon, France, Univ Lyon, Lyon, France, INSERM “Avenir” group, Lyon, France, Centre Léon Bérard, Lyon, France

    X
  • Published: July 24, 2009
  • DOI: 10.1371/journal.pgen.1000575
Corrections

6 Aug 2009: Vincent DF, Yan K-P, Treilleux I, Gay F, Arfi V, et al. (2009) Correction: Inactivation of TIF1γ Cooperates with KrasG12D to Induce Cystic Tumors of the Pancreas. PLoS Genet 5(8): 10.1371/annotation/7941c465-532f-4b42-b541-72d0810943b9. doi: 10.1371/annotation/7941c465-532f-4b42-b541-72d0810943b9 | View correction

Abstract

Inactivation of the Transforming Growth Factor Beta (TGFβ) tumor suppressor pathway contributes to the progression of Pancreatic Ductal AdenoCarcinoma (PDAC) since it is inactivated in virtually all cases of this malignancy. Genetic lesions inactivating this pathway contribute to pancreatic tumor progression in mouse models. Transcriptional Intermediary Factor 1 gamma (TIF1γ) has recently been proposed to be involved in TGFβ signaling, functioning as either a positive or negative regulator of the pathway. Here, we addressed the role of TIF1γ in pancreatic carcinogenesis. Using conditional Tif1γ knockout mice (Tif1γlox/lox), we selectively abrogated Tif1γ expression in the pancreas of Pdx1-Cre;Tif1γlox/lox mice. We also generated Pdx1-Cre;LSL-KrasG12D;Tif1γlox/lox mice to address the effect of Tif1γ loss-of-function in precancerous lesions induced by oncogenic KrasG12D. Finally, we analyzed TIF1γ expression in human pancreatic tumors. In our mouse model, we showed that Tif1γ was dispensable for normal pancreatic development but cooperated with Kras activation to induce pancreatic tumors reminiscent of human Intraductal Papillary Mucinous Neoplasms (IPMNs). Interestingly, these cystic lesions resemble those observed in Pdx1-Cre;LSL-KrasG12D;Smad4lox/lox mice described by others. However, distinctive characteristics, such as the systematic presence of endocrine pseudo-islets within the papillary projections, suggest that SMAD4 and TIF1γ don't have strictly redundant functions. Finally, we report that TIF1γ expression is markedly down-regulated in human pancreatic tumors by quantitative RT–PCR and immunohistochemistry supporting the relevance of these findings to human malignancy. This study suggests that TIF1γ is critical for tumor suppression in the pancreas, brings new insight into the genetics of pancreatic cancer, and constitutes a promising model to decipher the respective roles of SMAD4 and TIF1γ in the multifaceted functions of TGFβ in carcinogenesis and development.

Author Summary

Inactivation of the TGFβ tumor suppressor pathway contributes to the progression of Pancreatic Ductal AdenoCarcinoma (PDAC), a devastating malignancy. Transcriptional Intermediary Factor 1γ (TIF1γ) has recently been proposed to be involved in TGFβ signaling, a pathway inactivated in virtually all cases of this malignancy. To address the role of TIF1γ in pancreatic carcinogenesis, we used conditional Tif1γ knockout mice. In a genetic background expressing a constitutively active mutation of KRAS oncogene (KrasG12D) recurrently found in patients with PDAC, Tif1γ inactivation induces pancreatic precancerous lesions resembling those observed in the absence of Smad4, a key player involved TGFβ signal transduction. This observation strengthens the notion that TIF1γ plays an active role in TGFβ signaling. Interestingly, we also found that TIF1γ expression was markedly down-regulated in human pancreatic tumors supporting the relevance of our findings to human malignancy. Characterization of new players involved in the outbreak of early pancreatic lesions that will eventually evolve into invasive pancreatic cancer is crucial to detect the disease earlier and eventually develop new therapeutic drugs.

Introduction

Pancreatic Ductal AdenoCarcinoma (PDAC), characterized by a ductal cell-type differentiation pattern, is the most common type of pancreatic cancer, accounting for more than 85% of pancreatic neoplasms. PDAC is the fourth leading cause of cancer-related mortality and carries an overall 5-year-survival rate of less than 5% [1]. The poor outcome of these patients is due to late diagnosis and resistance to current therapies. PDAC appears to arise from precursor lesions known as Pancreatic Intraepithelial Neoplasia (PanINs) or from two types of cystic tumors: Mucinous Cystic Neoplasms (MCNs) and Intraductal Papillary Mucinous Neoplasms (IPMNs) [2]. Mucinous cystic neoplasms are cysts lined by mucin-producing epithelial cells usually associated with an ovarian-type of stroma. These cysts do not communicate with the larger pancreatic ducts. IPMNs form intraductal papillary projections replacing the normal duct epithelium, secrete mucin, and communicate with ducts. IPMNs are currently classified according to their pattern of apparent histological differentiation into three main subtypes: intestinal (with the neoplastic epithelium resembling the intestinal epithelium), the most frequent, pancreatobiliary and gastric [3].

Recurrent genetic alterations have been identified in human PDAC [4],[5]. Sporadic cases, which represent the vast majority of PDAC, are associated with activation of the KRAS oncogene (>90% of cases) and inactivation of the INK4A/ARF (>80% of cases), TP53 (>50% of cases) and SMAD4/DPC4 (>50% of cases) tumor suppressors. Inherited pancreatic cancers represent approximately 5–10% of all pancreatic cancers. In a high proportion of familial pancreatic cancers, the genetic alterations causing the disease are still unknown. However, several germinal mutations associated with complex familial syndromes have been shown to significantly increase the risk of developing pancreatic cancer (BRCA2, INK4A, STK11/LKB1, PRSS1, hMLH1 and hMSH2) [1],[6].

In the last five years, a series of genetically engineered mouse models of PDAC have been developed based on these signature gene mutations [7]. For instance, expression of a constitutively active Kras mutant protein (KrasG12D or KrasG12V) induces PanINs that eventually progress towards PDAC [8],[9],[10]. Kras activated mutants act in concert with inactivation of the p53 [11], Ink4A/Arf [12],[13], and TβRII [14] tumor suppressors to accelerate development of PDAC. These models and others [15][19] support the concept that progression towards invasive PDAC involves emergence from different precancerous lesions (PanINs, MCNs and IPMNs) depending on the associated genetic alterations.

The TGFβ pathway appears to be of particular importance to PDAC tumor suppression, since it is inactivated in virtually all cases of this malignancy [20], and since genetic lesions inactivating the pathway—inactivation of Smad4 or TβRII and over-expression of inhibitory Smad7— contribute to pancreatic tumor progression in mouse models [14], [21][24]. Transforming growth factor beta (TGFβ) is a secreted polypeptide belonging to a wide family of cytokines and growth factors including TGFβs, Bone Morphogenetic Proteins (BMPs) and activins [25],[26]. Upon binding to its receptors, TGFβ triggers phosphorylation of the SMAD2 and SMAD3 transcription factors. Phosphorylated SMAD2 and SMAD3 then interact with SMAD4. The SMAD2/3/4 complex accumulates within the nucleus, binds to DNA and activates the transcription of target genes leading to proliferative arrest or apoptosis of epithelial cells.

Transcriptional Intermediary Factor 1 gamma (also named TIF1γ/TRIM33/RFG7/PTC7/Ectodermin) [27],[28] appears to contribute to TGFβ signaling, although its precise functional role is not clear. Some data point toward TIF1γ as a negative regulator of the pathway through its capacity to mono-ubiquinate SMAD4 and limit SMAD4 nuclear accumulation [29],[30],[31]. In contrast, other studies have suggested that TIF1γ plays an important positive role in transducing TGFβ signaling through its interaction with SMAD2 and SMAD3 [32].

Here we wished to determine whether TIF1γ contributes to tumorigenesis consistent with a function within the TGFβ signaling pathway. We have focused on pancreatic exocrine tumors based on the prominent role played by TGFβ signaling in these malignancies. Using a conditional mouse strain, we show for the first time that Tif1γ is an important gene whose loss of function cooperates with KrasG12D activation to induce cystic pancreatic tumors resembling human IPMNs. We also report that TIF1γ expression is down-regulated in human PDAC and some types of precursor lesions, supporting the relevance of our mouse model to human malignancy.

Results/Discussion

To selectively abrogate Tif1γ expression in the pancreas, we crossed conditional Tif1γ knockout mice [33] with Pdx1-Cre mice [34]. Pdx1 is a gene expressed in the common progenitor to all pancreatic lineages during early embryogenesis, hence Pdx1-Cre transgenic mice exhibit recombination of floxed alleles in pancreatic cells from all lineages (endocrine, acinar, centroacinar and ductal cells) [35]. Pdx1-Cre;Tif1γlox/lox animals were born at expected ratios and showed normal lifespan without obvious developmental or physiological alterations. Live imaging techniques (Positron Emission Tomography, PET and Magnetic Resonance Imaging, MRI), histological techniques (immunodetection of insulin, glucagon, PPY, chymotrypsine, F4/80, CD3, MPO), metabolic tests (glucose tolerance) did not reveal any significant differences between wild-type and Pdx1-Cre;Tif1γlox/lox littermates (n>20, between 3 weeks and 2 years of age) (data not shown). As expected, immunohistochemistry experiments showed that Tif1γ was expressed in the nuclei of pancreatic cells in wild-type mice and that this staining was lost in the Pdx1-Cre;Tif1γlox/lox pancreas (Figure S1). In all, these observations show that Tif1γ is dispensable for normal pancreatic development and function in the mouse.

Activating KRAS mutations occur early in human PDAC pathogenesis and give rise to slowly progressing PanINs in mouse models [8],[10],[12]. We then asked whether Tif1γ inactivation could modify the phenotype or latency of the pancreatic lesions induced by KrasG12D. To that end, we generated Pdx1-Cre;LSL-KrasG12D;Tif1γlox/lox mice (n = 12, Table S1). All animals looked healthy at the time they were euthanized (the oldest animal was sacrificed at the age of 189 days). Since pancreatic lesions are often asymptomatic, we decided to explore in vivo the pancreas of these mutant mice (n = 4) by PET and MRI imaging techniques. Pdx1-Cre;LSL-KrasG12D;Ink4A/Arflox/lox mice, which exhibit rapid PDAC progression, were also employed in these studies. Strikingly, MRI imaging performed on Pdx1-Cre;LSL-KrasG12D;Tif1γlox/lox animals revealed an hypertrophic pancreas with multifocal cystic lesions exhibiting T2 hypersignals visible as early as 7 weeks after birth (T2 weighted scans allow detection of cysts as they are sensitive to water content) (Figure 1B). Such lesions were absent in the pancreas of wild-type and Pdx1-Cre;LSL-KrasG12D animals (Figure 1A) and were clearly different from those observed in Pdx1-Cre;LSL-KrasG12D;Ink4A/Arflox/lox mice, which harbor solid tumors exhibiting a T1 isosignals (T1 weighed scans allow detection of solid tumors) (Figure 1C). PET imaging did not show significant increased metabolic activity in the abdomen of wild-type (Figure 1D) and Pdx1-Cre;LSL-KrasG12D;Tif1γlox/lox mice (Figure 1E) whereas Pdx1-Cre;LSL-KrasG12D;Ink4A/Arflox/lox mice had abdominal lesions with readily detectable metabolic activity (Figure 1F). Macroscopic analysis of the Pdx1-Cre;LSL-KrasG12D;Tif1γlox/lox pancreas confirmed the presence of numerous cysts affecting the entire organ without macroscopic evidence of invasive carcinoma (Figure 1H and Figure S2) whereas the pancreas of Pdx1-Cre;LSL-KrasG12D;Ink4A/Arflox/lox mice was invaded by a firm and homogeneous mass (Figure 1I) and the pancreas from wild-type or Pdx1-Cre;LSL-KrasG12D animals had a normal macroscopic appearance (Figure 1G). The size of these cysts observed in the pancreas from Pdx1-Cre;LSL-KrasG12D;Tif1γlox/lox was variable and most of them contained papillary projections (Figure 1K). These lesions clearly contrast with the invasive tumors of ductal morphology identified as PDAC in the pancreas of Pdx1-Cre;LSL-KrasG12D;Ink4A/Arflox/lox (Figure 1L). The histological analysis of the 12 Pdx1-Cre;LSL-KrasG12D;Tif1γlox/lox animals (Table S1) revealed the presence of cystic lesions in 100% of these mice, such cysts being never observed in Pdx1-Cre;LSL-KrasG12D or wild type control animals. Quantitative analysis revealed that the area occupied by the abnormal pancreas exceeded 50% by the age 6 weeks in 6/7 Pdx1-Cre;LSL-KrasG12D;Tif1γlox/lox mice whereas it represented less than 20% in 6/6 Pdx1-Cre;LSL-KrasG12D mice (Table S1). Collectively these data demonstrate that inactivation of Tif1γ actively cooperates with activated KrasG12D to induce cystic tumors of the pancreas.

thumbnail

Figure 1. Homozygous deletion of Tif1γ cooperates with activated KrasG12D mutation to induce cystic tumors of the pancreas.

Magnetic Resonance Imaging (MRI) (A–C). Positron Emission Tomography (PET) (D–F). Gross anatomy of the pancreas after dissection (G–I). Hematoxylin Phloxine Saffron (HPS) staining of pancreas sections (J–L). Acinar tissue (stars), scattered endocrine islets (arrow) and exocrine ducts (arrowhead). When notified, pancreas is circled in red dashed lines. Pdx1-Cre;LSL-KrasG12D;Tif1γlox/lox used in this experiment was 154 days old and representative of the 4 mice we analyzed at the age of 18–27 weeks.

doi:10.1371/journal.pgen.1000575.g001

To carefully compare the pancreatic lesions observed in Pdx1-Cre;LSL-KrasG12D;Tif1γlox/lox with the Pdx1-Cre;LSL-KrasG12D controls, we performed a sequential histological analysis of pancreas from animals (n = 12) euthanized at different ages (Table S1). Contrary to wild-type mice (Figure 2A–2D), Pdx1-Cre;LSL-KrasG12D mice gradually developed focal PanINs by the age of about 10 weeks (Figure 2E–2H). Strikingly, in Pdx1-Cre;LSL-KrasG12D;Tif1γlox/lox pancreas, PanINs, signs of acute inflammation as well as enlarged and dilated ductal structure resembling budding cysts were observed as early as 3 weeks of age (Figure 2I and Figure S3). At later time points, inflammatory tissue and PanINs were mainly replaced with cystic lesions becoming more numerous and of larger size (Figure 2J–2L). Microscopic examination revealed that the lining of the cystic structures characteristically found in the pancreas from Pdx1-Cre;LSL-KrasG12D;Tif1γlox/lox mice was formed by epithelial cells with a cuboidal or cylindrical morphology. These cells formed numerous thick papillary projections in the cyst lumen. The axis of these projections usually contained masses of small monomorphic cells with an endocrine morphology (Insets in Figure 2I–2L).

thumbnail

Figure 2. Inactivation of Tif1γ accelerates activated KrasG12D–mediated pancreatic neoplasia.

Sections of wild-type (A–D), Pdx1-Cre;LSL-KrasG12D (E–H) and Pdx1-Cre;LSL-KrasG12D;Tif1γlox/lox (I–L) pancreas from mice at different ages were stained with HPS. Number of mice: 3 weeks, n = 3; 6–13 weeks, n = 5; 18–27 weeks, n = 4. Inset pictures: Higher magnification.

doi:10.1371/journal.pgen.1000575.g002

We performed immunohistochemical studies to characterized the evolving pancreatic lesions in these mice. We observed staining for chymotrypsin and insulin, which decreased with age, indicating a replacement of exocrine and endocrine components, together with abnormal ductal structures in Pdx1-Cre;LSL-KrasG12D;Tif1γlox/lox mice (Figure 3). There was a notable disappearance of well-organized endocrine islets with age coinciding with the accumulation of endocrine cells within the papillary projections bulging within the lumen of the cysts.

thumbnail

Figure 3. Endocrine and exocrine compartments destruction in the pancreas of Pdx1-Cre;LSL-KrasG12D;Tif1γlox/lox mice.

Wild-type (A–C′) and Pdx1-Cre; LSL-KrasG12D; Tif1γlox/lox (D–L′) pancreas sections were stained with HPS (A,D,G,J), immuno-revealed either with an anti-chymotrypsin antibody (B,E,H,K) or an anti-insulin antibody (C,F,I,L and C′,F′,I′,L′ that represent the boxed region in C,F,I,L at a higher magnification). For Pdx1-Cre;LSL-KrasG12D;Tif1γlox/lox mice, analysis were performed at different ages. Number of mice: 3 weeks, n = 3; 6–13 weeks, n = 5; 18–27 weeks, n = 4. Boxes: Higher magnification. The pictures show one representative mouse for each group.

doi:10.1371/journal.pgen.1000575.g003

To more precisely identify the nature of these lesions, several lineage markers were explored by immunohistochemistry. We first verified that Tif1γ expression was lost in pancreatic ducts from Pdx1-Cre;LSL-KrasG12D;Tif1γlox/lox mice (Figure 4E) compared to normal observed ducts in wild-type mice (Figure 4A). In the normal pancreas, cytokeratin 19 (CK19) is specifically expressed by ductal cells lining the secretory ducts (Figure 4B). We verified that most of the epithelial cells lining the cystic lesions observed in Pdx1-Cre;LSL-KrasG12D;Tif1γlox/lox mice were positive for CK19 (Figure 4F); this is consistent with a ductal phenotype for these cells. However, in contrast to the cells lining the normal secretory ducts (Figure 4C), many cells lining the cysts were mucus-secreting and stained for Alcian blue (Figure 4G). The cells with an endocrine appearance present within the intra-cystic papillary projections were CK19 and Alcian blue negative. There was no evidence of ovarian-type stroma or of invasive or microinvasive carcinoma, even on serial sections, suggesting that these cystic tumors resemble human IPMNs. In all, the cystic lesions observed in Pdx1-Cre;LSL-KrasG12D;Tif1γlox/lox mice show distinctive characteristics, including the presence of intra-epithelial endocrine pseudo-islets, suggestive of a mixed, endocrine-exocrine, lesion [7].

thumbnail

Figure 4. Immunophenotyping of pancreas cysts observed in Pdx1-Cre;LSL-KrasG12D;Tif1γlox/lox mice.

Wild-type (A–D) and Pdx1-Cre;Tif1γlox/lox;LSL-KrasG12D (E–H) pancreas sections were stained to detect glandular cells (CK19), mucus-secreting cells (Alcian Blue), Tif1γ and Smad4. The pictures show one representative mouse at the age of 22 weeks.

doi:10.1371/journal.pgen.1000575.g004

TIF1γ has recently been proposed to be involved in TGFβ signaling [29],[32]. The resemblance between the cystic lesions (either IPMNs [21],[23] or MCNs [22]) observed in Pdx1-Cre;LSL-KrasG12D;Smad4lox/lox mice, and the cystic lesions we observed in the Pdx1-Cre;LSL-KrasG12D;Tif1γlox/lox mice, reinforces an active role of TIF1γ in TGFβ signaling. However, we cannot rule out the possibility that TIF1γ could also be involved in other signaling pathways. Interestingly, we observed that Smad4 expression was almost undetectable in epithelial cells lining the papillary projections observed in Pdx1-Cre;LSL-KrasG12D;Tif1γlox/lox mice whereas it was detectable in epithelial cells lining the cysts (Figure 4H) or in normal ducts (Figure 4D). IPMNs observed in Pdx1-Cre;LSL-KrasG12D;Tif1γlox/lox mice always contain a significant endocrine component, a rare event in IPMNs observed in Pdx1-Cre;LSL-KrasG12D;Smad4lox/lox mice [21],[23]. This observation suggests that TIF1γ and SMAD4 could differentially regulate endocrine versus exocrine differentiation in a context of an activated KRAS oncogenic protein.

Based on the prominent cooperation noted between Kras activation and Tif1γ inactivation in promoting cystic pancreatic tumors in our mouse model, we speculated that TIF1γ expression may be down-regulated in human pancreatic tumors. To test this hypothesis, we analyzed by quantitative RT-PCR the expression level of TIF1γ and SMAD4 mRNA in 20 PDAC and 16 peritumoral tissues coming from surgical specimens removed for therapeutic purposes (peritumoral tissues were not available for 4 of these patients). The cellularity of the samples used for molecular analysis was verified histologically. Our results show that TIF1γ expression is significantly decreased in the tumors as compared to peritumoral tissues (P = 0.0054) (Figure 5A). We also compared TIF1γ expression levels in each individual tumor along with the peritumoral tissue from the same patient (n = 16). Our results show that TIF1γ expression is significantly down-regulated in most patients and is not up-regulated in any patient (Figure 5B). We next examined TIF1γ protein pattern of expression by immunohistochemistry in human pancreatic cancers and their precursors. In peritumoral tissues from PDAC, TIF1γ was detected in the majority of the nuclei of acinar, ductal and endocrine cells (Figure 5C). Centroacinar cells are more difficult to identify in routinely stained sections; however, since no epithelial cell population devoid of TIF1γ expression has been detected in the normal pancreas, it can be assumed that they also express TIF1γ. In PDAC (16 cases), TIF1γ nuclear expression level was significantly decreased as compared to the peritumoral tissue. In 8 cases, TIF1γ expression was heterogeneous, with large numbers of negative cells coexisting with scattered positive cells (Figure 5D). In 2 cases, TIF1γ was even undetectable (Figure 5E). In IPMNs (samples from 10 patients, all with the intestinal subtype according to current classifications [3]), almost all neoplastic cells in areas of low grade dysplasia (present in the 10 cases) displayed a weak nuclear positivity whereas in areas of high grade dysplasia (present in the 10 cases), more than 50% of cells were negative for TIF1γ; this was especially the case along the papillary projections (Figure 5F). In PanINs (samples from 15 patients, with grade 1 in 12, grade 2 in 10 and grade 3 in 8), the expression of TIF1γ was usually retained in grade 1 and 2 lesions (data not shown), but was undetectable in a variable proportion of cells in grade 3 lesions (Figure 5G). In MCNs (8 cases), TIF1γ protein was strongly expressed by all neoplastic cells, even in areas of high grade dysplasia and in foci of microinvasive carcinoma (Figure 5H). We showed in the same set of tumors that SMAD4 expression was also down-regulated in high-grade PanINs, IPMNs and PDAC, while remained highly expressed in MCNs (Figure S4).

thumbnail

Figure 5. TIF1γ expression in human pancreatic neoplasia.

The expression of TIF1γ from 20 human PDAC and 16 peritumoral tissues was determined by quantitative RT–PCR (A). The average expression (±sd) is plotted for each group (B). The significance (P value) for the difference between peritumoral and tumoral groups, determined by a Student's T test, is shown above the graphs. The expression of TIF1γ protein was evaluated by immunohistochemistry in different pancreatic lesions: PDAC (n patients = 20) (peritumoral (C) and tumoral tissues (D, E). IPMN (n patients = 10) of high grade dysplasia (F), high-grade PanINs (n patients = 8) (G) and MCN (n patients = 15) (H). Inset pictures: higher magnification.

doi:10.1371/journal.pgen.1000575.g005

None of the Pdx1-Cre;LSL-KrasG12D;Tif1γlox/lox mice (n = 4) sacrificed after the age of 13 weeks showed an aggressive cancer developed from IPMNs. This observation is consistent with epidemiological data in humans showing that IPMNs only rarely give rise to aggressive tumors [3]. Interestingly, IPMNs were reported in Pdx1-Cre;LSL-KrasG12D;Smad4lox/lox mice but no PDAC was found by the age of 13 weeks (n = 8) [21]. Another group observed in Pdx1-Cre;LSL-KrasG12D;Smad4lox/lox a significant proportion of PDAC between 23 and 33 weeks [23]. This suggests that the minimal latency period to see the onset of aggressive tumors may have not been reached in our study or the number of animals studied is too low. We have been in the process of “aging” a cohort of Pdx1-Cre;LSL-KrasG12D;Tif1γlox/lox mice to address this specific point even if we cannot rule out the possibility that these animals die before developing aggressive tumors because of pancreatic failure due to growing cysts. The molecular mechanism supporting the cooperation between activated KrasG12D mutation and Tif1γ inactivation to induce the formation of IPMNs is still unknown. TAK1 (TGFβ Associated Kinase 1), which has recently been proposed to explain R-Ras and TGFβ cooperation in breast tumors, is an interesting candidate [36].

The relationship between TIF1γ and SMAD4 and their respective role in TGFβ signaling have been a subject of extensive investigation and debate in the last few years [37]. Indeed, published data support distinct models whereby TIF1γ could function as either a negative regulator of TGFβ signaling [29],[30] or a complementary agonist of TGFβ signaling [32]. In the “antagonist model”, TIF1γ negative function relies on its ability to mono-ubiquitinate and relocate SMAD4 into the cytoplasm. In the “agonist model”, TIF1γ competes with SMAD4 for binding to SMAD2 and 3 and form TIF1γ-SMAD2/3 complexes regulating SMAD4-independent TGFβ responses. One can envision that these models, both supported by compelling biochemical and in vivo evidence, are not mutually exclusive and that one of them may be predominant depending on the cellular context. The experimental evidence we present here suggest that TIF1γ works with SMAD4 as a complementary agonist molecule during pancreatic tumorigenesis. In the presence of activated Kras, Tif1γ loss-of-function induces cystic lesions resembling those observed in the absence of Smad4 suggesting that both molecules act in concert to prevent tumor progression. Even if this hypothesis needs further demonstration, it is strengthened by the observation that TIF1γ expression is decreased in human pancreatic tumors and our observation that loss of Tif1γ does not significantly impair Smad4 expression level or Smad4 target genes expression (data not shown). The existence of a joint effort between TIF1γ and SMAD4 to maintain TGFβ-mediated tissue homeostasis has been proposed before. Indeed, during erythroid differentiation, TIF1γ mediates the differentiation response while SMAD4 mediates the antiproliferative response [32]. In a recent work, we showed that Tif1γ controlled iNKT (invariant Natural Killer T) cell expansion whereas Smad4 maintained their maturation state [33]. A recent comprehensive genetic analysis of >20,000 transcripts in 24 pancreatic cancers failed to identify point mutations, amplifications, deletion or translocations in the TIF1γ gene [20]. The present study strongly spurs us toward looking for TIF1γ genetic alterations in a larger set of pancreatic tumors. Besides, chromosomal breakpoints chromosome on 1p13.1 containing TIF1γ gene have been reported in acute megakaryocytic leukemias [38], osteochondromas [39], bronchial large cell carcinomas [40] and childhood papillary thyroid carcinomas [41]. Interestingly, we recently demonstrated that abrogation of the closely related Tif1α gene in mice caused hepatocellular carcinoma [42]. These observations reinforce the idea according to which TIF1γ loss of function could play an active protective role during tumorigenesis. TIF1γ overexpression has been suggested by others to facilitate tumorigenesis in other organs by inhibiting SMAD4-mediated growth inhibition and motility in response to TGFβ [29],[30]. This observation may reflect an active role of TIF1γ during tumor progression depending on the organ and involving a anti-SMAD4 mechanisms (“antagonist” model).

In conclusion, we demonstrated in a mouse model that inactivation of Tif1γ cooperates with activated KrasG12D to induce cystic pancreatic tumors. Characterization of new players involved in the outbreak of early pancreatic lesions that will eventually evolve into invasive pancreatic cancer is crucial to detect the disease earlier and eventually develop new therapeutic drugs. Further work to decipher the respective roles of SMAD4 and TIF1γ in PDAC as well as the functional cooperation between KRAS and TIF1γ could bring new insight into the etiology of pancreatic cancer, and generate a better understanding of the multifaceted role of TGFβ in carcinogenesis and development.

Materials and Methods

Mice

Tif1γlox/lox [32] mice harboring floxed exons 2–4 were generated by K.Y. and R.L. will be described elsewhere. Briefly, using a genomic clone that contains a portion of the Tif1γ gene, we generated a targeting vector in which a PGK Neo selection cassette flanked by two loxP sites was introduced into intron 1 and a third loxP site inserted into intron 4. This targeting vector was designed with the expectation that upon homologous recombination and subsequent Cre recombinase-mediated excision, exons 2, 3 and 4 along with the PGK-Neo cassette would be deleted, thereby causing a frameshift mutation with a premature termination codon in exon 5. The putative product of this deleted gene corresponds to a truncated protein lacking part of the RING finger-B box-coiled coil (RBCC) motif and the entire C-terminal region of the TIF1γ protein, which contains the conserved PHD finger/bromodomain unit.

Ink4A/Arflox/lox [12], LSL-KrasG12D [8], [43][45] and Pdx1-Cre [34] mouse strains done by others were previously described. Mice were maintained in a specific pathogen-free animal facility at the “Centre Léon Bérard” (Lyon, France) and handled in compliance with the institutional guidelines. All procedures were approved by an ethic committee under regulatory of governmental authority (CREEA).

Histology

Histology experiments were performed as previously described [46]. Primary antibody used were : TIF1γ (1/800; Euromedex), Chymotrypsin (1/1000; AbD Serotec), Insulin (1/1000; Dako), CK19 (1/10; Trauma III - Developmental Studies Hybridoma Bank - University of Iowa), SMAD4 (1/100; Santa Cruz). Alcian blue staining was performed as previously described [47].

Human samples

Cryopreserved tumoral and peritumoral tissue samples were obtained from an institutional tissue bank, the Tumorothèque des Hospices Civils de Lyon (Centre de Ressources Biologiques, Hospices Civils de Lyon). In accordance with French ethical rules, samples were from patients having given their informed consent or from deceased patients. Prior to molecular analysis, the quality and cellularity of tissue samples was verified histologically; tumor tissues were selected in order to contain a significant amount of neoplastic cells; peritumoral tissues were constantly altered by reactive fibrotic changes associated with a loss in acinar tissue and a massive ductular proliferation.

RNA analysis

Liquid nitrogen frozen human tumors were blended using a Pro200 homogenizer (Pro Scientific Inc.) in a 5 M guanidine solution. Total RNA was further purified by RNeasy mini kit (Qiagen). The cDNA was used as template with RT Kit SuperScript II (Invitrogen). quantitative RT-PCR was performed as previously described [48].

Live imaging

For MRI and PET experiments, mice were anesthetized using 3% isoflurane inhalation (TEM Sega, Lormont, France) and maintained in 1.5% isoflurane atmosphere during experiments.

For PET experiments, the mice were catherized in the caudal vein (24 gauge), injected with 250 µCi of 300 µL of radioactive 18-Fluorodéoxyglucose (FDG). After 90 minutes to allow FDG fixation, images were acquired during 15 minutes (constant 2% isoflurane atmosphere) using the “TEP clearPET” (Raytest, Inc.). MRI acquisitions were made with a BioSpec-7T system (Bruker, Ettlingen, Germany) using a 32 mm inner-diameter emission/reception volume coil (Rapid Biomedical, Würzburg, Germany). T1/T2-weighted contrast sequences synchronized to respiration were acquired for each mouse. A RARE (Rapid Acquisition with Relaxation Enhancement) sequence (TR/TE 3500/38.1 ms) with fat saturation was used. Geometric parameters were: a series of 18, 750 µm thick sections, 33 mm field of view, and 256×256 pixel matrix. Voxel size was therefore 129×129×750 µm3.

Supporting Information

Figure S1.

Loss of nuclear Tif1γ protein expression in the pancreas of Pdx1-Cre; Tif1γlox/lox mice. Immunohistochemistry showed that Tif1γ was expressed in the nuclei of pancreatic cells in wild-type mice and that this staining was lost in a Pdx1-Cre;TIF1γlox/lox pancreas.

doi:10.1371/journal.pgen.1000575.s001

(2.25 MB TIF)

Figure S2.

Image of a polycystic pancreas from a 154-day-old Pdx1-Cre; LSL-KrasG12D; Tif1γlox/lox mouse. Pancreas is circled in black. Note the presence of numerous translucent cysts (arrowhead).

doi:10.1371/journal.pgen.1000575.s002

(2.96 MB TIF)

Figure S3.

Presence of inflammatory cells in the pancreas associated with the cystic structures. Immunohistochemistry revealed the presence of different populations of leucocytes infiltrating the pancreas of a 20-day-old Pdx1-Cre; LSL-KrasG12D; Tif1γlox/lox mouse (D–F) compared to a wild-type littermate (A–C). MPO (A,D), F4/80 (B,E), and CD3 (C,F) were respectively used as specific markers for neutrophiles, macrophages and lymphocytes.

doi:10.1371/journal.pgen.1000575.s003

(2.71 MB TIF)

Figure S4.

SMAD4 expression in human pancreatic neoplasia. Quantitative RT–PCR to detect SMAD4 expression from the 16 PDAC for which peritumoral tissue was available (A). For each patient represented by an individual bar, SMAD4 expression is represented as a percentage of variation relative to mRNA expression in the peritumoral tissue from the same patient (B). SMAD4 protein expression pattern was also assessed by immunohistochemistry. In the normal pancreas (C), SMAD4 is strongly detected in endocrine islets, a faint labeling is visible in acinar cells and ductal cells. In adenocarcinomas (n patients = 20), no labeling for SMAD4 was detected whereas adjacent residual endocrine cells are positive (D). In another example of adenocarcinoma, SMAD4 is faintly but readily detectable in neoplastic cells; the labeling is both cytoplasmic and nuclear (E). In IPMN grade 3, SMAD4 expression is heterogeneous; most cells are negative, while a few scattered cells retain a faint expression, usually nuclear (F). In PanIN grade 3, SMAD4 expression is either undetectable, as exemplified in the largest figure (note the persistent expression in adjacent residual endocrine cells (arrow)), or heterogeneous, as shown in the inset (G). In MCN, SMAD4 expression is usually strong, in low grade (large figure) as well as in high grade (inset) lesions (H).

doi:10.1371/journal.pgen.1000575.s004

(3.34 MB TIF)

Table S1.

Measurement of the area occupied by the normal pancreatic tissue.

doi:10.1371/journal.pgen.1000575.s005

(0.06 MB DOC)

Acknowledgments

We thank G. Gouysse, J.B. Langlois, S. Martin, M. Janier, P. Meeus, and M. Peix for their expert assistance. We are deeply grateful to D. Wotton for advice, helpful discussions, and proofreading of the manuscript. We thank T. Jacks/R. DePinho, D.Tuveson, and D. Melton for providing Ink4A/Arflox/lox, LSL-KrasG12D, and Pdx1-Cre mouse strains.

Author Contributions

Conceived and designed the experiments: DFV SM LB. Performed the experiments: DFV KPY IT FG VA BK FL SM SGL SG JYS RL LB. Analyzed the data: DFV IT VA BK JCM SM JLI PD SG AP RR NB JYS RL LB. Contributed reagents/materials/analysis tools: DFV KPY IT VA JCM NB JYS RL LB. Wrote the paper: RL LB.

References

  1. 1. Maitra A, Hruban RH (2008) Pancreatic cancer. Annu Rev Pathol 3: 157–188.
  2. 2. Maitra A, Fukushima N, Takaori K, Hruban RH (2005) Precursors to invasive pancreatic cancer. Adv Anat Pathol 12: 81–91.
  3. 3. Furukawa T, Kloppel G, Volkan Adsay N, Albores-Saavedra J, Fukushima N, et al. (2005) Classification of types of intraductal papillary-mucinous neoplasm of the pancreas: a consensus study. Virchows Arch 447: 794–799.
  4. 4. Koorstra JB, Hustinx SR, Offerhaus GJ, Maitra A (2008) Pancreatic carcinogenesis. Pancreatology 8: 110–125.
  5. 5. Bardeesy N, DePinho RA (2002) Pancreatic cancer biology and genetics. Nature Reviews Cancer 2: 897–909.
  6. 6. Shi C, Hruban RH, Klein AP (2009) Familial pancreatic cancer. Arch Pathol Lab Med 133: 365–374.
  7. 7. Hruban RH, Adsay NV, Albores-Saavedra J, Anver MR, Biankin AV, et al. (2006) Pathology of genetically engineered mouse models of pancreatic exocrine cancer: consensus report and recommendations. Cancer Res 66: 95–106.
  8. 8. Hingorani SR, Petricoin EF, Maitra A, Rajapakse V, King C, et al. (2003) Preinvasive and invasive ductal pancreatic cancer and its early detection in the mouse. Cancer Cell 4: 437–450.
  9. 9. Guerra C, Schuhmacher AJ, Canamero M, Grippo PJ, Verdaguer L, et al. (2007) Chronic pancreatitis is essential for induction of pancreatic ductal adenocarcinoma by K-Ras oncogenes in adult mice. Cancer Cell 11: 291–302.
  10. 10. Grippo PJ, Nowlin PS, Demeure MJ, Longnecker DS, Sandgren EP (2003) Preinvasive pancreatic neoplasia of ductal phenotype induced by acinar cell targeting of mutant Kras in transgenic mice. Cancer Research 63: 2016–2019.
  11. 11. Hingorani SR, Wang L, Multani AS, Combs C, Deramaudt TB, et al. (2005) Trp53R172H and KrasG12D cooperate to promote chromosomal instability and widely metastatic pancreatic ductal adenocarcinoma in mice. Cancer Cell 7: 469–483.
  12. 12. Aguirre AJ, Bardeesy N, Sinha M, Lopez L, Tuveson DA, et al. (2003) Activated Kras and Ink4a/Arf deficiency cooperate to produce metastatic pancreatic ductal adenocarcinoma. Genes & Development 17: 3112–3126.
  13. 13. Bardeesy N, Aguirre AJ, Chu GC, Cheng KH, Lopez LV, et al. (2006) Both p16(Ink4a) and the p19(Arf)-p53 pathway constrain progression of pancreatic adenocarcinoma in the mouse. Proc Natl Acad Sci U S A 103: 5947–5952.
  14. 14. Ijichi H, Chytil A, Gorska AE, Aakre ME, Fujitani Y, et al. (2006) Aggressive pancreatic ductal adenocarcinoma in mice caused by pancreas-specific blockade of transforming growth factor-beta signaling in cooperation with active Kras expression. Genes Dev 20: 3147–3160.
  15. 15. Hezel AF, Gurumurthy S, Granot Z, Swisa A, Chu GC, et al. (2008) Pancreatic LKB1 deletion leads to acinar polarity defects and cystic neoplasms. Mol Cell Biol 28: 2414–2425.
  16. 16. Siveke JT, Einwachter H, Sipos B, Lubeseder-Martellato C, Kloppel G, et al. (2007) Concomitant pancreatic activation of Kras(G12D) and Tgfa results in cystic papillary neoplasms reminiscent of human IPMN. Cancer Cell 12: 266–279.
  17. 17. De La OJ-P, Emerson LL, Goodman JL, Froebe SC, Illum BE, et al. (2008) Notch and Kras reprogram pancreatic acinar cells to ductal intraepithelial neoplasia. Proc Natl Acad Sci U S A 105: 18907–18912. Epub 12008 Nov 18921.
  18. 18. Nolan-Stevaux O, Lau J, Truitt ML, Chu GC, Hebrok M, et al. (2009) GLI1 is regulated through Smoothened-independent mechanisms in neoplastic pancreatic ducts and mediates PDAC cell survival and transformation. Genes Dev 23: 24–36.
  19. 19. Shi G, Zhu L, Sun Y, Bettencourt R, Damsz B, et al. (2009) Loss of the Acinar-Restricted Transcription Factor Mist1 Accelerates Kras-Induced Pancreatic Intraepithelial Neoplasia. Gastroenterology 136(4): 1368–1378.
  20. 20. Jones S, Zhang X, Parsons DW, Lin JC, Leary RJ, et al. (2008) Core signaling pathways in human pancreatic cancers revealed by global genomic analyses. Science 321: 1801–1806.
  21. 21. Bardeesy N, Cheng KH, Berger JH, Chu GC, Pahler J, et al. (2006) Smad4 is dispensable for normal pancreas development yet critical in progression and tumor biology of pancreas cancer. Genes Dev 20: 3130–3146.
  22. 22. Izeradjene K, Combs C, Best M, Gopinathan A, Wagner A, et al. (2007) Kras(G12D) and Smad4/Dpc4 haploinsufficiency cooperate to induce mucinous cystic neoplasms and invasive adenocarcinoma of the pancreas. Cancer Cell 11: 229–243.
  23. 23. Kojima K, Vickers SM, Adsay NV, Jhala NC, Kim HG, et al. (2007) Inactivation of Smad4 accelerates Kras(G12D)-mediated pancreatic neoplasia. Cancer Res 67: 8121–8130.
  24. 24. Kuang C, Xiao Y, Liu X, Stringfield TM, Zhang S, et al. (2006) In vivo disruption of TGF-beta signaling by Smad7 leads to premalignant ductal lesions in the pancreas. Proc Natl Acad Sci U S A 103: 1858–1863.
  25. 25. Ross S, Hill CS (2008) How the Smads regulate transcription. Int J Biochem Cell Biol 40: 383–408.
  26. 26. Massague J, Gomis RR (2006) The logic of TGFbeta signaling. FEBS Lett 580: 2811–2820.
  27. 27. Venturini L, You J, Stadler M, Galien R, Lallemand V, et al. (1999) TIF1gamma, a novel member of the transcriptional intermediary factor 1 family. Oncogene 18: 1209–1217.
  28. 28. Yan KP, Dolle P, Mark M, Lerouge T, Wendling O, et al. (2004) Molecular cloning, genomic structure, and expression analysis of the mouse transcriptional intermediary factor 1 gamma gene. Gene 334: 3–13.
  29. 29. Dupont S, Zacchigna L, Cordenonsi M, Soligo S, Adorno M, et al. (2005) Germ-layer specification and control of cell growth by Ectodermin, a Smad4 ubiquitin ligase. Cell 121: 87–99.
  30. 30. Dupont S, Mamidi A, Cordenonsi M, Montagner M, Zacchigna L, et al. (2009) FAM/USP9x, a deubiquitinating enzyme essential for TGFbeta signaling, controls Smad4 monoubiquitination. Cell 136: 123–135.
  31. 31. Levy L, Howell M, Das D, Harkin S, Episkopou V, et al. (2007) Arkadia activates Smad3/Smad4-dependent transcription by triggering signal-induced SnoN degradation. Mol Cell Biol 27: 6068–6083.
  32. 32. He W, Dorn DC, Erdjument-Bromage H, Tempst P, Moore MA, et al. (2006) Hematopoiesis controlled by distinct TIF1gamma and Smad4 branches of the TGFbeta pathway. Cell 125: 929–941.
  33. 33. Doisne JM, Bartholin L, Yan KP, Garcia CN, Duarte N, et al. (2009) iNKT cell development is orchestrated by different branches of TGF-beta signaling. J Exp Med 206: 1365–1378.
  34. 34. Gu G, Dubauskaite J, Melton DA (2002) Direct evidence for the pancreatic lineage: NGN3+ cells are islet progenitors and are distinct from duct progenitors. Development 129: 2447–2457.
  35. 35. Jorgensen MC, Ahnfelt-Ronne J, Hald J, Madsen OD, Serup P, et al. (2007) An illustrated review of early pancreas development in the mouse. Endocr Rev 28: 685–705.
  36. 36. Erdogan M, Pozzi A, Bhowmick N, Moses HL, Zent R (2008) Transforming growth factor-beta (TGF-beta) and TGF-beta-associated kinase 1 are required for R-Ras-mediated transformation of mammary epithelial cells. Cancer Res 68: 6224–6231.
  37. 37. Heldin CH, Moustakas A (2006) A new twist in Smad signaling. Dev Cell 10: 685–686.
  38. 38. Ng YY, Fan JM, Mu W, Nikolic-Paterson DJ, Yang WC, et al. (1999) Glomerular epithelial-myofibroblast transdifferentiation in the evolution of glomerular crescent formation. Nephrol Dial Transplant 14: 2860–2872.
  39. 39. Sawyer JR, Thomas EL, Lukacs JL, Swanson CM, Ding Y, et al. (2002) Recurring breakpoints of 1p13 approximately p22 in osteochondroma. Cancer Genet Cytogenet 138: 102–106.
  40. 40. Johansson M, Dietrich C, Mandahl N, Hambraeus G, Johansson L, et al. (1994) Karyotypic characterization of bronchial large cell carcinomas. Int J Cancer 57: 463–467.
  41. 41. Klugbauer S, Rabes HM (1999) The transcription coactivator HTIF1 and a related protein are fused to the RET receptor tyrosine kinase in childhood papillary thyroid carcinomas. Oncogene 18: 4388–4393.
  42. 42. Khetchoumian K, Teletin M, Tisserand J, Mark M, Herquel B, et al. (2007) Loss of Trim24 (Tif1alpha) gene function confers oncogenic activity to retinoic acid receptor alpha. Nat Genet 39: 1500–1506.
  43. 43. Jackson EL, Willis N, Mercer K, Bronson RT, Crowley D, et al. (2001) Analysis of lung tumor initiation and progression using conditional expression of oncogenic K-ras. Genes & Development 15: 3243–3248.
  44. 44. Johnson L, Mercer K, Greenbaum D, Bronson RT, Crowley D, et al. (2001) Somatic activation of the K-ras oncogene causes early onset lung cancer in mice.[see comment]. Nature 410: 1111–1116.
  45. 45. Tuveson DA, Shaw AT, Willis NA, Silver DP, Jackson EL, et al. (2004) Endogenous oncogenic K-ras(G12D) stimulates proliferation and widespread neoplastic and developmental defects. Cancer Cell 5: 375–387.
  46. 46. Treilleux I, Blay JY, Bendriss-Vermare N, Ray-Coquard I, Bachelot T, et al. (2004) Dendritic cell infiltration and prognosis of early stage breast cancer. Clin Cancer Res 10: 7466–7474.
  47. 47. Albores-Saavedra J, Weimersheimer-Sandoval M, Chable-Montero F, Montante-Montes de Oca D, Hruban RH, et al. (2008) The foamy variant of pancreatic intraepithelial neoplasia. Ann Diagn Pathol 12: 252–259.
  48. 48. Bartholin L, Cyprian FS, Vincent D, Garcia CN, Martel S, et al. (2008) Generation of mice with conditionally activated transforming growth factor beta signaling through the TbetaRI/ALK5 receptor. Genesis 46(12): 724–731.