PTEN loss and KRAS activation cooperate in murine biliary tract malignancies

Journal of PathologyJ Pathol 2013; 230: 165–173Published online in Wiley Online Library(wileyonlinelibrary.com) DOI: 10.1002/path.4189

ORIGINAL PAPER

PTEN loss and KRAS activation cooperate in murine biliary tractmalignanciesVictoria Marsh,1 Emma J Davies,1 Geraint T Williams2 and Alan R Clarke1*

1 Cardiff School of Biosciences, Cardiff University, UK2 Institute of Cancer and Genetics, School of Medicine, Cardiff University, UK

*Correspondence to: Alan R Clarke, Cardiff School of Biosciences, Cardiff University, Museum Avenue, Cardiff CF10 3AX, UK. e-mail: [email protected]

AbstractCarcinomas of the biliary tract are aggressive malignancies in humans. Loss of the tumour suppressor PTENhas previously been associated with cholangiocarcinoma development in a murine model. Activation of KRAS isreported in up to one-third of human cholangiocarcinomas and 50% of gall bladder carcinomas. In this study weaimed to test the potential interaction between PTEN and KRAS mutation in biliary tract malignancy. We usedan inducible Cre–LoxP-based approach to coordinately delete PTEN and activate KRAS within the adult mousebiliary epithelium. We found that activation of KRAS alone has little effect upon biliary epithelium. Loss of PTENalone results in the development of low-grade neoplastic lesions, following long latency and at low incidence.Combination of both mutations causes rapid development of biliary epithelial proliferative lesions, which progressthrough dysplasia to invasive carcinoma. We conclude that activation of the PI3′K pathway following lossof PTEN is sufficient to drive slow development of low-grade biliary lesions in mice. In contrast, mutationalactivation of KRAS does not result in a similar phenotype, despite a prediction that this should activate boththe RAF–MEK–ERK and PI3′-kinase pathways. However, mutation of both genes results in rapid tumourigenesis,arguing that PTEN normally functions as a ‘brake’ on the PI3′-kinase pathway, limiting the influence of KRASactivation. Mutation of both genes creates a ‘permissive’ environment, allowing the full effects of both mutationsto be manifested. These data reveal an in vivo synergy between these mutations and provides a new mouse modelof biliary tract malignancy.Copyright 2013 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.

Keywords: PTEN; KRAS; mouse model; biliary tract malignancy

Received 14 November 2012; Revised 20 February 2013; Accepted 27 February 2013

No conflicts of interest were declared.

Introduction

Carcinomas of the biliary tract are rare malignanciesin humans, but are among the most difficult forms ofcancer to treat and generally have a poor prognosis[1]. Most cases are diagnosed late, once lymph node ordistant metastasis has occurred [2] and at stages whendisease is refractory to treatment [1]. Models of biliarytract cancers are potentially of great value to facilitatethe discovery and testing of novel chemotherapeuticagents, as well as to increase basic knowledge of thebiology and progression of disease.

Phosphatase and tensin homologue mutated onchromosome 10 (PTEN ) is a well characterizedhuman tumour suppressor [3,4] and antagonist of thephosphatidylinositol-3′ kinase (PI3′K)–Akt pathway[5]. Whilst loss of expression of PTEN itself is rare inhuman biliary tract malignancies [6], activation of thedownstream target of the PI3′K–Akt pathway, mam-malian target of rapamycin (mTOR), has previouslybeen associated with poor prognosis in human cases

of cholangiocarcinoma [7]. PTEN has also been linkedto development of cholangiocarcinoma following thediscovery that liver-specific deletion of both PTEN andSMAD4 in a mouse model leads to the developmentof intrahepatic cholangiocellular carcinoma [8].

The rat sarcoma virus oncogene homologue (Ras)family of GTPases function as intracellular signaltransducers and are activated in response to a widevariety of extracellular signals. Two of the best-characterized pathways activated downstream of Rasare the RAF–MEK–ERK MAPK signalling cascadeand the PI3′-kinase–Akt pathway [9,10]. All three Rasisoforms (HRAS , KRAS and NRAS ) have oncogenicpotential and are mutated in human cancers [11]. Muta-tional activation of KRAS is commonly observed innon-small cell lung cancers, colorectal cancers andpancreatic cancers [9]. Mutations in KRAS are alsoreported in up to one-third of intrahepatic cholangio-carcinomas [12,13] and about 50% of gall bladdercarcinomas [14] and are associated with poor prog-nosis [15]. Interestingly, the reported mutation rates of

Copyright 2013 Pathological Society of Great Britain and Ireland. J Pathol 2013; 230: 165–173Published by John Wiley & Sons, Ltd. www.pathsoc.org.uk www.thejournalofpathology.com

166 V Marsh et al

KRAS in cholangiocarcinomas vary widely, with muta-tional KRAS activation being much more common ineastern populations [16–18]. This suggests highly com-plex disease aetiology, likely related to ethnicity andenvironmental factors.

Given their convergent effects upon the PI3′K sig-nalling pathway, it might be predicted that loss ofPTEN function and activating KRAS mutations mightoccur with mutual exclusivity. However, evidencefrom mouse models of ovarian [19,20], endometrial[21] and lung cancers [22] indicate a clear synergybetween these mutations in promoting disease progres-sion. Since PTEN and KRAS mutations are associatedwith biliary tract disease, we aimed to test whethercombined deletion of Pten and activation of Kras coop-erate to promote biliary tract malignancy. We used con-ditional transgenesis to drive coordinate loss of PTENand activation of K-RAS within biliary epithelial cellsof the adult mouse.

Materials and methods

Experimental animalsProcedures were conducted within UK Home Officeregulations. Mice were maintained on an outbred back-ground, and were genotyped for the Lox-P-targetedPten allele [23], the k-RasV12 transgene [24], theRosa26R allele [25] and AhCreERT [26]. Cre activ-ity was induced in adults (aged 6–10 weeks) by fourdaily intraperitoneal (i.p.) injections with 80 mg/kg β-naphthoflavone and tamoxifen (Sigma) dissolved incorn oil. Kaplan–Meier survival analysis was per-formed, using SPSS Statistics v. 17.0. Tissue sampleswere frozen for cryosectioning or fixed in ice-cold 10%neutral buffered formalin (Sigma) overnight before pro-cessing and embedding in paraffin wax.

Cryosectioning and staining for β-galactosidaseactivityFor X-Gal staining, tissues were removed, fresh-frozenin OCT medium (R. A. Lamb), and cryosectioned(7 µm). Sections fixed in X-gal fixative [2% formalde-hyde, 0.2% glutaraldehyde, 0.02% NP-40 in phosphate-buffered saline (PBS)] for 10 min at room temperaturewere stained in 0.04% X-gal solution in X-gal stain-ing buffer (1 mM magnesium chloride, 3 mM potassiumferricyanide, 3 mM potassium ferrocyanide in PBS)overnight at 37 ◦C.

Histology and immunohistochemistryFormalin-fixed paraffin-embedded tissues were sec-tioned (5 µm) and stained with haematoxylin and eosin(H&E) for histological analysis or used for immuno-histochemistry (IHC). For histological analysis, tissueswere examined independently by ARC and GTW.For IHC, primary antibodies were: anti-Cytokeratin19 (1:100; Abcam), anti-Ki67 (1:100; Novacastra),anti-Pten (1:100), anti-phospho-p44/p42Thr202/Tyr204

(phospho-ERK1/2; 1:75) and anti-phospho-AktSer473

(1:50) (all Cell Signaling Technology). In all cases,heat-mediated antigen retrieval was performed.Binding of antibody was detected choromogenically,using diaminobenzidine substrate (Dako). Scoring forproliferation was performed from sections stained forKi67 antigen. Positively-stained nuclei were scoredas a percentage of total epithelial cells identified byhistology per field of view. A minimum of three indi-vidual animals were scored per genotype. Statisticalanalysis was performed using the Mann–Whitneytest. Immunofluorescent staining methods available asSupplementary Material.

Results

Activation of AhCreERT causes recombinationwithin biliary epitheliaRecombination within the gall bladder epithelium haspreviously been reported using the AhCreERT trans-gene [26]. To confirm this in our system, animalsbearing the AhCreERT transgene together with theROSA26R LacZ reporter transgene [25] were gener-ated. Maximal Cre activity was induced in adult ani-mals by administration of 80 mg/kg β-naphthoflavoneand tamoxifen (BNF/TAM). At day 5 after treatment,the animals were culled and recombination visualizedby exposure of the tissues to X-Gal substrate. Recombi-nation occurred within a significant number of epithe-lial cells of the gall bladder, as reported by blue stainingof the epithelium (Figure 1A).

AhCreERT-driven PTEN deficiency causes low-levelhyperplasia of the biliary epithelium anddevelopment of low-grade neoplasmsTo examine the consequences of AhCreERT-mediateddeletion of PTEN , animals bearing the Cre transgeneand targeted Pten alleles (AhCreERT+;Ptenf/f, referredto as ‘Ptenf/f’) along with wild-type (WT) controlsbearing only the Cre transgene (AhCreERT+;Pten+/+)were treated with BNF/TAM to induce Cre activity.Recombination within intrahepatic bile ducts was con-firmed using anti-PTEN immunohistochemistry (IHC),revealing absence of staining in bile ducts of Ptenf/f

animals compared to controls (Figure 1B). The animalswere monitored for signs of ill-health and euthanizedonce symptomatic of disease. Kaplan–Meier survivalanalysis of Ptenf/f animals (n = 30) revealed a signif-icant reduction in survival compared to WT controls(n = 29) (Figure 1C). Median survival times were 405and 801 days, respectively (log-rank test; p < 0.0001).

Analysis of tissues from Ptenf/f animals revealed arange of previously reported pathologies. The major-ity of males bore evidence of prostate neoplasia [27].In the liver, hepatic steatosis and focal steatohep-atitis was identified [28]. A spectrum of unreportedpathologies was also noted within the biliary system


PTEN and KRAS cooperate in biliary malignancies 167

A

C

E

F G

D

B

WT Ptenf/f

1.0

0.8

0.6

0.4

0.2

0.0

0.00 200.00 400.00 600.00 800.00 1000.00

Time (days)

Cum

ulat

ive

Sur

viva

l

(i) (ii) (iii) (iv)

(v) (vi) (vii) (viii)

WT Ptenf/f

Biliary System Pathology No Biliary System Pathology

Of 14 Ptenf/f mice bearing biliarypathologies:

BiliaryAdenoma

BiliaryHyper-plasia

Mixed cellneoplasm

1

1

10

23

6

Figure 1. PTEN deletion results in development of mixed hepatocyte/cholangiocyte lesions. (A) The ROSA26R LacZ reporter allele confirmedAhCreERT-driven recombination within biliary epithelial cells. LacZ activity (and hence recombination) was detected in the epithelial cellpopulation of the gall bladder (right panel is an enlargement of the area indicated in the left panel). (B) Anti-PTEN IHC confirmed deletionof PTEN protein in intrahepatic bile ducts (arrows indicate bile duct) in PtenF/F animals (right panel) compared to controls (left panel). (C)Kaplan–Meier analysis of Ptenf/f animals (blue line) and WT controls (red line; tick marks indicate censored data) revealed significantlyreduced survival of Ptenf/f animals. Median survival times were 405 and 801 days post-induction, respectively (p < 0.0001). (D) Histologicalanalysis of bile ducts (left panel) and gall bladder epithelium (right panel) of aged wild-type controls revealed no abnormalities. (E) Incontrast, a range of biliary system abnormalities was identified in aged Ptenf/f animals. Focal biliary hyperplasia of the epithelium ofinterlobular bile ducts (i) and gall bladder were identified. Clusters of branched and/or dilated bile ducts lined by bland epithelium, whichresembled microhamartomas, were also found (ii), together with similar lesions that appeared to be more adenoma-like and had scatteredmitoses (iii; inset shows mitoses indicated with arrows). For our analyses, the term ’biliary adenoma’ was used to encompass these twotypes of lesion. Finally, discrete, low-grade lesions composed of varying proportions of hepatocytes and cholangiocytes were identified,which we have classed as ’mixed-cell neoplasms’. In some cases, these lesions were composed exclusively of cholangiocytes and resembledcholangiocarcinomas (iv). Some lesions were purely composed of hepatocytes, and were determined to be hepatocellular adenomas orcarcinomas (v, vi). Most commonly, tumours had a mixed-cell histology, combining biliary and hepatocellular elements (vii, viii). (F) Scoringof the incidence of biliary system pathologies revealed a marked increase in disease incidence in Ptenf/f animals (14 of 24 mice, 58.3%)compared to aged WT controls (1 of 24 mice, 4.2%). (G) Of the 14 Ptenf/f mice that showed evidence of biliary system pathologies, theincidence of lesions present in individual animals was scored according to the categories described above and is presented as a Venndiagram. Many mice bore more than one category of biliary system pathology.


168 V Marsh et al

of aged Ptenf/f animals (Figure 1E). These patholo-gies were found to fall into three categories. First, wenoted focal hyperplasia of the epithelium of interlob-ular bile ducts, extrahepatic bile ducts and the gallbladder, which were manifested as epithelial crowd-ing and multilayering with occasional mitotic figures(Figure 1Ei). Second, aggregates of small, irregularlybranched and/or dilated bile ducts with rounded con-tours, lined by regular epithelial cells without overtmitotic activity, were identified (Figure 1Eii). Theseresembled so-called von Meyenburg complexes thatoccur in humans and are considered to be benign biliarymicrohamartomas. Many of these clusters were moreadenoma-like, with scattered mitoses (Figure 1Eiii).For the purposes of our analyses, we refer to thesetwo types of lesion as ‘biliary adenomas’. Finally, anumber of aged Ptenf/f animals bore clear neoplasticlesions within their livers. These lesions were welldemarcated from the adjacent liver parenchyma, whichwas not infiltrated. A number of these lesions resem-bled cholangiocarcinomas (Figure 1Eiv). Others werecomprised entirely of well-differentiated hepatocytesarranged in small solid masses or in trabeculae thatresembled liver cell adenomas or low-grade hepato-cellular carcinomas (Figure 1Ev, vi). Most, however,were composed of both hepatocytes and small branch-ing bile ducts that merged together (Figure 1Evii, viii).Mitotic figures were present in both cell components.In some lesions, bile duct elements greatly predom-inated, showing loss of epithelial polarity, irregularcontours, infiltration of neutrophil polymorphs, andoccasionally surrounded by a cellular stroma with alittle periductal collagen condensation. Overall, theselesions were determined to be multifocal low-gradeneoplasms showing bidirectional differentiation, mostlikely of hepatic progenitor cell origin. Here we referto these types of lesions as ‘mixed-cell neoplasms’.

In the Ptenf/f cohort (n = 24), 10 mice (41.7%)showed no evidence of biliary disease, whereas14 mice (58.3%) bore biliary system pathologies(Figure 1F). Of the 14 animals showing evidence ofbiliary pathologies, a number of mice showed morethan one type of pathology. In total, 11 animals borebiliary adenomas, four displayed biliary epithelialhyperplasia and seven bore mixed-cell, low-gradeneoplasms of the type described above (Figure 1G). Incontrast, just one of 24 control animals (4.2%) showedlimited evidence of biliary epithelium hyperplasia, andno other pathologies were noted.

Since low-grade mixed-cell neoplasms were themost advanced tumours identified in Ptenf/f ani-mals, we concentrated our analysis on this lesiontype. IHC against Cytokeratin 19 (CK19), a biliaryepithelium marker, was performed. WT bile ductand gall bladder epithelia stained strongly for CK19(Figure 2A). Lesions arising within the livers of agedPtenf/f animals were more varied in CK19 staininglevels and distribution. Hepatocyte-like componentswithin lesions were consistently negative for CK19staining. Cholangiocyte-like cells varied in staining

intensity for CK19, although in all cases stainingwas identified and was weaker than that of WTtissue. Anti-Ki67 staining of WT tissue indicated thatproliferating cells were rare within normal hepatocytesand cholangiocytes (Figure 2A). A significant increasein staining frequency was identified in Ptenf/f tissuescompared to controls (29.6 ± 13.4% versus 1.6 ± 1.4%positively-labelled epithelial cells). Positively-stainedcholangiocyte-like and hepatocyte-like nuclei wereidentified, indicating that proliferation of both celltypes contributed to the lesion, supporting our previoushistopathological observations. Co-staining was carriedout against CK19 and proliferating cell nuclear antigen(PCNA) to additionally confirm the proliferation ofcholangiocyte-like cells (see Supplementary material,Figure S1) and double-positive biliary epithelial cellswere readily identified.

To evaluate PTEN protein status and PI3′K–AKTpathway activity, we performed IHC against PTENand activated AKT (pAKTSer473) (Figure 2B). Hepa-tocytes and biliary epithelial cells in control sampleswere found to express PTEN. In contrast, bothhepatocyte-like and cholangiocyte-like componentsof lesions in aged Ptenf/f animals were negative foranti-PTEN staining. Cells comprising the stromalcomponent of the lesion retained PTEN expression, asexpected using this experimental approach.

Anti-pAKTSer473 staining revealed a reciprocalpattern of staining to that noted for anti-PTEN(Figure 2B). In controls, a very low level of activatedpAKTSer473 was detected. In Ptenf/f lesions, PTEN-deficient hepatocyte-like and cholangiocyte-like cellsshowed increased cytoplasmic staining for pAKTSer473

compared to surrounding normal cells. Stromal cellswithin lesions, previously found to retain PTENprotein, stained weakly for pAktSer473, indicatinglimited activation of AKT in these cells.

Combined KRAS activation and PTEN deletion, butnot activation of KRAS alone, causes biliary tractmalignancyGiven that mutational activation of KRAS is impli-cated in cholangiocarcinoma and gall bladder cancerdevelopment in humans [12–14], and that activatedKRAS is able to directly activate the PI3′K–Aktsignalling pathway [10], we next examined the effectsof combined activation of KRAS and deletion of PTENupon the biliary epithelium.

Experimental AhCreERT+;Ptenf/f;KrasLSL-V12/+ ani-mals (Ptenf/f;KrasV12/+) and AhCreERT+;KrasLSL-V12/+controls (KrasV12/+) were generated. Cre activity wasinduced in Ptenf/f;KrasV12/+ (n = 20) and KrasV12/+(n = 15) animals, which were aged until symptomaticof disease. Kaplan–Meier analysis (Figure 3A)revealed Ptenf/f;KrasV12/+ animals to have a significantreduction in survival compared to KrasV12/+ controls(medians of 43 and 449 days, respectively; log-ranktest, p < 0.0001). Comparison of the survival dataof these cohorts and the survival data of Ptenf/f and



WTA Ptenf/f

WTB Ptenf/f

Ant

i-CK

19A

nti-K

i67

Ant

i-Pte

nA

nti-p

hosp

ho-A

KT

(Ser

473)

Figure 2. PTEN-deficient lesions show CK19 and Ki67 positivityand activation of AKT. (A) IHC was used to assess expression ofCK19, Ki67, PTEN and pAKTSer473 within liver lesions of Ptenf/f

animals compared to WT controls. CK19 staining was found tobe less intense within cholangiocyte components of Ptenf/f lesionscompared to WT biliary epithelial cells (top row). A significantproportion of both cholangiocyte-like and hepatocyte-like cellscomprising lesions in Ptenf/f animals were positive for Ki67 antigen(bottom row), indicating the proliferative nature of these lesions.(B) Hepatocytes and cholangiocytes within lesions in Ptenf/f

animals were deficient for PTEN staining compared to controlbiliary epithelial cells, which express PTEN. Note that positive PTENstaining in this figure is of stromal cells (top row). IHC for phospho-AKTSer473 in adjacent sections revealed normal biliary epithelium toexpress low levels of activated AKT (bottom left panel). In contrast,lesions in Ptenf/f animals showed elevated expression of pAKT, witha direct correlation between PTEN deficiency and increased pAKTexpression (bottom right panel)

WT cohorts described above (Figure 3B) revealedPtenf/f and KrasV12/+ cohorts to have a similar survivalprobability (median survival of 407 and 420 days,respectively), both of which were reduced comparedto WT controls. Survival of Ptenf/f;KrasV12/+ animalswas the poorest of all cohorts.

Upon dissection, abnormalities weremacroscopically evident in Ptenf/f;KrasV12/+ animals.Some individuals developed a greasy coat condition,along with wart-like squamous papillomata of theskin. These lesions were not found in controls, but didnot significantly affect overall health and no individualwas euthanized due to this phenotype alone. In manycases the stomach was grossly enlarged, with marked

hyperplasia of the forestomach epithelium and floridluminal hyperkeratosis. This phenotype was notedto a less severe extent in a number of KrasV12/+animals. We believe that this phenotype contributedsignificantly to the morbidity of these animals, and isdescribed in Figure S2 (see Supplementary material).Finally, gall bladder enlargement was commonly notedin Ptenf/f;KrasV12/+ animals but not in controls.

Histopathology confirmed the presence of bil-iary tract abnormalities in Ptenf/f;KrasV12/+ animals(Figure 3D), not seen in KrasV12/+ controls (Figure 3C).In the gall bladder, multiple epithelial neoplasms werepresent within all animals analysed. Most frequently,these presented as extensive non-invasive papillaryneoplasms composed of multilayered, mitoticallyactive epithelial cells with loss of polarity, enlargedpleomorphic nuclei and a spectrum of epithelialdysplasia from low to high grade (Figure 3Di–iii).Invasive adenocarcinomas of the gall bladder that weremoderately differentiated with stromal desmoplasia,and which clearly invaded into surrounding tissues,were also identified in a number of individuals(Figure 3Div).

In addition to gall bladder lesions, widespreadmultifocal non-invasive papillary neoplasms werealso identified in the intrahepatic biliary system inPtenf/f;KrasV12/+ animals (Figure 3Dv, vi). Theseaffected bile ducts of all sizes, from major interlobularbile ducts to small bile duct radicles in portal tracts.Varying degrees of dysplasia, but mostly low-grade,were identified (Figure 3Dvii). No such lesions, withineither the gall bladder or bile ducts, were identifiedwithin KrasV12/+ control animals. Hepatocellular pro-liferative lesions were not found in Ptenf/f;KrasV12/+mice.

To quantify these observations, we scored theincidence of biliary tract lesions in Ptenf/f;KrasV12/+and KrasV12/+ mice. Where biliary abnormalities werenoted, we scored lesions according to three phenotypecategories: invasive carcinomas; non-invasive papillaryneoplasms with high-grade dysplasia; and non-invasivepapillary neoplasms with low-grade dysplasia.

No lesions were noted in any KrasV12/+ controls(n = 11). In contrast, every Ptenf/f;KrasV12/+ individ-ual analysed (n = 10) showed biliary tract neoplasia.All mice were found to harbour low-grade papillaryneoplasms; four individuals showed lesions with high-grade dysplasia, and three individuals showed clear evi-dence of invasive gall bladder carcinoma (Figure 3E).

IHC for CK19 revealed staining patterns inKrasV12/+ tissue sections that were identical to WTtissues (Figure 4A). Non-invasive papillary neo-plasms within the gall bladder and bile ducts ofPtenf/f;KrasV12/+ animals, and invasive adenocar-cinomas, showed strong CK19 immunopositivity,confirming their cholangiocyte nature. Anti-Ki67IHC revealed biliary tract lesions in Ptenf/f;KrasV12/+mice to be highly proliferative compared to thehistologically normal tissues of KrasV12/+ controls(Figure 4A) (positively stained nuclei: in intrahepatic


170 V Marsh et al

200.00100.000.00

1.0

0.8

0.6

0.4

0.2

0.0

300.00 400.00 500.00 600.00 0 200 400 600 800 1000

Time (days) Time (days)

A

D

(i)

(iv)

(vii) E

(v) (vi)

(ii) (iii)

Cum

ulat

ive

Sur

viva

l

1.0

0.8

0.6

0.4

0.2

0.0

B C

Cum

ulat

ive

Sur

viva

l

10 Ptenf/f-,KrasV12/+ mice analysed:

Low-gradelesions

Carcinoma

High-gradelesions

0

0

0

0

43

3

HL

Figure 3. Combined PTEN deletion and KRAS activation results in reduced survival, which is associated with widespread papillary neoplasiaof biliary epithelium. (A) Cre activity was induced in cohorts of Ptenf/f;KrasV12/+ mice (n = 20) and KrasV12/+ controls (n = 15), whichwere culled when symptomatic of disease. Kaplan–Meier survival analysis revealed a significant curtailment of lifespan in Ptenf/f;KrasV12/+

animals (black line) compared to KrasV12/+ animals (green line; tick marks indicate censored data, where cohort animals were euthanizedwithout reaching defined experimental endpoints). Median survival times were 43 and 449 days, respectively (log-rank test, p < 0.0001,χ2 = 29.72, DF = 1). (B) When overlaid with the survival curves of WT and Ptenf/f cohorts (also shown separately in Figure 1B), survivalof Ptenf/f;KrasV12/+ animals (black line) was significantly reduced compared to any other cohort (KrasV12/+, green line; Ptenf/f, blue line;WT, red line). (C) Gall bladder epithelium (top) and bile ducts (bottom) of KrasV12/+ animals was found to be grossly and histologicallynormal. (D) In contrast, Ptenf/f;kRasV12/+ mice bore marked abnormalities of the biliary tract epithelium. Grossly enlarged gall bladders inthese mice were found to be filled with papillary tumours, which carpeted the mucosa and extended into the cystic duct and intramuralsinuses (i). Neoplastic epithelia covering papillae (ii) were found to show a spectrum of dysplasia from low (L) to high (H) grade (iii).In some animals, invasive gall bladder carcinomas arising from intramucosal papillary high grade dysplasia were identified, which infiltratedthe adjacent liver (iv). Multifocal non-invasive papillary intrahepatic biliary neoplasms were also identified, affecting bile ducts of all sizes,including small intrahepatic ducts (v) and larger interlobular bile ducts (vi). Papillary lesions arising in intrahepatic bile ducts generallyshowed low-grade dysplasia (vii). (E) The incidence of pathologies within Ptenf/f;KrasV12/+ and KrasV12/+ animals was scored. Whilst noanimals within the KrasV12/+ control cohort (n = 11) had any type of lesion within the liver or biliary tract, every individual analysed withinthe Ptenf/f;KrasV12/+ cohort (n = 10) bore biliary tract neoplasms. The incidence of low-grade dysplastic lesions, high-grade dysplasticlesions and invasive carcinomas occurring in individual Ptenf/f;KrasV12/+ mice was scored and is presented as a Venn diagram. In manycases, mice showed evidence of more than one type of pathology.



bile ducts, 45.37 ± 14.3% versus 0.46 ± 0.80%; in gallbladder, 82.05 ± 9.44% versus 0.24 ± 0.32%).

PTEN IHC (Figure 4B) revealed WT PTENexpression in biliary epithelia of KrasV12/+ mice, asexpected. Detection of PTEN protein correlated with alack of staining for pAKTSer473. In contrast, all typesof biliary tract neoplasms within Ptenf/f;KrasV12/+ miceshowed reduced expression of PTEN. Loss of stainingfor PTEN correlated with an increase in cytoplasmicstaining for pAKTSer473. In the gall bladder epithe-lium, PTEN-deficient papillary neoplasms showed astrong increase in staining for pAKTSer473, coupledwith increased cell-surface localization. Invasive gallbladder carcinomas also showed strong membranousstaining for pAktSer473, although staining intensitywas not notably higher than that seen in non-invasivepapillary lesions.

IHC against activated (phospho-)ERK1/2 wasused to assess activation of the MAPK pathway(Figure 4C). In KrasV12/+ tissues, phospho-ERK wasdetected at very low levels within the cytoplasmof biliary epithelial cells. Increased staining forphospho-ERK was noted in all types of biliary lesionswithin Ptenf/f;KrasV12/+ animals. Papillary neoplasmsthroughout the biliary tree and invasive gall bladdercarcinomas showed both increased staining andnuclear localization of phospho-ERK, a feature notseen in any histologically normal KrasV12/+ sections.No significant difference in staining intensity wasnoted in invasive components of lesions compared tonon-invasive papillary counterparts.

Discussion

Human biliary tract malignancies are poorly character-ized in molecular terms, are difficult to treat and havea poor prognosis [2]. Good animal models of this dis-ease are therefore valuable to increase our knowledgeof basic disease biology and to facilitate the develop-ment of novel therapeutic agents.

Loss of PTEN is known to play a role in thedevelopment of cholangiocarcinoma in animal models[8], although is rarely noted in human disease [6].Oncogenic activation of KRAS contributes to bil-iary tract carcinogenesis in humans, with activatingmutations frequently detected [12–14]. Here, weinvestigated the effects of PTEN deletion and KRASactivation upon the murine biliary epithelium.

Upon deletion of PTEN alone, we found thatclearly identifiable lesions developed within the liverafter a long latency period. These lesions showeda spectrum of histopathologies. The most benignlesions identified were morphologically suggestiveof biliary microhamartomas or adenomas, and werecomposed of cholangiocytes. We also identifiedmore advanced lesion types, including intrahepatictumours showing a spectrum of hepatocytic and biliarydifferentiation without overt invasion of the adjacent

liver parenchyma. The presence of both cell types raisesthe hypothesis that they may originate from a hepaticprogenitor cell with the potential to differentiatetowards either a cholangiocyte or hepatocyte lineage.Hepatocytes and cholangiocytes within these lesionswere consistently negative for PTEN staining, support-ing the view that they arise from PTEN-deficient cells.Increased expression of activated AKT suggests thatthey are probably driven by activated PI3′K signalling.The lesions were found to be actively proliferatingand were partly composed of CK19-positive cholan-giocytes. CK19 staining was present within all lesions,although staining intensity was variable. Anecdotally,reduced CK19 expression has been shown to be afeature of human cholangiolocellular carcinomas,which show bidirectional differentiation and arepostulated to arise from liver progenitor cells [29,30].Interestingly, we also note reduced levels of CK19 inmixed hepatocyte/cholangiocyte lesions. However, inorder to identify the true cell of origin of the lesionswe describe, further studies must be conducted.

We next characterized the additional effects ofexpressing mutationally-activated KRAS (KrasV12)within this model. Double-mutant animals succumbedrapidly to ill-health, with a marked survival deficitcompared to WT and single-mutant animals. Everydouble-mutant animal examined in this study wasfound to bear some form of neoplasia of the biliarytract epithelium, involving either the gall bladder orbile ducts or commonly both. In these animals, diseasewas most frequently manifested as papillary low-gradedysplasia, but a significant number of individualshad high-grade dysplasia or invasive gall bladdercarcinoma. Interestingly, double-mutant animalsdeveloped biliary tract lesions with a completelydistinct histopathology from that observed in PTEN -deficient single-mutant animals. We believe that thisreflects the fact that, because disease progression isso rapid in double-mutant mice, they do not survivelong enough to develop the types of lesions seen inPtenf/f animals. Lesions arising in double mutantswere highly proliferative and showed strong activationof the PI3′K and MAPK pathways.

Our data therefore indicate that PTEN and KRASmutations cooperate within the mouse biliary epithe-lium, resulting in progressive hyperproliferation,dysplasia and, ultimately, invasive malignancy. Inhumans, loss of PTEN expression in biliary tract can-cers is uncommon, although we found in the mousethat deletion of PTEN alone was sufficient to causeslow development of low-grade neoplasms. In starkcontrast to this, whereas KRAS mutations are amongthe most common genetic abnormalities identifiedin human disease, in the mouse we found KRASactivation to be insufficient for biliary tumourigenesis.Finally, we showed that in the mouse, combinedKRAS and PTEN mutation has a more specific effectupon biliary epithelial cells, resulting in much moreovert and rapid transformation of these cells anddevelopment of carcinomas. Whilst this combination


172 V Marsh et al

KrasV12/+ Ptenf/f;KrasV12/+

CK

19K

i67

PT

EN

pER

KpA

KT

Ser

473

A

B

C

Figure 4. Biliary tract lesions within double-mutant animals show activation of the PI3′K–AKT and MAPK pathways. (A) Anti-CK19 stainingrevealed that non-invasive papillary neoplasms of intrahepatic bile ducts and gall bladder, and invasive gall bladder carcinomas (rightpanel, top row, left to right) in Ptenf/f;KrasV12/+ animals retained expression of CK19 at a similar level to that found in KrasV12/+ controlnormal gall bladder epithelium and bile duct epithelium (left panel, top row). All lesions in Ptenf/f;KrasV12/+ animals contained frequentKi67-positive nuclei (right panel, bottom row) compared to histologically normal tissue from KrasV12/+ controls, which showed rarepositively-stained cells (left panel, bottom row). (B) Anti-PTEN IHC revealed reduced staining of lesions within Ptenf/f;KrasV12/+ animals(right panel, top row) compared to KrasV12/+ controls, which retained PTEN staining (left panel, top row). Control tissues showed weakstaining for activated pAKTSer473 (left panel, bottom row), whereas Ptenf/f;KrasV12/+ lesions deficient for PTEN showed a marked increase inpAKT staining (right panel, bottom row). Papillary dysplasias and invasive carcinomas of the gall bladder additionally displayed cell surfacelocalization of pAKT staining (insets). (C) Anti-phospho-ERK IHC was used to assess MAPK signalling activity in Ptenf/f;KrasV12/+ lesions.Normal tissues from KrasV12/+ controls showed a low level of pERK staining in gall bladder and bile duct epithelia (left panel, left andright images, respectively). In contrast, increased staining intensity together with nuclear localization of staining for pERK was noted inall lesions of Ptenf/f;KrasV12/+ animals (right panel).

of mutations is extremely rare in human disease, themodel we present clearly recapitulates many of thekey histopathological features of human disease.

Mechanistically, it is well established that PTENantagonizes PI3′K pathway activity and preventsactivation of AKT . Consistent with this, we founddeletion of PTEN to cause modest activation ofAKT within slow-growing liver lesions. Mutationalactivation of KRAS is known to activate both theERK–MAPK cascade and the PI3′K pathway throughdirect interaction with p110 [10]. However, weshowed here that this alone was insufficient to drivecholangiocyte proliferation. This suggests that thepresence of functional PTEN maintains a ‘brake’ onthe PI3′K pathway and that ERK–MAPK pathwayactivation by mutant KRAS is insufficient to cause

neoplasia. The marked synergy between the twomutations therefore argues that, in this tissue, thePI3′K pathway is indeed activated by KRAS , but thatthis can only become manifest following additionalloss of the ‘brake’ function of PTEN .

In conclusion, we present evidence indicatingthat these two mutations may cooperate in biliarycarcinogenesis, and we also describe a novel model ofthis human disease.

Acknowledgments

This research was funded by a programme grantand studentship from Cancer Research UK and astudentship from the MRC. The authors wish to thank



Mark Bishop, Lucie Stocking, Derek Scarborough andMarc Isaacs for technical assistance.

Author contributions

ARC and VM designed the study; VM and EJDperformed experiments and collected and analyseddata; GTW provided histopathology expertise; andARC secured funding for these studies. All authorswere involved in writing the manuscript and reviewedand approved the final version.

References1. Anderson CD, Pinson CW, Berlin J, et al . Diagnosis and treatment

of cholangiocarcinoma. Oncologist 2004; 9: 43–57.2. Khan SA, Davidson BR, Goldin R, et al . Guidelines for the diag-

nosis and treatment of cholangiocarcinoma: consensus document.Gut 2002; 51(suppl 6): VI 1–9.

3. Li J, Yen C, Liaw D, et al . PTEN , a putative protein tyrosinephosphatase gene mutated in human brain, breast, and prostatecancer. Science 1997; 275: 1943–1947.

4. Steck PA, Pershouse MA, Jasser SA, et al . Identification ofa candidate tumour suppressor gene, MMAC1 , at chromosome10q23.3 that is mutated in multiple advanced cancers. Nat Genet

1997; 15: 356–362.5. Stambolic V, Suzuki A, de la Pompa JL, et al . Negative regulation

of PKB/Akt-dependent cell survival by the tumor suppressor PTEN.Cell 1998; 95: 29–39.

6. Pignochino Y, Sarotto I, Peraldo-Neia C, et al . TargetingEGFR/HER2 pathways enhances the antiproliferative effect ofgemcitabine in biliary tract and gallbladder carcinomas. BMC

Cancer 2010; 10: 631.7. Herberger B, Puhalla H, Lehnert M, et al . Activated mammalian

target of rapamycin is an adverse prognostic factor in patientswith biliary tract adenocarcinoma. Clin Cancer Res 2007; 13:4795–4799.

8. Xu X, Kobayashi S, Qiao W, et al . Induction of intrahepaticcholangiocellular carcinoma by liver-specific disruption of Smad4and Pten in mice. J Clin Invest 2006; 116: 1843–1852.

9. Malumbres M, Barbacid M. RAS oncogenes: the first 30 years. Nat

Rev Cancer 2003; 3: 459–465.10. Kodaki T, Woscholski R, Hallberg B, et al . The activation

of phosphatidylinositol 3-kinase by Ras . Curr Biol 1994; 4:798–806.

11. Bos JL. ras oncogenes in human cancer: a review. Cancer Res

1989; 49: 4682–4689.12. Abraham SC, Lee JH, Hruban RH, et al . Molecular and immunohis-

tochemical analysis of intraductal papillary neoplasms of the biliarytract. Hum Pathol 2003; 34: 902–910.

13. Nakanuma Y, Harada K, Ishikawa A, et al . Anatomic and molecularpathology of intrahepatic cholangiocarcinoma. J Hepatobil Pancreat

Surg 2003; 10: 265–281.

14. Kim YT, Kim J, Jang YH, et al . Genetic alterations in gallbladderadenoma, dysplasia and carcinoma. Cancer Lett 2001; 169: 59–68.

15. Isa T, Tomita S, Nakachi A, et al . Analysis of microsatelliteinstability, K-ras gene mutation and p53 protein overexpressionin intrahepatic cholangiocarcinoma. Hepatogastroenterology 2002;49: 604–608.

16. Tada M, Omata M, Ohto M. Analysis of ras gene muta-tions in human hepatic malignant tumors by polymerasechain reaction and direct sequencing. Cancer Res 1990; 50:1121–1124.

17. Tada M, Omata M, Ohto M. High incidence of ras gene mutationin intrahepatic cholangiocarcinoma. Cancer 1992; 69: 1115–1118.

18. Kiba T, Tsuda H, Pairojkul C, et al . Mutations of the p53tumor suppressor gene and the ras gene family in intrahepaticcholangiocellular carcinomas in Japan and Thailand. Mol Carcinog1993; 8: 312–318.

19. Mullany LK, Fan HY, Liu Z, et al . Molecular and functionalcharacteristics of ovarian surface epithelial cells transformed byKrasG12D and loss of Pten in a mouse model in vivo. Oncogene2011; 30: 3522–3536.

20. Richards JS, Fan HY, Liu Z, et al . Either Kras activation orPten loss similarly enhance the dominant-stable CTNNB1-inducedgenetic program to promote granulosa cell tumor development inthe ovary and testis. Oncogene 2012; 31: 1504–1520.

21. Dinulescu DM, Ince TA, Quade BJ, et al . Role of K-ras andPten in the development of mouse models of endometriosis andendometrioid ovarian cancer. Nat Med 2005; 11: 63–70.

22. Iwanaga K, Yang Y, Raso MG, et al . Pten inactivation acceleratesoncogenic K-ras-initiated tumorigenesis in a mouse model of lungcancer. Cancer Res 2008; 68: 1119–1127.

23. Suzuki A, de la Pompa JL, Stambolic V, et al . High cancersusceptibility and embryonic lethality associated with mutation ofthe PTEN tumor suppressor gene in mice. Curr Biol 1998; 8:1169–1178.

24. Guerra C, Mijimolle N, Dhawahir A, et al . Tumor induction by anendogenous K-ras oncogene is highly dependent on cellular context.Cancer Cell 2003; 4: 111–120.

25. Soriano P. Generalized lacZ expression with the ROSA26 Crereporter strain. Nat Genet 1999; 21: 70–71.

26. Kemp R, Ireland H, Clayton E, et al . Elimination of backgroundrecombination: somatic induction of Cre by combined transcrip-tional regulation and hormone binding affinity. Nucleic Acids Res2004; 32: e92.

27. Backman SA, Ghazarian D, So K, et al . Early onset of neoplasia inthe prostate and skin of mice with tissue-specific deletion of Pten .Proc Natl Acad Sci USA 2004; 101: 1725–1730.

28. Horie Y, Suzuki A, Kataoka E, et al . Hepatocyte-specific Ptendeficiency results in steatohepatitis and hepatocellular carcinomas.J Clin Invest 2004; 113: 1774–1783.

29. Komuta M, Spee B, Vander Borght S, et al . Clinicopathologicalstudy on cholangiolocellular carcinoma suggesting hepatic progen-itor cell origin. Hepatology 2008; 47: 1544–1556.

30. Komuta M, Govaere O, Vandecaveye V, et al . Histological diversityin cholangiocellular carcinoma reflects the different cholangiocytephenotypes. Hepatology 2012; 55: 1876–1888.

SUPPORTING INFORMATION ON THE INTERNETThe following supplementary material may be found in the online version of this article:

Supplementary methods

Figure S1. Co-stain of CK19 and PCNA confirms that cholangiocyte-like cells are proliferating within biliary tract lesions.

Figure S2. BNF/TAM-treated KrasV12/+ and Ptenf/f; KrasV12/+ animals, but not Ptenf/f or WT animals, are susceptible to squamous hyperplasiaof the forestomach driven by AhCreERT.


PTEN loss and KRAS activation cooperate in murine biliary tract malignancies

Documents

Transcript of PTEN loss and KRAS activation cooperate in murine biliary tract malignancies