p38delta/MAPK13 as a diagnostic marker for cholangiocarcinoma and its involvement in cell motility...

p38delta/MAPK13 as a diagnostic marker forcholangiocarcinoma and its involvement in cell motilityand invasion

Felicia Li-Sher Tan1,2,3, Aikseng Ooi3,4, Dachuan Huang3, Jing Chii Wong3, Chao-Nan Qian3,4, Cora Chao5, London Ooi2,

Yu-Meng Tan2, Alexander Chung1, Peng-Chung Cheow1, Zhongfa Zhang4, David Petillo4, Ximing J. Yang6 and Bin Tean Teh3,4

1 Department of General Surgery, Singapore General Hospital, Singapore2 Department of Surgical Oncology, National Cancer Centre, Singapore3 NCCS-VARI Translational Cancer Research Laboratory, National Cancer Centre, Singapore4 Laboratory of Cancer Genetics, Van Andel Research Institute, Grand Rapids, Michigan5 Department of Pathology, Singapore General Hospital, Singapore6 Department of Pathology, Northwestern University, Chicago, Illinois

Cholangiocarcinoma (CC) and hepatocellularcarcinoma (HCC) are two main forms of liver malignancies, which exhibit

differences in drug response and prognosis. Immunohistotochemical staining for cytokeratin markers has been used to some

success in the differential diagnosis of CC from HCC. However, there remains a need for additional markers for increased

sensitivity and specificity of diagnosis. In this study, we have identified a p38 MAP kinase, p38d (also known as MAPK13 or

SAPK4) as a protein that is upregulated in CC relative to HCC and to normal biliary tract tissues. We performed microarray

gene expression profiling on 17 cases of CC, 12 cases of adjacent normal liver tissue, and three case of normal bile duct

tissue. p38d was upregulated in 16 out of 17 cases of CC relative to normal tissue. We subsequently performed

immunohistochemical staining of p38d in 54 cases of CC and 54 cases of HCC. p38d staining distinguished CC from HCC with

a sensitivity of 92.6% and a specificity of 90.7%. To explore the possible functional significance of p38d expression in CC, we

examined the effects of overexpression and knockdown of p38d expression in human CC cell lines. Our results indicate that

p38d is important for motility and invasion of CC cells, suggesting that p38d may play an important role in CC metastasis. In

summary, p38d may serve as a novel diagnostic marker for CC and may also serve as a new target for molecular based

therapy of this disease.

Cholangiocarcinoma (CC) is an aggressive disease in need ofurgent investigative attention. It is the second most commontype of hepatobiliary cancer worldwide and it is endemic insoutheast Asia. Epidemiological studies have shown anincrease in the incidence of CC in the United States, UnitedKingdom, and Australia.1–4 Mortality rates have also risensharply in recent years.1

Cholangiocarcinomas are notoriously challenging to diag-nose and treat. In contrast to hepatocellular carcinoma (HCC),

which is responsive to targeted therapy and radiotherapy,5 CChas no proven adjuvant therapy. At present, only surgical resec-tion with tumor-free margins is associated with improvementin survival, with 5-year survival rates of 20–40%.6 However, dis-tant metastases, extensive regional lymph node metastasis andvascular encasement or invasion preclude resection. Further-more, neither chemotherapy nor radiotherapy has been eval-uated in randomized controlled trials and their efficacy for thisdisease remains to be explored.7 Although 5-fluorouracil andgemcitabine-based regimes have partial response rates of 20–30%, survival benefits have yet to be shown.8 Therefore, differ-entiating CC from HCC is essential in predicting clinical out-comes, patient counseling and deciding treatment modalities.

Carcinoembryonic antigen (CEA) and carbohydrate anti-gen 19-9 (CA19-9) are recognized as serum markers for CC,but they are also overexpressed in other malignant tumorsand are not specific to CC.7 Immunohistochemical stainingfor cytokeratin markers such as CK19 and CK7 has beenused to some success in distinguishing CC from HCC. How-ever, they are also not specific to CC, and are expressed inother cancers such as non-small-cell lung carcinoma. More-over, not all cases of CC express CK7 or CK19.9,10 Therefore,there remains a need for additional markers to increase sensi-tivity and specificity in the diagnosis of CC.

Key words: cholangiocarcinoma, p38delta, p38, MAPK, SAPK

Additional Supporting Information may be found in the online

version of this article

This paper is dedicated to the memory of Connie Low and Christian

Helmus

The first three authors contributed equally to this work

Grant sponsors: The Singapore Millennium Foundation, The

National Cancer Centre Research Foundation

DOI: 10.1002/ijc.24944

History: Received 26 May 2009; Accepted 24 Sep 2009; Online 8

Oct 2009

Correspondence to: Bin Tean Teh, Laboratory of Cancer Genetics,

Van Andel Research Institute, 333 Bostwick Ave NE, Grand Rapids,

MI 49503, USA, Tel: 616-234-5296, E-mail: [email protected]

Can

cerCellBiology

Int. J. Cancer: 126, 2353–2361 (2010) VC 2009 UICC

International Journal of Cancer

IJC

In this study, we have identified the p38d MAP kinase(also known as MAPK13 or SAPK4) as a novel marker todistinguish CC from HCC. We found that p38d is upregu-lated in CC compared to both HCC and normal liver/bileduct tissue. We further found that p38d is more sensitivethan the CK7 marker and of comparable sensitivity to theCK19 marker in identifying cases of CC. Functional charac-terization of p38d in human CC cell lines indicates that p38denhances cell migration and invasion, suggesting an impor-tant role in the metastasis of cholangiocarcinoma.

Material and MethodsFresh frozen and formalin fixed parafin embedded samples

This study has been approved by the Institutional ReviewBoard (IRB) of the National Cancer Centre Singapore(NCCS) and the Singapore General Hospital (SGH). Properconsents have also been obtained for all fresh-frozen samplesand formalin fixed parafin embedded (FFPE) tissue blocks.

Case selection was done based on a combination of serumtumor markers (CA19-9, CEA, a-fetoprotein and CA-125), an-atomical location, and immunohistological markers (CK7,CK19, CK20, hepatocyte antigen and TTF1). Fresh frozencholangiocarcinoma and cancer adjacent normal liver tissueswere obtained from surgical resection of primary tumors,while fresh frozen normal bile duct tissues were obtained fromliver transplant surgery. These tissues were collected retrospec-tively and stored in liquid nitrogen at the Tissue Repository ofthe NCCS. A total of 17 cases of histopathlogically confirmedcholangiocarcinoma, three samples of normal bile duct, and 12corresponding cancer adjacent normal liver tissues wereretrieved from the Tissue Repository of the NCCS. Since onlysurgical resection with tumor free margins is associated withimproved survival, patients involved in this study were nottreated with either chemo or radiotherapy prior to surgery.

Tissue blocks of 54 cases of histopathologically confirmedcholangiocarcinoma were also retrieved from the tissueblocks archive of the Department of Pathology, SingaporeGeneral Hospital. Representative sections were first takenfrom each tissue block; H&E staining was then performed onthe representative sections. The stained sections werereviewed by a board certified pathologist before the tissueblocks were released for this study. Histopathological infor-mation on cholangiocarcinoma cases used in this study aresummarized in Table S1. HCC tissue microarray (LV631)was purchased from Biomax Inc. USA. The tissue microarraycontains 54 cases of HCC in different grades, and 9 matchedand unmatched cancer adjacent normal liver tissues. Eachindividual case was represented on the tissue microarray as asingle core of 1.5mm diameter and 5-micron thickness. Thequality of the tissue microarray has been assessed throughstandard immunohistochemistry staining for b-actin.

RNA isolation

Total RNA were extracted from the fresh frozen tissues usingTrizol reagent (Invitrogen, CA) according to manufacturer’s rec-

ommendation. The isolated total RNA were then digested withRNase free DNase (RQ1 DNase, Promega) at 37�C for 30minutes and subsequently purified using RNeasy Mini kit (Qia-gen,CA). The quality and integrity of the purified RNA wereassessed using spectrophotometer and agarose gel electrophoresis.

Gene expression profiling

Purified RNA samples were used for oligonucleotide geneexpression profiling. An initial 4–10 lg of total RNA wasused to prepare antisense biotinylated RNA. A subset of sam-ples was spiked with external poly-A RNA positive controls(Affymetrix, CA). One-cycle synthesis of single-stranded anddouble-stranded complementary DNA was performed withthe use of T7-oligo (dT) primer (Affymetrix) and SuperScriptII reverse transcriptase (Invitrogen, CA). This cDNA wasthen used as a template for complementary RNA (cRNA)amplification and biotin labeling at 37�C for 15 hrs. Twentylg of purified, biotinylated cRNA were then fragmented, ofwhich 15 lg were hybridized to HGU133 Plus 2.0 GeneChipsat 45�C for 16 hours. Sample chip scanning and qualityassessment were performed with GeneChip Operating Soft-ware (GCOS) 1.4.0.036 (Affymetrix) using global scaling setto a target signal of 500, following the manufacturer’s recom-mended protocol (GeneChip Expression Analysis TechnicalManual, Affymetrix, Rev 5.0, November 2004). Gene expres-sion values were generated by using Microarray Suite v5.0software (Affymetrix). The probes were filtered according tothe methods of Dai et al.11 The hybridizations were normal-ized by using the robust multichip averaging (rma) algorithmcoded in the Bioconductor Affy package (see http://www.bio-conductor.org/) to obtain a summary expression level foreach retained gene. This resulted in more than 17,600 geneexpression levels per sample, each of which then had one nu-merical value to represent its relative expression intensity.

Statistical methods

Statistical and microarray analysis were carried out using theR statistical environment, version 2.7.0. Student t-test wasused when comparing mean values of two groups. Unlessspecifically stated in the results section, the significant levelwas set to be 0.05 throughout this paper.

Immunohistochemistry staining

Selected tissue blocks were sectioned into slices of 4-micronthickness and fixed onto charged slides for immunohisto-chemical staining. Immunohistochemical staining was per-formed using Dako REALEnvision Detection System (Dako,UK) using Novared (Vecta Lab, CA) and DAB (Vecta Lab,CA) as substrates. Briefly slides were deparafinized in twochanges of Formula 83 (CBG Biotech) for 5 minutes each,and then washed in 100% ethanol for 5 minutes, followed bystepwise re-hydration in 95%, and 80% ethanol (5 minutesfor each step) followed by 10 minutes in phosphate bufferedsaline (PBS). After the deparafinization and rehydration steps,antigen retrieval was carried out by incubating the slide in

Can

cerCellBiology

2354 p38d/MAPK13 as a diagnostic marker for cholangiocarcinoma


100�C citrate buffer pH 6.0 (Zymed laboratories Inc.) for 20minutes and allowing the slide to slowly cool down to roomtemperature. For CK7 and CK19 staining, antigen retrievalwas done by incubating slides in 0.1% Trypsin in PBS at37�C for 20 minutes. After antigen retrieval, slides werewashed twice in PBS and then immersed in PBS buffered 3%hydrogen peroxide for 30 minutes to block endogenous per-oxidase activity. The tissues were then washed twice in PBS,followed by blocking with 2.5% goat serum for 15 minutes atroom temperature. Subsequent steps of immunohistochemis-try staining were carried out according to manufacturer’s rec-ommendations. All primary antibodies were purchased fromAbcam. Primary CK7 antibody was used at 1:1000 dilution,CK19 antibody was used at 1:150 dilution, and p38d anti-body was used at 1:250 dilution.

Immunohistochemistry staining evaluation

Evaluation of immunohistochemically stained sections wasdone by a board certified pathologist.

Construction of p38d recombinant DNA construct

p38d from a human cDNA pool (Clontech, CA) was ampli-fied by PCR using forward primer (50-A GTTAAGCTTATGACCATGAGCCTCATCCGGAAA-30) and reverse primer(50-AGAATTCCACAGCTTCATGCCACTCCGGC-30). Theresulting p38d PCR product was inserted into pcDNA6/myc-His B vector (Invitrogen, CA) at the Hind III and Eco RIrestriction sites.

Cell culture

HuCCT1, EGI-1, HIGGK and TFK-1 CC cell lines weremaintained in RPMI1640 supplemented with 10% (v/v) fetalbovine serum (FBS; Invitrogen, CA). TGBCITKB CC cell linewas maintained in DMEM supplemented with 5% (v/v) FBS.PLC/RF5 and Chang liver HCC cell lines were maintained inRPMI1640 supplemented with 10% (v/v) FBS.

Transfection

Recombinant DNA constructs or siRNAs (Santa Cruz, CA)were transfected into various cholangiocarcinoma cell linesusing the Lipofectamine reagent (Invitrogen, CA) accordingto the manufacturer’s conditions.

Cell proliferation assay

Cells were seeded into 96-well plates (0.5 � 103 cells perwell) 24 hours prior to transfection with recombinant DNAconstructs or siRNAs. The growth of untransfected or 24, 48and 72 hours post-transfected cells were determined by thecolorimetric 3-(4,5-dimethylthiazol-2yl)-5-(3-carboxymethox-yphenyl)-(4-sulfophenyl)-2H-tetrazoluim assay according tothe manufacturer’s instruction (MTS; Promega, WI). Experi-ments were performed in triplicate and repeated at leasttwice.

Matrigel invasion assay and cell migration assay

2.0 � 105 cells in serum-free medium were seeded into theupper chamber of BioCoat inserts containing filters with 8lm pores for migration assay or onto Matrigel-coated filtersfor invasion assay (BD Pharmingen, CA). The lower chamberwas filled with 10% (v/v) serum-containing medium as at-tractant. Cells that did not migrate through the filters 22hours post-incubation were removed with cotton swabs. Cellsthat traversed or invaded through the filter/matrigel werefixed and stained by Diff-Quik Solution (Dade Behring, DE).After staining, cells in both control inserts and invasionchambers were counted under a microscope in 5 randomfields (magnification � 200). Each sample was assayed intriplicate and repeated at least twice. Student’s t-test was usedfor statistical analysis.

Western blot analysis

Cellular proteins were extracted with PBS containing 0.1%(v/v) Triton-100 in the presence of protease inhibitors. Pro-teins resolved by SDS-PAGE were electroblotted to a nitrocel-lulose membrane (Amersham, Buckinghamshire) and themembrane was incubated overnight at 4�C with blockingbuffer (PBS containing 5% (w/v) skim milk and 0.05% (v/v)Tween-20). Primary and secondary antibody incubationswere done in blocking buffer. Anti-p38d, anti-myc and anti-p38 antibodies were purchased from Santa Cruz Biotechnol-ogy (CA), whereas anti-b-actin antibody was purchased fromSigma (LA). The membrane was washed with PBS containing0.05% (v/v) Tween-20 followed by analysis using the Super-signal Chemiluminescent kit (Pierce, IL) according to themanufacturer’s recommendations.

Immunoprecipitation

Polyclonal anti-p38d, antibody and goat IgG (2lg) bound toprotein G Sepharose beads (Santa Cruz Biotechnology, CA)were each incubated overnight with 100lg of cell extracts inIP buffer (150mM NaCl, 20mM Tris-HCl, pH7.4, 2mMEDTA, 2mM DTT, 1% Triton X-100, plus protease inhibitors)at 4�C. Beads were then washed twice in buffer A (identical toIP buffer except that it contains 0.5% Triton X-100) and twicein buffer B (identical to IP buffer except that it contains 0.1%Triton X-100) before being re-suspended in SDS samplebuffer. Immunoprecipitated p38 proteins were separated onSDS-PAGE and analyzed by Western blot using monoclonalanti-p38d antibodies and anti phosphoryl-p38 antibody.

Resultsp38d mRNA levels in cholangiocarcinoma, bile duct and

cancer adjacent normal liver tissues

We used microarray gene expression profiling to search forgenes that were specifically upregulated in CC as comparedto cancer adjacent normal liver tissues, using LIMMA algo-rithm, with t-statistic p-value cutoff set at 0.01. The searchidentified a total of 2,431 genes; within this list of

Can

cerCellBiology

Tan et al. 2355


upregulated genes, we chose to focus on p38d due to theavailability of a good commercial antibody as well as the pre-viously documented importance of the p38 class of kinases inmalignant cholangiocarcinoma cell lines.12–14 The microarraygene expression profile showed that p38d was significantlyupregulated in 16 out of 17 cases of CC as compared to the12 cases of cancer adjacent normal liver tissues used in thisstudy (Welch Two Sample t-test, p-value ¼ 1.553e-06) (Fig.1). p38d transcript level was relatively higher in normal bileduct tissue as compared to cancer adjacent normal liver tis-sues, but lower than that in CC tissues (Welch Two Samplet-test, p-value ¼ 0.07).

Immunohistochemical staining of p38d in liver

malignancies

To determine whether upregulation of p38d was specific toCC, or generally present in both CC and HCC, we performedimmunohistochemical staining for p38d in formalin-fixed-paraffin-embedded CC and HCC tissues. Immunohistochemi-cal staining was done on 54 cases of HCC on a commercialtissue microarray and 54 cases of CC tissues obtained fromtissue block archives of Singapore General Hospital. Fifty outof 54 cases (92.6%) of CC samples showed positive cytoplas-mic staining for p38d (Fig. 2). In contrast, only five of 54cases of HCC (9.3%) exhibited positive staining for p38d.Furthermore, of the five HCC cases that stained positive forp38d, expression was confined to less than 10% of the cancercells. Thus, p38d expression is selectively expressed in casesof CC compared to HCC. The difference in the p38d protein

levels between these two malignancies was statistically signifi-cant (Pearson’s Chi square p-value ¼ 2.129e-09). p38d im-munostaining was also seen in some cases of cancer adjacentnormal liver tissue and bile duct, although at lower levelsthan in CC. The overall results of p38d staining on CC,HCC, and cancer adjacent normal liver tissue and bile ductare summarized in Table 1.

Comparison of p38d staining with CK7 and CK19 staining

in liver malignancies

Cytokeratin 7 (CK7) and cytokeratin 19 (CK19) are bio-markers currently used for the detection of cholangiocarci-noma. To compare the effectiveness of p38d and these knownmarkers in differentiating CC from HCC, CK7 and CK19staining was performed on 46 out of the 54 cases of CC whereextra tissue slides were available. A summary of the CK7,CK19 and p38d staining patterns in this subset of CC cases isgiven in Table 2. Twenty seven cases of CC stained positivefor both CK7 and CK19, while 17 cases stained positive foronly one of these cytokeratin markers. In all, 32 out of the 46cases stained positive for CK7 (sensitivity ¼ 69.6%), while 39out of the 46 cases stained positive for CK19 (sensitivity ¼84.8%). For this same subset, 89.1% of cases stained positivefor p38d, indicating that p38d is a sensitive marker for cholan-giocarcinoma (Table 2). The sensitivity of p38d in identifyingcholangiocarcinoma tissues determined with the cases used inthis study is significantly higher than that of CK7 (Chi-squarefor count data, p-value ¼ 0.0394), while it is comparable toCK19 (Chi-square for count data, p-value ¼ 0.7569). Next, wecompared the specificities of these markers in distinguishingCC from HCC. As stated earlier, five of the 54 cases of HCCstained positive for p38d (Table 1); thus, the specificity ofp38d staining in our study was 90.7%. The same set of HCCcases was also stained for CK7 and CK19. No cases of HCCshowed positive staining for either CK7 or CK19. Thus, p38dis less specific than either CK7 or CK19 in distinguishing CCfrom HCC (Table 2). The overall sensitivity and specificityscores of p38d, CK7 and CK19 are summarized in Table 3.

p38d protein levels in various cholangiocarcinoma

cell lines

We examined p38d protein levels in a panel of liver cancercell lines: five CC cell lines, TGBC1TKB, TFK-1, HuCCT1,EGI-1 and HIGGK; and two HCC cell lines, PLC/RF5 andChang liver. Our results showed that the CC cell linesTGBC1TKB, TFK-1, HuCCT1, EGI-1 and HIGGK expressedhigher levels of p38d protein than the HCC cell lines PLC/RF5 and Chang liver. There were no observable differencesin total p38 MAP kinase protein levels. Interestingly, the in-trahepatic CC cell lines HuCCT1, EGI-I and HIGGKexpressed higher levels of p38d protein compared to the extra-hepatic CC cell lines TGBC1TKB and TFK-1 (Fig. 3a). Tofurther examine the status of p38d activity, cell extractswere subjected to immunoprecipitation with anti-p38d anti-body. Activated p38d was then detected by a monoclonal

Figure 1. p38d mRNA levels in cancer adjacent normal liver,

normal bile duct and cholangiocarcinoma tissues. p38d is

upregulated in 16 out of 17 cases of cholangiocarcinoma, while its

level remains low in all 12 cases of cancer adjacent normal liver

tissues. The p38d mRNA level in normal bile duct is relatively

higher than in cancer adjacent normal liver tissues and relatively

lower than in cholangiocarcinoma tissues.

Can

cerCellBiology



phosphorylated p38a specific antibody that also recognizesphosphorylated p38d. Phosphorylated p38d was detected inall CC cell lines (Fig. 3b).

p38d regulates cholangiocarcinoma cell motility and

invasive migration

In view of our observation that p38d was upregulated in bothprimary CC tumors and in CC cell lines, we proceeded to

investigate the possible roles of p38d in cholangiocarcinoma.The family of p38 MAP kinases has been previously impli-cated in the transformation and growth of cholangiocarcino-mas.12,14–16 However, the specific roles of the individual p38MAP kinase isoforms in this process are not understood. Toascertain if the p38d isoform plays a role in cell motility inCC cell lines, we examined the effects of p38d siRNA-medi-ated knockdown on cell migration. TGBC1TKB and EGI-1

Figure 2. H&E staining (a), and immunohistochemical staining for CK7 (b), CK19 (c), and p38d (d). Immunohistochemical staining is shown

for two representatives from 17 cases of cholangiocarcinoma that showed positive staining only for either CK7 or CK19. The two different

cases were denoted by numbers (1 and 2) in front of the letters (a, b, c, and d) that denote the type of immunostaining. p38d antibody

showed intense cytoplasm staining in both of these cases. Scale bar ¼ 100lm

Can

cerCellBiology

Tan et al. 2357


cells were transfected with siRNA directed against p38dmRNA, or with a scrambled siRNA control (Fig. 4). Theresults showed that knockdown of p38d expression signifi-cantly decreased the number of cells that traversed through afilter in a standard migration assay (Figs. 4c and 4d). Addi-tionally, wound healing assays also demonstrated that knock-down of p38d led to a decrease in cell motility (Figs. S1a andS1b). The knockdown of p38d expression did not affect cellproliferation (Figs. 4e and 4f, Figs. S2a and 2b) indicating thatthe decreased number of traversed cells was due to inhibition

Table 1. Summary of p38d immunohistochemical staining results

Total positivecases/totalcases

Percentageof positivecells

StainingIntensity

Normal BileDuct

41/54 51–80% All intensities

32/54 1þ8/54 2þ1/54 3þ

*5 cases were unableto score due toinsufficient normalbile duct tissues

Cancer adjacentnormal tissues

17/54 10–50% All intensities

16/54 1þ0/54 2þ1/54 3þ

*4 cases were unableto score due toinsufficient normalcancer adjacentnormal liver tissues

CC 50/54 All intensities

16/54 1þ22/54 2þ12/54 3þ2/54 <10%

20/54 10–50%

21/54 51–80%

7/54 >80%

HCC 5/54 <10% 2þ

Staining intensity was scored according to a scale of 1 to 3. p38d wasstained positive in 50 out of 54 cases of CC tissues, but it was onlypositive in 5 out of 54 cases of HCC. Percentage of positively stainedcells was also lower in HCC as compared to CC, with less than 10% ofcancer cells showing p38d expression in positively stained HCCsamples. Besides cancer cells, cancer adjacent normal bile duct andliver tissues also showed a low level of p38d expression, with positivestaining scores around 1þ. The relative percentage of positivelystained cells was also lower in these normal tissues. Averagepercentage of positively stained cells in positively staining normal liverand normal bile duct tissues were around 10–50% and 51–80%,respectively.

Table 2. Comparison of CK7, CK19 and p38d staining on a subset ofcholangiocarcina (CC) tissues (46 cases)

Total#sampleswithindicatedpattern ofCK staining

# sampleswith indicatedpattern of CKstaining andp38d positivestaining

# Sampleswith indicatedpattern ofCK staining andp38d negativestaining

CK7�/CK19� 2 0 2

CK7þ only 5 5 0

CK19þ only 12 12 0

CK7þ/CK19þ 27 24 3

Total number of samples: 46

Table 3. Summary of sensitivity and specificity scores of p38d, CK7and CK19 in distinguishing CC from HCC (46 cases of CC and 54cases of HCC)

Markers Specificity Sensitivity

p38d 90.7% 89.1%

CK19 100% 84.8%

CK7 100% 69.6%

Figure 3. p38d is expressed in various cholangiocarcinoma cell

lines. (a) Cells were lysed and the immunoblots were probed with

antibody specific for p38d (upper panel) or a pan-p38 antibody to

detect all isoforms (middle panel). Blotting for b-actin was

performed as a loading control (lower panel). (b) The cell extracts

were immunopercipitated with polyclonal p38d antibody (p38) and

goat IgG (Iso) respectively and sequentially probed with

monoclonal phosphoryl-p38 (upper panel) and p38d antibody

(lower panel).

Can

cerCellBiology



of cell motility, not change in total cell number. The effect ofp38d on cell motility was further verified by the Matrigel inva-sion assay. As shown in Figure 5, the invasiveness ofTGBC1TKB and EGI-1 was significantly inhibited by knock-down of p38d (Figs. 5a and 5b). On the other hand, overex-pression of p38d in these cells led to enhanced invasiveness(Figs. 5c and 5d). These results suggest that p38d may play animportant role in the metastasis of cholangiocarcinoma.

Discussionp38d is a member of the p38 subfamily of MAP kinases. Thesekinases are activated by a range of cellular stresses and inflam-matory signals. Of note, chronic inflammation of the biliarytract predisposes to cholangiocarcinoma and is the leadingrisk factor for cholangiocarcinoma.17 An inflammatory cyto-kine associated with biliary tract inflammation, interleukin-6,has been shown to activate p38 MAPK in malignant cholan-giocytes.13 Activated p38 MAPK has been detected in humanCC tumor samples.18 A number of reports now support the

involvement of p38 MAPK in the development and growth ofCC. In vitro studies show that p38 MAPK is able to mediatecell proliferation, anchorage-independent growth, survival,migration, and invasive migration of CC cells.12,14–16 Xeno-graft studies in nude mice also indicate that p38 MAPK activ-ity is important for CC growth in vivo.14 However, the specificroles of individual p38 isoforms in CC are not understood.There are four isoforms of p38: p38a, p38b, p38d andp38c.19,20 p38a and p38b have been extensively studied, andwere specifically implicated in an interleukin-6 mediated cellsurvival pathway in cholangiocarcinoma cells.16 However, themajority of studies have not examined the specific roles ofp38d and p38c isoforms in CC growth. To date, the roles ofp38d and p38c in CC have remained unclear.

In our study, we have identified p38d as a protein that isupregulated in CC relative to normal biliary tract tissues. Inadditional to gene expression profiling, we also used immuno-histochemistry to evaluate the expression of p38d in CC. Weshow that p38d is specifically expressed in CC, but not HCC,

Figure 4. Knockdown of p38d inhibits cell migration. (a, b) Knockdown of p38d in TGBC1TKB and EGI-1 cells by siRNA. Western blot

analysis was performed to determine the expression level of p38d (upper panel) and total p38 (middle panel). b-actin was used as loading

control (lower panel). The same set of cells was subjected to migration assay and cell proliferation assay. (c, d) Migration assay. The

number of traversed cells was counted (magnification �200) in 5 fields. The results represent means 6S.D. of triplicates. Experimental

samples were statistically different from controls, p < 0.01. (e, f) The number of cells was determined by absorbance of O.D. 490nm. The

results represent means 6S.D. of triplicates. Knockdown samples showed no statistical difference from controls, p > 0.05.

Can

cerCellBiology

Tan et al. 2359


raising the possibility that p38d may serve as a novel diagnos-tic marker to distinguish CC from HCC. Comparative immu-nohistochemical staining on human CC cases showed thatp38d staining is more sensitive than CK7 and of comparablesensitivity to CK19 in the detection of CC. Incidentally, ourresults with CK7 and CK19 staining are consistent with previ-ous studies in which �60% of CC cases tested positive for CK7expression and �80% tested positive for CK19.10 However, wefound that p38d staining showed less specificity than CK7 orCK19 in distinguishing CC from HCC. These results suggestthat p38d staining may be used as a complement to the exist-ing panel of CC markers, but should not be used alone as asingle marker for the differential diagnosis of CC from HCC.

Functional characterization of p38d in human CC cell linesindicates that this protein is important for the motility andinvasive migration of CC cells. The p38 MAPK family hasbeen previously implicated in cell migration and invasion.Tanaka, et al. reported that the growth factor, WISP1v, indu-ces migratory invasive behavior in CC cells through a p38-de-

pendent mechanism.15 Similarly, Kim, et al. have implicatedp38 in Ras-induced migration and invasive behavior of humanbreast epithelial cells.21 However, these studies did not specifi-cally address the roles of individual p38 isoforms in cellmigration and invasion. Both groups used a p38a/b-specificinhibitor, SB203580,19,22,23 in their studies, implicating eitherp38a and/or p38b in cell migration and invasion. The poten-tial roles of other p38 isoforms in cell migration and invasionwere not examined and have remained unclear.15,21

Junttila et al. showed that the p38d isoform is involved ininvasive migration of head and neck squamous carcinomacells.24 Recently, by using gene knockout mice, it was demon-strated that p38d is essential for cell proliferation and tumordevelopment in the epidermis.25 We report here for the firsttime that the p38d isoform is also involved in cell migrationand invasion of cholangiocarcinoma cells. To our knowledge,this is the first report of a specific role for p38d incholangiocarcinoma. In addition to invasive behavior, thep38 MAP kinases have been implicated in other aspects of

Figure 5. Pro-invasion role of p38d in cholangiocarcinoma. (a, b) TGBC1TKB and EGI-1 cells were transfected with either scrambled siRNA

control or p38d siRNA and subjected to Matrigel invasion assay. The results represent means 6S.D. of triplicates. Knockdown samples are

statistically different from controls, p < 0.01. The inner panels indicate the knockdown efficiency of p38d. (c, d) TGBC1TKB and EGI-1 cells

were transfected with either empty vector or p38d-myc and subjected to Matrigel invasion assay. The results represent means 6S.D. of

triplicates. Experimental samples are statistically different from controls, p < 0.01. Inner panels indicate the overexpression of p38d.

Can

cerCellBiology



cholangiocarcinoma development and growth, including pro-liferation and survival. We did not find a role for p38d inCC cell proliferation in our study. The p38d isoform may notbe involved in CC proliferation, or may only be involvedunder specific conditions. A role for p38d in other aspects ofCC development and growth are yet to be verified. The abil-ity of p38d to mediate migration and invasion of cholangio-carcinoma cells suggests that p38d may be involved in chol-angiocarcinoma metastasis. More studies will be needed tounderstand the functional significance of p38d upregulationin this cancer. Currently, very little is known about the tran-scriptional and translational mechanisms controlling p38dexpression, and additional studies will also be required to

understand the mechanisms underlying selective upregulationof p38d expression in CC. In summary, our work indicatesthat p38d may serve as a novel marker for the differential di-agnosis of CC from HCC, useful as a complement to theexisting panel of CC markers. It is also possible that p38dplays a functional role in CC development, and may be anew target for molecular based therapy of this disease.

AcknowledgementsWe thank Vanessa Fogg for technical editing and Sabrina Noyes for assis-tance in preparation and submission of this manuscript.

References

1. Khan SA, Taylor-Robinson SD, ToledanoMB, Beck A, Elliott P, Thomas HC.Changing international trends in mortalityrates for liver, biliary and pancreatictumours. J Hepatol 2002;37:806–13.

2. Patel T. Increasing incidence and mortalityof primary intrahepaticcholangiocarcinoma in the United States.Hepatology 2001;33:1353–7.

3. Taylor-Robinson SD, Toledano MB,Arora S, Keegan TJ, Hargreaves S, BeckA, Khan SA, Elliott P, Thomas HC.Increase in mortality rates fromintrahepatic cholangiocarcinoma inEngland and Wales 1968–1998. Gut 2001;48:816–20.

4. West J, Wood H, Logan RF, Quinn M,Aithal GP. Trends in the incidence ofprimary liver and biliary tract cancers inEngland and Wales 1971–2001. Br JCancer 2006;94:1751–8.

5. Lau WY, Lai EC. Hepatocellularcarcinoma: current management and recentadvances. Hepatobiliary Pancreat Dis Int2008;7:237–57.

6. Jarnagin WR, Fong Y, DeMatteo RP,Gonen M, Burke EC, Bodniewicz BJ,Youssef BM, Klimstra D, Blumgart LH.Staging, resectability, and outcome in 225patients with hilar cholangiocarcinoma.Ann Surg 2001;234:507–17;discussion 17–9.

7. Khan SA, Davidson BR, Goldin R,Pereira SP, Rosenberg WM, Taylor-Robinson SD, Thillainayagam AV,Thomas HC, Thursz MR, Wasan H.Guidelines for the diagnosis and treatmentof cholangiocarcinoma: consensusdocument. Gut 2002;51 Suppl 6:VI1–9.

8. Gores GJ. Cholangiocarcinoma: currentconcepts and insights. Hepatology 2003;37:961–9.

9. Johnson DE, Herndier BG, Medeiros LJ,Warnke RA, Rouse RV. The diagnosticutility of the keratin profiles ofhepatocellular carcinoma and

cholangiocarcinoma. Am J Surg Pathol1988;12:187–97.

10. Stroescu C, Herlea V, Dragnea A,Popescu I. The diagnostic value ofcytokeratins and carcinoembryonic antigenimmunostaining in differentiatinghepatocellular carcinomas fromintrahepatic cholangiocarcinomas.J Gastrointestin Liver Dis 2006;15:9–14.

11. Dai M, Wang P, Boyd AD, Kostov G,Athey B, Jones EG, Bunney WE, MyersRM, Speed TP, Akil H, Watson SJ,Meng F. Evolving gene/transcriptdefinitions significantly alter theinterpretation of GeneChip data. Nucleicacids research 2005;33:e175.

12. Tadlock L, Patel T. Involvement of p38mitogen-activated protein kinase signalingin transformed growth of acholangiocarcinoma cell line. Hepatology2001;33:43–51.

13. Park J, Tadlock L, Gores GJ, Patel T.Inhibition of interleukin 6-mediatedmitogen-activated protein kinase activationattenuates growth of a cholangiocarcinomacell line. Hepatology 1999;30:1128–33.

14. Yamagiwa Y, Marienfeld C, Tadlock L,Patel T. Translational regulation by p38mitogen-activated protein kinase signalingduring human cholangiocarcinoma growth.Hepatology 2003;38:158–66.

15. Tanaka S, Sugimachi K, Kameyama T,Maehara S, Shirabe K, Shimada M,Wands JR, Maehara Y. Human WISP1v, amember of the CCN family, is associatedwith invasive cholangiocarcinoma.Hepatology 2003;37:1122–9.

16. Meng F, Yamagiwa Y, Ueno Y, Patel T.Over-expression of interleukin-6 enhancescell survival and transformed cell growthin human malignant cholangiocytes. JHepatol 2006;44:1055–65.

17. Ben-Menachem T. Risk factors forcholangiocarcinoma. Eur J GastroenterolHepatol 2007;19:615–7.

18. Agrawal S, Kuvshinoff BW, Khoury T,Yu J, Javle MM, LeVea C, Groth J,

Coignet LJ, Gibbs JF. CD24 expression isan independent prognostic marker incholangiocarcinoma. J Gastrointest Surg2007;11:445–51.

19. Cuenda A, Rousseau S. p38 MAP-kinasespathway regulation, function and role inhuman diseases. Biochim Biophys Acta2007;1773:1358–75.

20. Kumar S, Boehm J, Lee JC. p38 MAPkinases: key signalling molecules astherapeutic targets for inflammatorydiseases. Nat Rev Drug Discov 2003;2:717–26.

21. Kim MS, Lee EJ, Kim HR, Moon A. p38kinase is a key signaling molecule for H-Ras-induced cell motility and invasivephenotype in human breast epithelial cells.Cancer Res 2003;63:5454–61.

22. Gum RJ, McLaughlin MM, Kumar S,Wang Z, Bower MJ, Lee JC, Adams JL,Livi GP, Goldsmith EJ, Young PR.Acquisition of sensitivity of stress-activatedprotein kinases to the p38 inhibitor, SB203580, by alteration of one or moreamino acids within the ATP bindingpocket. J Biol Chem 1998;273:15605–10.

23. Eyers PA, Craxton M, Morrice N,Cohen P, Goedert M. Conversion of SB203580-insensitive MAP kinase familymembers to drug-sensitive forms by asingle amino-acid substitution. Chem Biol1998;5:321–8.

24. Junttila MR, Ala-Aho R, Jokilehto T,Peltonen J, Kallajoki M, Grenman R,Jaakkola P, Westermarck J, Kahari VM.p38alpha and p38delta mitogen-activatedprotein kinase isoforms regulate invasionand growth of head and neck squamouscarcinoma cells. Oncogene 2007;26:5267–79.

25. Schindler EM, Hindes A, Gribben EL,Burns CJ, Yin Y, Lin MH, Owen RJ,Longmore GD, Kissling GE, Arthur JS,Efimova T. p38delta Mitogen-activatedprotein kinase is essential for skin tumordevelopment in mice. Cancer Res 2009;69:4648–55.

Can

cerCellBiology

Tan et al. 2361


p38delta/MAPK13 as a diagnostic marker for cholangiocarcinoma and its involvement in cell motility...

Documents

Transcript of p38delta/MAPK13 as a diagnostic marker for cholangiocarcinoma and its involvement in cell motility...