EpCAM regulates cell cycle progression via control of cyclin ... - Nature

Oncogene (2013) 32, 641 -- 650 & 2013 Macmillan Publishers Limited All rights reserved 0950-9232/13 www.nature.com/onc

ORIGINAL ARTICLE

EpCAM regulates cell cycle progression via control of cyclin D1 expression A Chaves-Pe´rez1, B Mack2, D Maetzel3, H Kremling1, C Eggert1, U Harreús2 and O Gires1,2 The epithelial cell adhesion molecule (EpCAM) is an integral transmembrane protein that is frequently overexpressed in embryonic stem cells, tissue progenitors, carcinomas and cancer-initiating cells. In cancer cells, expression of EpCAM is associated with enhanced proliferation and upregulation of target genes including c-myc. However, the exact molecular mechanisms underlying the observed EpCAM-dependent cell proliferation remained unexplored. Here, we show that EpCAM directly affects cell cycle progression via its capacity to regulate the expression of cyclin D1 at the transcriptional level and depending on the direct interaction partner FHL2 (four-and-a-half LIM domains protein 2). As a result, downstream events such as phosphorylation of the retinoblastoma protein (Rb) and expression of cyclins E and A are similarly affected. In vivo, EpCAM expression strength and pattern are both positively correlated with the proliferation marker Ki67, high expression and nuclear localisation of cyclin D1, and Rb phosphorylation. Thus, EpCAM enhances cell cycle progression via the classical cyclin-regulated pathway. Oncogene (2013) 32, 641 -- 650; doi:10.1038/onc.2012.75; published online 5 March 2012 Keywords: EpCAM; EpICD; cyclin D1

INTRODUCTION The epithelial cell adhesion molecule (EpCAM) is a transmembrane protein with dual function as it is implicated in cell adhesion and mitogenic signalling.1 EpCAM is expressed in the basal membrane layers of simple squamous epithelia,2 and is overexpressed in various carcinomas3 and in tumour-initiating cells (also termed cancer stem cells).4,5 Recent publications uncovered the expression and tight correlation of EpCAM with the pluripotent and proliferative phenotype of human and murine embryonic stem cells.6 - 8 Elucidation of the mechanisms and compounds involved in EpCAM-mediated signalling provided a solid rationale for the observed expression patterns and potential functions of EpCAM in health and disease.1 One route of activation of EpCAM signalling relies on homophilic aggregation of EpCAM on opposing cells and results in a regulated intramembrane proteolysis.9 Initially, the extracellular domain termed EpEX is shed as an ectodomain, which can bind to intact EpCAM molecules itself to foster regulated intramembrane proteolysis. Ectodomain shedding is a prerequisite for the second step of activation that consists of an intramembrane cleavage of the remaining C-terminal fragment by g-secretase complexes containing presenilin-2 as a catalytic subunit.10 Upon intramembrane cleavage, the intracellular domain termed EpICD is released as a remarkably small 5-kDa protein, which is complexed with the adaptor protein FHL2 (four-and-ahalf LIM domains protein 2) and b-catenin in human carcinoma cells.10 EpICD-containing multiprotein complexes translocate into the nucleus and, upon interaction with the transcription factor Lef-1, are recruited to DNA to regulate gene transcription and subsequently cell proliferation. In embryonic stem cells, EpICD is directed to promoters of pluripotency genes such as oct4, nanog, c-myc, sox2 and klf4, and may thereby maintain the proliferative

and undifferentiated state of embryonic stem cells.8 Apparently, this feature of EpCAM is also important during reprogramming of differentiated cells into induced pluripotent stem cells (iPS). EpCAM and E-cadherin are markers of fully reprogrammed iPS,11 and EpCAM/EpICD foster the generation of iPS when coexpressed with classical iPS genes.12 In human cells, treatment with EpCAMspecific antibody G8.8 primarily altered the expression of genes involved in cell proliferation and cell cycle regulation.13 Although c-Myc, as a target of EpCAM, can explain some of the cellular effects of EpCAM, the actual mechanism behind EpCAM-mediated proliferation remained unexplored. It is still unclear whether EpCAM can directly activate components of the cell cycle machinery or if EpCAM-mediated proliferation is a secondary effect, for example, through repression of apoptosis, elevation of cell metabolism or interruption of antiproliferative signals. We show here that EpCAM controls cell cycle progression via the regulation of the key player cyclin D1 at the transcriptional level. As a result, downstream effects such as phosphorylation of retinoblastoma protein (Rb) and cell cycle progression are inhibited upon EpCAM downregulation and cells accumulated in G1 phase. Hence, EpCAM directly affects cell cycle regulation via the classical cyclin-regulated pathway. RESULTS Repression of EpCAM expression correlates with loss of cell proliferation and delayed cell cycle progression The de novo expression of EpCAM results in improved cell proliferation in HEK293 cells, especially under conditions of cellular stress such as growth factor withdrawal.14 Based on these findings, we tested the effects of EpCAM or FHL2 downregulation

1 Clinical Cooperation Group Molecular Oncology, Helmholtz-Zentrum Mu¨nchen, German Research Center for Environmental Health, and Head and Neck Research Department, Ludwig-Maximilians-University of Munich, Munich, Germany; 2Department of Otorhinolaryngology, Head and Neck Surgery, Grosshadern Medical Center, Ludwig-MaximiliansUniversity of Munich, Munich, Germany and 3The Whitehead Institute for Biomedical Research, Cambridge, MA, USA. Correspondence: Professor O Gires, Department of Otorhinolaryngology, Grosshadern Medical Center, Ludwig-Maximilians-University of Munich, Marchioninistr. 15, Munich 81377, Germany. E-mail: [email protected] Received 11 October 2011; revised 17 January 2012; accepted 3 February 2012; published online 5 March 2012

EpCAM and cyclin D1 A Chaves-Pe´rez et al

on FaDu hypopharynx carcinoma cells. FHL2 is a direct interacting partner of EpICD, which is required for growth-promoting properties of EpCAM in HEK293 cells.10 FaDu cells were transiently transfected with small interfering RNA (siRNA) oligonucleotides specific for EpCAM or FHL2. As a control, cells were transfected with a commercially available siRNA, which has no known targets in the human genome (control siRNA). Reduction of EpCAM total protein expression at the plasma membrane after treatment of cells with EpCAM-specific siRNA is exemplified in Supplementary

Figures S1A and B. EpCAM- and FHL2-specific siRNAs were similarly effective and resulted in knockdown efficiencies of 80% and 70% at the mRNA level and 66% and 70% at the protein level in immunoblots, respectively (Figures 3a and b). Experiments were conducted under standard conditions supplementing cells with 10% fetal calf serum (FCS) and under restrictive conditions with 1% FCS only. After 24, 48 and 72 h, microphotographs of cells were taken (Figure 1a). Especially at 48 h after transfection of siRNAs, reduction of EpCAM or FHL2 levels was associated with 46 -- 56%

1% FCS

FHL2 si

EpCAM si

Ctrl. si

10% FCS

24 hrs

48 hrs

72 hrs

24 hrs

48 hrs

72 hrs

** **

**

cells *105

* cells *105

642

** ** ** **

** ** * *

- 46%

- 53%

- 51%

- 56%

Figure 1. Inhibition of EpCAM expression results in decreased proliferation. (a) FaDu hypopharynx carcinoma cells were transiently transfected with control, EpCAM or FHL2 siRNAs in medium complemented with 1 or 10% FCS. Cells were visualised on microphotographs at the indicated time points. Shown are representative results from three independent experiments. (b) Cells were treated as described above. Cell numbers were counted at the indicated time points. Shown are mean values with s.d. from three independent experiments for cells kept in media containing 10% FCS (left panel) or 1% FCS (right panel). *P40.05; **P40.01. (c) Cell numbers at the time point 48 h after transient transfection were all standardised to values of control siRNA-treated cells for comparison. Shown are mean values with s.d. from three independent experiments for cells kept in media containing 10% FCS (left panel) or 1% FCS (right panel). (d) Doubling times of cells transiently transfected with control, EpCAM or FHL2 siRNAs were calculated and are given in hours with s.d. from three independent experiments. (e) Significance of the observed differences in doubling times was assessed with paired Student’s t-test comparing all three groups of cells under permissive and restrictive conditions as indicated. Oncogene (2013) 641 - 650

& 2013 Macmillan Publishers Limited


643

decreases in cell numbers under both growth conditions (Figures 1b and c). At day 3 after siRNA transfection, cells grown under nonrestrictive conditions, but not under restrictive conditions, were able to partially gain on control-treated cells (Figure 1b). Generally, cells grown under restrictive conditions displayed tenfold less cells at each of the three times points (24, 48 and 72 h). Calculated doubling times corroborated these findings. Inhibition of EpCAM or FHL2 expression under standard conditions increased doubling times by 2.3 and 2.5 h, respectively. Under restrictive growth conditions, EpCAM or FHL2 suppression induced an almost doubling of generation times from 20.25 to 39.47 h and 41.31 h, respectively (Figure 1d). The significance of the observed differences in doubling times was calculated across all three groups of cells under normal and restrictive conditions. All differences observed in Figure 1d were considered highly significant (Figure 1e), except between cells treated with EpCAM- or FHL2-specific siRNAs. Under restrictive conditions with supplementation of 1% FCS only, cells treated with EpCAM or FHL2 siRNA did not significantly differ in doubling times. Under permissive conditions (10% FCS), differences in doubling times between these two cell groups were considered significant (0.03; Figure 1e). Hence, knockdown of EpCAM and FHL2 similarly enhanced doubling times of FaDu cells as compared with control-treated cells. The repartition of cells in the different phases of the cell cycle was assessed in dependency of the expression of EpCAM and FHL2 in FaDu carcinoma cells. Cells were transiently transfected with control, EpCAM or FHL2 siRNAs, and cell cycle distribution was monitored upon staining with propidium iodide. Inhibition of EpCAM or FHL2 expression led to a reproducible accumulation in G1 phase, associated with significantly decreased percentages of cells in S phase and G2 phase (Figure 2). Effects of EpCAM and FHL2 inhibition on proliferation and cell cycle progression were similar in their amplitude. EpCAM regulates the transcription of the cyclin D1 gene In order to test if EpCAM affects progression through the cell cycle directly or indirectly, the expression of the essential regulator cyclin D1 was monitored in dependency of EpCAM and FHL2. FaDu cells were transiently transfected with control, EpCAM or FHL2 siRNAs, and levels of EpCAM and FHL2 visualised in immunoblots. EpCAM and FHL2 siRNA oligonucleotides induced reductions of mRNA and protein amounts after 48 h in the range of 65 -- 80% (Figures 3a and b). Messenger RNA levels of cyclin D1 were addressed upon quantitative real-time PCR in parallel. After knockdown of EpCAM or FHL2, cyclin D1 mRNA levels were reduced by 75% and 60%, respectively (Figure 3b). Similar levels of reduction of cyclin D1 were observed at the protein level, confirming results of quantitative PCR (Figure 3c). Comparable experiments were performed with A549 lung carcinoma cells. Reduction of EpCAM or FHL2 expression upon treatment of A549 with specific siRNA resulted in an average 55% decrease in cell numbers after 72 h of treatment and in 50% decrease in cyclin D1 mRNA levels (data not shown). The cyclin D1 gene is regulated via the canonical b-catenin/Lef-1 pathway, and Lef-1 consensus sequences within the promoter have been identified.15 Lef-1 consensus sequences originating from the cyclin D1 promoter (probes) were used to perform electromobility shift assays (EMSAs) with nuclear and cytoplasmic extracts of FaDu cells. Two major protein/DNA complexes (I and II) could be detected after incubation of radioactively labelled probes with FaDu nuclear extracts (Figure 3d, left panel). The specificity of protein/DNA complexes for Lef-1 sequences was addressed upon competition experiments with unlabelled oligonucleotides representing the Lef-1 consensus sequence of the cyclin D1 promoter (Top) or a mutated version to which Lef-1 no longer binds (Fop). Titration of cold Top-oligonucleotides but to a far lesser extent of Fop-oligonucleotides resulted in a gradual decrease of both major protein/DNA complexes (Figure 3d, left panel). Upon biochemical & 2013 Macmillan Publishers Limited

Figure 2. Cell cycle analyses in dependency of EpCAM and FHL2 expression. (a) FaDu cells were transiently transfected with control, EpCAM or FHL2 siRNAs and cell cycle distribution monitored with propidium iodide (PI). Shown are the mean percentages of cells in G1 phase (black bars), S phase (light grey bars) and G2 phase (dark grey bars) with s.d. from three independent experiments. Additionally, percentages of cells in G1 phase (b), S phase (c) and G2 phase (d) are shown independently. **P40.01.

and genetic approaches, EpICD was described as a component of a nuclear (but not cytoplasmic) protein complex involved in the formation of protein/DNA complex II.10 Hence, processing of EpCAM to generate the signalling moiety EpICD via regulated intramembrane proteolysis was hampered with presenilin-specific inhibitor N-[N-(3,5-Difluorophenacetyl)-L-alanyl]-S-phenylglycine tbutyl ester (DAPT). Nuclear and cytoplasmic fractions of treated and untreated cells were prepared and subjected to EMSA with cyclin D1 probes in duplicates. Formation of protein/DNA complex Oncogene (2013) 641 - 650


644 Ctrl si

Ctrl si Ep si actin EpCAM

FHL2 1

Ratio:

0.34

Ratio:

1

EpCAM

0.8

Cyclin D1

0.6 **

**

0.2 0

FHL2-si/Ctrl-si ratio

Ep si/Ctrl si ratio

GAPDH

1

0.3 FHL2 si

EpCAM si 1.2

0.4

FHL2 si

actin

GAPDH

1.2

FHL2

1 0.8 **

0.4 0.2 0 FHL2-si

EpCAM-si Csi

24h

48h

actin

actin

Cyclin D1

Cyclin D1

Ratio:

1

Cyclin D1

**

0.6

0.89 0.34

Ratio:

Csi

24h

48h

1

0.9

0.4

FaDu Nuc. extr.

-

+

DNA comp.

-

-

+

+

+

+

FaDu

+ nuc.

Top 1:0.8 1:2

Fop 1:4

DAPT

-

+

cyt. +

+

nuc. +

-

cyt. -

-

-

1:2 1:4

I II

I II

Figure 3. EpCAM regulates cyclin D1 expression. (a) FaDu cells were transiently transfected with control, EpCAM or FHL2 siRNAs and subjected to immunoblotting with antibodies specific for actin, EpCAM and FHL2 at 48 h after transfection. Ratios were calculated using Image Lab 3.01 software (Bio-Rad Laboratories GmbH, Mu¨nchen, Germany). Shown are representative results from three independent experiments. (b) FaDu cells were transiently transfected with control, EpCAM or FHL2 siRNAs and were subjected 48 h later to real-time PCR quantification of GAPDH, EpCAM, FHL2 and cyclin D1 using specific primer pairs. Shown are mean ratios of Ct values from specific siRNA- versus control siRNA-treated cells with s.d. from two independent experiments. **P40.01. (c) Lysates from FaDu cells treated as mentioned in (b) were subjected to immunoblotting with antibodies specific for actin and cyclin D1. Shown are representative results from two independent experiments. (d) Nuclear (nuc.) and cytoplasmic (cyt.) extracts from FaDu cells were incubated with radiolabelled probes representing Lef1 consensus sites of the cyclin D1 promoter (Top). Two major protein/DNA complexes I and II were detected on X-ray films. Unlabelled Top or a mutated version (Fop) probes were added in increasing concentrations (see ratios indicated in the figure) to the reaction (left panel). Where indicated with or þ , FaDu cells were treated with the g-secretase inhibitor DAPT before harvesting cells for the generation of nuclear and cytoplasmic extracts in order to inhibit cleavage of EpCAM (right panel). Shown are the representative results of three independent experiments.

II was exclusively seen with nuclear but not cytoplasmic extracts (Figure 3d, right panel), whereas complex I was formed in both cellular compartments. However, formation of complex I was stronger in the nuclear fraction. Importantly, formation of complex II in nuclear extracts was totally abrogated upon treatment of FaDu cells with the presenilin-specific inhibitor DAPT, whereas formation of complex I was only weakly affected (Figure 3d, right panel). Incubation of nuclear extracts with anti-EpICD antibodies did not result in a supershift of protein/DNA complexes, but rather Oncogene (2013) 641 - 650

hampered the formation of protein/DNA complexes if preincubated with nuclear extracts. Pre-incubation of nuclear extracts with EpICD-specific antibodies overnight at 41C resulted in a 40-52% reduction of signal intensity, which was in the same range as samples pre-incubated with Lef-1- or b-catenin-specific antibodies for 30 min at room temperature (Supplementary Figure 2A). These data suggested that EpICD is potentially incorporated in both complexes I and II, and pre-incubation with EpICD-specific antibody inhibits DNA binding of the complexes. However, EpICD & 2013 Macmillan Publishers Limited


appears essential for the generation of complex II, while being dispensable for complex I. Alternatively, to address the requirement for EpICD during the formation of protein/DNA complexes at the cyclin D1 promoter, we used HEK293 cells, which are negative for EpCAM but are responsive to EpCAM-mediated proliferation.14 Nuclear extracts of HEK293 cells allow for the formation of protein/DNA complex I but not II (Supplementary Figure 2B). HEK293 cells were transiently transfected with expression plasmid for EpICD-YFP, and nuclear extracts of transient transfectants were incubated with labelled cyclin D1 promoter probes to perform EMSAs. Alternatively, a synthetic peptide composed of the 26 amino acids of EpICD was spiked into nuclear extracts of untransfected HEK293 cells and used similarly. Reconstitution of protein/DNA complex II and a moderate increase in the formation of complex I was detected after expression of EpICD-YFP (Supplementary Figure 2B). Supplementation of synthetic EpICD was not functional, which is in line with its rapid and quantitative degradation in cell lysates (data not shown). Thus, EpICD is essential for the formation of nuclear complex II at Lef consensus sites of the cyclin D1 promoter. EpCAM affects Rb phosphorylation, cyclin A and E2 transcription Cyclin D1 is the essential cofactor of cyclin-dependent kinases (CDKs) 4 and 6, which upon activation phosphorylate retinoblas-

toma protein (pRb), resulting in the release of the E2F transcription factor and subsequent transcription of cyclins A and E, which are involved in the progression through cell cycle.16 Based on the regulation of cyclin D1 by EpCAM at the transcriptional level, we next analysed the effects of EpCAM and FHL2 on the phosphorylation status of Rb at serine residue 780, which represents a requirement for S-phase transition17 and depends on cyclin D1.18 FaDu cells were transiently transfected with EpCAM- or FHL2-specific siRNA and harvested 24, 48 and 72 h after transfection. Lysates of these cells were assayed for Rb phosphorylation using the phospho-Rb serine780 kit. For a control, total levels of Rb, actin and EpCAM were assessed in parallel. Reduction of EpCAM or FHL2 proteins levels resulted in concurrent reduction of phosphorylation of serine780 within pRb, but not in the reduction of overall Rb levels (Figure 4a). Release of E2F upon phosphorylation of Rb is a prerequisite for the induction of further regulators of cell cycle transition such as cyclins A and E. Samples of FaDu cells subjected to treatment with EpCAM and FHL2 siRNAs were next tested for levels of cyclin A and E2 mRNAs and proteins. Reduction of EpCAM mRNA levels by 75% resulted in mean 30% and 40% reduction of cyclin A and E2 mRNAs, respectively (Figure 4b, left panel). Comparably, reduction of FHL2 mRNA by 70% induced a 50% and 35% decrease of cyclin A and E2 mRNAs, respectively (Figure 4b, right panel). These findings were corroborated at the protein level. EpCAM

FHL2-si

EpCAM-si Csi

48h

72h

actin

actin

EpCAM

FHL2

Rb

Rb

Rb (P-ser)

Rb (P-ser)

EpCAM si RNA

1 **

**

0.8

GAPDH

0.6

EpCAM 0.4

Cyclin E2

**

Cyclin A

0.2

24h

48h

72h

FHL2 si RNA

1.2 si/ctrl.si mRNA ratio

1.2 si/ctrl.si mRNA ratio

24h

Csi

1 *

0.8 ** 0.6

GAPDH

**

FHL2

0.4

Cyclin E2 0.2

Cyclin A

0

0 EpCAM-si Csi

24h

Csi

48h

actin

actin

Cyclin E

Cyclin E

Ratio:

1

0.87

0.45

Cyclin A Ratio:

Ratio:

1

FHL2-si 24h 48h

0.9

0.39

Cyclin A 1

0.8

0.6

Ratio:

1

1

0.75

Figure 4. EpCAM expression regulates Rb phosphorylation. (a) FaDu cells were transiently transfected with control, EpCAM or FHL2 siRNAs and subjected to immunoblotting with antibodies specific for actin, EpCAM, Rb and serine 780 phosphorylation of Rb at 24, 48 and 72 h after transfection as indicated. Shown are representative results of three independent experiments. (b) FaDu cells were treated as mentioned in (a) and subjected to quantitative real-time PCR with primers specific for EpCAM, FHL2, cyclin E2 and cyclin A. Shown are mean ratios with s.d. of specific versus control siRNA-treated cells. *P40.05; **P40.01. (c) FaDu cells were treated as mentioned in (b) and subjected to immunoblotting with antibodies specific for actin, cyclin E and cyclin A. Ratios were calculated using Image Lab 3.01 software (Bio-Rad). Shown are representative results from three independent experiments. & 2013 Macmillan Publishers Limited

Oncogene (2013) 641 - 650

645


knockdown induced 40% and 55% reduction of cyclin A and E protein levels, respectively. FHL2 knockdown resulted in 25% and 60% reduction of cyclin A and E protein levels, respectively (Figure 4c).

EpCAM, Ki67, cyclin D1 expression and Rb phosphorylation status correlate in vivo The in vitro experiments disclosed a tight correlation between the expression of EpCAM and EpICD-YFP with transcription and expression of cyclin D1, and phosphorylation of Rb protein. In order to validate these data in vivo, serial sections of primary samples of normal mucosa and head and neck carcinomas were immunostained with antibodies specific for EpCAM, Ki67, cyclin D1, Rb and phospho-Rb. Seven cases also included tumour proximal areas of microscopically normal mucosa characterised by regular Oncogene (2013) 641 - 650

80.8

76

80

43.2 (**)

60

cell %

cell (10*4)

100

40 20

160 140 120 94.04 100 100 53.48 80 (**) 60 40 20 0

129.9 (*)

YF YF PP/ Ep s IC D i -E ps i YF YF P-C P/ si IC D -C si

YF YF P-E ps P/ IC i D -E ps i YF YF P-C si P/ IC D -C si

0

Cell population

Doubling time (hours)

YFP-Epsi YFP-ICD-Epsi YFP-Csi

13.98 ± 0.18 11.30 ± 0.13 11.06 ± 0.06 10.18 ± 0.24

YFP-IC-Csi p-Value YFP-Epsi vs. YFP-ICD-Epsi YFP-Epsi vs. YFP-Csi YFP-ICD-Epsi vs. YFP-Csi YFP-Csi vs. YFP-ICD-Csi

si -C

-E

D

D

IC

IC

P/

si

P/

YF

2

YF

YF

P

+

2.5

C

Ep

si

ps

i

2.09 (*)

P-

1

0.002 0.001 0.12 0.02

YF

0.56 0.92 (*)

3 Ratio Cyclin D1

actin

1.5 1

EpCAM

0.5 ICD-YFP si

si

-C D

Cyclin D1

YF

P/

IC

P-

C

ps YF

-E D

P/

IC

P-

Ep

si

i

0

YF

EpICD induces cyclin D1 transcription and expression The observed reductions of cell proliferation, cyclin D1, cyclin A, E2 transcript and protein expression, and Rb phosphorylation upon EpCAM knockdown can in principle be due to any function of EpCAM. In order to delineate which function of EpCAM is instrumental, effects of the signalling moiety EpICD were addressed. FaDu cells were transiently co-transfected with control or EpCAM-specific siRNA oligonucleotides and expression plasmids for yellow fluorescent protein (YFP) or an EpICD-YFP fusion protein. EpCAM siRNA binding site corresponds to nucleotides coding for amino-acid sequence 47QCTSVG52 (N- to C-terminus) within the extracellular domain of EpCAM. Ectopically expressed EpICD mRNA is not a target for EpCAM siRNA and hence EpICD expression remained unaffected (Supplementary Figure 1C). Numbers of vital cells were assessed after 48 h upon Trypan blue exclusion assay. Transfection of EpCAM siRNA with YFP induced a significant mean 47% decrease in cell numbers as compared with control siRNA and YFP-transfected cells (Figure 5a). Ectopic expression of EpICD-YFP instead of YFP together with EpCAM siRNA restored cell numbers to 94% of cells treated with control siRNA (Figure 5a). Expression of EpICD-YFP with control siRNA disclosed proliferative effects of EpICD in FaDu cells and resulted in a mean 30% increase in cell numbers (Figure 5a). Accordingly, cells treated with control siRNA and ectopically expressing EpICDYFP displayed the shortest doubling times (10.18±0.24 h). Cells treated with control siRNA and expressing YFP were similar to cells treated with EpCAM siRNA but expressing EpICD-YFP (11.06±0.06 and 11.30±0.13 h), whereas cells treated with EpCAM siRNA and expressing YFP had the longest doubling times (13.98±0.18 h) (Figure 5a, first table). The significance of the observed differences was calculated with paired Student’s t-test, where values from cells transfected with YFP and control siRNA served as a reference (Figure 5a, second table). According to these calculations, EpICD induced a significant increase in cell number, whereas EpCAM knockdown resulted in significant decrease of cell number, which was rescued upon EpICD co-transfection (difference to YFP-Csi considered not significant). Hence, EpICD-YFP enhances proliferation of FaDu cells and can complement for the lack of wild-type EpCAM. Samples of the abovementioned siRNA-treated cells were studied for expression levels of cyclin D1 mRNA transcripts and protein. For comparison, samples of FaDu cells treated with control siRNA and ectopically expressing YFP were set to one and served as a reference. EpICD-YFP expression induced a 2.09-fold increase in cyclin D1 mRNA transcript under these conditions (Figure 5b), which was paralleled by a substantial induction of cyclin D1 protein expression (Figure 5c). Knockdown of EpCAM in FaDu cells resulted in a 44% decrease in cyclin D1 transcript and protein levels, which could be complemented upon overexpression of EpICD-YFP, both at the mRNA transcript and protein expression levels (Figures 5b and c).

105 (*)

120

YF

646

Figure 5. EpICD induces cyclin D1 transcription and protein expression. (a) FaDu cells were transiently transfected with control, EpCAM or FHL2 siRNAs in combination with expression plasmids for YFP or a fusion of EpICD and YFP (YFP/ICD). After 48 h, numbers of vital cells were assessed upon Trypan blue exclusion. Results are given as mean cell numbers with s.d. from three individual experiments (left panel) and as mean percentages with s.d., where values from cells transfected with control siRNA and YFP were set to one (right panel). Significance of the results was calculated with paired Student’s t-test, where values from cells transfected with YFP and control siRNA were used as the reference. *Significant Po0.05; **Highly significant Po0.001. Mean doubling times and s.d. are given in hours in the lower table. (b) FaDu cells were transiently transfected with control, EpCAM or FHL2 siRNAs in combination with expression plasmids for YFP or EpICD-YFP. After 48 h, cyclin D1 mRNA levels were assessed upon quantitative real-time PCR and adjusted for levels of GAPDH as a control. For a comparison, all values were additionally set into relation to values of cells transfected with control siRNA and YFP. Shown are mean values with s.d. from two individual experiments. *P40.05. (c) FaDu cells were transiently transfected with control, EpCAM or FHL2 siRNAs in combination with expression plasmids for YFP or EpICD-YFP. After 48 h, protein levels of actin, EpCAM, EpICD-YFP and cyclin D1 were assessed upon immunoblotting with specific antibodies. Shown are representative results from two independent experiments.

stratification and no morphological signs of transformation. In normal tissue, EpCAM was primarily expressed in cells of the first suprabasal membrane layers. Expression patterns tightly correlated & 2013 Macmillan Publishers Limited


647 mucosa EpCAM

oropharynx carcinoma Cyclin D1

EpCAM

Cyclin D1

mucosa EpCAM

Cyclin D1

Ki67

Rb

pRb

Rb

pRb

oropharynx carcinoma EpCAM

Cyclin D1

Ki67

Figure 6. Expression of EpCAM, Ki67, cyclin D1, Rb and phospho-Rb correlate in vivo. (a; left two microphotographs) Serial cryo-sections of normal mucosa from the head and neck area were analysed upon immunohistochemistry with antibodies specific for EpCAM and cyclin D1 (red -- brown staining). Nuclei were visualised with haematoxylin staining (blue staining). (a; right two microphotographs) Serial sections from the autologous oropharynx carcinoma were analysed upon immunohistochemistry with antibodies specific for EpCAM and cyclin D1 (red -brown staining). Nuclei were visualised with haematoxylin staining (blue staining). (b) Magnified serial sections from normal mucosa (upper panels) and from the autologous oropharynx carcinoma (lower panels) stained with antibodies specific for EpCAM, cyclin D1, Ki67, Rb and phospho-Rb (serine 780) are depicted. Actual cell sizes are indicated with scales in each microphotograph.

with cyclin D1 and Ki67 expression (Figure 6a, left two panels and Supplementary Figure S3). Magnifications of normal mucosas and of head and neck carcinomas are shown in Figure 6b and comprised cellular infiltrate, tumour cells and stroma. In normal mucosas, Ki67 was primarily detected in nuclei of cells of the first and second suprabasal membrane layers, whereas Rb was detected more broadly in the suprabasal layers, but predominantly in a serine780 non-phosphorylated form (Figure 6b, upper panels). Staining intensities of EpCAM, cyclin D1 and ki67 in normal mucosas are given in Supplementary Table S1. It must be noted that staining intensities relate to the first two to three suprabasal cell layers only. Remaining epithelial cells of normal mucosas did generally not express EpCAM, cyclin D1 or ki67 (Supplementary Table S1). Autologous head and neck carcinomas (n ¼ 11) were characterised by strong overexpression of EpCAM and cyclin D1 throughout or in greatest parts of the tumour mass, with cyclin D1 displaying a primarily nuclear localisation (Figure 6a, right two panels and Supplementary Figure S3). Staining intensities of EpCAM, cyclin D1 and ki67 are given in Supplementary Table S1. Tumour areas consistently expressed high levels of EpCAM, cyclin D1 and Ki67 and phosphorylated Rb with coincident expression patterns within the tumour mass (Figure 6b and Supplementary Figure S3). DISCUSSION First descriptions of a correlation of EpCAM with a proliferative and de-differentiated state of epithelia date back to the mid 1990s, when Litvinov et al.19 and Schon et al.20 analysed cervical intraepithelial neoplasia and transformed epithelia, respectively. EpCAM expression was correlated with the proliferation marker Ki-67 at the expression level in primary cervical intraepithelial & 2013 Macmillan Publishers Limited

neoplasia samples and was expressed to lesser degree in welldifferentiated cervical intraepithelial neoplasia cells.19 In spontaneously transformed keratinocytes, high expression of EpCAM was associated with enhanced cell proliferation and EpCAM levels decreased with cell ageing.20 Subsequent work demonstrated oncogenic and proliferation supporting capacities of EpCAM, along with its ability to induce transcription of the c-Myc oncogene.14,21 Activation of gene transcription by EpCAM relies on regulated intramembrane proteolysis of the molecule by tumour necrosis factor-a-converting enzyme and a g-secretase complex containing presenilin 2 as a catalytic subunit.10 The resulting signalling moiety EpICD translocates into the nucleus, contacts DNA at Lef-1 consensus sites, and as such regulates transcription of genes primarily involved in proliferation and metabolism.10,13,22 However, it remained open whether these effects of EpCAM, which are in support of proliferation and an undifferentiated pluripotent phenotype,1 are direct or rather secondary effects resulting from the induction of target genes with mitogenic activities. In the present study, we demonstrate that EpCAM regulates initial steps in the transition of cells through the cell cycle via its ability to induce the transcription of cyclin D1. Cyclin D1 is the first cyclin required for entry in the cell cycle, as it is essential in the transition from G1 phase to S phase as cofactor for CDK4 and CDK6. In cooperation, CDK4/6 associated with cyclin D1 dictate the phosphorylation status of the tumour-suppressor protein Rb. Upon binding to cyclin D1, CDK4 and CDK6 become catalytically active and licensed to phosphorylate Rb and, thereby, induce the release of the transcription factor E2F from sequestration by Rb. EpCAM regulates the transcription and expression of cyclin D1, and consequently the phosphorylation of Rb was affected by variations in levels of EpCAM. Knockdown of EpCAM Oncogene (2013) 641 - 650


648

induced a decreased cell proliferation, decreased expression of cyclin D1 and Rb phosphorylation, increase of cells in the G1 phase and proportionally diminished numbers of cells in the S and G2 phases. The signalling moiety of EpCAM, that is EpICD, was able to complement for the lack of EpCAM with respect to cell proliferation and cyclin D1 gene transcription, suggesting that EpCAM affects cell cycle progression via signalling-dependent mechanisms rather than via its cell adhesive properties. This notion is further supported by the fact that inhibition of FHL2 expression has similar effects than the EpCAM knockdown. FHL2 is a direct interacting partner of EpCAM, which is essential for signalling.10 Owing to its modular structure composed of protein -protein interaction domains (LIM domains), FHL2 can contact various molecules involved in EpCAM signalling including EpCAM itself,10 b-catenin, tumour necrosis factor-a-converting enzyme and presenilin 2.23 Interestingly, FHL2 was shown to regulate the expression of cyclin D1 in mouse fibroblasts via its ability to associate with the cyclin D1 promoter at TCF consensus sites.24 Similar to EpCAM, inhibition of FHL2 expression led to a substantial impairment of cell proliferation and to hypophosphorylation of Rb.24 Based on these and our findings, we conclude that FHL2 is a central component of the EpCAM signalling pathway, which is mandatory for the transmission of EpCAM signals from the plasma membrane into the nucleus and, more specifically, to target genes such as cyclin D1. Similar functions of FHL2 have been reporter for Rho signalling and are in further support of central functions of FHL2 in cellular signal transduction.23 However, the sole presence of FHL2, for example in HEK293 cells, was not sufficient to form protein/DNA complex II at the cyclin D1 promoter. Expression and proteolytic activation of EpCAM as seen in carcinoma cells or complementation of HEK293 cells with the pre-cleaved moiety EpICD was mandatory to this end. Nonetheless, FHL2 appears as an essential mediator of EpCAM signalling, most probably because of its ability to interconnect various proteins, including those central to regulated intramembrane proteolysis of EpCAM. As has been shown for EpCAM, FHL2 is overexpressed in vivo in gastric and colon cancers, where its suppression results in the inhibition of carcinogenesis.25 Similar to EpCAM,26 FHL2 overexpression in ovarian and breast cancer is a negative prognostic marker.27,28 Interestingly, FHL2 is also overexpressed in hepatocellular carcinomas,29 in which EpCAM is a negative prognostic marker, which marks tumourinitiating cells.30 Thus, FHL2 and EpCAM are conjointly upregulated in a substantial number of tumour entities, where they might play an important role in promotion of tumour initiation and growth. At the molecular level, reduced expression of EpCAM or FHL2 further correlated with a reduction of cyclin A and E transcripts. However, this reduction was not quantitatively proportional to the reduction of EpCAM transcripts, as was the case for cyclin D1, and revealed repeatedly inferior to the amplitude of EpCAM knockdown. Hence, these data suggest an indirect regulation of cyclins A and E by EpCAM, very likely through hyperphosphorylation of Rb and ensuing release of E2F, which is a direct inducer of cyclins A and E.31,32 Additionally, c-Myc becomes activated by EpCAM10,14 and is itself an inducer of cyclins A and E.33 Thus, EpCAM possibly induces expression of cyclins A and E via an indirect route possibly involving E2F and/or c-Myc. EpCAM was furthermore shown to repress the expression of p53 and p21 during processes of reprogramming of murine embryonic fibroblasts.12 Although this effect was so far not investigated in tumour cells, EpCAMmediated repression of p53 and p21 might reveal to be another level of regulation of the cell cycle by EpCAM in cancer. In light of the reported induction of genes involved in reprogramming and formation of tumour-initiating cells such as c-myc14 and pluripotency genes (Oct4, Nanog and KLF4),6 - 8,12 EpCAM apparently acts at several crucial levels of regulation of the cell cycle and pluripotency. It is noteworthy that the positive and negative Oncogene (2013) 641 - 650

feedback loops take effect on the expression of EpCAM. These include the upregulation of the epcam gene by Tcf-4, a member of the Wnt-b-catenin signalling pathway.34 Oppositely, EpCAM expression is repressed by p53, which was reported as a contribution of p53 to the inhibition of invasive growth in breast cancer.35 Accordingly, loss of p53 resulted in DNA hypomethylation and amplification of the epcam gene.36 Together with the fact that p53 represses Nanog expression and induces a differentiation programme in embryonic stem cells,37 these data point toward opposite expressions and functions of p53 and EpCAM in proliferation and differentiation. EpCAM overexpression was first described in human melanoma and colon cancer. EpICD was preferentially observed in the nucleus of colon carcinoma samples, but not in normal colon mucosa, where EpICD was retained at the plasma membrane.10 Interestingly, regulation of cyclin D1 expression is fulfilled by b-catenin and Lef-1 in colon carcinoma,15 both interaction partners present in nuclear complexes comprising EpICD and FHL2.1,10 Oligonucleotide sequences used for EMSAs are composed of Lef-1 consensus sites of the cyclin D1 promoter region. Two major protein/DNA complexes were detected in nuclear extracts of EpCAM-positive cells, when incubating with cyclin D1specific oligonucleotide probes. Both complexes revealed sensitive to increasing concentrations of unlabelled wild type but far less to mutated oligonucleotides. However, primarily the lower migrating protein/DNA complex II was sensitive to treatment of cells with g-secretase inhibitors, which avert EpCAM cleavage, whereas the intensity of complex I only weakly diminished. Additionally, complex II was lacking in EpCAM-negative cells but was reconstituted upon transient overexpression of EpICD. Thus, EpICD appears essential for the formation of complex II, whereas it is dispensable for complex I, although data indicate an integration of EpICD in both complexes. In this context, it should also be noted that EpICD is a highly labile protein, which becomes rapidly and efficiently degraded in cell lysates. This is corroborated by the fact that a synthetic peptide composed of the 26 amino acids of EpICD is not instrumental in driving the formation of protein/DNA complex II in EpCAM-negative cells, when spiked into cell lysates. In line with this finding, EpICD peptide is readily visualised in immunoblots only as a pure synthetic peptide but not after incubation in cell lysates.9 Molecular mechanisms underlying this efficient degradation are so far unexplored, but are likely involved in the fine-tuning of EpCAM signalling at the level of EpICD stability. From the data presented, we conclude that EpICD decisively contributes to the formation of one out of two major protein/DNA complexes, which bind to the cyclin D1 promoter. Obviously, these data also suggest additional and EpCAM-independent levels of regulation of cyclin D1 by b-catenin and Lef-1. However, protein/DNA complex I contains proteins already assembled in the cytoplasm and in EpCAM-negative cells (see Figure 3d and Maetzel et al.10), which is not the case for EpICD-containing complexes, and might hence support distinct functions during cyclin D1 regulation. It is conceivable that overexpression and activation of EpCAM signalling directly contributes to the deregulation of cell cycle transitions and to proliferation in colon carcinoma and other tumour entities, through the induction of cyclin D1. Importantly, a correlation of EpCAM with the expression of cyclin D1, Ki67 and the phosphorylation of Rb was detected in primary samples of head and neck carcinomas, strongly suggesting in vivo relevance of our findings. Actually, this type of regulation represents the most direct option to control G1/S transition and thus proliferation, and will have strong impact on tumourigenesis in case of a deregulated activation of EpCAM. Head and neck squamous cell carcinomas are characterised by strong overexpression of EpCAM38 and cyclin D1.39 Both EpCAM and cyclin D1 are predictors of an invasive phenotype and poor prognosis of head and neck squamous cell carcinoma & 2013 Macmillan Publishers Limited


patients.38,40 These correlations and a relationship between both proteins can now be understood at the molecular level and explain the observed bad prognosis.

649 combination with horseradish peroxidase-coupled secondary antibody (1:5000) and the ECL system (Millipore) as described before.10

Quantitative real-time PCR MATERIALS AND METHODS Cell line, cell lines and plasmids Hypopharynx carcinoma FaDu cells were cultured in Dulbecco’s modified Eagle’s medium with 1 or 10% fetal calf serum. Transient cell clones were generated by transfection using MATra reagent according to the manufacturer’s protocols (IBA GmbH, Go¨ttingen, Germany). For rescue experiment, YFP (pEYFP-N1 plasmid) and EpICD-YFP (EpICD complementary DNA cloned in pEYFP-N1 plasmid) plasmids (1 mg each) were cotransfected with the indicated siRNA oligonucleotides.

siRNA transfection FaDu cells were plated in six-well plates (0.5 105 cells per well) and transiently transfected with 100 nM of control, EpCAM or FHL2 siRNAs, respectively, using MATra transfection reagent. The siRNA oligonucleotide sequences are: Control siRNA: 50 -UCGUCCGUAUCAUUUCAAU-30 EpCAM siRNA: 50 -UGCCAGUGUACUUCAGU UG-30 FHL2 siRNA: 50 -CUGCUUCUGUGACUUGUAU-30 EpCAM siRNA binding site corresponds to nucleotides coding for aminoacid sequence 47QCTSVG52 (N- to C-terminus) within the extracellular domain of EpCAM.

GAPDH forward (FW) 50 -TGCACCACCAACTGCTTAGC-30 ; reverse (RV) 50 -GGCATGGACTGTGGTCATGAG-30 EpCAM FW 50 -GCAGCTCAGGAAGAATGTG-30 ; RV 50 -CAGCCAGCTTTGAG CAAATGAC-30 FHL2 FW 50 -CAACCCATCAGCGGACTT-30 ; RV 50 -TTAAAGCAGTCGTTATGC CACT-30 CycD1 FW 50 -TATTGCGCTGCTACCGTTGA-30 ; RV 50 -CCAATAGCAGCAAA CAATGTGAAA-30 Cyc E FW 50 -AGGACGGCGAGGGACCAGTG-30 ; RV 50 -TTTGCCCAGCTCAG TACAGGCAGC-30 Cyc A FW 50 -CCGTCCAACAACCGCGGACC-30 ; RV 50 -TA AGGGGTGC AACCCGTCTCGTC-30 All real-time PCR reactions were performed in a 10-ml mixture containing 1 ml of complementary DNA, 5 ml of QuantiTect SYBR Green PCR Master Mix and 0.2 mM of each primer at 60 1C. Real-time quantisation was performed using Roche LightCycler- carousel-based system.

Rb protein assay

Cell counting and doubling time 5

FaDu cells were plated in six-well plates (0.5 10 cells per well) and treated with EpCAM, FHL2 siRNA or control siRNAs. Cells were grown in Dulbecco’s modified Eagle’s mediumsupplemented with 1 or 10% fetal calf serum. Counting of vital cells was performed at indicated time points by Trypan blue exclusion. Doubling time was calculated 48 h post transfection as was described before.14 Statistical significance was calculated using paired Student’s t-test. P-values of o0.05 were considered significant, Pvalues of o0.001 were considered highly significant.

Electromobility shift assays Nuclear and cytosolic extracts from FaDu cells were prepared according to the manufacturer’s protocol (NE-PER; Pierce, Thermo Fisher Scientific Inc., Rockford, IL, USA) and incubated with 32P-labelled duplex oligonucleotide probes following a published protocol.10 Where indicated, cells were treated with 10 mM of the g-secretase inhibitor DAPT (Sigma Aldrich Chemie GmbH, Taufkirchen, Germany) or nuclear extracts supplemented with 200 ng synthetic EpICD peptide. Protein-DNA complexes were separated upon native polyacrylamide gel electrophoresis and visualised with X-ray films.

Cell cycle analysis 3 105 cells were plated and transfected with control, EpCAM or FHL2 siRNAs. Cells were grown in 10% FCS. Cell cycle distribution was assessed after 2 days using propidium iodide staining. Cell cycle distribution of cells in G1, S and G2/M phases was analysed in a FACScalibur cytometer (Becton Dickinson, Heidelberg, Germany).

Immunoblot analysis Cells were transiently transfected with control, EpCAM or FHL2 siRNAs, and thereafter lysed in 40 ml of lysis buffer (Tris-buffered saline/1% Triton and Roche protease inhibitors, Roche, Penzberg, Germany). Proteins were separated by 15% sodium dodecyl sulphate - polyacrylamide gel electrophoresis and transferred to a 45-mm nitrocellulose membrane (polyvinylidene fluoride; Millipore GmbH, Schwalbach/Ts, Germany). EpICD,10 FHL2 (Cell Sciences, Canton, MA, USA) and b-catenin (BD Pharmingen, Heidelberg, Germany) antibodies were used in dilutions of 1:1000 in & 2013 Macmillan Publishers Limited

Total RNA from transiently transfected cells was isolated using the RNeasy kit (Qiagen Deutschland, Hilden, Germany). First-strand complementary DNA was synthesised using 1 mg of total RNA (DNase treated) in a 20-ml reverse transcriptase reaction mixture (QuantiTect Rev. Transcription Kit; Qiagen). The mRNA regions of GAPDH, EpCAM, FHL2, cyclin D1, cyclin E and cyclin A were amplified using following primers:

3 105 FaDu cells were plated in six-well plate and transfected with siRNA as described before. At different time points, cells were harvested and lysed in 40 ml of lysis buffer (phosphatase inhibitors were added: 5 mM NaF and 2 mM Na orthovanadate). Proteins were separated using immunoblot analysis described above and antibodies used were human retinoblastoma phosphoserine780 and human retinoblastoma antibody (Phospho Rb-kit; Cell Signalling Company, New England Biolabs GmbH, Frankfurt, Germany).

Human samples and in vivo immunohistochemistry Samples were obtained from head and neck squamous cell carcinoma patients (n ¼ 11) after informed written consent during routine surgery and in accordance with institutional ethics approval (Ethikkommision der Medizinischen Fakulta¨t der Ludwig-Maximilians-Universita¨t Mu¨nchen). Seven cases also included tumour proximal areas of microscopically normal mucosa characterised by regular stratification and no morphological signs of transformation. Haematoxylin - eosin-stained cryosections (4 mm) of these samples were reviewed to confirm diagnoses, document areas of squamous carcinoma cells and accordingly areas of normal mucosa. Sections of 4 mm thickness were cut on Superfrost plus slides and subjected to immunohistochemical analysis using the avidin-biotin-PO-Complex method. In brief, consecutive sections were fixed for 5 min at room temperature in aceton, followed by incubation with fresh 3.5% paraformaldehyde for additional 5 min at room temperature. Endogenous peroxidase activity was blocked using 0.3% H2O2 in phosphate-buffered saline for 10 min, followed by overnight incubation at 4 1C with mouse monoclonal antibodies specific for EpCAM (Cell Signaling; 1:2000 dilution), cyclin D1 (DCS Innovative Diagnostik Systeme, Hamburg, Germany, ready-to-use dilution), Ki67 (Dako Diagnostica GmbH, Hamburg, Germany; 1:800 dilution), Rb 4H1 (Cell Signaling; 1:500 dilution) and a rabbit polyclonal antibody for phosphoRb Ser807/811 (Cell Signalling; 1:100 dilution). Thereafter, sequential incubations with biotinylated anti-mouse, respectively anti-rabbit secondary antibody and peroxidase-labelled avidinbiotin complex were conducted (Vector Lab. Inc., Burlingame, CA, USA). Amino-ethyl-carbazole peroxidase substrate was used for the detection of antigen/antibody complexes, generating a red - brown staining. Counterstaining was achieved with haematoxylin (blue). Negative controls were conducted simultaneously using mouse and rabbit isotype control antibody, respectively (Cell Signaling). Oncogene (2013) 641 - 650


650

CONFLICT OF INTEREST The authors declare no conflict of interest.

REFERENCES 1 Munz M, Baeuerle PA, Gires O. The emerging role of EpCAM in cancer and stem cell signaling. Cancer Res 2009; 69: 5627 - 5629. 2 Balzar M, Winter MJ, de Boer CJ, Litvinov SV. The biology of the 17-1A antigen (Ep-CAM). J Mol Med 1999; 77: 699 - 712. 3 Went P, Vasei M, Bubendorf L, Terracciano L, Tornillo L, Riede U et al. Frequent high-level expression of the immunotherapeutic target Ep-CAM in colon, stomach, prostate and lung cancers. Br J Cancer 2006; 94: 128 - 135. 4 Visvader JE, Lindeman GJ. Cancer stem cells in solid tumours: accumulating evidence and unresolved questions. Nat Rev Cancer 2008; 8: 755 - 768. 5 Gires O, Klein CA, Baeuerle PA. On the abundance of EpCAM on cancer stem cells. Nat Rev Cancer 2009; 9: 143 author reply 143. 6 Ng VY, Ang SN, Chan JX, Choo AB. Characterization of epithelial cell adhesion molecule as a surface marker on undifferentiated human embryonic stem cells. Stem Cells 2009; 28: 29 - 35. 7 Gonzalez B, Denzel S, Mack B, Conrad M, Gires O. EpCAM is involved in maintenance of the murine embryonic stem cell phenotype. Stem Cells 2009; 27: 1782 - 1791. 8 Lu TY, Lu RM, Liao MY, Yu J, Chung CH, Kao CF et al. Epithelial cell adhesion molecule regulation is associated with the maintenance of the undifferentiated phenotype of human embryonic stem cells. J Biol Chem 2010; 285: 8719 - 8732. 9 Denzel S, Maetzel D, Mack B, Eggert C, Barr G, Gires O. Initial activation of EpCAM cleavage via cell-to-cell contact. BMC Cancer 2009; 9: 402. 10 Maetzel D, Denzel S, Mack B, Canis M, Went P, Benk M et al. Nuclear signalling by tumour-associated antigen EpCAM. Nat Cell Biol 2009; 11: 162 - 171. 11 Chen HF, Chuang CY, Lee WC, Huang HP, Wu HC, Ho HN et al. Surface marker epithelial cell adhesion molecule and E-cadherin facilitate the identification and selection of induced pluripotent stem cells. Stem Cell Rev 2011; 7: 722 - 735. 12 Huang HP, Chen PH, Yu CY, Chuang CY, Stone L, Hsiao WC et al. Epithelial cell adhesion molecule (EpCAM) complex proteins promote transcription factormediated pluripotency reprogramming. J Biol Chem 2011; 286: 33520 - 33532. 13 Maaser K, Borlak J. A genome-wide expression analysis identifies a network of EpCAM-induced cell cycle regulators. Br J Cancer 2008; 99: 1635 - 1643. 14 Munz M, Kieu C, Mack B, Schmitt B, Zeidler R, Gires O. The carcinoma-associated antigen EpCAM upregulates c-myc and induces cell proliferation. Oncogene 2004; 23: 5748 - 5758. 15 Shtutman M, Zhurinsky J, Simcha I, Albanese C, D’Amico M, Pestell R et al. The cyclin D1 gene is a target of the beta-catenin/LEF-1 pathway. Proc Natl Acad Sci USA 1999; 96: 5522 - 5527. 16 Weinberg RA. The retinoblastoma protein and cell cycle control. Cell 1995; 81: 323 - 330. 17 Mittnacht S. Control of pRB phosphorylation. Curr Opin Genet Dev 1998; 8: 21 - 27. 18 Geng Y, Yu Q, Sicinska E, Das M, Bronson RT, Sicinski P. Deletion of the p27Kip1 gene restores normal development in cyclin D1-deficient mice. Proc Natl Acad Sci USA 2001; 98: 194 - 199. 19 Litvinov SV, van Driel W, van Rhijn CM, Bakker HA, van Krieken H, Fleuren GJ et al. Expression of Ep-CAM in cervical squamous epithelia correlates with an increased proliferation and the disappearance of markers for terminal differentiation. Am J Pathol 1996; 148: 865 - 875. 20 Schon MP, Schon M, Klein CE, Blume U, Bisson S, Orfanos CE. Carcinomaassociated 38-kD membrane glycoprotein MH 99/KS 1/4 is related to proliferation and age of transformed epithelial cell lines. J Invest Dermatol 1994; 102: 987 - 991. 21 Osta WA, Chen Y, Mikhitarian K, Mitas M, Salem M, Hannun YA et al. EpCAM is overexpressed in breast cancer and is a potential target for breast cancer gene therapy. Cancer Res 2004; 64: 5818 - 5824.

22 Munz M, Hofmann T, Scheibe B, Gange M, Junghanns K, Zeidler R et al. The carcinoma-associated antigen EpCAM induces glyoxalase 1 resulting in enhanced methylglyoxal turnover. Cancer Genomics Proteomics 2004; 1: 241 - 247. 23 Johannessen M, Moller S, Hansen T, Moens U, Van Ghelue M. The multifunctional roles of the four-and-a-half-LIM only protein FHL2. Cell Mol Life Sci 2006; 63: 268 - 284. 24 Labalette C, Nouet Y, Sobczak-Thepot J, Armengol C, Levillayer F, Gendron MC et al. The LIM-only protein FHL2 regulates cyclin D1 expression and cell proliferation. J Biol Chem 2008; 283: 15201 - 15208. 25 Guo Z, Zhang W, Xia G, Niu L, Zhang Y, Wang X et al. Sp1 upregulates the four and half lim 2 (FHL2) expression in gastrointestinal cancers through transcription regulation. Mol Carcinog 2010; 49: 826 - 836. 26 Spizzo G, Went P, Dirnhofer S, Obrist P, Moch H, Baeuerle PA et al. Overexpression of epithelial cell adhesion molecule (Ep-CAM) is an independent prognostic marker for reduced survival of patients with epithelial ovarian cancer. Gynecol Oncol 2006; 103: 483 - 488. 27 Gabriel B, Fischer DC, Orlowska-Volk M, zur Hausen A, Schule R, Muller JM et al. Expression of the transcriptional coregulator FHL2 in human breast cancer: a clinicopathologic study. J Soc Gynecol Investig 2006; 13: 69 - 75. 28 Gabriel B, Mildenberger S, Weisser CW, Metzger E, Gitsch G, Schule R et al. Focal adhesion kinase interacts with the transcriptional coactivator FHL2 and both are overexpressed in epithelial ovarian cancer. Anticancer Res 2004; 24: 921 - 927. 29 Ng CF, Zhou WJ, Ng PK, Li MS, Ng YK, Lai PB et al. Characterization of human FHL2 transcript variants and gene expression regulation in hepatocellular carcinoma. Gene 2011; 481: 41 - 47. 30 Yamashita T, Forgues M, Wang W, Kim JW, Ye Q, Jia H et al. EpCAM and alphafetoprotein expression defines novel prognostic subtypes of hepatocellular carcinoma. Cancer Res 2008; 68: 1451 - 1461. 31 Soucek T, Pusch O, Hengstschlager-Ottnad E, Adams PD, Hengstschlager M. Deregulated expression of E2F-1 induces cyclin A- and E-associated kinase activities independently from cell cycle position. Oncogene 1997; 14: 2251 - 2257. 32 Schulze A, Zerfass K, Spitkovsky D, Middendorp S, Berges J, Helin K et al. Cell cycle regulation of the cyclin A gene promoter is mediated by a variant E2F site. Proc Natl Acad Sci USA 1995; 92: 11264 - 11268. 33 Jansen-Durr P, Meichle A, Steiner P, Pagano M, Finke K, Botz J et al. Differential modulation of cyclin gene expression by MYC. Proc Natl Acad Sci USA 1993; 90: 3685 - 3689. 34 Yamashita T, Budhu A, Forgues M, Wang XW. Activation of hepatic stem cell marker EpCAM by Wnt-beta-catenin signaling in hepatocellular carcinoma. Cancer Res 2007; 67: 10831 - 10839. 35 Sankpal NV, Willman MW, Fleming TP, Mayfield JD, Gillanders WE. Transcriptional repression of epithelial cell adhesion molecule contributes to p53 control of breast cancer invasion. Cancer Res 2009; 69: 753 - 757. 36 Nasr AF, Nutini M, Palombo B, Guerra E, Alberti S. Mutations of TP53 induce loss of DNA methylation and amplification of the TROP1 gene. Oncogene 2003; 22: 1668 - 1677. 37 Lin T, Chao C, Saito S, Mazur SJ, Murphy ME, Appella E et al. p53 induces differentiation of mouse embryonic stem cells by suppressing Nanog expression. Nat Cell Biol 2005; 7: 165 - 171. 38 Michalides R, van Veelen N, Hart A, Loftus B, Wientjens E, Balm A. Overexpression of cyclin D1 correlates with recurrence in a group of forty-seven operable squamous cell carcinomas of the head and neck. Cancer Res 1995; 55: 975 - 978. 39 Martin B, Schneider R, Janetzky S, Waibler Z, Pandur P, Kuhl M et al. The LIM-only protein FHL2 interacts with beta-catenin and promotes differentiation of mouse myoblasts. J Cell Biol 2002; 159: 113 - 122. 40 Yanamoto S, Kawasaki G, Yoshitomi I, Iwamoto T, Hirata K, Mizuno A. Clinicopathologic significance of EpCAM expression in squamous cell carcinoma of the tongue and its possibility as a potential target for tongue cancer gene therapy. Oral Oncol 2007; 43: 869 - 877.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene (2013) 641 - 650

& 2013 Macmillan Publishers Limited