| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
,
,1
* Inserm U602, Université Paris XI;
AP-HP, Hôpital Paul Brousse, Service de Biochimie et Biologie Moléculaire, Université Paris XI, Villejuif, France;
UMR 7091 CNRS, Université Paris VI, Génomique Fonctionnelle et Biologie des Systèmes pour la Santé, Villejuif, France;
AP-HP, Hôpital Ambroise Paré, Service d’Anatomie Pathologie, Université Versailles St. Quentin, Boulogne-Billancourt, France; and
|| Department of Research, Basel University Hospital, Basel, Switzerland
1Correspondence: Service de Biochimie et Biologie Moléculaire, Hôpital Universitaire Paul Brousse, Université Paris-Sud/XI, 14 Ave. Paul Vaillant Couturier, 94804 Villejuif Cedex, France. E-mail: antoinette.lemoine{at}pbr.ap-hop-paris.fr
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
50% of human hepatocellular carcinomas (HCCs). To elucidate T-cad functions in HCC, we examined T-cad protein expression in normal and tumoral human livers and hepatoma cell lines and investigated its influence on invasive potential of HCC using RNA interference silencing of T-cad expression in Mahlavu cells. Whereas T-cad expression was restricted to endothelial cells (EC) from large blood vessels in normal livers, it was up-regulated in sinusoidal EC from 8/15 invasive HCCs. Importantly, in three of them (38%) T-cad was detected in tumor cells within regions in which E-cadherin expression was absent. Among six hepatoma cell lines, only Mahlavu expressed T-cad but not E-cadherin. T-cad exhibited a globally punctuate distribution in quiescent Mahlavu and additionally it concentrated at the leading edge of migrating cells. Matrigel invasion assay revealed that Mahlavu possess a high invasive potential that was significantly inhibited by T-cad silencing. Wound healing and random motility assays demonstrated that inhibition of T-cad expression in Mahlavu significantly reduced their motility. We propose that T-cad expression in tumor cells might occur by cadherin-switching during epithelial-mesenchymal transition and may represent an additional mechanism contributing to HCC metastasis.—Riou, P., Saffroy, R., Chenailler, C., Franc, B., Gentile, C., Rubinstein, E., Resink, T., Debuire, B., Piatier-Tonneau, D., Lemoine, A. Expression of T-cadherin in tumor cells influences invasive potential of human hepatocellular carcinoma.
Key Words: primary tumors • cadherin switch • cell invasion • hepatoma cell lines • RNA interference
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. Improvements in early diagnosis and surgical techniques of HCC have contributed to decreased mortality. However, survival remains poor even for those patients with better clinical and pathological features treatable by surgical resection or orthotopic liver transplantation (OLT), due to a high incidence of recurrence of HCC after surgery (2)
. The poor prognosis of HCC patients is mainly affected by invasiveness of primary tumors. A better knowledge of the molecular mechanisms responsible for tumoral recurrence should lead to a more effective treatment for HCC patients. Furthermore, identification of molecular markers specific for the metastatic potential of HCC and predictive for tumoral recurrence (3)
should facilitate defining a subset of HCC patients for whom OLT could be an effective and curative treatment.
Cadherins are a superfamily of transmembrane glycoproteins that typically mediate calcium-dependent homophilic intercellular adhesion (4)
, and perturbations in cadherins have been associated with cancers and especially with invasion and metastasis (5)
. Decreased expression and abnormal cellular distribution of E-cadherin (E-cad), the prototypical member of the classic cadherin family and putative tumor suppressor, have been frequently found to be associated with dedifferentiation and invasiveness in a variety of human malignancies (6)
, including primary HCC (7
, 8)
. The loss of E-cad has been shown to be frequently associated with a "cadherin switch," corresponding to de novo expression of mesenchymal cadherins (9)
, such as N-cadherin, P-cadherin, R-cadherin, LI-cadherin, and cadherin-11, and has been found in tumors with enhanced invasiveness and poor prognosis (10)
. For T-cadherin (T-cad, also known as CDH13, H-cadherin), an atypical GPI-anchored member of the cadherin superfamily (11)
, we have shown a significantly higher gene expression level in HCC tumoral samples compared with adjacent nontumoral samples or normal livers (12)
. Comparable results have been recently reported (13)
. However, the functional relevance of elevated T-cad gene expression to HCC progression is not known.
This study examines expression and cellular distribution of T-cad in both human liver samples and human hepatoma cell lines and investigates whether T-cad influences invasive potential of HCC. Our data indicate that T-cad is involved in tumoral invasion of hepatoma cells by, at least in part, influencing cell motility.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Human hepatoma cell lines HepG2, Hep3B, and PLC/PRF/5 were obtained from the American Type Culture Collection (Rockville, MD, USA). TONG, HA22TNGH, and Mahlavu were provided by Sanofi-Synthelabo Recherche (Chilly-Mazarin, France).
Cell culture conditions
All cell lines except HA22TNGH were grown in Dulbecco’s modified Eagle’s medium (DMEM; Invitrogen, Cergy-Pontoise, France) supplemented with 10% fetal calf serum (FCS; Perbio Sciences, Helsingborg, Sweden), penicillin (100 IU/ml), and streptomycin (100 µg/ml); for HA22TNGH, DMEM was replaced by RPMI (Invitrogen). Primary endothelial cells (EC) were cultured in EGM2-MV complete medium (BioWhittaker/Clonetics, Walkersville, MD, USA) in 0.2% gelatin-coated culture vessels and fibroblasts in RPMI complete medium.
Perfusion of human normal livers for isolation of hepatocytes, endothelial cells, and fibroblasts
Perfusion of human normal livers (n=3) was performed as described previously (14)
. Briefly, cells were isolated after collagenase (1 g/l) perfusion of human liver fragments. EC were collected by centrifugation of recovered perfusion liquid; viable hepatocytes by sedimentation after total dissociation of tissue fragment and purification over a 40% Percoll gradient (Amersham Biosciences, Piscataway, NJ, USA); fibroblasts after a second sedimentation. Primary cultures of less than four passages were used in this study. Phenotype of isolated EC and fibroblasts are shown in Supplemental Fig. S1.
Antibodies
Rabbit polyclonal antibody (pAb; antibody) EC-1 specific for human T-cad was raised against the first extracellular domain of T-cad as described previously (15
, 16)
. Monoclonal antibodies (mAbs) were E-cad (BD Biosciences, Le Pont de Claix, France), vimentin (Biodesign, Saco), CD9 (clone SYB-9, ref 17
), and beta-actin (Sigma, St. Quentin-Fallavier, France). Secondary Abs were: HRP-conjugated goat immunoglobulin (Ig) specific for rabbit (Amersham Biosciences) or mouse IgG (Santa Cruz Biotechnology, Santa Cruz, CA), Alexa488-conjugated goat Ig specific for rabbit or mouse IgG (Molecular Probes, Cergy Pontoise, France), biotin-labeled goat Ig specific for rabbit (Dako, Glostrup, Denmark) or mouse IgG (Dako), and HRP-conjugated avidin (Dako). TRITC-conjugated phalloidin was from Molecular Probes.
Immunoblotting
The method for immunoblotting has been described previously (18)
. Frozen tissue samples ground under liquid nitrogen or fresh cells were lysed in Laemmli sample buffer, and protein concentration was determined using the Bio-Rad Protein Assay (Bio-Rad, Marnes-la-Coquette, France). Samples (30 µg per lane) were electrophoresed in 8% SDS-polyacrylamide gels under reducing conditions. A fixed amount of human aorta CHAPS extract enriched in T-cad (105 and 130 kDa isoforms; ref. 19
) was coelectrophoresed as a standard enabling interexperimental comparisons of T-cad expression. Primary Abs to T-cad (0.45 µg/ml), beta-actin (1:1000), CD9 (1:2000), E-cad (1:500), and vimentin (1:500), followed by appropriate secondary HRP-conjugated Abs (1:3000) and enhanced chemiluminescence (ECL; Amersham Biosciences) were used.
Immunohistochemistry
Paraffin embedded tissue sections (4 µm) were prepared according to classical methods and treated with blocking solution before being sequentially incubated with primary Abs (T-cad, 0.9 µg/ml; E-cad, 1:50) overnight at 4°C, followed by biotin-labeled secondary Ab and HRP-conjugated avidin (each for 60 min at room temperature). Detection was achieved with diaminobenzidine and hematoxylin counter staining. For negative controls, the primary Ab was substituted by same species normal IgG.
Immunofluorescence
Cell lines plated onto round 12-mm glass coverslips were cultured for selected time periods. For migrating cell analysis, confluent monolayers were scraped wounded and cultured further to allow migration out from the wound edge. Cell fixation/permeabilization was performed with 3.7% paraformaldehyde (Merck, Haar, Germany) and 0.1% Triton X-100 in PBS (each for 10 min) after or before staining of live or fixed cells, respectively. Staining buffers for live or fixed cells were culture medium containing 25 mM HEPES, 2 mM CaCl2, 0.03% sodium azide, or PBS containing 0.5% BSA, respectively. Ab dilutions were T-cad (4.5 µg/ml), Alexa488-conjugated goat anti-rabbit IgG (1:500), and normal rabbit IgG (4.5 µg/ml), each incubated for 45 min at room temperature. F-actin was stained with TRITC-phalloidin (25 ng/ml) for 1 h at room temperature. Nuclei were counterstained with TO-PRO3 (Molecular Probes). Coverslips were mounted with Mowiol (Sigma) and examined under a fluorescence microscope or a laser scanning confocal microscope (LSM 510, Zeiss, Esslingen, Germany). The excitation wavelengths for Alexa488 and TRITC were 488 and 543 nm, respectively. Image analysis was performed using LSM Image Examiner software (Zeiss).
siRNA transfection
Silencing of T-cad by RNAi was carried out using two different siRNAs (Ambion, Huntingdon, UK): T-cad siRNA-1 (5'GGACCAGUCAAUUCUAAAC3') and T-cad siRNA-2 (5'GGAACAAUGACUACUUUUU3'). A third T-cad siRNA without effects on the expression of T-cad at both transcriptional and protein levels was used as control (5'CUCUGUUCGUCCAUGCACG3'). Mahlavu (8x106 cells) grown to near confluence were harvested by brief trypsinization, washed, and resuspended in 400 µl of serum-free medium containing siRNA duplex (500 nM final). After two successive transfections (with a 24 h interval) by electroporation (300 V, 960 µF, 18–22 ms time constant), cells were cultured in complete medium and recovered appropriately for functional assays or immunoblotting.
Proliferation assays
Proliferation of cells plated at different densities (2.5x104 and 5x104 cells/well in 24-well dishes) over a 4 day period was determined by crystal violet dye assay. The relative cell number was calculated using a standard curve.
Invasion assays
Transwell chambers with polycarbonate membrane filters (6.5mm diameter, 8 µm pore size, Costar, Corning, NY) were coated with Matrigel (100 µg/cm2/100 µl PBS; BD Biosciences). Cells (1x105/200 µl DMEM-0.1% FCS) were added to the upper compartment and incubated for 24 h at 37°C. The lower compartment contained DMEM-1% FCS as chemoattractant. After removal of nonmigratory cells from the upper surface of the filter, invasive cells that had passed through to the lower surface of the filter were fixed and stained with crystal violet. Invasive cells were scored by counting at least five fields per filter. Counting accuracy was controlled by optical density (OD)570 quantification of methanol-solubilized dye.
Random cell motility assay
Random cell motility was examined by time lapse videomicroscopy. Cells plated at low density in complete medium (8000 cells/well; ref. 20
) into 24-well plates coated with Matrigel (10 µg/ml) were cultured for 14 h at 37°C in a 5% CO2 chamber placed on the motorized stage of a Leica inverted microscope equipped with a Princeton MicroMax CCD camera. Phase-contrast images were captured and analyzed with Metamorph software (Metamorph Imaging System, Molecular Devices Corp., Downingtown, PA, USA). The motility of individual cells was evaluated by tracking their movement from images recorded every 10 min. The average speed (µm/h) of locomotion was calculated as the total track length divided by the number of hours recorded. For each experimental condition, 30–50 cells were analyzed.
Wound healing assay
Confluent cell monolayers into 24-well plates (2.5x105 cells/well in complete media) were scrape-wounded using a micropipette tip. After being washed, cells were cultured in complete media for 10 h and time lapse videomicroscopy was performed as described above. Migration rate (average velocity) and degree of wound closure were assessed by measuring the distance between wound edges at time intervals of 2.5 h.
Statistics
Statistical analyses were performed by one-way ANOVA followed by post hoc Bonferroni’s multiple comparison when appropriate. A P value of <0.05 was considered significant.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. To confirm these results at the protein level, expression of T-cad in liver tissue samples derived from three patients with HCC (Fig. 1
A) and three healthy donors (normal liver, Fig. 1B
), and in whole-cell lysates of 6 hepatoma cell lines (Fig. 1C
) was examined by immunoblotting (15
, 16)
. Both forms of T-cad protein (mature 105 kDa and precursor 130 kDa) were overexpressed in tumoral samples (TL) as compared with same patient nontumoral samples (NTL; Fig. 1A
) and normal liver samples (Fig. 1B
). Among six hepatoma cell lines examined only Mahlavu expressed T-cad (Fig. 1C
) and at a higher level than in whole tissue tumoral samples as shown by variations in exposure times (Fig. 1, c
.f. intensities of the control T-cad standard in each blot). This confirms the differences observed at mRNA level (12)
. The precursor T-cad isoform was predominant in Mahlavu cells (Fig. 1C and D
), whereas the mature isoform was more abundant in whole liver extracts (Fig. 1A, B
). Trypsin/EDTA treatment of Mahlavu cells before lysis and immunoblot analysis resulted in the loss of most, if not all, 105 and 130 kDa proteins (Fig. 1D
). Thus both mature and precursor isoforms are expressed on the cell surface, as previously observed in various cell types (21
, 22)
. Incomplete loss of T-cad proteins even with extended trypsin/EDTA treatment, as observed in this and other studies (19
, 21)
, might reflect the relative insensitivity of T-cad to proteolytic degradation and/or the presence of incompletely processed intracellular T-cad.
|
Cellular distribution of T-cad in normal and tumoral livers
To evaluate the cellular distribution of T-cad within human liver, T-cad protein expression was investigated in the different cell types (hepatocytes, primary fibroblasts, and EC) isolated from three human normal liver biopsies (Fig. 1E
). Compared with its expression in whole normal liver (Fig. 1B
), T-cad protein was very weakly expressed in isolated hepatocytes, but strongly expressed in primary cultures of fibroblasts and EC (Fig. 1E
). Quantitative reverse transcriptase-polymerase chain reaction (RT-PCR) yielded similar results (data not shown). Phenotype characterization of primary cultures of fibroblasts, which exhibited the most abundant level of expression of T-cad (Fig. 1E
), revealed a strong expression of 
-smooth muscle actin, considered to be specific for myofibroblasts (Supplemental Fig. S1).
The distribution of T-cad in normal and tumoral human liver sections was examined immunohistochemically. For both normal and nontumoral liver tissues, T-cad was undetectable in hepatocytes, strongly expressed in EC of large blood vessels and occasionally a weak staining of sinusoidal EC was observed (Fig. 2
A). In 8/15 (53%) primary HCCs, a strong staining of sinusoidal EC was also observed (Fig. 2B
). T-cad-positive HCC were larger and characterized by higher tumoral vascular invasion than T-cad-negative HCC (Supplemental Table 1). Moreover, in three of the eight T-cad-positive primary HCC, within some areas T-cad expression was also detected in tumor cells (Fig. 2C, D
). T-cad was globally distributed over the cell body of tumor cells bordering sinusoids with T-cad-positive EC (Fig. 2C, D
) and without visible accumulation at cell-cell contacts. Thus, tumor-associated T-cad overexpression in sinusoidal EC and, in some cases, tumor cells largely accounts for the higher T-cad protein expression in tumoral whole-liver extracts (Fig. 1)
.
|
Cellular localization of T-cad in quiescent and migrating Mahlavu cells
Expression of T-cad in Mahlavu cultures was analyzed by fluorescence or confocal laser microscopy after staining with anti-T-cad Ab alone (Fig. 3
A) or costaining with TRITC-conjugated phalloidin (Fig. 3B
). Immunostaining for T-cad in both live-cells (Fig. 3A
, left panel) and fixed and permeabilized cells (Fig. 3A
, right panel) revealed a global and punctuate localization as previously shown for cultured EC and smooth muscle cells (16
, 23)
. No staining for T-cad was observed in hepatoma cells found to be T-cad negative according to RT-PCR and immunoblotting (examples shown in Supplemental Fig. S2A). Live-cell staining for T-cad in Mahlavu confirms its surface location, and the global localization in vitro resembles that observed for T-cad-positive cells in liver (Fig. 2)
and vascular (15)
tissue sections.
|
To investigate T-cad localization in migrating Mahlavu, wounded monolayers were fixed and permeabilized and then double-stained for both T-cad and F-actin and examined under confocal microscopy (Fig. 3B
). Marked changes of T-cad localization were observed in Mahlavu migrating into the wound area (Fig. 3B, d, g
) as compared with quiescent cells (Fig. 3B, a
). In addition to the global punctuate distribution, T-cad was distinctly enriched at the leading edge of migrating cells 2 h after wounding (Fig. 3B, d
) before concentrating within peudopodia-like structures 10 h after wounding (Fig. 3B, g
). Whereas quiescent Mahlavu exhibited abundant and uniformly distributed actin stress fibers (Fig. 3B, b
), migrating Mahlavu displayed typical cytoskeleton remodeling: 2 h after wounding actin bundles enriched at the leading edge and forming lamellipodia protruding into wound area were visible and associated with a significant decrease of actin stress fibers (Fig. 3B, e
; at 10 h condensed actin was concentrated in pseudopodia and stress fibers reappeared (Fig. 3B, h
). Colocalization of T-cad and condensed actin was visible within lamellipodia and pseudopodia of migrating Mahlavu (Fig. 3B, f, i
, arrows), whereas no colocalization was visible in quiescent Mahlavu (Fig. 3B, c
). This suggests a role for T-cad in cell invasion/migration.
Phenotype characterization of Mahlavu and hepatocytes within primary tumors
Expression of proteins known to switch during epithelial-mesenchymal transition (EMT), a cellular process used by cancer cells to convert to invasive cells (24)
, was investigated by immunoblotting and immunofluorescence. The six hepatoma cell lines tested fell into three categories based on their relative expression of E-cad and vimentin (Table 1
). PLC/PRF/5 and HepG2 were E-cad-positive and vimentin-negative, corresponding to an epithelial phenotype. TONG and Hep3B expressed both proteins, corresponding to an intermediate phenotype. Mahlavu and HA22TNGH were E-cad-negative and vimentin-positive, corresponding to a mesenchymal phenotype. Interestingly, HA22TNGH cells exhibited complete methylation of T-cad gene promoter (ref. 25
data not shown), which possibly explains their absence of T-cad protein. We conjecture that T-cad expression in hepatoma cells might be associated with a mesenchymal status and reflect a further cadherin switch during EMT.
|
To investigate this possibility, expression of E-cad in tumor cells was evaluated by immunohistochemical analysis on consecutive tumor liver sections from the 15 patients with HCC (Fig. 4
). Interestingly, the three HCCs displaying regions with T-cad-positive tumor cells (Fig. 4A, C
) also exhibited, in these regions, an overall pattern of aberrant weak cytoplasmic immunostaining for E-cad or a loss of E-cad expression in tumor cells (Fig. 4B, D
). In contrast, in the 12 other HCCs without T-cad-positive tumor cells, albeit exhibiting T-cad staining of some sinusoidal EC (Fig. 4E, G
), there was a strong typical intercellular membrane staining for E-cad (Fig. 4F, H
). Vimentin expression was undetectable in either the noninvasive or the invasive HCCs (data not shown). We suggest that the abnormal expression pattern of both T-cad and E-cad by tumor cells possibly reflects intermediate stages of EMT and a cadherin switch during this process in the three invasive HCCs.
|
Characterization of invasive properties of T-cad in hepatoma cells
To investigate the putative role of T-cad in the invasive potential of HCCs, the six human hepatoma cell lines were compared in Matrigel invasion assays. Mahlavu cells exhibited the strongest invasive potential (309±47 cells/field, P<0.0001; Fig. 5
). Among the T-cad negative cell lines only PLC/PRF/5 was invasive (201±34 cells/field, P<0.0001) but to a significantly lesser extent than Mahlavu (P<0.0001). HA22TNGH cells were poorly invasive (38±3 cells/field) as were HepG2, Hep3B or TONG (Fig. 5)
. Because both Mahlavu and HA22TNGH display mesenchymal features (i.e., fibroblastoid morphology, E-cad negative, and vimentin positive) but only Mahlavu express T-cad and are invasive, we speculate that T-cad protein expression in hepatoma cells with a mesenchymal phenotype could play some facilitatory role in invasion.
|
To study T-cad-specific effects in Mahlavu cells, we exploited RNAi. Two different siRNAs (T-cad siRNA-1 and T-cad siRNA-2) significantly inhibited the expression of T-cad in Mahlavu transfected cells, whereas T-cad expression was not affected by a further T-cad siRNA (control siRNA) or by irrelevant CD9 siRNA (Fig. 6
A). These siRNAs were used for functional tests to specifically determine T-cad related properties. Quantitative RT-PCR (data not shown) and immunoblot analysis (Fig. 6B
) showed that within 15 h after completion of transfection protocols, T-cad expression decreased by 80% in T-cad siRNA-transfected Mahlavu and remained suppressed for at least 4 days (Fig. 6B, C
). Levels of T-cad expression were unchanged in mock-transfected cells and control siRNA-transfected cells (data not shown). There was no induction of E-cad expression in T-cad siRNA-transfected Mahlavu (data not shown).
|
T-cad-dependent effects on the proliferation status of Mahlavu cells were investigated. Regardless of initial seeding density, cell growth over a period of 4 days was not different between cells in which T-cad expression was suppressed by siRNA transfections and the control cells (mock and control siRNA transfected; Fig. 6D
).
Involvement of T-cad in the invasive potential of Mahlavu was investigated using Matrigel invasion assay. For cells in which T-cad expression was suppressed, invasive potentials were significantly lower (40%) than those of control transfected (180 cells/field vs. 300 cells/field for T-cad siRNA1/2 and mock/control siRNA, respectively; Fig. 7
A) and untransfected Mahlavu (Fig. 5)
. T-cad effects on cell migration were investigated using random motility and wound assays (Fig. 7B-D
). Random motility assay showed significantly reduced (30%) mean velocities for Mahlavu cells with inhibited T-cad expression (30 µm/h) compared with mean velocities of control Mahlavu (42 µm/h; Fig. 7B
). Wound assay showed significant reductions (45%) in both mean velocities (Fig. 7C
) and degree (Fig. 7D
) of wound closure for Mahlavu cells with suppressed T-cad expression compared with those of control cells. Thus we conclude that T-cad expression in Mahlavu is an important determinant of their invasive potential and that T-cad-dependent migration in these cells is, at least in part, responsible for their invasive potential.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Although T-cad can mediate weak homophilic cell-cell aggregation (21)
, accumulating evidence suggests that T-cad might not function as a true intercellular adhesion molecule (23
, 26)
. Its function in cancer invasion is not known. Down-regulation of T-cad gene expression related to promoter hypermethylation has been frequently reported in breast, lung, and colon carcinomas (27)
. Contrastingly, in primary HCCs our previous (12)
and present results show that T-cad transcripts and proteins were globally overexpressed in about half of tumoral samples. Similar observations have been recently reported (13)
, reflecting a peculiarity of T-cad gene expression in HCC as compared with other tumor types. In further agreement with Adachi et al. (13)
we frequently found T-cad overexpression in sinusoidal EC in HCCs. Previous reports (28
, 29)
have shown that T-cad was increased in blood vessels penetrating lung and mammary tumors in mice, suggesting that T-cad is a marker of tumor angiogenesis. Similarly, overexpression of T-cad during pathological angiogenesis associated with atherosclerosis (15)
and restenosis (30)
has been described, and in vitro studies (23
, 31)
have established that T-cad overexpression in vascular cells promotes abnormal migration, growth, and phenotypic modulation. Together these results support the hypothesis that T-cad overexpression in HCC sinusoidal EC may represent a marker of tumor progression.
More remarkably, for three invasive HCCs of the eight exhibiting T-cad-positive sinusoidal EC, in some areas T-cad was also expressed in tumor cells. This observation contrasts with the results of Adachi et al. (13)
who did not detect T-cad expression in HCC tumor cells. Differing etiologies may explain this discrepancy since Adachi et al. (13)
focused on T-cad expression in HCC consecutive to viral infection, while we also examined HCC specimens associated with alcohol consumption. However, the three HCCs with T-cad-positive tumor cells were derived from both etiological types. Geographical variations in environmental and dietary carcinogens or even ethnic specific genetic polymorphisms of xenobiotic-metabolizing enzymes might be further considered. For example, aflatoxin B1 is known to play an important role in the etiology of HCC of Asiatic origin (1)
and the xenobiotic aryl hydrocarbon receptor ligands can repress T-cad gene expression (32)
.
T-cad overexpression was only found in tumor cells from tumoral areas of HCCs exhibiting aberrant E-cad. A similar T-cad expression associated with lost E-cad expression in Mahlavu, the most invasive cell line, substantiates use of Mahlavu as a relevant cellular model for investigating the function of T-cad in HCC. Exploitation of T-cad specific siRNAs in functional assays demonstrated the role of T-cad in invasiveness, at least in part through its influence on cell migration, a key step for invasion.
T-cad mediated effects on cellular migration might rely on the distribution of GPI-anchored T-cad to lipid rafts in the plasma membrane (33)
, where numerous signaling proteins are localized (34)
. T-cad enrichment to the leading edge of migrating cells, as shown for Mahlavu (this study) and for vascular EC and smooth muscle cells (16)
, supports this hypothesis, and concords with previous work demonstrating that lipid rafts redistribute to the leading edge of migrating cells during chemotaxis (35)
. Moreover, the colocalization of T-cad with condensed actin within lamellipodia and/or pseudopodia during the migration process suggests the existence of molecular adaptor(s) linking GPI-anchored T-cad to signaling partners involved in mediation of T-cad effects on cell migration. Homophilic ligation interactions between T-cad molecules expressed on the cell surface of EC and T-cad molecules immobilized on culture plates have been shown to induce a repulsion associated with enhanced cell migration in a RhoA- and Rac-dependent manner (23
, 36)
. Accordingly, we may speculate that interactions between T-cad molecules expressed at the surface of tumor cells might enhance their migration and influence their intravasation, as previously reported for N-cadherin, the prototypical mesenchymal cadherin (37
38
39)
. In contrast to its effects on migration/invasion, inhibition of T-cad expression did not influence Mahlavu proliferation, thus differing from previous observations in EC (31)
. The reason for this discrepancy was not investigated. However, the predominant expression of 130 kDa precursor isoform in Mahlavu vs. the more equivalent expression of mature and precursor forms in EC suggests different T-cad isoforms possibly mediate distinct cellular functions associated with proinvasive and proliferative phenotypes, respectively. Further investigations are required to clarify the mechanisms of action of T-cad in tumor cells and the role of its isoforms in cellular invasion.
Although the existence and importance of EMT in tumor progression and metastasis is a matter of debate (40)
, increasing evidence indicates that EMT enhances the progression of the epithelial cell’s gene expression program toward a mesenchymal phenotype and is involved in the progression of primary tumors toward metastases (24
, 41)
. Repression of E-cad by transcriptional regulators emerges as one critical step driving EMT and loss of classical cadherins is frequently associated with a cadherin-switch leading to overexpression of mesenchymal cadherins (10)
. The abnormal (i.e., either absent or cytoplasmic) expression of E-cad within tumoral areas of the three invasive HCCs containing T-cad positive tumor cells reflects a nonfunctional protein. Moreover, in these HCCs, the higher overall proportion of E-cad negative cells relative to T-cad-positive cells suggests that changes in E-cad expression occurring during initiation of EMT precede T-cad expression, which would result in a cadherin-switch taking place later in the EMT process and highlight T-cad as a mesenchymal cadherin as schematized in Fig. 8
. The lack of vimentin expression in tumor cells from primary HCCs was not surprising since vimentin expression by tumor cells has been rarely reported in human primary carcinoma but is rather associated with metastasis (42
, 43)
. Vimentin expression has been mostly observed in vitro whereby long term action of exogenous stimuli generate "complete EMT" leading to a fibroblastoid, migratory cell phenotype (41)
, as observed herein for Mahlavu. We propose (Fig. 8)
that GPI-anchored T-cad expression arising from cadherin-switching in tumor cells having undergone an EMT might influence cell motility and tumor malignancy by affecting signal transduction, as recently evidenced for other mesenchymal cadherins (9
, 44)
.
|
In conclusion, this study provides the first evidence that T-cad might represent a mesenchymal cadherin since it is expressed not only in hepatovascular EC but also in tumor cells that display EMT features. The observation in three of eight invasive HCCs of T-cad expression in tumor cells exhibiting nonfunctional E-cad is unprecedented. We suggest that T-cad expression, after the loss of E-cad, in tumor cells in vivo may correspond to a cadherin switch which can occur during EMT. Furthermore, the demonstration that T-cad is involved in invasion/migration of tumor cells suggests it plays an important role in tumor invasion and metastasis of HCC. T-cad might therefore represent a new marker for invasive tumor cells primed to intravasation.
|
ACKNOWLEDGMENTS |
|---|
Received for publication March 20, 2006. Accepted for publication June 23, 2006.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  M. Philippova, D. Ivanov, M. B. Joshi, E. Kyriakakis, K. Rupp, T. Afonyushkin, V. Bochkov, P. Erne, and T. J. Resink Identification of Proteins Associating with Glycosylphosphatidylinositol- Anchored T-Cadherin on the Surface of Vascular Endothelial Cells: Role for Grp78/BiP in T-Cadherin-Dependent Cell Survival Mol. Cell. Biol., June 15, 2008; 28(12): 4004 - 4017. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Y. Suehiro, T. Okada, T. Okada, K. Anno, N. Okayama, K. Ueno, M. Hiura, M. Nakamura, T. Kondo, A. Oga, et al. Aneuploidy Predicts Outcome in Patients with Endometrial Carcinoma and Is Related to Lack of CDH13 Hypermethylation Clin. Cancer Res., June 1, 2008; 14(11): 3354 - 3361. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. J. Wheelock, Y. Shintani, M. Maeda, Y. Fukumoto, and K. R. Johnson Cadherin switching J. Cell Sci., March 15, 2008; 121(6): 727 - 735. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. B. Joshi, D. Ivanov, M. Philippova, P. Erne, and T. J. Resink Integrin-linked kinase is an essential mediator for T-cadherin-dependent signaling via Akt and GSK3{beta} in endothelial cells FASEB J, October 1, 2007; 21(12): 3083 - 3095. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||
