Introduction
Formation of tumour cell aggregation/emboli in the circulation prolongs the survival of tumour cells and allows their physical trapping in the micro-vasculature and contributes to cancer cell hematogenous dissemination [
1,
2]. The formation of tumour emboli is heavily regulated by the expression, availability and activity of the cell surface adhesion molecules.
Galectin-3 is a multi-functional galactoside-binding protein that is expressed by many types of human cells. It is found inside and outside of the cells as well as in the circulation. Intracellular galectin-3 is an apoptosis inhibitor [
3] and mRNA splicing promoter [
4] whilst cell surface-associated extracellular galectin-3 acts as an adhesion molecule in cell-cell interactions [
5,
6] and promotes cancer progression and metastasis [
7,
8]. The concentration of free circulating galectin-3 is markedly increased in the sera of patients with breast, colorectal, lung [
9], head and neck [
10] cancers and melanoma [
11]. Patients with metastatic disease are seen to have higher concentrations of circulating galectin-3 than those with localized tumours. Recently, we have shown that the transmembrane mucin protein MUC1 is an endogenous ligand of galectin-3 in human colon cancer cells and that the interaction between MUC1 and galectin-3 occurs via binding of galectin-3 to the oncofetal Thomsen-Friedenreich carbohydrate (Galβ1,3GalNAcα-, T or TF) antigen on MUC1 [
12].
MUC1 is a large and heavily glycosylated transmembrane mucin protein that is expressed on the apical surface of normal secretory epithelia [
13]. In epithelial cancer cells, MUC1 is over-expressed [
14] and aberrantly glycosylated with short oligosaccharides such as GalNAcα- (Tn), sialylated GalNAcα- (sialyl-Tn) and TF antigen [
15,
16]. On cancer cells, MUC1 also loses its apical polarization and becomes expressed over the entire cell surface [
17,
18]. TF antigen is covered in normal epithelium by extensive glycosylation, sulphation and/or sialylation but expressed in unsubstituted form by most human cancer cells [
19,
20]. The increased expression of MUC1 and the increased occurrence of TF antigen are both associated, independently, with high metastatic potential of the cancer cells and poor prognosis of the patients [
21,
22]. MUC1 is an extremely elongated molecule which protrudes over 10 times further from the cell surface than the typical cell surface adhesion molecules [
14] and therefore influences cell adhesion when is present at high density at the cell surface [
23]. Thus, over-expression of MUC1 promotes tumour cell release from primary tumour sites by inhibiting E-cadherin-mediated cell-cell and integrin-mediated cancer-extracellular matrix interactions [
18,
24].
We have previously shown that the interaction between cell surface MUC1 and galectin-3 at concentrations similar to those found in the sera of cancer patients increases cancer cell heterotypic adhesion to endothelium as a result of MUC1 cell surface polarization which leads to exposure of heterotypic cell-cell adhesion molecules that are otherwise concealed by elongated structure of MUC1 [
12]. As change of MUC1 cell surface localization in response to galectin-3 binding may also expose the adhesion molecules that are essential to homotypic cancer cell interactions, we hypothesised that an increased interaction between circulating galectin-3 and cancer-associated MUC1 expressed on the surface of circulating tumour cells in cancer patients may promote the formation of cancer cell aggregates/emboli thus prolongs the survival of disseminated tumour cells in the circulation and contributes to cancer cell haematogenous dissemination.
We provide evidence in this study showing that the interaction between cell surface MUC1 and recombinant galectin-3 at pathologically-relevant circulating galectin-3 concentrations increases homotypic aggregation of human colon and breast cancer cells as a result of cell surface clustering of MUC1 and consequent exposure of the homotypic cell adhesion molecules including E-cadherin. The galectin-3-MUC-induced cell aggregation increases the survival of the cancer cells by preventing initiation cellular anoikis (suspension-induced apoptosis).
Discussion
This study shows that the presence of galectin-3 at pathologically-relevant circulating galectin-3 concentrations increases cancer cell homotypic aggregation by interaction with cancer-associated MUC1 that expresses the TF disaccharide. The interaction between galectin-3 and MUC1 causes MUC1 cell surface clustering and exposes smaller cell surface adhesion molecules including E-cadherin. The galectin-3-MUC1-induced cell aggregation enhances the survival of the cells under anchorage-independent conditions by preventing initiation of cellular anoikis. These discoveries have important implications for our understanding of the molecular mechanisms that underlie cancer metastasis.
Survival of tumour cells in the blood/lymphatic circulation is crucial in cancer cell metastatic spread [
1,
2]. Prolonging the survival of disseminating tumour cells in the circulation directly increases metastasis [
28]. Cells in aggregated form have been shown to have a much higher survival rate in the circulation than single cells [
29,
30]. Thus, an increase of tumour cell aggregation as a result of the increased interaction between circulating galectin-3 and cancer-associated MUC1 in cancer patients provides a survival advantage to the disseminating tumour cells in the circulation. Our results have shown that the enhanced survival of the tumour cells under anchorage-independent conditions by the galectin-3-MUC1 interaction is associated with an enhanced ability of the cell aggregates to escape initiation of anoikis. Anoikis is a specific type of apoptotic process induced by loss of cell adhesion or inadequate cell-matrix interactions [
31] and has been proposed to be the dominant mechanism in removing disseminating tumour cells from the circulation [
32]. Resistance to anoikis is considered to be a hallmark of metastatic cancer cells [
33]. As survival of the tumour cells in the circulation and being eventually able to produce metastasis at remote site occur only in less than 1 in 10,000 of the invaded cells [
29,
34], the enhanced survival of the disseminating tumour cells by the galectin-3-MUC1 interaction may have profound consequences on the metastatic potential of the cancer cells. The discovery by Topal
et al[
35] that intraportal injection of aggregated DHD/K12/TRb colon cancer cells into syngeneic BD IX rats produces over 4-times more liver metastases than injection with the same numbers of single cells is in consistence with this conclusion.
Furthermore, mechanical as well as "seed-soil" compatibility factors are both believed to contribute to the ability of specific types of cancer cells to spread to various target organs [
1,
36]. It is thought that the initial docking of specific cancer cells to the target organ is predominantly controlled by mechanical factors and that this is then followed by regulation of specific cell adhesion molecules on the cell surface. Thus, an increase of tumour cell homotypic aggregation as a result of the increased interaction between circulating galectin-3 and cancer-associated MUC1 in the bloodstream of cancer patients is likely to enhance physical trapping of the circulating tumour cells in the microvasculature at target organ and this also enhances metastasis.
It has been shown previously that cell surface-associated galectin-3 acts as a cell adhesion molecule and increases cancer cell homotypic aggregation by interaction with TF antigen expressed by unknown cell surface molecules on adjacent cells under anchorage independent conditions [
6,
37]. As cell surface-associated galectin-3 is relatively small in size, it is likely that such interaction may occur only after MUC1 cell surface polarization and exposure of the cell surface-associated galectin-3. It should be noted that clustering of the galectin cell surface receptors in response to galectin binding is probably not rare and has been shown to occur in some other cells types such as T cells [
38,
39]. Although the mechanism of galectin-ligand clustering remains unclear, clustering of galectin-3 ligands has been shown to enhance its binding affinity by as much as 10,000-fold [
40].
It should be noted that HT29-5F7 cells also express mRNAs for secreted MUC5AC and MUC5B mucins (with MUC5B the most highly expressed) and transmembrane MUC1 and MUC3 mucins (with MUC1 the most highly expressed) [
25]. Although it is unknown whether the transmembrane MUC3 also carries the TF structure and a similar mechanism of interaction between galectin-3 and MUC3 exists, the contribution of such an interaction to galectin-3-mediated cell aggregation in those cells is likely to be much less than that of the galectin-3-MUC1 interaction as the level of MUC1 expression in these cells is much greater than for MUC3.
Thus, the interaction between circulating galectin-3 and TF-expressing MUC1 on the surface of disseminating cancer cells promotes cell aggregation and embolus formation and enhances survival of disseminating tumour cells in the circulation. This, together with our earlier reports showing an enhanced cancer cell-endothelial adhesion also resulting from the galectin-3-MUC1 interaction [
12,
27], indicates that the increased circulation of galectin-3 in the bloodstream of cancer patients promotes several important steps of the metastatic cascade. The reduction of metastasis-associated survival of nude mice intravenously injected with galectin-3-pretreated than untreated MUC1-expressing human melanoma cells observed in our earlier study [
27] likely represents the consequence of galectin-3-MUC1 interaction on cancer cell heterotypic adhesion to endothelium as well as on cancer cell homotypic aggregation. This study provides new information into our understanding of the molecular mechanisms of cancer cell haematogenous dissemination and further highlights the functional importance of change of the cell surface glycosylation in cancer progression. It also implies that targeting the interaction of circulating galectin-3 with MUC1 in the circulation may represent an effective therapeutic approach for preventing metastasis.
Materials and methods
Materials
Recombinant full-length human galectin-3 and monoclonal antibody (mAb) against E-cadherin (MAB1838) were from R&D Systems (Abingdon, UK). B27.29 anti-MUC1 mAb was kindly provided by Dr. Mark Reddish (Biomira Inc, Canada). CT-2 anti-MUC1 cytoplasmic tail antibody was from NeoMarkers (Fremont CA). Biotin-conjugated peanut agglutinin (PNA) was purchased from Vector Laboratories Ltd (Peterborough, UK). Streptococcus pneumoniae Endo-N-acetyl-galactosaminidase (EC 3.2.1.97), O-glycanase, was obtained from Prozyme Inc (Oxford, UK). The Non-Enzymatic Cell Dissociation Solution (NECDS) was from Sigma. The Vybrant DiO and DiI Cell-labelling Solutions were from Molecular Probes (Cambridge, UK).
Cell lines
Human colon cancer HT29 and HT29-5F7 cells were cultured as previously described [
12]. Human colon cancer HT29-5F7 cells were kindly provided by Dr. Thecla Lesuffleur (INSERM U560, Lille, France). MUC1 transfection of HBL-100 human breast epithelial cells with full length cDNA encoding MUC1 resulted in the MUC1 positive transfectants HCA1.7+ and the subsequent bulk selection of the MUC1 negative revertants HCA1.7- were as described previously [
18].
Cell aggregation
Subconfluent cells were released from the culture plates with NECDS (which releases the cells while keeping the cell membrane proteins intact) and were dispersed in serum-free DMEM containing 0.5 mg/ml bovine serum albumin (BSA). Two 0.5 ml (1×10
6 cells) aliquots of the cell suspension were incubated, separately, with 5 μl/ml DIO and DiI Cell Labelling Solution for 30 min at 37°C. The DIO- and DiI-labelled cells were then mixed in the presence or absence of recombinant galectin-3, antibodies or lactose in DMEM in a rotating incubator (Julabo lab GMBH, Seelbach, Gemerney) at 100 rpm/min at 37°C for 1 hr followed by analysis with flow cytometry (Becton-Dickinson FACSVantage SE). Cells labelled only with DiI or DiO were used to identify the position of DiO- (upper-left panel) and DiI-(bottom-right panel) labelled cell populations in the bivariate correlation plot and cell populations containing both DiO and DiI fluorescence (top-right panel) in the correlation plot were defined as cell aggregates (Fig
1A).
This method measures cell aggregates containing both DiO- and DiI-labelled cells and does not include cell aggregates formed only by DIO- or DiI-labelled cells. Although this may underestimate the total cell aggregates, it provides a more accurate assessment of the effect of exogenous galectin-3 by reducing the variations hence the experimental errors due to the differences of initial existing numbers of cell aggregates in each individual assessment.
Reduction of TF expression by O-glycanase treatment
HCA1.7+ cells (105 cells) were incubated with 0.02U/ml O-glycanase for 2 hr at 37°C. The cells were either lysed directly with SDS-sample buffer for blotting with PNA or labelled with DiO and DiI and incubated with recombinant galectin-3 for 1 hr at 37°C before the cell aggregation assessments.
Electrophoresis and lectin/immunoblotting
SDS-PAGE (4% running gel and 3.5% stack gel for MUC1 blotting and 10% running gel and 4% stack gel for E-cadherin blotting) electrophoresis and immune/lectin blotting were performed as previously described [
12].
MUC1 cell surface localization
Cells released with NECDS were labelled, separately, with DiO and DiI as described above. Two equal cell populations (5×105) of the labelled cells were mixed in the presence or absence of 1 μg/ml galectin-3, B27.29 antibody or BSA for 1 hr at 37°C. The cell suspensions were applied to poly-lysine coated slides, fixed with 2% paraformaldehyde for 15 min, blocked with 5% normal goat serum for 0.5 hr, probed with B27.29 anti-MUC1 antibody (1:2,000 dilution in 1% goat serum/PBS) and followed by fluorescent-labelled secondary antibody (1:400 dilution in 1% goat serum/PBS). The slides were blinded and the number of cells lacking a continuous rim of MUC1 or the number of small or large aggregates in 500 adjacent cells in randomly selected low power fields was counted using an Olympus B51 fluorescent microscope with 20× objective.
In some experiments, HT29 or HT29-5F7 cells released by NECDS were fixed with 2% paraformaldehyde for 15 min, blocked with 5% normal goat serum for 0.5 hr, probed with B27.29 anti-MUC1 antibody (1:2,000 dilution in 1% goat serum/PBS) and followed by analysis of the MUC1 cell surface expression with flow cytometry.
In some other experiments, HT29-5F7 cells released by NECDS were treated with galectin-3 for 1 hr at 37°C before introduced and cultured in poly-HAMA-coated plates for 48 hr. The cells were then fixed with 2% paraformaldehyde for 15 min, blocked with 5% normal goat serum for 0.5 hr, probed with B27.29 anti-MUC1 antibody (1:2,000 dilution in 1% goat serum/PBS) and the MUC1 cell surface localization was observed under fluorescent microscope.
E-cadherin cell surface expression
Cells released with NECDS were fixed 2% paraformaldehyde for 15 min, blocked with 5% normal goat serum for 0.5 hr, incubated with anti-E-cadherin antibody (1:1000 dilution in 1% goat serum/PBS) or isotype control immunoglobulin and followed with fluorescent-conjugated secondary antibodies (1:400 dilution in 1% goat serum/PBS). The cell surface expression of E-cadherin was analysed by flow cytometry.
siRNA E-cadherin transfection
HT29-5F7 cells were treated with commercially available siRNA constructs (100 nM) against E-cadherin or scrambled control non-targeting siRNA (siCONTROL non-targeting siRNA, Dhamacon) for 48 hr at 37°C. The cells were either lysed and the expressions of E-cadherin was analysed by immunoblotting or the cells were labelled with DiO and DiI followed by the assessment of cell aggregation in the presence or absence of 1 μg/ml galectin-3 as described above.
In some experiments, the HT29 cells were seeded in 24-well plates with glass coverslips inserted for 18 hr at 37°C before treatment of the cells without or with E-cadherin siRNA or non-targeting control siRNA for 48 to 72 hr. The cells were then fixed with 2% paraformaldehyde for 15 min, blocked with 5% normal goat serum for 0.5 hr, incubated with anti-E-cadherin antibody (1:1000 dilution in 1% goat serum/PBS) for 1 hr. After application of fluorescent-conjugated secondary antibodies (1:400 dilution in 1% goat serum/PBS), the cell surface expression of E-cadherin was observed under fluorescent microscope.
Cell viability and anoikis in strained and aggregated cells under anchorage-independent condition
The cells were cultured in 96 or 24 well plates that were coated twice with 200 μl/well (96-well plates) or 5 ml/well (6-well plates) of 10 mg/ml poly-2-hydroxyethyl methacrylate (poly-HEMA, Sigma) in 95% ethanol over night at room temperature. After washing, the cells were treated with 1 μg/ml recombinant galectin-3 or BSA under suspension culture in serum-free medium for 6 to 48 hr at 37°C. The cells in suspension were collected and passed three times through a 40-μm cell strainer (BD Biosciences). The viability of the cells that did (strained) and did not (aggregates) pass through the strainers was both measured by the CellTiter-Glo® Luminescent Cell Viability Assay (Promega).
For the measurement of cellular anoikis, after separation of the cells by cell strainers, the strained cells and cell aggregates were resuspended in 0.05% trypsin in PBS and the apoptotic cells were measured by FITC-Annexin-V cell surface binding with a FITC-Annexin-V Apoptosis Detection Kit (Cambridge Biosciences, Cambridge, UK) by flow cytometry as the manufacture's instructions.
Statistical analysis
The statistical analyses were performed using the unpaired t test for single comparison, one-way analysis of variance (ANOVA) followed by Newman and Keuls test for multiple comparisons or Chi-Square test (StatsDirect for Windows, StatsDirect Ltd; Sale, UK) where appropriate. Differences were considered significant when p < 0.05.
Acknowledgements
The authors thank Dr. Thecla Lesuffleur (INSERM U560, Lille, France) for the HT29-5F7 cells and Dr. Mark Reddish (Biomira Inc, Edmonton, Canada) for the mAb B27.29. This work was supported by Cancer Research UK grant C7596 (to LGY) and North West Cancer Research Fund grant CR777 (to LGY).
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
QZ contributed to acquisition of data and experiment design. MB contributed to data analysis by flow cytometry. JH contributed to the generation of the HCA1.7+/- cells and interpretation of data. XG and HB participated in some experiments. JMR contributed to data analysis and assisted in drafting the manuscript. LGY contributed to conception, design and coordination of the study and drafting the manuscript. All authors read and approved the final manuscript.