Invasion is a complex process controlled by secretion and activation of proteases, alteration of integrin levels and GSL (glycosphingolipid) patterns. Differential organization of GSLs with specific membrane proteins and signal transducers in GEMs (GSL-enriched microdomains), initiates signalling events to modify cellular phenotype. Although the GSL monosialyl-Gb5 has been linked with invasion, its functional role in invasion is poorly described and understood. To investigate this problem, we induced the invasion of human breast cancer cells and subsequently explored the underlying mechanism. In the present study, the invasion of human MCF-7 breast cancer cells is highly dependent on clustering of monosialyl-Gb5, and the subsequent activation of monosialyl-Gb5-associated focal adhesion kinase and cSrc in GEM leading to the downstream activation of extracellular-signal-regulated kinase (ERK). As a result, we observed increased expression levels and activity of matrix metalloproteinases-2 and -9, which correlated with decreased expression of integrins α1 and β1. Together these results suggest that the organization of crucial molecules in GEMs of MCF-7 cells is critical for their invasive properties.
The majority of all cancer-related death arises from the metastatic spread of primary tumours. The mechanisms for the acquisition of invasive and metastatic potential of tumour cells are, however, not well understood at either the molecular or the biochemical level. Investigating these mechanisms is therefore one of the main challenges for exploratory and applied cancer research.
A number of complex invasion-associated cellular activities have been identified and characterized, including alterations in expression levels of adhesion molecules and proteolytic degradation of the extracellular matrix, along with changes in expression or activities of a variety of cellular proteins in multiple branched signalling pathways . In addition, previous studies  have highlighted the importance of aberrant glycosylation as a key event in the induction of invasion and metastasis. Altered glycosylation of cell-surface glycoproteins  and GSLs (glycosphingolipids) , which have been identified as tumour-associated antigens, are often found in malignant cells. Of particular importance is that GSLs can cluster and assemble with membrane proteins and signal transducers to form GEMs (GSL-enriched microdomains), where they are able to initiate signal transduction , and thus profoundly affect the cellular activities associated with tumour cell invasion. Roles for GSLs in invasion were demonstrated recently for GM3 (monosialoganglioside 3) and GM1 [6,7].
MSGb5 (monosialyl-Gb5), also known as SSEA-4 (stage-specific embryonic antigen-4) , is a GSL found in GEMs that are maximally expressed in human renal cell carcinomas and correlates with metastasis [9,10]. Furthermore, previous work  has shown that stimulation of MSGb5 by the mAb (monoclonal antibody) RM1 accounts for the invasive properties of MCF-7 human breast carcinoma cells. The molecular mechanisms regulating synthesis of MSGb5 have been studied previously ; however, no clear functional role of MSGb5 in the invasive and metastatic behaviour of tumour cells has been demonstrated.
In the present study, we have induced the invasion of MCF-7/AZ breast cancer cells by the ether lipid ET-18-OMe (1-O-octadecyl-2-O-methyl-glycerophosphocholine)  and analysed changes in organization of MSGb5, integrins and signal transducers. Subsequently, we identified the signalling pathway that leads to the activation of MMPs (matrix metalloproteinases). The findings of the present study may help to explain the mechanism by which GEM organization and composition results in the invasion of MCF-7 breast cancer cells.
MATERIALS AND METHODS
Abs raised against GSLs that were used in this study were: mouse IgM MBr1 against globo-H, mouse IgM 1A4 (E10) against Gb3, mouse IgG3 DH2 against GM3, mouse IgM MK1-8 against GM2, mouse IgM 9G7 against Gb4, mouse IgM SSEA-3 against Gb5 , mouse IgM RM1 against MSGb5  and mouse IgM 5F3 against disialyl-Gb5 . The mouse IgG3 against MSGb5 was generated by Dr D. Solter (Department of Developmental Biology, Max-Planck Institute of Immunobiology, Freiburg, Germany) and was obtained from the Development Studies Hybridoma Bank [developed under the auspices of the NICHD (National Institute of Child Health and Human Development) and maintained by the Department of Biological Sciences, University of Iowa, Iowa City, IA, U.S.A.]. Abs directed against integrins α1, α2 and β1, and anti-paxillin, anti-(phospho-paxillin), anti-p130Cas and anti-(phospho-p130Cas) were from Chemicon. Mouse anti-FAK (focal adhesion kinase) mAb and mouse anti-cSrc mAb were from Transduction Laboratories. Rabbit anti-(phospho-FAK) Abs (against phosphorylated residues Tyr398, Tyr577, Tyr861 and Tyr925) and anti-(phospho-cSrc) (Tyr416) were from Biosource. Rabbit anti-ERK (extracellular-signal-regulated kinase) and anti-(phospho-ERK) (Thr202/Tyr204) were obtained from Cell Signaling Technologies. Abs against MMP-2 and MMP-9 were from Sigma. Biotinylated anti-rabbit and anti-mouse, and FITC-labelled anti-mouse secondary Abs were from Vector Laboratories.
MCF-7/AZ and MCF-7/6 cells are variants of the human mammary carcinoma cell family MCF-7. The cells were maintained on tissue culture plastic substrate (Nunc) in a 50:50 (v/v) mixture of DMEM (Dulbecco's modified Eagle's medium) and Ham's F12 (Invitrogen) containing 250 units/ml penicillin, 100 μg/ml streptomycin (Invitrogen) and 10% (v/v) fetal bovine serum (Invitrogen) at 37 °C in a humidified 10% CO2/90% air atmosphere. Both cell variants have been studied in several invasion assays in vitro. MCF-7/AZ and MCF-7/6 cells are not invasive into collagen type I , whereas MCF-7/6, but not MCF-7/AZ cells, are invasive when tested in precultured heart fragments .
ET-18-OMe (clinical grade) was provided kindly by Dr P. Hilgard (ASTA MEDICA, Frankfurt am Main, Germany). Cells were cultured in the presence of 15 μg ET-18-OMe per ml of culture medium for the indicated times. Drug toxicity was evaluated through measurement of mitochondrial dehydrogenase activities with MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium bromide] reagent (Sigma) . PP1 (4-amino-1-tert-butyl-3-(1′-naphthyl)pyrazolo[3,4-d]pyrimidine), a cSrc kinase inhibitor, was from Calbiochem and was used at a final concentration of 20 μM. Micro-BCA (bicinchoninic acid) and BCA protein assay reagent kit were from Pierce. Vectastain® ABC-AmP™ kit was from Vector Laboratories.
Collagen type I invasion
Collagen type I invasion was determined as described previously . Briefly, six-well plates were filled with 1.25 ml neutralized collagen type I (0.09%; Upstate Biotechnology) and incubated for 1 h at 37 °C to allow gelification. Single-cell suspensions were prepared with trypsin/EDTA, mixed with or without ET-18-OMe and PP1, seeded on top of collagen type I gel and cultured at 37 °C for 24 h. The number of cells penetrating into the gel or remaining at the surface was counted in 12 fields of 0.157 mm2, using an inverted microscope controlled by a computer program (Dr L. Vakaet Jr, McForth, Creative Solutions, Rockville, MD, U.S.A.). The invasion index expresses the percentage of invading cells over the total number of cells.
GSL composition in GEMs and soluble fractions by TLC immunostaining
The composition of GSL in GEMs and in the soluble fraction of both MCF-7 cell lines after treatment with ET-18-OMe for 1 h was compared with untreated cells. The procedure was performed as described previously . Briefly, total GSLs were separated from phospholipids and other lipids by the acetylation procedure described previously . The GSLs were purified further by HPTLC (high-performance-TLC) in isopropanol/hexane/water (55:25:20 by vol.) system and identified with TLC immunostaining by specific Abs. The intensity of immunoblotted bands was quantified by densitometry using the Scion Image program (Scion).
Cell lysates were made from cells grown to 70% confluence. The cells were treated for the indicated times with ET-18-OMe and PP1, washed three times with PBS and subsequently lysed in 0.5 ml lysis buffer [PBS containing 1% (v/v) Triton X-100, 1% (v/v) Nonidet P40 and the following protease inhibitors: 10 μg/ml aprotinin, 10 μg/ml leupeptin, 1.72 mM PMSF, 100 mM NaF, 500 μM sodium metavanadate and 500 μg/ml sodium pyrophosphate]. Aliquots of lysates containing equal amounts of protein were boiled for 5 min in SDS/PAGE sample buffer [0.5 M Tris/HCl, pH 6.8, 5% (w/v) SDS, 20% (v/v)glycerol, 0.01% Bromophenol Blue and 5% (v/v) 2-mercaptoethanol], resolved by SDS/PAGE (7.5% gels) and transferred on to PVDF membranes (Bio-Rad Laboratories). After transfer, membranes were incubated with relevant Abs against FAK, phospho-FAK, cSrc, phospho-cSrc, ERK, phospho-ERK, MMP-2, MMP-9, integrins α1, β1, α2, p130Cas, phospho-p130Cas, paxillin and phospho-paxillin, followed by incubation with a secondary biotinylated Ab (1:1000) and developed by enhanced chemiluminescence, using the Vectastain® ABC-AmP™ detection kit. Images of the membranes were created and analysed using the BioChemi System and analysis software (UVP).
Integrins and signal transducers in GEMs and soluble fractions
Cells were treated for the indicated times with ET-18-OMe and postnuclear fractions were prepared in TNE (25 mM Tris/HCl, pH 7.5, 150 mM NaCl, 5 mM EDTA) supplemented with10 μg/ml aprotinin and 1.72 mM PMSF and 1% (v/v) Triton X-100, followed by sucrose-gradient density centrifugation to isolate 12 fractions [0.5 ml sample mixed with 0.5 ml 85% (w/v) sucrose solution, followed by 5.5 ml 35% (w/v) sucrose solution and 4.2 ml 5% (w/v) sucrose solution. Ultracentrifuge (Beckman) with swinging buckets (SW40 Ti) at 39000 rev./min, 4 °C for 17h], which included GEMs found in fractions 4–6, as described previously [22,23]. Integrins (α1, β1 and α2) and signal transducers (FAK, phospho-FAK, cSrc, phospho-cSrc, ERK and phospho-ERK) present in the GEM fractions and in the high-density soluble fractions (10–12) were identified further by SDS/PAGE (7.5% gels), followed by Western blotting as described in the Western blotting section of the Materials and methods section.
Immunoprecipitation of GSLs and signal transducers
Association of GSLs (Gb3, Gb5, MSGb5 or GM2) with signal transducers, FAK and cSrc, was analysed by immunoprecipitation. When cell cultures reached 70% confluence they were treated for the indicated times with ET-18-OMe and then lysed as described in the Western blotting section. Lysates containing 800 μg of protein, were mixed with 50 μl of packed Protein G–Sepharose beads (Amersham Biosciences) and placed on a rotator for 30 min at 4 °C, to preclear non-specific-binding proteins from the lysate. After centrifugation at approx. 70 g, the supernatant was collected and 2 μl of relevant Ab was added to the sample, which was then rotated at 4 °C overnight. After incubation Protein G–Sepharose beads were added, rotated at 4 °C for 2 h and centrifuged to collect the beads. The beads were then washed twice with lysis buffer and collected by centrifugation. SDS/PAGE sample buffer (100 μl) was added to the immunoprecipitates and then heated to 95 °C for 5 min. The immunoprecipitates were subjected to SDS/PAGE (7.5% gels), transferred electrophoretically on to PVDF membranes, incubated with relevant Abs and analysed and imaged as described in the Western blotting section.
Cells were suspended in trypsin/EDTA and stained using indirect immunofluorescence. Briefly, cells were incubated on a rotator at 4 °C for 1 h with IgG3 Ab against MSGb5, washed three times with CMF-HBSS (calcium- and magnesium-free Hanks balanced salt solution) and fixed with 3% (w/v) paraformaldehyde in CMF-HBSS for 20 min. Fixed cells were washed three times with CMF-HBSS and incubated with FITC-labelled anti-mouse secondary Ab. Stained cells were mounted with a drop of Glycergel mounting medium (Dako) containing 1% (w/v) 1,4-diazabicyclo[2.2.2]octane (fluorescence stabilizer). Fluorescence was observed using an Olympus IX51 microscope with an attached Olympus U-CMAD3 camera.
Cells were grown on 12-mm-diameter glass coverslips placed in the wells of a 24-well plate. The glass coverslips with cells attached were removed, washed three times with ice-cold PBS and fixed with 3% (w/v) paraformaldehyde in PBS. Fixed cells were washed three times with PBS, incubated with Abs against MSGb5, Gb3, Gb5 or GM2 and mounted as described above. Confocal laser scanning was with a Leica DM IRBE fluorescence microscope and a Leica TCS NT confocal unit.
In-gel gelatinase assay
Gelatin zymography was performed as described . Subconfluent cells were treated with ET-18-OMe and PP1 in DMEM/Ham's F12 medium for 0, 1, 2, 4, 6, 8, 12 and 18 h at 37 °C. The medium was collected, clarified by centrifugation, resolved in non-reducing gels containing 2 mg/ml gelatin and processed for activity by observing zones of gelatin degradation.
All treatments were performed at least three times. The Student's t-test (95%) was used for statistical evaluation. Levels were quantified using Scion Image statistical software.
Invasiveness of MCF-7/AZ but not of MCF-7/6 cells
The two human breast cancer cell variants, MCF-7/AZ and MCF-7/6, are non-invasive into a collagen type I gel-layer (Figure 1). ET-18-OMe treatment, however, increased the invasiveness of MCF-7/AZ cells, while the same treatment had no effect on MCF-7/6 cells (consistent with previous observations ). Pre-treatment with PP1 blocked the enhanced invasiveness of MCF-7/AZ cells (Figure 1).
Invasiveness of MCF-7/AZ and MCF-7/6 cells
Expression levels of activated cSrc, FAK and ERK in invasive MCF-7/AZ cells compared with non-invasive MCF-7/6 cells
The invasion-promoting effect of ET-18-OMe on MCF-7/AZ cells suggests that ET-18-OMe initiates cSrc-mediated signalling in MCF-7/AZ cells, but not in the variant MCF-7/6 cells (Figure 1). Therefore the effect of ET-18-OMe on phosphorylation of Tyr397 of FAK, the autophosphorylation site of FAK, and Tyr416 of cSrc kinase was examined. Expression levels of FAK and cSrc in both cell lines were unchanged, but kinase activity of cSrc (Figures 2A, left-hand panel and 2B) and FAK (Figures 2A, left-hand panel and 2C) were greatly enhanced in MCF-7/AZ cells 5–10 min after treatment. There was no such activation of cSrc and FAK in MCF-7/6 cells (Figure 2A, right-hand panel). The use of cSrc kinase inhibitor, PP1, blocked activation of cSrc but not Tyr397 phosphorylation on FAK, suggesting that the autophosphorylation of FAK promotes activation of cSrc. Next, the cSrc-dependent tyrosine phosphorylation sites on FAK (Tyr576, Tyr861 and Tyr925) were assayed. Time-dependent phosphorylation of FAK on Tyr925 (Figures 2A, left-hand panel and 2D) but not Tyr576 and Tyr861 (results not shown) was observed in ET-18-OMe-treated MCF-7/AZ cells, whereas no altered phosphorylation of FAK on Tyr925 (Figure 2A, right-hand panel) and no phosphorylation on Tyr576 and Tyr861 was detected in MCF-7/6 cells (results not shown). We addressed whether inhibition of cSrc prevented phosphorylation of FAK on Tyr925 further. Treatment with ET-18-OMe and PP1 reduced Tyr925 phosphorylation in MCF-7/AZ cells (Figures 2A, left-hand panel and 2D). These results suggest that the MAPK (mitogen-activated protein kinase)-pathway might be involved, since phosphorylation of FAK on Tyr925 initiates downstream signalling resulting in activation of ERK (by phosphorylation of Thr202 and Tyr204) [25,26]. Treatment with ET-18-OMe resulted in enhanced ERK activity in MCF-7/AZ cells after 10 min but not in MCF-7/6 cells (Figures 2A and 2E). p130Cas and paxillin are known substrates of cSrc that are associated with phosphorylation of FAK on Tyr861; in order to confirm the lack of phosphorylation on FAK Tyr861 and Tyr576, we determined the kinase activity of p130Cas and paxillin. Treatment with ET-18-OMe had no effect on phosphorylation of p130Cas or paxillin in both MCF-7/AZ and MCF-7/6 cells (results not shown).
Expression levels of activated cSrc (phospho-Tyr416), FAK (phospho-Tyr397/Tyr925) and ERK (phospho-Thr202/Tyr204) in whole-cell lysates from MCF-7/AZ and MCF-7/6 cells
Expression levels of activated cSrc, FAK and ERK in GEMs of invasive MCF-7/AZ and non-invasive MCF-7/6 cells
FAK, cSrc and ERK are associated with GEMs (low-density fractions 4–6) and the soluble high-density fractions (10–12) in both MCF-7/AZ (Figure 3A) and MCF-7/6 cells (results not shown). In both cell-line variants, expression levels of FAK, cSrc and ERK were unchanged after ET-18-OMe treatment (Figure 3A). Activity levels of cSrc, FAK and ERK were found in GEM as well as in the non-GEM fractions and phosphorylation levels of FAK, cSrc and ERK were enhanced after 15 min and decreased after 30 min in MCF-7/AZ cells (Figures 3B, 3C and 3D). PP1 blocked activation of cSrc and, subsequently, blocked activation of downstream ERK (Figures 3B and 3D). Activity levels in MCF-7/6 cells were essentially the same (results not shown).
Expression levels of activated cSrc (phospho-Tyr416), FAK (phospho-Tyr397) and ERK (phospho-Thr202/Tyr204) in GEM (fractions 4–6) and non-GEM (fractions 10–12) of MCF-7/AZ cells
GSL composition in MCF-7/AZ and MCF-7/6 cells
Major GSLs found in both MCF-7 cell lines treated with ET-18-OMe or untreated were identified as globo-series Gb3, Gb4, Gb5, MSGb5 and globo-H, and as the ganglio-series structure GM2. However, lacto-series structures were not detectable (Figure 4A). Composition of GSLs, as determined by their specific Abs, was compared for GEM fractions (4–6) and the soluble high-density fractions (10–12). Each GSL was present in fractions 4–6 but completely absent from fractions 10–12 (Figure 4B). No differences between the two cell-lines were observed in the conditions tested. The presence of Gb3 and other globo-series structures similar to Gb5 and globo-H in MCF-7 cells was reported previously . The presence of an abundance of GM3 and GM1 in MCF-7 cells has been observed previously ; however, the variants used in the present study lacked GM3 and GM1, which is in line with previous findings .
TLC of GSLs from whole-cell extracts (A) or GEMs (B) of MCF-7/AZ and MCF-7/6 cells with or without ET-18-OMe
Clustering of MSGb5 and association with cSrc and FAK in invasive MCF-7/AZ cells compared with non-invasive MCF-7/6 cells
Clustering of MSGb5
Alterations in the pattern of MSGb5 were observed by immunofluorescence and confocal microscopy (Figure 5A). MSGb5 was detected in both MCF-7/AZ and MCF-7/6 cells consistent with the above results. Fluorescence examination of MCF-7/AZ cells treated with ET-18-OMe revealed clustering of MSGb5 at the membrane 5–10 min after treatment compared with untreated MCF-7/AZ cells (Figure 5A, panels a and b). Clustering of MSGb5 in MCF-7/6 (results not shown), as well as clustering of the GSLs Gb3 and Gb5 in MCF-7/AZ cells after ET-18-OMe treatment was not observed (Figure 5A, panels c and d).
Clustering of MSGb5 and association with cSrc and FAK in MCF-7/AZ cells
Association of MSGb5 with cSrc and FAK
MSGb5 association with cSrc and FAK and phosphorylated-cSrc and -FAK was shown by immunoprecipitation. Aliquots of cell lysates were immunoprecipitated by incubation with anti-MSGb5 Ab and capture with protein G–Sepharose beads as described in the Materials and methods section. Both FAK and cSrc were detected in MSGb5 immunoprecipitates (Figure 5B, left-hand panel). In addition, kinase activity of FAK (phospho-Tyr397) and cSrc (phospho-Tyr416) was observed in the MSGb5-immunoprecipitates in MCF-7/AZ cells 5–10 min after ET-18-OMe treatment (Figure 5B), which is in line with the kinase experiments described above. No association of FAK and cSrc with Gb3, Gb5 or GM2 was observed (results not shown). In MCF-7/6 cells, cSrc and FAK were associated with MSGb5, and as expected no kinase activity was observed upon ET-18-OMe treatment (Figure 5B, right-hand panel). Equal levels of MSGb5 were detected in the immunoprecipitates of both cell lines (Figure 5B).
Activity and expression levels of MMPs in invasive MCF-7/AZ and non-invasive MCF-7/6 cells
Expression level of MMPs in MCF-7/AZ and MCF-7/6 cells
To address whether active ERK promotes MMP processing, we determined the expression levels of MMP-2, MMP-9 and MT1 (membrane type 1)-MMP. In MCF-7/AZ cells, expression of MMP-2 was greatly enhanced after 1 h and decreased 4 h after ET-18-OMe treatment (Figures 6A and 6B), whereas the level of MMP-9 increased after 1 h (Figure 6A and 6C). MT1-MMP expression was unaffected (results not shown). PP1 blocked expression of both MMP-2 and MMP-9 in MCF-7/AZ cells (Figure 6A, left-hand panel). ET-18-OMe had no effect on any of the MMPs tested in MCF-7/6 cells (Figure 6A, right-hand panel).
Activity and expression levels of MMPs in MCF-7/AZ and MCF-7/6 cells
Activity level of MMPs in MCF-7/AZ and MCF-7/6 cells
Since ET-18-OMe treatment resulted in increased expression of MMPs, we determined further the activity of MMP-2 in both cell-line variants after ET-18-OMe treatment. The results shown in Figure 6(D) revealed that MMP-2 activity increased significantly in MCF-7/AZ cells compared with MCF-7/6 cells.
Differences in integrin expression levels
Integrin expression in MCF-7/AZ compared with MCF-7/6 cells
Tumour-cell invasion includes not only secretion and activation of proteases but also alteration in integrin expression levels. To analyse the involvement of integrins, we determined the expression levels of integrins α1, α2 and β1. Expression of integrins α1 and β1 decreased over time in MCF-7/AZ cells (Figures 7A, left-hand panel, 7B and 7C), whereas no changes in expression of either integrin was observed after treatment of MCF-7/6 cells (Figure 7A, right-hand panel). In both cell lines integrin α2 was not detectable (results not shown).
Integrin expression in MCF-7/AZ and MCF-7/6 cells and expression levels of integrin α1 and β1 in GEM fractions (4–6) and non-GEM fractions (10–12) of MCF-7/AZ cells
Integrins in GEMs of MCF-7/AZ compared with MCF-7/6 cells
Additionally, we found integrins α1 and β1 associated with GEMs in both cell-lines. However, 15 min after treatment with ET-18-OMe, the two integrins were no longer found in the GEM fractions, but were present in the non-GEM fractions from MCF-7/AZ cells (Figure 7D). Moreover, immunoprecipitation revealed that both integrins were not associated with MSGb5, FAK and cSrc (results not shown). Similar treatment had no effect on the distribution of these integrins in MCF-7/6 cells (results not shown).
Tumour-cell invasion is a complex process controlled by several factors, many of which operate at the cell-surface membrane. This process includes secretion and activation of proteases , alteration of integrin receptor expression levels  and changes in GSL expression patterns [2,4]. In addition, GSLs and transmembrane glycoproteins, as well as signal transducers, are assembled in multiple types of GEMs, which are involved in GSL-dependent cell adhesion that induces the activation of signal transducers, thereby initiating signals that result in the modification of the cellular phenotype . This has been demonstrated for GM3 in mouse melanoma B16 cells [23,31] and in human bladder KK47 cancer cells , for DSGG (diasyl-GalNAcLc4) in renal cell carcinoma [16,32] and for GD3 (disialoganglioside 3) in melanoma cells .
MSGb5, described originally as SSEA/4 , is highly expressed in human renal cell carcinomas and correlates with metastatic potential [9,10]. Stimulation of MSGb5 by the mAb RM1 was suggested to be associated with invasiveness of MCF-7 human breast carcinoma cells, along with activation of FAK and cSrc . However, no other studies demonstrate a clear functional role for MSGb5 in invasion or a role in signal regulation leading to invasion.
In the present study, we have used two breast carcinoma cell variants MCF-7/AZ and MCF-7/6 that show similar composition in their GEM with regard to globo-H, Gb3, Gb4, Gb5, MSGb5 and GM2, and associated signalling molecules. Both variants are non-invasive into a collagen type I gel-layer; however, treatment with ET-18-OMe caused the phenotypic conversion of MCF-7/AZ cells from non-invasive to invasive, whereas no effect was observed in MCF-7/6 cells. This result enabled us to identify the key players in the activation of the signalling pathway that initiates invasion.
Next, enhanced activity levels of cSrc were detected in MCF-7/AZ but not in MCF-7/6 cells after treatment with ET-18-OMe. Suppression of cSrc with the tyrosine kinase inhibitor PP1, demonstrated clearly the prominent role for cSrc in invasion induced by ET-18-OMe. A probable mechanism by which cSrc promotes invasive behaviour is through formation of a transient complex with the protein tyrosine kinase, FAK [34,35]. A previous study  has described FAK association with a variety of cell surface molecules, resulting in its autophosphorylation at Tyr397 and thereby creating a high-affinity-binding site for cSrc. This specific organization has been shown to play a crucial role in cSrc-dependent phosphorylation of distinct tyrosine residues on FAK , leading to specific downstream signalling events. Our results support the formation of a FAK–cSrc complex upon FAK (Tyr397) activation and the subsequent cSrc-dependent phosphorylation of FAK on Tyr925, and the downstream phosphorylation of ERK, which is consistent with previous studies . The results of the present study appear to be similar with the observations made by Steelant et al.  and suggested the implication of GSLs, in particular MSGb5, in the initiation of downstream signalling events. Although both cell lines displayed no differences in their GSL composition upon treatment with ET-18-OMe, the possibility arose that changes in organization or clustering of their GSLs was responsible for the initiation and activation of associated signal transducers. Such an activation model was reported for disialylgalactosylgloboside in the renal cell carcinoma cell-line TOS-1  and for GM3 in B16 melanoma cells [23,31]. In the present study, we observed prompt clustering of MSGb5 in MCF-7/AZ cells and no changes in organization in MCF-7/6 cells upon treatment with ET-18-OMe. A close connection of FAK and cSrc with MSGb5 was suggested, since activated FAK and cSrc were found in GEMs of MCF-7/AZ cells, which was confirmed further by co-immunoprecipitation experiments that revealed the association of both FAK and cSrc, as well as their activated forms, with MSGb5 after ET-18-OMe treatment in MCF-7/AZ cells, whereas no association of the active forms of the proteins with MSGb5 was found in MCF-7/6 cells. Furthermore, we could exclude the involvement of other GSLs present in the GEM, since no clustering and no activation of these signal transducers were observed upon treatment with ET-18-OMe.
Another factor controlling tumour cell invasion is the secretion and activation of MMPs . Several studies [34,35] have shown that FAK–cSrc signalling complexes are associated with an invasive phenotype. In the present study, the results obtained using ET-18-OMe and PP1 support the importance of the downstream activation of ERK in the MAPK-pathway, in response to activation of cSrc and FAK (phospho-Tyr925), as being necessary for secretion and activation of MMP-2 and MMP-9. This is consistent with studies using MEK1AA (dominant negative MAPK/ERK kinase) expression or specific MEK1-inhibitor (PD98059) where secretion of MMP-2 and MMP-9 was suppressed dramatically [37,38]. Other signalling pathways that link the FAK–cSrc complex with MMPs were excluded, for example, via the activation of the distinct cSrc-dependent tyrosine residues on FAK and subsequent phosphorylation of the FAK-binding proteins p130Cas and paxillin [39,40], since no phosphorylation was observed on the cSrc-dependent tyrosine residues on FAK, and was confirmed by the lack of p130Cas and paxillin activation.
Since integrin receptors are crucial in cancer cell invasion , we searched for differences in integrin expression levels. Our results showed decreased expression levels of integrin subunits α1 and β1 after ET-18-OMe treatment of MCF-7/AZ cells compared with MCF-7/6 cells. These findings are in line with previous observations . Both receptor subunits associated initially with GEM, were undetectable in the GEMs of MCF-7/AZ cells, which suggests that they have been translocated, whereas the expression levels in MCF-7/6 cells remained unchanged. Moreover, co-immunoprecipitation experiments revealed that FAK is not associated with either integrin, which is in contrast with data reported by Schaller et al. . This confirmed the role of MSGb5 in the activation of downstream signalling events further.
In the present study, the multiple factors regulating the complex process of invasion can all be placed in the ‘concept’ of the glycosynapse . This new concept proposes the particular interactions and organization of integrins with tetraspanins and gangliosides, and has recently been extended to phenotypic conversion induced through the deletion or addition of a single component, resulting in a disorganized glycosynapse framework and initiating altered signalling events [6,43]. In the present study, we conclude that phenotypic conversion from non-invasive to invasive MCF-7/AZ breast cancer cells is induced by: (i) an aberrant MSGb5 pattern; (ii) loss of integrin receptor subunits α1 and β1; and (iii) high tetraspanin (CD9) expression levels , all of which are responsible for the formation of disorganized glycosynapse framework interfaces, thereby inducing activation of FAK, cSrc and downstream ERK, with consequent enhanced secretion and activity of MMP-2 and MMP-9, and thus leading to invasion (Figure 8).
Changes in GEM leading to cancer cell invasion
In conclusion, our studies are an extension of previous work on the glycosynapse , re-formulating the classic concept of integrin-dependent invasion of tumour cells and providing evidence that phenotypic conversion can be explained by differences in composition and organization of crucial molecules in the glycosynapse. At present, only a few studies have appeared that focus, in particular, on GM3 [6,44]. The present study reveals a novel insight into the composition and organization of the glycosynapse in MCF-7/AZ breast cancer cells, which explains phenotypic changes. We are currently investigating differences in the composition and organization of GSLs and associated molecules in other cancer cell-lines where similar phenotypic conversion has been observed. Further studies along this line are necessary to understand the complex interplay of distinct molecules in invasion, as well as other basic cellular mechanisms, and their implications on disease processes, which will be expected to lead to novel therapeutic approaches.
This work was supported by a grant from the U.S. National Institutes of Health (NIH) (RR-16480) under the BRIN (Biomedical Research Infrastructure Network)/INBRE (IDeA Network of Biomedical Research Excellence) programme of the National Center for Research Resources and the New Mexico Tech start-up funds.
calcium- and magnesium-free Hanks balanced salt solution
Dulbecco's modified Eagle's medium
extracellular signal-regulated kinase
focal adhesion kinase
mitogen-activated protein kinase
stage-specific embryonic antigen-4