Osteonectin downregulates E-cadherin, induces Osteopontin and Focal adhesion kinase activity stimulating an invasive melanoma phenotype



Osteonectin is recognised as a marker of metastasis progression in melanoma and has been implicated in the transition from radial to vertical growth phase. A Tetracycline-inducible system was used to regulate Osteonectin protein levels in melanoma cell lines to examine the morphological, biochemical and invasive changes that accompany its altered expression. Assay of protein and phosphorylation changes showed a downregulation of E-cadherin, upregulation of Osteopontin and a corresponding increase in phosphorylation of Focal Adhesion Kinase on Tyr397 and Tyr576 concomitant with Osteonectin induction. Melanoma cells overexpressing Osteonectin displayed increased invasive potential, whereas ablation of Osteonectin gene transcription using siRNA suppressed the invasive potential of these cells and resulted in the upregulation of E-cadherin. The recently described interaction of Osteonectin with Integrin Linked Kinase leading to modulation of its activity suggests a mechanism relevant to the loss of E-cadherin and cell adhesion that occurs during melanoma progression. These results indicate a central role for Osteonectin in the regulation of gene expression changes driving the progression of melanoma toward metastasis. © 2007 Wiley-Liss, Inc.

Cell–cell contacts and adhesion to the extracellular matrix (ECM) are important determinants of cell fate, proliferation, survival and tumour formation. Melanocytes of the epidermis are held in check via homeostatic interactions with the surrounding keratinocytes, but these regulatory contacts are overcome during the expansion of radial growth phase melanoma (RGP) and progression to the vertical growth phase (VGP).1 Some of the basic steps in the change in cell–cell adhesion molecules and cell communication responsible for melanoma progression have been characterised. First is the down regulation of receptors in melanocytes important for contact with keratinocytes, such as the loss of E-cadherin,2 this may be accompanied by the gain of receptors involved in the interaction between melanoma cells or with fibroblasts of the dermal compartment such as N-cadherin. Progression to metastasis involves the expression of molecules interacting with the ECM such as β3 integrin expression.3

Matricellular proteins are components of the ECM, which function as adaptors and modulators of cell–matrix attachment and include such structurally diverse proteins as thrombospondins-1 and -2, and Osteonectin (also known as SPARC and BM-40).4 These proteins are capable of supporting the initial and intermediate stages of cell adhesion with attachment and spreading. However, focal adhesion and stress fiber formation, characteristic of strong cell adhesion are not usually observed when cells are plated on these substrates and matricellular proteins can antagonize the proadhesive activities of other matrix proteins. For example, prolonged exposure to Osteonectin induces cell rounding similar to that seen by disruption of the ECM–integrin interactions by proteolysis or integrin antagonists.5

Although production of the β3 integrin protein has been found to be critical for melanoma cell progression to metastasis,6, 7 the genetic determinants that drive the metastatic process remain to be determined.8 A search for gene expression changes between β3-subunit positive and negative melanoma cell populations using a highly efficient subtractive hybridization method9 revealed that β3-integrin overexpression upregulates molecules associated with both adhesion and deadhesion. Notably, we found that the counter adhesive protein Osteonectin was induced during the progression of melanoma cells from RGP to invasive VGP. Osteonectin expression has previously been associated with progression in both human and mouse melanoma,10, 11 also as a marker correlated with increased incidence of distant metastases,12, 13 decreased survival14, 15 and recently with the induction of an epithelial–mesenchymal transition (EMT).16, 17

In addition to the role of cell adhesion, many growth factors and signaling pathways have also been implicated in the progression of melanocytic cells towards metastasis. Osteopontin is both a cytokine and cell adhesion protein acting as a ligand for various integrin receptors.18, 19 Using both DNA microarray and tissue section arrays the levels of Osteopontin protein have been found to be one of the earliest markers associated with progression to melanoma invasiveness.20 Normal melanocyte cell migration within the skin requires a transcriptional downregulation of E-cadherin, and there is evidence showing that the Snail family proteins act as direct repressors of E-cadherin gene activity and are involved in its suppression in melanoma cells.21 Another transcription factor implicated in gene expression changes underlying melanoma is MITF, which modulates various developmental, differentiation and survival genes and may act as an amplified oncogene in some melanomas.22, 23 It is clear that there is an important interplay between the cell adhesive properties that coordinate the signaling and transcriptional changes that regulate E-cadherin activity and result in an invasive phenotype in melanoma, with the mechanisms underlying this regulation yet to be fully described.

To further investigate the induction of Osteonectin in the RGP-VGP transition of melanoma we used a Tetracycline-inducible (Tet) system to regulate the production of Osteonectin in melanoma cells and have examined morphological, biochemical and invasive changes that accompany overexpression and ablation of Osteonectin gene transcription.


Dox, doxycycline; ECM, extracellular matrix; EMT, epithelial–mesenchymal transition; FAK, focal adhesion kinase; RGP, radial growth phase; Tet, tetracycline; VGP, vertical growth phase.

Material and methods

Cell culture

Human primary melanocytes (QF863 MC) and melanoblasts (QF1177 MB) were established from neonatal foreskin tissue and cultured as described.24 All human melanoma, HeLa and HEK293 cell lines were cultured in RPMI 1640 media supplemented with 5% FBS, 2 mM l-glutamine and penicillin (20 U/ml)/streptomycin (20 μg/ml). The origin of melanoma cells lines have been described previously,25 with WM1552C and SBcl2 a generous gift from Prof. M. Herlyn, The Wistar Institute Philadelphia. Purified Osteonectin protein from human blood plasma was purchased from Haematologic Technologies (Essex Junction, VT) and used as a supplement with serum free media.


The pBI-EGFP vector containing a modified polylinker, including the restriction sites AvrII, ClaI, NsiI, SphI and XhoI, was a generous gift from Dr. M. Whitelaw, University of Adelaide, SA. The pSilencer 3.1-H1 neo vector and pSilencer 3.1-H1 neo-negative control template was obtained from Ambion (Austin, TX).

The full length Osteonectin cDNA was excised as an EcoRI partial digest from pTRE-Osteonectin a kind gift from Dr. E. Thompson, St. Vincents Institute for Medical Research, VIC and ligated into the EcoRI site of pBlueScript. The pBI-EGFP-Osteonectin construct was generated by endfill of a XbaI-XhoI restriction fragment containing the full length Osteonectin gene from the pBluescript–Osteonectin clone. This fragment was subcloned into the endfilled AvrII restriction site of pBI-EGFP.

The generation of the pEF-tA-IRES-puro-6 construct involved the excision of an EcoRI-BamHI restriction fragment containing the tTA Tet-Off regulator from pUHD15.1neo26, 27 followed by directional ligation into the EcoRI-BamHI restriction sites of pEF-IRES-puro-6 kindly provided by Dr. M Whitelaw, University of Adelaide, SA.

The cloning of pSilencer–Osteonectin involved using the algorithm—Insert Design Tool for the pSilencer Vectors (Ambion) to obtain the sequence for sense and antisense 65mer hairpin siRNA encoding DNA oligonucleotides based on the predesigned Osteonectin siRNA target sequence (Ambion, ID No. 12,698). The oligonucleotides were synthesized by Geneworks (Adelaide, Australia) and subsequently annealed and ligated into the BamHI-HindIII site of pSilencer 3.1-H1 neo vector.


The generation of pSilencer permanently transfected MM96L cell clones required transfection of pSilencer constructs using Lipofectamine 2000 (Invitrogen, Melbourne, Australia). Clones were selected and maintained in 400 μg/ml of geneticin (Invitrogen). The generation of permanently transfected pBI-EGFP inducible (Tet-Off) MM96L cell clones involved cotransfection of the pBI-EGFP-Osteonectin construct with pEF-tA-IRES-puro in a ratio of 20:1. Clones were selected and maintained in 0.4 μg/ml puromycin (Sigma, Sydney, Australia). Addition of 1 μg/ml doxycycline hydrochloride (Dox) (Sigma) was used to suppress transcription of the transfected pBI-EGFP-Osteonectin DNA. Clones were expanded and split into +/−Dox growth media for 4–5 days. EGFP expression was assessed by live cell immunofluorescence then cell lysates subjected to western blot analysis for Osteonectin expression. Clones were selected for expansion based on a low to high range of inducible Osteonectin expression, clone 16 was graded as low (L), clone 22 as medium (M) and clone 13 as high (H).

Transient siRNA transfections of MM96L cells were performed using 200 pmol of siRNA molecules using Lipofectamine 2000 (Invitrogen) and cells were lysed after 48 hr. The predesigned annealed siRNA's were purchased from Ambion and directed against Osteonectin (Ambion, ID No. 12,698; 12,792), MITF (Ambion, ID No. 11,5142), SNAI2 (Ambion, ID No. 12,484), PTK2 (Ambion, ID No. 103,596) and a commercial negative control (Ambion, catalogue No. 4,611).

Western Blot analysis

Cell lysates were prepared using an SDS-extraction buffer (10 mM Tris-HCl pH 8.0, 20% Glycerol, 1% SDS) or the PARIS kit lysis buffer (Ambion). Denatured protein samples (∼10 μg) were resolved on either 8 or 10% SDS-PAGE gels and transferred to Hybond-C nitrocellulose membranes (Amersham Biosciences, Sydney, Australia). Overnight 4°C incubation of primary antibodies included anti-Osteonectin (1:5,000, Haematologic Technologies), anti-E-cadherin (1:3,000, Becton Dickinson, Sydney, Australia), anti-FAK (1:1,000, Lab Vision, Fremont, CA), anti-FAK[pY397] (1:1,000, Bioscource, Nivelles, Belgium), anti-FAK[pY576/577] (1:1,000, Cell Signaling Technologies, Danvers, MA), anti-MITF (1:400, Lab Vision), anti-SLUG G-18 (SNAI2; 1:200, Santa Cruz Biotechnologies, Santa Cruz, CA), anti-Osteopontin (1:1,000, R&D Systems, Minneapolis, MN), anti-intermediate filament antigen monoclonal supernatant (1:50,28), anti-GAPDH (1:8,000, R&D Systems). Membranes were washed followed by incubation with either HRP-conjugated anti-rabbit (1:8,000) or anti-mouse (1:10,000) (Upstate Biotechnology, Lake Placid, NY). Membranes were visualized using either Supersignal West Pico (Pierce Biotechnology, Rockford, IL) or Visualizer (Upstate Biotechnology).

RNA extraction and quantitative real time analysis

RNA isolation was performed using either the RNAqueous kit (Ambion) or PARIS (Ambion) as per manufacturers instructions. Samples were DNAseI treated using Turbo DNA-free kit (Ambion). First strand cDNA synthesis was performed on 0.65-μg RNA using Superscript III polymerase (Invitrogen). Quantitative Real Time PCR analysis using an ABI Prism 7000 or 7500 sequence detection system (Applied Biosystems, Melbourne, Australia) was performed on a 25-μl reaction containing 5% of the first strand cDNA synthesis solution, Taqman Universal PCR mix (Applied Biosystems) and Taqman Osteonectin primer/probe mix (ID No. Hs00277762_m1) (Applied Biosystems), with each sample analysed in duplicate. Gene expression levels were normalised to either, GAPDH (ID No. Hs99999905_m1) or a predeveloped 18S ribosomal RNA (Applied Biosystems).

Invasion assay

In vitro cell invasion was assayed by the membrane cell invasion system.29 Modified Boyden chambers containing polycarbonate membranes (8 μm, Transwell, Corning, Lowell, MA or 12 μm, Neuro probe, Gaithersburg, MD) were coated with 1 mg/ml diluted Matrigel (Becton Dickinson). Cells were resuspended in Serum free media +/−Dox as required, and equal cell numbers were applied to the top chamber. The Lower chamber contained Serum free media as a control or conditioned growth media containing 10% FBS, with +/−Dox as the attractant. Chambers were incubated for 20 hr at 37°C in 5% CO2. Cells that invaded through to the lower surface of the membrane were fixed in 4% paraformaldehyde and stained with 0.5% Toluidine blue in 1% borax solution. The cell nuclei were stained with DAPI (10 mg/ml, Sigma) and counted (5 different fields per membrane under 20× lens objective). Statistical analysis was performed using a Student's t-test and GraphPad Prism software, SanDiego, CA.


Endogenous protein markers associated with the transformed melanocytic phenotype

To begin an investigation of the action of Osteonectin in melanoma metastasis, total protein extracts were prepared from a range of melanocytic cells to first examine them for the expression pattern of markers previously associated with cellular transformation (Fig. 1). Western blot analysis indicated that the levels of Osteonectin expression in primary cell cultures of human melanoblasts and melanocytes were relatively low when compared to most melanoma cells, however melanoblasts contained levels similar to that seen in primary melanoma cell lines. Notably, this increase in Osteonectin expression correlated with a loss or down regulation of E-cadherin in melanoblasts. The melanoma cell lines that express high levels of Osteonectin also showed a general reduction or loss in E-cadherin, but as previously established30 E-cadherin expression may not be lost completely in all melanoma cells (lanes 5–7). Consistent with previous reports strong Osteopontin expression was observed in most of the melanoma cell lines tested as this is associated with progression to melanoma invasiveness,20 but in culture it was also found to be present at detectable levels in primary melanocyte and melanoblasts. An increase in phosphorylation of focal adhesion kinase (FAK) on tyrosine residues Tyr397 and Tyr576 was also observed in the melanoma cell lines, notably phosphorylation of FAK has been linked to an aggressive melanoma phenotype.31

Figure 1.

Endogenous protein expression of Osteonectin, E-cadherin, Osteopontin, FAK and GAPDH in melanocytic cells. Western blot on cell lysates from human primary melanocytes (lane 1), human primary melanoblasts (lane 2) and cell lines derived from RGP (lanes 3, 4), primary (lanes 5, 6), secondary (lanes 7, 8) human melanoma and human cervical cancer and human embryonic kidney (lanes 9, 10). Vertical lines indicate removal of intervening lanes or rearrangement of samples run on the same gel.

Exogenous regulation of Osteonectin protein levels and effect on cell morphology

The effects on cell function associated with changes in Osteonectin expression levels were assessed using clonal variants of the readily transfectable MM96L metastatic melanoma cell line,32 which while expressing significant levels of Osteonectin (Fig. 1) still retained some E-cadherin expression and intermediate phospho-FAK levels compared to other cell lines. Figure 2a shows the analysis of a representative pBI-EGFP vector control and 3 permanently transfected pBI-EGFP-Osteonectin Tet-Off inducible MM96L clones selected to represent high (H), medium (M) and low (L) Osteonectin induced expression levels. Although the H, M and L clones displayed different levels of basal Osteonectin expression, possibly due to clonal range variation upon selection, all demonstrated the capacity for a significant induction of Osteonectin when Dox was removed from the media. In contrast, analysis of 3 separate vector control cell lines showed no alteration of Osteonectin levels +/−Dox. The changes apparent in cellular morphology of the representative H, M and L clones corresponding to the removal of Dox were visualised by phase contrast brightfield microscopy (Fig. 2b), with these images showing distinct graded morphological changes as the level of Osteonectin expression increased. The clones acquire a less dendritic, flattened and larger cell body in L-clones, with even greater flattening leading to eventual cell rounding and detachment with M and H-clones. Similar morphology changes have previously been recorded using Tet-regulated Osteonectin expression in glioblastoma cell lines.33, 34 Examination of the same field of cells by fluorescent microscopy (data not shown) to observe the induction of EGFP expression showed this to correlate with the ectopic expression of Osteonectin as expected given both gene transcripts are driven through a bidirectional promoter regulated by the Tet responsive element. Although EGFP expression was indicative of Osteonectin expression, and useful in determining a time frame for functional studies, it was found not to be an absolute measure of protein levels as M and H-clone cells appeared with similar fluorescent intensities when detected with the same parameters set to detect L-clone fluorescence.

Figure 2.

Phenotypic changes in MM96L cells containing variable overexpressed or repressed Osteonectin protein levels. (a) A pBI-EGFP vector control and 3 permanently transfected pBI-EGFP-Osteonectin Tet-Off inducible MM96L cell lines representing total high (H), medium (M) and low (L) levels of ectopic Osteonectin expression after growth in +/−Dox conditions. Cell lysates were resolved by SDS-PAGE and probed with an Osteonectin antibody in the upper panel. Vertical lines indicate removal of intervening lanes between samples. (b) A comparison of phase contrast brightfield images (×20) showing morphological changes to transformed cells before (+) after (−) removal of Dox in the H, M and L Osteonectin inducible clones. (c) (i) Western blot of cell lysates from the parental MM96L cell line (P), a pSilencer negative control (Neg) and 2 pSilencer–Osteonectin MM96L clones (C1 and C4) using an Osteonectin antibody. (ii) Phase contrast brightfield images showing cell morphology of parental, pSilencer negative control (Neg) and pSilencer–Osteonectin (C1) representative of other clones tested. (d) Quantitative real time RT-PCR confirmation of Osteonectin expression in the H, M and L inducible and C1, C4 repressed MM96L clones from (a) and (b). Expression levels were normalised to GAPDH and shown relative to the parental MM96L, arbitrarily set to 1.0.

The Impact of Osteonectin repression on melanoma cell function was assessed using permanently transfected pSilencer-Osteonectin constructs [Fig. 2c(i)]. Western blot analysis showed strong repression of Osteonectin in representative clones C1 and C4 (lanes 3 and 4). A permanently transfected negative pSilencer vector control displayed similar expression of Osteonectin compared to the endogenous levels present in the parental MM96L cell line. Interestingly, the associated morphological changes although subtle [Fig. 2c(ii)], showed contrasting effects to that observed in the L, M and H MM96L Osteonectin expressing clones. Phase contrast brightfield images of C1, which were representative of other repression clones showed lengthening of dendrites accompanied by a slightly smaller cell body compared to parental and negative pSilencer vector control cell lines, possibly indicating greater cell attachment.

Quantitative real time RT-PCR analysis of the clonal variants in the basal, induced or repressed state was performed with Osteonectin mRNA levels corresponding with Osteonectin protein levels (Fig. 2d). This confirmed the designation of L, M and H MM96L clonal variants with the clonal effect on +Dox basal level of Osteonectin mRNA seen for the clonal variants also observed with the L-clone devoid of noninduced Osteonectin expression.

Osteonectin modulates protein markers associated with the transformed melanocytic phenotype

To assess the influence of Osteonectin overexpression in melanoma cells in vitro, the H, M and L MM96L clonal variants were used to examine expression changes of 3 key proteins associated with the progression to metastasis. The effects on E-cadherin regulation were first examined and the levels of this protein were found to be repressed in the presence of high levels of Osteonectin induced in –Dox cell growth conditions (Fig. 3a), a result consistent with the recent report of Robert et al.16 Notably, a higher basal level of E-cadherin in the L-clonal variant was observed consistent with the absence of Osteonectin in the uninduced state of these cells (Fig. 2a), however even with the lower level of Osteonectin induction a slight change in E-cadherin was still apparent. Moreover, this is also consistent with the increased levels of E-cadherin after Osteonectin ablation using siRNA techniques as discussed later (Fig. 5a). Separate analysis of 3 individual pBI-EGFP vector alone transfected control cell lines again showed no alteration of Osteonectin or E-cadherin levels +/−Dox (data not shown).

Figure 3.

Osteonectin overexpression correlates with changes in E-cadherin, Osteopontin and the phosphorylation state of FAK. (a) Western blot of cell lysates from the MM96L cell line (P) and high (H), medium (M) and low (L) Osteonectin inducible clones grown in +/−Dox were analysed with antibodies against E-cadherin and Osteonectin. (b) H, M and L Osteonectin inducible clones grown in +/− Dox then serum starved for 24 hr.35 Cell lysates were analysed by western blot using an Osteopontin antibody. (c) Cell lysates from H, M and L Osteonectin inducible clones grown in the +/−Dox were resolved by SDS-PAGE and probed with antibodies to FAK, FAK[pY397] and FAK[pY576]. Differences in expression levels were quantified by normalising to FAK, using imageJ software (http://rsb.info.nih.gov/ij/). Vertical lines indicate removal of intervening lanes between samples.

Concomitant with the downregulation of E-cadherin, examination of the same Osteonectin induced H, M and L MM96L cell extracts found that Osteopontin protein levels were upregulated (Fig. 3b). Assay of total protein levels and the phosphorylation status of FAK were also assessed in response to Osteonectin overexpression in the clonal variants. While little change in FAK protein levels was seen, this did illustrate that phosphorylation of FAK on both Tyr397 and Tyr576was apparent upon upregulation of Osteonectin (Fig. 3c), with the increase confirmed by normalisation to FAK levels.

Transient siRNA mediated knockdown of Osteonectin and FAK protein levels in the parental MM96L cell line were readily achieved (Fig. 4a). The removal of Osteonectin expression was not associated with a decrease in FAK phosphorylation relative to negative control siRNA treatment, in contrast ablation of FAK protein levels was concomitant with decreasing FAK Tyr397. This failure to observe changes in phospho-FAK, which may have been expected with loss of endogenous Osteonectin, may have been due to the presence of sufficient exogenous Osteonectin, which is present in serum used for cell growth. However, when 24-hr serum starvation conditions were used FAK levels were still unaffected after siRNA treatment of the cells (data not shown). In experiments testing the direct addition of purified Osteonectin protein to 24-hr serum starved cells, only a minor increase of FAK Tyr397 was found when normalised to FAK (Fig. 4b). It could be argued that the more dramatic affects observed in permanently transfected Osteonectin expressing cells (Fig. 3c) could be due to an autocrine mechanism acting in those cells, however this is unlikely given the stimulation of phospho-FAK levels reported to occur upon direct treatment of glioma cells.36

Figure 4.

Direct regulation of Osteonectin by ablation and addition of purified protein (a) MM96L cells were transiently transfected with siRNAs for negative control, Osteonectin and FAK. Western blot analysis was performed on cell lysates after 48 hr (lanes 1–3), using antibodies against Osteonectin, FAK[pY397] and FAK. (b) MM96L cells were passaged in normal growth conditions until ∼80% confluent, at which time cells were propagated for a further 24 hr in the absence of serum. Osteonectin supplemented serum free media was added (20 μg/ml), cells lysed at 15 min36 and FAK activity analysed by western blot (lanes 1, 2). Differences in expression levels were quantitated by normalising to FAK using imageJ software (http://rsb.info.nih.gov/ij/).

Changes accompanying repression of Osteonectin, MITF and SNAI2 in melanoma cells

To complement the regulated expression of Osteonectin in permanently transfected MM96L cell clones (Fig. 2), we also chose to ablate the Osteonectin transcript using transient siRNA transfection techniques to measure any changes in transcription factors implicated in melanoma, and in turn ablate the MITF and Snail family member SNAI2 (SLUG) transcription factors to assay for the effects on Osteonectin. The shorter time frame of these experiments allowed assessment of relative changes in transcription factors and Osteonectin that were not possible or that were not detectable in the permanently established MM96L clones. Knockdown of the Osteonectin transcript was confirmed by quantitative RT-PCR and lead to almost total loss of the Osteonectin protein (Fig. 5a, lane 1 vs. 4) compared to negative control siRNA treatment. The loss of Osteonectin was again accompanied by the reciprocal gain of E-cadherin, with slight increases in the level of MITF and SNAI2 proteins also apparent.

Figure 5.

Repression of Osteonectin, MITF and Snai2 in MM96L cells by siRNA treatment alters expression of E-cadherin. (a) MM96L cells were transiently transfected with negative control-siRNA, and siRNAs inhibiting MITF, SNAI2 and Osteonectin. Cell lysates were analysed by western blot, using antibodies against MITF, SNAI2, Osteonectin and E-cadherin. (b) Phase contrast images showing morphological changes to the cell after siRNA transfection. (c) Quantitative real time RT-PCR was performed on cDNA prepared from the same samples shown in (a). Expression levels were normalised to 18S and plotted relative to the negative control-siRNA, arbitrarily set to 1.0.

MITF siRNA knockdown (Fig. 5a), resulted in a clear upregulation of Osteonectin protein and mRNA relative to the negative control siRNA treated cells (Fig. 5c). Moreover, consistent with the data obtained from the H, M and L Osteonectin inducible permanently transfected MM96L cell lines, the increase in Osteonectin was associated with a transient decrease in E-cadherin expression and a flattened cell morphology (Fig. 5b). In contrast, the siRNA mediated knockdown of SNAI2 increased the expression of E-cadherin compared to the direct Osteonectin siRNA treated cells, secondly it was associated with a repression of Osteonectin, but no overt morphological changes were observed. Quantitative real time RT-PCR analysis performed on the transient knockdown of MITF, SNAI2 and Osteonectin in the transfected MM96L cells were consistent with those observed at the protein level (Fig. 5c).

Osteonectin increases the invasive potential of melanoma cells

To address whether overexpression of Osteonectin in melanoma contributes to a more invasive phenotype, the H and L Osteonectin inducible MM96L clonal variants were propagated in +/−Dox growth media and subjected to a matrigel coated modified Boyden chamber invasion assay. Figure 6a shows that in the absence of Dox the corresponding upregulation of Osteonectin enhanced the already high invasive properties of the H and L clonal variants. As a counterpoint to the relationship between overexpression of Osteonectin and increased invasiveness we wished to assess whether Osteonectin knockdown would reduce the invasive potential of melanoma. This was addressed by testing parental MM96L and pSilencer–Osteonectin expressing C1 MM96L melanoma cell lines (Fig. 2c) in the Boyden chamber invasion assay. As expected37 the ablation of Osteonectin in the C1-clone resulted in almost complete loss of its invasive potential in response to an attractant (Fig. 6b).

Figure 6.

Osteonectin affects the invasive potential of MM96L cells. (a) High (H) and low (L) Osteonectin inducible clones grown in +/−Dox were subjected to a modified Boyden chamber invasion assay. The attractant used was conditioned media containing 10% FBS. Columns represent the count of invaded cells on the underside of the membrane expressed relative to serum free conditions, arbitrarily set to 1.0 (not shown on graph). Error bars indicate SEM from 3 independent experiments (H, p = 0.3 relative to +Dox control; L, p = 0.1 relative to +Dox control). (b) The C1 pSilencer–Osteonectin and the MM96L (P) cell lines were subjected to a modified Boyden chamber invasion assay. The attractant was conditioned media containing 10% FBS. Error bars indicate SEM from 3 independent experiments (Parental, *p = 0.021 relative to control in serum free conditions.)


Our study utilised a Tet-inducible gene expression system to investigate the effects of Osteonectin in the transition to an invasive melanoma phenotype. Osteonectin induction was accompanied by significant changes in markers characterising the progression of melanoma. These included a downregulation of E-cadherin and an upregulation of Osteopontin, with a corresponding increase in phosphorylation of FAK on Tyr397 and Tyr576. Notably, downregulation of E-cadherin is a major determinant in the progression of melanoma,2 as when melanoma cells are forced to regain E-cadherin expression they become adhesive to keratinocytes, subject to growth control38, 39 and sensitized for apoptosis.40

Osteonectin has also previously been identified to correlate with human melanoma progression.9, 11, 12, 14, 15, 17 Using Boyden chambers we demonstrated that Osteonectin overexpression in MM96L melanoma cells leads to an increase in the invasive potential, reciprocally knockdown of Osteonectin suppressed the invasive potential of these cells further illustrating the importance of Osteonectin levels in melanoma progression.37 A recent report corroborates the findings of our study by demonstrating the increased invasive ability of both melanocytes and melanoma cells accompanied by the repression of E-cadherin expression upon Adenoviral mediated transduction of Osteonectin.16 Moreover, in vivo studies41 have demonstrated a resistance in developing carcinomas in Osteonectin-null mice after UV treatment, further implicating the involvement of Osteonectin in tumour progression.42

Integrin mediated signaling in cell-ECM interactions involves clustering at focal adhesions on the cell surface and subsequent FAK activation leading to cell migration. The significance of FAK signaling in melanoma43, 44, 45 has recently been established through linking the phosphorylation status of FAK with an aggressive melanoma phenotype.46, 47, 48, 49 A high level of FAK phosphorylation on Tyr397 and Tyr576 was observed in the collection of metastatic melanoma cell lines tested in our study. We have also demonstrated that FAK phosphorylation on Tyr397 and Tyr576 increased in response to overexpession of Osteonectin in MM96L cells, which correlated with an increase in invasive capacity.

Integrin-Linked Kinase (ILK) interacts with integrin subunits β1 and β3 when localised to focal adhesion complexes and has recently come to prominence as a key regulator of multiple signaling pathways.50 The recent discovery that Osteonectin can modulate ILK activity through direct protein–protein interaction in primary fibroblasts51 suggests a mechanism relevant to the loss of E-cadherin and cell adhesion, that occurs in the EMT of melanoma, shown using MM96L cells (Figs. 3 and 5) and reported for primary melanocytes.16 The observation that FAK activation increases in response to Osteonectin overexpression in MM96L cells, combined with our previously reported evidence that an increase in Osteonectin expression in RGP melanoma coincided with integrin β3 expression,9 allows us to propose that Osteonectin functions via interacting with ILK at focal adhesions to promote an invasive melanoma phenotype. The consequences of increased Osteonectin production observed during melanoma progression11, 14, 15 would result in a corresponding activation of ILK. Supporting this assumption is evidence linking upregulation of ILK with melanoma progression,52 though this has also been correlated with a loss of the ILK negative regulator ILK-associated phosphatase,53 these 2 possibilities are not mutually exclusive.

Upregulation of ILK initiates a cascade of signaling events,50 which are significantly associated with a malignant phenotype, the most notable being the downregulation of E-cadherin via Snail mediated repression,54 which is consistent with our finding E-cadherin increases in MM96L cells treated with SNAI2 siRNA (Fig. 5). Another ILK target includes PI3K, which activates AKT via phosphorylation of Ser473 (Ref.55). AKT activation in turn impacts upon the cell cycle by suppressing apoptosis and thereby increasing cell survival and has itself been associated with melanoma.56 PI3K has been shown to result in the activation of Osteopontin production,35 which has been associated with the earliest stages of melanoma tumour progression.18, 19, 20 The combined role of FAK and ILK stimulation in mediating an invasive phenotype mediated through Osteonectin expression has recently been established in studies of glioma cell lines.36

Our study has only briefly touched upon the regulatory role MITF may play in Osteonectin expression in melanoma. Many studies have shown MITF acts as a key regulator of melanocyte cell growth and functions in a protein dependent fashion resulting in both pro and antiproliferate effects on the cell cycle.22, 23 Our findings that MITF siRNA treatment of MM96L cells stimulates upregulation of Osteonectin and repression of E-cadherin supports a putative functional role for MITF in melanoma progression in part mediated through Osteonectin. In comparison siRNA mediated knockdown of SNAI2 increased the expression of E-cadherin to levels greater than that seen by Osteonectin siRNA treatment of cells. This result substantiates the work of Kuphal et al., 200557 who reported Osteonectin (SPARC) as a direct target of SNAI2 in permanently established antisense SNAI2 producing melanoma cell lines.

We have demonstrated that induction of Osteonectin in melanoma is linked with increased phosphorylation and subsequent activation of FAK resulting in a heightened invasive phenotype. We have also shown that downregulation of E-cadherin and overexpression of Osteopontin occurs in response to changes in Osteonectin. These results support a central role for Osteonectin regulation in the axis of gene expression changes that occur during melanoma progression. Future work on the signaling pathway surrounding Osteonectin, involving both FAK and ILK,36 will hopefully provide further insight into the regulatory processes associated with invasive tumours and perhaps lead to a related anti cancer drug therapy targeting these regulatory processes.58, 59


B.B.G. was a QCF PhD scholarship holder, and R.A.S. is an NHMRC Senior Research Fellow. The Institute for Molecular Bioscience incorporates the Centre for Functional and Applied Genomics as a Special Research Centre of the Australian Research Council.