Psoriasin (S100A7) associates with integrin &beta - Nature

Oncogene (2011) 30, 1422–1435

& 2011 Macmillan Publishers Limited All rights reserved 0950-9232/11 www.nature.com/onc

ORIGINAL ARTICLE

Psoriasin (S100A7) associates with integrin b6 subunit and is required for avb6-dependent carcinoma cell invasion MR Morgan1,4,6, M Jazayeri1,6, AG Ramsay1,2, GJ Thomas1,5, MJ Boulanger3, IR Hart1 and JF Marshall1 1 Centre for Tumour Biology, The London School of Medicine and Dentistry, John Vane Science Centre, London, UK; 2Centre for Medical Oncology, Institute of Cancer, Queen Mary University of London, Barts and The London School of Medicine and Dentistry, John Vane Science Centre, London, UK and 3Department of Biochemistry and Microbiology, Victoria, British Columbia, Canada

Expression of the integrin avb6 is upregulated in a variety of carcinomas where it appears to be involved in malignant progression, although the biology of this integrin is not fully explored. We have generated oral carcinoma cells that express avb6 composed of wild-type av and a mutant b6 that lacks the unique C-terminal 11 amino acids (aa). We found that these residues, although not required for avb6-dependent adhesion or migration, are essential for avb6-dependent invasive activity. We have used a proteomic approach to identify novel binding partners for the b6 subunit cytoplasmic tail and report that psoriasin (Psor) (S100A7) bound preferentially to the recombinant b6 cytoplasmic domain, though not in the absence of the unique C-terminal 11aa. Endogenous cellular Psor coprecipitated with endogenous b6 and colocalised with avb6 at the cell membrane and intracellular vesicles. Knockdown of Psor, with small interfering RNA, had no effect on avb6-dependent adhesion or migration but abrogated avb6mediated oral carcinoma cell invasion both in Transwell and, the more physiologically relevant, organotypic invasion assays, recapitulating the behaviour of the b6-mutant cell line. Membrane-permeant Tat-peptides encoding the unique C-terminal residues of b6, bound directly to recombinant Psor and inhibited cellular Psor binding to b6; this blocked avb6-dependent, but not avb6-independent, invasion. These data identify a novel interaction between Psor and b6 and demonstrate that it is required for avb6-dependent invasion by carcinoma cells. Inhibition of this interaction may represent a novel therapeutic strategy to target carcinoma invasion. Oncogene (2011) 30, 1422–1435; doi:10.1038/onc.2010.535; published online 6 December 2010 Correspondence: Dr JF Marshall, Institute of Cancer & CR-UK Clinical Centre, Centre for Tumour Biology, Queen Mary University of London, Barts and The London School of Medicine and Dentistry, John Vane Science Centre, Charterhouse Square, London EC1M 6BQ, UK. E-mail: [email protected] 4 Current address: Wellcome Trust Centre for Cell-Matrix Research, University of Manchester, Michael Smith Building, Manchester M13 9PT, UK 5 Current address: University of Southampton, School of Medicine, Cancer Sciences Division, Somers Building, Tremona Road, Southampton S016 6YD, UK 6 These authors contributed equally to this work. Received 19 February 2010; revised and accepted 6 October 2010; published online 6 December 2010

Keywords: avb6; integrins; invasion; psoriasin; cancer; oral SCC

Introduction Development of metastases requires that tumour cells initially must violate tissue boundaries to invade locally, disseminate and, after arrest at distant sites, must then extravasate and invade the colonised organ (Liotta and Kohn, 2001). The process of tumour cell invasion therefore is of major significance in oncology and increased understanding of the molecular basis of this phenomenon may lead to identification of novel antiinvasive targets. We suggest that the integrin avb6 constitutes such a target in carcinoma cells, which express this heterodimer selectively, and have sought to determine molecules associated with invasive activity mediated by this receptor. Integrin avb6 is a receptor for the extracellular matrix proteins fibronectin and tenascin, and also the latencyassociated peptide (LAP), which is the pro-peptide of latent tumour growth factor-b (Busk et al., 1992; Prieto et al., 1993; Weinacker et al., 1994; Munger et al., 1999; Thomas et al., 2002). Expression of avb6 is restricted to epithelial cells where, in most healthy adult tissue, it usually is absent or present only at low or undetectable levels (Breuss et al., 1993). However, avb6 is upregulated during processes of tissue remodelling, including wound healing, development and in many epithelial malignancies (reviewed in Thomas et al., 2006). Significantly, avb6 expression is increased in squamous cell carcinoma (SCC) cells (Breuss et al., 1995; Clark et al., 1996; Haapasalmi et al., 1996; Larjava et al., 1996; Jones et al., 1997; Regezi et al., 2002), often with the highest levels being localised at the invasive margins (Breuss et al., 1995; Jones et al., 1997; Thomas et al., 2006), suggesting that avb6 contributes to an invasive phenotype. Indeed we, and others, have shown that avb6 promotes invasion of transformed keratinocytes (Thomas et al., 2001a, b; Ramos et al., 2002). Moreover, strong expression of avb6 correlates with a significant reduction in median survival from various cancers (Bates et al., 2005; Elayadi et al., 2007; Hazelbag

Psoriasin mediates avb6-dependent invasion MR Morgan et al

1423

Figure 1 Generation of V3B6D11aa cells. (a) Aligned amino acid sequences of cytoplasmic domains of b1, b3, b6 and truncated b6 (b6D11aa) integrin subunits. Regions of high homology (cytodomains 1, 2 and 3) are underlined and the unique 11 amino acid C-terminal sequence of the wild-type b6 subunit is in bold. This unique sequence was replaced with a myc/his tag (italics) in b6D11aa.(b) Flow cytometric analysis of avb6 integrin expression in transfected oral SCC cell lines C1, VB6 and V3B6D11aa. Negative controls, IgG1k, are shown as filled, and avb6 expression (E7P6) as open, histograms. (c) Cell surface distribution of avb6 determined by indirect immunofluorescence using E7P6.

et al., 2007). Thus, avb6 acts as a pro-invasive integrin in several epithelial malignancies. Integrins modulate various cell functions, upon ligand binding, by generating intracellular signals through their cytoplasmic tails (Kumar, 1998). The cytoplasmic tails of most b integrin subunits, (including b1, b2, b3, b5 and b6), are highly conserved with three regions in particular (cytodomains 1, 2 and 3) having very high homology (Figure 1a) (Reska et al., 1992). However, the b6 integrin subunit has an unique 11 amino acid (11aa) C-terminal sequence (Figure 1a). These C-terminal 11aa are required for three-dimensional growth in type I collagen gels (Agrez et al., 1994; Niu et al., 2002) and are involved in density-dependent upregulation of b6 expression (Niu et al., 2001). Indeed, we have demonstrated previously that addition of these residues to the C-terminus of b3 was sufficient to promote avb3dependent invasive capacity (Morgan et al., 2004). It is clear, therefore, that these 11 aas are critical for the regulation of avb6 function in carcinoma cells, though the molecular basis of this regulation remains unclear. To examine the role of the C-terminus of b6 in modulating oral SCC invasion we have generated oral

carcinoma cells expressing an avb6 heterodimer containing a truncated form of b6 lacking the C-terminal 11aa. We have established that these C-terminal 11 aas are not required for avb6-dependent adhesion or migration but are essential for avb6-dependent oral carcinoma cell invasion. To identify integrin-associated proteins that could mediate the pro-invasive effects of the C-terminal 11aa, we used proteomics to identify proteins bound to recombinant b6 cytoplasmic domains. We found that psoriasin (Psor) (S100A7) bound preferentially to the wild-type b6 cytoplasmic tail, but not to an 11aa deletion mutant. Psor is a member of the S100 family of EF-hand proteins, first identified as being upregulated in psoriatic skin (Madsen et al., 1992). As discussed below, this sometimes secreted protein has been implicated as promoting anti-microbial activity and inflammation (Glaser et al., 2005; Wolf et al., 2008), regulating cell survival and signalling (Emberley et al., 2005; Peterson et al., 2007; Zhou et al., 2008) and being upregulated in oral and breast cancer (Al-Haddad et al., 1999; Jiang et al., 2004; Emberley et al., 2004; Krop et al., 2005; Banerjee et al., 2005; Ralhan et al., 2008; Wang et al., 2008; Paruchuri et al., 2008; Kesting et al., 2009). We Oncogene


1424

show that Psor was required for the pro-invasive activity of avb6 and that disruption of the association of Psor with the b6 cytoplasmic tail inhibited invasion of oral SCC cells. Previously avb6 and Psor have been implicated independently in carcinoma progression and invasion (Breuss et al., 1995; Clark et al., 1996; Haapasalmi et al., 1996; Larjava et al., 1996; Jones et al., 1997; Al-Haddad et al., 1999; Regezi et al., 2002; Jiang et al., 2004; Emberley et al., 2004; Krop et al., 2005; Banerjee et al., 2005; Ralhan et al., 2008; Wang et al., 2008; Paruchuri et al., 2008; Kesting et al., 2009), and our data identify a novel pro-invasive interaction that may explain some of these previous observations. The data further suggest that inhibition of this b6–Psor association may provide a novel therapeutic target to inhibit carcinoma invasion.

Results V3B6D11aa cells express functional avb6 heterodimers To elucidate the role of the 11aa C-terminal sequence that is unique to the b6 subunit, the V3 oral SCC cell line was infected with retroviruses encoding either wildtype b6, to create VB6 (Thomas et al., 2001a) or with a truncated-b6 lacking the C-terminal 11 aas, which were replaced with a myc/his tag, to create the V3B6D11aa cell line (Figure 1a). Flow cytometry and indirect immunofluorescence (Figures 1b and c) confirmed that both lines expressed similar avb6 levels at their cell surfaces; levels over 20-fold that found at the surface of the original C1 control cells. The subcellular localisation of avb6 in both VB6 and V3B6D11aa cells was similar; the punctate cell surface distribution of avb6 was not affected by the absence of the C-terminal 11aa. We have shown previously that LAP of tumour growth factor-b is an avb6-specific ligand in keratinocyte-derived cells (Thomas et al., 2002). We assessed the adhesion of the oral SCC cell lines to LAP in the presence and absence of integrin-blocking antibodies. VB6 and V3B6D11aa cells displayed similar levels of enhanced adhesion to this ligand, relative to the null infectants C1; these increased levels of adherence were reduced to basal levels in the presence of anti-avb6blocking antibody, 10D5 (Figure 2a). In haptotactic migration assays, toward fibronectin, C1 cells displayed almost no migration and this very limited activity was unaffected upon antibody blockade of avb6 or both avb6 and a5b1 (Figure 2b). In contrast, the VB6 and V3B6D11aa cells exhibited substantial levels of haptotaxis (10.69 and 10.99%, respectively) levels, which were abrogated by blockade of avb6 alone (Figure 2b). The residual migration of VB6 and V3B6D11aa in 10D5 was reduced further by the co-presence of P1D6, showing that there was a small component of migration towards fibronectin that was a5-dependent (Figure 2b). These results clearly demonstrate that the C-terminal 11 aas of the b6 integrin subunit are not required for either avb6-dependent adhesion to LAP or for migration Oncogene

towards fibronectin. The data also show that the modified avb6D11aa heterodimer, when expressed in oral SCC cells, can serve, at least partially, as a functional b6 integrin receptor. The C-terminal 11 aas of the b6 cytoplasmic tail are required for oral SCC invasion The invasive capacity of our panel of cells was investigated in Transwell assays. Although VB6 cells exhibited high levels of invasion, V3B6D11aa cells invaded Matrigel only as well as the low avb6expressing C1 cells (Figure 2c), even though, as we had established already, V3B6D11aa cells express as much functional avb6 as VB6 cells at their cell surface (Figure 1b). The substantially reduced invasive capacity of V3B6D11aa, relative to VB6, cells was highly significant (Po0.001). Moreover, although the enhanced invasion exhibited by VB6 is entirely avb6dependent, blockade of avb6 or a5b1 had no effect on the low levels of V3B6D11aa invasion (Figure 2d). Thus, the C-terminal 11 aas of the b6 cytoplasmic tail are essential for avb6-dependent SCC invasion. Cytoplasmic domain of b6 associates with Psor Given the essential role of the C-terminal 11aa of b6 in regulating avb6-dependent SCC invasion, we sought to identify molecules associated with this unique motif. The entire b6 subunit cytoplasmic tail and a C-terminal 11aa truncated version were expressed as glutathione-Stransferase (GST) chimeras, as described, and used to coat glutathione beads. These beads were used in pulldown assays, bound material eluted, separated by 2dimensional (2D)-gel electrophoresis and individual spots analysed by matrix assisted laser desorption ionisation time of flight mass spectrometry. All pulldowns were performed three times and analysed on three separate 2D-gels. The experiment was highly reproducible as, after colloidal blue staining of 2D-gels, the spot pattern was almost indistinguishable between gels (Figure 3a and data not shown). One particular protein spot co-precipitated with GSTb6WT but was almost absent from the GSTb6D11aa 2D precipitates (oval in Figure 3a, inset). Mass spectrometry of this spot identified Psor, also known as S100A7 (Madsen et al., 1992). In Figure 3b, we show that Psor co-immunoprecipitated with avb6, and b6 co-precipitated with Psor, from VB6 cell lysates. We repeated the GST pull-downs from VB6 lysates with GST, GST-b6 and GST-b6D11aa and showed again that Psor bound to the wild-type b6-cytoplasmic tail protein but not the truncated b6D11aa protein (Figures 3c and d). These data show that endogenous Psor interacts with wild-type avb6 in oral SCC cells. To establish whether Psor bound directly to the cytoplasmic tail of b6 we used solid-phase enzyme-linked immunosorbant assays. Recombinant GST-b6, GST-b6D11aa and GST alone was immobilised on 96-well plates and probed with increasing concentrations of Psor protein, in the presence or absence of 10 mM Tat-WT peptide. Figure 3e shows that


1425

Figure 2 Integrin b6 C-terminal 11aa are essential for avb6-dependent invasion. (a) Adhesion of C1, VB6 and V3B6D11aa cells to LAP was assessed by monitoring levels of 51Cr-labelled cells bound after 45 min. The experiment was performed in quadruplicate wells and repeated three times with similar results. Figure shows data from a single representative experiment. (b) Haptotactic migration of C1, VB6 and V3B6D11aa cells through Transwell filters coated with 10 mg/ml fibronectin in the presence of integrin inhibitory antibodies (10D5 anti-avb6; P1D6 anti-a5b1). Experiment was performed in quadruplicate wells three times and similar results were obtained on each occasion. (c) Mean invasion of C1, VB6 and V3B6D11aa cells through Matrigel-coated Transwell filters. (d) Matrigel invasion by VB6 (solid bars) and V3B6D11aa (empty bars) cell lines was assessed in the presence of integrin function blocking antibodies (10D5 anti-avb6; P1D6 anti-a5b1). Invasion assays were carried out in quadruplicate wells. Data presented are from a single representative experiment out of five similar repetitions. All error bars represent standard deviation.

only the GST-b6 bound to Psor in a dose-dependent manner suggesting interaction with the C-terminal 11aa. This conclusion was confirmed as the competitive peptide, Tat-WT, blocked the binding of psoraisin to GST-b6. In a similar experiment increasing concentrations of recombinant Psor were immobilised onto 96well plates and probed with a single concentration of GST, GST-b6 and GST-b6D11aa protein. Again, only the GST-b6 protein bound (Supplementary Figure 1). These data indicate that Psor can bind directly to the cytoplasmic tail of wild-type b6 but not the truncated b6D11aa protein. Finally, Psor was co-precipitated with avb6 from VB6 cells, but not from V3B6 D11aa cells (Figures 3f and g), confirming the absolute requirement for the C-terminal 11 aa for b6–Psor association to occur.

Psor is essential for avb6-mediated invasion of oral SCC cells Treatment with Psor small interfering RNA (siRNA) substantially reduced the levels of Psor protein expression, compared with control RNA interference-transfected cells, and this suppression persisted for at least 14 days (Figure 4a). Knockdown of Psor had no effect on total levels of the b6 subunit (Figure 4a). Control siRNA and Psor siRNA-treated VB6 cells were assayed for avb6-dependent adhesion to LAP. Figure 4b shows that there was no significant change in adhesion to this ligand. Similarly, treated cells were assessed for their capacity to migrate toward fibronectin and LAP. Figure 4c shows that suppression of Psor protein expression had no effect on avb6-dependent migration to the avb6 ligands, fibronectin and LAP. Oncogene


1426

Figure 3 Psoriasin (Psor) binds to the cytoplasmic tail of b6. (a) Proteins pulled down by GSTb6WT or GSTb6D11aa-coated beads were analysed by two-dimensional (2D)-gel electrophoresis. Two gels, from a single experiment are shown with a magnified region of interest. An oval identifies the differentially expressed spot that was excised, analysed by Mass Spectrometry and identified Psor (S100A7). (b) Lysates of VB6 cells were immunoprecipitated (IP) with anti-Psor (47C1068) or anti-b6 (71C5). Immunoprecipitates were detected by western blot (WB) with goat anti-b6 or anti-Psor, respectively. (c) Proteins were precipitated from VB6 lysates using GSTfusion proteins (GST, GST-b6 and GST-b6D11aa) and GST-fusion protein-bound Psor detected by western blot. Levels of Psor were expressed relative to levels of GST. Immunoblot is representative of three independent experiments and histogram (d) shows densitometric quantification of three separate experiments. (e) GST-fusion proteins (GST, GST-b6 and GST-b6D11aa) were immobilised on 96-well plates (10 mg/ml) and binding of purified recombinant Psor (0–10 mM) immunodetected after 10 min preincubation with, or without, 10 mM Tat-WT peptide. (f) Lysates of VB6 and V3B6D11aa cells were supplemented with 1.1 ng recombinant Psor and immunoprecipitated with rat anti-b6 (620 W). Immune complex-associated proteins were analysed by western blot (anti-psoraisin and anti-b6 (310 W)). (g) Levels of b6-associated Psor are expressed relative to levels of b6 integrin determined by densitometric analysis of three independent experiments. MW, Molecular Weight; TCL, total cell lysate.

Control or Psor siRNA-treated VB6 cells were tested for their ability to invade Matrigel-coated Transwell filters. Significantly, suppression of Psor expression reduced VB6 invasion by 60% (Po0.001) (Figure 4d). Thus, the pattern of behaviour, of VB6 cells treated with anti-Psor siRNA (in terms of cell adhesion, migration and invasion), recapitulates the behaviour of the V3B6D11aa cells that lack the C-terminal 11aa of the b6 subunit (Figure 2). Oncogene

Previously we reported that invasion by VB6 cells in organotypic cultures is avb6-dependent and that such cultures represent a physiologically relevant model of carcinoma invasion (Nystrom et al., 2002). As Psor knockdown in VB6 cells was long-lasting (Figure 4a) we were able to test Psor RNA interference treated VB6 cells in an organotypic invasion assay over a 10-day assay period. Figures 4e and f show cytokeratin-stained sections from control (4E) and Psor (4F) RNA


1427

Figure 4 Suppression of Psor expression inhibits avb6-dependent invasion in oral squamous cell carcinoma and breast carcinoma cells. (a) Levels of Psor and b6 integrin expression 72 h and 14 days after transfection with Psor or control (Con) RNAi. The ability of VB6 cells to promote adhesion to (b) or migration towards (c) avb6 ligands fibronectin and LAP was measured after treatment with control or Psor RNAi. (d) VB6 invasion through Matrigel was assessed following RNAi-mediated knockdown of Psor. (b–d)—average of three or more separate experiments. (e, f) Psor is required for organotypic invasion by VB6 cells. Cells were transfected with control (e) or Psor (f) RNAi and plated for 10 days in organotypic cultures that model in vivo oral SCC invasion (Prime et al., 1990) Sections were stained for cytokeratin (brown stain). Data show that Psor knockdown abrogates VB6 organotypic gel invasion. Image is one of seven separate experiments showing similar results. Suppression of invasion was highly significant (Po0.001). (g, h) Transwell Matrigel invasion assays were performed with MDA MB468 breast carcinoma cells treated with anti-aVb6 blocking antibody (g) or transfected with control siRNA or Psor siRNA (h). Figure shows means and standard deviation of four separate experiments each performed with triplicate or quadruplicate samples.

interference-treated cells. Strikingly, analysis of the Mean Invasion Index (Nystrom et al., 2002) from seven independent separate experiments, showed that Psor knockdown consistently and significantly reduced invasion of VB6 oral SCC cells by 495% (Po0.001). Thus, Psor is required for avb6-mediated invasion of matrigel and in the more physiological organotypic assay. To determine whether the requirement of Psor for avb6-dependent invasion was only specific for oral SCC,

the breast carcinoma cell line MDA MB468 was also analysed as western blotting and flow cytometry revealed that these cells expressed both avb6 and Psor (data not shown). Figure 4g shows that antibody blockade of avb6 inhibits MDA MB468 invasion through Matrigel (P ¼ 0.002) showing that invasion is avb6 dependent. Treatment of MDA MB468 with siRNA to Psor significantly inhibited (P ¼ 0.017; Figure 4h) invasion of Matrigel compared with control Oncogene


1428

siRNA-treated cells. Thus Psor regulation of avb6dependent invasion is not restricted to oral SCC cells. Binding of Psor to the C-terminus of b6 is required for invasion As Psor associates with the C-terminal 11aa of the b6 subunit, and suppression of Psor expression inhibits invasion of VB6 cells, we generated a cell permeable peptide to mimic the C-terminus of b6 (Tat-WT) and a scrambled Random control (Tat-Ran) peptide. N-terminal-biotinylation permitted confirmation that

both peptides were taken up similarly by cells, using immunochemistry against the biotin tag (data not shown). We first established whether the peptides interacted directly with Psor. Thus recombinant Psor was immobilised on 96-well plates and probed with increasing concentrations of Tat-WT or Tat-Ran peptides. Figure 5a shows that the Tat-WT peptide exhibited a dose-dependent binding to Psor whereas the Tat-Ran did not bind at all. A similar result was obtained when a single concentration of Tat-WT or Tat-Ran peptides was used to probe dishes coated with various concentrations of recombinant Psor (Figure 5b).

Figure 5 Cell permeable peptides mimicking b6 C-terminus bind to Psor and inhibit carcinoma invasion (a) Purified recombinant Psor (10 mM) was immobilised on 96-well plates and binding of Biotinyl-Tat-WT or Biotinyl-Tat-Ran (0–10 mM) assessed. Bound Tatlinked peptides were detected using Extravidin peroxidase and developed with ABTS. (b) A range of concentrations of recombinant Psor (0–10 mM) were immobilised on 96-well plates and binding of Biotinyl-Tat-WT or Biotinyl-Tat-Ran (1 mM) assessed. Bound Tatlinked peptides were detected using Extravidin peroxidase and developed with ABTS. (c and d) VB6 cells on fibronectin, some pretreated with 10 mM Tat-WT or Tat-Ran peptides, were treated with DTBP and lysates immunoprecipitated with rat anti-b6 (620 W) or non-immune rat IgG. Immune complex-associated proteins were analysed by western blot (anti-psoraisin and anti-b6). Immunoblot is representative of four independent experiments and histogram shows densitometric quantification of four experiments (d). (e) VB6 cells were incubated for 1 h with various concentrations of Tat-WT or Tat-Ran peptide before testing the cells in a Matrigel invasion assay. Data show mean invasion±s.d. from quadruplicate wells and are representative of three independent experiments. (f) The avb6negative H357 parental line of VB6 was treated for 1 h with Tat-WT or Tat-Ran peptides before testing the cells in a Matrigel invasion assay. Data show mean invasion±s.d. from quadruplicate wells and are representative of three independent experiments. The Tat-WT peptide had no effect on the invasive ability of H357. TCL, total cell lysate. Oncogene


1429

Thus a peptide encoding the C-terminus of b6 cytoplasmic tail is sufficient to mediate direct binding to Psor. To confirm whether these peptides competitively inhibited Psor binding to cellular b6 we used a modified co-immunoprecipitation method. As standard immunoprecipitation of b6 followed by western blotting for Psor was repeatedly, but not consistently, successful (Figure 3b), suggesting that the binding of Psor to cellular b6 probably was relatively weak, for these experiments we used the membrane-permeable crosslinker dimethyl 3,30 -dithiobispropionimidate 2 HCl (DTBP). Consistent with the data presented in Figure 3b, immunoprecipitation of b6 co-precipitated endogenous Psor (Figure 5c). However, pre-incubation of cells with a membrane permeable Tat-WT peptide (corresponding to the C-terminus of b6), but not with the control Tat-Ran peptide, inhibited the binding

of Psor to b6 (Figures 5c and d). These data further confirm the 2D-gel data and show that endogenous Psor binds to the C-terminus of the b6 subunit. Moreover, the membrane-permeable peptide Tat-WT can bind to Psor intracellularly and thus functions as an inhibitor of the association between Psor and the b6 cytoplasmic tail. To determine whether Psor must bind directly to b6 for invasion to occur, VB6 cells were treated with various concentrations of either Tat-WT or Tat-Ran before assaying invasion of Matrigel. Figure 5e shows a dose-dependent reduction in invasion caused by Tat-WT treatment suggesting that Psor must bind directly to the b6 subunit for avb6-dependent invasion to occur. To confirm that the inhibition of invasion caused by TatWT was avb6 specific, the invasive, but avb6-negative, H357 parental cell line of VB6 (Thomas et al., 2001a, b)

Figure 6 Cell permeable peptides mimicking b6 C-terminus inhibit avb6-dependent invasion of oral and lung carcinoma cells. Invasion of CA1 oral SCC cells (a) and H441 lung carcinoma cells (c) was assessed in the presence or absence of an anti-avb6 blocking antibody. Both lines exhibited avb6-dependent invasion. CA1 (b) and H441 (d) were pre-treated with 10 mM Tat-WT or Tat-Ran peptides and invasion through Matrigel was measured. The Tat-WT peptide blocked invasion. (e) The highly invasive avb6-negative HT-1080 fibrocarcoma cells were treated with 10 mM Tat-WT or Tat-Ran peptides and invasion through Matrigel was measured. The Tat-WT peptide had no effect on invasion. For all invasion assays, data show mean invasion±s.d. from quadruplicate wells and are representative of at least three independent experiments. Oncogene


1430

Discussion

Figure 7 Expression of avb6 and Psor in human oral squamous cell carcinoma (OSCC) tissues. Sections of oral hyperplasia and oral squamous cell carcinoma were immunolabelled for b6 integrin subunit or psoriasin on consecutive serial sections. Images show that there was little or no labelling of hyperplastic tissue, whereas in carcinoma sections there was strong labelling of both b6 and Psor. The table shows the expression (negative/weak ( þ ), moderate ( þ þ ) or strong ( þ þ )) of avb6 and Psor on 11 oral cancer patient samples.

were also treated with Tat-WT or Tat-Ran before testing invasion. Figure 5f shows that treatment with either Tat peptide had no effect on Matrigel invasion by H357 cells. We extended these studies by testing additional cell lines. Figure 6 shows that the oral carcinoma cell line CA1 (Figure 6a) and the lung carcinoma cell line H441 (Figure 6c) both invade Matrigel in an avb6-dependent manner. However, treatment of CA1 (Figure 6b) and H441 (Figure 6d) with Tat-WT peptide, but not Tat-Ran peptide, inhibited the avb6-dependent invasion. In contrast, treatment with either peptide of the avb6-negative fibrosarcoma cell line HT1080 had no effect on its ability to invade Matrigel. These data confirm that the ability of the Psor-binding Tat-WT peptide to inhibit invasion is restricted to cells that invade in an avb6-dependent manner. b6 integrin subunit and Psor colocalise in oral SCC We examined a small panel of oral SCC tissues (n ¼ 11) and non-malignant oral hyperplasias (n ¼ 2) for expression of avb6 and Psor. Figure 7 shows that, whereas the hyperplastic tissues exhibited little or no staining for avb6 or Psor the carcinoma tissues expressed significant levels of these proteins. Overall 81% of the oral SCC expressed high levels of avb6 and 63% had high levels of Psor. Psor staining was predominantly cytoplasmic whereas avb6 was predominantly membranous. Although, as shown in the central images of Figure 7, some areas of tumour had similarly high levels of both proteins, in other areas the expression of each protein varied. In fact some tumour cells were Psor-positive and avb6-negative (data not shown) indicating that regulation of Psor expression is not controlled by avb6 expression. Oncogene

The need to identify novel targets for anti-cancer therapy remains compelling. Data indicating a potential role for avb6 (Ramos et al., 2002; Bates et al., 2005; Elayadi et al., 2007; Hazelbag et al., 2007), suggesting that this adhesion receptor constitutes a promising anticancer target. Experimentally avb6 has been shown to mediate carcinoma invasion in vitro (Thomas et al., 2001a, b; Ramos et al., 2002) and promote carcinoma growth (Agrez et al., 1994) and invasion (Nystrom et al., 2006) in vivo. Moreover, strong avb6 expression correlates with a substantial reduction in median survival from colon cancer, non-small-cell lung cancer and cervical cancer (Bates et al., 2005; Elayadi et al., 2007; Hazelbag et al., 2007). As patients die from cancer, principally, because of metastasis, it seems likely that strong avb6 expression is associated with carcinoma dissemination. Therefore, it is of critical importance to understand the molecular processes that govern avb6dependent invasion and tumour progression. In this study, we have shown that the unique C-terminal 11aa of b6 are essential for avb6-dependent invasion and have described a novel pro-invasive interaction that requires the binding of Psor to these C-terminal 11aa. Our data suggests strongly that the interaction of endogenous Psor and the b6 subunit is direct, however, it is also possible that the association is further modulated by one or more other, as yet unidentified, molecules. This would certainly help explain the apparently ‘weak’ association of Psor revealed in standard co-immunoprecipitation studies. Regardless of whether the binding of Psor to b6 is direct or indirect, we demonstrate clearly that inhibition of Psor expression, or disruption of the b6–Psor association, substantially reduces avb6-dependent carcinoma invasion. Integrins mediate their activities, in large part, by structural and signalling molecules binding to discrete areas of their cytoplasmic domains (Kumar, 1998). The unique 11 aa C-terminal sequence of the b6 subunit already has been reported to be required for a number of avb6-mediated biological behaviours (Agrez et al., 1994; Niu et al., 2002; Morgan et al., 2004) whereas signalling through the Src family kinase Fyn has been shown to promote oral cancer progression as a consequence of avb6 ligation (Li et al., 2003). We generated a cell line (V3B6D11aa) that expressed the same high levels of avb6 as VB6 oral SCC cells, which was composed of a wild-type av molecule associated with a truncation mutant of b6 that lacked the C-terminal 11aa. Removal of this b6-specific motif completely abrogated avb6dependent oral SCC invasion. Indeed, we have also demonstrated that these C-terminal 11 aas of b6 are sufficient to transfer a pro-invasive potential to a completely different integrin, avb3 (Morgan et al., 2004). We postulated that the mechanism by which these residues promoted invasion was through the binding of unidentified molecules to this region. Previously, using the yeast two-hybrid assay, we had identified the HAX1 protein as a binding partner of the


1431

b6 cytoplasmic tail (Ramsay et al., 2007). In the present report we utilised another approach to attempt to isolate additional binding partners of this integrin subunit. Using proteomics we found that Psor (S100A7) bound recombinant b6 cytoplasmic domains, but that this association was almost completely lost in the absence of the unique C-terminal 11aa. We further confirmed that Psor can bind directly to the b6 cytoplasmic tail. Psor is a member of the S100 family of EF-hand proteins, first identified as being upregulated in psoriatic skin (Madsen et al., 1992). (Note that avb6 also shows a significant increased expression on psoriatic skin versus non-lesional skin from psoriatic patients. Dr Shelia Violette, Stromedix Inc., Cambridge, MA, USA; personal communication and Patent Number AA61K39395FI). Since then it has been shown that this protein, which also is overexpressed in other inflammatory diseases (Al-Haddad, 1999), is secreted, and can act as a chemoattractant and as an antibacterial defence mechanism (Glaser et al., 2005; Wolf et al., 2008). Although Psor can function as a secreted molecule, in our studies we examined the activity of the intracellular pool. We showed that Psor colocalised with avb6 at the cell membrane and in intracellular vesicles in VB6 cells. Very similar subcellular distribution of Psor was reported by Ruse and colleagues who showed that epidermal fatty acid-binding protein colocalised with psoriasin at membrane protrusions of keratinocytes and also at sites of cell matrix adhesion; these adhesion sites also included the focal adhesion proteins actinin and paxillin and, although not looked for, almost certainly integrins. In separate studies, the induction of anoikis, programmed cell death induced by loss of integrindependent adhesion was associated with the downregulation of Psor (Petersson et al., 2007). Thus Psor may regulate integrin-related processes other than those described here. Interestingly, like avb6, psoriasin expression has been reported to be up-regulated in epithelial cells of various oral squamous carcinomas (Banerjee et al., 2005; Zhou et al., 2008; Ralhan et al., 2008; Kesting et al., 2009) and subsets of breast cancers (Al-Haddad et al., 1999; Jiang et al., 2004; Emberley et al., 2004, 2005; Krop et al., 2005; Petersson et al., 2007; Wang et al., 2008; Paruchuri et al., 2008). The precise role of Psor in tumour behaviour is, however, unclear. In breast cancer, Psor was identified as being more frequently and more highly expressed in ductal carcinoma in situ than in invasive disease; that is, exhibiting an inverse correlation with tumour progression (Krop et al., 2005). Conversely, others have reported that the presence of the protein correlates directly with the degree of invasiveness of breast carcinoma cells (Al-Haddad et al., 1999; Jiang et al., 2004, Emberley et al., 2004). In SCCs of the oral cavity, recent reports suggest Psor expression correlates with more differentiated cancer phenotypes (Zhou et al., 2008; Kesting et al., 2009). Confusingly, Psor is reported to both promote survival through Jab1-activated phosphorylation of Akt (Emberley et al., 2005) and also suppress tumour progression by binding in a complex with b-catenin and promoting its degradation

and suppressing Wnt signalling (Zhou et al., 2008). Here we provide novel insight into the mechanism by which Psor can function in cancer, through interaction with avb6 integrin. In summary, it is clear that Psor can interact directly with the cytoplasmic tail of the integrin b6 subunit, binding to the C-terminal 11 aa. This binding promotes the avb6-dependent invasive activity of oral carcinoma cells and that intracellular delivery of peptides capable of disrupting this interaction led to a dramatic reduction in invasive behaviour. We further showed that these observations are not restricted to oral SCC as Psor also is required for avb6-dependent invasion of a breast and a lung carcinoma cell line. Thus in avb6-expressing tumours the targeting of Psor-mediated invasive activity represents a possible novel therapeutic approach and underscores the need for greater investigation into the role of this calcium-binding EF-hand family member in tumour biology. Materials and methods Cell lines and antibodies The oral SCC cell line V3 is a low avb6-expressing cell line derived by transfection of av complementary DNA into the av-negative oral SCC cell line H357 (Prime et al., 1990; Sugiyama et al., 1993; Jones et al., 1996) V3 was infected with pBabepuro retroviruses encoding the puromycin resistance gene alone to create C1 cells or, additionally, human b6 complementary DNA to create the VB6-cell population (Thomas et al., 2001a). Several mouse monoclonal antibodies were used. Anti-MHC Class I (W6/32) was provided by Professor Sir Walter Bodmer (University of Oxford). Antibodies to integrins a5b1 (P1D6) and avb6 (E7P6 and 10D5) were purchased from Chemicon International (Herts, UK). Mouse anti-b6 (71C5) was a gift from Dr Shelia Violette (Stromedix Inc.) and mouse anti-Psor (47C1068) was purchased from Abcam (Cambridge, UK). Rat monoclonal anti-b6 antibodies 620W, 310W and 53a2 were produced in-house. Goat anti-b6 antiserum (C19), purified and horseradish peroxidase-conjugated anti-GST (Z-5) were purchased from Santa Cruz Biotechnology, Inc. (Heidelberg, Germany). Mouse monoclonal anti-HAX1 (Clone 52) was purchased from BD Biosciences (Oxford, UK). Generating an oral SCC cell population expressing truncated b6 Using standard molecular techniques, the wild-type b6 sequence in pBabepuro (Thomas et al., 2001a) was used as a PCR template to generate a complementary DNA encoding a truncation mutant of the b6 subunit lacking the C-terminal 11 residues. A myc/his tag was encoded in place of the unique C-terminal sequence (Figure 1a). 50 EcoR1 and 30 Sal1 restriction sites were included in the complementary DNA to allow ligation of the insert into pBabepuro. Forward primer sequence: 50 -GCCACGGAATTCGCACAGCAAGAACTGA AACGAATGGGGATTGAACTGCTTTGCCTGTTCTTTC TATTTCTAGGAAGG-30 . Reverse primer sequence: 50 -CGA TGCGTCGACTCAATGGTGATGGTGATGATGACCGG TATGCATATTCAGATCCTCTTCTGAGATGAGTTTTT GTTCGAACCGCGGCTGTGTTTATAAGTTACATTTTT AAAAGTACTTGTGGATCCTCTGTAGAGTGGATTGG TTCCCGTTTGCCA-30 . Oncogene


1432 Retroviruses were generated (as described in Thomas et al., 2001a) and used to infect V3 cells. The puromycin-resistant cells were treated with anti-avb6 antibodies (E7P6 and 10D5) and sorted using magnetic beads (sheep anti-mouse IgG; Dynal AS, Oslo, Norway) to create the V3B6D11aa line with levels of avb6 expressed at the cell surface similar to those of VB6 cells (Jiang et al., 2004). Expression levels of avb6 integrins routinely were measured by flow cytometry (Figure 1b) at various time points throughout the experiments. Flow cytometry Cells were trypsinized and resuspended in phosphate-buffred saline (PBS) /0.1% bovine serum albumin (BSA)/0.1% NaN3 (0.1/0.1). E7P6 (10 mg/ml) was added for 45 min. Bound antibody was detected with fluorescein isothiocyanate (FITC)conjugated anti-mouse IgG (DAKO, Ely, UK) and cells analysed on a FACScalibur (Becton-Dickinson, Oxford, UK). Adhesion assay Briefly, chromium-[51Cr]-labelled cells were allowed to attach to substrate-coated 96-well plates for 45–60 min at 37 1C. After washing away unbound cells, adhesion was determined from the residual radioactivity associated with the wells. Migration assay The full protocol is described elsewhere (Thomas et al., 2001b). The undersides of Transwell (8 mm pore, Corning Inc., Corning, NY, USA) chambers were coated with 10 mg/ml human fibronectin, 0.5 ug/ml LAP or, as a control, 0.1% BSA (Sigma, Poole, UK). Cells (1 105 in 100 ml migration buffer) were added to the upper chamber and after 16 h at 37 1C all cells were harvested by trypsinization and counted on a CASY-1 cell counter (Scharfe Systems, Reutlingen, Germany). Invasion assays Invasion through Matrigel-coated Transwell filters was as described (Jones et al., 1996). Where appropriate, levels of invasion were normalised to control invasion levels. In antibody inhibition experiments, invasion was expressed relative to invasion in the presence of the negative control antibody, W6/32. Organotypic invasion assay gels were prepared, and invasive activity measured exactly as described previously (Thomas et al., 2006). Peptide synthesis Cell-permeable peptides were synthesized using solid-phase peptide synthesis by the Cancer Research UK Peptide Synthesis laboratory. Peptide sequences: Tat-WT GRKKRR QRRRPPQKHREKQKVDLSTDC Tat-Ran GRKKRRQR RRPPQSQVTCKRKDLHKED. Crude peptides were analysed by reverse phase high performance liquid chromatography and matrix assisted laser desorption ionisation time of flight mass spectroscopy then purified by reverse phase high performance liquid chromatography on an Aquapore ODS 20 micron 250 10 mm column. Some peptides were biotinylated in situ on resin support using standard procedures. Additional N-terminally biotinylated Tat-WT and Tat-Ran peptides were purchased from Bachem (Bachem Distribution Services GmbH, Weil am Rhein, Germany). All peptides were 495% pure. GST fusion protein constructs PCR primers were designed to amplify the full length b6 tail and an 11 aa truncated b6 tail with the addition of a 50 EcoRI Oncogene

and a 30 SalI site, shown in bold text in the primer sequences below. Primers used were as follows: Full length b6 tail Forward: 50 -GGGGAATTCCCAAGCT ACTGGTGTCATTTC-30 , Full length b6 tail Reverse: 50 -GG AGTCGACCTAGCAATCTGTGGAAAGGTC-30 , Truncated b6 tail Forward: 50 -GGGGAATTCCCAAGCTACTGGTGTC ATTTC-30 , Truncated b6 tail Reverse:50 -GGAGTCGACCTA CCTGTGTTTATAAGTTACATT-30 . Using standard molecular biology techniques, b6 PCR products were generated and ligated into pGEX-KG vector (Clontech-Takara Bio Europe, Saint-Germain-en-Laye, France). An additional construct containing GST alone was generated for use as a GST protein control. GST-fusion proein constructs were generated in BL21 bacteria upon 0.1 mM isopropyl-beta-D-thiogalactopyranoside induction and, following sonication and incubation with 1% (w/v) Triton X-100, purified using pre-washed glutathione sepharose beads (Sigma). Immobilised GST-fusion proteins were washed and stored at 80 1C in PBS or 4 1C in NP40 lysis buffer (40 mM HEPES pH7.8, 2% (v/v) NP40, 100 mM NaCl, 500 mM CaCl2 and 500 mM MgCl2) supplemented with 0.1% sodium azide. For solid phase assays, purified GST fusion proteins were eluted from sepharose beads with reduced glutathione (7.5 mM) for 90 min and stored at 80 1C. GST fusion protein pull-down assays Equal quantities (by mass) of each GST construct immobilised on glutathione sepharose beads were used to perform pulldowns from lysates of VB6 cells. Serum-starved VB6 cells were plated on petri-dishes (pre-coated with 5 mg/ml fibronectin) for 2 h before being scraped and lysed in NP40 lysis buffer for 30 min on ice then centrifuged at 14 000 g for 10 min at 4 1C. Lysates were pre-cleared twice for 1 h by tumbling with GSTalone beads then distributed equally to GST-, GSTb6WT- and GSTb6D11aa-coated beads for overnight tumbling at 4 1C. Bound complexes were washed with lysis buffer, and eluted in reducing sample buffer at 95 1C for 10 min. Proteins were detected by 2D electrophoesis and mass spectrometry or by SDS–polyacrylamide gel electrophoresis and immunoblotting (as described in co-immunoprecipitation section). 2D gel electrophoresis 2D-protein analysis was performed with the Cancer Research UK Proteomics service. Isolectric focussing utilised 13 cm pH 3-10 IEF strips (Immobiline DryStrip (IPG), GE Healthcare, Chalfont St Giles, Buckinghamshire, UK) on the Ettan IPGphor IEF System (GE Healthcare). Each 13 cm strip was then placed onto 16 cm 1.5 mm SDS–polyacrylamide gel electrophoresis gels (10% acrylamide). Gels were run at 15 1C, 6 mA per gel overnight. Fixed gels were stained with Coomassie Blue 2D and individual spots excised for mass spectrometry analysis. Detailed protocols are available upon request. Identification of proteins by mass spectrometry Protein identification was performed by the peptide analysis services laboratory at Cancer Research UK (Lincolns Inn Fields, London). Briefly, peptides eluted from spots were digested in trypsin and the peptide fragments analysed using a MALDI ABI 4700 TOF-TOF instrument. Peptide masses were then searched against databases to identify proteins. Co-immunoprecipitation of Psor and b6 Two methods were used. First, NP40 lysates from VB6 cells were immunoprecipiated with either antibodies to b6 (71C5) or Psor (47C1068). The immunoprecipitates were transferred to


1433 nitrocellulose and probed for Psor (47C1068) or b6 (C19), respectively. Although Psor co-precipitated repeatedly the amount of Psor precipitated was variable, suggesting a weak interaction. Thus, we also employed a cross-linker method. VB6 cells were serum-starved for 16 h, where appropriate treated with 10 mM Tat-peptides in aMEM (Invitrogen, Paisley, UK) in suspension for 1 h at 37 1C, and plated on fibronectincoated dishes (5 mg/ml) for 90 min at 37 1C. Cells were treated with the membrane permeable, thiol-cleavable, crosslinker DTBP (Perbio, Thermo Fisher Scientific, p/a Perbio Science UK, Ltd, Cramlington, Northumberland, UK) at 3 mM for 5 min at 37 1C, and washed twice with ice-cold TBS to quench the cross-linker before lysis. Cells were lysed with a buffer containing 1% Igepal CA-630, 20 mM Tris-HCl pH 7.5 (to further quench the cross-linker), 40 mM HEPES, 100 mM NaCl, 2 mM CaCl2, 6 mM MgCl2, 0.6 M Sucrose and protease inhibitors. After centrifugation at 2000 g for 10 min, lysates were pre-cleared with 40 ml recombinant Protein G sepharose beads (Zymed Laboratories, supplied by Invitrogen, Pasley, UK) twice for 30 min at 4 1C. Cleared lysates were rotated overnight at 4 1C with 30 ml Protein G beads pre-coated with 5 mg of either 620 W (anti-avb6) or rat IgG2b. Beads were washed four times with lysis buffer and proteins eluted with 45 ml reducing SDS sample buffer at 95 1C for 10 min. For coimmunoprecipitation of b6 and b6D11aa, from VB6 and V3B6D11aa cells, with recombinant Psor, cells were not treated with DTBP before lysis and the cleared lysate was supplemented with 1.1 ng recombinant Psor before incubation with antibodybound beads. Proteins were resolved by SDS–polyacrylamide gel electrophoresis on 4–12% Bis-Tris Novex gels (Invitrogen) and transferred to nitrocellulose. Immune complex-associated proteins were detected using the Odyssey infrared imaging system (LI-COR Biosciences UK Ltd, Cambridge, UK) and band intensity quantified using Odyssey 2.1 software. Solid-phase assays of Psor and b6 interaction The specificity of protein–protein interactions was assessed by solid-phase assays. To test association of Psor with GSTfusion proteins, 96-Well plates (Costar ½-area EIA/RIA, Corning Inc.) were coated with 10 mg/ml GST-b6, GSTb6D11aa, GST or BSA in PBS for 90 min at ambient temperature (RT). Wells were washed three times with PBS and blocked with 5% (w/v) BSA in PBS for 1 h at RT. After three washes with binding buffer (0.1% BSA in PBS), 0–10 mM recombinant Psor (Thomas et al., 2001c) as added and plates incubated for 2 h at RT. Plates were washed four times with binding buffer and bound Psor detected with a mouse monoclonal anti-Psor antibody (1 mg/ml for 30 min at RT followed by anti-mouse horseradish peroxidase (1 mg/ml for 30 min at RT). Peroxidase was developed with 2,20 -azinobis(3ethylbenzothiazoline-6-sulfonic acid) substrate (50 ml per well). Absorption at 405 nm was measured using a BioTek PowerWave 340 plate reader. Alternatively, plates were coated with 10 mM recombinant Psor or BSA and probed with 0–10 mg/ml GST-b6, GST-b6D11aa or GST. Bound GST-fusion proteins were detected with horseradish peroxidase-conjugated

anti-GST antibody (0.5 mg/ml) and developed with 3-ethylbenzothiazoline-6-sulfonic acid. For all experiments, background binding to wells coated with BSA was subtracted from measurements. Normalised mean absorbance of triplicate wells±s.d. are presented. Hyperbolic regression curves were fitted using Sigma Plot 8.0. All data presented are representative of three independent experiments. To test binding of Tat-linked peptides to Psor, a similar approach was employed with the following exceptions. Plates were coated with 10 mM recombinant Psor or BSA and probed with 0–10 mM Biotinyl-Tat-WT or Biotinyl-Tat-Ran. Alternatively, wells were coated with increasing concentrations of Psor (0–10 mM) and probed with 1 mM Biotinyl-Tat-WT or BiotinylTat-Ran. All washing steps were carried out six times with 0.5% Tween in PBS. Bound biotinyl-peptides were detected with Extravidin Peroxidase (Sigma; 1:500 for 20 min) before 3-ethylbenzothiazoline-6-sulfonic acid development. siRNA mediated RNA interference in cultured cell lines Custom SMARTpool siRNA reagents targeting Psor, b6 integrin subunit or control (non-targeting siRNA) were purchased from Dharmacon RNA Technologies (Chicago, IL, USA), resuspended in (1 ) siRNA buffer (100 mM KCl, 30 mM HEPES-pH 7.5, 1.0 mM MgCl2) to a 20 mM stock solution. At 24 h before transfection, cells were seeded in 6-well plates (100 000 cells per well). Cells were transfected with 100 nM RNA interference using Oligofectamine reagent (Invitrogen) according to the manufacturer’s instructions. After 4 h at 37 1C, cells were supplemented with serumcontaining growth media. Proteins were separated by SDS– polyacrylamide gel electrophoresis analysed by immunoblotting for psoraisin (1:1000) or b6 (goat anti-b6; 1:500) using standard methods (detailed protocols available by request). Loading was checked by stripping gels (Chemicon Blot Recycling Kit; Chemicon International, Millipore (UK) Ltd, Watford, UK) and re-probing for HSC70. Immunolabelling of human oral tissues A small series of 4 mm sections from formalin-fixed paraffinembedded human oral SCC tissues (n ¼ 11) and oral hyperplasias (n ¼ 2) were de-waxed, treated with pepsin (Digest-All; Zymed) for 50 at 37 1C and immunolabelled using standard avidin–biotin technique for b6 (6.2G2; a kind gift from Stromedix Inc.) and Psor (AbCam antibody 47C1068). Labelling with control, class-matched antibody showed no staining (data not shown). Protein expression was scored as negative/weak ( þ ), moderate ( þþ ) or strong ( þþ þ ). Statistical analysis Statistical analyses used Student’s t-test or analysis of varience as required. Conflict of interest The authors declare no conflict of interest.

References Agrez M, Chen A, Cone RI, Pytela R, Sheppard D. (1994. The alpha v beta 6 integrin promotes proliferation of colon carcinoma cells through a unique region of the beta 6 cytoplasmic domain. J Cell Biol 127: 547–556. Al-Haddad S, Zhang Z, Leygue E, Snell L, Huang A, Niu Y et al. (1999). Psoriasin (S100A7) expression and invasive breast cancer. Am J Pathol 155: 2057–2066.

Banerjee AG, Bhattacharyya I, Vishwanatha JK. (2005). Identification of genes and molecular pathways involved in the progression of premalignant oral epithelia. Mol Cancer Ther 4: 865–875. Bates RC, Bellovin DI, Brown C, Maynard E, Wu B, Kawakatsu H et al. (2005). Transcriptional activation of integrin beta 6 during the epithelial-mesenchymal transition defines a novel prognostic indicator of aggressive colon carcinoma. J Clin Invest 115: 339–347. Oncogene


1434 Breuss JM, Gallo J, DeLisser HM, Klimanskaya IV, Folkesson HG, Pittet JF et al. (1995). Expression of the beta 6 integrin subunit in development, neoplasia and tissue repair suggests a role in epithelial remodelling. J Cell Sci 108: 2241–2251. Breuss JM, Gillett N, Lu L, Sheppard D, Pytela R. (1993). Restricted distribution of integrin beta 6 mRNA in primate epithelial tissues. J Histochem Cytochem 41: 1521–1527. Busk M, Pytela R, Sheppard D. (1992). Characterization of the integrin alpha v beta 6 as a fibronectin-binding protein. J Biol Chem 267: 5790–5796. Clark RA, Ashcroft GS, Spencer MJ, Larjava H, Ferguson MW. (1996). Re-epithelialization of normal human excisional wounds is associated with a switch from alpha v beta 5 to alpha v beta 6 integrins. Br J Dermatol 135: 46–51. Elayadi AN, Samli KN, Prudkin L, Liu YH, Bian A, Xie XJ et al. (2007). A peptide selected by biopanning identifies the integrin alpha v beta 6 as a prognostic biomarker for non small cell lung cancer. Cancer Res 67: 5889–5895. Emberley ED, Murphy LC, Watson PH. (2004). S100A7 and the progression of breast cancer. Breast Cancer Res 6: R308–R315. Emberley ED, Niu Y, Curtis L, Troup S, Mandal SK, Myers JN et al. (2005). The S100A7-c-jun activation domain binding protein 1 pathway enhances prosurvival pathways in breast cancer. Cancer Res 65: 5696–5702. Glaser R, Harder J, Lange H, Bartels J, Christophers E, Schroder JM. (2005). Antimicrobial psoriasin (S100A7) protects human skin from Escherichia coli infection. Nat Immunol 6: 57–64. Haapasalmi K, Zhang K, Tonnesen M, Olerud J, Sheppard D, Salo T et al. (1996). Keratinocytes in human wounds express alpha v beta 6 integrin. J Invest Dermatol 106: 42–48. Hazelbag S, Kenter GG, Gorter A, Dreef EJ, Koopman LA, Violette SM et al. (2007). Overexpression of the alpha v beta 6 integrin in cervical squamous cell carcinoma is a prognostic factor for decreased survival. J Pathol 212: 316–324. Jiang WG, Watkins G, Douglas-Jones A, Mansel RE. (2004). Psoriasin is aberrantly expressed in human breast cancer and is related to clinical outcomes. Int J Oncol 25: 81–85. Jones J, Sugiyama M, Speight PM, Watt FM. (1996). Restoration of alpha v beta 5 integrin expression in neoplastic keratinocytes results in increased capacity for terminal differentiation and suppression of anchorage-independent growth. Oncogene 12: 119–126. Jones J, Watt FM, Speight PM. (1997). Changes in the expression of alpha v integrins in oral squamous cell carcinomas. J Oral Pathol Med 26: 63–68. Kesting MR, Sudhoff H, Hasler RJ, Nieberler M, Pautke C, Wolff KD et al. (2009). Psoriasin (S100A7) up-regulation in oral squamous cell carcinoma and its relationto clinicopathologic features. Oral Oncol 45: 731–736. Krop I, Marz A, Carlsson H, Li X, Bloushtain-Qimron N, Hu M et al. (2005). A putative role for psoriasin in breast tumour progression. Cancer Res 65: 11326–11334. Kumar CC. (1998). Signaling by integrin receptors. Oncogene 17: 1365–1373. Larjava H, Haapasalmi K, Salo T, Wiebe C, Uitto VJ. (1996). Keratinocyte integrins in wound healing and chronic inflammation of the human periodontium. Oral Dis 2: 77–86. Leoń R, Murray JI, Cragg G, Farnell B, West NR, Pace TC et al. (2009). Identification and characterization of binding sites on S100A7, a participant in cancer and inflammation pathways. Biochemistry 48: 10591–10600. Li X, Yang Y, Hu Y, Dang D, Regezi J, Schmidt BL et al. (2003). avb6-Fyn signalling promotes oral cancer progression. J Biol Chem 278: 41646–41653. Liotta LA, Kohn EC. (2001). The microenvironment of the tumourhost interface. Nature 411: 375–379. Madsen P, Rasmussen HH, Leffers H, Honore´ B, Dejgaard K, Olsen E et al. (1992). Molecular cloning, occurrence, and expression of a novel partially secreted protein ‘psoriasin’ that is highly up-regulated in psoriatic skin. J Invest Dermatol 97: 701–712. Oncogene

Morgan MR, Thomas GJ, Russell A, Hart IR, Marshall JF. (2004). The integrin cytoplasmic-tail motif EKQKVDLSTDC is sufficient to promote tumour cell invasion mediated by matrix metalloproteinase (MMP)-2 or MMP-9. J Biol Chem 279: 26533–26539. Munger JS, Huang X, Kawakatsu H, Griffiths MJ, Dalton SL, Wu J et al. (1999). The integrin alpha v beta 6 binds and activates latent TGF beta 1: a mechanism for regulating pulmonary inflammation and fibrosis. Cell 96: 319–328. Niu J, Dorahy DJ, Gu X, Scott RJ, Draganic B, Ahmed N et al. (2002). Integrin expression in colon cancer cells is regulated by the cytoplasmic domain of the beta 6 integrin subunit. Int J Cancer 99: 529–537. Niu J, Gu X, Ahmed N, Andrews S, Turton J, Bates R et al. (2001). The alpha v beta 6 integrin regulates its own expression with cell crowding: implications for tumour progression. Int J Cancer 92: 40–48. Nystrom ML, McCulloch D, Weinreb PH, Violette SM, Speight PM, Marshall JF et al. (2006). Cyclooxygenase-2 inhibition suppresses apha v beta 6 integrin-dependent oral squamous carcinoma invasion. Cancer Res 66: 10833–10842. Nystrom ML, Thomas GJ, Stone M, Mackenzie IC, Hart IR, Marshall JF. (2002). Development of a quantitative method to analyse tumour cell invasion in organotypic culture. J Pathol 205: 468–475. Paruchuri V, Prasad A, McHugh K, Bhat HK, Polyak K, Ganju RK. (2008). S100A7-downregulation inhibits epidermal growth factorinduced signalling in breast cancer cells and blocks osteoclast formation. PLoS One 3: e1741. Petersson S, Bylander A, Yhr M, Enerba¨ck C. (2007). S100A7 (Psoriasin), highly expressed in Ductal Carcinoma In situ (DCIS), is regulated by IFN-gamma in mammary epithelial cells. BMC Cancer 7: 205. Prieto AL, Edelman GM, Crossin KL. (1993). Multiple integrins mediate cell attachment to cytotactin/tenascin. Proc Natl Acad Sci USA 90: 10154–10158. Prime SS, Nixon SV, Crane IJ, Stone A, Matthews JB, Maitland NJ et al. (1990). The behaviour of human oral squamous cell carcinoma in culture. J Pathol 160: 259–269. Ralhan R, DeSouza LV, Malta A, Tripathi SC, Ghanny S, Gupta SD et al. (2008). Discovery and verification of head-and-neck cancer biomarkers by differential protein expression analysis using iTRAQ labelling, multidimensional liquid chromatography, and tandem mass spectrometry. Mol Cell Proteomics 7: 1162–1173. Ramos DM, But M, Regezi J, Schmidt BL, Atakilit A, Dang D et al. (2002). Expression of integrin beta 6 enhances invasive behaviour in oral squamous cell carcinoma. Matrix Biol 21: 297–307. Ramsay AG, Keppler MD, Jazayeri M, Thomas GJ, Parsons M, Violette S et al. (2007). HS1-associated protein X-1 regulates carcinoma cell migration and invasion via clathrin-mediated endocytosis of integrin alpha v beta 6. Cancer Res 67: 5275–5284. Regezi JA, Ramos DM, Pytela R, Dekker NP, Jordan RC. (2002). Tenascin and beta 6 integrin are overexpressed in floor of mouth in situ carcinomas and invasive squamous cell carcinoma. Oral Oncol 38: 332–336. Reska AA, Hayashi Y, Horwitz AF. (1992). Identification of amino acid sequences in the integrin beta 1 cytoplasmic domain implicated in cytoskeletal association. J Cell Biol 117: 1321–1330. Ruse M, Broome AM, Eckert RL. (2003). S100A7 (psoriasin) interacts with epidermal fatty acid binding protein and localizes in focal adhesion-like structures in cultured keratinocytes. J Invest Dermatol 121: 132–141. Sugiyama M, Speight PM, Prime SS, Watt FM. (1993). Comparison of integrin expression and terminal differentiation capacity in cell lines derived from oral squamous cell carcinomas. Carcinogenesis 14: 2171–2176. Thomas GJ, Hart IR, Speight PM, Marshall JF. (2002). Binding of TGF-beta 1 latency-associated peptide (LAP) to alpha(v) beta 6


1435 integrin modulates behaviour of squamous carcinoma cells. Br J Cancer 87: 859–867. Thomas GJ, Lewis MP, Hart IR, Marshall JF, Speight PM. (2001a). Alpha v beta 6 integrin promotes invasion of squamous carcinoma cells through up-regulation of matrix metalloproteinase 9. Int J Cancer 92: 641–650. Thomas GJ, Lewis MP, Whawell SA, Russell A, Sheppard D, Hart IR et al. (2001b). Expression of the alpha v beta 6 integrin promotes migration and invasion in squamous carcinoma cells. J Invest Dermatol 117: 67–73. Thomas GJ, Nystrom ML, Marshall JF. (2006). Alpha v beta 6 integrin in wound healing and cancer of the oral cavity. J Oral Pathol Med 35: 1–10. Thomas GJ, Poomsawat S, Lewis MP, Hart IR, Speight PM, Marshall JF. (2001c). Alpha v beta 6 integrin upregulates matrix metalloproteinase 9 and promotes migration of normal oral keratinocytes. J Invest Dermatol 116: 898–904.

Wang J, Barnes RO, West NR, Olson M, Chu JE, Watson PH. (2008). Jab1 is a target of EGFR signalling in ERalpha-negative breast cancer. Breast Cancer Res 10: R51. Weinacker A, Chen A, Agrez M, Cone RI, Nishimura S, Wayner E et al. (1994). Role of integrin alpha v beta 6 in cell attachment to fibronectin. Heterologous expression of intact and secreted forms of the receptor. J Biol Chem 269: 6940–6948. Wolf R, Howard OMZ, Dong H-F, Voscopoulos C, Boeshans K, Winston J et al. (2008). Chemotactic activity of S100A7 (Psoriasin) is mediated by the receptor for advanced glycation end products and potentiates inflammation with highly homologous but functionally distinct S100A15. J Immunol 181: 1499–1506. Zhou G, Xie TX, Zhao M, Jasser SA, Younes MN, Sano D et al. (2008). Reciprocal negative regulation between S100A7/psoriasin and beta-catenin signalling plays an important role in tumour progression of squamous cell carcinoma of oral cavity. Oncogene 27: 3527–3538.

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

Oncogene