Lysyl Oxidase Enzymatic Function Increases Stiffness to Drive Colorectal Cancer Progression Through FAK

SHORT COMMUNICATION Lysyl oxidase enzymatic function increases stiffness to drive colorectal cancer progression through FAK

A-M Baker1, D Bird1,2, G Lang1,2,TRCox1,2,3,4 and JT Erler1,2,3,4

The extracellular, matrix-modifying enzyme lysyl oxidase (LOX) has recently been linked to colorectal cancer (CRC) progression, in particular to the stages of invasion and metastasis. In this report, we use cell lines expressing a catalytically inactive mutant form of LOX to show that catalytic activity is required for LOX-mediated effects on proliferation and invasion in both in vitro and in vivo models of CRC. Furthermore, we use rheology to measure the relative stiffness of modified collagen matrices and subcutaneous tumors, and show that LOX-induced collagen cross-linking results in stiffening of the matrix both in vitro and in vivo. We observe a strong association between matrix stiffness and activation of the FAK (focal adhesion kinase)/SRC-signaling pathway, with a stiffer environment resulting in increased FAK/SRC phosphorylation and a more proliferative and invasive phenotype. We are the first to show a direct relationship between LOX enzymatic activity and tissue stiffness, and to demonstrate a role for stiffness in driving CRC progression. Our findings provide significant evidence to suggest that therapeutic inhibition of LOX activity may provide a novel effective treatment option for patients with metastatic CRC.

Oncogene (2013) 32, 1863–1868; doi:10.1038/onc.2012.202; published online 28 May 2012 Keywords: colorectal cancer; metastasis; lysyl oxidase; focal adhesion kinase

INTRODUCTION Early studies into the role of LOX in CRC gave conflicting results. 10 Cancer progression involves complex, dynamic interactions Csiszar et al. showed decreased LOX mRNA expression levels in between tumor cells and components of the surrounding CRC patients with non-metastatic disease, suggesting a role for extracellular matrix (ECM). The biomechanical and biochemical LOX as a tumor suppressor. However, a more recent study showed properties of the ECM are fundamental to the efficiency of that LOX mRNA expression was associated with a diffuse invasion and metastasis.1 In particular, the stiffness or elasticity of cytoplasmic expression of CEA (carcinoembryonic antigen) in the cellular environment has been shown to have a critical role in CRC patients suggesting LOX was associated with increased 2 invasive potential.11 Our previous study confirmed a role for LOX cancer progression. 12 Tumor cells have the potential to modify their environment in a in CRC cancer progression. Here, we now report that LOX number of ways. A notable example is the secretion of lysyl catalytic activity significantly increases tissue stiffness in a CRC oxidase (LOX), which catalyzes the covalent cross-linking of model, and use clinically relevant inhibitors to show that this collagens and elastin in the ECM, thereby increasing tissue contributes to tumor proliferation and metastasis in a FAK/SRC- 3 dependent manner. stiffness and enhancing tumor progression. Elevated LOX expression has been associated with invasion and metastasis in a number of solid tumors, notably breast,4 colorectal5 and head and neck.6 Inhibition of LOX has shown considerable promise in a RESULTS AND DISCUSSION number of preclinical models and has subsequently become a LOX activity promotes proliferation and increases stiffness of subcutaneous tumors novel target for therapeutic intervention. A central component of the signaling cascades responsible for To determine the effect of LOX activity on tumor stiffness, the CRC transducing extracellular mechanical force to changes in gene cell lines SW480, HT29 and LS174T were stably transfected with an expression within the cell is focal adhesion kinase (FAK).7 As such empty vector control, full-length LOX, or a catalytically inactive full-length LOX (containing the mutation K320A) as previously it has been studied intensively in the context of cancer, and has 5 7 described. Overexpression of mutant LOX resulted in an increase emerged as a notable target for drug development. FAK in protein expression (Supplementary Figure 1a) but no change in activation is closely associated with phosphorylation of SRC amine oxidase activity (Supplementary Figure 1b). These cell lines tyrosine kinase, with signaling through these kinases being were implanted into nude mice and allowed to grow as 8 frequently upregulated during the progression of cancer. SRC subcutaneous tumors. Consistent with our previous findings12 has been shown to be critical for LOX-mediated invasion of and previous work by Pez et al.,13 the tumors overexpressing LOX colorectal cancer (CRC) cells,5 and both SRC and FAK are grew significantly faster than control tumors (Figure 1a and implicated in LOX-mediated invasion of breast cancer cells.4,9 Supplementary Figure 1c). However, where the tumors were

1Hypoxia and Metastasis Team, Division of Cancer Biology, The Institute of Cancer Research, London, UK; 2CRUK Tumour Cell Signalling Unit, Division of Cancer Biology, The Institute of Cancer Research, London, UK and 3Biotech Research and Innovation Centre (BRIC), University of Copenhagen, Ole Maaløes Vej 5, Copenhagen, Denmark. Correspondence: Dr JT Erler, Biotech Research and Innovation Centre (BRIC), University of Copenhagen, Ole Maaløes Vej 5, Copenhagen N 2200, Denmark. Email: [email protected] 4These authors contributed equally to this work Received 9 December 2011; revised 19 March 2012; accepted 3 April 2012; published online 28 May 2012 Lysyl oxidase activity in tumor progression A-M Baker et al 1864

Figure 1. LOX activity leads to increased tumor stiffness. (a) Growth of SW480 cells as subcutaneous tumors in nude mice. Cells were stably transfected with an empty vector (‘Control’), full-length LOX (‘LOX’) or a catalytically inactive full-length LOX (‘mut LOX’) as previously described.5 In all, 4 Â 106 cells in 100 ml PBS were implanted subcutaneously into each flank of 6–8-week-old female MF1 nude mice (Harlan, Derby, UK; n ¼ 3 mice per group) and tumor growth was measured weekly with calipers. Bars represent mean±s.e.m. (b) Stiffness of SW480 subcutaneous tumors as measured by rheology. Briefly, 8 mm diameter, 1 mm thick tumor biopsies were measured over a range of 0.5–1.25% strain at a fixed angular frequency of 0.5 rad/s and temperature of 21 1C on an ARG2 controlled strain rotational rheometer (TA Instruments). The samples were found to be only minimally frequency dependent within the range of testing and showed linear viscoelastic response within the strain range evaluated. Bars represent mean±s.e.m., n ¼ 4 tumors per cell line. *Po0.05 using the two-sided Student’s t-test. (c) Collagen content of SW480 subcutaneous tumors. The total collagen content of fresh tumor samples was analyzed using the Sircol assay (BioColor Ltd) according to the manufacturers instructions. Bars represent mean±s.e.m.

expressing catalytically inactive LOX, the tumors grew at a similar to LOX-mediated modification and in particular stiffening of the rate to (in the cases of HT29 and LS174T) or even significantly surrounding collagen matrix, we pre-treated collagen matrices slower than the control tumors (in the case of SW480). These data with either recombinant human LOX (huLOX) or ribose, which is suggest that the catalytic activity of LOX is essential for LOX- also known to cross-link collagen.14 Both huLOX and ribose mediated promotion of tumor growth. treatment significantly increased the stiffness of collagen matrices, The stiffness of the tumors was measured using an ARG2 as measured by rheology (Figure 2b). SW480, HT29 or LS174T cells controlled strain rotational rheometer (TA Instruments, Crawley, were seeded onto the pre-modified matrices for 16 h, and then UK), and the results showed that SW480 tumors overexpressing analyzed for FAK and SRC phosphorylation by immunoblot. The full-length active LOX were significantly stiffer than the control results show that seeding cells onto a pre-modified stiffer collagen tumors or those expressing catalytically inactive mutant LOX surface is sufficient to activate the FAK-signaling pathway (Figure 1b). The same trend was observed in the HT29 and LS174T (Figure 2c and Supplementary Figure 2b). This recapitulates the tumors (Supplementary Figure 1d). The collagen content of the phenotype of the LOX-overexpressing cells, and confirms that tumors was measured using the Sircol assay (BioColor Ltd, LOX-mediated collagen cross-linking and subsequent stiffening is Carrickfergus, UK), and no significant difference in fibrillar collagen responsible for the increase in intracellular FAK and SRC content was observed (Figure 1c), suggesting that the changes in phosphorylation. stiffness can be attributed to a modification of existing collagen rather than an increase in collagen production. LOX activity and FAK/SRC activation are required for LOX-mediated promotion of proliferation in vitro LOX-mediated collagen modification increases phosphorylation of We have previously shown that overexpression of LOX promotes FAK at tyrosine 397 3D proliferation in collagen in vitro via SRC phosphorylation.5 Here, We have previously shown that active LOX increases SRC(Tyr418) we sought to determine whether LOX catalytic activity is essential phosphorylation and promotes 3D proliferation in collagen for these changes in 3D cultures, and whether FAK activation is in vitro.5 Here, we show that LOX catalytic activity in the SW480 also required. Cells were seeded into collagen and allowed to cell line promotes FAK(Tyr397) phosphorylation in vitro (Figure 2a), proliferate for 10 days before digestion of the collagen matrix and and confirm these findings in the HT29 and LS174T cell lines quantification of total viable cells. As previously reported,5 LOX (Supplementary Figure 2a). To confirm that these effects are due overexpression promotes growth in collagen, and this can be

Oncogene (2013) 1863 – 1868 & 2013 Macmillan Publishers Limited Lysyl oxidase activity in tumor progression A-M Baker et al 1865

Figure 2. LOX-mediated matrix stiffening increases FAK signaling in vitro.(a) Representative immunoblot analysis of phosphorylated FAK(Tyr397) (‘FAK-P’) levels in SW480 cells stably overexpressing LOX (‘ þ LOX’) or LTQ mutant LOX (‘ þ mut LOX’). Cells were cultured on 2.5 mg/ml collagen-coated wells for 16 h before treating with NP40 lysis buffer and analysis by immunoblotting as previously described.5 b actin was used as the loading control. (b) Relative stiffness of LOX- or ribose-modified collagen gels as measured by rheology. Wells were coated with 1.5 mg/ml collagen and allowed to set at room temperature (RT) for 30 min followed by 37 1C for 2 h before adding PBS only (‘Col I’), or PBS containing 60 mM ribose (‘Col I þ Rib’; Sigma-Aldrich, Dorset, UK) or 150 ng/ml human recombinant LOX (‘Col I þ LOX’; OriGene, Rockville, MD, USA) and incubating at 37 1C for 5 days. Stiffness was subsequently measured under the following conditions; briefly, 25 mm diameter, 1-mm thick collagen hydrogels were measured over a range of 0.5–1.25% strain at a fixed angular frequency of 0.5 rad/s and temperature of 21 1C on an ARG2 controlled strain rotational rheometer (TA Instruments). The samples were found to be only minimally frequency dependent within the range of testing and showed linear viscoelastic response within the strain range evaluated. Data are representative of two independent experiments. Bars represent mean±s.e.m., *Po0.05 and **Po0.01 using the two-sided Student’s t-test. (c) Representative immunoblot analysis of phosphorylated FAK(Tyr397) (‘FAK-P’) and phosphorylated SRC(Tyr418) (‘SRC-P’) levels in SW480 cells cultured on huLOX- or ribose-modified 1.5 mg/ml collagen matrix. After 5 days modification as described above, the wells were then washed with PBS for 1 h and SW480 cells were added for 16 h before treating with NP40 lysis buffer and analysis by immunoblotting as previously described.5 b actin was used as the loading control. (d) Effect of FAK and SRC inhibition on growth of SW480 cells in collagen. Cells were seeded into collagen gels as previously described5 and incubated at 37 1C for 10 days before digesting the collagen and counting viable cells. Media was supplemented with 1 mM dasatinib (LC Laboratories, Woburn, MA, USA), 5 mM FAK inhibitor 14 (Santa Cruz Biotechnology, Santa Cruz, CA, USA) or vehicle control (‘DMSO’). Bars represent mean±s.e.m., **Po0.01 and ***Po0.001 using the two-sided Student’s t-test. kDa, kilodalton. prevented by inhibition of SRC kinase with dasatinib (Figure 2d in nude mice (See Figure 1). Immunohistochemical staining was and Supplementary Figure 2c). In addition, we found that used to visualize LOX (Figure 3a), FAK (Tyr397; Figure 3b) and SRC inhibition of FAK-Tyr397 phosphorylation by the selective small (Tyr418; Figure 3c) in SW480 tumor sections, and as reported molecule inhibitor ‘FAK inhibitor 14’15 (Supplementary Figure 2d) previously,5 overexpression of LOX resulted in an increase in both also prevents LOX-mediated promotion of proliferation in collagen FAK and SRC phosphorylation. However, although overexpression (Figure 2d and Supplementary Figure 2c). Furthermore, over- of the catalytically inactive LOX (‘mut LOX’) was stable in vivo expression of inactive mutant LOX does not promote growth in (Figure 3a), it failed to induce FAK or SRC phosphorylation (Figures collagen (Figure 2d and Supplementary Figure 2c), suggesting that 3b and c), confirming that the enzymatic activity of LOX is LOX enzymatic activity is essential to confer an increase in responsible for the increase in FAK signaling in CRC tumors in vivo. proliferation. Furthermore, consistent results were observed in both the HT29 and LS174T subcutaneous tumor models (Supplementary LOX activity promotes activation of FAK signaling in vivo Figures 3a–c) suggesting that upregulation of FAK and SRC To confirm that LOX activity increases FAK and SRC phosphoryla- phosphorylation, as mediated by LOX activity, may be a general tion in vivo, the cell lines were implanted as subcutaneous tumors mechanism in CRC.

& 2013 Macmillan Publishers Limited Oncogene (2013) 1863 – 1868 Lysyl oxidase activity in tumor progression A-M Baker et al 1866

Figure 3. LOX-mediated matrix stiffening increases FAK signaling in vivo.(a) Representative immunohistochemical staining of LOX in subcutaneous tumors. Tumors were ﬁxed in 4% paraformaldehyde overnight before processing and parafﬁn embedding. LOX protein was detected by incubating sections overnight at 4 1C with mouse monoclonal anti-human LOX antibody as previously described.5 LOX is shown in brown and cell nuclei in blue. Scale bar represents 200 mm. (b) Representative immunohistochemical staining of phospho-FAK(Tyr397) in subcutaneous tumors. Phospho-FAK was detected by incubating sections overnight at 4 1C with rabbit anti-human phospho-FAK(Tyr397) (Abcam plc, Cambridge, UK) as previously described.5 Phospho-FAK is shown in brown and cell nuclei in blue. Scale bar represents 200 mm. (c) Representative immunohistochemical staining of phospho-SRC(Tyr418) in subcutaneous tumors. Phospho-SRC was detected by incubating sections overnight at 4 1C with rabbit anti-human phospho-SRC(Tyr418) antibody (Invitrogen Ltd, Paisley, UK) as previously described.5 Phospho-SRC is shown in brown and cell nuclei in blue. Scale bar represents 200 mm.

LOX activity promotes invasion and metastasis, via b1-integrin where the implanted cells were overexpressing the catalytically activation and FAK/SRC signaling inactive form of LOX, the cells were not able to metastasize We have previously reported that LOX overexpression can (Figures 4d and e), suggesting that LOX-mediated cross-linking promote invasion through collagen, and LOX knockdown can events are required for efficient metastasis to occur in this model. suppress invasion through collagen.5 Here, we use SW480 cells This is of significance as it confirms that therapeutic inhibition of overexpressing catalytically inactive mutant LOX to determine LOX activity may provide an effective treatment for patients with whether LOX catalytic activity is required for enhancing invasion. metastatic CRC. Alternatively, targeting the mechano-sensing Consistently, we found that overexpression of full-length LOX ability of tumor cells could prevent them from responding in a promoted invasion of the cells through collagen, but pro-metastatic manner to changes in tissue stiffness. overexpression of inactive LOX was not able to promote It is well established that mechanical stiffness of tumor ECM is a invasion (Figure 4a). Notably, overexpression of active or inactive property fundamental to tumor progression.2 Furthermore, the LOX did not affect migration of cells through a transwell analysis of tissue stiffness is already utilized in the clinic as a membrane (Supplementary Figures 4a and 4b). Furthermore, means to detect and monitor tumor growth.17 Tumors are known addition of dasatinib and FAK inhibitor 14 significantly reduced to increase and respond to changes in ECM stiffness via integrin the invasion of LOX-overexpressing cells through collagen, activation and subsequent phosphorylation of FAK, resulting in suggesting that FAK/SRC signaling is required for LOX-mediated Rho/ROCK activation and remodeling of the actin cytoskeleton.18 invasion through collagen (Figure 4a). It has previously been In addition, LOX-mediated cross-linking events have previously reported that FAK and SRC are activated by collagen via been proposed to promote breast cancer malignant progression b1-integrin signaling.16 Therefore, we carried out invasion assays upstream of this mechanism by increasing ECM stiffness.3 Here, in the presence of a b1-integrin blocking antibody and show that we use in vitro and in vivo models of CRC to show that LOX- inhibtion of the b1-integrin subunit results in a significant catalyzed cross-linking of collagen is sufficient to increase tumor decrease in invasion of high LOX-expressing cells (Supple- stiffness and induce FAK signaling. mentary Figure 4c). This is consistent with reports that LOX- Our findings support existing reports that elevated tumor mediated invasion in a breast cancer model requires b1-integrin stiffness increases malignancy2,3,19 and that LOX activity acts to activity.3 drive metastasis at the extracellular level. We also provide data to To confirm that it is LOX-mediated cross-linking, which support a key role for LOX catalytic activity in promoting stiffness- promotes invasion through collagen, we coated transwell inserts induced proliferation and invasion. These data have important with a thin layer of collagen, then pre-modified the matrix with clinical implications for the development of therapeutic LOX ‘huLOX’ or ribose for 5 days before seeding cells. We found that inhibitors, highlighting that the effective inhibition of LOX cells were more able to invade through collagen that had been catalytic activity is absolutely crucial to abrogate LOX-mediated pre-modified (Figures 4b and c), suggesting that cross-linking of ECM-driven proliferation and invasion. collagen promotes cell invasion. In summary, we show that LOX-mediated collagen cross-linking To investigate whether LOX catalytic activity is required for results in increased tissue stiffness and activation of the FAK/SRC promoting metastasis in vivo, we implanted the luciferase- signaling in in vitro and in vivo models of CRC. We further show expressing SW480 cell lines with manipulated LOX levels into that as a consequence, cells expressing high levels of enzymati- the spleen of nude mice and monitored the dissemination of cally active LOX protein have an increased capacity to proliferate, tumor cells. As previously reported, we found that SW480 cells invade and metastasize. This suggests that inhibition of LOX were poorly metastatic, and overexpression of active LOX resulted catalytic activity may provide a novel effective treatment option in an increased propensity for cells to metastasize.5 However, for patients with CRC to prevent metastatic progression.

Oncogene (2013) 1863 – 1868 & 2013 Macmillan Publishers Limited Lysyl oxidase activity in tumor progression A-M Baker et al 1867

Figure 4. LOX-mediated matrix stiffening drives invasion in vitro and metastasis in vivo. (a) Effect of LOX and mutant LOX overexpression, and FAK/ SRC inhibition on invasion of SW480 cells through collagen matrix. Invasion assays through collagen were carried out as previously described.5 The collagen matrix was supplemented with 1 mM dasatinib, 5 mM FAK inhibitor 14 or vehicle control. (b) Effect of LOX- or ribose-mediated collagen modification on invasion of SW480 cells through collagen matrix. Transwell inserts of 8 mm pore size (BD Bioscience, Oxford, UK) were coated with a thin layer of 1.8 mg/ml collagen (BD Bioscience) and allowed to set at RT for 30 min, then 2 h at 37 1CbeforeaddingPBScontaining60mM ribose (‘Ribose treated’) or 150 ng/ml huLOX (‘huLOX treated’) and incubating at 37 1C for 5 days. The PBS was then removed and the matrix was washed with fresh PBS for 1 h. Cells were serum starved for 24 h before harvesting in serum free Dulbecco’s modified Eagle medium (DMEM; Invitrogen) and seeding 2 Â 105 onto the collagen layer. DMEM containing 10% fetal bovine serum was used as a chemoattractant. Plates were incubated for 24 h at 37 1C before fixing in methanol for 5 min at RT and staining in 1% toluidine blue (Sigma-Aldrich) for 15 min at RT. Inserts were washed in water and the number of invaded cells per field of view was analyzed. Four fields of view per well and two wells per condition were counted. (c)Representative images of invaded cells after fixing and staining with toluidine blue. Images were taken using an Eclipse TS100 inverted light microscope (Nikon UK Ltd, Kingston upon Thames, UK). (d)Analysisofmetastaticburdeninanintrasplenic model. Luciferase-expressing SW480 cells (2 Â 106)were surgically implanted into the spleen of nude mice (n ¼ 6 for each condition) and metastatic burden was measured as total normalized luminescent signal from metastatic organs (liver, colon and stomach) at the experimental endpoint (4–5 weeks from surgery) as previously described.5 Data are representative of two independent experiments. Bars represent mean±s.e.m. (e) Representative images of mice surgically implanted with luciferase- expressing SW480 cells into the spleen, illustrating a high metastatic burden in mice implanted with cells expressing high levels of wild-type LOX (‘SW480 þ LOX’). Bars in (a)and(b)representmean±s.e.m., *Po0.05, **Po0.01 and ***Po0.001 using the two-sided Student’s t test.

& 2013 Macmillan Publishers Limited Oncogene (2013) 1863 – 1868 Lysyl oxidase activity in tumor progression A-M Baker et al 1868 CONFLICT OF INTEREST 9 Payne SL, Fogelgren B, Hess AR, Seftor EA, Wiley EL, Fong SF et al. Lysyl oxidase The authors declare no conflict of interest. regulates breast cancer cell migration and adhesion through a hydrogen peroxide-mediated mechanism. Cancer Res 2005; 65: 11429–11436. 10 Csiszar K, Fong SF, Ujfalusi A, Krawetz SA, Salvati EP, Mackenzie JW et al. Somatic mutations of the lysyl oxidase gene on chromosome 5q23.1 in colorectal tumors. ACKNOWLEDGEMENTS Int J Cancer 2002; 97: 636–642. We would like to thank the staff of the BSU and histopathology departments at the 11 Kim Y, Roh S, Park JY, Cho DH, Kim JC. Differential expression of the LOX family Institute of Cancer Research for assistance with in vivo work and processing of tissue genes in human colorectal adenocarcinomas. Oncol Rep 2009; 22: 799–804. samples. Funding was provided by the Medical Research Council (G0800102 to AMB), 12 Baker AM, Cox TR, Bird D, Lang G, Murray GI, Sun XF et al. The role of lysyl oxidase the Institute of Cancer Research (JTE), Cancer Research UK (C107/A10433 to TRC, DB, in SRC-dependent proliferation and metastasis of colorectal cancer. GL and JTE) and BRIC (TRC and JTE). J Natl Cancer Inst 2011; 103: 407–424. 13 Pez F, Dayan F, Durivault J, Kaniewski B, Aimond G, Le Provost GS et al. The HIF-1- inducible lysyl oxidase activates HIF-1 via the Akt pathway in a positive regulation loop and synergizes with HIF-1 in promoting tumor cell growth. Cancer Res 2011; REFERENCES 71: 1647–1657. 1 Suresh S. Biomechanics and biophysics of cancer cells. Acta Biomater 2007; 3: 14 Girton TS, Oegema TR, Tranquillo RT. Exploiting glycation to stiffen and 413–438. strengthen tissue equivalents for tissue engineering. J Biomed Mater Res 1999; 46: 2 Paszek MJ, Zahir N, Johnson KR, Lakins JN, Rozenberg GI, Gefen A et al. Tensional 87–92. homeostasis and the malignant phenotype. Cancer Cell 2005; 8: 241–254. 15 Golubovskaya VM, Nyberg C, Zheng M, Kweh F, Magis A, Ostrov D et al. A small 3 Levental KR, Yu H, Kass L, Lakins JN, Egeblad M, Erler JT et al. Matrix molecule inhibitor, 1,2,4,5-benzenetetraamine tetrahydrochloride, targeting the crosslinking forces tumor progression by enhancing integrin signaling. Cell 2009; y397 site of focal adhesion kinase decreases tumor growth. J Med Chem 2008; 51: 139: 891–906. 7405–7416. 4 Erler JT, Bennewith KL, Nicolau M, Dornhofer N, Kong C, Le QT et al. Lysyl oxidase 16 Barkan D, El Touny LH, Michalowski AM, Smith JA, Chu I, Davis AS et al. Metastatic is essential for hypoxia-induced metastasis. Nature 2006; 440: 1222–1226. growth from dormant cells induced by a col-I-enriched fibrotic environment. 5 Baker AM, Cox TR, Bird D, Lang G, Murray GI, Sun XF et al. The role of lysyl oxidase Cancer Res 2010; 70: 5706–5716. in SRC-dependent proliferation and metastasis of colorectal cancer. J Natl Cancer 17 Sinkus R, Lorenzen J, Schrader D, Lorenzen M, Dargatz M, Holz D. High-resolution Inst 2011; 103: 407–424. tensor MR elastography for breast tumour detection. Phys Med Biol 2000; 45: 6 Le QT, Harris J, Magliocco AM, Kong CS, Diaz R, Shin B et al. Validation of lysyl 1649–1664. oxidase as a prognostic marker for metastasis and survival in head and neck 18 Samuel MS, Lopez JI, McGhee EJ, Croft DR, Strachan D, Timpson P et al. squamous cell carcinoma: Radiation Therapy Oncology Group trial 90-03. J Clin Actomyosin-mediated cellular tension drives increased tissue stiffness and Oncol 2009; 27: 4281–4286. beta-catenin activation to induce epidermal hyperplasia and tumor growth. 7 Tilghman RW, Parsons JT. Focal adhesion kinase as a regulator of cell tension in Cancer Cell 2011; 19: 776–791. the progression of cancer. Semin Cancer Biol 2008; 18: 45–52. 19 Colpaert C, Vermeulen P, Van Marck E, Dirix L. The presence of a fibrotic focus is 8 Ishizawar R, Parsons SJ. c-Src and cooperating partners in human cancer. Cancer an independent predictor of early metastasis in lymph node-negative breast Cell 2004; 6: 209–214. cancer patients. Am J Surg Pathol 2001; 25: 1557–1558.

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

Oncogene (2013) 1863 – 1868 & 2013 Macmillan Publishers Limited