A Link Between Inflammation and Metastasis

Home , Calgranulin, Fibulin, FLNB, S100A12, S100 protein, Serum amyloid A1

ORIGINAL ARTICLE A link between inﬂammation and metastasis: serum amyloid A1 and A3 induce metastasis, and are targets of metastasis-inducing S100A4

MT Hansen1,8, B Forst1,8, N Cremers2,3, L Quagliata2, N Ambartsumian1,4, B Grum-Schwensen1, J Klingelho¨ fer1,4, A Abdul-Al1, P Herrmann5, M Osterland5, U Stein5, GH Nielsen6, PE Scherer7, E Lukanidin1, JP Sleeman2,3,9 and M Grigorian1,4,9

S100A4 is implicated in metastasis and chronic inflammation, but its function remains uncertain. Here we establish an S100A4- dependent link between inflammation and metastatic tumor progression. We found that the acute-phase response proteins serum amyloid A (SAA) 1 and SAA3 are transcriptional targets of S100A4 via Toll-like receptor 4 (TLR4)/nuclear factor-kB signaling. SAA proteins stimulated the transcription of RANTES (regulated upon activation normal T-cell expressed and presumably secreted), G-CSF (granulocyte-colony-stimulating factor) and MMP2 (matrix metalloproteinase 2), MMP3, MMP9 and MMP13. We have also shown for the first time that SAA stimulate their own transcription as well as that of proinflammatory S100A8 and S100A9 proteins. Moreover, they strongly enhanced tumor cell adhesion to fibronectin, and stimulated migration and invasion of human and mouse tumor cells. Intravenously injected S100A4 protein induced expression of SAA proteins and cytokines in an organ-specific manner. In a breast cancer animal model, ectopic expression of SAA1 or SAA3 in tumor cells potently promoted widespread metastasis formation accompanied by a massive infiltration of immune cells. Furthermore, coordinate expression of S100A4 and SAA in tumor samples from colorectal carcinoma patients significantly correlated with reduced overall survival. These data show that SAA proteins are effectors for the metastasis-promoting functions of S100A4, and serve as a link between inflammation and tumor progression.

Oncogene (2015) 34, 424–435; doi:10.1038/onc.2013.568; published online 27 January 2014 Keywords: inﬂammation; metastasis; S100A4; serum amyloid A

INTRODUCTION (MMPs).6–12 SAA can signal via Toll-like receptors (TLRs) and 9,13 The microenvironment surrounding tumor cells, including cells of nuclear factor-kB (NF-kB). In mice, SAA3 contributes to the 14 the immune system and proinflammatory factors, have a key role establishment of lung premetastatic niches, which are 15 in regulating metastasis formation.1,2 Although different microenvironments that foster metastasis formation. subclasses of immune cells can inhibit or promote metastasis, We have previously implicated the S100 family member S100A4 chronic inflammation in tumors generally predicts poor in both tumor progression and in inflammatory diseases such as 16–18 prognosis.3 Serum amyloid A (SAA) is a family of highly rheumatoid arthritis, psoriasis and dermato/polymyosites. homologous acute-phase proteins whose expression and S100 proteins require calcium-induced oligomerization for their accumulation in the blood is observed during inflammation and activity, and are involved in the regulation of proliferation, 19,20 has been associated with tumor progression and reduced survival survival, differentiation and motility. S100A4 expression is in many human cancers.4 SAA1 and SAA2 are the major acute- enhanced in highly metastatic CSML100 mouse mammary phase plasma isoforms. A third acute-phase isoform SAA3 is adenocarcinoma cells compared with poorly metastatic CSML0 expressed in rodents, but is a transcribed pseudogene in humans.5 cells, and functionally contributes to metastasis.21–25 Furthermore, In vivo concentrations of SAAs increase in the blood during the the metastatic behavior of the mouse adenocarcinoma cell line response to trauma, infection, inflammation and neoplasia, and VMR is determined by the stroma, and is stimulated by S100A4- serve to regulate homeostatically lipid metabolism and transport, expressing fibroblasts, indicating that S100A4 can promote immune cell chemotaxis and other inflammatory processes. SAA metastasis formation through both autocrine and paracrine proteins regulate the expression of cytokines including mechanisms.26 Clinical studies also demonstrate that augmented interleukins, granulocyte-colony-stimulating factor (G-CSF) and expression of S100A4 in primary tumors correlates with poor tumor necrosis factor-a, as well as matrix metalloproteinases prognosis (reviewed in Helfman et al.27). S100A4-mediated

1Danish Cancer Society Research Center, Copenhagen, Denmark; 2Universita¨tsmedizin Mannheim, University of Heidelberg, Mannheim, Germany; 3KIT Karlsruhe, Eggenstein- Leopoldshafen, Germany; 4Neuro-Oncology Group, Laboratory of Neural Plasticity, Institute of Neuroscience and Pharmacology, Faculty of Health Sciences, Copenhagen University, Copenhagen, Denmark; 5Experimental and Clinical Research Center, Charite´ Universita¨tsmedizin Berlin, at the Max-Delbru¨ck Center for Molecular Medicine, Berlin, Germany; 6Air Liquide Danmark A/S, Horsens, Denmark and 7Touchstone Diabetes Center, University of Texas Southwestern Medical Center, Dallas, TX, USA. Correspondence: Dr M Grigorian, Neuro-Oncology Group, Laboratory of Neural Plasticity, Institute of Neuroscience and Pharmacology, Faculty of Health Sciences, Copenhagen University, Blegdamsvej 3B, Copenhagen 2200, Denmark. E-mail: [email protected] 8These authors contributed equally to this work. 9These authors contributed equally to this work. Received 3 July 2013; revised 6 December 2013; accepted 7 December 2013; published online 27 January 2014 S100A4-induced SAA expression promotes metastasis MT Hansen et al 425 metastasis is associated with extensive T-cell infiltration into both expression by mouse recS100A4 (Figure 1e and Supplementary the primary tumor and sites of metastasis.28 Despite the extensive Figure S1D). In addition, mutant forms of hS100A4 (moA4mt1 and evidence implicating S100A4 in metastasis formation, the moA4mt2) that cannot form oligomers had an attenuated ability mechanism by which it exerts its effects remains uncertain. to induce SAA activation (Figure 1e and Supplementary Figure Here we found that SAA proteins are transcriptional targets of S1D). These data demonstrate that S100A4 specifically stimulates S100A4. In turn, SAA proteins stimulated both their own the expression of SAA in VMR cells. transcription and that of a variety of cytokines and MMPs, and promoted chemotactic migration and adhesion to fibronectin by tumor cells. Ectopic SAA expression in experimental tumors SAA augments the metastasis-associated properties of tumor cells sufficed to augment strongly metastasis formation, and was in vitro accompanied by a marked immune cell infiltration. Consistently, We next examined whether SAA proteins modify tumor cell coexpression of SAA and S100A4 correlated with poor prognosis properties associated with metastasis. VMR cells were transduced in human colorectal cancer patients. These data provide an with SAA1 and SAA3 retroviral expression vectors to obtain the important link between the metastasis-promoting and proinflam- stable SAA-expressing cell lines VMR/SAA1 and VMR/SAA3. Cells matory effects of S100A4, and suggest that the induction of SAA transduced with the empty vector (VMR/CTL) served as controls. expression is a major means by which S100A4 exerts these effects. Expression and secretion of SAA proteins was verified by western blotting (Supplementary Figure S2A). In subsequent experiments, the use of retrovirally transduced cells compared with exo- RESULTS genously added SAA protein (either as CM or as recombinant S100A4 induces expression of proinflammatory genes, including protein) allowed paracrine and autocrine effects to be compared. SAA family members Both VMR/SAA1 and VMR/SAA3 cells adhered much more To investigate how S100A4 promotes metastasis, we used strongly to fibronectin than the VMR/CTL cells (Figure 2a), but not microarrays to transcriptionally profile VMR tumor cells treated or to laminin or collagen (Supplementary Figure S2B). Effects of SAA non-treated for 24 h with the active multimeric form of the on cell motility were assessed by incubating wounded monolayers recombinant human S100A4 (hS100A4) protein. Of the 75 genes of mouse CSML100 and human MDA-MB-231 and SW480 tumor upregulated in response to hS100A4, more than 30% are cells with CM from VMR/SAA1, VMR/SAA3 and VMR/CTL cells, or associated with inflammation (Table 1). The most strongly with medium supplemented with recSAAs. Effects on the motility upregulated genes included the acute-phase reactants (SAA1 and of mouse embryonic fibroblasts were also examined, as S100A4 SAA3), cytokines and their receptors (CXCL1, CSF1, CCL4, INFAR2), stimulates the motility of activated fibroblasts and recruits them to the proinflammatory S100 family member S100A8, inflammation- tumors, promoting metastasis.29 In all cases, SAA-containing associated proteases and immune cell-specific genes. media induced a significant increase in motility compared with Given their upregulation in response to S100A4 and the control media (Figures 2b and c and Supplementary Figures S2C– literature correlating their expression with poor prognosis, we E). Both recSAA1 and recSAA3 also exerted chemotactic effects in determined whether SAA1 and SAA3 (hereafter referred to a dose-dependent manner on CSML100 tumor cells after 6 h in generically as SAA) might be mediators of S100A4-induced Transwell migration/invasion assays (Figure 2d). Significantly metastasis. First, we validated the transcriptional profiling. enhanced invasiveness of human MDA-MB-231 tumor cells was In VMR cells, SAA1 and SAA3 were both strongly transcriptionally also observed in three-dimensional Matrigel assays that examined induced in response to hS100A4 protein, as assessed the effect of CM from VMR/SAA1 and VMR/SAA3 cells, as well as by quantitative real-time polymerase chain reaction (qPCR) recSAAs compared with controls (Figure 2e and Supplementary (Figures 1a and b). SAA1/2 and SAA3 proteins were also detected Figure S2F). in conditioned medium (CM) from cells 24 h after induction SAA proteins induce MMP expression in monocytic and (Figures 1a and b). endothelial cells.11,12 MMPs are implicated in metastatic Next, we examined whether S100A4-mediated upregulation of progression. We therefore examined whether SAA proteins also SAA proteins also occurs in human tumor cells. As human SAA3 is upregulate MMP expression in VMR cells. Transcription of MMP2, a pseudogene, we only monitored the expression of SAA1. MMP3, MMP9 and MMP13 was significantly increased in VMR cells Treatment of a panel of human cancer cells with hS100A4 protein treated with recSAAs (Figure 3a). Consistently, zymography also resulted in the accumulation of SAA1/2 in the CM (Figure 1c). demonstrated that CM from VMR/SAA1 and VMR/SAA3 cells In primary mouse bone marrow-derived macrophages, we found contained increased proteolytic activity when compared with strong transactivation of SAA3 but not SAA1 by S100A4 controls. The size of the cleared bands corresponded to MMP2 (Supplementary Figure S1A). In T cells, basal expression of both and MMP9 (Supplementary Figure S3A). Increased levels of MMP2, SAAs was very weak, and their transactivation in response to MMP3, MMP9 and MMP13 proteins were also detected in CM from S100A4 was negligible (Supplementary Figure S1B). Polymixin B VMR cells treated with SAA3 (Supplementary Figure S3B). sulfate (PMB), a potent endotoxin inhibitor, had no significant Furthermore, CM taken from VMR cells treated with recS100A4 effect on the hS100A4-driven upregulation of SAA3 and recSAA3 proteins exhibited higher proteolytic activity in (Supplementary Figure S1C) but completely neutralized the even zymography assays (Supplementary Figure S3C), demonstrating stronger induction of SAA3 expression in response to lipopoly- direct effects of SAA proteins on protease production. Consis- saccharide, demonstrating that induction of SAA by recombinant tently, non-metastatic CSML0 cells that do not respond to hS100A4 protein was not due to endotoxin contamination. recS100A4 by upregulating SAA transcription (Supplementary To examine the specificity of the induction of SAA expression by Figure S4A) showed no increase in proteolytic activity in response S100A4, we first tested the ability of different S100 family proteins to S100A4 compared with VMR cells, but responded to recSAA to stimulate the accumulation of SAA in CM from VMR cells. (Supplementary Figure S4B). In addition to S100A4, the proinflammatory S100A8 and S100A9 SAA proteins regulate the expression of a variety of cyto- proteins also stimulated SAA expression, whereas S100A1, S100A2, kines.7,9,10 Furthermore, S100A4 induces the expression of RANTES S100A6, S100B, S100P and S100A12 had no effect (Figure 1d). The (regulated upon activation normal T-cell expressed and specificity was further confirmed by using two independent presumably secreted) and G-CSF.28,30 We therefore examined inhibitory anti-human S100A4 monoclonal antibodies that have whether SAAs induce the expression of cytokines and other no crossreactivity with mouse S100A4. They markedly inhibited proinflammatory proteins in VMR tumor cells. RecSAA1 or recSAA3 hS100A4-mediated SAA expression, but not the induction of SAA strongly stimulated transcription and accumulation of SAA

& 2015 Macmillan Publishers Limited Oncogene (2015) 424 – 435 S100A4-induced SAA expression promotes metastasis MT Hansen et al 426 Table 1. Genes upregulated in VMR tumor cells in response to S100A4 (24 h) identiﬁed by RNA microarray analysis

Gene Fold induction Description

CXCL1 9.61836 Cytokine CXCL1 SAA3 3.09445 Serum amyloid a-3 EXPI 3.04594 Extracellular proteinase inhibitor ALDOA-PS1 2.88877 Aldolase A, fructose-bisphosphate-pseudogene PDZRN3 2.76743 PDZ domain containing RING-finger 3 Gene CEBPD 2.14473 CCAAT/enhancer-binding protein (C/EBP), delta LRIG1 2.08822 Leucine-rich repeats and Ig-like domains 1 SAA1 1.91439 Serum amyloid a-1 D19WSU12E 2.08696 Endoplasmic reticulum metallopeptidase 1/ERMP1 PPAP2B 1.93407 Phosphatidic acid phosphatase type 2B CHI3L1 1.86881 Chitinase 3-like 1 gene PDE4B 1.76396 Phosphodiesterase 4B, camp specific MT1 1.75631 Metallothionein 1 D9UCLA1 1.75718 HCV NS5A-transactivated protein 13 target protein 2 HP 1.64029 Haptoglobin SLPI 1.62496 Secretory leukocyte peptidase inhibitor SOD2 1.70735 Superoxide dismutase 2, mitochondrial NFKB1 1.6833 Nuclear factor of kappa light polypeptide gene enhancer in B cells SCMH1 1.6032 Sex comb on midleg homolog 1 TRIM47 1.57104 Tripartite motif-containing 47 IER3 1.59972 Immediate-early response 3 gene CYP7B1 1.52857 Cytochrome P450, family 7, subfamily b BCL3 1.46493 B-cell leukemia/lymphoma 3 PIM1 1.57511 Proviral integration site 1 RIN2 1.50839 Ras and Rab interactor 2 ADAMTS1 1.54172 A disintegrin-like and metallopeptidase (reprolysin type) with thrombospondin type 1 NFKB2 1.49329 Nuclear factor of kappa light polypeptide gene enhancer in B cells 2, BID 1.47709 BH3-interacting domain death agonist RELB 1.40966 Transcription factor (v-rel reticuloendotheliosis viral oncogene homolog B) PIK3AP1 1.43798 Phosphoinositide-3 kinase adaptor protein 1 ANXA1 1.52158 Annexin A1 IFNAR2 1.48214 Interferon (alpha and beta) receptor 2 FTH1 1.45364 Ferritin heavy chain 1 ALDH3A2 1.44178 Aldehyde dehydrogenase family 3, subfamily A2 CP 1.44929 Ceruloplasmin GDNF 1.39213 Glial cell line-derived neurotrophic factor MMP2 1.39512 Matrix metallopeptidase 2 ATP2B4 1.46876 ATPase, Ca2 þ transporting, plasma membrane 4 FLNB 1.46582 Filamin, beta AXL 1.42897 AXL receptor tyrosine kinase SMOX 1.3777 Spermine oxidase SOD2 1.40841 Superoxide dismutase 2, mitochondrial FTH1 1.38935 Ferritin heavy chain 1 NADK 1.5183 NAD kinase S100A8 1.43334 S100 calcium-binding protein A8 (calgranulin A) THRAP4 1.43386 Thyroid hormone receptor-associated protein 4 FTH1 1.37284 Ferritin heavy chain 1 COX7B 1.43321 Cytochrome c oxidase subunit viib TNIP1 1.39607 TNFAIP3-interacting protein 1 ISP2 1.33707 Inplantation serine proteinase FTH1 1.33253 Ferritin heavy chain 1 LEPREL1 1.44379 Leprecan-like 1 SHB 1.36273 Src homology 2 domain-containing transforming protein B RETNLG 1.36848 Resistin-like gamma CTSL 1.32786 Cathepsin L MT-ATP6 1.32828 Mitochondrially encoded ATP synthase 6 DHRS8 1.37775 Estradiol 17-beta-dehydrogenase 11 MT-ATP6 1.32444 Mitochondrially encoded ATP synthase 6 FAAH 1.36722 Fatty acid amide hydrolase HIVEP1 1.36318 Human immunodeficiency virus type I enhancer-binding protein 1 OGFRL1 1.36524 Opioid growth factor receptor-like 1 GLUL 1.31464 Glutamate-ammonia ligase (glutamine synthetase NEFH 1.35809 Neurofilament, heavy polypeptide CSF1 1.30967 Colony-stimulating factor 1 (macrophage) GLIPR1 1.30676 GLI pathogenesis-related 1 (glioma PSD 1.34813 Pleckstrin and Sec7 domain containing PCOLN3 1.30149 Procollagen (type III) N-endopeptidase CCL4 1.28116 Chemokine (C–C motif) ligand 4 MRPL17 1.33649 Mitochondrial ribosomal protein L17 DHX40 1.33219 DEAH (Asp-Glu-Ala-His) box polypeptide 40

Oncogene (2015) 424 – 435 & 2015 Macmillan Publishers Limited S100A4-induced SAA expression promotes metastasis MT Hansen et al 427

Table 1. (Continued )

Gene Fold induction Description

ETS2 1.2775 E26 avian leukemia oncogene 2, 3 domain PNRC1 1.32848 Proline-rich nuclear receptor coactivator 1 NFE2L1 1.31799 Nuclear factor, erythroid-derived 2,-like 1 CYB5 1.27215 Cytochrome b-5 PHC2 1.27223 Polyhomeotic-like 2 (Drosophila) 2810439K08RIK 1.25865 Hypothetical protein TAGLN2 1.29413 Transgelin 2 FCNB 1.26924 Ficolin B The arrays were scanned by using an Axon GenePix 4000B microarray scanner (Axon Instruments Inc., Union City, CA, USA) and the GenePix Pro program (Biberach, Germany). Genes were considered as differentially expressed if they showed at least a 1.3-fold difference in signal intensity.

+S100A4 +S100A4 - + - + 600 SAA3 SAA1 *

15 400

* 10 ** ** (x folds) * 200 **

(x folds) ** 5 ** *

Transactivation of SAA3 ** ns

Transactivation of SAA1 0 0 0 1.5 3 6 12 24 0 1.5 3 6 9 12 24 hours hours

--+ +-+-+ -+-+-+ +hS100A4 CTL +Ab#1 +Ab#2 +IgG SAA1 SAA1

VMR SW480 SW620 MB-231 MB-435 MCF7 MCF7s1 SAA3 Tubulin Actin

CTL hA4 +Ab#1 +Ab#2 SAA1 recS100 CTL A4 A8 A12 A6 S100B S100P A2 A1 A9 +PMB CTL moA4 +Ab#1 +Ab#2 SAA3 SAA1

CTL A4 A8 A9 CTL moA4 moA4mt1 moA4mt2 SAA1 SAA1

Figure 1. S100A4 specifically upregulates SAA in VMR cells. SAA1 (A) and SAA3 (B) expression in VMR cells incubated with recS100A4 ( þ S100A4) for the indicated times was analyzed by qPCR, and normalized relative to non-treated controls (fold induction). Insets: Western blots of SAA1 and SAA3 in CM from VMR cells treated with ( þ ) and without ( À ) recS100A4 for 24 h. The experiments were performed independently three times, each with three replicates for each time point. Representative western blot analyses are presented in the insets. (C) Western blot of SAA1 in CM from the indicated human cell lines treated with ( þ ) and without ( À ) recS100A4 for 24 h. Loading control: actin. (D) SAA1 and SAA3 protein accumulation in CM from VMR cells incubated with the indicated recS100 proteins in the presence of PMB. (E) (a–c) Western blot of SAA in CM from VMR cells after stimulation with human S100A4 (hA4) in the presence of two different anti-human S100A4 monoclonal antibodies (Ab no. 1 and Ab no. 2) that inhibit SAA1 and SAA3 upregulation in response to hA4 but not mouse S100A4 (moA4). Non-treated cells (CTL) and isotype control antibodies (IgG) served as controls. Tubulin served as a loading control. (D) Mutated forms of moA4 (moA4mt1 and moA4mt2) unable to form oligomers and have an attenuated ability to induce SAA3. Representative western blot analysis is presented. proteins in CM from VMR cells (Figures 3b and c). Moreover, Molecular regulation of SAA expression by S100A4 RANTES, G-CSF, S100A8 and S100A9 were also transcriptionally How does S100A4 induce SAA expression? TLR-4 can mediate upregulated in response to recSAA1 and recSAA3 (Figures 3d and e). S100A8 and S100A9 signaling31 via NF-kB and IRAK (interleukin-1 Taken together, these data are consistent with the notion that receptor-associated kinase).32 Specific inhibitors for TLR-4 SAA proteins produced by tumor cells in response to S100A4 are (CLI-095), NF-kB(CAPE (caffeic acid phenethyl ester)) and IRAK able to modulate cell adhesion, migration and invasion, to (IRAK1/4 (interleukin-1 receptor-associated kinases 1/4 inhibitor)) transcriptionally upregulate MMPs, cytokines and other robustly suppressed SAA accumulation in the CM from S100A4- proinflammatory factors, and to regulate positively their own treated VMR cells (Figure 4a, Supplementary Figure S5A and expression (Figure 3f). Supplementary Figure S6), implicating these factors in the

& 2015 Macmillan Publishers Limited Oncogene (2015) 424 – 435 428 gr.Rc104sre sapstv oto.PBtetetcnrle o nooi otmnto.( the contamination. in presented endotoxin gel for the of controlled area the treatment outside PMB lies CM control. the to positive indicated a where added as recSAA1 tagged served the size, RecS100A4 its figure. to Owing presented. is analysis A1cls(mM/A1,C rmVRSA el cVRSA)o Mfo M el upeetdwt eSA (cmVMR recSAA1 with supplemented cells VMR/ VMR from ( from CM samples (cmVMR), CM triplicate CM or of control mean (cmVMR/SAA3) either the cells with represents performed VMR/SAA3 h. point were 22 from data Assays for CM assays. Each monitored Matrigel (cmVMR/SAA1), was closure three-dimensional cells Wound in SAA1 retroviruses. cells (VMR/SAA3) MDA-MB-231 SAA3 of or invasion (VMR/SAA1) SAA1 (VMR/CTL), vector ( empty either with transduced ( h 104rgltdepeso fSA riflmaoyctknsadMP,adtepstvl cigfefradlosdsrbdin described loops feedforward acting positively the and MPPs, and cytokines proinflammatory SAA, paper. of this expression S100A4-regulated the ( samples. ( replicate four (CTL). or cells three with mock-treated each times, ( three independently SAA3, performed were experiments The 3. Figure noee(05 2 435 – 424 (2015) Oncogene iue2. Figure lte.Oeo w needn xeiet sson ( shown. is experiments independent two of One plotted. M el rndcdwt ihrepyvco VRCL,SA VRSA)o A3(M/A3 ervrsst boetnwsquantified was fibronectin to retroviruses (VMR/SAA3) SAA3 or (OD (VMR/SAA1) absorbance SAA1 and (VMR/CTL), staining vector violet empty crystal either using with transduced cells VMR d b hmtci irto fCM10clsatr6hicbto ihteidctdrcAsi rnwl irto sas ( assays. migration Transwell in recSAAs indicated the with incubation h 6 after cells CSML100 of migration Chemotactic ) n ua D-B21cls( cells MDA-MB-231 human and ) d ATSadGCFad( and G-CSF and RANTES ) A rtisehnetmrcl deint boetn tmlt irto n nain n peuaeMP.( MMPs. upregulate and invasion, and migration stimulate fibronectin, to adhesion cell tumor enhance proteins SAA eSApoen peuaeteepeso fiflmainascae ee.VRclswr rae ihrcA1o recSAA3. or recSAA1 with treated were cells VMR genes. inflammation-associated of expression the upregulate proteins RecSAA 104idcdSAepeso rmtsmetastasis promotes expression SAA S100A4-induced Transactivation (fold induction) Adherent cells (fold induction) 10 20 30 40 50 0

Transactivation of MMPs VMR/CTL 0 2 4 6 8 (fold induction) c ATSG-CSF RANTES 0 1 2 3 4 etr lto A1i Mfo M el rae ihrcA1adrcA3 ersnaiewsenblot western Representative recSAA3. and recSAA1 with treated cells VMR from CM in SAA1 of blot Western ) VMR/SAA1 CTL SAA3 SAA1 ** M2MP M9MMP13 MMP9 MMP3 MMP2 ** Number of migrated cells *** ** **

100 200 300 VMR/SAA3 e n.s. 0 c 108adS0A nrsos a oioe yqC,adnraie fl nuto)rltv to relative induction) (fold normalized and qPCR, by monitored was response in S100A9 and S100A8 ) nmnlyrwudhaigasy.Wuddmnlyr eeicbtdwt Mfo M cells VMR from CM with incubated were monolayers Wounded assays. wound-healing monolayer in ) SAA3 2mg/mlCTL ** *** ***

SAA3 5mg/m ** *** SAA3 SAA1 CTL

* Relative area of wound (%)

SAA3 10mg/m CSML100 100 550 * 80 20 40 60 .Vle o enasrac iu notdwells uncoated minus absorbance mean for Values ).

Transactivation (fold induction) 0 100 150 SAA1 2mg/m Hansen MT 50 0 * *** cmVMR/SAA3 cmVMR/SAA1 cmVMR * 5 b 108S100A9 S100A8 SAA1 5mg/m and *** CTL SAA3 SAA1 b – SL0 MDA-MB-231 CSML100 Transactivation of SAA 10 e ** c tal et ). (x folds) Hours Mfo M/A ellnsicesstemtlt fmueCM10cells CSML100 mouse of motility the increases lines cell VMR/SAA from CM ) 10 0 2 4 6 8 15 A1SAA3 SAA1 *** SAA3 SAA1 CTL * Relative area of invasion *** *** in matrigel (%) 20 100 120 140 *** 60 80 25 RANTES G-CSF cmVMR ** cmVMR/SAA1 *** Relative area of wound (%) 100 20 40 60 80 MDA-MB-231 0 rSAA3 0 rS100A4 PMB rSAA1 cmVMR/SAA3 * cmVMR/SAA3 cmVMR/SAA1 cmVMR 5 *** *** ------S100A4 **

MMPs cmVMR+rSAA1 SAA + - - 10 Hours + + - - ± a xrsino Ms ( MMPs, of Expression ) ** + ...rltv oVRCLclsare cells VMR/CTL to relative s.e.m. - - & 15 05McilnPbihr Limited Publishers Macmillan 2015 f + + - - ceai iga showing diagram Schematic ) + - - - 20 SAA1 *** *** S100A9 S100A8 25 e A stimulates SAA ) a deinof Adhesion ) b A1and SAA1 ) þ SAA1). S100A4-induced SAA expression promotes metastasis MT Hansen et al 429

50 n.s. SAA1 ** 40 A4 - + - + + + --- 30 LPS - - + ---+ + +

Inhibitor --- TLR4 IRAK NF-kB TLR4 IRAK NF-kB 20 (fold induction) 10 Transactivation of SAA3 0

5 SAA3 CTL S100A4 - + + + TNF EGFR inh - - + - 4 S100A4 MEK inh - - - +

recSAA1 2 SAA1 (fold induction) S100A4 -+------Luciferase transactivation 1 SAA1 --++---- SAA3 ----++++ PMB ---+-+-- 0 Inhibitor ------TLR4 NFkB VMR-CTL VMR-I B dh-CI5 Figure 4. S100A4 and SAA signal via TLR4 and NF-kB. (a) VMR cells were treated with S100A4 (A4) or lipopolysaccharide in the presence and absence of TLR4, IRAK and NF-kB inhibitors. Expression of SAA1 in CM from the cells was assessed by western blotting. Representative western blot analysis from two independent experiments is presented. (b) VMR-I-kB and control cells (VMR-CTL) were engineered to express a NF-kB- driven luciferase reporter, and then were treated with S100A4 and tumor necrosis factor-a. Luciferase activity relative to mock-treated cells (CTL) was assessed. The experiments were performed independently three times, each with three replicate samples. (c) VMR cells were treated with S100A4 in the presence or absence of EGFR or MEK inhibitors. Expression of SAA3 relative to mock-treated cells was assessed by qPCR, and by western blot (insert). The experiments were performed independently four times, each with three replicate samples. A representative western blot is presented. (d) Western blots of SAA1 in CM from VMR cells treated with recS100A4, recSAA1 or recSAA3 in the presence and absence of PMB or inhibitors of TLR4 and NF-kB. Both the added recSAA1 and the endogenous cell-derived SAA1 are visible. A representative western blot analysis from two independent experiments is presented. upregulation of SAA in response to S100A4. Epidermal growth increased spontaneous and experimental metastases in both the factor (EGF)-induced extracellular signal-regulated kinase (ERK) lung and the liver (Figures 5a–d). Expression of SAA1 induced phosphorylation was not affected by these inhibitors, verifying the metastasis even more potently in the experimental metastasis specificity of the inhibition (Supplementary Figure S5B). Moreover, assay (Figures 5e and f). In addition, SAA-overexpressing VMR in VMR cells expressing a dominant-negative form of IkBa (VMR- cells instigated metastasis in other organs such as the spleen, IkBa-cI5 cells), the activity of an NF-kB-responsive luciferase lymph nodes, ovary, kidney and bones in 50% of the mice reporter in response to either S100A4 or tumor necrosis factor-a injected with VMR-SAA1 and in 80% of the mice injected with was significantly impaired compared with that in wild-type VMR VMR-SAA3. A marked infiltration of CD45-positive leukocytes into cells (Figure 4b). the lungs and livers of animals bearing VMR-SAA tumors was As S100A4 modulates EGF receptor (EGFR) signaling,33 we observed, but not in animals bearing control VMR tumors analyzed whether this pathway is also involved in S100A4-driven (Figure 5g). activation of SAA3. VMR cells were incubated with the tyrosine kinase inhibitors AG1478 (EGFR-specific) and AG879 (ErbB2- specific) before being treated with recS100A4. Inhibition of EGFR Systemic S100A4 induces SAA expression and other transcriptional strongly and significantly suppressed S100A4-dependent SAA3 targets in an organ-specific manner transcription and protein expression, whereas inhibition of the Intravenous injection of S100A4 increases metastasis formation in downstream kinase MEK did not influence S100A4-mediated the liver and lung by VMR cells.26 Furthermore, S100A8 induces upregulation of SAA3 mRNA (Figure 4c). SAA3 expression in the lung.14 We therefore examined whether Collectively, these data indicate a decisive role for NF-kBin increased systemic levels of proinflammatory S100A4, S100A8 or S100A4-driven transcriptional activation of SAA genes, and S100A9 proteins might facilitate metastasis by inducing the suggest an involvement of TLR-4 and EGFR(s) in S100A4-mediated expression of SAA in these organs. Significantly upregulated signal transduction. Although positive self-regulation by SAA also expression of SAA1 in both the liver and the lung was observed involves NF-kB, TLR-4 is not involved in this signaling pathway in after intravenous injection of S100A4 into mice for 2 weeks VMR cells (Figure 4d). (Figures 6a and e). SAA3 was also significantly upregulated in the liver, but only tendentially in the lung (Figures 6b and f). S100A8 significantly induced SAA1 expression in the lung but only Ectopic SAA expression in VMR cells promotes tumor metastasis tendentially in the liver (Figures 6a and e), while SAA3 was VMR-CTL and VMR-SAA cells were injected into experimental upregulated in both the liver and the lung (Figures 6b and f). animals to determine whether SAA expression is sufficient to S100A9 on the other hand had no significant effect on SAA stimulate metastasis in vivo. Histological evaluation of tissue expression in any organ. Of the three proteins, S100A4 and sections revealed that SAA3 expression resulted in significantly S100A8 most potently induced SAA expression. In addition to

& 2015 Macmillan Publishers Limited Oncogene (2015) 424 – 435 S100A4-induced SAA expression promotes metastasis MT Hansen et al 430

Lung metastases (s.c) Liver metastases (s.c) 150 150 *** ** 100 100

50 50 per tissue area unit per tissue area unit

Number of metastases 0 Number of metastases 0 VMR-CTL VMR-SAA3 VMR-CTL VMR-SAA3

Lung metastases (i.v) Lung metastases (i.v) 15 500 ** ** 400 10 300

200 5 100 per tissue area unit per tissue area unit

Number of metastases 0 Number of metastases 0 VMR-CTLVMR-SAA3 VMR-CTL VMR-SAA3

Lung metastases (i.v) Liver metastases (i.v)

250 ** 1500 *** 200 1000 150

100 500 50 per tissue area unit per tissue area unit

Number of metastases 0 Number of metastases 0 VMR-CTLVMR-SAA1 VMR-CTL VMR-SAA1

Lung CD45

Liver

VMR-CTL VMR-SAA1 VMR-SAA3 Figure 5. Ectopic SAA expression stimulates the metastatic potential of VMR tumor cells. VMR-SAA3 or VMR-CTL were used in spontaneous (a and b) and experimental (c and d) metastasis assays (nine mice per group). (e and f) VMR-SAA1 was used in experimental metastasis assays (eight mice per group). Numbers of metastases in liver and lungs were evaluated. (g) Immunohistochemical staining with CD45 antibodies of sections of lung (upper panel) and liver (lower panel) from VMR-CTL and VMR-SAA3 tumor-bearing mice. Scale bars, 50 mm. Immunohistochemical stainings were repeated two times and representative data are presented.

inducing expression of SAA in these organs, S100A4 and S100A8 Coincident expression of SAA and S100A4 in human colon both induced an increase in the level of SAA3 protein in the carcinomas is indicative of poor prognosis blood of the treated animals (Figure 6i). Differences in the To determine the relevance of our findings to human cancer, we specific activity of the recS100A proteins can be ruled out, as all examined whether expression of SAA alone and in combination three preparations induced SAA3 expression in VMR cells with S100A4 correlates with patient survival. Primary tumor (Figure 1d). specimens were analyzed from 60 colon adenocarcinoma patients In similar experiments, we also observed that S100A4 whose tumors had not metastasized at the time of surgery. significantly increased the expression of RANTES, G-CSF, S100A8 Of these patients, 23 subsequently developed distant metastases and S100A9 in the liver (Figure 6c and d). In the lung, it also metachronously. SAA and S100A4 expression as determined by increased the expression of S100A9, downregulated G-CSF and qPCR was independent of the UICC tumor stage, and the age and had a negligible effect on RANTES and S100A8 (Figure 6g and h). sex of the patients. There was no significant correlation between These data demonstrate that S100A4 can regulate the expression SAA expression and the metachronous development of meta- of RANTES, G-CSF, S100A8 and S100A9 in vivo, but does so in an stases, although a trend of low SAA expression together with organ-specific manner. better overall survival was observed (Figure 7a). However, S100A4

Oncogene (2015) 424 – 435 & 2015 Macmillan Publishers Limited S100A4-induced SAA expression promotes metastasis MT Hansen et al 431

SAA3 S100A9 SAA1 RANTES *** 10 100 * 10 *** 25 G-CSF 8 80 8 20 *** *** 6 60 6 15 S100A8 4 40 4 10 ** ** 2 20 2 5 (fold induction) (fold (fold induction) (fold (fold induction) (fold ns induction) (fold 0 0 0 0 Transactivation in liver Transactivation Transactivation in liver Transactivation Transactivation in liver Transactivation Transactivation in liver Transactivation

PBS PBS PBS PBS PBS PBS S100A4 S100A8 S100A9 S100A4 S100A8 S100A9 S100A4 S100A4 S100A4 S100A4

100 S100A9 SAA1 G-CSF 4 SAA3 5 5 ** 80 ** RANTES 4 ns 4 S100A8 3 60 *** 3 ns ** ns ns 3 2 40 2 2 1 20 1 (fold induction) (fold 1 (fold induction) (fold (fold induction) (fold (fold induction) (fold 0 0 0 0 Transactivation in lungs Transactivation Transactivation in lungs Transactivation Transactivation in lungs Transactivation Transactivation in lungs Transactivation PBS PBS PBS PBS PBS PBS S100A4 S100A8 S100A9 S100A4 S100A8 S100A9 S100A4 S100A4 S100A4 S100A4

PBS S100A4 S100A8 S100A9

SAA3

Figure 6. S100 proteins induce SAA and cytokine expression in an organ-specific manner. RecS100A4, recS100A8 and recS100A9 was injected intravenously into mice for 2 weeks. Expression of SAA1 (a and e), SAA3 (b and f), RANTES and G-CSF (c and g), and S100A8 and S100A9 (d and h) in lungs, liver and spleen was then assessed using qPCR, relative to the mean of the PBS control (fold induction). (i) SAA3 protein was detected in western blots of serum prepared from animals treated with recS100A4, recS100A8 and recS100A9 proteins. Seven or eight mice were used in each group. expression was higher in primary tumors that later developed chemokines that recruit and activate inflammatory cells, and distant metastases, and was significantly associated with reduced promote tumor cell proliferation, angiogenesis and metastasis.35 overall survival (Figure 7b). Importantly, high expression of both Several cytokines produced in response to S100A4/SAA are SAA and S100A4 was the best predictor of poor overall survival involved in the recruitment of myeloid cells and regulatory (Figure 7c). These findings are underscored by the 5-year survival T cells. Myeloid cells recruited to the tumor microenvironment rates, which showed substantially lower survival for patients suppress tumor immunity, and promote angiogenesis and with both high SAA and high S100A4 expression in their tumors metastasis.36 Regulatory T cells that infiltrate tumors can compared with patients with elevated levels of only one of promote metastasis by secreting RANKL.37 Furthermore, the these factors, or with low expression of both (Table 2). Thus, recruitment of myeloid cells is also a key event in the S100A4 transcript levels in primary colon tumors predict overall establishment of premetastatic niches,15 and S100A8 can induce survival, and this is further enhanced when combined with SAA3 during premetastatic niche formation in the lungs.14 SAA expression. SAA proteins promoted MMP expression in tumor cells, and increased their motility and adhesion to fibronectin. Fibronectin accumulates in premetastatic lungs, and promotes metastasis DISCUSSION formation.15 SAA-induced binding to fibronectin may therefore Correlative studies with human tumors and functional experi- support recruitment of tumor cells to sites of metastasis formation. ments in animals link S100A4 expression with metastasis.34 The Binding of tumor cells to fibronectin also releases them from mechanism remains unresolved. Here we show that S100A4 dormancy,38 suggesting that enhanced binding to fibronectin stimulates expression of inflammation-associated genes in tumor might promote outgrowth of metastases by suppressing cells in vitro and in non-transformed cells in vivo. Two of these quiescence. genes, namely SAA1 and SAA3, profoundly influenced the For the first time, we show that SAA proteins positively regulate metastatic properties of tumor cells in vitro and in vivo. their own expression, but in a TLR-4-independent manner in VMR Consistently, coordinate expression of S100A4 and SAA in cells. SAAs signal via several receptors, including TLR-2, TLR-4, human colorectal patient samples was indicative of poor RAGE, FPRL-1/ALX, SR-B1/ABCA1, CD55 and TANIS.4 Thus, a variety prognosis. The data suggest that S100A4 stimulates metastasis of alternative receptors other than TLR-4 could, in principle, be through triggering an inflammatory response, mediated at least in involved in self-regulated SAA expression. Expression of S100A8 part by SAA proteins. and S100A9 was also stimulated by SAA, identifying a positive T-cell infiltration into both primary tumors and metastases is a feedforward mechanism in which S100A4-mediated activation of feature of S100A4-mediated metastasis.28 The upregulation of SAA leads to self-amplifying accumulation of SAA and S100 inflammatory response proteins by S100A4, and the stimulation of proteins, and enhanced production of downstream targets such as metastasis by SAA we report here provide further evidence that metastasis-promoting cytokines and MMPs (Figure 3e). Thus, S100A4 promotes metastasis at least in part by acting on the S100A4 produced by activated fibroblasts, macrophages and immune system. These inflammatory response proteins could lymphocytes39,40 could trigger self-amplifying expression of SAA promote metastasis in several ways. SAA induces neutrophilia,9 and S100 proteins in both primary tumors and metastatic sites, in and tumor-associated neutrophils produce cytokines and turn promoting metastasis formation. This is a further example of

& 2015 Macmillan Publishers Limited Oncogene (2015) 424 – 435 S100A4-induced SAA expression promotes metastasis MT Hansen et al 432 SAA Table 2. Five-year survival (5YS) rates of patients with primary colon 1.0 adenocarcinomas correlated with the expression of SAA and S100A4

Marker Expression 5-Year survival (%) ±s.d. 0.8 low SAA Low 89.3 0.058 High 71.9 0.079 0.6 high S100A4 Low 86.2 0.048 0.4 High 44.4 0.166 based on SAA Overall survival Overall SAA þ S100A4 Both low 92.0 0.054 0.2 SAA low S100A4 high 66.7 0.272 SAA high S100A4 low 80.8 0.077 Both high 33.3 0.192 0 The combination of both, SAA and S100A4, add benefit for prognostica- 0 50 100 150 200 tion. If both markers are low, the 5YS is 92% compared with 89% and 86% Months for SAA and S100A4, respectively. If both markers are high, the 5YS is 33% compared with 72% and 44% for SAA and S100A4, respectively. The most S100A4 significant values are indicated in boldface. 1.0 P=0.001 Of the nine S100 family members tested, only the proinflam- 0.8 matory proteins S100A4, S100A8 and S100A9 were able to activate SAA expression in vitro in VMR tumor cells. In vivo, only S100A4 low and S100A8 had any significant effect on SAA expression. 0.6 Surprisingly, S100A12 is also a proinflammatory member of the S100 family, yet had no effect on SAA induction. These data 0.4 suggest that the ability to promote metastasis formation through high inducing expression of SAA proteins is restricted to only a few Overall survival Overall

based on S100A4 members of the S100 family. 0.2 Ectopic SAA expression in VMR tumor cells markedly extended the range of organs in which VMR metastases were found. 0 Consistently, mice bearing overt subcutaneous Lewis lung carcinoma tumors have been reported to exhibit enhanced 0 50 100 150 200 metastatic colonization of the lung when SAA1-expressing Lewis Months lung carcinoma cells were additionally injected intravenously.41 Neither the effect of SAA1 on metastasis in non-tumor-primed SAA and S100A4 mice, metastasis formation in organs other than the lung nor 1.0 P=0.007 immune cell inﬁltration into the metastatic lesions was examined in this study. In contrast to the broad range of organs in which SAA- 0.8 both low expressing tumor cells form metastases, the S100 proteins that induce SAA expression have a more organ-restricted effect on 0.6 metastasis formation. In studies on Lewis lung carcinoma, low S100A4, high SAA transactivation of SAA3 by S100A8 but not S100A9 has been reported during premetastatic niche formation.14 Furthermore, 0.4 low SAA, high S100A4 metastasis of VMR cells to the liver and lung is increased when

Overall survival Overall 26 both high S100A4 is injected intravenously. As S100 family members 0.2 interact with various receptors such as RAGE and TLR-4 to exert 20 based on SAA and S100A4 their effects, the ability of an individual S100 protein to regulate SAA expression is likely to be organ and cell-type dependent, and 0 determined by the specific S100 receptor(s) expressed. We 0 50 100 150 200 therefore suggest that the exact cells and organs in which SAA Months expression is induced by particular S100 proteins determines the organ specificity of metastasis formation by preconditioning these Figure 7. Coincident expression of SAA and S100A4 expression in organs for metastatic growth. Moreover, we suggest that S100A4 the primary tumors of patients with colon adenocarcinomas predict expression either by tumor cells or stromal cells may be involved poor overall survival. SAA and S100A4 expression in the primary in premetastatic niche formation in the liver and in the lung.28 tumors of 60 patients with primary colon adenocarcinomas that had S100A8 has been reported to induce SAA3 expression in lungs not metastasized at the time of surgery was determined by qPCR. 14 Kaplan–Meier analyses for overall survival were performed for SAA but not liver. Here we observed that S100A8 significantly expression (a), S100A4 expression (b) and coincident SAA and upregulated SAA3 in the liver, and only tendentially in the lung S100A4 expression (c). The survival time was set as the time (Figure 6). There are several conceivable reasons for this difference between resection of the primary tumor and the last discrepancy. The previously published study used ex vivo incuba- contact (censored data) or death. tion of the organs with recS100A8,14 whereas our experiments were performed in vivo. Furthermore, we analyzed SAA3 positively acting feedforward loops that establish an inflammatory expression using qPCR, whereas the other study used reverse milieu and provide a potent stimulus to tumor growth at primary transcription–polymerase chain reaction (RT–PCR). The in vivo and secondary sites.34 experiments in the previous study used heparin-conjugated

Oncogene (2015) 424 – 435 & 2015 Macmillan Publishers Limited S100A4-induced SAA expression promotes metastasis MT Hansen et al 433 S100A8,14 whereas we used native protein. Mouse strain-specific TCCTTC-30); S100A4 (forward, 50-TTGTGTCCACCTTCCACAAA-30; reverse, 50- differences may also be operative. GCTGTCCAAGTTGCTCATCA-30); S100A8 (forward, 50-CCTTGCGATGGTGATA The upregulation of genes involved in inflammation in response AAAGTG-30; reverse, 50-CCCAGCCCTAGGCCAGAA-30); S100A9 (forward, 50- CAAAGGCTGTGGGAAGTAATTAAGA-30; reverse, 50-AGCCATTCCCTTTAGACT to S100A4 is consistent with our findings that S100A4 is involved 0 in rheumatoid arthritis, psoriasis and idiopathic inflammatory TGGT-3 ). myopathies.17,18 Interestingly, SAA expression is also strongly associated with rheumatoid arthritis,42 and induces angiogenesis, Sequence of primers used for human qPCR cytokine expression, leukocyte recruitment and MMP-mediated For SAA and S100A4, amplicons of 153 and 124 bp were produced, matrix degradation in this context by NF-kB signaling.43 In further respectively. The primers used were as follows: SAA (forward, 50- work, we will determine whether S100A4 has a functional role in TGGTTTTCTGCTCCTTGGTC-30; reverse, 50-CCCGAGCATGGAAGTATTTG-30) rheumatoid arthritis by inducing the expression of SAA proteins. (SYBR green format); S100A4 (forward, 50-CTCAGCGCTTCTTCTTTC-30, primer, 50-GGGTCAGCAGCTCCTTTA-30); FITC-labeled probe (50-TGTGATGGT In conclusion, SAA proteins produced in response to S100A4 0 0 have the potential to stimulate metastasis formation in several GTCCACCTTCCACAAGT-3 ); LCRed640-labeled probe (5 -TCGGGCAAAGAG GGTGACAAGT-30) (hybridization probes format). ways. As S100A4 and SAA exhibit several overlapping effects including activation of MMPs and cytokines, we suggest that the metastasis-promoting effects of S100A4 are mediated at least in Recombinant proteins and in vivo application part by the SAA proteins they induce. Our data are consistent with Active oligomeric S100A4 protein was obtained as described.47 Murine the notion that systemic effects of S100A4 transcriptional targets, cDNAs for SAA1 and SAA3 were inserted with in-frame V5-His tags into the most notably SAAs, can act to induce an inflammatory reaction in bacterial expression vector pQE30 (Qiagen, Copenhagen, Denmark) and specific organs that prepares the organ microenvironment for used to produce recombinant (recSAA) protein in Escherichia coli according to the manufacturer’s protocol. Human recSAA1 was purchased from metastasis formation. Future studies will focus on understanding Prospec (East Brunswick, NJ, USA). Tail vein injections and intradermal how SAA acts on the immune system to promote metastasis injections using rodent intradermal delivery devices (Becton Dickinson, formation, and whether SAA could serve as a therapeutic target Albertslund, Denmark) were performed on C57Bl/6 mice aged 8–12 weeks for combating metastasis. five times per week for 2 weeks (25 mg protein per injection). Phosphate- buffered saline-injected animals served as controls. Excised organs and serum were snap frozen. MATERIALS AND METHODS Cell lines and reagents Antibodies and western blotting 21 VMR, CSML0 and CSML-100 cells were cultured as described. Human Monoclonal anti-human S100A4 antibodies39 were used in 10-fold excess breast (MB-MDA-231, MB-MDA-435, MCF-7 and MCF-s1) and colon cancer (10 mg/ml) over recS100A4 (incubation time 20 h) in S100A4 function cell lines (SW480 and SW620) were obtained from the Danish Cancer inhibition experiments. Other antibodies used were rabbit anti-mouse Society’s biobank. Mouse embryonic fibroblasts, bone marrow SAA3 serum,30 rabbit anti-mouse SAA1 antibodies (a kind gift from Lars Bo 2,9,44 macrophages and splenic T cells were isolated as described. Murine Nielsen, Copenhagen University, København K, Denmark),48 mouse SAA1 and SAA3 cDNAs (ImaGenes, Berlin, Germany) were inserted into the monoclonal anti-actin antibody AC-40 (Sigma-Aldrich), rabbit anti-MMP-3 pBABEpuro retroviral vector (Cell Biolabs Inc., San Diego, CA, USA). antibodies (Epitomics, Burlingame, CA, USA) and rabbit anti-NF-kB p65 Retroviral transduction and selection of infected cells was performed as antibodies (Santa Cruz, Heidelberg, Germany). The high homology 45 described. VMR-IkB clones were generated by transfection of VMR cells between SAA1 and SAA2 means that the SAA1 antibody also likely with a dominant-negative IkBa expression construct (kindly provided by detects SAA2; therefore, we refer to the protein detected by this antibody Dr M Glukhova, Institute Curie, Paris, France). PMB was from (Gibco, as SAA1/2. Western blotting was performed using standard methods. Naerum, Denmark). AG1478, AG879, CAPE and the IRAK 1/4 were from (Sigma-Aldrich, Broendby, Denmark). The TLR-4 inhibitor CLI-095 was from Adhesion and motility assays. Adhesion assays and monolayer wound Invitrogen (Taastrup, Denmark) and the MEK 1/2 inhibitor U0126 was from healing motility assays were performed as described.49,50 Relative Promega (Stockholm, Sweden). migration was calculated as the mean area of wound remaining at a given time point relative to the original wound area. The three- Microarray analysis dimensional Matrigel invasion assay has been described.51 RNA was prepared from snap-frozen cells using PeqGold according to the manufacturer’s instructions (PeqLab, Erlangen, Germany). The probe Chemotactic cell migration. Transwell plates with 12 inserts (pore size labeling, microarray chips, hybridization and scanning has been described 8 mm) were coated with fibronectin. CSML100 cells (5 Â 105 per well) in previously.46 serum-free Dulbecco’s modified Eagle’s medium were added. Lower chambers contained serum-free media without or with recombinant proteins. After incubation at 37 1C, cells penetrating the insert membrane RNA purification and quantitative real-time PCR were stained with Diff-Quick (Dade Behring, Marburg, Germany). For each First-strand cDNA synthesis with RNA extracted using the Nucleospin insert, migrated cells were counted in 20 independent microscopic fields. Triprep kit (Macherey-Nagel, Du¨ren, Germany) was performed using Two independent experiments were performed, each with triplicate SuperScript II RT (Invitrogen) with random primers, in duplicate or samples. triplicate. Each qPCR (Fast SYBR Green Master Mix kit (Roche Applied Science, Indianapolis, IN, USA) in a LightCycler 2.0 instrument (Roche DiagnosticsA/S, Hvidovre, Denmark)) was run at least in duplicate. Zymography Cells were grown to 90% confluence. The medium was exchanged with serum-free medium containing recombinant proteins. CM was harvested 6 Sequences of primer pairs used for mouse qPCR or 24 h later, filtered through 0.45 mm membrane filters and concentrated qPCR was carried out using the following primers: GAPDH (forward, 50-CCA using Vivapore or Vivaspin concentrators (Vivascience Ltd, Herlev, GCAAGGACACTGAGCAA-30; reverse, 50-GGGATGGAAATTGTGAGGGA-30); Denmark). The cells were subsequently trypsinized and counted to mouse SAA1 (forward, 50-AGTCTGGGCTGCTGAGAAAA-30; reverse, 50-GGCA normalize the quantity of proteins in the CM used. The concentrated CM GTCCAGGAGGTCTGTA-30); mouse SAA3 (forward, 50-ACATGTGGCGAGCCTA was assayed for protease activity using gelatin and casein zymography.52 CTCT-30; reverse, 50-GAGTCCTCTGCTCCATGTCC-30); MMP2 (forward, 50-CAC ACCAGGTGAAGGATGTG-30; reverse, 50-AGGGCTGCATTGCAAATATC-30); MMP9 (forward, 50-TGAATCAGCTGGCTTTTGTG-30; reverse, 50-ACCTTCCAGT Animal tumor experiments AGGGGCAACT-30); MMP13 (forward, 50-ACCCAGCCCTATCCCTTGAT-30; A/Sn mice aged 8–11 weeks were maintained according to European reverse, 50-TTTGGGATGCTTAGGGTTGG-30); G-CSF (forward, 50-CTCAACTTT Laboratory Animal Science Association guidelines. Tumor cells (0.5 Â 106 CTGCCCAGAGG-30; reverse, 50-TCCAGGGACTTAAGCAGGAA-30); RANTES cells per mouse) were injected subcutaneously or intravenously. Experi- (forward, 50-CATATGGCTCGGACACCACT-30; reverse, 50-ACACACTTGGCGGT ments were terminated when the first animal became moribund. Organs

& 2015 Macmillan Publishers Limited Oncogene (2015) 424 – 435 S100A4-induced SAA expression promotes metastasis MT Hansen et al 434 were macroscopically assessed for the presence of metastases. Lungs and 13 Okamoto H, Katagiri Y, Kiire A, Momohara S, Kamatani N. Serum amyloid A livers were analyzed using standard histology. Metastases per unit area activates nuclear factor-kappaB in rheumatoid synovial fibroblasts through were counted in tissue sections. Immunohistochemistry was performed binding to receptor of advanced glycation endproducts. J Rheumatol 2008; 35: using anti-CD45 antibodies (BD Pharmingen, San Diego, CA, USA). 752–756. 14 Hiratsuka S, Watanabe A, Sakurai Y, Akashi-Takamura S, Ishibashi S, Miyake K et al. Human tumor analysis The S100A8-serum amyloid A3-TLR4 paracrine cascade establishes a pre-metastatic phase. Nat Cell Biol 2008; 10: 1349–1355. Primary adenocarcinomas from 60 colon cancer patients were obtained 15 Kaplan RN, Riba RD, Zacharoulis S, Bramley AH, Vincent L, Costa C et al. VEGFR1- after written consent and ethical approval, and snap frozen. All patients positive haematopoietic bone marrow progenitors initiate the pre-metastatic were untreated previously, did not have a history of familial colon cancer niche. Nature 2005; 438: 820–827. and underwent surgical resection at the Robert-Ro¨ssle Cancer Hospital, 16 Grigorian M, Ambartsumian N, Lukanidin E. Metastasis-inducing S100A4 protein: Berlin, Germany. None had metastases at the time of surgery, but 23 implication in non-malignant human pathologies. Curr Mol Med 2008; 8: 492–496. developed distant metastases metachronously. Tumor cells were micro- 17 Zibert JR, Skov L, Thyssen JP, Jacobsen GK, Grigorian M. Significance of the dissected from serial cryosections, and then total RNA was isolated. 53 S100A4 protein in psoriasis. J Invest Dermatol 2010; 130: 150–160. Duplicate qRT–PCR was performed as described previously. Calibrator 18 Cerezo LA, Kuncova´ K, Mann H, Tomcı´kM,Za´mecnı´k J, Lukanidin E et al. The cDNAs for the SAA-RT–PCR were derived from HepG2 cells, and for the metastasis promoting protein S100A4 is increased in idiopathic inflammatory S100A4-RT–PCR from HCT116 cells. These were used in serial dilutions myopathies. Rheumatology (Oxford) 2011; 50: 1766–1772. simultaneously in each run. 19 Marenholz I, Heizmann CW, Fritz G. S100 proteins in mouse and man: from evolution to function and pathology (including an update of the nomenclature). Statistical analysis Biochem Biophys Res Commun 2004; 322: 1111–1122. 20 Fritz G, Botelho HM, Morozova-Roche LA, Gomes CM. Natural and amyloid self- Metastatic burden comparisons used the Mann–Whitney test. P-values for the other analyses used the two-tailed unpaired Student’s t-test. P-values assembly of S100 proteins: structural basis of functional diversity. FEBS J 2010; 277: 4578–4590. o0.05 were regarded as significant. *Po0.05, **Po0.01 and ***Po0.001. Statistical evaluation of human colorectal samples used the non- 21 Ebralidze A, Tulchinsky E, Grigorian M, Afanasyeva A, Senin V, Revazova E et al. parametric two-sided Mann–Whitney rank-sum test. Kaplan–Meier curves Isolation and characterization of a gene specifically expressed in different were evaluated with the log-rank test. Cutoffs were calculated by receive metastatic cells and whose deduced gene product has a high degree of operating characteristic analyses. homology to a Ca2-binding protein family. Genes Dev 1989; 3: 1086–1093. 22 Grigorian M, Ambartsumian N, Lykkesfeldt AE, Bastholm L, Elling F, Georgiev G et al. Effect of mts1 (S100A4) expression on the progression of human breast cancer cells. Int J Cancer 1996; 67: 831–841. CONFLICT OF INTEREST 23 Maelandsmo GM, Hovig E, Skrede M, Engebraaten O, Flørenes VA, Myklebost O The authors declare no conflict of interest. et al. Reversal of the in vivo metastatic phenotype of human tumor cells by an anti-CAPL (mts1) ribozyme. Cancer Res 1996; 56: 5490–5498. 24 Ambartsumian NS, Grigorian MS, Larsen IF, Karlstrøm O, Sidenius N, Rygaard J ACKNOWLEDGEMENTS et al. Metastasis of mammary carcinomas in GRS/A hybrid mice transgenic for the mts1 gene. Oncogene 1996; 13: 1621–1630. We thank Ekaterina Dulina, Inge Skibshøj and Lene Bregnholt Larsen for careful 25 Davies MP, Rudland PS, Robertson L, Parry EW, Jolicoeur P, Barraclough R. technical assistance. We gratefully acknowledge funding by the European Union Expression of the calcium-binding protein S100A4 (p9Ka) in MMTV-neu (TuMIC, Health-F2-2008-201662) and INARMERA (FP7-INCO-2010-6), the Danish transgenic mice induces metastasis of mammary tumours. Oncogene 1996; 13: Cancer Society and the Dansk Kræftforsknings Fond. 1631–1637. 26 Schmidt-Hansen B, Klingelho¨fer J, Grum-Schwensen B, Christensen A, Andresen S, Kruse C et al. Functional significance of metastasis-inducing S100A4 (Mts1) in REFERENCES tumor–stroma interplay. J Biol Chem 2004; 279: 24498–24504. 1 Langley RR, Fidler IJ. The seed and soil hypothesis revisited—the role of tumor– 27 Helfman DM, Kim E, Lukanidin E, Grigorian M. The metastasis associated protein stroma interactions in metastasis to different organs. Int J Cancer 2011; 128: S100A4. Br J Cancer 2005; 92: 1955–1958. 2527–2535. 28 Grum-Schwensen B, Klingelho¨fer J, Grigorian M, Almholt K, Nielsen BS, 2 DeNardo DG, Johansson M, Coussens LM. Immune cells as mediators of solid Lukanidin E et al. Lung metastasis fails in MMTV-PyMT oncomice lacking S100A4 tumor metastasis. Cancer Metast Rev 2008; 27: 11–18. due to a T-cell deficiency in primary tumors. Cancer Res 2010; 70: 936–947. 3 Fridman WH, Galon J, Page`s F, Tartour E, Saute`s-Fridman C, Kroemer G. Prognostic 29 Klingelho¨fer J, Grum-Schwensen B, Beck MK, Knudsen RS, Grigorian M, and predictive impact of intra- and peritumoral immune infiltrates. Cancer Res Lukanidin E et al. Anti-S100A4 antibody suppresses metastasis formation by 2011; 71: 5601–5605. blocking stroma cell invasion. Neoplasia 2012; 14: 1260–1268. 4 Malle E, Sodin-Semrl S, Kovacevic A. Serum amyloid A: an acute-phase protein 30 Lin Y, Rajala MW, Berger JP, Moller DE, Barzilai N, Scherer PE. Hyperglycemia- involved in tumour pathogenesis. Cell Mol Life Sci 2009; 66: 9–26. induced production of acute phase reactants in adipose tissue. J Biol Chem 2001; 5 Larson MA, Wei SH, Weber A, Weber AT, McDonald TL. Induction of human 276: 42077–42083. mammary-associated serum amyloid A3 expression by prolactin or lipopoly- 31 Ehrchen JM, Sunderko¨tter C, Foell D, Vogl T, Roth J. The endogenous Toll-like saccharide. Biochem Biophys Res Commun 2003; 301: 1030–1037. receptor 4 agonist S100A8/S100A9 (calprotectin) as innate amplifier of infection, 6 Furlaneto CJ, Campa A. A novel function of serum amyloid A: a potent stimulus for autoimmunity, and cancer. J Leukoc Biol 2009; 86: 557–566. the release of tumor necrosis factor-alpha, interleukin-1beta, and interleukin-8 by 32 O’Neill LA. Signal transduction pathways activated by the IL-1 receptor/toll-like human blood neutrophil. Biochem Biophys Res Commun 2000; 268: 405–408. receptor superfamily. Curr Top Microbiol Immunol 2002; 270:47–61. 7 He R, Shepard LW, Chen J, Pan ZK, Ye RD. Serum amyloid A is an endogenous 33 Klingelho¨fer J, Møller HD, Sumer EU, Berg CH, Ambartsumian N, Grigorian M et al. ligand that differentially induces IL-12 and IL-23. J Immunol 2006; 177: 4072–4079. EGFR ligands as new extracellular targets for the metastasis-promoting S100A4 8 Song C, Hsu K, Yamen E, Yan W, Fock J, Witting PK et al. Serum amyloid A protein. FEBS J 2009; 276: 5936–5948. induction of cytokines in monocytes/macrophages and lymphocytes. Athero- 34 Lukanidin E, Sleeman JP. Building the niche: The role of the S100 proteins in sclerosis 2009; 207: 374–383. metastatic growth. Semin Cancer Biol 2012; 22: 216–225. 9 He RL, Zhou J, Hanson CZ, Chen J, Cheng N, Ye RD. Serum amyloid A induces 35 Gregory AD, Houghton AM. Tumor-associated neutrophils: new targets for cancer G-CSF expression and neutrophilia via Toll-like receptor 2. Blood 2009; 113: 429–437. therapy. Cancer Res 2011; 71: 2411–2416. 10 Yamada T. Serum amyloid A (SAA): a concise review of biology, assay methods 36 Schmid MC, Varner JA. Myeloid cells in the tumor microenvironment: modulation and clinical usefulness. Clin Chem Lab Med 1999; 37: 381–388. of tumor angiogenesis and tumor inflammation. J Oncol 2010; 2010: 201026. 11 Lee HY, Kim MK, Park KS, Bae YH, Yun J, Park JI et al. Serum amyloid A stimulates 37 Tan W, Zhang W, Strasner A, Grivennikov S, Cheng JQ, Hoffman RM et al. Tumour- matrix metalloproteinase-9 upregulation via formyl peptide receptor like-1- infiltrating regulatory T cells stimulate mammary cancer metastasis through mediated signaling in human monocytic cells. Biochem Biophys Res Commun RANKL-RANK signalling. Nature 2011; 470: 548–553. 2005; 330: 989–998. 38 Aguirre-Ghiso JA, Liu D, Mignatti A, Kovalski K, Ossowski L. Urokinase receptor 12 Zhao Y, Zhou S, Heng CK. Celecoxib inhibits serum amyloid a-induced matrix and fibronectin regulate the ERK(MAPK) to p38(MAPK) activity ratios that metalloproteinase-10 expression in human endothelial cells. J Vasc Res 2009; determine carcinoma cell proliferation or dormancy in vivo. Mol Biol Cell 2001; 12: 46: 64–72. 863–879.

Oncogene (2015) 424 – 435 & 2015 Macmillan Publishers Limited S100A4-induced SAA expression promotes metastasis MT Hansen et al 435 39 Cabezoń T, Celis JE, Skibshøj I, Klingelho¨fer J, Grigorian M, Gromov P et al. 47 Novitskaya V, Grigorian M, Kriajevska M, Tarabykina S, Bronstein I, Berezin V et al. Expression of S100A4 by a variety of cell types present in the tumor micro- Oligomeric forms of the metastasis-related Mts1 (S100A4) protein stimulate environment of human breast cancer. Int J Cancer 2007; 121: 1433–1444. neuronal differentiation in cultures of rat hippocampal neurons. J Biol Chem 2000; 40 Klingelhofer J, Sˇenolt L, Baslund B, Nielsen GH, Skibshøj I, Pavelka K et al. Up- 275: 41278–41286. regulation of metastasis-promoting S100A4 (Mts1) in rheumatoid arthritis: putative 48 Bang CA, Bro S, Bartels ED, Pedersen TX, Nielsen LB.. Effect of uremia on HDL involvement in the pathogenesis of RA. Arthritis Rheum 2007; 56: 779–789. composition, vascular inflammation, and atherosclerosis in wild-type mice. 41 Sung HJ, Ahn JM, Yoon YH, Rhim TY, Park CS, Park JY et al. Identification and Am J Physiol Renal Physiol 2007; 293: 1325–1331. validation of SAA as a potential lung cancer biomarker and its involvement in 49 Baumann P, Cremers N, Kroese F, Orend G, Chiquet-Ehrismann R, Uede T et al. metastatic pathogenesis of lung cancer. J Proteome Res 2011; 10: 1383–1395. CD24 expression causes the acquisition of multiple cellular properties associated 42 Cunnane G. Amyloid precursors and amyloidosis in inflammatory arthritis. with tumor growth and metastasis. Cancer Res 2005; 65: 10783–10793. Curr Opin Rheumatol 2001; 13: 67–73. 50 Forst B, Hansen MT, Klingelho¨fer J, Møller HD, Nielsen GH, Grum-Schwensen B 43 Mullan RH, Bresnihan B, Golden-Mason L, Markham T, O’Hara R, FitzGerald O et al. et al. Metastasis-inducing S100A4 and RANTES cooperate in promoting tumor Acute-phase serum amyloid A stimulation of angiogenesis, leukocyte recruitment, progression in mice. PLoS One 2010; 5: e10374. and matrix degradation in rheumatoid arthritis through an NF-kappaB-dependent 51 Ambartsumian N, Christensen CRL, Lukanidin E. Endothelial cell invasion signal transduction pathway. Arthritis Rheum 2006; 54: 105–114. assay. Cell Biology, Vol. 1. Academic Press: Waltham, MA, USA 2005, pp 44 Grigorian M, Tulchinsky E, Burrone O, Tarabykina S, Georgiev G, Lukanidin E. 363–366. Modulation of mts1 expression in mouse and human normal and tumor cells. 52 Schmidt-Hansen B, Klingelho¨fer J, Grum-Schwensen B, Christensen A, Andresen S, Electrophoresis 1994; 15: 463–468. Kruse C et al. Functional significance of metastasis-inducing S100A4(Mts1) in 45 Pear W, Nolan GP, Scott ML, Baltimore D. Production of high titer helper-free tumor–stroma interplay. J Biol Chem 2004; 279: 24498–24504. retroviruses by transient transfection. Proc Natl Acad Sci USA 1993; 90: 8392–8396. 53 Stein U, Walther W, Arlt F, Schwabe H, Smith J, Fichtner I et al. MACC1, a newly 46 Møller HD, Ralfkjær U, Cremers N, Frankel M, Pedersen RT, Klingelho¨fer J et al. Role identified key regulator of HGF-MET signaling, predicts colon cancer metastasis. of fibulin-5 in metastatic organ colonization. Mol Cancer Res 2011; 9: 553–563. Nat Med 2009; 15: 59–67.

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

& 2015 Macmillan Publishers Limited Oncogene (2015) 424 – 435