SMYD3 Promotes Cancer Invasion by Epigenetic Upregulation of the Metalloproteinase MMP-9

Cancer progression is a multistep process resulting from altered gene expression through genetic and epigenetic mechanisms (1, 2). As epigenetic events might be reversed using targeted therapies, there is intensive effort to understand the interplay between these mechanisms in the transcriptional control of key regulatory molecules and the altered epigenetic landscape of cancer cells (2). To investigate epigenetic events in cancer progression, we studied a transformation model offering several unique experimental features. Infection by intracellular Theileria parasites causes a lymphoproliferative disease in cows with clinical features similar to some human leukemia. T. annulata infects mainly B cells and macrophages, whereas T. parva infects B and T lymphocytes (3, 4). Theileria-infected cells are transformed and immortalized; displaying uncontrolled proliferation independent of exogenous growth factors in vitro and increased ability to migrate and form metastases in immunodeficient mice (5, 6, 7). Interestingly, Theileria-dependent transformation is reversible; animals treated with the theilericidal drug buparvaquone are cured in most cases (3, 8, 9). When Theileria-infected cells are treated in vitro with buparvaquone, the intracellular parasite disappears from the host leukocyte, which either enters apoptosis or stops proliferating, regaining growth factor dependence (8). Studies showed that Theileria alters host signal transduction pathways, but there is no evidence of permanent host genome changes (4). Theileria infection activates the c-Jun-NH₂-kinase (JNK) pathway and the activator protein (AP-1) transcription factor (10, 11), which are critical for both the survival and invasiveness of transformed B lymphocytes (5). Theileria also induces the IkappaB kinase (IKK) pathway leading to NF-κB activation (12) and the TGF-β pathway implicated in virulence (6). This system offers a unique model to address reversible, epigenetic changes involved in tumorigenesis.

Matrix metalloproteinase (MMP) proteins may function as mediators of metastasis of Theileria-infected cells (13–15). The large family of MMP proteases play critical functions in development and cancer progression (16). In particular, MMP-9 plays an important role in tumorigenesis and metastasis formation by regulating tumor growth, angiogenesis, cell migration, and invasion (16). MMP-9 is a collagenase that degrades extracellular matrix (ECM) components including collagens and laminin. Its role in cancer progression is related to ECM degradation and growth factor release (17). The MMP-9 gene is mainly regulated at the transcriptional level by transactivators binding to the promoter, including AP-1 and NF-κB (18, 19). Recent studies provided insights into how MMP-9 expression is regulated in different cancer types and the role of chromatin modifications (20). MMP-9 transcription involves a stepwise, coordinated recruitment of activators, chromatin modifiers, coactivators, and general transcription factors to its promoter (21, 22).

Although MMP-9 was linked to the invasive phenotype of Theileria-transformed cells (13–15), the involvement of epigenetic regulatory mechanisms (and their reversibility) has never been explored in Theileria-infected leukocytes. We studied MMP-9 expression in Theileria-infected cells as a model for epigenetic regulation in leukocyte transformation. The invasive B-cell lymphosarcoma cell line TBL3 was generated by in vitro infection, with T. annulata, of immortalized BL3, a bovine B lymphosarcoma cell line (23). TBL3 cells have a higher invasive capacity than BL3 cells and elevated MMP-9 expression (15). Here, we show that Theileria infection induces the reversible expression of the histone methyltransferase SMYD3 (SET and MYND domain containing 3). SMYD3 encodes a histone H3K4 di- and trimethyltransferase involved in tumor proliferation in colorectal, hepatocellular, and breast carcinomas (24, 25). SMYD3 binds specific DNA sequences, 5′-CCCTCC-3′ or 5′-GGAGGG-3′, in the promoter region of target genes, leading to transcriptional activation (24). However, the role of SMYD3 in cancer progression is unclear and few direct target genes are known. We show that SMYD3 protein binds to the MMP-9 promoter and regulates expression in Theileria-infected cells and human tumor models. Moreover, SMYD3 knockdown led to decreased H3K4me3 marks on the MMP-9 promoter and reduced MMP-9. Interestingly, SMYD3 levels impact the transformed phenotypes and tumor invasiveness. Thus, our work identifies SMYD3 as a novel regulator of the MMP-9 metastatic gene affecting tumor migration and invasion in vitro and in vivo, thereby highlighting its potential as a therapeutic target to treat metastatic disease.

Parental BL3 and infected TBL3 cells, previously described (23), were cultured at 37°C in RPMI-1640 (Gibco) supplemented with 2 mmol/L l-glutamine, 10% fetal calf serum (FCS), 100 units/mL penicillin/streptomycin, 10 mmol/L HEPES, and 50 mmol/L 2-mercaptoethanol. HT1080-2.2 cells (with an integrated 2.2 Kb MMP9 promoter-luciferase) obtained from Dr. C. Yan (Albany Medical College, Albany, NY; ref. 26) were cultured at 37°C in Dulbecco's Modified Eagle's Medium (DMEM) with glutamax and high glucose, supplemented with antibiotics and 10% FCS. HT1080 shRNA cell lines were generated by transfecting the pSuper vector (Ctrl-sh) or the SMYD3-sh pSuper vector, and selecting with neomycin (300 μg/mL). Appropriate HT1080 clones were selected for further use following SMYD3-immunoblotting analysis. Cells were treated with buparvaquone (BW720c; Calbiochem) at 50 ng/mL for 64 hours or with phorbol 12-myristate 13-acetate 9 (PMA; Sigma p1585) 50 ng/mL for 8 hours. Compositions of all buffers are detailed in Supplementary Material.

A total of 1 × 10⁴ cells were plated in 96-well plates in triplicate, and buparvaquone was added at different concentrations. After 48 hours, cell viability was measured by the Cell proliferation Kit II–XTT (Roche) and the GloMax-Multi Detection System (Promega).

Bovine TBL3 and BL3 cells (5 × 10⁴) were incubated with/without buparvaquone, resuspended in 1% FCS and seeded onto Matrigel (BD Biosciences; 70 μL) in Boyden chamber, BD Falcon Cell Culture Inserts (8-μm pores; BD Biosciences). Lower chambers contained 20% FCS medium as a chemoattractant. After 24 hours at 37°C, migrated cells were stained with crystal violet and counted under the microscope (40× objective, at least 10 fields per filter). To test migration of human HEK293 or HT1080 cells, cells were trypsinized and placed in the upper chamber (1 × 10⁴ cells) coated with 10 μg/mL collagen I in media containing 0.3% and 20% FCS in the lower chamber. After 16 hours, cells were fixed with 4% formaldhehyde and migrated cells were stained and counted. All experiments were carried out in triplicate. Migration index is the number of migrated cells per filter (control set to 1).

Cells were suspended in 0.35% agarose containing 10% FCS and plated above 0.5% agarose at 1 × 10⁴ cells per 6-well plate. After 10 to 14 days, colonies were stained with crystal violet and scored. Experiments were carried out in triplicate.

Total RNA was extracted by an RNeasy Plus Mini Kit (Qiagen). cDNA synthesis was carried out with Reverse Transcriptase (RT) Superscript III (Invitrogen) and semiquantitative PCR analysis with GoTaq polymerase (Promega). Quantitative PCR (qPCR) amplification was carried out in the ABI7500 machine (Applied Biosystems) using Sybr Green (Applied Biosystems). See Supplementary Table S1 for primer sequences. Relative quantities of mRNA were calculated with the ΔC_t method.

Proteins were extracted with Laemmli lysis buffer, resolved by SDS-PAGE, and transferred to nitrocellulose membranes, before incubation overnight at 4°C with antibodies against MMP-9, SMYD3, or β-actin (see Supplementary data), followed by secondary antibodies. Membranes were developed by SuperSignal (Thermo Scientific).

Supernatants from cells grown in serum-free media were mixed with lysis buffer and loaded into SDS-PAGE gels, copolymerized with 0.1% gelatin. Gels were washed twice (30 minutes) in renaturing buffer (2.5% Triton X-100) and incubated at 37°C, 18 hours in a gelatinase solution. Gels were stained with 0.5% Coomassie Blue and destained with methanol:acetic acid (50:10).

HT1080 and HEK293T (human embryonic kidney transformed with the AgT) cells were transfected with Lipofectamine 2000 (Invitrogen). The pGL3-Nkx2.8–WT-SBE and pcDNA-SMYD3 vectors were obtained from Dr. Y. Furukawa (Tokyo University, Tokyo, Japan; ref. 24). The MMP-9 promoter-luciferase construct was obtained from Dr. D. Boyd (MD Anderson Cancer Center, Houston, TX).

Dual luciferase assays (Promega) were conducted by the GloMax-Multi Detection System (Promega) to measure firefly and Renilla luciferase activity. The pGL4 hRluc/TK vector was used for luciferase assay normalization.

For short-term depletion experiments, cells were transfected with siRNA oligonucleotides (60 nmol/L) Supplementary Table S1. The short hairpin RNA (shRNA) vector targeting human/bovine SMYD3 was constructed by cloning double-stranded oligonucleotides into the pSuper-vector (Oligoengine), generating siRNA directed against the sequence AGCCTGATTGAAGATTTGA. shCtrl plasmid is an empty pSuper vector.

Chromatin was extracted from 5 × 10⁶ to 20 × 10⁶ cells and crosslinked with 1% formaldehyde (10 minutes, 37°C). After incubation in cell lysis buffer 30 minutes at 4°C, nuclei were resuspended in nuclei lysis buffer containing glass beads (400 μL per mL; Sigma) before sonication (4–5 cycles of 4 minutes) using the Bioruptor (Diagenode). The INPUT sample (5%) was removed before the immunoprecipitation. Chromatin aliquots (20 or 50 μg for transcription factors) were immunoprecipitated with specific antibodies overnight at 4°C and recovered with preblocked Protein G/A sepharose beads (Sigma). Washes were as following: twice Tris/Sucrose/EDTA (TSE)-150 mmol/L NaCl buffer, once TSE-500 mmol/L NaCl buffer, once washing buffer, and twice Tris/EDTA (TE) buffer. Beads were diluted in 100 μL of dilution buffer plus 150 μL of TE/1% SDS. After decrosslinking, DNA was analyzed by qPCR. The immunoprecipitation enrichment was calculated with the ΔC_t method. Results were presented as “percentage of enrichment” of INPUT sample [100× (Output/Input)] or as “fold-enrichment” [100× (Output/Input) normalized to binding on MMP-9 promoter in control cells].

Cells were fixed in 70% ethanol and washed twice in PBS, incubated at room temperature for 20 minutes in propidium iodide buffer (50 μg/mL PI; Sigma) containing 100 μg/mL RNaseA. Cells were analyzed using the FACSCalibur flow cytometer (BD Biosciences), with data acquisition using CellQuest software.

Zebrafish (Danio rerio; wild-type AB strain, and Casper, generously given by GIS AMAGEN, Gif sur Yvette) were handled according to the European Union guidelines. Fish were kept at 28°C in aquaria with day/night light cycles. HT1080 cells were transfected with siRNA oligonucleotides. After 24 hours, cells were incubated with Dulbecco's PBS (DPBS) containing 10 μmol/L fluorescent CellTracker CM-Dil green (Invitrogen). A total of 20 to 30 cells were injected into the yolk of dechorionated and anesthetized (Tricaine methanesulfonate) 48 hours postfertilization embryos using a FemtoJet microinjector (Eppendorf). Embryos were incubated at 34°C for 3 days and analyzed by a microscope Station cell observer ZEISS and an Axio Vision software.

As MMP-9 expression has been linked to Theileria transformation (7, 15, 27), we analyzed mRNA and protein in Theileria-infected TBL3 cells and the noninfected BL3 parental cell line. We observed high MMP-9 mRNA levels in TBL3 cells but not in BL3 cells (Fig. 1A). To verify reversal by the theilericidal drug buparvaquone, we tested different concentrations of buparvaquone treatment on cell proliferation. The drug caused dramatic growth arrest in TBL3 cells, whereas BL3 growth was unaffected (Fig. 1B). Using the standard dose concentration of 50 ng/mL for 64 hours (5), we observed a 5-fold reduction in MMP-9 transcript levels in treated TBL3 cells but no effect in BL3 cells (Fig. 1C). Furthermore, we observed a similar Theileria induction and buparvaquone reversibility of MMP-9 protein and protease activity by gelatin zymography (Fig. 1D and E). Finally, we studied their invasive capacity by in vitro Matrigel invasion assays. TBL3 cells were more invasive than BL3 cells, and buparvaquone treatment reversed the phenotype (Fig. 1F). Thus, the Theileria-inducible and buparvaquone-reversible regulation of MMP-9 expression offers a model system to explore epigenetic control of this metastasis gene.

To explore epigenetic changes linked to MMP-9 regulation, we studied histone marks on the MMP-9 promoter focusing on a regulatory region (0.5 kb upstream from the transcription start site (TSS) containing transcription factor binding sites AP-1, NF-κB, and Sp1) conserved between the bovine and human promoters (Fig. 2A and Supplementary Fig. S2). We conducted chromatin immunoprecipitation (ChIP) studies in TBL3 and BL3 cells, with and without buparvaquone treatment, using antibodies specific for activating histone marks (i.e., Acetyl-H3, H3K4me2, and H3K4me3) and repressive marks (i.e., H3K9me3 and H3K27me3; Fig. 2B–F). We observed enrichment of H3K4me2 and H3K4me3 marks in this region in TBL3 as compared with BL3 cells (Fig. 2B and C). Furthermore, buparvaquone treatment dramatically decreased H3K4me3 marks in TBL3 cells (but not H3K4me2), almost to the levels observed in BL3 cells (Fig. 2B). In contrast, we observed no significant changes for Acetyl-H3, nor the repressive H3K27me3 and H3K9me3 marks, upon Theileria infection or buparvaquone treatment (Figs. 2D–F). Thus, reversible Theileria-induced MMP-9 transcription is associated with dramatic changes in H3K4 trimethylation on the MMP-9 promoter.

To identify enzymes responsible for the H3K4-methyl marks on the MMP-9 promoter, we examined expression of a panel of H3K4 methyltransferases. Strikingly, SMYD3 was the only methyltransferase gene clearly induced by Theileria in TBL3 cells (over 4-fold induction; Fig. 3A). Furthermore, buparvaquone treatment decreased SMYD3 expression to BL3 levels (Fig. 3A). The bovine SMYD3 gene encodes a protein very similar to the human enzyme, with almost complete conservation of key SET domain residues (Supplementary Fig. S1). To test SMYD3 activity in these 2 cell lines, we transfected cells with a SMYD3 reporter construct containing the Nkx2.8 promoter, a known SMYD3 target gene, coupled to luciferase (24). This SMYD3 reporter was 6-fold more active in TBL3 cells (Fig. 3B). Furthermore, there was a significant reduction (50%) in the luciferase levels when the SMYD3-binding site in the promoter was mutated (Fig. 3B; ref. 24). Thus, SMYD3 was a promising candidate as the methyltransferase responsible for elevated H3K4me2/me3 marks on the MMP-9 promoter.

SMYD3 binds to DNA via specific binding motifs (24) and bioinformatic analyses identified putative SMYD3-binding motifs in both the human and bovine MMP-9 promoters (Fig. 3C, Supplementary Fig. S2). ChIP-qPCR analysis of the bovine MMP-9 promoter, using oligonucleotides that amplify a cluster of putative SMYD3-binding sites, revealed a marked enrichment for H3K4me3 modifications in this region in Theileria-infected cells (Fig. 3D). Enrichment was reversed by buparvaquone treatment and absent in BL3 cells (Fig. 3D). These results suggested that SMYD3 binds to sites in the MMP-9 promoter to drive MMP-9 expression. Deeper investigation in TBL3/BL3 cells was hindered by the poor cross-reactivity of commercially available antibodies with bovine SMYD3.

The putative SMYD3-binding motifs are conserved in the human MMP-9 promoter (Fig. 3C, Supplementary Fig. S2), so we investigated the SMYD3-MMP-9 link further in human cancer cells. We studied a fibrosarcoma cell line, HT1080, which expresses MMP-9 and displays high invasive properties in vitro and in vivo (28, 29). We used an HT1080 derivative containing a 2.2 Kb MMP-9-Luc reporter stably integrated into the genome (26) to test SMYD3 effects on promoter activity in a chromosomal context (Fig. 4A). We also tested SMYD3 function by transient cotransfection with a MMP-9-Luc reporter plasmid (Fig. 4B). In both cases, SMYD3 transfection caused a modest, but statistically significant, increase in MMP-9 promoter activity (1.5–2.5 fold), which was enhanced by PMA treatment (2–5 fold; Fig. 4A and B). We observed similar synergistic effects on MMP-9 protein levels when SMYD3 was transfected with PMA stimulation (Fig. 4C). We carried out knockdown experiments using 2 different siRNA oligonucleotides that led to marked reduction in SMYD3 mRNA and protein (Fig. 5A–C). SMYD3 knockdown caused a significant decrease in MMP-9 mRNA and protein levels (Fig. 5B and C). Furthermore, siRNA SMYD3 caused a reduction in the activity of the MMP-9-Luc reporter integrated into the HT1080 genome (Fig. 5D). These results support the hypothesis that SMYD3 regulates MMP-9 expression in human cancer cells.

We generated stable HT1080 cell lines expressing shRNA against SMYD3 and control. SMYD3 reduction caused a decrease in MMP-9 mRNA and protein levels (Supplementary Fig. S3A and S3B). We carried out ChIP and observed SMYD3 binding to the MMP-9 promoter in control cells but not in knockdown cells (Fig. 5F). Moreover, SMYD3 inhibition caused a significant decrease in the H3K4me3 marks on the MMP-9 promoter but not on the control glyceraldehyde-3-phosphate dehydrogenase (GAPDH) promoter (Fig. 5F). Also, SMYD3 inhibition did not affect Acetyl-H3 marks on the MMP-9 promoter (Fig. 5F). We excluded the possibility of an indirect effect of SMYD3 knockdown on MMP-9 transcription via the modulation of chromatin modifiers known to participate in the regulation of MMP-9 (22). We found no evidence for altered expression of these regulators upon SMYD3 knockdown (Supplementary Fig. S4). Thus, SMYD3 binds directly to the human MMP-9 promoter and appears responsible for H3K4 methylation and transcriptional activation.

To test whether SMYD3 regulation of MMP-9 contributes to tumor phenotypes, we examined the effect of SMYD3 knockdown by in vitro migration assays with collagen-coated filters. The downregulation of SMYD3 using 2 independent siRNAs caused a significant reduction in the migratory capacity of HT1080 cells, similar to knocking down MMP-9 alone (Fig. 6A and B). Conversely, we also tested the effect of overexpressing SMYD3 on cell migration; we transfected SMYD3 into human transformed HEK293T cells, which express low levels of endogenous SMYD3 (Fig. 6C) and observed an upregulation of MMP-9 mRNA (Fig. 6D) and a consequent increase in invasiveness (Fig. 6E).

SMYD3 has also been implicated in tumor cell growth (24, 25). We found that knocking down SMYD3 (using siRNA or shRNA) HT1080 cells provoked a decrease in proliferation (Supplementary Fig. S3C), associated with an increased G₁ phase and reduced S-phase (Supplementary Fig. S3D). Strikingly, SMYD3 knockdown also reduced colony formation in soft-agar growth assays, a standard in vitro test for tumorigenic potential (Fig. 7A–D). The few colonies that formed were small, with few dispersed cells (Fig. 7C). Moreover, this effect could be phenocopied by depleting MMP-9 (Fig. 7A and B), suggesting that these transformed phenotypes are linked to the regulation of MMP-9.

Finally, we tested SMYD3 knockdown in vivo using a zebrafish xenotransplantation model (30) as a simple and effective system for testing cell invasion and metastasis formation of HT1080 cells. The embryos that developed tumors 72 hours postinjection were scored for the dissemination of tumor cells (Fig. 7E, Table 1). We detected widely dispersed fluorescent spots in the yolk and in several cases in the tail region of zebrafish embryos in control xenografts, whereas the tumor mass detected in almost all of the embryos injected with SMYD3-siRNA or MMP9-siRNA–treated cells exhibited no dissemination (Fig. 7E, Table 1). Thus, SMYD3 expression levels affect the invasiveness and metastatic behavior of human tumor cells in vitro and in vivo.

The discovery of a large number of histone lysine methyltransferases and demethylases has revealed that histone methylation is a dynamic process directing the fine regulation of chromatin structure and gene expression (2, 31, 32). Yet, it remains a significant challenge to dissect the precise series of epigenetic events in tumor progression, which is critical to develop targeted therapeutic strategies. It is particularly difficult to reconcile the global changes in histone modifications and enzymes, with localized effects on specific gene promoters (33, 34, 35). H3K4 methylation remains particularly intriguing, with examples of deregulated H3K4 methyltransferases (e.g., SMYD3; ref. 24) and demethylases (e.g., PLU-1/JARID1B; refs. 36–38). SMYD3 is interesting in this respect, as it binds to a specific DNA sequence, allowing it to target particular promoters (24). Here, we present evidence that SMYD3 is specifically upregulated upon Theileria transformation and targets the promoter of the MMP-9 gene. This study offers insight into how Theileria parasites might mold the host cell epigenome and provides the first direct link between SMYD3 and a specific target gene in invasion and metastasis.

We focused on the MMP-9 gene as a paradigm to explore parasite-induced chromatin changes associated with host cell gene expression and invasion. MMP-9 is regulated extensively at the transcriptional level by classical transcription factors such as AP-1 and NF-κB (19). An elegant study by Ma and colleagues showed that these transcription factors lie downstream of signal transduction pathways and are recruited to the MMP-9 promoter in a stepwise process coordinated by chromatin-modifying complexes (21). Although AP-1 activation is important for MMP-9 regulation upon Theileria infection, it is not sufficient, as some Theileria-transformed cell lines express AP-1 without activating MMP-9 (data not shown; ref. 14). This drove us to search for epigenetic modulators that contribute to host gene regulation. We found that H3K4me3 was specifically associated with inducible and reversible MMP-9 expression (Fig. 2). Surprisingly, we saw no changes in Acetyl-H3, previously linked to MMP-9 regulation (21), nor in any repressive histone marks. SMYD3 was the only H3K4 methyltransferase whose expression correlated with the marks observed on the MMP-9 promoter. Our gain-of-function and loss-of-function studies showed that SMYD3 plays a direct role in MMP-9 regulation. This is likely via binding to putative SMYD3 recognition motifs in the MMP-9 promoter and a multiple repeat (5′-CCCTCCCTCCC-3′) that is conserved between the bovine and human MMP-9 promoters (Supplementary Fig. S2). SMYD3 binding to this region seems essential for MMP-9 gene regulation. The effects of SMYD3 knockdown appear to be independent of the expression of other chromatin modifiers (21, 39; Supplementary Fig. S4). But it is formally possible that SMYD3 and associated H3K4 methylation recruit additional regulatory proteins required for full MMP-9 induction (22). Another possibility is that promoter-bound SMYD3 contributes to gene expression by methylating nonhistone proteins locally, such as the AP-1 transcription factor. There are precedents for transcription factor regulation by methylation and this merits further investigation (40). Our study highlights the dynamic nature of H3K4 methylation in transcriptional regulation and how methyltransferase binding to select promoters can control gene expression.

We provide the first evidence that Theileria parasites influence host cell gene expression by modulating epigenetic regulators. The inducible and reversible nature of MMP-9 expression provided a model gene to focus on chromatin events induced by the parasite, leading to the discovery of SMYD3 as a promising candidate. It is likely that other chromatin-modifying enzymes contribute to Theileria-induced transformation, just as multiple modifiers cooperate in human tumors (2). The Theileria system offers an attractive model to search for novel SMYD3 target genes, by comparing infected and buparvaquone-treated transcriptomes accompanied by genome-wide ChIP analysis. One intriguing aspect is the striking upregulation of solely SMYD3 and not other H3K4 methyltransferases, such as mixed-lineage leukemia (MLL) members that are deregulated in human cancers and regulate MMP-9 (41). SMYD3 is overexpressed in human colorectal and hepatocellular carcinomas (24) and breast cancer cells (25). Elevated SMYD3 expression was recently observed in chronic lymphocytic leukemia (CLL) and in highly metastatic pancreatic cancers (42). However, little is known about SMYD3 transcriptional control (43) and our model provides a tool for studying the inducible and reversible regulation of SMYD3 in cancer cells. Our results show that the SMYD3-MMP9 circuit is functional in human cells, thereby demonstrating that the Theileria model can be effectively used to generate mechanistic hypotheses to be tested in human cancers.

Our study directly links the SMYD3 methyltransferase to the MMP-9 gene, which is important in angiogenesis and metastasis. MMP-9 expression is linked to lymphoma, leukemia, and metastasis (44, 45). There are few characterized targets for SMYD3 in tumorigenesis, except for the hTERT telomerase and WNT10B genes (25, 39). Previous studies highlighted the role of SMYD3 in cancer proliferation, but those linking SMYD3 to invasion have not identified a direct molecular target (24, 25, 46, 47). SMYD3 was also identified in gene signatures in metastatic pancreatic tumors (42). The urgent need to develop better therapies for metastatic disease makes it important to identify targets of regulators induced in metastatic cells. Clinical trials using MMP inhibitors have been hampered by problems with selectivity (48). Thus, our finding that a histone methyltransferase, which is induced in metastatic tumors, specifically regulates MMP-9 expression and tumor invasion in vivo, offers a target to develop therapeutic strategies aimed at preventing cancer progression.

No potential conflicts of interest were disclosed.

The authors thank Claire Rougeulle and Slimane Ait-Si-Ali for critical reading of the manuscript, members of UMR7216 for helpful discussions, Carmen Garrido for suggesting the zebrafish experiments, and André Bouchot for fluorescent microscopy assistance.

The Weitzman laboratory was supported by Fondation de France (FdF #2102), Association pour le Recherche contre le Cancer (ARC Fixe #4975, ARC-Equipement #7990), and Association for International Cancer Research (AICR, #08-0111). The Langsley laboratory acknowledges funding from Wellcome Trust grant (075820/A/04/Z) and Inserm/CNRS infrastructure support. A.M. Cock-Rada was supported by Colfuturo (Colombia), ARC, and Fondation pour la Recherche Médicale (FRM #FDT20100920170). N. Janski was supported by Université Paris-Diderot and AXA Research Fund.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.