MODULATION OF HUMAN STROMELYSIN 3 PROMOTER ACTIVITY AND GENEEXPRESSION BY HUMAN BREAST CANCER CELLSAthar AHMAD1*, John F. MARSHALL1, Paul BASSET2, Patrick ANGLARD2 and Ian R. HART1

1Division of Oncology, United Medical and Dental School, London, UK2IGBMC, Strasbourg, France

The matrix-degrading enzyme family of matrix-metallopro-teinases (MMPs) has been implicated in the process oftumour metastasis. Cellular protein and RNA localisationtechniques have been used to show that, whilst several MMPgenes are expressed in both cancer and stromal cells, strome-lysin 3 is expressed only in stromal fibroblasts adjacent tocancer cells. Immunohistochemical and in situ hybridisationevidence suggests that neoplastic cells can stimulate stromalcell MMP production either in a paracrine fashion or by acell–cell contact mechanism. Using 2 different lengths of thehuman stromelysin 3 (ST3) gene 58 flanking sequence clonedupstream of luciferase and CAT reporter genes, we now showthat human breast cancer cells can directly activate the ST3promoter. The putative response element in the ST3 pro-moter, which lies between 0.46 and 3.4 kb upstream of thetranscription start site, is able to effect a 2- to 3-fold increasein downstream gene expression. We further show that thistranscriptional up-regulation definitely occurs via a paracrine,and possibly via a cell–cell contact, mechanism. Confirmationthat this ST3 promoter activation results in ST3 gene induc-tion of a similar magnitude was shown using Northernblotting of stimulated fibroblasts. Our data provide furtherevidence that cancer cells can induce fibroblast MMP expres-sion and help to explain the in vivo expression pattern of ST3in breast cancer. Int. J. Cancer 73:290–296, 1997.r 1997 Wiley-Liss, Inc.

The human stromelysin 3 (ST3, MMP-11) gene, a member of thematrix metalloproteinase (MMP) family, was first identified incancer tissues by differential cDNA library screening of surgicalspecimens of human breast cancer and fibro-adenoma (Bassetetal., 1990). Indeed, levels of ST3 RNA in breast cancer have beenreported to predict a greater likelihood of recurrent disease (Engelet al.,1994), and ST3 protein levels in breast cancer sections havebeen shown to be associated with poor prognosis (Chenardet al.,1996). In all breast cancers studied so far, expression of ST3 RNAand protein is limited to the stromal fibroblasts within the tumourmasses that surround invasive breast cancer cells (Rouyeret al.,1994–1995 and refs. therein). This stromal fibroblastic pattern ofST3 expression implies that breast cancer cells may produce adiffusible factor(s) that can act upon neighbouring fibroblasts toinduce ST3 gene expression. Breast cancer cells can secrete avariety of growth factorsin vitro, and many of these can induce theexpression of a number ofMMP genes (Lippman and Dickson,1989; Gullick, 1990; Okadaet al.,1990; Itoet al.,1995; Himelsteinet al.,1994). Furthermore a factor termed EMMPRIN (extracellu-lar matrix metalloproteinase inducer), previously known as humantumour cell–derived collagenase-stimulating factor, has been puri-fied from both cancer cell membranes and the media conditionedby cancer cells (Biswaset al., 1995). This epithelial cell–derivedproduct can stimulate the expression of interstitial collagenase(MMP-1), stromelysin 1 (MMP-3) and 72 kDa gelatinase (MMP-2)in cultured fibroblasts (Guoet al., 1997) and provides furtherevidence that cancer cell–stromal cell interactions are important intumour progression. These interactions need not only be mediatedby soluble factors secreted by the neoplastic cells. Direct contactbetween cancer cells and surrounding fibroblasts is an importantmechanism mediating increased stromal fibroblast MMP expres-sion (Himelsteinet al.,1994; Segainet al.,1996).

To study regulation of theST3gene in these interactions, wehave utilised 2 different lengths of the 58 flanking sequence of thehumanST3gene to drive reporter genes. Using 2 different assay

systems and both transient and stable transfectants, we show that anumber of cytokines and the phorbol ester tumour promoter12-O-tetradecanoylphorbol-13-acetate (TPA), known to induceexpression of mostMMP genes, do not induce humanST3genepromoter activity. However, we show that certain breast cancercells up-regulate reporter gene activity under the transcriptionalcontrol of 3.4 kb of the human ST3 58 flanking sequencevia either adiffusible factor or a cell–cell contact mechanism. This promoterup-regulation was confirmed by Northern blotting, which showedinduction of ST3 mRNA expression.

MATERIAL AND METHODS

Cell culture

The breast cancer cell lines MDA-MB-231, MCF7 and ZR75were obtained from Central Cell Services at the Imperial CancerResearch Fund (ICRF, London, UK) and grown in DMEMcontaining 10% FCS and 4 mM glutamine; ZR75 cells were grownin this medium supplemented with 1028 M oestradiol (Sigma-Aldrich, Poole, UK).

The SV40-immortalised normal human mammary epithelial cellline MTSV1-7 (a gift from Dr. J. Taylor-Papadimitriou, ICRF) wasgrown in DMEM supplemented with 10% FCS, 4 mM glutamine,10 µg/ml insulin and 5 µg/ml hydrocortisone. The murine (NIH3T3)and human (HFF, human foreskin fibroblasts) fibroblast cell lines(obtained from ICRF Central Cell Services) were grown in DMEMsupplemented with 10% FCS and 4 mM glutamine. All cell lineswere grown as monolayer cultures on tissue culture plastic in 8%CO2 at 37°C. Conditioned media from the breast cell lines werecollected by washing a 90% confluent cell monolayer with PBStwice and then culturing the cells for a further 48 hr in serum-free,growth factor-free medium. The harvested culture medium wasthen centrifuged to remove cellular debris, filter-sterilised (0.45 µmfilters; Millipore, Watford, UK) and stored at220°C until use.

The various additives and cytokines used (phorbol ester TPA,tumour necrosis factor [TNFa], epidermal growth factor [EGF],basic fibroblast growth factor [bFGF] and platelet-derived growthfactor [PDGF]) were obtained from Sigma-Aldrich and used at theconcentrations indicated under ‘‘Results’’.

Reporter gene constructsHuman ST3 promoter-CAT (chloramphenicol-acetyl transfer-

ase) reporter gene constructs were based on the enhancerless,promoterless pBLCAT6 reporter vector and contained either 0.46kb (restriction sites XbaI–XhoI) or 3.4 kb (SphI–XhoI) of the 58

Contract grant sponsor: Institut National de la Sante´ de la RechercheMedicale; Contract grant sponsor: Centre Nationale de la Recherche Scien-tifique; Contract grant sponsor: Centre Hospitalier Universitaire Re´gional;Contract grant sponsor: Bristol-Myers Squibb Pharmaceutical Research In-stitute; Contract grant sponsor: Association pour la Recherche sur le Cancer;Contract grant sponsor: Ligue Nationale Franc¸aise contre le Cancer.

*Correspondence to: Richard Dimbleby Department of Cancer Research/ICRF Laboratory, The Rayne Institute, UMDS, St. Thomas’ Hospital,Lambeth Palace Road, London SE1 7EH, UK. Fax: 0171 922 8216.E-mail: ahmad@icrf.icnet.uk

Received 14 April 1997; Revised 11 June 1997

Int. J. Cancer:73,290–296 (1997)

r 1997 Wiley-Liss, Inc.

Publication of the International Union Against CancerPublication de l’Union Internationale Contre le Cancer

flanking sequence of the humanST3gene as detailed by Anglardetal. (1995). Both of these promoter sequences then were sub-clonedinto the identical restriction sites in the pSP73 shuttle vector(Promega, Southampton, UK) and thence into the SacI–XhoIrestriction sites within the promotorless, enhancerless luciferasereporter vector pGL3Basic (Promega). Negative and positivecontrol vectors comprised the promoterless, enhancerlessPGL3Basic/pBLCAT6 or SV40-driven CAT and luciferase reportervectors, respectively.

Transient transfections and CAT assaysTransient transfections were performed on NIH3T3 fibroblasts

grown in 10 cm diameter cell culture dishes. Cells were grown soas to reach 75% confluence at the time of transfection. After amedium change, 12 µg of purified plasmid DNA were precipitatedusing a calcium phosphate mammalian cell transfection kit (Pro-mega), and the precipitate was added to the cells for 16 hr. Cellswere washed twice with PBS, and then either serum-free condi-tioned media, control medium or the cytokine-containing mediawas added. After a further 24 hr, cells were harvested, and 13 106

cells from each plate (cell counting performed using a Casy 1 cellcounter [Scharfe, Reutlingen, Germany]) were disrupted using 3freeze-thaw cycles. CAT assays were performed on the cellextracts. CAT activity was determined either by separation of thereaction products using thin layer chromatography (TLC, on platespre-coated with silica gel [Sigma-Aldrich]) followed by visualisa-tion using autoradiography or by liquid scintillation counting(Wallac 1409.001 scintillation counter, Milton Keynes, UK) afteracetylated reaction product extraction using xylene. CAT activity isexpressed as the percentage of the total14C-labelled chlorampheni-col which was acetylated.

Preparation of clones of NIH3T3 fibroblasts stably expressingluciferase under the transcriptional control of SV40,0.46 kb ST3 or 3.4 kb ST3 promoters

NIH3T3 fibroblasts which had achieved 75% confluence wereco-transfected with each luciferase promoter construct (10 µg) andthe pBabe Puro vector (1 µg) (Morgenstern and Land, 1990).Sixteen hours later, cells were washed twice with PBS and thencultured for 2 days in normal growth medium. Cells were split intoselection medium containing 10 µg/ml puromycin (Sigma-Aldrich)at ratios ranging from 1:5 to 1:20. After 4 days, the medium was

replaced with fresh selection medium, and dishes were cultured fora further 14–20 days before the puromycin-resistant colonies werepicked into 24-well plates for expansion. A total of 130 colonies(combined figure for all 3 promoter constructs) were selected. Cells(1 3 106) from each colony subsequently were assayed for lucifer-ase activity using the Promega luciferase assay system according tothe manufacturer’s instructions.

Luminosity was measured using a BioOrbit 1251 Luminometer(Labtech, Uckfield, UK). From the 130 clones 91 luciferase-expressing colonies were isolated, of which the highest expressingclones (defined as luciferase units/106 cells) resulting from transfec-tion with each promoter construct (SV40, 0.46 kb or 3.4 kb humanST3 promoter lengths) were selected for subsequent use inco-culture experiments.

Co-culture experimentsCells (105) of each NIH3T3 sub-clone and breast epithelial cell

line (105) were grown together in individual wells of 6-well platesin 2 ml serum-containing medium for 16 hr to allow the cells toattach. Cells were washed with PBS and co-cultured in serum-freemedium for 24 hr, after which co-cultured cells were lysed and eachcell lysate assayed for luciferase activity. Cell lysate proteinconcentration was measured using the Bio-Rad protein assay kit(Hemel Hempstead, UK), and results are expressed as luciferaseunits/µg total protein.

RNA isolation and Northern blottingTotal RNA from HFF or breast carcinoma cell lines was isolated

using the RNAzol B (guanidium thiocyanate; Biogenesis, Poole,UK) kit by following the manufacturer’s instructions. RNA wasquantitated by UV absorbance. Fifteen micrograms of total RNAwere size-fractionated on a 1% agarose-formaldehyde gel andtransferred onto Hybond N filters (Amersham, Aylesbury, UK).Probes were [a-32P] dCTP-labelled using an oligolabelling kit(Pharmacia Biotech, St. Albans, UK) and hybridised to filtersovernight at 42°C according to standard procedures (Sambrooketal., 1989). Filters were washed once in 53 SSC for 15 min at 42°Cand then twice in 0.13 SSC/0.1% SDS for 15 min at 42°C andautoradiographed at270°C for 1–6 days. The humanST3cDNAprobe (nucleotides 346–2105) was provided by Dr. A. Docherty(Cell Tech, Slough, UK). Filters subsequently were stripped with

FIGURE 1 – NIH3T3 fibroblasts were grown to 75% confluence and then transfected transiently with the 0.46 and 3.4 kbST3–CATconstructs;16 hr post-transfection, cells were washed in PBS, and either serum-free medium or serum-free medium containing the relevant agent, at thefollowing concentrations, was added (TPA: 20 ng/ml; TNFa: 20 ng/ml; EGF: 20 ng/ml; bFGF: 10 ng/ml; PDGF: 20 ng/ml). After a further 24 hr,cells were harvested and CAT assays performed on 13 106 cells from each transfection. CAT activity was assayed as described in ‘‘Material andMethods’’.(a, b)CAT activity for each transfection; values shown are mean6 SD for triplicate plates.

291STROMELYSIN 3 REGULATION

FIGURE 2

292 AHMAD ET AL.

boiling 0.1% SDS and reprobed using theb-actin cDNA probe,which served as an internal control for RNA loading and transfer.

Quantification of ST3 mRNA expression relative tob-actinexpression was performed using digital image analysis (NIH ImageSoftware, Bethesda, MD) from scanned images of Northernautoradiographs (Apple Color One Scanner 600/27, London, UK).

RESULTS

The results from a representative experiment in which the effectof certain cytokines and the tumour promoter TPA on modulatinghumanST3promoter activity were assessed by performing tran-sient transfections into mouse NIH3T3 fibroblasts are shown inFig. 1a,b. The data show no evidence of any cytokine-inducedactivation of either the 0.46 kb or the 3.4 kbST3promoter drivingtheCAT reporter gene. Here, we also used, as a positive control, aTPA-inducible promoter (3 consecutive TPA-response elementscloned into the HindIII–BamHI sites of the pBLCAT2 vector;Luckow and Schutz, 1987) to show that in conditions where thispromoter responded to TPA the activity of theST3promoter wasunaffected (data not shown).

To assess whether breast cancer cells secrete a factor(s) that maybe important in regulation ofST3promoter activity, the effects ofconditioned medium (CM) from 3 breast cancer cell lines ontransiently transfected NIH3T3 fibroblasts were studied (Fig. 2a,b).Media from both MCF7 (Fig. 2a) and MDA-MB-231 (Fig. 2b)cells consistently resulted in a significant (p , 0.01) 2.1- to2.3-fold upregulation of 3.4 kb ST3-mediated CAT activity,

whereas ZR75-derived CM (or CM from BT20, BT474 or T47Dbreast cancer cells; data not shown) showed no such enhancementwhen the reporter gene was driven by the 3.4 kb ST3 promoter. Noupregulation of CAT activity was obtained when the reporter genewas under the transcriptional control of either the SV40 or 0.46 kbST3 promoter (Fig. 2a,b). Figure 2c shows a representativeautoradiograph that shows a 2-fold increase in 3.4 kbST3promoteractivity as a result of MCF7 CM as demonstrated by a 2-foldincreased intensity of the 1,3-diacetylated chloramphenicol deriva-tive (arrowed in Fig. 2c, lane 8). These changes inST3promoteractivity as a result of a factor(s) from breast cancer cells also wereobserved when stable transfectants of NIH3T3 fibroblasts weregrown in the presence of CM from MCF7 and MDA-MB-231 cells(data not shown).

To assess whether this enhancement in 3.4 kb ST3 promoteractivity resulted in an increase inST3gene expression in humanfibroblasts, Northern blots were performed using total RNAisolated from HFF grown in either benign breast epithelial cell orbreast cancer cell CM for 48 hr; data presented in Figure 3 show anincrease in fibroblastic steady-state ST3 mRNA levels as a conse-quence of either MCF7 (Fig. 3a, lane 3) or MDA-MB-231 (Fig. 3b,lane 3) CM. ZR75 CM did not result in any detectable changes inactivity when compared to control lanes (Fig. 3a,b, lane 4vs. lane1, respectively), while CM from a normal human mammaryepithelial cell line, MTSV1-7, also failed to have any discernibleeffect on ST3 RNA levels (Fig. 3a,b,lane 2). Digital image analysisrevealed that when the small differences in RNA loading weretaken into account, up-regulation of ST3 mRNA levels shown wereapprox. 5-fold increased in response to MCF7 CM (Fig. 3a) and2.5-fold in response to MDA-MB-231 CM (Fig. 3b).

To assess whether cancer cell–fibroblastic cell contact may playa role in the regulation of humanST3 promoter activity, stableNIH3T3 transfectants expressing the luciferase gene driven bySV40, 0.46 kb or 3.4 kbST3 promoters were co-cultured witheither normal (MTSV1-7) or breast cancer (MCF7, MDA- MB-231and ZR75) cells. Both MCF7 and MDA-MB-231 cells induced asignificant 2- to 3-fold increase (p , 0.01) in luciferase activityrelative to control values (Fig. 4) when the 3.4 kb promoter wasused to drive the luciferase gene. In the experiment shown,co-cultivation with ZR75 cells did induce a slight increase in 3.4 kbST3promoter activity, but this was not significant (p , 0.12) and,though this trend was observed with repeat experiments, themagnitude of the response never achieved statistical significance,unlike the results obtained with MCF7 and MDA-MB-231 cells.Again, no increase in reporter gene activity was observed when theSV40 or 0.46 kb ST3 promoters were utilised or when the NIH3T3stable transfectants were co-cultured with MTSV1-7 cells.

A B

FIGURE 3 – Human foreskin fibroblasts (HFF) were grown either in serum-free medium or medium conditioned (CM) by the displayed breastepithelial cell lines. After 48 hr, total RNA was prepared as described in ‘‘Material and Methods’’ and Northern blot analysis was performed.Filters were hybridized initially with a humanST3cDNA probe. After stripping, filters were reprobed using theb-actin cDNA probe to ensureequal RNA loading and transfer.(a) Lane 1, HFF grown in serum-free medium; lane 2, MTSV1-7 CM; lane 3, MCF7 CM; lane 4, ZR75 CM.(b)Lane 1, HFF grown in serum-free medium; lane 2, MTSV1-7 CM; lane 3, MDA-MB-231 CM; lane 4, ZR75 CM.

FIGURE 2 – Transient transfections were performed on NIH3T3fibroblasts. Cells were washed, 16 hr post-transfection and eitherserum-free medium or serum-free medium conditioned by the dis-played breast epithelial cells was added to the transfected fibroblastsfor 24 hr; 13 106 cells from each transfection were assayed for CATactivity. (a, b) CAT activity as mean6 SD for triplicate plates,representative of 3 separate experiments.(c) Transient transfections onNIH3T3 fibroblasts were performed, and 16 hr post-transfection cellswere washed and either serum-free medium or serum-free conditionedmedium from MCF7 cells was added to the transfectants. CAT assayswere performed and CAT activity measured by determination ofseparation of reaction products using thin layer chromatographyfollowed by autoradiographic exposure. Lanes 1–4 represent cellsgrown in serum-free medium; lanes 5–8 represent cells grown inserum-free conditioned medium from MCF7 cells. Transfections wereperformed with the following vectors: lane 1, promoterless/enhancer-less pBLCAT6; lane 2, SV40–CAT; lane 3, 0.46 kb ST3; lane 4, 3.4 kbST3; lane 5, pBLCAT6; lane 6, SV40–CAT; lane 7, 0.46 kb ST3; lane 8,3.4 kb ST3.

293STROMELYSIN 3 REGULATION

Northern blots, using a humanST3 cDNA probe, were thenperformed on total RNA extracted from breast or fibroblast celllines alone (as a control) or from the co-cultures. As observed in theprevious experiments using CM from breast cancer cells, inductionof ST3 mRNA was seen as a result of co-culture with MCF7(2.8-fold) and MDA-MB-231 (3.7-fold) cells (Fig. 5, lanes 7 and 8,respectively), as seen with the monitored reporter gene activity. Wealso noted that, while CM from ZR75 cells did not lead to any increasein fibroblastic ST3 mRNA levels, their co-culture with fibroblastsresulted in a 2.4-fold ST3 mRNA level induction (Fig. 5, lane 9).

DISCUSSION

Our results demonstrate that the humanST3promoter can beactivated by factor(s) secreted by certain breast cancer cell lines orby the cancer cells themselves and that the putative response

element to such factors is located between 0.46 and 3.4 kbupstream of the transcription start site. The observed levels ofST3promoter up-regulation are shown, by Northern blotting, to resultin induction of ST3 mRNA levels in human fibroblasts. These datasuggest that at least part of the up-regulation of fibroblastic ST3expression seen in breast carcinoma may be a consequence oftranscriptional regulation through theST3 promoter. Such amechanism of enhancement of ST3 expression in fibroblasts couldhelp to explain thein vivo localisation of ST3 mRNA and protein,as shown by in situ hybridisation and immunohistochemicaltechniques, respectively, in human breast cancer (Rouyeret al.,1994–1995). Thus, ST3 is restricted in its expression to the stromalfibroblasts within the tumour mass that are in juxtaposition with thecancer cells (Bassetet al.,1990).

The promoter regions of a number ofMMP genes (includinggelatinase B, stromelysin 1 and 2 and interstitial collagenase)

FIGURE 4 – Clones of NIH3T3 fibroblasts (105 cells) stably expressing luciferase under the transcriptional control of either SV40 promoter or0.46 or 3.4 kb of theST3promoter were co-cultured with 105 breast epithelial cells as described in ‘‘Material and Methods’’. Luciferase assayswere performed after 24 hr of co-culture in serum-free medium. Luciferase activity is shown as luciferase units/µg cell lysate proteinconcentration. Values given are mean6 SD for triplicate plates and are representative of 3 separate experiments.

FIGURE 5 – Breast epithelial or human foreskin fibroblast (HFF) cell lines were grown either alone in serum-free medium or in co-culture (106

human fibroblasts with 106 breast epithelial cells) for 48 hr. Total RNA was isolated and Northern blotting performed. Filters were probed initiallywith a human ST3 cDNA probe and subsequently stripped and reprobed with ab-actin cDNA probe to ensure equal loading and transfer of RNA.Lanes 1–5 represent RNA from the following cell lines in serum-free medium: lane 1, HFF; lane 2, MDA-MB-231; lane 3, MCF7; lane 4, ZR75;lane 5, MTSV1-7. Lanes 6–9 represent RNA from the following co-cultures: lane 6, HFF and MTSV1-7; lane 7, HFF and MCF7; lane 8, HFF andMDA-MB-231; lane 9, HFF and ZR75.

294 AHMAD ET AL.

contain response elements that are sensitive to a variety of stimuli,including cytokines and growth factors (Gaireet al.,1994; Benbowand Brinckerhoff, 1997). In particular, the presence of an activator-protein 1 (AP-1) site, also known as the TPA-response element, isimportant for the transcriptional up-regulation of many MMPpromoters in response to cytokines and the tumour promoter TPA(Angel et al.,1987; Windsoret al.,1993). In transient transfectionassays usingST3promoter–CAT constructs (utilising 0.46 kb and3.4 kb of the 58 flanking sequence of the humanST3gene), we haveshown that, unlike many other MMP promoters, theST3promoteris non-responsive to a variety of cytokines and the tumour promoterTPA (Fig. 1a,b). This observation is of some importance sincebreast carcinoma cells can produce a number of cytokines (Lip-pman and Dickson, 1989; Bassetet al.,1990; Gullick, 1990) whichcould act on neighbouring fibroblasts to up-regulateMMP geneexpression. Our results make it seem likely that the observedincreases inST3promoter activity were not a consequence of thecytokines used. These data are consistent with earlier studies,where, having found no ST3 promoter up-regulation in response toTPA, no consensus sequence for an AP-1 binding site was foundwithin 1.4 kb of the transcription start site of the humanST3gene(Anglardet al.,1995).

The results presented here suggest strongly that there is (are) abreast cancer cell–derived soluble factor(s) capable of transcription-ally up-regulating the humanST3 gene, as shown in Figure 2,where CM from MCF7 and MDA-MB-231 breast cancer cellsresulted in a significant increase in 3.4 kbST3 promoter–CATactivity. The putative response element(s) in theST3promoter tosuch a factor(s) appears to be located between 0.46 and 3.4 kbupstream of the transcription start site since no transcriptionalup-regulation was demonstrated for the 0.46 kbST3–CATcon-struct. Northern blotting was used to confirm increased ST3 mRNAexpression in human fibroblasts in response to breast cancer cellCM, and it appeared that this enhanced fibroblastic ST3 mRNAexpression was specific to certain breast cancer cell lines only.Others have reported that soluble factors from MCF7 cells canenhance production of other MMPs and that such soluble factorsare likely to be distinct from the several cytokines known to beproduced by cancer cells (Itoet al., 1995). We further show thatMDA-MB-231, a highly metastatic tumour cell line, also canstimulateST3 gene expressionvia a paracrine mechanism. CMfrom ZR75 cells (as well as from BT20, BT474 and T47D breastcancer cells; data not shown) did not result in any increased ST3promoter activity or mRNA expression; thus, this capacity toinduce ST3 expression is by no means retained by all establishedcell lines.

Cancer–stromal cell interactions responsible for enhanced fibro-blast MMP production are not limited to tumour cell–derivedsoluble factors. Cancer cell–stromal cell contact is an important

mechanism by which neoplastic cells can promote extracellularmatrix degradationvia increased stromal fibroblast MMP expres-sion (Itoet al.,1995; Segainet al.,1996). To address this question,we created clones of NIH3T3 fibroblasts, engineered to stablyexpress the reporter gene luciferase under the transcriptionalcontrol of 0.46 and 3.4 kb of theST3promoter. Using the SV40promoter to drive reporter gene activity as a control, we were able,using co-culture experiments, to confirm that MCF7 and MDA-MB-231 cells up-regulate luciferase activity from the 3.4 kb promoter.This up-regulation was of the same order as that observed with CMderived from these cell lines (i.e., 2- to 3-fold). Northern blotanalysis was used (Fig. 5) to demonstrate that co-culture of humanfibroblasts with MCF7 and MDA-MB-231 cell lines also resultedin enhanced ST3 mRNA expression. This up-regulation ofST3promoter activity and mRNA levels seen in the co-culture experi-ments may relate either to a breast cancer cell–derived solublefactor(s) being secreted during co-culture and acting on thefibroblasts or to a cancer cell–fibroblast interaction. However, datashown in Figures 4 and 5 using ZR75 cells suggest that cancer–stromal cell contact may play a role in the regulation of the humanST3gene. Thus, whilst no ST3 promoter up-regulation could beseen in response to ZR75 CM (Fig. 2a,b), there was evidence thatco-culture with ZR75 cells resulted in a small, though notstatistically significant, switch-on in 3.4 kb ST3 promoter–reporteractivity (Fig. 4). In addition, Northern blot analysis revealed that,whilst co-culture with ZR75 cells resulted in an approx. 2-fold ST3mRNA induction (Fig. 5, lane 9), no such induction was seen infibroblasts grown in ZR75 CM.

It is apparent that the magnitude of induction of fibroblastic ST3mRNA by MCF7 and MDA-MB-231 CM and co-culture (approxi-mate range 2- to 5-fold) appears weak in comparison to thefibroblastic ST3 levels commonly observed in human carcinomas,suggesting that theex vivo conditions used in our study onlypartially mimic thein vivosituation.

In summary, we have shown that certain breast cancer cells caninduce stromal cell ST3 expression by a soluble-factor and,possibly, a cell–cell contact mechanism. Such mechanisms mightpartially explain why, in breast cancer, ST3 expression is found inthe fibroblasts around cancer cells. Our data provide furtherevidence that cancer cell–stromal cell interactions are important inthe regulation of proteolytic enzymes strongly implicated incarcinoma progression.

ACKNOWLEDGEMENTS

This work was supported by funds from INSERM, CNRS,Centre Hospitalier Universitaire Re´gional, Bristol-Myers SquibbPharmaceutical Research Institute, Association pour la Recherchesur le Cancer, and Ligue Nationale Franc¸aise contre le Cancer.

REFERENCES

ANGEL, P., IMAGAWA , M., CHIU, R., IMBRA, R.J., RAHMSDORF, H.J., JONAT,C., HERRLICH, P. and KARIN, M., Phorbol ester-inducible genes contain acommoncis element recognised by a TPA-modulatedtrans-acting factor.Cell, 49,729–739 (1987).ANGLARD, P., MELOT, T., GUERIN, E., THOMAS, G. and BASSET, P., Structureand promoter characterisation of the human stromelysin-3 gene.J. biol.Chem.,270,20337–20344 (1995).BASSET, P., BELLOCQ, J.P., WOLF, C., STOLL, I., HUTIN, P., LIMACHER, J.M.,PODHAJCER, O.L., CHENARD, M.P., RIO, M.C. and CHAMBON, P., A novelmetalloproteinase gene specifically expressed in stromal cells of breastcarcinomas.Nature(Lond.), 348,699–704 (1990).BENBOW, U. and BRINCKERHOFF, C.E., The AP-1 site and MMP generegulations: what is all the fuss about?Matrix Biol. 15 (In press) (1997).BISWAS, C., ZHANG, Y., DECASTRO, R., GUO, H., NAKAMURA , T., KATAOKA ,H. and NABESHIMA, K., The human tumour-cell derived collagenasestimulatory factor (renamed Emmprin) is a member of the immunoglobulinsuperfamily.Cancer Res.,55,434–439 (1995).CHENARD, M.P., O’SIORAN, L.O., SHERING, S., ROUYER, N., LUTZ, N.Y.,WOLF, C., BASSET, P., BELLOCQ, J.P. and DUFFY, M.J., High levels ofstromelysin 3 correlate with poor prognosis in patients with breastcarcinoma.Int. J. Cancer,69,448–451 (1996).

ENGEL, G., HESELMEYER, K., AUER, G., BACKDAHL , M., ERIKSSON, E. andLINDER, S., Correlation between stromelysin 3 mRNA levels and outcomeof human breast cancer.Int. J. Cancer,58,830–835 (1994).

GAIRE, M., MAGBUNUA, Z., MCDONNELL, S., MCNEIL, L., LOVETT, D.H. andMATRISIAN, L.M., Structure and expression of the human gene for thematrix metalloproteinase matrilysin.J. biol. Chem.,269,2032–2040 (1994).

GULLICK , W.J., Growth factors and oncogenes in breast cancer.Progr.Growth Factor Res.,2, 1–13 (1990).

GUO, H.M., ZUCKER, S., GORDON, M.K., TOOLE, B.P. and BISWAS, C.,Stimulation of matrix metalloproteinase production by recombinant extra-cellular matrix metalloproteinase inducer from transfected Chinese hamsterovary cells.J. biol. Chem.,272,24–27 (1997).

HIMELSTEIN, B.P., CANETE-SOLER, R., BERNHARD, E.J. and MUSCHEL, R.J.,Induction of fibroblast 92kDa gelatinase/type IV collagenase expression bydirect contact with metastatic tumor cells.J. Cell Sci.,107,477–486 (1994).

ITO, A., NAKAJIMA , S., SASAGURI, Y., NAGASE, H. and MORI, Y., Coculture ofhuman breast adenocarcinoma MCF7 cells and human dermal fibroblastsenhances production of matrix metalloproteinases 1, 2 and 3 in fibroblasts.Brit. J. Cancer,71,1039–1045 (1995).

LIPPMAN, M.E. and DICKSON, R.B., Mechanisms of growth control in

295STROMELYSIN 3 REGULATION

normal and malignant breast epithelium.Rec. Progr. Hormone Res.,45,383–440 (1989).

LUCKOW, B. and SCHUTZ, G., CAT constructions with multiple uniquerestriction sites for the functional analysis of eukaryotic promoters andregulatory elements.Nucleic Acids Res.,15,5490 (1987).

MORGENSTERN, J.P. and LAND, H., Advanced mammalian gene transfer:high titre retroviral vectors with multiple drug selection markers andcomplementary helper-free packaging cell line.Nucleic Acids Res.,18,3587–3596 (1990).

OKADA , Y., TSUCHIYA, H., SHIMIZU , H., TOMITA, K., NAKANISHI , I., SATO, H.,SEIKI, M., YAMASHITA , K. and HAYAKAWA , T., Induction and stimulation of92 kDa gelatinase/type IV collagenase production in osteosarcoma andfibrosarcoma cell lines by tumor necrosis factora. Biochem. biophys. Res.Comm.,171,610–617 (1990).

ROUYER, N., WOLF, C., CHENARD, M.P., RIO, M.C., CHAMBON, P., BELLOCQ,J.P. and BASSET, P., Stromelysin-3 gene expression in human cancer: anoverview.Invasion Metastasis,14,269–275 (1994–1995).SAMBROOK, J., FRITSCH, E.F. and MANIATIS, T., Molecular cloning: alaboratory manual,Cold Spring Harbor Laboratory Press, Cold SpringHarbor, NY (1989).SEGAIN, J.P., HARB, J., GREGOIRE, M., MEFLAH, K. and MENANTEAU, J.,Induction of fibroblast gelatinase B expression by direct contact with celllines derived from primary tumour but not from metastases.Cancer Res.,56,5506–5512 (1996).WINDSOR, L.J., GRENETT, H., BIRKEDAL-HANSEN, B., BODDEN, M.K.,ENGLER, J.A. and BIRKEDAL-HANSEN, H., Cell type-specific regulation ofSL-1 and SL-2 genes. Induction of the SL-2 but not SL-1 gene by humankeratinocytes in response to cytokines and phorbol esters.J. biol. Chem.,268,17341–17347 (1993).

296 AHMAD ET AL.