metastasisisstronglyreducedbythematrix ... · metastasisisstronglyreducedbythematrix...

Metastasis is strongly reduced by the matrixmetalloproteinase inhibitor Galardin in the MMTV-PymTtransgenic breast cancer model

Kasper Almholt,1 Anna Juncker-Jensen,1

Ole Didrik Lærum,3 Keld Danø,1 Morten Johnsen,2

Leif Røge Lund,1 and John Rømer1

1Finsen Laboratory, Rigshospitalet, and 2Department of MolecularBiology, University of Copenhagen, Copenhagen Biocenter,Copenhagen, Denmark and 3The Gade Institute, Section ofPathology, Haukeland University Hospital, Bergen, Norway

AbstractMatrix metalloproteinases (MMP) have several roles thatinfluence cancer progression and dissemination. However,low molecular weight metalloproteinase inhibitors (MPI)have not yet been tested in transgenic/spontaneousmetastasis models. We have tested Galardin/GM6001,a potent MPI that reacts with most MMPs, in the MMTV-PymT transgenic breast cancer model. We followed acohort of 81 MMTV-PymT transgenic mice that receivedGalardin, placebo, or no treatment. Galardin treatment wasstarted at age 6 weeks with 100 mg/kg/d, and all micewere killed at age 13.5 weeks. Galardin treatmentsignificantly reduced primary tumor growth. Final tumorburden in Galardin-treated mice was 1.69 cm3 comparedwith 3.29 cm3 in placebo-treated mice (t test, P =0.0014). We quantified the total lung metastasis volume inthe same cohort of mice. The median metastasis volumewas 0.003 mm3 in Galardin-treated mice compared with0.56 mm3 in placebo-treated mice (t test, P < 0.0001).Thus, metastasis burden was reduced more than 100-fold,whereas primary tumor size was reduced only 2-fold. Wealso found that primary tumors from Galardin-treated miceexhibited a lower histopathologic tumor grade, increasedcollagen deposition, and increased MMP-2 activity. MMPsare known to have tumor-promoting and tumor-inhibitory

effects, and several clinical trials of broad-spectrum MPIshave failed to show promising effects. The very potentantimetastatic effect of Galardin in the MMTV-PymTmodel does, however, show that it may be possible tofind broad-spectrumMPIs with favorable inhibition profiles,or perhaps combinations of monospecific MPIs, for futureclinical application. [Mol Cancer Ther 2008;7(9):2758–67]

IntroductionMatrix metalloproteinases (MMP) are potentially involved atseveral stages of cancer progression including tumorinitiation, vascularization, invasion, and metastasis (1, 2).The roles of MMPs in cancer reflect their physiologic ability,through limited proteolysis, to restructure the tissue micro-environment and to convey extracellular signaling. In thehealthy organism, MMPs are involved in several tissueremodeling processes during ontogenesis and adult life,including embryo implantation (3), bone development (4),mammary gland development (5, 6), and wound healing (7).

The MMPs constitute a large family of 22 extracellularmetalloproteinases in mice (24 in humans) and are eithersecreted (15 MMPs) or membrane-anchored (7 MMPs)(1, 8, 9). The MMPs are synthesized as inactive proformsthat require proteolytic activation. Ten of the 22 pro-MMPs,including the membrane-anchored MMPs, have a furincleavage site and may be activated by furin-like proproteinconvertases before they are exported from the cell. Theremaining pro-MMPs are activated extracellularly typicallyby the serine protease plasmin or by other MMPs. TheMMP system is counterbalanced by a group of four tissueinhibitors of metalloproteinases (TIMP) that have varyingspecificities for individual MMPs (10). In addition, themembrane-anchored MMP inhibitor RECK regulates asubgroup of MMPs including MMP-2, -9, and -14 (11).The MMPs are collectively able to degrade any componentof the extracellular matrix. Important substrates for theMMPs as a group include the native fibrillar collagens, allmajor basement membrane components, and severalextracellular signaling molecules.

The structural integrity of the stroma is an essentialfactor in preventing carcinogenesis (12). In the mammarygland, disruption of the extracellular matrix and cellularmicroenvironment caused by transgenic overexpressionof MMP-3 in its active form or of MMP-14 leads to tumordevelopment even in the absence of a transgene-suppliedoncogene (13, 14). In the presence of a transgene-suppliedoncogene, tumor development, growth, and metastasisare in most cases promoted by increased MMP activityand inhibited by decreased MMP activity (reviewed inref. 15). A couple of examples, here limited to studiesof transgenic breast cancer models, illustrate the general

Received 3/14/08; revised 6/13/08; accepted 6/15/08.

Grant support: Copenhagen Hospital Corporation, Danish Cancer ResearchFoundation, European Commission contract LSHC-CT-2003-503297, and‘‘Aase og Ejnar Danielsens’’ Foundation.

The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely toindicate this fact.

Note: Current address for K. Almholt and J. Rømer: Histology andInflammation, Novo Nordisk A/S, Novo Nordisk Park, Building F6.2.30,DK-2760 Maløv, Denmark.

Requests for reprints: Leif Røge Lund, Finsen Laboratory, Rigshospitalet3735, Copenhagen Biocenter, Ole Maaløes Vej 5, DK-2200 Copenhagen,Denmark. Phone: 45-3545-6047; Fax: 45-3545-3797.E-mail: [email protected]

Copyright C 2008 American Association for Cancer Research.

doi:10.1158/1535-7163.MCT-08-0251

2758

Mol Cancer Ther 2008;7(9). September 2008

Research. on December 30, 2020. © 2008 American Association for Cancermct.aacrjournals.org Downloaded from

http://mct.aacrjournals.org/

finding: (a) tumor onset is promoted by MMP-7 over-expression (16) and delayed by TIMP-2 overexpression(17), (b) tumor growth is reduced by overexpression of eitherTIMP-1 (18) or TIMP-2 (17), and (c) tumor metastasis isreduced by overexpressing TIMP-1 through a liver-specifictransgene (18).

Several small molecular weight metalloproteinase inhib-itors (MPI) have been in clinical trials, but none of thesecompounds have, to our knowledge, been tested in atransgenic metastasis model (15). We have now used theMMTV-PymT transgenic breast cancer model (19) toanalyze the effect of Galardin/GM6001/Ilomastat (Fig. 1;ref. 20). Galardin is a potent and broad-spectrum hydrox-amate-type MPI that is designed as a molecular mimic ofMMP substrates, which allows it to enter the active site ofMMPs, where it strongly, but reversibly, binds the criticalzinc atom (21). The MMTV-PymT transgene combines themammary-specific LTR promoter from the MMTV viruswith the pluripotent middle T oncogene from the mousepolyomavirus to induce breast tumors with 100% pene-trance in virgin females (19). Virtually all MMTV-PymTmice also develop lung metastases (19, 22, 23). Thehistopathology (19, 24, 25) and expression of breastcancer biomarkers in the MMTV-PymT tumors, includinghormone receptors, cytokeratins, ErbB2, and h1-integrin,are consistent with that seen in aggressive (human)breast cancers (24, 25). In contrast to several transplantedtumors, the expression patterns of MMPs (15, 26, 27) in theMMTV-PymT tumors are also similar to ductal mammaryadenocarcinomas. In the present study, we show adramatic reduction in the metastatic burden of MMTV-PymT mice after treatment with Galardin.

Materials andMethodsGalardinThe MMP inhibitor 3-(N -hydroxycarbamoyl)-2(R )-

isobutylpropionyl-L-tryptophan methylamide (Galardin/GM6001/Ilomastat; molecular weight: 388.47) was synthe-sized essentially as described (21). The following IC50

values have been reported for Galardin against humanMMPs: MMP-1 (6 nmol/L), MMP-2 (7/17 nmol/L), MMP-3(28 nmol/L), MMP-7 (41 nmol/L), MMP-8 (1.4 nmol/L),MMP-9 (4.1/15 nmol/L), MMP-12 (23 nmol/L), MMP-13(3.2 mmol/L), MMP-14 (23/33 nmol/L), MMP-15(6 nmol/L), MMP-16 (8 nmol/L), and MMP-26 (17 nmol/L;

refs. 28, 29). Galardin was administered in one of two ways:bydaily injections (i.p.) at 100mg/kg as a 20mg/mLslurry in0.9% saline, 4% clinically formulated carboxymethylcelluloseor by surgical insertion of custom-made 60-day slow releasetablets containing 150 mg Galardin each (Innovative Re-search of America). The tablets were inserted s.c. adjacent tothe dorsal midline (MMTV-PymT model) or in the dorsalneck region (wound-healing analysis), and the insertion sitewas closed with wound clips (Roboz Surgical Instrument).The amount of Galardin per tablet corresponds roughly to100 mg/kg/d for a 25 g mouse [150 mg / (0.025 kg � 60days)]. Tablets were inserted under general anesthesiainduced by s.c. injection of 0.03 mL/10 g of a 1:1 mixture ofDormicum (Midazolam 5 mg/mL) and Hypnorm (Fluanison5 mg/mL and Fentanyl 0.1 mg/mL).

In vivo AnalysesAll animal experiments were conducted according to

institutional guidelines and were approved by the DanishAnimal Experiments Inspectorate. A concurrent healthreport compliant with the guidelines of the Federation ofEuropean Laboratory Animal Science Associationsrevealed no infections.MMTV-PymT Tumor Model. Hemizygous FVB/N-

TgN(MMTVPyVT)634 Mul (hereafter FVB-PymT ; ref. 19)female virgins and nontransgenic controls were usedthroughout the study, all of which were generated fromthe same cohort of breeding pairs. All animals wereweaned and tail-clipped at age 4 weeks, genotyped byPCR using a primer pair specific for the MMTV-PymTtransgene (23), and distributed randomly into the treatmentgroups. Tumor onset and growth were monitored asdescribed previously (22). Slow-release tablets containingGalardin or placebo were inserted at age 43 F 1 days(range, 41-45 days). The mice receiving tablets were cagedindividually until sacrifice to allow fast and uniformclosure of the tablet insertion site. To terminate theexperiment, the mice were anesthetized as above, tumortissue was removed and immediately placed on dry ice forbiochemical analyses, and the mice were perfusion fixedas described (22). The lungs were removed, air-evacuated,fixed, cryoprotected with sucrose, and cryoembedded inOCT compound (Sakura Finetek) for stereologic metastasisquantification (30). Regional lymph nodes and tumor tissuefor histopathology were removed and placed in freshlyprepared PBS with 4% paraformaldehyde overnight at 4jCbefore automated paraffin embedding. Tumor tissue wascut into f5-mm-thick slabs for optimal access of fixative.Wound-Healing Model. Six- to 8-week-old FVB mice

were anesthetized as described above, and 20-mm-longfull-thickness incisional skin wounds were made mid-dorsally with a scalpel in each of 75 FVB mice randomlydistributed with 15 in each of five groups: Galardin orplacebo injections, Galardin or placebo tablets, and notreatment. The wounds were neither dressed nor sutured.The mice were caged individually and scored as describedpreviously (7, 31). The administration of Galardin orplacebo, as i.p. injections or slow-release tablets, wasstarted 2 days before wounding.Figure 1. Galardin/GM6001/Ilomastat.

Molecular Cancer Therapeutics 2759




HistopathologyFibers of collagens I, II, and III were stained in 5-Am

paraffin sections using PicroSirius Red (32), counterstainedwith Weigert’s hematoxylin, and visualized using polariz-ing microscopy. Histopathologic analysis of tumor stagewas done by an experienced pathologist (O.D.L.) on codedslides containing two Mayer’s H&E-stained sections madethrough the center of each (right side) abdominal gland/tumor and using the nomenclature introduced by Lin et al.(25, 33). In short, the development of mammary tumors isdivided into five different classes or progression stages:normal gland, hyperplasia, adenoma, early carcinoma, andlate carcinoma. The highest grade present in the twosections determined the score.

Stereologic Quantification of Lung MetastasisVolume

For volumetric measurement of total lung metastasis, thecryoembedded lungs were cut transversally to the tracheainto uniform 2-mm-thick slabs. The five to eight slabsobtained from each lung were re-embedded in OCTcompound into a single block as specified previously (30).From each block, 8-Am cryostat sections were obtained andstained with H&E. Stereologic determination of themetastasis volumes was done by computer-assisted pointcounting on coded slides (30).

Quantification of Tumor Cell Dissemination to LymphNodes

Three sections (3 Am), each separated by 15 Am, ofbrachial and axillary lymph nodes were stained with ananti-cytokeratin-8 antibody as described previously (23).Coded slides were analyzed under the microscope for thepresence of cancer cells. A lymph node specimen wasconsidered positive if one or more cytokeratin-8-positivecells were clearly identified beneath the node capsule inany one of the three sections.

Gelatin ZymographyFrozen tumor tissue from the left side abdominal gland/

tumor was crushed using a mortar and pestle, lysed in 5 ALlysis buffer [0.5 mol/L Tris-HCl, 0.2 mol/L NaCl, 10 mmol/LCaCl, 1% Triton-X-100 (pH 7.6) per milligram weight oftissue, and homogenized for 2� 8 min in an ultrasound bath.The resulting supernatants were collected after centrifuga-tion at 12,000 � g for 30 min at 4jC. Equal amounts of totalprotein were loaded onto a 10% Bis-Tris Novex ZymogramGel (Invitrogen) containing 0.1% gelatin as a substrate. Thegel was run under nonreducing conditions, renatured,developed, and stained with Coomassie blue according tothe manufacturer’s protocol. Mouse pro-MMP-2 and -9(Calbiochem) were used as markers for gelatinases.

In situ HybridizationTissue for in situ hybridization studies was obtained from

perfusion-fixed FVB-PymT and nontransgenic FVB females.MMP mRNAs were detected in tissue samples using[35S]UTP-labeled RNA probes at 105 cpm/AL on 3-Am-thick paraffin sections essentially as described (34). Toprovide positive and negative controls for hybridizationspecificity, the RNA probes were transcribed from separateplasmids carrying two nonoverlapping cDNA fragments of

mouse MMP-2, -3, -7, -9, -10, and -13, and the plasmidswere transcribed in both antisense and sense orientation.The six plasmids for MMP-2, -3, and -13 have beendescribed (26). The remaining plasmids were pG7-MMAT-5¶ RACE 2-1 (35) containing the 1 to 546 fragment(GenBank L36244) in pGEM7Zf+ and pKSMMATAH (35)containing the 406-ApaLI(1062) fragment in pBluescript II

Figure 2. Primary tumor growth and metastasis in Galardin-treatedmice. Tumor size was monitored weekly by palpation in FVB-PymT micethat were randomly distributed to receive the MMP inhibitor Galardin(n = 26), placebo (n = 26), or no treatment at all (n = 27). Treatmentwas started at age 41 to 45 d by s.c. insertion of 60-d slow-releasetablets containing 150 mg Galardin or placebo. The tablets were left inplace until the mice were killed at age 92 to 96 d and analyzed for lungmetastasis. The lengths (l ) and widths (w ) of every tumor in each mousewas used to estimate the total tumor volume: V = f(lw2p / 6) asdescribed (22). A, geometric mean of tumor volumes plotted versus agefor untreated (o), placebo-treated (4), and Galardin-treated (E) mice.B, tumor volumes at sacrifice. Columns, geometric means; bars, 95%confidence interval. C, total metastasis volume was quantified with anunbiased stereologic technique (30) on five to eight evenly spaced H&E-stained sections of the lungs from each mouse. Columns, medians; bars,95% confidence interval.

MMP Inhibition Reduces Spontaneous Metastasis2760




KS+ for MMP-7; pmSP6592b (36) containing theBamHI(1999)-2258 fragment (GenBank Z27231) in pSP65,pmSP6492b (36) containing the AvaI(1934)-2258 fragment inpSP64, and pKaA207 containing the 261 to 966 fragment inpBluescript KS+ for MMP-9; and pKaA225 containing the633-BamHI(1162) fragment (GenBank Y13185) in pBlue-script II KS- and pKaA222 containing the 1,162 to 1,693fragment in pCRII for MMP-10. The MMP-3 and -10plasmids were constructed to avoid the 325-bp-longsequences that are 99% identical in these two cDNAs.

Statistical AnalysesHypotheses on differences between treatment groups

were tested by two-sided Student’s t tests for tumorvolume and with Mann-Whitney U test for metastasisvolume due to a nonnormal distribution in the Galardintreatment group. Tumor volumes had a log-normaldistribution and were logarithmically transformed beforeanalysis. To allow logarithmic transformation, tumorvolumes of zero were assigned a value of 10�4 cm3. Lungand lymph node metastasis incidences were analyzed byFisher’s exact tests. Tumor grade was scored using anordinal scale (grades A-E) for the right side no. 4 mammarygland from each mouse and analyzed by Kruskal-Wallistest. In all analyses, P < 0.05 was considered significant. Allvalues are given as average F SD if not otherwise stated.The SAS software package (version 8.0; SAS Institute) wasused to perform the statistical analyses.

ResultsReducedTumor Growth in Galardin-TreatedMiceTo obtain a reproducible administration of the MPI

Galardin with minimal trauma to the mice during the

50-day drug exposure, we used slow-release tabletsestimated to release 100 mg/kg/d. We have foundpreviously that this dose of Galardin, given as once dailyi.p. injections, results in a maximal delay of skin woundhealing in mice (7). To validate the in vivo efficacy of theslow-release tablets, we did the same wound-healing assayon 75 mice that were randomly distributed with 15 in eachof five groups: Galardin or placebo injections, Galardin orplacebo tablets, and an untreated group. We found thatthe slow-release tablets were at least as effective as thei.p. injection route, causing a somewhat greater delay inwound healing (time to healing delayed from 17.2 F 4 to26.5 F 6 days) compared with the injection route (timeto healing delayed from 18.0 F 2 to 22.9 F 5 days). Time tohealing in untreated mice was 16.4 F 2 days. Thedifference between the two Galardin administration routeswas not significant (26.5 F 6 versus 22.9 F 5 days; t test,P = 0.099).

We then generated a cohort of 81 FVB-PymT micerandomly distributed with 27 in each of three groupsreceiving Galardin or placebo slow-release tablets or notreatment at all. Treatment was started at age 43 F 1 days(range, 41-45 days), which was chosen because it is theapproximate age of tumor onset in this model and becausethe mammary epithelium is almost fully developed by age6 weeks. Importantly, Galardin administration at an earlierage would delay mammary development (6). There weretwo unscheduled deaths in the cohort leaving 79 FVB-PymT mice to be analyzed: one mouse in the Galardintablet group was found dead and one mouse in the placebotablet group was killed prematurely due to severe edemasurrounding the tablet. Seven nontransgenic control micewere included among the untreated mice.

Table 1. Galardin treatment of MMTV-PymT transgenic mice

Treatment group

Untreated Placebo Galardin

No. mice 27 26 26Age at sacrifice (d) 94.1 F 1 93.8 F 1 94.4 F 1Tumor incidence at sacrifice (%) 100 100 100Final primary tumor volume (geometric mean; cm3)* 3.37 3.29 1.69Final primary tumor volume (95% confidence interval of the geometric mean; cm3) 2.50-4.56 2.45-4.41 1.28-2.23Dissemination to lymph nodes, brachial + axillary (incidence; %)c 30 (31/105) 29 (30/102) 21 (22/103)Dissemination to brachial lymph nodes (incidence; %)c 25 (13/52) 29 (15/51) 22 (11/51)Dissemination to axillary lymph nodes (incidence; %)c 34 (18/53) 29 (15/51) 21 (11/52)Lung metastasis (incidence; %)b 100 (27/27) 92 (24/26) 58 (15/26)Lung metastasis volume (median; mm3)x 1.39 0.56 0.0030Lung metastasis volume (95% confidence interval of the median; mm3) 0.23-3.05 0.19-2.34 0-0.048Lung metastasis volume (geometric mean; mm3)k 0.92 0.26 0.0033

NOTE: A cohort of FVB-PymT mice were either left untreated or implanted with placebo or Galardin pellets and monitored weekly for tumor onset and killedat age 92 to 96 d. Primary tumor volume and tumor dissemination to regional lymph nodes and to lungs were assessed at the time of sacrifice.*t test on log-transformed data (placebo versus Galardin): P = 0.0014.cNot significant.bFisher’s exact test (placebo versus Galardin): P = 0.009.xMann-Whitney U test (placebo versus Galardin): P < 0.0001.kThe geometric mean is also reported for direct comparison with earlier work (22, 23). To allow logarithmic transformation, metastasis volumes of zero wereassigned a value of 10-4 mm3. t test on log-transformed data (placebo versus Galardin): P < 0.0001.





All the MMTV-PymT transgenic mice developed tumorsirrespective of treatment group. Macroscopic tumor onsetoccurs from age 5 to 8 weeks in FVB-PymT mice, aroundthe same time as treatment was started in this experiment.We are therefore unable to determine whether Galardin hasany effect on tumor onset. Galardin reduced the growthrate of the MMTV-PymT tumors, although the tumors stillfollow an exponential growth pattern in the presence ofGalardin (Fig. 2A). All mice were killed at age 13.5 weeks(range, 92-96 days). The geometric mean of the tumorburden measured before sacrifice was 3.37 cm3 in untreatedmice, 3.29 cm3 in placebo-treated mice, and 1.69 cm3 inGalardin-treated mice (Table 1). Galardin treatment thusreduced final tumor size significantly by 49% comparedwith placebo-treated mice (t test, P = 0.0014). Tumor size inthe placebo-treated group was not significantly differentfrom the untreated group (Fig. 2B; t test, P = 0.90).

Reduced LungMetastasis in Galardin-TreatedMiceWe quantified the total volume of lung metastases in each

mouse using an unbiased stereologic technique (30). Theincidence of lung metastases was greatly reduced byGalardin treatment compared with either of the two controlgroups. The incidence of lung metastases was 100% inuntreated mice (27 of 27) and 92% in placebo-treated mice(24 of 26) but only 58% in Galardin-treated mice (15 of 26).Lungs from the nontransgenic control mice (n = 7) wereincluded randomly among the analyzed samples; none ofthese were scored positive for metastases. Galardintreatment thus significantly reduced lung metastasisincidence compared with placebo-treated mice (Fisher’sexact test, P = 0.009). The metastatic burden, expressed astotal volume of tumor tissue in the lungs, was alsosignificantly reduced by Galardin administration. Themedian metastasis volume was 1.39 mm3 in untreatedmice and 0.56 mm3 in placebo-treated mice but only0.003 mm3 in Galardin-treated mice (Fig. 2C; Table 1).The effect of Galardin was highly significant comparedwith the placebo group (Mann-Whitney U test, P < 0.0001),whereas the placebo-treated and the untreated mice did notdiffer significantly (Mann-Whitney U test, P = 0.27). Wealso analyzed the metastasis volume-to-primary tumorvolume ratios. This fraction was significantly reduced in

Galardin-treated mice. The median of the ratios was 39 �10�5 in untreated mice and 17 � 10�5 in placebo-treatedmice but only 0.16 � 10�5 in Galardin-treated mice (over a100-fold difference between placebo- and Galardin-treatedmice; Mann-Whitney U test, P < 0.0001).

Dissemination to Lymph Nodes in Galardin-TreatedMice

The MMTV-PymT model exhibits regional lymph nodedissemination (23). We recovered the bilateral brachial andaxillary lymph nodes from each mouse (total of 4 nodes) inthe experimental cohort described above. The small size ofthe nodes caused a few of them not to be scored fortechnical reasons. In total, 154 of 158 brachial and 156 of 158axillary nodes were scored. The incidence of tumor cellswas very similar in brachial versus axillary lymph nodesfor all treatment groups (Table 1). Treatment with Galardindid not significantly affect the incidence of tumor cells inlymph nodes, although there was a trend toward a lowerincidence of positive lymph nodes in the Galardin group(Table 1). Galardin-treated mice thus contained tumor cellsin 21% (22 of 103) of all their lymph nodes (brachial +axillary) compared to 29% (30 of 102) of the lymph nodesfrom placebo-treated mice (Fisher’s exact test, P = 0.20).The placebo group was very similar to the group ofuntreated mice that showed a 30% (31 of 105) incidence oftumor cells in lymph nodes.

Increased Collagen Deposition in Galardin-TreatedTumors

Gelatin zymography of tumor extracts from placebo- andGalardin-treated tumors revealed that levels of pro-MMP-2and active MMP-2 were increased in Galardin-treatedtumors. Pro-MMP-9 was seen sporadically in tumors fromboth groups, whereas active MMP-9 could not be detected(Fig. 3). The increase in MMP-2 enzyme levels is probably aresult of substrate accumulation caused by the continuousinhibition of metalloproteinases including the collageno-lytic and gelatinolytic MMPs. To directly evaluate theeffect of Galardin treatment on collagen turnover, westained tumors from the no. 4 mammary glands ofplacebo-treated (n = 9) and Galardin-treated (n = 8) FVB-PymT mice with PicroSirius Red for the fibrillar collagenstype I, II, and III. In normal mammary glands from

Figure 3. Increased MMP-2 activity in Galardin-treated tumors. Gelatin zymography of tumor extractsfrom placebo- and Galardin-treated tumors. The levelsof pro-MMP-2 and active MMP-2 are increased inGalardin-treated tumors. Pro-MMP-9 is seen sporadical-ly in tumors from both groups.





age-matched nontransgenic FVB mice (n = 2), collagen wasprimarily confined to the wall of milk ducts and to the thinfibrous septae that subdivide the organ (Fig. 4A and B). Bycomparison, MMTV-PymT tumors contain a very dense butalso highly disorganized collagen matrix. In placebo-treatedtumors, large cancer cell nodules are surrounded bycollagen-rich stroma but are almost devoid of collagenwithin the nodules (Fig. 4C, C ¶, C 00). Similar collagen-freetumor areas are infrequent in tumors of comparable sizefrom Galardin-treated mice (Fig. 4D, D ¶, D 00). At high

magnification, the collagen fibers also appear thicker intumors from Galardin-treated mice (data not shown).

Histopathologic Progression in Galardin-TreatedMMTV-PymTPrimaryTumors

To determine whether Galardin treatment affected thehistopathologic progression of the primary tumors, we dida histologic evaluation of the abdominal (no. 4) mammarygland from all mice in the cohort described above. Theglands were scored on an ordinal scale as either normalgland (A), hyperplasia (B), adenoma (C), early carcinoma

Figure 4. Collagen deposition in healthymammary glands and Galardin-treatedmammary tumors. The no. 4 (right side)mammary glands were obtained from 92-to 96-d-old FVB-PymT mice that had beenrandomly distributed to receive the MMPinhibitor Galardin or placebo starting at age41 to 45 d. The no. 4 mammary gland wasalso obtained from tumor-free nontransgenicmice matched for strain and age. Sections ofthe tissue samples were stained with Pic-roSirius Red to label collagen type I (yellowto red ) and III (green ) under polarizingmicroscopy. Sections were counterstainedwith Weigert’s hematoxylin (black nuclei).A and B, no. 4 mammary gland of a tumor-free nontransgenic mouse under bright-field(A) and polarized (B) illumination for com-parison. C, C¶, C00, representative tumorsfrom placebo-treated mice. D, D¶, D00,representative tumors from Galardin-treatedmice. The collagen-rich skin is visible inseveral panels. Bar, 500 Am.





(D), or late carcinoma (E). Primary tumors from eitheruntreated or placebo-treated mice were primarily of the‘‘early carcinoma’’ grade, and the very similar distributionof tumor grades in these two groups served as an internalcontrol for the reproducibility of this technique. Unlike thecontrol groups, tumors from the Galardin-treated micedistributed evenly into the ‘‘adenoma’’ and ‘‘early carci-noma’’ grades (Fig. 5). The Galardin-treated tumors thustended to be of lower histopathologic grade, although thedifference was not statistically significant (Kruskal-Wallistest, P = 0.13).

MMP Expression in MMTV-PymT-Induced Tumorsand LungMetastases

We used in situ hybridization to analyze the mRNAexpression patterns of the gelatinases MMP-2 and -9, thestromelysins MMP-3 and -10, collagenase-3/MMP-13, andmatrilysin/MMP-7. The gelatinases and stromelysins wereselected because they are two of the main MMP groups andbecause they are thought to play key roles in extracellularmatrix turnover, mammary gland morphogenesis, andmetastasis (2, 9). Collagenase-3 was selected due to itspresumed relevance for microinvasion (37). MMP-7 waschosen for its unique expression pattern in human breastcancer (38). Membrane-type MMPs were excluded becausetheir expression patterns have been reported (27). Weanalyzed primary tumors and lungs with metastases fromuntreated 13-week-old FVB-PymT mice. In primary tumors,we found that MMP-2, -3, -10, and -13 were expressed instromal cells. MMP-2 was expressed throughout the tumorstroma, whereas MMP-3 and -13 were restricted to certainsubcompartments of the stroma as we have reportedpreviously (26). We found that the expression of MMP-10(n = 4) was far less abundant than the former three, withprominent expression only in a few narrow streaks ofconnective tissue in areas of compact tumor tissue (Fig. 6A).Expression of MMP-7 and -9 was even more restricted. Wefound sporadic expression of MMP-7 (3 of 4 four cases) in apopulation of cancer cells exclusively located adjacent tonecrotic areas (Fig. 6B). We found sporadic expression ofMMP-9 (5 of 9 cases) often near or adjacent to necrotic areasin a pattern consistent with mesenchymal or inflammatorycells (Fig. 6C). We also analyzed metastasis-containinglungs (n = 5-7) for expression of these six MMPs. MMP-2was expressed in the larger bronchii (data not shown) andMMP-9 was expressed throughout the lung tissue inuniformly scattered cells (Fig. 6D). MMP-2 and -9 expres-sion was, however, not particularly associated with lungmetastases. We did not detect expression of MMP-3, -7, -10,or -13 in the lungs.

DiscussionWe have shown previously that the MPI Galardinefficiently suppresses metalloproteinase activity in vivo ,inhibiting wound healing (7, 39) and mammary glanddevelopment (6), although these processes are not com-pletely blocked by Galardin. We now report that tumorgrowth and metastasis in the MMTV-PymT mouse model

are also significantly inhibited, but not completely blocked,by continuous systemic administration of Galardin. Impor-tantly, lung metastasis volume was reduced f100-fold inGalardin-treated MMTV-PymT mice, whereas primarytumor volume was reduced only 2-fold. This suggests thatGalardin-sensitive metalloproteinases are directly involvedin spontaneous metastasis. Interestingly, there was only amodest reduction in the number of cytokeratin-8-positivetumor cells observed in the regional lymph nodes,suggesting that the effect of Galardin-sensitive metal-loproteinases on metastasis is primarily on hematogenousdissemination or perhaps on a later step of the metastaticprocess after escape of the cancer cells from the primarytumor.

The collagen network of the Galardin-treated MMTV-PymT mammary tumors was much more dense comparedwith the untreated tumors that contain large areasapparently devoid of collagen matrix. This difference isarguably a direct consequence of reduced collagen degra-dation in the presence of Galardin, which may limit theexpansion of tumor nodules due to physical constraintsand lead to preservation of the overall structure of thegland for a longer time. The trend toward lower histopath-ologic tumor grade in the Galardin-treated mice could thusbe an indirect consequence of reduced collagen degrada-tion. We observed a marked increase in pro-MMP-2 andactive MMP-2 in the Galardin-treated tumors concomitantwith the accumulation of collagen. A correspondingincrease in MMP-2 expression was found previously inGalardin-treated skin wounds (7). MMP-2 is one of themajor gelatinolytic enzymes. It is effectively inhibited by

Figure 5. Histopathologic progression of Galardin-treated tumors. Theno. 4 (right side) mammary glands were obtained from 92- to 96-d-oldFVB-PymT mice that had been randomly distributed to receive the MMPinhibitor Galardin (n = 26), placebo (n = 26), or no treatment at all(n = 27), starting at age 41 to 45 d. H&E-stained sections of the tissuesamples were scored according to the following defined histopathologicstages (25, 33): normal gland (A), hyperplasia (B), adenoma (C), earlycarcinoma (D), or late carcinoma (E). Data are presented as superimposedhistogram traces indicating the percentage of tissue samples thatreceived tumor grade scores A to E for each of the three groups of mice:untreated (o), placebo-treated (4), and Galardin-treated (E). Theobserved distribution differences were not significant (Kruskal-Wallistest, P = 0.13).





Galardin with an IC50 value in the range of 7 to 17 nmol/L(28, 29) and may thus be up-regulated in the tumors as aconsequence of substrate accumulation. The widespreadexpression of the MMP-2 mRNA in the MMTV-PymTtumor stroma (26) is consistent with an important role ofMMP-2 in extracellular matrix turnover. In fact, almost allof the MMPs, which have been analyzed by in situhybridization, were found to be expressed in the primaryMMTV-PymT tumors albeit at greatly varying levels. Wehave previously submitted data for MMP-2, -3, and -13 (26).In the present report, we add data for MMP-7, -9, and -10,whereas others have submitted data for MMP-14, -15, -16and -17 (27). From this total of 10 MMPs analyzed by in situhybridization in the MMTV-PymT tumors, MMP-2, -3, -9,-10, -13, -14, and -16 were found predominantly in thestromal cells and MMP-7 and -15 predominantly in the

cancer cells, whereas MMP-17 was not detected. Thesefindings agree well with the mRNA expression data forhuman breast cancer, because MMP-2 (40), MMP-3 (41),MMP-9 (42), MMP-13 (37), and MMP-14 (43) are all foundpredominantly in the stroma and MMP-7 predominantly incancer cells (38), suggesting that the MMTV-PymT model isa highly relevant model (reviewed in ref. 15). Thus far,we have not identified any MMP, whose expression isparticularly associated with lung metastases, althoughMMP-9-expressing cells were widespread in the lung tissueand could be important for tumor cell colonization.

Taken together, our findings indicate that tumor angio-genesis, reorganization of the mammary tissue microenvi-ronment, and even metastasis can occur despite continuousand systemic suppression of the Galardin-sensitive MMPsalthough with reduced efficiency. All of the protease

Figure 6. Expression of MMP-7, -9, and -10 mRNA in MMTV-PymT-induced mammary cancers and lung metastases. MMP mRNAs were detected byin situ hybridization in primary tumors (A-C) and lung metastases (D) with antisense RNA probes. Expression is seen as silver grains in the bright-fieldimages (row 1) and white reflections in the dark-field images (row 2). Hybridization with a nonoverlapping antisense RNA probe (row 3) and thecorresponding sense RNA probe (row 4) on adjacent sections provided positive and negative controls. A, MMP-10 mRNA was infrequent in theprimary breast tumors but was occasionally detected quite prominently in narrow bands of connective tissue in areas of compact tumor tissue (arrows ).B,MMP-7 mRNA was practically absent in the primary breast tumors and was restricted to occasional cancer cells (arrows ) adjacent to necrotic areas (n ).C, MMP-9 mRNA was practically absent in the primary breast tumors but was found in scattered stromal cells (arrows ) near or adjacent to necroticareas. D, MMP-9 mRNA was present in scattered cells throughout the entire lung tissue but was not convincingly associated with lung metastases(compare rows 2 and 3 with row 4 ). Sections were counterstained with H&E. Bars, 100 Am (A and D), 50 Am (C), and 25 Am (B).





intervention studies done in transgenic breast cancermodels report similar partial phenotypes such as delayedcarcinogenesis or slower tumor growth, if they report anyphenotype at all (15). These partial phenotypes areconsistent with individual proteases having mutuallyoverlapping functions. A functional overlap may existwithin the MMP family or even between different proteasefamilies as illustrated by the synergistic inhibition ofwound healing (7) and embryo implantation (3) obtainedby superimposing metalloproteinase inhibition (Galardintreatment) and genetic deficiency of the serine proteaseplasmin.

The MMPs have been linked to invasion and metastasisfor some time (reviewed in refs. 1, 44). Especially MMP-2and -9 have been associated with invasion and metastasisin studies that primarily use colonization assays of injectedcells or spontaneous metastasis from transplanted s.c.tumors (44). One such example involves Galardin, whichreduces the formation of osteolytic bone colonies fromcardially injected breast cancer cells (45). Unfortunately,the direct evidence regarding metastasis in transgenicbreast cancer models is rather limited (reviewed in ref. 15).Three examples almost exhaust the available evidence andat the same time illustrate the conflicting results that havebeen obtained. Metastasis incidence is thus reduced in theMMTV-PymT model by overexpression of TIMP-1 (18). Incontrast, MMP-7 overexpression has no effect on lungmetastasis in the MMTV-neu model (16), and MMP-11deficiency unexpectedly results in a higher number ofmetastases in the MMTV-ras model (46), suggesting asuppressive role of this MMP in cancer metastasis. Asimilar suppressive role against metastasis has beenreported in other models for MMP-3 (47), MMP-8 (48),and MMP-12 (49). Based on the limited evidence availablefor each MMP, it is probably premature to assign a generalmetastasis-suppressive role to any of these MMPs. Thebroader picture is that tumor-suppressive roles of MMPscan be identified at several stages of cancer progressionextending from carcinogenesis to metastasis (2), but thebulk of available evidence still identifies MMPs primarilyas tumor-promoting enzymes. The dual role of MMPs intumor progression is probably one of several reasons thatseveral clinical trials of broad-spectrum MPIs have failedin the past (2, 50). The promising effect observed withGalardin in the MMTV-PymT mouse model providesrenewed proof of principle and suggests that the potentialof broad-spectrum MPIs may not have been fully explored.The broad inhibition profile of Galardin (see Materials andMethods) does not allow identification of the criticalMMP(s), and it will undoubtedly be a challenge to identifya broad-spectrum MPI with a favorable inhibition profile.An alternative approach to selective MMP inhibitionwould be to use a combination of highly specific MMPinhibitors (e.g., inhibitory monoclonal antibodies). TheMMTV-PymT model, or other transgenic models with awell-characterized involvement of MMPs, could be auseful tool for the industry in testing the next generationof MPIs.

Disclosure of Potential Conflicts of InterestNo potential conflicts of interest were disclosed.

Acknowledgments

We thank Dr. William J. Muller for generously the MMTV-PymT transgene,Dr. Lars Bo Hansen for synthesizing Galardin for this study, Ib J.Christensen for assistance with statistical analyses, and AgnieszkaIngvorsen and John Post for expert technical assistance.

References

1. Egeblad M, Werb Z. New functions for the matrix metalloproteinases incancer progression. Nat Rev Cancer 2002;2:161–74.

2. Martin MD, Matrisian LM. The other side of MMPs: protective roles intumor progression. Cancer Metastasis Rev 2007;26:717–24.

3. Solberg H, Rinkenberger J, Danø K, Werb Z, Lund LR. A functionaloverlap of plasminogen and MMPs regulates vascularization duringplacental development. Development 2003;130:4439–50.

4. Stickens D, Behonick DJ, Ortega N, et al. Altered endochondral bonedevelopment in matrix metalloproteinase 13-deficient mice. Development2004;131:5883–95.

5. Fata JE, Leco KJ, Moorehead RA, Martin DC, Khokha R. TIMP-1 isimportant for epithelial proliferation and branching morphogenesis duringmouse mammary development. Dev Biol 1999;211:238–54.

6. Wiseman BS, Sternlicht MD, Lund LR, et al. Site-specific inductive andinhibitory activities of MMP-2 and MMP-3 orchestrate mammary glandbranching morphogenesis. J Cell Biol 2003;162:1123–33.

7. Lund LR, Rømer J, Bugge TH, et al. Functional overlap between twoclasses of matrix-degrading proteases in wound healing. EMBO J 1999;18:4645–56.

8. Jodele S, Blavier L, Yoon JM, DeClerck YA. Modifying the soil to affectthe seed: role of stromal-derived matrix metalloproteinases in cancerprogression. Cancer Metastasis Rev 2006;25:35–43.

9. Page-McCaw A, Ewald AJ, Werb Z. Matrix metalloproteinases and theregulation of tissue remodelling. Nat Rev Mol Cell Biol 2007;8:221–33.

10. Nagase H, Visse R, Murphy G. Structure and function of matrixmetalloproteinases and TIMPs. Cardiovasc Res 2006;69:562–73.

11. Oh J, Takahashi R, Kondo S, et al. The membrane-anchored MMPinhibitor RECK is a key regulator of extracellular matrix integrity andangiogenesis. Cell 2001;107:789–800.

12. Radisky DC, Bissell MJ. Matrix metalloproteinase-induced genomicinstability. Curr Opin Genet Dev 2006;16:45–50.

13. Sternlicht MD, Lochter A, Sympson CJ, et al. The stromal proteinaseMMP3/stromelysin-1 promotes mammary carcinogenesis. Cell 1999;98:137–46.

14. Ha HY, Moon HB, Nam MS, et al. Overexpression of membrane-typematrix metalloproteinase-1 gene induces mammary gland abnormalitiesand adenocarcinoma in transgenic mice. Cancer Res 2001;61:984–90.

15. Almholt K, Green KA, Juncker-Jensen A, Nielsen BS, Lund LR, RømerJ. Extracellular proteolysis in transgenic mouse models of breast cancer.J Mammary Gland Biol Neoplasia 2007;12:83–97.

16. Rudolph-Owen LA, Chan R, Muller WJ, Matrisian LM. The matrixmetalloproteinase matrilysin influences early-stage mammary tumorigen-esis. Cancer Res 1998;58:5500–6.

17. Blavier L, Lazaryev A, Dorey F, Shackleford GM, DeClerck YA. Matrixmetalloproteinases play an active role in Wnt1-induced mammarytumorigenesis. Cancer Res 2006;66:2691–9.

18. Yamazaki M, Akahane T, Buck T, et al. Long-term exposure toelevated levels of circulating TIMP-1 but not mammary TIMP-1 suppressesgrowth of mammary carcinomas in transgenic mice. Carcinogenesis 2004;25:1735–46.

19. Guy CT, Cardiff RD, Muller WJ. Induction of mammary tumors byexpression of polyomavirus middle T oncogene: a transgenic mouse modelfor metastatic disease. Mol Cell Biol 1992;12:954–61.

20. Grobelny D, Poncz L, Galardy RE. Inhibition of human skin fibroblastcollagenase, thermolysin, and Pseudomonas aeruginosa elastase bypeptide hydroxamic acids. Biochemistry 1992;31:7152–4.

21. Galardy RE. GalardinTM. Drugs Future 1993;18:1109–11.

22. Almholt K, Nielsen BS, Frandsen TL, Brunner N, Danø K, Johnsen M.





Metastasis of transgenic breast cancer in plasminogen activator inhibitor-1gene-deficient mice. Oncogene 2003;22:4389–97.

23. Almholt K, Lund LR, Rygaard J, et al. Reduced metastasis oftransgenic mammary cancer in urokinase-deficient mice. Int J Cancer2005;113:525–32.

24. Maglione JE, Moghanaki D, Young LJ, et al. Transgenic polyomamiddle-T mice model premalignant mammary disease. Cancer Res 2001;61:8298–305.

25. Lin EY, Jones JG, Li P, et al. Progression to malignancy in the polyomamiddle T oncoprotein mouse breast cancer model provides a reliable modelfor human diseases. Am J Pathol 2003;163:2113–26.

26. Pedersen TX, Pennington CJ, Almholt K, et al. Protease mRNAs arepredominantly expressed in the stromal areas of microdissected mousebreast carcinomas. Carcinogenesis 2005;26:1233–40.

27. Szabova L, Yamada SS, Birkedal-Hansen H, Holmbeck K. Expres-sion pattern of four membrane-type matrix metalloproteinases in thenormal and diseased mouse mammary gland. J Cell Physiol 2005;205:123–32.

28. Ma D, Wu W, Yang G, Li J, Ye Q. Tetrahydroisoquinoline basedsulfonamide hydroxamates as potent matrix metalloproteinase inhibitors.Bioorg Med Chem Lett 2004;14:47–50.

29. Uttamchandani M, Wang J, Li J, et al. Inhibitor fingerprinting of matrixmetalloproteases using a combinatorial peptide hydroxamate library. J AmChem Soc 2007;129:7848–58.

30. Nielsen BS, Lund LR, Christensen IJ, et al. A precise and efficientstereological method for determining murine lung metastasis volumes. AmJ Pathol 2001;158:1997–2003.

31. Rømer J, Bugge TH, Pyke C, et al. Impaired wound healing in micewith a disrupted plasminogen gene. Nat Med 1996;2:287–92.

32. Junqueira LCU, Bignolas G, Brentani RR. PicroSirius staining pluspolarization microscopy, a specific method for collagen detection in tissuesections. Histochem J 1979;11:447–55.

33. Lin EY, Nguyen AV, Russell RG, Pollard JW. Colony-stimulating factor1 promotes progression of mammary tumors to malignancy. J Exp Med2001;193:727–40.

34. Kristensen P, Eriksen J, Danø K. Localization of urokinase-typeplasminogen activator messenger RNA in the normal mouse by in situhybridization. J Histochem Cytochem 1991;39:341–9.

35. Wilson CL, Heppner KJ, Rudolph LA, Matrisian LM. The metal-loproteinase matrilysin is preferentially expressed by epithelial cells in atissue-restricted pattern in the mouse. Mol Biol Cell 1995;6:851–69.

36. Lund LR, Rømer J, Thomasset N, et al. Two distinct phases ofapoptosis in mammary gland involution: proteinase-independent and-dependent pathways. Development 1996;122:181–93.

37. Nielsen BS, Rank F, Lopez JM, et al. Collagenase-3 expression in

breast myofibroblasts as a molecular marker of transition of ductalcarcinoma in situ lesions to invasive ductal carcinomas. Cancer Res2001;61:7091–100.

38. Heppner KJ, Matrisian LM, Jensen RA, Rodgers WH. Expression ofmost matrix metalloproteinase family members in breast cancer representsa tumor-induced host response. Am J Pathol 1996;149:273–82.

39. Lund LR, Green KA, Stoop AA, et al. Plasminogen activationindependent of uPA and tPA maintains wound healing in gene-deficientmice. EMBO J 2006;25:2686–97.

40. Wolf C, Rouyer N, Lutz Y, et al. Stromelysin 3 belongs to a subgroupof proteinases expressed in breast carcinoma fibroblastic cells andpossibly implicated in tumor progression. Proc Natl Acad Sci U S A1993;90:1843–7.

41. Nakopoulou L, Giannopoulou I, Gakiopoulou H, Liapis H, Tzonou A,Davaris PS. Matrix metalloproteinase-1 and -3 in breast cancer: correlationwith progesterone receptors and other clinicopathologic features. HumPathol 1999;30:436–42.

42. Nielsen BS, Sehested M, Kjeldsen L, Borregaard N, Rygaard J, DanøK. Expression of matrix metalloprotease-9 in vascular pericytes in humanbreast cancer. Lab Invest 1997;77:345–55.

43. Okada A, Bellocq J-P, Rouyer N, et al. Membrane-type matrixmetalloproteinase (MT-MMP) gene is expressed in stromal cells of humancolon, breast, and head and neck carcinomas. Proc Natl Acad Sci U S A1995;92:2730–4.

44. Deryugina EI, Quigley JP. Matrix metalloproteinases and tumormetastasis. Cancer Metastasis Rev 2006;25:9–34.

45. Winding B, NicAmhlaoibh R, Misander H, et al. Synthetic matrixmetalloproteinase inhibitors inhibit growth of established breast cancerosteolytic lesions and prolong survival in mice. Clin Cancer Res 2002;8:1932–9.

46. Andarawewa KL, Boulay A, Masson R, et al. Dual stromelysin-3function during natural mouse mammary tumor virus-ras tumor progres-sion. Cancer Res 2003;63:5844–9.

47. McCawley LJ, Crawford HC, King LE, Jr., Mudgett J, Matrisian LM. Aprotective role for matrix metalloproteinase-3 in squamous cell carcinoma.Cancer Res 2004;64:6965–72.

48. Montel V, Kleeman J, Agarwal D, Spinella D, Kawai K, Tarin D.Altered metastatic behavior of human breast cancer cells after experimen-tal manipulation of matrix metalloproteinase 8 gene expression. CancerRes 2004;64:1687–94.

49. Houghton AM, Grisolano JL, Baumann ML, et al. Macrophageelastase (matrix metalloproteinase-12) suppresses growth of lung metas-tases. Cancer Res 2006;66:6149–55.

50. Overall CM, Kleifeld O. Tumour microenvironment. Opinion: validatingmatrix metalloproteinases as drug targets and anti-targets for cancertherapy. Nat Rev Cancer 2006;6:227–39.





2008;7:2758-2767. Mol Cancer Ther Kasper Almholt, Anna Juncker-Jensen, Ole Didrik Lærum, et al. transgenic breast cancer modelmetalloproteinase inhibitor Galardin in the MMTV-PymT Metastasis is strongly reduced by the matrix

Updated version

http://mct.aacrjournals.org/content/7/9/2758

Access the most recent version of this article at:

Cited articles

http://mct.aacrjournals.org/content/7/9/2758.full#ref-list-1

This article cites 50 articles, 20 of which you can access for free at:

Citing articles

http://mct.aacrjournals.org/content/7/9/2758.full#related-urls

This article has been cited by 9 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. (CCC)Click on "Request Permissions" which will take you to the Copyright Clearance Center's

.http://mct.aacrjournals.org/content/7/9/2758To request permission to re-use all or part of this article, use this link



http://mct.aacrjournals.org/content/7/9/2758.full#ref-list-1

http://mct.aacrjournals.org/content/7/9/2758.full#related-urls

http://mct.aacrjournals.org/cgi/alerts

mailto:[email protected]



metastasisisstronglyreducedbythematrix ... · metastasisisstronglyreducedbythematrix...

Documents