Inhibition of Invasion and Metastasis in Cells Transfected with an Inhibitor of Metalloproteinases1

1992;52:701-708. Cancer Res Yves A. DeClerck, Norma Perez, Hiroyuki Shimada, et al. an Inhibitor of MetalloproteinasesInhibition of Invasion and Metastasis in Cells Transfected with

Updated version

http://cancerres.aacrjournals.org/content/52/3/701

Access the most recent version of this article 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

[email protected] at

To request permission to re-use all or part of this article, contact the AACR Publications

on April 10, 2014. © 1992 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from on April 10, 2014. © 1992 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from



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

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

mailto:[email protected]




http://cancerres.aacrjournals.org/




[CANCER RESEARCH 52, 701-708. February 1. 1992]

Inhibition of Invasion and Metastasis in Cells Transfected with an Inhibitor ofMetalloproteinases1

Yves A. DeClerck,2 Norma Perez, Hiroyuki Shimada, Thomas C. Boone, Keith E. Langley, and Shirley M. Taylor

Division of Hematology/Oncology. Department of Pediatrics [Y. A. D., N. P.], the Department of Pathology [H. S.], and the Department of Microbiology [S. M. T.],University of Southern California, Los Angeles, California 90033, and AMGEN, Inc. [T. C. B., K, E. L.J, Thousand Oaks, California 91320

ABSTRACT

The balance between levels of metalloproteinases and their corresponding inhibitors is a critical factor in tumor invasion and metastasis. Down-regulation of the activity of these proteases was achieved by transfectionof invasive and metastatic rat cells with the complementary DNA formetalloproteinase inhibitor/tissue inhibitor of metalloproteinase 2 (MI/TIMP-2), a novel inhibitor of metalloproteinases recently described. (Y.A. DeClerck et al., J. Biol. Chem., 264: 17445-17453, 1989; W. G.Stetler-Stevenson et al., J. Biol. Chem., 264: 17374-17378, 1989). Secretion of functional MI/TIMP-2 protein in stably transfected cellsresulted in a marked decrease in metalloproteinase activity. Partialsuppression of the formation of lung colonies after i.v. injection in nudemice was observed in a transfected clone expressing high levels of MI/TIMP-2. Production of MI/TIMP-2 in four clones markedly reducedtumor growth rate in vivo after s.c. injection and completely suppressedlocal tissue invasion. Thus, down-regulation of metalloproteinase activityhas a striking effect on local invasion and partially suppresses hematog-

enous metastasis.

INTRODUCTION

Metastasis, a major cause of mortality and morbidity incancer patients, is a complex multistep process, during whichtumor cells locally invade the surrounding tissues, penetrateblood or lymphatic vessels (intravasation), and exit vessels atdistant sites (extravasation) to form secondary tumors (1).Proteolytic degradation of the ECM3 is an important part of

this process (2), and several classes of enzymes, includingMMP, serine proteinases (plasminogen activators), and cathep-sins, all abundantly secreted by a variety of tumor cells, havebeen implicated (3).

MMP are a family of Zn2+ dependent endopeptidases with a

broad spectrum of proteolytic activity for several componentsof the ECM (4). The family includes interstitial collagenase,type IV collagenases (M, 72,000 progelatinase, and M, 92,000progelatinase) and stromelysin (rat transin). These proteasesare substrate specific. Interstitial collagenase (MMP-1) degrades collagen types I, II, and III; M, 72,000 progelatinase(MMP-2) and M, 92,000 progelatinase (MMP-9) degrade collagen types IV and V; and denatured collagen (gelatin) andstromelysin (MMP-3) degrades fibronectin, laminin, andproteoglycans.

The secretion of MMP in an inactive proenzyme form is animportant feature that regulates their activity in the extracellular milieu. A second mechanism regulating the extracellular

Received 3/26/91 ; accepted 11/11/91.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 inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1This work was supported by Grant CA 42919 from the NIH, Department ofHealth and Human Services, and Grant BE-84 from the American Cancer Societyto Y. A. D. and in part by the Neil Bogart Memorial Laboratories of the T. J.Marteli Foundation for Leukemia, Cancer and AIDS Research.

! To whom requests for reprints should be addressed, at Children's Hospital

of Los Angeles, Division of Hematology/Oncology, 4650 Sunset Boulevard, LosAngeles, CA 90027.

3The abbreviations used are: ECM, extracellular matrix; MI, metalloproteinase inhibitor; MMP, matrix metalloproteinases; TIMP, tissue inhibitor ofmetalloproteinases; cDNA, complementary DNA; SSC, standard saline citrate.

activity of these enzymes is provided by inhibitors of metalloproteinases. TIMP, a M, 28,000 ubiquitous glycoprotein, isconsidered to be a major regulator of metalloproteinase activityin tissues. This inhibitor is secreted by many cells in cultureincluding fibroblasts (5), endothelial cells (6, 7), chondrocytes(8), and vascular smooth muscle cells (9) and is present in bone,cartilage, and amniotic fluid (10-13). TIMP inhibits MMP byforming an irreversible 1:1 stoichiometric complex with theactivated enzyme (14). We have recently reported the purification (15) and cloning (16) of a MI related to but distinct fromTIMP and other investigators have described an inhibitor thatforms a complex with M, 72,000 progelatinase in human tumorcells (17, 18). This latter protein, designated TIMP-2, has anucleotide sequence identical to the nucleotide sequence of MI(19).

Under normal conditions, the balance between MMP andtheir corresponding inhibitors is a critical factor that maintainsthe homeostasis of connective tissue proteins (4). We havetherefore postulated that under pathological conditions associated with excessive degradation of the ECM such as tumorinvasion and metastasis, there is an imbalance between metalloproteinases and metalloproteinase inhibitors. The observationof a direct correlation between the secretion of MMP by tumorcells and their invasive and metastatic potential suggests thatsuch imbalance can be achieved by increased production ofMMP by tumor cells (20-22). Alternatively, this imbalancecould be created by a decreased production of inhibitors, assuggested by the observation of an inverse correlation betweenlevels of TIMP production and the invasive potential of tumorcells (23, 24).

In this report, we have investigated whether down-regulationof the extracellular activity of metalloproteinases in tumori-genie, invasive, and metastatic cells could reestablish the balance between proteases and inhibitors and subsequently altertheir phenotype. c-Ha-ras I expressing rat embryo cells [clone4R (25)] were selected because of the large amount of metalloproteinases secreted (26) and the absence of detectable inhibitor.Down-regulation of their metalloproteinase activity wasachieved by stable transfection and expression of the cDNA ofMI/TIMP-2. Our previous studies have shown that recombinant MI/TIMP-2 can inhibit the degradation of ECM and theinvasion of artificial tissue substrates by these rat embryo cellsin vitro (27). Here we examine the effect of increased expressionof MI/TIMP-2 in the same cells on their in vitro and in vivoproperties.

MATERIALS AND METHODS

Cell Culture. c-Ha-ras-1 transfected rat embryo cells (4R) were obtained from Dr. L. A. Liotta (NIH, Bethesda, MD). Cells were culturedin Eagle's minimal essential medium (Grand Island Biological, Santa

Clara, CA) containing 10% (v/v) fetal bovine serum (Irvine Scientific,Irvine, CA), penicillin (100 units/ml), and streptomycin (100 Mg/ml).

Construction of pcDNA.HMI Vector and Cell Transfection. The human MI/TIMP-2 cDNA was inserted into an expression vector containing the cytomegalovirus promoter and enhancer elements (pcDNA;

701

on April 10, 2014. © 1992 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from



INHIBITION OF METALLOPROTEINASES IN TUMOR CELLS

Invitrogen, Inc., San Diego, CA). Human MI/TIMP-2 cDNA wasisolated as a Ncol/Stul fragment from plasmid pUC.HMI (28), bluntended, and ligated into plasmid pcDNA to form plasmid pcDNA.HMI(Fig. 1). Correct orientation was verified by restriction endonucleasedigestion. Cotransfection using calcium phosphate precipitation (29)was carried out with plasmid pY3 which encodes the hygromycin Bphosphotransferase gene as dominant selectable marker (30).

Metalloproteinase-Metalloproteinase Inhibitor Assays. Concentrated(50x) serum free conditioned media from cell cultures were tested forproteolytic activity for type I and type IV collagen and for inhibitoryactivity against rabbit fibroblast collagenase. Proteolytic activity inconditioned medium was measured after activation of metalloprotei-nases with 1 mM p-aminophenylmercuric acetate for 30 min at 37Â°C.

Aliquots of activated conditioned medium were then incubated in 96-well microtiter plates coated with 14C-labeled type I or type IV collagen

as described previously (9, 26). Inhibitory activity was measured byincubating various amounts of concentrated serum free conditionedmedium in the presence of rabbit fibroblast collagenase (0.05 unit) for15 min at 22'C followed by the determination of residual proteolytic

activity. One unit is defined as the amount of inhibitor that inhibits theproteolytic activity of 2 units of collagenase by 50% (15). The methodmeasures the level of inhibitory activity that is not already complexedwith secreted enzymes.

Acrylamide Gel Electrophoresis and Reverse Zymogram. Sodiumdodecyl sulfate-polyacrylamide gel electrophoresis was performed according to the method of Laemmli (31) using 12.5% (w/v) acrylamidegels. Reverse zymograms were used to detect the presence of gelatinasesand metalloproteinase inhibitors in culture media as reported previously(9, 15). In brief, gelatin (0.1%, w/v) was incorporated into the acrylamide mixture prior to polymerization. Samples were electrophoresedat 4'C without heat treatment. After electrophoresis, gels were incu

bated in Triton X-100 (2.5%, v/v) for l h with 2 changes to removesodium dodecyl sulfate, followed by 3 h at 37Â°Cin the presence of

gelatinase to allow partial proteolysis of gelatin. After overnight incubation at 37Â°Cin 50 mM Tris-200 mM NaCl-10 mM CaCl2, pH 7.5, the

gels were stained with 0.25% (w/v) Coomassie brilliant blue and des-tained in methanol:acetic acid:water (50:10:40). A clear zone indicatesthe presence of a protein with gelatinolytic activity, whereas a darkzone indicates the presence of an inhibitor of gelatinase.

Immunological Methods. Immunoblots were performed according tothe method of Burnette (32) using a rabbit polyclonal antiserum againstrecombinant MI/TIMP-2 as primary antibody (dilution, 1:200). Im-munocomplexes were stained with goat anti-rabbit IgG conjugated tohorseradish peroxidase (Bio-Rad Laboratories, Richmond, CA).Screening of transfected clones was done by immunodot-blot analysison a 96-well filtration manifold (Schleicher and Schnell. Keene, NH).Immunocomplexes were identified using a biotinylated affinity-purifiedanti-immunoglobulin, avidin, and biotinylated alkaline phosphatase(Vectastain ABC-AP kit; Vector Laboratories, Burlingame, CA).

Northern Blot Analysis. Cytoplasmic RNA was prepared from sub-confluent cultures after removal of nuclei by lysing cells in 0.5% (v/v)Nonidet P-40 and centrifugation at 2,000 x g for 10 min (33). TotalRNA from frozen lung nodules was extracted using the method ofChirgwin et al. (34). Polyadenylated RNA was separated from cellularRNA by chromatography on oligodoxythymidylate cellulose. RNAsamples (20 /ig) were electrophoresed in 1% formaldehyde-agarose gelsand blotted to nylon membranes (Irvine Scientific, Irvine, CA). Afterbaking at 80Â°Cfor 2 h, blots were hybridized in 5x SSC (lx SSC =0.15 M NaCl plus 0.015 M sodium citrate) at 42Â°Cin the presence of"P-labeled cDNA probes for 48 h. Blots were then washed in 3x SSCand 0.3x SSC at room temperature and autoradiographed at -80Â°C.

Tumorigenicity and Metastatic Assays. For tumorigenic assay, cellssuspended in 0.1 ml of sterile phosphate buffered saline (140 mM NaCl,4 mM KC1, 0.5 mM Na2HPO4, and 0.15 mM KH2PO4) were injecteds.c. into each flank of athymic (nude) mice. Tumor volume was calculated from three dimensional measurements using a caliper (35). After14 days, the animals were sacrificed and the tumor and adjacent tissueswere resected, fixed in 10% (v/v) buffered formalin, and embedded inparaffin for analysis by light microscopy. In 1 set of experiments (5animals), after s.c. injection of 4R tumor cells, the tumor on the right

flank received a daily local injection of 0.1 mg (in 0.1 ml of phosphatebuffered saline) of sterile recombinant MI/TIMP-2 obtained fromChinese hamster ovary cells (28). This injection was done initially atthe site of the injection of tumor cells and later at the posterior port ofthe tumor as it developed. The tumor on the left flank received a dailyinjection of 0.1 ml of sterile phosphate buffered saline (control). Forexperimental metastatic assays, 6-10-week-old athymic (nude) micewere given injections in the lateral tail vein of 2 x IO4cells suspendedin 0.1 ml of sterile phosphate-buffered saline. Cell viability, determinedby trypan blue exclusion prior to i.v. injection, was higher than 85%.After 2 weeks, animals were sacrificed and the number of colonies atthe surface of the lungs was determined under a dissecting microscopeafter endotracheal injection of India ink (26).

RESULTS

Transfection of 4R Cells with a Full Length MI/TIMP-2cDNA. We constructed an expression vector, pcDNA.HMI,containing human MI/TIMP-2 cDNA cloned from humanheart tissue (16), and placed under the control of a cytomeg-alovirus promoter and enhancer (Fig. 1). This expression plasmid was introduced into highly invasive and metastatic 4Rcells. Fifty-two clones cotransfected with pcDNA.HMI and pY3plasmids (hygromycin resistance marker) were established inthe presence of 0.4 mg/ml of hygromycin B and screened forthe secretion of MI/TIMP-2 by immunodot-blot analysis. Un-transfected cells and cells transfected with pY3 alone werenegative for the production of recombinant human MI/TIMP-2 and were used as control. We selected 5 clones cotransfectedwith pcDNA.HMI and pY3 (2 strongly positive, 2 weaklypositive, and 1 negative) and 2 clones transfected with pY3plasmid alone for a more detailed analysis. Northern analysisdemonstrated the presence of a 1.85-kilobase message in fourclones positive for MI/TIMP-2 production by immunodot-blotanalysis (Fig. 2, A and C). A 1.3-kilobase message was alsodetected in parental cells and in all transfected clones indicatinga low level of endogenous expression. The level of c-Ha-ras-Itranscript was generally unaffected by transfection and expression of MI/TIMP-2 (Fig. 2, B and D). Immunoblot analysis ofserum free conditioned medium of these clones (Fig. IE) indicated a correlation between the presence of the 1.85-kilobasetranscript and the detection of a M, 21,500 immunologicallyreactive protein of expected size for MI/TIMP-2 (15, 17). Byreverse zymogram analysis (Fig. 2F), we also demonstrated the

Kpnl

Kpnl

Kpnl

SV40Splice &poly A

M13

SV40

Fig. 1. Construction of pcDNA.HMI. Plasmid pcDNA.HMI was made byinserting the Ncol-Stul fragment of plasmid pUC.HMI into plasmid pcDNA byblunt end ligation. (Ml. cytomegalovirus; poly A, polyadenylation site; kh,kilobases.

702





secretion of two type IV collagenases (M, 92,000 and M, 72,000gelatinases) by all clones selected and the presence of functionalMI/TIMP-2 in clones expressing the 1.85-kilobase transcript(clones 8.27, 8.39, 8.60, and 8.68). The small amount of MI/TIMP-2 detected by reverse zymogram in clones 4.17 and 8.71(Fig. IF) and not detected by Western blot analysis (Fig. IE)likely represents endogenous (rat) MI/TIMP-2 that does notcross react with our anti-human MI/TIMP-2 antiserum.

Phenotypic Analysis of Transfected Clones. Analysis of theeffect of increased MI/TIMP-2 expression on proteolytic activity, growth rate in vitro, and tumorigenic and metastatic poten

tial was performed in the same 7 clones described above (Table1). The level of free inhibitor activity measured in the conditioned medium closely correlated with the presence of the 1.85-

kilobase message in transfected clones and no free inhibitoractivity was detected in parental cells or clones transfected withthe pY3 plasmid alone. As anticipated, the level of free inhibitoractivity detected in conditioned media of positive clones inversely correlated with the level of proteolytic activity for typeI and type IV collagens. Clone 4.6 which was found to have thehighest level of proteolytic activity for type IV collagenase(Table 1), secreted smaller amounts of M, 72,000 and M, 92,000

A 1234567844- â€ž¿� * â€¢¿�

2A-1.4- t â€¢¿� i

B4.4-

2.4-

1.4Â»

23 5678-130- 75

- 50

- 39

-Â«27

â€¢¿�Â«17

C 4A-

24-

6 5 6

â€¢¿�*MI/TIMP-2

Fig. 2. Analysis of production of functional MI/TIMP-2 in transfected 4R cells. A-D, Northern analysis of 4R cells and 4R cells transfected with pY3 andpcDNA.HMI. Samples of cytoplasmic RNA (20 Â¿ig;A and B) and polyadenylated RNA (5 ng; C and D) were electrophoresed on I % formaldehyde-agarose gels andblotted to nylon membranes. Probes used were MI/TIMP-2 oligonucleotide probe 198.29 (16) which does not cross-hybridize with TIMP (A and C) and H-RASprotooncogene from American Tissue Culture Collection (B and D). Left, positions of standard RNA markers (kilobases). E, immunoblot analysis of concentratedserum free conditioned media. Samples (unreduced) corresponding to 8.5 X 10' cells were loaded on sodium dodecyl sulfate-polyacrylamide. The gel was electroblottedand incubated with polyclonal rabbit antiserum against human recombinant MI/TIMP-2. Righi, positions of prestained molecular weight markers (in thousands). F,reversed zymogram analysis of concentrated conditioned media. Samples (unreduced and not heated) corresponding to 4.2 X Id' cells were electrophoresed on gelatin-containing sodium dodecyl sulfate-polyacrylamide and treated with gelatinase as described in material and methods. Right, positions of M, 92,000 and M, 72,000gelatinases (Gel.) and MI/TIMP-2. For all sections (A to F): Lane I, 4R parental cells; Lane 2, clone 4.6; Lane 3, clone 4.17; Lane 4, clone 8.27; Lane 5, clone 8.39;Lane 6, clone 8.60; Lane 7, clone 8.68; Lane S, clone 8.71. Clones 4.6 and 4.17 were transfected with the pY3 plasmid alone; clones 8.27, 8.39, 8.60, 8.68, and 8.71were cotransfected with pcDNA.HMI and pY3 plasmids.

Table 1 In vitro and in vivo characteristics of4R cells and 4R cells transfected with pcDNA.HMI andpÂ¥3

CloneParent

cells4.64.178.278.398.608.688.71Secreted

inhibitoryactivity"

(milliunits/IO6cells)None

NoneNone1103518

6285719

NoneSecreted

collagenaseactivity*Type

I(xg/ 10'cells)189

Â±1.6112.2 Â±9.570.5 Â±227.42 Â±522.16 Â±033.54 Â±2.86.12 Â±1.824.8 Â±9.7Type

IV(ng/10'cells)158

Â±10.6597 Â±12.1123 Â±14.619.5 Â±08.7 Â±4.24.5 + 6.4

25.2 Â±5.9119.7 + 3.3Doubling

mm''

(h)19.2

18.015.616.819.215.618.019.2Tumor1'8/8

8/88/88/88/88/88/88/8Metastastic

potential'234+11

(10)175Â± 16(11)483 + 32(11)391 + 19(11)150Â± 14(11)80 + 20(11)

295 + 33(10)351 +37(10)

" Free inhibitory activity in serum free conditioned medium was calculated in the presence of rabbit fibroblast interstitial collagenase.* Values represent amount of collagen degraded over 24 h at 37'C. Data points represent mean + SD of triplicate samples and were corrected for cell numbers.' Doubling time was calculated from growth curves obtained from cultures of cells plated at Id' cells in 35-mm tissue culture dishes. Values represent the doubling

time measured during the exponential phase of growth.d Tumor cells were injected at 5 x 10*(4 injections) and 1.2 x 10*(4 injections) s.c. in athymic (nude) mice and animals were observed for the formation of tumors.

Data represent number of tumors formed per number of injections. In all animals a tumor was detected within 10 days.' Data represent mean number of nodules + SE. Numbers in parentheses, total number of animals given injections during 2 separate experiments.

703





gelatinases than other clones (Fig. 2F). The absence of production of endogenous MI/TIMP-2 in this clone (as detected byreverse zymogram; Fig. 2F, Lane 2) may partially explain thisapparent discrepancy. It is also possible that this clone producesother proteases (such as plasminogen activators) that may beinvolved in the proteolytic degradation of type IV collagen (36).

Since it had previously been shown that down-regulation ofTIMP in Swiss 3T3 cells can affect tumorigenicity (37), we alsoexamined growth rate in vitro and tumorigenicity in vivo in allclones as well as parental 4R cells. Expression of MI/TIMP-2had no effect on these parameters. We then used an experimental metastatic model to determine whether inhibition of metalloproteinase activity in transfected clones would affect theirability to form lung colonies after i.v. injection into nude mice.The number of lung colonies formed by these clones was ingeneral not significantly altered by expression of MI/TIMP-2with one exception. Clone 8.60 showed a 66% decrease in thenumber of metastatic nodules (Fig. 3). The level of inhibitoryactivity produced by this clone was 6-12-fold higher than thelevels detected in the three other producing clones.

Expression of MI/TIMP-2 in Metastatic Tumor Nodules. Theabsence of complete suppression of metastasis raised the possibility that there was heterogenous expression of MI/TIMP-2in transfected clones and that these lung colonies may haveoriginated from subpopulations of cells that did not expressfunctional MI/TIMP-2. Northern analysis of RNA isolatedfrom metastatic lung nodules derived from clone 8.60 cells andsubcultured in vitro demonstrated that this was not the case,since the levels of MI/TIMP-2 and c-Ha-rai-I transcripts infive cultured mÃ©tastaseswere similar to the levels observed inclone 8.60 (Fig. 4, A and B). The same results were obtainedwhen RNA was extracted directly from tumor nodules (Fig.4C). Furthermore, cultured mÃ©tastasessecreted levels of im-munologically reactive and functional MI/TIMP-2 similar toparental clone 8.60 (Fig. 4, D and E), eliminating the possibilitythat the transfected cDNA may have been mutated to a nonfunctional gene.

Effect of MI/TIMP-2 Expression on Tumor Growth and LocalInvasion in Vivo. Since the experimental metastatic model usedonly allows for the investigation of a late step in metastasis(tumor cell extravasation), we also examined the effect of MI/TIMP-2 expression on an early step of tumor progression, i.e.,local tumor growth and invasion. We injected transfected cloness.c. into the flank of nude mice and found a significant difference in the growth rate in vivo of tumors formed by clones thatexpressed recombinant MI/TIMP-2 in comparison with 4Rparental cells and clones that did not express MI/TIMP-2 (Fig.5). The latency period for the appearance of a small tumornodule (0.1-0.2 cm') was the same for all clones and 4R

parental cells (8 days) but the average tumor size measured 14days after injection was less than 1 cm' in the 4 clones express

ing MI/TIMP-2. In nonexpressing clones and parental cells,the average tumor size ranged between 1.66 and 5.12 cm').

Furthermore, these tumors showed a striking difference in theirinvasiveness in vivo. HistolÃ³gica! analysis demonstrated thattumors derived from 4R cells and from clones not expressingMI/TIMP-2 had rapidly invaded the muscle of the abdominalwall, had penetrated the peritoneal cavity, and in one case hadpenetrated the kidney (Fig. 6). In contrast, tumors derived fromclones expressing MI/TIMP-2 were entirely confined to the s.c.tissue layer by a capsule of dense connective tissue and had notinvaded the muscle layer. The data therefore demonstrate that

871

8 60

827

417

Fig. 3. Analysis of metastatic nodules in nude mice given injections of 4Rclones. Athymic (nude) mice 6-8 weeks old were given i.v. injections of 2 x IO4cells. After 2 weeks animals were sacrificed and lungs were stained by intratrachealinjection of India ink. Right, clones analyzed.

elevated MI/TIMP-2 expression in 4R cells has a major effecton the process of local growth and invasion.

Effect of Local Administration of Recombinant MI/TIMP-2on Tumor Invasion. To further demonstrate that the formationof a fibrotic capsule around tumors expressing MI/TIMP-2was due to high MI/TIMP-2 production and not to othergenetic alterations that may have been associated with DNAtransfection, recombinant MI/TIMP-2 was administered locally by daily s.c. injection in tumors formed by parental 4Rcells. The recombinant inhibitor was initially administered atthe site of injection of the tumor cells and later as a tumordeveloped (days 8 to 9), between the posterior aspect of thetumor mass and the underlying adjacent tissues. Control tumorson the other flank of the animal were given daily injections of0.1 ml of sterile phosphate buffered saline (Fig. 7). Whereasthese injections had no effect on the overall tumor growth (datanot shown), histolÃ³gica! analysis of the tumors after 14 daysshowed the formation of a fibrotic capsule between the tumorcells and the adjacent muscle layer in tumors treated withrecombinant MI/TIMP-2. This fibrotic capsule was particularlyprominent in the posterior part of the tumor near the site ofinhibitor administration but was much less obvious in theanterior part of the tumor where invasion of the muscle layerwas detected (Fig. 7, A and C). In control tumors the musclelayer was extensively invaded (Fig. IB). These data thereforeare consistent with our hypothesis that local production of MI/TIMP-2 in tumor can alter the protease/protease inhibitorbalance in favor of the inhibitors. This change in balance resultsin the increased deposition of matrix proteins and the formationof a fibrotic capsule that limits tumor cell invasion.

DISCUSSION

We have demonstrated that it is possible to restore themetalloproteinase-metalloproteinase inhibitor balance in tumorcells and to markedly decrease the activity of their secretedmetalloproteinases by transfection with the cDNA of an inhibitor of these enzymes. In addition, restoration of this balancehas a significant impact on the behavior of these cells in vivo.In particular, we have observed that the local production ofMI/TIMP-2 in 4R cells markedly limited their growth in vivo

704





Fig. 4. Analysis of metastacic lung nodulesfrom 8.60 cells injected i.v. in nude mice. Individual lung nodules were obtained 14 daysafter clone 8.60 cells were injected i.v. intonude mice and were subcultured in vitro orquick frozen in liquid nitrogen. .â€¢/and B, cyto-plasmic RNA (20 <ig)from five cultured metastatic lesions (Ml-m5) was analyzed byNorthern blot analysis. Probes used were198.29 MI/TIMP-2 (A) and H RAS (B). C,Northern analysis of RNA from five frozenmetastatic nodules (nl to n5) hybridized withprobe 198.29. Right, positions of endogenous(1.3) and transfected (1.85) MI/TIMP-2 transcripts (in kilobases). D, immunoblot analysisof conditioned media from samples describedin A. Samples of concentrated (SOx) mediumcorresponding to 8.5 x IO4cells were analyzed.Ml, 0.2 Mg of recombinant MI/TIMP-2 ascontrol. Right, positions of prestained molecular weight standards (in thousands). Â£,reverse zymogram of samples described in D.corresponding to 4.2 x l(>4cells. Gel, gelati-

nasi* (M, 92,000 and M, 72,000).

4R 8.60m1 m2 m3 m4 m5

B

*

Ml aeo ml m2 m3 m4 m5

-75

-50

-39

k _-27

'Gel

n1 n2 n3 n4 n 5

ir Â»w -185

-.13.MI/TIMP-2

and completely suppressed their ability to invade surroundingtissues.

Our data show that despite induction of a significant changein the metalloproteinase-metalloproteinase inhibitor balance in4R cells, the metastatic behavior of these cells examined withan experimental model was not completely suppressed. Considering the highly complex and multistep nature of metastasis(1), a complete suppression of this process in 4R cells overex-pressing MI/TIMP-2 would be unexpected. Previous observations using similar experimental models have also demonstratedthat administration of recombinant TIMP was unable to completely suppress metastasis in mice (26,38). Since the inhibitionof metalloproteinases by MI/TIMP-2 is of a stoichiometricnature (17, 28), it is anticipated that the degree of inhibitionwill depend on the amount of inhibitor produced. This isconsistent with our observation showing that significant inhibition of metastasis could be achieved only in the presence ofvery high levels of secreted inhibitor. However, despite highlevels of MI/TIMP-2 secretion, a significant number of tumorcells are able to successfully metastasize. This finding suggeststhe presence of an escape mechanism, perhaps involving com-partmentalization of metalloproteinases and/or MI/TIMP-2(39, 40). For example, preferential localization of collagenaseat the cytoplasmic face of the plasma membrane of humanpancreatic carcinoma cells has been reported (41) and a similarlocalization in 4R cells may have prevented the enzyme frominteracting with the inhibitor. Alternatively, a proteolytic cascade other than metalloproteinases, such as the plasminogenactivator-plasmin system, may have been responsible for thisescape (36, 42). Our data suggest that in order to completelysuppress metastasis, other steps and/or proteolytic pathwaysmay have to be inhibited.

Khokha et al. (37) have previously shown that down-regulation of TIMP in nontumorigenic and nonmetastatic Swiss 3T3cells makes these cells invasive, tumorigenic, and metastatic.Although their data initially suggested that the acquisition ofone step (i.e., increased metalloproteinase activity) can allowcells to further advance through the last stage of tumor progression, their more recent clonal analysis of these cells indicates that down-regulation of TIMP is associated with furtherchanges in gene expression that may have accounted for theirtumorigenic potential (43). These data suggest that reductionin TIMP secretion enhances the oncogenic capacity of the cellnot only by changing the metalloproteinase activity but also by

altering the extracellular environment in multiple ways. It istherefore not surprising that in our studies inhibition of onlyone proteolytic pathway involved in local invasion and penetration through the basement membrane was insufficient to suppress distant metastasis in an experimental model.

The effect of MI/TIMP-2 expression on local tumor growthand invasiveness in vivo is a remarkable observation implyingthat this inhibitor may play an important role in an earlier stage

â€¢¿�4RV 4-6T 4-17D 8-27â€¢¿�8-39A 8-60A 8-68O 8-71

o>o;O

DAYS POST INJECTION

Fig. 5. In viro growth of tumor cells injected s.c. into nude mice. Parental 4Rcells and cells from clones transfected with pY.1plasmid or pY3 and pcDNA.HMIplasmids (10*cells suspended in 0.1 ml of phosphate buffered saline) were injected

s.c. in each flank of athymic mice (2 animals for each clone). After the appearanceof small nodules (0.1-0.2 cm') on day 8. tumor size was measured every 2 days

with a calipcr. Animals were sacrificed after 14 days and tumors were resectedand analyzed microscopically (Fig. 6). Data represent the mean tumor volume offour separate tumors for each cell line. Standard deviations were less than 20%of mean tumor volume.

705





rx^%* -,_;â€¢,:;Y|p ^:

^5'^^^"^ Ã•Ã”Ã‰i;^?:>.Â«Ã•MfiAk,Â¿I^F^̂ .Â«v*--;--i,.isÂ£nÅ“.vf ? :v*-*â€¢¿�'â€¢':â€¢'*â€¢â€¢.â€¢¿�-Â«v-^vAv/"V1-'^ â€¢¿�" " Pfjtf'>â€¢Â».**'.??\^: jf-'-Â«1'S*^. l "iifÃ„.:>Â¿-

'â€¢â€¢Â»â€¢W;'.^Ã^-^aÃfe*1feÃ¡r't^.l*%* '--AV*-â€¢:â€¢'- Ayf'Â«Ãr!Â»Ã;.â€¢*,â€¢â€¢â€¢¿�*-tei.â€¢¿�'^i>' JÃto^ 'â€¢';;â€¢Â»/>'/Â« "â€¢â€¢Si .Â»â€¢'*.â€¢ â€¢¿�"' '' - J * â€¢¿�T* â€¢¿�â€¢"T^.r.T^Dt*'*

^^^^^^^fe l;^Ã¨^ "â€¢y^Y

SA&-v^ ^ ^Ã¬^* * ;^r-'

-'^&s' L~

Fig. 6. Microscopic analysis of tumors derived from 4R cells and transfected clones injected s.c. into nude mice. Representative photomicrographs of thin sectionsof tumors stained with hematoxylin-eosin (original, x 100). .I. 4R parental cells. Note the invasion of the muscle. B, clone 4.6. Note invasion of the kidney on theright side of the photomicrograph. C, clone 8.27. Note absence of invasion of the muscle and presence of a fibrotic capsule that separates the tumor (left) from theunderlying muscle (right). D, clone 8.39. Note absence of invasion of s.c. tissue. E, clone 8.60. Note the presence of a similar fibrotic capsule between tumor cells(left) and muscle (right). F, clone 8.71. Note extensive invasion of the muscle as seen with parental 4R cells. Tumors from clone 4.17 (not shown) had invaded themuscle as seen with 4R parental cells, clone 4.6, and clone 8.71. Tumors derived from clone 8.68 (not shown) were similar to tumors from clone 8.60.

of tumor progression toward the metastatic phenotype. Thedata indicate that local production of this inhibitor causes adense fibrotic reaction that prevents tumor cells from invadingadjacent tissues and limits tumor growth in vivo. It is noteworthy that the accumulation of connective tissue proteins wasspecifically localized to the interface between the tumor cellsand the normal tissue. This suggests that the predominant effect

of MI/TIMP-2 was on increasing the deposition of ECM

proteins by mesenchymal cells, most likely by impairing theproteolytic degradation of these proteins by metalloproteinasessecreted by tumor cells and normal inflammatory cells. Sincelocal invasion is the first requirement for distant metastasis,inhibition of this step may have a significant impact on theprevention of distant metastasis.

706





Fig. 7. Microscopic analysis of tumors derived from 4R cells and treated locally with recombinant MI/TIMP-2. Photomicrographs of thin sections of tumorsstained with hematoxylin-eosin. A, 4R parental cells injected daily with 0.1 mg of sterile recombinant MI/TIMP-2. Note the presence of a thick capsule of connectivetissues (I'ih.) that separate tumor cells from the adjacent muscle in the posterior part of the tumor (post) near the site of inhibitor administration, whereas tumor

cells had invaded through the muscle in the anterior (ant.) of the tumor (original, x 20). B. 4R parental cells injected daily with O.I ml of sterile phosphate bufferedsaline. Note extensive invasion through the muscle as seen in Fig. <>.f (original, x 100). C, section of the posterior part of a tumor treated with recombinant MI/TIMP-2 showing the presence of a fibrotic reaction that separates tumor cells from muscle (original, x 100).

In this study, we have used cDNA transfection to express ametalloproteinase inhibitor in tumor cells and modulate theprotease-protease inhibitor balance. Our data are consistentwith a negative effect of MI/TIMP-2 expression on the invasivebehavior of tumor cells. However, final proof of the directinvolvement of MI/TIMP-2 in this process awaits the demonstration that the effect can be reversed by either antibody orantisense to MI/TIMP-2. Nonetheless, the presence of lowlevels of endogenous transcript for MI/TIMP-2 in these tumorcells suggests that our ability to modulate and enhance endog

enous expression will result in a similar effect on proteolyticactivity and may allow us to significantly alter tumor progression in vivo.

ACKNOWLEDGMENTS

The authors wish to thank B. Wilson for typing the manuscript, D.T. Yean for technical assistance, and K. Cheng-Chen for photographic

work.

707





REFERENCES

1. Fiedler, I..).. Balch, C. M. The biology of cancer metastasis and implicationsfor therapy. In: M. M. Ravitch (ed.). Current Problems in Surgery, pp. 131-209. Chicago: Year Book Medical Publishers, Inc., 1987.

2. Liotta, L. A. Tumor invasion and mÃ©tastases:role of the extracellular matrix.Cancer Res., 46:1-7, 1986.

3. Goldfarb, R. II. and Liotta, L. A. Proteolytic enzymes in cancer invasionand metastasis. Semin. Thromb. Hemost., 12: 294-307, 1986.

4. Matrisian, L. M. Metalloproteinases and their inhibitors in matrix remodeling. Trends Genet., 6: 121-125, 1990.

5. Stricklin, G. P., and Welgus, H. G. Human skin fibroblast collagenaseinhibitor. Purification and biochemical characterization. J. Biol. Chem., 258:12252-12258, 1983.

6. HerrÃ³n, G. S., Banda, M. J., Clark, E. J., Gavrilovic, J., and Werb, Z.Secretion of metalloproteinases by stimulated capillary endothelial cells. J.Biol. Chem., 261: 2814-2818, 1986.

7. DeClerck, Y. A., and Laug, W. E. Inhibition of tumor cell collagenolyticactivity by bovine endothelial cells. Cancer Res., 46:3580-3586, 1986.

8. Gavrilovic, J., Hembry, R. M., Reynolds, J. J., and Murphy, G. Tissueinhibitor of metalloproteinases (TIMP) regulates extracellular type I collagendegradation by chondrocytes and endothelial cells. J. Cell Sci., 87:357-362,1987.

9. DeClerck, Y. A. Purification and characterization of a collagenase inhibitorproduced by bovine vascular smooth muscle cells. Arch. Biochem. Biophys.,765:28-37, 1988.

10. Cawston, T. E., Galloway, W. A., Mercer, E., Murphy, G.. and Reynolds. J.J. Purification of rabbit bone inhibitor of collagenase. Biochem. J., 195:159-165, 1981.

11. Murphy, G., Cawston, T. E., and Reynolds, J. J. An inhibitor of collagenasefrom human amniotic fluid. Biochem. J., 195: 167-170, 1981.

12. Bunning, R. A. D., Murphy, G., Kumar, S., Phillips, P., and Reynolds, J. J.Metalloproteinase inhibitors from bovine cartilage and body fluids. Eur. J.Biochem., 139: 75-80, 1984.

13. Welgus, H. G., and Siriklin, G. P. Human skin fibroblast collagenase inhibitor. Comparative studies in human connective tissues, serum and amnioticHuid. J. Biol. Chem., 25Â«:12259-12264, 1983.

14. Welgus, H. G., Jeffrey, J. J., Eisen, A. Z., Roswit. W. T., and Stricklin. G.P. Human skin fibroblast collagenase: interaction with substrate and inhibitor. Coll. Relat. Res., 5: 167-179, 1985.

15. DeClerck, Y. A., Yean, T. D., Ratskin. B. J., Lu, H. S., and Langley, K. E.Purification and characterization of two related but distinct metalloprotei-nase inhibitors secreted by bovine aortic endothelial cells. J. Biol. Chem.,264: 17445-17453, 1989.

16. Boone. T. C, Johnson, M. J, DeClerck, Y. A., and Langley, K. E. cDNAcloning and expression of a metalloproteinase inhibitor related to tissueinhibitor of metalloproteinases. Proc. Nati. Acad. Sci. USA, 87: 2800-2804,1990.

17. Stetler-Stevenson, W. G., Krutzch, H. C.. and Liotta, L. A. Tissue inhibitorof metalloproteinase (TIMP-2). J. Biol. Chem., 264: 17374-17378, 1989.

18. Goldberg, G. I., Marmer, B. L., Grant, G. A., Eisen, A. Z., Wilhelm, S., andHe, C. Human 72-kilodalton type IV collagenase forms a complex with atissue inhibitor of metalloproteinases designated TIMP-2. Proc. Nati. Acad.USA, 86:8207-8211, 1989.

19. Stetler-Stevenson, W. G., Brown, P. D., Onisto, M., Levy, A. T., and Liotta,L. A. Tissue inhibitor of metalloproteinases-2 (TIMP-2) mRNA expressionin tumor cell lines and human tumor tissues. J. Biol. Chem., 265: 13933-13938, 1990.

20. Liotta, L. A., Tryggvason, K., Garbisa, S., Hart, L, Foltz, C. M., and Shafie,S. Metastastic potential correlates with enzymatic degradation of basementmembrane collagen. Nature (Lond.), 284: 67-68, 1980.

21. Nakajima, M., Welch, D. R., Belloni, P. N., and Nicolson, G. L. Degradationof basement membrane type IV collagen and lung subendothelial matrix byrat mammary adenocarcinoma cell clones of differing metastatic potentials.Cancer Res., 47:4869-4876, 1987.

22. Murphy, G., Reynolds, J. J., and Hembry, R. M. Metalloproteinases andcancer invasion and metastasis. Int. J. Cancer, 44: 757-760, 1989.

23. Halaka, A., Bunning, R. A. D., Bird, C. C., Gibson, M., and Reynolds, J. J.Production of collagenase and inhibitor (TIMP) by intracranial tumors anddura in vitro. J. Neurosurg., 49: 461-466, 1983.

24. Hicks, N. J., Ward, R. V., and Reynolds, J. J. A fibrosarcoma model derivedfrom mouse embryo cells: growth properties and secretion of collagenase andmetalloproteinase inhibitor (TIMP) by tumour cell lines. Int. J. Cancer, 33:835-844, 1984.

25. Pozzati, R., Muschel, R., Williams, J.. Padmanabhan, R., Howard, B., Liotta,L., and Khoury, G. Primary rat embryo cells transformed by one or twooncogenes show different metastatic potentials. Science (Washington DC),232:223-227, 1986.

26. Alvarez, O. A., Carmichael, D. F., and DeClerck, Y. A. Inhibition of collagenolytic activity and metastasis of tumor cells by a recombinant humantissue inhibitor of metalloproteinases. J. Nati. Cancer Inst., 82: 589-595,1990.

27. DeClerck, Y. A., Yean, D. T., Chan, D., Shimada, H., and Langley, K. E.Inhibition of tumor invasion of smooth muscle cell layers by recombinanthuman metalloproteinase inhibitor (rMI/rTIMP-2). Cancer Res., 51: 2151-2157, 1991.

28. DeClerck, Y. A., Yean, T. D., Lu, H. S., Ting, J., and Langley, K. E.Inhibition of autoproteolytic activation of interstitial procollagenase by re-combinant melalloproteinase inhibitor MI/TIMP-2. J. Biol. Chem., 266:3893-3899, 1991.

29. Cullen, B. R. Use of eukaryotic expression technology in the functionalanalysis of cloned genes. In: S. L. Berger and A. R. Kimmel (cds.), Methodsin Enzymology, pp. 684-704. New York: Academic Press, 1987.

30. Blochlinger, K., and Diggelman, H. Hygromycin B phosphotransferase as aselectable marker for DNA transfer experiments with higher eukaryotic cells.Mol. Cell. Biol., 4: 2929-2931, 1984.

31. Laemmli, U. K. Cleavage of structural proteins during the assembly of thehead of bacteriophage T4. Nature (Lond.), 227: 680-685, 1970.

32. Burnette, W. N. Western blotting: electrophoretic transfer of proteins fromsodium dodecyl sulfate-polyacrylamide gels to unmodified nitrocellulose andradiographie detection with antibody and radioiodinated protein A. Anal.Biochem., 112: 195-203, 1981.

33. Sambrook, J., Fritsch, E. F., and Maniatis, T. Molecular Cloning, a Laboratory Manual. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory,1989.

34. Chirgwin, J. M., Przybyla, A. E., McDonald, R. J.,and Rutter, W. J. Isolationof biologically active ribonucleic acid from sources enriched in ribonuclease.Biochemistry, 18: 5294-5299, 1979.

35. Tomayo, M. M., and Reynolds, P. Determination of subcutaneous tumorsize in athymic (nude) mice. Cancer Chemother. Pharmacol., 24: 148-152,1989.

36. Mackay, A. R., Corbitt, R. H., Hartzler, J. L., and Thorgeirsson, U. P.Basement membrane type IV collagen degradation:Evidence for the involvement of a proteolytic cascade independent of metalloproteinases. CancerRes., 50:5997-6001, 1990.

37. Khokha, R., Waterhouse, P., Yagel, S., Lala, P. K., Overall, C. M., Norton,G., and Denhardt, D. T. Antisense RNA-induced reduction in murine TIMPlevels confers oncogenicity on Swiss 3T3 cells. Science (Washington DC),2-tt: 947-950, 1989.

38. Schultz, R. M., Silberman. S., Persky. B., Bajkowski, A. S., and Carmichael.D. F. Inhibition by human recombinant tissue inhibitor of metalloproteinasesof human amnion invasion and lung colonization by murine Hid I Id melanoma cells. Cancer Res., 48: 5539-5545, 1988.

39. Moscatelli, D., and Rifkin, D. B. Membrane and matrix localization ofproteinases: a common theme in tumor cell invasion and angiogenesis.Biochim. Biophys. Acta, 948: 67-85, 1988.

40. Unemori, E. N., Bouhana, K. S., and Werb, Z. Vectorial secretion ofextracellular matrix proteins, matrix-degrading proteinases, and tissue inhibitor of metalloproteinases by endothelial cells. J. Biol. Chem., 265:445-451,1990.

41. Moll, U. M., Lane, B., Zucker, S., Suzuki, K., and Nagase, H. Localizationof collagenase at the basal plasma membrane of a human pancreatic carcinoma cell line. Cancer Res., 50:6995-7002, 1990.

42. Mignatti, P., Robbins, E., and Rifkin, D. B. Tumor invasion through thehuman amniotic membrane:requirement for a proteinase cascade. Cell, 47:487-498, 1986.

43. Khokha, R., Waterhouse, P., Lala, P., Zimmer, M., and Denhardt, D. T.Increased proteinase expression during tumor progression of cell lines down-modulated for TIMP levels: a new transformation paradigm? J. Cancer ResClin. Oncol., 117: 333-339, 1991.

708




Inhibition of Invasion and Metastasis in Cells Transfected with an Inhibitor of Metalloproteinases1

Documents

Transcript of Inhibition of Invasion and Metastasis in Cells Transfected with an Inhibitor of Metalloproteinases1