Synthetic Matrix Metalloproteinase Inhibitor,BB-94, Inhibits the Invasion of Neoplastic Human

Prostate Cells in a Mouse Model

J. David Knox,1* Louis Bretton,2 Tamara Lynch,2 G. Tim Bowden,3 andRay B. Nagle2

1Department of Pharmacology, McGill University, Montreal, Quebec, Canada2Department of Pathology, College of Medicine, University of Arizona, Tucson, Arizona

3Department of Radiation Oncology, College of Medicine, University of Arizona, Tucson, Arizona

BACKGROUND. It has been suggested that increased metalloproteinase activity is a criticalevent in neoplastic progression leading to the initiation of local invasion and ultimately to thedissemination of neoplastic cells. This has led to an interest in testing the ability of metallo-proteinase inhibitors to prevent the progression of carcinoma in situ into invasive and, there-fore, more malignant tumors. One such agent is the synthetic matrix metalloproteinase in-hibitor, BB-94.METHODS. The effect of BB-94 on the intrinsic invasive potential of matrilysin-transfectedDu-145 cells was evaluated by an in vitro invasion assay. In addition a diaphragm invasionmodel, which provides an easily oriented structure in which the earliest penetration of thebasal lamina can be observed, was used to investigate the effect of BB-94 on the invasion andgrowth of tumors formed by these cells when injected into S.C.I.D. mice.RESULTS. The synthetic matrix metalloproteinase inhibitor, BB-94, was shown to effectivelyinhibit the invasion of matrigel and murine diaphragm.CONCLUSIONS. Metalloproteinase inhibitors, such as BB-94, that are able to limit tumorgrowth, and local invasion, may decrease the invasion of invasive carcinomas. Prostate35:248–254, 1998. © 1998 Wiley-Liss, Inc.

KEY WORDS: batimastat; matrilysin; DU-145; matrigel

INTRODUCTION

In most cancers lethality is the result of local inva-sion and metastasis of neoplastic cells from the pri-mary tumor. Tumor cell invasion comprises threesteps: 1) attachment to the underlying basal lamina, 2)proteolysis of the basal lamina, and 3) migration of theneoplastic cells through the basal lamina and into theinterstitial stroma [1]. The proteolytic phase of tumorcell invasion is thought to result from the cooperationof multiple proteases produced by both the neoplasticcells and the surrounding stroma [2]. The metallopro-teinases, a multigene family of metal requiring en-zymes, specifically degrade components of the basallamina and are active at physiologic pH. These char-acteristics and the results of correlative immunohisto-chemical and in situ hybridization studies have sug-

gested that the metalloproteinases play a pivotal rolein tumor cell invasion [3–7].

The metalloproteinases are secreted in an inactiveform. Thus, their enzymatic activity is regulated bytheir production and activation, and by the relativeabundance of a family of specific endogenous inhibi-tors of metalloproteinases known collectively as tissueinhibitors of metalloproteinases (TIMPs). Functionalstudies have been performed where the relative levelsof metalloproteinases and TIMPs were modulated bysense and antisense expression vectors. These experi-ments demonstrated that a net increase in metallopro-

*Correspondence to: J. David Knox, Department of Pharmacology,McGill University, 3655 Drummond Ave., Montreal, Quebec H3G1Y6, Canada. E-mail: dknox@pharma.mcgill.caReceived 18 June 1997; Accepted 30 January 1998

The Prostate 35:248–254 (1998)

© 1998 Wiley-Liss, Inc.

teinase activity is associated with increased invasivepotential [8–12].

Collectively, these studies support the hypothesisthat increased metalloproteinase activity is a criticalevent in neoplastic progression, leading to the initia-tion of local invasion and ultimately to the dissemina-tion of neoplastic cells. Consequently, agents that in-hibit metalloproteinase activity should either block orretard invasion and arrest neoplastic progression atthe stage of hyperproliferative noninvasive cancer,which may be as clinically effective as blocking neo-plastic progression at an earlier stage [13].

One such agent is BB-94 (batimastat, [(4-N-hydroxyamino)-2R-isobutyl-3S-(thienyl-thiomethyl)succinyl]-L-phenyl-alanine-N-methylamide), a syn-thetic matrix metalloproteinase inhibitor [14]. Severalprevious studies clearly demonstrated that BB-94 re-stricts the growth of a variety of different tumor types[15–19]. An inhibition of tumor growth at both theprimary and metastatic sites was observed in all in-stances, and in two studies it was shown that the re-duced rate of tumor growth resulted in an increasedmedian survival time [15,17]. The effect of BB-94 ontumor invasion and metastasis is less clear. BB-94treatment did not influence the number of spontane-ous metastases observed in a murine melanoma study[16]. However, in an orthotopic colon carcinoma studyan apparent reduction in local invasion and spreadwas reported [17].

The purpose of this study was to rigorously exam-ine the ability of BB-94 to inhibit tumor invasion. Theinvasive properties of the human prostate tumor cellline, Duc-26, have previously been correlated with itsstable transfection and subsequent expression of themetalloproteinase matrilysin [11]. To evaluate the ef-fect of BB-94 on the intrinsic invasive potential of theDuc-26 cells, an in vitro assay was used. The invasivepotential of the tumor cells in vivo may be positivelyinfluenced by metalloproteinases secreted by the ad-jacent stromal cells or invading endothelial cells dur-ing angiogenesis. Our previously described dia-phragm invasion model was used to assay the effect ofBB-94 on the invasive ability of prostate tumor cells invivo because it provides an easily oriented structure inwhich the earliest penetration of the basal lamina canbe observed [12]. The effect of three different BB-94treatment regimens on the invasiveness, number, andsize of tumor foci were examined.

MATERIALS AND METHODS

Cells

Duc-26 cells were generated by transfecting the Du-145 cell line derived from a central nervous system

metastasis of a primary prostate adenocarcinoma(American Type Culture Collection, Rockville, MD)with a vector conferring neomycin resistance and abeta-actin promoter to drive the expression of a cDNAfor wild-type matrilysin [12]. The resultant stablytransfected cell line was grown in cell culture underselection in MEM (Gibco, Grand Island, NY) supple-mented with 10% fetal calf serum (JRH Biosciences,Lenexa, KS), 100 U/ml penicillin/streptomycin(Gibco), and G418 (350 active units/ml) at 37°C and5% CO2.

Batimastat, [4-(N-hydroxyamino)-2R-isobutyl-3S-(thienyl-thiomethyl)-succinyl]-L-phenylalanine-N-methylamide, was provided by British BioTech Phar-maceuticals, Ltd. (Oxford, UK).

In Vitro Invasion Assay

In a volume of 1.5 ml serum-free medium, 4.0 × 105Duc-26 cells were plated in the top chamber of Col-laborative Biomedical Products (Bedford, MA) cellculture inserts containing 8.0-mm pores that were oc-cluded with matrigel (catalogue no. 40481). In the bot-tom chamber, a chemoattractant consisting of 2.0 ml ofmedium was placed, conditioned by a confluent 100-mm plate by NIH 3T3 cells. The conditioned mediumin the bottom chamber was replaced by 2 ml of freshlyconditioned medium after 24 hr. The media of boththe top and bottom chambers of the treatment wellscontained 60 nM of BB-94 prepared by making serialdilutions from a 3-mM stock solution of BB-94 dis-solved in DMSO. After incubation for 48 hr, the mem-branes were cut out with a scalpel and stained usingthe Diff-Quik Stain Set (Baxter Scientific Products, Mc-Gaw Park, IL). The number of cells on the top andbottom sides of the membrane was determined usinga 40× objective. The narrow depth of focus of thisobjective resulted in only one side of the membrane ofa given field being in focus at a time, enabling one todistinguish between invasive and noninvasive cells.

Animals

The S.C.I.D. mouse colony at the University of Ari-zona are BALB-C/B-17 mice that were originally pro-vided by L. Shultz of Jackson Laboratories (Bar Har-bor, ME). They were maintained in a specific patho-gen-free environment in compliance with USPHSguidelines governing the care and maintenance of ani-mals.

Experimental Design

Six groups, 3–4 mice per group, of 4-week-oldS.C.I.D. mice were injected on day 0 with 5 × 106 Duc-26 cells i.p. A BB-94 dose of 30 mg/kg, which was

Inhibition of Prostate Tumor Cell Invasion 249

previously shown to result in serum levels of 12–30ng/ml in BalbC mice over 24 hr, was chosen [17]. TheBB-94 (suspended in phosphate-buffered saline, pH7.4, containing 0.01% Tween-80) was administeredthree times/week i.p. commencing 1, 2, or 3 weeksafter injection of the tumor cells and then continuedfor the duration of the 4-week experiment. As shownin Figure 1, this protocol generates experimentalgroups of mice that have been treated with BB-94 forthe last 3 weeks of the experiment, the last 2 weeks ofthe experiment, or just the last week of the experiment,respectively. Vehicle alone was administered to the 2-and 1-week experimental groups as well as all thecontrol groups, starting at week 2 of the experimentand continuing until their BB-94 treatment was com-menced or their diaphragms were collected. The dia-phragms of the 2- and 3-week control groups werecollected 2 and 3 weeks, respectively, after injection ofthe tumor cells. The diaphragms of the 4-week controlgroup and all the BB-94-treated groups were collected4 weeks after the injection of the tumor cells.

Diaphragm Analysis

The two caudal-most ribs and the portion of thespine to which they attach were removed along withthe diaphragm and fixed in 10% buffered formalinovernight. The samples were then transferred to 70%ethanol until being run through a tissue processor.This method of sample collection ensures that the dia-phragm is fixed in a distended configuration.

Upon removal from the tissue processor, the dia-phragm was detached from the ribs and spine and cutinto three or four strips which were then embedded ina paraffin block perpendicular to the cutting surface.Sections were then cut and immunohistochemicallystained using the monoclonal antibody, 10/11, whichstains human cytokeratins 8 and 18 as the primaryantibody, and indirect biotin-streptavidin peroxidase

staining, which highlighted the human tumors withinthe unstained mouse tissue. The sections were thencounterstained with hematoxylin.

The sections were examined with an Olympus BH-2microscope equipped with an SPlan 10× objective. Thenumbers of tumors observed in each section werequantified and scored as invasive or noninvasive. Inspecimens where no tumors were observed in the ini-tial section, an additional five sections were cut at 50-mm intervals. The first section containing at least onetumor was scored. If all six sections were negative, theanimal was considered negative and excluded fromthe study.

For further analysis the images of the observed tu-mors were captured with a Kontron Elektronik camera(ProgRes 3012, Kontron Elektronic Corp., NewportBeach, CA) connected to a computer equipped withImage Manager software (version 2.2, Roche ImageAnalysis Systems, Inc., Eton College, NC). Follow-ing conversion to PICT file format, the imageswere analyzed using NIH Image (version 1.55,http://woolsey-indigo2.wustl.edu/jon/LISTS/nihimage.htm). This software enabled a determinationof the tumor area as well as the calculation of the ratioof tumor area beneath and above the basal lamina.

Statistical Analysis

A single-factor ANOVA was performed to deter-mine if there were significant differences between thegroups, using Microsoft Excel Version 5.0a (MicrosoftCorp., Redmond, WA). If a significant difference wasobserved, an LSD test was used to identify whichgroups differed significantly from one another (Stat-soft, Inc., Tulsa, OK). The data plotted are the mean ±the standard error of the average values determinedfor each individual within the treatment group.

RESULTS

In Table I, the invasive potential of the Duc-26 cellsis compared to that of the parental Du-145 cell line.The results of two in vitro invasion assays demon-strate that the Duc-26 cells are more invasive. In TableII, the effect of the presence of 60 nM BB-94 in themedia on the invasive potential of the Duc-26 cells ispresented. The presence of BB-94 had no effect on theability of Duc-26 cells to migrate through 8-mm pores(data not shown). However, it clearly blocked the abil-ity of the Duc-26 cells to invade through pores oc-cluded with matrigel.

The diaphragm invasion model provides an easilyoriented structure in which the earliest penetration ofthe basal lamina can be observed. An example of anoninvasive and an invasive tumor formed by Duc-26

Fig. 1. Flow chart showing experimental design. All mice areinjected at T = 0. Open bar indicates that mice are being injectedwith vehicle alone. Solid bar indicates that mice are being injectedwith BB-94.

250 Knox et al.

cells in a control and a BB-94-treated mouse are shownin Figure 2A,B, respectively. The unstained basallamina against the stained background of an invasivetumor makes it easy to distinguish between invasiveand noninvasive tumors as well as to delineate theinvasive and noninvasive portions of a particular tu-mor.

Assuming a random distribution, the number oftumors observed in any given individual will be re-lated to the area of the diaphragm examined. All the

sections were of a standard thickness, 5 mm, but thelength of diaphragm represented in the sections var-ied depending upon the size of the diaphragm and thenumber of strips into which the diaphragm was cut. Inorder to determine the influence of BB-94 treatment ontumorigenicity, the number of tumors observed wasnormalized to the measured length of the diaphragmpresent in the section scored. These data are shown inFigure 3, where it can be seen that the average numberof tumors/centimeter of diaphragm examined rangedfrom 0.21–0.55; however, no significant difference wasobserved between the treated and control groups(P = 0.848).

The effect of BB-94 on tumor size was determinedby measuring the area of each tumor present in thecross section scored. The results of this analysis are

TABLE I. Matrilysin Expression Is Correlated WithIn Vitro Invasive Potential of Human Prostate

Cancer Cells*

Cell lineNumber of

invasive cells

Number ofnoninvasive

cells % invasion

Experiment 1Duc-26 244 1,288 18.9Du-145 58 1,456 4.0Experiment 2Duc-26 99 691 12.5Du-145 15 596 2.5

*Random fields of each membrane were examined. The numberof cells observed on the top side of the membrane was countedfirst. Then by changing the plane of focus, the correspondingnumber of cells on the bottom of the membrane in the same fieldwas noted. The percent invasion figures presented were calcu-lated by dividing the total number of cells on top of the mem-brane by the total number of cells on the bottom of the mem-brane.

TABLE II. Inhibition of Metalloproteinase Activityby BB-94 Reduces the In Vitro Invasive Potential of

Duc-26 Cells*

Cell lineNumber of

invasive cells

Number ofnoninvasive

cells % invasion

Experiment 1Duc-26 31 272 11.9Duc-26 + BB-94 0 337 0.0Experiment 2Duc-26 124 437 32.4Duc-26 + BB-94 4 508 0.6

*The cells and the conditioned medium used in these experi-ments were matched, the sole difference being the presence orabsence of 60 nM BB-94. Random fields of each membrane wereexamined. The number of cells observed on the top side of themembrane was counted first. Then by changing the plane offocus, the corresponding number of cells on the bottom of themembrane in the same field was noted. The percent invasionfigures presented were calculated by dividing total number ofcells on top of the membrane by the total number of cells on thebottom of the membrane.

Fig. 2. Paraffin sections of a noninvasive (A) and an invasive (B)tumor were cut and immunohistochemically stained using themonoclonal antibody, 10/11, which stains human cytokeratins 8and 18 as the primary antibody, and indirect biotin-streptavidinperoxidase staining, which highlighted human tumors within theunstained mouse tissue. Sections were then counterstained withhematoxylin. The unstained basal lamina (arrows) against thestained background of an invasive tumor makes it easy to distin-guish between invasive and noninvasive tumors, as well as to de-lineate the invasive and noninvasive portions of a particular tumor.

Inhibition of Prostate Tumor Cell Invasion 251

shown in Figure 4. As was the case with tumor num-ber, treatment with BB-94 had no significant effect ontumor area (P = 0.835).

The percent invasion determined as the average ofthe percent invasion observed in each individualmouse within a group is shown in Figure 5. Tumors inall three of the BB-94-treated groups as well as in the2-week control group were significantly less invasivethan in the 3- and 4-week control groups (P = 0.011).Interestingly, mice that received 1 week of BB-94 treat-ment beginning 3 weeks after tumor cell inoculationhad fewer invasive tumors than control mice sacri-ficed at 3 weeks after tumor inoculation. The percentinvasion demonstrated by the control group sacrificedat 3 weeks was 58.8 ± 18.1%, while the percent invasionobserved in the mice that received 1 week of BB-94treatment 3 weeks after tumor cell inoculation was 6.6± 6.4%.

DISCUSSION

In the in vitro invasion assay, the tumor cells are theonly potential source of metalloproteinases. The in-creased invasive potential demonstrated by the Duc-26 cells relative to the parental Du-145 cell line sug-gested that the forced expression of matrilysin is re-sponsible for the more invasive phenotype displayedby the Duc-26 cells. The ability of 10 times the re-ported in vitro IC50 of BB-94 against matrilysin [14] tocompletely abrogate the observed invasive potentialfurther strengthens the argument that the synthesisand secretion of metalloproteinases is a required char-acteristic of the invasive phenotype of Duc-26 cells.

The administration of BB-94 commencing 1 weekafter the i.p. injection of tumor cells had no significanteffect on the number of diaphragm tumors observed.

This suggests that either most of the tumors observedwere already established 1 week after injection, or thatmetalloproteinases are not involved in tumor estab-lishment at this site. This is in contrast to an earlierreport where treatment with BB-94 beginning 10 daysafter intraperitoneal tumor cell inoculation of humancolon carcinoma cells significantly reduced tumorige-nicity [18].

Regression or eradication of established tumorswas not expected because BB-94 does not demonstrateany direct cytotoxic effect [15,18,20]. However, if theBB-94 treatment had a significant effect on tumorgrowth and thereby reduced the cross-sectional areaof the tumors, then the number of tumors observed ina random sampling of cross sections would be re-duced. As predicted by the tumor number data, BB-94

Fig. 3. A random section of each specimen was examined andthe number of tumors observed in each individual was noted. Thefirst section containing at least one tumor was scored. The dataplotted are the mean ± standard error of the average values de-termined for each individual within the treatment group. No sig-nificant difference in tumor number was observed (P = 0.848).

Fig. 4. A random section of each specimen was examined andthe cross-sectional area of the tumors observed in each individualwas determined. The first section containing at least one tumorwas scored. The data plotted are the mean ± standard error of theaverage values determined for each individual within the treatmentgroup. BB-94 had no significant effect on the cross-sectional areaof the tumors observed (P = 0.835).

Fig. 5. A random section of each specimen was examined andthe number of tumors observed in each section was noted andscored as invasive or noninvasive. The first section containing atleast one tumor was scored. The data plotted are the mean ±standard error of the average values determined for each indi-vidual within the treatment group. The 3- and 4-week controlswere found to be significantly different from the other four groups(P = 0.011).

252 Knox et al.

did not have a significant effect on the cross-sectionalarea of the Duc-26 diaphragm tumors.

In contrast to our results, a negative effect of BB-94on the growth of a variety of tumor types has beenreported in earlier studies, in which it was suggestedthat BB-94 prevented tumor growth either by limitingthe ability of the tumor to break down and expandinto the adjacent tissue, or by inhibiting tumor vascu-larization [15–19]. There are significant data support-ing the hypothesis that the growth of solid tumors isdependent upon angiogenesis [21], and angiogenesisshares the same steps of adhesion, proteolysis, andmigration as tumor invasion. Specifically, metallopro-teinases play a key role in both processes [22], and ithas been proposed that matrix metalloproteinase in-hibitors such as BB-94 could inhibit tumor growth byinhibiting angiogenesis [13].

Neither of these arguments contradicts the lack oftumor growth inhibition observed in this study. Thetumors analyzed in this study were situated on theperitoneal side of the diaphragm, where growth is notconstrained by a surrounding stroma. Immunohisto-chemical analysis of diaphragm tumors formed byDu-145 variants with an endothelial cell-specificmarker, PECAM [23], has shown an absence of endo-thelial cells in tumors of the small size formed by theprotocol used in this study (A.E. Cress, personal com-munication). Because these tumors are likely to be inthe preangiogenic phase of tumor development, theymay be refractory to the growth-inhibitory effects ofBB-94.

An interesting result was that treating the animalswith BB-94 for just the last week of the experimentresulted in significantly fewer invasive tumors thanobserved in the 3-week control mice. This was unex-pected because, having no direct cytotoxic effect, BB-94 was not expected to eradicate established invasivetumors. However, at the time that it was first admin-istered, ∼50% of the tumors should have already pen-etrated the basal lamina and should have been invad-ing the skeletal musculature of the diaphragm, as wasobserved in the 3-week controls. This result suggeststhat the BB-94 treatment resulted in the reduced num-ber of invasive tumor foci observed.

Previously, we demonstrated that invasive neoplas-tic prostate cells are surrounded by a basal laminasimilar in composition to that surrounding normalprostate cells [24]. This observation indicated that thedegradation and deposition of the basal lamina at theleading edge of the invading tumor is a dynamic pro-cess. We propose that the inhibition of basal laminadegradation following the administration of BB-94leads to the rapid reformation of a basal lamina at theinterface between the tumor cells and the stroma.Once reformed, it is possible that the tautness and

movement of the diaphragm smoothed out the inter-face between the tumor and the diaphragm, produc-ing the observed result. Alternatively, in contrast tothe tumor cells growing on the peritoneal side on thediaphragm, it is possible that neovascularization ofthe invasion front of the tumor is required to sustainthe invasive cells. In this case, BB-94 may have causeda regression of the tumors by inhibiting angiogenesis.This hypothesis could be tested by increasing the pe-riod between the injection of the tumor cells and thecommencement of BB-94 administration. Once thevasculature of the invasion front of the tumor wasestablished, the hypothesis predicts that no regressionof the invasive cells would be observed.

In conclusion, this report demonstrates that BB-94 isable to inhibit local invasion, as evidenced by the in-hibition of matrigel and of diaphragm invasion by theDuc-26 cells. These characteristics suggest that metal-loproteinase inhibitors, such as BB-94, that are able tolimit tumor growth and local invasion, may be effec-tive therapeutic agents.

REFERENCES

1. Liotta LA, Stetler-Stevenson WG: Tumor invasion and metasta-sis: An imbalance of positive and negative regulation. CancerRes 1991;51:5054–5059.

2. Schmitt M, Janicke F, Graeff H: Tumor-associated proteases.Fibrinolysis [Suppl] 1992;4:3–26.

3. D’Errico A, Garbisa S, Liotta LA, Castronovo V, Stetler-Stevenson WG, Gigioni WF: Augmentation of type IV collage-nase, laminin receptor, and Ki67 proliferation antigen associatedwith human colon, gastric, and breast carcinoma progression.Mod Pathol 1991;4:239–246.

4. Levy AT, Cioce V, Sobel ME, Garbisa S, Grigioni WF, Liotta LA,Stetler-Stevenson WG: Increased expression of the Mr 72,000type IV collagenase in human colonic adenocarcinoma. CancerRes 1991:51:439–444.

5. Stetler-Stevenson WG, Liotta LA, Kleiner DE Jr: Extracellularmatrix 6: Role of matrix metalloproteinases in tumor invasionand metastasis. FASEB J 1993;7:1434–1441.

6. Basset P, Wolf C, Rouyer N, Bellocq JP, Rio MC, Chambon P:Stromelysin-3 in stromal tissue as a control factor in breast can-cer behavior. Cancer 1994;74:1045–1049.

7. Knox JD, Wolf C, McDaniel K, Clark V, Loriot M, Bowden GT,Nagle RB: Matrilysin expression in human prostate carcinoma.Mol Carcinog 1996;15:57–63.

8. Khokha R, Waterhouse P, Yagel S, Lala PK, Overall CM, NortonG, Denhardt DT: Antisense RNA-induced reduction in murineTIMP levels confers oncogenicity on Swiss 3T3 cells. Science1989;243:947–950.

9. Noel AC, Lefebvre O, Maquoi E, VanHoorde L, Chenard MP,Mareel M, Foidart JM, Basset P, Rio MC: Stromelysin-3 expres-sion promotes tumor take in nude mice. J Clin Invest 1996;97:1924–1930.

10. Witty JP, McDonnell S, Newell KJ, Cannon P, Navre M, TresslerRJ, Matrisian LM: Modulation of matrilysin levels in colon car-cinoma cell lines affects tumorigenicity in vivo. Cancer Res 1994;54:4805–4812.

11. Powell WC, Knox JD, Navre M, Grogan TM, Kittelson J, Nagle

Inhibition of Prostate Tumor Cell Invasion 253

RB, Bowden GT: Expression of the metalloproteinase matrilysinin DU-145 cells increases their invasive potential in severe com-bined immunodeficient mice. Cancer Res 1993;3:417–422.

12. Knox JD, Mack CF, Powell WC, Bowden GT, Nagle RB: Prostatetumor cell invasion: A comparison of orthotopic and ectopicmodels. Invasion Metastasis 1993;13:325–331.

13. Kohn EC, Liotta LA: Molecular insights into cancer invasion:Strategies for prevention and intervention. Cancer Res 1995;55:1856–1862.

14. Botos I, Scapozza L, Zhang D, Liotta LA, Meyer EF: Batimistat,a potent matrix metalloproteinase inhibitor, exhibits an unex-pected mode of binding. Proc Natl Acad Sci USA 1996;93:2749–2754.

15. Davies B, Brown PD, East N, Crimmin MJ, Balkwill FR: A syn-thetic matrix metalloproteinase inhibitor decreases tumor bur-den and prolongs survival of mice bearing human ovarian car-cinoma xenografts. Cancer Res 1993;53:2087–2091.

16. Chirivi RGS, Garofalo A, Crimmin MJ, Bawden LJ, StoppacciaroA, Brown PD, Giavazzi R: Inhibition of the metastatic spreadand growth of B16-BL6 murine melanoma by a synthetic matrixmetalloproteinase inhibitor. Int J Cancer 1994;58:460–464.

17. Wang X, Fu X, Brown PD, Crimmin MJ, Hoffman RM: MatrixMetalloproteinase inhibitor BB-94 (batimistat) inhibits humancolon tumor growth and spread in a patient-like orthotopicmodel in nude mice. Cancer Res 1994;54:4726–4728.

18. Watson SA, Morris TM, Robinson G, Crimmin MJ, Brown PD,

Hardcastle JD: Inhibition of organ invasion by the matrix me-talloproteinase inhibitor batimistat (BB-94) in two human coloncarcinoma metastasis models. Cancer Res 1995;55:3629–3633.

19. Taraboletti G, Garofalo A, Bellotti D, Drudis T, Borsotti P, Scan-ziani E, Brown PD, Giavazzi R: Inhibition of angiogenesis andmurine hemangioma growth by batimistat, a synthetic inhibitorof matrix Metalloproteinases. J Natl Cancer Inst 1995;87:293–298.

20. Sledge GW Jr, Qulali M, Goulet R, Bone EA, Fife R: Effect ofmatrix metalloproteinase inhibitor batimistat on breast cancerregrowth and metastasis in athymic mice. J Natl Cancer Inst1995;87:1546–1550.

21. Folman J, Klagsbrun M: Angiogenic factors. Science 1987;235:442–447.

22. Liotta LA, Steeg PS, Stetler-Stevenson WG: Cancer metastasisand angiogenesis: An imbalance of positive and negative regu-lation. Cell 1991:63:327–336.

23. DeLisser HM, Chilkotowsky J, Yan HC, Daise ML, Buck CA,Albelda SM: Deletions in the cytoplasmic domain of platelet-endothelial cell adhesion molecule-1 (PECAM-1, CD31) result inchanges in ligand binding properties. J Cell Biol 1994;124:195–203.

24. Knox JD, Cress AE, Clark V, Manriquez L, Affinito K-S, DalkinBL, Nagle RB: Differential expression of extracellular matrixmolecules and the a-integrins in the normal and neoplastic pros-tate. Am J Pathol 1994;145:167–174.

254 Knox et al.