(CANCER RESEARCH 53. 1409-1415. March 15. 1993] Identification and Characterization of a Novel Matrix-degrading from Hormone-dependent Human Breast Cancer Cells

Yuenian E. Shi,1 Jeff Torri, Lynn Yieh, Anton Wellstein, Marc E. Lippman, and Robert B. Dickson

Vincent T. Lombardi Cancer Rese arch Center, Georgetown University Medical Center. Washington. D.C. 20007

ABSTRACT inactive, undipped are generally secreted as a complex with a TIMP (6, 19). activity depends upon secreted ratios of A novel matrix-degrading enzyme was identified from human breast /TIMP and activational process. Measurements of the 72- cancer cells. This enzyme appears as major gelatinase in hormone-depen kDa collagenase itself in breast cancer are developing as useful prog dent breast cancer cell lines and has as an apparent molecular mass of 80 nostic indicators (13, 19-22). kDa on gelatin zymography. The 80-kDa enzyme has a unique metal ion Fully metastatic models of hormone-responsive breast cancer have specificity. In addition to calcium ions, the gelatinolytic activity can be supported by manganese and/or magnesium. Unlike 92- and 72-kDa ge- only recently been described, and some progress has been made in in latinases and other known members of the family, the vitro invasion systems to evaluate regulatory mechanisms (23, 24). 80-kDa protease is not activated by p-aminophenylmercuric acetate and The reconstituted basement membrane extract, Matrigel, has been its gelatinolytic activity is not inhibited by tissue inhibitor of metallopro utilized in assessing invasive potential of cancer cells (25). Tumor cell teinase 2. It is active over the pH range 7.5-9.5 with an optimum at pH 8.5. invasiveness into a Matrigel barrier in a Boyden chamber is thought to The enzyme degrades gelatin and type IV collagen. The proteolytic activity depend on a proteolytic cascade involving collagenases and possible of the enzyme is inhibited by EDTA and leupeptin. These unique features other and is also subjected to cell motility (25, 26). Invasion clearly distinguish the 80-kDa protease from the known 92-and 72-kDa of hormone-dependent breast cancer cells in vitro is stimulated by gelatinases. The expression of 80-kDa enzyme can be detected in hormone- estrogen or tamoxifen (a weakly estrogenic, nonsteroidal antiestrogen) dependent human breast cancer cell lines in vitro and in tumors grown from these cells in athymic nude mice. (23) but not by the steroidal pure antiestrogen ICI 164,384 (24). The pure antiestrogen was capable of blockade of both estrogen- or tamox- ifen-induced invasion suggesting a role for the estrogen receptor in INTRODUCTION mediating these effects. We now report the identification, partial pu rification, and characterization of a 80-kDa matrix metalloprotease- Métastasesof epithelial tumors, such as breast cancer, occur when like enzyme from hormone-dependent human breast cancer cells. Our the epithelial cells are no longer contained by the basement membrane data indicate that the 80-kDa enzyme is a novel enzyme which rep boundary to the underlying stromal compartment. The process of resents a major component of gelatin-degrading proteases in hormone- metastasis does not appear to be governed by precisely the same mechanisms as tumor growth, although its progressive deregulation dependent human breast cancer cells. sometimes occurs in parallel with deregulated proliferation (1,2). It is this disconcordance which helps to make breast cancer an unpredict MATERIALS AND METHODS able disease; even very small malignant lesions at the limit of detec Reagents. The protease inhibitors phenylmethylsulfonyl fluoride, leupep tion by mammography or palpation can already be metastatic (1, 3). tin, and pepstatin were obtained from Boehringer Mannhem. Indianapolis, IN. Tumor cell invasion is thought to critically depend upon proteolytic EDTA. benzamidine, dithiothreitol, type 1 collagen, and iodoacetamide were events (4, 5). The proteases are a large family, grouped into four main from Sigma, St. Louis, MO. Type I 'H-collagen was obtained from Dupont, classes: serine-proteases, such as plasminogen activators, plasmin, Wilmington, DE. Recombinant 72-kDa type IV collagenase, recombinunt and elastase; cysteine proteases, principally cathepsins B and L; met- TIMP-2, and type IV 'H-collagen were kindly provided by Dr. Roben Bird. alloproteinases, which comprise interstitial or type I collagenases, Molecular Oncology. Inc., Gaithersburg, MD. Transferrin, laminin. and fi- type IV collagenases or gelatinases, and stromelysins; and aspartyl bronectin were purchased from Collaborative Research; gelatin-Sepharose and proteinases, such as cathepsin D (6). Although from all four chelate-Sepharose were from Pharmacia, Piscataway, NJ. protease classes have been implicated in the process of cancer inva Cell Lines and Culture Conditions. T47Dco cells were kindly provided sion and metastasis (6-11 ), the proteolytic processes of metastasis are by Dr. Dean Edwards (University of Colorado). MCF-7 cells were obtained from Dr. Marvin Rich (Michigan Cancer Foundation), and MDA-MB-435 was thought to initially depend upon type IV collagenases which degrade kindly supplied by Dr. Janet Price, The University of Texas M. D. Anderson type IV collagen, the major structural component of basement mem Cancer Center, Houston, TX. All other breast cancer cell lines were obtained brane and, presumably, one of the first barriers to métastases(1, 12, from the American Type Culture Collection (Rockville. MD). All cell lines 13). At present, 72- and 92-kDa type IV collagenases are both known were maintained in Costar T75 flasks with Richters IMEM (Biofluides, Rock to exist in breast cancer (6, 14, 15). Each is secreted in an inactive ville. MD) supplemented with 10% fetal calf serum (GIBCO. New York, NY). form and requires an activational cleavage yielding 68- and 86-kDa Preparation of Conditioned Medium and Plasma Membranes. In order active enzymes, respectively. //; vivo, activational cleavage is thought to remove the soluble gelatinases present in the serum, experiments were to be either autocatalytic or dependent upon other proteases in the performed with subconfluent monolayers obtained by culturing the cells for 3 tumor environment (16, 17). Two natural inhibitors of these enzymes days in IMEM supplemented with 10% fetal calf serum that was previously have been characterized: TIMP-I2 and TIMP-2 (14, 15-17). The exposed to gelatin affinity chromatography. The medium was discarded and the monolayers were washed twice with phosphate-buffered saline. The monolay ers were cultured in the absence of serum, in IMEM supplemented with transferrin ( I mg/liter), fibronectin ( 1 mg/liter), and trace elements (Biofluides, Received 9/9/92; accepted 1/6/93. Rockville. MD). After 24 h, the serum-free medium was discarded, and the The costs of publication of this article were defrayed in part by the payment of page cells were replenished with the fresh serum-free medium and cultured for charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this faci. another 48 h. At the end of this period, the conditioned medium was collected. 1To whom requests for reprints should be addressed. The medium was then centrifuged at 1200 x g. and supernatants were saved 2 The abbreviations used are: TIMP. tissue inhibitor of metalloproteinase; Con A, and concentrated 20-fold by Ultrafiltration using Centripreps (Amicon Divi concanavalin A; PMSF. phenylmethylsulfonyl fluoride; SDS. sodium dodecyl sulfate; sion, molecular weight cutoff. 10,000) at 4°C.Isolation of plasma membrane PAGE, polyacrylamide gel electrophoresis; APMA. /j-aininophenylmercuric acetate; IMEM. improved minimum essential medium; CM. conditioned medium; MMP. matrix from tumors was performed as previously described (27). Briefly, the tumor metalloproteinase(s). samples were homogenized with cold phosphate-buttered saline containing 1 1409 Downloaded from cancerres.aacrjournals.org on October 1, 2021. © 1993 American Association for Cancer Research. IDENTIFICATION Ol A NOVhl MAl'KIX DICK \DISC, I'ROIIASI

IHMPMSFand l IHMben/.amidine. After centrifugation al low speed (800 x g) Inhibition Studies. All the inhibitors were evaluated in gelatin substrate for 15 min. the supernatant was further centrifuged at HX).(XX)x g for I h. The /ymography, and TIMP-2 was also subjected to the soluble gelatin degradation membrane pellet was extracted overnight at 4°Cin a 50 imi Tris (pH 7.4) assay. Aliquots of partially purified 80-kDa protease and 72-kDa gelatinase huiler containing 2(X) IHMNaCl. l IHMPMSF, l IHMbenzamidine, and 2<7c were subjected to gelatin zymography. In order to allow the inhibitors to Triton X-100. sufficiently diffuse into the gel following electrophoresis, the gels were washed (ielatin-Sepharose Chromatograph}. The 80-kDa protease from T47Dco 3 times (10 min/wash) with cold 2.5% Triton X-100 buffer and sliced into individual lanes, and each slice was incubated with protease inhibitor at 4°C cells was first separated from other minor gelatinases by virtue of its lack of affinity for gelatin-Sepharose chromatography. This was accomplished by for 2 h. Under these conditions, the inhibitors gradually diffuse into the gel slice while the enzymes remain inactive. After overnight incubation at 37°C, passing the concentrated conditioned medium over a column ( 1.5 x 4 cm) of gelatin-Sepharose equilibrated with collagenase buffer (buffer C) containing inhibition of gelatin-degrading activities of the proteases was evaluated by the 50 imi Tris (pH 7.5). 0.15 MNaCl. 5 IHMCa2*, and 0.05% Brij-35. When the loss of the zone of degradation seen on zymography compared to that of the column flow-through sample was tested on gelatin /ymography. no detectable untreated sample. 92- kDa and 72-kDa gelatinases were observed. Therefore, this gelatin- Tumor (¡rowth in Athvmic Nude Mice. MCF-7 cells were maintained in Sepharose flow-through sample was considered as the "starting material" for IMEM containing 10% fetal calf serum in T-175 flasks and were harvested by scraping. Cells were resuspended in growth medium at a concentration of 2.7 subsequent purification. X 10" cells/ml and 0.15 ml (4 X IO6 cells/injection) of cell suspension was Con A-Sepharose. The concentrated conditioned medium depleted of 92- and 72-kDa gelatinases was first applied to a Con A column ( 1.5 X 4.5 cm) injected s.c. into ovariectomized athymic nude mice (Charles River Breeding Laboratory. Inc., Wilmington, MA) treated with 17ß-estradiolpellet (0.7 ing/ equilibrated with buffer C. The column was washed with three column vol pellet). Each mouse received two injections of tumor cells, with one on each umes of this buffer. Adsorbed proteins were eluted with 2 column volumes of side. Tumor size was determined at intervals by measurement of two right- buffer C containing O.I M methyl mannopyranoside (Sigma) and 3 column angle diameters with a caliper. volumes of buffer C containing 0.5 Mmethyl mannosylpyranoside. Data Analysis and Reproducibility. Except where indicated, the results of /.ine Chelation Sepharose Chromatography. The zinc chelation each experiment discussed are representative of the results obtained from Sepharose column was prepared by passing 20 ml of 30 IHMZnCl2 through a several replicate experiments using 80-kDa proteases isolated from different column ( 1.5 x 5 cm) of chelate-Sepharose. The active fractions from the above experiments. Con A column were pooled, dialy/ed against buffer C without Ca2 ', and passed over the column equilibrated with buffer C without Ca: *. The column was eluted as previously described (28). RESULTS Substrate Zymograms. Preparation of substrate gels and subsequent de Identification of 80-kDa Protease as a Major Gelatinase from tection of proleolytic activity were conducted as previously described (29). Hormone-dependent Human Breast Cancer Cells. Proteolytic en Protease substrates utili/.ed were gelatin (1 mg/ml). Preparation of substrate zymes that are capable of degrading gelatin were first identified in gels was carried out in Laemmli buffer system except that the amount of ammonium sulfate was doubled. Proteins were first separated by eleclrophore- serum-free conditioned medium of parental T47D human breast can sis on an 10% polyacrylamide gel containing different substrates without cer cells and the subline, T47Dco cells. The T47D cells are weakly reducing agent. The electrophoresis was performed at 4°Cat a constant current tumorigenic and poorly invasive; T47Dco cells are highly tumorigenic of 40 mA. The gel was washed 3 times for 20 min in 50 mvi Tris buffer (pH and more invasive. The gelatinolytic activity from CM was analyzed 7.5) containing 5 IHMCa2*, 150 IHMNaCl, and 2.5% Triton X-100. After by zymography on a gelatin-containing SDS-PAGE (Fig. 1). When the overnight incubation at 37°C.the gels were stained with 0.25% Coomussie gelatin gel was incubated in a reaction mixture containing 5 ITIMCa2+, blue in 5% acetic acid and 30% methanol for l h and destained in the same a prominent negative staining complex (double bands) with apparent buffer. Proteolytic activities were detected by clear /ones indicating the lysis of molecular mass of 80 kDa was present in the conditioned medium. the substrate. The gelatinolytic activity in the CM of T47Dco cells was significantly Kn/\ malic Degradation Assay. Enzymatic activity was assayed by mea higher than that of T47D cells suggesting a potential role of this suring degradation of 'H-gelatin in a manner conceptually similar to that of 80-kDa gelatinase in the progression of breast cancer. Steiler-Stevenson (30) with some minor but critical modifications. Briefly, Interestingly, the 80-kDa enzyme may be preferentially expressed tritiated type 1collagen (NEN) was diluted with nonradioactive type 1collagen as the predominant gelatinase in estrogen receptor-positive human and then denatured at 55"C for IO min. Enzyme samples and gelatin were breast cancer cell lines. As demonstrated in Fig. 2, the expression of added lo buffer C to a final volume of 55 ul. After overnight incubation at 37°C.the reaction was stopped by the addition of EDTA at a final concentra 80-kDa gelatinase was observed in T47D, MCF-7, ZR-75-1, and BT474 hormone-dependent breast cancer cells but not in MDA-MB- tion of 18 IHM.Proteins were precipitated on ice with 0.3% of trichloroacetic 231, MDA-MB-435, MCF-7adrR, MDA-MB-436. and BT549 hor acid and ().(X)14% of tannic acid. The radioactive soluble peptide fragments were detected in a liquid scintillation counter. The critical modification of the mone-independent breast cancer cells. The 80-kDa enzyme had no published procedure (30) for detection of 80-kDa enzymatic activity in this apparent gelatinolytic activity in a reaction mixture containing EDTA soluble assay was the very low concentration of trichloroacetic acid in the final (Fig. 3), a characteristic of . precipitation step. When the final concentration of trichloroacetic acid is in creased to 1%, the apparent enzymatic activity of the 80-kDa protease (deter 1 mined by soluble counts) decreased dramatically. This may be due to the proteolytic generation of relatively large gelatin fragments which can be pre cipitated with higher concentration of Irichloroacetic acid, in contrast to other collagenases. for which the 1% trichloroacetic acid assay was originally de scribed (30). The enzymatic activity of 80-kDa against type IV collagen was also eval uated in soluble reaction mixture containing type IV "H-collagen and 80-kDa sample in the buffer C. After overnight incubation at 37°C.degradation of the type IV collagen was analyzed by SDS-PAGE followed by autoradiography. Fig. I. Gelatin zymogruphy of proteases secreted by T47D and T47Dco cells. Conflu ent cultures of T47D and T47Dco cells were washed three times with phosphate-buffered Determination of Optimal pH Range. The optimal pH range was deter saline and incubated in serum-free IMEM for 2 days. The resultant conditioned medium mined by incubating slices of gel (each slice corresponding to one lane) in was concentrated 30-fold by ullrafiltration, normali/ed by cell number, and subjected to Hanks' balanced salt solution adjusted to pH values of 5, 6, 7, 8, 9, and 10. gelatin zyrnogruphy. After overnight incubation at pH 7.5, proteolytic activities were After an overnight incubation in separate dishes at 37°Cwith shaking, the pH visualized following Coomassie blue staining. ¡MtieI, 92- and 72-kDa type IV collage values of the Hanks' balanced salt solution were measured again to ensure that nases from combined conditioned medium of HT 1080 and Hs578T cells; laines 2-5, conditioned medium from T47D cells; tones 6-9. conditioned medium from T47Dco cells the pH values had not been changed. (different preparations). 1410 Downloaded from cancerres.aacrjournals.org on October 1, 2021. © 1993 American Association for Cancer Research. IDENTIUCATION OF A NOVEL MATRIX DEGRADING PROTEASE

Isolation of the 80-kDa Gelatinolytic Enzyme. To isolate and characterize the 80-kDa gelatinolytic enzyme, 2.5 liters of the serum- free CM of T47Dco cells were concentrated l(X)-fold by ultrafiltra- tion. The concentrated CM was first applied to gelatin-Sepharose. Unexpectedly, the 80-kDa gelatinolytic activity was detected in the flow-through fractions, unlike the 92- and 72-kDa gelatinases, which are absorbed by gelatin-Sepharose and can be eluted with Me2SO. When gelatin-Sepharose T47Dco conditioned medium was subjected to standard elution with Me2SO, 92- and 72-kDa gelatinase bands were observed in gelatin /ymography (data not shown). These results Fig. 2. Gelatin zymographic analysis of secreted proteases from different breast tumor suggest that in addition to 80-kDa as a dominant gelatinase, T47Dco cell lines. Conditioned medium was concentrated 20-fold and subjected to gelatin zymog- raphy. The proteolytic activities were detected following overnight incubation at pH 8.5. cells also express low levels of 92- and 72-kDa gelatinases. Therefore we currently use gelatin-Sepharose as a first purification step to sep Ca Mg EDTA arate 80-kDa gelatinase from other minor gelatinases by virtue of its lack of affinity for gelatin-Sepharose. Subsequent purifications were conducted by using Con A and Zn2f chelation chromatography. The specific activity of the recovery of enzyme activity against gelatin is summarized in Table 1. There was a decrease in the total enzymatic activity after gelatin affinity chromatography which indicates the re moval of some gelatin-binding gelatinases. After Zn2 ' chelation chro matography, we achieved an approximate 85-fold increase in specific activity. Catalytic Properties of 80-kDa Gelatinase: Unique Metal Ion Specificity. The 92- and 72-kDa gelatinases are well studied metal- loproteinases which are widely expressed in human breast cancer cells. The characterization of 80-kDa gelatinase was carried out in Ca Mg Mn comparison with 92- and 72-kDa gelatinases. Mammalian metallo- Ca EDTA EDTA EDTA EDTA proteinases generally possess consensus metal ion-, termed the "zinc-binding site." Operationally, calcium, as well as zinc. is usually used to study the proteolytic activity of enzymes. While the 80-kDa enzyme can also utilize 5 ITIMCa2' to yield maximal activity, closer examination (Fig. 3/4) demonstrated a unique metal ion spec ificity. EDTA. at 5 mM. inhibited the activity of all three gelatinases. In contrast to 92- and 72-kDa gelatinases which were able only to utilize calcium, the 80-kDa gelatinase showed a broader range of divalent cation dependence. In addition to calcium ions, the gelati nolytic activity of 80-kDa protease can be supported by manganese

B and/or magnesium ions. Fig. 3. Metal ¡onspecificity of gelatinolytic activity of 80-kDa protease. KO-kDa protease partially purified by Zn*- chelation chromatography from T47Dco cells and To further confirm that the blocking effect of EDTA on the gelat inolytic activity of 80-kDa gelatinase is due specifically to its metal- combined conditioned medium from HTI080 (92 kDa) and Hs578T (72 kDa) were subjected to gelatin /ymography analysis for metal specificity. A. metal ion dependency chelating activity, a reversibility assay was carried out. As demon for gelatinolytic activity. Slices of gelatin gels containing 92-, 72-, and 80-kDa gelatinases were prepared and incubated with Hanks' balanced salt solution at pH 7.4 with 5 m.M strated in Fig. 3fi, as little as 0.5 ITIMEDTA totally blocked the EDTA or 5 mw of different metal ions. EDTA inhibited all three enzymes. The gelati gelatinolytic activity of all three gelatinases. This inhibitory effect of nolytic activity of the 92- and 72-kDa proteases was observed only in the incubation buffer EDTA on the gelatinolytic activity of 80-kDa gelatinase was com containing calcium, but the 80-kDa protease showed activity in the preparations of either calcium, manganese, or magnesium. B. reversal of EDTA-induced inhibition of gelati pletely reversed by 5 mM of calcium, magnesium, and manganese. nolytic activity by different metal ions. Prior to incubation, slices of gelatin gels contain Unexpectedly, under the same conditions, these metal ions failed to ing 92-, 72-, and 80-kDa gelatinases were prepared and preincubated with 0.5 mM EDTA restore the gelatinolytic activity of 92- and 72-kDa gelatinases. One in the incubation buffer for 30 min before 5 niM addition of different metal ions. All the metal ions tested restored the gelatinolytic activity of 80-kDa protease but failed to restore likely possibility for this failure is that besides calcium, zinc is also the gelatinolytic activity of 92- and 72-kDa proteases. required for enzymatic activity of 92- and 72-kDa gelatinases.

Table I Partial purification of XU-kDiiprotease The activity of the enzyme was determined in the gelatin degradation assay as described in "Materials and Methods." Protein present in the different steps of purification was estimated from the absorbance at 280 nm using bovine serum albumin as a standard. Each point represents the value for a single measurement. Similar results were detected in two separate experiments.

PurificationstepConcentrated (mg)84793.90.136Totalactivity(units")11,2341,522412221Specificactivity'1(units/mg)133191051,625Purification(-fold)15.585Recovery(%)KKC2716 mediumGelatin-Sepharose flow-throughCon A-SepharoseZn2* chelationTotalprotein.Aim " One unit activity is defined as KXX)dpm of soluble gelatin (total soluble count subtracted with background count) in 20 ul of pooled active fraction. h Specific activity was calculated by dividing total activity by total protein. '" 92- and 72-kDa collagenases and other gelatinase were removed by gelatin-Sepharose; thus the activity of the gelatin-Sepharose flow-through sample represents the 80-kDa enzyme. 1411 Downloaded from cancerres.aacrjournals.org on October 1, 2021. © 1993 American Association for Cancer Research. IDENTIFICATION OF A NOVEL MATRIX DEGRADING PROTEASE

Optimal pH for Gelatinolytic Activity of 80-kDa Protease. Since the 92- and 72-kDa gelatinases are neutral metalloproteinases, we determined the optimal pH of 80-kDa gelatinase compared to 92- and 72-kDa gelatinases. As shown in Fig. 4, the 80-kDa gelatinase from T47Dco cells was active in the range from pH 7.5 to pH 9.5, with its optimal activity around pH 8.5-9. In contrast, the optimal pH for 92- and 72-kDa gelatinases was 7.5, and gelatinolytic activity of 92- and 72-kDa gelatinase was dramatically diminished at pH 8.5-9. None of the gelatinases had acidic pH optima, suggesting that they were not lysosomal enzymes released from dying cells. Stability. The enzymatic activity of the 80-kDa protease was ac tive when the samples were incubated at room temperature or at 55°C for 25 min. The activity was lost dramatically after heating the 80-kDa enzyme at 100°Cfor 5 min. No activity of 80-kDa enzyme was observed after heating at 100°Cfor 15 min (data not shown). Effect of AI'MA on the Activation of 80-kDa Gelatinase. Mam malian metalloproteinases are usually secreted as latent proenzymes (zymogen) and require activation for their enzymatic activity. In vitro, the zymogen forms of metalloproteinases can usually be activated by proteolytic cleavage induced either by serine proteases such as trypsin and plasmin or by autocatalytic cleavage induced by organomercurial 20000 compounds such as APMA. To test the effects of such an activator, the 80-kDa enzyme was incubated in the presence or absence of 1 mm APMA at 37°Cfor 30 min. The gelatinolytic activity was measured both by gelatin substrate zymography and by soluble gelatin degra dation assay. As shown in Fig. 5/4, incubation of 92- and 72-kDa gelatinases with APMA resulted in conversion of high molecular mass of proenzymes to low molecular mass enzyme, as determined by zymography. The intensities of the zymographic bands of the 72- and 92-kDa enzymes were not significantly affected by the presence of APMA, as expected, since the latent proenzyme is known to be activated by a conformational change induced during SDS-PAGE. No such molecular mass conversion was observed in the 80-kDa enzyme, indicating that the 80-kDa enzyme was not responsive to APMA treatment. This lack of modification of 80-kDa gelatinase by APMA was further confirmed by a soluble gelatin degradation assay. The results in Fig. 5ßconfirm that the gelatinolytic activity of the 80-kDa gelatinase was not changed after incubation with APMA. As a positive control, the gelatinolytic activity of purified recombinant 72-kDa en B Fig. 5. Analysis of effect of APMA on activation of 80-kDa gelatinase. Partially zyme was dramatically increased upon APMA treatment. purified 80-kDa protease and combined conditioned medium containing 92- and 72-kDa In order to rule out the possibility that the activation of the 80-kDa proteases (see Fig. 3) were incubated with or without 1 mv APMA at 37°Cfor 30 min. enzyme by APMA may take place during the longer time period of Samples were then subjected to gelatin zymography (A ) for analysis of possible auto- cleavage and soluble gelatin degradation assay (ß|for determining gelatinolytic activity. incubation, the enzyme was incubated with 1 miviAPMA overnight at A. Lane 1. control 92- and 72-kDa gelatinases; Lane 2, APMA-treated 92- and 72-kDa gelatinases; Lane 3, control 80-kDa gelatinase; Lane 4. APMA-treated 80-kDa gelatinase. room temperature. Neither change of gelatinolytic activity in soluble In B. soluble gelatin degradation assay was conducted as described in "Materials and Methods." Bars, means ±SD of triplicate measurements.

gelatin degradation assay nor induced autocleavage as determined in gelatin zymography of the 80-kDa enzyme was observed (data not shown). Substrate Specificity of 80-kDa Protease. We next investigated the ability of the 80-kDa protease to degrade substrates other than gelatin. The substrate specificity of the 80-kDa protease when tested against the type IV collagen (native type IV collagen purified from the Engelbreth-Holm-Swarm tumor) was examined in the soluble assay. As shown in Fig. 6, when the reaction mixtures were incubated at 37°Covernight, the a-chain of type IV collagen was completely digested by both recombinated 72-kDa type IV collagenase and 80-

Fig. 4. pH dependence of the 80-kDa gelatinase. Zn2* chelation partially purified kDa protease; no specific degradation fragment pattern was observed, 80-kDa protease and combined conditioned medium containing 92- and 72-kDa gelati presumably due to the excess of enzymes over the substrate. When the nases (see Fig. 3) was subjected to gelatin zymography. After electrophoresis, slices of 80-kDa sample was diluted 20-fold, partial digestion was observed. gelatin gel containing the 92-. 72-, and 80-kDa proteases were incubated in Hanks' balanced salt solution containing 5 mvi calcium at the indicated pH. The 80-kDa gelatinase Inhibitors for 80-kDa Protease. We examined the effects of var was active in the range from pH 7.5 to pH 9.5. with its optimal pH at 8.5. ious protease inhibitors specific for serine, aspartate, cysteine. and 1412 Downloaded from cancerres.aacrjournals.org on October 1, 2021. © 1993 American Association for Cancer Research. IDENTIFICATION OF A NOVEL MATRIX DEGRADING PROTEASE

DISCUSSION

A general aspect of malignant neoplasia is the capacity to invade and degrade tissue barriers, such as basement membrane, by enhanced proteolysis. Evidence has accumulated that important proteases in the metastatic process include metalloproteinases, serine, cysteine, and aspartyl proteases ( 11). Of these, most of the evidence to date in early stages of breast cancer metastasis implicates metalloproteinases; both

Fig. 6. Degradation of type IV coligan by 80-kDa gclatinasc. Partially purified 80-kDa was subjected to type IV 'H-collagen soluble assay as described in "Materials and Methods." After incubation of type IV 'H-collagen with proteases, 20 pi of reaction mixture were subjected to SDS-PAGE followed by autoradiography. Lane I, nontreated type IV collage; Lane 2. incubated with 35 ng of recomhinant APMA-activaled 72-kDa protease; Lane 3, incubated with partially purified SO-kDa protease; Lane 4. incubated with 20-fold diluted 80-kDa protease.

metalloproteinases on gelatin-degrading activity of 80-kDa enzyme (Fig. 7). Although the activity of the 80-kDa gelatinase was totally inhibited by the metal chelator EDTA. the specific metalloproteinase inhibitors TIMP-2 and phosphoramidon failed to inhibit the activity of 80-kDa enzyme (Fig. 1A). The inability of TIMP-2 to inhibit the 80-kDa gelatinolytic activity was further confirmed in the soluble gelatin degradation assay. As demonstrated in Fig. IB, although the gelatinolytic activity of 72-kDa enzyme was completely inhibited by TIMP-2 at the molar ratio of 1 to 3.5, the gelatinolytic activity of 80-kDa enzyme was not inhibited by TIMP-2 at the concentrations tested. These results did not rule out the possibility that the failure of 20000 TIMP-2 to inhibit 80-kDa protease is due to an insufficient molar ratio of inhibitor to enzyme, because we did not know the molar ratio between the TIMP-2 and the 80-kDa protease in our test. However, this possibility is unlikely based on the fact that no TIMP-2-mediated I T concentration-dependent inhibition of the 80-kDa enzyme was ob served within the range of 10-fold of the tested concentrations of T TIMP-2. Inhibitors of the aspartate, serine, and cysteine classes of 8 10000- protease, including pepstatin, benzamidine, PMSF, and iodoacetamide had no discernable effect on the activity of the 80-kDa enzyme. The proteolytic activity of the 80-kDa enzyme was completely blocked by the cysteine protease inhibitor leupeptin and the reducing agent dithio- threitol. Although these inhibition data do not directly demonstrate to which protease family this 80-kDa enzyme belongs, the inability of TIMP-2 to inhibit the 80-kDa protease together with the data on the inability of APMA to activate the enzyme suggest that this metal- 8 § dependent, matrix-degrading protease does not belong to the previ D- S Sí CL ously defined family. Detection of 80-kDa Protease in Breast Tumors. As a prelimi nary study to develop the methodology for evaluation of levels of the B 80-kDa enzyme in primary human breast tumor tissues, we have Fig. 7. Effects of various protease inhibitors on gelatinolytic activity of 80-kDa pro carried out studies of its expression in MCF-7 cells grown as tumors tease. A, gelatin zymographic analysis of protease inhibitors specific for serine, cysteine, asparatic, and metalloproteinases. After electrophoresis. the 80-kDa protease was sub in nude mice. Tumor membrane fractions were prepared and subjected jected to the incubations with different protease inhibitors as described in "Materials and to gelatin zymography analysis. As shown in Fig. 8, the 80-kDa Methods." The inhibitors used are: dithiothreitol, 2 HIM;leupeptin, O.I HIM;phosphora enzyme was clearly visible in detergent-solubilized plasma membrane midon. UK)ug/ml; iodoacetamide. 10 HIM:TIMP-2. 8 ug/ml; ETDA, 5 HIM;PMSF, 1 HIM; benzamidine. 2 HIM;pepstatin. O.I HIM.R, analysis of effects of TIMP-2 on gelatìnolytic fraction. The lack of doublet of the MCF-7 tumor derived 80-kDa activities of 72- and 80-kDa proteases in soluble gelatin degradation assay. After titration. the submaximal dose of 80-kDa protease and 20 ng of recombinant APMA-activated protease may be due to the degradation of the lower and usually 72-kDa gelatinase were incubated with or without TIMP-2 at various amounts at 37°C weaker gelatinase band during the preparation of tumor membrane overnight. The gelatinolytic activity was measured as described in "Materials and Meth fractions. ods." Bars, means ±SD of triplicate measurements. 1413 Downloaded from cancerres.aacrjournals.org on October 1, 2021. © 1993 American Association for Cancer Research. IDENTIFICATION OF A NOVEL MATRIX DEGRADING PROTEASE

pattern of protease inhibitors when compared to known members of the metalloproteinase family. These results clearly demonstrate that although the 80-kDa enzyme is a metal-dependent matrix-degrading

<0 protease, it may not belong to the previously defined MMP family. 3 The 80-kDa protease has been shown to degrade gelatin and type IV collagen. In addition, by using substrate zymography, our preliminary results suggest that the other major structural components of basement membrane such as fibronectin and laminin may also be the substrates for this novel protease (data not shown). This broad range of pro teolytic activity on multiple basement membrane components could indicate the potential of the 80-kDa enzyme to contribute to basement membrane dissolution and metastasis. In addition, since 72- and 92- kDa type IV collagenases and plasminogen are secreted in latent form, the 80-kDa enzyme also has the potential to act as a master switch to activate these zymogens by a proteolytic clip. Such an action could contribute to a proteolytic cascade, widely speculated to occur in

Fig. 8. Detection of 80-kDa protease in MCF-7 tumors grown in nude mice. Plasma métastases(35, 37). membranes of MCF-7 tumors were prepared as described in "Materials and Methods." Unlike the most metalloproteases in which the optimal enzymatic After extraction of membranes with 2% Triton X-100 overnight, the gelatinolytic activity activity occur at neutral pH, the 80-kDa protease seems to prefer a of the plasma membrane extract was analyzed in gelatin zymography at the conditions of pH 8.5. Partially purified 80-kDa protease was used as positive control. slightly alkaline pH for its proteolytic activity in vitro. However, this iw vitro alkaline optimal pH for the 80-kDa enzyme may not be relevance to the situation in vivo because (a) the pH range for pro 72- and 92-kDa species have been described in breast cancer. In the teolytic activity of the 80-kDa enzyme is characterized with gelatin present work we have isolated and characterized an apparently novel (denatured collagen) which may not be a physiological substrate for matrix degrading protease from hormone-dependent breast cancer the enzyme in vivo and (h) the optimal pH is determined in gelatin cells. The novelty of this 80-kDa protease is based on four main lines zymography in which the substrate is partially denatured by detergent of evidence: (a) its metal ion specificity, since the gelatinolytic ac SDS. In addition, at present, it is not know whether the high pH tivity of 80-kDa protease is not exclusively calcium dependent but can environments could occur in some necrotic areas of tumors, as DNA is released, which may facilitate the proteolytic activity of the 80-kDa also function in the presence of manganese and magnesium; (b) its basic optimal pH; (r) its inability to be activated by APMA; and (d) enzyme. its inability to be inhibited by TIMP-2. To our knowledge, 80-kDa Recently, a family of membrane-bound metalloproteinases termed "" has been proposed (38). Although the members in this new protease is the only extracellular matrix-degrading enzyme which family are Zn2 '-dependent metalloproteinases, they are catalytically shows a specific metal dependency on manganese and magnesium. This unique metal ion specificity suggests a distinction for the 80-kDa different from previously described matrix-degrading metalloprotein ases in which their enzymatic activities are not inhibited by phos- protease from other members in the metalloproteinase family. All members of the matrix metalloproteinase family are secreted as phoramidon and TIMP. Most importantly, there is no similarity be inactive zymogens and require an extracellular activation. Recent tween the primary sequences of the astacin family and the family of models for this proenzyme activation emphasize the importance of the MMP. The 80-kDa protease shares some physical and catalytical interaction between a sulfhydryl side chain and the active zinc atom properties with this new family of metalloproteinases, including its that results in a catalytically inert active center (17, 31). The sulfhy membrane-association, basic pH optima, inability to be activated by dryl group in this interaction is donated by the cysteine 73 within a APMA. and inability to be inhibited by phosphoramidon and TIMP-2. highly conserved sequence PRCGVPDV located immediately adja The relevance of this 80-kDa protease to human breast tumor in cent to the proenzyme cleavage site (1, 32, 33). Disruption of this vasion and metastasis remains to be determined. Our preliminary interaction by proteolytic or autocatalytic removal of the cysteine 73 results demonstrate that the expression of 80-kDa protease is elevated containing sequence results in conformational rearrangement and in the more aggressive T47Dco subline. In addition, the 80-kDa pro rapid attainment of protease activity ( 1). So far we have demonstrated tease may be preferentially expressed in hormone receptor-positive two striking features of 80-kDa enzyme which are inconsistent with human breast cancer cells, since the 80-kDa protease was the major metalloproteinase family. First, the 80-kDa protease appears to be gelatinolytic activity in T47D. MCF-7. ZR-75-1. and BT474 hor secreted as an active form, since APMA treatment of 80-kDa protease mone-dependent breast cancer cells but was absent in MDA-MB-231 neither altered gelatinolytic activity in soluble gelatin degradation and other hormone-independent breast cancer cells. The invasiveness assay nor induced autocleavage as determined in gelatin zymography. of steroid receptor-positive breast cancer cells can be influenced by However, the inability of APMA to activate the 80-kDa enzyme does steroid hormones (23, 24). Tumor cell invasiveness into a Matrigel not exclude the possibility that the 80-kDa enzyme may undergo some barrier was shown to be dependent on a proteolytic cascade involving activation cascade through presently undefined mechanisms. The pos collagenases and other enzymes (25, 39). Since the 80-kDa enzyme sibility that the 80-kDa enzyme was activated during isolation and was expressed as dominant matrix-degrading protease in hormone- storage was unlikely, because the zymographic appearance of the dependent cells, the possibility that the 80-kDa protease may play a 80-kDa complex during isolation was similar and no molecular shift role in steroid-increased invasiveness of hormone-dependent breast of the 80-kDa enzyme was observed. Second, the gelatinolytic activity cancer cells is under investigation. Steroid hormones may either di of 80-kDa protease is not inhibited by TIMP-2. This result is striking rectly regulate 80-kDa protease expression or indirectly influence the in terms of the fact that the activity of all known members of the activation status of the 80-kDa protease to facilitate the increased metalloproteinase family can be inhibited by TIMP family members invasiveness observed in response to steroid treatment. It remains to (17, 34-36). The inability of TIMP-2 to inhibit the activity of 80-kDa be seen to what extent the 80-kDa protease contributes to the meta- protease indicates that the 80-kDa protease may respond to a distinct static progression of breast cancer. 1414 Downloaded from cancerres.aacrjournals.org on October 1, 2021. © 1993 American Association for Cancer Research. IDENTIFICATION OF A NOVEL MATRIX DEGRADING PROTEASE

ACKNOWLEDGMENTS 20. Ponton, A., Coulombe. B.. and Skup. D. Decreased expression of tissue inhibitor of metalloproteinases in metastatic tumor cells leading to increased levels of collagenase activity. Cancer Res.. 51: 2138-2143. 1991 We are most grateful to Dr. Robert Bird for helpful advice and for providing 21. Nakajima. M.. Welch. D.. Belloni. B. N.. and Nicholson, G. L. Degradation of recombinant 72-kDa type IV collagenase and TIMP-2. We wish to thank Drs. basement membrane type IV collagen and lung subendothelial matrix by rat mam William Stetler-Stevenson and Erik Thompson for critical discussion and re mary adenocarcinoma cell clones of differing metastatic protentials. Cancer Res.. 47: view of this work. We thank Dr. Rafi Fridman for excellent technical assis 4869-1876. 19X7. 22. Overall. C. M.. Wrana. J. R., and Tryggvason, K. Independent regulation of colla tance. genase. 72-kDa progelatinase, and metalloendoproteinase inhibitor expression in hu man fibrohlasls by transforming growth factor-0. J. Biol. Chem., 264: 1860-1869. REFERENCES 1989. 23. Thompson, E. W., Reich. R.. Snima, T. B., Albini. A., Graf. J., Martin, G. R.. Dickson, 1. Liotta. L. A., Steeg. P. S.. and Stetler-Stevenson, W. G. Cancer metastasis and R. B., and Lippman, M. E. Differential regulation of growth and invasiveness of angiogenesis: an imbalance of positive and negative regulation. Cell. 64: 327-336, MCF-7 breast cancer cells by antiestrogens. Cancer Res., 48: 6764-6768, 1988. 1991. 24. Thompson, E. W.. Katz, D., Shima, T B.. Wakeling, A.. Lippman. M. E., and Dickson. 2. Muschel. R. J.. and Liotta. L. A. Role of oncogenes in metastasis. Carcinogenesis R. B. ICI 164.384. a pure antagonist of estrogen-stimulated MCF-7 cell proliferation (Lond.), 9: 705-710, 1988. and invasiveness. Cancer Res.. 49: 6929-6934. 1989. 3. Poste. G., and Fidler. 1. The pathogenesis of cancer metastasis. Nature (Lond.), 283: 25. Albini. A., Iwamoto, Y., Kleinman. H. K., Martin, G. R., Aaronson. S. A.. Kozlowski. 139-146. 1980. J. M., and McEwan, R. N. A rapid in vitro assay for quantitating the invasive potential 4. 4. Liotta, L. A. Tumor invasion and metastasis-role of the extracellular matrix: Rhoads of tumor cells. Cancer Res., 47: 3239-3245, 1987. Memorial Award Lecture. Cancer Res., 46: 1-7, 1986. 26. Lester. B. R.. and McCarthy J. B. Tumor cell adhesion to the extracellular matrix and 5. Gottesman, M. The role of proteases in cancer. Semin. Cancer Biol., /: 97-KM). 1990. signal transduction mechanisms implicated in tumor cell motility, invasion and me 6. Goldberg. G. I., and Eisen. A. Z. Extracellular matrix metalloproleinase in tumor tastasis. Cancer Metastasis Rev.. //: 31—44,1991. invasion and metastasis. In: M. E. Lippman and R. B. Dickson (eds.). Regulatory 27. Shi, Y. E., and Yager, J. D. Regulation of rat hepalocyte epidermal growth factor Mechanisms in Breast Cancer, pp. 421—MO.Boston, MA: Kluwer Academic Pub receptor by the liver tumor promoter ethinyl estradiol. Carcinogenesis (Lond.). II: lishers. 1990. 1103-1109. 1990. 7. Kane. S. E.. and Gottesman. M. M. The role of cathepsin L in malignant transfor 28. Cawston. T. E.. and Tyler, J. A. Purification of pig synovial collagenase to high mation. Semin. Cancer Biol.. /. 127-136, 1990. specific activity. Biochem. J.. 183: 647-656. 1979. 8. Sloane, B. F. Cathepsin B and cystatins: evidence for a role in cancer progression. 29. Heussen, C.. and Dowdle. E. B. Electrophoretic analysis of plasminogen activation in Semin. Cancer Biol., /: 137-152. 1990. polyacrylamide gels containing sodium dodecyl sulfate and copolymerized substrates. 9. Rochetbrt. H.. Capony, F., and Garcia. M. Cathepsin D: a protease involved in breast Anal. Biochem.. 102: 196-202. 1980. cancer metastasis. Cancer Metastasis Rev., 9: 321-331, 1990. 30. Kleiner, D. E., Unsworth, E. J., Krutzsch, H. C., Stetler-Stevenson, W. G. Higher- 10. Lola, P. K., and Graham. C. H. Mechanisms of trophoblast in invasiveness and their order complex formation between the 72-kilodallon type IV collagenase and tissue control: the role of proteases and protease inhibitors. Cancer Metastasis Rev.. 9: inhibitor of metalloproteases-2. Biochemistry. 31: 1665-1672. 1992. 369-380, 1990. 31. Nagase. H.. Enghild. J. J.. Suzuki. K.. and Salveson. G. Stepwise activation mecha 11. Duffy. M. J. The role of proteolytic enzymes in cancer invasion and metastasis. Clin. nisms of the precursor of matrix metalloproteinase 3 istromelysin) by proteinases and Exp. Metastasis, 10: 145-155. 1992. (4-aminophenyl) mercuric acetate. Biochemistry, 29: 5783-5789, 1990. 12. Liotta, L. A.. Abe. S.. Robey. P. G.. and Martin, G. R. Preferential digestion of Vallee. B. L., and Auld. D. S. Zinc coordination, function and structure of zinc basement membrane collagen by an enzyme derived from a metastatic murine tumor. enzymes and other proteins. Biochemistry. 29: 5647-5659. 1990. Proc. Nati. Acad. Sci. USA, 76: 2268-2272, 1979. Van Wart. H. E., and Birkedal-Hansen, H. The cysteine switch: a principle of regu 13. Liotta. L. A.. Tryggvason, K.. Garbisa. S., Hart. I.. Foltz. C. M.. and Schaffie, S. lation of metalloproteinase activity with potential applicability to the entire matrix Metastasis potential correlates with enzymetic degradation of basement membrane gene family. Proc. Nati. Acad. Sci. USA. 87: 5578-5582. 1990. collagen. Nature (Lond.), 284: 67-68. 1980. 34. Stetler-Stevenson. W. G.. Brown, P. D.. Onisto, M., Levy, A., and Liotta, L. A. Tissue 14. Docherty, A. J. P.. and Murphy. G. The tissue metalloproteinase family and the inhibitor of metalloproteinase-2 (TIMP-2) mRNA expression in tumor cell lines and inhibitor T1MP: a study using cDNAs and recombinant proteins. Ann. Rheum. Dis., human tumor tissue. J. Biol. Chem.. 265: 13933-13938. 1990. 49: 469-479, 1990. 35. Suziki, K.. Enghild, J. J., Morodarni. T. Salvesen, G.. and Nagase, H. Mechanisms of 15. Frederlick, J., and Woessner. J. R. Matrix metalloproteinase and their inhibitors in activation of tissue procollagenase by matrix metalloproteinase 3 (stromelysin). Bio connective tissue remodeling. FASEB J., 5: 2145-2154, 1991. chemistry. 29: 10261-10270, 1990. 16. Stetler-Stevenson, W. G.. Krutzsch. H. C.. Wacher. M. P., Margulies, I. M. K., and 36. Goldberg. G. I.. Marmer. B. L., Grant. G. A.. Eisen. A. Z.. Wilhelm. S., and He, C. Liotta. L. A. The activation of human type IV collagenase proenzyme: sequence Human 72-kilodalton type IV collagenase forms a complex with a tissue inhibitor of identification of the major conversion product following organomercurial activation. metaiioproteinase designated TIMP-2. Proc. Nati. Acad. Sci. USA. 86: 8207-8211. J. Biol. Chem.. 264: 1353-1356. 1989. 1989. 17. Stetler-Stevenson, W. G.. Krulzsch, H. C.. and Liotta. L. A. Tissue inhibitor of 37. Brinrckerhoff. C. E.. Suzuki. K.. Mitchell. T. I., Dram. F., Coon, C. !.. Palmiter, R. D.. metalloproteinase (TIMP-21: a new member of the metalloproteinase inhibitor family. and Nagase. H. Rabbit procollagenase synthesized and secreted by a high-yield J. Biol. Chem.. 264: 17374-17378. 1989. mammalian expression vector requires stromelysin (matrix metalloproteinase-3) for 18. Wilhelm, S. M., Collier, I. E., Marmer, B. L., Eissen, A. Z., Grant, G. A., and maximal activation. J. Biol. Chem., 265: 22262-22269, 1990. Goldberg, G. 1. SV40-transtbrmed human lung fibroblasts secrete a 92-kDa type IV 38. Dumermuth. E.. Sterchi. E. E.. Jiang, W., Wolz, R. L.. Bound, J. S.. Flannery. A. V. collagenase which is identical to that secreted by normal human macrophages. J. Biol. and Beynon, R. J. The astacin family of . J. Biol. Chem.. 266: Chem.. 264: 17213-17221. 1989. 21381-21385, 1991. 19. Unemori, E. N., Bouhana. K. S.. and Werb. Z. Vectorial secretion of extracellular 39. Kramer, R. H., Bensch, K. G., and Wong, J. Invasion of reconstituted basement matrix proteins, matrix-degrading proteinases, and tissue inhibitor of metalloprotein- membrane matrix by metastatic human tumor cells. Cancer Res.. 46: 1980-1989. ases by endothelial cells. J. Biol. Chem., 265: 445^451. 1990. 1990.

1415 Downloaded from cancerres.aacrjournals.org on October 1, 2021. © 1993 American Association for Cancer Research. Identification and Characterization of a Novel Matrix-degrading Protease from Hormone-dependent Human Breast Cancer Cells

Yuenian E. Shi, Jeff Torri, Lynn Yieh, et al.

Cancer Res 1993;53:1409-1415.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/53/6/1409

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

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/53/6/1409. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.

Downloaded from cancerres.aacrjournals.org on October 1, 2021. © 1993 American Association for Cancer Research.