Communication Vol. 267, No. 20, Issue July 15, OF pp ...Communication Vol. 267, No. 20, Issue of July 15, pp. 13803-13806,1992 THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1992 by The American

Communication Vol. 267, No. 20, Issue of July 15, pp. 13803-13806,1992 THE JOURNAL OF BIOLOGICAL CHEMISTRY

0 1992 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A.

The Matrix Metalloproteinase Pump- 1 Catalyzes Formation of Low Molecular Weight (Pro)urokinase in Cultures of Normal Human Kidney Cells*

(Received for publication, March 19, 1992) Patrick A. MarcotteS, Irene M. Kozan, Sarah A. Dorwin, and Jon M. Ryan From the Thrombolytics Venture, Pharmaceutical Products Division, Corporate Molecular Biology, and Biological Development Department, Chemical and Agricultural Products Division, Abbott Laboratories, Abbott Park, Illinois 60064

The enzyme responsible for the metalloproteinase activity which cleaves the G l ~ ' ~ ~ - L e u ' ~ ~ bond of (pro)urokinase has been isolated from the conditioned medium of cultured normal human kidney cells. Using S-Sepharose and Cibacron Blue-agarose chromatog- raphy, then C-4 reversed phase high pressure liquid chromatography, a protein of about 20,000 Da was isolated. Through an identical amino-terminal se- quence, the protein was shown to be the matrix metal- loproteinase previously referred to in the literature as "pump- 1" (putative metalloproteinase). When apro- tinin was added during the course of the purification, the major species isolated was the zymogen form (28,000 Da) of pump-1. Pump-1 has been shown to efficiently cleave the susceptible bond of both pro- urokinase (single-chain) and active (two-chain) uro- kinase and thereby produce the corresponding low mo- lecular weight forms. The amino-terminal sequences of the A and B chains of low molecular weight uroki- nase prepared by action of pump-1 on recombinant high molecular weight urokinase are identical to those of the low molecular weight urokinase isolated from human kidney cell culture. Since the reaction of uro- kinase with this metalloproteinase results in separa- tion of its serine proteinase region from the domain which mediates binding to the urokinase receptor, it may be of importance in the regulation of the func- tional activity of the plasminogen activator in cellular processes.

We have previously described (1) a metalloproteinase activ- ity in human kidney cell culture which cleaves with high site specificity the G l ~ ' ~ ~ - L e u ' ~ ~ bond of high M, (pro)urokinase. This proteolysis results in separation of the serine proteinase (B-chain) region of the enzyme from the growth factor and kringle domains of the A-chain, which mediate binding of the protein to the cell-surface urokinase receptor (2,3). Therefore, this reaction may be important in the regulation of the func-

* 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" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

9A, One Abbott Park Rd., Abbott Park, IL 60064-3500. $ To whom correspondence should be addressed Dept. 48R, AP-

tional activity of the urokinase molecule and may be involved in urokinase-mediated processes related to cell migration and tumor invasion (4, 5).

Specific proteolysis of this glutamyl-leucyl bond was in- ferred from the identification of Leu144 as the amino terminus of the A-chain of kidney cell culture low M, urokinase (6). Low M , pro-urokinase with an amino terminus of Leu144 has also been isolated from cultures of the lung adenocarcinoma cell line CALU-3 (7). Our earlier work (1) demonstrated that the activity was the property of a soluble metalloproteinase, stimulated by calcium or zinc ions, but neither its complete purification nor its identification had been achieved. Isolation of the pure protein by reversed phase HPLC,' and determi- nation of its amino-terminal sequence, has demonstrated its identity with "pump-1'' (putative metalloproteinase). This protein was so named because its existence was first postu- lated as the result of the isolation of a novel cDNA with sequence homology to those encoding proteins of the colla- genase family (8). The protein was subsequently expressed through recombinant technology (9), and still later it was isolated from the culture medium of rectal carcinoma cells (10). These studies demonstrated that a metalloproteinase is indeed the product of this gene, and provided protein for its characterization (9, 10, 11). As with other matrix metallopro- teinases (12), pump-1 is known to exist in zymogen and active forms (9), and we have isolated both from the culture medium of human kidney cells.

EXPERIMENTAL PROCEDURES

Normal diploid human kidney cells were grown in modified Me- dium 199 containing 8% serum and bovine embryonic extract. After 7 days of growth, the medium was removed, the cells were rinsed with phosphate-buffered saline, and serum-free urokinase production me- dium was added to the confluent cultures. After a 21-day incubation at 36 "C without medium renewal, the cells were removed by centrif- ugation. Purification of the metalloproteinase from the medium was accomplished by chromatography on S-Sepharose (Pharmacia) (I), then on Cibacron Blue-agarose (Sigma), using a gradient of 0.2-1.0 M NaCl in 20 mM Tris-HC1, pH 7.5, to elute the proteins. The final step employed reversed phase C-4 HPLC (Vydac), using a gradient of 20-50% acetonitrile in water (with 0.1% trifluoroacetic acid), and resulted in the isolation of homogeneous proteins.

Recombinant TIMP-2 (13) was a gift of Dr. K. E. Langley of Amgen. Human sequence r-pro-urokinase (A-74187) (14) and high M, r-urokinase (A-76120) (15), derived through expression of the pro- urokinase gene in mouse SP 2/0 cells (14), were produced by Abbott Laboratories. Human kidney cell culture low M, urokinase was puri- fied from Abbokinase" (15).

Metalloproteinase activity was estimated through analysis by SDS- PAGE of the products of its reaction with r-pro-urokinase, as has been previously described (1). Either oxidized glutathione (GSSG) or 4-aminophenylmercuric acetate (pAPMA) at 1 mM was used as an activator of the proteinase. Preparative scale reactions of r-pro- urokinase and high M, r-urokinase with the metalloproteinase, and isolation of the products, were carried out by methods analogous to those described earlier (1).

Amino-terminal sequences were determined on an Applied Biosys- tems model 477 Protein Sequencer. Electrophoretic analyses were carried out using either the Pharmacia Phast system for activity determinations or SDS mini-gels (10-20% gradient) from Integrated

The abbreviations used are: HPLC, high pressure liquid chro- matography; pump-1, putative metalloproteinase; SDS, sodium do- decyl sulfate; PAGE, polyacrylamide gel electrophoresis; TIMP, tis- sue inhibitor of metalloproteinases; pAPMA, 4-aminophenylmercuric acetate; GSSG, oxidized form of glutathione.

13803

13804 Pump-1 Converts High M, to Low M, (Pro)urokinase Separation Systems for protein characterizations. Protein sequence searches, as well as calculations of molecular weights and isoelectric points of proteins and peptides, were performed on a VAX 6420 computer using UWGCG software (16).

RESULTS

Identification of the Urokinase Cleaving Enzyme as Pump- 1-The metalloproteinase activity was monitored by its activ- ity in cleaving pro-urokinase into two fragments, which we have previously shown (1) to be the result of specific hydrol- ysis of the G l ~ ' ~ " - L e u ' ~ ~ bond. Extraction from the culture medium was carried out using S-Sepharose (1); in some in- stances aprotinin was added to the collected fractions. We have found that a preferred second step is chromatography on Cibacron Blue-agarose, which is more reproducible than on zinc-Sepharose (l) , a method found to often result in low recovery of total activity. This may be due to interaction of the active site of the protein, or the thiol group of the zymogen, with the zinc-chelate support. As these studies were carried out before the identity of the enzyme was determined, these were empirical observations as to the manner in which the enzyme activity could be most readily purified.

The major species recovered after the Cibacron Blue-aga- rose chromatography were proteins of 30,000-35,000 Da, with minor components observed from 15,000-40,000 Da on SDS- PAGE (Fig. 1). The pattern observed was dependent on whether the plasmin inhibitor aprotinin had been added to the prior chromatography. From a purification in which no inhibitor had been added, a distinct band of about 20,000 Da was observed (Fig. 1, lane 2). Separation of the proteins using reversed phase HPLC (Fig. 2, upper) resulted in a correspond- ence between certain bands on the gel and the chromatogra- phy peaks. The first of the two components which eluted at 55-60 min was thereby shown to be the protein of 20,000 Da, which, after reconstitution, had detectable activity in cleaving pro-urokinase. The recovery of activity of the purified protein was not complete, but was specific in its reaction with pro- urokinase and inhibited by EDTA. Therefore a sample was subjected to amino-terminal sequence analysis, and the results are reported in Table I (Analysis A).

Da - - - T- "

66,000-

45,000- a 36,000- - 29,000- .I

24,000- .-

20,100- a

14,200- a 1 2 3 4 5 6 7

"

FIG. 1. SDS-PAGE analyses of preparations of kidney cell culture pump-1 and its zymogen before and after HPLC pu- rification. Lane I , molecular weight markers; lane 2, proteins after Cibacron Blue-agarose (no aprotinin); lane 3, component a recovered after HPLC (Fig. 2, upper); lane 4, component b recovered after HPLC (Fig. 2, lower); lane 5, proteins after Cibacron Blue-agarose (with aprotinin); lane 6, proteins in lane 5 after a 20-h treatment with 1 mM GSSG; lane 7, proteins in lane 5 after a 20-h treatment with 1 mM pAPMA. The gels were stained with the Dai-Ichi silver stain.

€ 0 N N c m W c e B a a

C a

l?

I I I I I . I I I I I I I . 1

0 10 20 30 40 50 60 70 Retention tine (minutes)

FIG. 2. HPLC purification (Vydac C-4 column) of kidney cell culture proteins after S-Sepharose and Cibacron Blue- agarose chromatography. Upper, preparation without aprotinin; Lower, analogous preparation with aprotinin added in the S-Sepha- rose chromatography. The eluted proteins were collected and the molecular mass values determined by SDS-PAGE: a, 20,000 Da; b, 29,000 Da; c, 30,000 Da; d, 17,000 Da; e, broad band near 30,000 Da.

The 20 cycles of amino acid sequence were compared with the 28,232 sequences in the Protein Sequence Data Library using the University of Wisconsin Genetics Computer Group (UWGCG) program (16). One identical match was found, to residues 95-114 of the protein known as "putative metallo- proteinase" (pump-1) (8,9). No other sequence was retrieved, even allowing for up to five mismatches in the 20 residues. As the molecular weights, inhibitor specificities, and amino-ter- minal sequences are identical, we tentatively concluded that the activity which produces low M, urokinase with an amino terminus of Leu'44 is the property of the matrix metallopro- teinase pump-1.

The purification protocol was repeated, with aprotinin being added to the fractions collected in the S-Sepharose chromatography. After the second step (Cibacron Blue-aga- rose chromatography), analysis by SDS-PAGE revealed a heterogeneous band near 30,000 Da, and no visible protein of 20,000 Da (Fig. 1, lune 5). The material was resolved into two major fractions using HPLC (Fig. 2, lower). The minor com- ponent, about 10% of the total, corresponded to a peak in the preparation without aprotinin, and was shown to be a protein of about 29,000 Da by SDS-PAGE. Analysis of the major sequence (Table I, Analysis B) demonstrated that this mate- rial is the secreted zymogen form of pump-1, which begins at Leu" (9). A minor sequence found was that of the same protein, but starting at Leuz0, the 1st 2 amino acids (Leu-Pro)

Pump-1 Converts High M, to Low M , (Pro)urokinase 13805

TABLE I Amino-terminal seqeunce analyses of active and zymogen forms of the

proteinase

Analyzer Analysis A Analysis B

cyc1e Amino acid Sequence no.” Amino acid Sequence no?

1 TYr 95 Leu 18 2 Ser 96 Pro 19 3 Leu 97 Leu 20 4 Phe 98 Pro 21 5 Pro 99 Gln 22 6 Asn 100 Glu 23 7 Ser 101 Ala 24 8 Pro 102 G ~ Y 25 9 LY 5 103 G ~ Y 26 10 Trpb 104 Met 27 11 Thr 105 Ser 28 12 Ser 106 Glu 29 13 LYS 107 Leu 30 14 Val 108 Gln 31 15 Val 109 - 32 16 Thr 110 Glu 33 17 TYr 111 Gln 34 18 Arg 112 Ala 35 19 Ile 113 Gln 36 20 Val 114 ASP 37

“ Amino acid residues numbered as in Refs. 8 and 9. ’ Low level of tryptophan derivative observed in cycle. Tryptophan amino acid in sequence; not observed in analysis.

being lost probably through a diketopiperazine reaction. The major peak, 90% of the total protein, which corresponded to the broad band at -30,000 Da on SDS-PAGE, was also devoid of activity, and by sequence analysis comprised of a mixture of proteins possibly related to collagen or elastin.

In the absence of an activator, no pro-urokinase cleaving activity could be detected in the HPLC-purified zymogen. A low level of activity could be observed in a 20-h incubation which included 1 mM pAPMA, in a preparation where the HPLC eluate had been rapidly evaporated and the protein immediately reconstituted in buffer. However, the recovered protein of 29,000 Da was found by SDS-PAGE to be un- changed in a 24-h incubation with 1 mM pAPMA. Surpris- ingly, the recovered 29,000-Da protein could not be substan- tially hydrolyzed by the HPLC-purified active enzyme, during a 24-h incubation in which a similar quantity of pro-urokinase was completely cleaved. Since activation of the zymogen is readily effected in the preparation prior to HPLC, and con- version to the 20,000-Da active form easily observed, either an irreversible reaction occurs in most of the zymogen protein molecules during the final chromatography, which employs organic solvent and strong acid, or some other component (another protease or an effector such as collagen) must be present for facile activation to proceed. Earlier work demon- strated (9) that a purified recombinant protein A pump-1 fusion protein undergoes apparent autocatalytic reaction upon treatment with pAPMA. Since the addition of other molecules was not necessary to effect this reaction, the more likely explanation of our results is that irreversible denaturation of most of the zymogen takes place during HPLC.

Activation of the Zymogen Form of Human Kidney Culture Pump-1-Incubation of the Cibacron Blue-agarose purified zymogen preparation with pAPMA or GSSG resulted in tran- sient formation of a protein of 20,000 Da (Fig. 1, lanes 6 and 7) corresponding to active pump-1, followed by its disappear- ance over a 48-h incubation. The broad band on SDS-PAGE of -30,000 Da decayed progressively over time, consistent with it being comprised of collagen fragments, which had copurified with the zymogen and which were substrates for the active enzyme.

This zymogen preparation was also incubated with recom- binant high M, pro-urokinase or high M, urokinase either alone or with the metalloproteinase activator pAPMA. Greater than 100-fold stimulation of reaction was observed upon addition of the activating agent. The cleavage of pro- urokinase was completely inhibited by addition of 50 pg/ml r-TIMP-2 (13). The single-chain and two-chain forms of urokinase were cleaved with equal efficiency by the activated pump-1. From r-pro-urokinase (14), the amino-terminal frag- ment and low M , pro-urokinase are isolated, as previously reported (1). From reaction with high M , r-urokinase (15), affinity chromatography on benzamidine-Sepharose allows isolation of low M , urokinase, identical in its amino-terminal sequences to the enzyme purified from kidney cell culture (6, 15).

DISCUSSION

Since the discovery of the matrix metalloproteinase known as pump-1 (8), it has been proposed to be involved in processes related to tumor development or invasion (17, 18). As with the related metalloproteinases stromelysin (18, 19) and type- IV collagenase (20), expression of pump-1 has been found to be elevated in invasive relative to normal cells and tissues (17, 18, 21, 22), although pump-1 mRNA could also be de- tected in some normal prostate (17) and bronchial mucosa samples (22). The protein has been found in the medium of cultured rectal carcinoma cells (lo), but pump-1 has not been previously isolated from cultures of non-malignant cells. The kidney cells used for these studies were obtained from infant cadavers with no evidence of kidney disease. Characteristic of normal, nontransformed cells, they have a diploid karyotype and limited proliferative lifespan. The cells have undergone 12-15 population doublings (three subcultures) prior to being used for these studies.

A matrix metalloproteinase of similar size, but as yet un- determined sequence, has been isolated from rat uterus (23), and others have suggested (24,25) that it is the rat homologue of pump-1. The active form of the rat protein has a more acidic isoelectric point (PI = 5.9) (23) than that calculated for active human pump-1 (PI = 9.6, residues 95-267) or for its zymogen (PI = 7.8, residues 18-267). The cDNA for human pump-1 has been transfected into COS (9) and mouse mye- loma cells (ll), and the protein thereby expressed and char- acterized. The reaction specificity of pump-1 with macromo- lecular substrates has been studied using collagens, gelatin, elastin, and fibronectin (9 , l l ) ; it has also been shown to have caseinolytic activity (10). A synthetic peptide substrate for the enzyme has recently been reported (26). The finding that this proteinase catalyzes hydrolysis of (pro)urokinase rapidly at only the G l ~ ’ ~ ~ - L e u ’ ~ ~ bond implies that bonds of this type are preferred sites of reaction of pump-1. The reported mech- anism of autolytic activation of pump-1 is through cleavage of the G1ug4-Tyrg5 bond (9). Also, from the deduced amino acid sequence (8), there are 8 glutamyl residues in the pro- segment of the protein (residues 18-94), but only 4 in the active form (residues 95-267). We propose that pump-1 may react most rapidly at Glu-X sites, where the residue X is derived from a hydrophobic amino acid, although other pep- tide bonds can also be cleaved by the enzyme (26).

This metalloproteinase (EC 3.4.24.23), recently named “matrilysin” (27), is likely also expressed by other cells in culture, as the lung adenocarcinoma cell line CALU-3 has been shown to produce low M, pro-urokinase with an amino terminus of Leu’44 (7). The amino-terminal fragment of uro- kinase has also been isolated as a mitogenic factor from prostatic cancer (PC-3) cell culture (28), possibly generated

13806 Pump-1 Converts High M,

from high M, urokinase by action of pump-1. As the concerted action of matrix-localized metalloproteinases and plasmino- gen activators (29,30) has been implicated in processes of cell migration and tumor invasion, the identification of pump-1 as the enzyme responsible for truncation of high M, urokinase may be important in understanding part of the relationship between these classes of enzymes. Studies of the level of this metalloproteinase as a function of differentiation, transfor- mation, or growth of various cell lines; and of whether there is a correspondence between expression of matrilysin (pump- 1) and the observed forms or fragments of urokinase, will be the subject of future studies.

Acknowledgments-We thank our colleagues, Drs. A. Mazar, J. Henkin, and R. B. Credo, for helpful discussions as this work pro- gressed. Conditioned medium of cultured human kidney cells was prepared by K. Jeep and M. Barron, and r-pro-urokinase and high M , r-urokinase were provided by Dr. D. Eisenhauer. Amino acid sequence analyses were carried out in the laboratory of Dr. J. Meuth. We are also grateful for the advice of Drs. W. Holleman, M. Long, and R. Schultz. We were kindly informed by Dr. L. Matrisian of the new name and EC number of the metalloproteinase. We thank a reviewer of this paper for the suggestion that the reaction of the HPLC-purified active and zymogen forms of the protein be studied.

REFERENCES 1. Marcotte P. A,, Dudlak, D., Leski, M. L., Ryan, J., and Henkin, J. (1992)

Fibrinolysis 6, Suppl. 1,57-62 2. Stoppelli, M. P., Corti, A., Soffientini, A., Cassani, G., Blasi, F., and

Assoian, R. K. (1985) Proc. Natl. Acad. Sci. U. S. A. 82,4939-4943 3. Appella, E., Robinson, E. A,, Ullrich, S. J., Stoppelli, M. P., Corti, A.,

4. Blasi, F., and Verde, P. (1990) Semin. Cancer Biol. 1, 117-126 Cassani, G., and Blasi, F. (1987) J. Bid. Chem. 262,4437-4440

5. Pollanen, J., Stephens, R. W., and Vaheri, A. (1991) Adu. Cancer Res. 57, 273-328

to Low M, (Pro)urokinase 6. Hommandberg, G. A,, and Wai, T. (1990) Biochim. Biophys. Acta 1038,

209-215

17120-17126

and Breathnach, R. (1988) Biochem. J . 252, 187-192

6327-5334

. .

7. Stump, D. C., Lijnen, H. R., and Collen, D. (1986) J. Bid. Chern. 261,

8. Muller, D., Quantin, B., Gesnel, M.-C., Millon-Collard, R., Abecassis, J.,

9. Quantin, B., Murphy, G., and Breathnach, R. (1989) Biochemistry 28,

10. Mi azaki, K., Hattori, Y., Umenishi, F., Yasumitsu, H., and Umeda, M.

11. Murphy, G., Cockett, M. I., Ward, R. V., and Docherty, A. J. P. (1991)

~.~ ~ ~ . .

6990) Cancer Res. 50, 7758-7764

Rcochem. .I. 277. 277-279 12. Woessner, JYFr(1997) FASEB J. 5, 2145-2154 13. Boone, T. C., Johnson, M. J., DeClerck, Y. A., and Langley, K. E. (1990)

14. Lo, K.-M., and Gillies, S. D. (1991) Blochim. Bcophys. Acta 1088, 217-224 Proc. Natl. Acad. Sci. U. S. A. 87,2800-2804

15. Marcotte, P. A,, Henkin, J., Credo, R. B., and Badylak, S. F. (1992)

16. Devereux, J., Haeberli, P., and Smithies, 0. (1984) Nucleic Acids Res. 12, Fibrinolysis 6, 69-78 '2Q7-90K

17. Pajouh M. S. Na el R B Breathnach. R. Finch J. S., Brawer, M. K.,

18. McDonnell, S., Navre, M., Coffey, R. J., and Matrisian, L. M. (1991) Mol.

""0 YY"

and Bowde;, G.%.'(1991jJ. CancerRes. d in . On'col. 117, 144-150

19. Matrisian, L, M., McDonnell, S., Miller, D. B.! Navre, M., Seftor, E. A.,

20. Ballin. M.. Gomez. D. E.. Sinha. C. C.. and Thoreeirsson. U. P. (1988)

Carcinog. 4, 527-533

and Hendnx, M. J. C. (1991) Am. J. Med. Scc. 302, 157-162

Biochem. Bio h$ Res Commun. 154; 832-838 21. Basset, P., Belf&,'J. P:, Wolf, C., Stoll, I., Hutin, P., Limacher, J. M.,

Podhajcer, 0. L., Chenard, M. P., Rio, M. C., and Chambon, P. (1990) Nature 348,699-704

22. Muller, D., Breathnach, R., Angelmann, A,, Millon, R., Bronner, G., Flesch, H.. Dumont, P., Eber, M., and Abecassis, J. (1991) Int. J. Cancer 48,

. .

23. Woessner, J. F., and Ta lin, C. J. (1988) J. Biol. Chem. 263,16918-16925 24. Matrisian, L. M., and #owden, G. T. (1990) Semin. Cancer Biol. 1, 107-

25. Emonard H. and Grimaud, J.-A. (1990) Cell. Mol. Biol. 36,131-153 26. Knight, d. G:, Willenbrock, F., and Murphy, G. (1992) FEES Lett. 296,

27. International Union of Biochemistry and Molecular Biolo eds) (1992) 263-266

28. Rabbanl S. A,, Desjardlns, J., Bell, A. W., Banvdle, D., Mazar, A,, Henkin, Enzyme Nomenclature 1992, p. 412, Academic Press, San%i(ego

J., and Goltzman, D. (1990) Biochem. Blophys. Res. Commun. 173,1058-

550-556

115

1064

907, 191-217

85

29. Tryggvason, K., Hoyhtya, M., and Salo, T. (1987) Biochim. Biophys. Acta

30. Moscatelli, D., and Rifkin, D. B. (1988) Biochim. Biophys. Acta 948, 67-