Site-directed Mutagenesis, Proteolytic Cleavage, and Activation of Human Proheparanase*

Heparanase is an endo-β-d-glucuronidase that degrades heparan sulfate in the extracellular matrix and cell surfaces. Human proheparanase is produced as a latent 65-kDa polypeptide undergoing processing at two potential proteolytic cleavage sites, located at Glu109-Ser110 (site 1) and Gln157-Lys158 (site 2). Cleavage of proheparanase yields 8- and 50-kDa subunits that heterodimerize to form the active enzyme. The fate of the linker segment (Ser110-Gln157) residing between the two subunits, the mode of processing, and the protease(s) engaged in proheparanase processing are currently unknown. We applied multiple site-directed mutagenesis and deletions to study the nature of the potential cleavage sites and amino acids essential for processing of proheparanase in transfected human choriocarcinoma cells devoid of endogenous heparanase but possessing the enzymatic machinery for proper processing and activation of the proenzyme. Although mutagenesis at site 1 and its flanking sequences failed to identify critical residues for proteolytic cleavage, processing at site 2 required a bulky hydrophobic amino acid at position 156 (i.e. P2 of the cleavage site). Substitution of Tyr156 by Ala or Glu, but not Val, resulted in cleavage at an upstream site in the linker segment, yielding an improperly processed inactive enzyme. Processing of the latent 65-kDa proheparanase in transfected Jar cells was inhibited by a cell-permeable inhibitor of cathepsin L. Moreover, recombinant 65-kDa proheparanase was processed and activated by cathepsin L in a cell-free system. Altogether, these results suggest that proheparanase processing at site 2 is brought about by cathepsin L-like proteases. The involvement of other members of the cathepsin family with specificity to bulky hydrophobic residues cannot be excluded. Our results and a three-dimensional model of the enzyme are expected to accelerate the design of inhibitory molecules capable of suppressing heparanase-mediated enhancement of tumor angiogenesis and metastasis.

Heparanase is an endo-␤-D-glucuronidase that degrades heparan sulfate (HS) 1 side chains of heparan sulfate proteoglycans (HSPGs) (1)(2)(3). HS interacts with a wide variety of proteins, including major components of the extracellular matrix (ECM) such as collagen IV, laminin, and fibronectin, thus playing an important role in ECM organization, self-assembly, and insolubility (4 -6). Moreover, by binding a multitude of proteins, HSPGs ensure that bioactive molecules such as growth factors, chemokines, lipoproteins, and enzymes are localized to the cell surface and ECM and function in the control of normal and pathological processes (5)(6)(7)(8)(9).
Heparanase activity is involved in cell invasion associated with tumor metastasis, angiogenesis, autoimmunity, and inflammation (1)(2)(3)10). Heparanase is up-regulated in a variety of human tumors, correlating with increased tumor vascularity and poor postoperative survival (11)(12)(13)(14). Cloning of the same heparanase cDNA by several groups (2,3,(15)(16)(17) suggests that a single dominant functional HS-degrading endoglycosidase is expressed by mammalian cells. The heparanase cDNA encodes for a latent 65-kDa proheparanase of 543 amino acids. It undergoes proteolytic processing that is likely to occur at two cleavage sites, yielding an 8-kDa peptide, corresponding to the N terminus of the protein, a 50-kDa polypeptide at the C terminus, and a 6-kDa linker peptide between these regions (18,19). Heparanase activity is readily detected in mammalian cells transfected with plasmid vectors encoding the full-length heparanase cDNA (2,3,15). The active form of the enzyme has first been thought to be a 50-kDa polypeptide, isolated from cells and tissues. However, attempts to obtain heparanase activity in cells transfected with the 50-kDa polypeptide have failed (18,19), indicating that the N-terminal portion of the proenzyme is required for heparanase enzymatic activity. Purified active heparanase appears in SDS-PAGE as a 50-kDa protein accompanied by an 8-kDa polypeptide, suggesting that the active form of the enzyme exists as a heterodimer composed of a 50-kDa subunit (Lys 158 -Ile 543 ) noncovalently associated with an 8-kDa subunit (Gln 36 to Glu 109 ) (20). Indeed, coexpression of the 8-and 50-kDa heparanase subunits demonstrated that heterodimer formation is necessary and sufficient for heparanase enzymatic activity (18,19,21).
Based on sequence analysis of the N terminus of the 50-kDa heparanase subunit and mass spectrometry analysis of the 8-kDa subunit (16,20), two potential cleavage sites were predicted in the human proheparanase molecule: one site (site 1) located between Glu 109 and Ser 110 , and the other site (site 2) between Gln 157 and Lys 158 (Fig. 1). Yet, the mechanism of proteolytic processing and activation of the proenzyme is largely unknown.
We applied site-directed mutagenesis to study the amino acid sequence context at cleavage sites 1 and 2 in the human proheparanase. Our results indicate that cleavage at site 2 is brought about by an endoproteolytic activity, most likely cathepsin L, with specificity to a bulky hydrophobic amino acid at position P2 (i.e. Tyr 156 , one amino acid N-terminal to the P1 cleavage site).
Preparation of Dishes Coated with ECM-Bovine corneal endothelial cells (second to fifth passages) were plated onto 35-mm tissue culture dishes at an initial density of 2 ϫ 10 5 cells/ml and cultured as described above, except that 4% dextran T-40 was included in the growth medium. Na 2 35 SO 4 (25 Ci/ml, Amersham Biosciences) was added on days 2 and 5 after seeding, and the cells were incubated with the label without medium change. On day 12, the sub-endothelial ECM was exposed by dissolving the cell layer with phosphate-buffered saline containing 0.5% Triton X-100 and 20 mM NH 4 OH, followed by four washes with phosphate-buffered saline (10,23). The ECM remained intact, free of cellular debris, and firmly attached to the entire area of the tissue culture dish.
Generation of Heparanase Mutant cDNAs and Expression Constructs-Mutants of the human heparanase cDNAs were constructed according to site-specific mutagenesis by overlap extension approach (25). Briefly, for each mutation, two fragments were amplified in two separate PCR reactions using the heparanase cDNA as a template. PCR 1 amplified the fragment that contains the mutation site together with the upstream sequence, utilizing reverse primer containing the mutation and a common forward primer containing the wild type heparanase sequence (1UERIhpa). PCR 2 amplified the DNA fragment that contains the mutation site together with downstream sequences, utilizing the forward primer containing the mutation and a common reverse primer containing the wild type heparanase sequence (4Lhpa). The two PCR-purified (JETquick gel extraction spin kit, Genomed, BadOeynhausen, Germany) fragments were mixed, denatured, annealed, and extended, and the product was amplified in a third PCR reaction using 1UERIhpa and 4Lhpa primers. The PCR 3 product was subcloned into Hpa-pcDNA3 digested with EcoRI and AflII, using a TaKaRa DNA ligation kit II (Takara Bio Inc., Shiga, Japan). and 5Ј-AGTGTTCTCGGGCTGCCAATTGCTCCT-3Ј. All PCR reactions were performed using the Pwo DNA polymerase (Roche Applied Science) and the following cycling conditions: 95°C for 3 min followed by 32 cycles at 96°C for 18 s, 60°C for 90 s, and an elongation step at 72°C for 70 s.
Transfection-Jar cells were transfected (48 h at 37°C) with pcDNA3 vectors containing the wild type or mutated full-length heparanase cDNA, using FuGENE transfection reagent (Roche Applied Science) according to the manufacturer's (Invitrogen) instructions. Transfected cells were then selected with 600 g/ml G418, and stable populations of heparanase expressing cells were obtained (10,15).

SDS-PAGE and Western Blot Analysis-Transfected
Jar cells (1 ϫ 10 7 ) were lysed in buffer containing 150 mM NaCl, 50 mM Tris-HCl, pH 8, and 1% Brij supplemented with a mixture of protease inhibitors (Roche Applied Science). Cell debris was removed by centrifugation, and protein concentration in the supernatants was determined using Bradford reagent (Bio-Rad). Samples (3 mg of total protein) were added to 60 l of heparin-Sepharose beads (Amersham Biosciences) and incubated overnight at 4°C. Beads were then washed once in 0.6 M NaCl followed by two washes in saline and boiling (5 min) in SDS-PAGE sample buffer. To detect the 8-kDa subunit, the 0.6 M NaCl wash was avoided, and the beads were washed twice with saline alone. Supernatants were then subjected to SDS-PAGE (4 ml of 10% acrylamide and 4 ml of 18% acrylamide) under reducing conditions, and proteins were transferred to a polyvinylidene difluoride membrane (Bio-Rad). Western blot analysis was performed using anti-heparanase polyclonal antibodies raised in rabbits against the full-length 65-kDa human proheparanase (Ab1453), synthetic peptides residing in the 8-kDa subunit (Ab810; 96 GTKTDFLIFDPKK 108 ), or the linker segment (AbCKLE; 127 CKYGSIPPDVEEKLRLE 143 ). Antibody #1453 was affinity-purified using an immobilized 50-kDa subunit (AP1453). The polyvinylidene difluoride membrane was probed with the appropriate antibody, followed by horseradish peroxidase-conjugated secondary antibody and a chemiluminescent substrate (Pierce) (18,26,27).
Computer Modeling of Heparanase-A thre-dimensional model of heparanase was generated using Modeler (DS Modeling, Accelrys Inc., San Diego, CA) and the structure of ␤-D-xylosidase (1px8, PDB code) as a template. Alignment was provided by the threading server 3D-PSSM (28,29), giving an identity value of 10.8% and a similarity value of 24.1%. Model building was followed by energy minimization using CHARMm (DS Modeling, Accelrys, Inc.), choosing CHARMm22 as a force field. The model depicts the linker segment and the 50-kDa subunit, but not the 8-kDa subunit that could not retrieve any protein structure.

RESULTS
Processing of Human Heparanase at Site 1-The human proheparanase sequence spanning the two predicted proteolytic cleavage sites (sites 1 and 2), is presented in Fig. 1. Multiple sequence alignment revealed a high conservation of the two cleavage sites among the mammalian (i.e. human, bovine, mouse, and rat) proheparanases (not shown). Mutations were generated in the sequence spanning the predicted cleavage site 1 ( 105 DPKKESTFEER 115 ) of the full-length ϳ65-kDa proheparanase ( Fig. 2A). These included double mutations at Glu 109 and Ser 110 (mutation #1), Lys 107 and Lys 108 (mutation #2), and Asp 105 and Pro 106 (mutation #3). Each of these amino acids was replaced by glutamine (Gln). We have also deleted six amino acids ( 110 STFEER 115 ) spanning Ser 110 to Arg 115 (mutation #4) and 11 amino acids ( 110 STFEERSY-WQS 120 ) spanning Ser 110 to Ser 120 (mutation #5) ( Fig. 2A). The mutated Hpa-pcDNAs were transfected into Jar cells devoid of endogenous heparanase protein and enzyme activity ( Fig. 2) but possessing the enzymatic machinery for proper processing and activation of the proenzyme. Expression of the wild type and mutants proheparanases, production of the 8-and 50-kDa subunits, and enzyme activity were evaluated by Western blot analysis and heparanase enzyme assay. All mutants exhibited heparanase activity comparable to cells transfected with the full-length wild type enzyme ( Fig. 2A), as revealed by release of HS degradation fragments from the ECM substrate. We have previously demonstrated that labeled fragments eluted in fractions 20 -35 are degradation products of HS, as they were (i) 5to 6-fold smaller than intact HS side chains; (ii) resistant to further digestion with papain and chondroitinase ABC; and (iii) susceptible to deamination by nitrous acid (24). The mutants were all processed into the 8-and 50-kDa subunits, comparable to the wild type heparanase (Fig. 2B). Cells transfected with vector alone (V o ) showed no detectable heparanase activity and protein. Altogether, applying the mutations and deletions described above, we were unable to identify a specific amino acid or sequence context necessary for cleavage at site 1.
Glutamine 157 at P1 of Cleavage Site 2 Is Not Required for Processing of Human Proheparanase-Sequence analysis of the N terminus of the 50-kDa subunit isolated from cells and tissues revealed that proheparanase cleavage at site 2 occurs between Gln 157 and Lys 158 (20,30,31). A point mutation was generated at Gln 157 , the precise predicted site 2 cleavage, by substituting this P1 amino acid to alanine (Q157A, mutation #1). Surprisingly, the proheparanase Q157A mutant showed, upon transfection into Jar cells, enzymatic activity (Fig. 3A) and processing into the 8-kDa (not shown) and 50-kDa (Fig. 3B) subunits, similar to the native 65-kDa proenzyme. This result prompted us to investigate further the requirements and amino acids essential for proper proteolytic cleavage at site 2.
Mutagenesis of Conserved Amino Acids in the Flanking Sequence of Cleavage Site 2-We investigated the possibility that proteolysis at site 2 does not occur at the carboxyl side of Gln 157 but rather at the N-terminal side of the adjacent Lys 158 (Fig. 1). Proteolytic cleavage at the N-terminal side of an amino acid is uncommon. Nevertheless, Nardilysin, for example, cleaves at the amino side of two adjacent lysine residues in a target protein (32). Two adjacent lysines, Lys 158 and Lys 159 , found at FIG. 1. Human proheparanase processing and amino acid sequence of the linker segment and two predicted cleavage sites. The 65-kDa human proheparanase undergoes proteolytic cleavage at two predicted sites, site 1 and site 2 (arrows), generating 8-kDa subunit at the N terminus, 50-kDa subunit at the C terminus, and a linker peptide of 48 amino acids (ϳ6 kDa) in between (linker). Excision of the linker segment is followed by heterodimerization of the two subunits, forming the active heparanase enzyme. ; lysines 107 and 108 were replaced by glutamine (f, 107 KK 108 :QQ), aspartic acid 105 and proline 106 were replaced by glutamine (•), or 6 (Ⅺ, ⌬STFEER), and 11 (᭜, ⌬STFEERSYWQS) amino acids were deleted. Cell lysates were incubated with 35 S-labeled ECM and tested for heparanase activity by gel filtration (Sepharose 6B) analysis of released sulfate-labeled material, as described under "Experimental Procedures." Jar cells transfected with native proheparanase or each of the mutated constructs exhibited the same heparanase enzymatic activity. B, Western blot analysis. Lysates (3 mg of protein each) of transfected Jar cells were incubated with heparin-Sepharose beads. The beads were boiled in SDS-PAGE sample buffer, and the solubilized proteins were subjected to SDS-PAGE and Western blot analysis with antibodies raised against the 65-kDa (AP 1453) and 8-kDa (Ab 810) proteins, as described under "Experimental Procedures." Proheparanase is properly processed into the 50-and 8-kDa subunits, in all mutants. cleavage site 2 of the human proheparanase ( Fig. 1) were separately substituted to alanine. As demonstrated in Fig. 4, both the K158A and K159A mutants showed heparanase activity (Fig. 4A) and processing into the 8-kDa (not shown) and 50-kDa (Fig. 4B, left) subunits, in a manner indistinguishable from that observed with the native proenzyme.
We have also substituted the conserved Arg 153 with glutamine, assuming that it may constitute a potential RXXX2 recognition motif for a subtilisin/kexin isozyme-1 convertase (33)(34)(35). Again, the R153A mutant (mutation #10) was found to be normally processed and enzymatically active (Fig. 4, A and  B). The results described above exclude a role of either Lys 158 , Lys 159 , or Arg 153 in recognition of the protease at site 2 and subsequent processing of the human proheparanase. However, a triple (RKK:QQQ) mutant in which each of the three basic amino acids, Arg 153 , Lys 158 , and Lys 159 , was replaced by glutamine (mutation #7) showed no heparanase activity when expressed in transfected Jar cells (Fig. 4A). We have previously demonstrated that sulfate-labeled material released from ECM and eluted in fractions 3-15 (Sepharose 6B, peak I) is produced by proteolytic enzymes degrading the proteoglycan core protein and residing in the ECM and cell lysates (24). Western blot analysis of the RKK mutant revealed abnormal processing at site 2, manifested by the appearance of a slightly higher molecular mass (ϳ51 kDa) protein rather than the 50-kDa polypeptide, whereas generation of the 8-kDa subunit appeared normal (Fig. 4B, right). Apparently, the sequence context of positively charged amino acids at site 2 is critical for proper processing of proheparanase.
Additional point mutations were generated in other conserved amino acids (Glu 154 , Leu 151 , and Leu 152 ) located upstream of Gln 157 (mutations #14 and #15). Substitutions of each of these amino acids by alanine had no effect on both heparanase activity (Fig. 5A) and processing of the 65-kDa proheparanase (Fig. 5B), determined in extracts of transfected Jar cells. In contrast, substitution of Tyr 156 to alanine (Y156A; mutation #11) altered the correct processing of proheparanase and resulted in the appearance of a slightly higher molecular mass (ϳ51 kDa) protein instead of the characteristic 50-kDa subunit (Fig. 6B). Accordingly, lysates of Jar cells expressing the Y156A mutant showed no heparanase activity, as indicated by the lack of labeled HS degradation fragments eluted in fractions 20 -35 (Fig. 6A). Tyr 156 was also substituted to valine (Y156V; mutation #13), a hydrophobic amino acid, or glutamic acid (Y156E; mutation #12), a negatively charged amino acid. Whereas the Y156E mutant showed no heparanase enzymatic activity (Fig. 6A) and yielded a protein of ϳ51 kDa (Fig. 6B), the Y156V mutant was normally processed to the native subunits, yielding heparanase activity, similar to that observed with Jar cells expressing the wild type enzyme (Fig. 6). These results indicate that Tyr 156 has a critical role in processing of proheparanase at site 2. Our finding that Tyr 156 can be replaced by valine, but not by alanine or glutamic acid, emphasizes the importance of a bulky hydrophobic amino acid at position 156 for proper processing and activation of the latent 65-kDa proenzyme.
Activation of Proheparanase Is Inhibited by a Cathepsin L Inhibitor-The requirement for an aromatic amino acid (Tyr 156 ) at position P2 of cleavage site 2 and the cleavage specificity (N-terminal side of paired basic residues) suggest that cleavage of proheparanase at site 2 is brought about by cathepsin L-like proteases of the papain family (36,37). This possibility was investigated by applying the Jar cell system, used throughout the present study. Briefly, Jar cells, capable of proheparanase processing but devoid of heparanase gene expression and enzymatic activity, were transiently transfected with the full-length 65-kDa proheparanase cDNA in the absence or presence of Z-Phe-Phe-CH 2 F (Z-FF-FMK, cathepsin L inhibitor I, Calbiochem, cat. no. 421419), a cell-permeable and irreversible inhibitor of cathepsin L (38). The cell layer was then washed, and cell lysates were tested for heparanase activity using sulfate-labeled ECM as a substrate. Transient transfection was applied in these experiments, taking into account the relatively long half-life (ϳ 30 h) of the heparanase protein (26). Stable transfection involving a long term exposure to the cathepsin inhibitor was toxic and may result in accumulation of both the proheparanase and active enzyme. As demonstrated in Fig. 7, Jar cells transfected with the proenzyme in the presence of 7.2 M cathepsin L inhibitor failed to express heparanase activity, similar to mock transfected cells. Cells transfected with proheparanase in the presence of 0.72 M inhibitor exhibited a high heparanase activity (i.e. release of HS degradation fragments eluted in peak II), similar to cells trans- FIG. 3. Point mutation in Gln 157 has no effect on heparanase processing and activation. A, enzymatic activity. Jar cells were stable transfected with pcDNA plasmid alone (V o ) (✕ | ), plasmid encoding the full-length heparanase cDNA (Hpa) (E), or plasmids encoding point mutated heparanase cDNA in which glutamine 157 was replaced by alanine (᭜, Q157A). Cell lysates were tested for heparanase activity as described in the legend to Fig. 2 and under "Experimental Procedures." Mutant Q157A was as active as the wild type heparanase. B, Western blot analysis. Lysates (3 mg of protein each) of transfected Jar cells were incubated with heparin-Sepharose beads. The beads were boiled in sample buffer, and soluble proteins were subjected to SDS-PAGE and Western blot analysis with antibodies raised against the 65-kDa (AP 1453) proheparanase, as described under "Experimental Procedures." Cells transfected with the full-length (Hpa) or Q157A point-mutated enzyme yielded the characteristic 50-kDa heparanase subunit.
fected with the proenzyme in the absence of the inhibitor (Fig. 7). The transfection step itself was not affected by the cathepsin L inhibitor as indicated by similar levels of proheparanase mRNA expression detected by RT-PCR and a similar ␤-Gal enzymatic activity determined in Jar cells transfected with ␤-galactosidase cDNA in the absence or presence of the Z-FF-FMK cathepsin L inhibitor (not shown). In a subsequent experiment, recombinant proheparanase was incubated (1 h, 37°C) with a cytosolic fraction of Jar cells, in the absence or presence of various protease inhibitors. The reaction mixture was then incubated (5 h, 37°C, pH 5.8) with sulfate-labeled ECM and analyzed for heparanase activity. Activation of the proenzyme was partially inhibited by an inhibitor (E-64, Calbiochem, cat. no. 324890) of cysteine pro-teases (58% inhibition), or by pepstatin A (an inhibitor of cathepsin D and other aspartic proteases) (47% inhibition). An almost complete inhibition (98%) was obtained by an inhibitor of cathepsins L, B, and S (inhibitor III, Calbiochem, cat. no. 219419) and by the Z-FF-FMK inhibitor of cathepsin L (82% inhibition). In contrast, there was no inhibition by a mixture of metalloproteinase inhibitors (Sigma, A4336) and by several serine peptidase inhibitors (4-(2-aminoethyl)benzenesulfonylfluoride hydrochloride (AEBSF), 3,4-dichloroisocoumarin, and chymostatin). Similar results were obtained when the Jar cell lysate was incubated with proheparanase and each of the protease inhibitors at pH 5.5 or 6.8 (not shown).
Next, recombinant 65-kDa proheparanase was incubated (1 ). Cell lysates were tested for heparanase activity as described under "Experimental Procedures." Cells transfected with each of the mutated heparanase constructs were as active as cells transfected with the wild type enzyme. B, Western blot analysis. Lysates of transfected Jar cells were incubated with heparin-Sepharose beads. The beads were boiled in sample buffer, and soluble proteins were subjected to SDS-PAGE and Western blot analysis with antibodies AP-1453, as described in the legend to Fig. 2 and under "Experimental Procedures." Proheparanase appears properly processed into the 50-kDa subunits, in both mutants.
h, 37°C) with a purified preparation of cathepsin L (Sigma, cat. no. C-6854) in the absence and presence of the Z-FF-FMK cathepsin L inhibitor. As demonstrated in Fig. 8, the proenzyme was processed into the active enzyme, as revealed by heparanase enzymatic activity (Fig. 8A) and Western blot anal-ysis showing the 50-and apparent 8-kDa subunits (Fig. 8B). The FF-FMK inhibitor itself had no effect on heparanase enzymatic activity (not shown). Processing and activation took place at both pH 5.5 (in the presence of EDTA and dithiothreitol) (Fig. 8) and pH 6.8 (not shown) and was readily inhibited by the cathepsin L inhibitor (Fig. 8). Altogether these results suggest that processing and activation of proheparanase is, to a large extent, brought about by a cathepsin L-like activity. DISCUSSION We have shown that substitution and deletion mutations at cleavage site 1 (i.e. the N terminus of the linker segment) of the human proheparanase and its flanking amino acids (6 and 11 residues toward the N and C termini, respectively) had no effect on processing and generation of an active enzyme. It therefore appears that the sequence context at cleavage site 1 does not constitute a strict recognition motif for a specific protease (36). Clearly, additional mutations should be applied outside this region to verify the mode of processing at site 1 and generation of the 8-kDa subunit. In contrast, site-directed mutagenesis at site 2 revealed that substitution of the conserved Tyr 156 to alanine or glutamic acid abolished normal processing and activation of proheparanase. No effect was seen when Tyr 156 was replaced by valine, indicating that the specificity of proteolytic cleavage at site 2 depends on the bulky hydrophobic nature of the amino acid residue at P2 of the cleavage (i.e. Tyr 156 ). Replacement of Tyr 156 by alanine or glutamic acid yielded a slightly higher molecular mass (ϳ51 kDa) subunit, most likely due to an alternative cleavage site specified by the next tyrosine at position 146, 10 amino acids upstream the original site, within the linker segment. Interestingly, both the actual results and proposed three-dimensional model (Fig. 9, see below) indicate that such a small extension of the 50-kDa subunit abolishes heparanase enzymatic activity, apparently because the HS substrate is no longer accessible to the enzyme active site.
The role of Tyr 156 in proper cleavage of site 2 implicates cathepsin L-like proteolytic activity in proheparanase processing (37-39). Cathepsins are a prominent subgroup of the papain superfamily of peptidases, among which cathepsin L preferentially recognizes aromatic residues at position P2 of the cleavage site (37, 40 -45). We have demonstrated that Jar cells transfected with the proheparanase cDNA in the presence of a cell-permeable irreversible inhibitor (Z-Phe-Phe-CH 2 F) of cathepsin L (38), failed to express heparanase enzymatic activity upon lysis and incubation with sulfate-labeled ECM. Moreover, the proenzyme was processed and activated by a purified preparation of cathepsin L in a cell-free system, and this was inhibited by the Z-FF-FMK cathepsin L inhibitor. Similarly, in vitro processing and activation of recombinant proheparanase by lysates of Jar cells were inhibited by the cathepsin L inhibitor, but there was no effect to serine protease inhibitors and to a mixture of metalloproteinase inhibitors. These results, the requirement for a hydrophobic amino acid at P2, and the cleavage specificity (N-terminal side of paired basic residues) suggest that cleavage of proheparanase at site 2 is brought about by cathepsin L. Besides their role inside lysosomes, cysteine proteases degrade and process proteins outside lysosomes (41). Cathepsin L plays a role in processing of diverse proteins, including yolk (46), interleukin-8 (47), and proenkephalin (40), as well as in the generation of endostatin from collagen XVIII (48). Cathepsin D has long been shown to be involved in breast cancer progression and metastasis (49). Because cathepsin D, an aspartic protease, exhibits preference to bulky hydrophobic residues at P1 and P1Ј, its potential involvement in proheparanase processing cannot be excluded. In fact, our preliminary results indicate that in a cell-free system, recombinant proheparanase is activated by cathepsin D, although, unlike cathepsin L, several cleavage fragments were detected by Western analysis. A possible involvement of cathepsin D is supported by the partial inhibition of proheparanase activation in Jar cell lysates by pepstatin, an inhibitor of cathepsin D, and other aspartic proteases. An alternative possibility is that in some cells cleavage at site 2 may take place at Tyr 156 , serving as P1, by chymotrypsin-like proteases known for their specificity to aromatic hydrophobic amino acids (37,50,51). In this case, it is predicted that, following cleavage at Tyr 156 , the adjacent Gln 157 is removed by an exoproteolytic activity, leaving Lys 158 at the N terminus of the 50-kDa subunit, as revealed by sequence analysis of the heparanase protein isolated from cells and tissues (2,20,30,31). Such a combination of endo-and exo-proteolytic activities has been reported for convertases (52). Again, our preliminary results show that recombinant 65-kDa proheparanase is activated in a cell-free system by kallikrein human 3, a member of the chymotrypsin family of proteases, also known as prostate-specific antigen (53-55) (data not shown). It should be noted, however, that results obtained with cell lysates are not representative of intact cells, due to the presence of a variety of proteolytic enzymes that under physiological conditions may not be accessible to the proheparanase protein. The actual involvement of cathepsin D and kallikrein human 3 in heparanase processing by intact breast and prostate cancer cells is being investigated.
It is remarkable that all the mutations applied in this study and introduced at sites 1 and 2 failed to yield a product composed of the 50-kDa subunit attached to an intact linker segment with a predicted molecular mass of ϳ6 kDa. Moreover, a linker peptide of ϳ6-kDa could not be detected even in our analysis of processing products of the wild type proheparanase, using antibodies (Ab CKLE) that specifically recognize the linker segment (data not shown). Instead, substitution of the essential Tyr 156 yielded a protein of ϳ51-kDa, which failed to form an active enzyme. This result is in agreement with previous studies indicating that the linker segment must be removed from the 50-kDa subunit to enable proper heterodimerization and generation of an active heparanase enzyme (19). On the other hand, our studies, applying transfected cells, as well as a cell-free system, suggest that such a restriction may not take place at the N terminus of the linker segment so that the 8-kDa subunit may still operate even when extended by a few amino acids at its C terminus.
A computer-generated three-dimensional structure of the human proheparanase (see "Experimental Procedures") revealed that the linker segment masks the catalytic site of the proenzyme but is exposed to the outer surface the 50-kDa protein (Fig. 9A). The model predicts that complete removal of the linker segment is necessary to obtain active heparanase (Fig. 9B), as also substantiated by the mutational analysis. In fact, the three-dimensional model (Fig. 9C) and theoretical analysis of denaturation/renaturation dynamics of the 50-kDa subunit indicate that an extra ϳ1-kDa peptide is enough to hinder accessibility of the HS substrate to the active site of the enzyme, thus maintaining the enzyme in an inactive configuration (not shown). We have recently identified three heparinbinding domains along the 50-kDa subunit. 2 Only the one residing in the N terminus of this subunit is critical for enzy- FIG. 9. A three-dimensional model of heparanase, highlighting amino acids that participate in heparanase processing, activity, and substrate recognition. The heparanase model was predicted based on similarity to ␤-D-xylosidase, as described under "Experimental Procedures." A, presented are the linker segment (white ribbon) and the 50-kDa subunit (purple ribbon). Cleavage site 1 is located at the N terminus of the linker (yellow arrow). Cleavage site 2 (white arrow) is located between Gln 157 (blue) and Lys 158 (orange). Tyr 156 (yellow) is exposed while Gln 157 (blue) is buried, implying that Tyr 156 is available for proteolytic cleavage, in agreement with the mutagenesis results. B, magnification of the enzyme active site. Presented are the substrate (red), Glu 225 and Glu 343 (green), identified as the proton donor and nucleophile residues, respectively (31), as well as Lys 158 and Lys 159 (blue) located in the substrate binding domain. C, side image of the heparanase model. A 1-kDa peptide at the C terminus of the linker protein is left bound to the 50-kDa heparanase subunit when Tyr 156 is substituted by glutamic acid or alanine, and the proenzyme is improperly processed at site 2. The enzyme is inactive, apparently because the HS substrate is no longer accessible to the enzyme active site. matic activity, so that binding to heparin and HS may still take place, in a manner similar to that observed with the latent enzyme. A complex mode of processing and activation of the latent proenzyme ensures a tight regulation of active heparanase expression under normal conditions and escape in disease situations. The present results and the proposed three-dimensional model are expected to facilitate the design of peptides and small molecules capable of blocking heparanase processing and activation, toward inhibition of heparanase-mediated acceleration of tumor angiogenesis and metastasis.