CATHEPSIN L IS RESPONSIBLE FOR PROCESSING AND ACTIVATION OF PROHEPARANASE THROUGH MULTIPLE CLEAVAGES OF A LINKER SEGMENT

Heparanase is an endo-beta-d-glucuronidase that degrades heparan sulfate in the extracellular matrix and on the cell surface. Human proheparanase is produced as a latent protein of 543 amino acids whose activation involves excision of an internal linker segment (Ser(110)-Gln(157)), yielding the active heterodimer composed of 8- and 50-kDa subunits. Applying cathepsin L knock-out tissues and cultured fibroblasts, as well as cathepsin L gene silencing and overexpression strategies, we demonstrate, for the first time, that removal of the linker peptide and conversion of proheparanase into its active 8 + 50-kDa form is brought about predominantly by cathepsin L. Excision of a 10-amino acid peptide located at the C terminus of the linker segment between two functional cathepsin L cleavage sites (Y156Q and Y146Q) was critical for activation of proheparanase. Matrix-assisted laser desorption ionization time-of-flight mass spectrometry demonstrates that the entire linker segment is susceptible to multiple endocleavages by cathepsin L, generating small peptides. Mass spectrometry demonstrated further that an active 8-kDa subunit can be generated by several alternative adjacent endocleavages, yielding the precise 8-kDa subunit and/or slightly elongated forms. Altogether, the mode of action presented here demonstrates that processing and activation of proheparanase can be brought about solely by cathepsin L. The critical involvement of cathepsin L in proheparanase processing and activation offers new strategies for inhibiting the prometastatic, proangiogenic, and proinflammatory activities of heparanase.


Heparanase
is an endo-β-Dglucuronidase that degrades heparan sulfate in the extracellular matrix and on the cell surface. Human proheparanase is produced as a latent protein of 543 amino acids whose activation involves excision of an internal linker segment (Ser 110 -Gln 157 ), yielding the active heterodimer composed of 8-and 50-kDa subunits. Applying cathepsin L knockout tissues and cultured fibroblasts, as well as cathepsin L gene silencing and over-expression strategies, we demonstrate, for the first time, that removal of the linker peptide and conversion of proheparanase into its active 8 + 50 kDa form is brought about predominantly by cathepsin L. Excision of a 10 amino acid peptide located at the C-terminus of the linker segment between two functional cathepsin L cleavage sites (Y 156 Q-and Y 146 Q) was critical for activation of proheparanase. MALDI-TOF mass spectrometry (MS) demonstrates that the entire linker segment is susceptible to multiple endocleavages by cathepsin L, generating small peptides. MS demonstrated further that an active 8-kDa subunit can be generated by several alternative adjacent endocleavages, yielding the precise 8-kDa subunit and/or slightly elongated forms. Altogether, the mode of action presented here demonstrates that processing and activation of proheparanase can be brought about solely by cathepsin L. The critical involvement of cathepsin L in proheparanase processing and activation offers new strategies for inhibiting the prometastatic, proangiogenic and proinflammatory activities of heparanase.
Heparanase is an endo-β−D-glucuronidase that degrades heparan sulfate (HS) side chains of heparan sulfate proteoglycans (HSPGs) (1)(2)(3)(4). Enzymatic cleavage of HSPGs by heparanase contributes to disassembly of the extracellular matrix (ECM) and basement membranes, resulting in accelerated cell invasion (1)(2)(3)(4). Heparanase activity has been traditionally correlated with cell invasion associated with cancer metastasis, a consequence of structural modifications that loosen the ECM barrier (5,6). This notion gained further support by employing siRNA and ribosome technologies, clearly demonstrating heparanasemediated HS cleavage and ECM remodeling as a critical prequisite for cancer metastasis, angiogenesis and inflammation (7). Heparanase up-regulation has been documented in an increasing number of human carcinomas (3,8). In many cases, heparanase induction correlates with increased tumor metastasis, tumor vascular density, and shorter post operative survival of cancer patients, providing strong clinical support for the pro-metastatic and pro-angiogenic functions of the enzyme (3,9,10).
Given the ability of heparanase to affect cell and tissue function in a variety of normal and pathophysiological processes, its activity must be kept under tight regulation. Control of heparanase post-translational modifications (4,11,12), including proteolytic processing and activation (13)(14)(15)(16), represent important regulatory mechanisms. The human heparanase cDNA has an open reading frame that encodes for 543 amino acids yet the latent proheparanase is 65-kDa in size due to glycosylation (1,11). Proheparanase undergoes proteolytic processing involving the removal of an internal segment of 48 amino acids residing between Ser 110 (site 1) and Gln 157 (site 2), yielding an active heterodimer composed of 8-kDa and 50-kDa subunits (13)(14)(15)(16). Until now, the mode of processing and the fate of the internal linker segment (Ser 110 -Gln 157 ) were only partially elucidated (13). Mutagenesis at site 1 and its flanking sequences failed to identify critical residues for proteolytic cleavage, while processing at site 2 required a bulky hydrophobic amino acid (Tyr 156 ) at position P2 of the cleavage site (13). The critical involvement of Tyr 156 in accurate cleavage at site 2 and the inhibitory effect of a cell permeable inhibitor of cathepsin L implicate a cathepsin L-like proteolytic activity in the processing of proheparanase (13).
Cathepsin L, a papain-like cysteine proteinase, is synthesized as a preproenzyme (17). The prepeptide is removed in the endoplasmic reticulum and the pro-enzyme undergoes autoactivation in the acidic environment of late endosomes and/or lysosomes. The mature form is stored within lysosomes where it functions as an endopeptidase (17). In addition to its role in terminal protein degradation inside lysosomes, cathepsin L is implicated in multiple physiological and pathological processes (18), primarily by virtue of its involvement in pro-enzyme processing and protein maturation (19). For example, in the thymic cortex, cathepsin L plays a key role in cleaving the CLIP peptide thereby allowing for antigen presentation in the context of MHC class II molecules (20). Cathepsin L within secretory vesicles is responsible for the generation of several peptide neurotransmitters and hormones, including enkephalin, β-endorphin and ACTH (21)(22)(23). Recently, a cathepsin L isoform that lacks a signal peptide was reported to function in processing of the CDP/Cux transcription factor and thereby in cell cycle regulation (24). Moreover, cathepsin L and other cysteine proteases have been implicated in tumor development and progression through degradation of protein constituents of the plasma membrane and the extracellular matrix (ECM) (25)(26)(27)(28). It was recently shown that deletion of cathepsin L in the RIP1-Tag2 (RT2) mouse model of pancreatic islet cell tumorigenesis resulted in a profound reduction in tumor growth (average decrease of 88%), and a significant impairment in tumor invasion (25).
The present study was undertaken to characterize the mechanism of proheparanase processing. Applying cathepsin L knockout fibroblasts and cathepsin L gene silencing and over-expression strategies, we demonstrate that conversion of proheparanase into its active form is brought about almost exclusively by cathepsin L. Furthermore, the precise mode of action of cathepsin L in proheparanase processing was elucidated in a cell free system.
Tissue samples. The generation of cathepsin L knockout mice has been previously described (32). Liver and spleen, as well as serum, were collected from wild-type and cathepsin L knockout mice. Tissues were homogenized in lysis buffer (1% Brij 35, 150 mM NaCl, 50 mM Tris-HCl, pH 7.5) supplemented with a mixture of protease inhibitors (Sigma) (13) and the supernatant fraction was subjected to heparanase enzymatic activity, as described below.
Preparation of dishes coated with ECM. Bovine corneal endothelial cells (second to fifth passage) were plated into 35 mm tissue culture dishes at an initial density of 2 x 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 mCi/ml) (Amersham Pharmacia Biotech, Buckinghamshire, UK) was added on days 2 and 5 after seeding and the cultures were incubated with the label without medium change. On day 12, the subendothelial ECM was exposed by dissolving the cell layer with PBS containing 0.5% Triton X-100 and 20 mM NH 4 OH, followed by four washes with PBS (6,30). The ECM remained intact, free of cellular debris and firmly attached to the entire area of the tissue culture dish.

SDS-PAGE and Western blot analysis.
Cells (1 x 10 6 ) were lysed in buffer containing 1% Brij 35, 150 mM NaCl, 50 mM Tris-HCl, pH 7.5, or in buffer containing 1% NP-40, 10 mM EDTA in PBS, both supplemented with a mixture of protease inhibitors (Sigma) (13). Cells were incubated with the lysis buffer for 15 to 30 min on ice, cell debris were removed by centrifugation and the protein concentration in the supernatants was determined using the Bradford protein assay (Bio-Rad Laboratories, Hercules, CA). Samples of up to 100 µg total protein were subjected to SDS _ PAGE (10% acrylamide) under reducing conditions and proteins were transferred to a polyvinylidene difluoride membrane (Millipore Corporation, Billerica, MA). The membrane was probed with the appropriate antibody, followed by horseradish peroxidase-conjugated secondary antibody and a chemiluminescent substrate (Westone, iNtron Biotechnology, Gyeonggi-do, Korea) (13). For detection of heparanase, samples containg up to 1 mg total protein were first added to a mixture of 30 µl of heparin agarose beads and 30 µl Concanavalin A agarose beads (Sigma) and incubated with rotation for 2 h at room temperature, followed by overnight incubation at 4 0 C. Beads were then washed twice with saline, boiled (7 min) in sample buffer, and the supernatants were subjected to SDS-PAGE and immunoblotting, as described above.

Generation of heparanase deletion mutant.
Deletion mutant of human heparanase cDNA was constructed according to site-specific mutagenesis by the overlap extension approach (13). Two fragments were amplified in two separate PCR reactions using the full-length heparanase cDNA as a template. PCR 1 amplified the fragment that contains the deletion mutation together with the upstream sequence, utilizing a reverse primer containing the deletion mutation and a common forward primer containing the wild type heparanase sequence (1UERIhpa). PCR 2 amplified the DNA fragment that contains the deletion mutation together with downstream sequences, utilizing the forward primer containing the deletion mutation and a common reverse primer containing the wild type heparanase sequence (4LHpa). The two PCR purified fragments (JETquick gel extraction spin kit, Genomed, Lohne, Germany) were mixed, denatured, annealed and extended, and the product was amplified in a third PCR reaction, using the by guest on July 8, 2020 http://www.jbc.org/ Downloaded from 1UERIhpa and 4Lhpa primers. The PCR 3 product was subcloned into pcDNA3-Hepa (containing the full-length heparanase) digested with EcoR I and Afl II using the TaKaRa DNA ligation kit II (Takara Bio Inc, Shiga, Japan). The external primers are: 1UERIHpa, 5`-CTTCAGCATCTTAGCCGTCTTT-3` and 4LhHpa, 5`-GCAGCCAGGTGAATTCCCAAGAT-3`.
The reverse and forward primers for the mutation are: 5`-GGGCCATTCCAACCGTAA-3` and 5`-TTACGGTTGGAATGGCCC-3`, respectively. All PCR reactions were performed using the TaKaRa Ex Taq TM polymerase (TaKaRa Bio Inc.) and the following cycling conditions: 95 0 C for 3 min followed by 32 cycles of 96 0 C for 18 s, 60 0 C for 90 s, then an elongation step of 72 0 C for 70s (13).
siRNA Transfection. Anti human cathepsin L siRNA was designed according to Zheng X. et al. (34) from the human cathepsin L cDNA sequence AAGTGGAAGGCGATGCACAAC (91-111) and was synthesized by Dharmacon RNA technologies (Lafayette, CO). MDA-MB-435 breast cacrcinoma or JAR choriocarcinoma cells were seeded into 6 well plates and grown in 2.5 ml DMEM supplemented with 10% fetal calf serum. After 24 h in culture, MDA-MB-435 (20-30% confluent) cells were transfected with 25 µl of 20 µM stock solution of siRNA duplexes (0.2 µM final concentration) using GeneSilencer siRNA transfection reagent kit, according to the manufacturer's protocol (Genlantis, San Diego, CA). Forty eight hours later, the cells were retransfected with another dose of siRNA and incubated for additional 48 h. JAR cells were transfected with cathepsin L siRNA as described above and 24 h later were retransfected with siRNA plus pcDNA containing the full-length heparanase or empty vector. The cells were harvested by a rubber policeman, solubilized in lysis buffer containing protease inhibitors and subjected to Western blot analysis and heparanase activity assay, as described above.
Mass spectrometry. Recombinant, full-length proheparanase, or linker truncated 50+8 kDa linear heparanase (hepa-3GS), generated by replacing the linker segment with a spacer of three glycineserine pairs (15), were incubated in 100 mM Tris buffer, pH 6.8, supplemented with 8 mM EDTA and 4 mM DTT, for 80 minute at 37 0 C, with or without human liver derived cathepsin L (Sigma). The reaction mixtures were purified on HPLC converse-phase C-18 column and subjected to matrix-assisted laser desorption/ionization-time of flight (MALDI-TOF) mass analysis. Masses of peptides (P1-P13) determined by MALDI-TOF corresponded to the calculated masses (± 20 δ) of peptides derived from the linker segment and adjacent portions of the 8-kDa subunit, taking into account phosphorylation modifications on particular tyrosines, serines and threonines. The peptides (P1-P13), their corresponding sequences and calculated masses (see Figure 6) (including additional 80 δ for each phosphate group), were:

Results
Knockdown of cathepsin L markedly suppresses proheparanase processing. To investigate the role of cathepsin L in proheparanase processing, we utilized siRNA mediated silencing of cathepsin L (34) in JAR human choriocarcinoma cells capable of by guest on July 8, 2020 http://www.jbc.org/ Downloaded from proheparanase processing, but devoid of heparanase gene expression and enzymatic activity (13). Cells were first transfected with anticathepsin L siRNA, or were mock transfected. Twenty four hours later the cells were transiently co-transfected for an additional 24 h with a pcDNA-Hepa vector that encodes for the fulllength 65-kDa heparanase. The marked decrease (~75%) in the level of procathepsin L in the siRNA transfected cells (Fig. 1A) was accompanied by an almost complete inhibition of proheparanase processing into the 50-kDa heparanase subunit (Fig. 1B) and by a profound decrease in heparanase enzymatic activity, compared to mock transfected cells (Fig. 1C). The lack of detectable 50-kDa subunit, representing active heparanase, in cathepsin L siRNA transfected JAR cells (Fig. 1B), indicates that in this experimental system (i.e., cells devoid of endogenous heparanase activity) proteolytic enzymes other than cathepsin L do not play a significant role in heparanase processing and activation.
Next, the effect of siRNA mediated cathepsin L silencing on proheparanase processing was examined in MDA-MB-435 human breast carcinoma cells expressing high levels of endogenous heparanase (30). RT-PCR analysis of anti-cathepsin L siRNA transfected cells showed a 54% knockdown of procathepsin L mRNA compared to mock transfected cells ( Fig. 2A). Heparanase mRNA levels remained unchanged, excluding the possibility of an off-target effect of cathepsin L siRNA on heparanase mRNA. At the protein level, procathepsin L was reduced by 71% in cathepsin L-siRNA transfected vs. mock transfected cells as revealed by Western blot analysis (Fig. 2B) and densitometric analysis (not shown). Knockdown of cathepsin L resulted in a marked decrease (65%) in the level of the processed form of heparanase (represented by the 50-kDa subunit) as determined by Western blot and densitometric analysis (Fig.  2C). Consequently, heparanase enzymatic activity was markedly reduced (65%) in lysates obtained from cells transfected with anti-cathepsin L siRNA as compared to mock transfected cells (Fig. 2D). These results demonstrate that cathepsin L mediates the processing of proheparanase in MDA-MB-435 cells, although to a lesser extent than in JAR cells. This difference may be due to the long half life of the already processed active heparanase (35) that masks the actual decreased level of processing obtained during the short term treatment with cathepsin L siRNA, incomplete depletion of Cathepsin L, and/or the presence of additional cathepsin L-like proteases in MDA-MB-435 vs. JAR cells. Altogether, these results clearly demonstrate that cathepsin L mediates the processing of endogenous and acquired proheparanase in MDA-MB-435 and JAR carcinoma cells, respectively.
Cathepsin L knockout fibroblasts are incapable of proheparanase processing. Fibroblasts derived from cathepsin L knockout RT2 pancreatic islet tumors (25) were tested for their ability to process proheparanase in comparison with fibroblasts from wild type (WT) RT2 tumors (Fig. 3). WT fibroblasts expressed high levels of properly processed highly active heparanase, as revealed by Western blot analysis (Fig. 3B) and heparanase enzymatic activity (Fig. 3C). In contrast, cathepsin L knockout fibroblasts (Fig. 3A) expressed high levels of the full length 65-kDa pro-enzyme (Fig.  3B) and almost no heparanase enzymatic activity (Fig. 3C). Next, WT and cathepsin L knockout fibroblasts were stably transfected with pcDNA-Hepa. While in cathepsin L knockout fibroblasts the over-expressed heparanase accumulated as a 65-kDa latent proheparanase, in the WT fibroblasts proheparanase underwent proper processing into a highly active enzyme (Fig.  3D,E). In subsequent studies, heparanase enzymatic activity was examined in tissue extracts derived from WT and cathepsin L knockout mice. As demonstrated in Fig. 3F, samples (i.e., liver, spleen) derived from cathepsin L knockout mice exhibited very little or no detectable heparanase activity as opposed to high activity levels in corresponding samples derived from wild type mice. Collectively, these results emphasize the critical role of cathepsin L as the principal and likely sole protease responsible for proheparanase processing and activation.  (36,37) and heparanase (30). Semiquantitative RT-PCR demonstrated that MCF-7 cells stably transfected with pcDNA-Hepa express high levels of proheparanase mRNA (Fig. 4A) and protein (not shown), comparable to those found in wt MDA-MB-435 cells. Nevertheless, little or no significant increase in heparanase activity was detected in heparanase transfected MCF-7 cells (Fig. 4B) despite their high heparanase mRNA expression level (Fig. 4A). These results indicate that increased expression of heparanase is not sufficient to confer a comparable increase in heparanase enzymatic activity; rather that it is processing of the proheparanase protein which is limiting.
We next tested whether the high level of cathepsin L activity in MDA-MB-435 cells (36,37) is responsible for a more effective processing and generation of active heparanase in these cells than in MCF-7 cells. For this purpose, MCF-7 cells were transiently doubly transfected with vectors encoding for heparanase and cathepsin L, and compared to cells doubly transfected with heparanase and control empty vector. As depicted in Figure 4C, transfection with both heparanase and cathepsin L resulted in a marked increase in heparanase enzymatic activity as compared to MCF-7 cells transfected with heparanase alone. Notably, this activity was similar in magnitude to that observed with MDA-MB-435 cells (Fig. 4C). These results clearly demonstrate that the limiting factor in heparanasetransfected MCF-7 cells is cathepsin L, emphasizing again its critical involvement in proheparanase processing.
A ten amino acid peptide at the C-terminus of the linker segment is critical for proheparanase processing and activation. We have previously demonstrated that substitution of Tyr 156 with alanine (Y156A) altered the correct processing of proheparanase, resulting in generation of a 51-kDa inactive polypeptide instead of the characteristic 50-kDa subunit (13). We speculated that a point mutation (Y156A) in the potential cathepsin L motif Y 156 Q-at the C-terminus of the linker peptide will expose the next potential cleavage site (i.e., Y 146 Q-), along the linker segment (internal cleavage site) resulting in a 10 amino acid peptide (E148-Q157) which remains attached to the 50-kDa subunit and thereby blocks access to the enzyme's active site, as predicted by the 3D model of heparanase (13). To validate this mode of action, we generated a deletion mutant ∆E148-Q157 (Fig. 5A) that lacks the 10 amino acid peptide (E148-Q157) at the C-terminus of the linker which precedes the 50-kDa subunit, and consequently eliminates the cathepsin L cleavage site (Y 156 Q-) but still contains the remaining 38 amino acid portion of the linker (between Ser 110 -Gln 147 ), including the internal Y 146 Q-cathepsin L cleavage motif. To investigate the processing and enzymatic activity of this deletion mutant vs. the point mutated (Y156A) proheparanase, we transfected JAR cells with vectors encoding either the wild-type proheparanase, the deletion mutant ∆E148-Q157, or the Y156A mutant (Fig.  5A). Western blot analysis revealed that unlike the Y156A mutant which was improperly processed, yielding a 51-kDa subunit (Fig. 5B) with very low heparanase activity (Fig. 5C, о), similar to that of mock transfected cells (Vo), the deletion mutant ∆E148-Q157 was properly processed, yielding a 50-kDa subunit (Fig. 5B) and a highly active enzyme (Fig. 5C, •). Thus, while the Y156A point mutant resulted in an improperly processed (elongated 51-kDa subunit) and inactive heparanase due to the presence of a 10 amino acid peptide that blocks access to the active site, the ∆E148-Q157 mutated proheparanase (lacking the 10 amino acid peptide at the C-terminus of the linker segment) underwent proteolytic cleavage at an internal cleavage site (Y 146 Q), yielding a properly processed 50-kDa subunit and active heparanase. Therefore, these results indicate that activation of proheparanase requires proteolytic excision of the 10 amino acid peptide at the Cterminus of the linker segment. Moreover, these results demonstrate that the Y 146 Q-motif serves as a functional cathepsin L endocleavage site located within the linker segment upstream to the Y 156 Q site, yet could be identified only when the latter motif is mutated (Fig. 5B).
To further validate the functionality of these two sites (Y 156 Q-; Y 146 Q-) as cathepsin L cleavage motifs, JAR cells transfected with vectors encoding either the wt proheparanase or the Y156A mutant were grown in the presence or absence of the cell permeable cathepsin L inhibitor by guest on July 8, 2020 http://www.jbc.org/ Downloaded from I (Calbiochem, Merck, Darmstadt, Germany) and subjected to Western blot analysis (Fig. 5D). In the absence of the cathepsin L inhibitor the wt proheparanase was properly processed, while the Y156A mutant yielded the 51-kDa subunit (Fig.  5D), as expected. In contrast, in the presence of the inhibitor, processing was abolished at both sites resulting in accumulation of the 65-kDa proenzyme (Fig. 5D); further verifying that these sites are indeed cathepsin L cleavage motifs.
Since the intact linker segment could not be detected by pAb CKLE antibodies that specifically recognize the linker segment (not shown), we assumed that other endocleavage sites (apart from the Y 146 Q site) exist along the linker segment.
Accumulation of the 65-kDa proheparanase in the presence of cathepsin L inhibitor (Fig. 5D) further suggest susceptibility of the linker peptide to multiple endocleavages by cathepsin L, functioning as the primary protease responsible for removing the entire linker segment (see below).

MALDI-TOF analysis of proheparanase cleavage peptides generated by cathepsin L.
We have previously demonstrated that recombinant latent proheparanase is properly processed by cathepsin L in a cell free system, yielding an active enzyme composed of the 8-and 50-kDa subunits (13). In order to verify the accurate size of the 8-kDa heparanase subunit generated by cathepsin L and the fate of the linker segment, a reaction mixture consisting of recombinant proheparanase incubated with cathepsin L was subjected to MALDI-TOF mass spectrometry (MS) analysis. Incubation of the full length 65-kDa proheparanase with cathepsin L yielded several peptides of different masses, none of which was detected upon incubation of proheparanase in reaction buffer alone (not shown), indicating that these are cleavage products of proheparanase generated by cathepsin L. Five main masses (P1-P5) ranging from ~ 8 to ~ 9 kDa were revealed by this analysis (Fig. 6A). The determined mass of 8244.29 daltons (P1) is in agreement with the expected calculated mass of the full length 8-kDa subunit (Gln 36 -Glu 109 ) including Gln 36 that resides after cleavage of the signal peptide at the consensus site Gln 34 -Ala-Gln 36 (16). This result indicates that cathepsin L yields the precise 8-kDa subunit. Other masses were of 8624.76 (P2), 8771.58 (P3), 8917.89 (P4) and 9410.66 (P5) daltons (Fig. 6A), in agreement with masses that correspond to Gln 36 -Phe 112 , Gln 36 -Glu 113 , Gln 36 -Glu 114 and Gln 36 -Tyr 117 , respectively (see Materials and Methods). These are species of the 8-kDa subunit with extended carboxy termini, likely generated by several endocleavages that take place upstream to site 1 within the N-terminus of the linker segment. The occurrence of multiple adjacent endocleavages upstream to site 1 is in agreement with sitedirected mutagenesis indicating tolerance for point mutation substitutions in this site and its flanking regions, yielding an active enzyme (13). Thus, it appears that abolishing the exact endocleavage at site 1 can be compensated by adjacent upstream endocleavages, yielding a slightly elongated 8-kDa subunit that is still functional.
MS analysis revealed that in the presence of cathepsin L, shorter heparanase peptides (~2 to ~5 kDa) were generated, in addition to the principal peptides (P1 -P5) corresponding to the 8-kDa subunit (see above). Attenuated processing of recombinant proheparanase yielded up to 8 peptides (Fig. 6B)  Notably, when recombinant single chain heparanase in which the linker segment was replaced by 3 glycine-serine residues (hepa-3GS) (15) was incubated with cathepsin L, none of these peptides were detected by mass spectrometry (not shown). Altogether, the MS analysis suggests that cathepsin L acts at multiple endocleavage sites along the linker segment, thereby removing it in small fragments rather than as an intact peptide. This observation is supported by the observation that the linker segment could not be detected by an antibody (α-CKEL) that specifically recognizes the linker peptide in the context of the intact proenzyme (not shown). Notably, determination of a particular MS peptide sequence took into account the occurrence of phosphorylation modification (additional 80 δ) on particular tyrosine or serine residues along the linker (Fig.  7), consistent with our observation that the linker by guest on July 8, 2020 http://www.jbc.org/ Downloaded from segment is preferentially phosphorylated on serine and tyrosine residues (our unpublished results).
Proposed mode of action. The 13 peptides identified by the MS analysis were aligned on the basis of overlapping sequences (Fig. 7). Interestingly, five putative phosphorylation sites (Ser 110 , Ser 116 or Tyr 117 , Tyr 129 or Ser 131 , Tyr 146 and Tyr 156 ) are in accordance with the overlap among the different peptides, supporting their existence (Fig. 7). This alignment highlighted seven internal cleavage sites with characteristic cathepsin L cleavage motifs, all containing an aromatic or hydrophobic amino acid at positions P2 or P3 of the cleavage site (Fig. 7). Altogether, the mass spectrometry data suggest that the entire linker segment can be removed solely by cathepsin L through multiple endocleavages.

Discussion
Cathepsin L is a characteristic lysosomal cysteine proteinase of the papain superfamily of peptidases (18) that functions primarily as an endoprotease in the lysosome. In addition, the enzyme has been implicated in multiple physiological and pathological processes (18). Applying gene silencing, knockout and overexpression approaches, we demonstrate that alterations in the endogenous cellular levels of cathepsin L markedly affect the processing of proheparanase in carcinoma cells and immortalized fibroblasts. Of particular significance is the lack of detectable processing and activation of the latent 65-kDa enzyme in JAR cells co-transfected with the full-length latent enzyme and anti-cathepsin L siRNA (Fig. 1), and in tissues and cultured fibroblasts derived from cathepsin L knockout mice (Fig. 3E, F), suggesting that proheparanase cannot be properly processed and activated by proteolytic enzymes other than cathepsin L.
Unlike JAR cells, MDA-MB-435 breast carcinoma cells express high levels of endogenous heparanase. The lack of complete inhibition of heparanase processing and enzymatic activity in MDA-MB-435 cells transfected with anticathepsin L siRNA (Fig. 2) may be attributed to the exceedingly long half life (30 h) of the endogenous active enzyme in these cells (35). Alternatively, the presence of additional cathepsin L-like proteases in MDA-MB-435 cells compared to JAR cells could contribute to this difference. The critical role of cathepsin L in proheparanase processing in breast carcinoma cells is further supported by the finding that over-expression of cathepsin L markedly increased the processing and activation of proheparanase when over-expressed in co-transfected MCF-7 cells, which otherwise express low endogenous levels of the proenzyme (Fig. 4B). By contrast, MCF-7 cells overexpressing the 65-kDa proenzyme alone, exhibited a very low heparanase enzymatic activity. Interestingly, MCF-7 cells exhibit a low cathepsin L activity primarily due to their high content of endogenous cathepsin L inhibitors (36,37). This balance may be altered upon cathepsin L overexpression. In contrast, MDA-MB-435 cells express relatively low levels of endogenous cathepsin L inhibitors (36,37). and hence proheparanase is readily processed and activated by these cells. Altogether, our results indicate that cathepsin L is the principal and most likely the sole enzyme responsible for processing and activation of proheparanase.
Applying deletion and point mutations, we show that a 10 amino acid peptide at the Cterminus of the linker segment, located between the two functional cathepsin L cleavage sites Y 156 Q-and Y 146 Q-, abolished heparanase activity and that its removal by cathepsin L is critical for proheparanase activation. These results are consistent with the predicted 3D model of heparanase, demonstrating that a 10 amino acid peptide at the C-terminus of the linker is sufficient for blocking the accessibility of the substrate (heparan sulfate) to the active site of heparanase (13). This peptide can therefore be applied as a prototype to design peptides and peptidomimetics that inhibit heparanase enzymatic activity.
According to mass spectrometry analysis (Fig. 7), processing of recombinant proheparanase with cathepsin L generated the precise 8-kDa subunit (P1) plus four slightly elongated forms (P2, P3, P4 and P5), suggesting the occurrence of five endocleavage sites targeted by cathepsin L upstream of site 1 (Glu 109 ) at Phe 111 , Glu 112 , Glu 113 and Tyr 117 , respectively (Figs. 6 & 7). The feasibility of multiple adjacent endocleavages upstream of site 1 is in agreement with site-directed mutagenesis indicating tolerance for point mutations substitutions in this site and its flanking region, yielding an active enzyme (13). Thus, it appears that abolishing the primary endocleavage at site 1 can be compensated by adjacent upstream endocleavages, yielding a slightly elongated 8-kDa subunit that is still functional. Moreover, mass spectrometry analysis (Fig. 6C) revealed up to 8 peptides (P6-P13) all originating from the linker segment, indicating that cathepsin L attacks at multiple endocleavage sites along the linker segment (Figs. 6 & 7), namely, Trp 118 , Gln 119 , Gly 130 , Leu 140 , Glu 143 , Tyr 146 , Gln 147 , Glu 148 , and Tyr 156 , suggesting that the linker peptide is not removed as an intact segment. In support of this hypothesis is the observation that an intact linker segment could not be detected by Western blot analysis (not shown).
Among the eleven endocleavage sites revealed by the MS analysis (Figs. 6 & 7), nine are typical recognition motifs of cathepsin L which preferentially include bulky (aromatic or hydrophobic) amino acids at the P2 or P3 position of the cleavage sites (38,39). Y 156 Q-, Y 146 Q-(aromatic acid (Tyrosine) at P2 of the cleavage), Y 146 QE-(tyrosine at P3 of the cleavage), L 143 E-(hydrophobic acid (Leucine) at P2 of the cleavage), Y 129 G-(tyrosine at P2), W 118 Q-(tryptophane at P2) or YWQ-(tyrosine at P3) and F 111 E-(Phenyl-alanine at P2) or F 111 EE-(Phenyl alanine at P3). A similar cleavage pattern was noted for processing of proenkephalin by cathepsin L at particular mono or dibasic amino acids where an aromatic acid is located at P2 or P3 of the cleavage site (21). The other cleavages at Glu 109 -(site 1), Phe 111 -, Tyr 117 -, Leu 140 -and Tyr 146 do not exhibit typical cathepsin L motifs. Although atypical cleavages by cathepsin L have been reported (39), an alternative explanation is that the Glu 109 , Phe 111 , Tyr 117 , Leu 140 and Tyr 146 may be regarded as the C-terminus of peptides generated by endocleavage at an adjacent upstream typical motif (Fig. 7) which is subsequently blunted by the very weak carboxypeptidase activity of cathepsin L itself (40), in a stepwise manner. Thus, the 8-kDa subunit and four other peptides derived from the linker (e.g., P2, P5, P10 and P12) may be generated by either direct endocleavage at atypical sites, or by endocleavage at cathepsin L typical sites followed by a carboypeptidase activity at the peptide C-terminus. The mass spectrometry analysis also demonstrated that the linker segment is preferentially phosphorylated on serine and tyrosine residues, suggesting their possible regulatory role in heparanase processing through an effect on the accessibility of particular cathepsin L cleavage sites, as reported for HSP-70 (41).
The following observations have enabled us to determine herein the detailed mechanism of proheparanase processing. Applying MDA-MB-435 and MCF-7 breast carcinoma cells we have demonstrated that proheparanase processing is mediated by cathepsin L. Moreover, applying JAR cells devoid of endogenous heparanase, and cathespin L knockout fibroblasts and tissues, we showed that proheparanase processing is brought about by cathepsin L as the primary and possibly sole protease. Applying site-directed mutagenesis we have demonstrated that proper proheparanase processing and activation requires proteolytic removal of a 10 amino acid peptide located between two functional cathepsin L cleavage sites, Y 156 Q-at the C-terminus of the linker segment and Y 146 Q-inside the linker segment, emphasizing the key role of cathepsin L in proheparanase processing as it removes the most critical part of the linker segment and exposes the heparanase active site. Moreover, mass spectrometry analysis revealed that the entire linker segment is susceptible to multiple cleavages by cathepsin L and that the 8-kDa subunit can be generated by several alternative adjacent endocleavages, yielding the precise 8-kDa subunit or slightly elongated, but still active forms. Altogether, these results indicate that proper processing and activation of proheparanase can be brought about solely by cathepsin L.
Clearly, a better understanding of the mechanism of proheparanase processing requires elucidation of heparanase and cathepsin L cellular trafficking in order to more accurately identify the cellular site of processing. We have noticed that despite incomplete cathepsin L gene silencing in JAR cells (Fig. 1A), processing of proheparanase was fully abrogated (Fig. 1B,C), suggesting that high levels of cathepsin L are needed for proheparanase processing, as previously demonstrated for other physiological functions of cathepsin L (42,43). Over-expression of cathepsin L and other lysosomal proteases affects their sorting from the lysosomes and subsequent targeting into specific granules or secretory vesicles, where they perform a particular by guest on July 8, 2020 http://www.jbc.org/ Downloaded from physiological function. For example, in transformed fibroblasts, up-regulated procathepsin L is targeted into dense core multi-vesicular bodies (MVBs), also known as secretory lysosomes. These dense bodies are generated by fusion of vesicles of endosomal origin with either lysosomes or the plasma membrane where they are secreted as single vesicles termed exosomes (44). Our preliminary results suggest that heparanase and cathepsin L are most probably co-localized in MVBs (not shown).
The involvement of cathepsin L in tumor progression through proteolytic degradation of structural constituents of the ECM is well documented (25)(26)(27). Our results indicate that the tumor promoting effect of cathepsin L may be due, at least in part, to its role in proheparanase processing. Elucidation of heparanase trafficking and secretion may offer new tools for suppressing the proangiogenic and prometastatic properties of heparanase. Figure 1: Knockdown of procathepsin L inhibits processing of exogenous proheparanase by JAR choriocarcinoma cells. Human choriocarcinoma JAR cells, devoid of endogenous heparanase, were subjected to two sequential transfections with 2 µM anti-procathepsin L siRNA at a 48 h interval (siCat L), or were mock transfected (Control). 72 h after the first transfection, cells were transiently transfected with a pcDNA plasmid encoding the full-length heparanase. 24 h later the cells were lysed and subjected to Western blot analysis of A. procathepsin L, and B. heparanase, as described in "Materials and Methods". C. Heparanase activity. Lysates of siCat L (о) and mock (•) transfected cells (both also transfected with the full-length heparanase) were lysed and incubated (7 h, 37 o C, pH 6.0) with sulfate labeled ECM. Sulfate labeled degradation fragments released into the incubation medium were analyzed by gel filtration on Sepharose 6B, as described under "Materials and Methods".  Fibroblasts derived from either cathepsin L knock-out RT2 tumors (KO, о) or wild-type RT2 tumors (WT, •) were lysed (1% NP-40, 10 mM EDTA in PBS supplemented with protease inhibitors) and subjected for Western blot analysis of A. cathepsin L (39, 26 and 21 kDa forms) and α-tubulin; and B. heparanase, as described in "Materials and Methods". C. Cell lysates were also analyzed for heparanase activity assay, as described in "Materials and Methods". D,E. Cathepsin L knockout fibroblasts (KO, □) and wild type fibroblasts (WT, ) were also stably transfected with heparanase and subjected to Western blotting for heparanase (D) and determination of heparanase enzymatic activity (E). F. Tissues (liver, о, •; spleen ∆, ) derived from cathepsin L knockout (•, ) or wt (о, ∆) mice were homogenized in lysis buffer and the supernatant fraction (500 µg) was analyzed (37 o C, 18 h, pH 6.0) for heparanase enzymatic activity, as described in Figure 1.   The primary sequence of the linker segment is shown, flanked by the 8-kDa subunit at the N-terminus, and by the 50-kDa subunit at the Cterminus. Sequences of peptides 1 to 13 established by mass spectrometry (Fig. 7) were overlapped, highlighting nine typical cathepsin L endocleavage sites ( ) taking place at the first or second amino acid upstream to a bulky amino acid (bold), and five atypical cathepsin L endocleavage sites ( ) (C-terminus of Ser 109 , Phe 111 , Tyr 117 , Leu 130 and Tyr 146 ). PO 4 indicates predicted phosphorylated sites. Phosphorylation of the underlined amino acids in peptides 1-13 was taken into account in calculating their molecular mass.  Abboud-Jarrous et al. Fig. 7