After diabetes, the heart has a singular reliance on fatty acid (FA) for energy production, which is achieved by increased coronary lipoprotein lipase (LPL) that breaks down circulating triglycerides. Coronary LPL originates from cardiomyocytes, and to translocate to the vascular lumen, the enzyme requires liberation from myocyte surface heparan sulfate proteoglycans (HSPGs), an activity that needs to be sustained after chronic hyperglycemia. We investigated the mechanism by which endothelial cells (EC) and cardiomyocytes operate together to enable continuous translocation of LPL after diabetes. EC were cocultured with myocytes, exposed to high glucose, and uptake of endothelial heparanase into myocytes was determined. Upon uptake, the effect of nuclear entry of heparanase was also investigated. A streptozotocin model of diabetes was used to expand our in vitro observations. In high glucose, EC-derived latent heparanase was taken up by cardiomyocytes by a caveolae-dependent pathway using HSPGs. This latent heparanase was converted into an active form in myocyte lysosomes, entered the nucleus, and upregulated gene expression of matrix metalloproteinase-9. The net effect was increased shedding of HSPGs from the myocyte surface, releasing LPL for its onwards translocation to the coronary lumen. EC-derived heparanase regulates the ability of the cardiomyocyte to send LPL to the coronary lumen. This adaptation, although acutely beneficial, could be catastrophic chronically because excess FA causes lipotoxicity. Inhibiting heparanase function could offer a new strategy for managing cardiomyopathy observed after diabetes.
In diabetes, because glucose uptake and oxidation are impaired, the heart is compelled to use fatty acid (FA) exclusively for ATP generation (1). Multiple adaptive mechanisms, either whole-body or intrinsic to the heart, operate to make this achievable, with hydrolysis of triglyceride-rich lipoproteins being the major source of FA to the diabetic heart (2). This critical reaction is catalyzed by the vascular content of lipoprotein lipase (LPL), and we were the first to report significantly higher coronary LPL activity after diabetes (3). In the heart, LPL is synthesized by cardiomyocytes, transported to heparan sulfate (HS) proteoglycan (HSPG) binding sites on the myocyte surface, and from this temporary reservoir, the enzyme is transferred across the interstitial space to reach endothelial cells (EC) (4,5). Before this transfer, liberation of HSPG-sequestered LPL is a prerequisite and is facilitated by heparanase, an EC endoglycosidase that can cleave HS side chains on HSPGs in the extracellular matrix and on the cell surface to release bound proteins (6).
Heparanase is synthesized as a latent 65-kDa precursor. After its secretion and reuptake (7), heparanase enters the lysosome and is cleaved into a 50-kDa active form by cathepsin L (8,9). Active heparanase is stored in this organelle until secreted. In addition to an extracellular function, by entry into the nucleus to regulate histone acetylation/methylation, heparanase is also capable of modulating gene transcription (10,11). Thus, tumor cells, by expressing higher levels of heparanase, are more invasive because secreted heparanase can break down extracellular matrix in addition to promoting gene expression related to an invasive phenotype (12). For example, matrix metalloproteinase-9 (MMP-9) gene expression induced by nuclear heparanase can further degrade extracellular matrix and basement membranes (13). In addition, MMP-9 also causes accelerated shedding of HSPGs, such as syndecan-1 (14,15), thereby liberating HSPG-bound ligands, including angiogenic and growth factors. At present, whether heparanase can influence gene transcription in the heart is unclear.
In the heart, heparanase is synthesized predominantly in EC, and after diabetes, we reported a high-glucose–induced secretion of active heparanase (16). That high glucose also stimulated secretion of latent heparanase, whose function in the heart is unknown, should be noted. Interestingly, human fibroblasts that do not express heparanase can take up exogenous latent heparanase and convert it into active heparanase (17). Thus, an immediate response to hyperglycemia is an EC secretion of active heparanase to facilitate translocation of LPL from the myocyte surface to the coronary lumen (18). Given that the active heparanase pool in EC is limited and may not meet the demand for a sustained LPL transfer after diabetes, it is possible that latent heparanase could be taken up and converted to active heparanase in cardiomyocytes. Such an effect could turn on MMP-9 expression, thereby stimulating shedding of HSPGs that bind LPL and thus maintaining vascular LPL increase chronically. Results from the present studies suggest that high-glucose–induced secretion of EC heparanase can regulate gene expression of MMP-9 in cardiomyocytes, an effect that may sustain transfer of LPL to the coronary lumen after chronic diabetes.
Research Design and Methods
This investigation conformed to the Guide for the Care and Use of Laboratory Animals published by National Institutes of Health and the University of British Columbia.
Male Wistar rats (250–320 g) were injected intravenously with 55 mg/kg streptozotocin (STZ). These animals (D55) were kept for 4 days before isolation of cardiomyocytes. To induce acute hyperglycemia, diazoxide (100 mg/kg, intraperitoneally) was injected in male Wistar rats (250–320 g), and they were kept for 4 h.
Rat aortic EC (RAOEC) were obtained from Cell Applications. STZ, filipin, genistein, and chloroquine were obtained from Sigma-Aldrich. Heparin (HEPALEAN; 1,000 units/mL) was from Organon Canada, Ltd. Heparinase III (IBEX Technologies) was purified from recombinant Flavobacterium heparinum. Purified latent heparanase was prepared as described (19). Small interfering (si)RNA for rat MMP-9 was obtained from Qiagen (SI02004247). Control siRNA consisted of a scrambled sequence purchased from Santa Cruz Biotechnology. Lipofectamine RNAiMAX was from Invitrogen. Purified MMP-9 and p-aminophenylmercuric acetate (AMPA) were from RayBiotech (Norcross, GA). Anti-LPL 5D2 antibody was a gift from Dr. J. Brunzell, University of Washington, Seattle. Anti-heparanase antibody mAb 130, which recognizes the active (50 kDa) and latent (65 kDa) forms of heparanase, was from InSight (Rehovot, Israel). An antibody that recognizes the cleaved form of HSPGs was purchased from Seikagaku (Tokyo, Japan). Rat MMP-9 antibody was from Millipore. All other antibodies were obtained from Santa Cruz Biotechnology. A rat syndecan-1 ELISA kit was obtained from MyBioSource (San Diego, CA). A histone acetyltransferase (HAT) activity kit was obtained from BioVision (Milpitas, CA).
Isolation of Cardiomyocytes
Ventricular calcium-tolerant myocytes were prepared by a previously described procedure (20). Cardiomyocytes were isolated from adult male Wistar rats (250–320 g), were plated on laminin-coated culture dishes, and allowed to settle for 3 h. Before treatment, unattached cells were washed away using Medium 199.
Coronary LPL Activity
To measure coronary LPL, hearts were perfused retrogradely with heparin (5 units/mL) (3). Coronary effluents were collected (for 10 s/fraction) at different time points over 5 min, and LPL activity in each fraction was determined by measuring the in vitro hydrolysis of a sonicated [3H]triolein substrate emulsion (3). Coronary LPL activity was calculated as the area under curve of LPL activity in each fraction over time.
RAOEC were cultured at 37°C in a 5% CO2 humidified incubator alone or cocultured with adult rat cardiomyocytes. RAOEC from the fifth to the eighth passage were used.
To test whether exogenous heparanase can bind and be taken up by cardiomyocytes, isolated myocytes were treated with 500 ng/mL recombinant latent heparanase for different intervals at 37°C. This experiment was also conducted at 15°C to inhibit heparanase uptake. In the presence of 10 IU/mL heparin to compete for HSPGs binding sites, or 10 IU/L heparinase III to digest HS side chains (2 h), the contribution of myocyte surface HSPGs in heparanase uptake was determined. For studying the mechanism of internalization, 350 mmol/L sucrose or 1 μg/mL filipin was applied to myocytes for 15 min before latent heparanase to block clathrin-coated pits and caveolae-dependent internalization, respectively (21,22). Dynasore (0–25 μmol/L) or genistein (0–100 μmol/L) for 1 h were also used to inhibit dynamin and tyrosine kinase activity, respectively, which could be involved in internalization of exogenous heparanase (23,24). Chloroquine (200 μmol/L) was used to inhibit lysosomal enzymes, and myocytes were incubated for 2 h. Anacardic acid (10 μmol/L) was used to inhibit HAT activity. Purified MMP-9 proenzyme was activated by 10 μmol/L AMPA at 37°C for 1 h (25) and added into the myocyte culture medium at a final concentration of 1, 5, and 10 ng/mL for 30 min to validate the effect of MMP-9 on syndecan-1 shedding and LPL release from the cardiomyocyte surface. To knockdown MMP-9 expression, cardiomyocytes were cultured in a six-well plate and transfected with 100 pmol siRNA using lipofectamine.
Uptake of Exogenous Heparanase
Myocytes were incubated with 500 ng/mL purified latent heparanase and washed three times with cold PBS. To determine binding and uptake of heparanase by myocytes, total cell lysates were collected to detect heparanase by Western blot. For measuring internalization of heparanase, plasma membrane was removed by a procedure described previously (24). Briefly, myocytes were lysed in 50 mmol/L Tris-HCl (pH 7.5), 150 mmol/L NaCl, 0.2 mol/L sucrose, 2 mmol/L EDTA, 2 mmol/L EGTA, and protease inhibitor cocktail (Roche) and centrifuged at 10,000g at 4°C for 10 min. The supernatant was further centrifuged at 100,000g at 4°C for 1 h to spin down the plasma membrane. The resulting supernatant (devoid of plasma membrane) containing the cytosolic fraction was used to monitor heparanase internalization (24).
Total RNA was extracted from isolated myocytes or RAOEC using Trizol reagent (Invitrogen), and 1 µg total RNA was used for RT-PCR. PCR primers were as follows (26,27):
Heparanase (forward, 5′-CAAGAACAGCACCTACTCACGAAGC-3′; reverse, 5′-CCACATAAAGCCAGCTGCAAAGG-3′; 616 bp product);
MMP-9 (forward, 5′-CCCCACTTACTTTGGAAACGC-3′; reverse, 5′-ACCCACGACGATACAGATGCTG-3′; 686 bp product);
18S rRNA (forward, 5′-CGGCTACCACATCCAAGGAA-3′; reverse, 5′-GCTGGAATTACCGCGGCT-3′; 187 bp product).
Isolation of Lysosomes and Nuclear Fractions
Lysosome-enriched fractions were isolated using a kit from Sigma-Aldrich. Nuclear and cytosolic fractions were separated using the nuclear/cytosol fractionation kit from BioVision. To validate the purity of proteins, we used cytosolic (GAPDH) and nuclear (histone H3) protein markers to detect their presence in cytosolic and nuclear fractions.
Western blotting was done as described previously (28). In some experiments, cell culture media was concentrated by trichloroacetic acid precipitation or Amicon centrifuge filter (Millipore) before detection of heparanase or LPL.
Values are means ± SE. Wherever appropriate, one-way ANOVA, followed by the Bonferroni test, was used to determine differences between group mean values. The level of statistical significance was set at P < 0.05.
Heparanase Present in Cardiomyocytes Originates Predominantly From an Exogenous Source
In the heart, heparanase is expressed predominantly by EC. Undeniably, our RT-PCR results indicated that heparanase mRNA was abundant in RAOEC when total RNA from EC was used as a template. However, when the same amount of total RNA was used, heparanase mRNA was not measurable in freshly isolated cardiomyocytes, implying low transcription activity (Fig. 1A, right panel). Regardless of this indeterminable gene expression, the 65-kDa latent and the 50-kDa active forms of heparanase protein were both detected in isolated myocytes (Fig. 1A, left panel). We confirmed that the presence of heparanase in cardiomyocytes was not due to EC contamination because we failed to detect the EC marker CD31 in this preparation (Fig. 1A, left panel). Interestingly, unlike RAOEC, the dominant heparanase present in cardiomyocytes was the active form (Fig. 1A, left panel). Culturing cardiomyocytes for 36 h reduced the content of both heparanase forms, suggesting that the protein cannot be efficiently synthesized de novo when being turned over (Fig. 1B). Introducing latent heparanase into the culture medium rapidly increased the level of latent heparanase in the total cell lysate of myocytes (within 5 min), an effect that continued over time. This increase was inhibited when the temperature was lowered to 15°C. Unlike latent heparanase, the increase of active heparanase was gradual and only significant after 4 h (Fig. 1C), suggesting an intracellular conversion of latent to active heparanase in cardiomyocytes. Our data imply that the heparanase content in cardiomyocytes originates largely from uptake of exogenous protein.
Internalization of Latent Heparanase by Cardiomyocytes Is Through a Caveolae-Dependent Pathway That Requires HSPGs, Dynamin, and Tyrosine Kinase Activation
Heparanase in the total cell lysates could consist of two parts: heparanase bound to the myocyte surface or that which had been internalized. Samples devoid of plasma membrane were prepared by ultracentrifugation and validated by the absence of membrane protein Na+-K+ ATPase (Fig. 2A, right panel). After incubation of myocytes with latent heparanase, a robust increase of this exogenous protein was observed in the plasma membrane free fraction, indicating that it had been internalized (Fig. 2A). Using heparin to competitively inhibit the binding of latent heparanase to HSPGs, we were able to reduce the amount of heparanase that was internalized (Fig. 2A). Because similar results were seen with heparinase III, which digests the HS chains of HSPGs (Fig. 2B), our data imply that exogenous heparanase may be taken up by myocytes by binding to the HS on the cell surface. Downstream, HSPG-dependent internalization can occur through caveolae or clathrin-coated pits. Because the internalization of latent heparanase was blocked by filipin, but insensitive to sucrose (Fig. 3A), HSPGs-mediated endocytosis of heparanase is likely a caveolae-dependent rather than a clathrin-mediated event. It should be noted that at the concentration of sucrose used, the endocytosis of epidermal growth factor receptor (EGFR), which is typically internalized through formation of clathrin-coated pits (29), was blocked (Fig. 3A, right panel). The involvement of dynamin and tyrosine kinase in this endocytotic process was apparent because dynasore (Fig. 3B) and genistein (Fig. 3C) reduced the amount of heparanase internalized.
Internalized Latent Heparanase Is Activated in Lysosomes and Enters the Nucleus
On incubation of cardiomyocytes with latent heparanase, we were able to detect this enzyme in lysosomal fractions within 30 min. With time, latent heparanase in this lysosomal fraction progressively increased. Unlike latent heparanase, lysosomal active heparanase only increased at a later time (after 3 h; Fig. 4A). Because this increase in lysosomal active heparanase was prevented by chloroquine, which inhibits lysosomal proteases, our data suggest lysosomal conversion of latent to active heparanase in cardiomyocytes (Fig. 4B).
Nuclear Entry of Heparanase Is Accompanied by Increased MMP-9 Expression
Assuming that conversion of latent to active heparanase in the lysosomes is to fulfill a biological function, we determined the nuclear content of heparanase and observed the presence of active heparanase in this organelle even in the absence of exogenous heparanase addition. However, after the addition of latent heparanase into the culture medium, there was a dramatic increase of latent and active heparanase in the nucleus after 4 h (Fig. 5A). In agreement with previous studies (13), addition of active heparanase to nuclear fractions of cardiomyocytes increased HAT activity (Fig. 5B). Importantly, nuclear entry of active heparanase was associated with an increase in MMP-9 expression 18 h after heparanase treatment (Fig. 5C). This increase in MMP-9 could be inhibited using the HAT activity inhibitor anacardic acid (Fig. 5D). Using purified MMP-9, this sheddase was capable of detaching syndecan-1 from the myocyte surface (Fig. 6A, upper right panel), together with LPL (Fig. 6A, lower right panel). In line with its augmented expression, MMP-9 secretion into the medium also increased after latent heparanase incubation (Fig. 6A), which was accompanied by an accelerated shedding of syndecan-1 (Fig. 6B) and release of HSPG-bound LPL (Fig. 6C). Because knockdown of MMP-9 effectively damped these effects (Fig. 6B and C), our data suggest that the heparanase/MMP-9 axis may play an important role in facilitating LPL translocation from cardiomyocytes to EC.
Increased Coronary LPL Activity After Diabetes Is Related to Endothelial Heparanase-Induced MMP-9 Expression in Cardiomyocytes
In the heart, heparanase is predominantly expressed in EC. As we previously reported (18), glucose stimulated secretion of the latent and active forms of heparanase from EC (Fig. 7A). In addition, coculturing of cardiomyocytes with EC in the presence of high glucose increased the amount of latent heparanase in cardiomyocytes. Because this effect was absent when EC were removed from the top chamber (Fig. 7B), our data suggest that cardiomyocytes can take up endothelial-derived heparanase, a process accelerated by high glucose. We reasoned that after prolonged hyperglycemia, such an uptake would result in augmented conversion of latent to active heparanase, together with its nuclear entry. Indeed, cardiomyocytes from D55 animals demonstrated a 1.5-fold increase in the nuclear content of active heparanase (Fig. 7C) that was associated with cleavage of nuclear HSPGs (Supplementary Fig. 1A). Notably, similar to our in vitro observations, MMP-9 expression (Fig. 7D and Supplementary Fig. 1B) and shedding of syndecan-1 (Fig. 7E) in D55 myocytes was significantly higher compared with the control. Because increased coronary LPL activity was observed in D55 hearts (Fig. 7F), our data imply that EC promote sustained translocation of LPL from cardiomyocytes to the coronary lumen by a heparanase/MMP-9 axis.
To maintain its energy supply after impaired glucose utilization during diabetes, the heart has a higher demand for FA, and much of it is provided by accelerated LPL-mediated hydrolysis of plasma triglyceride at the coronary lumen. Although functional at the apical side of EC, heart LPL is synthesized in cardiomyocytes, resides at a temporary reservoir on myocyte surface HSPGs, and is fast-tracked to the vascular lumen by high-glucose–induced release of endothelial heparanase, which cleaves these HSPGs (30). Results from the current study suggest that in addition to this rapid adaptation, cardiomyocytes can take up endothelial heparanase, which activates gene expression to maintain LPL trafficking during chronic hyperglycemia (Fig. 8).
Of the multiple cell types in the heart, the EC is likely the major source of heparanase. Surprisingly, in the presence of undetectable heparanase gene expression in cardiomyocytes, we detected protein in these cells that was predominantly active heparanase. Given that the half-life of active heparanase is ∼30 h (7) and that in vitro culturing of cardiomyocytes for this time decreased its heparanase content, our data suggest that the presence of heparanase in cardiomyocytes is not reliant on de novo synthesis but likely due to extracellular uptake. We confirmed this process using exogenous latent heparanase and observed that the uptake was rapid and likely HSPG-dependent. In CHO cells, latent heparanase triggers clustering of both syndecan-1 and -4, followed by rapid internalization of the heparanase-HSPG complex (31). Because we have previously reported clustering of syndecan-4 when cardiomyocytes are exposed to latent heparanase (18), it is possible that such a heparanase-HSPG complex is also responsible for endocytosis of this enzyme in cardiomyocytes. Nevertheless, we cannot rule out the contributive effects of other receptors, such as LDL receptor–related proteins and mannose 6-phosphate receptors, in this heparanase-uptake process (32). Of the two novel endocytotic pathways, HSPGs are involved more with a caveolae-dependent rather than clathrin-mediated endocytosis (33,34). For example, syndecan-1 has been shown to mediate apolipoprotein E-VLDL uptake in the human fibroblast cell line GM00701, a process that was clathrin-independent, but inhibited by nystatin, an inhibitor of the lipid raft–caveolae endocytosis pathway (35). Our data demonstrate that internalization of latent heparanase by cardiomyocytes relies on caveolae to form endocytotic vesicles and dynamin for fission of these vesicles from the plasma membrane. The participation of tyrosine kinase in this process has also been reported in the syndecan family–mediated endocytosis (36,37). Clustering is believed to lead to redistribution of syndecans to lipid rafts, where their cytoplasmic domain can be phosphorylated by tyrosine kinase (34,38). Recruitment of cortactin to phosphorylated syndecans promotes actin polymerization at the actin cortex, thus bringing HSPGs-mediated endocytotic vesicles into the cells (39). Collectively, our data imply that exogenous latent heparanase enters cardiomyocytes through the HSPG-dependent caveolae pathway.
Comparable to cardiomyocytes, cells such as fibroblasts, which do not express heparanase, can also take up latent heparanase (17). Interestingly, in fibroblasts, this latent heparanase can be converted into active heparanase, likely through lysosomal processing (8). Our results also indicate a lysosome enzyme–dependent conversion of latent to active heparanase in cardiomyocytes. In EC, active heparanase stored in lysosomes can be secreted in response to high glucose or enter the nucleus in the presence of FA (40). Data from the current study demonstrate cardiomyocyte nuclear entry of both latent and active heparanase. Two potential nuclear localization signals (residues 271–277; PRRKTAK and residues 427–430; KRRK) are present in the human heparanase sequence, which could directly mediate its nuclear entry (41). Additionally, Hsp90 has been proposed as a chaperone that mediates nuclear translocation of heparanase (42). However, the exact mode of heparanase translocation is currently unknown but of crucial importance given that nuclear protein transport, especially proteins that serve as transcriptional regulators, have been implicated in the pathogenesis of certain diseases (43). In HL-60 cells (42) and bovine coronary artery EC (40), active heparanase in the nucleus, by regulating gene expression, influences cell differentiation and glucose metabolism, respectively. Likewise, nuclear entry of active heparanase in myocytes was accompanied by an increase in MMP-9 gene expression. It is possible that by cleaving nuclear HSPGs, active heparanase relieves the suppression that HSPGs have on HAT activity, leading to acetylation of histone proteins required to promote gene transcription of MMP-9 (11,13). This effect has been observed in several cancer cell lines upon transfection of heparanase (13,44). Because MMP-9 is capable of shedding syndecan-1 and -4 to generate soluble ectodomains carrying HSPGs-bound ligands (45), upregulating its expression could conceivably release LPL from the myocyte surface for translocation to the coronary lumen. Our studies illustrate a novel mechanism by which latent heparanase, by modulating myocyte MMP-9 expression, can have a long-term effect on LPL trafficking (Fig. 8).
Although multiple sources could explain the presence of heparanase in cardiomyocytes, we assumed that it is secondary to uptake of latent heparanase after its secretion from EC. Indeed, high-glucose–induced secretion of latent heparanase from EC was efficiently taken up by cardiomyocytes in our coculture system. To extend these observations to a model of diabetes, we used STZ animals that were hyperglycemic for 72 h (D55) and observed accumulation of active heparanase in the myocyte nucleus, together with a robust increase in MMP-9 expression and an accelerated shedding of HSPGs. These animals also exhibited increased coronary LPL activity with heparin perfusion. Our results suggest that in addition to the high-glucose–induced secretion of active heparanase to initiate an immediate transfer of LPL from the myocyte surface to the vascular lumen, activation of MMP-9 expression by latent heparanase taken up into the myocytes is pivotal for sustaining trafficking of LPL during chronic hyperglycemia (Fig. 8). Interestingly, we have previously shown that when compared with D55, animals with acute hyperglycemia (28.2 ± 2.3 mmol/L compared with 5.9 ± 1.0 mmol/L in controls) induced by diazoxide (4 h) are unable to maintain their high LPL activity at the vascular lumen after enzyme release by heparin (46). Because cardiomyocytes from diazoxide hearts do not increase MMP-9 expression (Supplementary Fig. 2), the sustained increase of coronary LPL seen with D55 can likely be attributed to upregulation of MMP-9 by latent heparanase.
Overall, results from the present study suggest that in response to hyperglycemia, EC-derived heparanase is a key mediator that controls the ability of the cardiomyocyte to send LPL to the coronary lumen. Although this adaptation might be beneficial in the short-term to compensate for energy deficit after diabetes, it is potentially catastrophic over a protracted duration because superfluous FA causes lipotoxicity in the heart, ultimately leading to diabetic cardiomyopathy (Fig. 8). Inhibiting heparanase secretion or function could offer a new strategy for managing the cardiomyopathy after diabetes.
One limitation of this study is the lack of mouse models supporting the role of MMP-9 in regulating coronary LPL. MMP-9 knockout mice are available and could be potentially used in this study. However, coronary LPL activity remains unchanged in both type 1 and type 2 diabetic mouse hearts. This could be a consequence of genetic adaptation or the excessive heart rate in control animals (∼600 bpm), which, unlike in humans (parasympathetic tone), is under sympathetic control. This permits prior translocation of LPL from the cardiomyocyte to the coronary lumen to saturate all of the LPL binding sites. Hence, we have largely depended on rats when studying cardiac metabolism and LPL.
See accompanying article, p. 2600.
Funding. The current study is supported by an operating grant from the Canadian Diabetes Association to B.R., and in part by a grant to I.V. from the Israel Ministry of Health and a grant to G.L. from the National Natural Science Foundation. Y.W. and D.Z. are the recipients of Doctoral Student Research Awards from the Canadian Diabetes Association.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. Y.W. conceived the idea, generated most of the data, and wrote the manuscript. A.P.-L.C., K.N., F.W., D.Z., B.H., N.L., and A.W. helped with obtaining some of the data. B.R. helped with writing the manuscript. G.L. provided valuable suggestions and technical support on this work. I.V. assisted with valuable suggestions and the preparation of the highly purified latent and active heparanase. B.R. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.