Proteoglycans in health and disease: new concepts for heparanase function in tumor progression and metastasis

Authors


I. Vlodavsky, Cancer and Vascular Research Center, Rappaport Faculty of Medicine, Technion, P. O. Box 9649, Haifa 31096, Israel
Fax: +972 4 8510445
Tel: +972 4 8295410
E-mail: vlodavsk@cc.huji.ac.il

Abstract

Heparanase is an endo-β-d-glucuronidase capable of cleaving heparan sulfate side chains at a limited number of sites, yielding heparan sulfate fragments of still appreciable size. Importantly, heparanase activity correlates with the metastatic potential of tumor-derived cells, attributed to enhanced cell dissemination as a consequence of heparan sulfate cleavage and remodeling of the extracellular matrix and basement membrane underlying epithelial and endothelial cells. Similarly, heparanase activity is implicated in neovascularization, inflammation and autoimmunity, involving the migration of vascular endothelial cells and activated cells of the immune system. The cloning of a single human heparanase cDNA 10 years ago enabled researchers to critically approve the notion that heparan sulfate cleavage by heparanase is required for structural remodeling of the extracellular matrix, thereby facilitating cell invasion. Progress in the field has expanded the scope of heparanase function and its significance in tumor progression and other pathologies. Notably, although heparanase inhibitors attenuated tumor progression and metastasis in several experimental systems, other studies revealed that heparanase also functions in an enzymatic activity-independent manner. Thus, inactive heparanase was noted to facilitate adhesion and migration of primary endothelial cells and to promote phosphorylation of signaling molecules such as Akt and Src, facilitating gene transcription (i.e. vascular endothelial growth factor) and phosphorylation of selected Src substrates (i.e. endothelial growth factor receptor). The concept of enzymatic activity-independent function of heparanase gained substantial support by the recent identification of the heparanase C-terminus domain as the molecular determinant behind its signaling capacity. Identification and characterization of a human heparanase splice variant (T5) devoid of enzymatic activity and endowed with protumorigenic characteristics, elucidation of cross-talk between heparanase and other extracellular matrix-degrading enzymes, and identification of single nucleotide polymorphism associated with heparanase expression and increased risk of graft versus host disease add other layers of complexity to heparanase function in health and disease.

Abbreviations
ECM

extracellular matrix

EGFR

epidermal growth factor receptor

FGF

fibroblast growth factor

HS

heparan sulfate

HSPGs

heparan sulfate proteoglycans

HSulf-1

human Sulf1

MMP

matrix metalloproteinase

TIM

triosephosphate isomerase

VEGF

vascular endothelial growth factor

Introduction

Proteoglycans are composed of core protein to which glycosaminoglycan (GAG) side chains are covalently attached. GAGs are linear polysaccharides consisting of a repeating disaccharide, generally of an acetylated amino sugar alternating with uronic acid. Units of N-acetylglucosamine and glucuronic/iduronic acid form heparan sulfate (HS). The polysaccharide chains are modified at various positions by sulfation, epimerization and N-acetylation, yielding clusters of sulfated disaccharides separated by low or nonsulfated regions [1,2]. The sulfated saccharide domains provide numerous docking sites for a multitude of protein ligands, ensuring that a wide variety of bioactive molecules (i.e. cytokines, growth factors, enzymes, protease inhibitors, extracellular matrix proteins) binds to the cell surface and extracellular matrix (ECM) [3–6] and thereby functions in the control of normal and pathological processes, among which are morphogenesis, tissue repair, inflammation, vascularization and cancer metastasis [1–3]. Two main types of cell-surface HS proteoglycan (HSPG) core proteins have been identified: the transmembrane syndecan with four isoforms, carrying HS near their extracellular tips and occasionally also chondroitin sulfate chains near the cell surface [3]; and the glycosylphosphatidyl inositol-linked glypican with six isoforms, carrying several HS side chains near the plasma membrane and often an additional chain near the tip of its ectodomain [7]. Two major types of ECM-bound HSPG are found: agrin, abundant in most basement membranes, primarily in the synaptic region [8]; and perlecan, with a widespread tissue distribution and a very complex modular structure [9]. Accumulating evidences indicate that HSPGs act to inhibit cellular invasion by promoting tight cell–cell and cell–ECM interactions, and by maintaining the structural integrity and self-assembly of the ECM [10,11]. Notably, one of the characteristics of malignant transformation is downregulation of GAGs biosynthesis, especially of the HS chains [10,11]. Low levels of cell-surface HS also correlate with high metastatic capacity of many tumors. For example, reduced syndecan-1 levels on the cell surface of colon, lung, hepatocellular, breast, and head and neck carcinomas was associated with increased tumor metastasis [10]. In other cases, syndecan-1 was nonetheless overexpressed, and appeared to promote metastasis [12]. This behavior is attributed mostly to HSPGs within the ECM, exemplified by the protumorigenic function of shed syndecan-1 in multiple myeloma [10,13] (see below).

In addition to modulation of HSPG levels, expression of enzymes involved in GAGs biosynthesis and modification is impaired during cell transformation. Hereditary multiple exostosis provided the first direct evidence linking an aberrant HS structure to tumorigenesis. Hereditary multiple exostosis is an autosomal-dominant disorder characterized by the presence of multiple bony outgrowths (exostoses), a consequence of mutation in EXT family members. These genes encode an enzyme (GlcA/GlcNAc transferase) required for chain elongation and synthesis of HS in the Golgi apparatus [14,15]. Bone outgrowths as a result of mutation and inactivation of these enzymes imply their function as tumor-suppressors. HS can similarly be modified extracellularly by secreted enzymes such as heparan sulfate 6-O-endosulfatases which selectively remove the 6-O-sulfate groups from HS. Human Sulf-1 (HSulf-1) appears to be misregulated in cancer; it is present in a variety of normal tissues but is downregulated in cell lines originating from ovarian, breast, pancreatic, renal and hepatocellular carcinomas [16]. Loss of HSulf-1 expression results in increased sulfation of HSPGs, sustained association of heparin-binding growth factors with their cognate receptors and augmented downstream signaling. Expression of HSulf-1 in cell lines derived from head and neck carcinoma inhibits cell growth, motility and invasion in vitro [17]. Similarly, overexpression of HSulf-1 and HSulf-2 in CAG myeloma cells inhibits tumor xenograft development and the assembly of fibroblast growth factor (FGF)-2 signaling complex on the cell surface [18], supporting its function as negative regulator of cancer.

Whereas the activity HSulf-1 appeares to attenuate tumor progression, cleavage of HS by the endo-β-glucuronidase heparanase is strongly implicated in cell dissemination associated with tumor metastasis. Cloning of the heparanase gene 10 years ago [19–22] and the generation of specific tools (i.e. molecular probes, antibodies, siRNA) enabled researchers to critically approve the notion that HS cleavage by heparanase is required for structural remodeling of the ECM underlying tumor and endothelial cells, thereby facilitating cell invasion [23–25]. Progress in the field and the generation of genetic tools (i.e. heparanase transgenic and knockout mice) [26–29] have led in recent years to the discovery of new concepts which expand the scope of heparanase function and its significance in tumor progression and other pathologies.

In this minireview we discuss recent progress in heparanase research, focusing on enzymatic activity-dependent and -independent functions mediated by defined protein domains and splice variants, and cross-talk between heparanase and proteases. Aspects such as heparanase gene regulation, proteolytic processing, cellular localization and the development of heparanase inhibitors have been the subject of several recent review articles [23,25,30,31] and are not discussed in detail here.

Heparanase in tumor progression and metastasis

Enzymatic activity capable of cleaving glucuronidic linkages and releasing polysaccharide chains resistant to further degradation by the enzyme was first identified by Ogren & Lindahl [32]. The physiological function of this activity was initially implicated in the degradation of macromolecular heparin to physiologically active fragments [32,33]. The activity of the newly discovered endo-β-glucuronidase, referred to as heparanase, was soon after shown to be associated with the metastatic potential of tumor-derived cells such as B16 melanoma [34] and T-lymphoma [35]. These early observations gained substantial support when specific molecular probes became available shortly after cloning of the heparanase gene. Both overexpression and silencing of the heparanase gene clearly indicate that heparanase not only enhances cell dissemination, but also promotes the establishment of a vascular network that accelerates primary tumor growth and provides a gateway for invading metastatic cells [23,25]. Although these studies provided a proof-of-concept for the prometastatic and proangiogenic capacity of heparanase, the clinical significance of the enzyme in tumor progression emerged from a systematic evaluation of heparanase expression in primary human tumors. Immunohistochemistry, in situ hybridization, RT-PCR and real time-PCR analyses revealed that heparanase is upregulated in essentially all human carcinomas examined [23,25]. Notably, increased heparanase levels were most often associated with reduced patient survival post operation, increased tumor metastasis and higher microvessel density [23–25]. We choose to highlight the role of heparanase in human cancer by focusing on head and neck carcinoma and multiple myeloma as examples of solid and hematological malignancies.

Heparanase in head and neck carcinoma: signaling in motion

Squamous cell carcinoma of the head and neck continues to be the sixth most common neoplasm in the world, with > 500 000 new cases projected annually [36]. Approximately 200 000 deaths occur yearly as the result of cancer of the oral cavity and pharynx, and the outcome has not improved significantly in the past 25 years [37]. Tumor metastases are common among patients with head and neck cancer with uncontrolled local or regional disease, and autopsy studies revealed 40–47% overall incidence of distant metastases [38,39]. Applying immunohistochemistry, no staining of heparanase was detected in normal epithelium adjacent to the tumor lesions (Fig. 1A), likely due to methylation of the gene and its repression by p53 [40–43]. By contrast, heparanase upregulation was found in the majority of head and neck [44], salivary gland [45], tongue [46] and oral [47] carcinomas. Notably, respective patients that exhibit no or weak heparanase staining are endowed with a favorable prognosis and prolonged survival post operation [44–46,48]. For example, 70% of the patients with salivary gland carcinoma that stained negative for heparanase were still alive 300 months (25 years) following diagnosis, whereas none of patients stained strongly for heparanase survived at 300 months [45]. Somewhat surprising, heparanase upregulation in head and neck and tongue carcinomas was associated with larger tumors [44,46]. This association was also seen in hepatocellular, breast and gastric carcinomas [49–51]. Likewise, heparanase overexpression enhanced [52–55], whereas local delivery of antiheparanase siRNA inhibited, the progression of tumor xenografts [56]. These results imply that heparanase function is not limited to tumor metastasis but is engaged in progression of the primary lesion.

Figure 1.

 (A) Immunohistochemical staining of heparanase in squamous cell carcinoma of the head and neck (SCCHN) tumor specimens. Formalin-fixed, paraffin-embedded 5 μm sections of head and neck tumors were subjected to immunostaining of heparanase, applying anti-heparanase polyclonal Ig #733. Shown are representative photomicrographs of positively stained specimens exhibiting cytoplasmic (Cyto, middle) and nuclear (Nuc, lower) heparanase localization. Normal-looking tissue adjacent to the tumor lesion stained negative for heparanase (upper). Nuclear heparanase is associated with decreased levels of phospho-EGFR, lower lymph vessel density, and favorable prognosis of head and neck cancer patients (see text for details). (B) Heparanase expression associates with tumor cell invasion into lymph vessels. Head and neck tumor specimen was stained with anti-heparanase polyclonal (green, upper) and D2-40 monoclonal (a marker for human lymphatics; red, middle) Ig, illustrating heparanase-positive tumor cells inside a lymphatic vessel lumen (merge, lower).

Heparanase and tumor vascularization

The cellular and molecular mechanisms underlying enhanced tumor growth by heparanase are only starting to be revealed. At the cellular level, both tumor cells and cells that comprise the tumor microenvironment (i.e. endothelial, fibroblasts, tumor-infiltrating immune cells) are likely to be affected by heparanase. The proangiogenic potency of heparanase has been established clinically [23,25,31] and in several in vitro and in vivo model systems, including wound healing [29,57], tumor xenografts [52,55], Matrigel plug assay [57] and tube-like structure formation [58]. Moreover, microvessel density was significantly reduced in tumor xenografts developed by Eb lymphoma cells transfected with antiheparanase ribozyme [59]. The molecular mechanism by which heparanase facilitates angiogenic responses has traditionally been attributed primarily to the release of HS-bound growth factors such as vascular endothelial growth factor (VEGF)-A and FGF-2 [60,61], a direct consequence of heparanase enzymatic activity. In addition, enzymatically inactive heparanase was noted to facilitate adhesion and migration of primary endothelial cells [58] and to promote phosphorylation of signaling molecules such as Akt and Src [53,55,58,62,63], the latter found to be responsible for VEGF-A induction following exogenous addition of heparanase or its overexpression [55]. Furthermore, heparanase was also noted to facilitate the formation of lymphatic vessels. In head and neck carcinoma, high levels of heparanase were associated with increased lymphatic vessel density, increased tumor cell invasion to lymphatic vessels (Fig. 1B) and increased expression of VEGF-C [64], a potent mediator of lymphatic vessel formation [65]. Heparanase overexpression by melanoma, epidermoid, breast and prostate carcinoma cells induced a three- to fivefold elevation of VEGF-C expression in vitro, and facilitated lymph angiogenesis of tumor xenografts in vivo, whereas heparanase gene silencing was associated with decreased VEGF-C levels [64]. These results suggest that enhanced lymph angiogenesis by heparanase is not specific for head and neck carcinoma, but rather is a common trait. Upregulation of VEGF-C was greatly dependent on the cellular localization of heparanase. Whereas localization of heparanase to the cytoplasm (representing secreted heparanase and predicting poor prognosis of cancer patients; Fig. 1A, Cyto) was associated with increased VEGF-C staining, nuclear localization of heparanase (Fig. 1A, Nuc), shown to correlate with a favorable prognosis of head and neck cancer patients [44], was associated with low levels of VEGF-C [64]. Similarly, localization of heparanase in the cell cytoplasm was associated with activation of the epidermal growth factor receptor (EGFR) in head and neck carcinoma [66].

Heparanase and EGFR activation

Decorin, a chondroitin sulfate/dermatan sulfate proteoglycan directly interacts with EGFR and this evokes a downregulation of the receptor and inhibition of its downstream signaling. The antiproliferative effect of decorin on cancer cells via EGFR is reviewed by Iozzo & Schaefer [67]. By contrast, EGFR phosphorylation is markedly increased in cells overexpressing heparanase or following its exogenous addition, whereas heparanase gene silencing is accompanied by reduced EGFR and Src phosphorylation levels [66]. Notably, EGFR activation was observed following the addition or overexpression of mutated, enzymatically inactive heparanase protein. Although inactive, double-mutated (Glu225, Glu343) [68] heparanase retains its high affinity towards HS and hence may facilitate signaling by ligation and activation of membrane HSPGs such as syndecan [69,70]. This however appears not to be the case because heparanase deleted for its heparin-binding domain (Δ10) [71] efficiently stimulated EGFR phosphorylation [66]. Notably, enhanced EGFR phosphorylation by heparanase was restricted to selected tyrosine residues (i.e. 845, 1173) thought to be direct targets of Src rather than a result of receptor autophosphorylation [72]. Indeed, enhanced EGFR phosphorylation of tyrosine residues 845 and 1173 in response to heparanase was abrogated in cells treated with Src inhibitors or antiSrc siRNA [66]. The functional significance of EGFR modulation by heparanase emerged by monitoring cell proliferation. Thus, heparanase gene silencing was accompanied by a decrease in cell proliferation, whereas heparanase overexpression resulted in enhanced cell proliferation and the formation of larger colonies in soft agar, in a Src- and EGFR-dependent manner [66]. The clinical relevance of the heparanase–Src–EGFR pathway has been elucidated for head and neck carcinoma. Notably, heparanase expression in head and neck carcinomas correlated with phospho-EGFR immunostaining, and even more significant was the correlation between heparanase cellular localization (i.e. cytoplasmic versus nuclear) and phospho-EGFR levels [66]. These studies provide a more realistic view of heparanase function in the course of tumor progression. Thus, while heparanase enzymatic activity has traditionally been implicated in tumor metastasis, the current view points to a multifaceted protein engaged in multiple aspects of tumor progression, combining enzymatic activity-dependent and -independent activities of heparanase and affecting two systems critical for tumor progression, namely tumor vascularization and EGFR activation.

Signaling by the heparanase C-domain

The concept of enzymatic activity-independent function of heparanase gained substantial support by the recent identification of the heparanase C-domain as the molecular determinant behind its signaling capacity. The existence of a C-terminus domain (C-domain) emerged from a prediction of the 3D structure of a single-chain heparanase enzyme [73]. In this protein variant, the linker segment was replaced by three glycine–serine repeats (GS3), resulting in a constitutively active enzyme [74]. The structure obtained clearly illustrates a triosphosphate isomerase (TIM)-barrel fold, in agreement with previous predictions [68,75]. Notably, the structure also delineates a C-terminus fold positioned next to the TIM-barrel fold [73]. The predicted heparanase structure led to the hypothesis that the seemingly distinct protein domains observed in the 3D model, namely the TIM-barrel and C-domain regions, mediate enzymatic and nonenzymatic functions of heparanase, respectively. Interestingly, cells transfected with the TIM-barrel construct (amino acids 36–417) failed to display heparanase enzymatic activity, suggesting that the C-domain is required for the establishment of an active heparanase enzyme, possibly by stabilizing the TIM-barrel fold [73]. Deletion and site-directed mutagenesis further indicated that the C-domain plays a decisive role in heparanase enzymatic activity and secretion [73,76,77]. Notably, Akt phosphorylation was stimulated by cells overexpressing the C-domain (amino acids 413–543), whereas the TIM-barrel protein variant yielded no Akt activation compared with control, mock-transfected cells [73]. These findings clearly indicate that the nonenzymatic signaling function of heparanase leading to activation of Akt is mediated by the C-domain. Notably, the C-domain construct lacks the 8 kDa segment (Gln36–Ser55) which, according to the predicted model, contributes one beta strand to the C-domain structure (reviewed in [78]). Indeed, Akt phosphorylation was markedly enhanced and prolonged in cells transfected with a mini gene comprising this segment linked to the C-domain sequence (8-C) [73,78]. This finding further supports the predicted 3D model, indicating that the C-domain is indeed a valid functional domain responsible for Akt phosphorylation. The cellular consequences of C-domain overexpression were best revealed by monitoring tumor xenograft development. Remarkably, tumor xenografts produced by C-domain-transfected glioma cells grew faster and appeared indistinguishable from those produced by cells transfected with the full-length heparanase in term of tumor size and angiogenesis, yielding tumors sixfold bigger than control. By contrast, progression of tumors produced by TIM-barrel-transfected cells appeared comparable with control mock-transfected cells [73,78]. These results show, that in some tumor systems (i.e. glioma), heparanase facilitates primary tumor progression regardless of its enzymatic activity, whereas in others (i.e. myeloma) heparanase enzymatic activity dominates (see below). Enzymatic activity-independent function of heparanase is further supported by the recent identification of T5, a functional human splice variant of heparanase.

T5, a functional human heparanase splice variant

Almost all protein-coding genes contain introns that are removed in the nucleus by RNA splicing and are often alternatively spliced. Alternative splicing increases the coding capacity of the genome, generating multiple proteins from a single gene. The resulting protein isoforms frequently exhibit different biological properties that may play an essential role in tumorigenesis [79,80]. A splice variant of human heparanase which lacks exon 5 has been described [81,82]. This splice variant fails to get secreted and lacks enzymatic activity and its biological significance remains unclear. Additional human heparanase splice variants have been predicted in silico [83]; the expression of one, termed T5 (Fig. 2A), was found to be enriched in lung carcinoma and chronic myeloid leukemia compared with control tissue and cells. In this splice variant, 144 bp of intron 5 are joined with exon 4, resulting in a 169-amino-acids protein that lacks the enzymatic activity typical of heparanase [83]. Unlike previously identified splice variants of heparanase, T5 is secreted and facilitates Src phosphorylation [83]. Furthermore, Src phosphorylation was markedly reduced in cells treated with antiT5 siRNA [83]. Overexpression of T5 by pharynx (FaDu), myeloma (CAG) and embryonic kidney (293) cells resulted in enhanced proliferation and larger colony formation in soft agar, which was attenuated by Src inhibitor (Fig. 2B) [83]. Likewise, T5 gene silencing was associated with reduced cell proliferation, indicating that endogenous levels of T5 and heparanase affect tumor cell proliferation. Moreover, development of tumor xenografts produced by heparanase- and T5-infected myeloma cells was markedly enhanced compared with xenografts generated by control cells (Fig. 2C) [83]. Tumors developed by T5-expressing cells exhibited a higher density of blood vessels decorated with smooth muscle actin-positive cells (pericytes) [83], an indication of vessel maturation. The clinical relevance of T5 emerged from analysis of renal cell carcinoma biopsies, in which T5 and heparanase expression appeared to be induced in 75% of cases [83]. Thus, although inhibitors directed against the enzymatic activity of heparanase are being currently evaluated in clinical trials [84–87], T5 and the heparanase C-domain are not expected to be affected by these inhibitors. It appears, therefore, that a well-defined enzymatic activity thought to be relatively easy to target, turned, at least in certain tumor systems, into a complex objective as more knowledge accumulates and the biology of the protein is being elucidated.

Figure 2.

 Heparanase splice variant, T5, endowed with protumorigenic characteristics. (A) Schematic structure of wild-type (WT) and heparanase splice variant, T5. SP-signal peptide; glutamic acids residues 225 and 343 critical for heparanase enzymatic activity, are detonated (see text for details). (B) Colony formation in soft agar. Control (Vo) heparanase (Hepa)-, and T5-infected myeloma (CAG, upper), pharynx (FaDu, second panels) and embryonic kidney (293, third panels) cells (5 × 103 cells·dish−1) were mixed with soft agar and cultured for 3–5 weeks. CAG cells were similarly grown in the absence (dimethylsulfoxide; fourth panels) or presence of Src inhibitor (PP2, 0.4 nm; lower panels). Shown are representative photomicrographs of colonies at high (×100) magnification. (C) Tumor xenograft development. Control (Vo), heparanase-, and T5-infected CAG myeloma cells were injected subcutaneously (1 × 106 0.1 mL−1). At the end of the experiment on day 37, tumors were harvested and photographed.

Multiple myeloma: moving antiheparanase therapy closer to reality

Multiple myeloma is the second most prevalent hematologic malignancy. This B-lymphoid malignancy is characterized by tumor cell infiltration of the bone marrow, resulting in severe bone pain and osteolytic bone disease. Although progress in the treatment of myeloma patients has been made over the last decade, the overall survival of patients is still poor.

Heparanase enzymatic activity was elevated in the bone marrow plasma of 86% of myeloma patients examined [88], and gene array analysis showed elevated heparanase expression in 92% of myeloma patients [89]. Heparanase upregulation in myeloma patients was associated with elevated microvessel density and syndecan-1 expression [88]. Although heparanase is proangiogenic in myeloma, which is a common feature shared with solid tumors, heparanase regulation of syndecan-1 shedding has emerged as highly relevant to multiple myeloma progression.

Syndecan-1 is particularly abundant in myeloma, and is the dominant and often the only HSPG present on the surface of myeloma cells [90]. Cell-surface syndecan-1 promotes adhesion of myeloma cells and inhibits cell invasion in vitro [13]. By contrast, high levels of shed syndecan-1 are found in the serum of some myeloma patients and are associated with poor prognosis [91]. The multiple roles of syndecans in cancer progression and strategies for their targeting is presented in the accompanying minireview by Theocharis et al. [92]. Shed syndecan-1 becomes trapped within the bone marrow ECM where it likely acts to enhance the growth, angiogenesis and metastasis of myeloma cells within the bone [13,93,94]. This is supported by the finding that enhanced expression of soluble syndecan-1 by myeloma cells promotes tumor growth and metastasis in a mouse model [13,94]. Notably, heparanase upregulates both the expression and shedding of syndecan-1 from the surface of myeloma cells [89,95]. In agreement with this notion, heparanase gene silencing was associated with decreased levels of shed syndecan-1 [89]. Importantly, both syndecan-1 upregulation and shedding require heparanase enzymatic activity, because overexpression of mutated inactive heparanase failed to stimulate syndecan-1 expression and shedding [95]. Syndecan-1 shedding was similarly augmented by the addition of recombinant active heparanase to CAG myeloma cells, and even more dramatic shedding was observed following the addition of bacterial heparinase III (heparitinase) [95]. These findings indicate that cleavage of HS by heparanase or heparinase III may render syndecan-1 more susceptible to proteases mediating the shedding of syndecan-1. However, it appears that heparanase may play an even more direct role in regulating shedding of syndecan-1, by facilitating the expression of proteases engaged in syndecan shedding.

Heparanase–matrix metalloproteinase cooperation in myeloma progression

It was recently demonstrated that enhanced expression of heparanase leads to increased levels of matrix metalloproteinase (MMP)-9 (a syndecan-1 sheddase), whereas heparanase gene silencing resulted in reduced MMP-9 activity [96]. Upregulation of MMP-9 expression has significant biological relevance because inhibition of MMP-9 reduces syndecan-1 shedding [96]. For the importance of syndecan shedding in diseases see the accompnaying minireview by Manon-Jensen et al. [97]. Moreover, not only MMP-9, but also urokinase-type plasminogen activator and its receptor, molecular determinants responsible for MMP-9 activation, are upregulated by heparanase. These findings provided the first evidence for cooperation between heparanase and MMPs in regulating HSPGs on the cell surface and likely in the ECM, and are supported by the recent generation and characterization of heparanase knockout mice. HS chains isolated from these mice were longer, critically supporting the notion that heparanase is the only functional endoglycosidase capable of degrading HS [26]. Despite the complete lack of heparanase gene expression and enzymatic activity, heparanase knockout mice develop normally, are fertile and exhibit no apparent anatomical or functional abnormalities [26]. Interestingly, heparanase deficiency was accompanied by a marked elevation of MMP family members such as MMP-2, MMP-9 and MMP-14, in an organ-dependent manner. Thus, MMP-14 levels were increased eightfold in the liver of heparanase knockout mice compared with control littermates, whereas MMP-2 levels were increased 2.5-fold in the mammary gland [26], suggesting that MMPs provide tissue-specific compensation for heparanase deficiency. This is likely the reason for over-branching of the mammary gland in heparanase-knockout mice [26], a phenotype also noted in heparanase transgenic mice [27]. Collectively, these results suggest that heparanase is intimately engaged in the regulation of gene transcription and acts as a master regulator of protease expression, mediating gene induction or repression, depending on the biological setting.

The heparanase–syndecan axis is a target for therapy

Results from studies using several in vivo model systems support the notion that enzymatic activities responsible for syndecan-1 modification are valid targets for myeloma therapy. For example, enhanced expression of either HSulf-1 or HSulf-2 attenuated myeloma tumor growth [18]. Even a more dramatic inhibition of tumor growth was noted following administration of bacterial heparinase III (heparitinase) to SCID mice inoculated with either CAG myeloma cells or cells isolated from the bone marrow of myeloma patients [98]. Although heparinase III and human heparanase both degrade HS chains, their cleavage products are distinct. Whereas heparinase III is a β-eliminase that extensively degrades HS, heparanase is an endo-β-d-glucuronidase whose substrate-recognition sites were recently characterized [99]. Unlike the bacterial enzyme, heparanase cleaves HS more selectively and generates fragments of 4–7 kDa, yielding strictly distinct outcomes in the context of tumor progression. Although administration of heparinase III is associated with reduced tumor growth, heparanase activity is elevated in many hematological and solid tumors, correlating with poor prognosis and shorter post-operative survival rate (see above). Accordingly, inhibition of heparanase enzymatic activity is expected to suppress tumor progression. To examine this in myeloma, a chemically modified heparin, which is 100% N-acetylated and 25% glycol-split was tested. This flexible molecule is a potent inhibitor of heparanase enzymatic activity, lacks anticoagulant activity typical of heparin, and does not displace ECM-bound FGF-2 or potentiate its mitogenic activity [30,31,100,101]. The modified heparin profoundly inhibits the progression of tumor xenografts produced by myeloma cells [30,98]. These studies support the notion that heparanase enzymatic activity not only facilitates tumor metastasis, but also promotes the progression of primary tumors.

Conclusions and perspective

Although much has been learned in the last decade, the repertoire of heparanase functions in health and disease is only starting to emerge. Clearly, from activity implicated mainly in cell invasion associated with tumor metastasis, heparanase has turned into a multifaceted protein that appears to participate in essentially all major aspects of tumor progression. In this regard, evidence now supports a concept by which growth of the primary tumor is fueled by circulating metastatic tumor cells [102,103]. According to this notion, tumor cells are present in the circulation in large numbers even at the early stages of cancer and long before metastatic growth at distant sites can be detected [103]. These cells can reinfiltrate and promote growth and angiogenesis of the primary tumor [102]. The possible involvement of heparanase in tumor self-seeding is supported by the timing of its induction during tumorigenesis and its prometastatic function. Using the RIP-Tag2 tumor model, it was demonstrated that heparanase mRNA and protein are elevated upon the transition from normal to angiogenic islets, followed by a further increase when solid tumors were detected [104]. Furthermore, heparanase expression is elevated already at the early stages of human neoplasia. In the colon, heparanase gene and protein are expressed already at the stage of adenoma [105], and during esophageal carcinogenesis heparanase expression is induced in Barrett’s epithelium (Fig. 3), an early event that predisposes patients to the formation of dysplasia which may progress to adenocarcinoma [106]. Tumor self-seeding also facilitates the recruitment of stromal components. Although the proangiogenic capacity of heparanase has been established, its likely impact on other components of the tumor microenvironment (i.e. fibroblasts, macrophages) awaits thorough investigation.

Figure 3.

 Immunohistochemical staining of esophageal specimens. Formalin-fixed, paraffin-embedded 5 μm sections of normal (upper panel), Barrett’s (second panel), low-grade (third panel), high-grade (fourth panel) and adenocarcinoma (lower panel) esophageal biopsies were subjected to immunostaining of heparanase, applying anti-heparanase polyclonal Ig #733 (left panels) or anti-(Ki-67), a marker of cell proliferation (right panels).

Heparanase expression at the early stages of tumor initiation and progression, and by the majority of tumor cells (evident by a high extent of immunostaining), can be utilized to turn the immune system against the very same cells. Accumulating evidence suggests that peptides derived from human heparanase can elicit a potent antitumor immune response, leading to lysis of heparanase-positive human gastric (KATO III), colon (SW480) and breast (MCF-7) carcinoma cells, as well as hepatoma (HepG2) and sarcoma (U-2 OS) cells [107–109]. By contrast, no killing effect was noted towards autologous lymphocytes [107–109]. Notably, the development of tumor xenografts produced by B16 melanoma cells was markedly restrained in mice immunized with peptides derived from mouse heparanase (i.e. amino acids 398–405; 519–526) compared with a control peptide in both immunoproection and immunotherapy approaches [109]. T-regulatory cells are frequently present in colorectal cancer patients. Interestingly, T-regulatory cells against heparanase could not be found [110]. Antiheparanase immunotherapy is thus expected to be prolonged and more efficient due to the absence of T-suppressor cells. A related treatment approach is being tested in advanced metastasized breast cancer patients [111]. Although this immunotherapeutic concept, together with available heparanase inhibitors, is hoped to advance cancer treatment, the identification of single nucleotide polymorphism associated with heparanase expression and increased risk for graft versus host disease following allogeneic stem cell transplantation [112–114] offers a genetic concept which can potentially be translated into patients’ diagnosis. Studies in these directions, identification of heparanase receptor(s) mediating its signaling function, and elucidation of heparanase route and function in the cell nucleus, will advance the field of heparanase research and reveal its significance in health and disease. Resolving the heparanase crystal structure will accelerate the development of effective inhibitory molecules and neutralizing antibodies paving the way for advanced clinical trials in patients with cancer and other diseases (i.e. colitis, psoriasis, diabetic nephropathy) involving heparanase.

Acknowledgements

We thank Prof. Benito Casu (‘Ronzoni’ Institute, Milan, Italy) for his continuous support and active collaboration. This work was supported by grants from the Israel Science Foundation (grant 549/06); National Institutes of Health (NIH) grants CA138535 (RDS) and CA106456 (IV); the Israel Cancer Research Fund (ICRF); and the Juvenile Diabetes Research Foundation (JDRF grant 1-2006-695). I. Vlodavsky is a Research Professor of the ICRF. We gratefully acknowledge the contribution, motivation and assistance of the research teams in the Hadassah-Hebrew University Medical Center (Jerusalem, Israel) and the Cancer and Vascular Biology Research Center of the Rappaport Faculty of Medicine (Technion, Haifa). We apologize for not citing several relevant articles, due to space limitation.

Ancillary