Proteoglycans in health and disease: novel roles for proteoglycans in malignancy and their pharmacological targeting

Abstract

The expression of proteoglycans (PGs), essential macromolecules of the tumor microenvironment, is markedly altered during malignant transformation and tumor progression. Synthesis of stromal PGs is affected by factors secreted by cancer cells and the unique tumor-modified extracellular matrix may either facilitate or counteract the growth of solid tumors. The emerging theme is that this dual activity has intrinsic tissue specificity. Matrix-accumulated PGs, such as versican, perlecan and small leucine-rich PGs, affect cancer cell signaling, growth and survival, cell adhesion, migration and angiogenesis. Furthermore, expression of cell-surface-associated PGs, such as syndecans and glypicans, is also modulated in both tumor and stromal cells. Cell-surface-associated PGs bind various factors that are involved in cell signaling, thereby affecting cell proliferation, adhesion and motility. An important mechanism of action is offered by a proteolytic processing of cell-surface PGs known as ectodomain shedding of syndecans; this facilitates cancer and endothelial cell motility, protects matrix proteases and provides a chemotactic gradient of mitogens. However, syndecans on stromal cells may be important for stromal cell/cancer cell interplay and may promote stromal cell proliferation, migration and angiogenesis. Finally, abnormal PG expression in cancer and stromal cells may serve as a biomarker for tumor progression and patient survival. Enhanced understanding of the regulation of PG metabolism and the involvement of PGs in cancer may offer a novel approach to cancer therapy by targeting the tumor microenvironment. In this minireview, the implication of PGs in cancer development and progression, as well as their pharmacological targeting in malignancy, are presented and discussed.

Abbreviations

ADAMTS

a disintegrin and metalloprotease domain with thrombospondin motifs

CS

chondroitin sulfate

DS

dermatan sulfate

ECM

extracellular matrix

EGF

epidermal growth factor

EGFR

epidermal growth factor receptor

FGF

fibroblast growth factor

GAG

glycosaminoglycan

HA

hyaluronan

HC

hepatocellular carcinoma

Hh

hedgehog

HS

heparan sulfate

HSPGs

heparan sulfate proteoglycans

MMP

matrix metalloproteinases

PDGF

platelet-derived growth factor

PG

proteoglycan

SLRPs

small leucine-rich proteoglycans

VEGF

vascular endothelial growth factor

VEGFR

vascular endothelial growth factor receptor

Introduction

Proteoglycans (PGs) are macromolecules composed of a specific core protein substituted with covalently linked glycosaminoglycan (GAG) chains named chondroitin sulfate (CS), dermatan sulfate (DS), keratan sulfate, heparin and heparan sulfate (HS). Hyaluronan (HA) is the only GAG synthesized in a free form not covalently bound on a core protein. GAGs are linear, negatively charged polysaccharides comprised of repeating disaccharides of acetylated hexosamines (N-acetyl-galactosamine or N-acetyl-glucosamine) and mainly by uronic acids (d-glucoronic acid or l-iduronic acid) being sulfated at various positions. Keratan sulfate is the only GAG to be comprised of repeating disaccharides containing N-acetyl-glucosamine and galactose.

PGs can be classified into three main groups according to their localization, extracellularly secreted, those associated with the cell surface and intracellular. Each main group is further classified into subfamilies according to their gene homology, core protein properties, size and modular composition. Secreted PGs involve large aggregating PGs, named hyalectans, small leucine-rich PGs (SLRPs) and basement membrane PGs. Cell-surface-associated PGs are divided into two main subfamilies (syndecans and glypicans), whereas serglycin is the only intracellular PG characterized to date [1,2]. The wide molecular diversity of PGs derives from the multitude of possible combinations of protein cores, O-linked and N-linked oligosaccharides, and various types and numbers of GAG chains. The specific structural characteristics of GAG types provide some of the structural basis for the multitude of their biological functions [3]. PGs exhibit numerous biological functions acting as structural components in tissue organization, and affect several cellular parameters, such as cell proliferation, adhesion, migration and differentiation. PGs interact with growth factors and cytokines, as well as with growth factor receptors, and are implicated in cell signaling. The catalog of physiological functions and the pathobiological roles in which PGs are involved have grown rapidly.

During carcinogenesis, malignant cells secrete soluble growth factors that stimulate cell growth and activate stromal cells to secrete effectors that in turn stimulate further tumor cell growth. Both activated stromal and tumor cells are implicated in the reorganization of the extracellular matrix (ECM) to facilitate tumor cell growth, migration and invasion. PG expression is markedly modified in the tumor microenvironment. Altered expression of PGs on tumor and stromal cell membranes affects cancer cell signaling, growth and survival, cell adhesion, migration and angiogenesis. The type and fine structure of GAG chains attached to PGs are markedly affected in the context of malignant transformation as result of the altered expression of GAG-synthesizing enzymes. Structural modifications of GAGs may facilitate tumorigenesis in various ways, modulating the functions of PGs. Our rapid increase in knowledge that PGs are among the key players in the tumor microenvironment and can modulate tumor progression, suggests their potential as pharmacological targets. Pharmacological treatment may target PG metabolism, their utilization as targets for immunotherapy or their direct use as therapeutic agents. In this minireview, we focus on the roles of PG in cancer development and progression, as well as their pharmacological targeting in malignancy.

Extracellular matrix PGS

Hyalectans

The subfamily of hyalectans includes versican, aggrecan, neurocan and brevican. Hyalectans have the ability to bind HA through their N-terminal globular domain (G1), the central domain carries most of the GAG chains and exhibits lectin-like activity located in the C-terminal globular domain (G3) (Fig. 1) [1].

Figure 1

Schematic representation of hyalectans (hyaluronan binding PGs) and SLRPs found in ECM.

Versican expression and functions in cancer and inflammation

Versican is expressed throughout the body and provides ECM with hygroscopic properties creating a loose and hydrated matrix that is necessary to support key events in development and disease. Four splice-variants of human versican (V0, V1, V2 and V3) have been identified. The differences among the versican splice-variants are found in the central portion of the protein core, and they vary in the number and presence of GAGs. Versican is able to regulate many cellular processes including adhesion, proliferation, apoptosis, migration, invasion and ECM assembly via the highly negatively charged CS/DS side chains, and by interactions of the G1 and G3 domains with other proteins [4–6]. Versican binds to ECM components such as HA, type I collagen, tenascin-R, fibulin-1 and -2, fibrillin-1, fibronectin, P- and L-selectin and chemokines. Versican also binds to the cell-surface proteins CD44, integrin β1, epidermal growth factor receptor (EGFR), P-selectin glycoprotein ligand-1 [6,7] and toll-like receptor 2 [8].

V0 and V1 are the predominant isoforms present in cancer tissues [9–12]. Overexpression of the V3 isoform in melanoma cells markedly reduces cell growth in vitro and in vivo [13], and promotes pulmonary metastases [14]. V3 isoform may have a dual role as an inhibitor of tumor growth and a stimulator of metastasis. Elevated levels of versican have been reported in most malignancies to date and have been associated with cancer relapse and poor patient outcome in breast, prostate and many other cancer types [10,15–21] (Table 1). We have seen that versican is accumulated in the preclinical phase of breast cancer in nonpalpable breast carcinomas (Lambropoulou et al., personal data). This accumulation seems to be associated with risk factors because increased mammographic density and malignant appearing microcalcifications are found in these cases. Versican appears to be most commonly secreted by the activated peritumoral stromal cells in adenocarcinomas [20,21]. However, human pancreatic cancer cells can also secrete versican [22].

Table 1. Expression of proteoglycans in malignancies and possible pharmacolοgical targeting. ADAMTS, a disintegrin and metalloprotease domain with thrombospondin motifs; HA, hyaluronan; MMP, matrix metalloproteinases.

Proteoglycan subfamilies	Expression in malignancies/roles	Possibilities for potential pharmacological targeting
Hyalectans
Versican	↑ Brain tumors, melanomas, osteosarcomas, lymphomas, acute monocytic leukemia, testicular tumors, breast, prostate, colon, lung, pancreatic, endometrial, ovarian and oral cancers Prognostic factor, proliferation, adhesion, motility, metastasis, angiogenesis	Inhibition of synthesis (a) Blocking of growth factors; blocking antibodies, kinase inhibitors, antisense oligonucleotides, (b) miRNA-based therapeutics; miR-199a* Inhibition of degradation (a) ADAMTS, (b) MMPs inhibitors Inhibition of interaction with HA (a) HA oligomers
Aggrecan	↓ Laryngeal cancer Cartilage degradation
Brevican	↑ Glioma Diagnostic marker, adhesion, motility	Inhibition of cleavage (a) ADAMTS, (b) MMPs inhibitors Immunotherapy
SLRPs
Decorin	↑ Osteosarcoma, testicular tumors, ovarian, colon, gastric, pancreatic, laryngeal, breast cancer Prognostic factor, inhibition of growth and migration (↓ ErbB and Met signaling)	Upregulation of synthesis (a) Viral-mediated delivery, (b) Demethylation agents, (c) Proteasome inhibitors Admistration of decorin protein or synthetic peptides
Lumican	↑ Osteosarcoma, melanoma, breast cancer Prognostic factor, inhibition of growth and migration	Upregulation of synthesis (a) Viral-mediated delivery Admistration of lumican protein or synthetic peptides
Basement membranes
Collagen XVIII	↑ Ovarian, pancreatic cancer ↓ Liver and oral cancer Promoter of growth and angiogenesis Inhibitor of angiogenesis (endostatin)	Admistration of endostatin fragment or synthetic peptides Endostatin producing cells in immunoisolation devices
Perlecan	↑ Liver, oral tumors, melanoma Promoter of growth and angiogenesis Inhibitor of angiogenesis (endorepellin)	Admistration of endorepellin fragment or synthetic peptides
Agrin	↑ Liver tumors. Promoter of growth and angiogenesis? Inhibitor of angiogenesis (C-terminus)?	Admistration of C-terminal fragment or synthetic peptides
Cell surface
Glypican-1	↑ Breast, pancreatic cancer Proliferation, growth factor signaling	Inhibition of synthesis (a) Blocking of growth factors; blocking Abs, kinase inhibitors, antisense oligonucleotides, (b) siRNA-based therapeutics Immunotherapy
Glypican-3	↓ Lung, gastric, ovarian, breast cancer and mesothelioma Inhibition of proliferation and induction of apoptosis ↑ Neuroblastoma, Wilm’s tumor, hepatocellular carcinoma, melanoma Promoter of tumor cell growth	Admistration of soluble fragments Immunotherapy
Syndecan-1	↑ Breast, prostate, head and neck, myeloma ↓ Lung, colorectal, endometrial, cervical gastric, pancreatic cancer Prognostic factor, proliferation, adhesion, migration, differentiation, angiogenesis	Immunotherapy Heparanase inhibitors
Syndecan-2	↑ Lung, ovarian, osteosarcoma, brain tumors, colon, mesothelioma Proliferation, adhesion, migration, angiogenesis	Immunotherapy Heparanase inhibitors
Syndecan-4	↑ Breast cancer, melanoma Adhesion and migration	Immunotherapy Heparanase inhibitors
Intracellular
Serglycin	↑ Hematological malignancies Diagnosis	?

Functional studies have demonstrated that versican can increase cancer cell motility [12,23–25], proliferation [26] and metastasis [27]. Cancer cells can form a polarized pericellular sheath through compartmentalized cell-surface CD44 expression and subsequent assembly of HA/versican aggregates that promotes their motility (Fig. 2) [25]. Soluble versican is able to reduce the attachment of prostate cancer and melanoma cells to fibronectin-coated surfaces in vitro [28,29]. Both the G1 and G3 versican domains have been shown to promote cell proliferation in NIH-3T3 fibroblasts and tumor cells [23,24,30,31] (Table 1). The G1 domain of versican stimulates proliferation by destabilizing cell adhesion, whereas the G3 domain induces proliferation, at least in part, by activating EGFR via the action of epidermal growth factor (EGF)-like motifs (Fig. 2). G1 and G3 domains may differentially control tumor growth rate and have interactive roles to promote tumor development and metastasis. Notably, several protease families that include a disintegrin and metalloprotease domain with thrombospondin motifs (ADAMTS), matrix metalloproteinases (MMPs) and plasmin can cleave versican generating fragments containing the G1 domain in some cases [4,5]. Thus, regulation of G1 and G3 versican levels by proteases is an important factor in cancer cell motility and metastasis (Table 1).

Figure 2

Versican interacts with hyaluronan creating large aggregates. CD44 interacts with HA and versican forming a polarized and highly hydrated pericellular sheath that destabilizes cell adhesion and promotes cell motility. These aggregates facilitate further cell-shape changes and promote cell proliferation. EGF-like motifs present in the G3 domain of versican activate EGFR promoting cell proliferation and motility.

Versican may also promote the formation of an inflammatory microenvironment in the tumor stroma. The interaction between versican and toll-like receptor-2 links inflammation and metastasis [8,32]. Ligation of toll-like receptor-2 present on endothelial cells and fibroblasts by versican activates these cells and triggers the secretion of inflammatory cytokines. This is mechanistically important insofar as inflammation is often associated with cancer initiation and promotion, and inflammatory cells (primarily macrophages) are consistently regarded as critical mediators in the development of malignancies. Indeed, tumor-associated macrophages can enhance angiogenesis, ECM degradation and remodeling, and can promote cancer cell invasion [32].

Notably, the versican gene promoter contains a p53-binding site in its first intron and can be directly activated by the tumor suppressor p53 in a dose-dependent manner [33,34]. It was found to be a target gene of Wnt signaling in human embryonic carcinoma cells [35].

Versican regulation and targeting

Matrix effectors including platelet-derived growth factor (PDGF), transforming growth factor β, interleukin 1β, angiotensin II and steroid hormones affect versican synthesis in various normal cell lines [6]. Transforming growth factor β has been found to regulate synthesis of versican in fibrosarcoma, osteosarcoma and glioma cells [11,12,36]. The expression of versican can be also modified by EGF, insulin-like growth factor I and PDGF-BB in malignant mesothelioma cells [37]. The effect of protein tyrosine kinase signaling pathways on versican synthesis can be reversed following treatment with various tyrosine kinase inhibitors [38]. Therefore, targeting versican synthesis may be a potential mechanism for reducing this powerful tumor-promoting agent. In agreement with this concept, the tyrosine kinase inhibitor genistein can block versican expression induced by growth factors in malignant mesothelioma cell lines [37]. Genetic and preclinical studies support the targeting of growth factor [transforming growth factor β, PDGF, EGF and vascular endothelial growth factor (VEGF)] signaling as a therapeutic strategy for combating cancer. To date, several approaches to inhibit growth factor signaling pathways in cancer have been investigated. These approaches mainly include: (a) inhibition at the translational level using antisense oligonucleotides that can be engineered into immune cells or delivered directly into tumors, (b) inhibition of the ligand–receptor interaction using monoclonal antibodies, and (c) inhibition of the receptor-mediated signaling cascade using specific tyrosine kinase inhibitor. However, there are no data to show that such approaches are effective in inhibiting the effects of versican in cancer cell models (Table 1). Interestingly, versican synthesis is regulated by microRNAs (miRNA). miR-199a* function as a guide molecule in post-transcriptional gene silencing by binding to the 3′-untranslational region (3′-UTR) of versican mRNA, leading to translational repression. miR-199a* is considered to be an onco-suppressor, targeting molecules critically involved in the promotion of tumor growth and is often downregulated in malignancies [39,40]. Several companies have developed miRNA-based therapeutics and a strategy that increases the natural dose of miR-199a*, by introducing a short double-stranded synthetic RNA that is loaded into RNA-induced silencing complex or by utilizing expression of the hairpin pre-miRNA in a viral vector expression system, may be useful for targeting versican among the other genes involved in tumorigenesis [41] (Table 1).

The use of an antibody against to the ADAMTS-specific versican cleavage site inhibits glioma cell migration in a dose-dependent manner, suggesting that the local accumulation of versican fragments may also promote cancer cell motility and invasion (Fig. 2) [12]. Concurrent with this hypothesis, processed versican fragments have been identified in the peritumoral stroma of prostate cancer in association with higher ADAMTS1 and ADAMTS4 levels, but not in the interface surrounding normal prostate glands where full-legth versican is present [42]. Notably, the broad-spectrum MMP inhibitor GM6001 (Galardin), which inhibits the activity of MMPs and ADAMTS proteases, has been shown to inhibit cancer cell invasion and metastasis in a transgenic breast cancer model [43]. Other protease inhibitors such as catechin gallate esters, present in natural sources (green tea) have been shown to selectively inhibit ADAMTS-1, -4 and -5 and aggrecan catabolism in cartilage [44]. The ability of MMPs and ADAMTS inhibitors to prevent versican catabolism and versican-induced motility and metastasis may be an interesting area of future study (Table 1).

Manipulation of the versican catabolic pathways may also provide novel therapeutic targets for cancer invasion and metastasis. For example, formation of a pericellular matrix rich in HA and versican implicated in cancer cell motility, could be inhibited by treatment with HA oligomers. Disruption of the HA–CD44 interaction with HA oligomers has been shown to markedly inhibit the growth of B16F16 melanoma cells [45]. In addition, HA oligomers inhibit the formation of receptor tyrosine kinases complexes and their phosphorylation in prostate, colon and breast carcinoma cells [46]. Thus, the use of HA oligomers is a potentially attractive agent to block the formation of large versican–HA aggregates and HA–CD44 interactions, as well as local tumor invasion (Table 1).

Aggrecan degradation during cartilage destruction

Aggrecan is a large molecule found mainly in cartilage and brain, almost exclusively in the form of aggregates with HA (Fig. 1). It is a main matrix organizer also involved in the regulation of cartilage development, growth and homeostasis. Several studies describe its involvement in malignancies. Aggrecan is markedly reduced and degraded at specific sites during cartilage destruction in the progression of laryngeal squamous cell cancer [47,48] (Table 1).

Brevican: roles in gliomas

Brevican and neurocan, found in the central nervous system, affect neuronal attachment and neurite outgrowth (Fig. 1). Upregulation and proteolytic cleavage of brevican increase the aggressiveness of glial tumors significantly [49] and enhance cell adhesion and motility. Brevican promotes EGFR activaton, increases the expression of cell-adhesion molecules and promotes the secretion of fibronectin and accumulation of fibronectin microfibrils on the cell surface [50]. The expression of a novel tumor-specific isoform of brevican that is localized on the cell membrane has been found in all high-grade gliomas and is suggested to play a significant role in glioma progression. In addition, the absence of brevican from benign gliomas prompts its use as a diagnostic marker to distinguish primary brain tumors of similar histology, but different pathologic course [51]. Inhibition of the expression of the tumor-specific isoform of brevican and inhibition of brevican cleavage may be a potential pharmacological target for the treatment of brain tumors (Table 1).

Small leucine-rich PGs

SLRPs are characterized by a protein core with leucine-rich repeats, the presence of N-terminal cysteine clusters and C-terminal ‘ear repeats’ (classes I–III), and at least one GAG side chain [52]. The family of SLRPs is now divided into five classes (see the minireview by Iozzo & Schaefer [53]). Most of these SLRPs have a consensus sequence for modification with GAGs, but some exist as glycoproteins in the tissues. SLRPs are important regulators of various biological processes because leucine-rich repeats are particularly relevant for protein–protein interactions [52]. SLRPs have a curved molecular architecture and bind to collagen fibrils through a core protein, participating in the formation and spacing of the fibrils (Fig. 1). Most studies on the role of SLRPs in malignancy have focused on decorin and lumican, which are typically located around tumor microenvironment.

Decorin signaling in cancer

Decorin represents a powerful tumor cell growth and migration inhibitor by modulating both tumor stroma deposition and cell-signaling pathways. In addition, genetic evidence suggests that a lack of decorin is ‘permissive’ for tumor development [54]. Growth arrest of various tumor cell lines has been associated with induced expression of inhibitors of cyclin-dependent kinase p21 (Fig. 3) [55,56]. Notably, decorin binds directly to EGFR and downregulates its activity as well as the activity of other members of the ErbB family (Fig. 3) [55,57]. These receptors are overexpressed and/or mutated in many cancers driving tumor progression [58]. Decorin can compete with EGF for receptor binding on the cell surface of tumor cells. After binding, the receptor dimerizes and is subsequently internalized and degraded in the lysosomes via caveolin-mediated internalization (Fig. 3) [59]. Decorin inhibits tumor cell proliferation by evoking a signaling cascade that is different from the one evoked by EGF, possibly by inducing a different EGFR conformation and selectively activating phosphotyrosines in the receptor autophosphorylation domain. Decorin also suppresses the activity of ErbB2 and ErbB4 receptors via degradation (Fig. 3) [57,60]. The only exception to this model reported to date has been in MG-63 human osteosarcoma cells, in which the decorin-expressing tumor cells show overexpression paired to constitutive activation of the EGFR favoring cell migration [61]. Decorin also interacts with Met, the receptor for hepatocyte growth factor, and induces transient receptor activation, recruitment of the E3 ubiquitin ligase c-Cbl, and rapid intracellular degradation of the receptor (Fig. 3). Decorin suppresses intracellular levels of β-catenin, one of the key downstream effectors of Met, and inhibits cell migration and growth (Fig. 3). Thus, by antagonistically targeting multiple tyrosine kinase receptors, decorin contributes to the reduction in primary tumor growth and metastastic spreading [62].

Figure 3

SLRPs have a curved molecular architecture and bind to collagen fibrils through the core protein participating in the formation and spacing of the fibrils. Decorin binds to the EGFR and evokes a unique signaling cascade that causes EGFR dimerization, caveolin-mediated internalization and lysosomal degradation. EGFR is phosphorylated upon decorin binding, and induces the expression of p21 leading to growth arrest. Decorin binds also to the c-Met inducing transient activation of the receptor recruitment of c-Cbl and rapid intracellular degradation of the receptor by the proteasome. Upon decorin binding to c-Met, intracellular levels of β-catenin are suppressed inhibiting tumor cell growth and migration.

Decorin as a marker for prognosis and aggressiveness

Decorin expression is altered in various types of cancer (Table 1). Therefore, decorin has been taken into consideration as a possible prognostic marker in cancer patients. It is important to note that decorin expression levels are directly proportional to the amount of tumor stroma. There are very few studies that analyze the prognostic significance of decorin expression levels in terms of patient survival. High expression of decorin in patients with advanced ovarian cancer was associated with a poor response to treatment and a higher incidence of relapse for those patients that initially responded [63]. Reduced amounts of decorin were associated with poor prognosis in node-negative invasive breast cancer [64] and some types of soft tissue tumors [65]. Low decorin levels in liposarcomas and malignant peripheral nerve sheath tumors are also associated with lower disease-free and survival rates (Table 1). Decorin is also deposited in the tumor stroma in nonpalpable breast carcinomas as versican. The accumulation of decorin is more pronounced than that of versican and is also associated with cancer-suggesting mammogram findings (personal data).

In normal tissues, the ratio of CS and DS side chains associated with decorin is balanced. In tumor tissue, such as colon, ovarian, gastric and pancreatic carcinoma, the CS chains become predominant [66–69]. DS is chemically more complex and requires additional enzymes to be synthesized, compared with CS. Therefore, it is reasonable to think that the synthesis of the chemically simpler CS is favored in tumor tissue. In line with this concept, CS side chains have been proposed as being more permissive to cell migration favoring tumor aggressiveness [66]. It is not surprising that, tumor cells, with very few exceptions, do not express decorin, but the mechanism by which decorin expression is switched off is unclear and needs further investigation. Decorin expression in cancer can also be altered by transcriptional, post-transcriptional and post-translational modifications. Notably, decorin production is re-established following treatment with a proteasome inhibitor, suggesting that ovarian cancer cells have developed a mechanism for the rapid degradation of decorin [67]. Potentially, epigenetic control, including hypermethylation of the promoter region, might play a role in silencing the decorin gene. In the tumor stroma, by contrast, hypomethylation of the decorin promoter has been previously demonstrated [1]. In addition, tumor cells can synthesize soluble factors that repress decorin expression by stromal cells [70].

Decorin as potential anticancer agent

Decorin will attract more interest in future as an anticancer therapeutic. Considering that chemotherapy is still the leading therapy for cancer patients and that decorin could be administered in concomitance with various compounds, it is relevant to understand their biological interaction. Decorin shows a synergistic effect with carboplatin in inhibiting ovarian cancer cell growth [71], whereas it antagonizes carboplatin and gemcitabine effects against pancreatic cancer cells [72]. Because of a lack of data and somewhat contrasting results, this area definitely needs more investigation. Considerable effort has been recently applied to prove that decorin can be an antitumor therapeutic in vivo. Adenoviral-mediated delivery of decorin slows the growth of lung, squamous and colon carcinoma tumor xenografts in immunocompromised mice [73], retards mammary adenocarcinoma growth and prevents metastatic spreading to the lungs reducing ErbB2 receptor levels [66]. Ectopic expression of decorin in a rat glioma model prolongs the survival of the animals and the size of the tumor is directly proportional to how early and how much decorin is expressed [74]. Administration of decorin protein core showed that it specifically localizes within the tumor, antagonizes EGFR activity and induces apoptosis in the A431 squamous carcinoma model [75]. The outcome is primary tumor growth inhibition because of the slower growth rate combined with apoptosis and impaired tumor metabolism. Decorin injected systemically can reduce breast tumor growth and metabolism and halt metastatic spread to the lungs [76]. The finding that ectopic expression of decorin can revert the malignant phenotype in several cell lines of various histogenetic backgrounds [56] and can antagonize primary tumor growth and metastases in vivo further raises a hope for the postulated clinical application of decorin and related molecules. Decorin might be utilized in the near future as an adjunct ‘protein therapeutic’ for solid tumors in which receptor tyrosine kinases play a key role. Additional in vivo studies with various tumor models are desirable. Furthermore, studies to elucidate the functional interaction between decorin and existing anticancer chemotherapeutics in order to evaluate potential limitations are needed (Table 1).

Lumican: a promising agent in the cancer fight

Lumican is substituted with keratan sulfate side chains, and inhibits cell proliferation by inducing p21 expression and antagonizes anchorage-independent cell growth as well as cell migration [77]. Cleavage of lumican by the membrane-type matrix metalloprotease-1 abrogates lumican-evoked suppression of colony formation [77]. Interestingly, and on the whole contrary to decorin, lumican is expressed by some tumor cell lines [78] and its inhibition promotes cancer cell growth in some cases. Transfection of B16F1 mouse melanoma cells to express lumican or treatment with recombinant protein induces impaired anchorage-independent growth and the capacity to invade the ECM [79]. Lumican has been reported to be more abundant than decorin in human breast carcinomas [80], even though the clinical relevance of this observation has not yet been explained. More importantly, lumican expression has been proposed as a prognostic factor in lymph node-negative breast cancer [64]. A recent study tested different recombinant and synthetic peptides and found an active site in the leucine-rich repeat 9 domain of the lumican core protein, which is able to inhibit melanoma cell migration [81]. It will be interesting in future to test the effects of recombinant lumican on tumor growth and metastasis. Additional studies carried out with different cancer types are required to dissect the mechanism of action of lumican. However, it appears that lumican targeting is promising and more specific information about its mode of action is needed. SLRPs or peptides derived therefrom could be applied in the fight against cancer, because they represent a class of natural inhibitors of cancer growth (Table 1).

Basement membrane PGs

Basement membranes are thin layers of specialized ECM underlying epithelial and endothelial layers; a specific class of matrix PGs located in basement membranes with particular structural and functional characteristics. Basement membranes are elongated molecules with a collage of domains that share structural and functional homology with numerous ECM proteins, growth factors and surface receptors. This class involves three main, well-characterized members: perlecan, collagen type XVIII and agrin, which are almost universally decorated with HS side chains (Fig. 4) [82]. Their genes are highly conserved and carry disparate biological functions for the maintenance of basement membrane homeostasis, modulation of growth factor activity and angiogenesis. Basement membrane heparin sulfate proteoglycans (HSPGs) have a dual function as pro- and anti-angiogenic factors participating in cancer growth and angiogenesis [82]. They can stimulate angiogenic signaling by sequestering, protecting and concentrating HS-binding growth factors, such as fibroblast growth factor-2 (FGF-2), VEGF and PDGF, through which the HSPG growth factor complex may be presented in a ‘biologically active’ form to the cognate receptors (Fig. 5). By contrast, they contain powerful angiostatic fragments, such as endostatin and endorepellin at their C-termini that are released by proteolytic processing and can act in a paracrine function on sprouting endothelial cells, either locally or distantly (Fig. 5).

Figure 4

Schematic representation showing the structural domains of the basement membrane PGs perlecan, agrin and collagen XVIII.

Figure 5

PGs of the basement membrane bind HS-binding growth factors and stimulate angiogenesis and tumorigenesis by sequestering, protecting and concentrating growth factors. HSPG–growth factor complexes are presented in a ‘biologically active’ form to the cognate receptors promoting their signaling. By contrast, they contain angiostatic fragments, such as endostatin and endorepellin, which are released by proteolytic cleavage. Endorepellin interacts with α2β1 integrin inducing the interaction and phosphorylation of src homology-2 protein phosphatase-1 with integrin α2, triggering a signaling cascade that leads to disruption of the actin cytoskeleton and thus to cytostasis. Endostatin binds to integrins (α5β1, αvβ3, αvβ5), glypican and VEGFR2 affecting several key components of VEGF signaling cascade, stimulating synthesis of thrombospondin (a powerful angiostatic protein) and suppressing c-myc. Endostatin induces reprogramming of the gene expression and disrupts endothelial cell migration.

Perlecan: a critical regulator of growth factor-mediated signaling and angiogenesis

Perlecan binds FGF-2 via HS, promotes receptor activation and ultimately downstream signaling, which supports mitogenesis and angiogenesis (Fig. 5). The lack of HS in perlecan inhibits wound healing and FGF-2-induced angiogenesis and tumor growth [82,83]. Targeted knockdown of perlecan reduced the growth factor response, as revealed by decreased tumor growth and angiogenesis [82,84]. In hepatoblastoma xenografts treated with anti-vascular endothelial growth factor receptor (VEGFR) therapy, vessel recovery over time was associated with an increase in perlecan and heparanase expression around tumor vessels [85], suggesting a synergistic role of heparanase [86] and/or proteases [87] in the liberation of HS-bound VEGF and subsequent VEGFR2 activation (Fig. 5). The C-terminal domain of perlecan, called endorepellin, blocks endothelial cell migration and capillary morphogenesis both in vitro and in vivo [82]. Endorepellin interacts specifically with the α2β1 integrin [88], and stimulation with endorepellin induces the interaction and phosphorylation of Src homology-2 protein phosphatase-1 with integrin α2 in a dynamic fashion in endothelial cells [89], triggering a signaling cascade that leads to disruption of the actin cytoskeleton and thus to cytostasis (Fig. 5) [88]. Systemic delivery of human recombinant endorepellin to tumor xenograft-bearing mice causes a marked suppression of tumor growth and metabolic rate mediated by a sustained downregulation of the tumor angiogenic network [90]. The distal laminin-like globular domain (LG3) possesses most of the biological activity [82,88] and can be released from the parent molecule by bone morphogenetic protein-1/Tolloid-like metalloproteinases [91]. LG3 has been detected in pathological conditions including pancreatic, colon and breast cancer [82]. It has been proposed that endorepellin/LG3 is liberated via partial proteolysis during tissue remodeling and cancer growth, thereby representing an additional layer of control for angiogenesis [82]. One possibility is that tumor growth might be enhanced in vivo by a lack of circulating LG3. In line with this, circulating LG3 levels are reduced in patients with breast cancer, suggesting that reduced titers might be a useful biomarker for cancer progression and invasion [92]. The fragment in circulation might be continuously released to add an additional layer of control for angiogenesis during cancer progression. Utilization of the antiangiogenic fragment as either a protein- or peptide-based pharmacological agent might represent a novel therapeutic rationale, especially when provided in combination with other tumor-suppressive compounds (Table 1).

Collagen XVIII–endostatin: innovative angiogenic regulators

Collagen XVIII has been suggested to play a negative regulatory, but not essential, role in angiogenesis in certain contexts. Collagen type XVIII harbors the C-terminal antiangiogenic fragment, endostatin, proteolytically derived from the C-terminus of collagen type XVIII [82,93]. The expression of collagen XVIII and endostatin has been studied in ovarian, hepatocellular, pancreatic and oral squamous cell carcinoma and found to vary among different cancer types [94]. Endostatin is a potent antiangiogenic molecule reducing tumor growth, choroidal neovascularization and wound healing [82]. High levels of circulating endostatin reduce tumor burden, block the formation of pulmonary metastases [95] and, notably, induce a total gene expression reprogramming [96], which ultimately disrupts endothelial cell migration (Fig. 5) [97]. Endostatin has been linked to cell-surface receptors including various integrins, glypican and VEGFR2. Specifically, endostatin downregulates several key components of the VEGF signaling cascade and, at the same time, stimulates the synthesis of thrombospondin, a powerful angiostatic protein and suppresses c-myc (Fig. 5) [82,96]. Endostatin has been studied in phase I and II clinical trials for patients with metastatic cancer and has shown low efficacy, however, a new more stable version of endostatin has re-entered the clinic and is now used in certain countries for the treatment of lung and gastric cancer [94]. Recently, an immunoisolation device that contains endostatin-expressing cells was used effectively for the treatment of melanoma and Ehrlich tumors in mice [98]. This suggests that the macroencapsulation of engineered cells that produce endostatin may be innovate therapeutic strategy for treatment of malignancies (Table 1).

Agrin: roles in tumor angiogenesis

Agrin is strongly expressed around blood vessels [82]. Agrin can be considered to be a marker of tumor angiogenesis in the liver because it is markedly deposited in proliferating bile ductules, in newly formed septal vessels in hepatic cirrhosis and in the angiogenic network of malignant hepatocellular carcinomas (HC) [99]. Agrin overexpression in newly formed blood vessels is also present in cholangiocarcinoma [100]. The role of agrin in tumor angiogenesis is not yet established, but initial studies suggest that agrin’s effect on tumor angiogenesis is likely context dependent. It seems that the expression of agrin may be protective against disorganized angiogenesis in glioblastomas, whereas in hepatic malignancies it seems to support tumor angiogenesis, at least in the initial stages of tumor development. There remains very limited knowledge of whether C-terminal endorepellin-like fragment of agrin signals through the integrin receptors. Further research on this area will improve our knowledge for possible targeting of agrin in tumors (Table 1).

Cell-surface PGs

The cell-surface PGs include two major subfamilies: the syndecans, with four members cloned in mammals, which are type I transmembrane proteins mostly substituted with HS chains; and glypicans linked with a glycosyl-phosphatidylinositol anchor to the cell membrane, with six members cloned in mammals substituted with HS chains (Fig. 6). Syndecans and glypicans are generally considered to act as coreceptors for heparin-binding mitogenic growth factors [2].

Figure 6

Localization and classification of cell-surface associated and intracellular PGs. The cell-surface PGs encompass mainly the transmembrane PGs syndecans and the glycosyl-phosphatidylinositol-anchored glypicans. Serglycin is the only characterized intracellular PG.

Glypicans influence tumor development and progression

Glypicans show cell-type and developmental-stage-specific expression. They are involved in fundamental biological processes such as cell–ECM interactions and the control of cellular division, differentiation and morphogenesis [101]. At the level of signaling, they are involved in the regulation of pathways including Wnt, FGF, Hedgehog (Hh), bone morphogenic protein, Slit and insulin-like growth factor [101]. Therefore, depending on the biological context, glypicans can either stimulate or inhibit signaling activity. Notably, the HS chains are essential for the glypican-induced stimulation of FGF activity, and partially required for the regulatory activity of glypicans in Hh, Wnt and bone morphogenetic protein signaling [101–104].

Glypicans influence tumor development and progression, and their expression is abnormal in various human tumors (Table 1). For example, glypican-1 is upregulated in breast and pancreatic cancer, and it has been postulated that this aberrant expression of glypican-1 may play a key role in promoting growth factor signaling in cancer cells [105,106]. By contrast, glypican-3 gene is mutated in patients with Simpson–Golabi–Behmel syndrome, deregulating the balance between cell proliferation and apoptosis [107]. The loss of glypican-3 induces overgrowth observed in this syndrome. Patients with Simpson–Golabi–Behmel syndrome are at increased risk of embryonic tumors. Glypican-3 inhibits the signaling of Hh by competing with Patched, the Hh receptor, for Hh binding [102]. The binding of Hh to the receptor Patched triggers the signaling pathway by blocking the inhibitory effect of Patched on Smoothened. Glypican-3 competes with Patched for Hh binding allowing to ligand-free Patched to inhibit Smoothened, reducing signaling and cell growth (Fig. 7). The binding of Hh to glypican-3 triggers its endocytosis and degradation further strengthens the negative regulatory role of glypican-3 in Hh signaling (Fig. 7). Glypican-3 plays a negative role in cell proliferation and induces apoptosis in mesothelioma and breast cancer that is only dependent on the protein core [108]. Glypican-3 expression is downregulated in tumors with different histogenetic backgrounds as a result of hypermethylation of the glypican promoter, and its expression can be restored by treatment with demethylation agents [109–112]. Hh signaling pathway is hyperactivated in a proportion of these cancers [113,114] and it is therefore possible that the loss of glypican-3 may contribute to cancer progression by inducing the activation of Hh signaling similarly to Simpson–Golabi–Behmel syndrome. By contrast, upregulation of glypican-3 is observed in embryonic tumors, such as neuroblastoma and Wilm’s tumor [115]. In tumors originating from tissues that express glypican-3 only in the embryo, such as HC and melanoma, its expression tends to reappear with malignant transformation [101]. The canonical Wnt pathway plays a key role in most HC and overexpression of glypican-3 promotes tumor growth-promoting Wnt signaling. The binding of Wnt to glypican-3 facilitates and/or stabilizes the interaction of Wnt with the receptor Frizzled with the consequent increment on signaling (Fig. 7). HC cells forced to overexpress a soluble form of glypican-3 showed lower tumorogenicity because soluble glypican-3 blocked the activity of several pro-tumorigenic growth factors. Glypican-3 may promote local cancer growth in some cancer tissues, whereas it inhibits tissue invasion and metastasis in others.

Figure 7

The binding of Wnt to glypican-3 facilitates and/or stabilizes the interaction of Wnt and Frizzled with the consequent increament on signaling. The binding of Hh to the receptor Patched triggers the signaling pathway by blocking the inhibitory effect of Patched on Smoothened. Glypican-3 competes with Patched for Hh binding allowing ligand-free Patched to inhibit Smoothened, reducing signaling and tumor growth. The binding of Hh to glypican-3 triggers endocytosis and degradation of the complex. Serglycin interacts with and stores bioactive molecules inside storage granules and secretory vesicles. Upon secretion, serglycins modulate their activities through protection, transport, activation and delivery of the bioactive molecules.

Glypican-3 is a novel tumor marker for early-stage melanoma [116] and early-stage HC [117]. It is not known, however, whether its elevated serum levels are important in tumor progression or are simply a reflection of aggressive tumors. The high levels of shed glypican may arise from increased expression of the PG or increased activity of tumor proteases. In this light, glypican-3 may be a candidate antigen for cancer immunotherapy in HC (Table 1). It is strongly expressed exclusively in HC, is highly immunogenic and stimulates eradication by T cells of tumors expressing glypican-3 in mice and markedly inhibited growth of an established tumor. Glypican-3-derived peptide-pulsed vaccination is a novel strategy to prevent HC and melanoma in patients and needs to be further developed as an anticancer therapy [116]. Glypican-1 on target cells is recognized by some natural cytotoxicity receptors of natural killer cells and the suppression of glypican-1 in pancreatic cancer led to lower cytotoxic effects of natural killer cells [118] (Table 1).

Syndecans: multiple roles in cancer progression and strategies for their targeting

Syndecan-1 is present at early stages during development and on epithelial and cancer cells in adults. Syndecan-2 is distributed in mesenchymal tissues, liver and neuronal cells. Syndecan-3 is mainly associated with neural tissues, whereas syndecan-4 is ubiquitously distributed [2]. Syndecans are involved in complex signaling events through which they regulate cell proliferation, differentiation, adhesion and migration [119]. For the multiple roles of syndecan shedding see the accompanying minireview by Manon-Jensen et al. [120].

Syndecan-1 is expressed in various human cancers and correlates with tumor recurrence in human prostate cancer [121]. Furthermore, it is upregulated in human breast cancer and correlates with poor prognosis [122]. Syndecan-1 expression by fibroblasts in the tumor-associated stroma appears to be necessary for breast carcinogenesis [123] and may promote tumorigenesis by regulating tumor cell spreading and adhesion [124], proliferation and angiogenesis [125,126]. It is suggested that cell-surface syndecans promote proangiogenic signaling by binding FGF-2 and VEGF and presenting them to their high-affinity receptors and also by protecting them from inactivation (Fig. 8) [127,128]. Notably, syndecan-1 also mediates cell adhesion in cooperation with integrins by activation of actin and fascin bundling (Fig. 8) [119,124]. By contrast, syndecan-1 is downregulated in various malignancies and a possible inhibitory function is suggested [119,129–131]. Syndecan-1 may serve as a prognostic marker in a cancer-type-specific manner (Table 1). The effects of syndecan-1 on tumor progression and prognosis probably also depend upon whether syndecan-1 is tumor cell derived or synthesized by the host stroma, because its expression by reactive tumor stromal cells promotes tumor progression [123,125,126,132–135]. Soluble syndecan-1 ectodomain is present in the serum of various malignancies. A variety of proangiogenic/protumor growth factors and enzymes that are upregulated in solid tumors in vivo (e.g. FGF-2 and MMPs) accelerate syndecan-1 shedding in vitro [136,137], presumably by promoting cleavage of the ectodomain [138]. Soluble syndecan-1 accelerates tumor growth in vivo and increased cell invasion through collagen in vitro [139]. Soluble syndecan-1 may activate proangiogenic/protumor factors, like FGF-2 [140], also bind to mitogens, effectively creating a chemotactic gradient [141] and/or interact with proteases protecting them from contact with endogenous inhibitors (Fig. 8) [119].

Figure 8

Syndecans interact with matrix molecules in cooperation with integrins, inducing membrane ruffling and stress fiber formation, promoting cell adhesion and migration. Syndecans bind to growth factors presenting them to their high-affinity receptors, activating signaling pathways that promote cell proliferation and motility. Soluble syndecans interact with growth factors protecting them from degradation and creating a promigratory chemotactic gradient. Soluble syndecans present growth factors to their high-affinity receptors, activating signaling pathways that promote cell proliferation and motility. Soluble syndecans bind to matrix molecules, competitively inhibiting cell-surface syndecans, and thereby stimulating tumor and endothelial cell migration.

Overexpression of syndecan-2 has been found in numerous malignancies, such as Lewis lung carcinoma, ovarian, osteosarcoma, brain tumors, mesothelioma and colon cancer [119,142–144] (Table 1). This PG promotes tumorigenesis in colorectal cancer and Lewis lung carcinoma by regulating cell adhesion, spreading and cytosketetal organization [142,144]. Recently, we found that expression of syndecan-2 in breast cancer cells is regulated by estradiol through the action of estrogen receptor alpha [145]. Syndecan-2 is highly expressed in the microvasculature of mouse glioma tumors and promotes angiogenesis, at least in part, by modulation of endothelial cell adhesion [143]. Exogenous addition of EGF, FGF-2, VEGF, MMP-2 and MMP-9 to brain microvascular endothelial cells increases syndecan-2 ectodomain shedding [143]. The released ectodomain may act on tumor cells and/or the endothelium to modulate tumorigenesis. Soluble syndecan-2 ectodomain causes significant increase in endothelial cell membrane protrusion, migration, capillary tube formation and cell–cell interactions most probably because of competitive inhibition of cell-surface syndecan- 2 (Fig. 8) [119,143]. Colorectal carcinoma cells demonstrate reduced cell adhesion in the presence of soluble syndecan-2 ectodomain [143], whereas overexpression of full-length syndecan-2 in lung carcinoma cells induces stress fiber formation and cytoskeletal structures indicative of strong adhesion; cells also become nonmotile [142]. Cell-surface syndecan-2 is also needed for the correct organization of laminin and fibronectin, and its reduced interaction with matrix ligands because of the presence of soluble ectodomain would be expected to decrease basement membrane assembly by competing for laminin and fibronectin binding [146]. Syndecan-2 ectodomain may exert its proangiogenic effect by binding to growth factors and cytokines [119], effectively creating a chemotactic gradient similar to that produced by soluble syndecan-1 (Fig. 8) [141].

Syndecan-4 is a focal adhesion component in a range of cell types and mediates breast cancer cell adhesion and spreading [119]. It also forms a complex with the proangiogenic molecule, FGF-2, and its receptor, fibroblast growth factor receptor (FGFR-1) promoting FGF signaling (Fig. 8) [147]. The attachment of syndecan-4 to fibronectin initiates intracellular signaling, including protein kinase Cα and focal adhesion kinase activation, to promote focal adhesion formation (Fig. 8) [148]. FGF-2 treatment of melanoma cells resulted in reduction in the expression of syndecan-4, a decrease in cell attachment on fibronectin and promoted cell migration [149]. Syndecan-4 overexpressing cells form larger and more dense focal adhesions, correlated with stronger attachment and decreased cell migration [150], whereas lack of syndecan-4 engagement promotes an amotile fibroblast phenotype in which focal adhesion kinase and Rho signaling are downregulated and filopodia are extended [151] (Table 1). Syndecan-4 is highly expressed in less-aggressive seminomatous testicular tumors. Reduced levels are detected in more-aggressive non-seminomatous germ cell tumors and it correlates with the metastatic potential of these tumors (personal data). Interestingly, syndecan-4 is found also in the tumor stroma in some testicular tumors and the presence of stromal-associated syndecan-4 correlates with increased neovascularization. These results indicate that a directed homeostasis in syndecan-4 levels supports the optimal migration of tumor cells and its expression by stromal cells may facilitate cancer cell growth and angiogenesis, similarly to syndecan-1.

The antitumor effect of murine/human chimeric syndecan-1-specific monoclonal antibody nBT062 conjugated with highly cytotoxic maytansinoid derivatives against multiple myeloma cells in vitro and in vivo was examined. Treatment significantly inhibited multiple myeloma tumor growth in vivo and prolonged host survival in both the xenograft mouse models of human multiple myeloma and the SCID-hu mouse model. These results provide a preclinical framework supporting the evaluation of nBT062-maytansinoid derivatives in clinical trials to improve patient outcome in multiple myeloma [152] (Table 1). Another strategy for syndecan targeting in the treatment of malignancies is the inhibition of heparanase an enzyme that degrades HS (see the accompanying minireview by Barash et al. [153]). Heparanase activity is strongly implicated in structural remodeling of the extracellular matrix, a process which can lead to invasion by tumor cells. In addition, heparanase augments signaling cascades leading to enhanced phosphorylation of selected protein kinases and increased gene transcription associated with aggressive tumor progression. This function is mediated by a novel protein domain localized at the heparanase C-terminus. Heparanase is also an important regulator of syndecan clustering, shedding and mitogen binding. Recent data indicate that modified glycol-split heparin, as well as other heparanase inhibitors, can profoundly inhibit the progression of tumor xenografts produced by myeloma and carcinoma cells, thus moving antiheparanase therapy closer to reality [154] (Table 1).

Intracellular PGs

Serglycin is the only characterized intracellular PG found in hematopoietic and endothelial cells (Fig. 6). It has important functions related to the formation of several types of storage granules. Serglycin interacts with several bioactive molecules (e.g. histamine, tumor necrosis factor α and various proteases) and is important for the retention of these key inflammatory mediators inside storage granules and secretory vesicles. Serglycin can further modulate the activities of partner molecules in different ways after secretion from activated immune cells, through protection, transport, activation and interactions with substrates or target cells (Fig. 7) [155]. Few reports involve serglycin in malignancies. Niemann et al. [156] demonstrated the expression of serglycin in various hematologic malignancies. Serglycin expression was found to distinguish acute myeloid leukemia from acute lymphoblastic leukemia. It was found to be a selective marker for immature myeloid cells, distinguishing acute myeloid leukemia from Philadelphia chromosome-negative chronic myeloproliferative disorders. The expression and constitutive secretion of serglycin has also been reported in multiple myeloma, where the released serglycin inhibits bone mineralization [157]. The role of serglycin in hematological malignancies remains unkown and further studies will improve our knowledge of its involvement and potential pharmacological targeting (Table 1).

Acknowledgement

We thank Prof. R.V. Iozzo for critical reading and valuable advice.