Role of hyaluronan in glioma invasion

Abstract

Gliomas are the most common primary intracranial tumors. Their distinct ability to infiltrate into the extracellular matrix (ECM) of the brain makes it impossible to treat these tumors using surgery and radiation therapy. A number of different studies have suggested that hyaluronan (HA), the principal glycosaminoglycan (GAG) in the ECM of the brain, is the critical factor for glioma invasion. HA-induced glioma invasion was driven by two important molecular events: matrix metalloproteinase (MMP) secretion and up-regulation of cell migration. MMP secretion was triggered by HA-induced focal adhesion kinase (FAK) activation, which transmits its signal through ERK activation and nuclear factor kappa B (NF-κB) translocation. Another important molecular event is osteopontin (OPN) expression. OPN expression by AKT activation triggers cell migration. These results suggest that HA-induced glioma invasion is tightly regulated by signaling mechanisms, and a detailed understanding of this molecular mechanism will provide important clues for glioma treatment.

Malignant gliomas are highly invasive and infiltrative tumors that have a poor prognosis with a median survival of only one year.1,2 A major barrier to effective malignant glioma treatment is the invasion of these cells into brain parenchyma. Because of this fact, local therapies such as surgery or radiation therapy are not effective.3 Glioma cells invade through the ECM of the brain by activating a number of coordinated cellular programs, which include those necessary for migration and invasion.3 Therefore, a detailed understanding of the mechanisms underlying this invasive behavior is essential for the development of novel effective therapies.

During glioma invasion, tumor cells closely interact with the ECM. Although brain tumor cells may share some of the invasive characteristics with tumors that arise outside of the central nerve system (CNS), the particular structure and composition of the brain ECM suggest the existence of unique invasive mechanisms for brain tumors.4

Brain ECM is composed of typical ECM proteins and a HA scaffold with associated glycoproteins and proteoglycans.5 Typical ECM proteins such as laminin, type-IV collagen and fibronectin have been implicated in the invasion of other tumors by regulating cell adhesion and migration.6 However HA, which is associated with proteoglycans and GAGs, is especially abundant in the brain parenchyma compared to other tissues.7 Furthermore, malignant gliomas contain higher amounts of HA than normal brain tissue.7 These facts raise the possibility that HA might play an important role in glioma invasion, a process that is distinct from other non-CNS derived tumors.

HA Structure and Function

HA is a simple but unusual polysaccharide. It is a member of GAG family and is synthesized as a large, negatively charged, unbranched polymer that is composed of repeating disaccharides of glucuronic acid and N-acetylglucosamine.8–10 It has a simple chemical structure but differs in many ways from other GAGs. First, there are 10,000 or more disaccharide repeating units in HA, making it 10³–10⁴ kDa in molecular weight. Second, unlike other GAGs, which include heparan sulphate and chondroitin sulphate, HA contains no sulfate groups or epimerized uronic acid residues. Third, HA is synthesized at the inner face of the plasma membrane as a free linear polymer without any protein core, while other GAGs are synthesized by resident Golgi enzymes and covalently attached to core proteins.10–12

As an ECM component, HA plays a dual function. In normal ECM, HA provides tissue homeostasis, biomechanical integrity, and structure and assembly of tissues.13 However in malignant tumor tissues, HA transmit signals into cytoplasm and increases tumor cell proliferation, motility and invasion.14,15 However, the molecular mechanism of HA-induced tumor progression is divergent and tissue specific.

Expression of HA in Cancer

HA contributes to certain types of cancer development. In addition to extracellular HA, intracellular and nuclear forms of HA have been detected. Intracellular HA is involved in cell signaling, whereas nuclear HA could promote chromatin condensation and thus facilitate mitosis.16 HA expression is frequently increased in malignant tumors.12 Histological studies demonstrate that HA concentrations are usually higher in malignant tumors than in corresponding benign or normal tissues.12 HA levels can be increased around tumor cells themselves or within the tumor stroma.7 Boregowda et al. reported that highly differentiated tumors such as lung, breast, colon, kidney, prostate, astrocytomas and non-Hodgkin's lymphoma expressed increased amounts of HA in both tumor epithelia and the intratumoral areas.17

HA and Prognosis

The level of HA in tumor cells is predictive of malignancy and often correlates with cancer aggressiveness in patients with breast cancers, ovarian carcinomas, non-small-cell lung cancers and prostate cancer.18–22 For instance, a high proportion of HA-positive cancer cells and a high intensity of the HA-signal predicts a poor survival rate in colon carcinomas.23 In addition, Auvinen et al. reported that both the intensity of stromal HA signal alone and that combined with the HA positivity in tumor cells were independent prognostic factors for overall survival in breast carcinoma patients.18

HAS in Cancer

HA is made by HA synthases (HAS1, HAS2 and HAS3). Dysregulation of HAS genes results in abnormal production of HA and promotion of abnormal biological processes such as transformation and metastasis. It is thought that HA facilitates tumor growth by opening up spaces for the tumor to migrate to and by interacting with HA binding molecules, assisting in tumor cell adhesion and migration.24 HA-induced tissue hydration physically creates spaces through which tumor cells may migrate and invade. HA-rich matrices within the tumor-associated stroma are also infiltrated by newly forming blood vessels.12 A number of cancers are associated with elevated expression of HAS.16,25 In animal models, the overexpression of HAS promotes growth of fibrosarcoma and prostate carcinoma, as well as metastasis of mammary carcinoma.26–28

HA as a Cancer Marker

HA levels are often increased in the sera of patients with a variety of different tumors.29–31 It has also been shown that HA is significantly elevated in the sera of patients with metastatic disease compared to the sera of patients without metastatic disease.29 In addition, Urinary HA and HAase levels correlate with levels that can be detected in tissues.32 Likewise, elevated levels of HA and HAase in the urine form a clinically reliable marker for the presence and grade of bladder cancer and prostate cancer.22,32–36 Therefore, HA is of considerable interest as a tumor marker in different cancers.

HA Receptors and Intracellular Signaling

Newly synthesized HA may be secreted and subsequently interact with several cell surface receptors such as cluster determinant 44 (CD44), receptor for hyaluronate-mediated motolity (RHAMM), lymphatic vessel endothelial HA receptor (LYVE-1), hyaluronan receptor for endocytosis (HARE), liver endothelial cell clearance receptor and Toll-like receptor-4. Of these, CD44 and RHAMM are well known as signal-transducing receptors that influence cell proliferation, survival and motility. Furthermore, they are known to be closely related to tumor progression.37–43

CD44

CD44, considered the principal receptor for HA, is a multifunctional single-pass transmembrane glycoprotein consisting of four functional domains: an amino terminal domain, stem structure, transmembrane domain and cytoplasmic domain.44 The amino terminal domain contains motifs that provide docking sites for the ECM components such as HA and other GAGs.45 The stem structure domain links the amino-terminal domain and transmembrane domain. Stem structure is enlarged by alternative splicing in cancer cells. Although some splice-forms include motifs for specific post-translational modification, little is known about the functional significance of an enlarged segment. The transmembrane region consists of 23 hydrophobic amino acids and cysteine residues which seem to be involved in oligomerization. This region might be responsible for the association of CD44 with lipid rafts. The cytoplasmic domain of CD44 has been linked to the cytoskeleton through interactions with proteins such as ankyrin and ERM proteins.46,47 This interaction has been implicated in cell adhesion and motility. Moreover, the cytoplasmic tail of CD44 interacts with many regulatory and adaptor molecules of cell signaling such as Src kinases, Rho GTPases, VAV2, GAB1, etc.48 The interaction of CD44 with the cytoskeleton and various signaling molecules plays a pivotal role in promoting metastatic-specific tumor phenotypes such as MMP-mediated matrix degradation, tumor cell growth, migration and invasion.12,39,41,49–51

RHAMM

RHAMM is also well-known as a HA-binding protein. RHAMM is expressed on the cell surface and in the cytoplasm, as well as in the cytoskeleton and nucleus. Like CD44, RHAMM is subject to alternative splicing, particularly during tissue repair or in cancer cells.40,52–54 Interactions of HA with RHAMM can trigger a number of cellular signaling pathways including those that involve protein kinase C, FAK, MAP kinases, NFκB, RAS, phosphatidylinositol kinase (PI3K), tyrosine kinases and cytoskeletal components.12,48,55–58 Although it is clear that CD44 and RHAMM can participate independently in proliferative and migratory phenomena, their relative contributions to any given event have not been fully resolved in most cases, and it is likely that they have redundant or overlapping functions in some situations. In general, the interactions of HA with CD44 and RHAMM are especially important for tumorigenesis and tumor progression by activating downstream signaling molecules, but how this receptor transmits signal to downstream targets is still unclear.

Coupling with Other Receptors

CD44 is tightly coupled with receptor tyrosine kinases, p185HER2 and non-receptor tyrosine kinases, c-Src family kinases.59,60 CD44 and p185HER2 are physically linked to each other via interchain disulfide bonds and HA can stimulate CD44-associated p185HER2 tyrosine kinase activity that leads to increased tumor cell growth.59 The cytoplasmic domain of CD44 binds to c-Src kinase at a single site with high affinity.60,61 Importantly, HA interaction with CD44 stimulates c-Src kinase activity, increasing tyrosine phosphorylation of cytoskeletal proteins such as cortactin. The binding of HA to CD44 isoforms, which complex with p185HER2 and c-Src kinase, likely triggers direct coupling with two tyrosine kinase-linked signaling pathways during tumor progression. This pathway can explain partly how HA receptors activate downstream targets to promote tumor progression.

Lipid Rafts

Although a number of binding partners of CD44 have been reported, the mechanisms of signal transduction via CD44 remain poorly understood. One of the most important signaling events following stimulation via CD44 is tyrosine phosphorylation of intracellular protein substrates. The Src-family non-receptor PTKs such as Lck, Fyn, Lyn and Hck were shown to be coupled to CD44.62–65 Moreover, CD44 resists solubilization in nonionic detergents, and recent investigations have shown that CD44 molecules lacking the cytoplasmic tail are still detergent insoluble, possibly because of their association with Triton X-100 insoluble lipids.66,67 These results provide an important link between the CD44 signaling pathway and lipid rafts because this behavior is similar to that of most glycosylphosphatidyl inositol (GPI) anchored glycoproteins, which are mostly confined to lipid rafts.68 Co-isolation of CD44 with micro-domains strongly suggests that CD44 generates cellular activation signals utilizing the signaling machinery of the lipid rafts.

HA-induced MMP Secretion in Glioma Invasion

Glioblastoma is a severe type of primary brain tumor and irreversibly infiltrate the normal CNS by the interaction with ECM. One of the major component in brain ECM is HA, and glioma cells express high levels of the HA receptors, CD44 and RHAMM.7,69,70 Both of these HA receptors were known to play an important role for glioma migration and invasion.53,69–72

Another important player of glioma invasion is of MMPs. Several studies have demonstrated the increased expression of MMPs in gliomas, and GBM and anaplastic astrocytomas. They express higher levels of MMP-2 and -9 than do low-grade astrocytomas.73 However, the relationship between HA signaling and MMP secretion was not well understood.

PTEN is one of the most frequently mutated tumor suppressors in GBM. PTEN also is known to suppresses glioma invasion by regulating MMP secretion.74 Moreover, we reported that PTEN suppresses HA-induced secretion of MMP-9 and inhibits HA-induced invasion in U87MG cells, probably via dephosphorylation of FAK (see Fig. 1).74 We further addressed that FAK is direct substrate of PTEN as a protein phosphatase.74 These results suggest that PTEN mutation is inevitable for the HA-induced glioma invasion by regulating MMP-9 secretion.

FAK, one of main substrates of the PTEN protein, is upregulated in GBMs, particularly in invasive zones. FAK overexpression has been correlated with the invasive potential of a tumor and poor patient prognosis.75,76 In HA-stimulated glioma invasion, FAK is required for the Ras-ERK 1/2 signaling pathway and modulates MMP-9 secretion (Fig. 1).77,78

Another important signaling event which regulates MMP expression is the transcriptional activation of genes whose promoter sequences contain putative binding sites for AP-1 and/or NFκB.79 NFκB is in a family of transcription factors that have been shown to be involved in gene regulation of cellular processes like inflammation, immune response, cell proliferation and apoptosis.80 NFκB activity is regulated by the endogenous inhibitor IκBα; interaction of NFκB with IκBα blocks the nuclear transport signal and keeps it sequestered in the cytoplasm. Following any kind of stimulation, IκBα is phosphorylated at serine residues 32 and 36, which leads to its ubiquitination and degradation. The free NFκB then translocates to nucleus and activates the transcription of target genes.81 A number of upstream kinases such as NFκB-inducing kinase (NIK), PI3K and MEKK play significant roles in regulation of activation of IKK. In a recent study, we have demonstrated that HA elevates the levels of MMP-9 mRNA through NFκB activation (Fig. 1).82 We further showed that HA-induced NFκB activation through phosphorylation and degradation of IκBα by activating IKK was mediated by FAK phosphorylation (Fig. 1). These results suggest that regulation of FAK is the most important step in blocking HA-induced MMP-9 secretion.

HA-Induced OPN Expression in Glioma Migration

As dicussed above, glioma invasion is enhanced not only by MMP secretion but also by upregulation of motility. Several studies have suggested that HA plays an important role in glioma cell motility and invasion, but the molecular mechanisms of HA-associated motility were not well understood.74,82 Recently, we identified group of genes which can be differentially regulated by HA treatment and demonstrated that OPN, one of the transcriptional targets of HA, is responsible for the stimulation of glioma motility by HA treatment (Fig. 1).83 OPN, a highly phosphorylated glycoprotein of the ECM, is known to promote cell attachment, spreading and motility, as well as more complex events like vascular remodeling, bone mineralization and tumor metastasis by interacting with its cell surface receptors CD44 and the α-containing integrins.84,85 Interestingly, many of the basic cellular functions affected by OPN are also controlled by HA. Moreover, OPN and HA are frequently overexpressed in a variety of malignant cells and the expression levels of OPN and HA have been correlated with degrees of glioma malignancy.86 These results suggest that OPN induction might be involved in the stimulation of HA-induced cell motility.

PTEN also suppresses migration, as genetic deletion of the PTEN tumor suppressor gene promotes cell motility and PTEN reconstitution or overexpression inhibits cell motility in a variety of cell type.87,88 The signaling mechanism that upregulates OPN expression was also under the control of PTEN activity.85 OPN expression was upregulated by the PI3K/AKT/mTOR pathway in U87MG cells after treatment by HA in PTEN deleted glioma cells and reconstitution of PTEN suppressed OPN expression (Fig 1).83 These results suggest that PTEN suppress glioma motility and invasion by the inhibition of HA-induced OPN expression and MMP-9 secretion.

Emodin as an Anti-Cancer Drug Against Glioma Invasion

Since HA is a primary ECM component that can stimulate glioma invasion, it makes inhibitors of this signaling as attractive candidates for therapeutic agents. Among protein tyrosine kinase inhibitors, we found that emodin (3-methyl-1,6,8-trihydroxyanthraquinone) has strong anti-invasive activity for HA-induced glioma invasion (Fig. 1).82 Emodin is one of the main active components contained in the root and rhizome of Rheum palmatum L. Emodin has been shown to have a number of biological activities, including antiviral, antimicrobial, immunosuppressive, hepatoprotective, anti-inflammatory and anticancer effects.89–91 Especially in glioma cells, emodin was known to suppress the TGFbeta and FGF-2 induced expression of syndecan-1, a major cell surface heparin sulfate proteolycan, which was specific to malignant gliomas and was absent in normal brain tissues and it may participate in the motility of glioma cells.92 Emodin suppressed the expression of MMP-2 and MMP-9 at both transcriptional and translational levels.82 Pharmacological studies indicated that emodin can significantly decrease the activation of FAK, AKT and ERK1/2, thereby suppressing MMP production (Fig. 1).82 We further observed that emodin efficiently suppressed the HA-stimulated AP-1 and NFκB promoter activities.82 In an in vivo tumor progression model, the activity of MMP-9 was suppressed by emodin treatment.82 These results indicate that oral administration of emodin effectively suppresses MMP-9 expression in human glioma through inhibition of AKT and ERK1/2 activation in vivo. These findings suggest that emodin may be a clinically valuable anti-cancer agent against gliomas by blocking FAK and AKT activation, which is the key signaling event of HA-induced MMP secretion and cell motility.

Conclusions and Future Directions

Understanding of molecular mechanisms of glioma invasion is critical in developing novel therapeutic strategies or treatments because major therapeutic approaches such as surgery and radiotherapy are useless without blocking glioma invasion. In this review, we have demonstrated that HA is a major player in the brain ECM that cooperates with PTEN-deletion during glioma invasion. These results provide strong evidence for the emerging concept in cancer research that the microenvironment modulates the effects of genetic alterations.

PTEN is a tumor suppressor that plays important roles by blocking HA-induced FAK activation and AKT activation. FAK and AKT activation is important for the secretion of MMP and motility of cells. By using emodin, a protein tyrosine kinase inhibitor, we have demonstrated that blocking HA signaling partially reduced tumor size. These results suggest that chemotherapeutic treatments that can effectively block HA induced signaling might be clinically valuable as an anti-invasive agent for gliomas.

Figures and Tables

Figure 1 Inhibition mechanism of emodin on HA-induced glioma invasion and motility. HA induced the invasion of glioma cells by the induction of MMP-9 through the RaS/FAK/ERK 1, 2 activation. In addition, NFκB translocation by FAK activation is also important for the MMP-9 expression. PTEN can effectively modulate the expression of MMP-9 by the dephosphorylation of FAK. HA induced the motililty of glioma cells by the induction of OPN through the PI3K/AKT/mTOR pathway. PTEN can effectively modulate the expression of OPN, and induced OPN contributes to HA-induced cell migration in glioma cells. Emodin suppresses HA-induced MMP-9 and OPN expression through the inhibition of FAK, and AKT activation. Emodin as a protein tyrosine kinase inhibitor can block protein kinases which are important for the HA-induced glioma invasion and motility.

Acknowledgements

This work was supported by the National Cancer Center Grant 0410052-3.