Tumor cells use a wide variety of post-translational mechanisms to modify the functional repertoire of their transcriptome. One emerging but still understudied mechanism involves the export of cytoplasmic proteins that then partner with cell-surface receptors and modify both the surface-display kinetics and signaling properties of these receptors. Recent investigations demonstrate moonlighting roles for the proteins epimorphin, FGF1, FGF2, PLK1 and Ku80, to name a few, during oncogenesis and inflammation. Here, we review the molecular mechanisms of unconventional cytoplasmic-protein export by focusing on the mitotic-spindle/hyaluronan-binding protein RHAMM, which is hyper-expressed in many human tumors. Intracellular RHAMM associates with BRCA1 and BARD1; this association attenuates the mitotic-spindle-promoting activity of RHAMM that might contribute to tumor progression by promoting genomic instability. Extracellular RHAMM-CD44 partnering sustains CD44 surface display and enhances CD44-mediated signaling through ERK1 and ERK2 (ERK1/2); it might also contribute to tumor progression by enhancing and/or activating the latent tumor-promoting properties of CD44. The unconventional export of proteins such as RHAMM is a novel process that modifies the roles of tumor suppressors and promoters, such as BRCA1 and CD44, and might provide new targets for therapeutic intervention.
Introduction
Structural predictions of protein function are based upon rules of precedence and are not always useful for identifying novel protein functions. As an example, protein export is considered to be dependent upon access to the classical secretory pathway of the Golgi/ER (Hathout, 2007; Simpson et al., 2007). A large body of evidence suggests that secretion through this route, which requires an N-terminal signal peptide, accounts for most of the constitutive export of proteins (Hathout, 2007; Simpson et al., 2007). However, proteomic analyses of cultured plant and mammalian cell surfaces unexpectedly demonstrate the presence of a heterogeneous group of cell-surface proteins that lack signal peptides and are traditionally predicted to be cytoplasmic proteins (Chivasa et al., 2006; Chivasa et al., 2005; Tjalsma et al., 2006). Emerging evidence indicates that these exported `cytoplasmic' proteins neither transit through the Golgi/ER nor are released as a result of cell death, but are exported in response to specific stimuli (e.g. non-constitutively) by unconventional mechanisms that do not require known structural secretion cues. These unconventionally exported proteins perform unexpected extracellular functions that often have little to do with their intracellular duties (Nickel, 2005; Prudovsky et al., 2007; Radisky et al., 2003). The cell-surface display of these `cytoplasmic' proteins appears to be important for the adaptation of cells to stress, and tumor cells commonly use this adaptive mechanism.
The hyaluronan-mediated motility receptor (RHAMM) is an example of an oncogenic `cytoplasmic' protein with unexpected extracellular functions that are not defined by protein-structure `rules'. Although RHAMM resembles a quintessential cytoplasmic protein, its oncogenic effects have been attributed to both intracellular and extracellular functions. As an extracellular hyaluronan (HA)-binding protein with CD44-activating functions, it controls signaling through RAS proteins, which are molecular switches that commonly acquire gain-of-function mutations in human cancer. CD44 is a cell-adhesion and HA receptor that can promote invasion and metastasis in experimental tumor models and is a surface marker for aggressive tumor-progenitor-cell subsets in leukemia and in breast, prostate and other human cancers. As a mitotic-spindle or centrosomal protein, RHAMM acts on the breast cancer 1 early onset (BRCA1)–BRCA1-associated RING domain 1 (BARD1) pathway that controls mitotic-spindle integrity.
Germ-line loss-of-function mutations in BRCA1 result in genetic susceptibility to breast, ovarian and prostate cancers. Given that there is evidence that both genetic mutations and extracellular cues are driving forces for the genomic instability that fuels tumor progression, it is timely to consider how these apparently disparate `inside-outside' mechanisms are integrated. Inside-out signaling by integrins, a family of cell-adhesion receptors, is undoubtedly one such mechanism. However, the unconventional export of cytoplasmic proteins to perform `moonlighting' functions that are distinct from their normal cytoplasmic duties is another less-well-noticed mechanism. Here, we consider the cytoplasmic and extracellular functions of RHAMM, a tumor antigen encoded by a novel breast-cancer susceptibility gene, as a timely example of how intracellular and/or extracellular compartmentalization of a single protein might permit the simultaneous sensing of microenvironment cues and modification of genomic stability.
Non-classical or unconventional export
The primary structure of RHAMM lacks sequences that predict either classical secretion or integration into the plasma membrane. This is also typical for the primary structures of other unconventionally exported cytoplasmic proteins (Nickel, 2005; Prudovsky et al., 2007; Radisky et al., 2003), including growth factors [e.g. fibroblast growth factor 1 and 2 (FGF1 and FGF2)], cytoplasmic galectins [e.g. GAL1 (LGALS1) and GAL3 (LGALS3)], trafficking proteins [e.g. epimorphin (syntaxin 2)], metabolic enzymes (e.g. phosphoglucose isomerase, also co-discovered as neuroleukin and autocrine motility factor), heat shock proteins (e.g. HSP70) (reviewed in Radisky et al., 2003; Takenaka et al., 2004) as well as transcription factors and/or DNA-repair proteins [e.g. Ku (Paupert et al., 2007)]. Current knowledge of the molecular mechanisms for unconventional export is summarized in Fig. 1. Cytoplasmic proteins are exported in response to specific stimuli, and many (e.g. FGF1, Ku, RHAMM, epimorphin and GAL3) stimulate cell adhesion and/or motility upon export, facilitating specific steps in progression and metastasis – such as localized invasion, arrest and extravasation (Nickel, 2005; Prudovsky et al., 2007; Radisky et al., 2003).
Routes for non-conventional protein export. (A) Transporter channels. Cytoplasmic proteins can be exported through transporter-protein channels and then captured by cell-surface counter receptors. As an example, FGF2 is shown being exported through the ABC transporter and then captured by a heparin sulphate proteoglycan. (B) Protein-release complex. Other cytoplasmic proteins, such as epimorphin (or FGF1, not shown), associate with a protein-release complex that is composed of S100A13 and synaptotagmin 1 that then binds to annexin 2. The epimorphin protein-release complex is localized to, and flipped through, the cytoplasmic membrane in response to Ca2+ and phosphatidylserine. Released epimorphin then binds to αv-integrin, resulting in activation of morphogenic signaling cascades. (C) Flippase activity. An alternate route of export results from cytoplasmic proteins binding to transporter proteins that have intrinsic flippase activity when stimulated by phosphatidylserine. (D,E) Exocytosis and membrane blebbing. Additional mechanisms of non-conventional export include exocytosis (D) and membrane blebbing (E). Cytoplasmic and/or nuclear proteins such as Ku are released by exocytosis of exosomes, but cytoplasmic proteins can also be exported in vesicles formed by membrane blebbing.
Routes for non-conventional protein export. (A) Transporter channels. Cytoplasmic proteins can be exported through transporter-protein channels and then captured by cell-surface counter receptors. As an example, FGF2 is shown being exported through the ABC transporter and then captured by a heparin sulphate proteoglycan. (B) Protein-release complex. Other cytoplasmic proteins, such as epimorphin (or FGF1, not shown), associate with a protein-release complex that is composed of S100A13 and synaptotagmin 1 that then binds to annexin 2. The epimorphin protein-release complex is localized to, and flipped through, the cytoplasmic membrane in response to Ca2+ and phosphatidylserine. Released epimorphin then binds to αv-integrin, resulting in activation of morphogenic signaling cascades. (C) Flippase activity. An alternate route of export results from cytoplasmic proteins binding to transporter proteins that have intrinsic flippase activity when stimulated by phosphatidylserine. (D,E) Exocytosis and membrane blebbing. Additional mechanisms of non-conventional export include exocytosis (D) and membrane blebbing (E). Cytoplasmic and/or nuclear proteins such as Ku are released by exocytosis of exosomes, but cytoplasmic proteins can also be exported in vesicles formed by membrane blebbing.
Numerous unconventionally exported proteins exert their extracellular functions by pairing with integral cell-surface `counter' receptors. For example, in addition to documented RHAMM-CD44 (Hamilton et al., 2007; Tolg et al., 2006) and FGF-CD44 pairings, GAL1 pairs with surface molecules that contain β-galactoside, whereas GAL3 interacts with the NG2 cell-surface proteoglycan α3β1-integrin (Nangia-Makker et al., 2007), epithelial growth-factor receptor (EGFR) and transforming growth factor-β (TGFβ) receptors (Partridge et al., 2004). Additional examples include the partnering of epimorphin with αv-integrins (Hirai et al., 2007) and associations between Ku and matrix metalloproteinases [e.g. MMP9 (Paupert et al., 2007)] (Table 1). Extracellular expression of cytoplasmic proteins, such as RHAMM and Ku, results from a redistribution of intracellular pools to the extracellular compartment that is not necessarily associated with increased synthesis or stability of mRNA or protein. Thus, the redistribution of RHAMM or Ku, and their respective regulation of CD44 and MMP9 activity, cannot be predicted by the analysis of gene-expression signatures.
Known intracellular and extracellular functions of some non-conventionally exported proteins
Gene symbol(s) . | Commonly used names . | Intracellular functions . | Extracellular functions . | Cell-surface `counter' receptors . |
---|---|---|---|---|
EPIM, STX2 | Epimorphin, syntaxin | Vesicle fusion | Morphogen signaling | αv integrins |
RHAMM, HMMR | RHAMM, CD168 | Mitotic-spindle integrity | Motogenic signaling | CD44 |
GAL1, LGALS1 | Galectin 1 | Multiple signaling functions | Proliferation, migration | CA125, β-galactoside glyco-lipids |
GAL3, LGALS3 | Galectin 3 | E.g. suppress anoikis | Induction of T-cell apoptosis; cell adhesion/migration | N-glycans on, e.g. TGFβ/EGF receptors and NG2/α3β1 complexes |
FGF1 | bFGF-1 | Signaling functions | Proliferation/angiogenesis | Heparan sulfate proteoglycans |
FGF2 | bFGF-2 | |||
XRCC5 | Ku-80 | Transcription factor | Chemoattractant | MMP9 |
Gene symbol(s) . | Commonly used names . | Intracellular functions . | Extracellular functions . | Cell-surface `counter' receptors . |
---|---|---|---|---|
EPIM, STX2 | Epimorphin, syntaxin | Vesicle fusion | Morphogen signaling | αv integrins |
RHAMM, HMMR | RHAMM, CD168 | Mitotic-spindle integrity | Motogenic signaling | CD44 |
GAL1, LGALS1 | Galectin 1 | Multiple signaling functions | Proliferation, migration | CA125, β-galactoside glyco-lipids |
GAL3, LGALS3 | Galectin 3 | E.g. suppress anoikis | Induction of T-cell apoptosis; cell adhesion/migration | N-glycans on, e.g. TGFβ/EGF receptors and NG2/α3β1 complexes |
FGF1 | bFGF-1 | Signaling functions | Proliferation/angiogenesis | Heparan sulfate proteoglycans |
FGF2 | bFGF-2 | |||
XRCC5 | Ku-80 | Transcription factor | Chemoattractant | MMP9 |
The non-conventional export of cytoplasmic proteins can increase with cellular transformation. For example, the export of both RHAMM and Ku is specifically associated with cellular transformation in multiple myeloma (Adamia et al., 2005; Muller et al., 2005). Ku and RHAMM are detected on the surface of stimulated multiple-myeloma cells ex vivo, but neither protein is detected on the surface of resting or CD40 or HA-stimulated primary B cells. The association of cell-surface RHAMM with CD44 is also restricted to highly invasive/migratory breast-cancer cell lines (Hamilton et al., 2007), such as MDA-MB-231 cells, which exhibit surface markers that are associated with tumor-initiating progenitor cells (e.g. CD44+CD24–) (Neve et al., 2006; Sheridan et al., 2006). Moonlighting functions of proteins are not necessarily associated with unconventional export. For example, polo-like kinase 1 (PLK1), a member of the conserved polo-like kinase family that has classically been characterized for its intracellular role in cell division and/or genomic stability, has recently been linked to breast-cancer-cell invasion in vitro and in vivo by promoting phosphorylation of vimentin, which associates with integrins to modify cell invasion (Rizki et al., 2007). Such studies highlight the importance of unexpected, multiple functions that many proteins acquire during tumor progression. Some unexpected functions of `cytoplasmic' proteins are only realized, however, if the protein is unconventionally exported.
In addition to contributing to tumorigenesis, the unconventional export of proteins such as RHAMM also contributes to inflammatory disorders (Nedvetzki et al., 2004) and to repair of excisional wounds (Tolg et al., 2006). Cell migration during these processes requires the moonlighting extracellular functions of RHAMM. However, unbiased gene-expression screens do not detect this location and function change. By contrast, RHAMM is often identified in genetic screens as an intracellular mitotic-spindle protein. Biological characterization was necessary to identify RHAMM as an extracellular protein that signals via an association with HA-CD44 complexes to promote cell motility and invasion (Toole, 2004; Turley et al., 2002). The process of unconventional export greatly increases the functional complexity of cytoplasmic proteins and provides an example of a process that is not predicted by currently understood `rules' of protein structure. Unconventional export appears to be used especially during neoplastic progression. RHAMM is an excellent example of a protein for which unconventional export contributes to oncogenesis.
RHAMM expression and oncogenesis
RHAMM was originally identified as a soluble protein, released by sub-confluent migrating cells (Hardwick et al., 1992; Turley, 1982; Turley et al., 1991), that promoted cell motility and invasion via interactions with HA and the cell surface (Tolg et al., 2006; Turley et al., 2002). In a seminal paper, Hall et al. (Hall et al., 1995) showed that RHAMM overexpression is transforming and is required for maintaining RAS transformation. RHAMM mRNA and protein are poorly expressed in most normal human tissues (Evanko et al., 2007; Slevin et al., 2007; Toole, 2004; Turley et al., 2002) and RHAMM expression is not essential to either mouse embryonic development or normal adult-mouse homeostasis. RHAMM expression is increased during wound repair in response to hypoxia and fibrogenic factors such as TGFβ1 (Samuel et al., 1993), and, to date, its only known physiological function is to promote wound repair. Thus, genetic deletion of RHAMM (HMMR) results in slow healing of skin wounds, owing to an essential role in fibrogenesis and/or mesenchymal differentiation (Tolg et al., 2006).
RHAMM expression is increased during the neoplastic progression of a variety of human tumors (Giannopoulos and Schmitt, 2006; Toole, 2004; Turley et al., 2002), which is consistent with its ability to transform cells in experimental models. In particular, high mRNA levels (Pujana et al., 2007; Wang et al., 1998) and protein hyperexpression in breast-tumor-cell subsets is associated with poor outcome and increased peripheral metastasis (Wang et al., 1998). RHAMM is a novel breast-cancer susceptibility gene, and a significant association between homozygous variation in this gene and early-onset breast cancer has been reported (Pujana et al., 2007). RHAMM is also a tumor-associated antigen found in solid and blood tumors (Greiner et al., 2002). Extracellular RHAMM that is produced by, and displayed on, tumor cells is the basis for the successful clinical use of RHAMM peptide vaccines in reducing the disease activity of acute myeloid leukemia (AML) and multiple myeloma; this vaccine is currently in phase 1 clinical trials (Schmitt et al., 2007).
Experimentally, mutation of key basic residues in the C terminus of RHAMM, causing disrupted HA binding (Yang et al., 1994), also abolished the transforming ability of RHAMM and inhibited RAS transformation (Hall et al., 1995). These results, and others, provide evidence that at least some of the oncogenic effects of RHAMM result from its extracellular HA-binding properties (Hall et al., 1995). The occurrence of cell-surface RHAMM (designated CD168), although once considered to be controversial because of its lack of a signal peptide, has been extensively documented by many independent laboratories over the past 10 years as a cell-surface HA receptor for a variety of cell types in culture (Adamia et al., 2005; Evanko et al., 2007; Giannopoulos and Schmitt, 2006; Slevin et al., 2007; Toole 2004; Turley et al., 2002). The HA-binding extracellular functions of RHAMM contribute to normal wound repair in culture and in vivo (Slevin et al., 2007; Tolg et al., 2006), and to the progression of diseases such as arthritis in animal models (Naor et al., 2007). Mechanistically, extracellular RHAMM `activates' pro-migration and invasion functions of the transmembrane-adhesion and HA receptor CD44. This RHAMM-regulated activation process results in increased cell-surface expression of CD44 and an increase in the activation of ERK1 and ERK2 (ERK1,2) by CD44 (Hamilton et al., 2007; Tolg et al., 2006) (Fig. 2A).
Extracellular and intracellular functions of RHAMM. (A) Cell-surface RHAMM promotes the activation of signaling cascades. Shown is one molecular mechanism for this. Cell-surface RHAMM, which is not an integral membrane protein, partners with CD44 and, in the presence of hyaluronan, activates ERK1/2 [indicated as phosphorylated (PO4) ERK1,2], which results in the expression of genes that are required for motility and invasion. (B) In X. laevis egg extracts, a RAN-GTP gradient, which is established by chromosome-bound guanine nucleotide-exchange factor RCC1 activity, is required for anastral mitotic-spindle assembly. RAN-GTP activity regulates the function of a number of mitotic-spindle proteins, including importins, that then form inhibitory complexes with both spindle-assembly factors and TPX2. For example, by binding importins (indicated as step 1), RAN-GTP releases TPX2 (step 2), which is a major activator of Aurora kinase A (AURKA). TPX2 directly activates AURKA by protecting an autophosphorylated residue (step 3). AURKA, in turn, can phosphorylate (PO4) BRCA1 to facilitate G2-M transition (step 4). Via an interaction with the dynein complex, RHAMM localizes to the spindle pole, at which it interacts with γ-tubulin (step 5). RHAMM also interacts with TPX2 and dynein, thereby having the potential to localize TPX2 to spindle poles (step 5). The BRCA1-BARD1 complex modifies TPX2 localization and spindle assembly by attenuating RHAMM function through ubiquitylation (Ub) (step 6). Ubiquitylation of RHAMM, and subsequently its degradation, probably releases TPX2 from the spindle pole (step 7), thus affecting AURKA activation and G2-M progression.
Extracellular and intracellular functions of RHAMM. (A) Cell-surface RHAMM promotes the activation of signaling cascades. Shown is one molecular mechanism for this. Cell-surface RHAMM, which is not an integral membrane protein, partners with CD44 and, in the presence of hyaluronan, activates ERK1/2 [indicated as phosphorylated (PO4) ERK1,2], which results in the expression of genes that are required for motility and invasion. (B) In X. laevis egg extracts, a RAN-GTP gradient, which is established by chromosome-bound guanine nucleotide-exchange factor RCC1 activity, is required for anastral mitotic-spindle assembly. RAN-GTP activity regulates the function of a number of mitotic-spindle proteins, including importins, that then form inhibitory complexes with both spindle-assembly factors and TPX2. For example, by binding importins (indicated as step 1), RAN-GTP releases TPX2 (step 2), which is a major activator of Aurora kinase A (AURKA). TPX2 directly activates AURKA by protecting an autophosphorylated residue (step 3). AURKA, in turn, can phosphorylate (PO4) BRCA1 to facilitate G2-M transition (step 4). Via an interaction with the dynein complex, RHAMM localizes to the spindle pole, at which it interacts with γ-tubulin (step 5). RHAMM also interacts with TPX2 and dynein, thereby having the potential to localize TPX2 to spindle poles (step 5). The BRCA1-BARD1 complex modifies TPX2 localization and spindle assembly by attenuating RHAMM function through ubiquitylation (Ub) (step 6). Ubiquitylation of RHAMM, and subsequently its degradation, probably releases TPX2 from the spindle pole (step 7), thus affecting AURKA activation and G2-M progression.
In the context of breast tumorigenesis, activation of extracellular HA binding by CD44-RHAMM complexes confers malignant potential. For example, HA accumulates in tumors of high-risk breast-cancer patients (Auvinen et al., 2000; Wernicke et al., 2003) and is a major polysaccharide ligand for CD44 (Toole, 2004), which itself is a defined marker of cellular subtypes with increased tumorigenic potential [i.e. CD44+CD24low breast-tumor-initiating stem-like cells (Al-Hajj et al., 2003; Liu et al., 2007; Shipitsin et al., 2007)]. The strong link between RHAMM-CD44 and activation of the RAS/ERK1,2 kinase pathways implies that these HA-receptor complexes control signaling pathways that are hyper-activated in many cancers, including that of the breast, and permit risk stratification based on the de-regulated `oncogenic-pathways' gene signatures seen in patients with a variety of cancers (Bild et al., 2006).
Intracellular RHAMM decorates interphase microtubules, centrosomes and the apex of mitotic spindles, and is also located in the cell nucleus (Assmann et al., 1999; Entwistle et al., 1996; Joukov et al., 2006; Maxwell et al., 2003). Because intracellular RHAMM affects mitotic-spindle integrity (Groen et al., 2004; Joukov et al., 2006; Maxwell et al., 2003) and high levels of RHAMM correlate with genomic instability in multiple myeloma (Maxwell et al., 2004), RHAMM–centrosome–mitotic-spindle associations have the potential to affect cell transformation and tumor progression by promoting genomic instability (Fig. 2B). RHAMM interacts with BRCA1 (Joukov et al., 2006; Pujana et al., 2007); this is consistent with a proposed role of RHAMM in driving centrosome amplification and mitotic-spindle aberrations in breast-cancer progression.
In summary, RHAMM might potentiate oncogenesis by controlling both CD44-related signaling functions via the tumor microenvironment (Fig. 2A) and BRCA1-related mechanisms for mitotic spindle/centrosome integrity (Fig. 2B). But how does RHAMM regulate these disparate functions and when are they relevant in the tumorigenic process?
Extracellular RHAMM: an activator of latent oncogenic properties of CD44?
CD44 is an integral membrane protein that is subject to extensive alternative splicing (reviewed in Naor et al., 2007). The mechanism by which CD44 binds to HA has been well described (Banerji et al., 2007) and is due to a `link module' that is common to most HA-binding proteins (Day and Prestwich, 2002; Milner et al., 2006). The link module bears no structural resemblance to the highly conserved basic HA-binding region of RHAMM (Day and Prestwich, 2002; Yang et al., 1994). CD44-HA interactions affect cell adhesion to extracellular matrix components and are implicated in aggregation, proliferation, migration and invasion of tumor cells in culture and in some (Evanko et al., 2007; Toole, 2004), but not all (Lopez et al., 2005), mouse models of tumor susceptibility. Although CD44 has oncogenic potential (Gotte and Yip, 2006; Toole, 2004), its lack of a clear association with clinical outcome has resulted in controversy over its role in tumor progression (Agnantis et al., 2004; Kuhn et al., 2007; Vargo-Gogola and Rosen, 2007; Wielenga et al., 2000). For example, CD44 expression declines with prostate-tumor progression, raising the possibility that it can act under certain circumstances as a tumor suppressor (Epstein et al., 2005; Kauffman et al., 2003).
Two recent articles (Liu et al., 2007; Shipitsin et al., 2007) have again highlighted CD44 as a significant factor in tumor progression and clinical outcome. These studies identify high CD44 expression on a subset of breast-tumor cells, the gene signature for which predicts poor clinical outcome. Shipitsin et al. (Shipitsin et al., 2007) used a technique called SAGE (serial analysis of gene expression) to provide detailed molecular characterization of CD44+ breast-tumor stem-cell-like subsets, originally identified by Al-Hajj et al. (Al-Hajj et al., 2003), and correlated CD44+ cell-specific genes with increased expression of stem-cell markers and decreased patient survival. Among the genes that are elevated in metastatic CD44+ breast-tumor cells is hyaluronan synthase 1 (HAS1), which provides a potential mechanism to explain the increase in HA that is associated with poor outcome in breast cancer patients (Shipitsin et al., 2007). Complimentary findings demonstrate that the gene signature from CD44+CD24– cells isolated from pleural effusions from breast-cancer patients predicted poor outcome in breast, prostate and other cancers (Liu et al., 2007). Although these results support experimental evidence that CD44 functions as a pro-metastatic gene, they are still discordant with equally convincing evidence that CD44 expression suppresses metastases in animal models of breast cancer (Lopez et al., 2005), can be absent in breast metastatic lesions (Shipitsin et al., 2007) and is lost during the progression of other cancers, such as prostate cancer (Tang et al., 2007).
These contradictions might be reconciled if CD44 `oncogenicity' is determined by the regulated expression or availability of partner proteins in combination with specific variant expression. CD44 function is regulated by differential expression of functionally diverse CD44 isoforms (Naor et al., 2007) as well as via the actions of partner proteins. Examples of CD44 partner proteins include HAS1, MMP9 and MMP14 (Bourguignon et al., 2007; Toole, 2004; Wang et al., 2007), each of which is elevated in the highly tumorigenic CD44+ subsets (Liu et al., 2007; Shipitsin et al., 2007). Independent studies provide insight into a novel, tumor-specific mechanism that CD44+CD24– progenitor breast-tumor cell lines [e.g. MDA-MB-231 (Neve et al., 2006; Sheridan et al., 2006)] use to `activate' the tumor-promoting potential of CD44 (Hamilton et al., 2007; Tolg et al., 2006). Mechanistically, extracellular RHAMM can enhance the localization of CD44 to the surface of invasive/metastatic breast-tumor cells. Prolonged surface localization of CD44 enhances the subsequent activation of CD44-dependent signaling, including the formation of CD44-ERK1/2 complexes and the targeting of these complexes to the nucleus (Tolg et al., 2006), which is a requirement for ERK1/2-mediated cell transformation (Liu et al., 2007; Shipitsin et al., 2007). In aggressive progenitor breast-cancer cells in vitro, extracellular RHAMM-CD44 complexes coordinate HA-dependent ERK1/2 activation and motility/invasion, providing plausible support for a CD44-activating function of cell-surface RHAMM (Hamilton et al., 2007). Unlike HAS1, MMP9 and MMP14, RHAMM mRNA is not elevated within progenitor subsets (Liu et al., 2007; Shipitsin et al., 2007). Rather, an alteration in localization (and not necessarily abundance) of RHAMM protein regulates CD44 activation, and subsequent signaling pathways and cellular migration. Proteomic analyses or unbiased gene-expression screens would not predict this extracellular role for RHAMM in tumor progression.
These and other studies establish a clear link between transformation, tumor progression and the extracellular functions (HA binding, signaling and compensation) of cell-surface RHAMM. These functions are shared by CD44, and the evidence that these two proteins are functionally and physically linked in aggressive breast-cancer cell lines provides plausible support for a CD44-activating function of cell-surface RHAMM. RHAMM surface expression might even partially compensate for loss of CD44, which can occur, for example, in advanced prostate cancer, and maintain the activation of HA-driven signaling pathways in the absence of CD44 by partnering with other integral HA receptors. These extracellular functions of RHAMM are probably instrumental to the oncogenic effects of this unconventional protein but are not the complete story. The mitotic-spindle and centrosomal-binding properties of RHAMM also have the potential to influence tumor progression.
Intracellular RHAMM: an antagonist of BRCA1 and promoter of oncogenic genomic instability?
Network modeling of co-expression profiles in functional genomic and proteomic data sets recently identified RHAMM as a novel breast-cancer susceptibility gene (Pujana et al., 2007). Variations in the RHAMM locus were linked to an elevated risk for breast cancer, which was most pronounced for carriers under the age of 40 (Pujana et al., 2007). Mechanistically, RHAMM was shown to interact with BRCA1, at physical and genetic levels, in the regulation of centrosome amplification (Pujana et al., 2007). In combination with a recent report describing RHAMM-BRCA1-BARD1 interactions in frog extracts (Joukov et al., 2006), these studies offer significant additional insights into how the intracellular functions of RHAMM might impact tumor progression.
Intracellular forms of RHAMM have been shown to localize to interphase microtubules, the nucleus, mitotic spindles and the centrosome (Hofmann et al., 1998; Maxwell et al., 2005; Maxwell et al., 2003). RHAMM expression is cell-cycle regulated and peaks at G2-M. RHAMM was one of only 96 genes identified as being cell-cycle regulated in both HeLa cells and synchronized primary human fibroblasts (Cho et al., 2001; Whitfield et al., 2002). In Xenopus laevis meiotic extracts, RHAMM is a component of the microtubule-associated proteome (Liska et al., 2004). Within HeLa S3 cells, phosphoproteome analyses identified two RHAMM peptides among the 260 proteins identified that are phosphorylated during mitosis (Nousiainen et al., 2006). Other reports demonstrate that RHAMM participates in mitotic-spindle assembly, centrosome integrity and passage through the G2-M boundary of the cell cycle (Groen et al., 2004; Maxwell et al., 2003). Therefore, transcriptome, proteome and phosphoproteome analyses argue that the regulation of RHAMM localization to microtubules can be important during mitosis. Experimental analyses suggest that intracellular RHAMM can participate in both centrosomal (astral) and chromatin (anastral) mitotic-spindle formation.
Primary-structure homology between the conserved C-terminal domain of RHAMM (which overlaps with the HA-binding domain) and X. laevis kinesin-like protein 2 (XKLP2), a plus-end-associated kinesin-like protein, is consistent with the centrosomal localization of RHAMM and its interaction with TPX2 (targeting protein for XKLP2), a spindle-assembly protein (Groen et al., 2004; Maxwell et al., 2005; Maxwell et al., 2003). The X. laevis ortholog of RHAMM, XRHAMM, functions in RAN-dependent mitotic-spindle assembly in X. laevis egg extracts, which are a model of centrosomal-independent spindle assembly (Groen et al., 2004). Depletion of XRHAMM from X. laevis extracts impairs TPX2 localization and anastral spindle assembly in a manner that is overcome by active RAN (Groen et al., 2004). Alteration of leucine residues within the C-terminal domain disrupts aster structure and TPX2 localization in X. laevis extracts (Groen et al., 2004; Maxwell et al., 2003). The interaction of RHAMM with TPX2 is also observed in human cells (Maxwell et al., 2005) and truncation of the C-terminal domain of RHAMM impairs centrosome localization, which is consistent with an effect of RHAMM on astral spindle formation (Maxwell et al., 2003). In addition to TPX2 and γ-tubulin (a centrosomally restricted isoform of tubulin), XRHAMM also associates with the BRCA1-BARD1 heterodimer (Joukov et al., 2006; Pujana et al., 2007).
The association between the loss of BRCA1 function and elevated risk for breast and/or ovarian cancer is often attributed to the role for BRCA1 in DNA double-strand-break repair and cell-cycle arrest (reviewed in Venkitaraman, 2002). Additionally, BRCA1 localizes to centrosomes during mitosis (Hsu et al., 2001), and impaired BRCA1 function results in centrosome amplification in vivo (Xu et al., 1999) and in vitro (Starita et al., 2004). BRCA1 heterodimerizes with BARD1 to form a tumor-suppressor complex with ubiquitin E3 ligase activity (Baer and Ludwig, 2002). Within cell lines from mammary tissue in vitro, BRCA1-BARD1 regulates centrosome amplification via ubiquitylation of γ-tubulin (Starita et al., 2004). Recent evidence demonstrates that BRCA1-BARD1 also regulates anastral spindle assembly by attenuating excessive XRHAMM mitotic activity (Joukov et al., 2006). Within human cells, RHAMM and BRCA1-BARD1 physically interact, and RHAMM protein levels are regulated by BRCA1-BARD1 as a result of its ubiquitylation by BRCA1-BARD1 (Joukov et al., 2006; Pujana et al., 2007). Silencing of RHAMM or BRCA1 can induce centrosome amplification, which is not observed when both gene products are silenced (Pujana et al., 2007). Although appropriate expression levels of RHAMM appear to be required for mitotic-spindle formation and centrosome structure, either overexpression of the C-terminal RHAMM sequence or inhibition of RHAMM-spindle interactions (achieved by mutation of the conserved C-terminal leucine zipper) results in aberrant mitotic-spindle assembly, particularly when BRCA1-BARD1 is absent (Joukov et al., 2006; Maxwell et al., 2005).
The implications of these studies are that the loss of BRCA1-BARD1 activity, which might occur in inherited forms of breast and ovarian cancers, combined with elevated RHAMM expression, which is a prognostic factor for poor outcome (Pujana et al., 2007; Wang et al., 1998), might contribute to chromosomal instability during progression. Indeed, high levels of RHAMM, elevated centrosome volumes and cytogenetic aberrations all correlate with poor prognosis in some cancers, such as multiple myeloma (Maxwell et al., 2005). Finally, elevated expression of RHAMM in primary sporadic breast tumors is also associated with an early age of diagnosis (Pujana et al., 2007).
Conclusions and perspectives
Collectively, these studies illuminate RHAMM as an oncogene and indicate that its product can enhance tumor invasion and progression by multiple functional mechanisms that manifest in distinct subcellular locations. These functions include intracellular regulation of mitotic-spindle assembly, and extracellular activation of CD44 motogenic/invasion functions and of RAS-regulated signaling pathways. Importantly, regulated secretion of RHAMM occurs during wound repair, is used by tumor cells during disease progression and appears to involve the redistribution of existing cellular pools rather than new, or increased, transcription. It may be that the increased proliferation associated with cancer cells facilitates the secretion of intracellular proteins, such as RHAMM and Ku, and provides for outside-in and inside-out control of genetic stability; the identification of genes with dual functions in regulating the sensing of the extracellular matrix and in controlling genetic instability [e.g. PLK1, XRCC3, CENPA and Ku80 (Rizki et al., 2007)] supports this hypothesis. An understanding of the multiplicities of function for many proteins will not only highlight novel mechanisms of tumorigenesis, but might also provide novel therapeutic targets.
The validity of scientific `truths' is often questioned when they are contradicted by data. A decade ago, the lack of signal peptide within the primary structure of RHAMM and epimorphin called into question their extracellular functions. However, it has now become clear that they are not the only oncogenic `cytoplasmic' proteins with moonlighting extracellular functions. For example, similar dual inside-out function has been described for Ku (Gullo et al., 2006; Muller et al., 2005); additionally, FGF1 and FGF2 (Nickel, 2005; Prudovsky et al., 2007), and GAL1 and GAL3 (Califice et al., 2004; Takenaka et al., 2004), are examples of leader-less proteins that are unconventionally exported and that therefore perform distinct intracellular and extracellular functions in cancer.
Although not yet well-characterized, the molecular mechanisms that propel unconventional export are being defined and have been particularly well-described for FGF2 (Rajalingam et al., 2005; Zehe et al., 2006) and epimorphin (Hirai et al., 2007). Non-classical secretion of `cytoplasmic' proteins is likely to be a widespread phenomenon that is common to the tissue-repair and morphogenetic processes that tumor cells use to modify their microenvironment. Clarification of the mechanisms that are used for unconventional export will undoubtedly expand our perspective of protein compartmentalization and function, and, ultimately, of the relationship between tumor microenvironment and tumor initiation and progression.