Hyaluronan (HA) is a structurally simple polysaccharide, but its ability to act as a template for organizing pericellular matrices and its regulated synthesis and degradation are key to initiating repair responses. Importantly, these HA functions are usurped by tumor cells to facilitate progression and metastasis. Recent advances have identified the functional complexities associated with the synthesis and degradation of HA-rich matrices. Three enzymes synthesize large HA polymers while multiple hyaluronidases or tissue free radicals degrade these into smaller bioactive fragments. A family of extracellular and cell-associated HA-binding proteins/receptors translates the bioinformation encrypted in this complex polymer mixture to activate signaling networks required for cell survival, proliferation, and migration in an actively remodeling microenvironment. Changes in HA metabolism within both the peritumor stroma and parenchyma are linked to tumor initiation, progression, and poor clinical outcome. We review evidence that metastatic tumor cells must acquire the capability to autonomously synthesize, assemble, and process their own “portable” HA-rich microenvironments to survive in the circulation, metastasize to ectopic sites, and escape therapeutic intervention. Strategies to disrupt the HA machinery of primary tumor and circulating tumor cells may enhance the effectiveness of current conventional and targeted therapies. Cancer Res; 76(9); 2507–12. ©2016 AACR.
Tumor initiation and progression result not only from mutant signaling networks but also from key host microenvironmental factors. A host-derived stroma develops concomitantly with primary tumor initiation and provides a nurturing or “cancerized” microenvironment that supports tumor cell survival, growth, and invasion (1–4). Aggressive tumor cell subsets become increasingly autonomous and “stromal independent” by establishing autocrine networks permitting them to produce essential microenvironmental factors in an absence of host primary peritumor stroma. This autonomy facilitates the survival of immigrant metastatic cells during their colonization of foreign microenvironments. The tissue polysaccharide, hyaluronan (HA), is one such microenvironmental factor. Its autocrine production, cell surface retention, and processing/degradation by aggressive carcinoma cells are major and targetable factors in metastasis. Evidence from multiple model systems and clinical analyses support a key role for cancerized stromal HA in the initiation of malignant tumors. However, metastatic tumor cells appear to acquire the ability to synthesize and assemble their own pericellular HA-rich matrices that function to maintain the activation of oncogenic driver and survival signaling pathways. These matrices also support adhesion and extravasation at ectopic sites of metastasis, and they function to condition the microenvironment surrounding metastatic lesions. Here we propose that these carcinoma cell–associated pericellular HA-rich matrices represent a “portable cancerized microenvironment” in circulating tumor cells that can be effectively targeted to limit metastasis and improve patient outcome.
HA Pericellular Matrices, Fragments, and Metastasis
Native HA consists of repeating β (1–3) linked disaccharides of d-glucuronic acid and N-acetyl-d-glucosamine, which are knitted together as large (107 daltons or higher) polymers by the formation of hexosaminidic β (1–4) linkages. HA synthesis results from the action of three plasma membrane associated HA synthases (HAS 1, 2, and 3), which are homologous but encoded by distinct genes and are differentially expressed (1, 5). Growing HA polymers are thought to be extruded through pores in the plasma membrane (Fig. 1) created by HAS oligomerization (1, 5). HAS expression and synthesis of HA are regulated by cytokines and growth factors such as PDGF-BB, TGFβ1, and bFGF (5). Extruded high molecular weight HA polymers are initially captured by cellular HA receptors and are assembled and stabilized into pericellular coats by a family of extracellular HA-binding proteins including tenascin, inter-α-trypsin inhibitor, versican, and TSG-6 (6–8). These HA-rich pericellular matrices are easily visualized in culture as pericellular coats (9) and are also clearly present in vivo (10, 11). In primary tumors, the cancerized stroma is a rich source of growth factors and cytokines that both stimulate HA synthesis and enhance carcinoma survival and progression. Autonomous metastatic tumor cells have developed an autocrine ability to produce these factors, which also promote synthesis and assembly of HA-rich pericellular matrices coating these aggressive tumor cells. These HA pericellular matrices facilitate several steps required by circulating tumor cells in the metastatic cascade. Compelling data also suggest that the autocrine production of HA pericellular matrices by themselves is not sufficient for successful metastatic colonization. Rather, it appears that tumor cells must also acquire an ability to fragment and metabolize HA to form metastatic colonies.
Synthesis of pericellular HA functions to organize and cluster HA receptors on the plasma membrane and sequester growth factors and cytokines near the cell membrane (Fig. 1). Receptor clustering induced by pericellular HA organizes plasma membrane microdomains such as lipid rafts, which together with the sequestered growth factors and cytokines promote coupling and sustained activation of driver oncogenic antiapoptotic, and proinvasion/migration signal transduction networks (12–14). Thus, HA-rich pericellular matrices facilitate intravasation and protect immigrant cells from anoikis during circulatory transit. They also contribute to the ectopic tissue colonization of foreign microenvironments (15). For example, they can promote the adhesion of tumor cells to microvessel endothelium as has been demonstrated for metastatic prostate carcinoma cells in culture (9). Furthermore, breast tumor cell subpopulations that bind high levels of HA adhere to vessels and extravasate ex ovo more rapidly than tumor cells that bind little or no HA (16). Pericellular HA matrices may participate in the conditioning of foreign microenvironments by stimulating the release of microvesicles from HAS and HA-rich cell microvilli (17). These microvesicles contain numerous proteins involved in organizing and remodeling extracellular matrices (e.g., CD44, MMPs, and EMPPRIN) as well as other regulatory factors that are known to produce cancer friendly microenvironments. This model of HA pericellular matrices as portable microenvironments that protect and nurture trafficking and extravasated metastatic tumor cells is summarized in Fig. 2. However, an excellent example for the requirement of both HA synthesis and metabolism in tumorigenesis is provided by the naked mole rat, which is resistant to tumors (18). Fibroblasts isolated from the naked mole rat are resistant to transformation yet actively produce very high molecular weight HA produced via a mutant HAS. These same cells can be rendered sensitive to transformation by oncogenes if fragmentation of HA is stimulated by introduction of hyaluronidases.
HA fragments appear to be required for further conditioning of the metastatic microenvironment acting in part to attract host cells that remodel the microenvironment of metastatic colonies. HA is fragmented by enzymatic attack from one or more of several hyaluronidases (e.g., Hyal 1, 2, or 3). It can also be chemically cleaved by oxygen and nitrogen free radicals that are elevated in tumor microenvironments (19–21). This two-pronged attack on HA polymers results in a complex mixture of oligosaccharides and larger fragments with a broad range in sizes (e.g., 4 mer to 300,000 daltons). HA fragments perform a range of essential functions necessary for successful metastasis and include effects on both tumor cell and host cell functions. Evidence is emerging that specific size ranges of HA fragments exert differential effects on tumor cell migration and growth (4, 22). In addition, HA fragmentation within ectopic sites promotes the influx of host protumorigenic macrophages and activates endothelial cells to stimulate angiogenesis (2). Thus, aggressive immigrant tumor cells that have developed the ability to autonomously synthesize HA assemble it into pericellular matrices and fragment/process the HA polymers that have a selective advantage in metastasis. HA fragmentation is associated with more aggressive disease and studies using prostate cancer cell lines demonstrate that metastasis requires both elevated HA synthesis and Hyal 1–mediated HA fragmentation. Furthermore, targeting hyaluronidase-1 effectively limits prostate carcinoma invasion and metastasis in xenograft models (19, 22, 23). Analysis of the naked mole rat genome has identified candidate genomic adaptations in the HA receptor genes CD44 and RHAMM, which may modify signaling through HA and partially account for the extraordinary longevity of the naked mole rat and its resistance to tumors (24).
CD44 and RHAMM as Mediators of HA-Promoted Metastasis
One family of proteins that has been extensively studied for their role in metastasis are the CD44 isoforms, which are produced by alternative splicing as well as changes in glycosylation of the core protein. In particular, the standard CD44 isoform (CD44s), which binds to HA and lacks alternatively spliced domains, is structurally and functionally well characterized (4, 13, 15, 25). CD44s is a type I transmembrane protein with an HA binding amino terminal link module that is structurally homologous to link modules found in other HA receptors such as LYVE-1 and in extracellular HA-binding proteins such as TSG-6, versican (26, 27). X-ray crystallography and nuclear magnetic resonance of the CD44 link module has guided the development of specific small molecules to disrupt binding of HA within its long shallow binding groove, which accommodates 8 mer regions of the HA polymer (28).
CD44 colocalizes with HAS enzymes in lipid rafts where it is clustered by high molecular weight HA polymers and functions as a coreceptor for growth factor receptors to lower the activation threshold of oncogenic driver signaling networks (5, 12, 13). This organizing influence of HA on cell surface CD44 also promotes the plasma membrane localization of multidrug resistance proteins and monocarboxylate transporters. Membrane-localized multidrug transporters promote the efflux of therapeutic drugs and xenobiotics (29) while monocarboxylate transporters lower intracellular lactate concentrations in tumor cells to favor their survival in hypoxic microenvironments (30). Specific sizes of HA fragments (<20 kDa) can disrupt HA-induced CD44 clusters to block these effects (31–33). For example, these small HA fragments limit oncogenic pathway activation and reverse drug resistance in CD133-positive highly tumorigenic subpopulations of ovarian carcinoma cells (31). Disrupting HA–CD44 interactions may also be effective in promoting anoikis of circulating tumor cells and limiting HA-mediated extravasation and ectopic colonization. Therefore, HA fragments, mimics, and small molecules that target module HA binding to link modules are likely to have enormous potential for restricting tumor cell metastasis and/or restoring their sensitivity to currently used targeted- and chemotherapies.
Unlike CD44, RHAMM (HMMR) is an HA receptor that is poorly expressed in most homeostatic tissues but is generally upregulated as well as cell surface displayed during tissue repair (4). It is consistently overexpressed in many tumors, and its high expression is linked to the progression of multiple carcinomas (e.g., lung, pancreatic, breast, prostate, glioblastoma, and colorectal cancers), and mesenchymal tumors including fibromatoses and sarcomas (34–43). RHAMM is considered to be a tumor marker for leukemia (44) and colorectal subtypes (42). Elevated RHAMM expression is associated with castration-resistant disease in patients harboring prostatic metastases (36) and elevated levels of RHAMM and HA area associated with a likelihood of biochemical failure in at-risk (Gleason 7) prostatectomized patients (45). RHAMM hyperexpression in breast primary tumor cell subpopulations is linked to increased peripheral metastases (46) while increased RHAMM expression in colorectal tumor budding cells at the invasive front of primary tumors is linked to frequent lymphatic invasion, high tumor grade, and nodal metastasis (47). RHAMM is also one of 4 genes elevated in circulating lung adenocarcinoma cells, and patients exhibiting an increase in these markers at 3 months after their first detection had a significantly shorter survival time than patients exhibiting a decrease in these markers (41). These results predict that aggressive RHAMM-positive subpopulations emerge early in the development of primary tumors and contribute to metastasis. Several clinical studies also suggest a role for RHAMM in drug resistance that may determine patient outcome. Thus, the therapeutic sensitivity of peripheral nerve sheath tumors to therapy targeting AURKA depends in part upon RHAMM expression (39), and multidrug resistance of some human leukemia cell lines is mediated by RHAMM (48). These and other studies suggest RHAMM hyperexpression in aggressive tumor cell subsets has promise as a biomarker for assessing patient risk. These studies also indicate that RHAMM is an attractive target both for limiting spread of these cancers and for treating residual disease.
Both cell surface and intracellular RHAMM contribute to tumor cell aggression (4). RHAMM was originally identified as an extracellular HA motility promoting receptor on the surface of RAS-transformed fibroblasts (49). Extracellular RHAMM binds to HA and functionally associates with CD44 and growth factor receptors such as EGFR, PDGFR, and RON to coordinate activation kinetics of kinase signaling networks (4). Subsequent studies demonstrated that RHAMM also has diverse and complex subcellular functions. The binding partners of intracellular RHAMM vary with cell and tissue background but include kinases/activators such as ERK1, 2/MEK1 (4), SRC (3), and AURKA/TPX2 (38) as well as nuclear proteins, microtubule and actin-binding proteins such as spectrin alpha (50) and supervillin (51). The functional association of intracellular RHAMM with AURKA/TPX2 is linked to breast cancer risk in BRCA1, 2 carriers predicting key functional roles for these intracellular RHAMM complexes in tumor susceptibility and aggression (38). In experimental models, intracellular RHAMM and its binding partners regulate tumor cell extravasation, mitotic spindle integrity/chromosome segregation, and cytokinesis. Proliferative responses mediated by cell surface CD44/EGFR can be induced via intracellular RHAMM/ERK-dependent mechanisms, indicating that cell surface CD44 and intracellular RHAMM can also functionally integrate to promote mitosis (52). Intracellular RHAMM may therefore contribute to tumor aggression by affecting genomic stability, and it may contribute to tumor cell intravasation and extravasation by regulating both gene expression (53) and cytoskeleton dynamics (53, 54).
Tumor cells also export RHAMM to the cell surface using not well-characterized secretion mechanisms stimulated by growth factors such as TGFβ1 (55). Cell surface RHAMM binds both fragmented and high molecular weight HA polymers (4), and it often acts as a coreceptor for CD44 to activate kinase cascades (4). It is therefore tempting to speculate that cell surface RHAMM functions as an “early warning system” for promoting cell migration during tissue damage, including that resulting from tumor formation (4). The HA-binding domains of RHAMM contain key positively charged amino acids originally described as BX7B motifs. These are structurally distinct from HA binding CD44 link domain (4, 14). Peptides mimicking the RHAMM BX7B motif and peptides mimicking the size of HA fragments, which specifically block cell surface RHAMM functions (56, 57) along with anti-RHAMM antibodies, inhibit early responses relevant to both wound repair and tumorigenesis. These responses include RHAMM-stimulated mesenchymal cell motility, tumor cell invasion, and angiogenesis (57, 58). Early clinical studies also suggest that targeting cell surface RHAMM is a potential effective cancer therapy. Thus, immune responsiveness to cell surface RHAMM positively correlates with good clinical outcome and immunization with a RHAMM-derived peptide evokes clinical responses in patients with acute myeloid leukemia and multiple myeloma (34, 59). These studies demonstrate that cell surface RHAMM is both biologically functional and immunogenic, and predict that cell surface RHAMM-directed vaccines may be useful for treating RHAMM-expressing malignancies.
Conclusions and Opportunities for Therapeutic Intervention
There is now substantial evidence that elevated HA synthesis, fragmentation, and organization as pericellular coats are important steps in the development of tumor cell aggression and primary tumor stromal autonomy required for establishing distant colonies. We propose that autocrine HA matrices produced by circulating metastatic carcinoma cells are utilized as “portable cancerized microenvironments.” These matrices function to organize carcinoma cell plasma membrane microdomains and receptors, which through interactions with CD44 facilitate oncogenic prosurvival mechanisms, trap and sequester essential nutrients/cytokines/growth factors and promote functioning and cell surface localization of multidrug and monocarboxylate transporters. In the context of HA-rich matrices, CD44 and RHAMM either act independently or as coreceptors to enhance pro-oncogenic signaling activity, promote motility, and modify the transcriptome of these cells. The HA-induced functional integration of CD44 and cell surface and intracellular RHAMM is an area of active research that will lead to improved targeting of circulating metastatic tumor cells. Inhibiting HA synthesis by using 4-methylumbilliferone inhibits metastasis of melanoma and prostate carcinoma cells (60, 61). Furthermore, synthetic peptides, antibodies, or small HA oligomers that that inhibit CD44 and RHAMM function have also proven effective for inhibiting tumor metastasis in multiple model systems. Given recent studies identifying the importance of RHAMM (in contrast to CD44) in mediating HA binding by suspended mesenchymal cells (62), it is worth considering that agents that specifically interfere with RHAMM/HA functions may be particularly effective at disrupting metastasis of circulating tumor cells that overexpress RHAMM (41).These inhibitors have the potential to be adopted as either single or neoadjuvant therapies to improve patient outcome by limiting metastasis recurrence and relapse, which remain major clinical challenges in cancer treatment.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
This work was supported by the Chairman's Fund Professor in Cancer Research (J.B. McCarthy), American Heart Association Scientist Development Grant 13SDG6450000 (D.K. Wood), Prostate Cancer Canada, LRCP catalyst and Canadian Breast Cancer Foundation (E.A. Turley).
- Received November 11, 2015.
- Revision received January 10, 2016.
- Accepted January 14, 2016.
- ©2016 American Association for Cancer Research.