[Frontiers in Bioscience 3, d604-615, July 1, 1998]
Reprints
PubMed
CAVEAT LECTOR




Table of Conents
 Previous Section   Next Section

THE ROLE OF CD44 AS A CELL SURFACE HYALURONAN RECEPTOR DURING TUMOR INVASION OF CONNECTIVE TISSUE

Warren Knudson

Department of Biochemistry, Department of Pathology, Rush Medical College, Rush-Presbyterian-St. Luke's Medical Center, Chicago, IL 60612-3864

Received 4/27/98, Accepted 5/15/98

4. THE FUNCTION OF TUMOR-ASSOCIATED HYALURONAN

Although hyaluronan concentrations are elevated in many tumors and, may even be prognostic for some, it remains to be determined whether this hyaluronan is important or serves any relevant function. Does the presence of hyaluronan facilitate tumor progression and if so, by what mechanism(s)? Alternatively, is the hyaluronan synthesized as a host response to tumor invasion, i.e., a desmoplastic response to present a matrix barrier or even to "wall-off" or encapsulate the growing tumor? Although no definitive answers are available, considerable research and speculation have been directed toward these questions. Whatever, the function, the association of hyaluronan with embryonic development, wound healing and regeneration suggests that the accumulation of this glycosaminoglycan is a common feature of tissue remodeling in physiologic as well as pathologic settings.

4.1. Hyaluronan and cell migration: A function by association

The potential importance of hyaluronan to aspects of cellular function, were first noted by Toole and his co-workers in embryonic systems. Hyaluronan was observed to be actively synthesized and deposited nearly coincident with the onset of migration of particular embryonic cells (43). Furthermore, the cessation of migration was tightly correlated with a dramatic reduction in the levels of hyaluronan within the connective tissue. This pattern was documented in the migration of dorsal trunk and cranial neural crest cells (e.g., presumptive corneal fibroblasts) as well as the migration of endocardial cushion cells of the developing heart (44-46). Subsequent to these early observations, increasing hyaluronan deposition was also found in regenerating and wound healing systems in adult tissues. For example, large concentrations of hyaluronan are deposited precisely coincident with the extensive migration of blastemal cells during limb regeneration in the newt (47). Results such as these led to the suggestion that the presence of large concentrations of extracellular hyaluronan were, in some way, functioning to facilitate cell migration. However, the evidence for such a role was still circumstantial. One attempt at a direct approach was the injection of Streptomyces hyaluronidase into the sub-blastodisc of stage-8 chick embryos, a region known to be involved in the migration of dorsal neural crest cells. Although the hyaluronidase treatment did not totally prevent the migration of the neural crest cells, the migration pathway was significantly truncated (48). In still another system, high concentrations of hyaluronan were also expressed in a temporal fashion within the developing embryonic chick limb. However, little overt cell migration is occurring at this developmental stage—being a stage better characterized by extensive cell proliferation. So, the original model for hyaluronan was expanded to include facilitation of cell division as well as migration (49).

When investigators re-examined tumor tissues and found an enrichment in hyaluronan, a function was proposed based on the association of hyaluronan during development, wound healing and regeneration discussed above, i.e., to provide an embryonic-like extracellular environment, conducive to cell migration and/or proliferation. An early attempt to validate this role was made by Toole and co-workers using a highly invasive rabbit carcinoma. When inoculated at either a subcutaneous or intramuscular site in the nude-mouse, non-invasive tumors were established (37). These well-encapsulated benign tumors contained 3-4 fold less hyaluronan than the same tumor grown in the rabbit (the natural host), where the cells behaved as an aggressively invasive and metastatic tumor. Thus, an accumulation in hyaluronan appeared to be associated with the active invasion of cells into adjacent tissue. In a more recent work, a time-dependent increase in stromal staining for hyaluronan was observed in lining mesentery tissue following metastatic attachment and growth of murine ovarian carcinoma cells (50). Similar accumulations were observed in the metastatic seeding of murine breast carcinoma cells (50). Hyaluronan concentration increased each day coordinate with the invasion and growth of these cells at the metastatic site. The mesenteric cells were suggested as the source of the hyaluronan as neither the ovarian nor the mammary carcinoma cells exhibited a capacity to synthesize hyaluronan in vitro. Thus, it appeared that the host tissue was responding to the attachment, proliferation and invasion of the malignant cells. The stromal association of hyaluronan, in the human tumors described above, as well as those illustrated in figure 2C and D, suggest that there is a host response to the presence of actively migrating cells. The question remains whether this is an effort to mount a "protective" response, induced by the host tissue cells or, a process of "matrix engineering," or tissue remodeling, directed by the infiltrating tumor cells, i.e., an effort to establish a more conducive extracellular environment. Unfortunately, observations again appear to point in both directions. In the benign rabbit V2 carcinoma tumor, established in the nude mouse, large concentrations of hyaluronan were apparently not necessary for the host to mount an effective response to the presence of a growing tumor mass. Second, several studies have demonstrated that tumor cells have the capacity in vitro to stimulate hyaluronan synthesis by normal connective tissue cells such as fibroblasts (50-53). Some of the tumor-directed hyaluronan stimulatory activities, especially those involving soluble factors, have been defined (e.g., tumor cell-derived PDGF) (54), while others remain to be elucidated (55). Many of the same tumor cells that exhibit a capacity to induce increases in hyaluronan synthesis by adjacent fibroblasts also exhibit a capacity to induce fibroblast production of matrix metalloproteinases (56). Thus, these malignant cell types have a (potentially) tremendous capacity for matrix engineering or directing the remodeling of adjacent extracellular matrix. Such remodeling would be predicted to result in an increase in tissue hydration, likely reminiscent of the embryonic tissue environment—an environment conducive to cell migration and proliferation.

Alternatively, some investigators have correlated the cellulular capacity for CD44-mediated endocytosis and degradation of hyaluronan (assayed in vitro), to tumor metastatic aggressiveness (53, 57). That is, the malignant cells that are best equipped to internalize and degrade hyaluronan appear to be the most efficient at metastasis. One interpretation of these results is that tumor-associated hyaluronan presents a barrier to invasion, that is breached effectively by cells with a capacity to bind, internalize and degrade the glycosaminoglycan. However, others have suggested binding and endocytosis of a substratum ligand as part of the mechanism for cell locomotion (58). Thus, the relationship between endocytosis capacity and its involvement in tumor progression remain to be better defined. This may require methods to selectively inhibit CD44-mediated endocytosis and determine whether tumor cells maintain hyaluronan-mediated locomotion, both in vivo and in vitro.

4.2. Proposed mechanisms for the involvement of hyaluronan in cell migration and proliferation

One mechanism proposed to explain how hyaluronan could potentially facilitate cell migration and proliferation was based on the intrinsic physicochemical nature of hyaluronan, i.e., deposition of high concentrations of hyaluronan result in an expansion of, or "loosening," of spaces within the connective tissue (59). In the embryonic systems described above, the deposition of hyaluronan was associated with an overall expansion of the tissue (43). Conversely, the removal of hyaluronan was associated with a dramatic shrinkage of the tissue and coincident with loss of tissue hydration (43). The injection of Streptomyces hyaluronidase into the chick embryo described above resulted in a complete collapse of tissue spaces between somites and somitomeres as well as the overlying ectoderm, neural tube and endoderm (48). Thus, the hydrodynamic effects of hyaluronan, i.e., the hydration and subsequent expansion that occurs in the presence of a high molecular mass, negatively charged polysaccharide, were often suggested as the primary mechanism for how hyaluronan may facilitate cell migration and/or proliferation. Again however, this model was based on evidence of coincidence. To address this question directly, investigators established model systems for cell migration. Nonetheless, the results obtained were ambiguous. For example, the migration of neutrophils is completely inhibited in gels highly enriched in hyaluronan (60). On the other hand, the migration in hyaluronan gels of other cell types such as fibroblasts, appears to be enhanced (61, 62). This suggested that hyaluronan involvement in cell migration was somehow cell-specific, serving as a favorable milieu for some cell types, an impediment for others. This also led investigators to look for the nature of these cell-specific differences, possibly due to the presence or absence of cell surface hyaluronan receptors (e.g., CD44), or perhaps again, to differences in CD44-mediated capacity for hyaluronan endocytosis.

The role of direct tumor cell - hyaluronan interactions in facilitating cell migration and/or cell proliferation fall primarily into two categories, i) hyaluronan as a matrix support for haptotaxic cell locomotion as depicted in figure 2B and, ii) hyaluronan serving as a cell surface-bound, or pericellular, coating. Concerning a role of hyaluronan as cell locomotion substratum, Thomas et al. demonstrated that the stable transfection of human melanoma cells with pCD44 resulted in an enhanced motility of the cells on hyaluronan-coated surfaces as compared to the parental, CD44-negative, cells (63). This enhanced motility was inhibited by anti-CD44 antibodies or, by the presence of a CD44 ligand competitor, soluble CD44-immunoglobulin fusion protein. Thus, hyaluronan could function as an insoluble matrix ligand support for the locomotion of tumor cells, mediated via CD44. However, regardless of this in vitro potential the question remains whether CD44 - hyaluronan interactions participate in cell adhesion, migration or proliferation in vivo In a recent study, Yu et al., attempted to answer this question by transfecting a CD44-positive murine mammary carcinoma cell line (TA3/St) with cDNA encoding soluble CD44 (53). The transfectants spontaneously release soluble CD44 that, in turn, competes with the activity of endogenous cell surface-localized CD44. Whereas, following tail vein injection into syngeneic mice, control cells formed massive lung metastases, metastases were reduced to near zero in the transfectants expressing soluble CD44. When lung metastases did become established, stimulated deposition of hyaluronan was observed. These data serve to support the suggestion that, in vivo, some form of tumor cell CD44-mediated interaction is important for the stabile establishment and growth of distant metastases. That the critical partner in this interaction with CD44 is hyaluronan, although highly likely, still remains to be definitively determined.

Tumor cells expressing CD44 have the capacity to bind hyaluronan and, in the presence of other hyaluronan binding proteins (i.e., proteoglycans such as aggrecan, versican or hyaluronectin), assemble the components into a pericellular matrix shell or coat (64). Analogous to a bacterial glycocalyx, this pericellular coat may serve to cocoon the malignant cells, or serve as a shield, to prevent undesirable cell-cell interactions. Undesirable cell-cell interactions would included homotypic as well as heterotypic cell-cell interactions. Providing a homotypic cell-cell barrier may help in facilitating cell proliferation. Providing a heterotypic barrier may help malignant cells evade immune surveillance or inhibit T-cell mediated cell lysis. Some cells such as synovial fibroblasts synthesize sufficient hyaluronan and hyaluronan binding-proteins to exhibit an endogenous pericellular matrix. Early studies demonstrated that the pericellular coat exhibited by synovial fibroblasts served as an effective barrier to lymphocyte-mediated cell lysis in vitro (65). Following treatment with hyaluronidase, this barrier was removed and the lymphocytes were allowed direct cell-cell contact with the synovial cells (65, 66). Many tumor cells, particularly those derived from carcinomas, do not make the levels of extracellular matrix components necessary to establish a pericellular coat such as that of a synovial lining cell (64). However, these same tumor cells do express CD44 and can bind hyaluronan synthesized and secreted by other cell types. Given the appropriate extracellular macromolecules, epithelial tumor cells can assemble a pericellular matrix nearly identical to that of synoviocytes (64). An example of such is illustrated in figure 3. Panel A depicts anti-CD44 staining of the MCF-7 human mammary carcinoma cell line in co-culture with a rat fibroblast-like cell line that synthesizes copious amounts of hyaluronan. This in vitro system was designed to represent a simplified model of the tumor tissue shown in figure 2. As can be seen, the MCF-7 cells display a rich immunostaining for CD44. However, in the process of plating the cells in co-culture an unfortunate accident occurred—the cells became contaminated with bacteria. As shown in panel B, the MCF-7 cells cultured separately were completely encompassed by contaminating bacteria. However, in the co-cultures (panel C), the fibroblast-like cells exhibited a pericellular matrix barrier that prevented bacterial encroachment. Even more interesting was that the tumor cells within the culture (the more polygonal to rounded cells) also exhibited a protective pericellular coat. Similar results were obtained in co-cultures of human colon carcinoma cells (HT29) and fibroblasts (panel D). Addition of 1 U/ml of Streptomyces hyaluronidase for 30 minutes, resulted in the complete disillusionment of this pericellular matrix and now, the bacteria were able to come into direct contact with both the tumor cells and the fibroblasts (panel E). Considering that these tumor cell lines have little capacity to synthesize hyaluronan or, at least enough to establish an endogenous pericellular matrix (panel B), a most likely explanation is CD44-mediated assembly of fibroblast-derived matrix macromolecules. We have demonstrated previously that COS-7 cells transfected with pCD44, or tumor cells already exhibiting a high level of CD44 expression, have the capacity to assemble similar pericellular matrices, in the presence of exogenously added hyaluronan and proteoglycan (64, 67). These pericellular matrices were visualized by the exclusion of uniform particles (i.e., glutaraldehyde-fixed, horse red blood cells). The ability of the cells in figure 3 to exclude live bacteria serves to demonstrate the barrier capacity of such hyaluronan-rich cell coats. Whether this mechanism occurs in vivo, as with other CD44:hyaluronan-mediated activities, remains to be determined. Nonetheless, the potential for establishing a glycosaminoglycan-rich protective cocoon exists.

Figure 3. Formation of hyaluronan-rich pericellular matrices around CD44-positive tumor cells in vitro. Co-cultures of MCF-7 human mammary carcinoma cells (round-polygonal shaped cells) together with cells derived from a rat fibrosarcoma cell line (slender, elonged cells, (64)) were fixed and incubated with a biotinylated anti-human CD44H monoclonal antibody (A3D8). Following incubation the cells were washed and processed using a Vectastain ABC kit (Vector laboratories). The human tumor cells displayed prominent staining for CD44 (Panel A). All of these cultures then became heavily contaminated with bacteria. Panel B depicts cultures of MCF-7 cells grown independently. The cells are completely engulfed with bacteria. However, the same cells in co-culture with fibroblasts display excluded zone of pericellular matrix to which the bacteria cannot penetrate (Panel C). Co-cultures of human colon carcinoma cells, HT-29 display similar barrier matrices (Panel D). Following treatment of MCF-7 containing co-cultures with 1U/ml Streptomyces hyaluonidase (Panel E), the protective barrier is lost and the bacteria are allowed direct contact with both cell types within the co-culture.

It should also be noted that elaborate pericellular coats may function to reduce cell-matrix interactions and work to inhibit migration, i.e., masking receptors interactions necessary for cell locomotion. For example, we have shown previously that, in the presence of hyaluronan alone, the hyaluronan becomes bound via CD44 but is insufficient to establish a pericellular coat capable of excluding particles (64, 68). However, when aggrecan proteoglycan is added to the cells in addition to hyaluronan, pericellular matrices form within two hours of culture. Why is this important? Addition of aggrecan or versican proteoglycan to cultures of neural crest cells inhibited their migration on fibronectin and/or laminin-coated substrata (69). The authors suggested that the deposition of hyaluronan-binding proteoglycans into the extracellular matrix may serve as natural matrix cues to signal a migration stop during development. Putting the two sets of data together, we hypothesized that it is the assembly of a proteoglycan plus hyaluronan pericellular matrix cocoon that functions to prevent migration (1). Taken further, this model would predict that during development (and possibly tumor invasion), cells expressing CD44 migrate along hyaluronan-containing matrix tracts until they encounter areas rich in hyaluronan as well as proteoglycan. After encountering the proteoglycan-enriched matrix, the cells become coated in a pericellular matrix and their migration ceases. Taken another step, this would predict that in healthy, adult connective tissues (containing hyaluronan and proteoglycans such as versican), cell migration is actively inhibited by these natural stop signals. However, following proteolytic activity and subsequent removal of the proteoglycan, coupled with elevated hyaluronan synthesis, a migratory-favorable extracellular environment would be established. In the study of brain tumors described above, the tumor-associated hyaluronan levels were elevated, but varied little with tumor grade (36). However, the level of the hyaluronan-binding proteoglycan hyaluronectin (related to versican) dropped precipitously with tumor grade. The ratio of hyaluronectin to hyaluronan was reduced from 1.3 – 6.5 in normal adult brain tissue to 0.8 – 4.3 in grade II and III tumors, to 0.01 – 0.4 in grade IV tumors. The grade IV ratios were similar to ratios present in fetal brain tissues.