[Frontiers in Bioscience 3, d604-615, July 1, 1998] |
THE ROLE OF CD44 AS A CELL SURFACE HYALURONAN RECEPTOR DURING TUMOR INVASION OF CONNECTIVE TISSUE
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
3. TUMOR-ASSOCIATION HYALURONAN
In order to assess the significance of CD44 as a hyaluronan receptor during tumor progression, it is first necessary to understand the relationship between hyaluronan itself and tumor progression.
3.1. Structure and synthesis of tumor-associated hyaluronan
All evidence to date suggests that there is no difference in the basic structure of tumor-associated hyaluronan and hyaluronan synthesized by normal tissues. Hyaluronan is a member of the glycosaminoglycans--long nonbranching aminosugar polysaccharides that reside and function primarily within the extracellular space (7). Several properties distinguish hyaluronan as a unique member of the glycosaminoglycan family including: 1) It is the only member that does not contain esterified sulfate residues; 2) Individual chains of hyaluronan are of very high molecule mass, in the range of 1-6 x 106 Daltons; 3) Hyaluronan is not synthesized via attachment to a protein core, nor is it synthesized within intracellular organelles and; 4) Hyaluronan is the only glycosaminoglycan synthesized by eukaryotes as well as some prokaryotes. As shown in figure 1, hyaluronan contains a typical glycosaminoglycan repeating disaccharide structure comprised of beta-1,4-N-acetylglucosamine residues linked beta-1,3 to residues of glucuronic acid. Hyaluronan polysaccharide chains containing thousands of these disaccharide units are believed to be synthesized by a single enzyme termed "hyaluronan synthase," a multipass protein localized within the plasma membrane (16-18). As the hyaluronan is synthesized from individual UDP-sugar precursors, the elongated chain is simultaneously extruded through the plasma membrane and eventually released into the extracellular space (18, 19). Hyaluronan is synthesized, to varying degrees, by nearly all cell types (20). Some of the more avid hyaluronan-synthesizing cell types include fibroblasts and related fibrosarcoma, glioma and mesothelioma cells (15).
Figure 1. Disaccharide structure of hyaluronan. Hyaluronan is a linear, non-branching polysaccharide chain consisting of the disaccharide unit beta-1,4-N-acetylglucosamine linked beta-1,3 to residues of glucuronic acid. This disaccharide unit is repeated from approximately 2000-13000 times resulting in hyaluronan chains with molecular mass ranging from 1 - 6 x 106 Dalton.
3.2. Preferential enrichment of tumor tissue with hyaluronan
Recent studies, as will be discussed below, have made attempts to correlate levels of tumor-associated hyaluronan to other parameters of tumor progression, i.e., its use as a prognostic indicator. With this, an inherent assumption has already been made in these studies that hyaluronan is preferentially enriched within tumor extracellular matrix, i.e., it is the glycosaminoglycan of choice to examine. Is this a valid assumption? Is hyaluronan enriched in tumors over other glycosaminoglycans such as chondroitin sulfate or heparan sulfate? That an enrichment in tumor tissues of some form of polysaccharide-like substance, has long been recognized. A pathology textbook from 1907 describes a "mucin-like" substance commonly associated with malignant breast carcinoma (21). The mucin was described in the text as; acid in reaction, to have an affinity for basic dyes, and to be analogous to the mucinous substance present in the umbilical cord. Although well before the chemical characterization of hyaluronan in 1934 (22), the mucinous substance described was almost certainly hyaluronan. As better techniques became available, investigators re-investigated the composition of tumor-associated extracellular matrix. From the early use of chemical and spectroscopic analysis evolved methods such as cellulose acetate electrophoresis, differential reactivity to basic dyes, differential susceptibility to enzymes (i.e., the advent of purified hyaluronidases, chondroitinases and heparinases) and, HPLC separation of individual glycosaminoglycan disaccharides. Current approaches incorporate the use of specific RIAs, ELISAs and morphological probes. As these techniques evolved, so did the accuracy/validity of the detection. Nonetheless, the overall conclusions remained the same. Many tumors, both of epithelial and connective tissue origins, appear to be selectively or preferentially enriched in hyaluronan. Example tumors include human carcinomas (i.e., mammary, lung, pancreatic, parotid, prostatic, hepatic, esophageal, gastric and colonic carcinoma), as well as human gliomas, nephroblastomas, and mesotheliomas (23). Animal models of chicken sarcoma, fibrosarcoma, melanona and lymphosarcoma have also been described as enriched in hyaluronan (23). Three tumor types in particular within this listing, human mesotheliomas, nephroblastomas (Wilms' tumor) and, to a somewhat lesser extent, breast carcinomas, are considered to have the highest enrichment in hyaluronan. In these tumors, significant increases in hyaluronan are not only present within the primary tumor, but also readily detected in serum and other body fluids using current ELISA and RIA techniques (24-28). Although an increase in the proportion of hyaluronan to other glycosaminoglycans is common to most solid human tumors (as compared to the uninvolved normal tissue), there are notable exceptions. For example, studies on colon and bladder cancer have shown no change, or even a reduction in tumor-associated hyaluronan as compared to corresponding normal tissues (29, 30). However, even in these tumors there are reports to the contrary (25, 31).
Current methodologies rely on ELISAs, RIAs and morphological probes that specifically detect deposition of hyaluronan in situ, as a protein would be detected and quantified using monoclonal antibodies. Because hyaluronan is essentially non-immunogenic, the use of anti-hyaluronan antibodies is not available. Thus, indirect methods have been developed using connective tissue proteins such as the major proteoglycan of cartilage, aggrecan, and link protein, that naturally bind hyaluronan with high affinity and specificity via a hyaluronan binding region (HABR) of the protein. Using a biotinylated, partially digested aggrecan/link protein complex (HABR complex), De la Torre et al. demonstrated intense hyaluronan staining within sections of human breast carcinoma (32). Most of the hyaluronan was found deposited within the tumor-associated stromal elements. The nests of malignant infiltrating cells were essentially negative. Normal breast connective tissue was totally negative as was the uninvolved connective tissue adjacent to the carcinoma margin. That is, a clear diminution in staining was observed at the tumor margin. As shown in figure 2, panel C, we observed a similar staining pattern in one example of human breast carcinoma (using a similar biotinylated HABR probe). The tumor-involved areas are richly stained for hyaluronan. The tumor cells appear to be essentially unstained, with most reactivity occurring within the stroma. Panel D of figure 2 depicts a higher power micrograph of a different, more cellular, human breast carcinoma. Again, it appears that the predominant deposition of hyaluronan is within the residual stromal connective tissue. Staining within the tumor mass can also be seen however, but at this stage of tumor progression, the tumor stroma and parenchyma become more highly intermixed. In another study, hyaluronan distribution within normal and cancerous human gastrointestinal tissues was examined, again utilizing a biotinylated HABR probe (33). In the normal tissues, only the stratified squamous epithelia of the esophagus exhibited prominent hyaluronan staining, the simple epithelia of the stomach and large intestine were completely negative. Like the mammary cancers described above, the stroma of all of the gastrointestinal tumors displayed intense positive staining for hyaluronan stained as compared to normal connective tissue. Although hyaluronan deposition is often associated within stromal elements within the tumor tissue, there are notable exceptions. For example, in Wilm’s tumor (nephroblastoma) hyaluronan was localized within the epithelial blastemal cells, with little reaction in the adjacent stromal compartment (34).
Figure 2. Localization of hyaluronan and CD44 within sections of human breast carcinoma. Sections were made from two different samples of formaldehyde-fixed, paraffin-embeded human breast carcinoma. The samples were deparaffinized and incubated with biotinylated anti-human CD44H antibody (A3D8, Panel A) or biotinylated HABR complex (Panels C and D). Following incubation the sections were washed and processed using a Vectastain ABC kit (Vector laboratories). Following development of peroxidase reaction (brownish color) the sections were counterstained with Mayer’s hematoxylin. Malignant cells invading into adjacent stroma stain positively for CD44 expression (Panel A). Sections from the same tumor display prominent staining for hyaluronan (Panel D) with the majority of the staining associated with the stroma. Sections from a different tumor also display prominent staining for hyaluronan (Panel C). In the lower power view depicted in Panel C, clear demarcation of hyaluronan from more distant uninvolved mammary connective tissue can clearly be seen. Panel B represents an idealized cartoon of the interactions that are occurring in panels A, C and D.
Thus, a somewhat generalized conclusion can be made that hyaluronan is elevated in most solid tumor tissues, at least with respect to comparable normal tissues. Typically, but not always, this elevation is at the expense of other glycosaminoglycans (35, 36). However, in cases where other glycosaminoglycans are upregulated, hyaluronan is often upregulated as well. The particular tumor-association glycosaminoglycan composition may be dependent on the particular host connective tissue in which the tumor resides (especially if the source of the glycosaminoglycan is the host connective tissue cells). For example, when rabbit V2 carcinoma invades into rabbit muscle there is a nine-fold increase in total glycosaminoglycan composed predominately of (71%) hyaluronan (37). However, when the same V2 carcinoma cells invades into rabbit mesenteric tissue, there is a nine-fold increase in chondroitin sulfate but only a three-fold increase in hyaluronan, resulting in a tumor-associated matrix composed of approximately equal concentrations of hyaluronan and chondroitin sulfate (38).
3.3. Is the level of tumor hyaluronan content of diagnostic or prognostic value?
As discussed above, in some cancers such as mesothelioma and nephroblastoma, hyaluronan levels are elevated to such an extent that the elevation can be detected in the serum (27) or urine (39). In the study of mesothelioma patients, elevated serum hyaluronan was found in all patients at time of presentation. Patients that presented with serum hyaluronan at, or above, 250 mg/l showed significant likelihood to fall into the progressive disease group whereas, those with lower values typically fell into a group that responded to therapy. It was concluded that serum hyaluronan in this malignancy was predictive of progressive disease. In a study of Wilm’s tumor patients, urine hyaluronan was significantly elevated in 74% of the patients pre-operatively as compared to normal control volunteers (39). Post-operatively, urine hyaluronan values were significantly reduced, returning to near normal in disease-free patients. In post-operative relapse patients, urine hyaluronan levels began to again increase. Thus, for Wilm’s tumor there also appears to be a close correlation between urine hyaluronan levels and disease progression.
Serum hyaluronan was also monitored in malignant breast carcinoma. Initial results from one group of investigators suggested a significant elevation in serum hyaluronan in patients with disseminated metastatic disease as compared to patients with malignant disease without metastasis or benign disease of the breast (26). However, a subsequent study, with a larger population of patients, showed no prognostic significance of serum hyaluronan in breast cancer (40). In the latter study, serum hyaluronan levels were similar to those of control patients.
In a small number of studies, attempts have been made to quantify hyaluronan content within the primary tumor and compare these values to tumor grade or stage. In a study of 35 cases of malignant brain tumors (astrocytomas, gliomas and meningiomas) the investigators demonstrated that hyaluronan levels within tumor tissues were substantially elevated as compared to normal brain tissue (36). However, the elevated hyaluronan levels showed little statistical change with tumor grade. The hyaluronan levels in brain tumors grades II through IV were all elevated (similar to the concentrations present in fetal brain). A similar finding was obtained in a study of breast carcinoma (41). Although the hyaluronan levels were substantially elevated in the breast carcinoma tissue as compared to normal tissues, there was little change in hyaluronan in malignancies between grades I-III. A more recent study examined the significance of hyaluronan in colorectal cancer (42). In this study of 202 samples of patients with colorectal adenocarcinoma, all adjacent normal intestinal epithelium, in all samples, was negative for hyaluronan. However, 93% of the tumor tissues analyzed were positive for hyaluronan (i.e., 7% of the cases were hyaluronan negative). It was also observed that high grade tumors had moderate to strong hyaluronan staining intensity while lower grades were typically weakly stained. Patients where the tumor-associated hyaluronan was absent or weakly stained were noted to have a higher survival rate than patients with moderate or strong hyaluronan staining. However, as will be discussed below, a better prognostic indicator that was revealed from this study was the association of hyaluronan with tumor epithelium versus tumor stroma (i.e., the relative distribution of hyaluronan within tumor tissue). The authors speculated that the tumor epithelial-association of hyaluronan could have been the result of an interaction with tumor cell CD44. When more studies are performed it may be found that tumor-associated hyaluronan levels, although elevated, are not varying appreciably with tumor grade or stage. Instead, what may be of more prognostic value are the expression levels of proteins that interact with the hyaluronan, either hyaluronan-binding proteins within the matrix or, the expression of hyaluronan receptors such as CD44.