We previously identified a 90-kD (GP90), collagen-binding, membrane glycoprotein, termed extracellular matrix receptor III (ECMR III), that is homologous to the lymphocyte homing receptor and CD44 antigen (Gallatin, W. M., E. A. Wayner, P. A. Hoffman, T. St. John, E. C. Butcher, and W. G. Carter. 1989. Proc. Natl. Acad. Sci. USA. 86:4654-4658). CD44 is abundantly expressed in many epithelial tissues, and is localized predominantly to filopodia in cultured keratinocytes. Here we establish CD44 as a polymorphic family of related membrane proteoglycans and glycoproteins possessing extensive diversity in both glycosylation and core protein sequence. Human neonatal foreskin keratinocytes (HFKs) and QG56 lung squamous carcinoma cells express an alternatively spliced form of the CD44 core protein (termed CD44E) that contains an additional 132 amino acids in the carbohydrate attachment region of the extracellular domain. HFKs, HT1080 fibrosarcoma and QG56 cells, as well as many other human cells, contain varying ratios of GP90 and structurally related, higher molecular mass forms of CD44 that express the following characteristics: (a) each form reacted with anti-CD44 (mAbs) P1G12, P3H9, and P3H5. Each of these mAbs recognized a distinct, nonoverlapping epitope present on each CD44 form. (b) Differences in mass were due primarily to variation in carbohydrate moieties, including sulfated asparagine-linked glycopeptides (GP), chondroitin sulfate (CS), and heparan sulfate (HS) glycosaminoglycans, as well as O-linked mucin and polylactosamine structure(s). The major polymorphic forms were designated HT1080 GP90 and CS180, QG56 GP230, and HFK HS/CS250, based on dominant carbohydrate moieties and relative mass. (c) The polymorphic forms use CD44 and CD44E core proteins, each containing a unique set of potential attachment sites for O- and N-glycosides and glycosaminoglycans. (d) Immunofluorescence microscopy, differential extraction with Triton-X-114 detergent, and incorporation into liposomes indicated that all the forms were membrane bound glycoconjugates. These results define CD44 as a structurally diverse, but immunologically related, set of intrinsic membrane macromolecules, and suggest that these structurally varied forms might be expected to manifest multiple functions.
