Hyaluronan
and its binding
proteins,
the hyaladherins
B.P. Toole Department
of Anatomy
and Cellular Biology, Tufts University Boston, Massachusetts, USA
Health
Science
Schools,
Current Opinion in Cell Biology 1990, 2:839-844
Introduction
Hyaluronan (HA) is a high-molecular-weight, highly anionic polysacchande found in extracellular matrices and at ceU surfaces. It is composed of 200-10000 disaccharides of P-1,4-glucuronate-P-1,3-N-acetylglucosaminegiving a molecular weight of between 1 x 105 and 5 x 106, usually at the higher end of this range. No branching of the polymer or variations in composition of the repeating disaccharide have been demonstrated convincingly. Despite the apparent monotony of this polymer, its unique physicochemical properties and its consequent role in tissue structure and function have fascinated polymer biophysicists and physiologists for decades. These properties include its voluminous structure in solution, its ability to form continuous networks at low concentration, and its consequent viscoelasticity, osmotic properties, and effects on molecular exclusion, diffusion, flow resistance, etc. (reviewed in Comper and Laurent, Pbysiol Rev 1978, 58:255-315). In addition to its long-established physicochemical function, HA is important in regulating ceU behavior. HA is enriched in extracellular matrices in which cells migrate or proliferate (reviewed by Toole, In Cell Biology of the Extracellular Matrix edited by Hay E. Plenum, 1981, pp 259-294), and recent evidence supports a role for HA in these cellular processes. In addition, HA has been shown to ineuence cell interactions and differentiation in some cellular systems. Underlying its role in ceU behavior are its unique physicochemical properties and its interactions with hyaluronan-binding proteins (HABPs), both in exttaceUu& matrices and on cell surfaces. Recent progress in this field has emphasized the molecular nature and role of I-L4BPs. Until this year, it was thought that there were at least two classes of HABPs: structural HABPs such as HAbinding proteoglycans and link proteins; and cell surfaceassociated HABPs that have properties suggesting that they are HA receptors. It has now become apparent that these two groups of HABPs have homologous I-IA-binding domains and belong to a single family of proteins; I suggest that this family be termed the hyaladherins. In this review I will discuss the recent work demonstrating the inter-relationships between the various I-L4BPs,
the importance of HA interaction with ceU surface-associated HABPs in ceU behaviour, and some preliminary, but promising, work on the potential importance of HA stimulatory factors in regulating these events. Several interesting aspects of current research on the structural properties and metabolism of HA, and its role in ceU behavior and pathology have been discussed in recent Ciba Foundation Symposium ( 7%eBiology ofHyaluronan edited by Evered D and Whelan J. Ciba Found Sjmp, 1989).
Hyaluronan-binding
proteins
The most thoroughly studied HA-binding macromolecules are the link proteins and large proteoglycan, termed aggrecan, of cartilage (reviewed in Gallagher, Cr.07 Opin Cell Bioll989,
1:1201-1218). Goetinck et al. (I Cell Biol
1987,105:2403-2408) have provided convincing evidence that the tandemly repeated domains IIa and IIb (B and B’ loops) at the carboxy terminus of cartilage link protein contain the HA-binding region. Four different peptides, derived from two corresponding regions of each of the tandem repeats, were found to block binding of HA to link protein, implying that all four regions were involved in the binding. Each of these peptides contains a cluster of positively charged amino acids and binding decreases with increase in ionic strength, suggesting that binding is ionic in nature. The amino-terminal region of the core protein of aggrecan contains a domain homologous to the tandem repeats in link protein and thus is assumed to mediate HA binding to aggrecan (Neame et al, J Biol Chem 1987, 262:1776%17778). Last year, similar homologies were described in several other macromolecules: near the amino-terminus of the HA-binding proteoglycan, versican, which is produced by human iibroblasts [l] ; in an HABP produced by human glial cells [2] that is related to the HABP, hyaluronectin (Delpech and Halavent, J Neurocbem 1981, 36:855-859), and to versican [ 11; and at the N-terminus of the adhesive protein(s) variously known as CD44, Pgpl, Hermes antigen, ECMRIII coUagen receptor, HCAIvl, etc.-here termed CD44 [3,4]. The homologies between these proteins are further discussed by Zimmerman and Ruoslahti [ 11.
Abbreviations bFG&basic HA-hyaluronan
fibroblast (hyaluronate,
growth factor; BHK-baby hamster hyaluronic acid); HABP-hyaluronan-binding
@ Current
Biology
kidney;
Ltd ISSN 0955*74
FSH-follicle-stimulating protein; TCF-transforming
hormone; growth
factor.
839
840
Cell-to-cell
contact
and extracellular
matrix
It is generally accepted that the link proteins and aggrecan of cartilage are structural macromolecules that, through interaction with HA, form gigantic ternary complexes that contribute to the physical properties of cartilage. Similar structural complexes of HA, link protein and proteoglycans such as versican presumably participate in building the matrices of several other connective tissues. The structure and distribution of these complexes were reviewed by Gallagher (1989). However, there is a separate group of HABPs that apparently serve as cell-surface receptors for HA and mediate the effects of HA on cell behavior that are described below. Two putative cellsurface receptors of this type have been studied extensively. The first of these was originally described as HAbinding sites of high afhnity on the surface of SV-3T3 cells (Underhill and Toole, J Cell Biol 1979, 82:475-484; Underhill and Toole, J Biol C&em 1980,255:4544-4549) and has now been identified in baby hamster kidney (BHK) cells as an 85 kD glycoprotein recognized by the K.3 monoclonal antibody (Underhill et al, J Biol Chem 1987, 262:13142-13146; herein termed the 85 kD HA receptor). Another 85kD HABP present at the surface of cultured libroblasts is derived from serum [5] but this protein has properties distinct from the BHK receptor. The second putative receptor was iirst isolated from the culture medium of 3T3 cells and chick embryo libroblasts as a mixture of 56-70 kD proteins (Turley et al, Biocbemisttty 1987, 26:2997-3005) but has been shown subsequently to form large aggregates at the cell surface [ 6). Both these putative HA receptors interact directly or indirectly with the cytoskeleton (Iacy and Underhill, J Cell Biol 1987, 105:1395-1404) [7] and mediate various effects of HA on cell behavior (see below) but their biochemical inter-relationship is not yet known. Recently, my colleagues and I have obtained monoclonal antibodies to two classes of chick embryo HABPs that are widely distributed on the surfaces of embryonic and transformed or tumor cells (Banerjee and Toole, unpublished). One group of antibodies recognizes proteins of molecular weights of 93, 90, and 69kD, respectively, that may be related to the 85kD HA receptor. The other group of antibodies ret ognizes a protein of molecular weight of approximately 300kD; the relationship of this protein to other HA receptors is not yet known. A cell-surface HABP is also involved in uptake of HA from the circulation. Clearance of HA takes place mainly in lymph nodes (Fraser etal, Biocbem J1988,256:153158) and in the liver where the site of uptake is the endothelium (Fraser et al, Cell TissRes 1985,242:505-510). Binding, internalization and degradation of HA by liver endothelial cells has been demonstrated in culture (Smedsrod et al, Biocbem J 1984, 223~617626; Laurent et al., Biocbem J 1986, 234:653658). The HA-binding sites in volved have now been shown to be recycled during endocytosis of I-IA and it appears that 75% of these sites are internal during cycling [8]. Internalization and degradation of HA by other vascular endothelium has also been demonstrated (McGuire et al., J Cell Physoll987, 133:267-276). The HABP that mediates endocytosis has not yet been identihed but it is probably different from
the 85kD HA receptor (Raja et al., J Biol Chem 1988, 263:16661-16668) An important development is the recent iinding that the adhesive protein CD44 contains a domain that has approximately 30% homology with the HA-binding regions of link protein and aggrecan [3,4]. Significantly, however, the basic residues previously implicated in HA binding to link protein are not conserved in CD44. CD44 is very similar or identical to the Hermes antigen, Pgp-1 and ECMRUI (Picker et al, J Cell Bioll989, 109:927-937; Gallatin et al., Proc Natl Acud Sci c/sA 1989, 86:4654-4658; Wolffe et al., J Biol Chem 1990, 265:341-347). These molecules are involved in T-lymphocyte activation and adhesion, and homing of circulating lymphocytes via interaction with high endothelial venules, the entry site to organized lymphoid tissues. Although mainly studied in the hematopoietic system, CD44 and its homologues are widely distributed (Picker et al, 1989; Carter and Wayner, J Biol Chem 1988, 263:4193-4201). They also exist in varied molecular forms: in hematopoietic cells, CD44 is synthesized as a 37 kD protein that is processed to form a glycoprotein with a molecular weight of 80-95 kD [3,4] or a minor chondroitin sulfate proteoglycan form with a molecular weight 180-200 kD &lkanen et al, J Jmmunoll988, 141:1615-1623) [3]. Epithelial and carcinoma cells produce yet another form, of approximately 160 kD molecular weight (Picker et al., 1989) [ 31. CD44 presumably mediates adhesion of cells to extracellular matrices because it exhibits binding to collagen (Carter and Wayner, 1988) as well as containing the putative HA-binding domain. Interestingly, two other HABPs, link protein (Chandrasekhar et al., J Biol cbem 1983,258:622&6231) and aggrecan (Toole and Iowther, Biocbem J 1968, 109:857-866), also bind to collagen. Evidence published this year indicates that CD44 is in fact the same as, or very similar to, the 85kD HA receptor of BHK cells. Aruffo et al. [9] transfected hamster CD44 cDN& into COS cells and demonstrated expression of cell-surface reactivity with K3 antibody to the 85 kD HA receptor. They also showed that hyaluronidase treatment of various tissues, or competition with HA, blocks binding of soluble forms of 044. M Culty, K Miyake, P Kincade, E Sikorski, E Butcher and C Underhill (personal communication) have obtained additional evidence for identity of CD44 and the 85kD HA receptor; this evidence includes immunoprecipitation of HA-binding activity from cell extracts with antibodies to CD44, and inhibition of HA binding and HA-mediated cell aggregation with these antibodies. Supporting these lindings is the work of Lesley et al. [ 101 who showed that antibodies to CD44 block binding of HA to lymphocytes and HA-mediated lymphocyte aggregation. It is not yet clear, however, if the HA-binding function of CD44 is involved in the homing of lymphocytes to high endothelial venules. Although binding of soluble CD44-immunoglobulin fusion proteins to high endothelial cells is blocked by treatment of the cells with hyaluronidase or by competition with HA (91, binding of intact lymphocytes is not (C. Underhill, personal communication).
Hyaluronan
Because CD44 and the 85 kD HA receptor are very similar, if not identical, the latter must also share homologies with the HA-binding domains of the structural HABPs: aggrecan, versican and link protein. However, other data distinguish the HA-binding properties of these structural molecules and the HA receptor (see Toole et al, in 7&e Role of Extracellular Matrix in Development edited by Tretstad R. Alan I&s, 1984 pp 43-66). The first and most marked distinction is in the length of HA oligosaccharide required for recognition by the HABPs. Link protein, aggrecan, hyaluronectin, and the proteoglycan, PG-M, of chick embryo fibroblasts require a HA decasaccharide for binding. Oligosaccharides of smaller lengths exhibit negligible binding, and the affinities of binding of HA decasaccharide and HA polymer do not dilfer greatly. However, cell-surface HA receptors recognize HA hexasaccharide and their afftnity of binding increases dramatically with the length of the ligand. Second, the HA receptor is not as highly speciiic for HA as link protein and the proteoglycans. Third, increases in ionic strength enhance the affiity of interaction of HA with the HA receptor but decrease a&&y for link protein and aggrecan. Presumably, sequence and conformational differences in the HA-binding regions of the structural HABPs and CD44/HA receptor allow for these differences in ligand recognition. Despite the differences in binding behavior it is clear that these macromolecules make up a family that share related HA-binding motifs. Therefore, they deserve a family name: the hyaladherins.
Hyaluronan
and cell behavior
HA has been shown to influence several types of ceU behavior, but its effects depend on at least three parameters: the size and concentration of HA and the ceU type in question (discussed in Goldberg and Toole, A&r Rbeum
1987, 30:76%778). A positive correlation has been demonstrated between HA synthesis, HA synthetase activity and ceU proliferation (Brecht et al, Biocbem J 1986, 239445-450;). Inhibition of HA synthesis leads to arrest of cells in mitosis prior to rounding, possibly because HA-mediated detachment from the substratum is required (Brecht et al, 1986). Recently, the 85 kD HA receptor and, in some cases, HA itself have been localized to the surface of a variety of proliferating epithelial cells [ 111; however, other work suggests that HA accumulation is not restricted to the proliferating zone of the epidermis [12]. Also, endothelial cell proliferation is stimulated by HA-derived oligosaccharides but is inhibited by polymeric HA [ 131. Clearly, much further work is required to establish the precise relationship of HA to proliferation. Many studies have suggested a role for HA in cell rngration, both in creating hydrated pathways that facilitate penetration of tissues (Toole, 1981) and in direct influences on ceU locomotion (Turkey, Ciba Found 5” 1989, 143121-137). Turkey and colleagues (I Cell Sci 1985, 78:133145) [14] have shown that addition of an
and its binding
proteins,
the hyaladherins
Toole
HABP plus HA to fibroblasts stimulates ceU movement and that this HABP is concentrated in the rt&ling lamellae of actively locomoting cells. The endogenous ceU surface HABP of these libroblasts is associated with protein kinase activity that is stimulated by addition of HA to the cells or to the isolated HABP complex, pointing to the possibility that instructive signal transduction results from interaction of HA with ceU surface HABPs [6]. Grey et al. [ 151 have recently purilied a 70 kD ceU migration-stimulating factor from fetal libroblasts and fibroblasts derived from cancer patients. This factor stimulates confluent cells to penetrate collagen gels by an HA-dependent mechanism. The ability of confluent cells to penetrate collagen gels was shown to correlate with the level of HA synthesis and size of HA produced at conlluence [ 161. Factor-induced stimulation of confluent adult fibroblasts to invade collagen gels was neutralized by treating the cells with HA-specilic hyaluronidase, as was the spontaneously elevated ability of fetal and cancer patient fibroblasts to invade the collagen gels [ 171. Possibly the best-studied iniluence of HA is in modulation of cell aggregation. Addition of low concentrations of HA to lymphoma cells or macrophages induces their aggregation. On the other hand, the divalent cation-independent aggregation of several transformed ceU lines is inhibited by treatment with hyaluronidases or high concentrations of HA (Wright et al, Cancer Res 1981,41:5107-5113; Underhill and Toole, Eap Cell Res 1981, 131:419-423). Underhill and Toole (1981) demonstrated that these aggregation phenomena are dependent on crossbridging by HA of HA receptor sites on adjacent cells; inhibition of crossbridging at high HA concentrations is due to saturation of these receptors. It has been shown since that HA-mediated aggregation of virally transformed ceU lines and macrophages is inhibited by addition of K3 antibody to the 85kD HA receptor (Green et al, Eqi~ Cell Res 1988, 178:224-232) and that HA-mediated lymphocyte aggregation is inhibited by antibodies to CD44 [ 101, a molecule now known to be related to the HA receptor [9]. The very recent observation that transfection of libroblasts with CD44 cDNA induces their ability to selfaggregate (St John et al, Cell 1990, 60:45-52) may then be another example of H@IABP-mediated adhesion. Related to these in vitro aggregation phenomena, but of importance in vivo, would be the HA-mediated adhesion of macrophages to the peritoneal wall following antigen injection (Shannon et al, Immunol Cbmmun 1980, 9:357-370), the possible involvement of HA-mediated interactions in lymphocyte homing [9], and HA-mediated condensation of embryonic cells during differentiation as discussed below. During differentiation of mesodermal cells in the embryonic limb, HA-binding sites appear on the surface of these cells at the time of cellular condensation, just prior to onset of cartilage and muscle differentiation (Knudson and Toole, Dev Biol1987, 124:82-90). In vitro studies have demonstrated that HA-HABP interactions mediate aggregation of these condensation-stage mesodermal cells (Knudson, Development, in press). Recent studies by my colleagues and myself (Turner, Bane+ and
841
842
Cell-to-cell
contact
atid
extracellular
matrix
Toole, unpublished) and from the labomtory of C Knudson (personal communication) have shown that interference with HA binding to the surface of these mesodermal cells, via treatment with either HA oligosaccharides or monoclonal antibody to the HA receptor, blocks chondrogenesis. These experiments indicate that cell-surface binding of HA is crucial to differentiation of chondrocytes. We are in the process of characterizing the HABPs involved and these studies have revealed that they recognize HA hexasaccharides in similar manner to the 85 kD HA receptor. Another HA-binding proteoglycan, PG-M, may also be involved in mesodermal cell condensation (Kimata et al, J Bid C&em 1986, 261:13517-13525) but this protein requires HA decasaccharide for binding (Yamagata et al, J Biol C%em 1986, 261:13526-13535). Additional HABPs requiring decasaccharide for HA recognition are also involved in retaining newly synthesized proteoglycan-HA complexes in the pericellular matrix of differentiated chondrocytes [18]. Crossman and Mason [19] recently isolated a 102 kD HABP from rat chondrosarcoma that may mediate this retention but this has not yet been demonstrated directly.
Regulation of HA synthesis and cell-cell interactions
by growth
factors
HA synthesis is unusual in that it takes place at the inner face of the plasma membrane (Philipson and Schwartz, J Biol Gkm 1984, 2595017-5023; Prehm, Biocbem J 1984, 220:597&X). Nascent HA is apparently extruded through the membrane during synthesis and may be retamed at the cell surface via attachment to the synthetase complex. Nascent HA remains bound to the synthetase complex throughout synthesis [20] and then dissociates from the synthetase by an unknown mechanism [ 211. The molecular mechanism of regulation of synthesis of HA is unknown but many agents have profound effects on HA production. Several growth factors stimulate HA synthesis, for example see [22], but these effects often vary according to cell type, age, and condition of culture, for example see [23]. My colleagues and I have examined early embryonic limb for endogenous growth factors that might stimulate HA production during development and found that, at the earliest stages, when both cell proliferation and HA production are maximal, basic Iibroblast growth factor (bFGF) is a major factor present in the limb (Munaim et al., Proc Nat1 Acud Sci USA 1988,85:8091-8033). Basic FGF was shown to stimulate HA production and formation of HA-dependent pericellular matrices in the embryonic limb mesoderm [24] and thus appears to be a major regulatory factor in early limb development. We also found that, during differentiation, the limb ectoderm secretes another HA-stimulatory factor, possibly related to transforming growth factor (TGF)P [24]. We believe that this factor regulates differential events in the sub-ectodemral versus central regions of the limb.
Another developmental system in which HA production is regulated by cell-cell interactions is the mucification of cumulus cells that is caused by follicle-stimulating hormone (FSH) at ovulation. In this case, the oocyte produces a HA-stimulatoty factor that, in concert with FSH, interacts with the surrounding cumulus cells to induce IL4 synthesis and expansion of the cumulus-oocyte complex, thus providing an appropriate environment for sperm penetration and fertilization subsequent to ovulation [25]. Tumor cells also produce HA-stimulatory factors [26]. These factors interact with fibroblasts causing them to produce a HA-rich peritumoral extracellular matrix that may facilitate tumor cell invasion through normal tissues (reviewed in Biswas and Toole, In Cell Membranes, Volume 3 edited by Elson et al. Plenum, 1987, pp 341-363). Elevated levels of HA-stimulatory activity have also been detected in the sera of cancer patients but not in controls or patients in remission [ 271. The tumor-derived factors have not yet been fully characterized but preliminary results have ruled out major involvement of FGF or TGFP [ 26,271. Interestingly, however, chick embryonic limb ectodermal cells and human carcinoma cells both stimulate HA production in either chick limb mesodermal cells or adult human Iibroblasts (Knudson and Knudson, Cancer Lett, in press).
Conclusions
The past year has been a particularly exciting period in this field as it has brought together areas that were previously quite disparate. The homology shown between CD44, link protein and aggrecan has led to the discovery of the similarity between, and possible common identity of, CD44 and the 85 kD HA receptor. This in turn has consolidated previous claims that the latter molecule is indeed a receptor as well as revealing possible roles for HA-cell interactions in the immune system. In addition, this work has pointed to the existence of a family of proteins having homologous HA-binding domains, which we have named the hyaladherins because they are involved in binding of HA and because they mediate or modulate adherence between cells, between extracellular matrix macromolecules, and between cells and extracellular matrices. The following year should see significant advances in our understanding of the functions of CD44, the 85 kD HA receptor, and their relatives in cell behavior and embryonic development, the scope of and relationships within the hyaladherin family, and the regulation of their synthesis and action.
Acknowledgements
I wish to thank C Knudson, W Knudson and C Underhill for providing unpublished data.
Hyaluronan
Annotated reading 0 ma
references
and recommended
Of interest Of outstanding
interest
1.
Z~MMERMANN DR, Ruosvwn E: Multiple domains of the large libroblast proteoglycan, versican. EMBO J 1989, 8:297+2981. The HA-binding proteoglycan of libroblasts, versican, was sequenced and its several domains compared with other proteins. Homologies be tween the N-terminal, putative HA-binding domain of versican and those of several other HA-binding macromolecules are discussed. 2. a
PERIDES G, IANE WS, ANDRNVS D, DAHL D, BICNAMI A: Isolation and partial characterization of a glial hyaluronate-binding protein. J Biol C5em 1989, 264:5981-5987. An HABP was isolated from human brain white matter and partially sequenced. This protein Seems to be restricted to the central nervous system and tumors thereof. The sequences obtained show strong homolo. gies with the HA-binding domains of link protein and aggrecan. and are almost identical to sequences in versican as described by Zimmerman and Ruoslahti ( 11. I, MOT M, PEsANDO JM, SEED B: A lymphocyte molecule implicated in lymph node homing is a member of the cartilage link protein family. Cell 1989, 56:1057-1062. This and the following paper [4] give the sequences for the putative lymphocyte homing receptor, known most commonly as Hermes antigen, CD44 or Pgpl. These two publications are the first to describe homologies between this group of molecules and the HA-binding domains of cartilage link protein and aggrecan. STAMENKOVIC
4.
GOLDSTEIN
0
DING JF, BUTCHER
DPH, PICKER LJ. MN-N CN, BARGAIZE RF, EC: A human lymphocyte homing recep for, the Hermes antigen, is related to cartilage proteoglycan core and link proteins. Cell 1989, 56:1063-1072. This and the above paper [3] report the cloning and sequencing of cDNAs for the putative homing receptor, CD44, and show homology with the HA-binding domains of link protein and the proteoglycan, agg-.
IA, 2.~0~
M, SUZUKI S, KIMATA K: Hyaluronic acid associated with the surfaces of cultured fibroblasts is linked to a serum-derived 85-kDa protein. J Biol them 1990, 265:5247-5257. An 85kD HABP was purified from fibroblasts but this protein could not be metabolically labeled and was also found to be present in sera The authors present evidence suggesting that this protein can mediate attachment of HA to the surface of the fibroblasts but indicating distinc tions between the properties of this protein and the 85 kD HA receptor identified by Underhill ef al (1987).
5. 0
YONEDA
6. 0
TUR&Y EA: Hyaluronic acid stimulates protein k&se activity in intact cells and in an isolated protein complex. J Biol &em 1989, 260951-8955. This study demonstrates phosphorylation of tyrosine and serine/thre onine residues in tibroblast proteins in response to treatment of the cells with small amounts of HA fn addition, a complex was isolated from the fibroblasts that contains tightly associated HABP and HA&mulated protein kinase. 7. *
TUR~EV EA, BRA.%EL P, MOORE D: A hyaluronan-binding protein shows a partial and temporally regulated codistribution with actin on locomoting chick heart fibroblasts. ,Qp Cell Res 1990, 187:243249. Co-distribution of HABP and actin filaments in fibroblast cellular processes was shown by double immunolabelUng techniques. Promotion of the association of HABP with T&on-insoluble cytoskeleton preparations by addition of HA to the cells was also demonstrated. 8.
CT, RAJA RH, WEIGEL PH: Endocytosis of hyahtronic acid by rat liver endothelial cells. Evidence for receptor recycling. Biocbem J 1989, 257:87-. Detailed kinetics and distribution data are given for the internalization of HA and cycling of the HA-binding sites that mediate HA endocytosis. The results are consistent with receptor-mediated endocytosis of HA by the coated-pit pathway. l
MCGARY
ARUFFO
4
proteins,
STAMENKOVIC
I, MEINICK
the hyaladherins M, UNDEIU-IU
Toole
CB. SEED B:
CD44 is the principal cell surface receptor for hyafuronate. Cell 1990, 61:13031313. Evidence is given suggesting that CD44 and the 85 kD HA receptor of BHK cells are similar or identical. Soluble CD44 fusion proteins were shown to bind to various tissues, including high endothelial vent&s, in a HA-dependent manner. Ttansfection of CD44 cDNA into COS cells induced ceU surface reactivity with K3 antibody to the 85 kD HA recep tar. 0
l
3. 0
9.
and its binding
10. 0
LESlEv J, ScHULTE
R, HYMAN R Binding of hyaluronic acid to ceU lines is inhibited by monoclonal antibodies a@nst Pgp-I. .EQ Cell Res 1990, 187:224-233. Several lines of evidence were presented suggesting that Pgpl, which is closely related or identical to CD44, mediates HA-binding to some lymphoid ceU lines. Antibodies to Pgp-1 were shown to inhibit HA biding to the cell lines, attachment of the cells to HA and HA-mediated ceU aggregation. However, many Pgpl positive ceU lines and normal cells did not exhibit HA-binding or HA-mediated aggregation. lymphoid
11. 0
ALHO AM, UNDERHIU CB: The hyaluronate receptor is preferentially expressed on proliferating epithelial cells. J G-11 Babl 1989, 108:1557-1565. The 85 kD HA receptor was found to be present around epithelial cells in a tide variety of tissues. In those sites where both proliferating and non-proliferating cells were present, the receptor appeared to be restricted to the former, e.g. in the basal layer of epidermis and the base of intestinal crypts. HA was colocaUzed with HABP in most, but not all, cases. 12.
TAMM~ R, RI~EUINO JA, MARGOUS RU, MAIBACH HI, TAMM~ M: Hyahtronate accumulation in human epidermis treated with retinoic acid in skin organ culture. J invest Dermutol 1989, 92:326-332. These authors have also demonstrated HA around epidermal cells and here show increased accumulation of HA in epidermis after treatment of skin cultures with retinoic acid. However, the HA was found to accumulate in the differentiating as weU as basal layers of the epidermis. l
WEST DC, KUMAR S: The effect of hyahtronate and oligosaccharides on endothelial cell proliferation monolayer integrity. E&O Cell Res 1989, 183:1791%. These investigators previously showed that l-L4 oligosaccharides, not H.4 polymer, stimulate angiogenesis. fn this paper they show oligosaccharides composed of 3-16 disaccharide repears stimulate dothelial ceU proliferation whereas HA polymer is inhibitory. They demonstrate binding and uptake of HA by the endothelial cells. 13.
l
14. 0
its and but that enalso
E, AUEWPERG N: A hyahuonate binding protein transiently cod&tributes with p21k.m in cuhured cell lines. Eq Cell Res 1989, 182340-348. This study demonstrates transient co-localization of HABP and p21 in the cell processes and ruffles of motile cells. However, whereas p21 remains in tuflles as motility ceases, HABP relocates to the ceU body. Tuw
15. 0
GREY AM, SCHOR AM, RUSHTON G, Ews 1, SCHOR SL Purilication of the migration stimulating factor produced by fetal and breast cancer patient fibroblasts. Proc Nat1 Acud Sci L’U 1989, 86:243%2442. This group has shown previously that con&rent fetal or cancer patient fibroblasts invade collagen gels to a greater extent than normal adult fibroblasts. In the work described here they have purilied a 70 kD protein from fetal and cancer patient fibroblasts that induces enhanced invasive behavior in adult libroblasts. 16. 0
CHEN
WYJ, GRANI’ ME, SCHOR AM, ~CHOR SL Dierences between adult and foetal fibroblasts in the regulation of hyaluronate synthesis: correlation with migratory activity. J cell Sci 1989, 943577-584. The same group as in [ 151 has shown that a major difference between adult and fetal fibroblasrs is that, when confluent, the latter produce higher levels of HA of greater average molecular weight than the former.
17.
SCHOR Sl+ SCHOR AM, ME, ELIIS 1: Mechanism
l
ing factor produced effect on hyaluronic 1989, 25~737-746.
CHEN J, of action of the by fetal and cancer acid synthesis. In GREY AM,
RUSHTON G, Grt,un’ migration stimulatpatient fibroblasts: Vim Ceif Dev Bid
843
844
Cell-to-cell
contact
and extracellular
matrix
The work presented in [ 15,161 is extended in this paper to show that the migration-stimulating activity of the 70 kD factor is dependent on HA production, both with respect to the enhancement of in\-asive behavior of adult fibroblasts on addition of the factor and the intrinsic difference between fetal or cancer patient fibroblasts and adult fibroblasts. 18. 0
MCCARTHY MT, Tools BP: Membrane-associated hyaluronatebinding activity of chondrosarcoma chondrocytes. J Cell P&o/ 1989, 141:191-202. This study demonstrates that cell-surface HABP retains HAproteogtycan complexes in the pericellular matrix and that the HABPfs) is not link protein or aggrecan. 19. 0
CROSSMAN MV. Mao~ RM: Purification and characterization of a hyaluronan-binding protein from rat chondrosarcoma Biccbem J 1990, 266399-406. A 102kD HABP was purified from extracts of Swarm rat chondrosarcoma and shown to react with HA-proteoglycan (aggrecan) complexes. The protein has some properties similar to aggrecan but was shown to be a distinct protein by peptide mapping.
20.
NG RF, SCHWARTZ NB: Solubilization and partial puritication of hyaluronate synthetase from oligodendroglioma cells. J Biol Cbem 1989, 264:11n611783. The molecular characteristics of hyluronate synthetase have been very difficult to elucidate because of problems in obtaining active prepara tions in solution. Although solubilization had been achieved for bacterial synthetase, this is the first report for eukaryotic synthetase. The solubilized enzyme was shown to have similar properties to membrane. bound enzyme, to retain nascent HA, and to produce polymeric HA 0
nan in human lung but not in skin tibroblasts. ,?$I Cell Res 1‘990. 186:192-195. TGFg %L< shown to stimulate HA synthesis three to fourfold in lungderived libroblasts but had no effect on skin fibroblasts. Proteo&can .synthesis was stimulated in both cell types This study extends the growing literature showing heterogeneity in fibroblasts from different sources. 24. 0
TOOLE BP, MUIWM SI, WEUIS S. KNUII~ON CB: Hyaluronatecell interactions and growth factor regulation of hyaluronate synthesis during limb development. 0% Fotorrtuhtion $,alrp 1989, 143:138149. This article reviews work on changes in HA .synthesis. HA-binding sites and organization of pencellular HA during differentiation of limb mesoderm. It al.so provides data on the role of HAstimulatory factors, especially bFGF which is enriched in the developing limb prior to differentiation and an ectodemlaltyderived factor that appears to sustain high levels of HA in the subectodermal mesoderm. The action of the latter factor is blocked by antibody to TGFP.
25.
SALLISIRJ 4 YANAGISHITA M. HAXAU. VC: Mouse oocytes regulate hyaluronic acid synthesis and mucilication by FSHstimulated cumulus cells. Dee, Biol 1990. 138:2632. Prior to ovulation, in response to increased FSH levels, the granulosa cells surrounding the oocyte (cumulus cells) change from a compact organization to a separated or ‘expanded’ configuration as a result of production of a HA-rich pencellular matrix. This report shows that the oocyte produces and secretes an HAstimulatory factor that, in concert with FSH, acts on the cumulus cells. 0
21. PREHM P: Release of hyaluronate from eukaryotic cells. 0 Biccbem J 1990, 267:185-189. This paper examines the mechanism of release of newiy synthesized HA from the cell surface. It was shown that, for the B6 hybrid carcinomaembryo fibroblast fine, HA is released as an intact polymer of the same size range as cell surface HA; in the case of SV3T3 cells, partial degra dation due to radical action also occurred.
W. BISWA-~ C. LI XQ, NEMEC RE. TOOIJI BP: The regulation of tumour-associated hyaluronan. Ciba Fowdafion $wnlp 1989, 143:15&169. The potential role of HA and HABPs in tumor cell invasion is reviewed. In particular, the nature of tumor~derived factors that stimulate libroblast HA production is discussed.
22.
27.
HEUXN P, LAURENT TC, HE~DIN CH: Effect of growth factars on hyaluronan synthesis in cultured human libroblasts. Biocbem J 1989, 258:919922. Platelet-derived growth factor, epidermal growth factor, bFGF and TGF8 were shown to stimulate HA synthesis in human fibroblasts. The level of stimulation of synthesis did not correlate with mitogenic activity of the factors. a
23. 0
WESTERGREN-THORSSON MALhLvROM A TGF-8
G, SARNSTRAND B, enhances the production
FRANSSON LA. of hyaluro-
26. l
KNUTSON role and
DECKHI M, CHILI ES, DOUHAL~I C. ~MOIIN A. HAU J. SPENDLO\‘E R, b3NGAKU(ER MT, STIXN R: Hyaluronic acid-stimulating activity in sera from the bovine fetus and from breast cancer patients. Gztrcer Ra 1989. 49:349+3505. This study documents the presence of an HA-stimulatory gtycoprotein in bovine fetal sent that decreases in amount with age of the fetus. Similar HA-stimulatory activiv is also present in elevated amounts in the set-a of cancer patients compared with that of normal controls and cancer patients in remission. 0
and its binding
proteins,
the hyaladherins
B.P. Toole Department
of Anatomy
and Cellular Biology, Tufts University Boston, Massachusetts, USA
Health
Science
Schools,
Current Opinion in Cell Biology 1990, 2:839-844
Introduction
Hyaluronan (HA) is a high-molecular-weight, highly anionic polysacchande found in extracellular matrices and at ceU surfaces. It is composed of 200-10000 disaccharides of P-1,4-glucuronate-P-1,3-N-acetylglucosaminegiving a molecular weight of between 1 x 105 and 5 x 106, usually at the higher end of this range. No branching of the polymer or variations in composition of the repeating disaccharide have been demonstrated convincingly. Despite the apparent monotony of this polymer, its unique physicochemical properties and its consequent role in tissue structure and function have fascinated polymer biophysicists and physiologists for decades. These properties include its voluminous structure in solution, its ability to form continuous networks at low concentration, and its consequent viscoelasticity, osmotic properties, and effects on molecular exclusion, diffusion, flow resistance, etc. (reviewed in Comper and Laurent, Pbysiol Rev 1978, 58:255-315). In addition to its long-established physicochemical function, HA is important in regulating ceU behavior. HA is enriched in extracellular matrices in which cells migrate or proliferate (reviewed by Toole, In Cell Biology of the Extracellular Matrix edited by Hay E. Plenum, 1981, pp 259-294), and recent evidence supports a role for HA in these cellular processes. In addition, HA has been shown to ineuence cell interactions and differentiation in some cellular systems. Underlying its role in ceU behavior are its unique physicochemical properties and its interactions with hyaluronan-binding proteins (HABPs), both in exttaceUu& matrices and on cell surfaces. Recent progress in this field has emphasized the molecular nature and role of I-L4BPs. Until this year, it was thought that there were at least two classes of HABPs: structural HABPs such as HAbinding proteoglycans and link proteins; and cell surfaceassociated HABPs that have properties suggesting that they are HA receptors. It has now become apparent that these two groups of HABPs have homologous I-IA-binding domains and belong to a single family of proteins; I suggest that this family be termed the hyaladherins. In this review I will discuss the recent work demonstrating the inter-relationships between the various I-L4BPs,
the importance of HA interaction with ceU surface-associated HABPs in ceU behaviour, and some preliminary, but promising, work on the potential importance of HA stimulatory factors in regulating these events. Several interesting aspects of current research on the structural properties and metabolism of HA, and its role in ceU behavior and pathology have been discussed in recent Ciba Foundation Symposium ( 7%eBiology ofHyaluronan edited by Evered D and Whelan J. Ciba Found Sjmp, 1989).
Hyaluronan-binding
proteins
The most thoroughly studied HA-binding macromolecules are the link proteins and large proteoglycan, termed aggrecan, of cartilage (reviewed in Gallagher, Cr.07 Opin Cell Bioll989,
1:1201-1218). Goetinck et al. (I Cell Biol
1987,105:2403-2408) have provided convincing evidence that the tandemly repeated domains IIa and IIb (B and B’ loops) at the carboxy terminus of cartilage link protein contain the HA-binding region. Four different peptides, derived from two corresponding regions of each of the tandem repeats, were found to block binding of HA to link protein, implying that all four regions were involved in the binding. Each of these peptides contains a cluster of positively charged amino acids and binding decreases with increase in ionic strength, suggesting that binding is ionic in nature. The amino-terminal region of the core protein of aggrecan contains a domain homologous to the tandem repeats in link protein and thus is assumed to mediate HA binding to aggrecan (Neame et al, J Biol Chem 1987, 262:1776%17778). Last year, similar homologies were described in several other macromolecules: near the amino-terminus of the HA-binding proteoglycan, versican, which is produced by human iibroblasts [l] ; in an HABP produced by human glial cells [2] that is related to the HABP, hyaluronectin (Delpech and Halavent, J Neurocbem 1981, 36:855-859), and to versican [ 11; and at the N-terminus of the adhesive protein(s) variously known as CD44, Pgpl, Hermes antigen, ECMRIII coUagen receptor, HCAIvl, etc.-here termed CD44 [3,4]. The homologies between these proteins are further discussed by Zimmerman and Ruoslahti [ 11.
Abbreviations bFG&basic HA-hyaluronan
fibroblast (hyaluronate,
growth factor; BHK-baby hamster hyaluronic acid); HABP-hyaluronan-binding
@ Current
Biology
kidney;
Ltd ISSN 0955*74
FSH-follicle-stimulating protein; TCF-transforming
hormone; growth
factor.
839
840
Cell-to-cell
contact
and extracellular
matrix
It is generally accepted that the link proteins and aggrecan of cartilage are structural macromolecules that, through interaction with HA, form gigantic ternary complexes that contribute to the physical properties of cartilage. Similar structural complexes of HA, link protein and proteoglycans such as versican presumably participate in building the matrices of several other connective tissues. The structure and distribution of these complexes were reviewed by Gallagher (1989). However, there is a separate group of HABPs that apparently serve as cell-surface receptors for HA and mediate the effects of HA on cell behavior that are described below. Two putative cellsurface receptors of this type have been studied extensively. The first of these was originally described as HAbinding sites of high afhnity on the surface of SV-3T3 cells (Underhill and Toole, J Cell Biol 1979, 82:475-484; Underhill and Toole, J Biol C&em 1980,255:4544-4549) and has now been identified in baby hamster kidney (BHK) cells as an 85 kD glycoprotein recognized by the K.3 monoclonal antibody (Underhill et al, J Biol Chem 1987, 262:13142-13146; herein termed the 85 kD HA receptor). Another 85kD HABP present at the surface of cultured libroblasts is derived from serum [5] but this protein has properties distinct from the BHK receptor. The second putative receptor was iirst isolated from the culture medium of 3T3 cells and chick embryo libroblasts as a mixture of 56-70 kD proteins (Turley et al, Biocbemisttty 1987, 26:2997-3005) but has been shown subsequently to form large aggregates at the cell surface [ 6). Both these putative HA receptors interact directly or indirectly with the cytoskeleton (Iacy and Underhill, J Cell Biol 1987, 105:1395-1404) [7] and mediate various effects of HA on cell behavior (see below) but their biochemical inter-relationship is not yet known. Recently, my colleagues and I have obtained monoclonal antibodies to two classes of chick embryo HABPs that are widely distributed on the surfaces of embryonic and transformed or tumor cells (Banerjee and Toole, unpublished). One group of antibodies recognizes proteins of molecular weights of 93, 90, and 69kD, respectively, that may be related to the 85kD HA receptor. The other group of antibodies ret ognizes a protein of molecular weight of approximately 300kD; the relationship of this protein to other HA receptors is not yet known. A cell-surface HABP is also involved in uptake of HA from the circulation. Clearance of HA takes place mainly in lymph nodes (Fraser etal, Biocbem J1988,256:153158) and in the liver where the site of uptake is the endothelium (Fraser et al, Cell TissRes 1985,242:505-510). Binding, internalization and degradation of HA by liver endothelial cells has been demonstrated in culture (Smedsrod et al, Biocbem J 1984, 223~617626; Laurent et al., Biocbem J 1986, 234:653658). The HA-binding sites in volved have now been shown to be recycled during endocytosis of I-IA and it appears that 75% of these sites are internal during cycling [8]. Internalization and degradation of HA by other vascular endothelium has also been demonstrated (McGuire et al., J Cell Physoll987, 133:267-276). The HABP that mediates endocytosis has not yet been identihed but it is probably different from
the 85kD HA receptor (Raja et al., J Biol Chem 1988, 263:16661-16668) An important development is the recent iinding that the adhesive protein CD44 contains a domain that has approximately 30% homology with the HA-binding regions of link protein and aggrecan [3,4]. Significantly, however, the basic residues previously implicated in HA binding to link protein are not conserved in CD44. CD44 is very similar or identical to the Hermes antigen, Pgp-1 and ECMRUI (Picker et al, J Cell Bioll989, 109:927-937; Gallatin et al., Proc Natl Acud Sci c/sA 1989, 86:4654-4658; Wolffe et al., J Biol Chem 1990, 265:341-347). These molecules are involved in T-lymphocyte activation and adhesion, and homing of circulating lymphocytes via interaction with high endothelial venules, the entry site to organized lymphoid tissues. Although mainly studied in the hematopoietic system, CD44 and its homologues are widely distributed (Picker et al, 1989; Carter and Wayner, J Biol Chem 1988, 263:4193-4201). They also exist in varied molecular forms: in hematopoietic cells, CD44 is synthesized as a 37 kD protein that is processed to form a glycoprotein with a molecular weight of 80-95 kD [3,4] or a minor chondroitin sulfate proteoglycan form with a molecular weight 180-200 kD &lkanen et al, J Jmmunoll988, 141:1615-1623) [3]. Epithelial and carcinoma cells produce yet another form, of approximately 160 kD molecular weight (Picker et al., 1989) [ 31. CD44 presumably mediates adhesion of cells to extracellular matrices because it exhibits binding to collagen (Carter and Wayner, 1988) as well as containing the putative HA-binding domain. Interestingly, two other HABPs, link protein (Chandrasekhar et al., J Biol cbem 1983,258:622&6231) and aggrecan (Toole and Iowther, Biocbem J 1968, 109:857-866), also bind to collagen. Evidence published this year indicates that CD44 is in fact the same as, or very similar to, the 85kD HA receptor of BHK cells. Aruffo et al. [9] transfected hamster CD44 cDN& into COS cells and demonstrated expression of cell-surface reactivity with K3 antibody to the 85 kD HA receptor. They also showed that hyaluronidase treatment of various tissues, or competition with HA, blocks binding of soluble forms of 044. M Culty, K Miyake, P Kincade, E Sikorski, E Butcher and C Underhill (personal communication) have obtained additional evidence for identity of CD44 and the 85kD HA receptor; this evidence includes immunoprecipitation of HA-binding activity from cell extracts with antibodies to CD44, and inhibition of HA binding and HA-mediated cell aggregation with these antibodies. Supporting these lindings is the work of Lesley et al. [ 101 who showed that antibodies to CD44 block binding of HA to lymphocytes and HA-mediated lymphocyte aggregation. It is not yet clear, however, if the HA-binding function of CD44 is involved in the homing of lymphocytes to high endothelial venules. Although binding of soluble CD44-immunoglobulin fusion proteins to high endothelial cells is blocked by treatment of the cells with hyaluronidase or by competition with HA (91, binding of intact lymphocytes is not (C. Underhill, personal communication).
Hyaluronan
Because CD44 and the 85 kD HA receptor are very similar, if not identical, the latter must also share homologies with the HA-binding domains of the structural HABPs: aggrecan, versican and link protein. However, other data distinguish the HA-binding properties of these structural molecules and the HA receptor (see Toole et al, in 7&e Role of Extracellular Matrix in Development edited by Tretstad R. Alan I&s, 1984 pp 43-66). The first and most marked distinction is in the length of HA oligosaccharide required for recognition by the HABPs. Link protein, aggrecan, hyaluronectin, and the proteoglycan, PG-M, of chick embryo fibroblasts require a HA decasaccharide for binding. Oligosaccharides of smaller lengths exhibit negligible binding, and the affinities of binding of HA decasaccharide and HA polymer do not dilfer greatly. However, cell-surface HA receptors recognize HA hexasaccharide and their afftnity of binding increases dramatically with the length of the ligand. Second, the HA receptor is not as highly speciiic for HA as link protein and the proteoglycans. Third, increases in ionic strength enhance the affiity of interaction of HA with the HA receptor but decrease a&&y for link protein and aggrecan. Presumably, sequence and conformational differences in the HA-binding regions of the structural HABPs and CD44/HA receptor allow for these differences in ligand recognition. Despite the differences in binding behavior it is clear that these macromolecules make up a family that share related HA-binding motifs. Therefore, they deserve a family name: the hyaladherins.
Hyaluronan
and cell behavior
HA has been shown to influence several types of ceU behavior, but its effects depend on at least three parameters: the size and concentration of HA and the ceU type in question (discussed in Goldberg and Toole, A&r Rbeum
1987, 30:76%778). A positive correlation has been demonstrated between HA synthesis, HA synthetase activity and ceU proliferation (Brecht et al, Biocbem J 1986, 239445-450;). Inhibition of HA synthesis leads to arrest of cells in mitosis prior to rounding, possibly because HA-mediated detachment from the substratum is required (Brecht et al, 1986). Recently, the 85 kD HA receptor and, in some cases, HA itself have been localized to the surface of a variety of proliferating epithelial cells [ 111; however, other work suggests that HA accumulation is not restricted to the proliferating zone of the epidermis [12]. Also, endothelial cell proliferation is stimulated by HA-derived oligosaccharides but is inhibited by polymeric HA [ 131. Clearly, much further work is required to establish the precise relationship of HA to proliferation. Many studies have suggested a role for HA in cell rngration, both in creating hydrated pathways that facilitate penetration of tissues (Toole, 1981) and in direct influences on ceU locomotion (Turkey, Ciba Found 5” 1989, 143121-137). Turkey and colleagues (I Cell Sci 1985, 78:133145) [14] have shown that addition of an
and its binding
proteins,
the hyaladherins
Toole
HABP plus HA to fibroblasts stimulates ceU movement and that this HABP is concentrated in the rt&ling lamellae of actively locomoting cells. The endogenous ceU surface HABP of these libroblasts is associated with protein kinase activity that is stimulated by addition of HA to the cells or to the isolated HABP complex, pointing to the possibility that instructive signal transduction results from interaction of HA with ceU surface HABPs [6]. Grey et al. [ 151 have recently purilied a 70 kD ceU migration-stimulating factor from fetal libroblasts and fibroblasts derived from cancer patients. This factor stimulates confluent cells to penetrate collagen gels by an HA-dependent mechanism. The ability of confluent cells to penetrate collagen gels was shown to correlate with the level of HA synthesis and size of HA produced at conlluence [ 161. Factor-induced stimulation of confluent adult fibroblasts to invade collagen gels was neutralized by treating the cells with HA-specilic hyaluronidase, as was the spontaneously elevated ability of fetal and cancer patient fibroblasts to invade the collagen gels [ 171. Possibly the best-studied iniluence of HA is in modulation of cell aggregation. Addition of low concentrations of HA to lymphoma cells or macrophages induces their aggregation. On the other hand, the divalent cation-independent aggregation of several transformed ceU lines is inhibited by treatment with hyaluronidases or high concentrations of HA (Wright et al, Cancer Res 1981,41:5107-5113; Underhill and Toole, Eap Cell Res 1981, 131:419-423). Underhill and Toole (1981) demonstrated that these aggregation phenomena are dependent on crossbridging by HA of HA receptor sites on adjacent cells; inhibition of crossbridging at high HA concentrations is due to saturation of these receptors. It has been shown since that HA-mediated aggregation of virally transformed ceU lines and macrophages is inhibited by addition of K3 antibody to the 85kD HA receptor (Green et al, Eqi~ Cell Res 1988, 178:224-232) and that HA-mediated lymphocyte aggregation is inhibited by antibodies to CD44 [ 101, a molecule now known to be related to the HA receptor [9]. The very recent observation that transfection of libroblasts with CD44 cDNA induces their ability to selfaggregate (St John et al, Cell 1990, 60:45-52) may then be another example of H@IABP-mediated adhesion. Related to these in vitro aggregation phenomena, but of importance in vivo, would be the HA-mediated adhesion of macrophages to the peritoneal wall following antigen injection (Shannon et al, Immunol Cbmmun 1980, 9:357-370), the possible involvement of HA-mediated interactions in lymphocyte homing [9], and HA-mediated condensation of embryonic cells during differentiation as discussed below. During differentiation of mesodermal cells in the embryonic limb, HA-binding sites appear on the surface of these cells at the time of cellular condensation, just prior to onset of cartilage and muscle differentiation (Knudson and Toole, Dev Biol1987, 124:82-90). In vitro studies have demonstrated that HA-HABP interactions mediate aggregation of these condensation-stage mesodermal cells (Knudson, Development, in press). Recent studies by my colleagues and myself (Turner, Bane+ and
841
842
Cell-to-cell
contact
atid
extracellular
matrix
Toole, unpublished) and from the labomtory of C Knudson (personal communication) have shown that interference with HA binding to the surface of these mesodermal cells, via treatment with either HA oligosaccharides or monoclonal antibody to the HA receptor, blocks chondrogenesis. These experiments indicate that cell-surface binding of HA is crucial to differentiation of chondrocytes. We are in the process of characterizing the HABPs involved and these studies have revealed that they recognize HA hexasaccharides in similar manner to the 85 kD HA receptor. Another HA-binding proteoglycan, PG-M, may also be involved in mesodermal cell condensation (Kimata et al, J Bid C&em 1986, 261:13517-13525) but this protein requires HA decasaccharide for binding (Yamagata et al, J Biol C%em 1986, 261:13526-13535). Additional HABPs requiring decasaccharide for HA recognition are also involved in retaining newly synthesized proteoglycan-HA complexes in the pericellular matrix of differentiated chondrocytes [18]. Crossman and Mason [19] recently isolated a 102 kD HABP from rat chondrosarcoma that may mediate this retention but this has not yet been demonstrated directly.
Regulation of HA synthesis and cell-cell interactions
by growth
factors
HA synthesis is unusual in that it takes place at the inner face of the plasma membrane (Philipson and Schwartz, J Biol Gkm 1984, 2595017-5023; Prehm, Biocbem J 1984, 220:597&X). Nascent HA is apparently extruded through the membrane during synthesis and may be retamed at the cell surface via attachment to the synthetase complex. Nascent HA remains bound to the synthetase complex throughout synthesis [20] and then dissociates from the synthetase by an unknown mechanism [ 211. The molecular mechanism of regulation of synthesis of HA is unknown but many agents have profound effects on HA production. Several growth factors stimulate HA synthesis, for example see [22], but these effects often vary according to cell type, age, and condition of culture, for example see [23]. My colleagues and I have examined early embryonic limb for endogenous growth factors that might stimulate HA production during development and found that, at the earliest stages, when both cell proliferation and HA production are maximal, basic Iibroblast growth factor (bFGF) is a major factor present in the limb (Munaim et al., Proc Nat1 Acud Sci USA 1988,85:8091-8033). Basic FGF was shown to stimulate HA production and formation of HA-dependent pericellular matrices in the embryonic limb mesoderm [24] and thus appears to be a major regulatory factor in early limb development. We also found that, during differentiation, the limb ectoderm secretes another HA-stimulatory factor, possibly related to transforming growth factor (TGF)P [24]. We believe that this factor regulates differential events in the sub-ectodemral versus central regions of the limb.
Another developmental system in which HA production is regulated by cell-cell interactions is the mucification of cumulus cells that is caused by follicle-stimulating hormone (FSH) at ovulation. In this case, the oocyte produces a HA-stimulatoty factor that, in concert with FSH, interacts with the surrounding cumulus cells to induce IL4 synthesis and expansion of the cumulus-oocyte complex, thus providing an appropriate environment for sperm penetration and fertilization subsequent to ovulation [25]. Tumor cells also produce HA-stimulatory factors [26]. These factors interact with fibroblasts causing them to produce a HA-rich peritumoral extracellular matrix that may facilitate tumor cell invasion through normal tissues (reviewed in Biswas and Toole, In Cell Membranes, Volume 3 edited by Elson et al. Plenum, 1987, pp 341-363). Elevated levels of HA-stimulatory activity have also been detected in the sera of cancer patients but not in controls or patients in remission [ 271. The tumor-derived factors have not yet been fully characterized but preliminary results have ruled out major involvement of FGF or TGFP [ 26,271. Interestingly, however, chick embryonic limb ectodermal cells and human carcinoma cells both stimulate HA production in either chick limb mesodermal cells or adult human Iibroblasts (Knudson and Knudson, Cancer Lett, in press).
Conclusions
The past year has been a particularly exciting period in this field as it has brought together areas that were previously quite disparate. The homology shown between CD44, link protein and aggrecan has led to the discovery of the similarity between, and possible common identity of, CD44 and the 85 kD HA receptor. This in turn has consolidated previous claims that the latter molecule is indeed a receptor as well as revealing possible roles for HA-cell interactions in the immune system. In addition, this work has pointed to the existence of a family of proteins having homologous HA-binding domains, which we have named the hyaladherins because they are involved in binding of HA and because they mediate or modulate adherence between cells, between extracellular matrix macromolecules, and between cells and extracellular matrices. The following year should see significant advances in our understanding of the functions of CD44, the 85 kD HA receptor, and their relatives in cell behavior and embryonic development, the scope of and relationships within the hyaladherin family, and the regulation of their synthesis and action.
Acknowledgements
I wish to thank C Knudson, W Knudson and C Underhill for providing unpublished data.
Hyaluronan
Annotated reading 0 ma
references
and recommended
Of interest Of outstanding
interest
1.
Z~MMERMANN DR, Ruosvwn E: Multiple domains of the large libroblast proteoglycan, versican. EMBO J 1989, 8:297+2981. The HA-binding proteoglycan of libroblasts, versican, was sequenced and its several domains compared with other proteins. Homologies be tween the N-terminal, putative HA-binding domain of versican and those of several other HA-binding macromolecules are discussed. 2. a
PERIDES G, IANE WS, ANDRNVS D, DAHL D, BICNAMI A: Isolation and partial characterization of a glial hyaluronate-binding protein. J Biol C5em 1989, 264:5981-5987. An HABP was isolated from human brain white matter and partially sequenced. This protein Seems to be restricted to the central nervous system and tumors thereof. The sequences obtained show strong homolo. gies with the HA-binding domains of link protein and aggrecan. and are almost identical to sequences in versican as described by Zimmerman and Ruoslahti ( 11. I, MOT M, PEsANDO JM, SEED B: A lymphocyte molecule implicated in lymph node homing is a member of the cartilage link protein family. Cell 1989, 56:1057-1062. This and the following paper [4] give the sequences for the putative lymphocyte homing receptor, known most commonly as Hermes antigen, CD44 or Pgpl. These two publications are the first to describe homologies between this group of molecules and the HA-binding domains of cartilage link protein and aggrecan. STAMENKOVIC
4.
GOLDSTEIN
0
DING JF, BUTCHER
DPH, PICKER LJ. MN-N CN, BARGAIZE RF, EC: A human lymphocyte homing recep for, the Hermes antigen, is related to cartilage proteoglycan core and link proteins. Cell 1989, 56:1063-1072. This and the above paper [3] report the cloning and sequencing of cDNAs for the putative homing receptor, CD44, and show homology with the HA-binding domains of link protein and the proteoglycan, agg-.
IA, 2.~0~
M, SUZUKI S, KIMATA K: Hyaluronic acid associated with the surfaces of cultured fibroblasts is linked to a serum-derived 85-kDa protein. J Biol them 1990, 265:5247-5257. An 85kD HABP was purified from fibroblasts but this protein could not be metabolically labeled and was also found to be present in sera The authors present evidence suggesting that this protein can mediate attachment of HA to the surface of the fibroblasts but indicating distinc tions between the properties of this protein and the 85 kD HA receptor identified by Underhill ef al (1987).
5. 0
YONEDA
6. 0
TUR&Y EA: Hyaluronic acid stimulates protein k&se activity in intact cells and in an isolated protein complex. J Biol &em 1989, 260951-8955. This study demonstrates phosphorylation of tyrosine and serine/thre onine residues in tibroblast proteins in response to treatment of the cells with small amounts of HA fn addition, a complex was isolated from the fibroblasts that contains tightly associated HABP and HA&mulated protein kinase. 7. *
TUR~EV EA, BRA.%EL P, MOORE D: A hyaluronan-binding protein shows a partial and temporally regulated codistribution with actin on locomoting chick heart fibroblasts. ,Qp Cell Res 1990, 187:243249. Co-distribution of HABP and actin filaments in fibroblast cellular processes was shown by double immunolabelUng techniques. Promotion of the association of HABP with T&on-insoluble cytoskeleton preparations by addition of HA to the cells was also demonstrated. 8.
CT, RAJA RH, WEIGEL PH: Endocytosis of hyahtronic acid by rat liver endothelial cells. Evidence for receptor recycling. Biocbem J 1989, 257:87-. Detailed kinetics and distribution data are given for the internalization of HA and cycling of the HA-binding sites that mediate HA endocytosis. The results are consistent with receptor-mediated endocytosis of HA by the coated-pit pathway. l
MCGARY
ARUFFO
4
proteins,
STAMENKOVIC
I, MEINICK
the hyaladherins M, UNDEIU-IU
Toole
CB. SEED B:
CD44 is the principal cell surface receptor for hyafuronate. Cell 1990, 61:13031313. Evidence is given suggesting that CD44 and the 85 kD HA receptor of BHK cells are similar or identical. Soluble CD44 fusion proteins were shown to bind to various tissues, including high endothelial vent&s, in a HA-dependent manner. Ttansfection of CD44 cDNA into COS cells induced ceU surface reactivity with K3 antibody to the 85 kD HA recep tar. 0
l
3. 0
9.
and its binding
10. 0
LESlEv J, ScHULTE
R, HYMAN R Binding of hyaluronic acid to ceU lines is inhibited by monoclonal antibodies a@nst Pgp-I. .EQ Cell Res 1990, 187:224-233. Several lines of evidence were presented suggesting that Pgpl, which is closely related or identical to CD44, mediates HA-binding to some lymphoid ceU lines. Antibodies to Pgp-1 were shown to inhibit HA biding to the cell lines, attachment of the cells to HA and HA-mediated ceU aggregation. However, many Pgpl positive ceU lines and normal cells did not exhibit HA-binding or HA-mediated aggregation. lymphoid
11. 0
ALHO AM, UNDERHIU CB: The hyaluronate receptor is preferentially expressed on proliferating epithelial cells. J G-11 Babl 1989, 108:1557-1565. The 85 kD HA receptor was found to be present around epithelial cells in a tide variety of tissues. In those sites where both proliferating and non-proliferating cells were present, the receptor appeared to be restricted to the former, e.g. in the basal layer of epidermis and the base of intestinal crypts. HA was colocaUzed with HABP in most, but not all, cases. 12.
TAMM~ R, RI~EUINO JA, MARGOUS RU, MAIBACH HI, TAMM~ M: Hyahtronate accumulation in human epidermis treated with retinoic acid in skin organ culture. J invest Dermutol 1989, 92:326-332. These authors have also demonstrated HA around epidermal cells and here show increased accumulation of HA in epidermis after treatment of skin cultures with retinoic acid. However, the HA was found to accumulate in the differentiating as weU as basal layers of the epidermis. l
WEST DC, KUMAR S: The effect of hyahtronate and oligosaccharides on endothelial cell proliferation monolayer integrity. E&O Cell Res 1989, 183:1791%. These investigators previously showed that l-L4 oligosaccharides, not H.4 polymer, stimulate angiogenesis. fn this paper they show oligosaccharides composed of 3-16 disaccharide repears stimulate dothelial ceU proliferation whereas HA polymer is inhibitory. They demonstrate binding and uptake of HA by the endothelial cells. 13.
l
14. 0
its and but that enalso
E, AUEWPERG N: A hyahuonate binding protein transiently cod&tributes with p21k.m in cuhured cell lines. Eq Cell Res 1989, 182340-348. This study demonstrates transient co-localization of HABP and p21 in the cell processes and ruffles of motile cells. However, whereas p21 remains in tuflles as motility ceases, HABP relocates to the ceU body. Tuw
15. 0
GREY AM, SCHOR AM, RUSHTON G, Ews 1, SCHOR SL Purilication of the migration stimulating factor produced by fetal and breast cancer patient fibroblasts. Proc Nat1 Acud Sci L’U 1989, 86:243%2442. This group has shown previously that con&rent fetal or cancer patient fibroblasts invade collagen gels to a greater extent than normal adult fibroblasts. In the work described here they have purilied a 70 kD protein from fetal and cancer patient fibroblasts that induces enhanced invasive behavior in adult libroblasts. 16. 0
CHEN
WYJ, GRANI’ ME, SCHOR AM, ~CHOR SL Dierences between adult and foetal fibroblasts in the regulation of hyaluronate synthesis: correlation with migratory activity. J cell Sci 1989, 943577-584. The same group as in [ 151 has shown that a major difference between adult and fetal fibroblasrs is that, when confluent, the latter produce higher levels of HA of greater average molecular weight than the former.
17.
SCHOR Sl+ SCHOR AM, ME, ELIIS 1: Mechanism
l
ing factor produced effect on hyaluronic 1989, 25~737-746.
CHEN J, of action of the by fetal and cancer acid synthesis. In GREY AM,
RUSHTON G, Grt,un’ migration stimulatpatient fibroblasts: Vim Ceif Dev Bid
843
844
Cell-to-cell
contact
and extracellular
matrix
The work presented in [ 15,161 is extended in this paper to show that the migration-stimulating activity of the 70 kD factor is dependent on HA production, both with respect to the enhancement of in\-asive behavior of adult fibroblasts on addition of the factor and the intrinsic difference between fetal or cancer patient fibroblasts and adult fibroblasts. 18. 0
MCCARTHY MT, Tools BP: Membrane-associated hyaluronatebinding activity of chondrosarcoma chondrocytes. J Cell P&o/ 1989, 141:191-202. This study demonstrates that cell-surface HABP retains HAproteogtycan complexes in the pericellular matrix and that the HABPfs) is not link protein or aggrecan. 19. 0
CROSSMAN MV. Mao~ RM: Purification and characterization of a hyaluronan-binding protein from rat chondrosarcoma Biccbem J 1990, 266399-406. A 102kD HABP was purified from extracts of Swarm rat chondrosarcoma and shown to react with HA-proteoglycan (aggrecan) complexes. The protein has some properties similar to aggrecan but was shown to be a distinct protein by peptide mapping.
20.
NG RF, SCHWARTZ NB: Solubilization and partial puritication of hyaluronate synthetase from oligodendroglioma cells. J Biol Cbem 1989, 264:11n611783. The molecular characteristics of hyluronate synthetase have been very difficult to elucidate because of problems in obtaining active prepara tions in solution. Although solubilization had been achieved for bacterial synthetase, this is the first report for eukaryotic synthetase. The solubilized enzyme was shown to have similar properties to membrane. bound enzyme, to retain nascent HA, and to produce polymeric HA 0
nan in human lung but not in skin tibroblasts. ,?$I Cell Res 1‘990. 186:192-195. TGFg %L< shown to stimulate HA synthesis three to fourfold in lungderived libroblasts but had no effect on skin fibroblasts. Proteo&can .synthesis was stimulated in both cell types This study extends the growing literature showing heterogeneity in fibroblasts from different sources. 24. 0
TOOLE BP, MUIWM SI, WEUIS S. KNUII~ON CB: Hyaluronatecell interactions and growth factor regulation of hyaluronate synthesis during limb development. 0% Fotorrtuhtion $,alrp 1989, 143:138149. This article reviews work on changes in HA .synthesis. HA-binding sites and organization of pencellular HA during differentiation of limb mesoderm. It al.so provides data on the role of HAstimulatory factors, especially bFGF which is enriched in the developing limb prior to differentiation and an ectodemlaltyderived factor that appears to sustain high levels of HA in the subectodermal mesoderm. The action of the latter factor is blocked by antibody to TGFP.
25.
SALLISIRJ 4 YANAGISHITA M. HAXAU. VC: Mouse oocytes regulate hyaluronic acid synthesis and mucilication by FSHstimulated cumulus cells. Dee, Biol 1990. 138:2632. Prior to ovulation, in response to increased FSH levels, the granulosa cells surrounding the oocyte (cumulus cells) change from a compact organization to a separated or ‘expanded’ configuration as a result of production of a HA-rich pencellular matrix. This report shows that the oocyte produces and secretes an HAstimulatory factor that, in concert with FSH, acts on the cumulus cells. 0
21. PREHM P: Release of hyaluronate from eukaryotic cells. 0 Biccbem J 1990, 267:185-189. This paper examines the mechanism of release of newiy synthesized HA from the cell surface. It was shown that, for the B6 hybrid carcinomaembryo fibroblast fine, HA is released as an intact polymer of the same size range as cell surface HA; in the case of SV3T3 cells, partial degra dation due to radical action also occurred.
W. BISWA-~ C. LI XQ, NEMEC RE. TOOIJI BP: The regulation of tumour-associated hyaluronan. Ciba Fowdafion $wnlp 1989, 143:15&169. The potential role of HA and HABPs in tumor cell invasion is reviewed. In particular, the nature of tumor~derived factors that stimulate libroblast HA production is discussed.
22.
27.
HEUXN P, LAURENT TC, HE~DIN CH: Effect of growth factars on hyaluronan synthesis in cultured human libroblasts. Biocbem J 1989, 258:919922. Platelet-derived growth factor, epidermal growth factor, bFGF and TGF8 were shown to stimulate HA synthesis in human fibroblasts. The level of stimulation of synthesis did not correlate with mitogenic activity of the factors. a
23. 0
WESTERGREN-THORSSON MALhLvROM A TGF-8
G, SARNSTRAND B, enhances the production
FRANSSON LA. of hyaluro-
26. l
KNUTSON role and
DECKHI M, CHILI ES, DOUHAL~I C. ~MOIIN A. HAU J. SPENDLO\‘E R, b3NGAKU(ER MT, STIXN R: Hyaluronic acid-stimulating activity in sera from the bovine fetus and from breast cancer patients. Gztrcer Ra 1989. 49:349+3505. This study documents the presence of an HA-stimulatory gtycoprotein in bovine fetal sent that decreases in amount with age of the fetus. Similar HA-stimulatory activiv is also present in elevated amounts in the set-a of cancer patients compared with that of normal controls and cancer patients in remission. 0
