Distinct Interaction of Versican/PG-M with Hyaluronan and Link Protein
The proteoglycan aggregate is the major structural component of the cartilage matrix, comprising hyaluronan (HA), link protein (LP), and a large chondroitin sulfate (CS) proteoglycan, aggrecan. Here, we found that another member of aggrecan family, versican, biochemically binds to both HA and LP. Functional analyses of recombinant looped domains (subdomains) A, B, and B′ of the N-terminal G1 domain revealed that the B-B′ segment of versican is adequate for binding to HA and LP, whereas A and B-B′ of aggrecan bound to LP and HA, respectively. BIAcore™ analyses showed that the A subdomain of versican G1 enhances HA binding but has a negligible effect on LP binding. Overlay sensorgrams demonstrated that versican G1 or its B-B′ segment forms a complex with both HA and LP. We generated a molecular model of the B-B′ segment, in which a deletion and an insertion of B′ and B are critical for stable structure and HA binding. These results provide important insights into the mechanisms of formation of the proteoglycan aggregate and HA binding of molecules containing the link module. The proteoglycan aggregate is the major structural component of the cartilage matrix, comprising hyaluronan (HA), link protein (LP), and a large chondroitin sulfate (CS) proteoglycan, aggrecan. Here, we found that another member of aggrecan family, versican, biochemically binds to both HA and LP. Functional analyses of recombinant looped domains (subdomains) A, B, and B′ of the N-terminal G1 domain revealed that the B-B′ segment of versican is adequate for binding to HA and LP, whereas A and B-B′ of aggrecan bound to LP and HA, respectively. BIAcore™ analyses showed that the A subdomain of versican G1 enhances HA binding but has a negligible effect on LP binding. Overlay sensorgrams demonstrated that versican G1 or its B-B′ segment forms a complex with both HA and LP. We generated a molecular model of the B-B′ segment, in which a deletion and an insertion of B′ and B are critical for stable structure and HA binding. These results provide important insights into the mechanisms of formation of the proteoglycan aggregate and HA binding of molecules containing the link module. The proteoglycan aggregate is the major structural component of the extracellular matrix of the cartilage (1Hascall V.C. Heinegard D.K. Wight T.N. Hay E.D. Proteoglycans: Metabolism and Pathology in Cell Biology of Extracellular Matrix. Plenum Press, New York1991: 149-175Google Scholar). It consists of hyaluronan (HA), 1The abbreviations used are: HA, hyaluronan; LP, link protein; CS, chondroitin sulfate; TBS-T, Tris-buffered saline with Tween 20; TSG, tumor necrosis factor-stimulated gene; PMSF, phenylmethanesulfonyl fluoride; PVDF, polyvinylidene difluoride; b-, biotinylated; HRP, horseradish peroxidase.1The abbreviations used are: HA, hyaluronan; LP, link protein; CS, chondroitin sulfate; TBS-T, Tris-buffered saline with Tween 20; TSG, tumor necrosis factor-stimulated gene; PMSF, phenylmethanesulfonyl fluoride; PVDF, polyvinylidene difluoride; b-, biotinylated; HRP, horseradish peroxidase. link protein (LP), and a large chondroitin sulfate (CS) proteoglycan, aggrecan. In this form, HA and CS chains attain stable deposition and exert their functions in the extracellular matrix. The aggregate provides cartilage with a unique gel-like property and resistance to deformation through water absorption, and regulates chondrocyte differentiation by storing extracellular signaling molecules involved in differentiation and distributing them to target cells (2Watanabe H. Yamada Y. Nat. Genet. 1999; 21: 225-229Crossref PubMed Scopus (159) Google Scholar). Thus, the proteoglycan aggregate plays important roles in the development and homeostasis of the cartilage. Formation of the proteoglycan aggregate in the cartilage requires specific interactions of aggrecan, HA, and LP. Both LP and the N-terminal G1 domain of aggrecan consist of three looped subdomains; A, B, and B′ (3Doege K.J. Sasaki M. Kimura T. Yamada Y. J. Biol. Chem. 1991; 266: 894-902Abstract Full Text PDF PubMed Google Scholar, 4Neame P.J. Christner J.E. Baker J.R. J. Biol. Chem. 1987; 262: 17768-17778Abstract Full Text PDF PubMed Google Scholar, 5Perkins S.J. Nealis A.S. Dudhia J. Hardingham T.E. J. Mol. Biol. 1989; 206: 737-753Crossref PubMed Scopus (70) Google Scholar). Each of the B and B′ subdomains contains a structure, termed a link module, which is believed to bind to HA (6Kohda D. Morton C.J. Parkar A.A. Hatanaka H. Inagaki F.M. Campbell I.D. Day A.J. Cell. 1996; 86: 767-775Abstract Full Text Full Text PDF PubMed Scopus (266) Google Scholar, 7Day A.J. Prestwich G.D. J. Biol. Chem. 2002; 277: 4585-4588Abstract Full Text Full Text PDF PubMed Scopus (467) Google Scholar). Functional analyses using recombinant aggrecan domains and subdomains have revealed that a stretch of B-B′ is the minimal segment for HA binding and a single link module does not bind to HA (8Watanabe H. Cheung S.C. Itano N. Kimata K. Yamada Y. J. Biol. Chem. 1997; 272: 28057-28065Abstract Full Text Full Text PDF PubMed Scopus (120) Google Scholar), contrasting with tumor necrosis factor-stimulated gene-6 (TSG-6), which can bind to HA with a single link module (9Kahmann J.D. O'Brien R. Werner J.M. Heinegard D. Ladbury J.E. Campbell I.D. Day A.J. Struct. Fold. Des. 2000; 8: 763-774Abstract Full Text Full Text PDF Scopus (90) Google Scholar). The A subdomains of both aggrecan G1 and LP, having an immunoglobulin (Ig)-fold structure (5Perkins S.J. Nealis A.S. Dudhia J. Hardingham T.E. J. Mol. Biol. 1989; 206: 737-753Crossref PubMed Scopus (70) Google Scholar), are thought to interact with each other (4Neame P.J. Christner J.E. Baker J.R. J. Biol. Chem. 1987; 262: 17768-17778Abstract Full Text PDF PubMed Google Scholar, 10Grover J. Roughley P.J. Madry H. Padera R. Seidel J. Langer R. Freed L.E. Trippel S.B. Vunjak-Novakovic G. Biochem. J. 1994; 300: 317-324Crossref PubMed Scopus (34) Google Scholar). Both the G1 and LP are associated side by side (11Perkins S.J. Nealis A.S. Dunham D.G. Hardingham T.E. Muir I.H. Biochem. J. 1992; 285: 263-268Crossref PubMed Scopus (13) Google Scholar), and each binds to the other and to HA, accomplishing the stable structure of the aggregate. The proteoglycans of the aggrecan family such as versican (12Zimmermann D.R. Ruoslahti E. EMBO J. 1989; 8: 2975-2981Crossref PubMed Scopus (501) Google Scholar, 13Shinomura T. Nishida Y. Ito K. Kimata K. J. Biol. Chem. 1993; 268: 14461-14469Abstract Full Text PDF PubMed Google Scholar), neurocan (14Rauch U. Karthikeyan L. Maurel P. Margolis R.U. Margolis R.K. J. Biol. Chem. 1992; 267: 19536-19547Abstract Full Text PDF PubMed Google Scholar), and brevican (15Yamada H. Watanabe K. Shimonaka M. Yamaguchi Y. J. Biol. Chem. 1994; 269: 10119-10126Abstract Full Text PDF PubMed Google Scholar) comprise similar domains to aggrecan. Because the G1 domain is highly conserved among the members, they may form aggregates with HA and LP. Whereas both neurocan and brevican are exclusively expressed in the brain, versican is expressed in various tissues including the central and peripheral nervous systems, the luminal surface of glandular epithelia, blood vessels in normal and tumor tissues, dermis, and the proliferative zone of the epidermis, and embryonic tissue (16Bode-Lesniewska B. Dours-Zimmermann M.T. Odermatt B.F. Briner J. Heitz P.U. Zimmermann D.R. J. Histochem. Cytochem. 1996; 44: 303-312Crossref PubMed Scopus (162) Google Scholar). In these tissues, versican plays important roles in adhesion (17Yamagata M. Saga S. Kato M. Bernfield M. Kimata K. J. Cell Sci. 1993; 106 (Pt 1): 55-65Crossref PubMed Google Scholar), migration (18Landolt R.M. Vaughan L. Winterhalter K.H. Zimmermann D.R. Development. 1995; 121: 2303-2312Crossref PubMed Google Scholar), proliferation, and differentiation (19Kishimoto J. Ehama R. Wu L. Jiang S. Jiang N. Burgeson R.E. Proc. Natl. Acad. Sci. U. S. A. 1999; 96: 7336-7341Crossref PubMed Scopus (191) Google Scholar) of cells by interacting with cell surfaces and extracellular matrix molecules. Although versican is transiently expressed during development, it is constitutively expressed in some adult tissues such as the heart, blood vessels, and the brain. As both HA and LP are ubiquitously expressed (20Binette F. Cravens J. Kahoussi B. Haudenschild D.R. Goetinck P.F. J. Biol. Chem. 1994; 269: 19116-19122Abstract Full Text PDF PubMed Google Scholar), versican may exert its functions on them in the aggregate form. However, a proteoglycan aggregate composed of any other member of the aggrecan family has not been identified. In the present study, we demonstrate that versican biochemically binds to both HA and LP at a segment of B-B′ of the G1 domain, contrasting with aggrecan which binds to LP at the A subdomain. Using the link module structure of TSG-6 and CD44, we performed molecular modeling of the B-B′ segment, which may explain the requirement of tandemly repeated link modules. These results may provide a clue to the mechanism of the specific HA-interactions of molecules with link modules. Construction of Expression Vectors—A 325-amino acid cDNA fragment of the versican G1 domain was generated by polymerase chain reaction (PCR), using human versican cDNA (gift from Isogai), Pfu DNA polymerase (Stratagene), and a primer set (Table I) that are attached by XhoI and BamHI linker sites, respectively. The reaction program was 25 cycles of 94 °C for 20 s, 60 °C for 30 s, and 72 °C for 2 min. The amplified DNA fragment was inserted at XhoI and BamHI sites of pBFX (8Watanabe H. Cheung S.C. Itano N. Kimata K. Yamada Y. J. Biol. Chem. 1997; 272: 28057-28065Abstract Full Text Full Text PDF PubMed Scopus (120) Google Scholar). The resulting plasmid designated as VerABB′/pBFX was used as the basic vector for further versican gene constructions. A cDNA fragment encoding the G1 subcloned into pBluescript II SK(–) as a template and a primer set as described in Table I used for of of versican The encoding subdomains inserted at XhoI and BamHI sites of The DNA of was using an DNA Expression of human LP and B-B′ of LP by using was performed to a cDNA and a segment encoding using polymerase with the reaction The DNA fragment encoding LP was inserted at XhoI and BamHI sites of The resulting designated as plasmid encoding a protein of the B-B′ of LP and chain was by by insertion of the fragment into for polymerase chain in a Expression and of cells in containing and The cells at in with 2 of vector DNA and of to the from the of the stable cells in the of for and a of stable was further to of recombinant in of the cells The cells in for and the was used as the of phenylmethanesulfonyl and the was for 60 at and the was to an with 20 and the was three with 2 of the The was with of the containing and using a with the recombinant by and at The of recombinant was by and of each recombinant protein was from of of versican was as described M. Dours-Zimmermann M.T. Winterhalter K.H. Zimmermann D.R. J. Biol. Chem. Full Text Full Text PDF PubMed Scopus Google Scholar). of was in a using of containing 25 PMSF, and The was for at °C and at for from the by the of an at by the sulfate in the to to form at °C and by at for min. The was in A and to at °C the was with A containing and was by from to 2 on a with an and and of LP and HA was performed as described 1995; Scholar). versican and LP and in a at °C the was with protein for 30 at 2 of an was at °C for 2 was performed by with of protein for 2 at The with on a a and to a polyvinylidene The was for in 20 Tween 20 containing and with 2 for at The reaction was by The was by a or a and a The was for at in containing the was with an or an three with TBS-T, the was with three with TBS-T, the reaction was by the was with was 25 LP or 25 HA for 2 at (8Watanabe H. Cheung S.C. Itano N. Kimata K. Yamada Y. J. Biol. Chem. 1997; 272: 28057-28065Abstract Full Text Full Text PDF PubMed Scopus (120) Google Scholar). The was with 2 for three with TBS-T, and with to that with or analyses performed using a BIAcore™ at 25 of recombinant LP and to and to a of LP, and at in containing into the cell at a of and respectively. The in each cell was by the of the In the the binding performed with a of in both and In a of protein from to in and was into the and the in was each of the surface was by of of The for by analyses using by the and as described in (8Watanabe H. Cheung S.C. Itano N. Kimata K. Yamada Y. J. Biol. Chem. 1997; 272: 28057-28065Abstract Full Text Full Text PDF PubMed Scopus (120) Google Scholar). from the of the during the from the of the the of the during the The was as the of The by of of three we used the HA and the In a of protein from to in was into the and the in was each of the surface was performed by of of and respectively. and of of the link module from acid of versican, aggrecan, LP, CD44, and TSG-6 of both and and neurocan and brevican of and was performed using an of and S.B. J. Mol. Biol. PubMed Scopus Google Scholar). of the of human versican B and B′ with human TSG-6 was using an T. and M. in which of deletion and insertion acid is and critical for molecular of the of the link of human versican B and B′ domains and human modeling of B and B′ was performed using K. H. J. Mol. 2000; Scholar) with the structure of TSG-6 as a The link module contains As the program does not into the side chains of may In this we bound sulfate of their in and performed using the of the D. M. J.D. S. J. H. S. D. L. K. S. T. B. B. M. R. J. Watanabe M. J. D. M. J. Chem. B. PubMed Scopus Google Scholar) was Both B and B′ domains that the insertion of B and the deletion of B′ are involved in the of both domains and the surfaces are in a The model of was both of and are which is of the stable A. A. J. Mol. Biol. PubMed Scopus Google Scholar). to Both LP and HA at the B-B′ Whereas versican has been to interact with HA Zimmermann D.R. Ruoslahti E. J. Biol. Chem. 1992; 267: Full Text PDF PubMed Google Scholar), the versican and LP has not been We versican binds to LP by from was with LP and with The of versican in the was by using the was to using horseradish was that versican bound to LP in We their by and versican to a a and to a was performed using and versican as a and a and the was with by with a and a to of versican and of the was using a recombinant protein of G1 as a and These results that LP with versican on the we the subdomains for the with LP by using recombinant domains and subdomains of LP, and the G1 domain of versican and aggrecan as with a at the and chain at the for the LP, which was expressed as a protein recombinant showed a single for which an that to a LP bound to and both of which B and B′ subdomains of the versican G1 domain, whereas and not These results that the B-B′ segment of the versican G1 domain is for with LP. In and bound to LP, but that the A subdomain of the aggrecan G1 domain with LP. LP and not bind to LP. we the subdomains that bind to HA, using a LP, and bound to HA in to and not as (8Watanabe H. Cheung S.C. Itano N. Kimata K. Yamada Y. J. Biol. Chem. 1997; 272: 28057-28065Abstract Full Text Full Text PDF PubMed Scopus (120) Google Scholar). These results that B-B′ of versican, aggrecan, and LP, are for with of LP or HA Using BIAcore™ the of of recombinant with LP and HA, using BIAcore™ and bound to recombinant LP on the and the including and LP results by (Table The and with LP similar and that the A subdomain of versican G1 has a negligible effect on the with LP. Although both and with LP, associated and resulting in a of that the B-B′ segment enhances of the A subdomain with LP. G1 domains of versican and aggrecan aggrecan G1 showed to a with at a of that the B-B′ of both versican G1 and LP interact with each of various recombinant domains with in a we the of HA of these recombinant LP, and with HA, but and not (Table with the Although three molecules with HA at their B-B′ segment, LP bound to HA, by versican and aggrecan the A associated and with resulting in as by the A subdomain showed with HA LP. and with HA at a of and with (8Watanabe H. Cheung S.C. Itano N. Kimata K. Yamada Y. J. Biol. Chem. 1997; 272: 28057-28065Abstract Full Text Full Text PDF PubMed Scopus (120) Google Scholar). These results that the A HA binding enhances of the B-B′ segment with HA, its are among versican aggrecan and of various recombinant domains with in a LP, and HA a the BIAcore™ we versican G1 with both LP and HA to form a stable or with LP for Although of three molecules not using the BIAcore™ the of by the of We to a cell of HA and LP. associated with HA, and LP further that LP with the complex or a of the HA by The of three molecules was not by of any for which was used in of similar In was used in of they not form a stable complex not we and their was not These results that HA and the B-B′ of both versican and LP bind to each other to form a stable both and LP and the HA the bound to HA, and their binding was not by not These results that of their interactions may not critical for formation of the stable of aggrecan, both versican G1 and LP both B and B′ domains for whereas TSG-6 requires a single link module. Both B and B′ domains and of to the link module of that both B and B′ from an domain of the link module. We of the link module of versican, aggrecan, LP, and of both and and neurocan and brevican of In the of the human versican B domain, the acid and to and the in TSG-6 Day A.J. J. Biol. Chem. Full Text Full Text PDF PubMed Scopus Google Scholar), and the acid and to and the for HA in J. B. S.B. Day A.J. A. J. Biol. Chem. Full Text Full Text PDF PubMed Scopus Google Scholar) In the of the B′ domain, and to and of the of and and to and for HA in The acid and of the B domain and and of the B′ on a that these acid form a surface of B-B′ In the acid of TSG-6 showed a of and of B or B′ to Thus, we of B and of to as the acid the The a deletion of three acid in the B′ domain and an insertion of or three acid in the B domain The deletion in B′ to the of acid for HA binding in both TSG-6 and CD44, which may explain the of B′ to interact with As the deletion and insertion not found in TSG-6 and containing a single link module, they to associated with of a single link module to B and B′ on these we performed molecular modeling of the B-B′ Because was linker stretch B and and both involved in HA they thought to side by In this the insertion and deletion generated to the of both B and B′ domains a and a chain of was to the it was the it a of modeling of the B and B′ A, the molecular model of human versican B and B′ domains with surfaces are in acid that to involved in HA binding in TSG-6 are in the as in B, the and and are The B and B′ domain are in and respectively. and an insertion of B and a deletion of respectively. the deletion in the the surface which stable for the of B and the insertion in the B, the surface the complex model of the B-B′ segment and In this the of with the as We have the molecular interactions in the formation of proteoglycan aggregate and its in As has been versican forms proteoglycan aggregates with LP and The versican G1 domain, aggrecan binds to HA at a minimal segment of which is by the A subdomain. Whereas aggrecan G1 binds to LP at the A versican G1 does at the B-B′ segment, that the B-B′ is adequate for stable aggregate formation of binding and HA by BIAcore™ analyses revealed important roles of each in the aggregate the requirement of both B and we further performed molecular modeling of the tandemly repeated link of B and B′ of versican a deletion of the B′ and an insertion of the B, they are side by which the with These results provide important insights into the mechanisms of HA binding of molecules with the link module. of HA recombinant expressed and in the We used the BIAcore™ which any such that and can We have found that and results similar to but that HA and may interactions HA and basic acid of the (8Watanabe H. Cheung S.C. Itano N. Kimata K. Yamada Y. J. Biol. Chem. 1997; 272: 28057-28065Abstract Full Text Full Text PDF PubMed Scopus (120) Google Scholar). The HA binding of versican G1 was that of aggrecan G1 of HA binding and for versican Zimmermann D.R. Ruoslahti E. J. Biol. Chem. 1992; 267: Full Text PDF PubMed Google Scholar) and aggrecan T. H. A.A. D. S. T. H. T. J. Y. 1987; PubMed Scopus Google and Hardingham T.E. Muir H. Biochem. J. PubMed Scopus Google may the of and of which enhances HA binding (8Watanabe H. Cheung S.C. Itano N. Kimata K. Yamada Y. J. Biol. Chem. 1997; 272: 28057-28065Abstract Full Text Full Text PDF PubMed Scopus (120) Google Scholar). We present for the the molecular model of the B-B′ segment, which is in with the that a module of versican does not bind to HA and that the minimal for is a deletion in the B′ and an insertion in the B to the of the aggrecan family and LP, that they an link module was tandemly in It tandemly repeated link are for HA binding. As versican G1 and LP bind to HA of and the link module of TSG-6 from (9Kahmann J.D. O'Brien R. Werner J.M. Heinegard D. Ladbury J.E. Campbell I.D. Day A.J. Struct. Fold. Des. 2000; 8: 763-774Abstract Full Text Full Text PDF Scopus (90) Google Scholar), these large proteoglycans have HA binding by of the link module in G1 and LP interact at their B-B′ and their does not with their HA binding. These that the of the versican B-B′ segment for LP binding is of the surface for HA binding. Whereas the B-B′ segment of LP is versican B-B′ is their However, their binding was the was performed with not Thus, their binding specific The acid of the versican B-B′ segment to that of aggrecan The critical acid for LP binding of versican B-B′ to We have at of functions and of the A the A domains of versican and LP of the B-B′ segment, as for aggrecan G1 (8Watanabe H. Cheung S.C. Itano N. Kimata K. Yamada Y. J. Biol. Chem. 1997; 272: 28057-28065Abstract Full Text Full Text PDF PubMed Scopus (120) Google Scholar), the of is among LP recombinant protein is expressed as a as by a not and it binds to LP to its B-B′ These results with the that LP forms a in the of and to a bound to HA L. S. J. Biol. Chem. Full Text PDF PubMed Google Scholar). the A subdomain of aggrecan G1 can bind to LP, which we demonstrated for the provides from of complex (4Neame P.J. Christner J.E. Baker J.R. J. Biol. Chem. 1987; 262: 17768-17778Abstract Full Text PDF PubMed Google Scholar), and using with J. Roughley P.J. Madry H. Padera R. Seidel J. Langer R. Freed L.E. Trippel S.B. Vunjak-Novakovic G. Biochem. J. 1994; 300: 317-324Crossref PubMed Scopus (34) Google Scholar). the A subdomain of versican to bind to LP. The A subdomains of versican aggrecan and LP a conserved results that a in the acid may for Although we have demonstrated the requirement of an for the mechanism of of the to We have described the structure for the formation of proteoglycan aggregate and its in results of may to of the mechanism of HA and to tissue of the extracellular matrix to the resistance to other of the LP family have been such as the link S. T. H. T. J. Y. Biochem. 2000; PubMed Scopus Google Scholar), resulting in members, which are designated as HA and link protein A. J. Biol. Chem. Full Text Full Text PDF PubMed Scopus Google Scholar). The of the aggrecan family may from LP family members, and other of proteoglycan aggregate with functions may As the A subdomains of both aggrecan G1 and LP interact with each the A subdomain is to a in of the However, the that the B-B′ of both versican and LP interact with each other of the is constitutively expressed in some tissues such as the and the other the cartilage LP may in the versican aggregate formation in these In the versican is present in the zone of the and the is in Y. T. H. T. Kimata K. 1994; Full Text PDF PubMed Scopus Google Scholar). versican may form aggregates with LP and HA, and the versican aggregate may have from the aggrecan aggregate. on the structure and in of these aggregates to We A. J. A. P. and P. J. for critical and and M. and for and versican