IE!
`
`ANNUAL REVIEW OF
`
`NEUROSCIENCE
`
`VOLUME 13, 1990
`
`W. MAXWELL COWAN, Editor
`Howard Hughes Medical Institute
`
`ERIC M. SHOOTER, As3oc'i'are Editor
`Stanford University School of Medicine
`
`CHARLES F. STEVENS, Associate Editor
`Yale University School of Medicine
`
`RICHARD F. THOMPSON, Associiote Editor
`University of Southern California
`
`ANNUAL REVIEWS INC‘
`
`41.19 EL CAMINO WAY
`
`PD. BOX H.513‘?
`
`PALO ALTO. CALIFORNIA 943034339?
`
`Page 1 of31
`
`ILMN EXHIBIT 1040
`
`
`
`Page 1 of 31
`
`ILMN EXHIBIT 1040
`
`

`
`ANNUAL REVIEWS INC.
`Palo Alto, California, USA
`
`COPYRIGHT c'_t;j,u I991} av ANNUAL REVIEWS lN(T.., PALO ALTO, CALl!"URNl.‘\. Usit.
`ALL atoms aasaavizo. The appearance of the code at the bottom of the first page
`of an article in this serial indicates the copyright owner’s consent that copies
`of the article may be made for personal or internal use, or For the personal or
`internal use of specific clients. This consent is given on the condition. however.
`that the copier pay the stated per-copy fee of $2.00 per article through the
`Copyright Clearance Center. Inc. (El Congress Street. Salem, MA 0l9?0] for
`copying beyond that permitted by Section 10? or 108 of the US Copyright Law.
`The per—co-py Fee ol" $2110 per article also applies to the Copying, under the
`stated conditions, ofartieles published in any Artnntn’ Rena-tr serial before January
`I.
`l9'?S. Individual readers. and nonprofit libraries acting For them, are
`permitted to make a single copy of an article without charge for use in research
`or teaching. The consent does not extend to other kinds of copying. such as
`copying For general distribution. for advertising or promotional purposes. For
`creating new collective works, or For resale. For such Ltses. written pennission
`is required. Write to Permissions Dept.. Annual Reviews Inc. 4139 El Camino
`Way. PO. Box IIIIB9, PHIO Alto, CA 94303-0397 USA.
`
`fntcnrarinnal Standard Scrin! Number : I’).-'47—tfiU6X
`
`litre:-itatiririrri Srnrrtthrri Book Navriher : t‘.l'-8'24}-24'i'3—?
`
`Annual Review and publication titles are registered trademarks of Annual Reviews
`Inc.
`
`@ The paper used in this publication meets the minimum requirements of Amer-
`ican National Standard for Information Scienccs— Permanence oi‘ Paper for Printed
`Library Materials. ANSI Z39.4E—l934.
`
`Annual Reviews Inc. and the Editors ofits publications assume no responsibility
`For the statements expressed by the contributors to this Review.
`
`T‘r'PESET BY Alli’ 'I"i"PESETTERS (GLASGOW! LTD.. SCO'l'LfitND
`PRINTED AND BOUND IN THE UNITED STATES DF AMERICA
`
`Page 2 of31
`
`
`
`Page 2 of 31
`
`

`
`.-'3.'22?—.55
`Anna. Rec. Neuroscf. J'99'D.
`Ct:Ip_}'r1'g.Ir.'
`',u;'.‘,' 1990 by Annual Re.-.Jfew.r Inc. AH rr'gl::.r reserved
`
`CARBOHYDRATES AND
`
`CARBOHYDRATE—BINDING
`
`PROTEINS IN THE
`
`NERVOUS SYSTEM
`
`T. M. Jesseflf‘ M. A. Hynes,* and J. Doddi
`
`Center for Neurobiology and Behavior, Departments of * Biochemistry
`and Molecular Biophysics, Howard Hughes Medical Institute,
`and ‘rPhysiology and Cellular Biophysics. Columbia University,
`New York, New York 10032
`
`INTRODUCTION
`
`The possibility that cellular interactions within the vertebrate nervous
`system are mediated by cell surface carbohydrates has been considered on
`numerous occasions. The initial suggestions that carbohydrate structures
`might mediate neural cell adhesion were based on the expression of com-
`plex oligosaccharides, in particular gangliosides, by neural cells, and the
`detection of cell surface glycosyltransferases that were proposed to func-
`tion nonenzymatically as receptors for surface oligosaccharides (Roseman
`1970, Roth et al 1971, Marchase l9Tt', Shur tit Roth 1975}. Subsequent
`progress in elucidating the function of cell surface carbohydrates in the
`nervous system has. however, been slow, in part because of the dilficulties
`(a) in purifying and characterizing complex oligosaccharides that are
`expressed on small subsets of neural cells and (E:-} in generating these
`structures synthetically. In addition, the identification of cell surface mol-
`ecules such as neural cell adhesion molecules (NCAM), N-cadherin, and
`integrins (Edelman 1986, Talteichi 1988, Ruoslahti & Pierschbacher l98?,
`Rutishauser 3; Jessell 1983) has focused attention on mechanisms ofneural
`cell adhesion that involve direct protein-protein interactions.
`Increasing evidence indicates that interactions between surface oligo-
`saccharides and carbohydrate-binding proteins mediate cell adhesion and
`recognition between nonneural cells. Thus, the stage- and species-specific
`227
`
`0147-006X ,u‘90/030 l—022'."$D2.00
`
`Page 3 of31
`
`
`
`Page 3 of 31
`
`

`
`223
`
`IESSELL, HYl*~JES at noon
`
`t
`
`binding of a sperm to the zona peltucida coat that surrounds mammalian
`oocytes has been shown to result from the interaction of a sperm receptor
`with an 0-linked oligosaccharide that is present on the zona pellucida
`glycoprotein, ZP-3 (Wassarman I987). Evidence also indicates that the
`polylactosamine oligosaccharides present on the blastomeres of pre-
`implantation mouse embryos are involved in adhesive interactions that
`occur during compaction (Fenderson et al 1984, Rastan et al 1985, Bayna
`et al 1983). In addition, the specific homing of recirculating lymphocytes
`to peripheral lymphoid targets (Rosen et al 1935, Rosen 3: Yednock 1986,
`Gallatin et al 1986, Brandley et al 1987) appears to depend, at least in
`part, on receptors that recognize specific carbohydrate structures. Finally,
`the hepatic asialoglycoprotein receptor that is responsible for the clearance
`of circulating serum glycoproteins (Ashwell & Harford I982} represents
`the bestwcharacterized surface protein with a defined physiological role in
`the recognition of carbohydrate structures.
`With the availability of monoclonal antibodies (MAbs), the complex
`expression patterns of oligosaccharides on neural cells has become more
`readily apparent. Some of the defined carbohydrate antigens are highly
`restricted to subsets of vertebrate neurons and reveal molecular gradients
`in developing neural tissues. In addition, several carbohydrate-binding
`proteins with specificity for neural cell surface oligosaccharides have
`recently been detected in the vertebrate nervous system. In this review we
`discuss briefly the evidence emerging from both neural and nonneural
`systems that indicates that cell surface carbohydrate structures may indeed
`play important roles in mediating neural cell recognition and adhesion.
`
`DIVERSITY OF CARBOHYDRATE STRUCTURES
`
`ON VERTEBRATE CELLS
`
`The complex oligosaccharides expressed by vertebrate cells are associated
`with ceramides in glycolipids or attached via N- or O—Iinkages to protein
`backbones. Several classes of carbohydrate structures can be defined on
`the basis of their polysaocharide backbone sequences (Table I). Lactoseries
`
`I Oligosaocharide
`Table
`charide backbone sequence
`
`classification
`
`by
`
`polysac-
`
`Lactc-series
`
`(Type I}
`{Type 2)
`Globoseries
`Cianglioseries
`
`Gal(fil—3)CilcNAc[,|H'l-3]GaIflI-4G1c—R
`Gal(,[il -4)GlcNAc(,E I —3}Ga|fl I -4G1c-R
`Gal Nfitcfl I -3Ga]t:¢ I -4Gal,6'l-4C'rlc-R
`Galfl I -3GalNAc,G i -4GalB I -4Glc— R
`
`Page 4 of31
`
`
`
`Page 4 of 31
`
`

`
`CARBOHYDRATE RECOGNITION
`
`229
`
`carbohydrates contain the [Galfil-3(4}GlcNAc-R] structure, globoseries
`carbohydrates contain the [GalNAcf)’l-3GalaI-4Gal-R] backbone and
`ganglioseries structures contain the [Galfil-3GalNAc_l3’l—R] sequence
`(E-Iakornori I931, Feizi 1985). These backbone sequences can be modified
`extensively by the addition of branched or terminal saccharides,
`thus
`generating many structurally distinct members of each class. In addition,
`
`the attachment of the same saccharide via multiple linkages. the existence
`of branched carbohydrate chains of‘ the same or diflering structure. and
`extensive variation in sialic acid content {Schauer I982} provide the poten-
`tial for an enormous diversity of complex oligosaccharide structures.
`The assembly of complex oligosaccharides is achieved by the coor-
`dinated and sequential activity of glycosyltransferase enzymes (Beyer 3.:
`Hill
`I982}. Each enzyme is capable of adding specific saccharides via
`defined linkages to a highly restricted set of oligosaccharide substrates.
`The structural diversity in cell surface oligosaccharides must therefore be
`defined in large part by the cellular expression and substrate specificity of
`these glycosyltransferascs.
`
`CARBOHYDRATE-BINDING PROTEINS
`
`A large number of endogenous proteins have been characterized that bind
`to distinct surface oligosaocharides on vertebrate cells. These carbo-
`hydrate-binding proteins can be subdivided into several categories on the
`basis oftheir primary structure and biochemical properties (see Drickamer
`1988, Barondes 1988):
`l. Calcium-dependent carbohydrate-binding proteins {C-type lectins):
`Lectins of this class require the presence of calcium for carbohydrate-
`binding activity and are defined on the basis of the homology with the
`carbohydrate-recognition domain of the rat hepatic asialoglycoprotein
`receptor (Drickamer I983). The common structural organization of the
`binding domain results from a conserved set of IS amino acids that includes
`cysteine residues involved in disulphide bond formation, which is essential
`For carbohydrate-binding activity (Dl‘iCl<fll'E'l€l‘ 1988). Diflbrent members
`of this class exhibit distinct sugar-binding specificities that may result from
`nonconserved amino acids within the binding domain. This class of lectins
`also includes the chicken hepatic N-acetylglycosamine receptor. the soluble
`rat mannose-binding proteins (Drickarner et al I986, Ezel-towitz et al i988),
`the pulmonary surfactant apoprotein (I-Iaagsman et al 198?}, cartilage pro-
`teoglycan (Krusius et al 1987, Halbcrg et al I938), and a lymphocyte hom-
`ing receptor (Siegelrnan et al I989. Lasky et al I939). Several other proteins
`have been identified that exhibit
`this conserved carbohydrate-binding
`domain but that have not yet been demonstrated to function as lectins.
`
`Page 5 of31
`
`
`
`Page 5 of 31
`
`

`
`230
`
`JESSELL, Hvmas a Door)
`
`I
`
`2. Calcium-independent, soluble carbohydrate-binding proteins (S-lac
`type lectins}: Several low-molecular-weight soluble carbohydrate-binding
`proteins have been isolated from a wide range of tissues and constitute a
`separate class of lectins {Barondes I984. 1988). S-lac lectins form a struc-
`turally homologous class, sharing a number ofconserved amino acids that
`may be critical for carbol1ydrate—binding function (see Drickamer I983).
`In contrast to C~type lectins. S~lac lectins are inhibited by oxidation and
`do not contain invariant cysteine residues. They exhibit a degree of con-
`servation similar to that of the C-type lectins, and most of them share
`binding specificity for fi-galactosides. Some of‘ these lectins, however, have
`markedly different binding allinities for more complex lactosarnine-based
`oligosaccharides (Lefiler dz Barondes 1986}. Included in this class are
`two proteins termed RL-l-4.5 and RL-29 that are expressed with striking
`selectivity within the developing nervous system. Soluble lectins with man-
`nose-binding properties have also been identified in neural tissues (Zanetta
`et al 1987}, although the structure of these proteins has not yet been
`established.
`
`3. Membrane-bound or soluble glycosyltransferases: Cilycosyltrans-
`ferases have been proposed to function as carbohydrate-binding proteins
`when their appropriate nucleotide sugar donors are not available {Bayna
`et al 1936}. The cell surface expression of glycosyltransferases, though
`still controversial. appears likely, and represents a situation in which the
`enzyme may bind carbohydrate substrates without enzymatic transfer of
`
`additional saccharides. Comparison ofthe primary structure of galactosyl-
`transferases {Shaper et al I986. I988. Narimatsu et al 1986) and sialyl-
`transferase (Weinstein et al 1987), predicted by CDNA clones, does not
`indicate any striking sequence similarities between glycosyltransferases
`with distinct saccharide-binding properties.
`
`CARBOHYDRATE-MEDIATED CELL ADI-IESION
`
`AND RECOGNITION
`
`In several nonneural systems there is now quite compelling evidence that
`interactions between carbohydrates and carbohydrate-binding proteins
`mediate cell adhesion and recognition. In this section we discuss briefly
`the evidence for carbohydrate recognition in nonneural systems with the
`intention of providing a framework for assessing the role of similar or
`identical molecules expressed within the nervous system.
`
`Feriiiiznrion
`
`The binding of sperm to mammalian eggs exhibits a striking species-
`and stage-specificity. This simple example of selective cell recognition has
`
`Page 6 of31
`
`
`
`Page 6 of 31
`
`

`
`CARBOHYDRATE RECDGNTITON
`
`23]
`
`provided an accessible system with which to examine the molecular basis
`of intercellular recognition in vertebrates. The initial specificity in mam-
`malian sperm-egg interactions appears to result from the binding of recep-
`tors on the sperm surface to a ligand present on the zona pellucida mem-
`brane that surrounds the egg (Wassann-an 1987). In the mouse. the zona
`pellucida consists of only three major glycoproteins, termed ZP-1, ZP-2
`and ZP-3. It has been established that the ZP-3 glycoprotein mediates the
`binding of sperm (Bleil & Wassarman 1930. Wassarman I987). The ZP~3
`protein contains both N-linked and O-linked oligosaccharides. By com-
`paring the effect of proteolytic degradation of ZP-3 with the selective
`cleavage of N- and O-linked oligosaccharides from the protein. was
`sarman and colleagues have shown that the O-linked oligosaceharides on
`ZP-3 serve as the ligand recognized by sperm receptors {see Wassarman
`I987). The precise structure of the ZP-3 oligosaccharides that function as
`the sperm receptor is not yet known. The addition of fucans and other
`fucose polymers. however, inhibits the binding of sperm in a wide variety
`of species, suggesting that sperm binding may be dependent on a fucose
`
`group (Huang 3:. Yanagimachi 1934). The use of selective glycosidases
`
`has also implicated an unlinked terminal galactose as a crucial structural
`component of the ZP-3 Cl-linked oligosaccharide receptor (Bleil & Was-
`sarrnan 1988).
`
`The nature of receptors on the sperm surface that bind to the ()-linked
`oligosaccharide on ZP-3 is not clear. One candidate is a galactosyl-
`transferase {CialTase) (Lopez et al I985]. Several lines of evidence are
`consistent with this possibility. Inhibitors ofGalTase activity. in particular
`sc-lactalbumin and UDP-dialdehyde. inhibit sperm binding when assayed
`in vitro. The addition of soluble GalTase or anti-GalTase antibodies also
`
`inhibits the binding of sperm. In addition. UDP-galactose (the nucleotide
`sugar donor for the enzyme). but not other nucleotide sugars. is able to
`dissociate sperm from the egg zona pellucida (Lopez et al 1935). Since this
`GalTase catalyzes
`the transfer of galactose to terminal N-acetyl-
`glucosarnine, the ZP—3 oligosaccharide might be expected to exhibit a
`terminal Macetylglucosamine residue. The demonstration that a pre-
`existing terminal ct-galactose on ZP~3 is required For sperm receptor activity
`(Bleil 3: Wassarman I988), together with the evidence that the receptor
`involves a fucose residue, suggests that multiple saocharide determinants
`are involved in sperm recognition of ZP-3. In some invertebrates.
`the
`sperm receptor appears to be a leetin termed bindin (Glabe et al 1982). It
`is still possible. therefore, that lectins on the surface of mammalian sperm
`contribute to interactions with the ZP—3 O-linked oligosaccharides. Recent
`studies have identified the asialoglycoprotein receptor (ASGP-R) or a
`closely related protein on the surface of mammalian sperm (Abdullah &
`
`Page 7 of31
`
`
`
`Page 7 of 31
`
`

`
`232
`
`JESSELL, HYNES & oooo
`
`Kierszenbaurn 1989). Further characterization of the O-linked oligo-
`saccharides on ZP-3 should provide information on functionally relevant
`carbohydrate-binding proteins on the sperm surface.
`
`Adhesion of Biostomeres in Cieoooge-stage Mouse Embryos
`
`Studies of preimplantation mouse embryos have provided evidence for
`roles of both calcium-dependent adhesion molecules and oligosaccharides
`in blastomere adhesion. At the early 8-cell stage, individual mouse blaste-
`meres exhibit a low degree of cell-cell contact and retain defined cell
`boundaries (Calarco-Gillam I935). A striking increase in blastomere
`adhesion and in cell-cell contact occurs at this stage of development, a
`process termed compaction. The earliest phase of compaction appears to
`be Ca“-dependent, whereas at later times in the compaction process,
`adhesion becomes progressively less dependent on the presence of Ca“
`(Ducibella 3: Anderson 1975). The Ca“ -dependent cell adhesion molecule,
`E-cadherin, is expressed on blastomeres at the 8-cell stage and mediates
`the early, Ca: ’' -dependent phase of compaction (Hyafil et al i980, Takeichi
`I986}. Several studies have suggested that oligosaccharides play a role in
`the later, Ca“—independent adhesive events that undertie compaction.
`The possibility that oligosaccharides are involved in blastomere
`adhesion derived initially from analysis of a set of Eactoseries oligo-
`saocharides, originally defined as
`stage-specific embryonic antigens
`(SSE!-ts) (Sc-Iter & Knowles l9T8). These lactoseries structures, in par-
`ticular the fucosylated SSEA-I structure. are expressed on mouse blaste-
`meres immediately preceding compaction. The addition of soluble, multi-
`valent neo-glycoproteins that express the SSEA-l-reactive trisaccharide
`inhibits the compaction of early mouse embryos (F-enderson et al 1984}.
`Inhibition of compaction is specific to the SSEPL-l determinant, since
`structural analogues do not inhibit compaction. in addition, when embry-
`onal cell surface polylactosaminoglycans are cleaved by the bacterial
`enzyme endo-beta-galactosidase,
`the recompaction of previously dis-
`sociated blastomeres is significantly delayed (Rastan et al 1985). Similar
`
`enzyme treatment does not appear to affect earlier stages of compaction,
`
`a finding that suggests that the cell surface oligosaccharides sensitive to
`endo-beta-galactosidase are involved only in the later stages of com-
`paction.
`Mouse blastomeres and teratocarcinoma cells express leetin-like mol-
`ecules on their surfaces. A 56 kDa lectin purified from mouse terato-
`carcinoma cells (Grabel I984, Grabel et al 1935) has been reported to react
`with fucosylated oligosaccharides similar to those present on B-cell stage
`embryos. There is also evidence for the involvement of cell surface Gal-
`
`Page 8 of31
`
`
`
`Page 8 of 31
`
`

`
`CARBOHYDRATE RECOGNITION
`
`233
`
`Tases in embryonic compaction (Bayna et al I988). GalTase activity on
`the surface of mouse blastorneres increases markedly during the later
`phases of compaction. Moreover, addition of aulactalbumin or antibodies
`that eliminate GalTase activity result in the decompaction of mouse morn-
`lac. The compaction of preimplantation mouse embryos may therefore
`prove useful as a system with which to analyze the respective contributions
`of cadherin- and carbohydrate-mediated adhesion on a single mammalian
`cell.
`
`Lymphocyte Homing
`
`Cine of the initial events in the recirculation of lymphocytes is their
`migration from the bloodstream to lymphatic ducts in peripheral lymphoid
`tissues. This process has been termed lymphocyte homing and involves the
`adhesion of lymphocytes to a specialized set of high endothelial venules
`(I-IEV} that are located in the post-capillary beds (Ga!latin et al 1986).
`Different subsets of recirculating lymphocytes interact in an organ-specific
`
`manner with HEV in peripheral lymph nodes and in mucosa-associated
`lymphoid tissues such as Peyer’s Patches.
`There is now considerable evidence that lymphocyte homing to HEV is
`mediated by a set of lymphocyte cell surface proteins that interact with
`oligosaccharide structures located on the HEV cell surface. The binding
`of lymphocytes to HEV cells in vitro is inhibited selectively by mannose-
`Es-phosphate and by yeast phosphomannan polysaccharides (Stoolrnan et
`al 1934, Yednoelt et al
`l987a.b). Sialidase treatment of HEV cells also
`perturbs lymphocyte adhesion (Rosen et al 1985], a result that suggests
`that circulating lymphocytes recognize a complex set of oligosaccharides
`on HEV cells. Polyacrylamide surfaces derivatized with carbohydrates
`have been used to define, more precisely, the carbohydrate-binding speci-
`ficity of homing receptors on the lymphocyte surface (Brandley et al 1987).
`Lymphocytes adhere selectively to polyacrylamide gels derivatized with
`either phosphomannans or Fucose sulfate polymers such as fucoidan
`(Brandley et al I987). The binding of lymphocytes to both saccharides
`suggests that distinct classes of homing receptors with differing saccharide
`specificities may exist on a single population of lymphocytes. Phospho-
`mannans have been detected in a variety of mammalian glycoconjugates,
`
`and HEY cells are known to express high levels of sulphated glyco-
`conjugates {Andrews et al 1983), although their identity is unclear.
`The structure ofa murine lymphocyte homing receptor with peripheral
`lymph node specificity has recently been determined (Siegelman et al I989,
`Lasky et al I989). This receptor contains a carbohydrate-binding domain
`with structural homology to the hepatic asialoglycoprotein receptor {see
`
`Page 9 of31
`
`
`
`Page 9 of 31
`
`

`
`234
`
`JESSELL, HYNES at DODD
`
`the receptor contains multiple EGF-like domains.
`below). In addition,
`Collectively. these findings provide strong evidence for the involvement of
`cell surface oligosaocharides and lectins in selective cell recognition by
`lymphocytes.
`
`Glycoprorein Recognition by Hepatic Lectins
`
`The endocytosis of circulating serum glycoproteins by hepatocytes
`revealed one of the first physiological roles for carbohydrate recognition
`in vertebrate species. In rat, the removal of terminal sialic acid residues
`from native serum glycoproteins exposes penultimate galactose residues
`and results in the accelerated clearance of these glycoproteins from the
`circulation {Ashwell 8: Harford I982). This process results from the recog-
`nition of desialylated glycoproteins by an integral membrane receptor on
`the hepatocyte surface, the ASGPLR. The specificity of the rat ASGP-R
`for galactose residues can be demonstrated by blocking the endocytosis
`of glycoproteins by enzymatic alteration of terminal saccharide residues
`(Ashwell & Harford 1982}.
`Studies on the interactions between hepatic lectins and their carbo-
`hydrate ligands have provided important insights into the factors that may
`regulate the affinity of carbohydrate-mediated interactions in vertebrate
`cells. A direct correlation between binding affinity and the degree of sac-
`charide substitution has been demonstrated with model neoglycoproteins
`that vary in the number of saccharides substituted on the protein (Kawa-
`guchi et al 1980, Lee & Lee 1982, Kuhlenschrnidt et al I984, Lehrman et
`al I986). The extent of oligosaccharide branching also significantly affects
`lectin binding affinity. For example. triantennary oligosaccharides exhibit
`a markedly greater inhibitory potency at the ASGP-R than biantennary
`oligosaccharides of the same linear structure (Lee 1989). The degree of
`oligosaccharide branching may therefore be critical in determining the
`aflinity of binding to protein receptors.
`Although both the chick and rat hepatic ASGP-Rs are single subunit
`transmembrane proteins, they exist in the membrane as multimers. The
`active species of the chicken hepatic lectin appears to be a ltexamer (Drick-
`amer 1988). Thus, although the receptor monomer is capable of binding
`to its carbohydrate ligand. the multimeric nature of these receptors in the
`plasma membrane may contribute to their enhanced affinity for ligands
`that contain clusters of terminal sugars {Kuhlenschmidt et al 1984]. The
`multimeric composition of these carbohydrate-binding receptors may also
`extend the range of their carbohydrate-binding capabilities, either by pro-
`ducing a shift in sugar binding specificity andfor by increasing the aflinity
`of binding of these proteins to branched oligosaceharides with different
`backbone structures {Lee 1939).
`
`Page 10 of31
`
`
`
`Page 10 of 31
`
`

`
`DIVERSITY OF CARBOHYDRATE EXPRESSION IN
`THE NERVOUS SYSTEM
`
`CARBOHYDRATE RECOGNITION
`
`235
`
`The analysis ofthe comparatively simple interactions between cells in early
`embryos and in nonneural tissues has provided evidence that surface
`carbohydrates are involved in cell recognition and adhesion. Within the
`nervous system there is no direct evidence that carbohydrates subserve
`similar adhesive or recognition functions. Indirect support for this idea
`has, however, derived from the restricted expression patterns ofcell surface
`carbohydrates, the identification of carbohydrate-binding proteins in sub-
`sets of neurons, and preliminary functional studies on gangliosides and
`glycosyltransferases. The complexity of cell interactions in the nervous
`system has made determination of the precise function of these carbo-
`hydrates and binding proteins substantially more difficult
`than,
`for
`example,
`in sperm-egg interactions or blastomere compaction. Despite
`this, progress has been made in ascribing potential functions to these
`molecules in several regions of the nervous system.
`
`Carbohydrate-Mediated Adhesion in the Retino- Team!
`System
`
`The topographic projection of retinal ganglion cell axons onto the tectal
`surface represents one of the best-studied neural systems with which to
`examine the formation of specific connections in the nervous system.
`Biochemical support for the existence of carbohydrate gradients in the
`retino-tectal system was originally obtained by measuring the adhesion of
`dissociated dorsal or ventral retinal cells to topographically appropriate
`tectal regions in vitro (Barbera et al I973, Barbers 1975, Marchase 1977).
`Marchase (I9?7} established that
`treatment of ventral retinal cells or
`ventral tectum with protease blocked the preferential adhesion of the
`retinal cells to their matching tectal halves, whereas similar protease treat-
`ment of dorsal retinal cells or dorsal tectum did not alter adhesion. ln
`contrast, treatment of dorsal retinal or dorsal tectal cells with N-ace-
`tylhexosatninidase or sialidase resulted in a decrease in specific retino-
`tectal adhesion (Marchase l9'?'i). Marchase (IEJT7) proposed the existence
`of two opposing gradients of complementary molecules: a protease~it1sen-
`sitive molecule containing a terminal 13-N-acetylgalactosamine residue that
`is more concentrated in dorsal retina and tectum, and a protease-sensitive
`molecule that is more concentrated in the ventral retina and tectum.
`Support for this model was obtained with the demonstration that the
`binding of Gm ganglioside to retina and tectum exhibits a ventral-to-
`dorsal gradient, thus suggesting that Gm may be the substrate for the
`ventrally located protease-sensitive molecule (Marchase 1977}. One can-
`
`Page 11 of31
`
`
`
`Page 11 of 31
`
`

`
`236
`
`JESSELL, HYNES at noon
`
`;
`
`didate for the protease-sensitive molecule is therefore a galactosyl trans-
`ferase (GalTase). GM; synthetase, which recognizes Gm and converts it to
`GM, by the addition of a terminal galactose residue. Enzymatic activity
`similar to that of GM. synthetase was detected in a ventral
`to dorsal
`gradient in the retina {Marchase 1977). Functional evidence that gan-
`gliosides may be involved in the specification of retino-tectal projections
`has been provided by independent experiments that revealed selective
`adhesion ofneural retinal cells to immobilized gangliosides, including GM;
`(Blackburn et al I936). These adhesive interactions appear to be specific
`in that they are not detected between neural retinal cells and other charged
`lipids (e.g. sulphatides. phospholipids} or between gangliosides and other
`cell types, such as hepatocytes. In addition, of several gangiiosides tested,
`Gm, G93, and GD,“ supported a greater strength and extent of adhesion
`than Gm” Gm, and Gmb, a result that suggested the presence of a receptor
`on retinal cells that can distinguish between gangtiosides (Blackburn et al
`I986}, Additional evidence that Gm and GM. synthetase contribute to
`specific retino-tectal connections has not emerged. Furthermore, although
`the experiments described above establish the presence within the retino-
`tectal
`system of complementary patterns of potentially interactive
`molecules, a problem in any functional interpretation lies in the fact that
`cell-to-cell or cell-to«substrate adhesion was examined and may not reflect
`the properties of retinal growth cones. Nonetheless, it remains a viable
`idea that a graded distribution of oligosaccharides and carbohydrate-
`binding molecules on the surface of retinal axons interacts with a reverse
`gradient on the tectum to produce specific connections.
`In further support, immunocytochemical studies have revealed selective
`expression patterns of several gangliosides in the retino-tectal system. The
`monoclonal antibody 1838 detects a number of developmentally regulated
`ganglioside species in chicken retina and brain (Dubois et al 1986}. The
`major ganglioside species recognized by antibody [BB8 in retina. GT3, is
`associated with the cell bodies of most neurons in the retina in early
`development but becomes progressively restricted to synaptic layers during
`later development (Grunwald et al I985). Other gangliosides, distinct from
`those detected with monoclonal antibody IBBS, are recognized by Mfitb
`JONES (Constantine-Paton et al 1986, Blum 8: Barnstable I937). The
`JONES antigen is a 9-O-acetylated derivative of the GD; ganglioside and
`is similar or identical to the Dl.l-reactive ganglioside that is expressed in
`the developing rat neuroectoderm (Levine et al 1984). The pattern of
`expression ol'the 9-O-acetylated form of the GD; ganglioside is independent
`of that of the nonacetylated GD; ganglioside recognized with MAb R24
`{see Table 2). In retina and tectum, the JONES antigen is distributed on
`neurons and glia in a dorso-ventral gradient (Constantine-Paton et al
`
`Page 12 of31
`
`
`
`Page 12 of 31
`
`

`
`CARBOHYDRATE RECOGNITION
`
`237
`
`1936), whereas Gm staining appears to be uniformly distributed, thus
`suggesting that the selectivity ofexpression of the acetylated form is depen-
`dent on the spatial restriction of specific biosynthetic or degradation
`enzymes (Blum & Barnstable l937).
`The Jones antigen, Gm, and two other gangliosides with a Jones-reactive
`epitope are found in a number of other developing neural tissues. The
`distribution correlates with the presence of migrating neuroblasts {Meme
`dez-Otero et 2:! I938}. which suggests that the function ofthe Dl.lf.lones
`antigen in the retina and other regions of the embryonic central nervous
`system is in the regulation of early neuroepithelial cell migration. The
`distribution of the Dl.l antigen has been shown to overlap with that of
`fibronectin and the fibronectin receptor (Stallcup I988, Stallcup et al 1989).
`The adhesion of post-natal rat cerebellar cells to fibronectin is inhibited by
`antibodies to the fibronectin receptor, by Arg-Gly-Asp peptides that block
`the recognition of fibronectin by its receptor, and also by antibodies to Di .1
`itself. The DI .l ganglioside may therefore be required to enhance the attach-
`ment to fibronectin of cells expressing the fibronectin receptor. The tem1ina-
`tion of expression of Dl.l on neural epithelial cells at the cessation of
`mitosis may permit post—mitotic neuroblasts to detach from fibronectin
`and migrate away from germinal zones (Stallcup I988, Stallcup et al I989}.
`Recent biochemical and functional studies have extended the original
`observation (Roth et al
`I971) that N-acetylgalactosaminyl-transferase
`(GalNac'l'ase) is found on the surface of embryonic chick neural retinal
`cells (Balsamo et al 1936}. GalNacTase from embryonic chick neural retina
`can be isolated both as a soluble protein and as a particulate complex
`associated with its endogenous acceptor (Balsamo et al I986). The two
`C-alNacTase forms are immunologically cross-reactive but have different
`molecular masses. Under most conditions, the enzyme remains associated
`with its endogenous carbohydrate acceptor {Balsamo et al 1986).
`thus
`suggesting the possibility of a leetin-like function for this enzyme in the
`retina. lmmunochemical analysis, using antibodies that do not distinguish
`the two forms of the enzyme, reveals that at least one form of the enzyme
`is associated with cells throughout the retina in the early embryo but
`becomes restricted to synaptic layers and to the outer segment of the
`photoreceptor in the retina of adult animals. The isolation of both soluble
`and particulate forms of the enzyme from the retina, together with the
`developmental change in its anatomical localization, raises the possibility
`that one form of the enzyme c