Perspectives Series: Cell Adhesion in Vascular Biology

Less than a decade ago, the cloning of three seemingly disparate vascular cell surface molecules uncovered the selectin family of receptors that are now known to mediate critical cell-cell interactions in processes such as leukocyte trafficking, thrombosis, acute and chronic inflammation, and ischemia-reperfusion injury (1–15). L-, E-, and P-selectin are type-1 membrane glycoproteins sharing a similar structure, comprising an NH 2 terminal calcium-dependent “C type” lectin domain, an EGFlike domain, variable numbers of short consensus repeats domains, a single-pass transmembrane-domain, and a cytosolic carboxyl terminal tail. Abundant evidence indicates that selectins recognize specific carbohydrate ligands. Extensive studies involving mutagenesis, domain swapping, antibody inhibition, etc., have helped to define the functional domains of the selectins. It appears that selectin recognition of carbohydrate ligands involves primarily the amino-terminal C-type lectin domain, influenced to a substantial extent by the EGF-like domain, and to a much lesser degree by the short consensus repeats (1–20). As with most other adhesion receptors, it is also evident that selectins are signaling molecules. This perspective focuses on one of the most puzzling and controversial aspects of selectin biology—the structure of the natural ligands recognized by these receptors. Mammalian lectin domains typically recognize specific oligosaccharide sequences with considerable selectivity, but relatively low affinity. Thus, for example, the cation-independent M6P receptor (CI-M6PR) recognizes mannose 6-phosphate a 1-2 linked to another mannose (21), CD22 recognizes an a 2-6–linked sialic acid residue attached to Gal b 1-4GlcNAc (22), and the selectins share the ability to recognize the tetrasaccharide sialyl-Lewis x (SLe x ) 1 and its isomer sialyl-Lewis a . In the first two instances, K d values for the basic recognition sequence are in the low micromolar range, and higher affinity binding (nanomolar range) is achieved either by multiple binding sites within a single lectin molecule (CI-M6PR) or by molecular aggregation of the monovalent lectin on the cell surface (CD22). This phenomenon of multivalency-based avidity seems to apply to most other naturally occurring lectins (23). With the selectins, the binding affinities for sialyl-Lewis x are very poor, being in the mid micromolar to millimolar range. However, molecular aggregation (e.g., demonstrable by chemical cross-linking) has not been reported for any of the selectins (regional clustering has been reported for L-selectin on the pseudopods of neutrophils). Instead, there are some reports of high affinity binding of recombinant soluble monovalent selectins to natural ligands (16, 17). Thus, the issue of how selectins generate functional high affinity binding remains unresolved (see discussion below). Regardless, the recombinant selectins used to identify biological ligands are multimeric in structure and/or are usually presented in multivalent clustered arrays on solid supports or in soluble complexes. Thus, it should not be surprising that a large number and variety of molecules have been reported to bind to these multimeric recombinant selectins in a glycosylation-dependent manner. To date, very few of these “ligands” have actually been shown to mediate biological interactions. Thus, the current challenge is to tell the difference between what can bind to a recombinant selectin in vitro, and what does bind in vivo to a cell surface selectin in a biologically relevant manner. What should be the criteria for a real selectin ligand? First and foremost, it should be present in the right place at the right time; e.g., a true P-selectin ligand should be present on the surface of a mature circulating blood cell at a time when P-selectin is being actively expressed on an endothelial or platelet surface somewhere in the vasculature. Second, selective removal or blockade of the putative ligand on the intact cell should abrogate biologically relevant interactions. Next, absence of the ligand should abrogate the function of the relevant selectin in the specific situation in question. Finally, the ligand should be recognized with some selectivity by the specific selectin in question, with a relatively high affinity, and preferably with a well-defined stoichiometry. Of course, some of these criteria might not be fulfilled if other relevant ligands share partial functional redundancy and/or overlap in recognition by other selectins. Let us consider as an example the potential ligands on the surface of a circulating neutrophil for P-selectin (which is itself expressed on activated endothelium or platelets). A large body of evidence indicates that sugar chains called sialylated fucosylated lactosamines (of which SLe x is the simplest example) are critical components of the ligands (1–14). The relevant fucosylAddress correspondence to Ajit Varki, Glycobiology Program, UCSD Cancer Center, and Division of Cellular and Molecular Medicine, University of California, San Diego, La Jolla, CA 92093. Phone: 619-534-3296; FAX: 619-534-5611; E-mail: [email protected] Accepted for publication 3 December 1996.