The position of carbohydrates on the cell surface renders them uniquely poised to engage with proteins on the surfaces of other cells or pathogens. Indeed, protein–carbohydrate interactions have been implicated in physiological processes ranging from fertilization to development to immune system function. Given these important functions, it seems surprising that individual protein-carbohydrate interactions are so weak: Their association constants are typically 1000- to 1,000,000-fold poorer than those of protein–protein interactions. To compensate for their low affinity, most protein–carbohydrate interactions are multivalent. Multivalent interactions occur when multiple binding groups (e.g., carbohydrates) on one cell bind to multiple copies of a receptor (e.g., a protein) on another cell. By combining energetic contributions of multiple individual complexes, multivalent interactions are kinetically labile yet exhibit high apparent affinity. These attributes of multivalent interactions complicate the study of protein–carbohydrate interactions. It is challenging to determine if a protein–carbohydrate interaction is relevant, to assess the molecular mechanisms that contribute to formation and stabilization of protein–carbohydrate complexes, and to design potent inhibitors of protein–carbohydrate interactions. To address all of these issues, we developed novel and general synthetic routes to multivalent carbohydrate displays (Fig. 1)
We reasoned polymer chemistry would be a powerful approach to assemble multivalent carbohydrate derivatives. We sought to devise a method to synthesize multivalent ligands with control over length and valency. Most polymerization reactions are incompatible with the polar functional groups of carbohydrates and give rise to mixtures of polymers that vary widely in valency and length. We surmised that the ring opening metathesis polymerization (ROMP) could be ideal for generated defined multivalent carbohydrate displays. ROMP can be a living polymerization, a reaction in which chain elongation occurs more readily than termination. As a result, we can use ROMP to assemble carbohydrate-substituted polymers with defined lengths, valencies, and arrangements of functional groups (Fig. 2). Our first indication of the power of ROMP for making bioactive polymers was our finding that ROMP can be used to generate carbohydrate-substituted polymers that block cell–cell interactions. Our subsequent studies underscore the versatility of this synthetic strategy for generating multivalent carbohydrate derivatives with tailored biological activities.
We used ROMP to create polymers that varied in chemical structure to probe the mechanisms underlying multivalent binding. Multivalent interactions had been thought to be non-selective, but our data challenge this paradigm. We found that multivalent ligands can by highly specific. Moreover, we used our synthetic expertise to carry out the first comparison of how the architecture of multivalent ligand influences its function. One finding to emerge from this investigation is that multivalent ligands generated by ROMP are especially effective at clustering proteins. We subsequently synthesized multivalent carbohydrate displays that inhibit a target carbohydrate-binding protein (L-selectin) by a novel mechanism: they cluster the carbohydrate-binding protein and induce it proteolytic release from the cell surface. This method of altering protein function had been unprecedented and it prompted us to consider using multivalent ligands to explore signaling pathways.