The eosinophil cationic protein (ECP)' is one of several small, distinct argininerich proteins that have been isolated from the eosinophil's large specific granule. The molecular mass of human ECP has been estimated at 18-21 kD (1-3), depending on the degree of glycosylation (3). ECP has been shown to possess a wide variety of biological activities, including the ability to stimulate factor XII-dependent coagulation pathways (4), to neutralize the anticoagulant effects of heparin (5), and to inhibit lymphocyte proliferation induced by PHA or mixed lymphocyte reactions (6) . ECP is a potent cytotoxin; ECP-containing supernatants of activated eosinophils have been shown to be toxic to isolated myocardial cells in vitro (7), and both ECP and the eosinophil granule major basic protein (MBP) were found in the endothelial and endomyocardial lesions characteristic of the hypereosinophilic syndrome (8). Furthermore, ECP is also a potent helminthotoxin; destruction of schistosomula of Schistosoma mansont was reported at concentrations as low as 10' M (8-10-fold more active than MBP) (9, 10) . ECP has also been shown to kill trypomastigote and amastigote stages of Trypanosoma cruzi (11) . Both ECP and the related granule protein, eosinophil-derived neurotoxin (EDN), induce the neurotoxic effect known as the Gordon phenomenon (12)

The role(s) of the eosinophil Charcot-Leyden crystal (CLC) protein in eosinophil or basophil function or associated inflammatory processes is yet to be established. Although the CLC protein has been reported to exhibit weak lysophospholipase activity, it shows virtually no sequence homology to any known member of this family of enzymes. The X-ray crystal structure of the CLC protein is very similar to the structure of the galectins, members of a beta-galactoside-specific animal lectin family, including a partially conserved galectin carbohydrate recognition domain (CRD). In the absence of any known natural carbohydrate ligand for this protein, the functional role of the CLC protein (galectin-10) has remained speculative. Here we describe structural studies on the carbohydrate binding properties of the CLC protein and report the first structure of a carbohydrate in complex with the protein. Interestingly, the CLC protein demonstrates no affinity for beta-galactosides and binds mannose in a manner very different from those of other related galectins that have been shown to bind lactosamine. The partial conservation of residues involved in carbohydrate binding led to significant changes in the topology and chemical nature of the CRD, and has implications for carbohydrate recognition by the CLC protein in vivo and its functional role in the biology of inflammation.

No abstract

Purkinje cell toxicity is one of the characteristic features of the Gordon phenomenon, a syndrome manifested by ataxia, muscular rigidity, paralysis, and tremor that may lead to death (Gordon, 1933). Two members of the RNase superfamily found in humans, EDN (eosinophil-derived neurotoxin) and ECP (eosinophil cationic protein), cause the Gordon phenomenon when injected intraventricularly into guinea pigs or rabbits. We have found that another member of the RNase superfamily, an antitumor protein called onconase, isolated from Rana pipiens oocytes and early embryos, will also cause the Gordon phenomenon when injected into the cerebrospinal fluid of guinea pigs at a dose similar to that of EDN (LD50, 3-4 micrograms). Neurologic abnormalities of onconase-treated animals were indistinguishable from those of EDN-treated animals, and histology showed dramatic Purkinje cell loss in the brains of onconase-treated animals. The neurotoxic activity of onconase correlates with ribonuclease activity. Onconase modified by iodoacetic acid to eliminate 70% and 98% of the ribonuclease activity of the native enzyme displays a similar decrease in ability to cause the Gordon phenomenon. In contrast, the homologous bovine pancreatic RNase A injected intraventricularly at a dose 5000 times greater than the LD50 dose of EDN or onconase is not toxic and does not cause the Gordon phenomenon. A comparison of the RNase activities of EDN, onconase, and bovine pancreatic RNase A using three pancreatic RNA substrates demonstrates that onconase is orders of magnitude less active enzymatically than EDN and RNase A. Thus, another member of the RNase superfamily in addition to EDN and ECP can cause the Gordon phenomenon.(ABSTRACT TRUNCATED AT 250 WORDS)