Advanced protein structure prediction methods combined with structure modeling show that the mammalian proteins, described until now as calcium-activated chloride channels (CLCAs), appear in fact to be membrane anchored metal-dependent hydrolases, possibly proteases. A metallohydrolase structural domain was predicted, unexpectedly, in the CLCA sequences. The well-conserved active site in the modeled structure of this hydrolase domain allows the prediction of catalytic action similar to that of metalloproteases. A number of protein structure prediction methods suggest the overall fold of the N-terminal hydrolase domain to be most similar to that of zinc metalloproteases (zincins), notably matrixins. This is confirmed by analysis of the three-dimensional structure model of the predicted CLCA1 hydrolase domain built using the known structure of the MMP-11 catalytic domain. Fragments of CLCA1 corresponding to the modeled hydrolase domain were expressed in Escherichia coli, and the resulting proteins were readily refolded into monomeric soluble protein, indicating formation of stable independent domains. The homology model was used to predict putative substrate sequences. Homologs of mammalian CLCA genes were detected in the genomes of a vast array of multicellular animals: lower vertebrates, tunicates, insects, crustaceans, echinoderms, and flatworms. The hydrolase prediction is discussed in the context of published experimentally determined effects of CLCA proteins on chloride conductance. Altered proteolytic processing of full-length CLCA1 containing a mutation abolishing the predicted hydrolase activity is shown as initial experimental evidence for a role of the hydrolase domain in processing of mature full-length CLCA1. The hydrolase prediction together with the presented experimental data add to doubts about the function of CLCAs as chloride channels and strengthen the hypothesis of channel-activating and/or channel-accessory roles.
(c) 2006 Wiley-Liss, Inc.