From the lysosomal cysteine proteinase cathepsin B, isolated from human liver in its two‐chain form, monoclinic crystals were obtained which contain two molecules per asymmetric unit. The molecular structure was solved by a combination of Patterson search and heavy atom replacement methods (simultaneously with rat cathepsin B) and refined to a crystallographic R value of 0.164 using X‐ray data to 2.15 A resolution. The overall folding pattern of cathepsin B and the arrangement of the active site residues are similar to the related cysteine proteinases papain, actinidin and calotropin DI. 166 alpha‐carbon atoms out of 248 defined cathepsin B residues are topologically equivalent (with an r.m.s. deviation of 1.04 A) with alpha‐carbon atoms of papain. However, several large insertion loops are accommodated on the molecular surface and modify its properties. The disulphide connectivities recently determined for bovine cathepsin B by chemical means were shown to be correct. Some of the primed subsites are occluded by a novel insertion loop, which seems to favour binding of peptide substrates with two residues carboxy‐terminal to the scissile peptide bond; two histidine residues (His110 and His111) in this “occluding loop' provide positively charged anchors for the C‐terminal carboxylate group of such polypeptide substrates. These structural features explain the well‐known dipeptidyl carboxypeptidase activity of cathepsin B. The other subsites adjacent to the reactive site Cys29 are relatively similar to papain; Glu245 in the S2 subsite favours basic P2‐side chains. The above mentioned histidine residues, but also the buried Glu171 might represent the group with a pKa of approximately 5.5 near the active site, which governs endo‐ and exopeptidase activity. The “occluding loop' does not allow cystatin‐like protein inhibitors to bind to cathepsin B as they do to papain, consistent with the reduced affinity of these protein inhibitors for cathepsin B compared with the related plant enzymes.

The topology predictor SPLIT 4.0 (http://pref.etfos.hr) predicts the sequence location of transmembrane helices by performing an automatic selection of optimal amino acid attribute and corresponding preference functions. The best topological model is selected by choosing the highest absolute bias parameter that combines the bias in basic charge motifs and the bias in positive residues (the "positive inside rule") with the charge difference across the first transmembrane segment. Basic charge motifs, such as the BBB, BXXBB, and BBXXB motifs in alpha-helical integral membrane proteins, are significantly more frequent near cytoplasmic membrane surface than expected from the Arg/Lys (B) frequency. The predictor's accuracy is 99% for predicting 178 transmembrane helices in all membrane proteins or subunits of known 3D structure.

No abstract

No abstract