SummaryThe trimeric protein LamB of Escherichia coli K-12 (maltoporin) speci®cally facilitates the diffusion of maltose and maltooligosaccharides through the outer membrane. Each monomer consists of an 18-stranded antiparallel b-barrel with nine surface loops (L1 to L9). The effects on transport and binding of the deletion of some of the surface loops or of combinations of several of them were studied in vivo and in vitro. In vivo, single-, DL4, DL5, DL6, and double-loop deletions, DL4 DL5 and DL5 DL6, abolished maltoporin functions, but not the double deletion DL4 DL6 and the triple deletion DL4 DL5 DL6. While deletion of the central variable portion of loop L9 (DL9v) affected maltoporin function only moderately, the combination of DL9v with the double deletion of loops L4 and L6 (triple deletion DL4 DL6 DL9v) strongly impaired maltoporin function and resulted in sensitivity to large hydrophilic antibiotics without change in channel size as measured in vitro. In vitro, the carbohydrate-binding properties of the different loop mutants were studied in titration experiments using the asymmetric and symmetric addition of the mutant porins and of the carbohydrates to one or both sides of the lipid bilayer membranes. The deletion of loop L9v alone (LamBDL9v), of two loops L4 and L6 (LamBDL4 DL6), of three loops L4, L5 and L6 (LamBDL4 DL5 DL6) or of L4, L6 and L9v (LamBDL4 DL6 DL9v) had relatively little in¯uence on the carbohydrate-binding properties of the mutant channels, and they had approximately similar binding properties for carbohydrate addition to both sides compared with only one side. The deletion of one of the loops L4 (LamBDL4) or L6 (LamBDL6) resulted in an asymmetric carbohydrate binding. The in vivo and in vitro results, together with those of the puri®cation across the starch column, suggest that maltooligosaccharides enter the LamB channel from the cell surface side with the non-reducing end in advance. The absence of some of the loops leads to obstruction of the channel from the outside, which results in a considerable difference in the on-rate of carbohydrate binding from the extracellular side compared with that from the periplasmic side.