The maltoporin of Salmonella typhimurium: sequence and folding model.

The sequence of the lamB gene from Salmonella typhimurium was determined. It encodes the precursor to the LamB protein from S. typhimurium (pre-LamBS.t.; 452 residues) which presents extensive homologies with the pre-LamB protein from Escherichia coli (pre-LamBE.c.; 446 residues). The first third of pre-LamBS.t. is the most conserved, with 4% changes and strict identity between the signal peptides. The last two-third contains five "variable" segments where more than 50% of the residues are changed with respect to LamBE.c.. The three first variable segments are 8 to 14 residues long and contain only substitutions, while the two more distal ones are 24 and 29 residues long and also include insertions and deletions. It is remarkable that the variable segments correspond essentially to regions predicted to be extramembranous loops on our 2D folding model for LamBE.c.; they alternate with conserved predicted transmembranous segments. Four of the variable regions were predicted to be cell-surface-exposed loops on the basis of genetic and immunological data, while one of them (region II) was predicted to be periplasmic on the sole basis of folding rules. The LamB protein from S. typhimurium can substitute for the LamB protein from E. coli for maltodextrins binding and transport, but not for infection by any of the known E. coli phages using LamBE.c. for adsorption. A tetrapeptide, RGDS, assumed to be responsible for mammalian cell aggregation by LamBE.c. is conserved in LamBS.t., suggesting that it could have a functional role. The conservation of the binding and transport activity can be accounted for by the conservation of the regions known to be directly involved, namely the first third of the protein and a region corresponding to 352 to 374 of LamBS.t.. The phage resistance can be attributed to the variability of the four cell-surface-exposed loops previously identified as essential for phage adsorption. These results, together with those obtained with polyclonal and monoclonal antibodies directed against known LamB regions, strongly support the folding model presented for LamBE.c. and the idea that it can essentially be extended to LamBS.t., except perhaps for a region between residues 155 and 245. We propose that the existence of variable regions is due essentially, and perhaps only, to the local lack of structural constraints in the protein. The intergenic region between lamB and the following gene, malM, comprises conserved segments, including one palindromic unit.