Porins are membrane proteins of the outer membrane of Gram-negative bacteria, with an b-barrel architecture (1, 2). Although much more is known about bacterial porins, there are functionally and structurally similar proteins, known as voltage-dependent anion channels (VDACs), that are believed to be present in the outer membrane of mitochondria (3). Known porins are composed of 250 to 450 amino acid residues. Their function is to facilitate diffusion of small molecules across the membrane by allowing solutes to pass through an aqueous channel in the middle of the transmembrane b-barrel. Some porins are nonspecific and permeable to any solutes smaller than 600 Da, but they can be cation- or anion-selective, favor polar solutes over nonpolar ones, or have specific substrates, such as maltodextrins (a1-4-polyglucose) or sucrose. The rate of transport in the nonspecific porins is a linear function of the concentration gradient of the solute. In contrast, the specific porins follow Michaelis-Menten kinetics, indicating initial binding of the solute.
The porins for which three-dimensional protein structures have been determined are homotrimers containing three identical transmembrane channels. The OmpA protein of Escherichia coli is thought to be a monomeric porin with a barrel comprised of eight b-strands. The membrane-bound form of a-hemolysin, a bacterial toxin, belongs to the same structural class, because its transmembrane region comprises a 14-stranded antiparallel b-barrel (4). A similar structure has been predicted for the channel assembled by the bacterial toxin aerolysin in its heptameric membrane-bound form.
The size of the pore is determined by one or more extracellular surface loops of polypeptide chain that fold back into the channel. Although the 18-stranded b-barrel of glycoporins is clearly wider than the 16-stranded barrel of general porins, the latter have a wider channel because their intrachannel loop structures are less extensive. The width of the channel mouth (the eyelet) of the general porins is determined by one long and structured loop between adjacent b-strands, which controls access to the channel. In some porins, additional control is provided by a transverse electric field, which is generated by an uneven distribution of positively and negatively charged residues inside the channel mouth. In maltoporins and sucrose porin, the channel mouth is constricted by three surface loops. For this reason, it is much narrower than the eyelet in the nonspecific porins. There is an extensive binding site or an aromatic path for the sugar rings to be translocated (5-8). Suitably positioned partners for hydrogen bonds to the sugar hydroxyl groups assist the translocation process.
Porins have a number of peculiar properties. First, their sequences are at least as hydrophilic as are those of soluble proteins. Yet, the lipid-exposed surface of the b-barrel is highly hydrophobic, and the proteins are only soluble in the presence of a detergent. Second, the primary structures of porins form at least 10 families that show no clear sequence homology with each other. It appears that highly diverse primary structures can fold into very similar three-dimensional structures, as shown by the eleven porin structures presently known at atomic resolution (see Table 1 of Membrane Proteins). The sequences of the mutually homologous nonspecific porins of E. coli (OmpF and PhoE) are not related to those of the general diffusion porins from the phototrophic bacteria Rhodobacter capsulatus and Rhodopseudomonas blastica, or from Paracoccus denitrificans, despite the fact that all five fold into 16-stranded b-barrels. The maltoporin family has an 18-stranded topology and is not related to the nonspecific porins. Third, porins are unusually stable, forming trimers that are resistant toward denaturation by SDS, even at elevated temperatures. Perhaps owing to this high stability, porins appear to be relatively easy to crystallize, as shown by the existence of a number of high-resolution structures.
In fact, the first well-ordered crystals of a membrane protein were those of the E. coli OmpF porin grown in the late 1970s in Jurg Rosenbusch’s laboratory in Basel. Owing to an unfavorable crystal symmetry, however, this crystal structure was not determined until 1995.
The diverse nature of porin primary structures, along with their hydrophilic nature, makes the assignment of a novel sequence as a porin problematic (see Homology modeling). The hydrophobic transmembrane b-barrel is composed of b-strands whose lengths range from 6 to 17 residues. Moreover, only every second residue of each strand faces the lipid core of the membrane and is consistently hydrophobic. Owing to the fluid nature of the bilayer core, only the hydrophobic nature of the lipid-facing residues needs to be conserved. The rest of the residues in the barrel can be either hydrophilic or hydrophobic, depending on whether they are exposed to the aqueous pore or buried in the protein structure. As with helical and monotopic membrane proteins, the strands are rich in aromatic residues in the region that interacts with the lipid headgroups (see a-Helix). A protein structure prediction method based on the detection of these interfacial aromatic residues and on the "every second residue hydrophobic" pattern correctly assigned 16 out of the 18 strands in maltoporin (9).
Because porins reside in the outer bacterial membrane, they must first be translocated through the bacterial inner membrane to the periplasm. To this end, the porin sequences carry an N-terminal signal sequence. The subsequent folding pathway is not known with certainty, but it has been proposed that the folding of porins begins in the aqueous periplasm by the formation of the trimer interface (2). The interface resembles the hydrophobic core of soluble proteins and could, therefore, form spontaneously before insertion into the membrane. The lipid phase would then induce the formation of the three b-barrels, because the barrel both satisfies the hydrogen-bonding requirement of the peptide groups and uses the hydrophobic effect in the interaction with the oily core of the bilayer. A trimeric porin that is indistinguishable from the natural molecules has been produced by heterologous expression by protein engineering, followed by protein folding in vitro from inclusion bodies in the presence of detergents.
Much of the biological interest in porins stems from the observation that in pathogens porins often are the antigens against which the host’s antibodies are directed. Thus, porins from pathogenic bacteria could be used to raise antibodies and make vaccines. Maltoporin of E. coli is also called LamB, because it is the receptor for lambda phage.
