Outer Membrane
Chapter
5
HIROSHI NIKAIDO
The outer membrane (OM) of gram-negative bacteria has been studied in many laboratories, and this chapter can give only a brief outline of our current knowledge. More details can be found in the more extensive, but somewhat outdated, reviews published about 10 years ago (103, 142). Two books published in 1979 and 1987 (76, 77) also deal with various aspects of bacterial OMs. More recent reviews on special topics are cited below.
Electron-microscopic examination of thin sections of gram-negative bacteria showed already in the 1960s that these cells are covered by an extra membrane layer, OM, which is located outside the peptidoglycan layer. The OM was then isolated in several laboratories (119, 144, 164) from Escherichia coli and Salmonella typhimurium cell lysates by sucrose equilibrium density centrifugation, which took advantage of the fact that the OM has a higher buoyant density (close to 1.22) than the cytoplasmic or inner membrane with a density of 1.15.
Equilibrium density centrifugation still remains the most reliable method for the isolation of OM. The disruption of the cells can be carried out either by EDTA-lysozyme lysis (144) or with a French pressure cell (164, 176). The latter method can be used on any strain, but the recovery of inner membrane may become poor because it is converted into very small vesicles. Sonication can be used if done judiciously, but extensive treatment should be avoided because it may result in the formation of hybrid vesicles.
When one is interested only in protein composition, extraction of cell envelope with mild detergents such as sodium dodecyl sarcosinate (Sarkosyl) can be used because these detergents extract preferentially the cytoplasmic membrane proteins (52). Conclusions based on this method, however, must be interpreted with care, because some OM and inner membrane proteins may show atypical behavior.
Procedures for the isolation of OM have been reviewed recently (132).
The OM contains two types of lipids, lipopolysaccharide (LPS) and phospholipids, as well as a set of characteristic proteins (Table 1). In addition, the OM of Enterobacteriaceae contains a unique polysaccharide, enterobacterial common antigen (ECA).
Table 1Major components of the outer membranea |
The phospholipid composition of the OM is similar to that of the cytoplasmic membrane, with a slight enrichment in phosphatidylethanolamine (144).
LPS is a unique constituent of the bacterial OM. The following points are important. (i) LPS is composed of three parts: the proximal, hydrophobic lipid A region; the distal, hydrophilic, O-antigen polysaccharide region that protrudes into the medium; and the core oligosaccharide region that connects the two (Fig. 1). (ii) The lipid A is a polar lipid of unusual structure, in which a backbone of glucosaminyl-β-(1→6)-glucosamine is substituted with six or seven fatty acid residues, all of them saturated. (iii) The proximal part of the core and the lipid A backbone contain a large number of charged groups, most of them anionic. (iv) The properties of mutants producing incomplete LPS molecules suggest the biological functions performed by the various parts of the LPS molecule. Loss of O antigen results in loss of virulence, suggesting that this portion is important in escaping phagocytosis. (Laboratory strains of E. coli such as K-12 and B completely lack the O antigen, with the B strain lacking also the distal part of the core.) Loss of the more proximal part of the core, as in the "deep rough" mutants (i.e., in Rd1, Rd2, and Re chemotypes [Fig. 1]), makes the strains exceptionally sensitive to a wide range of hydrophobic compounds including dyes, antibiotics, bile salts, other detergents, and mutagens (see reference 198); thus this area is involved in maintaining the barrier property of the OM (see below). Only conditional mutants in lipid A biosynthesis are available; this part of the LPS is presumably essential for the assembly of the OM.
Important advances in this area during the past decade or two include the resolution of naturally occurring, heterogeneous LPS into individual components, followed by the determination of their structure (186), complete chemical synthesis of lipid A and its analogs (88), elucidation of the steps of lipid A biosynthesis (152), and studies on the molecular genetics of LPS biosynthesis (153, 166). The details of these advances will be described elsewhere in this book (chapter 69).
ECA is an acidic polysaccharide containing N-acetyl-d-glucosamine, N-acetyl-d-mannosaminuronic acid, and 4-acetamido-4,6-dideoxy-d-galactose (102). It represents up to 0.2% of the dry weight of E. coli (112) and is anchored to the OM through a covalently linked phospholipid moiety (87) in most strains. There has been much recent progress in the molecular genetics of ECA synthesis (115).
Two types of capsular polysaccharides are found in enteric bacteria. The M (mucous) antigen, or colanic acid of an identical structure, is produced by most enteric bacteria, presumably as a means of protection against desiccation. In contrast, K-antigen polysaccharides have structures specific to each serotype within a species and presumably help these bacteria evade phagocytosis based on specific recognition of surface molecules. These polysaccharides are discussed in detail elsewhere in this book (chapter 9).
The protein pattern of the OM is dominated by a few major proteins that fall into the classes described below.
Murein Lipoprotein.
Murein lipoprotein is a small (7,200-Da) protein that exists in a large number of copies, 7 × 105 per cell. The N-terminal residue, cysteine, is modified at two places. Its sulfhydryl group is substituted with a diglyceride, and its amino group is substituted by a fatty acyl residue. These three fatty acid groups are thought to penetrate into the inner leaflet of the OM (Fig. 2). The rest of the protein, mostly in α-helical form, is hydrophilic and is most likely located in the periplasm. About one-third of the murein lipoprotein molecules are bound covalently to the peptidoglycan layer through the ε-amino group of its C-terminal lysine, thus fixing the OM to the underlying peptidoglycan.
Consistent with this structure, mutants with deletions through the lipoprotein gene (lpp) produce unstable OMs, resulting in the release of OM vesicles and periplasmic enzymes into the growth medium (71). Their OM is unaltered in its permeability to hydrophilic solutes (134). Murein lipoprotein is discussed in more detail in chapter 66.
Classical Porins.
The proteins coded for by ompF, ompC, and phoE genes in E. coli and the homologous genes plus ompD in S. typhimurium are trimeric porins that produce relatively nonspecific pores or channels that allow the rapid passage of small hydrophilic molecules across the OM (Table 2). Among these porins, the PhoE protein is unique in that it is produced only under conditions of phosphate starvation. Thus, in the usual culture media, only the OmpF and OmpC (and in S. typhimurium also OmpD) porins are produced, but the relative abundance of these porins is under an efficient regulation by environmental signals (see below). The total amount of these porins present is relatively constant and is very large (Table 2), making porins some of the most abundant proteins in E. coli and S. typhimurium in terms of mass; they can represent up to 2% of the total protein of the cell. The structure and function of porins have been reviewed (11, 36, 130, 169).
Table 2Classical porins of E. coli K-12 and S. typhimurium LT2a |
When the OM-peptidoglycan complex is extracted with sodium dodecyl sulfate (SDS) at temperatures below 60°C, much of the porin is left behind in the insoluble fraction (156). Porins can be solubilized as undenatured, tightly associated trimers by including 1 M NaCl in the SDS solution. It requires heating in SDS at temperatures above 70°C to denature the porins and dissociate them into monomeric subunits (see reference 142).
Porins are not hydrophobic proteins in terms of amino acid composition, and do not contain any long stretches of hydrophobic amino acids as typically found in integral proteins of the plasma membrane. These observations are now explained by the three-dimensional structures of porins, elucidated by electron (78) and X-ray (37, 208) crystallography. Weiss, Schulz, and coworkers (209) solved the structure of the trimeric porin from Rhodobacter capsulatus at 1.8 Å (0.18 nm) resolution, and this was soon followed by the solution of E. coli OmpF and PhoE porin structures by Cowan and coworkers (37). Although OmpF was an early example of an intrinsic membrane protein to be crystallized (59), it was difficult to solve its structure apparently because crystal forms with unfavorable symmetry properties were used. Recently the structure of Rhodopseudomonas blastica porin was also determined (86). There is little sequence homology between the E. coli porins and that of R. capsulatus, yet all these trimeric porins fold in a pattern remarkably similar to each other (Fig. 3). The polypeptide chain of each subunit traverses the membrane 16 times as antiparallel β-strands, forming a β-barrel structure surrounding a large channel. The β-strands are tilted by 30 to 60° in relation to the trimer axis. Each subunit produces a channel, and the trimer therefore contains three channels. Because every other amino acid side chain protrudes into the hydrophilic channel and therefore tends to be hydrophilic, there is no stretch of consecutive hydrophobic residues in transmembrane β-strands; rather, there is a tendency to have hydrophobic and hydrophilic residues in alternating positions. Among the hydrophobic residues facing the lipid bilayer, there is a strong preference for aromatic amino acids close to the surfaces of the bilayer.
The loops connecting membrane-traversing β-strands are quite short (one to four residues) on the periplasmic side, but they are often long on the external side. External loops show much variability in sequence among porins from various enteric bacteria. Many external loops contain a large number of charged, mostly acidic, amino acid residues, which are thought to interact strongly with the negatively charged groups of LPS, presumably through divalent cation bridges. Most importantly, in each of these porins the pores are narrowed by the inward folding of the third external loop (L3) into the lumen of the β-barrel (Fig. 3). Thus the solute-discriminating property of the porin channel is influenced greatly by this infolding loop, rather than by the overall dimensions of the β-barrel. The loop produces a narrowing of less than 10 Å in depth ("eyelet"), with a cross-section of 8 × 10 Å (R. capsulatus) or 7 × 11 Å (E. coli OmpF) (Fig. 3B). The latter value is in excellent agreement with the size determined from the relative diffusion rates of sugars, mentioned below. This design produces a channel with a wide entrance, a wide exit, and a short central constriction. Such a channel with a narrow bottleneck is effective in excluding larger solutes while minimizing the frictional interactions between the solutes and the walls of the pore. Thus small molecules can diffuse across the OM with near-maximal rates.
The three-dimensional structures of porins provide explanations for their other functional properties. For example, PhoE porin shows a preference for anions, in contrast to the cation preference of OmpF. Site-directed mutagenesis showed that the most important residue for this alteration in charge preference was Lys-125 of PhoE, which replaces Gly-131 of OmpF (3). These residues are located in the eyelet, where a strong influence on ion selectivity is expected (37) (Fig. 4).
All three porins of E. coli are known to exclude lipophilic solutes (see above). This was explained by Schulz (169) as follows. The eyelet contains several negatively charged residues on one side and several positively charged residues on the opposite side (Fig. 4). Because of the electrostatic forces, the side chains of these charged residues are maximally extended, and thus this arrangement produces a structure of well-defined and rigid shape, a necessity for a pore that would exclude larger solutes efficiently. In addition, this high electric field will make the entry of nonpolar molecules energetically unfavorable because they have to displace water molecules arranged to minimize their potential energy (see Fig. 4).
Although the structure of OmpC has not been elucidated, the major difference in OmpC sequence is the presence of a 15-residue insertion in the external loop 4 of OmpF. Possibly the narrower channel size of OmpC (see below) is produced by the infolding of this loop, in addition to the infolding of loop 3 already present in OmpF.
Mutant forms of E. coli porins producing larger channels were isolated by selecting for strains that are able to grow on maltodextrin oligosaccharides in the absence of the specific channel, LamB. Remarkably, most of the point mutants alter one of the charged residues that form the electrostatically rigidified eyelet mentioned above (37), changing Arg residues at 42, 82, and 132 in OmpF (or corresponding Arg residues at positions 37, 74, and 105 in OmpC) into uncharged residues, or changing Asp-113 in OmpF (or corresponding Asp-105 in OmpC) into an uncharged residue (10, 117). These results are consistent with the hypothesis that electrostatic extension of charged side chains is essential for rigidity of the narrow constriction. As expected, deletions within the eyelet-forming loop (L3) also produce widening of the channel (37).
The center of the porin trimer, the intersubunit region, is completely hydrophobic. Schulz (169) suggests that this feature may be important in the assembly and folding of porin trimers. Thus monomeric porins may form an immature trimer in the aqueous phase, driven by the aggregation of hydrophobic parts of the subunits to produce the central intersubunit region. The rest of the monomeric unit is imagined to exist as a giant loop containing about 200 residues. This hydrophilic trimer, upon insertion into the outer membrane, will refold to produce the rest of the transmembrane β-strands. Since the C-terminal phenylalanine residue of porins occupies an important position in this central core, perhaps we can understand why the deletion or substitution of this residue prevents the trimerization (and also OM insertion) of PhoE protein (180).
Some strains of E. coli are known to produce alternative or additional porins. Thus several clinical isolates of E. coli contain an additional porin, called protein 2 or Lc, which is encoded by a prophage genome (19, 151). On the chromosome of E. coli K-12, there is a defective prophage containing a gene for nmpC porin. This porin, which differs from Lc porin only in four amino acid residues, is not normally expressed because of an insertion of the IS5 element (19). Selection for pore-containing cells in an ompF – background resulted in the discovery of a normally repressed porin, OmpG, in E. coli K-12 (118). Encapsulated strains of E. coli often produce a characteristic porin, protein K (184, 210). Sequences of porins have been compared (79).
OmpA Protein.
The OmpA proteins have monomer molecular weights (35,159 in E. coli K-12) similar to those of porins, but they are extracted as monomeric proteins by SDS at low temperature. SDS, however, does not denature OmpA completely unless the temperature is increased to 100°C. Thus the mobility of OmpA protein in SDS-polyacrylamide gel electrophoresis decreases upon its denaturation caused by heating of the sample. This behavior, often referred to as "heat-modifiability," is frequently observed with OM proteins that are not easily denatured by SDS (165). SDS probably mimics LPS, the lipid that normally interacts with OM proteins, because both SDS and LPS contain one (or about one) negative charge per hydrocarbon chain.
The number of OmpA molecules per cell approaches 105. There is little evidence that it exists as homooligomers or as stoichiometric complexes with other proteins.
The N-terminal domain of OmpA is believed to fold into a transmembrane β-barrel, reminiscent of trimeric porins (described above) but containing only eight antiparallel β-strands (122, 204). A convincing evidence for this structure is the demonstration that all of the mutations that diminish the ability of OmpA to serve as a phage receptor occur in the four external loops predicted by the model (122). This transmembrane domain is then followed by an Ala-Pro-Val-Val-Ala-Pro-Ala-Pro-Ala-Pro-Ala-Pro sequence at residues 176 through 187, a sequence resembling the "hinge" region of immunoglobulins. This region is in turn followed by the C-terminal domain, presumed to reside in the periplasmic space. Indeed, proteases cleave native OmpA near or in the hinge region, leaving the N-terminal transmembrane domain intact (170).
Mutants lacking OmpA protein are extremely poor recipients in conjugation, and they tend to produce spherical cells with unstable OM, especially when combined with defects in murein lipoprotein. OmpA is thus thought to have a role in stabilizing the structure of the OM (reviewed in reference 142). Earlier, it was suggested that OmpA is involved in the uptake of amino acids and peptides, but the probable reason for this result is that populations of ompA mutants tend to contain a large fraction of nonviable cells. More recently, purified OmpA was found to produce nonspecific diffusion channels of about the same size as OmpF (181). The pore-forming activity, as well as the size of the channel, was confirmed by reconstitution into planar lipid membrane (160). (Although the authors conclude that the pore diameter was only 0.7 nm, the single channel conductance observed, 0.6 nS in 1 M NaCl, is somewhat larger than that of the single OmpF channel, 0.4 nS, obtained by dividing by 3 the conductance for the trimer, 1.2 nS [13].) The penetration of solutes through the OmpA channel, however, is about two orders of magnitude slower than that through the OmpF channel, and thus OmpA does not make a significant contribution to the permeability of E. coli OM. In contrast, Pseudomonas aeruginosa OM is devoid of any OmpF/OmpC homolog and instead expresses an OmpA homolog, OprF, as the major nonspecific porin. OprF is also monomeric and shows very low permeability although its pore size is quite large (8, 134). This explains the low general permeability of P. aeruginosa OM.
The reason why these monomeric porins, OmpA and OprF, produce such low rates of penetration has remained an enigma. With OmpA, however, experimental evidence has been obtained recently that only a small fraction of the protein population (2 to 3%) folds in a manner that results in functional channels (182).
Specific Channels.
The proteins forming specific channels in E. coli and S. typhimurium are listed in Table 3. The LamB protein (phage λ receptor) of E. coli is responsible for allowing the passage of maltose and maltodextrins across the OM. Although S. typhimurium is not sensitive to phage lambda, an OM protein similar to LamB has been reported (167). The LamB protein is a porinlike trimeric protein predicted (51), and recently shown (163a), to be constructed as a β-barrel.
Table 3OM proteins involved in specific transporta |
The phage T6 receptor protein of E. coli, coded for by gene tsx, is an OM protein involved in the specific diffusion of nucleosides. The sequence of the protein reveals no long hydrophobic stretches (25), and the protein therefore is likely to produce a β-barrel as well. Purified Tsx protein was shown also to produce a channel with a specific binding site for nucleosides in black lipid film reconstitution assay (107).
A third representative of the specific channel in E. coli is the ScrY protein, which is encoded by a plasmid gene and produces sucrose-specific channels (168). This protein was crystallized in a form that diffracts up to 2.3 Å (53).
High-Affinity Receptors in OM.
High-affinity receptor proteins also catalyze specific transport of ligands across the OM (Table 2), but are different in several respects from the specific channels described above. First, these proteins bind the ligands much more tightly and hence are called "receptors." For example, BtuB binds vitamin B12 with a Kd of 3 nM. Second, the transport of ligands through these proteins requires the presence of TonB, a protein anchored in the cytoplasmic membrane(149). Finally, at least the BtuB-TonB system is known to accumulate its ligand (vitamin B12) in the periplasm at concentrations at least 1,000-fold higher than the external concentration, using proton motive force across the cytoplasmic membrane as the energy source(23). These properties are suited for the uptake of ligands that are too large for passage through the porin channels and exist in extremely low external concentrations.
One model of TonB-dependent transport assumes that the TonB protein extends through the periplasmic space and physically interacts with OM receptors (149). A conserved sequence called the "TonB box" is present in all of the receptors, and a pentapeptide corresponding to this sequence inhibits TonB-dependent transport functions at remarkably low concentrations (down to 1 μg/ml) (193). Furthermore, TonB can be chemically cross-linked to the FhuA receptor (149). These data support this physical interaction model. On the other hand, TonB function was not impaired (in at least one assay) by deletion of the presumed rigid "arm" that is thought to extend through the thickness of the periplasm (91).
An internal deletion of fepA gene, coding for the enterobactin receptor, converted FepA protein into a large pore (97, 159). A similar finding was reported for FhuA, the ferrichrome receptor (82). These results suggest that the receptors are also basically open channels which are closed by the presence of the additional, ligand-binding domain. Such a model, however, is undoubtedly an oversimplification, and perhaps a more dynamic feature is needed to explain the high degree of accumulation of ligands achieved by these proteins (149).
FadL protein is known to catalyze the diffusion of long-chain fatty acids across the OM (18). FadL has a high affinity for ligands, yet this transport process does not appear to require the participation of TonB protein, in contrast to the other high-affinity receptors.
Proteins Involved in Direct Import/Export of Proteins and Possibly Drugs.
A minor 50-kDa OM protein, TolC, known to be involved in the entry of some colicins into E. coli cells (207), was found to serve as a channel for the direct export of hemolysin into the medium, together with an accessory factor, HlyD, and the inner membrane transporter HlyB (206). TolC produces ion-permeable channels upon reconstitution (14).
Hemolysin and other proteins secreted through TolC (and its homologs, see below) cross the cytoplasmic membrane without going through the Sec-dependent pathway. There are, however, many secreted proteins of gram-negative bacteria that utilize the Sec-dependent pathway, also called the general secretory pathway, and then eventually get secreted into the extracellular space (reviewed in reference 150). In E. coli, components of the P pilus of uropathogenic strains are examples of this class and are secreted through an outer membrane protein, PapC, which appears to recognize specifically the various subunits of this pilus as complexes with the periplasmic chaperonin (44). Another branch of the general secretory pathway for secretion across OM, the "main branch," utilizes PulD of Klebsiella oxytoca and its homologs; many proteins are secreted through this branch by a variety of bacteria, but this branch is usually absent in E. coli, Salmonella spp., or Shigella spp. (150). Nevertheless, this branch may be used by exogenous genetic elements. Thus, filamentous phages of E. coli insert their protein IV, a PulD homolog, into the outer membrane most probably for phage export (158), and Shigella flexneri secretes plasmid-coded invasins through MxiD, another PulD homolog (reviewed in reference 150).
Because the outer surface of the OM constitutes the outermost area of the cell, it has various properties that are characteristic of the cell surface of all unicellular organisms. These properties include the presence of hydrophilic, usually negatively charged carbohydrate chains that allow the organisms to avoid surface phagocytosis. Clearly the O-antigen chains of LPS and capsular polysaccharides fulfill this requirement, and most of the major OM proteins are acidic proteins. Furthermore, as is common in microorganisms with extensive contact with higher animals, the surface structure has undergone extensive structural diversification, as seen in the variety of O-antigen and capsular polysaccharide structure; this presumably helps in avoiding attack by antibodies and digestive enzymes.
Asymmetric Bilayer.
Asymmetric Bilayer. Many pieces of evidence indicate that the lipid bilayer forms the basic continuum of the OM (142). However, the distribution of lipids in the bilayer is highly asymmetrical. Most of the LPS molecules are located in the outer leaflet of the bilayer, as shown by electron microscopy after antibody labeling (123) and by enzymatic modification of LPS in intact cells (57). Furthermore, as the number of hydrocarbon chains in phospholipids is about equal to that in LPS in the OM (Table 1), most of the inner leaflet must be occupied by phospholipids and there should be few, if any, phospholipid molecules in the outer leaflet (176). This hypothesis was tested (81) by treating intact S. typhimurium cells with an impermeable reagent, CNBr-labeled dextran. There was no labeling of phosphatidylethanolamine, which should have occurred if some of these molecules were located in the outer leaflet. However, in deep rough mutants significant amounts of phosphatidylethanolamine were found in the outer leaflet (see Fig. 5B and C) and were efficiently labeled. It should be emphasized, however, that phospholipids may be present in the outer leaflet in species not closely related to E. coli or S. typhimurium (146).
Physical Properties of the LPS Leaflet.
The outer leaflet containing LPS alone must be instrumental in lowering the permeability of OM to lipophilic solutes (see below). In order for lipophilic solutes to traverse any lipid bilayer, they must first enter the hydrocarbon interior of the bilayer. This is thought to occur by the formation of transient lacunae in the bilayer, either by intramolecular movement of the hydrocarbon chains or by intermolecular movement of one lipid molecule against another. The significance of the latter factor is suggested by the finding that Thermus aquaticus, which grows at temperatures of 70 to 80°C, makes its membrane sufficiently impermeable even at these temperatures by using a glycolipid containing three rather than two fatty acid chains (reviewed in reference 129). Thus, increases in the number of hydrocarbon chains within a single lipid molecule increase the contact between the lipid molecules and presumably prevent the formation of lacunae that arise from the movement of one lipid molecule away from another. Each LPS molecule contains six to seven fatty acid chains, in contrast to the glycerophospholipid molecule which contains only two such chains. This results in a very large increase in the energy of interaction between neighboring LPS molecules, as can be seen from the following experiment. When an aggregate of LPS is diluted with glycerophospholipids in the presence of Mg2+, islands of LPS persist for several days or more in a sea of phospholipids even at 37°C, presumably because of the strong interaction between the LPS molecules (187). However, the inner core and lipid A portions of LPS contain many negatively charged groups, and the electrostatic repulsion between the charges in neighboring LPS molecules must be neutralized by divalent cations. Indeed, the isolated E. coli OM contains 5.3 atoms of Mg2+ plus Ca2+ per LPS molecule, if the data of Coughlin et al. (35) are corrected for the presence of phospholipids in the membrane (Table 1).
The other factor that contributes to the lower permeability is the decreased intramolecular mobility of hydrocarbon chains. This is caused by the high content of saturated fatty acid residues, which favor the quasicrystalline, tight packing of the chains. This principle is clearly involved in the low permeability of the LPS leaflet, as its fatty acid chains are saturated. Thus, the interior of the isolated LPS has very low fluidity, as shown by experiments using spin labels (141) and X-ray diffraction (89, 90). X-ray diffraction of intact OM also shows a strong, sharp reflection at 4.2 Å (194), indicative of a quasicrystalline arrangement of hydrocarbon chains. Consistent with this low fluidity, hydrophobic probes partition much less (by a factor of about 10) into the hydrophobic interior of LPS bilayer than into the interior of phospholipid bilayers (200).
Although there is general agreement that the interior of LPS is of low fluidity, there is little agreement on the details of the phase transition behavior of LPS. Some workers detected sharp liquid-crystalline to fluid transitions of isolated LPS at temperatures ranging from 22 to 37°C (see reference 175). However, cells usually avoid having sharp thermal transitions of their lipids, especially near their growing temperatures, because such transitions make the membranes unstable and leaky (129). It is therefore difficult to believe that such responses of isolated LPS reflect its behavior in living cells. Possibly the sharp, cooperative melting is a laboratory artifact caused in part by the absence of proteins. It is also likely that the transition temperature was lowered by the presence of insufficient amounts of divalent cations, as even the preparations of "native" LPS contain much fewer divalent cations than the OM (3.3 atoms per LPS, in contrast to 5.3 atoms in the intact OM [35]). In fact, no cooperative melting of LPS was observed in the range of 15 to 75°C in a study carried out in the presence of 1 mM Mg2+ (141).
Asymmetric Bilayer as Permeation Barrier.
Many gram-negative bacteria, including E. coli and S. typhimurium, are intrinsically resistant to hydrophobic antibiotics (e.g., macrolides, novobiocins, rifamycins, actinomycin D, fusidic acid), detergents (e.g., bile salts, SDS, Triton X-100), and hydrophobic dyes (e.g., eosin, methylene blue, brilliant green, acridine dyes) (142). In fact, the resistance to dyes and bile salts has been exploited in the formulation of a number of selective media favoring the growth of these organisms but suppressing that of gram-positive bacteria. This phenotype is due, at least in part, to the low permeability of the OM bilayer to these lipophilic solutes, although recent study suggests that active efflux mechanisms make synergistic contributions in many cases (see below).
The role of OM in such resistance is suggested by the observation that E. coli and S. typhimurium cells can be made much more susceptible to these lipophilic agents by EDTA treatment that weakens the lateral interaction between LPS molecules, or by mutational alteration of LPS (see below). The diffusion of highly lipophilic solutes across the lipid bilayer domains of the OM has been measured (147) based on the approach of Zimmermann and Rosselet (211), which was used extensively in measuring the permeability of the porin pathway (142). Intact cells of gram-negative bacteria were incubated in solutions containing 3-oxosteroid molecules, which diffused spontaneously across the OM bilayer and then rapidly across the usual phospholipid bilayer of the cytoplasmic membrane, to be oxidized finally by the 3-oxosteroid dehydrogenase cloned from Pseudomonas testosteroni. This assay showed that the OM bilayer is indeed less permeable, probably by a factor of 50 to 100, in comparison with the cytoplasmic membrane bilayer. Such low permeability is consistent with the low fluidity and tight architecture of the LPS leaflet of the OM, discussed above. However, the membrane still allows an influx of highly lipophilic solutes at a significant rate. For example, androstanediol diffuses with the permeability coefficient of 10–5 cm/s, a value comparable to the diffusion rate of cephalothin (a typical monoanionic cephalosporin) through the OmpF porin channels (142).
The OM bilayer thus can only slow down the penetration of uncharged, hydrophobic agents, and this is why gram-negative bacteria need additional mechanisms, such as active efflux, in order to develop significant resistance to these inhibitors (see below). The OM bilayer, however, is a very effective barrier against amphiphilic compounds that contain charged groups. With such compounds, the intrinsic low permeability of the bilayer is made far more effective because at a physiological pH only a small fraction of these molecules exist in uncharged forms that can traverse lipid bilayers; this is probably the major reason that S. typhimurium OM was earlier found to be essentially impermeable to nafcillin, a hydrophobic penicillin containing a carboxyl group (128). Similar reasoning shows that taurocholic acid, with its strongly acidic sulfate group, should essentially fail to penetrate the OM bilayer of enteric bacteria, in contrast to cholic acid, which is a weak acid. This makes ecological sense, because it is known that most of the bile acids exist as conjugated bile acids in the intestinal tract of vertebrates.
Perturbation of Bilayer Due to Deep Rough Mutations.
The permeability of the OM does not become altered when the structure of LPS becomes defective by the loss of the entire O polysaccharide and even of the distal portion of the core. However, when the sugars in the more proximal portion of the core are lost (chemotypes Rd1, Rd2, and Re [Fig. 1]), these mutants (deep rough mutants) become dramatically more sensitive to hydrophobic dyes, detergents, and antibiotics (142, 198). They also become hypersensitive to fatty acids, phenol, and polycyclic hydrocarbons. Although these results clearly indicate the importance of LPS in preventing the diffusion of hydrophobic solutes in the wild-type OM, the reason for the permeability increase in the mutants is more complex than suspected at first.The simplest hypothesis is that the leaflet composed of these defective LPS molecules allows an easier penetration of lipophilic solutes (Fig. 5C). However, there was no difference in the partition of such solutes into the lipid interior of wild-type and deep rough LPS (200). (Although a recent review [166] supports this model because tolC mutants, which appear to be altered in their LPS, show a deep-rough-like phenotype, this phenotype is more easily explained on the basis of defective active efflux mechanism [see below].)
An alternative hypothesis consistent with all available data was suggested by the observation that deep rough mutants produce OMs with lower buoyant density, containing drastically decreased levels of porins (see reference 142). Recent in vitro studies showed that the trimerization of OmpF porin is not facilitated by deep rough LPS (173), and this explains why these porin proteins fail to become incorporated into the OM. The resultant decrease in the protein content leads to incorporation of more phospholipids into the OM, and importantly, into the outer leaflet, creating phospholipid bilayer patches that allow the rapid transmembrane diffusion of lipophilic solutes (Fig. 5B).
Perturbation of Bilayer Due to Other LPS Mutations.
A temperature-sensitive mutation in lpxA, coding for UDP-N-acetylglucosamine O-acyltransferase, the first enzyme of lipid A biosynthesis, makes E. coli extremely susceptible to hydrophobic antibiotics (205). The mutation decreases significantly the rate of synthesis of LPS even at permissive temperatures, and the main cause for hypersusceptibility is likely to be the same as in deep rough mutants, that is, the production of phospholipid bilayer patches. The mutant is hypersusceptible to large, hydrophilic peptide antibiotics as well, and this suggests that the OM is unstable and may go through transient rupture and resealing (198).
Mutations in another gene, firA, also produce deep-rough-type hypersusceptibility in E. coli and S. typhimurium (72). The sequence of this gene is homologous to lpxA, and so it is likely to be involved in lipid A synthesis as well (198).
The envA1 mutant, which also produces decreased amounts of LPS (64), shows deep-rough-type hypersusceptibility, again presumably due to the formation of phospholipid bilayer domains in the OM.
Perturbation of Bilayer by Chelators.
Leive (95) found in 1965 that a brief treatment of E. coli with EDTA in Tris buffer liberated about one-half of the LPS from the cells and at the same time made them much more sensitive to actinomycin D, a hydrophobic agent. Later studies showed that EDTA-treated cells became hypersusceptible to a wide range of hydrophobic antibiotics, dyes, and detergents and behaved very much like the deep rough mutants described above (96). Clearly, removal of divalent cations that neutralized the electrostatic repulsion between LPS molecules results in the destabilization of the LPS monolayer portion of the membrane. The increase in permeability is likely to be due to the filling in of the space, formerly occupied by LPS, by phospholipid molecules, thereby creating phospholipid bilayer domains (Fig. 5D). This notion is also supported by the observation that the cells must continue to synthesize macromolecules (presumably LPS) in order to reestablish the barrier property of the OM (96).
In another use of EDTA in bacterial physiology, E. coli or S. typhimurium cells are incubated in 20% sucrose containing EDTA and Tris buffer and then rapidly diluted in water, in order to release the periplasmic proteins (127). In this "osmotic shock" procedure, the periplasmic space first becomes filled with 20% sucrose, which produces plasmolysis. Upon dilution, the osmotic imbalance will induce the influx of water into the periplasm as well as the efflux of sucrose from the periplasm into the medium. Because the latter process is limited by the poor permeability of porin channels for sucrose (see below), the OM, which has become weakened by EDTA-Tris treatment, becomes ruptured momentarily by the massive influx of water into the periplasm.
In both of these procedures, Tris, a bulky primary amine, contributes to the weakening of the lateral interaction between LPS molecules by partially replacing other cations tightly bound to LPS. Amines like Tris presumably inhibit the tight association between LPS molecules through steric hindrance; it is known that triethylamine is an excellent agent for dissociating the intermolecular aggregation of purified LPS (58). Tris alone, at high concentrations (50 mM or higher), is sufficient, without EDTA, to cause the release of some LPS and to make bacterial cells somewhat more sensitive to various agents (197).
Other chelating agents that are known to increase the lipophile permeability of the OM include nitrilotriacetate (66) and sodium hexametaphosphate (199). The latter is an especially convenient agent in the laboratory because it can sensitize E. coli to a wide range of hydrophobic agents without interfering much with its viability.
Ascorbate and acetylsalicylate were reported to permeabilize P. aeruginosa OM, supposedly owing to their chelator activity (66). However, the assay was carried out in an extremely dilute buffer, and repetition of the experiments under more standard conditions failed to show any permeabilizing activity (195). Similarly, although fluoroquinolones were claimed to permeabilize the OM through metal chelation (28), the effect was observed only after a long incubation at very high concentrations of the drugs, and it is unlikely that this mechanism has much relevance to the antibacterial action of these drugs.
Effects of Divalent Metal Cations.
As stated above, the LPS monolayer portion of the OM must be stabilized by divalent cations. In mutants producing unstable OMs, such as those lacking murein lipoprotein or deep rough mutants, the addition of millimolar concentrations of Mg2+ to the growth medium stabilizes the membrane structure, as judged by the decreased release of periplasmic enzymes (142).
In contrast, very high concentrations of Ca2+ perturb the structure of the OM. Thus, DNA is not expected to penetrate through the normal OM, but treatment of cells with 20 mM or higher concentrations of Ca2+ at 0°C can convert E. coli and S. typhimurium cells into recipients in transformation and transfection. The maltose-binding protein in the periplasmic space also becomes accessible for antibody added from the outside (24). Ca2+, presumably because it interacts less strongly with the surrounding water molecules than Mg2+ does, tends to precipitate anions (for example, CaSO4 is insoluble but MgSO4 is very water soluble). Similarly, Ca2+ "freezes" acidic phospholipids by raising their melting temperature, in some cases precipitating them out of solution (145). Most probably Ca2+ influences the LPS monolayer in the same manner and produces "cracks" by freezing the LPS leaflet. Very high (100 mM) concentrations of Mg2+ in the cold are also known to increase OM permeability (197).
Perturbation of LPS Leaflet with Polycations.
Polymyxin is a cyclic decapeptide antibiotic with a fatty acid tail, five positively charged groups, and no negative charge. It first binds to the OM of E. coli or S. typhimurium, presumably by binding to LPS, and then goes through the OM by disrupting this barrier. This conclusion is supported by the study of a polymyxin-resistant (pmrA) mutant of S. typhimurium. This mutant contains four- to sixfold higher amounts of a positively charged sugar, 4-aminoarabinose, in the lipid A portion of its LPS (203). Consequently, the mutant LPS is less acidic and binds polymyxin much less readily than the wild-type LPS. Among members of the Enterobacteriaceae, there are also naturally occurring "mutants" that are intrinsically resistant to polymyxin. Proteus mirabilis, one of these species, is known to produce LPS that contains much higher amounts of 4-aminoarabinose substituent than does that of E. coli or S. typhimurium (see reference 197). These data indicate that the binding to LPS, followed by the disruption of the OM barrier, is an essential step in the action of polymyxin.
Polymyxin is thus a prototype of polycationic agents which disorganize and disrupt the OM. However, its action is complicated by the presence of a hydrophobic fatty acid tail. Vaara and Vaara (201, 202) studied the action of polymyxin nonapeptide, a papain-cleaved derivative of polymyxin that has lost the N-terminal diaminobutyric acid residue and the fatty acyl residue attached to it. This cyclic polycation is very active in sensitizing the wild-type strains of E. coli and S. typhimurium to a number of hydrophobic agents such as novobiocin, fusidic acid, erythromycin, clindamycin, rifampin, actinomycin D, cloxacillin, and nafcillin, reducing their MIC sometimes by a factor of 300 or more at concentrations of 3 μg/ml or less. It must also be emphasized that polymyxin nonapeptide remains active in the presence of physiological concentrations of Na+ (about 150 mM) and Mg2+ (1 to 2 mM). It does not cause any release of LPS. It produces long, fingerlike projections involving only the outer leaflet of the OM. This expansion of the outer leaflet undoubtedly is the cause of the increase in permeability, although the precise molecular mechanism of the process is not yet clear.
The remarkable activity of polymyxin nonapeptide is presumably related to its rigid, cyclic structure and to the large number (five) of cationic groups. This is suggested by the observation that ring opening drastically decreases its activity. Among linear oligolysines, Lys4 shows no permeabilizing activity (196). Lys5, with its five net positive charges, shows only a marginal permeabilizing activity, which becomes suppressed in the presence of monovalent cations and divalent cations in the physiological range (196). It is not until one reaches molecules such as Lys20, with a much larger number of cationic groups, that permeabilizing activity comparable to polymyxin nonapeptide is found. In contrast to the latter compound, Lys20 released about 30% of the LPS from the cell, and thus its sensitizing action could have the same basis as that of EDTA, already discussed.
During recent years, other small cationic peptides such as cecropins (from insect hemolymph) (21), magainins (from frog skin) (17), melittin (from bee venom) (42), and defensins (from mammalian phagocytic cells) (92) have been studied extensively as antibacterial compounds. Their ultimate target appears to be the cytoplasmic membrane, in which they produce channels. In order to reach this target, they are likely to cross the OM by binding to LPS and by perturbing the structure of the OM, just like polymyxin B. Although some of these compounds were shown to bind to LPS, the OM permealization effect was rather small (197); this is expected because the first three groups are linear peptides. Defensins have three disulfide bonds to rigidify their structure, but the number of net positive charges is rather small in many of them (three in human neutrophil defensins, for example). Some workers believe that these compounds can be developed into useful antibacterial agents, but this may not be easy. First, all these agents show their effect only at much higher concentrations (in the range of 10 to 30 mg/liter) than do typical antibiotics. Second, the activity of most of these agents is drastically decreased at physiological concentrations of Na+ and Mg2+. Among other cationic peptides, protegrins from porcine neutrophils appear very potent, presumably due to the large number of positive charges (seven in one compound) and disulfide cross-links (116). Bactenecins from bovine neutrophils have also been reported to increase the OM permeability (see reference 197).
In contrast to these small peptides, the bactericidal/permeability-increasing protein present in the azurophil granules of neutrophils is a large protein (about 58 kDa) (48). It binds to LPS with a strong affinity (Kd = 23 nM), permeabilizes OM, and kills specifically gram-negative bacteria (48). Interestingly, it is homologous in sequence to a serum protein, the LPS-binding protein, which has an even higher affinity to LPS but does not kill the bacteria.
It has been proposed that aminoglycosides cross the P. aeruginosa OM mainly by binding to LPS, followed by the disorganization of the membrane (65). It was also reported that aminoglycosides increase the OM permeability, as judged by the influx of nitrocefin and an uncharged fluorescent probe, N-phenyl-1-naphthylamine (66). The permeabilizing effect was seen only in an extremely dilute (5 mM) buffer. Adding 1 mM Mg2+ totally abolished this permeabilizing effect (66). However, in the body fluids and cytosol of vertebrates, the salt concentration is above 100 mM and the Mg2+ concentration is around 1 to 2 mM. It is thus doubtful whether aminoglycosides exert significant levels of generalized permeabilizing effects in the tissues of patients, although it is still possible that even an undetectably low level of permeabilization is sufficient to allow the penetration a few molecules of aminoglycosides, which could later produce bactericidal consequences through misreading (38). Similar criticisms apply to the claim that azithromycin (containing only two positive charges) permeabilizes the E. coli OM and thereby promotes its own uptake (49).
Drug Hypersusceptibility through Presumed Defects in Active Efflux.
There are drug-hypersusceptible mutations of E. coli that were thought to produce hyperpermeable OM, but were more recently shown to cause defective active efflux. The acrA mutation is a classic example. It is hypersusceptible to acridine dyes, phenethyl alcohol, SDS (126), and various lipophilic antibiotics. No reproducible alteration has been observed in LPS and the outer membrane proteins. DNA sequencing (104; J. Xu, M. L. Nilles, and K. P. Bertrand, Abstr. Gen. Meet. Am. Soc. Microbiol. 1993, abstr. K-169, p. 290) showed the presence of an operon acrAB (called acrAE in reference 104). AcrB has the sequence expected for an active efflux pump located in the cytoplasmic membrane, and its efflux function was demonstrated experimentally (104). The AcrAB system appears to have an unusually broad specificity which can transport novobiocin, erythromycin, fusidic acid, mitomycin C, and tetracycline in addition to the dyes and detergents (104).
AcrA codes for a periplasmic lipoprotein, essential for efflux (104). Interestingly, similar multidrug efflux systems of E. coli K-12, such as EmrAB (99) and AcrEF (previously EnvCD) (94, 104), also contain similar periplasmic lipoproteins, EmrA and AcrE, and these are homologous to the periplasmic components of systems that export proteins directly into the medium, such as HlyD (43). Furthermore, they are homologous to a membrane fusion protein of simian virus 5, a paramyxovirus (43). These data suggest that proteins of this "membrane fusion protein" family bring cytoplasmic membrane and OM together to enable the direct excretion of proteins and drugs into the medium (43).
The presumed function of the membrane fusion proteins implies the involvement of an OM channel for efflux. Indeed, the hemolysin export system of E. coli uses an OM channel, TolC (206). Possibly TolC also serves as the OM channel for other drug efflux systems. If so, the hypersusceptibility of tolC mutants to various dyes, detergents, and hydrophobic antibiotics is caused by defects in the active efflux of these agents, rather than increased OM permeability.
Interestingly, some gene clusters coding for protein/drug export systems contain genes for OM channels as well. Thus the protease export gene cluster in Erwinia chrysanthemi (93), the cyclolysin export gene cluster in Bordetella pertussis (61), and alkaline protease export gene cluster in P. aeruginosa (46) all contain tolC homologs in addition to the genes for transporter proteins and the membrane fusion proteins. Similarly, in P. aeruginosa, a single gene cluster codes for all three components of an assembly that pumps out various antibiotics into the medium (148).
The antibiotic-hypersusceptible mutant DC2 (154) contains the abs mutation that makes the strain more sensitive to dyes, detergents, and lipophilic antibiotics (32). Possibly abs also codes for one of the efflux pumps. Although it has been reported that the mutant LPS contains 30% more amino groups (32), it is not clear whether this difference explains the hypersusceptibility phenotype, because such a phenotype is not found in pmrA mutants of S. typhimurium which contain even higher amounts of amino substituents (203).
Apparent Effect of Energization of the Cytoplasmic Membrane.
It is well known that deenergization of the cytoplasmic membrane results in the apparent permeation of lipophilic probes, such as N-phenyl-1-napththylamine, 8-anilino-1-naphthalenesulfonic acid, and azidopyrene, into E. coli cells (reviewed in reference 142). It was believed until recently that the OM bilayer was essentially impermeable to these highly lipophilic agents, and thus the phenomenon suggested an implausible scheme in which the energization state of the cytoplasmic membrane somehow influenced the fluidity of the OM bilayer.
It is now clear that the OM allows the penetration of these solutes at slower, but still significant rates (147). Thus the apparent impermeability observed in energized cells must be due to the process of active efflux. Uncouplers simply inhibit the active efflux process, thereby increasing the cellular concentration of these probes. Indeed, E. coli is now known to have several chromosomally coded efflux systems of very broad specificity, some of which have been mentioned above (94, 131).
To understand the role of OM permeability in bacterial physiology, it is essential to have a quantitative estimate of this parameter. Knowing qualitatively that the OM is or is not permeable to a compound under an arbitrary set of conditions tells us very little, because, if one waits long enough and if a sensitive enough assay is available, most flexible solute molecules will eventually be found to traverse the porin channel. Quantitative determination of permeability coefficients for nutrient molecules can conveniently be carried out in mutant cells producing decreased levels of porins, where the diffusion through the OM becomes the rate-limiting step in nutrient transport (5). Similarly, the permeability of β-lactam antibiotics can be assessed quantitatively if excess β-lactamase is present in the periplasmic space to hydrolyze incoming antibiotic molecules (211). In the latter procedure, it is common to assume that the periplasmic β-lactamase behaves in a way similar to the enzyme in free solution. Although this is not the case (109), the coefficients obtained will be fairly close to the correct ones, as long as the β-lactamase activity is in large excess over the rate of entry of the drug (for example, see reference 140). When the conditions are pushed to the limit, however, for example by using very low drug concentrations and cells containing marginal enzyme activity (for example, see reference 9), the error could become significant. In P. aeruginosa, the OM permeability coefficients obtained by using the method described above could not explain the extent of resistance of the bacteria to β-lactams (98), but this is likely to be due to the contribution of the active efflux mechanism to resistance (96a).
For studying the properties of individual porin species, it is convenient to use systems reconstituted from purified porins, such as proteoliposomes (137, 138), planar lipid bilayers (162, 163), or black lipid films (13). These studies showed that OmpF and OmpC, the E. coli porin channels expressed under the usual growth conditions, show no true specificity for any solute class. Nevertheless, the diffusion rates through these channels are greatly affected by the gross physicochemical properties of the solutes, such as size, electric charge, and hydrophobicity.
Size of the Solute.
As predicted for diffusion through narrow channels, the chance that the solute molecule successfully enters the channel by random collision (and therefore the macroscopic permeability coefficient toward that particular solute) is strongly influenced by the size of the solute molecule. Thus, the rate of diffusion of disaccharides through the E. coli OmpF channel is nearly two orders of magnitude lower than the rate of diffusion of a pentose, although the size of disaccharides, 342 Da, places them well within the "exclusion limit" of 600 Da of this porin (137, 138). The dependence of the permeability on solute size is somewhat different for the OmpC channels. From the slopes of the permeability-versus-solute-size curves one can calculate the approximate diameters of OmpF and OmpC pores as 1.16 and 1.08 nm (138). As noted earlier, the figure for OmpF is in excellent agreement with the 1.1 × 0.7 nm dimension of the eyelet region of the channel, determined by X-ray crystallography (37). (Earlier, the pore diameter was also estimated from single channel conductance in lipid bilayer experiments; this practice cannot be justified theoretically, and indeed yields incorrect values [see reference 130 for discussion].)
The pore size may be larger in some other gram-negative bacteria; for example, P. aeruginosa produces porins with significantly larger channels (8, 136). In these larger channels, the diffusion rates of solutes are not as strongly influenced by their size.
Hydrophobicity of the Solute.
Studies with monoanionic cephalosporins showed that the permeability through the OmpF channel of E. coli was affected negatively by the hydrophobicity of the solute, with a 10-fold increase in the 1-octanol/water partition coefficient of the uncharged form of the solute producing a four- to fivefold decrease in the penetration rate (137). I have described above how this property of the channel can be explained by the structure of the eyelet region of the channel.
Electrical Charge.
In black lipid film studies (13), OmpF and OmpC porins showed a preference for cations over anions. Furthermore, liposome studies showed that the diffusion of uncharged sugars though these channels was several times faster than that of corresponding negatively charged sugar acids and that the addition of another negative charge further slowed penetration (138). The retardation effect of the negative charge is further enhanced in intact cells (140), presumably because of the presence of interior-negative Donnan potential (171, 179) and of the presence of a high concentration of negatively charged LPS molecules on the surface.
In contrast to the behavior of OmpF and OmpC channels, the presence of negative charges (such as carboxylate and sulfate) does not hinder, and usually accelerates, the diffusion through PhoE porin (138). The PhoE channel also shows anion selectivity in black lipid film studies (12). Although some workers hypothesized that this channel is specific for phosphate, and the protein is called rather misleadingly phosphoporin by some, clearly there is no basis for such an assumption (4), and the channel simply seems to favor any negatively charged solute.
The structure of porin channel helps us to understand the mechanism for ionic discrimination by these porins. Thus the eyelet region of OmpF channel contains several charged amino acid residues (Fig. 4). As described above, PhoE contains an additional, positively charged residue, Lys-125, right at the eyelet.
Regulation of Porin Synthesis.
Enteric bacteria survive in an environment that is full of detergents, e.g., the bile salts, and the properties of the porin channels are ideal for excluding these large, negatively charged, hydrophobic compounds. Furthermore, the synthesis of OmpF, the porin with a larger channel, is repressed by high osmolarity as well as by high temperature (reviewed in reference 142). This means that practically no OmpF porin is synthesized at 37°C in the presence of 0.15 M NaCl, i.e., under conditions found inside the mammalian body. By switching from the wider OmpF channel to the narrower OmpC channel, the bacteria lose only 50% or so of the permeability toward small nutrients such as glucose, whereas permeability to larger, more hydrophobic, or negatively charged compounds (or toward compounds with a combination of all these properties, such as bile salts) is much more drastically reduced because of the more restrictive properties of the OmpC channel (142).
The exclusive synthesis and use of OmpC inside the human body was supported by the observation (114) that cephalosporin therapy of a patient infected by an ompF+ ompC+ strain of S. typhimurium resulted in the selection of an ompF+ ompC– mutant strain. This would not make sense if OmpF porin were produced in the body of the patient, because in media allowing the production of OmpF, the mutant is as susceptible as the parent strain to cephalosporins, which penetrate readily through the OmpF channel. However, the mutant is much more resistant to cephalosporins in media containing 0.15 M NaCl (the conditions mimicking those in human body fluids) because it lacks OmpC due to the mutation and because OmpF synthesis is completely repressed under these conditions. Thus the selection of the mutant strain in the patient must have occurred under the conditions in which OmpC was the only porin expressed by S. typhimurium.
On the other hand, the OmpF porin is probably beneficial when the bacteria are out of the animal body. In ponds and streams, low osmolarity and low temperature will result in the derepression of OmpF synthesis, and the wider channel of OmpF will allow the more efficient uptake of nutrients from a very dilute environment. The PhoE porin, which is induced by phosphate starvation (see reference 142), may also be useful in a similar environment, by allowing a more rapid diffusion of phosphorylated compounds.
The molecular mechanism of osmotic regulation is complex. High osmolarity, through unknown mechanisms, is known to increase the negative supercoiling of DNA, which in turn increases the expression of ompC (62). This global mechanism of adaptation is presumably also involved in the regulation of porin levels by other signals, such as anaerobiosis. A more specific regulatory mechanism through EnvZ-OmpR, a "two-component system" (121, 178), is superimposed on this general mechanism. High osmotic pressure produces the autophosphorylation of a histidine residue of the sensor protein EnvZ, which then transfers phosphate onto the response regulator, OmpR. In the simplest model, low concentrations of phosphorylated OmpR will bind to a high-affinity site upstream of ompF and activate its transcription. However, high concentrations of phosphorylated OmpR will bind to the low-affinity site, again in front of ompF, and inhibit its transcription; at the same time, its binding to a low-affinity site(s) upstream of ompC will activate its transcription. Experimental evidence, however, is needed to substantiate the model. Another complication is the presence of micF gene, which is transcribed divergently from the same operator together with ompC. micF transcript is a natural antisense messenger that is expected to bind to the ompF mRNA, and high-level expression of micF indeed decreases the expression of OmpF (120). Although the deletion of micF did not seem to have much effect on OmpF expression (110), micF RNA appears to play a more important role in response to temperature (1) and other signals, as described below.
OmpF synthesis is also known to be repressed by acidic pH (70). This repression is now known to require general catabolite repression (189).
Other signals suggestive of the host environment also repress OmpF synthesis. Thus oxygen stress, which would occur by the response of phagocytic cells of host animals to invading organisms, produces this repression (31), as does salicylate (157, 161), which is now known to be produced by plant tissues in response to microbial invasion (30). The decrease in OmpF must make the bacteria more resistant to various antimicrobial compounds present in higher animals and plants, and is known to make them also more resistant to various antibiotics (31, 157). Interestingly, the Mar regulatory system of E. coli responds to the presence of some antibiotics (2), causing the repression of OmpF porin and up-regulation of a drug efflux pump(s), producing a higher level of antibiotic resistance (33, 60). Lowering of outer membrane permeability has a particularly strong synergistic effect on antibiotic resistance based on active efflux (131), and thus it is not surprising that some antibiotic resistance mutations (e.g., nfxC [75] and marA, described above in E. coli) and R factors are known to repress the synthesis of OmpF porin (see reference 131 for review). In most of the cases described above, the repression of OmpF synthesis appears to require the synthesis of micF antisense RNA.
One aspect of porin physiology that puzzles many scientists is why such a large number of porin molecules, up to 105 per cell, are necessary. This can be understood when one remembers that porin channels (i) are nonspecific and (ii) mediate simple diffusion processes. Because the channels are narrow, diffusion of even moderately large molecules, such as disaccharides, becomes very slow in comparison with that of very small molecules, as described already. Furthermore, because the rate of simple diffusion is proportional to the concentration difference between the external medium and the periplasm, the rate, which could be quite high in laboratory media often containing a 0.01 M carbon source, drops more than three orders of magnitude at micromolar concentrations of nutrients, the kind of environment for which most of the active transport systems of E. coli seem to have been designed (85). Thus, we can calculate that 105 molecules of porin per cell are indeed necessary for the transport of moderate-sized carbon source molecules such as glucose, at rates that would support the maximal growth rate of E. coli, at external concentrations in the micromolar range (142). For larger molecules such as disaccharides or nucleosides, diffusion through even 105 porin channels is not rapid enough to support rapid growth, which is why specific diffusion systems are needed for these compounds in spite of their ability to diffuse through the porin channels.
Possibility of Dynamic Alterations in Porin Channels.
In an early study of OmpF porin in black lipid films, acidification of the medium to pH 3 significantly decreased the pore size (13). More recently, the decrease in pore size was shown to occur already at pH 5.4 (190). Since some cultures of E. coli may reach this pH value during growth, it will be important to see whether this pH-mediated alteration of pore size occurs in intact cells.
Experiments with planar bilayer systems have shown that the porin channels are voltage-regulated (162, 163). These results have been confirmed in other laboratories (111). Nevertheless, it seems unlikely that this phenomenon plays a physiological role in intact cells, because electrical potential of significant size is unlikely to exist across the OM with such a high density of open channels. The only potential that is known to exist is Donnan potential, but this was shown to have no effect on the OM permeability in intact E. coli cells (169). Perhaps one physiological function of voltage-gating is to close the porin channels when they become erroneously incorporated into the cytoplasmic membrane, as was suggested by Tommassen (J. Tommassen, personal communication, cited in reference 130). It should be emphasized that the glucose permeability of OM of growing E. coli cannot be explained unless most of the porin channels are open (5). Because of the crudeness of the calculations, we cannot rule out the possibility that a fraction of the pores may be closed, but the experimental data are certainly incompatible with models assuming that only a small fraction, say 1%, of the channels are open.
Crude extracts of E. coli were found to contain materials that lower the threshold voltage for the closing of porin channel (41). The channel-closing potential must have a polarity opposite of the Donnan potential across the OM, but it is argued that an interior-positive potential may be generated by the rapid influx of cations when cells suddenly face a high salt concentration (41).
Application of patch-clamping methods to spheroplasts of E. coli showed the existence of pressure-sensitive channels, which were thought to correspond to the porin channels (26). However, fractionation of the E. coli membranes showed that such channels are derived from the cytoplasmic membrane (16), and cloning and sequencing of the gene coding for a pressure-sensitive channel of E. coli (183) showed that the protein is not related to porins.
Among the specific diffusion channels, the LamB channel has been studied most intensively. In both intact cells and reconstituted systems, this protein produces open channels that allow the diffusion of structurally unrelated small solutes, in addition to maltose (15, 100). The structural similarity between LamB and classical porins suggests that in both channels the solute diffusion occurs by an essentially similar mechanism, i.e., through water-filled channels. However, there is a specific binding site for maltose and maltodextrins within each LamB channel, with the estimated Ks value of 0.1 (15) and 1 mM (101) for maltoheptaose. Theory predicts that such a channel would produce Michaelis-Menten- or saturation-type kinetics of diffusion (15). Thus the downhill diffusion of specific ligands through such a channel would become accelerated at low external concentrations. Indeed, this saturation phenomenon was demonstrated experimentally (56). Similar saturation kinetics were found in another specific channel, OprD of P. aeruginosa (191), which facilitates the diffusion of basic amino acids (192). Interestingly, a recent study suggests that under carbon starvation conditions the biosynthesis of LamB becomes derepressed, and that LamB plays a major role in the influx of small sugars and sugar alcohols such as glucose and glycerol. The binding site of the LamB channel does bind these compounds, albeit with low affinity (39).
Because the LamB channel is exposed on the cell surface, it binds even macromolecular ligands, such as starch. Ferenci’s group have devised ingenious methods to use starch binding for the isolation of lamB mutants altered in the binding of the ligands, either with affinity columns (50) or by retardation of chemotaxis in starch-containing plates (68). Studies of these mutants showed that the maltose binding was affected by mutations in the presumed third external loop (68). Interestingly, in porins this is the loop that folds into the β-barrel and produces the solute-discriminating constriction (37, 208).
The recent determination of the structure of LambB protein by X-ray crystallography (163) showed that each subunit of a trimer traverses the OM 18 times as β-strands, and the central channel is narrowed by the infolding of several external loops, including the third loop. All of the known mutations affecting ligand affinity are at the site of the construction of the channel, which has a diameter of about 0.5 nm. Interestingly, the channel is lined by a succession of aromatic residues, which are hypothesized to act by binding to the pyranose rings of the ligands, facilitating their movement through the channel as a "greasy slide."
Bayer (6) discovered sites at which inner membranes are apparently fused to the OM in thin sections of plasmolyzed E. coli cells. Although such sites could not be detected under a special condition of fixation (73), a more recent work suggests that this was the result of the impact freezing method used in that particular study (7).
In addition to the Bayer adhesion sites that are dispersed all over the cell surface (6), a ringlike adhesion zone, the periseptal annulus, was discovered by Rothfield and coworkers (34, 105). Biochemical studies suggest that the direct export of proteins (206) or drugs (131) into the medium requires the physical association of cytoplasmic-membrane pumps and OM channels. A class of periplasmic proteins, showing homology with a viral membrane fusion protein (43), are known to be involved in this assembly, as described above. Thus a third group of possibly transient fusion sites is likely to exist in cells of E. coli and S. typhimurium, although it is not known whether they can be detected by the conventional morphological methods.
By introducing labeled phospholipid molecules into the OM of intact cells, Jones and Osborn (80) showed that there is a rapid exchange between the phospholipids of the outer and the inner membranes. Attempts to detect the presence of phospholipid exchange (or carrier) proteins in the periplasm have been unsuccessful (80), and the exchange likely occurs through some fusion sites. The kinetics of translocation of newly synthesized phospholipids from the inner membrane to the OM was studied by Donohue-Rolfe and Schaechter (45). Phosphatidylglycerol and cardiolipin are translocated much more rapidly than phosphatidylethanolamine. The translocation of the latter compound is inhibited under conditions that decrease the proton motive force.
As described elsewhere in this treatise (chapter 69), LPS is synthesized on the cytoplasmic membrane. Core oligosaccharide is built via successive extension by transfer from nucleotide sugars to incomplete LPS molecules, presumably anchored onto the inner surface of the cytoplasmic membrane. The repeating unit of the O antigen is made independently in a form anchored to the membrane via a carrier lipid, undecaprenol pyrophosphate. There are several possibilities as to how polymerization of the O-antigen polysaccharide and its transfer to the core LPS may be done, in relation to the export of LPS to the outer leaflet of the OM (143). Mulford and Osborn (125) showed by immunoelectron microscopy that in mutants defective in core synthesis, the newly made O chain, presumably still linked to the carrier lipid, is located on the outer surface of the inner membrane. Since the sugars are transferred from nucleotide sugars, which are components of the cytoplasm, O repeat units are probably assembled on the inner face of the cytoplasmic membrane, and the completed repeat unit, linked to the lipid carrier, is translocated or "flipped" across the membrane. The O units are then polymerized by a polymerase, and finally the long O chains are transferred to the core by a ligase, both reactions presumably taking place on the periplasmic face of the inner membrane. Subsequent studies from the Osborn laboratory support this hypothesis. Thus the ligase reaction in intact cells, but not that in isolated membrane fractions, is inhibited by dinitrophenol, suggesting that the flipping of the core LPS requires proton motive force (113). Further, the transfer of galactose 1-phosphate, containing the first sugar of the O unit, to the lipid carrier is again sensitive to dinitrophenol inhibition only in intact cells, a result suggesting that perhaps in the absence of flipping of the O unit there is no regeneration of lipid phosphate that can act as the acceptor of galactose 1-phosphate (108). The gene cluster for O-antigen synthesis, rfb, contains a gene for a putative 12-transmembrane helix protein, RfbX (153). It is possible that this protein is the flippase, because in its absence the polymerization of O the unit is completely inhibited and the ligation of monomeric O unit to the R core is also greatly decreased (83).
Newly made LPS molecules were seen to appear on the outer surface of the OM, as patches over the Bayer fusion sites (124). The molecular mechanism of this export process is not yet clear. In contrast to phospholipids, the translocation of LPS is unidirectional. LPS added to the OM from the external medium was not found to move to the inner membrane (80).
OM proteins are synthesized initially with signal sequences and are exported through the Sec-dependent secretion pathway. Although these features are shared with the periplasmic proteins, the OM proteins do not remain in the periplasm. How does the cell distinguish proteins destined for the OM from those destined for the periplasm? Much work has been carried out in this area, and the results have been summarized in an excellent review (69). Following is only a sketch of some of the more pertinent results.
One of the central questions is whether the proteins are first exported into the periplasm, or remain associated with the membranes throughout the export process. On balance, currently available data favor the former alternative. First, practically all OM proteins are as hydrophilic as an average cytosolic or periplasmic protein and lack hydrophobic stretches that would induce their insertion into the membranes. Furthermore, insertion of such a long, hydrophobic sequence into the membrane domain of OmpA resulted in the anchoring of the altered protein within the cytoplasmic membrane, thus preventing its export to the OM (106). Second, whenever internal deletions or truncations of OmpA and PhoE interfered with their export to the OM, the defective proteins accumulated in the periplasm (22, 84). Third, when a mutation slows down the cleavage of signal sequence from pre-LamB, and thus presumably the release of LamB into the periplasm, the trimerization and insertion of LamB into the OM becomes impaired (27). Finally and most importantly, unaltered OmpA in its immature conformation was found in the periplasm, especially when the protein was overproduced (55), and a metastable trimeric assembly intermediate of OmpF was recovered in the periplasmic fraction by osmotic shock procedure (54).
OM proteins can possibly exist in soluble forms in the periplasm, owing to their hydrophilic nature. How then do they become inserted into the OM? Most of the OM proteins that have been sequenced appear to have β-barrel structures (79, 130). Inspection of the three-dimensional structure of porins shows that the outside surface of β-barrels is covered by hydrophobic amino acid side chains, as expected. However, individual β-strands in OM proteins tend to be amphiphilic, and they cannot penetrate into the hydrophobic interior of membranes until the complete β-barrel becomes assembled. This forms a striking contrast to cytoplasmic membrane proteins, whose α-helices can be successively inserted into the membrane one by one.
Hints on the refolding process can be obtained from in vitro refolding studies. Thus the immature folding intermediate of OmpA, which accumulates in the periplasm, is more easily denatured by SDS, soluble in Sarkosyl, and completely digested by trypsin, all properties suggesting a more open conformation of the protein. Addition of LPS refolds this open conformer into a tighter, mature form that is not denatured by SDS easily, is insoluble in Sarkosyl, and is only partially digested by trypsin (55). Similarly, nascent OmpF porins secreted into medium by spheroplasts behave as open monomers and are rapidly trimerized in the presence of excess Triton X-100 and small amounts of SDS, the latter probably acting as an analog of LPS because both compounds contain about one net negative charge per hydrocarbon chain (172). Mature OmpF trimers can be denatured in guanidium hydrochloride and then renatured in the presence of SDS, detergent octyl-polyoxyethylene, and lecithin (47).
Similar studies were also carried out with hopes of mimicking the reactions occurring in intact cells. The monomeric, nascent OmpF becomes correctly folded and trimerized rapidly and efficiently when fragments of OM or LPS bilayers are supplied (172). Interestingly, this requires a very low concentration (0.01%) of Triton X-100 or other nonionic detergents. LPS of the very deficient structure, i.e., those of the deep rough type that do not allow the efficient incorporation of porins in intact cells (see above), are not active in stimulating the trimerization reaction, suggesting the physiological nature of the reaction observed (173). When OmpF and PhoE proteins lacking most of the signal sequences are made by in vitro translation, such proteins are trimerized only rather inefficiently and tend to fold into "dead-end" conformations (40, 174).
All these results suggest that the immature conformers of OM proteins become refolded in the periplasm (or upon entry into the OM), often through interactions with LPS. In this connection, it is noteworthy that even the extremely hydrophobic, deep rough LPS has a solubility of 3 × 10–8 M in aqueous media (185), a concentration far higher than the concentrations of phospholipid monomers in an aqueous environment (5 × 10–10 M for phosphatidylcholine [188]). Thus, in principle some LPS should be available in the periplasm to facilitate the folding of these proteins. Unfortunately, lateral interaction between LPS molecules is extremely strong (187), and thus the LPS monomers are not in equilibrium with the LPS in the membrane (see reference 185). This may be the reason why the assembly of porins in intact cells requires nascent LPS: inhibition of fatty acid synthesis by cerulenin inhibits the trimerization of porins (but not insertion of the OmpA) (20, 155). Several studies also suggest the strong and specific nature of interaction between porin trimers and LPS (see, for example, reference 74).
Electron microscopy suggested that newly inserted trimeric porins of S. typhimurium appeared at discrete sites corresponding to Bayer fusion sites (177). Perhaps one way to reconcile this finding with the presumed periplasmic pathway of protein export is to assume that the insertion of the new proteins into the OM requires sites of exceptionally high plasticity and fluidity and that fusion sites fulfill this requirement. This is not an unreasonable hypothesis, considering the low fluidity of the LPS leaflet, discussed already. Possibly Triton X-100 is needed to produce similar sites in vitro.
Covalent modification and export of lipoproteins will not be discussed in this chapter, as these topics are covered elsewhere in this book (chapter 66).
This chapter is based on the earlier version, which was published in the first edition of this book (1987) and was written with the inestimable contribution from Martti Vaara as the coauthor. I apologize for the inability to cite a very large number of important studies, simply because of the lack of space. Research in my laboratory has been supported by grants from the National Institutes of Health (AI-09644) and from the Lucille P. Markey Charitable Trust.
References
1. Andersen, J., S. A. Forst, K. Zhao, M. Inouye, and N. Delihas. 1989. The function of micF RNA. J. Biol. Chem. 264:17961–17970.
2. Ariza, R. F., S. P. Cohen, N. Bachhawat, S. B. Levy, and B. Demple. 1994. Repressor mutations in the marRAB operon that activate oxidative stress genes and multiple antibiotic resistance in Escherichia coli. J. Bacteriol. 176:143–148.
3. Bauer, K., M. Struyve, D. Bosch, R. Benz, and J. Tommassen. 1989. One single lysine residue is responsible for the special interaction between polyphosphate and the outer membrane porin PhoE of Escherichia coli. J. Biol. Chem. 264:16393–16398.
4. Bauer, K., P. van der Ley, R. Benz, and J. Tommassen. 1988. The pho-controlled outer membrane porin PhoE does not contain specific binding sites for phosphate or polyphosphate. J. Biol. Chem. 263:13046–13053.
5. Bavoil, P., H. Nikaido, and K. von Meyenburg. 1977. Pleiotropic transport mutants of Escherichia coli lack porin, a major outer membrane protein. Mol. Gen. Genet. 158:23–33.
6. Bayer, M. E. 1979. The fusion sites between outer membrane and cytoplasmic membrane of bacteria: their role in membrane assembly and virus infection, p. 167–202. In M. Inouye (ed.), Bacterial Outer Membranes. John Wiley & Sons, Inc., New York.
7. Bayer, M. E. 1991. Zones of membrane adhesion in the cryofixed envelope of Escherichia coli. J. Struct. Biol. 107:268–280.
8. Bellido, F., N. L. Martin, R. J. Siehnel, and R. E. W. Hancock. 1992. Reevaluation, using intact cells, of the exclusion limit and role of porin OprF in Pseudomonas aeruginosa outer membrane permeability. J. Bacteriol. 174:5196–5203.
9. Bellido, F., J.-C. Pechere, and R. E. W. Hancock. 1991. Novel method for measurement of outer membrane permeability to new β-lactams in intact Enterobacter cloacae cells. Antimicrob. Agents Chemother. 35:68–72.
10. Benson, S. A., J. L. L. Occi, and B. A. Sampson. 1988. Mutations that alter the pore function of the OmpF porin of Escherichia coli K-12. J. Mol. Biol. 203:961–970.
11. Benz, R. 1988. Structure and function of porins from gram-negative bacteria. Annu. Rev. Microbiol. 42:359–393.
12. Benz, R., R. P. Darveau, and R. E. W. Hancock. 1984. Outer-membrane protein PhoE from Escherichia coli forms anion-selective pores in lipid-bilayer membranes. Eur. J. Biochem. 140:319–324.
13. Benz, R., K. Janko, and P. Läuger. 1979. Ionic selectivity of pores formed by the matrix protein (porin) of Escherichia coli. Biochim. Biophys. Acta 551:238–247.
14. Benz, R., E. Maier, and I. Gentschev. 1993. TolC of Escherichia coli functions as an outer membrane channel. Int. J. Med. Microbiol. Virol. Parasitol. Infect. Dis. 278:187–196.
15. Benz, R., A. Schmidt, and G. H. Vos-Scheperkeuter. 1987. Mechanism of sugar transport through the sugar-specific LamB channel of Escherichia coli outer membrane. J. Membr. Biol. 100:21–29.
16. Berrier, C., A. Coulombe, C. Houssin, and A. Ghazi. 1989. A patch-clamp study of ion channels of inner and outer membranes and of contact zones of E. coli, fused into giant liposomes: pressure-activated channels are localized in the inner membrane. FEBS Lett. 259:27–32.
17. Bevins, C. L., and M. Zasloff. 1990. Peptides from frog skin. Annu. Rev. Biochem. 59:395–414.
18. Black, P. N., and C. C. DiRusso. 1994. Molecular and biochemical analyses of fatty acid transport, metabolism, and gene regulation in Escherichia coli. Biochim. Biophys. Acta 1210:123–145.
19. Blasband, A. J., W. R. Marcotte, Jr., and C. A. Schnaitman. 1986. Structure of the lc and nmpC outer membrane porin protein genes of lambdoid bacteriophages. J. Biol. Chem. 261:12723–12732.
20. Bolla, J. M., C. Lazdunski, and J.-M. Pages. 1988. The assembly of the major outer membrane protein OmpF of Escherichia coli depends on lipid synthesis. EMBO J. 7:3595–3599.
21. Boman, H. G. 1991. Antibacterial peptides: key components needed in immunity. Cell 65:205–207.
22. Bosch, D., J. Leunissen, J. Verbakel, M. de Jong, H. van Erp, and J. Tommassen. 1986. Periplasmic accumulation of truncated forms of outer membrane protein PhoE of Escherichia coli K-12. Mol. Gen. Genet. 189:449–455.
23. Bradbeer, C. 1993. The proton motive force drives the outer membrane transport of cobalamin in Escherichia coli. J. Bacteriol. 175:3146–3150.
24. Brass, J. M., W. Boos, and R. Hengge. 1981. Reconstitution of maltose transport in malB mutants of Escherichia coli through calcium-induced disruptions of the outer membrane. J. Bacteriol. 146:10–17.
25. Bremer, E., J. Middendorf, J. Martinussen, and P. Valentin-Hansen. 1990. Analysis of the tsx gene, which encodes a nucleoside-specific channel-forming protein (Tsx) in the outer membrane of Escherichia coli. Gene 96:59–65.
26. Buechner, M., A. H. Delcour, B. Martinac, J. Adler, and C. Kung. 1990. Ion channel activities in the Escherichia coli outer membrane. Biochim. Biophys. Acta 1024:111–121.
27. Carlson, J. H., and T. J. Silhavy. 1993. Signal sequence processing is required for the assembly of LamB trimers in the outer membrane of Escherichia coli. J. Bacteriol. 175:3327–3334.
28. Chapman, J. S., and N. H. Georgopapadakou. 1988. Routes of quinolone permeation in Escherichia coli. Antimicrob. Agents Chemother. 32:438–442.
29. Charbit, A., K. Gehring, W. Saurin, H. Nikaido, T. Ferenci, and M. Hofnung. 1988. Maltose transport and starch binding in phage resistant point mutants of maltoporin: functional and topological implications. J. Mol. Biol. 201:487–496.
30. Chen, Z. X., H. Silva, and D. F. Klessig. 1993. Active oxygen species in the induction of plant systemic acquired resistance by salicylic acid. Science 262:1883–1886.
31. Chou, J., J. T. Greenberg, and B. Demple. 1993. Posttranscriptional repression of Escherichia coli OmpF protein in response to redox stress: positive control of the micF antisense RNA by the soxRS locus. J. Bacteriol. 175:1026–1031.
32. Clark, D. 1984. Novel antibiotic hypersensitive mutants of Escherichia coli: genetic mapping and chemical characterization. FEMS Microbiol. Lett. 21:189–195.
33. Cohen, S. P., L. M. McMurry, D. C. Hooper, J. S. Wolfson, and S. B. Levy. 1993. Cross-resistance to fluoroquinolones in multiple-antibiotic-resistant (Mar) Escherichia coli selected by tetracycline or chloramphenicol: decreased drug accumulation associated with membrane changes in addition to OmpF reduction. Antimicrob. Agents Chemother. 33:1318–1325.
34. Cook, W. R., T. J. MacAlister, and L. I. Rothfield. 1986. Compartmentalization of the periplasmic space at division sites in gram-negative bacteria. J. Bacteriol. 168:1430–1438.
35. Coughlin, R. T., S. Tonsager, and E. J. McGroarty. 1983. Quantitation of metal cations bound to membranes and extracted lipopolysaccharide of Escherichia coli. Biochemistry 22:2002–2007.
36. Cowan, S. W, and T. Schirmer. 1994. Structures of non-specific diffusion pores from Escherichia coli, p.353–362. In J.-M. Ghuysen and R. Hakenbeck (eds.), Bacterial Cell Wall. New Comprehensive Biochemistry, vol. 27. Elsevier, Amsterdam.
37. Cowan, S. W., T. Schirmer, G. Rummel, M. Steiert, R. Ghosh, R. A. Pauptit, J. N. Jansonius, and J. P. Rosenbusch. 1992. Crystal structures explain functional properties of two E. coli porins. Nature (London) 358:727–733.
38. Davis, B. D. 1987. Mechanism of bactericidal action of aminoglycosides. Microbiol. Rev. 51:341–350.
39. Death, A., L. Notley, and T. Ferenci. 1993. Derepression of LamB protein facilitates outer membrane permeation of carbohydrates into Escherichia coli under conditions of nutrient stress. J. Bacteriol. 175:1475–1483.
40. de Cock, H., R. Hendriks, T. de Vrije, and J. Tommassen. 1990. Assembly of an in vitro synthesized Escherichia coli outer membrane porin into its stable trimeric configuration. J. Biol. Chem. 265:4646–4651.
41. Delcour, A. H., J. Adler, C. Kung, and B. Martinac. 1992. Membrane-derived oligosaccharides (MDO’s) promote closing of an E. coli porin channel. FEBS Lett. 304:216–220.
42. Dempsey, D. E. 1990. The actions of melittin on membranes. Biochim. Biophys. Acta 1031:143–161.
43. Dinh, T., I. T. Paulsen, and M. H. Saier, Jr. 1994. A family of extracytoplasmic proteins that allow transport of large molecules across the outer membranes of gram-negative bacteria. J. Bacteriol. 176:3825–3831.
44. Dodson, K. W., F. Jacob-Dubuisson, R. T. Striker, and S. J. Hultgren. 1993. Outer membrane PapC molecular usher discriminately recognizes periplasmic chaperone-pilus subunit complexes. Proc. Natl. Acad. Sci. USA 90:3670–3674.
45. Donohue-Rolfe, A. M., and M. Schaechter. 1980. Translocation of phospholipids from the inner to the outer membrane of Escherichia coli. Proc. Natl. Acad. Sci. USA 77:1867–1871.
46. Duong, F., A. Lazdunski, B. Cami, and M. Murgier. 1992. Nucleotide sequence of a cluster of genes involved in synthesis and secretion of alkaline protease in Pseudomonas aeruginosa: relationship to other secretion pathways. Gene 121:47–54.
47. Eisele, J.-L., and J. P. Rosenbusch. 1990. In vitro folding and oligomerization of a membrane protein. Transition of bacterial porin from random coil to native conformation. J. Biol. Chem. 265:10217–10220.
48. Elsbach, P., and J. Weiss. 1993. The bactericidal/permeability-increasing protein (BPI), a potent element in host defense against gram-negative bacteria and lipopolysaccharide. Immunobiology 187:417–429.
49. Farmer, S., Z. S. Li, and R. E. W. Hancock. 1992. Influence of outer membrane mutations on susceptibility of Escherichia coli to the dibasic macrolide azithromycin. J. Antimicrob. Chemother. 29:27–33.
50. Ferenci, T., and K.-S. Lee. 1982. Directed evolution of the lambda receptor of Escherichia coli through affinity chromatography selection. J. Mol. Biol. 160:431–444.
51. Ferenci, T., W. Saurin, and M. Hofnung. 1988. Folding of maltoporin with respect to the outer membrane. J. Mol. Biol. 201:493–496.
52. Filip, C., G. Fletcher, J. L. Wulff, and C. F. Earhart. 1977. Solubilization of the cytoplasmic membrane of Escherichia coli by the ionic detergent sodium lauryl sarcosinate. J. Bacteriol. 115:717–722.
53. Forst, D., K. Schulein,T. Wacker, K. Diedrichs, W. Kreutz, R. Benz, and W. Welte. 1993. Crystallization and preliminary X-ray diffraction analysis of ScrY, a specific bacterial outer membrane porin. J. Mol. Biol. 229:258–262.
54. Fourel, D., S. Mizushima, and J.-M. Pages. 1992. Dynamics of the exposure of epitopes on OmpF, an outer membrane protein of Escherichia coli. Eur. J. Biochem. 206:109–114.
55. Freudl, R., H. Schwarz, Y.-D. Stierhof, K. Gamon, I. Hindennach, and U. Henning. 1986. An outer membrane protein (OmpA) of Escherichia coli K-12 undergoes a conformational change during export. J. Biol. Chem. 261:11355–11361.
56. Freundlieb, S., U. Ehmann, and W. Boos. 1988. Facilitated diffusion of p-nitrophenyl-α-d-maltohexaoside through the outer membrane of Escherichia coli. J. Biol. Chem. 263:314–320.
57. Funahara, Y., and H. Nikaido. 1980. Asymmetric localization of lipopolysaccharides on the outer membrane of Salmonella typhimurium. J. Bacteriol. 141:1463–1465.
58. Galanos, C., and O. Lüderitz. 1975. Electrodialysis of lipopolysaccharides and their conversion to uniform salt forms. Eur. J. Biochem. 54:603–610.
59. Garavito, R. M., and J. P. Rosenbusch. 1980. Three-dimensional crystals of an integral membrane protein: an initial X-ray analysis. J. Cell. Biol. 86:327–329.
60. George, A. M., and S. B. Levy. 1983. Amplifiable resistance to tetracycline, chloramphenicol, and other antibiotics in Escherichia coli: involvement of a non-plasmid-determined efflux of tetracycline. J. Bacteriol. 155:531–540.
61. Glaser, P., H. Sakamoto, J. Bellalou, A. Ullman, and A. Danchin. 1988. Secretion of cyclolysin, the calmodulin-sensitive adenylate cyclase-haemolysin bifunctional protein of Bordetella pertussis. EMBO J. 7:3997–4004.
62. Graeme-Cook, K. A., G. May, E. Bremer, and C. F. Higgins. 1989. Osmotic regulation of porin expression: a role for DNA supercoiling. Mol. Microbiol. 3:1287–1294.
63. Grewal, K. K., P. J. Warner, and P. H. Williams. 1982. An inducible outer membrane protein involved in aerobactin-mediated iron transport by colV strains of Escherichia coli. FEBS Lett. 140:27–30.
64. Grundstrom, T., S. Normark, and K.-E. Magnusson. 1980. Overproduction of outer membrane protein suppresses envA-induced hyperpermeability. J. Bacteriol. 144:844–847.
65. Hancock, R. E. W. 1981. Aminoglycoside uptake and mode of action—with special reference to streptomycin and gentamicin. II. Effects of aminoglycosides on cells. J. Antimicrob. Chemother. 8:429–445.
66. Hancock, R. E. W., and P. G. W. Wong. 1984. Compounds which increase the permeability of Pseudomonas aeruginosa outer membrane. Antimicrob. Agents Chemother. 26:48–52.
67. Hantke, K. 1983. Identification of an iron uptake system specific for coprogen and rhodotorulic acid in Escherichia coli K12. Mol. Gen. Genet. 191:301–306.
68. Heine, H.-G., G. Francis, K.-S. Lee, and T. Ferenci. 1988. Genetic analysis of sequences in maltoporin that contribute to binding domains and pore structure. J. Bacteriol. 170:1730–1738.
69. Henning, U., and R. Koebnik. 1994. Outer membrane proteins of Escherichia coli: mechanism of sorting and regulation of synthesis, p. 381–395. In J.-M. Ghuysen and R. Hakenbeck (ed.), Bacterial Cell Wall. New Comprehensive Biochemistry, vol. 27. Elsevier, Amsterdam.
70. Heyde, M., and R. C. Portalier. 1987. Regulation of outer membrane porin proteins of Escherichia coli by pH. Mol. Gen. Genet. 208:511–517.
71. Hirota, Y., H. Suzuki, Y. Nishimura, and S. Yasuda. 1977. On the process of cellular division in Escherichia coli: a mutant of E. coli lacking a murein-lipoprotein. Proc. Natl. Acad. Sci. USA 74:1417–1420.
72. Hirvas, L., P. Koski, and M. Vaara. 1991. Identification and sequence analysis of the gene mutated in the conditionally lethal outer membrane permeability mutant SS-C of Salmonella typhimurium. EMBO J. 10:1017–1023.
73. Hobot, J. A., E. Carlemalm, W. Villiger, and E. Kellenberger. 1984. Periplasmic gel: new concept resulting from the reinvestigation of bacterial cell envelope ultrastructure by new methods. J. Bacteriol. 160:143–152.
74. Hoenger, A., H. Gross, U. Aebi, and A. Engel. 1990. Localization of the lipopolysaccharides in metal-shadowed reconstituted lipid-porin membranes. J. Struct. Biol. 103:185–195.
75. Hooper, D. C., J. S. Wolfson, M. A. Bozza, and E. Y. Ng. 1992. Genetics and regulation of outer membrane protein expression by quinolone resistance loci nfxB, nfxC, and cfxB. Antimicrob. Agents Chemother. 36:1151–1154.
76. Inouye, M. (ed.). 1979. Bacterial Outer Membranes. John Wiley & Sons, Inc., New York.
77. Inouye, M. (ed.). 1987. Bacterial Outer Membranes as Model Systems. John Wiley & Sons, Inc., New York.
78. Jap, B. K., P. J. Walian, and K. Gehring. 1991. Structural architecture of an outer membrane channel as determined by electron crystallography. Nature (London) 350:167–170.
79. Jeanteur, D., J. H. Lakey, and F. Pattus. 1994. The porin superfamily: diversity and common features, p. 363–380. In J.-M. Ghuysen and R. Hakenbeck (ed.), Bacterial Cell Wall. New Comprehensive Biochemistry, vol. 27. Elsevier, Amsterdam.
80. Jones, N. C., and M. J. Osborn. 1977. Translocation of phospholipids between the outer and inner membrane of Salmonella typhimurium. J. Biol. Chem. 252:7405–7412.
81. Kamio, Y., and H. Nikaido. 1976. Outer membrane of Salmonella typhimurium: accessibility of phospholipid head groups to phospholipase C and cyanogen bromide activated dextran in the external medium. Biochemistry 15:2561–2570.
82. Kilmann, H., R. Benz, and V. Braun. 1993. Conversion of the FhuA transport protein into a diffusion channel through the outer membrane of Escherichia coli. EMBO J. 12:3007–3016.
83. Klena, J. D., and C. A. Schnaitman. 1993. Function of the rfb gene cluster and the rfe gene in the synthesis of O antigen by Shigella dysenteriae 1. Mol. Microbiol. 9:393–402.
84. Klose, M., H. Schwarz, S. MacIntyre, R. Freudl, M.-L. Eschbach, and U. Henning. 1988. Internal deletions in the gene for an Escherichia coli outer membrane protein define an area possibly important for recognition of the outer membrane by this polypeptide. J. Biol. Chem. 263:13291–13296.
85. Koch, A. L. 1971. The adaptive response of Escherichia coli to a feast and famine existence. Adv. Microb. Physiol. 6:147–217.
86. Kreusch, A., A. Neubuser, E. Schlitz, J. Weckesser, and G. E. Schulz. 1994. Structure of the membrane channel porin from Rhodopseudomonas blastica at 2.0 Å resolution. Protein Sci. 3:58–63.
87. Kuhn, H. M., E. Neter, and H. Mayer. 1983. Modification of the lipid moiety of the enterobacterial common antigen by the "Pseudomonas factor." Infect. Immun. 40:696–700.
88. Kusumoto, S., N. Kusunose, M. Imoto, and T. Shiba. 1990. Chemical synthesis of endotoxin. Adv. Exp. Med. Biol. 256:3–11.
89. Labischinski, H., G. Barnickel, H. Bradaczek, D. Naumann, E. T. Rietschel, and P. Giesbrecht. 1985. High state of order of isolated bacterial lipopolysaccharide and its possible contribution to the permeation barrier property of the outer membrane. J. Bacteriol. 169: 9–20.
90. Labischinski, H., D. Naumann, C. Schulz, S. Kusumoto, T. Shiba, E. T. Rietschel, and P. Giesbrecht. 1989. Comparative X-ray and Fourier-transform-infrared investigations of conformational properties of bacterial and synthetic lipid A of Escherichia coli and Salmonella minnesota as well as partial structures and analogues thereof. Eur. J. Biochem. 179:659–665.
91. Larsen, R. A., G. E. Wood, and K. Postle. 1993. The conserved proline-rich motif is not essential for energy transduction by Escherichia coli TonB protein. Mol. Microbiol. 10:943–953.
92. Lehrer, R. I., A. K. Lichtenstein, and T. Ganz. 1993. Defensins: antimicrobial and cytotoxic properties of mammalian cells. Annu. Rev. Immunol. 11:105–128.
93. Letoffe, S., P. Delepelaire, and C. Wandersman. 1990. Protease secretion by Erwinia chrysanthemum: the specific secretion functions are analogous to those of Escherichia coli α-hemolysin. EMBO J. 9:1375–1382.
94. Lewis, K. 1994. Multidrug resistance pumps in bacteria: variations on a theme. Trends Biochem. Sci. 19:119–123.
95. Leive, L. 1965. Release of lipopolysaccharide by EDTA treatment of E. coli. Biochem. Biophys. Res. Commun. 21:290–296.
96. Leive, L. 1974. The barrier function of gram-negative envelope. Ann. N. Y. Acad. Sci. 235:109–127.
96a. Li, X.-Z., D. Ma, D. M. Livermore, and H. Nikaido. 1994. Role of efflux pump(s) in intrinsic resistance of Pseudomonas aeruginosa: active efflux as a contributing factor to β-lactam resistance. Antimicrob. Agents Chemother. 38:1742–1752.
97. Liu, J., J. M. Rutz, J. B. Feix, and P. E. Klebba. 1993. Permeability properties of a large gated channel within the ferric enterobactin receptor, FepA. Proc. Natl. Acad. Sci. USA 90:10653–10657.
98. Livermore, D. M., and K. W. M. Davy. 1991. Invalidity for Pseudomonas aeruginosa of an accepted model of bacterial permeability to β-lactam antibiotics. Antimicrob. Agents Chemother. 35:916–921.
99. Lomovskaya, O., and K. Lewis. 1992. Emr, an Escherichia coli locus for multidrug resistance. Proc. Natl. Acad. Sci. USA 89:8938–8942.
100. Luckey, M., and H. Nikaido. 1980. Specificity of diffusion channels produced by phage receptor protein of Escherichia coli. Proc. Natl. Acad. Sci. USA 77:167–171.
101. Luckey, M., and H. Nikaido. 1980. Diffusion of solutes through channels produced by phage lambda receptor protein of Escherichia coli: inhibition by higher oligosaccharides of maltose series. Biochem. Biophys. Res. Commun. 64:687–693.
102. Lugowski, C., E. Romanowska, L. Kenne, and B. Lindberg. 1983. Identification of a trisaccharide repeating unit in the enterobacterial common antigen. Carbohydr. Res. 118:173–181.
103. Lugtenberg, B., and L. van Alphen. 1983. Molecular architecture and functioning of the outer membrane of Escherichia coli and other gram-negative bacteria. Biochim. Biophys. Acta 737:51–115.
104. Ma, Z., D. N. Cook, M. Alberti, N. G. Pon, H. Nikaido, and J. E. Hearst. 1993. Molecular cloning and characterization of acrA and acrE genes of Escherichia coli. J. Bacteriol. 175:6299–6313.
105. MacAlister, T. J., B. MacDonald, and L. I. Rothfield. 1983. The periseptal annulus: an organelle associated with cell division in gram-negative bacteria. Proc. Natl. Acad. Sci. USA 80:1372–1376.
106. MacIntyre, S., R. Freudl, M.-L. Eschbach, and U. Henning. 1988. An artificial hydrophobic sequence functions as either an anchor or a signal sequence at only one of the two positions within the Escherichia coli outer membrane protein OmpA. J. Biol. Chem. 263:19053–19059.
107. Maier, C., E. Bremer, A. Schmid, and R. Benz. 1988. Pore-forming activity of the Tsx protein from the outer membrane of Escherichia coli. Demonstration of a nucleoside-specific binding site. J. Biol. Chem. 263:2493–2499.
108. Marino, P. A., B. C. McGrath, and M. J. Osborn. 1991. Energy dependence of O-antigen biosynthesis in Salmonella typhimurium. J. Bacteriol. 173:3128–3133.
109. Martinez, M. B., F. J. Schendel, M. C. Flickinger, and G. L. Nelsestuen. 1992. Kinetic properties of enzyme populations in vivo: alkaline phosphatase of the Escherichia coli periplasm. Biochemistry 31:11500–11509.
110. Matsuyama, S.-I., and S. Mizushima. 1985. Construction and characterization of a deletion mutant lacking micF, a proposed regulatory gene for OmpF synthesis in Escherichia coli. J. Bacteriol. 162:1196–1202.
111. Mauro, A. M. Blake, and P. Labarca. 1988. Voltage gating of conductance in lipid bilayers induced by porin from outer membrane of Neisseria gonorrhoeae. Proc. Natl. Acad. Sci. USA 85:1071–1075.
112. Mayer, H., and G. Schmidt. 1979. Chemistry and biology of the enterobacterial common antigen (ECA). Curr. Top. Microbiol. 85:99–153.
113. McGrath, B. C., and M. J. Osborn. 1991. Evidence for energy-dependent transposition of core lipopolysaccharide across the inner membrane of Salmonella typhimurium. J. Bacteriol. 173:3134–3137.
114. Medeiros, A. A., T. F. O’Brien, E. Y. Rosenberg, and H. Nikaido. 1987. A strain of Salmonella typhimurium became more resistant to cephalosporins during therapy because it lost its OmpC porin. J. Infect. Dis. 156:751–757.
115. Meier-Dieter, U., K. Barr, R. Starman, L. Hatch, and P. D. Rick. 1992. Nucleotide sequence of the Escherichia coli rfe gene involved in the synthesis of enterobacterial common antigen. Molecular cloning of the rfe-rff gene cluster. J. Biol. Chem. 267:746–753.
116. Migorodskaya, O. A., A. A. Shevchenko, K. O. M. A. Abdalla, I. V. Chernushevich, T. A. Egorov, A. X. Musoliamov, V. N. Kokryakov, and O. V. Shamova. 1993. Primary structure of three cationic peptides from porcine neutrophils. Sequence determination by the combined usage of electrospray ionization mass spectrometry and Edman degradation. FEBS Lett. 330:339–342.
117. Misra, R., and S. A. Benson. 1988. Genetic identification of the pore domain of the OmpC porin of Escherichia coli K-12. J. Bacteriol. 170:3611–3617.
118. Misra, R., and S. A. Benson. 1989. A novel mutation, cog, which results in production of a new porin protein (OmpG) of Escherichia coli K-12. J. Bacteriol. 171:4105–4111.
119. Miura, T., and S. Mizushima. 1968. Separation by density gradient centrifugation of two types of membranes from spheroplast membrane of Escherichia coli K12. Biochim. Biophys. Acta 150:159–161.
120. Mizuno, T., M.-Y. Chou, and M. Inouye. 1984. A unique mechanism regulating gene expression: translational inhibition by a complementary RNA transcript (micRNA). Proc. Natl. Acad. Sci. USA 81:1966–1970.
121. Mizuno, T., and S. Mizushima. 1990. Signal transduction and gene regulation through the phsophorylation of two regulatory components: the molecular basis for the osmotic regulation of the porin genes. Mol. Microbiol. 4:1077–1082.
122. Morona, R., M. Klose, and U. Henning. 1984. Escherichia coli K-12 outer membrane protein (OmpA) as a bacteriophage receptor: analysis of mutant genes expressing altered proteins. J. Bacteriol. 159:570–578.
123. Mühlradt, P. F., and J. R. Golecki. 1975. Asymmetric distribution and artificial reorientation of lipopolysaccharide in the outer membrane bilayer of Salmonella typhimurium. Eur. J. Biochem. 51:343–352.
124. Mühlradt, P. F., J. Menzel, J. R. Golecki, and V. Speth. 1973. Outer membrane of Salmonella. Site of export of newly synthesized lipopolysaccharide on the surface. Eur. J. Biochem. 35:371–481.
125. Mulford, C. A., and M. J. Osborn. 1983. An intermediate step in translocation of lipopolysaccharide to the outer membrane of Salmonella typhimurium. Proc. Natl. Acad. Sci. USA 80:1159–1163.
126. Nakamura, H. 1968. Genetic determination of resistance to acriflavine, phenethyl alcohol, and sodium dodecyl sulfate in Escherichia coli. J. Bacteriol. 96:987–996.
127. Neu, H. C., and L. A. Heppel. 1965. The release of enzymes from Escherichia coli by osmotic shock and during the formation of spheroplasts. J. Biol. Chem. 240:3685–3692.
128. Nikaido, H. 1976. Outer membrane of Salmonella typhimurium: transmembrane diffusion of some hydrophobic substances. Biochim. Biophys. Acta 433:118–132.
129. Nikaido, H. 1990. Permeability of lipid domains of bacterial membranes, p. 165–190. In R. C. Aloia, C. C. Curtain, and L. M. Gordon (ed.), Membrane Transport and Information Storage. Advances in Membrane Fluidity, vol. 4. Alan R. Liss, New York.
130. Nikaido, H. 1992. Porins and specific channels of bacterial outer membranes. Mol. Microbiol. 6:435–442.
131. Nikaido, H. 1994. Prevention of drug access to bacterial targets: permeability barriers and active efflux. Science 264:382–388.
132. Nikaido, H. 1994. Isolation of outer membranes. Methods Enzymol. 235:225–234.
133. Nikaido, H. 1994. Porins and specific diffusion channels in bacterial outer membrane. J. Biol. Chem. 269:3905–3908.
134. Nikaido, H., P. Bavoil, and Y. Hirota. 1977. Outer membrane of gram-negative bacteria. XV. Transmembrane diffusion rates in lipoprotein-deficient mutants of Escherichia coli. J. Bacteriol. 132:1045–1047.
135. Nikaido, H., and T. Nakae. 1979. The outer membrane of gram-negative bacteria. Adv. Microb. Physiol. 20:163–250.
136. Nikaido, H., K. Nikaido, and S. Harayama. 1991. Identification and characterization of porins in Pseudomonas aeruginosa. J. Biol. Chem. 266:770–779.
137. Nikaido, H., and E. Y. Rosenberg. 1981. Effect of solute size on diffusion rates through the transmembrane pores of the outer membrane of Escherichia coli. J. Gen. Physiol. 77:121–135.
138. Nikaido, H., and E. Y. Rosenberg. 1983. Porin channels in Escherichia coli: studies with liposomes reconstituted from purified proteins. J. Bacteriol. 153:241–252.
139. Nikaido, H., and E. Y. Rosenberg. 1990. Cir and Fiu proteins in the outer membrane of Escherichia coli catalyze the transport of monomeric catechols: study with beta-lactam antibiotics containing catechol and analogous groups. J. Bacteriol. 172:1361–1367.
140. Nikaido, H., E. Y. Rosenberg, and J. Foulds. 1983. Porin channels in Escherichia coli: studies with β-lactams in intact cells. J. Bacteriol. 153:232–240.
141. Nikaido, H., Y. Takeuchi, S. Ohnishi, and T. Nakae. 1977. Outer membrane of Salmonella typhimurium: electron spin resonance studies. Biochim. Biophys. Acta 465: 152–164.
142. Nikaido, H., and M. Vaara. 1985. Molecular basis of bacterial outer membrane permeability. Microbiol. Rev. 49:1–32.
143. Osborn, M. J. 1979. Biosynthesis and assembly of the lipopolysaccharide of the outer membrane, p. 15–34. In M. Inouye (ed.), Bacterial Outer Membranes. John Wiley & Sons, Inc., New York.
144. Osborn, M. J., J. E. Gander, E. Parisi, and J. Carson. 1972. Mechanism of assembly of the outer membrane of Salmonella typhimurium. Isolation and characterization of cytoplasmic and outer membrane. J. Biol. Chem. 247:3962–3972.
145. Papahadjopolos, D., and A. Portis. 1978. Calcium-induced lipid phase transition and membrane fusion. Ann. N.Y. Acad. Sci. 308:50–65.
146. Paul, S., K. Chaudhuri, A. N. Chatterjee, and J. Das. 1992. Presence of exposed phospholipids in the outer membrane of Vibrio cholerae. J. Gen. Microbiol. 138:755–761.
147. Plésiat, P., and H. Nikaido. 1992. Outer membranes of gram-negative bacteria are permeable to steroid probes. Mol. Microbiol. 6:1323–1333.
148. Poole, K., K. Krebes, C. McNally, and S. Neshat. 1993. Multiple antibiotic resistance of Pseudomonas aeruginosa: evidence for involvement of an efflux operon. J. Bacteriol. 175:7363–7372.
149. Postle, K. 1993. TonB protein and energy transduction between membranes. J. Bioenerg. Biomembr. 25:591–601.
150. Pugsley, A. P. 1993. The complete general secretory pathway in gram-negative bacteria. Microbiol. Rev. 57:50–108.
151. Pugsley, A. P., and C. A. Schnaitman. 1978. Outer membrane proteins of Escherichia coli. VII. Evidence that bacteriophage-directed protein 2 functions as a pore. J. Bacteriol. 133:1181–1189.
152. Raetz, C. R. H. 1990. Biochemistry of endotoxins. Annu. Rev. Biochem. 59:129–170.
153. Reeves, P. 1994. Biosynthesis and assembly of lipopolysaccharides, p. 281–317. In J.-M. Ghuysen and R. Hakenbeck (ed.), Bacterial Cell Wall. New Comprehensive Biochemistry, vol. 27. Elsevier, Amsterdam.
154. Richmond, M. H., D. C. Clark, and S. Wotton. 1976. Indirect method for assessing the penetration of beta-lactamase-nonsusceptible penicillins and cephalosporins in Escherichia coli strains. Antimicrob. Agents Chemother. 10:215–218.
155. Ried, G., I. Hindennach, and U. Henning. 1990. Role of lipopolysaccharide in assembly of Escherichia coli outer membrane proteins OmpA, OmpC, and OmpF. J. Bacteriol. 172:6048–6053.
156. Rosenbusch, J. P. 1974. Characterization of the major envelope protein from Escherichia coli. Regular arrangement on the peptidoglycan and unusual dodecylsulfate binding. J. Biol. Chem. 249:8019–8029.
157. Rosner, J. L. 1985. Nonheritable resistance to chloramphenicol and other antibiotics induced by salicylates and other chemotactic repellents in Escherichia coli K-12. Proc. Natl. Acad. Sci. USA 82:8771–8774.
158. Russel, M. 1991. Filamentous phage assembly. Mol. Microbiol. 5:1607–1613.
159. Rutz, J. M., J. Liu, J. A. Lyons, S. K. Goranson, M. A. McIntosh, J. B. Feix, and P. E. Klebba. 1992. Formation of a gated channel by a ligand-specific transport protein in the bacterial outer membrane. Science 258:471–475.
160. Saint, N., E. De, S. Julien, N. Orange, and G. Molle. 1993. Ionophore properties of OmpA of Escherichia coli. Biochim. Biophys. Acta 1145:119–123.
161. Sawai, T., S. Hirano, and Y. Yamaguchi. 1987. Repression of porin synthesis by salicylate in Escherichia coli, Klebsiella pneumoniae, and Serratia marcescens. FEMS Microbiol. Lett. 40:233–237.
162. Schindler, H., and J. P. Rosenbusch. 1978. Matrix protein from Escherichia coli outer membranes forms voltage-controlled channels in lipid bilayers. Proc. Natl. Acad. Sci. USA 75:3751–3755.
163. Schindler, H., and J. P. Rosenbusch. 1981. Matrix protein in planar membranes: clusters of channels in a native environment and their functional reassembly. Proc. Natl. Acad. Sci. USA 78:2302–2306.
163a. Schirmer, T., T. A. Keller, Y.-F. Wang, and J.P. Rosenbush. 1995. Structural basis for sugar translocation through maltoporin channels at 3.1 Å resolution. Science 267:512–514.
164. Schnaitman, C. A. 1970. Protein composition of the cell wall and cytoplasmic membrane of Escherichia coli. J. Bacteriol. 104:890–901.
165. Schnaitman, C. A. 1973. Outer membrane proteins of Escherichia coli. I. Effect of preparation conditions on the migration of protein in polyacrylamide gels. Arch. Biochem. Biophys. 157:541–552.
166. Schnaitman, C. A., and J. D. Klena. 1993. Genetics of lipopolysaccharide biosynthesis in enteric bacteria. Microbiol. Rev. 57:655–682.
167. Schülein, K. A., and R. Benz. 1990. LamB (maltoporin) of Salmonella typhimurium: isolation, purification and comparison of sugar binding with LamB of Escherichia coli. Mol. Microbiol. 4:625–632.
168. Schülein, K., A. Schmid, and R. Benz. 1991. The sugar-specific outer membrane channel ScrY contains functional characteristics of general diffusion pores and substrate-specific porins. Mol. Microbiol. 5:2233–2241.
169. Schulz, G. E. 1993. Bacterial porins: structure and function. Curr. Opin. Cell Biol. 5:701–707.
170. Schweizer, M., I. Hindennach, W. Garten, and U. Henning. 1978. Major proteins of the Escherichia coli outer cell envelope membrane. Interaction of protein II* with lipopolysaccharide. Eur. J. Biochem. 82:211–217.
171. Sen, K., J. Hellman, and H. Nikaido. 1988. Porin channels in intact cells of Escherichia coli are not affected by Donnan potentials across the outer membrane. J. Biol. Chem. 263:1182–1187.
172. Sen, K., and H. Nikaido. 1990. In vitro trimerization of OmpF porin secreted by spheroplasts of Escherichia coli. Proc. Natl. Acad. Sci. USA 87:743–747.
173. Sen, K., and H. Nikaido. 1991. Lipopolysaccharide structure required for in vitro trimerization of Escherichia coli OmpF porin. J. Bacteriol. 173:926–928.
174. Sen, K., and H. Nikaido. 1991. Trimerization of an in vitro synthesized OmpF porin of Escherichia coli outer membrane. J. Biol. Chem. 266:11295–11300.
175. Seydel, U., H. Labischinski, M. Kastowsky, and K. Brandenburg. 1993. Phase behavior, supramolecular structure, and molecular conformation of lipopolysaccharide. Immunobiology 187:191–211.
176. Smit, J., Y. Kamio, and H. Nikaido. 1975. Outer membrane of Salmonella typhimurium: chemical analysis and freeze-fracture studies with lipopolysaccharide mutants. J. Bacteriol. 124:942–958.
177. Smit, J., and H. Nikaido. 1978. Outer membrane of Gram-negative bacteria. XVII. Electron microscopic studies on sites of insertion of porins and growth of cell surface in Salmonella typhimurium. J. Bacteriol. 135:687–702.
178. Stock, J. B., A. J. Ninfa, and A. M. Stock. 1989. Protein phosphorylation and regulation of adaptive responses in bacteria. Microbiol. Rev. 53:450–490.
179. Stock, J. B., B. Rauch, and S. Roseman. 1977. Periplasmic space in Salmonella typhimurium and Escherichia coli. J. Biol. Chem. 252:7850–7861.
180. Struyvé, M., M. Moons, and J. Tommassen. 1991. Carboxy-terminal phenylalanine is essential for the correct assembly of a bacterial outer membrane protein. J. Mol. Biol. 218:141–148.
181. Sugawara, E., and H. Nikaido. 1992. Pore-forming activity of OmpA protein of Escherichia coli. J. Biol. Chem. 267:2507–2511.
182. Sugawara, E., and H. Nikaido. 1994. OmpA protein of Escherichia coli outer membrane occurs in open-channel and closed-channel forms. J. Biol. Chem. 269:17981–17987.
183. Sukharev, S. I., P. Blount, B. Martinac, F. R. Blattner, and C. Kung. 1994. A large conductance mechanosensitive channel in E. coli encoded by mscL alone. Nature (London) 368:265–268.
184. Sutcliffe, J., R. Blumenthal, A. Walter, and J. Foulds. 1983. Escherichia coli outer membrane protein K is a porin. J. Bacteriol. 156:867–872.
185. Takayama, K., Z. Z. Din, P. Mukerjee, P. H. Cooke, and T. N. Kirkland. 1990. Physicochemical properties of the lipopolysaccharide unit that activates B lymphocytes. J. Biol. Chem. 265:14023–14029.
186. Takayama, K., and N. Qureshi. 1992. Chemical structure of lipid A, p. 44–65. In D. C. Morrison and J. L. Ryan (ed.), Bacterial Endotoxic Lipopopolysaccharides, vol. I. CRC Press, Boca Raton, Fla.
187. Takeuchi, Y., and H. Nikaido. 1981. Persistence of segregated phospholipid domains in phospholipid-lipopolysaccharide mixed bilayers: studies with spin-labeled phospholipids. Biochemistry 20:523–529.
188. Tanford, C. 1973. The Hydrophobic Effect. Formation of Micelles and Biological Membranes, p. 100. Wiley-Interscience, New York.
189. Thomas, A. D., and I. R. Booth. 1992. The regulation of expression of the porin gene ompC by acid pH. J. Gen. Microbiol. 138:1829–1835.
190. Todt, J. C., W. J. Rocque, and E. J. McGroarty. 1992. Effects of pH on bacterial porin function. Biochemistry 31:10471–10478.
191. Trias, J., J. Dufresne, R. C. Levesque, and H. Nikaido. 1989. Decreased outer membrane permeability in imipenem-resistant mutants of Pseudomonas aeruginosa. Antimicrob. Agents Chemother. 33:1201–1206.
192. Trias, J., and H. Nikaido. 1990. Protein D2 channel of the Pseudomonas aeruginosa outer membrane has a binding site for basic amino acids and peptides. J. Biol. Chem. 265:15680–15684.
193. Tuckman, M., and M. S. Osburne. 1992. In vivo inhibition of TonB-dependent processes by a TonB box consensus peptapeptide. J. Bacteriol. 174:320–323.
194. Ueki, T., T. Mitsui, and H. Nikaido. 1979. X-ray diffraction studies of outer membrane of Salmonella typhimurium. J. Biochem. (Tokyo) 85:173–182.
195. Vaara, M. 1990. Do salicylates and ascorbate increase the outer membrane permeability to hydrophobic antibiotics in Pseudomonas aeruginosa? Drugs Exp. Clin. Res. 16:569–574.
196. Vaara, M. 1990. The effect of oligolysines Lys-3, Lys-4, and Lys-5 on the outer membrane permeability of Pseudomonas aeruginosa. FEMS Microbiol. Lett. 67:15–20.
197. Vaara, M. 1992. Agents that increase the permeability of the outer membrane. Microbiol. Rev. 56, 395–411.
198. Vaara, M. 1993. Antibiotic-supersusceptible mutants of Escherichia coli and Salmonella typhimurium. Antimicrob. Agents Chemother. 37:2255–2260.
199. Vaara, M., and J. Jaakkola. 1989. Sodium hexametaphosphate sensitizes Pseudomonas aeruginosa, several other species of Pseudomonas, and Escherichia coli to hydrophobic drugs. Antimicrob. Agents Chemother. 33:1741–1747.
200. Vaara, M., W. Z. Plachy, and H. Nikaido. 1990. Partitioning of hydrophobic probes into lipopolysaccharide bilayers. Biochim. Biophys. Acta 1024:152–158.
201. Vaara, M., and T. Vaara. 1983. Polycations sensitize enteric bacteria to antibiotics. Antimicrob. Agents Chemother. 24:107–113.
202. Vaara, M., and T. Vaara. 1983. Polycations as outer membrane disorganizing agents. Antimicrob. Agents Chemother. 24:114–122.
203. Vaara, M., T. Vaara, M. Jensen, I. Helander, M. Nurminen, E. T. Rietschel, and P. H. Mäkelä. 1981. Characterization of the lipopolysaccharide from the polymyxin-resistant pmrA mutants of Salmonella typhimurium. FEBS Lett. 129:145–149.
204. Vogel, H., and F. Jähnig. 1986. Models for the structure of outer-membrane proteins of Escherichia coli derived from Raman spectroscopy and prediction methods. J. Mol. Biol. 190:191–199.
205. Vuorio, R., and M. Vaara. 1992. The lipid A biosynthesis mutation lpxA2 of Escherichia coli results in drastic antibiotic supersusceptibility. Antimicrob. Agents Chemother. 36:826–829.
206. Wandersman, C., and P. Delepelaire. 1990. TolC, an Escherichia coli outer membrane protein required for hemolysin secretion. Proc. Natl. Acad. Sci. USA 87:4776–4780.
207. Webster, R. E. 1990. The tol gene products and the import of macromolecules into Escherichia coli. Mol. Microbiol. 5:1005–1011.
208. Weiss, M. S., U. Abele, J. Weckesser, W. Welte, E. Schiltz, and G. E. Schulz. 1991. Molecular architecture and electrostatic properties of a bacterial porin. Science 254:1627–1629.
209. Weiss, M. S., and G. E. Schulz. 1992. Structure of porin refined at 1.8 Å resolution. J. Mol. Biol. 227:493–509.
210. Whitfield, C., R. E. W. Hancock, and J. W. Costerton. 1983. Outer membrane protein K of Escherichia coli: purification and pore-forming properties in lipid bilayer membranes. J. Bacteriol. 156:873–879.
211. Zimmermann, W., and A. Rosselet. 1977. Function of the outer membrane of Escherichia coli as a permeability barrier to beta-lactam antibiotics. Antimicrob. Agents Chemother. 12:368–372.
Chapter5
