Solute and Ion Transport: Outer Membrane Pores and Receptors

SATOSHI YAMASHITA AND SUSAN K. BUCHANAN*

[SECTION EDITOR: PETER MALONEY]

Posted 31 March, 2010

Laboratory of Molecular Biology, National Institute of Diabetes and Digestive and Kidney Disease, National Institutes of Health, Bethesda, MD 20892

*Corresponding author. Mailing address: Laboratory of Molecular Biology, National Institute of Diabetes and Digestive and Kidney Disease, National Institutes of Health, Bethesda, MD 20892. Phone: (301) 594-9222, E-mail: This e-mail address is being protected from spambots. You need JavaScript enabled to view it

INTRODUCTION

The cell envelopes of gram-negative bacteria consist of two membrane systems: the inner membrane (IM) and outer membrane (OM). The two membranes have chemically different characteristics; while the IM is composed of various phospholipids, the OM has an asymmetric structure with the inner leaflet composed of phospholipids and the outer leaflet composed primarily of lipopolysaccharide molecules. The lipopolysaccharide molecules are linked together through divalent cations such as Mg²⁺, creating an impermeable barrier that protects the bacterium from toxins and harsh environmental conditions (71). Because of the impermeable nature of the OM, it contains many kinds of transporters for selective transport of various solutes into the periplasm. Structural studies of OM channels indicate that almost all OM channels form transmembrane β-barrels, rather than bundles of α-helices as are found in receptors, channels, and transporters in the IM. This means that the transport mechanisms of OM channels are unique and differ from those found in the IM. In this chapter, we describe four categories of transport systems across the OM: (i) general and specific porins; (ii) specific transporters; (iii) TonB-dependent transporters; and (iv) export channels. Porins are the simplest channels, transporting small solutes down a concentration gradient by passive or facilitated diffusion. Specific transporters facilitate the uptake of targeted solutes by intimately interacting with the substrates. TonB-dependent transporters are involved in high-affinity active transport of iron chelates and vitamin B₁₂ against a concentration gradient. OM export channels serve to defend the cell from environmental stresses such as toxic compounds. In the section on export channels we will discuss type I secretion in terms of toxin export, but the remaining OM secretion systems are discussed in other chapters.

PORINS

The outer membranes of gram-negative bacteria contain many diffusion channels called porins (58). Porins are integral membrane proteins that are composed of antiparallel β-strands that fold into a β-barrel, such that the first strand interacts with the last strand. The β-strands are amphipathic; the inside is hydrophilic and the outside is hydrophobic. The pore formed by the barrel is filled with water and allows diffusion of general nutrients such as sugars, ions, and amino acids across the OM by a gradient of solute concentration. Porins allow the passage of solutes with molecular mass up to about 600 Da (73). In Escherichia coli, OmpF, OmpC, and PhoE are known as general porins. Although the three proteins have high sequence similarity (about 60% identical), there is a solute preference based on size or electrostatics for each of the porins. OmpF and OmpC prefer cations, whereas OmpF allows diffusion of somewhat larger molecules. Because of the larger pore size, production of OmpF is more highly regulated than OmpC when the bacteria are stressed through osmotic strength, acidic pH, and/or toxic chemicals (16, 32). PhoE prefers to transport anions and it is expressed only under phosphate starvation (72). The porin channels appear to be open most of the time. It is known that the channels can be closed either at high voltage (~100 mV) or in the presence of membrane-derived oligosaccharides in in vitro experiments (24). However, the physiological significance of these phenomena is not clear. Crystallographic studies have shown that the general porins form 16-stranded β-barrels with eight extracellular loops and they exist as trimers, consisting of three β-barrels with three independent pores (Fig. 1) (6, 21). Extracellular loop 3 is especially long and folds into the barrel to constrict the pore and confer substrate selectivity. The OmpF pore has a diameter of 7 × 11 Å at its most narrow point. Mutagenesis studies have shown that mutations in loop 3 critically affect the rate of substrate diffusion (68, 78). In Salmonella enterica serovar Typhimurium the major porin is called OmpD. However, OmpD functions to export substances rather than to import them: OmpD works with the IM protein pump YddG to export toxic compounds out of the cell (53, 69).

Fig. 1Crystal structure of OmpC. (a) Monomer structure as viewed from the side. (b) Trimer structure as viewed from extracellular surface. Extracellular loop 3 (red) folds into the barrel to constrict the pore.

LamB and ScrY are specific porins: they use facilitated diffusion to transport a narrow range of desirable substrates. They consist of 18-stranded instead of 16-stranded β-barrels, but they form trimers just as general porins do. LamB is thought to primarily import maltose and higher oligosaccharides of the maltose series (61). The transported sugars are captured by a periplasmic binding protein (PBP) and imported by an IM ATP-binding cassette (ABC)-type transport complex (31, 65). Because LamB expression levels increase when nutrients are scarce, it is thought that this protein can transport a wide variety of sugars such as glucose, lactose, arabinose, and even glycerol (23). Crystallization studies of E. coli LamB resulted in structures with bound substrates such as maltose, trehalose, melibiose, and sucrose, although the latter is not a preferred substrate for LamB (29, 86). The channel is partially closed (narrowing to about 5 to 6 Å in diameter) by three extracellular loops that fold into the barrel pore similarly to extracellular loop 3 in general porins. The pore is lined with aromatic residues that form a “greasy slide,” contributing to facilitated diffusion of sugars through hydrophobic interactions (29). ScrY is known as a sucrose channel that has approximately 25% sequence identity with LamB. The crystal structure of ScrY from serovar Typhimurium showed that it has a wider pore than LamB, with a diameter of approximately 11 Å (37). Reflecting its larger pore size, functional studies showed that ScrY can transport a wide variety of sugars in addition to sucrose (79). Therefore, substrate recognition for both LamB and ScrY is not very specific.

In addition to the porins described above, there are also putative porins that have been identified through sequence similarity or porin-like activities. BglH from E. coli is a 58-kDa protein that forms a trimer (3). This protein is thought to be expressed as a part of the bglGFB operon that is related to the degradation of aryl-β-glucosides. RafY is a 50.7-kDa protein that exists in E. coli strains that contain raffinose degradation genes (2). However, the specificity for trisaccharides, including raffinose, is not clear because the channel is not blocked by any potential substrates. OmpG is a monomeric protein that shows no sequence homology to the other porins and exists in both E. coli and Salmonella species. When LamB is disrupted, the uptake of oligosaccharides is facilitated by OmpG (54). The crystal structure of OmpG was recently solved and it was shown that this protein forms a 14-stranded β-barrel (Fig. 2) (84). The open and closed conformations of OmpG and functional studies suggested that this protein regulates solute diffusion in a pH-dependent manner. OmpL from E. coli K-12 has homology to KdgM, which is an oligosaccharide channel of Erwinia. In one study, null mutations in the ompL gene made a dsbA mutant more sensitive to DTT (22), but another study concluded that null mutations in the ompL do not exhibit unusual phenotypes (70).

Fig. 2Crystal structure of OmpG. (a) Side view of the structure in closed conformation. (b) The structures in closed (left) and open (right) conformations as viewed from the extracellular surface.

A typical gram-negative bacterial genome also codes for a number of smaller OM proteins, having only 8 or 10 β-strands. OmpA is classic 8-stranded β-barrel protein with a large globular domain in the periplasm. This protein is expressed at very high levels and has been identified as a major component of the outer membrane of E. coli (13). The primary function of OmpA is to maintain the integrity of the cell envelope by binding peptidoglycan, but other possible functions include adhesion, invasion, and participation in biofilm formation (5, 64). Although in vitro experiments have shown that OmpA forms ion-permeable pores in lipid bilayers, the crystal structure suggests that no pore exists in the barrel interior (62). However, a nuclear magnetic resonance spectrometry structure, which reveals both static and dynamic regions of a protein, showed that the β-barrel domain of OmpA has flexibility along the axis of the barrel and could allow the barrel to form a pore for channel function (36). Another small β-barrel proposed to function as a channel is OmpW of E. coli. This protein is also an 8-stranded β-barrel and is suggested to be involved in protecting bacteria from environmental stresses (56). This protein has sequence similarity to proteins that are related to alkane and naphthalene degradation in Pseudomonas species (30, 80). These similarities therefore suggest a role for OmpW in the transport of hydrophobic molecules. Single-channel conductance experiments of OmpW showed that this protein forms channels and that the channels could be blocked by the detergent molecules (40). The crystal structure of OmpW did not have a continuous channel spanning the membrane, but it had a deep hydrophobic binding pocket on the extracellular side that contained a detergent molecule (Fig. 3). The interior of OmpW is unusually hydrophobic and there is a gap in the middle of barrel that is connected to the binding pocket for the detergent molecule. The authors suggested that the gap might allow lateral passage of small hydrophobic molecules through the barrel directly into the lipid bilayer, followed by diffusion into the periplasm, that would possibly be assisted by a PBP. Other small OM proteins of E. coli are OmpX (adhesion, 8-stranded) (83), PagP (acyltransferase, 8-stranded) (1), OmpT (protease, 10-stranded) (82), and OmpLA (phospholipase, 10-stranded obligate dimer) (76).

Fig. 3Crystal structure of OmpW. (a) The structure with a bound lauryldimethylamine-N-oxide (LDAO) molecule (space filling representation) as viewed from the side. The position of gap in the middle of barrel wall is in red. The putative position of the OM is indicated by horizontal lines. (b) Surface representation as viewed from the same direction and in the same colors as (a). The gap is connected to the binding pocket of the bound LDAO molecule and is thought to form a tunnel for passage of the putative substrate.

SPECIFIC TRANSPORTERS

The specific transporters are integral membrane proteins that have obvious binding sites for specific substrates. Two protein families in this category have been characterized: Tsx for nucleoside transport and FadL for fatty acid transport. These proteins are smaller than general porins but are mechanistically similar to porins, since substrate transport does not require an energy source.

Tsx is a nucleoside-specific import channel. Tsx from E. coli is approximately 31 kDa and was identified as a receptor for phage T-six. For subsequent import of the substrates into the cytoplasm, only the IM proteins NupC and NupG are needed (60). Tsx preferentially imports nucleosides and deoxynucleosides, in preference to free nucleobases or monophosphate nucleosides (8). Purified Tsx reconstituted into planar lipid membranes shows very low channel conductance (52). This observation indicates a narrow pore size that was confirmed by the crystal structure (Fig. 4). It was shown that Tsx folds into a 14-stranded β-barrel with seven extracellular loops (88). Tsx has a continuous narrow channel with dimensions of 10 to 12 Å in the long direction and 3 to 5 Å in the short direction. The Tsx pore is lined by a number of aromatic residues that narrow the channel and contribute to nucleoside binding. The structures of Tsx bound with different nucleosides revealed two different binding sites: one is located close to the extracellular side of the channel and the other is in the middle of the channel. Mutagenesis studies showed that both aromatic residues and ionizable residues in the channel contribute to substrate binding. It was suggested that binding and release of nucleosides by the weak binding sites would result in the movement substrates through the channel. Unlike general porins, no extracellular loops fold inward to constrict the pore.

Fig. 4Crystal structure of Tsx. (a) Side view of the structure bound with two thymidine molecules (space filling representation). (b) Surface representation viewed from the extracellular surface without bound substrates.

The FadL family of proteins functions to transport hydrophobic compounds. FadL from E. coli was identified as a transporter of long-chain fatty acids (LCFAs), which are important sources of metabolic energy and carbon. The uptake of LCFAs is also needed when bacteria infect a host cell because of the high concentration of LCFAs released from a host cell by the action of bacterial phospholipases. Efficient long-chain fatty acid uptake by FadL is predicted to suppress the local inflammatory host response (9). The affinity of FadL for LCFAs is in the submicromolar range (47). It is not clear whether there is a PBP for this system, but it has been suggested that the periplasmic protein Tsp, which has homology to the retinoid-binding protein, may be involved in fatty acid binding (4). After transiting the periplasm, LCFAs insert into the outer leaflet of the IM. A flip-flop mechanism is thought to reposition LCFAs in the inner leaflet of the IM, where they are activated by the IM-associated fatty acyl-CoA synthetase FadD (43).

E. coli FadL has a molecular mass of 46 kDa and exists as a monomer. The crystal structure showed that this protein forms a 14-stranded β-barrel (Fig. 5) (81). There is no continuous channel thorough the barrel. The N-terminal 42 residues of FadL form a small domain with short helices that plugs the barrel. A hydrophobic groove binds detergent molecules in the extracellular region. On the inside, FadL has a hydrophobic pocket containing another detergent molecule that constitutes the putative high-affinity binding site. A second crystal structure of FadL in a different conformation was also determined. Conformational differences between the two structures were found in the N-terminal region near the hydrophobic pocket and suggested the formation of a channel for passage through the barrel. Based on these observations, substrate movement was proposed as follows: (i) substrate is captured by the hydrophobic groove; (ii) substrate diffuses into the high-affinity binding pocket; and (iii) conformational changes in the hydrophobic pocket create the initial part of the transport pathway. It is not clear how additional conformational changes occur to complete the movement of LCFAs to the periplasm. Other homologs of FadL transport various aromatic compounds (for example, TbuX of Ralstonia picketii is a toluene channel [41] and XylN of Pseudomonas putida is a xylene channel [42]). The mechanistic differences between FadL and these homologs would be of interest.

Fig. 5Crystal structure of FadL. The bound detergent molecules are indicated in space filling representation. The positions of detergent molecules indicate a putative movement of the substrate from the extracellular side to the middle of barrel. The periplasmic side of the barrel is completely closed by the N-terminal hatch domain (blue). C₈E₄, tetraethylene glycol monooctyl ether; LDAO, lauryldimethylamine-N-oxide.

TonB-DEPENDENT TRANSPORTERS

Gram-negative bacteria, like other organisms, require some nutrients that are scarce in the environment (e.g., iron). Because the process of facilitated diffusion cannot meet the cell's requirements for nutrients like iron, bacteria have evolved active transport systems called TonB-dependent transporters to accumulate sufficient Fe³⁺ and vitamin B₁₂ against a concentration gradient. Energy is required for transport of the substrate into the periplasm, and it is supplied through association with a complex of three proteins, TonB, ExbB, and ExbD, that form an inner membrane complex (Fig. 6) (10). Because of the low availability and insolubility of Fe³⁺ at physiological pH in environment, bacteria produce and secrete iron-chelating compounds called siderophores. E. coli produces a single siderophore, enterobactin, that is transported by the ferric enterobactin transporter FepA. FepA imports Fe³⁺ chelated to enterobactin (a cyclic triester of 2,3-dihydroxybenzoylserine with a molecular mass of 719 Da). There are other transporters of this family in E. coli: FhuA for ferrichrome, FecA for diferric dicitrate, Cir (colicin I receptor) for Fe³⁺ catecholates or Fe³⁺ dihydroxybenzoyl-serine and BtuB for vitaminB₁₂ (cobalamin) (20, 38, 39, 85). Distinct from other categories of OM transport, TonB-dependent transporters bind substrate with very high affinity, with dissociation constants reported in the nanomolar range. This transport system also requires proton motive force that is transduced by the IM-anchored periplasmic protein TonB, which is, in turn, stabilized by the IM proteins ExbB and ExbD. There is a PBP for this transport system (77), and transport across the inner membrane into the cytoplasm proceeds via an ABC-type transport complex (14). In the cytoplasm, Fe³⁺ is freed by various mechanisms and released (49). Crystal structures of TonB-dependent transporters showed that the receptors fold into a 22-stranded β-barrel (12, 33, 34), one of the largest barrels known to date, with a globular N-terminal domain (fimbrial ushers are larger, having 24 β-strands and a plug domain with a more complex fold) (66). The N-terminal domain consists of 150 to 200 amino acids and folds into the barrel from the periplasmic side to plug the pore. The N-terminal domain also has a conserved sequence of five residues called the TonB box that interacts with TonB prior to transport. The structures of several TonB-dependent transporters have been solved in the presence and absence of substrate (15, 17, 18, 33, 34, 35, 89). Binding of the correct siderophore or vitamin B₁₂ induces conformational changes in the transporter that may act to initiate binding to the TonB complex. An important conformational change is observed in the TonB box, which becomes more mobile and more exposed to the periplasm upon substrate binding (87).

Fig. 6Mechanism of TonB-dependent ferric siderophore transport. The transporter (green) is shown to interact with a TonB complex (yellow) at the TonB box motif (red). Transport of a ferric siderophore across the IM requires a PBP and an ABC transporter (blue). Once the ferric siderophore enters the cytoplasm, the ferric ion (Fe³⁺) is reduced to the ferrous ion (Fe²⁺).

The interaction between TonB and two transporters has been determined through crystal structures of the TonB C-terminal domain in complex with either FhuA or with BtuB (Fig. 7) (63, 75). These structures provide us with a picture of the transporter-TonB complex just before transport, but it still is not clear how the substrate is transported into the periplasm or what happens when energy is applied to the system. In all of the structures determined so far, the plug domain blocks the pore of the β-barrel, leaving no transport channel for the metal chelate. The plug domain must either undergo a conformational rearrangement to allow the passage of the substrate or must be pulled partially out of the barrel, but the mechanism remains unsolved. Recent spectroscopic experiments suggest that the plug undergoes a physical displacement during the transport process (51).

Fig. 7Crystal structure of BtuB-TonB complex. The C-terminal domain of TonB (green) is attached to the N-terminal end of BtuB (red) at the periplasmic side. The BtuB plug domain is in cyan. Bound cobalamine is shown in space filling representation in orange.

In addition to transporting metal chelates, many TonB-dependent transporters are misappropriated by toxic proteins called bacteriocins to gain entry into the cell. The bacteriocins are synthesized and secreted by certain bacteria to kill other bacteria in time of stress (59). The crystal structures of colicins bound to BtuB and Cir showed snapshots of large protein molecules bound to these transporters (11, 48, 74) (Fig. 8). While Cir apparently does not use a coreceptor to transport the colicin into the cell, BtuB-specific colicins have been shown to recruit OmpF for colicin transport, assembling a BtuB-OmpF-colicin translocon (50). Details of protein import across the OM are currently under investigation.

Fig. 8Crystal structures of two colicin receptor binding domains bound to their TonB-dependent transporters. Left, the complex between colicin E2 R-domain and its receptor, BtuB. Right, the complex between colicin Ia R-domain and its receptor, Cir. The receptors are green, and the colicin molecules are purple.

EXPORT CHANNELS

We have already described how the porins OmpD and OmpW can function as export channels. However, gram-negative bacteria have more efficient efflux systems for unwanted solutes. The TolC proteins are ubiquitous OM channels that form protein secretion complexes with many different IM proteins and periplasmic proteins (57). For example, TolC of E. coli forms a complex with AcrA (periplasmic) (90) and AcrB (inner membrane) (44), which belongs to the resistance-nodulation cell-division superfamily of transporters (Fig. 9). This complex exports a variety of compounds such as organic solvents, detergents, and antibiotics out of bacterial cells. The E. coli genome codes for three homologs of TolC and approximately 30 IM translocases belonging to the resistance-nodulation cell-division superfamily, the major facilitator family, and the ABC family (67). Crystal structures have been solved for TolC and AcrB, but the complete complex structure has not been visualized (45, 55). TolC was previously thought to form just an OM spanning structure like the β-barrels of porins. Intriguingly, however, the structure revealed that TolC consists of a 12-stranded β-barrel that is formed from three protomers that extend into a 100-Å-long α-helical barrel domain in the periplasm (Fig. 10). In the crystal structure, the channel is fully open at the extracellular side; it does not have inward-folded loops to constrict the channel. In contrast, the periplasmic end of the α-barrel was closed almost completely. It was proposed that the α-helical barrel could be opened by allosteric changes in the structure when other components of the efflux complex interact with TolC. TolC also works as a component of the type I secretory pathway (25). One example of a secreted substrate is E. coli toxic protein α-hemolysin (HlyA) (46). For secretion of HlyA, an inner membrane ABC transporter, HlyB, and the inner membrane-anchored periplasmic protein, HlyD, form a complex with TolC. The α-hemolysin, a 110-kDa protein, is thought to at least partially unfold to pass through the channel of TolC.

Fig. 9Schematic model of the TolC-AcrAB efflux complex. TolC (red) and AcrB (orange) form a complex to export substrates from the cytoplasm to the extracellular space. AcrA molecules (green) are predicted to bind the side walls of AcrB and TolC in the periplasm, and are thought to behave as molecular clamps. The export of substrate (pink) is driven by a proton/substrate antiport mechanism.

Fig. 10Crystal structure of TolC. The β-barrel channel (OM) and the α-helical tunnel (periplasmic) are formed by three protomers that are colored by monomer.

The extracellular surface of gram-negative bacteria is covered by capsular polysaccharides or exopolysaccharides. E. coli is known to produce and secrete more than 80 different capsular polysaccharides, called K antigens. The K antigen in E. coli O9a:K30 is assembled and exported by proteins encoded by a 12-gene operon (27), whereby the export process across the OM requires the Wza protein from the outer membrane auxiliary family (28). Wza is a 359-residue lipoprotein that forms a stable octamer. Electron microscopy (EM) studies showed octameric ring-like structures with dimensions 90 × 90 × 100 Å (7). The crystal structure of the 340-kDa Wza octamer was recently solved by the same group to 2.26-Å resolution (26). The structure is composed of a novel transmembrane α-helical barrel and an extensive periplasmic domain with a large central cavity (Fig. 11). Therefore, Wza represents the first example of a transmembrane α-helical barrel in the outer membrane. The eight helices (one from each protomer) forming the barrel are amphipathic: hydrophilic on the inside and hydrophobic on the outside. Wza is open on the extracellular side and closed on the periplasmic side in the crystal structure. A recent 3D EM structure of a complex between Wza and Wzc (the IM partner of Wza) showed that Wza changes conformation and opens at the periplasmic side of the channel upon binding to Wzc (19).

Fig. 11Crystal structure of Wza. (a) The octameric structure of Wza as viewed from the side. Each monomer is a different color. The α-barrel spans the OM, and three large domains extend to the periplasm. (b) View of the continuous pore of Wza octamer from the extracellular surface.

SUMMARY

We have provided an overview of OM transport systems based on recent research on outer membrane pores and receptors. Modes of transport are categorized by specificity, affinity for substrate, energy dependence, and direction of transport. Experiments using X-ray crystallography, nuclear magnetic resonance spectrometry, and EM have allowed the structure determination of most types of OM transporters and have helped to explain some of the mechanistic principles for transport across the OM. Porins are the most abundant OM proteins and their distribution differs according to bacterial strains. Because they form rather nonspecific channels, their expression is tightly controlled through multiple regulatory mechanisms. Further study of porins will show how OM permeability differs by each bacterial species and how it changes dynamically in response to environmental stresses. The existence of specific transporters, including TonB-dependent transporters, suggests the importance of selectively importing substrates into bacteria. Indeed, these transport systems are critically related to survival of the bacteria during infection of host cells. This means that these specific transport systems are potentially good targets for vaccine and drug development. The fatty acid transporter FadL is also attractive as the prototype of a protein family that transports various hydrophobic organic compounds. Further investigation of this family could be useful for studying environmental biodegradation.

TonB-dependent transport is the most dynamic transport system across the OM. Future elucidation of the transport mechanism, including structural rearrangements in the transporter, will reveal novel aspects of molecular transport. The studies of TonB-dependent transporters also serve as examples for structures of OM proteins in complexes with other protein components. Together with OM export channel studies, the accumulation of the knowledge should lead to establishment of general methodologies for visualization of more complicated, larger complex structures in bacterial membranes. Studies of export channels have shown how varied OM transporters can be. TolC is interesting from the point of view of evolution of β-barrel proteins because the β-barrel domain is built from three protomers. Wza is the first example of a transporter using an octameric α-helical barrel to span the OM. The biogenesis of these exporters is interesting because it could be quite different from that of other OM transporters in terms of insertion and folding in the OM. From a mechanistic point of view, each OM export channel seems to have unique gating properties. Since TolC is related to multidrug resistance in bacteria, further elucidation of the export mechanism and the regulation are urgently needed to design new drugs to specifically block this channel.

In conclusion, the study of OM transporters and receptors is still one of the frontiers not only for structural and functional studies of transport, but also for understanding biogenesis and pathogenesis of gram-negative bacteria. We hope that significant findings in both microbiology and clinical applications will be achieved by studies of this area over the next decade.