OUP user menu

A molecular Swiss army knife: OmpA structure, function and expression

Stephen G.J. Smith, Vivienne Mahon, Matthew A. Lambert, Robert P. Fagan
DOI: http://dx.doi.org/10.1111/j.1574-6968.2007.00778.x 1-11 First published online: 1 August 2007


The OmpA outer membrane protein of Escherichia coli and other enterobacteria is a multifaceted protein. This protein is expressed to very high levels and ompA is tightly regulated at the posttranscriptional level. It can function as an adhesin and invasin, participate in biofilm formation, act as both an immune target and evasin, and serves as a receptor for several bacteriophages. Many of these properties are due to four short protein loops that emanate from the protein to the outside of the cell. Herein it is described how the structure of this protein relates to its many functions.

  • OmpA
  • outer membrane protein
  • β-barrel
  • invasion
  • adhesion
  • pore-forming protein
  • immune evasion


It is now three decades since the OmpA protein was identified as a major component of the outer membrane of Escherichia coli (Chai & Foulds, 1977). Since then OmpA has been the subject of intense study establishing it as a multifaceted molecule with many diverse roles. Furthermore, fundamental studies of the assembly and structure of OmpA have had far-reaching implications for the understanding of outer membrane proteins in general (Kleinschmidt, 2003). In addition, the regulation of ompA gene expression is a paradigm of ribo-regulation. Biotechnological applications for this robust protein include surface display of antigens (Hobom et al., 1995; Lang, 2000). Here, the current understanding of the structure and functions of the OmpA from E. coli and other enterobacteria is reviewed. The reader is directed to the following reviews for further information on the assembly and insertion of OmpA (Kleinschmidt, 2003, 2006).

Structure of OmpA

The ompA gene (originally tolG) was identified in E. coli in 1974 when it was noted that bacteria with a mutation in this gene displayed a tolerance to bacteriocin JF246 (a type L colicin) (Foulds & Chai, 1978). The fact that ompA encoded a major outer membrane protein was not discovered until 1977 (Chai & Foulds, 1977). The presence of a two-domain structure was first proposed by Bremer (1982); their work showed that truncated OmpA (or Σ) lacking as much as 132 amino acids from the C-terminus were still capable of inserting into the outer membrane and conferred all of the known OmpA phenotypes (Bremer et al., 1982). On the basis of the colicin resistance pattern of various mutated OmpA proteins, Morona (1984) predicted that OmpA crossed the outer membrane several times. This was followed by the surprisingly accurate prediction by Vogel & Jähnig (1986), based on Raman spectroscopy, of the membrane topology of the N-terminal domain of OmpA (Fig. 1). The Vogel and Jähnig prediction called for an N-terminal β-barrel domain composed of eight membrane-spanning antiparallel, amphipathic β-strands connected by four long loops at the outer surface of the membrane and three short turns at the periplasmic face. The C-terminal portion of the protein was proposed to reside in the periplasm and perhaps interact with peptidoglycan (De Mot et al., 1994; Koebnik & Kramer, 1995). Additional evidence for this two-domain structure came from the use of circular dichroism on purified OmpA to determine the secondary structure of both the full-length OmpA and a truncated N-terminal domain (Sugawara et al., 1996). This study demonstrated that the protein consisted of a predominantly β-stranded N-terminal domain with a globular C-terminal domain containing a high proportion of α-helices.


Structure of OmpA. (a) A topological model OmpA showing the major features of the protein. Residues in β-sheets are in diamonds, those coloured in red make up the aromatic girdle while those in yellow are the alternating hydrophobic residues. Loops are numbered. Residues in loops 1 and 2 are important for binding endothelial cells and C4bp are highlighted in blue. Positions that may differ between enterobacteria are asterixed (Gophna et al., 2004; Power et al., 2006). The position of an insertion of three to four amino acids in loop 3 of invasive enterobacteria is indicated (Gophna et al., 2004). (b) Three-dimensional structure of OmpA. The view was generated using PyMOL (PDB-code 1BXW, Pautsch & Schulz, 1998). The aromatic girdle is shown in red and the first and last β-strands are highlighted in brown and blue to aid in the location of the termini, respectively. Loops are arrowed and numbered. Regions of loops 1 and 2 required for binding to endothelial cells and C4bp are highlighted in blue.

In 1998, in a landmark publication, the first crystal structure of OmpA was described (Pautsch & Schulz, 1998); within 2 years, an exceptional refinement of this structure was produced (Fig. 1) (Pautsch & Schulz, 2000). These crystal structures demonstrated that the N-terminal 171 residues of OmpA adopt an eight-stranded all-next-neighbour amphipathic β-barrel as predicted by Vogel and Jähnig. The barrel may be likened to an inverted micelle with a hydrophobic surface and a largely polar core stabilized by extensive hydrogen bonding, with salt bridges separating several large water-filled cavities. Despite the presence of these cavities, the crystal structures confirmed the lack of a continuous passage for water or solutes. This finding seemingly contradicted earlier reports of ion-permeable pores formed in lipid bilayers by OmpA (Sugawara & Nikaido, 1992). Sugawara and Nikaido demonstrated that OmpA-containing proteoliposomes were nonspecifically permeable to small molecules and estimated the pore size at c. 1 nm. They later showed that only a minor subset of OmpA molecules (2–3%) contained an open channel. High-resolution nuclear magnetic resonance (NMR) structural data have gone some way towards answering this apparent contradiction. Besides confirming the structure of the OmpA N-terminal β-barrel domain, dynamic NMR also revealed a degree of flexibility along the axis of the barrel, which could perhaps allow for a channel function (Arora et al., 2001). Molecular dynamics studies also predicted mobility within the barrel, similar to that observed with NMR (Bond et al., 2002). In particular, it was suggested that a gating mechanism involving the alteration of a critical central salt bridge from Glu52–Arg138 to Glu128–Arg138 could allow for transient pore formation with a predicted membrane conductance similar to those observed experimentally (Arora et al., 2000). However, Hong et al. demonstrated in an elegant series of experiments that the Glu52–Arg138 salt bridge identified in the crystal and NMR structures could switch to form alternate bridges with Lys82 and Glu52, respectively, and open a membrane-traversing ion channel (Fig. 2). A functional channel gating mechanism was also shown to contribute to the survival of bacteria growing under osmotic stress (Hong et al., 2006). These data provide a model capable of reconciling the earlier contradictory evidence for and against the existence of an OmpA pore. It is now possible to speculate that the crystal structure as solved by Pautsch and Schulz represents a snapshot of the OmpA pore in its more abundant closed form.


Pore formation by OmpA. (a) Viewing the OmpA molecule from above the backbone of the barrel is represented by the grey circle. Arg138 forms a strong electrostatic interaction with Glu52, the salt bridge formed across the inside of the barrel blocks ion transport. Lys82 may impinge on this interaction and stochastic movement of this residue towards Glu52 can result in the formation of another mutually exclusive ion pair. In this conformation, Glu52 does not block the movement of ions through the pore. (b) The eight-stranded β-barrel of OmpA is depicted. The β-sheets are indicated by rectangles and the external loops are also indicated. In the eight-stranded conformation, a large carboxy-terminal domain remains in the periplasm. Upon elevation of the temperature a 16-stranded β-barrel is formed, with eight of the strands from the domain formerly located in the periplasm.

However, the argument surrounding a pore function for OmpA has recently become more complicated and controversial. Before detailed high-resolution structural data were available for OmpA, it had been proposed that the protein could adopt an alternative conformation, consisting of a much larger 16-stranded β-barrel involving an additional eight β-stands from the C-terminal domain (Fig. 2) (Stathopoulos, 1996). This model has received some support in recent years from measurements of single-channel conductance due to OmpA in planar bilayers (Arora et al., 2000). These experiments indicated that incorporation of OmpA into a bilayer resulted in two distinct but interconvertible conductance states, one of 50–80 pS and a second of 260–320 pS, corresponding to a small and large channel, respectively. Interestingly, whereas a full-length OmpA was required to observe both channel sizes, a truncate containing just the N-terminal domain only gave rise to the smaller channels. One conclusion that could be drawn from this finding is that the small pore corresponds to the standard eight-stranded β-barrel, gated in the manner proposed by Hong et al., while the larger pore corresponds to a larger, perhaps 16-stranded, β-barrel. The formation of these two distinct pore sizes has also been shown to be temperature dependent, with the larger form predominant at physiological temperatures (Fig. 2) (Zakharian & Reusch, 2003). It has since been demonstrated that the temperature-dependent switch from small to large pore is irreversible (Zakharian & Reusch, 2005), prompting Zakharian and Reusch to propose that the large pore conformation may actually be the native fully folded form of the protein with the small pore variant an exceptionally stable folding intermediate. The membrane association and insertion of OmpA has been shown to be a multi-step process involving several partially folded intermediates, the last of which was only observed at close to physiological temperatures (Kleinschmidt & Tamm, 1996, 1999; Kleinschmidt et al., 1999). Zakharian and Reusch draw several parallels between these findings and their own model, and offer an alternative interpretation whereby the final temperature-dependent insertion step equates to the change from a small to large pore conformer. Despite the advances made in the study of membrane proteins, and OmpA in particular, it is clear that there are several pressing issues that are yet to be resolved.

Regulation of OmpA expression

OmpA is an abundant protein and a predominant antigen in enterobacterial outer membranes occurring at a copy number of ∼100 000 per cell (Koebnik et al., 2000). Given the expense of expressing OmpA to such a high level, it is hardly surprising that its gene expression is highly regulated and is environmentally responsive. Several stimuli and factors that control OmpA expression in E. coli are summarized in Table 1.

View this table:

Regulation of OmpA Expression in Escherichia coli

Stimulus or regulatorEffect on expressionReference
Acid challenge of nonacid adapted strainsDecreasedSainz et al. (2005)
Adhesion to abiotic surfacesDecreasedOtto et al. (2001)
Anaerobic culture in acidified mediaIncreasedYohannes et al. (2004)
Antimicrobial peptide – attacinDecreased due to destabilization of lipopolysaccharide with resultant feedback on OmpACarlsson et al. (1991, 1998)
Bacteriophage T2, T4 or T7 infectionHalf-life of ompA message substantially decreased, mediated through RNaseE and RNaseGUeno & Yonesaki (2004)
Chromate stressReducedAckerley et al. (2006)
Cyclic AMPCatabolite repressionGibert & Barbe (1990)
Defective lipopolysaccharideDecreased translationRied et al. (1990)
Growth in urineDecreasedSnyder et al. (2004)
Growth rateDecreases with growth rateLugtenberg et al. (1976)
Growth phaseDecreased in stationary phaseLugtenberg et al. (1976)
Nitrogen shortageIncreasedBaev et al. (2006)
Starvation in lakewaterDecreasedOzkanca & Flint (2002)
PolyaminesIncreasedYohannes et al. (2005)
MicA/Hfq/RNaseESee text
RNaseR/SigmaESee text
HhaRepressedBalsalobre et al. (1999)

The control of ompA mRNA turnover has been intensively researched and is a paradigm of ribo-regulation. Synthesis of OmpA is growth rate dependent (Lugtenberg et al., 1976) such that the ompA mRNA half-life increases proportionally with growth rate (Nilsson et al., 1984). During optimal growth conditions the half-life of the ompA message is ∼15 min but with a slower growth rate; for example at lower temperatures, this half-life drops to ∼4 min (Nilsson et al., 1984). The stability and longevity of ompA mRNA are due to a highly folded 5′-untranslated region (5′-UTR) of the transcript (Fig. 3) (Emory & Belasco, 1990; Emory et al., 1992; Hansen et al., 1994). The single-stranded regions of the 5′-UTR (ss1 and ss2) are targeted by RNaseE; however, ribosomal occupation of the ribosome-binding site (RBS) mitigates against cleavage at ss2. The protein Hfq is required for destabilizing the ompA transcript and has been shown to bind in the vicinity of ss2 causing a decrease in the transcript stability (Vytvytska et al., 1998, 2000). Recently, two papers have demonstrated base-pairing of a small RNA, MicA (SraD), to the ompA mRNA in the vicinity of the RBS and that Hfq is required to favour this interaction and results in translational inhibition with subsequent RNA decay (Fig. 3) (Rasmussen et al., 2005; Udekwu et al., 2005). Hfq is crucial in stabilising this interaction, inhibiting translation and inducing RNA decay (Rasmussen et al., 2005; Udekwu et al., 2005). The dependency of ompA mRNA stability on growth phase is underpinned by the fact that MicA levels inversely correlate with those of the ompA transcript. MicA expression is positively regulated by the extracytoplasmic σ factor SigmaE in response to envelope stress i.e. the overexpression of outer membrane proteins (Udekwu & Wagner, 2007). Envelope stress thus reduces OmpA levels via SigmaE and MicA. The ompA transcript is also targeted by RNaseR, a 3′–5′ exoribonuclease that is up-regulated in response to stress conditions and the entry into stationary phase (Zuo & Deutscher, 2001). In the absence of RNaseR, at stationary phase, higher levels of ompA transcript (and consequently OmpA protein) are produced (Andrade et al., 2006). Interestingly, RNaseR appears to act on full-length ompA mRNA; before this, RNase R was thought only to act on transcripts preprocessed by PNPase (Andrade et al., 2006).


Ribo-regulation of ompA. (a) The leader sequence of ompA mRNA can fold into two stem-loop structures hp1 and hp2. RNaseE cleaves the mRNA at ss1 and ss2 (yellow stars). (b) The micA mRNA forms a highly folded mRNA. The micA mRNA can base pair with ompA mRNA in the vicinity of the RBS This binding is facilitated by Hfq. (c) During the logarithmic phase or when growth rate is high, less MicA is produced and translation of ompA mRNA proceeds. (d) At the stationary phase or under conditions of low growth rate, micA copy number is elevated and it binds to the RBS of ompA mRNA, preventing ribosomal recruitment and facilitating ribonucleolytic degradation. Binding is facilitated by the protein Hfq.

Given that OmpA expression is controlled by two stress-responsive ribonucleolytic mechanisms, it is conceivable that many of the environmental stimuli regulating OmpA expression (summarized in Table 1) are transduced through these pathways in response to membrane stress.

Functions and properties of OmpA

Given its copy number and location, it is not surprising that the OmpA protein serves a multitude of functions. However, this is not always of benefit to the bacterium. It was recognized soon after its discovery that it is a receptor for several bacteriophages, particularly phages K3, Ox2 and M1 (Schwarz et al., 1983; Morona et al., 1984; Riede et al., 1984; Montag et al., 1987). The p38 tail fibre protein of these viruses recognize OmpA through its surface-exposed loops 1, 2 and 3 (Morona et al., 1984, 1985; Koebnik, 1999). Recently, Power (2006) have identified two alleles of ompA, ompA1 and ompA2. The nucleotide sequence differences of these alleles translate into several amino acid changes in the second and third loops (see Fig. 4). Interestingly, bacteria that possess the ompA2 allele are less sensitive to a panel of 24 bacteriophages (Power et al., 2006). A number of bacteriocins exploit OmpA as a receptor, namely colicin U of Shigella boydii (Smajs et al., 1997), bacteriocin 28b of Serratia marcescens (Enfedaque et al., 1996) and colicin L (Foulds & Chai, 1978). Furthermore, the F plasmid-encoded outer membrane protein TraN functionally interacts with OmpA, thus increasing the frequency of conjugal mating (Klimke & Frost, 1998; Klimke et al., 2005). An amino acid substitution in the surface-exposed loop 4, G154D, of OmpA results in decreased conjugation frequency. Notwithstanding this, a loop-deletion study by Koebnik (1999) indicated that all loops were required for efficient conjugal transfer suggesting that all loops are required for conjugation in common with the interaction with phage ligands.


Comparison of the ompA1 and ompA2 alleles. The translated sequences of the ompA1 and ompA2 alleles in the vicinity of loops 2 and 3 are aligned (Power et al., 2006). Perfect matches are shaded black.

OmpA: an adhesin/invasin

In 1991, Weiser & Gotschlich described a role for OmpA in the pathogenesis of E. coli K1 in chick egg and rat models. Escherichia coli K1 is an organism capable of causing sepsis and meningitis in neonates (Kim, 2000, 2002). Key to the ability of this microbe to cause meningitis is the capacity to cross the blood–brain barrier (BBB). The BBB is composed of brain microvascular endothelial cells (BMECs) that have tight junctions and low permeability; in contrast to epithelial cells these cells are invaded by relatively few microbial species. Meningitic strains of E. coli can invade both epithelial and endothelial cells (Meier et al., 1996). Prasadarao (1996b) demonstrated that OmpA was essential in invasion of BMECs in that ompA mutants were 25–50-fold less invasive than the parental strain. Furthermore, peptides corresponding to loops 1 and 2 (see Fig. 1) blocked invasion by bacteria-bearing OmpA. Intriguingly, the peptide corresponding to loop 2 maps to the positions of OmpA that differ between ompA1 and ompA2 alleles (Power et al., 2006). This may indicate that specific subsets of OmpA exist that may be more or less invasive, depending on the loop 2 amino acid sequence. Purified amino-terminal OmpA can bind directly to BMECs while a derivative lacking all four extracellular loops does not (Shin et al., 2005). The receptor for OmpA on BMECs has been identified and is a gp96 homologue termed Ecgp that displays extensive similarity to the heat shock protein Hsp90 (Prasadarao, 2002, 2003). The interaction between OmpA and its receptor occurs via GlcNAc1,4-GlcNAc epitopes on the Ecgp glycoprotein (Prasadarao et al., 1996a). Two parts of OmpA, termed regions 1 and 2, have been proposed to interact with chitobiose on the basis of docking experiments (Datta et al., 2003). Region 1 corresponds to loops 1 and 2 and in particular Asn 27 and 28 of loop 1 are crucial for initial sugar binding. Region 2 is more extensive and encompasses a water-filled pocket between residues from loops 1, 3 and 4. Mechanistically, the basis for internalization of bacteria expressing OmpA involves actin condensation (Prasadarao et al., 1999) and requires the activation of cellular signalling pathways involving the focal adhesion kinase (FAK) and protein kinase C (Sukumaran & Prasadarao, 2002).

Besides direct binding to BMECs, OmpA exerts its influence on bacterial binding by modulating type 1 fimbrial expression (Teng et al., 2006). Type 1 fimbriae are required for binding to BMECs (Teng et al., 2005) and their expression is controlled by the inversion of a promoter-containing element upstream of the fim operon (Smyth et al., 1996; Smith & Dorman, 1999). In ompA mutants this invertible element is predominantly in the OFF orientation, leading to decreased fimbrial expression and thus to less binding to BMECS mediated by these organelles (Teng et al., 2006).

The utility of OmpA in cellular interactions is not solely restricted to meningitic E. coli binding to BMECs. Enterohaemorrhagic E. coli (EHEC) of serotype O157:H7 utilize OmpA in adhering to HeLa epithelial cells and Caco-2 colonic epithelial cells (Torres & Kaper, 2003). This pathogen is often transmitted via contaminated plant foodstuffs, e.g. alfalfa sprouts. OmpA also appears to be absolutely critical in adherence to plant surfaces as an ompA mutant of E. coli O157 did not colonize alfalfa bean sprouts (Torres et al., 2005). Finally, together with the Hek protein (Fagan & Smith, 2007), we established a role for OmpA in invasion of colonic epithelial cells by meningitic E. coli (R.P. Fagan and S.G. Smith, in preparation) (Fagan, 2006). It is not known at this stage what parts of OmpA are required for binding to epithelial cells or if Ecgp also acts as a receptor for OmpA on these cells.

OmpA: an immune evasin

An essential attribute in the manifestation of meningitis by E. coli is the ability to survive in the bloodstream; higher counts of bacteria in the bloodstream are prognostic of more severe disease symptoms (Glode et al., 1977a, b). OmpA expression has been associated with serum resistance in a neonatal rat model (Weiser & Gotschlich, 1991). The basis for avoidance of host defenses is manifold and includes interference with complement activation, inhibition of cytokine induction and the ability to multiply within macrophages. OmpA of E. coli K1 binds to complement-binding protein 4 (C4bp) (Prasadarao et al., 2002). C4bp is an inhibitor of C3b activation via the classical pathway and is composed of eight distinct complement control proteins (CCPs). OmpA binds to C4bp at CCP3 using the sequences in loops 1 and 2 that are also responsible for BMEC binding (Prasadarao et al., 2002). C4bp bound to E. coli acts as cofactor to factor 1 promoting the cleavage of complement factors C3b and C4b, which ultimately arrests the complement cascade (Wooster et al., 2006). Logarithmic phase bacteria appear to be more effective at abrogating the activation of the complement cascade by this method (Wooster et al., 2006).

Escherichia coli K1 expressing OmpA can bind, enter and survive intracellularly within macrophages and monocytes. OmpA+ bacteria can multiply within the phagosome and attain very high numbers, eventually bursting the cell (Sukumaran et al., 2003). By contrast, ompA mutants do not survive within macrophages (Sukumaran et al., 2003). Bacteria expressing OmpA prevent macrophage apoptosis by inducing the antiapoptotic factor Bclxl (Sukumaran et al., 2004). In addition, E. coli K1 expressing OmpA, which had infected monocytes, suppressed the expression of chemokines and cytokines (Selvaraj & Prasadarao, 2005). At a molecular level, OmpA interferes with the activation of both nuclear factor-kappa B and mitogen-activated protein kinase (Selvaraj & Prasadarao, 2005). While OmpA obviously has a major bearing on the ability of meningitic E. coli to thwart the immune system (and mediate invasion of endothelial cells) other factors expressed by pathogenic bacteria are likely to be required concomitantly since laboratory strains of E. coli that express OmpA are not particularly immune evasive or invasive.

Escherichia coli can cause pneumonia in the immunocompromised. Both innate and adaptive immune mechanisms keep the lower lobe of the lung virtually sterile and among the innate defenses the surfactant family has been recently described. Two lung surfactants, surfactant proteins (SP), SP-A and SP-D (Wu et al., 2003), are members of the collectin family (Crouch et al., 2000). In addition to reducing the surface tension of the airway surface fluid of the lung, these proteins also serve as antimicrobial factors. In an animal experimental model the pulmonary clearance of E. coli was increased in mice overexpressing SP-D and reduced in SP-A knockout mice. However, bacteria that did not express OmpA were more effectively controlled by the surfactants implying that OmpA can be protective against these collectins (Wu et al., 2003).

OmpA: a target of the immune system

The expression of OmpA is however a double-edged sword, while it can function in avoidance of host defenses it can also be a target for the innate immune system. Bacteria that have been engulfed by neutrophils are killed by oxidative and nonoxidative methods. Nonoxidative methods employ proteins and peptides that permeabilize the bacteria. Neutrophil elastase (NE) is a major component of neutrophils (Belaaouaj et al., 2000). OmpA is a known target for NE and bacteria expressing this protein are severely damaged by elastase (Belaaouaj et al., 2000). Neutrophils derived from mice that were NE+/+ displayed faster killing kinetics for OmpA+ bacteria than their transgenic NE−/− counterparts (Belaaouaj et al., 2000). Bacteria that did not express OmpA were equally susceptible to both NE+/+ and NE−/− neutrophils. Invasive strains are more likely to encounter neutrophil elastase, thus it might be in the interests of these bacteria to express OmpA derivatives that could resist NE activity. Invasive isolates tend to express the ompA2 allele (Fig. 4) (Gophna et al., 2004). However, NE does not appear to have any preference for the gene products of either allele of ompA (Gophna et al., 2004). However, septicaemic strains secrete large amounts of OmpA and one strain of serotype O78 has been shown to resist NE (Gophna et al., 2004). Presumably the large amounts of exogenous OmpA saturate NE. Indeed, secretion of OmpA into serum has been shown to occur in vivo (Hellman et al., 2000).

During systemic inflammation two proteins are produced at high levels in serum, serum amyloid A (SAA) and the C-reactive protein. SAA binds to a wide variety of enterobacteria and this is mediated via OmpA (Hari-Dass et al., 2005). SAA bound to E. coli can act as an innate immune opsonin (Shah et al., 2006). SAA-opsonization of E. coli expressing OmpA increased uptake of bacteria by both neutrophils and peripheral macrophages (Shah et al., 2006). Furthermore, neutrophil reactive oxygen intermediate production was stimulated in response to SAA-opsonization of the bacteria.

Cells of the innate immune system recognize structures on microbial cells known as pathogen-associated molecular patterns (PAMPS) (Greene & Smith, 2007). PAMPS are recognized by pattern-recognition receptors (PRRs) such as Toll-like receptors (TLRs) (Greene & Smith, 2007). The recognition of such molecules by the innate cells leads to the initiation of an adaptive immune response. OmpA from Klebsiella pneumoniae binds to a wide range of immune effector cells (Jeannin et al., 2000, 2002; Soulas et al., 2000; Blacklaws, 2001; Godefroy et al., 2003). Furthermore, K. pneumoniae OmpA (KpOmpA) activates both macrophages and dendritic cells (Soulas et al., 2000; Blacklaws, 2001; Jeannin et al., 2002). KpOmpA signals through but does not bind to TLR2 (Jeannin et al., 2002). Instead KpOmpA binds to scavenger receptors LOX-1 and SREC-I (Jeannin et al., 2005). Lox-1 colocalizes with and contributes to TLR2-dependent signalling (Jeannin et al., 2005). In fact, KpOmpA may be used as a carrier for vaccines given its effectiveness in delivering antigens to antigen-presenting cells (Jeannin et al., 2002). Similarly, OmpA from EHEC is also effective in stimulating dendritic cells (Torres et al., 2006).

OmpA in biofilm formation

In addition to its role as a virulence factor, recent evidence suggests that OmpA has a role in E. coli biofilm formation. Biofilm formation by E. coli is clinically relevant when patients have in-dwelling urinary cathethers that may become colonized by bacteria. A proteomic analysis by two dimensional polyacrylamide gel electrophoresis of E. coli grown in biofilms within hydrogels has revealed that OmpA is up-regulated in both a uropathogenic isolate and a laboratory strain (Orme et al., 2006). OmpA has been shown to bind to abiotic surfaces (Lower et al., 2005). Furthermore, Barrios (2006) have demonstrated that ompA mutants formed biofilms of much reduced thickness in flow cell experiments. Thus, OmpA may be a potential target for inhibition of biofilm formation in vivo (Orme et al., 2006).

Conclusions and perspective

Much data regarding the structure, function and expression of OmpA have accumulated over the last three decades. The structure of OmpA has given many insights into the possible porous nature of this protein. Key to the further characterization of the pore is the definition of molecules that utilize this conduit into the cell. The various topological models and solved structures have widened the knowledge of the organization of the external loops that are critical to many of the cellular interactions in which this protein engages.

Although OmpA is highly conserved, amino acid differences occur in the loops between invasive and noninvasive E. coli strains, it is tempting to speculate that these changes may contribute to the invasivity of these strains.

The formation of the large pore proposed by Zakharian and Reusch provides an alternative model for the spatial organization of OmpA in the membrane. Does the putative 16-stranded β-barrel contribute to the functionality of this protein in host–pathogen interactions? Could this alternate structure be differentially recognized by the immune systems or by bacteriophages in comparison to the eight-stranded molecule?

It is hoped that the reader is convinced of the pleiotropic nature of the OmpA protein and agree that this protein is indeed the cell's Swiss army knife.


This work was supported by grants from The Health Research Board of Ireland (Grant number HRB2001/RP85), Enterprise Ireland (Grant SC2002-535) and the Irish Higher Education Authority (HEA) to support the ITTAC project TCD and a start-up fund from Trinity College Dublin.


  • Editor: Ian Henderson


View Abstract