The role of ATP-binding cassette transporters in bacterial pathogenicity

Abstract

The ATP-binding cassette transporter superfamily is present in all three domains of life. This ubiquitous class of integral membrane proteins have diverse biological functions, but their fundamental role involves the unidirectional translocation of compounds across cellular membranes in an ATP coupled process. The importance of this class of proteins in eukaryotic systems is well established as typified by their association with genetic diseases and roles in the multi-drug resistance of cancer. In stark contrast, the ABC transporters of prokaryotes have not been exhaustively investigated due to the sheer number of different roles and organisms in which they function. In this review, we examine the breadth of functions associated with microbial ABC transporters in the context of their contribution to bacterial pathogenicity and virulence.

Introduction

The ATP-binding cassette (ABC) transporter superfamily is an ancient family of proteins with phylogenetic evidence indicating that significant functional diversification had already emerged prior to the evolutionary divergence of prokarya, archaea and eukarya (Davidson et al. 2008; Fuellen et al. 2005). The extensive functional flexibility of this family is such that evolution adopted these transporters to such an extent that ABC transporters are now present throughout all domains of life. Canonical ABC transporters are highly efficient translocation systems that couple the hydrolysis of ATP to the unidirectional movement of compounds across the phospholipid bilayer of cellular membranes. Two broad functional categories can be assigned to these transporters based on the vector of transport, ABC exporters and importers. ABC exporters are present in all known species and facilitate the translocation of compounds out of the cytoplasm of a cell. However, ABC importers, which are found in prokaryotes, archaea and more recently in plants (Shitan et al. 2003; Terasaka et al. 2005), drive the acquisition and transport of compounds from the extracellular milieu into the cytoplasm. In addition to these two main classes of transporters, the ABC superfamily also includes non-canonical energy-coupling factor importers and the non-transport systems that lack a transmembrane component, both of which are outside the scope of this review. For overviews of these systems, see Eitinger et al. (2011) and Hopfner and Tainer (2003), respectively.

ABC transporters share a common architecture

Despite the differing vectors of transport between the canonical ABC transporters, they have many conserved features. All canonical ABC transporters share a conserved core of four domains comprising two transmembrane domains (TMDs) and two nucleotide-binding domains (NBDs). These domains can be encoded in a variety of ways, generating components as single polypeptides to multi-domain fusions such as half-transporters encoded by TMD–NBD fusions (Fig. 1). The TMDs traverse the lipid bilayer in a series of α-helices, with the overall structure providing a channel through which compounds are transported. Attached to the cytosolic side of the TMDs are two NBDs that act as “motor” domains that hydrolyze ATP molecules to drive the overall translocation cycle in an energy-dependent manner. The compounds translocated by ABC transporters are often referred to as allocrites, as ATP is the classical substrate molecule that is hydrolyzed to the product ADP (Holland et al. 1990), whilst the allocrite is unchanged during transport.

Fig. 1

Domain organization schematic of ABC transporters associated with bacterial pathogenicity. In Gram-negative bacteria, ABC transporters are localized to the cytoplasmic membrane (CM) and, in some circumstances, interact with other protein components embedded in the outer membrane (OM), the periplasm and CM. a The transmembrane domains (TMD—rectangles) and nucleotide binding domains (NBD—spheres/hexagons) for ABC exporters are shown for a homodimeric TMD–NBD fusion (i), a heterodimeric TMD–NBD fusion (ii), a homodimeric TMD–NBD Type I secretion system in complex with a membrane fusion protein (MFP) and outer membrane porin (OMP) (iii) and a heterooligomeric complex of two NBDs and two TMDs (iv). b A soluble substrate-binding protein (SBP—ovals), as would be found in the periplasmic compartment of Gram-negative bacterial pathogens, interacting with canonical ABC importers composed of a heterooligomeric complex of two NBDs and two TMDs (v), a heterooligomeric complex of two non-identical NBDs and two TMDs (vi), a heterooligomeric complex of two NBDs and two non-identical TMDs (vii), a heterooligomeric complex of two NBDs and a TMD fusion (viii) and a heteroligomeric complex of non-identical NBDs and TMDs (ix). Domains shaded in grey indicate encoding by a distinct polypeptide

The nucleotide binding domains

The NBDs are the conserved domain of the ABC transporter superfamily. They are often thought of as ‘motors’ attached to different TMDs to facilitate transport, although this is an oversimplification of the tight integration that couples these two domains. NBDs are typically between 200 and 300 amino acids in length and consist of a larger RecA-like catalytic domain and a smaller helical domain, the latter of which is unique to ABC transporters (for detailed reviews of NBD structure and function, see Davidson et al. (2008), Kerr (2002) and Moody and Thomas (2005)). The first structural information for the NBDs was obtained in 1998 (Hung et al. 1998). This work provided the foundation from which the dimerization of the NBDs in a ‘head-to-tail’ arrangement with motifs from each NBD contributing to the nucleotide binding sites was determined (Jones and George 1999). These early insights have been confirmed by numerous crystal structures of dimeric NBDs (Chen et al. 2003b; Hopfner et al. 2000) and NBDs associated with intact ABC transporters (Dawson and Locher 2006; Oldham et al. 2007; Pinkett et al. 2007).

Essential to the function of the NBDs are a number of conserved motifs that establish the interfacial contacts present in the NBD dimer, provide the nucleotide binding sites, facilitate ATP hydrolysis and couple thermodynamic work to the TMDs (for further details on motifs and their roles, see Davidson et al. (2008), Jones and George (2004), and Schneider and Hunke (1998)). The most important motifs for nucleotide binding are the P-loop and the ABC signature motif. The P-loop, also called the Walker A-motif, is located in the RecA-like subdomain and binds the β- and γ-phosphate groups of ADP or ATP. The ABC signature motif, which is also referred to as the LSGGQ motif (due to the consensus amino acid coding sequence), is located in the helical subdomain and contacts the nucleotide of the ATP-bound state. These motifs are arranged such that the P-loop from one monomer is positioned adjacent to the ABC signature sequence motif of the opposing monomer. Thus, the nucleotide-binding site is formed by the interaction of motifs between two NBD monomers. The presence of these motifs in each monomer of the NBD dimer thereby provides two ATP binding sites in the transporter. The catalytically functional conformation of the NBD dimer is induced by ATP binding. The consequence is a tight dimerization of two NBD monomers with ATP ‘sandwiched’ in between, hence the sandwich–dimer model presented in the literature (Smith et al. 2002). It is highly likely that ATP binds at both NBD binding sites as the hydrolysis of ATP has been observed to demonstrate strong positive cooperativity (Davidson et al. 1996; Greller et al. 1999; Liu et al. 1997; Zaitseva et al. 2005b).

In addition to the ATP-binding motifs, the Walker B and the H-loop motifs are present at the dimer–ATP-binding interface to assist in the hydrolysis of the bound nucleotide. The Walker B-motif is arranged as a β-strand in the RecA-like subdomain and coordinates a magnesium ion (Mg²⁺), via a water molecule essential for ATP hydrolysis. The H-loop, also referred to as the Switch motif, contains a conserved histidine residue that contacts the γ-phosphate of ATP and has a role in hydrolysis. Together these motifs facilitate the nucleophilic attack of a water molecule upon the γ-phosphate of the ATP molecule to release ADP and Pi (Hollenstein et al. 2007a; Zaitseva et al. 2005a). ATP hydrolysis only occurs in the tightly dimerized nucleotide-bound state of the NBDs. After ATP hydrolysis, the NBDs re-open to allow Pi to dissociate whilst the NBD dimer resides in a short-lived ADP-bound state until the nucleotide subsequently dissociates (Lu et al. 2005). Although nucleotide binding is a prerequisite for the formation of a catalytically functional NBD dimer, there are some notable exceptions in the nucleotide hydrolysis event. In some instances, one of the nucleotide-binding sites has a degenerate arrangement of motifs, rendering the site impaired or incapable of ATP hydrolysis, such as in the xenobiotic transporters LmrCD (Lubelski et al. 2006b) and TmrAB (Zutz et al. 2011). However, perturbation of the theoretically ‘optimal’ catalytic arrangement of the NBDs is not a widespread feature in prokaryotic ABC transporters. Although the functional basis for the distortion of a nucleotide-binding site, resulting in a non-equivalence of sites within an otherwise symmetrical transporter, is an issue of continuing research, the catalytic function of the NBDs is maintained by virtue of the remaining consensus ATP-binding site.

Additional motifs are also present in the NBDs to ensure efficient NBD–NBD interactions and interdomain coupling between the NBDs and the TMDs. The Q-loop, which is located in the loop region that connects the RecA-like and helical subdomains, has been shown to mediate interactions between the NBDs and TMDs (Dawson and Locher 2006; Hollenstein et al. 2007b; Pinkett et al. 2007). Located at the dimer interface as well as at the interface to the TMDs, this region is likely involved in interdomain signalling (Dawson et al. 2007). Another example is the D-loop, which has roles in NBD–NBD interactions and ensuring tight coupling of the ATP-binding sites in the NBDs to the allocrite binding sites within the TMDs. In addition to the motifs described above, some NBDs, notably those involved in import of compounds, also have accessory domains that may be involved in regulating the transport process (Biemans-Oldehinkel et al. 2006).

The transmembrane domains

The integral membrane components of the ABC transporters have relatively low sequence similarity and, accordingly, show greater diversity than the NBDs. ABC exporters have a conserved TMD core with each TMD containing six transmembrane (TM) helices, for a total of twelve helices. The crystal structures of prokaryotic ABC exporters indicate that the TMDs extend significantly beyond the boundary of the membrane into the cytoplasm. In the high-resolution structure of Sav1866 from Staphylococcus aureus, this extension was approximately 25 Å (Dawson and Locher 2006). Although lacking in a large cytoplasmic extension, ABC importers show considerable variety in the number and arrangement of their TM helices. This range spans from five TM helices per TMD, in the methionine importer MetI from Escherichia coli (Rees et al. 2009), to ten helices per TMD, in the vitamin B₁₂ transporter BtuCD from E. coli (Locher et al. 2002). The number of helices is postulated to reflect the size of the channel required for allocrite transport by each ABC transporter, with larger compounds requiring transporters with larger channels. Allocrite translocation through the TMDs remains poorly understood and, until recently, it was thought that allocrites moved through the TMDs without directly interacting with them. However, recent studies on the E. coli maltose importer, MalFGK₂A, indicated that this was not true for all ABC importers. In MalF, the TMD-encoding protein of the maltose transporter, an allocrite-binding site was clearly identified where one maltose molecule was bound by interactions with residues in the TMDs (Oldham et al. 2007). This suggests that, at least for some larger allocrites, there may be a necessary interaction between the TM helices and allocrites that occurs as part of the transport process.

The transport process involves the transmission of thermodynamic work from the NBDs to the TMDs. The coupling of this transmission involves a number of features, notably the Q-loop in the NBDs and the structurally conserved ‘coupling helices’ of the TMDs. The coupling helices show very little sequence conservation but are an architecturally conserved feature present in all ABC transporters. The coupling helices form the NBD–TMD interface (Locher 2009), where it is thought that nucleotide-induced conformational changes are transmitted via non-covalent interactions from the Q-loop region of the NBDs.

ABC exporters associated with bacterial virulence

ABC exporters are the ubiquitous subfamily of ABC transporters present throughout the three domains of life. They are, in contrast to eukaryotes and prokaryotes, not as highly represented amongst archaea (Davidson et al. 2008). Numerous studies have presented classification schemes for this broad superfamily (Dassa and Bouige 2001; Davidson et al. 2008; Eitinger et al. 2011). The most recent classification scheme placed the majority of ABC exporters in the Class 1 ABC transporter family. For an excellent overview of the bacterial ABC transporter superfamily see Davidson et al. 2008. In this review, we have focussed on virulence-associated ABC transporters (Fig. 2). ABC transporters with an apparent efflux function exist in both Class 1 and Class 3 according to this scheme. Based on the transporter classification (TC) database (http://www.tcdb.org) (Saier et al. 2006), there are 40 subfamilies of prokaryotic ABC exporters, annotated with the TC identification (TCID) as the 3.A.1. family, that have been identified to date. These can be broadly subdivided into transporters associated with glycoconjugate biosynthesis, Type I protein secretion or xenobiotic efflux. The prokaryotic xenobiotic transporters are highly represented in recent literature because of their homology to eukaryotic transporters, which are predominantly Class 1 ABC transporters. However, the other transporters also play important roles in bacterial physiology where they are intimately involved in biosynthetic pathways, such as in glycan export (Cuthbertson et al. 2009), and in protein secretion, such as that of surface (S)-layer proteins, toxins and competence factors (Otto and Gotz 2001; Zaitseva et al. 2005a). In this section, we will examine the contribution of ABC exporters to the virulence of bacterial pathogens in mammalian hosts.

Fig. 2

Roles associated with ABC transporters involved in bacterial pathogenicity in a model Gram-negative cell. ABC exporters include (a) Type I secretions systems associated with toxin, S-layer protein, siderophore, hydrolytic enzyme or antimicrobial peptide secretion, which have roles in adhesin and colonization of the host; (b) glycoconjugate and polysaccharide biogenesis pathways, which are involved in membrane biogenesis and immune evasion; and (c) xenobiotic efflux pathways involved in host environment tolerance and with postulated roles in antibiotic efflux. ABC importers (d) are associated with processes such as nutrient acquisition (e.g. metal ions, amino acids, vitamins and oligopeptides) and osmoprotection, which have crucial roles for successful colonization and promulgation in the host environment

Glycoconjugate biosynthesis

ABC transporters have a major role in prokaryotic glycoconjugate biosynthesis with four glycoconjugate-annotated transporter subfamilies with TC identifications (TCID) 3.A.1.101–4. Prokaryotes have up to three cell surface glycoconjugate biosynthetic pathways, which are referred to as the Wzx/Wzy-dependent, ABC transporter-dependent and synthase-dependent pathways, based on their inner membrane export mechanism. Importantly, these three general assembly export strategies are not confined to the assembly of any one class of glycoconjugate. The ABC transporter-dependent pathway mediates the translocation of glycans from all major classes of cell surface glycoconjugates including glycoproteins, teichoic acids, capsular polysaccharides (CPS), exopolysaccharide (EPS) and the O-antigen polysaccharide portion of lipopolysaccharide (LPS). For an excellent review on the role of ABC transporters in the broader aspects of glycoconjugate biosynthesis, see Cuthbertson et al. (2010).

All polysaccharides assembled by ABC transporters are fully polymerized by sequential glycosyl transfer that occurs at the cytoplasmic face of the inner membrane prior to translocation. The ABC transporter then exports the completed glycan across the inner membrane in an ATP-dependent manner. Further export to the outer membrane requires additional transport components such as a polysaccharide co-polymerase protein and an outer membrane polysaccharide export protein (for detailed reviews, see Cuthbertson et al. (2009) and Whitfield (2006)). The involvement of ABC transporters in glycan export was first recognized in studies of CPS biosynthesis in E. coli (Pavelka et al. 1991; Smith et al. 1990) and Haemophilus influenzae (Kroll et al. 1990), where loss of the ABC exporter resulted in the accumulation of the synthesized polymer in the cytoplasm. Since then, the dependence of other glycan biosynthetic pathways on ABC transporters has also been reported (Alaimo et al. 2006; Lazarevic and Karamata 1995; Zhang et al. 1993). Polysaccharides such as O-antigen, teichoic acids, CPS and EPS all have important roles in the virulence of pathogenic bacteria, where they fulfil a diverse range of roles including cell signalling, biofilm formation and evasion of the host immune responses.

Lipopolysaccharide biogenesis

The outer membrane of many Gram-negative bacteria is densely coated with LPS, a polysaccharide consisting of lipid A and a core oligosaccharide (the complex of which we will refer to as the lipid A core) and a repeated O-antigen polysaccharide tail. LPS biogenesis utilizes ABC transporters for translocation of both of these components. The O-antigen of LPS is one of the most variable Gram-negative cell constituents and a major contributor to the antigenic variability of the cell surface. The O-antigen consists of repeating oligosaccharide units (O-units) of two to eight different sugar residues. Antigenic variation arises from the different sugar moieties and the arrangements and linkages within and between the O-units. The O-antigen is presented on the cell surface and, due to its high immunogenicity, undergoes intense selection pressure from environmental factors and the host immune response. Consequently, this has led to the maintenance of diverse O-antigen isoforms within individual species, such as E. coli (Kauffmann 1947), and forms the basis of serotyping classifications for Gram-negative bacteria. The ABC transporter Wzt/Wzm, which has a respective TMD/NBD architecture, facilitates O-antigen biogenesis in a number of pathogenic and non-pathogenic bacterial species (Cuthbertson et al. 2010; Whitfield 2006). This pathway is not essential for cell viability, but the loss of the export pathway does result in cytotoxic accumulation of polymer unless the stress is relieved by other mutations (Clarke et al. 2004; Cuthbertson et al. 2005). The O-antigen is a major virulence factor in the host–pathogen interaction, such as in pathogenic E. coli and Klebsiella pneumoniae, where it protects the invading bacterium from killing by membrane-directed compounds, such as antimicrobial peptides, and aids in the evasion of complement-mediated killing and bacterial colonization (Achtman and Pluschke 1986; Bengoechea et al. 2004; Pluschke et al. 1983a, b; Whitfield et al. 1991). Mutagenesis studies in pathogenic E. coli have also shown that loss of O-antigen reduces survival and decreases the virulence of the bacterium (Achtman and Pluschke 1986; Pluschke et al. 1983a, b).

The other constituent of LPS, the lipid A core, is transported across the inner membrane by the ABC transporter MsbA (Raetz et al. 2007). Although classified in the 3.A.1.106.1 group, its physiological role is in LPS biosynthesis and so it will be addressed in this section. The lipid A anchors LPS to the outer membrane and is covalently attached to the O-antigen via its core oligosaccharide domain. Lipid A is highly immunogenic and is recognized by the innate immune system by cell surface receptors including Toll-like receptor 4, MD-2 and CD14 (Beutler and Poltorak 2000; Park et al. 2009; Poltorak et al. 1998). MsbA has also been proposed to be a general phospholipid flippase, but there remains a paucity of definitive biochemical evidence to support this interpretation. The possible involvement of MsbA in phospholipid export has arisen from the observation that loss of MsbA in Gram-negative bacteria, notably E. coli and Pseudomonas aeruginosa, was lethal (Doerrler et al. 2004, 2001; Ghanei et al. 2007). However, in Neisseria meningitidis, loss of MsbA did not affect viability but instead correlated with a very low LPS content (Steeghs et al. 1998). Thus, it has since been proposed that the effects observed on phospholipid transport, in msbA conditional mutant strains, may have been a secondary effect due to accumulation of toxic concentrations of lipid A precursors in the cytoplasm (Tefsen et al. 2005). An intriguing facet of MsbA function observed in in vitro studies is its capacity for xenobiotic binding and efflux (Reuter et al. 2003; Siarheyeva and Sharom 2009). However, the significance of this xenobiotic efflux capability in the context of bacterial physiology remains to be elucidated.

Teichoic acid

In Gram-positive pathogens peptidoglycan (PG) is the most conserved component of the cell envelope and is densely populated with anionic polymers of teichoic acids. These polymers, which can comprise nearly 50% of the cell wall mass, are translocated to the cell surface via an ABC transporter where they are covalently attached to PG (Lazarevic and Karamata 1995). The ABC transporters responsible for teichoic acid translocation, TagGH in Bacillus subtilis (Lazarevic and Karamata 1995) and TarGH in Staphylococcus aureus (Schirner et al. 2011), have been shown to be essential for bacterial viability. Teichoic acids are a virulence factor in several bacterial pathogens where they have been shown to function as non-proteinaceous adhesins. In Sta. aureus, teichoic acids play an important role in the colonization of both abiotic (Gotz 2002) and biotic surfaces, most notably in the adherence to human epithelial and endothelial cells during host colonization (Weidenmaier et al. 2005). In the opportunistic pathogen Sta. epidermidis, teichoic acid has been shown to be involved in adherence to fibronectin (Hussain et al. 2001). However, the teichoic acid secretion system of Sta. epidermidis has not yet been defined.

Capsular and exopolysaccharides

Secreted prokaryotic polysaccharides include the tightly attached CPS, which form a discrete surface layer also known as the capsule, and the secreted EPS, which only retains a limited association with the cell surface. CPS and EPS are large hydrophilic polymers with masses in the range of 10⁵ to 10⁶ Da whose challenging biosynthesis involves a complex inner membrane assembly process prior to translocation across the periplasm and the outer membrane. In E. coli, CPS is translocated across the inner membrane by ABC transporter KpsMT where KpsM is the TMD and KpsT is the NBD (Kroncke et al. 1990a, b; Pavelka et al. 1994, 1991; Pigeon and Silver 1994). Polymer translocation across the periplasm and the outer membrane requires KpsE, a polysaccharide copolymerase, and KpsD, an outer membrane export protein (Arrecubieta et al. 2001; Bronner et al. 1993; McNulty et al. 2006; Pazzani et al. 1993; Silver et al. 1987). These proteins are proposed to interact, generating a structure functionally analogous to a tripartite drug efflux pump (Higgins et al. 2004). CPS are also known to act as virulence determinants in pathogenic bacteria where they have roles in resisting complement-mediated killing, preventing adhesion of monocytes and neutrophils during host invasion, and in resisting cationic peptides. The role of secreted CPS in virulence has been observed in N. meningitidis (Frosch et al. 1991), H. influenzae (Kroll et al. 1990; Moxon and Vaughn 1981), Campylobacter jejuni (Keo et al. 2011) and Salmonella enterica serovar Typhimurium (Hashimoto et al. 1993).

Secreted EPS is also known to be an important virulence determinant in many pathogenic bacteria. In Brucella abortus, a facultative intracellular bacterial pathogen which causes abortion in domestic animals and undulant fever in humans, the ABC transporter ExsA was identified as a putative extracellular polysaccharide exporter (Rosinha et al. 2002). ExsA was first identified in Rhizobium meliloti where the ExsA gene was responsible for the secretion of the polysaccharide succinoglycan (Becker et al. 1995). Although the virulence mechanisms of Brucella spp. are not fully understood, decreased survival of the ΔexsA mutant Brucella in mice compared with the survival of the parental strain demonstrated that ExsA was required for full virulence (Rosinha et al. 2002).

Protein secretion

Prokaryotic protein secretion predominantly occurs via the Type II secretory pathway but a number of additional translocation systems, referred to as Types I and III through VI, also provide important protein secretory pathways (for reviews, see Fronzes et al. (2009) and Saier et al. (2008)). Notably, the Type I secretory pathway is an ABC transporter-dependent mechanism. To date, more than 25 families of ABC exporters (TCIDs: 3.A.1.109–113, 116, 118, 119, 123, 124, 126–134) have been associated with Type I protein secretion functionality. These families translocate a broad range of exoproteins including toxins, antimicrobial peptides, siderophores, hydrolytic enzymes and S-layer proteins. Many of the secreted proteins have roles in facilitating invasion or survival in the host, while the antimicrobial peptides have roles in competition against other bacteria but are not directly implicated in their virulence.

Type I secretion systems have been studied in detail in Gram-negative bacteria where they operate in a continuous process to translocate proteins across both the inner and the outer membrane (Zaitseva et al. 2005a). The Type I translocon, which facilitates multi-membrane translocation, is a complex of three proteins that includes an ABC transporter, a membrane-fusion protein (MFP) and an outer membrane protein (OMP). The ABC transporter recognizes the target protein and hydrolyzes ATP to drive protein translocation. The MFP is a periplasmic spanning protein, which has one transmembrane segment anchored in the inner membrane and connects the complex to the OMP during formation of the translocon (Schulein et al. 1992). The OMP is an outer membrane porin protein, such as TolC, that has an extended periplasmic domain and forms the tunnel across the outer membrane (Koronakis et al. 2000). The majority of Type I secreted proteins of Gram-negative bacteria feature a ~60 amino acid carboxyl terminal secretion sequence, which is not removed during translocation and is recognized by the ABC transporter. The secretion process is assumed to occur in a post-translational manner, with the allocrite protein in a largely unfolded conformation (Bakkes et al. 2010; Zaitseva et al. 2005a).

Toxins

The best-characterized Type I secretion systems belong to the repeats-in-toxin (RTX) exoprotein family. Their name is derived from the repeated glycine- and aspartate-rich sequences located in the carboxy-terminal region of the protein. These repeats form binding sites for calcium (Ca²⁺) ions and adopt a β-roll structure upon ion binding in the extracellular space (Baumann et al. 1993). Members of this protein family exhibit a broad range of sizes (from 40 to >600 kDa) and biological activities, and although the nomenclature of this family is synonymous with the pore-forming cytotoxins, a number of other secreted proteins, proteases and lipases also belong to the RTX family (for a detailed review, see Linhartova et al. (2010)). These other exoproteins can also be significant virulence factors in mammalian diseases.

RTX toxins are Type I-dependent secreted proteins that are present in a broad range of Gram-negative mammalian pathogens such as HlyA and EhxA from E. coli (Goebel and Hedgpeth 1982; Schmidt et al. 1995), CyaA from Bordetella pertussis (Glaser et al. 1988), PvxA from Proteus vulgaris (Welch et al. 1981), ApxIA, ApxIIA, ApxIIIA and ApxIVA from Actinobacillus pleuropneumoniae (Chang et al. 1989; Frey et al. 1991; Jansen et al. 1993; Schaller et al. 1999), VcRtxA or MARTX _VC from Vibrio cholerae (Lin et al. 1999), MbxA from Moraxella bovis (Angelos et al. 2003) and MmxA from Mannheimia varigena (Chang et al. 1993). The best-characterized Type I secretion system is the RTX α-hemolysin (Hly) secretion system of E. coli. It is commonly found in uropathogenic E. coli (Welch et al. 1981) where it is an important virulence factor due to its cytolytic activity against a broad range of mammalian cell types (e.g. erythrocytes, granulocytes, monocytes, endothelial cells and renal epithelial cells from mice, ruminants and primates). The hlyCABD operon encodes the genes required for the Type I secretion system and the synthesis of HlyA (Hess et al. 1986). The location of this operon, either on chromosome-embedded pathogenicity islands or on transmissible plasmids, suggests possible transmission by horizontal gene transfer between Gram-negative bacteria (Knapp et al. 1986; Ludwig and Goebel 1999).

The Hly translocon consists of the ABC transporter HlyB (Wang et al. 1991), the MFP HlyD (Johnson and Church 1999; Schulein et al. 1992) and the OMP TolC (Koronakis et al. 2000). HlyC is an acyl transferase involved in the maturation of HlyA. HlyA is a 107-kDa cytolysin and is thought to be largely unfolded in the cytoplasm due to the tightly restricted availability of Ca²⁺ ions. The interaction of the allocrite HlyA with HlyB and HlyD triggers recruitment of TolC and creates a transient channel–tunnel complex from the bacterial cytosol to the extracellular milieu (Wickner and Schekman 2005; Zaitseva et al. 2005a). Assembly of the Hly translocon occurs independent of nucleotide binding, but ATP hydrolysis by HlyB is required for HlyA export (Thanabalu et al. 1998). Once exported, HlyA folds in the extracellular space to form the biologically active toxin.

HlyA mediates its cytolytic activity by inserting asymmetrically into the outer leaflet of eukaryotic plasma membranes in which it generates hydrophilic transmembrane pores (Menestrina et al. 1995a, b). These pores are cation selective, have a diameter of 1–2 nm and open in response to a transmembrane potential (Menestrina et al. 1995a). HlyA contributes to virulence through cytolysis of cells and a combination of secondary effects. These secondary effects have been proposed to have major contributions to bacterial virulence. The pore formation allows for passive influx of extracellular Ca²⁺ that, in nucleated cells, stimulates arachidonate metabolism leading to the production of lipid inflammatory mediators. Such mediators have deleterious effects on neighbouring cells resulting in organ dysfunction, release of granular constituents from granulocytes and release of pro-coagulatory substances from platelets.

The pore-forming function of HlyA is a conserved aspect of the haemolytic and leukotoxic mechanism of the RTX toxins (Benz et al. 1994; Clinkenbeard and Thiessen 1991; Maier et al. 1996; Menestrina et al. 1994; Schmidt et al. 1996). However, significant variation amongst the toxins exists such as the bifunctional toxin CyaA from Bor. pertussis (Glaser et al. 1988), which consists of a fusion of an adenylate cyclase domain and a pore-forming RTX moiety and recently identified multifunctional autoprocessing RTX toxins from V. cholerae (Lin et al. 1999). An unusual toxin is cytolysin (CylB) of Enterococcus faecalis, a causative agent of surgical wound infections (Callegan et al. 2002). CylB, which is exported by the ABC transporter CylT (Gilmore et al. 1990), is distantly related to bacteriocins and has been shown in numerous models to contribute to the severity of Ent. faecalis infections (Garsin et al. 2001; Huycke et al. 1991; Ike et al. 1984, 1987; Schlievert et al. 1997). The cytolysin operon is found in the Ent. faecalis pathogenicity island and is associated with a number of other virulence determinants, including aggregation substance and enterococcal surface protein.

Antibiotic and antimicrobial peptide efflux transporters

Numerous ABC transporters, functioning effectively as Type I secretion systems, have been identified in antibiotic, lantibiotic and bacteriocin-producing prokaryotes. These types of proteins do not mediate virulence towards mammalian species, but they should be mentioned as they are critical for inter- and intra-species competition against bacteria colonizing eukaryotic host niches. Thus, they can alter the behaviour and composition of the endogenous bacterial flora of the host. Notably, several ABC transporter genes have been observed to cluster with antibiotic biosynthetic genes. Examples of these ABC transporters include DrrAB, which secretes daunorubicin and doxorubicin, from Streptomyces peucetius (Guilfoile and Hutchinson 1991), the cyclic peptide antibiotic microcin J25 exporter McjD from E. coli (Solbiati et al. 1999), the OleC4-OleC5 oleandomycin transporter of Stp. antibioticus (Olano et al. 1995; Rodriguez et al. 1993), CmrAB of Stp. griseus which transports a chromoycin precursor (Menendez et al. 2007), the pyoluteorin efflux pump from Pseudomonas (Huang et al. 2006), kasugamycin resistance of Stp. kasugaensis (Ikeno et al. 2000), and the spiramycin transporter of Stp. ambofaciens (Schoner et al. 1992). In these organisms, it is highly likely that the primary physiological role of the ABC transporter is active efflux of the antibiotic, thereby providing self-resistance (Mendez and Salas 2001).

A number of ABC transporters involved in bacteriocin and lantibiotic export have been characterized in Gram-positive and Gram-negative bacteria, such as the Type I secretion system, CvaAB-TolC, which is dedicated to the export of the antibacterial toxin colicin V bacteriocin (Zhong et al. 1996). Colicin V kills sensitive bacterial cells by disrupting their membrane potential (Yang and Konisky 1984). There are numerous other ABC transporters associated with bacteriocin export such as the aureocin A70 transporter AurT from Sta. aureus (Netz et al. 2001), the PepT lanitibiotic exporter (Meyer et al. 1995) and epidermin secretion system (Peschel and Gotz 1996) of Sta. epidermidis, the subtilin transporters SpaB and SpaEFG of Bac. subtilis (Klein and Entian 1994), the nisin transporters NisT and NisFEG (Siegers and Entian 1995) and bacteriocin secretion system (Rince et al. 1997) of Lactococcus lactis, the macedocin exporter McdEFG from Str. macedonicus (Papadelli et al. 2007), the salivaricin exporter SboEFG from Str. salivarius (Hyink et al. 2007), and the gallidermin and epidermin transporter GdmT of Str. gallinarum (Hille et al. 2001). Some of these transporters may also provide some resistance to xenobiotics such as drugs. One notable example of this is the SmbG bacteriocin immunity protein from Str. mutans which is capable of providing resistance to antibiotics such as tetracycline and penicillin and is upregulated in response to exposure to these compounds (Matsumoto-Nakano and Kuramitsu 2006).

The non-proteinaceous β-exotoxin I is an unusual toxin that merits inclusion. Although it is produced by the Gram-positive insect pathogen Bac. thuringiensis, this Type I secreted toxin is highly toxic to mammalian cells (Beebee et al. 1972; Mackedonski et al. 1972). Bac. thuringiensis is often used as a biological control agent against insects. However, strains that produce β-exotoxin I, which mediates toxicity towards dipteran, coleopteran and lepidopteran species, cannot be used due to their potential toxicity towards humans and the persistence of toxin in the environment (Benz 1966). β-Exotoxin I is an adenine nucleotide analogue that is secreted by the ABC transporter BerAB (Espinasse et al. 2002). β-Exotoxin I is thought to act by inhibiting RNA polymerase function, which disrupts insect larval development (Sebesta and Horska 1970).

Proteases

The RTX family also includes the Type I-dependent secreted zinc metalloproteases (Hooper 1994; Miyoshi and Shinoda 2000) that are secreted by a variety of pathogens. These RTX proteases are ~50 kDa metzincin metalloendopeptidases that consist of an N-terminal proteolytic domain and a C-terminal calcium-binding RTX domain. The proteolytic domains share a conserved methionine, located on a turn near the catalytic site, and an elongated carboxy-terminal HEXXHXXGXXH motif. The three histidine residues in this motif are known to bind the catalytic zinc ion and the catalytically important glutamic acid residue (Stocker and Bode 1995; Stocker et al. 1995). RTX proteases involved in virulence have been identified in Pseudomonas (Duong et al. 1992; Guzzo et al. 1990; Liao and McCallus 1998); (Chabeaud et al. 2001; Woods et al. 2001) and Proteus (Walker et al. 1999; Wassif et al. 1995) species.

The RTX protease AprA from P. aeruginosa has been implicated in P. aeruginosa septicaemia (Fetzer et al. 1967) due to its proteolysis of fibrin and fibrinogen (Shibuya et al. 1991) and its anticoagulant activity in human plasma, which was attributed to its direct fibrinogenolytic function. AprA also readily cleaves soluble laminin (Heck et al. 1986), which suggests a direct role (together with elastase) in both tissue invasion and hemorrhagic tissue necrosis in P. aeruginosa infections. AprA has also been implicated in modulating the efficacy of the host immune response by proteolytically degrading human γ-interferon (Horvat and Parmely 1988) and components of serum complement (Hong and Ghebrehiwet 1992), the latter of which has been suggested to impair complement-mediated opsonization of P. aeruginosa. The RTX protease ZapA, from Proteus mirabilis, was observed to be involved in the proteolytic degradation of the immunoglobulins (Ig) IgG (Loomes et al. 1993), IgA1 and IgA2 (Wassif et al. 1995). The degradation of these Igs could confer protection against opsonization during urinary tract infections.

Surface layer proteins

The S-layer proteins form two-dimensional crystalline arrays that encompass the bacterial cell surface, playing roles in proteolysis, peptidoglycan catabolism, adhesion and immune evasion (Bahl et al. 1997; Beveridge et al. 1997). These surface arrays are self-assembling and are often composed of a single protein or glycoprotein species. The majority of bacterial S-layer proteins are exported via the Type II secretion pathway, but numerous Type I-dependent S-layer proteins, including some RTX family proteins, have also been identified. In C. rectus, a Gram-negative pathogen associated with human periodontal disease, three Type I secreted S-layer proteins, Crs, CsxA and CsxB, are known (Braun et al. 1999; LaGier and Threadgill 2008; Miyamoto et al. 1998; Wang et al. 1998). These proteins are proposed to be virulence factors and have been implicated in the evasion of host defences, such as phagocytosis and the bactericidal activity of serum (Okuda et al. 1997; Thompson 2002). C. fetus, a bacterial pathogen of ungulates and humans, also resists serum killing due to its S-layer which contains the Type I secreted SapA protein (Thompson et al. 1998)

Hemophore and siderophore secretion

Nutrient acquisition is a central facet of the host–pathogen interaction. Iron, which has important roles in many aspects of prokaryotic cellular metabolism and respiration, is highly constrained in the host environment, in part to restrict its availability to invading pathogens. Consequently, invasive pathogens utilize a range of mechanisms to scavenge and liberate iron from the host environment. Haemoglobin is one potential source of heme iron for bacterial pathogens. HasA is one of many secreted heme-binding proteins that scavenge free or protein-bound heme with high affinity and return it to bacterial cell surface receptors for use by the pathogen (Letoffe et al. 1994a). Serratia marcescens is an opportunistic pathogen responsible for a number of opportunistic and nosocomial infections, including those of the urinary and respiratory tracts, septicaemia and meningitis (Hejazi and Falkiner 1997). The hemoprotein from Ser. marcescens is secreted by an archetypal Type I secretion system comprised of the ABC transporter (HasD), a MFP (HasE) and an OMP of the TolC family (HasF) (Letoffe et al. 1994b). HasA is an atypical Type I secretion system allocrite because, unlike the majority of other translocated proteins, it requires a molecular chaperone, SecB, to prevent its folding in the cytoplasm prior to export (Debarbieux and Wandersman 2001; Delepelaire and Wandersman 1998). However, despite compelling biochemical data for the function of HasA, its contribution to bacterial pathogenicity remains ambiguous as a hasA deletion strain of Y. pestis (Rossi et al. 2001) showed no reduction in virulence. This is most likely due to the presence of additional iron scavenging pathways employed by pathogens.

Extracellular iron is also acquired by siderophores, such as pyoverdine (Pvd), which are known virulence factors of P. aeruginosa. This Fe-scavenging involves both ABC importer-mediated uptake (described in “Iron”) and, in P. aeruginosa, recycling of Pvd by an ABC exporter (Imperi et al. 2009). Upon binding of extracellular iron, Fe³⁺–Pvd is transported by the membrane-bound receptor FpvA into the periplasm, where the Fe³⁺ is reduced to Fe²⁺ and bound by an alternative binding protein for use within the cell (Greenwald et al. 2007; Schalk et al. 2002). Upon Fe²⁺ release, the iron-free siderophore, apo-Pvd, is recycled and secreted into the extracellular environment to scavenge additional iron. Recycling of Pvd is thought to occur due to the complex and energetically expensive production of the siderophore by non-ribosomal protein synthetic pathways (Andrews et al. 2003; Imperi et al. 2009). As a consequence, recycling of Pvd reduces cellular energy costs and avoids depletion of cellular resources. Pvd recycling occurs via the tripartite ABC efflux complex of PvdT–PvdR and OmpQ, which acts to efflux Pvd from the periplasm to the extracellular milieu where it associates with the membrane-bound FpvA receptor (Imperi et al. 2009). The contribution of this pathway to P. aeruginosa virulence was shown in growth competition assays where strains compromised for Pvd recycling were outcompeted by those competent for recycling (Griffin et al. 2004). Interestingly, both Pvd-producing and non-producing strains are isolated from biofilms in the lungs of cystic fibrosis patients (De Vos et al. 2001), which may suggest that competition for Pvd occurs within biofilm communities (Griffin et al. 2004).

Xenobiotic efflux

The active extrusion of xenobiotics is an essential protective mechanism for organisms from all domains of life to defend against dangerous compounds, whether endogenous metabolic by-products or pervasive environmental toxins that would otherwise damage organelles, DNA, proteins or cellular processes. Prokaryotic drug resistance facilitated by xenobiotic exporters is a by-product of this otherwise essential defence mechanism. Since the introduction of antibiotics in the 1930s, clinical drug resistance has reached the point where it has become so prevalent that overcoming bacterial resistance to drugs is now likened to an unwinnable war. Prokaryotes have an extraordinary array of different drug efflux systems, with many pathogens containing multiple copies of diverse drug efflux transporters. ABC exporters, renowned for their role in drug resistance in eukaryotic cancers, have also been implicated in contributing to prokaryotic drug resistance (Higgins 1992). The logic underlying this reasoning is based on the consistent ability of many ABC exporters to efflux a plethora of structurally unrelated compounds. However, individual ABC transporters have not yet been observed to be a major factor of prokaryotic multi-drug resistance (MDR) in mammalian pathogens. In this section, we will examine the function of ABC multi-drug transporters that are, by virtue of their function, often implicated in bacterial virulence.

LmrA

Multi-drug efflux by a bacterial ABC exporter was first demonstrated by the Gram-positive Lac. lactis transporter, LmrA. LmrA was identified on the basis of its high sequence homology with the human chemotherapeutic drug resistance ABC exporter ABCB1 (van Veen et al. 1996). LmrA transports allocrites that are partitioned into the inner leaflet of the cell membrane, thereby preventing compounds from accessing the cytoplasm and exhibiting cytotoxicity (Bolhuis et al. 1996). LmrA has a broad substrate range consistent with a polyspecific xenobiotic transporter (Poelarends et al. 2000). Heterologous expression of LmrA in E. coli confers resistance to a broad range of clinically relevant antibiotics, including those from the aminoglycoside, β-lactam, macrolide, quinolone and tetracycline families (Poelarends et al. 2002). Recent studies of LmrA have demonstrated H⁺–Na⁺–Cl⁻ symport activity and this has led to the suggestion that symport function is most likely the physiological role of LmrA, allowing cells to survive in high osmolarity environments (Velamakanni et al. 2009).

BmrA

BmrA is an ABC efflux protein from the non-pathogenic Gram-positive bacterium Bac. subtilis that was identified on the basis of its significant homology (42% identity) to LmrA (Steinfels et al. 2004). BmrA is a homodimeric ABC efflux protein capable of transporting a range of allocrites including Hoechst 33342, mitoxantrone, 7-aminoatinomycin and doxorubicin (Chami et al. 2002; Dalmas et al. 2005; Ravaud et al. 2006; Steinfels et al. 2004). However, Bac. subtilis has not been observed to alter BmrA expression in response to ethidium bromide exposure during growth as has been observed for other multi-drug resistance transporters (Steinfels et al. 2004). This has led to the inference that BmrA may function as a constitutively expressed ABC efflux system responsible for the efflux of low levels of toxic compounds (Steinfels et al. 2004). Thus, the precise physiological role of BmrA in Bac. subtilis remains to be elucidated. Homologs of LmrA and BmrA have been identified through genomic screens of other organisms, such as HorA from Lactobacillus brevis (Sakamoto et al. 2001) and VcaM from the pathogen V. cholerae, but the majority of these ABC transporters lack detailed biochemical characterization and determination of their physiological functions.

LmrCD

LmrCD, also expressed by Lac. lactis, was the first characterized heterodimeric bacterial ABC multi-drug exporter (Lubelski et al. 2004). The role of LmrCD in multi-drug resistance has been implicated by the observation that expression of LmrCD was higher in four different MDR strains of Lac. lactis compared with the wild type (Lubelski et al. 2006a). LmrCD deletion mutants were also observed to be hypersensitive to a number of drug molecules including Hoechst 33342, rhodamine, cholate and daunomycin (Lubelski et al. 2006a). LmrC and LmrD are TMD–NBD fusion proteins but the NBDs are structurally distinct, with non-canonical residues present in the nucleotide-binding sites. Consequently, it has been suggested that the non-equivalence of binding sites in the NBDs establish distinct functional roles for these sites in the ATP hydrolysis process (Lubelski et al. 2006b). The physiological function of LmrCD has been proposed to be bile tolerance via the active efflux of cholate (Zaidi et al. 2008). Genomic screens have identified homologs of LmrCD in other bacterial species, such as EfrAB from Ent. faecalis (Lee et al. 2003), and although their actual contribution to in vivo drug resistance remains unclear, their role in bile tolerance may be important for pathogenic bacteria that colonize the intestine (Zaidi et al. 2008).

BilE

The bile exporter BilE is an atypical efflux pump of the human pathogen Listeria monocytogenes. Originally annotated as a glycine betaine/carnitine/choline ABC importer (Glaser et al. 2001), subsequent studies have shown that this ABC transporter is not involved in compatible solute-mediated environmental tolerance (Sleator et al. 2001) but has an important role in the active extrusion of toxic bile salts from the cytoplasm and is linked to virulence. Its similarity to the compatible solute family of transporters has been proposed to have arisen from the ability of BilE to sense environmental stimuli at the biochemical level in a manner similar to that of OpuA, through changes in membrane fluidity and stability. This could be a critical factor in Lis. monocytogenes ability to rapidly adapt to changes that perturb the bacterial membrane, the primary target of bile activity.

SmdAB

Ser. marcescens has a high level of intrinsic resistance to antibiotics similar to that of P. aeruginosa (Chen et al. 2003a). The genome of Ser. marcesens features a number of genes predicted to encode putative MDR pumps, including the ABC transporter SmdAB. Similar to LmrCD, SmdAB is thought to have functionally distinct NBDs (Matsuo et al. 2008). Studies of SmdAB indicate that it is capable of conferring resistance to a broad array of compounds including ciprofloxacin, norofloxacin, DAPI, tetracycline, Hoechst 33342 and TPPCI (Matsuo et al. 2008). However, despite its apparent in vitro capacity to mediate MDR, the role of SmdAB in virulence or in vivo drug resistance in this human pathogen has not yet been established.

Sav1866

Sav1866 from Sta. aureus is the only bacterial ABC transporter associated with multi-drug efflux for which high-resolution structural data is available. The structure of Sav1866 has been solved to 3.0 Å with ADP bound (Dawson and Locher 2006) and to 3.4 Å when complexed with the non-hydrolyzable ATP analogue AMP-PNP (Dawson and Locher 2007). By virtue of its canonical, homodimeric TMD–NBD arrangement, with each TMD comprised of six α-helices, Sav1866 is often viewed as the archetypal bacterial ABC exporter. However, despite the insights obtained from the high-resolution structure of Sav1866, knowledge of its actual role in MDR is more limited. To date, Sav1866 has been shown to be capable of exporting Hoechst 33342 and ethidium bromide (Velamakanni et al. 2008), but its role in in vivo antibiotic resistance remains to be elucidated. Consequently, although the structure of Sav1866 provides an ideal template for studies into bacterial multi-drug resistance ABC exporters, it may not be able to provide insights into the subtleties and differences found in all xenobiotic efflux proteins.

ABC importers associated with bacterial virulence

ABC importers facilitate the acquisition of essential compounds from the extracellular environment. This class of ABC transporters is found in prokaryotes, archaea and plants but is absent from other members of eukarya. ABC importers belong to the Class 3 subgroup of ABC transporters (Davidson et al. 2008) and have been further sub-divided by structural features into two classes, the Type I and Type II importers (Locher 2009). Type I ABC importers are associated with small molecule transport, such as ions and amino acids, and have five to eight TM helices in each TMD subunit. Type II importers are associated with large molecule transport, such as metal chelates and vitamins, and have ten TM helices in each TMD subunit (Locher et al. 2002). Although ABC importers show significant diversity in the number of TM helices, all ABC importers, with only a few notable exceptions, are synthesized as four distinct polypeptides, i.e. two NBD and two TMD polypeptides. In addition to the ABC transporter, a substrate binding protein (SBP) is required to form the ABC importer complex, also referred to as the ABC permease. However, a distinct class of non-canonical ABC importers that acquires compounds via a SBP-independent mechanism was recently identified (for a review, see Eitinger et al. (2011)). SBPs are high-affinity allocrite binding proteins with affinities that span from the nanomolar to the micromolar concentration range. Their role is to recruit and present the allocrite to the ABC importer. In general, SBPs are clustered within the same operon as the ABC transporter. However, microbial genome sequencing has identified numerous SBPs without an apparent cognate ABC transporter. These ‘orphan’ SBPs have recently been shown to interact with ABC transporters within pre-existing ABC permease complexes. This indicates that the transport capability of these complexes may be considerably larger than originally thought and is not solely governed by the SBP clustered within the operon (Chen et al. 2010). The nomenclature in this area warrants mention as it is an issue for potential confusion. The majority of genes have been named in sequential order such that the ‘A’ gene in some operons is the SBP, while in other transporters it refers to the NBD gene.

SBPs are responsible for the recruitment of a broad array of different compounds from the extracellular environment. Consequently, they have different structural and functional features to facilitate their roles (for a thorough review, see Berntsson et al. (2010)). In Gram-negative organisms, SBPs are located in the periplasmic compartment, while in Gram-positive organisms they are typically located on the external face of the cytoplasmic membrane tethered by a lipoprotein anchor or directly fused to the transporter (van der Heide and Poolman 2002). A similar organization exists in archaea, although some SBPs also have a type III signal sequence and have been proposed to form an extracellular oligomeric complex referred to as a ‘bindosome’ (Zolghadr et al. 2007). ABC permeases are often the primary uptake pathway for nutrients with limited extracellular concentrations. A central innate defensive mechanism employed in the host environment is the restriction of nutrient bioavailability to prevent colonization by pathogenic organisms. As a consequence, a number of ABC importers also have important roles in colonization and propagation of infection. In this section, we will review the ABC importers associated with bacterial virulence.

Metal ion acquisition

Nearly 30% of all proteins require interaction with a metal in order to facilitate biological activity (Andreini et al. 2008). The vast majority of metalloproteins use a first row transition metal, e.g. Mn, Fe, Cu or Zn. As a consequence, the host environment has evolved a number of strategies to restrict the bioavailability of these metal ions to concentration ranges of micromolar and below to limit the capacity for bacterial invasion and colonization. ABC importers associated with acquisition of the transition row metals Fe, Mn and Zn are highly represented in studies of bacterial pathogenicity.

Iron

Iron was integrated into biological processes early in evolution due to its environmental abundance and range of different oxidation states. In a biological context, iron has key roles in a range of pathways by virtue of its flexible redox chemistry including respiration, photosynthesis and nitrogen fixation. Iron acquisition is also of major importance for bacterial pathogens and, consequently, the animal or human host employs a range of strategies to limit its bioavailability (in the order of ~10⁻¹⁸ M) (Braun 2001), such as sequestering it in lactoferrin or transferrin to restrict its availability to non-host processes. There are three classes of iron-specific uptake systems: the ferric iron transporters (TCID 3.A.1.10), the iron chelate transporters (3.A.1.14) and the ferric siderophore uptake transporters (3.A.1.21). In addition, some members of the Mn/Zn/Fe chelate transporter family (3.A.1.15) have been implicated in iron acquisition.

The ferric transporters facilitate the import of ions in a Fe³⁺ chelate/siderophore-independent manner into the cytosol. The Fe³⁺-ABC permeases, FbpABC in N. gonorrhoeae and meningitidis (Adhikari et al. 1996; Chen et al. 1993) and HitABC in H. influenzae (Adhikari et al. 1995; Sanders et al. 1994), employ the Fe³⁺ binding protein, FbpA or HitA, respectively, to recruit Fe³⁺ ions to the ABC transporter FbpBC or HitBC. In Neisseria spp., the affinity of FbpA is such that it is capable of stripping the Fe³⁺ directly from transferrin for use by the pathogen (Anderson et al. 2004; Shouldice et al. 2004; Taboy et al. 2001). As Fe³⁺ ions would not occur, under normal physiological conditions, in a ‘free’ or unbound state, it is likely that in pathogenic bacteria, ABC permeases of this class would be involved in directly acquiring Fe³⁺ from host proteins and carrier molecules.

The iron chelate transporters and ferric–siderophore transporters are responsible for the acquisition of iron from the host via a bacterial chelating compound. A variety of different chelating siderophores are known and the ABC permease import-dependent siderophores include ferric-hydroxamates, -bactins and -dicitrates. In Gram-positive bacteria, the ABC permease import process involves an extracellular SBP that binds the siderophore that then delivers it via its ABC transporter into the cytosol. In Gram-negative bacteria, as the siderophores are too large for passage through the OM porins, they interact with an outer membrane transporter for delivery into the periplasmic compartment via a TonB system (Faraldo-Gomez and Sansom 2003; Paulsen et al. 1997; Postle 1990; Wang and Newton 1971), whereupon they are then able to interact with a siderophore-specific SBP and its cognate ABC transporter (Sutcliffe and Russell 1995). In both Gram-positive and -negative bacteria, the iron-bound siderophore interacts with cytosolic proteins, causing the release of iron. The archetypal Gram-negative siderophore ABC permease is the FepBCDG system of E. coli which transports enterobactin. In this pathway, the Fe³⁺-loaded siderophore is bound on the outer membrane by FepA (Braun et al. 1998; Buchanan et al. 1999). The ferric–siderophore complex is then translocated into the periplasm in a process involving TonB, ExbB and ExbD (Moeck and Coulton 1998), where it is bound by the SBP FepB (Braun et al. 1998). FepB delivers the ferric–siderophore complexes to the ABC transporter FepCDG (Chenault and Earhart 1991; Shea and McIntosh 1991). Once localized to the cytoplasm, Fe³⁺ is removed from the siderophore by a cytoplasmic enzyme, Fes.

The iron chelate transporters (TCID 3.A.1.21), as typified by the archetypal E. coli FepBCDG or FecBCDE iron chelate transporters, have been observed in the genomic screens of almost all prokaryotes. Notably, in studies of pathogenic bacteria, these transporters often contribute to the virulence capacity of these organisms. Notable examples include the PiaABCD ABC permease of Str. pneumoniae (Brown et al. 2001), the ABC permease FtsABC of Str. pyogenes (Hanks et al. 2005), the HtsABC, FhuCBG-D1-D2, SirABC and SstABCD ABC permeases of Sta. aureus (Beasley et al. 2009; Dale et al. 2004; Morrissey et al. 2000; Speziali et al. 2006) and the ViuPDCG ABC permease of V. cholera (Butterton et al. 1992; Henderson and Payne 1993, 1994; Occhino et al. 1998; Wyckoff et al. 1999).

A subset of these transporters is involved in the scavenging of iron from haemoglobin released by the Type I secreted toxin HlyA from erythrocytes. The HlyA-induced lysis of erythrocytes allows secreted hemophores, such as HasA, to bind heme groups released from haemoglobin degradation (Cescau et al. 2007). Hemophore transporters then either strip the iron from the hemophores directly or transport them into the cells, where the porphyrin ring is opened by oxygenase to release iron (Tong and Guo 2009). These ABC-dependent heme transporters have been identified in a number of virulence studies of pathogenic organisms such as Str. pyogenes (HtsABCD or SiaABC) (Bates et al. 2003; Lei et al. 2003; Nygaard et al. 2006; Sun et al. 2010), V. cholerae (HutABCD) (Henderson and Payne 1993, 1994; Occhino et al. 1998), P. aeruginosa (PhuSTUVW) (Ochsner et al. 2000; Tong and Guo 2009), Shigella dysenteriae (ShuSTUV) (Burkhard and Wilks 2008) and Sta. aureus (HtsABC) (Grigg et al. 2007; Tong and Guo 2009).

Y. pestis has three known siderophore transport systems, two of which have been shown to be associated with virulence. The iron chelate transport pathway YfeABCD, which transports the siderophore yersiniabactin, employs the SBP YfeA to acquire yersiniabactin in the periplasm, which is then transported into the cytoplasm via YfeBCD (Perry et al. 2007). Strains lacking a functional YfeBCD ABC permease were avirulent in mouse models (Bearden and Perry 1999). The second yersiniabactin transporter identified is the YbtPQ ABC permease. This transporter belongs to the ferric–siderophore uptake group (TCID 3.A.1.21) and is structurally unique among ABC importers as it is comprised of two fusion polypeptides featuring an amino-terminal TMD and a carboxy-terminal NBD. In Y. pestis, loss of this ABC permease also resulted in an attenuation of virulence in a mouse infection model (Fetherston et al. 1999). Intriguingly, although both transporters translocate yersiniabactin into the cytoplasm, loss of either ABC permease was sufficient to attenuate in vivo virulence in their respective mouse models. This indicates that, during infection, iron acquisition by Y. pestis is essential for invasion and/or colonization of the host and that both pathways are necessary for this process. As iron accumulation was reduced, but not abrogated in both studies, this could suggest that yersiniabactin uptake is a rate-limited step. A homolog of the YbtPQ system has also been identified in M. tuberculosis, the IrtAB ABC permease. Loss of this pathway resulted in decreased survival in a mouse lung in vivo infection model (Rodriguez and Smith 2006; Ryndak et al. 2010).

In addition to the major classes of iron transporters, members of the Mn/Zn/Fe chelate transporters (3.A.1.15) are also involved in iron acquisition. However, phenotypic studies of this class of transporters indicate that its members are predominantly involved in manganese and/or zinc acquisition and, consequently, the actual physiological contribution to iron acquisition is less certain. Despite this, there is evidence that these transporters are regulated in response to extracellular iron concentrations, for example, in Salmonella Typhimurium, where the SitABCD ABC permease transporter was up-regulated under iron-restricted conditions (Cockayne et al. 1998) and deletion of the ABC permease genes attenuated virulence in a mouse model (Janakiraman and Slauch 2000; Wandersman and Delepelaire 2004). However, these studies did not unequivocally demonstrate that SitABCD was required for iron acquisition in vivo. Similarly, in Sta. epidermidis, the SitABCD permease which is capable of both iron and manganese uptake is expressed during the implantation phase of infection (Massonet et al. 2006), but the metal transported in vivo was not established.

The studies of iron transport systems in pathogenic bacteria, in general, indicate that although multiple iron transport pathways exist, loss of a single one is often sufficient to reduce or attenuate virulence in vivo. This has great significance for understanding how pathogens respond to the limitation of essential nutrients during invasion and colonization of the host.

Manganese

Although Mn is essential for almost all forms of life, its biological importance was overlooked for many years. Mn has known roles in a wide range of enzymes involved in phosphorylation, hydrolysis, carbon metabolism, decarboxylation and oxidative stress response (Papp-Wallace and Maguire 2006). More recently, its importance in bacterial pathogenesis has become increasingly apparent, where it is implicated in having a number of roles complementary to those of iron. Mn ABC permeases belong to the broad family of Mn/Zn/Fe chelate transporters (3.A.1.15). ABC permeases involved in Mn transport were first identified in Gram-positive bacteria as putative adhesion proteins belonging to the lipoprotein receptor-associated antigen I family (Jenkinson 1994). Subsequent studies found that they also had the capacity to function as specific Mn transporters (Kitten et al. 2000; Kolenbrander et al. 1998). Since then, ABC permeases of the Mn/Zn/Fe chelate transporter family have been identified in almost all pathogenic bacteria, but in many cases the physiological function of these transporters remains nebulous and inferred largely based on sequence comparisons of the SBPs. An additional complication in establishing the significance of this group of permeases is that many early studies often misidentified these transporters due to their ability to bind or transport other metal ions, albeit at lower affinities than their cognate ligand (Janakiraman and Slauch 2000).

The best-characterized Mn transporter is the PsaBCA permease from the major human pathogen Str. pneumoniae (Berry and Paton 1996; Dintilhac et al. 1997; Lawrence et al. 1998; McAllister et al. 2004; Tseng et al. 2002). Its role in Mn transport has been definitively established and loss of any component of this Mn uptake system has been shown to compromise growth on Mn-restricted media and results in complete attenuation of virulence in vivo (McAllister et al. 2004; McDevitt et al. 2011). Similar phenotypes have been observed in Str. pyogenes, where deletion of the high-affinity Mn ABC permease MtsABC resulted in a 30-fold reduction in virulence in a mouse infection model, and in Str. mutans, where deletion of the Mn ABC permease SloABC reduced endocarditis virulence in a rat infection model (Paik et al. 2003). Similar findings were observed in Str. suis where loss of the Mn SBP TroA in a murine Str. suis infection model resulted in an avirulent phenotype (Schreur et al. 2011) and in N. gonorrhoeae (MntABC) where permease was required for biofilm formation and intracellular survival (Lim et al. 2008).

In addition to the Mn ABC permease, some prokaryotes also utilize natural resistance-associated macrophage protein (NRAMP) systems as secondary Mn transporters. However, these proton-coupled transporters have not, in general, been found to serve as primary Mn uptake pathways. As a consequence, the requirement of the Mn ABC permease for virulence in organisms that also possess a NRAMP system is not unequivocal. Nevertheless, the overall trend in the literature suggests that in the majority of cases the ABC permease is the primary virulence determinant. In Sta. aureus, which possesses both a high-affinity Mn ABC permease (MntABC) and a NRAMP system (MntH), phenotypic studies found that the ABC permease was the high-affinity Mn transporter (Horsburgh et al. 2002). However, in a mouse abscess model, both systems had to be disrupted before a reduction in virulence was observed (Ando et al. 2003; Horsburgh et al. 2002). In contrast, in the Gram-negative pathogen Salmonella Typhimurium, mutation of the Mn permease SitABCD, alone or in conjunction with the NRAMP system, reduced virulence in a Nramp1 ^G169 murine typhoid model of infection (Kehres et al. 2002; Zaharik et al. 2004). Similarly, in avian pathogenic E. coli O78 strain chi7122, the SitABCD transporter was found to be more important for virulence in a chicken infection model than the NRAMP system (Sabri et al. 2008).

The recurring theme from these studies is that the Mn ABC permease has an important role in protection from oxidative stress and that in vivo loss of this function is associated with a compromised ability to mediate virulence. This strongly indicates that, during host colonization, many pathogens have a critical dependence on Mn acquisition to manage oxidative stress independent of iron-based protective mechanisms. Furthermore, this is significant in both aerobes and facultative aerobes and suggests that it may not simply be due to a compromised ability to effectively manage oxidative stress arising from growth under aerobic conditions but may also have a role in protection from exogenous redox stress mediated by innate immune defences.

Zinc

Zn is also an essential metal ion and is highly abundant in biology due to its single oxidation state in solution (Andreini et al. 2006b). In humans, it is the second most abundant transition row element and recent studies have suggested that zinc-binding proteins could constitute about 10% of the entire human proteome (Andreini et al. 2006a). Zn has critical structural and/or catalytic roles in all the major classes of enzymes and is also important for transcription and replication factors (Andreini et al. 2008; Coleman 1998). As a consequence, Zn acquisition in vivo has been found to be a virulence determinant in a number of pathogenic bacteria.

The Zn ABC permease, ZnuABC, has been observed to be involved in the virulence of E. coli O157:H7 (Gabbianelli et al. 2011), E. coli CFT073 (Gunasekera et al. 2009; Sabri et al. 2009) and Salmonella Typhimurium (Ammendola et al. 2007; Campoy et al. 2002). In addition to the Zn ABC permease, an additional Zn recruitment factor ZinT, or YodA, has been found in some Gram-negative pathogens, such as Salmonella Typhimurium, where it is thought to function as a Zn chaperone aiding in the recruitment of Zn to ZnuA in Zn-restricted environments (Petrarca et al. 2010). The Zn ABC permease in Str. pneumoniae, although not independently shown to mediate virulence (Claverys 2001), has an unusual organization. The permease is encoded by the adcABC operon that includes the high-affinity Zn SBP AdcA. However, a second Zn SBP AdcAII was recently identified clustered with a group of cell surface-localized histidine triad proteins expressed in response to Zn restriction (Loisel et al. 2008). The physiological significance of this apparent redundancy in Zn recruitment is currently unclear. Intriguingly, Str. pyogenes encodes an orphan Zn-binding protein, Lsp, that is homologous to AdcAII both in sequence and genomic context, i.e. clustered with the histidine triad proteins. Mutation of the Lsp metal-binding site leads to a decrease in virulence in mouse soft tissue, implicating an important role for Lsp in Zn homeostasis. Intriguingly, Str. pyogenes lacks a homolog of the primary Zn SBP AdcA, which is clustered with the cognate ABC transporter AdcBC in Str. pneumoniae. As Str. pyogenes retains a homolog of the ABC transporter, it is likely that Lsp mediates Zn transport through AdcBC, but this has not yet been demonstrated (Weston et al. 2009). The specific requirement of Zn to mediate virulence, if any, remains unclear. However, in addition to its general requirement for enzyme structure and function, it has been suggested that it is required for activation of copper/zinc superoxide dismutase in invasive pathogens to help defend against host-induced oxidative and nitrosative stress (Claverys 2001).

Amino acid transporters

Prokaryotes acquire amino acids from the host environment to use as carbon and/or nitrogen sources or because they are required due to auxotrophic growth conditions. Although amino acid transporters have been observed to be up-regulated in numerous bacterial pathogens in response to colonization of the host, there are only a few specific examples of amino acid ABC permeases being required for survival in the host environment. The glutamine (Gln) ABC permease GlnPQ has been observed to be required for Str. pneumoniae (Hartel et al. 2011) and Group B Streptococcus (GBS). The Str. pneumoniae genome encodes six Gln ABC permeases, two of which lack the putative Gln SBP GlnH, which is suggestive of significant functional redundancy. However, in vivo studies of two of these transporters found that deletion of a single Gln ABC permease significantly affected virulence, resulting in moderate to near-complete attenuation (Hartel et al. 2011). In GBS, loss of the GlnPQ ABC permease resulted in decreased in vitro and in vivo virulence due to impaired fibronectin adherence, although the underlying relationship between cellular Gln levels and fibronectin adhesins remains unclear. C. jejuni also contains a Gln ABC permease, PaqPQ, but strangely, loss of this permease was associated with an enhancement of resistance to aerobic and organic peroxide stress and an improved survival in intracellular macrophage and epithelial cell models. Other amino acid transporters have been observed to have roles in the virulence of specific pathogens, such as the glutamate (Glu) transporter of N. meningitidis GlnT (Colicchio et al. 2009), and no doubt more will be uncovered through targeted in vivo studies.

Peptide transporters

In addition to amino acids, another source of carbon and/or nitrogen utilized by prokaryotes are peptides that are acquired from the environment by di- or tripeptide (Dpp) ABC permeases and/or oligopeptide (Opp) ABC permeases belonging to the peptide/opine uptake family (TCID 3.A.1.5). In addition to their role in cellular metabolism, these peptide ABC permeases are also implicated in signalling processes and in virulence (reviewed in Detmers et al. (2001) and Doeven et al. (2005)). In Group A Streptococcus (GAS) strain CS101, the Dpp ABC permease is associated with uptake of amino acids, heme production, chemotaxis, sporulation and expression of the SpeB cysteine protease, a major virulence factor. In the human pathogen Str. agalactiae, the Opp ABC permease was observed to be involved in regulating virulence, observed as a decrease in adherence to epithelial cells, via reduced expression of fibrinogen-binding protein, FbsA. These observations show that peptide acquisition has an important role in pathogen invasion, although the mechanism, possibly through environmental sensing, remains to be determined.

Vitamins

In the host, vitamin B₁₂ has an important role in natural killer cell activation and lymphocyte production. Bacteria have been proposed to scavenge vitamin B₁₂ from the host environment and, thereby, reduce the hostility of the host environment for themselves (Tamura et al. 1999). The TonB-dependent vitamin B₁₂ importer BtuCED (TCID 3.A.1.13) has been identified in E. coli (de Veaux et al. 1986) and imports vitamin B₁₂ in corrinoids (Borths et al. 2002; Chimento et al. 2003; Locher et al. 2002) to the periplasmic binding protein BtuF (Hvorup et al. 2007). However, the requirement of a vitamin B₁₂ ABC permease for full virulence has not yet been established in an in vivo infection model.

Osmoprotectants

Often during infection, there is increased osmotic pressure at the site of infection. Consequently, to avoid membrane damage from osmotic stress, bacteria acquire osmoprotectants from the environment. These include proline, choline, ectoine, glycine betaine and proline betaine, which are acquired by ABC permeases (TCID 3.A.1.12). In Lac. lactis, OpuABC imports glycine betaine and proline betaine from the extracellular environment (Horn et al. 2005, 2006; van Der Heide and Poolman 2000). These osmolyte transporters are present in a diverse range of human pathogens, such as M. tuberculosis, which expresses the ProXVWZ betaine transporter (Price et al. 2008), and P. aeruginosa, which expresses the ABC permease CbcXWV that has been shown to import choline, betaine and carnitine (Chen et al. 2010). In Sta. aureus, the choline ABC permease OpuABC was found to be important for resistance to salt stress via choline uptake. Loss of the ABC permease rendered Sta. aureus less pathogenic and defective in biofilm formation in vitro and in vivo (Kiran 2009). In Lis. monocytogenes, growth at high salt concentrations is attributed to the accumulation of organic solutes such as glycine betaine and carnitine. In the Lis. monocytogenes LO28 strain, a deletion of three of its osmolyte ABC permease systems (BetL, Gbu and OpuC) resulted in a significant reduction in its ability to cause systemic infection following peroral coinoculation with the wild-type parent in a murine model (Wemekamp-Kamphuis et al. 2002). However, deletion of OpuC alone resulted in similar effects (Sleator et al. 2001), leading to the inference that the OpuABC permease may play an important role in listerial pathogenesis (Wemekamp-Kamphuis et al. 2002).

Antimicrobial peptides

Animals and humans produce small cationic peptides of varying lengths and sequences with antimicrobial properties, which assist in a strong non-specific innate immune response. Such peptides include protamine, melitlin and polymyxin B. It is thought that these peptides destabilize the outer cytoplasmic membrane of pathogens. Many pathogens have evolved methods to withstand the effects of these cationic peptides (Otto 2009) via ABC permeases (TCID 3.A.1.5.1), which internalize these peptides for detoxification. Salmonella Typhimurium has two known systems: the first, SapABCDF, has been associated with protamine detoxification, involving SapA binding the peptide and delivering it to the ABC permease SapBCDF to be internalized (Groisman 1994; Groisman et al. 1992; Parra-Lopez et al. 1993). Once acquired, the current models hypothesize that the bacteria decomposes the peptide, rendering it ineffective. The second known system is the YjeABEF ABC permease that confers resistance to protamine, melittin, polymyxin B and human defensin-1 and -2. Mutations in this permease decreased proliferation in activated macrophages and attenuated survival in a mouse typhoid fever model (Eswarappa et al. 2008). The presence of two permeases in Salmonella Typhimurium indicates that there is functional overlap of these antimicrobial peptide transporters, but the significance of this has not been established outside of their respective infection models. Examples of similar antimicrobial peptide ABC permeases have been found in other pathogens. For example in H. influenzae, the sap operon was upregulated when exposed to cationic peptides and was found to be required for resistance (Mason et al. 2006). In E. coli, the YejABEF importer has also been identified, but a direct role in the virulence of pathogenic strains has not been established (Novikova et al. 2007; Stumpe and Bakker 1997).

Other transporters

In addition to the general classes of transporters above, there is also an increasing number of ABC permeases that have been found to have specialized functions associated with the virulence of pathogenic bacteria. These include M. tuberculosis and the ABC permease LpqY/SugABC responsible for the import of the disaccharide trehalose. Trehalose, which is not present in mammals, arises as a byproduct of mycolic acid biosynthesis, a key constituent of the mycobacterial cell envelope. The trehalose ABC permease is proposed to mediate the retrograde transport of released trehalose, thereby functioning as a recycling system. Perturbation in this recycling process strongly impaired M. tuberculosis virulence, implying a central role for the trehalose ABC permease in pathogenicity (Kalscheuer et al. 2010).

Conclusions

Prokaryotic ABC transporters show a remarkable diversity of functions, some of which are essential for the virulence of pathogenic bacteria. However, the physiological function of many ABC transport systems remains to be elucidated. Indeed many of the biochemically characterized systems, such as the xenobiotic transporters, may have distinct primary physiological functions that supersede their observed roles in drug efflux. Intriguingly, although ABC transporters have a clear contribution to virulence, in contrast to the role of ABC transporters in eukaryotic diseases, these roles are rarely associated with multi-drug efflux. In pathogenic bacteria, these roles are often evasion of or resistance to host defences, coupled with functions that aid colonization of the host environment, such as nutrient scavenging. As a consequence, many of the cell surface or secreted factors may prove to be useful targets for antimicrobial therapeutics or vaccine development. However, their applications in these roles would rely on a clear understanding of their physiological function and role in virulence, and so clearly more research into this ‘frontier territory’ of proteins is needed.