The tripartite ATP-independent periplasmic (TRAP) transporters are the best-studied family of substrate-binding protein (SBP)-dependent secondary transporters and are ubiquitous in prokaryotes, but absent from eukaryotes. They are comprised of an SBP of the DctP or TAXI families and two integral membrane proteins of unequal sizes that form the DctQ and DctM protein families, respectively. The SBP component has a structure comprised of two domains connected by a hinge that closes upon substrate binding. In DctP-TRAP transporters, substrate binding is mediated through a conserved and specific arginine/carboxylate interaction in the SBP. While the SBP component has now been relatively well characterized, the membrane components of TRAP transporters are still poorly understood both in terms of their structure and function. We review the expanding repertoire of substrates and physiological roles for experimentally characterized TRAP transporters in bacteria and discuss mechanistic aspects of these transporters using data primarily from the sialic acid-specific TRAP transporter SiaPQM from Haemophilus influenzae, which suggest that TRAP transporters are high-affinity, Na+-dependent unidirectional secondary transporters.
tripartite ATP-independent periplasmic
periplasmic binding protein
extracytoplasmic solute receptor
Bacteria and archaea live in diverse and often constantly changing environments, where nutrients are usually very scarce. To be competitive in these niches, they have evolved active uptake systems that are able to scavenge low concentrations of nutrients efficiently. These high-affinity transporters share a common feature, the presence of a substrate-binding protein (SBP, often also known as a periplasmic binding protein or extracytoplasmic solute receptor), which is either free in the periplasm of Gram-negative bacteria or anchored to the cytoplasmic membrane in Gram-positive bacteria and archaea. The purpose of the SBP is to bind the substrate with high affinity and specificity, and then to present it to the translocation machinery for unidirectional transport across the membrane (Wilkinson & Verschueren, 2003; Doeven et al., 2004; Mulligan et al., 2009). SBPs are found ubiquitously in bacteria and archaea and were first found to function with ATP-binding cassette (ABC) transporters that are primary transporters driven by direct ATP hydrolysis. The ABC transporters are the subject of an accompanying review by Schneider and colleagues. Only much more recently have SBPs been discovered to function with secondary transporters that utilize pregenerated electrochemical gradients across the cytoplasmic membrane to catalyse the transmembrane movement of substrates (Fig. 1).
Composition of archetypal ABC, secondary and TRAP transporters. ABC transporters, represented by maltose transporter MalEFGK2 (PDB code 2R6G), are composed of an SBP (red), a TMD (blue) and a NBD (green). Conventional secondary transporters, represented here by lactose permease (PDB code 1PV6), are composed of a TMD with no need for ancillary components. TRAP transporters are composed of an SBP, represented by SiaP (PDB code 3B50), and two unequally sized integral membrane components. The small membrane component (DctQ, yellow) contains four TMHs and the large membrane component (DctM, blue) is predicted to have 12 TMHs.
ABC transporters are characterized by the presence of a highly conserved nucleotide-binding domain (NBD) and a less well-conserved pair of transmembrane domains (TMDs) (Fig. 1) (Davidson et al., 2008). They utilize the NBD to couple the free energy associated with the binding and hydrolysis of ATP to the ‘energetically uphill’ accumulation of substrate, resulting in the formation of high substrate gradients (up to 106: 1) across the membrane (Dean et al., 1989; Dippel & Boos, 2005). ABC transporters that use an SBP function almost exclusively in uptake processes, although there is evidence for an SBP-dependent ABC transporter from Sinorhizobium meliloti being able to function in export (Hosie et al., 2001). ABC importers catalyse the uptake of a wide range of substrates that differ considerably in their structure and physiological function, including sugars, amino acids, polypeptides, vitamins and metal–chelate complexes (Davidson et al., 2008).
The known SBP-dependent secondary transporters are currently placed in two families, the tripartite ATP-independent periplasmic (TRAP) transporters (TC: 2.A.56) (Fig. 1), the subject of this review, and a smaller family called the tripartite tricarboxylate transporters (TTT) (TC: 2.A.80), which have a similar tripartite structure, but are not related to TRAP transporters at the level of sequence (Kelly & Thomas, 2001; Winnen et al., 2003). The defining characteristic of secondary transporters in general is that they energize the thermodynamically uphill transport of one solute using the energy derived from the thermodynamically favourable transport of another (or more than one other) solute (also known as the counter-ion, usually H+ or Na+). This is a simplified view of an incredibly diverse family of transporters that not only vary widely in the substrate specificity, but can also vary in the counter-ion specificity and mode of action (Poolman & Konings, 1993). A unifying feature of classical secondary transporters is that, depending on the direction and the magnitude of the electrochemical gradient applied, the substrate can move bidirectionally through the transporter (Poolman & Konings, 1993). This is in contrast to ABC transporters that are generally considered to be unidirectional (Davidson et al., 2008). There are now a number of high-resolution structures of secondary transporters, many of which are Na+ dependent, having between 10 and 14 transmembrane helices (TMH), and function without the need for ancillary proteins (Yamashita et al., 2005; Faham et al., 2008; Weyand et al., 2008). These transporters are proposed to go through four major conformations: outward-facing open, outward-facing occluded, inward-facing occluded and an inward-facing open conformation (Weyand et al., 2008). Occlusion of the substrate-binding site is an essential step in the mechanism to prevent the leakage of molecules and ions across the membrane. In the proposed mechanism, substrate binds to the outward-facing open conformation, which induces a conformation change resulting in closure of the outer opening of the protein and formation of the outward-facing occluded state. The movement of a helical bundle results in conversion from the outward-facing to the inward-facing occluding state, which must then undergo a final structural change to form the inward-facing open state, although the nature of this cannot yet be directly inferred from the currently available structures. Interestingly, these recent structures are from transporters of the SSS, NCS1 and NSS families, all of which share no apparent sequence identify and yet all structurally use the same fold, which consists of an unexpected inverted repeat of a core five-TMH region including a discontinuous or a broken helix that is important for substrate and Na+ binding (Yamashita et al., 2005; Faham et al., 2008; Weyand et al., 2008).
While secondary carriers have been studied for many decades, the discovery of a secondary transporter that utilizes an SBP was not revealed until the 1990s through the work of Prof. David Kelly and colleagues at the University of Sheffield in the United Kingdom. During the characterization of a C4-dicarboxylate transporter from Rhodobacter capsulatus, they purified what appeared to be a normal SBP and characterized ligand binding to this protein in detail (Shaw et al., 1991; Walmsley et al., 1992a). However, when they sequenced the genes flanking the SBP gene, dctP, they could not find any genes for the rest of this assumed ABC transporter. Instead, they found genes encoding two membrane proteins: one a predicted four TMH protein (DctQ) and the other a predicted 12 TMH protein (DctM) (Forward et al., 1997) (Fig. 1). All three components were shown to be essential for the transport of C4-dicarboxylates in R. capsulatus (Forward et al., 1997). Analysis of the transport characteristics revealed that transport by DctPQM was insensitive to orthovanadate, a potent inhibitor of ABC transporters, but sensitive to uncouplers that dissipate the membrane potential, indicating that this transporter was an SBP-dependent secondary transporter (Forward et al., 1997). The dctP gene, the first in the operon, is expressed at much greater levels than dctQM, probably due to the occurrence of a partial transcriptional terminator sequence between the dctP and the dctQM genes (Shaw et al., 1991; Forward et al., 1997), and this is reflected by the fact that DctP can accumulate to high levels in the R. capsulatus periplasm (Shaw et al., 1991).
It was clear that TRAP transporters were not limited to R. capsulatus, but were widespread across the bacteria and archaea (Rabus et al., 1999; Kelly & Thomas, 2001; Mulligan et al., 2007), and sequence analysis provided some initial clues about the functions of the different subunits. The SBP protein did not share obvious sequence homology to SBPs from ABC transporters, but their structures revealed that they share the same overall fold (see the later section on TRAP SBP structures). Two families of SBPs are known to function with TRAP transporters, the most common being related at the sequence level to DctP from R. capsulatus (forming DctP-TRAP systems) and the other being the TRAP-associated extracytoplasmic immunity (TAXI) proteins (forming TAXI-TRAP systems), which are found more rarely in bacteria, but are the only type of TRAP transporter encoded in the genomes of archaea (Kelly & Thomas, 2001). The large membrane component (DctM family) is likely to contain a core of 12 TMHs and is a member of the ion transporter (IT) superfamily (Rabus et al., 1999; Prakash et al., 2003). Sequence analysis has revealed that DctM shares a motif with DcuC – a C4-dicarboxylate secondary transporter of the IT superfamily that requires no ancillary components to function – indicating that DctM has a weak, but significant relationship with classical secondary transporters (Rabus et al., 1999; Prakash et al., 2003). Consequently, and although it has not been explicitly demonstrated experimentally, DctM is predicted to form the membrane translocation pathway of this transporter through which the substrate passes to cross the membrane. DctQ, the small integral membrane component, has four TMHs as determined using reporter fusion experiments (Wyborn et al., 2001) and is the subunit with the lowest level of sequence identity across the family (Rabus et al., 1999). Although it has been shown to be essential for transport in the DctPQM system (Forward et al., 1997), the function of DctQ remains unknown.
The aim of this article is to review the progress made in our knowledge and understanding of TRAP transporters since the first major review of these transporters in 2001 (Kelly & Thomas, 2001). The article has been divided into three main sections; in the first, we review the growing literature on functionally characterized TRAP transporters, in the second, we review the insights into the function of TRAP transporters that have emerged from the structure of the SBP proteins and present the limited amount of information known about the membrane domains. Finally, we review the mechanistic aspects of these transporters by integrating the limited experimental data available into a hypothetical transporter cycle.
The biological diversity of TRAP transporters: variations upon a theme
When this subject was first reviewed in 2001, there were only three TRAP transporters that had been studied in any detail, including the archetypal DctPQM system of R. capsulatus. One was a similar C4-dicarboxylate transporter from Wolinella succinogenes (Ullmann et al., 2000), which is responsible for the majority of transport of the C4-dicarboxylates fumarate and malate for use in carbon metabolism (Ullmann et al., 2000), while the second was a Na+-dependent TRAP transporter for l-glutamate in the cyanobacterium Synechocystis sp. strain PCC 6803 (Quintero et al., 2001). This latter system turns outs to be a rather atypical TRAP transporter as it contains an SBP that is of the type used by ABC transporters rather than a DctP-TRAP SBP (see the later section on the structure and function of TRAP SBPs). In addition, there was a report of an SBP-dependent secondary transporter for l-glutamate in the bacterium Rhodobacter sphaeroides 2.4.1 (Jacobs et al., 1996), which was osmotic shock sensitive (indicative of the presence of an SBP) and stimulated by the presence of Na+; however, the genes encoding this activity were never identified and so it is not clear whether this is a TRAP transporter or not. Finally, there was an interesting report in 2000 of an orphan DctP-TRAP SBP, YbdE, which functions not as part of a TRAP transporter, but rather as a sensor for C4-dicarboxylates. It is divergently transcribed from the genes encoding a two-partner sensor–regulator pair, ybdF and ybdG, and a gene encoding a C4-dicarboxylate transporter of the DcuA family (ybdH). YbdE has been experimentally demonstrated to be an extracytoplasmic C4-dicarboxylate sensor that interacts with YbdFG to activate the transcription of ybdH (Asai et al., 2000). While the R. capsulatus DctPQM system is demonstrably specific for C4-dicarboxylates, it was clear that there were likely to be a range of other substrates for TRAP transporters (Kelly & Thomas, 2001) and this has been proven to be correct. We will now review all the additional TRAP systems that have been identified since 2001 (Table 1) and conclude this section with a perspective on the potential range of TRAP substrates yet to be discovered.
The SiaPQM system has attracted attention due to its role in the virulence of H. influenzae, as sialic acid is an important sugar acid that is essential for colonization of the middle ear by this bacterium and that must be acquired from an exogenous source (Bouchet et al., 2003; Severi et al., 2007; Jenkins et al., 2010). After uptake into the cell, the sialic acid is used to sialylate the lipopolysaccharide, which has a function in immune evasion and hence colonization (Hood et al., 1999; Vimr et al., 2000; Jenkins et al., 2010). Sialic acid is a term encompassing over 40 different naturally occurring nine-carbon sugar acids, and the most common of these is N-acetylneuraminic acid (Neu5Ac) (Table 1) (Angata & Varki, 2002), which is the substrate for SiaPQM. A feature of all higher metazoan cells is the presence of a sialic acid molecule as the outermost constituent on surface-exposed glycoconjugates (glycoproteins and glycolipids). These sialylglycoconjugates have a number of important functions on the surface of eukaryotic cells, including cell-to-cell communication, adhesion and preventing complement activation and autoimmunity (Angata & Varki, 2002; Schauer et al., 2004; Vimr et al., 2004). It has become apparent that a number of human commensal and pathogenic bacteria are able to utilize sialic acids as a source of carbon, nitrogen and amino sugars for use in cell wall synthesis (Plumbridge & Vimr, 1999), in addition to its role in immune evasion.
The siaPQM genes have been studied in both an unencapsulated laboratory strain of H. influenzae RW118 (Rd) (Severi et al., 2005) and the nontypeable strain, 2019 (Allen et al., 2005). Disruptions in either siaP, encoding the SBP or siaQM (termed siaT in strain 2019), encoding the fused integral membrane protein, led to a cessation of radiolabelled Neu5Ac transport activity (Severi et al., 2005) and prevented lipopolysaccharide sialylation with decreased persistence in human serum, indicating that SiaPQM is the sole uptake route for Neu5Ac in H. influenzae (Allen et al., 2005; Severi et al., 2005). The siaPQM genes are regulated by CRP and the presence of sialic acid through the SiaR protein (Johnston et al., 2007) to enable H. influenzae to ensure that sufficient sialic acid taken up by SiaPQM is available for lipopolysaccharide sialylation, where the excess can be catabolized as a carbon and nitrogen source (Vimr et al., 2000; Severi et al., 2005).
The SiaP protein binds Neu5Ac with high affinity [calculated KD using a number of different methods between ∼30 and 120 nM for Neu5Ac (Table 2)] as well as a range of other sialic acids such as Neu5Gc (Muller et al., 2006). Neu5Ac binding to SiaP has also been observed using electrospray ionization (ESI)-MS (Severi et al., 2005) and isothermal titration calorimetry (Johnston et al., 2008), which both confirmed that SiaP binds a single molecule of Neu5Ac. The pre-steady-state kinetics of Neu5Ac binding to SiaP have been established using stopped-flow fluorescence spectroscopy (Muller et al., 2006). The data obtained were consistent with a simple bimolecular interaction similar to that seen in ABC SBPs (Miller et al., 1980, 1983). The implications of this and comparisons with other characterized TRAP SBPs will be covered in the later section of the mechanism of TRAP transporters. Structures of SiaP were determined in 2006 and later in 2008, which has provided considerable insight into function (Muller et al., 2006; Johnston et al., 2008), and the complete SiaPQM system has also been fully reconstituted and characterized in vitro (Mulligan et al., 2009), also discussed in more detail in later sections. The H. influenzae SiaPQM system is now arguably the most well-understood TRAP transporter.
Substrate-binding affinities of characterized TRAP SBPs
Severi et al. (2005), Muller et al. (2006), Johnston et al. (2008)
Walmsley et al. (1992)
Thomas et al. (2006)
Thomas et al. (2006)
Gonin et al. (2007)
Mulligan et al. (2009)
Lecher et al. (2009)
Kuhlmann et al. (2008)
The orthologous sialic acid TRAP transporter in the animal pathogen Pasteurella multocida (Steenbergen et al., 2005) has also been demonstrated to be critical for infection in both mouse (Fuller et al., 2000) and turkey models of infection (Tatum et al., 2009). Similarly, the orthologous genes in the pathogen Vibrio cholerae have also been studied as they are present with other sialic acid catabolic genes in a Vibrio pathogenicity island (Jermyn & Boyd, 2002, 2005). The SBP, VC1779 (VcSiaP), has been purified and characterized in vitro, revealing it to be a high-affinity binding protein for Neu5Ac with a KD of 111 nM (Mulligan et al., 2009). Like in H. influenzae, the genes for the transporter and the catabolism of sialic acid are induced by the addition of Neu5Ac to the growth medium. Real-time PCR was used to demonstrate that the transcription levels of the gene encoding VcSiaP increased 85-fold in the presence of Neu5Ac compared with the presence of glucose (Almagro-Moreno & Boyd, 2009). Unlike H. influenzae, V. cholerae only catabolizes Neu5Ac, although this does confer a significant competitive advantage to V. cholerae during colonization of the gut – a sialic acid-rich environment (Almagro-Moreno & Boyd, 2009).
Compatible solutes such as ectoine and hydroxyectoine are important in balancing osmotic pressure in many organisms that live in high-salt environments, such as halophilic or marine-dwelling organisms. These compatible solutes accumulate in response to osmotic stress and can reach very high concentrations in the cytoplasm without perturbing metabolic functions. Compatible solutes have other beneficial effects by protecting proteins from the detrimental effects of UV radiation and extremes of temperature. Many organisms are able to synthesize their own ectoine and other compatible solutes; however, under certain conditions, it is more economical to sequester compatible solutes from the environment using specialized transporters (Oren et al., 1999).
Transport of ectoine and hydroxyectoine by TRAP transporters has been studied in two different organisms: the halophile Halomonas elongata DSM 2581 and the marine bacterium Silicibacter pomeroyi DSS-3 (Grammann et al., 2002; Kuhlmann et al., 2008; Lecher et al., 2009). The TRAP transporter studied in S. pomeroyi DSS-3 is encoded by the genes uehABC that are coregulated with the ectoine utilization genes, whereas the genes for the ectoine-specific TRAP transporter from H. elongata, teaABC, are not. This demonstrates the fundamental difference between these two organisms with regard to their use of ectoine; H. elongata uses ectoine as a compatible solute, whereas S. pomeroyi dissimilates ectoine and can use it as the sole carbon and nitrogen source (Lecher et al., 2009). This potentially confers an advantage to S. pomeroyi over its marine-dwelling neighbours as the marine environment is poor in nutrients and the use of a transporter with a high-affinity binding protein will allow efficient scavenging of ectoine. In the case of the TeaABC system from H. elongata, it can function to take up ectoine from the environment; however, its primary role is thought to be in the recovery of endogenously synthesized ectoine that has leaked through the membrane (Grammann et al., 2002; Kuhlmann et al., 2008).
Both TeaA and UehA SBPs have been purified and characterized and the crystal structures of both SBPs have also been determined with bound ligand (Kuhlmann et al., 2008; Lecher et al., 2009). UehA recognizes both ectoine and hydroxyectoine with an affinity of approximately 1 μM, while TeaA binds ectoine with a higher affinity (KD=0.19 μM) and also hydroxyectoine with a lower affinity (KD=3.8 μM) (Kuhlmann et al., 2008; Lecher et al., 2009) (Table 2).
Gluconate/malonate and 2,3-diketo-l-gulonate (2,3-DKG)
A study of chemicals that induce expression of the SBP genes for all the TRAP and ABC transporters in the symbiotic nitrogen-fixing bacterium S. meliloti 1021 led to the prediction of a number of new TRAP substrates (Mauchline et al., 2006). Two of the TRAP transporters included in this study have been studied further: one is a predicted malonate-specific TRAP transporter (SMa0151,-55,-57) (Mulligan et al., 2007) and the other is a gluconate transporter, GntABC (SMa0249,-50,-52) (Steele et al., 2009).
Evidence for the malonate-specific TRAP transporter is still only indirect, and comes from the induction (≥3-fold) of SMa0157 in the presence of malonate (Mauchline et al., 2006) and the observation that the genes for this transporter are genetically linked to a gene encoding malonate-CoA synthetase, the first gene in the catabolic pathway of malonate (Kim et al., 2002). Unexpectedly, the gene encoding the large TRAP membrane component was found to be fused to the gene encoding a malonyl-CoA decarboxylase (Mulligan et al., 2007). These findings corroborate the transcriptional inducer study and suggest that this atypical TRAP transporter is specific for malonate. Presumably, malonate is taken into the cell by this TRAP transporter, converted to malonyl-CoA by the malonyl-CoA synthetase enzyme, which is in turn converted to acetyl-CoA by malonyl-CoA decarboxylase, for use in the tricarboxylic acid cycle. Malonate, a C3-dicarboxylate, is abundant in the bacteroides and is an important metabolite in the development of mature nodules, as demonstrated for Rhizobium trifolii (An et al., 2002). However, it is also a known inhibitor of succinate dehydrogenase, and so rapid conversion once inside the cell is likely to be advantageous, and hence a possible reason for this unique fusion. However, there is as yet no experimental demonstration to support the hypothesis that Sma0151-57 can act as both a transporter and a malonyl-CoA decarboxylase.
The gluconate TRAP transporter (Table 1) from S. meliloti has been characterized experimentally and is essential for growth on gluconate (Steele et al., 2009). The structural genes are encoded on the symbiotic megaplasmid pSymA, which also includes the gluconate utilization operon. This system is unusual in that gluconate does not induce expression of the transporter genes (Sma0249-52). In two independent studies, l-arabinose was demonstrated to be able to induce expression (Mauchline et al., 2006; Steele et al., 2009). The reason why the transporter substrate is not the inducer for this transporter and the connection to l-arabinose is not yet clear.
A compound that is also a linear C6-monocarboxylate-like gluconate is 2,3-DKG (Table 1). 2,3-DKG is formed by the hydrolysis of dehydro-l-ascorbate (the oxidized form of l-ascorbate), which occurs spontaneously in aqueous environments at neutral pH (Bode et al., 1990; Deutsch et al., 1998). 2,3-DKG is the substrate of the sole TRAP transporter found in Escherichia coli K-12 encoded by the yiaMNO genes. The genes yiaM, yiaN and yiaO (encoding the large membrane component, the small membrane component and the SBP, respectively) form part of a nine-gene operon (yiaK-S) implicated in carbohydrate metabolism (Badia et al., 1998; Ibanez et al., 2000). The substrate for the YiaMNO transporter has been difficult to identify. First, it was proposed to transport l-lyxose, but l-lyxose transport was unaffected by deletion of the yiaMNO genes (Plantinga et al., 2004). The substrate was then thought to be l-xylulose based on genome context analysis and by indirect binding and transport assays (Plantinga et al., 2004), although the yiaK-S operon was not induced by l-xylulose (Plantinga et al., 2005). The strongest evidence for the identity of the substrate came from the biophysical analysis of the SBP, YiaO (Thomas et al., 2006). ESI-MS, steady-state tryptophan fluorescence and circular dichroism were used to unequivocally demonstrate that 2,3-DKG binds to YiaO, whereas l- and d-xylulose do not bind (Thomas et al., 2006). Although this is the only TRAP transporter in E. coli K-12, the nature of the substrate, 2,3-DKG, is unstable, and the lack of a clear growth phenotype upon deletion of the transporter genes has precluded its further study.
Two orthologous TRAP transporters from R. capsulatus and R. sphaeroides have been characterized independently and the purified SBPs bind α-keto acids such as pyruvate and 2-oxobutyrate (Gonin et al., 2007; Thomas et al., 2006) (Table 1). The SBPs from these TRAP transporters were originally predicted to bind sorbitol and mannitol due to the proximity of the R. sphaeroides SBP gene, originally named smoM, to an operon containing sorbitol and mannitol catabolic genes (Stein et al., 1997). Subsequently, both SBP proteins, RRC01191 from R. capsulatus and TakP from R. sphaeroides, have been purified and characterized in vitro and the real nature of their substrate has been elucidated (Thomas et al., 2006; Gonin et al., 2007). Tryptophan fluorescence spectroscopy of RRC01191 found no evidence for binding of sorbitol or mannitol, but did detect binding of pyruvate and 2-oxobutyrate with a high affinity (3.4 and 0.32 μM, respectively) (Table 2) (Thomas et al., 2006). Using a series of structurally related compounds, it was demonstrated that ligands with longer aliphatic chains such as α-ketovalerate bound with an even higher affinity (Table 2) (Thomas et al., 2006). Subsequently, the TakP protein was also characterized using tryptophan fluorescence spectroscopy revealing similar binding affinities and substrate preferences as RRC01191 (Gonin et al., 2007). Surprisingly, TakP was proposed to form a dimer based on its elution profile from size-exclusion chromatography and this novel observation was confirmed by the solution of the crystal structure of TakP, details of which will be discussed later.
Recently, a crystal structure of TTHA0766, a TRAP SBP from the hyperthermophile Thermus thermophilus HB8, has been described that is most closely related to TakP at the sequence level (Akiyama et al., 2009). This SBP, which we will refer to as LakP for simplicity, crystallized with l-lactate coordinated in the binding site, suggesting that lactate is the physiological ligand for this TRAP transporter.
Taurine (2-aminoethanesulphonic acid)
Taurine is an abundant organic solute in many environments and is utilized by many organisms including bacteria, algae and marine invertebrates (Novak et al., 2004). It can also constitute a component of bacterial cell walls (Kelly & Weed, 1965) and can be used as the sole source of carbon, nitrogen, sulphur and energy in many terrestrial and marine bacteria (Kertesz et al., 2000; Denger et al., 2006).
The taurine utilization pathway has been established in S. pomeroyi DSS-3, which includes an ABC transporter orthologous to a characterized taurine-specific ABC transporter from E. coli (Eichhorn et al., 2000; Gorzynska et al., 2006). A second pathway of taurine dissimilation has been described for R. sphaeroides 2.4.1. This pathway includes a TRAP transporter whose genes are in close proximity to the taurine catabolic genes (Bruggemann et al., 2004). It has now been shown that the transport and catabolic genes are coregulated in response to the addition of taurine, strongly suggesting the involvement of this TRAP transporter in taurine uptake; however, taurine binding or transport by this TRAP transporter has not been demonstrated explicitly. Orthologues of tauKLM have also been predicted in Paracoccus denitrificans NKNIS and Desulfotalea psychrophila LSv54 (Bruggemann et al., 2004; Denger et al., 2006). If taurine is in fact a TRAP substrate, which appears likely, it will be unusual in that it does not contain a carboxylate group, but rather a sulphonate group instead.
There is evidence for the recognition of two aromatic compounds by TRAP transporters, the first of which is the halogenated substrate 4-chlorobenzoate (4-CBA), which is transported by the gene products of fcbT1-3 from Comamonas sp. strain DJ-12 (Chae et al., 2000). These TRAP transporter genes form part of an operon that also contains genes encoding enzymes responsible for the hydrolytic dechlorination of 4-CBA to 4-hydroxybenzoate (4-HBA). 4-HBA can then be metabolized by a host of other enzymes, ultimately resulting in succinyl-CoA and acetyl-CoA (Nichols & Harwood, 1995). Thus, transport of these compounds is an important step in their degradation. The fcbT1-3 genes were also shown to be induced by the benzoate derivatives, 4-bromobenzoate, 4-iodobenzoate, and weakly inducible by 2-chlorobenzoate and 4-fluorobenzoate, indicating a wider potential substrate range encompassing a variety of aromatic compounds (Chae et al., 2000). These aromatic compounds also inhibited the transport of 4-CBA, confirming that they are genuine substrates of this TRAP transporter (Chae et al., 2000).
Another TRAP transporter implicated in the transport of aromatic compounds has been identified in Agrobacterium radiobacter (Contzen & Stolz, 2000). This TRAP transporter is predicted to transport protocatechuate and 4-sulphocatechol as the genes for the TRAP transporter are closely associated with genes encoding the enzyme protocatechuate 2,3-dioxygenase that is responsible for the degradation of these compounds (Contzen & Stolz, 2000). Unlike the 4-CBA TRAP transporter, this has not been confirmed experimentally.
Amino acid-specific TRAP transporters
The final class of compounds that have been identified as substrates for TRAP transporters are the amino acids. There are three examples of amino acid-specific TRAP transporters: pyroglutamate in Bordetella pertussis Tohama I, l-glutamate from Synechocystis sp. strain PCC 6803 and finally l-glutamate/l-glutamine from T. thermophilus HB8, which is a TAXI-TRAP system, which will be reviewed in turn.
Pyroglutamate is an amino acid derivative formed from glutamine, which is present in some proteins including the filamentous haemagglutinin of B. pertussis. The gene encoding the 5-oxoprolinase, which catalyses the conversion of pyroglutamate to l-glutamate in an ATP-dependent manner, is flanked by genes encoding consecutive TRAP transporters (Rucktooa et al., 2007). The SBPs encoded by these two sets of genes, DctP6 (BP1887) and DctP7 (BP1891), have been crystallized and their structures were determined, revealing pyroglutamate coordinated in each binding site. DctP7 has also been demonstrated to bind pyroglutamate with a KD of 0.3 μM (Rucktooa et al., 2007) (Table 2). However, in B. pertussis, both the large membrane protein (dctM-like) genes are frameshifted and this strain was unable to transport pyroglutamate in vivo (Rucktooa et al., 2007). The genes are not frameshifted in the related bacterium Bordetella bronchiseptica RB50, which has a wider host range, and there are orthologous systems in a number of related bacteria that are associated with a 5-oxoprolinase gene including Bordetella parapertussis 12822, Acidovorax avenae ssp. citrulli AAC00-1, Polaromonas sp. JS666, Aurantimonas sp. SI85-9A1, Bradyrhizobium japonicum USDA 110, Hahella chejuensis KCTC 2396 and Roseobacter sp. MED193 (Rucktooa et al., 2007; C. Mulligan & G.H. Thomas, unpublished data). We would predict that all of these TRAP transporters can also transport pyroglutamate; however, experimental evidence would be required to confirm this.
The l-glutamate TRAP transporter from the cyanobacterium Synechocystis sp. strain PCC 6803 has been mentioned earlier (Table 1), but is still unique as it combines the integral membrane components of a TRAP transporter (GtrBC) with an SBP (GtrA) that is homologous to glutamine (GlnH from E. coli)-, glutamate (GluB from Corynebacterium glutamicum)- and glutamate/glutamine/aspartate/asparagine (BztA from R. capsulatus)-binding SBPs from ABC transporters (Quintero et al., 2001). Deletion of the gtrABC genes revealed that this transporter is responsible for Na+-dependent glutamate transport. This is the only example to date of an ABC-type SBP functioning with a TRAP transporter integral membrane component. More recently, we observed the reciprocal combination of a TRAP-type SBP gene associated with the genes encoding the TMD and NBD components of an ABC transporter in the genome of the bacterium Magnetospirillum magneticum AMB-1, which requires experimental verification (Mulligan et al., 2007). However, it does support the idea that ABC and TRAP SBPs are distantly related and have retained enough structural and functional similarity to very occasionally evolve to function with their ‘noncognate’ transporter system.
TAXI-TRAP transporters and Archaea
All of the TRAP transporters described above contain SBPs that are homologous to DctP, but there is a second family of SBPs, the TAXI SBPs, which are less abundant than the DctP-TRAP SBPs in bacteria, but are the only type of TRAP transporters seen in the Archaea, and invariably contain fused membrane domains (Kelly & Thomas, 2001). Evidence as to the substrate range of these TRAP transporters is limited; however, there is a single structure of a TAXI-TRAP SBP, TTHA115, from the bacterium T. thermophilus HB8, which crystallized with a ligand that could be interpreted as either an l-glutamate or an l-glutamine (Takahashi et al., 2004), and more recently, a TAXI-TRAP transporter from Psychrobacter arcticus 273-4, a bacterium adapted for growth down to temperatures of −10 °C (Bakermans et al., 2009), has been partially characterized. Deletion of the genes encoding this transporter led to decreased growth on glutamate, acetate, butyrate and fumarate (Bakermans et al., 2009). It has been speculated previously that the substrate range of TAXI-TRAP transporters may be limited to glutamate/glutamine or closely related compounds (Mulligan et al., 2007), but these data suggest a transporter with a potential wider range of substrate and awaits further characterization.
Although no TRAP transporters from Archaea have been functionally characterized as yet, the TAXI-TRAP systems are present across the Archaea, including euryarchaeota, crenarchaeota and korarchaeota, but with only between one and three systems encoded on any one genome, compared with some bacteria, which have over 20 different systems (Mulligan et al., 2007). Examination of the genome context of the transporter genes for these systems reveals some clues to the potential functions of these TAXI-TRAP systems. The TRAP transporters encoded by the genes, pcal_0207 and pcal_0208 in the crenarchaeon Pyrobaculum calidifontis JCM 11548, are adjacent to a gene (pcal_0209) encoding a predicted glutamine amidotransferase, which could catalyse the transfer of the ammonium group from glutamine to a specific substrate. The TRAP transporter encoded by AF0466-AF0467 in Archaeoglobus fulgidus is in close proximity to the genes encoding the 2-oxoglutarate oxidoreductase subunits α–δ (otherwise known as 2-oxoglutarate synthase). Finally, the transporter gene NC_000854 from Aeropyrum pernix K1 is divergently transcribed from a gene encoding a predicted glutamate-1-semialdehyde aminotransferase. This enzyme converts l-glutamate-1-semialdehyde to 5-aminolevulinate, which is an intermediate in the haem biosynthetic pathway. This enzyme catalyses the reversible carboxylation of 2-oxoglutarate to form its CoA derivative. While this in silico data are only suggestive, it supports the data from the study of bacterial TAXI-TRAP systems that they are likely to transporter glutamate/glutamine or closely related C5-dicarboxylates such as 2-oxoglutarate. However, a clear biochemical demonstration of the function of TAXI-TRAP transporter from either a bacterium or an archaea is still absent from the literature.
How functionally diverse are TRAP transporters likely to be?
Since the first review of TRAP transporters in 2001 (Kelly & Thomas, 2001), the number of known or very likely substrates for TRAP transporters has grown considerably, primarily for DctP-TRAP systems. Structurally, the known substrates are united by the presence of a carboxylate group, which turns out to be important for the recognition of the substrate by the DctP-type SBPs (see next section), although in the case of taurine, the carboxylate is replaced by a sulphonate group. While this clear requirement for a carboxylate (or sulphonate) group in the substrate will ultimately constrain the range of substrates that are recognized by TRAP transporters compared with ABC transporters, there are still many hundreds of organic acids that are potential TRAP substrates. The large numbers of evolutionarily diverse TRAP transporters that are now being found in marine environments, especially in the SAR11 clade of Alphaproteobacteria (Morris et al., 2002), suggest that they are likely to have much greater diversity of function than previously suspected.
It is also important that genome annotators are fully aware of TRAP transporters, as there are many examples where either the SBPs are annotated as components of ABC transporters, or the substrate is stated as C4-dicarboxylates, even though it is now well established that C4-dicarboxylate transporters form only a small subset of the known TRAP transporters.
The structures of TRAP transporters
One of the biggest improvements in our knowledge of TRAP transporter function since 2001 has come through the resolution of the crystal structures of a number of TRAP SBP proteins, predominantly of the DctP-TRAP family, but also the TAXI-TRAP family (Table 1). This has provided considerable insight into how these proteins recognize their ligands with a high affinity and specificity, and has revealed some general principles of DctP-TRAP SBPs. The structures of the membrane domains are not known, although some limited experimental and in silico data are available, which will be considered later in this section.
The structures of TRAP SBP proteins
The first structure of a TRAP SBP to be determined was the SiaP protein from H. influenzae, a DctP-type SBP (Muller et al., 2006) (Fig. 2). Structurally, SiaP can roughly be divided into two domains that are composed of α-helices and β-sheets (Fig. 2c). They are connected by a characteristic ‘mixed hinge’, unique to DctP-type TRAP SBPs, that is comprised of two β-strands and a long α-helix that forms the backbone of the structure by spanning the whole breadth of the molecule (Fig. 2a). Ligand binding occurs in the cleft that separates the two domains (Fig. 2a). According to our current working model, the carboxylate group of the ligand, Neu5Ac in this case, forms a salt bridge with a highly conserved arginine residue in the SBP (Arg147 in SiaP) (Fig. 2d). Once initiated by this interaction, the protein likely undergoes a rigid body conformational rearrangement through an unknown sequence of events. The result is that the ligand is buried within the binding cavity (Fig. 2b), which ensures high-affinity binding by involving a distinct set of residues from both domains that are in proximity for interaction in the closed form of the protein (Fig. 2d).
The structural basis of substrate recognition by SiaP, a typical TRAP SBP. The conformational change that SiaP undergoes in the transition from (a) the open unliganded to (b) the closed liganded form. Structurally, SiaP can be divided into two domains that are each comprised of α-helices (red) and β-sheets (yellow), where loops are shown in green. Upon ligand binding (Neu5Ac shown in blue sticks) to the cleft, the equilibrium is changed towards the closed conformation. (c) A schematic topology diagram of SiaP, modified from Muller et al. (2006), with domain I in light blue and domain II in dark blue. α-Helices are in circles (α1–α11) and β-strands are in triangles (1–11), while short 310 helices are numbered η1–η5. Helix 9 is in both colors as this spans the whole length of the protein. (d) A zoom of the Neu5Ac-binding site in SiaP, illustrating the conserved salt bridge between the carboxylate and a highly conserved arginine residue in domain II (Arg147 in SiaP).
There are now structures for a range of different DctP-type TRAP SBPs, including a further structure of Neu5Ac-SiaP (Johnston et al., 2008), the ectoine-binding proteins TeaA from H. elongata (Kuhlmann et al., 2008) and UehA from S. pomeroyi (Lecher et al., 2009), the pyruvate-binding protein TakP from R. sphaeroides (Gonin et al., 2007), the pyroglutamate-binding proteins DctP6 and DctP7 from B. pertussis (Rucktooa et al., 2007a), an uncharacterized TRAP SBP from Thermotoga maritima (TM0322) (Cuneo et al., 2008) and a lactate-binding protein TTHA0766 (LakP) from T. thermophilus (Akiyama et al., 2009) (Table 1). By comparing all the known structures of TRAP SBPs (M. Fischer & G.H. Thomas, unpublished data), it is clear that they exhibit a very similar overall fold (Table 3). Structurally, the SBPs superimpose with a root mean square deviation between 0.9 and 3.3 Å (Table 3). Remarkably, the amino acid sequence identity between the SBPs is only around 20% in most cases, where TakP and TM0322 are the most dissimilar, with 12% identity. In terms of how they all recognize their substrates, it is clear that the Arg/carboxylate salt bridge is strictly conserved, while the rest of the binding site shows little conservation, which is a reflection of the structural variety of the ligands that are bound by these proteins.
Root mean square deviations (RMSDs) and sequence identities (SI) for DctP-type TRAP SBPs on the basis of Secondary Structure Matching superpositions using http://www.ebi.ac.uk/msd-srv/ssm
The PDB code used for the superpositioning is indicated in the vertical column.
SiaP is the only monomeric DctP-type SBP for which both the open unliganded and the closed liganded structures are available, which is partly due to the fact that most of the SBPs studied copurify with their substrate bound, unless this is removed explicitly, and are therefore likely to crystallize in a substrate-bound form. Because the open and closed forms of SiaP are only crystallographic snapshots of the unliganded and liganded states, respectively, the structural dynamics of the transition upon substrate binding and release remains to be elucidated.
While the structures of monomeric DctP-TRAP SBPs are basically the same and all use the conserved Arg/carboxylate pair in substrate recognition, the structure of TakP was interesting as it revealed two important variations in the typical SiaP-like structure. The first was that the substrate, pyruvate in this case, was bound in a complex with a sodium ion, forming part of the ligand-binding site (Gonin et al., 2007). This corecognition of the substrate with a cation has also been seen in LakP, where calcium lactate was observed in the binding site (Akiyama et al., 2009). These observations are particularly pertinent now that evidence has been presented that TRAP transporters use the electrochemical sodium gradient in their transporter mechanism (see the next section) (Mulligan et al., 2009), suggesting a possible use of this bound ion in the transport cycle. However, we think that the cation is more likely to serve a simpler structural function, allowing binding site plasticity (M. Fischer & G.H. Thomas, unpublished data). We propose that the metal ion facilitates the binding of relatively small substrates such as pyruvate and lactate, whereas in TRAP SBP structures with larger substrates bound the function of the cation is replaced by amino acid side chains emerging from the protein. Akiyama et al. (2009) were also able to crystallize LakP with zinc coordinated in the binding pocket instead of calcium, or with calcium totally removed and replaced by water, strengthening the suggestion of a solely structural role for the coordinated metal ions.
The second important difference between TakP/LakP and other TRAP SBPs, including SiaP, is that both TakP and LakP form dimers in solution and in their respective crystal structures (Gonin et al., 2007; Akiyama et al., 2009), although the functional consequences of this have not been investigated to date. We have analysed the distribution of electrostatic surfaces on all DctP-type SBP structures and observed that TakP and LakP exhibit a hydrophobic patch on their dimerization interface (Fig. 3), which is not present in the monomeric structures. In both TakP and LakP, the proteins form a ‘back-to-back’ dimer (Gonin et al., 2007; Akiyama et al., 2009), but the position of the extended C-terminal helix, common to both these proteins, is different (M. Fischer & G.H. Thomas, unpublished data). Differences in the surface charges of SBPs may also be a reflection of the natural habitat of the organism from which the protein was derived. As an example of this, the ectoine- and hydroxyectoine-binding proteins, TeaA and UehA, exhibit a high degree of negative surface charge, suggesting an adaption to the saline environment of these marine organisms (Kuhlmann et al., 2008; Lecher et al., 2009). In addition to solution and crystallographic evidence for TakP and LakP dimer formation, TM0322, an SBP of unknown substrate specificity from T. thermophilus HB8, was proposed to form a higher oligomeric state (Cuneo et al., 2008). These oligomers were constructed from the monomeric crystal structure on the basis of size exclusion chromatography and small-angle X-ray scattering data. It is apparent from the surface representation of TM0322 (Fig. 3) that it resembles the monomeric SiaP-like structures more closely than the TakP/LakP protomers, and hence the mode of oligomerization is likely to be different. While the exact benefits of forming dimers/higher oligomers remain to be elucidated, the back-to-back dimerization of TakP and LakP is at least in agreement with the proposed working model of monomers interacting with the membrane component and the productive delivery of the substrate to the translocation channel (see the next section).
Electrostatic surfaces of monomeric and dimeric TRAP SBPs. All structures are shown in the same orientation based on an SSM superposition (Table 3) and are viewed towards the dimer interface in TakP. The overall architecture is generally very similar, whereas the hydrophobic patch forms the dimer interface of TakP and LakP is evident for these two proteins, but not in monomeric TRAP SBPs such as SiaP and TeaA.
The structures of the membrane domains
While there are no crystal structures of the membrane domains from TRAP transporters, there are limited experimental and computational data that are pertinent to describe in relation to their organization and possible functions. As the large membrane domain, DctM, is a member of the IT superfamily (Rabus et al., 1999), an assignment that is also supported by hydropathy profile analysis (Lolkema & Slotboom, 2003), it is very likely to form the substrate translocation pathway. A number of studies have provided evidence supporting a conserved core of 12 TMH's in the DctM family with an NOUT COUT topology (Rabus et al., 1999; Kelly & Thomas, 2001; Prakash et al., 2003), but this has not been demonstrated experimentally to date.
Analysis of the Pfam entry for DctM (PF06808) (Finn et al., 2010) reveals only a small number of highly conserved residues in the DctM proteins, despite the DctM subunit being the most conserved of the three components (Rabus et al., 1999). Here, we report a previously unrecognized pair of conserved twin proline residues, which occur in helices IV and XI of the DctM subunit (highlighted on the topological model of SiaQM in Fig. 4). These residues are likely to induce discontinuity or breaking of their respective TMH, which is a key structural feature observed in other Na+-dependent secondary transporters (Yamashita et al., 2005; Faham et al., 2008; Weyand et al., 2008). In fact, these other Na+-dependent transporters all share a common 10 TMH (5+5) topology that is not apparent at the level of sequence, which would be consistent with the location of the conserved prolines in helices IV and XI if the DctM proteins had evolved from an ancestral 5+5 duplication (which would now be helices I–V and VIII–XII in Fig. 4) and then inserted two additional helices between the repeats. This hypothesis is also supported by the observation that extra helices in DctM family proteins have only been observed at the N- or C-termini or between helices V–VI and VII–VIII (Prakash et al., 2003). Clearly, only a high-resolution 3D-structure of a DctM protein will provide an unambiguous answer to the functional organization of the TMH within the DctM proteins, although there is much that could be learnt from traditional molecular approaches if a high-resolution structure is not immediately achievable.
Schematic model of the predicted structure of SiaQM, a natural DctQ/DctM fusion protein. The small subunit is in yellow, and the large subunit is in dark blue. The additional linking helix present in this protein is indicated in green. The pairs of highly conserved twin prolines present in helices IV and XI (of the DctM portion) are indicated in red. The N- and C-termini of the protein are indicated. The topology model for DctQM was constructed by performing a multiple sequence alignment of seven DctQM members predicted to transport Neu5Ac, which includes four fused DctQM proteins and three separate DctQ and DctM components. The topology model for all seven membrane components was determined using TMHMM (Krogh et al., 2001) and this was overlaid onto the multiple sequence alignment to define the consensus helix boundaries that are mapped onto the sequence of Haemophilus influenzae SiaQM.
While considering the evolution of the DctM subunit, it is worth mentioning that in the TTT transporters, which have an organization analogous to that of TRAP transporters with an SBP and a large (12 TMH) and small (four TMH) membrane subunit, the large subunit is not defined as a member of the IT superfamily of proteins (Prakash et al., 2003), and so appears to have evolved from a different ancestral secondary transporter.
Regarding the small subunit, DctQ, of the TRAP transporter, there is one experimentally determined topology for the DctQ protein from R. capsulatus, which confirmed an NIN CIN topology with four TMH for this protein (Wyborn et al., 2001). The sequence identity between DctQ subunits is the lowest in the TRAP transporter subunits, although all members of this family are around the same length as R. capsulatus DctQ. However, while there is no experimental evidence to suggest that the topology of R. capsulatus DctQ is conserved throughout the family, this topology is supported by detailed analysis of the sialic acid clade of TRAP transporters (C. Mulligan & G.H. Thomas, unpublished data), as illustrated in Fig. 4. The SiaQM protein from H. influenzae is one member of this clade where both subunits are fused, and in this and other ‘fusion’ proteins, there is in silico evidence for an additional predicted helix that would be required to connect the cytoplasmic C-terminus of the small subunit to the periplasmic N-terminus of the large subunit (Fig. 4).
As noted previously, analysis of the genetic organization of TRAP transporter genes, from over 1000 different regions, revealed that the gene encoding the small subunit is almost always encoded upstream of the gene for the large subunit (Kelly & Thomas, 2001; Mulligan et al., 2007), which could suggest a chaperone-like function for the small subunit in the assembly of the large subunit, or at least a requirement for the small subunit to fold before the large subunit. A more obvious role for the DctQ subunit is to act as a ‘landing pad’ for the SBP (Kelly & Thomas, 2001). However, there is currently no evidence to support either of these hypotheses about the function of the small subunit and experimental evidence for the biological function of this essential subunit is still almost totally lacking.
How might TRAP transporters work?
The transport mechanisms of selected SBP-dependent ABC transporters and certain conventional secondary transporters are now relatively well understood, especially with the elucidation of a number of high-resolution crystal structures at various stages of the transport cycle (Yamashita et al., 2005; Davidson et al., 2008; Faham et al., 2008; Weyand et al., 2008). However, the functioning of an SBP with a secondary transporter raises a number of mechanistic questions that are currently unanswered. In this final section, we consider the different stages in the likely transporter cycle using the limited experimental data from known TRAP transporters, which comes primarily from the H. influenzae SiaPQM system.
Binding of substrate by the SBP
As we have described in this article, the structures of TRAP SBP proteins have been resolved in either closed conformations with a bound substrate or open substrate-free conformations. A number of groups have determined the binding affinities of the SBPs for their substrates, which are usually in the submicromolar range (see Table 2 for details and references).
For ABC transporters, ligand binding by selected SBP has been studied using stopped-flow fluorescence spectroscopy to determine pre-steady-state kinetics of binding (Miller et al., 1980, 1983), exploiting the large conformation change that these proteins undergo when they close to bind ligand, which is reflected in changes in the overall fluorescence of the tryptophan residues in the protein. In all cases for ABC SBPs, the rate of fluorescence change was linearly dependent on the concentration of the substrate, consistent with a simple bimolecular association with the ligand binding to a single site on the protein via a one-step mechanism [Eqn. (1), where P is the protein and L is the ligand (Miller et al., 1980, 1983)].
These data imply that in the absence of a ligand, the majority of the ABC SBP is in an open conformation ready for binding, as opposed to a mixed population of open and closed unliganded SBPs, and upon binding of the ligand, it adopts the closed conformation. For TRAP SBPs data from multiple DctP-TRAP SBPs suggest two different kinetic schemes that describe the mechanism of binding. Both the Neu5Ac-specific SiaP protein and the 2,3-DKG-specific YiaO protein exhibit kinetic behaviour similar to the ABC transporter SBPs mentioned previously (Muller et al., 2006; Thomas et al., 2006), suggesting that binding of ligand by these two TRAP SBPs is a simple bimolecular interaction via a one-step mechanism.
Conversely, analysis of the ligand-binding kinetics of DctP and RRC01191 (the TakP homologue), both from R. capsulatus, revealed that the observed rate of fluorescence change upon ligand binding decreased hyperbolically with increasing substrate concentration, indicating that a different ligand-binding mechanism was being used in these SBPs. These data were interpreted to suggest that an initial isomerization of the protein from a closed to an open conformation is required before binding of ligand and formation of the closed liganded complex (Walmsley et al., 1992) [Eqn. (2)].
The proposed mechanism in Eqn. (2) suggests the presence of two populations of unliganded SBPs: the closed unliganded form (P1), which cannot bind ligand, and the open unliganded form (P2). The equilibrium between the open and the closed unliganded SBP in these cases favours the closed unliganded state (96%), thus requiring the SBP to open first before binding ligand (Walmsley et al., 1992b). This is compared with SiaP and YiaO (and the ABC SBPs mentioned previously), where the equilibrium predominantly favours the open conformation precluding the need for the initial isomerization of the protein. These data clearly show a mechanistic variation in ligand binding to SBPs within the TRAP transporters. The significance of the two mechanisms is not known; however, one of the proteins exhibiting the more complex kinetics, RRC01191, is a homologue of TakP, which is a known dimeric TRAP SBP (Gonin et al., 2007). Therefore, it is possible that the more complex kinetics are connected to the oligomeric states of the SBP, which have previously always been assumed to be monomers. Clearly, structures of either YiaO or DctP where the kinetics are known could support or refute the hypothesis that the complex kinetics are seen in dimeric SBPs, where simple kinetics are seen in monomeric SBPs.
The Na+ dependence of TRAP transporters
The energetic requirements of transport via TRAP transporters have been studied in a number of different systems. The first TRAP transporter to be described, DctPQM from R. capsulatus, was initially differentiated from SBP-dependent ABC transporters by the negative effect of uncouplers on transport of succinate (Forward et al., 1997). The identity of the counter-ion was not elucidated; however, various data for other TRAP transporters suggested that Na+ might be the coupling ion in TRAP transporters. The characterization of an SBP-dependent secondary transporter from R. sphaeroides, which is presumed to be a TRAP transporter (Jacobs et al., 1996), demonstrated that transport of glutamate in whole cells using this transporter was Na+ dependent. Also, the in vivo characterization of the GtrABC TRAP transporter from Synechocystis sp. strain PCC 6803 was found to be Na+ dependent (Quintero et al., 2001).
More recent in silico analyses have also suggested that TRAP transporters in general may be Na+ dependent. By calculating the TRAP densities (number of TRAP transporters per Mb genomic DNA) of close to 200 prokaryotic genomes and analysing the lifestyles of the organisms, we discovered that the four organisms with the highest TRAP densities (total numbers of complete TRAP transporters ranging from 21 to 28) were all marine-dwelling organisms (Mulligan et al., 2007). For two of these organisms, S. pomeroyi DSS-3 and Jannaschia sp. CCS1, data were available from TransportDB revealing that TRAP transporters make up 10% and 9% of the total transporter complement in these organisms, respectively (Mulligan et al., 2007). The average TRAP density from 70 organisms where the data were available was calculated to be 2%. Other marine-dwelling organisms with higher than average percentage contributions of TRAP transporters were identified and found to span a number of phyla, indicating that this enrichment is not due to lineage-dependent expansion (Mulligan et al., 2007). This observation fits with previous work showing that marine bacteria prefer to utilize Na+ as a counter-ion due to its abundance and the fact that seawater is slightly alkaline (Tsuchiya & Shinoda, 1985; Kogure et al., 1998).
Recently, the purification and reconstitution of the Neu5Ac-specific TRAP transporter SiaPQM into liposomes allowed analysis of Neu5Ac transport in vitro, revealing that substrate translocation is powered by an inwardly directed Na+ gradient (ΔμNa+) (Mulligan et al., 2009). The use of SiaPQM-containing proteoliposomes allows specific gradients to be applied across the liposome membrane, thus facilitating a thorough dissection of the energetic requirements of transport by SiaPQM. Application of a pH gradient (ΔpH, inside alkaline compared with outside) or the membrane potential (ΔΨ, inside negative compared with outside) individually resulted in no transport of radiolabelled Neu5Ac into the lumen of the liposome. However, upon addition of an inwardly directed ΔμNa+, rapid transport was observed, indicating that SiaPQM requires ΔμNa+ to power transport. When applied in combination with other electrochemical gradients, it was revealed that the rate of transport increased when the ΔμNa+ was applied with the ΔΨ, suggesting that transport of Neu5Ac is electrogenic (Mulligan et al., 2009). Neu5Ac is negatively charged at physiological pH (pKa=2.6) and the ΔΨ that is applied produces a charge separation across the membrane with the inside negative. The fact that Neu5Ac is negatively charged and that the addition of a ΔΨ increases the transport rate, despite it making the inside of the proteoliposome negative compared with the outside, suggests that more than one Na+ ion is transported per molecule of Neu5Ac (Mulligan et al., 2009). However, the exact stoichiometry of Neu5Ac to Na+ is not known.
The SBP plays an essential role in the transporter cycle and confers unidirectionality on a secondary carrier
In vivo studies of transport by DctPQM from R. capsulatus and other TRAP systems have shown that the deletion or the removal of the SBP compromises the transport function of the system (Forward et al., 1997). This observation has now been shown to be true in vitro using membrane-reconstituted SiaPQM (Mulligan et al., 2009). These data confirm that the SBP is an essential part of the transporter, and under the conditions tested, the integral membrane components could not transport substrate without it. While it is then clear that the SBP must interact with the membrane domains of the transporter, the subunit with which it interacts is not known and is an important question that needs to be addressed in elucidating the transporter mechanism.
This same study also showed that SiaPQM does not behave like a classical secondary transporter (Mulligan et al., 2009). Conventional secondary transporters such as lactose permease from E. coli are able to transport substrate in both directions across the membrane depending on the magnitude and the direction of the electrochemical gradients. Solute counterflow and exchange reactions are used to characterize secondary transporters and indicate whether a transporter is capable of bidirectional substrate movement (Wong & Wilson, 1970; Marty-Teysset et al., 1996). When these procedures were applied to SiaPQM-containing proteoliposomes, no bidirectional movement of substrate was observed, indicating that SiaPQM is a unidirectional secondary transporter.
The nature of the interaction between the SBP and the membrane domains is still unknown and there are no data as yet to suggest whether the SBP interacts with the small or large subunits or both subunits during the transport cycle. The interaction between the SBP and the membrane must clearly be very specific to each system and this has been demonstrated using the V. cholerae SiaP orthologue, VcSiaP, which is 49% identical to SiaP, but is unable to catalyse uptake in vitro using the reconstituted H. influenzae SiaQM proteins (Mulligan et al., 2009). It remains unknown which residues impart the specificity of this interaction, but it is clearly an important question to the understanding of the transport mechanism.
An interesting observation from the in vitro characterization of SiaPQM was that addition of excess unliganded SiaP could induce efflux of preaccumulated radiolabelled Neu5Ac from the proteoliposomes (Mulligan et al., 2009). This was dependent on a specific interaction between SiaP and SiaQM because VcSiaP could not induce the same phenotype. This rather artificial set-up implies that the unliganded SBP must be able to dock onto SiaQM to initiate this reverse transport reaction. While this efflux phenomenon is an interesting observation, it is unlikely to be physiologically relevant as in vivo transported Neu5Ac will be rapidly catabolized, but does demonstrate that under extreme conditions, the TRAP transporter can work in reverse as would be expected for a secondary transporter.
Working model for the catalytic cycle of TRAP transporters
The current working model proposed for the TRAP transporter cycle is summarized in Fig. 5, and is based on our knowledge of the SBP subunit, the energetics of the reconstituted SiaPQM system and analogy to known secondary transporters, although there is no experimental support for many of the stages of the cycle. The periplasmic DctP subunit first binds a substrate that induces a conformational change to produce the closed liganded form of the SBP (step 1). This form is recognized by DctQM in an assumed ‘resting state’ (step 2). The ‘resting state’ of DctM, which is the component predicted to form the translocation channel, is assumed to be such that the periplasmic opening is closed (outside-closed). This assumption is made because the structures of some other secondary transporters, LacY and GlpT, are both in the outside-closed conformation (Abramson et al., 2003), but there is no experimental support for this assumption. Either concomitant with or subsequent to the docking of the substrate-loaded DctP protein, the binding and/or the translocation of at least two external Na+ ions would somehow induce the opening of the SBP and subsequently the rearrangement of DctM to the outside-open conformation (step 3). The substrate would be released into the open cleft of DctM, at which point DctM would then reorientate back to the initial outside-closed conformation (step 4), allowing the substrate to diffuse into the cytoplasm and releasing unliganded DctP to bind more substrate (step 5). The precise order and timing of these latter steps has no experimental support, although clearly, these steps must occur for the transporter to eventually complete a cycle to repeat stage 1 again. Elucidating the details of these stages, especially how the transporter couples the electrochemical gradient to the opening of the SBP and ligand release, awaits further experimental study.
Hypothetical working model of the TRAP transport mechanism. The mechanism of transport by TRAP transporters is represented by five discrete steps; (1) the extracytoplasmic SBP binds substrate, which induces a large conformational change in the SBP forming the closed liganded form of the SBP. (2) The closed liganded form of the SBP interacts with DctQM in its resting state (outside closed). (3) The binding and/or the translocation of two Na+ ions induces the outside open conformation in DctM, which also opens the SBP, releasing the substrate into the translocation channel. (4) The two Na+ ions are released into the cytoplasm. DctM returns to its resting state, releasing the substrate into the cytoplasm. (5) Unliganded SBP is released and is able to bind the substrate to start the cycle again.
In conclusion, we might consider why SBP-dependent secondary transporters are utilized by bacteria instead of the more abundantly used ABC transporters or conventional secondary transporters. If, as is suggested, TRAP transporters cotransport two Na+ ions with one substrate molecule, the energetic cost will be lower than that of ABC transporters (Mulligan et al., 2009), while the use of the SBP allows the transporter to function with an affinity similar to an ABC transporter. TRAP transporters then appear to provide the best of both worlds, and emerging metagenomic and metaproteomic data suggest that they are widely used in marine bacteria that live in Na+-rich environments and that are nutritionally poor (Morris et al., 2002). Also, the plethora of new TRAP transporters being discovered in these bacteria, some of which have over 20 different systems in the same organism, also suggests that there are many more substrates out there that are recognized by TRAP transporters than we currently know.
M.F. is funded by the Wild fund and BBSRC. C.M. is funded through a BBSRC grant (BBF0147591), and the Thomas lab acknowledges its ongoing funding by the BBSRC.
(1998) A rare 920-kilobase chromosomal inversion mediated by IS1 transposition causes constitutive expression of the yiaK-S operon for carbohydrate utilization in Escherichia coli. J Biol Chem 273: 8376–8381.
(2008) Structural analysis of a periplasmic binding protein in the tripartite ATP-independent transporter family reveals a tetrameric assembly that may have a role in ligand transport. J Biol Chem 283: 32812–32820.
(1997) TRAP transporters: a new family of periplasmic solute transport systems encoded by the dctPQM genes of Rhodobactercapsulatus and by homologs in diverse gram-negative bacteria. J Bacteriol 179: 5482–5493.
(2007) Crystal structures of an Extracytoplasmic Solute Receptor from a TRAP transporter in its open and closed forms reveal a helix-swapped dimer requiring a cation for alpha-keto acid binding. BMC Struct Biol 7: 11.
(2002) New type of osmoregulated solute transporter identified in halophilic members of the bacteria domain: TRAP transporter TeaABC mediates uptake of ectoine and hydroxyectoine in Halomonas elongata DSM 2581(T). J Bacteriol 184: 3078–3085.
(2006) Conservation of structure and mechanism in primary and secondary transporters exemplified by SiaP, a sialic acid binding virulence factor from Haemophilus influenzae. J Biol Chem 281: 22212–22222.
(2007) Tripartite ATP-independent periplasmic transporters: application of a relational database for genome-wide analysis of transporter gene frequency and organization. J Mol Microb Biotech 12: 218–226.
(2005) Sialic acid transport in Haemophilus influenzae is essential for lipopolysaccharide sialylation and serum resistance and is dependent on a novel tripartite ATP-independent periplasmic transporter. Mol Microbiol 58: 1173–1185.
(1997) Cloning, nucleotide sequence, and overexpression of smoS, a component of a novel operon encoding an ABC transporter and polyol dehydrogenases of Rhodobacter sphaeroides Si4. J Bacteriol 179: 6335–6340.
(2003) Crystal structures of periplasmic solute-binding proteins in ABC transport complexes illuminate their function. ABC Proteins: From Bacteria to Man (HollandIB, ColeSPC, KuchlerK, HigginsCF, eds). Elsevier, Amsterdam.