Introduction

The view that amino acid efflux in industrial processes is artificial and therefore specific mechanisms for amino acid excretion are unlikely originally prevented investigations on the efflux of amino acids as occurs, for instance, with mutant strains of Escherichia coli and Corynebacterium glutamicum. However, Krämer and coworkers succeeded in demonstrating that the excretion of l-threonine from E. coli and the excretion of l-glutamate, l-lysine, l-isoleucine, and l-threonine from C. glutamicum is an active process (Krämer 1994). Based on these seminal studies, it has recently become possible to identify specific export carriers at the molecular level. This has led to the identification of novel transporter families and to basic insights into the bacterial amino acid balance. This mini-review summarizes what is known on these new types of carriers of C. glutamicum and E. coli. As yet, functional identifications of the carriers identified are only available for these two organisms since they are the main bacteria used for amino acid production. As evident from the numerous genome sequences now available, export carriers are widespread. Recent reviews are available on the production of amino acids by C. glutamicum and E. coli and on aspects of amino acid export by these bacteria (Burkovski and Krämer 2002; Eggeling and Sahm 2001; Eggeling et al. 2001a).

The cell wall of C. glutamicum

While E. coli is probably quite familiar to the reader, this may not be the case with C. glutamicum; therefore, the special features of the cell wall structure of these two organisms will be briefly compared here. This is meaningful since several components of the cell wall are of significance for metabolite transport (Eggeling and Sahm 2001). Within the actinomycetes, C. glutamicum and closely related Corynebacterium species such as C. melassecola, C. lactofermentum, C. flavum, C. efficiens, and Brevibacterium thiogenitalis belong to the suborder Corynebacterianeae (Stackebrandt et al. 1997). This taxon also includes, for example, Mycobacterium tuberculosis. The Corynebacterianeae are characterized by an unusual cell wall (Brennan and Nikaido 1995; Puech et al. 2001). Whereas the classification according to Gram-staining only recognizes two types of bacterial cell walls, various biochemical analyses have led to a much more differentiated picture (Fig. 1). Thus, in addition to the cytoplasmic membrane, bacteria belonging to the Corynebacterianeae have a second lipid layer representing a pseudo outer membrane. In this respect, they resemble bacteria such as E. coli. The major fraction of this outer lipid layer consists of special branched fatty acids, the mycolic acids. The cytoplasmic membrane represents the major barrier to amino acid export, but the outer lipid layer can additionally control the efflux of certain amino acids. This will be discussed below in connection with l-glutamate efflux. At least for M. tuberculosis and Mycobacterium smegmatis, the outer lipid layer is regarded as a considerable permeation barrier that, in this case, presents a serious obstacle to the diffusive access of antibiotics (Brennan and Nikaido 1995). Current research on amino acid transport concerns novel carriers from E. coli and C. glutamicum that mediate export across the cytoplasmic membrane.

Fig. 1.
figure 1

The cell envelope of Corynebacterianeae (right) differs substantially from the canonical cell-wall structures of gram-positive (left) and gram-negative (middle) bacteria. The models do not contain all components. In most of the gram-positive bacteria (left), the cytoplasmic membrane is covered by a porous peptidoglycan layer which does not represent a permeability barrier. Gram-negative bacteria (center) are surrounded by two membranes. The outer membrane functions as an efficient permeability barrier and contains lipopolysaccharides and porins. In Corynebacterianeae (right) the peptidoglycan is linked to the heteropolysaccharide arabinogalactan, which in turn is attached to the mycolic acids. Together with soluble mycolic acid derivatives, they form an outer lipid bilayer that in mycobacteria serves as an exceptionally efficient permeability barrier. Since there are considerable variations of this outer lipid layer within the Corynebacterianeae (Guerardel et al. 2002), the influence of this barrier on the export of metabolites in C. glutamicum is not yet well defined. The structure of the glycolipid (lipomannan) anchored in the cytoplasmic membrane is also unknown

Full size image

The l-lysine exporter

The cloning of the l-lysine export carrier gene lysE was a breakthrough in the molecular analysis of amino acid export (Vrljic et al. 1996). In contrast to the cloning of drug exporters that confer resistance, cloning of lysE from C. glutamicum required inducible lysine synthesis (Vrljic et al. 1995). The lysE-encoded polypeptide is a small membrane protein of 25.0 kDa that exhibits five transmembrane-spanning helices. By analogy with other small transporter proteins, e.g. Emr or AQP1 (Murata et al. 2000), LysE might be active as an oligomer, in this case a dimer. In addition to the transmembrane-spanning helices, a sixth hydrophobic segment is present in LysE that may dip into the membrane or be surface-localized (Vrljic et al. 1999). The energy source driving l-lysine export is the proton motive force (H+ antiport or OH− symport). The presence of this type of exporter obviously requires strict control of the export process. There are two mechanisms that serve this purpose. At the genetic level, the regulator LysG only drives transcription of the carrier gene at an elevated intracellular l-lysine concentration (≥35 mM) (Bellmann et al. 2001), and, at the protein level, the carrier exhibits a rather weak affinity for l-lysine (K m≈20 mM) (Bröer and Krämer 1991). Both features ensure that, under normal conditions, where the intracellular concentration of l-lysine is about 5 mM, no substantial excretion occurs.

In addition to l-lysine, LysE also exports l-arginine. Both amino acids are exported at a rate of 0.75 nmol min−1 (mg dry wt)−1. Therefore, LysE is in fact an exporter of basic amino acids. l-Histidine, l-citrulline, and l-ornithine are not substrates of the carrier. Interestingly, in addition to l-lysine and l-arginine, also l-histidine and l-citrulline induce lysE expression, although the latter two amino acids do not serve as a substrate for the carrier (Bellmann et al. 2001).

Why amino acid export?

In environments rich in peptides, such as complex medium, peptide uptake may result in increased intracellular l-lysine concentrations. Since C. glutamicum is not able to degrade this amino acid, it must be pumped out of the cell to reduce the intracellular level. Consequently, when a lysE deletion mutant is grown on salt medium with glucose there is no observable phenotype. However, when a low concentration of the peptide Lys-Ala is added to the same medium (3 mM is sufficient), the intracellular l-lysine concentration accumulates to more than 1 M, resulting in growth arrest (Vrljic et al. 1996; Bellmann et al. 2001). Therefore, as a general conclusion, both the regulation of the synthesis of amino acids and also their export are mechanisms for achieving homoeostasis of the intracellular amino acid concentration. It may be anticipated that many bacteria will prove to exhibit this type of intracellular amino acid control. Indeed, amino acid export has been observed, for instance, with Streptococcus faecalis (Nisbet and Payne 1982) and with Lactococcus lactis during growth on milk (Juillard et al. 1995), although the exporters responsible for this activity have not yet been identified.

The LysE family of translocators

At the time LysE was identified, only one other corresponding sequence was known. In the meantime, however, genome sequencing projects have shown that LysE proteins occur in many bacteria. At present, more than 30 functionally uncharacterized proteins with identities to LysE are known. They are found, for example, in Bacillus subtilis, Aeromonas salmonicida, Helicobacter pylori, Vibrio cholerae, and Yersinia pestis (Vrljic et al. 1999). Thus, LysE of C. glutamicum is the prototype of a new large family of translocators. Since members of a single transporter family seldom catalyze the transport of structurally divergent types of compounds (i.e. amino acids versus sugars), and, moreover, function with strongly preferential polarity of transport direction (i.e. outward versus inward) (Saier 2000), it is assumed that the majority of members of the LysE family export small positively charged molecules. In addition to l-lysine and l-arginine, these could also be structurally related substances or derivatives of basic amino acids. In fact, it has recently been shown that in the att locus of Rhodococcus fasciens, which is essential for the synthesis of an unknown compound, the LysE homologue AttX is present; this protein is associated with the virulence of the bacterium (Maes et al. 2001). Although the substance causing the virulence has not yet been identified, the attA, attB and attH genes, which are involved in its synthesis, share identities with l-arginine biosynthesis genes. This scenario makes it possible for AttX of R. fasciens to extrude an arginine derivative as the virulence-inducing factor.

The LysE superfamily

The CadD and the RhtB families are structurally very closely related to the LysE family. Together, these three families constitute the LysE superfamily of translocators (Vrljic et al. 1999). Members of the CadD family are present in gram-positive bacteria, e.g. Staphylococcus species, where they function in cadmium resistance (Chaouni et al. 1999). Another member of this family is QacF of Bacillus firmus, which possibly functions as an exporter of quaternary ammonium ions.

The RhtB family and amino acid export by E. coli

The RhtB family consists of at least 30 proteins. These proteins are present in gram-negative and gram-positive bacteria as well as in archaea (Aleshin et al. 1999). Interestingly, E. coli has as many as five paralogues, which were discovered and investigated in connection with l-threonine production using E. coli mutants (Aleshin et al. 1999; Zakataeva et al. 1999). We systematically studied all five paralogues using deletion mutants and overexpressing strains made from E. coli MG1655. The assay consisted of growth response on plates with a gradient of increasing metabolite concentrations. As shown in Table 1 and in accordance with the data of Zakataeva et al. (1999), overexpression of rhtB confers resistance to l-threonine, l-homoserine, and l-homoserine lactone, and a deletion mutant exhibits increased sensitivity to these compounds. A similar response is present with yeaS (Table 1). With yfiK and rhtC a growth response was observed only in the presence of l-threonine. However, it has recently been shown that YfiK exports l-cysteine and O-acetylserine (Franke et al. 2003). YfiK together with YdeD, a major facilitator protein, augment l-cysteine production by E. coli (Daßler et al. 2000). The specificity of the RhtB exporters and their true substrates are not known. The effects found with l-homoserine lactone led to the suggestion that they may be involved in quorum sensing (Aleshin et al. 1999). The recently completed genome sequence of Shewanella oneidensis revealed that this organism contains genes for seven RhtB paralogues and one LysE protein (Heidelberg et al. 2002).

Table 1. Growth response of ten isogenic Escherichia coli strains with altered gene expression to the presence of increasing concentrations of l-threonine, l-homoserine, and l-homoserine lactone. The response of clones with the denoted genes deleted (Δ) or overexpressed in pUC18 (pUC) is given in relation to the parental strain MG1655. The concentrations were 0–100 mM for l-threonine, 0–20 mM for l-homoserine, and 0–20 mM for l-homoserine lactone. ● Growth better than control (strain MG1655), ○ less growth, − no effect, nd not done
Full size table

Although it is probable that none of the five RhtB carriers of E. coli is the major carrier for active l-threonine export (Kruse et al. 2002), they are nevertheless relevant for the excellent properties of producer strains. Producers accumulate l-threonine concentrations in the medium ranging from more than 100 g l−1 up to the limit of solubility (Debabov 2002). However, the addition of 5 mM l-threonine to a culture of E. coli wild-type (MG1655) reduces the growth rate from 0.42 to 0.37 h−1. Thus, at the exceptionally high concentrations that producer strains have to tolerate even a small difference in the movement of l-threonine over the membrane might contribute to the stability of strains and their accumulation properties. This view is supported by the fact that a producer strain has been found to be impaired in its l-threonine uptake (Okamoto et al. 1997), and that upon rhtB or rhtC overexpression the specific productivities in producer strains were slightly increased (Kruse et al. 2002).

The ThrE family and l-threonine export by C. glutamicum

The recently identified l-threonine exporter ThrE of C. glutamicum also represents the prototype of a new translocator family (Simic et al. 2001). This is in agreement with the notion that most eukaryotic transporters belong to already recognized families, whereas prokaryotic systems frequently belong to uncharacterized families of which no functionally characterized members are known (Saier 2000). The ThrE family is relatively small, with about 20 identified homologues. Members of the family are found in Bacteria, Archaea, and the fungal kingdom. It has been suggested that the carrier is active as a ten-transmembrane-spanning transporter. An interesting phylogenetic feature of the members of the ThrE family is that they exist either as a single long polypeptide chain or as two short polypeptides (Fig. 2). Together with the weak sequence similarities between the amino-terminal and carboxy-terminal parts of the large polypeptide, this is evidence that the proteins resulted from a gene duplication event (Yen et al. 2002). A similar situation is found with carriers of the drug-metabolite-exporter family and the major-intrinsic-protein family (Murata et al. 2000).

Fig. 2.
figure 2

Depiction of the domain structure of selected proteins of the ThrE family of exporters. Arrows indicate the C-termini. ThrE of C. glutamicum is given at the top; the hydrophobic part is shown as two hatched rectangles, which represent the carrier part with 10-transmembrane helices and illustrate its origin by duplication. In Campylobacter jejeuni, two adjacent genes encode the exporter, in which the two polypeptides, each with 5-transmembrane helices, are not fused. A hydrophilic domain probably not involved in export is indicated as a black box. Mycobacterium tuberculosis has two such hydrophilic domains. For further details see text

Full size image

The ThrE exporters are characterized by another interesting structural feature. All of them show an extended hydrophilic N-terminal domain (black box in Fig. 2). This domain exhibits weak sequence similarity with portions of hydrolases (proteases, peptidases, and glycosidases). The occurrence of this domain is not likely to be accidental and must have functional significance. According to the topology prediction, it is located at the cytoplasmic side of the membrane. Thus, the hydrophilic domains of these exporters may be involved in generating the transport substrate by a hydrolytic activity present on the polypeptide. Furthermore, this domain supports the view that ThrE has an additional function and that l-threonine is not the true substrate but accepted due to a side activity of ThrE.

Nevertheless, at an intracellular l-threonine concentration of 170 mM, 59% of the l-threonine excretion in C. glutamicum is driven by ThrE activity. In addition, 22% of the total efflux is due to passive diffusion and the remaining 19% is due to at least one other as yet unidentified carrier (Simic et al. 2001). As far as biotechnology is concerned, it is interesting that increased export can be achieved in C. glutamicum by thrE overexpression, i.e. l-threonine accumulation is increased by up to 40 % upon thrE overexpression. Further increased accumulation is achieved by reduced cell-internal degradation of l-threonine. This degradation to glycine proceeds by aldol cleavage catalyzed by serine hydroxymethyl transferase (Simic et al. 2002).

The LIV-E family of translocators

As shown for C. glutamicum producer strains, l-isoleucine efflux is mediated both by diffusion and by active export. The export carrier involved has recently been identified (Kennerknecht et al. 2002). It exports l-isoleucine or l-leucine at comparable rates of about 7 nmol min−1 (mg dry wt)−1, whereas l-valine is excreted at a significantly reduced rate. The exporter is a two-component permease encoded by brnFE, and similar proteins in Bacillus subtilis have been found to be related to 4-azaleucine resistance (Belitsky et al. 1997). The BrnF polypeptide is predicted to span the membrane seven times and the polypeptide of the smaller BrnE protein four times. The genes brnF and brnE are widespread in prokaryotes, but lacking in eukaryotes, and generally map together in the same order in operons. Together, BrnF and BrnE comprise the members of the novel LIV-E family of exporters. A paralogue of this carrier is present in C. glutamicum. Some α-Proteobacteria, such as Agrobacterium tumefaciens and Sinorhizobium meliloti contain up to three paralogues. The fact that the phylogeny of LIV-E family members does not correlate to that of the organismal 16S rRNAs leads to the suggestion that these carriers do not play a role as essential "housekeeping" proteins. This suggestion is further supported by the fact that in both C. glutamicum and B. subtilis the LIV-E family carriers are nonessential.

Outlook

Novel families of export carriers have been identified in connection with bacterial amino acid excretion. The l-glutamate exporter of C. glutamicum may represent another new carrier family. However, this exporter has not yet been identified, although an average of almost 3,000 tonnes of sodium glutamate are produced daily using C. glutamicum and the process has been studied for almost 50 years. l-Glutamate excretion is a very special case, since additional treatments involving the entire cell wall, such as penicillin addition, are always necessary (Eggeling et al. 2001b). Also, the cell wall's outer lipid layer, consisting of mycolic acids, could in part limit l-glutamate efflux (Fig. 1). Mutants with a reduced mycolic acid content have altered permeability (Puech et al. 2000), and mutants defective in trehalose synthesis exhibit significantly increased l-glutamate production, from 40.2 to 45.6 g l−1 (Nakamura et al. 2002).