Energization of Outer Membrane Transport by the ExbB ExbD Molecular Motor

ABSTRACT

The outer membranes (OM) of Gram-negative bacteria contain a class of proteins (TBDTs) that require energy for the import of nutrients and to serve as receptors for phages and protein toxins. Energy is derived from the proton motif force (pmf) of the cytoplasmic membrane (CM) through the action of three proteins, namely, TonB, ExbB, and ExbD, which are located in the CM and extend into the periplasm. The leaky phenotype of exbB exbD mutants is caused by partial complementation by homologous tolQ tolR. TonB, ExbB, and ExbD are genuine components of an energy transmission system from the CM into the OM. Mutant analyses, cross-linking experiments, and most recently X-ray and cryo-EM determinations were undertaken to arrive at a model that describes the energy transfer from the CM into the OM. These results are discussed in this paper. ExbB forms a pentamer with a pore inside, in which an ExbD dimer resides. This complex harvests the energy of the pmf and transmits it to TonB. TonB interacts with the TBDT at the TonB box, which triggers a conformational change in the TBDT that releases bound nutrients and opens the pore, through which nutrients pass into the periplasm. The structurally altered TBDT also changes the interactions of its periplasmic signaling domain with anti-sigma factors, with the consequence being that the sigma factors initiate transcription.

OUTER MEMBRANE PROTEINS (OM): STRUCTURES AND FUNCTIONS

Cells communicate with their environments through membranes, in which proteins specify the permeability. Gram-negative bacteria are surrounded by two membranes: an outer membrane (OM) and an inner, cytoplasmic membrane (CM). The CM is equipped with energy-generating systems that power the active transport of substrates across the membrane with a high specificity. The OM is less selective than the CM, regarding the structure of the molecules that pass through. Molecules up to a size of approximately 600 Daltons diffuse through proteins that are designated porins (1). They consist of 16 β-strands that form a β-barrel with a pore in the center, through which substrates diffuse along a concentration gradient (1). In a second class of OM proteins, a 22-stranded β-barrel encloses a pore that is closed by an N-terminal plug that must be opened for the substrates to pass through. In a third class, an extension that is adjacent to the plug approaches the periplasm. These proteins actively transport substrates and signal the presence of substrates at the cell surface into the periplasm and further into the cytoplasm, where gene transcription is initiated (2–5). In Escherichia coli (E. coli) K-12, eight outer membrane proteins, designated TonB-dependent transporters (TBDT), consume energy while they are active as the substrate transporters and colicin or phage receptors. However, the OM contains no energy-generating system. Energy is derived from the proton motif force (pmf) of the CM. Protons are built up in the periplasmic space by ATP hydrolysis through the CM-bound ATPase or by CM-coupled redox reactions. The energy of the pmf is transferred from the CM to the substrate loaded TBDTs by three proteins: TonB, ExbB, and ExbD (Fig. 1). ExbB and ExbD harvest the energy that is transmitted to the substrate loaded TBDTs by TonB. The TBDTs and the colicins (see below) contain a conserved sequence of five to eight amino acid residues (6), called the TonB box, to which TonB binds. Commonly studied TBDTs include the Fe³⁺-ferrichrome transporter FhuA (7) and the vitamin B₁₂ transporter BtuB (8).

FIG 1.

A structural representation of the Ton complex (TonB, ExbB, and ExbD) in Gram-negative bacteria. Left: a cartoon representation of the Ton uptake system. Five monomers of ExbB are represented at the inner membrane (purple, pink, green, blue, orange). ExbB interacts with the TM domains of ExbD (black and white cylinders) that contain the essential Asp residue that is responsible for proton (black sphere with white + symbol) translocation across the inner membrane. Due to a larger concentration of protons in the periplasmic space than in the cytoplasm, the proton motive force (pmf) is used. Electrophysiological studies show that the Ton subcomplex forms pH-selective channels, thereby providing insight into the mechanism by which it may harness the pmf for energy production (68). The movement of protons is symbolized with black arrows. It is hypothesized that the ExbB ExbD subcomplex forms the proton channel and that the energy that is harnessed from the proton motive force is translocated through TonB (yellow cylinder) and across the periplasmic space to the TonB-dependent transporter (TBDT, blue cylinder). TBDTs span the outer membranes of Gram-negative bacteria. When a ligand binds on the extracellular surface of the TBDT, a conformational change occurs, thereby exposing a conserved binding domain, called the TonB box, into the periplasm (blue rectangle). The TonB box is the binding site for the C-terminal region of TonB (yellow rectangle), which opens a channel in the TBDT and allows the bound ligand to diffuse into the periplasm. Right: molecular representations of known components of the Escherichia coli Ton complex. The same color scheme as was used in the left panel is used here. Cartoons and molecular surfaces are used to depict each protein. The subunits of the ExbB pentamer forms a central pore in which a dimer of ExbD subunits are located (PDB: 6TYI) (71). For simplicity, we removed two of the ExbB monomers to better visualize the ExbD dimer within the central pore. ExbD extends into the periplasmic space, has an ordered C-terminal domain (black and white ovals), and is predicted to self-dimerize (PDB: 2PFU). Some disordered residues of ExbD in the periplasmic space have not yet been resolved, and these are represented as black and white dashed lines. Although the binding site has not yet been determined, the transmembrane domain of the TonB subunit (yellow) is predicted to bind at the periphery of ExbB. The oligomeric state and most of the structure of TonB is unknown (the transmembrane domain and the flexible periplasmic domain), but the C-terminal globular domain has been determined in complex with multiple TBDTs. Here, the C-terminal of TonB (yellow) is bound to the TonB box of FhuA (blue; PDB, 2GRX) (85). FhuA was selected as a representative TBDT. The molecular graphics were created using UCSF ChimeraX (91).

In recent years, considerable progress has been made in understanding the structure of isolated complete and truncated TonB, ExbB, and ExbD proteins. Electron-microscopy, cryo-electron microscopy (cryoEM), mass spectroscopy, and cross-linking studies resulted in reasonable models that explained the derivation of energy from the pmf and its use in fueling OM transport. In the following, we focus on ExbB and ExbD, their physiological activities, and the relationship between the structures of these proteins and their modes of action.

TONB IS INVOLVED IN ALL TBDT FUNCTIONS

Luria and Delbrück (9) observed that mutants of E. coli that are resistant to phage T1 arise spontaneously in the absence of selection. The mutations were mapped to two loci: tonA (now named fhuA) and tonB (ton is derived from T-one). tonA encodes the OM receptor protein FhuA (10), whereas TonB provides the energy that is required for the irreversible adsorption of T1 to FhuA. Energy-depleted cells are devoid of productive T1 adsorption. T1 binds reversibly to tonB mutants and to the isolated FhuA protein uncoupled from TonB (11). In contrast, T1 host range (T1h) mutants infect tonB mutants and are inactivated by purified FhuA and unenergized cells. T1h mutants behave like phage T5, which uses FhuA as its adsorption site and infects cells independent of TonB and an energy supply. T5 binding to the isolated FhuA receptor triggers the release of DNA from the phage head (7), whereas the incubation of T1 with OM does not change T1 morphology. The receptor properties of FhuA coupled to TonB differ from those of uncoupled FhuA. Ferrichrome, which is an iron carrier (siderophore) that is translocated across the OM in a TonB-dependent manner by FhuA, inhibits T5 adsorption to tonB mutants and to energy-depleted tonB wild-type cells. Ferrichrome at a 1,000-fold higher concentration only slightly inhibits T5 adsorption to energized tonB⁺ cells (12). A lack of FhuA energization is required to observe the Ton-mediated ferrichrome inhibition of wild-type cells. TonB is involved in the response of FhuA to energization that results in at least two different functional FhuA states: an energized state that supports T1 infection and an unenergized state that allows ferrichrome to inhibit T5 infection. An inactive fhuA mutant carrying a valine to aspartate replacement at position 11 was T1 sensitive only in the presence of ferrichrome. Ferrichrome apparently induces a conformation in the FhuA mutant protein to which T1 was able to bind (13). These data indicate that TBDTs undergo conformational changes that are induced by substrates and TonB.

TONB ACTIVITY DEPENDS ON EXBB EXBD ACTIVITY

Colicins are toxic proteins that kill E. coli (14). They are used to readily isolate mutants in OM functions, as colicin-resistant mutants frequently lose the ability to bind and subsequently to take up colicins. Exb was discovered as a resistance marker against colicin B, and it was used in a study to learn how specific proteins are imported with high efficiency and how these proteins kill cells. E. coli mutants that became resistant to colicin B were designated exbB, although, at the time, the number of involved genes was unknown (15). Two loci, designated exbB and exbA (now tonB), were subsequently mapped (16). A typical phenotype of colicin B-resistant mutants is the overproduction and secretion of enterochelin, which is also called enterobactin (siderophore), that solubilizes Fe³⁺ in the culture medium. The Fe³⁺enterochelin complex is imported by the FepA colicin B receptor (17) across the OM (17). Mutants lacking the receptor and mutants devoid of ExbB are deficient in Fe³⁺enterochelin uptake, which leads to an iron shortage. Iron limitation derepresses the transcription of the enterochelin synthesis genes via the inactivation of the Fur repressor (18), which results in the synthesis of a surplus of enterochelin, which is secreted.

exbB mutants were accidentally isolated during a screening for iron-regulatory mutants (19). Some of the mutants weakly overproduced OM proteins in the size range of 80 kDa, in contrast to the strong overproduction that was observed in cells that were grown under iron deficiency. The mutants were sensitive to phage T1, which excluded tonB mutations. However, they were resistant to colicin M, less sensitive to colicins B and Ia, showed a reduced transport of ferrichrome, Fe³⁺citrate, and vitamin B₁₂, and were partially resistant to albomycin, which is an antibiotic that is derived from ferrichrome (3). tonB mutants are usually completely resistant to all of these compounds, in contrast to the different levels of resistance that are exhibited by exb mutants. Indeed, the mutations were not mapped to tonB but to a locus called exbB, which apparently encoded a function that is closely associated with TonB.

THE EXBB LOCUS IS COMPOSED OF TWO GENES: EXBB AND EXBD

To identify the products encoded by the exb locus, DNA fragments of an E. coli K-12 exbB⁺ strain were cloned into a plasmid and transferred into an E. coli exbB::Tn10 insertion mutant that was unable to grow on an iron-restricted medium (20, 21). Growth-positive transformants transported ferrichrome. The expression of the plasmid DNA by an in vitro transcription-translation system and in minicells identified two proteins, one of 26 kDa in size and the other, in smaller amounts, of 17.8 kDa in size. The nucleotide sequence revealed two genes, namely, exbB and 9 bp downstream of exbD (20). Both proteins were found to be located in the cytoplasmic membrane (21, 22). exbD encodes proteins of approximately 150 residues. In contrast, the size of ExbB is variable across species of Gram-negative bacteria. ExbB starts with the N-terminus in the periplasm, and it is followed by three short transmembrane segments. ExbD is anchored once by the N-proximal region in the CM and extends into the periplasm, similar to the TonB protein (23) (Fig. 1). Of particular interest is a single charged amino acid residue at position 25 in the transmembrane segment of ExbD. The replacement of this Asp residue by Asn inactivates ExbD in all TonB-dependent reactions and is linked to proton translocation across the CM (24).

The complementation of various chromosomal exb mutants with plasmids carrying wild-type exbB, exbD, or exbBD showed that both genes are required to restore partial sensitivity to colicins B, D, M, and albomycin (24). To test ExbBD activity in colicin translocation across the OM, cells were treated with an osmotic shock to bypass the OM receptor and TonB (21). The exposure of exbB, tonB, or exbB tonB mutants to osmotic shock and colicin M resulted in 0.3% survivors, in contrast to 90% survivors in the absence of the osmotic shock. Among the wild-type cells, 0.1% survived the colicin M treatment in the absence of the osmotic shock, whereas 0.02% survived in the presence of the osmotic shock. The results of the bypass experiment support a function of ExbBD in the OM import of colicin M.

THE PARTIAL COMPLEMENTATION OF EXBBD MUTANTS BY WILD-TYPE TOLQR AND VICE VERSA

The amino acid sequence of ExbB is 26% identical and 79% similar to that of TolQ. The amino acid sequence of ExbD is 25% identical and 70% similar to that of TolR. The nucleotide sequence homology between exbB and tolQ is 51%, and that between exbD and tolR is 49% (21). The striking homologies suggest similar functions for the ExbBD and TolQR proteins. TolQ and TolR form, together with TolA (TonB equivalent, but no distinct sequence homology), an uptake system for the group A colicins A, E1, E2, E3, and K (14), as well as for the DNA of the infecting filamentous bacteriophages f1, fd, and M13. In addition, Tol is required for cell division and to maintain the integrity of the cell envelope (25, 26).

exbBD mutants are only partially resistant to various colicins and albomycin and sensitive to the phages T1 and φ80. The partial loss of function is confined to cellular activities that require ExbBD. These properties stand in contrast to the complete inactivity of tonB mutants in all TonB-dependent import systems. Partial resistance led to the view that ExbB and ExbD are “auxiliary proteins” (27). However, this interpretation was incompatible with the following findings: the simultaneous mutation of tolQ and exbBD conferred full resistance to all tested ligands (colicins B, D, and M and albomycin; phages T1 and φ80) (28, 29). Consistently, the partial sensitivity of the exbBD mutant was the result of complementation by wild-type TolQ (28, 29). Complementation with wild-type exbB exbD resensitized tolQ mutants to colicins E1 and E2. An exbB exbD tolQ triple mutant that was resistant to all ligands, including the group A colicins E1 and E2, became fully sensitive to all ligands upon complementation with either exbB exbD or tolQ tolR. The overexpression of wild-type tolQR conferred sensitivity not only to group A colicins but also to group B colicins, phage φ80, and albomycin. Nevertheless, these results were questioned by data that were obtained from phage φ80 adsorption experiments (27). When phage suspensions were incubated with wild-type E. coli or exbB or tolQ mutants, the phage titers decreased to the same levels as were observed in the wild-type and in the tolQ mutant, albeit only slightly in the exbB mutant, and they remained constant in the exbB tolQ double mutant. As the lack of tolQ affected phage adsorption only marginally but added to exbB-induced resistance, it was concluded that TolQ cannot replace ExbB in TonB-dependent φ80 infections. However, a spot titration assay clearly revealed the contribution of TolQ to ExbBD activity (28, 29). The stock solutions of toxin that were used in this assay had a high activity such that 10³-fold to 10⁴-fold diluted samples still produced phage plaques and colicin lysis zones. A 10-fold dilution series of phage and colicin suspensions were deposited on nutrient agar plates that were seeded with the strain to be tested, after which phage plaques and colicin lysis zones were counted. Plaques formed after manyfold repetitions of whole cycles of phage proliferation. Colicins and albomycin must enter cells to kill them. Phage adsorption assays differ, as they measure single events against a relatively high background of unspecific phage binding that does not lead to infection. In contrast, the killing assays are sensitive and detect low levels of activity. Double exbB tolQ mutants of E. coli were completely resistant to the bacteriophages T1 and φ80, in contrast to strains with single exbB or tolQ mutations, which were sensitive. Cells carrying mutations in exbB were partially tolerant to colicins B, D, and M, and they became fully tolerant via the introduction of tolQ mutations. This suggested the involvement of both exbB and tolQ in tonB-dependent uptake. The results were supported by studies with phage fd, which infects cells containing an active Tol system. fd does not infect an exbBD tolQR quadruple mutant. Complementation with wild-type exbB exbD fully replaced tolQ tolR and restored phage uptake to the wild-type level (30). The mutual functional replacement of ExbBD and TolQR explains the partial resistance of exbBD mutants through cross-reactivity with TolQR and vice versa. Additional evidence for a functional role of TolQ in TonB-dependent transport came from experiments in which cobalamin transport rates were determined (31). The transport of cobalamin into the periplasm is powered by the pmf and requires TonB. The transport rate into an exbB mutant is approximately 20% of that of an exbB⁺ wild-type strain, 65% of that of a tolQ mutant, and <5% of that of a tonB mutant or an exbB tolQ double mutant. The transport of the exbB mutant is complemented by wild-type TolQ.

Cross-complementation between the energy-driven Ton and Tol import systems might play a relevant physiological role if one system becomes defective, in which case the complementary system maintains the cell survival for some time. For phage infection, the activity of the complementary system is sufficient to support phage proliferation. Homologous sequences and similar functional mechanisms suggest a common origin for the Ton and Tol systems. Their partial cross-reactivity is a remnant of the original system. A comparison of the respective structures at an atomic resolution may allow for the tracing of the specification pathways to Ton and Tol during evolution.

FUNCTIONS OF EXBB AND EXBD

Once it became clear that ExbB and ExbD were not auxiliary proteins but were essential for the energization of OM import, numerous studies attempted to uncover their functions. ExbB and ExbD are required for the energy-consuming OM transport, as they generate the power that is subsequently transmitted to TonB via the transfer of protons from the periplasm to the cytoplasm. Compared to the sophisticated modern automatic transmissions in cars, ExbB and ExbD are considered to act like a motor. Of the radiolabeled ferrichrome that was taken up into an exbB mutant, 70% was chased out by unlabeled ferrichrome, compared to a tonB mutant, in which the released ferrichrome amounts to 95% (32). No ferrichrome was released from wild-type cells. Ferrichrome bound to FhuA or taken up into the periplasm of the exbB mutant was chased out of the cells, whereas ferrichrome that had entered the cytoplasm of the wild-type cells did not escape. The three proteins were shown to interact with each other. Specifically, ExbB bound to a Ni-NTA agarose column via a C-terminal (His)₆ retained radiolabeled ExbD and TonB on the column, which were coeluted with ExbB (24). Of note, the ExbB(His)₆ protein that was used in this study was active, as it restored sensitivity to colicins B and M and growth on iron siderophores to the wild-type level in an exbB mutant (24). ExbD was degraded in the absence of ExbB (24). ExbB physically stabilized ExbD and TonB (32). ExbB and ExbD did not prevent the degradation of TonB by trypsin and proteinase K in spheroplasts, but ExbB inhibited the degradation of ExbD (32). Furthermore, the 90 min chemical half-life of chromosomally encoded TonB was reduced to 5 min in an exbB mutant, and the TonB activity was reduced 18-fold (27).

To examine the formation of oligomers in cells of E. coli, the size and shape of ExbB and ExbD solubilized in decyl maltoside were analyzed via blue native electrophoresis size exclusion chromatography, small-angle X-ray scattering, and transmission electron microscopy. ExbB adopted stable homooligomers with four to six monomers. Mass spectrometry revealed stable ExbB6 ExbD1 and ExbB5 ExbD1 oligomers as well as some copies of smaller oligomers. ExbB seemed to form a defined core around which ExbD and TonB were assembled (33, 34). The crystal structures that are discussed later demonstrate the ExbB5 ExbD2 stoichiometry of the ExbB ExbD complex.

Kathleen Postle and her group systematically investigated the interaction of the TonB, ExbB, and ExbD proteins via in vivo cross-linking with formaldehyde and photo-cross-linking through residue-specific sequential bPpa substitutions in energized and unenergized wild-type cells, mutants, and spheroplasts. The formaldehyde cross-linking revealed the pmf-dependent binding of the periplasmic regions of ExbD and TonB. Inactive ExbD(D25N) and TonB(H20A) mutant proteins in the transmembrane domains failed to cross-link. The pmf dependent cross-linking and the failure of the mutants to interact indicated the functional relevance of the identified interaction regions (35–40). Three stages of interactions between the ExbB ExbD and TonB proteins were proposed (36). The highly dynamic, disordered ExbD domain that is immediately distal to the transmembrane domain was considered to be of particular importance in this process (36). This domain contains a conserved motif V45, V47, L49, and P50, which is required for signal transduction to TonB and to the C-terminal end of ExbD. Small deletions in this motif inactivated TonB-dependent reactions, as TonB no longer responded to changes in the pmf (36). pBpa substitutions at the ExbD codons V45 and V47, as well as the alanine replacements of V45, V47, L49, and P50, eliminated ExbD activity. pBpa substitutions reacted with five complexes, namely, an ExbD homodimer, two ExbD-ExbB complexes, and two ExbD-TonB complexes. The V45A and V47A double substitutions also resulted in photo-cross-links of ExbD to approximately eight unknown proteins (35). During the TonB energization cycle, residues within the disordered domain reacted with multiple proteins of undefined function, which poses the question of how specificity in the energy transmission is established. In contrast to ExbD substitutions, replacements in individual transmembrane domains of ExbB identified residues that were critical for signal transduction, none of which participated in the proton pathway. Taken together, these studies suggested that ExbB serves as a scaffolding protein for ExbD and TonB assembly and mediates ExbB-TonB interactions (35).

Some cross-links do not fit crystal structures. Cross-links identify the nearest neighbors of proteins as they occur in cells. They also demonstrate structural changes within proteins and between proteins as they might occur during reactions or steps in transport cycles. In contrast, crystal and cryoEM structures represent one form that is probably among the most stable. Different structures obtained from various growth and crystallization/freezing conditions may show various stages of a protein’s activity. However, the results obtained via cross-linking and crystal structure determination should be compatible. Although we see some compatibility between the experiments, there are some major differences when analyzing the energy-harnessing ExbB ExbD TonB system. For example, in the cross-linking experiments (34–39), ExbB forms disulfide-linked homodimers between D211C, but, in the crystal and cryo-EM structures, the D211 residues are too far apart to form a disulfide bond. However, the D211C mutation in E. coli does break an important hydrogen bond that likely stabilizes the complex. This hydrogen bond appears to be conserved across Gram-negative bacteria, and it is seen in the structures of E. coli and S. marcescens ExbB ExbD. The broken hydrogen bond likely disrupts the structure of the complex and explains why this mutation inhibits normal activity levels. Additionally, in the cross-linking experiments, the transmembrane domains of TonB and ExbD interact in vivo. However, according to the crystal and cryo-EM structures, this is not energetically favorable. It should be emphasized that the 5:2 ExbB:ExbD ratio is now established (see the next section) in vivo and in vitro, and it represents at least one stable conformation of this complex.

Taken together, the data support a model in which ExbB and ExbD harness the pmf of the CM and convert TonB into an energized form that binds to the TonB boxes of TBDTs, resulting in structural changes of the TBDTs. As a result, the substrates dissociate from their initial binding sites and are transported through the opened pore and into the periplasm. More data are required, particularly regarding the localization of the cross-links in high resolution structures of ExbB, ExbD, and TonB complexes, to arrive at a model that illustrates, in molecular terms, the response of ExbB, ExbD, TonB, and TBDTs to the pmf.

TRANSCRIPTION CONTROL BY CELL SURFACE SIGNALING (ECF) REQUIRES TONB, EXBB, AND EXBD

The TonB ExbB ExbD protein complex not only powers active transport across the OM but also is involved in the regulation of gene transcription from the cell surface into the cytoplasm (40, 41). The transcription regulatory signal is initiated at the cell surface and then transmitted across the OM, periplasm, and cytoplasmic membrane into the cytoplasm, where specific sigma factors are activated to direct the RNA polymerase to selected promoters. This has, at first and most clearly, been shown in the iron citrate transport system of E. coli K-12 (41). Citrate forms, with iron, a (Fe³⁺citrate)₂ complex, for which E. coli expresses a transport system that consists of the FecA OM transport protein, the periplasmic FecB protein, and the FecCDE ABC transporter in the cytoplasmic membrane. FecA is a typical TBDT (2, 4) that consists of a β-barrel with 22 antiparallel β-strands (residues 223 to 741 of the mature protein), a globular plug (residues 81 to 222) that fills the lumen of the barrel, a TonB box located at the end of the plug domain (residues 80 to 84), and an N-terminal extension (residues 1 to 79) designated the signaling domain. The binding of (Fe³⁺citrate)₂ induces large structural changes in FecA that extend from the cell surface to the periplasm. Extracellular loops, in particular loops 7 and 8, clamp down over (Fe³⁺citrate)₂ as it binds to FecA. Furthermore, the TonB box becomes highly mobile, thereby enabling TonB to bind FecA in response to the ExbBD-mediated pmf. It is assumed that this interaction opens the plug that (Fe³⁺citrate)₂ can pass through into the periplasm. FecA, TonB, ExbB, ExbD, and the pmf are required for the initiation of transcription. However, the import of (Fe³⁺citrate)₂ into the periplasm is not required to start transcription. Two additional proteins, namely, the FecR anti-σ factor, which, in fact, is a pro-σ factor, and the FecI σ factor, display fec gene regulation. The fecR and fecI genes are located in tandem 5′ to the fecABCDE transport genes. Rather, (Fe³⁺citrate)₂ binds with high affinity and specificity to FecA, which triggers the binding of the periplasmic FecA signaling domain to the C-terminal part of FecR that triggers the proteolysis of FecR via the periplasmic Prc protease and, subsequently, via the RseP protease in the cytoplasmic membrane (41–44). RseP releases the N-terminal FecR fragment into the cytoplasm, of which residues 1 to 59 bind to the FecI σ factor, which induces fecA transcription (44). In most ECF-type transcription regulatory devices, the anti-σ factor is completely degraded, and the released free sigma factor binds to the RNA polymerase and induces transcription. This type of transcription regulation applies to many TBDT-mediated import systems (41) as well as to many pathways that alter the bacterial envelope to adapt to changing environments, such as osmolarity, pressure, temperature, and iron availability (45).

EXBB EXBD GENES IN THE GENOMES OF BACTERIA

Most papers that deal with outer membrane transport analyze the TBDT proteins and TonB, but they rarely analyze ExbBD. ExbBD was mostly identified as a factor that was required for the growth of pathogenic bacteria in natural, usually iron-poor, conditions. In many genomes of Gram-negative bacteria, the exbB and exbD genes are located close to each other and to the tonB gene. However, E. coli K-12 already encodes exbB exbD, separately from tonB. There are only a few organisms in which the biochemical activities of the TonB, ExbB, and ExbD proteins were studied.

Serratia marcescens.

An exceptional example is S.marcescens (Sm), which was studied via genetic, physiological, biochemical, biophysical, and bioinformatic methods (46–49). Sm encodes tonB and a tonB paralog, designated hasB, that is part of the heme import system (46). ExbBD_Sm was functional with HasB and TonB_Ec but growth with ExbBD_Sm HasB started after 3 to 4 h, in contrast to ExbBD_Sm TonB_E.c, which started after 20 h. HasB did not interact with ExbBD_Ec (49). In S. marcescens, TonB and HasB are present, and the interaction of ExbBD with TonB or HasB may adapt cells to various iron supply conditions. Compared to ExbB_Ec the ExbB_Sm contains an N-terminal periplasmic extension of 81 residues, of which presumably 43 residues function as a signal sequence that is cleaved off during export. Cells expressing mutant ExbB devoid of the N-terminal extension started growth 2 h later than did cells expressing wild-type ExbB. NMR measurements revealed that the N-terminal extension of ExbB_Sm interacts with the C-terminal globular domain of HasB. Although not essential, the N-terminal extension does play a role in heme acquisition via the Has system. ExbB proteins of different length are frequently found, but no unique function has been assigned to them.

Vibrio.

The first example of more than one set of tonB exbB exbD genes in a strain was described in Vibrio cholerae (50). The transcription of tonB1 exbB1 exbD1 and tonB2 exbB2 exbD2 was coregulated by iron via the Fur repressor. tonB1 exbB1 exbD1 are part of an operon that encodes three heme transport genes. Chromosomal mutants in exbB1 or exbB2 showed no defect in iron uptake via heme, vibriobactin, or ferrichrome, but exbD1 exbB2 double mutants were transport deficient.

Vibrio vulnificus contains three tandemly arranged tonB exbB exbD genes, of which two loci encode a fourth protein that is designated TtpC (51). One of them, namely, TtpC2, together with TonB2, is required for the uptake of various Fe³⁺ siderophores.

Methylococcus trichosporium OB3b.

M. trichosporium OB3b is a methanotroph that oxidizes methane. The particulate monooxygenase requires copper, which is taken up bound to methanobactin, which is a small, ribosomally synthesized polypeptide (52). The transporter gene that encodes a TBDT is located downstream of two genes that encode proteins with sequence similarity to the E. coli fecR and fecI genes that regulate the transcription of ferric citrate transport genes. They suggest a transcription regulation by an extracytoplasmic function sigma factor (ECF). Copper uptake via methanobactin resembles heme uptake via hemophores. However, an ECF-type regulation and a TonB-dependent import of methanobactin loaded with copper still have to be demonstrated.

Acinetobacter baumannii.

The genome of A. baumannii encodes 3 tonB genes, of which tonB1 is located upstream of exbB1 exbD1.1 exbD1.2. tonB2 is monocistronic, and tonB3 is located 3′ of exbB3 exbD3. An extra exbD (exbD1.2) is frequently encountered, but its roles in energy collection and energy transmission have not yet been studied. Growth on an iron-poor medium depends on the tonB3 operon. TonB3 is essential for A. baumannii virulence in insect and mammalian infection models (53).

Xanthomonas campestris pv. Campestris.

The genome of X. campestris pv. Campestris encodes a tonB exbB exbD1 exbD2 cluster (54). tonB exbB exbD1 is required for the induction of a hypersensitive response (HR) on the nonhost plant pepper (Capsicum annuum) and for the induction of typical black rot symptoms on the host plant cauliflower (Brassica oleracea). exbD2 plays a role in HR but not in iron supply. In fact, the low iron concentration in the pepper leaves restricted the growth of the tonB exbB exbD1 mutant, whereas the exbD2 mutant grew even better than did the wild-type in pepper leaf tissue. Substrates other than the known iron complexes should be tested to identify TBDTs that are powered by ExbD2.

Sphingomonas.

Sphingomonas was defined in 1990 as a group of Gram-negative, rod-shaped, chemoheterotrophic, aerobic bacteria. They typically contain glycosphingolipids that protect cells from antibacterial substances. The genus Sphingomonas includes more than 20 species, and these species are quite diverse in terms of their phylogenetic, ecological, and physiological properties. They are widespread in nature, occur in highly polluted environments, and use rare, recalcitrant, and toxic compounds. Accordingly, they metabolize a wide variety of carbon sources of low concentration, such as oligosaccharides and polysaccharides, oligopeptides and polypeptides, polyethylene glycol, phenanthrene, lignin-derived aromatic compounds, sulfanilic acid, hydrophobic polycyclic hydrocarbons, dibenzofuran, and dibenzo-p-dioxin. Sphingobium sp. strain SYK-6 served as a model organism with which to study the uptake of such compounds (55, 56). It encodes 74 TBDTs, of which 12 were specifically expressed by growth on lignin-derived aromatic compounds. Of the six tonB-like genes, only tonB1 in the gene cluster tonB1 exbB1 exbD1 exbD2 was required for the uptake of the lignin derivatives. An aspartate residue corresponding to Asp25 of the E coli ExbD was present in ExbD1, ExbD2, and ExbD3/TolR, which supports its essential role in signal transfer to TonB. The TBDTs probably transport the lignin derivatives into the periplasm at the expense of the pmf, which is transmitted to the TBDTs by the TonB1 ExbB1 ExbD1 protein complex.

BACTEROIDETES

Bacteroidetes represent one of the dominant phyla of bacteria and comprise the largest number of Gram-negative organisms in the gut (57–61). They degrade dietary glycans that cannot be metabolized by the host. A vast repertoire of enzymes serves to breakdown carbohydrates. Glycoside hydrolases, glycan binding proteins, and TonB-dependent transporters work together in the outer membrane and are encoded in discrete gene clusters that are termed polysaccharide utilization loci (PUL). Some PULs are highly specific for distinct glycan substructures, whereas others can target a range of structures. The high percentage of PUL genes, 18% of the genome, indicates the outstanding importance of a large variety of polysaccharides for the nutrition of Bacteroides. The starch utilization system has been studied quite early and was named Sus (57). Large glycans are degraded outside cells to a size such that they can be imported by TBDTs and then further degraded inside the cells. The envelopment of the entire ligand by a surface-located, endo-acting enzyme precludes the import of undigested, high molecular weight glycans. The transport of oligosaccharides across the outer membrane requires the activity of two proteins: a TBDT (SusC) and a lipoprotein (SusD). Bacteriodes spp encode over 100 SusC-SusD pairs. Not only sugars but also other substrates, such as iron compounds and vitamins, are taken up by TBDTs. Unlike other TBDTs, SusC forms homodimers, of which each barrel is capped by SusD. SusC transporters contain TonB boxes that are similar to each other and to those of other bacteria. Ligands are bound at the SusCD interface in a large, solvent-excluded cavity. In the absence of a substrate, the SusD lid of the empty transporter is mobile and assumes a number of conformational states. Upon substrate binding, the closed state is stabilized by ligand interactions both with SusC and SusD (59). It is assumed that the binding of TonB to the TonB box of SusC, a process that is energized by the pmf through ExbB ExbD, generates the conformational changes in SusC that are required for the plug to move and open a pore, through which the substrate enters the periplasm. The TonB-induced substrate dissociation reverts the transporter back to the dynamic open state.

Bacteroides fragilis.

B. fragilis is an opportunistic, anaerobic pathogen and a commensal of the human large intestinal tract. The genome of strain 638R encodes six predicted TonB proteins (TonB1 to 6), four ExbB orthologs (ExbB1 to 4), and five ExbD orthologs (ExbD1 to 5). Of the 104 predicted TBDTs, 33 are composed of a 22-strand β-barrel with a plug typical for the iron siderophore transporters, and 71 belong to the SusC-like nutrient transporters (62). Since there are many more transport proteins than potential energy transmission proteins, more than one transporter must be coupled to a TonB ExbB ExbD complex. In fact, only one of the six TonB proteins, namely, TonB3, was found to be involved in the import of any of the tested substrates, heme, vitamin B₁₂, ferrichrome, starch, mucin-glycans, and N-linked glycans (62). In a defined medium, only the growth of the tonB3 mutant was completely abolished and restored by complementation with wild-type tonB3. In a mouse model of intestinal colonization, only the tonB3 mutant was rapidly outcompeted by the surviving parent strain. After 6 days of colonization, the tonB3 mutant showed a 4 log-fold decrease, compared to the parent. Various combinations of the tonB1 to tonB6 mutants revealed a synergistic growth effect of tonB6 and tonB3 in the intestinal tract, which was not observed in vitro. The studies did not include ExbB1 to 4 or ExbD1 to 5. The specificity of TonB3 suggests the selection of only a single type of ExbB and ExbD to form an active transport system. Since none of the exbB and exbD genes map close to tonB3, it is not obvious which one of the ExbB and ExbD proteins forms an active complex with TonB3 (62). Distinct TBDTs will act in concert with certain TonB ExbB ExbD complexes. One must also take into consideration that B. fragilis grows under anaerobic conditions, which may result in a different pmf, compared to the aerobic growth to which TonB ExbB ExbD complexes may be adapted. The TonB proteins differ substantially. For example, the N-terminus of TonB5 is much longer (200 to 300 residues) than the N termini of TonB1 to 4. It folds into 4 transmembrane domains in contrast to the single helix of TonB1 to 4. Detailed physiological experiments are required to assign substrates to TBDTs and to assign TBDTs to TonB ExbB ExbD complexes.

Bacteroides thetaiotaomicron.

B. thetaiotaomicron is numerically one of the most abundant species in the lower gastrointestinal tract. The PUL of B. thetaiotaomicron 2 encodes eight proteins, susRABCDEFG, for the adherence of starch to the cell, the hydrolysis of starch by a surface exposed amylase, and the transport of maltooligosaccharides into the periplasm via a TBDT (60, 61). SusC serves as a transporter that receives the oligosaccharides from SusD. At least 120 paralogs of SusC are encoded in the genome of B. thetaiotaomicron. In contrast, only 17 SusD paralogs are predicted. The structure of the TBDTs differs from the TBDTs that transport iron compounds. They form dimers and contain N-terminal extensions of 200 to 300 amino acids that interact with SusD at the extracellular face of the β-barrel. SusD caps the barrel of SusC, thereby forming a large interface of approximately 4000 Ų. In the absence of a glycan ligand, SusD undergoes a hinge-like motion to expose the binding site to the external environment (59). The ligand binds to SusC and to the cap of SusD, thereby resulting in a stable conformation.

A genome analysis of B. thetaiotaomicron DSM2079 predicts five exbB, seven exbD, and three tonB-like genes. The KEGGH database contains 10 paralogs of TonB, including proteins that are much longer (up to 600 residues) and more complex (4 transmembrane segments) than, for example, the E. coli TonB with 263 residues and a single transmembrane segment. The exbB and exbD genes are frequently located in tandem. The number of TBDTs that transport the degradation products of complex glycans by far exceeds the number of possible TonB ExbB ExbD energy transmission complexes, provided that TonB1 interacts only with ExbB1 ExbD1, TonB2 with ExbB2 ExbD2, and so on. The few data in the literature suggest that, usually, no mixed complexes are formed between the TonB, ExbB, and ExbD of different operons (see, for example, the Sphingomonas and Xanthomonas sections), but this would have to be shown for B. thetaiotaomicron. Compared with the few other systems in which the activities of multiple tonB exbB exbD energy transmission proteins have been measured, one may predict that a single TonB ExbB ExbD transmission system is used for most transport activities.

TONB, EXBB, AND EXBD ARE INVOLVED IN THE TRANSCRIPTION REGULATION OF PULS GENES

The transcription of genes that are located in PULs are regulated by various mechanisms (63). A prominent way to adapt the expression of proteins in PULs to the available glycans involves transcription regulation via extracytoplasmic function σ factors (ECF) and anti-σ factors. ECF regulation through TBDTs implies the function of TonB, ExbB, ExbD, and the pmf, although this has not been shown experimentally for PULs. At least 12 of the 26 putative ECF-type regulatory systems that catabolize mucin O-glycans were identified in B. thetaiotaomicron (63). Cells survey their environments for the various glycans and generate a rapid response. For the ECF-type transcription, regulation substrates do not enter the cells but act from the cell surface. Minor cross talk occurred between simultaneously active ECF systems, suggesting species adaptation to the life in the gut via the coregulation of nutrient intake and surface antigenicity. In a few selected systems, specific interactions between the periplasmic signaling domains of SusC-like transporters and anti-σ factors as well as between anti-σ factors and ECF σ factors were demonstrated. The deletion of the C-proximal region of an anti-σ factor resulted in the constitutive expression of Pul genes in the absence of inducing glycans. This property resembles the Fec system, in which the N-terminal FecR fragment stimulates the FecI sigma factor in the absence of the inducing (Fe³⁺citrate)₂. This type of regulation was designated “trans-envelope signaling” (63), a term which was adopted from “cell surface signaling” of the Fec type transcription regulation (41).

EXBB AND EXBD ARE INVOLVED IN PROTEIN SECRETION

Myxococcus xanthus initiates a multicellular developmental program in response to starvation. Starving cells secrete the protease PopC, and this secretion depends on the TBDT Oar and on the proteins that are encoded by exbB1 and exbD1, which are adjacent to the oar gene (64). Secretion requires the pmf, as the dissipation of the pmf reduced PopC in the supernatant. Mutants in tonB1 were blocked in PopC secretion. Mutants in exbB1, exbD1, and exbD2 displayed an aberrant development but still secreted reduced amounts of PopC. Since one of the additional Ton systems in M. xanthus could have complemented the function of TonB1 ExbB ExbD1 ExbD2, Oar-dependent PopC secretion was studied in wild-type E. coli expressing single ExbBD proteins. When coexpressed with Oar, PopC was exclusively detected in the supernatant of the E. coli cell culture. Obviously, the E. coli Ton system energizes Oar to secrete PopC. PopC secretion also demonstrates that a substrate as large as 50.8 kDa can pass through the pores of TBDTs.

Sphingobium fuliginis produces an organophosphate hydrolase (OPH) to degrade organophosphate insecticides and nerve agents. OPH is targeted to the CM of E. coli when it is coexpressed with either ExbD or ExbB ExbD (65). In the absence of ExbD, OPH remains in the cytoplasm. The precursor of OPH contains a TAT motif that was not sufficient to transfer OPH to the CM. The coexpression of ExbB and ExbD was required, probably because both proteins stabilize each other. OPH that was detergent-solubilized from a membrane fraction was copurified on an OPH specific immune affinity column together with TonB, ExbB, and ExbD. A pulldown and two hybrid assays confirmed the interaction of OPH with TonB and ExbD. Since no TBDT was involved, it is likely that the pmf, through the action of ExbD, activates an unknown protein that transfers OPH from the cytoplasm and inserts it into the CM.

ANCESTRAL EXBB EXBD ACTIVITIES

Ancestral cyanobacteria are believed to have predated by a billion of years the Great Oxygenation Event, during which they produced the oxygen on Earth. Unexpectedly, most isolated ancestral cyanobacteria do not encode FeoB, which is the transporter of Fe²⁺ (18) and must have been the dominant iron oxidation state in the anoxic sea. Rather, Pseudoanabena sp. PCC 7367 and all of their basal homologs that were isolated from marine habitats encode ExbBD and various TBDTs, but only a few species encode TonB (66). A phylogenetic study on the relationships of ExbBD and MotAB, using Bayesian inference and the maximum likelihood method, came to the conclusion that ExbBD and MotAB originated from a common system and have separated into distinct monophyletic groups (67). ExbBD proteins in Firmicutes are the deepest branching lineage. Among Archaebacteria, only Euryarchaeta encode ExbBD.

STRUCTURES OF THE EXBB EXBD TONB COMPLEX

A decade ago, the stoichiometry and structure of ExbB and ExbD were unknown. The first high-resolution X-ray structure of the E. coli ExbB ExbD subcomplex (3.5 Å) showed ExbB as a pentameric structure with a single ExbD inside the hydrophobic central pore of ExbB (68). Not long after, Maki-Yonekura et al. (69) used X-ray crystallography and single particle cryoEM to show that hexameric and pentameric complexes of E. coli ExbB coexist, with the proportion of hexamer increasing with the pH (69). The hexameric complex consisted of six ExbB subunits and three ExbD subunits, whereas the pentameric complex consisted of five ExbB subunits and one ExbD subunit. However, these complexes were not resolved to high resolution, 6.7 Å and 7.1 Å, respectively, making it difficult to identify regions of density. The functional stoichiometry of the ExbB ExbD subcomplex has been highly disputed, well before the first two high resolution structures were published. Postle et al. calculated the ratio of ExbB:ExbD:TonB to be 7:2:1, which has not been confirmed by other in vivo or in vitro studies (70). In contrast, recent publications strongly support the ExbB ExbD subcomplex at a 5:2 ratio (Fig. 2).

FIG 2.

ExbB ExbD and homologous motor proteins oligomerize at a 5:2 ratio. A structural comparison of Escherichia coli ExbB/ExbD (PDB, 6TYI), Campylobacter jejuni MotA/MotB (PDB, 6YKM) and Escherichia coli TolQ/TolR from (A) a top-down view (from the periplasm), (B) a side-view, and (C) a side-view with the removal of two ExbB/MotA/TolQ subunits to allow for the better visualization of ExbD/MotB/TolR (71, 81, 92). The all colors are consistent with those used in Fig. 1. The ExbB/MotA/TolQ subunits are purple, pink, green, blue, and orange. The pentameric subunits form a central pore in which a dimer of ExbD/MotB/TolR subunits is shown in black and white. All molecular graphics were created using UCSF ChimeraX (91). Here, Campylobacter jejuni MotA/MotB has been depicted; however, similar structures have been determined but are not shown (Vibrio mimicus MotA MotB [PDB, 6YSL] and Bacillus subtilis [PDB, 6YSL]) (73). No complete structure of the TolQ-TolR complex exists; however, the Tol complex was modeled upon the 5:2 structure of ExbB-ExbD (PDB, 6TYI) (71, 92). This model was generated using the SWISS-MODEL Workspace (https://swissmodel.expasy.org/).

First, Celia et al. used cryoEM in lipid nanodiscs to determine a 3.3 Å structure of E. coli ExbB forming a pentameric, hydrophobic central pore surrounding a dimer of ExbD (71). This pentameric stoichiometry was supported by mass spectrometry experiments that were performed on E. coli native membranes (72). Although Celia et al. (68) found an X-ray crystallography structure of 5:1 (discussed above) in 2016, the stoichiometry was likely due to a major truncation in ExbD and/or crystal lattice packing that forced this oligomeric state. Since 2019, several other ExbB ExbD subcomplexes from various Gram-negative bacterial species were determined to have a 5:2 ratio of ExbB ExbD, including P. savastanoi and S. marcescens (47, 71, 73).

In these structures, only the transmembrane domains of ExbD are resolved. A large portion of the periplasmic domain of ExbD is disordered and dynamic. Thus, the complete density has not been determined. However, Garcia-Herrero et al. determined the NMR E. coli structure in a solution of the periplasmic N-terminal domain of ExbD (74). The folded C-terminal region is comprised of two α-helices that are located on one side of five β-strands with a mixed parallel and antiparallel arrangement. Structurally, ExbD closely resembles the C-terminal lobe of the siderophore-binding proteins FhuD and CeuE (74).

Several models for the mechanism of the Ton complex have been proposed. Although the hypothesized mechanisms vary, they all propose the association of the TonB C-terminal domain with the TBDT TonB box and the application of force to alter the conformation of the TBDT plug domain. The displacement of the TBDT plug domain by TonB is proposed to occur through a rotation, pulling, or hybrid wrap and pull mechanism (75). Although several hypotheses have been proposed, there is little evidence as to how the rotation or conformational changes of ExbB and ExbD effect TonB and, subsequently, TBDTs. Multiple homologous motor protein complexes share a 5:2 ratio with ExbB ExbD, including the recently published cryoEM structures of the Mot complex (MotA MotB). MotA and MotB form a stator complex that uses the proton motor force to generate a flagellar torque. MotA has 22% identity to ExbB, and MotB has 16% identity to ExbD. For decades, the Mot complex was thought to be a 4:2 complex (76, 77). Many X-ray crystallography structures confirmed MotB as a dimer, and it was predicted that MotA was at a 2:1 ratio with MotB (76, 78–80). Yet, cryoEM structures solved the Mot complex from C. jejuni, Bacillus subtilis, and Clostridium sporogenes demonstrated a 5:2 stoichiometry ratio of MotA and MotB, in addition to them having a shared subunit arrangement (73, 81). However, there are some noticeable differences between the MotAB subcomplex and the ExbBD subcomplex. MotAB was shown to have a slightly larger extracellular domain, and ExbD is shorter than MotB and other homologous proteins (Fig. 2).

Based on their structures, Santiveri et al. (81) proposed a mechanism for proton translocation across the inner membrane in the MotAB system. We propose that ExbB and ExbD undergo a similar mechanism for proton translocation (Fig. 3). Like MotB, ExbD contains a charged amino acid, namely, D25, in the otherwise hydrophobic transmembrane domain (21), which may respond to the pmf. Indeed, the replacement of D25 by N25 inactivated ExbD (21).

FIG 3.

Predicted mechanistic model of ExbB around ExbD adapted from the Santiveri et al. model of MotA and MotB. For simplicity, only TM2 and TM3 are shown in a top-down view for each monomer of ExbB, based on the protein complex in E. coli (depicted in purple, pink, green, blue, and orange). (A) A proton from the periplasm (black sphere with a + symbol) interacts with the D25 of ExbD chain Y (black). ExbD chain Z (white) is already interacting with a proton bound to ExbB chain D (blue). (B) ExbB rotates 36° clockwise by a power stroke involving ExbD chain Y D25. (C) The rotation causes a conformation change to the ExbB pentamer, thereby allowing for the D25 of ExbD chain Z (white) to be accessible to the cytoplasm. (D) ExbD chain Z D25 releases its proton, which exits the protein complex into the cytoplasm. (E) After the deprotonation of ExbD chain Z, a new proton binds from the periplasm. The cycle repeats with ExbD chain Y and ExbB chain A (purple). A total of 10 cycles must occur for the ExbB ExbD complex to return to its starting position.

First, a proton (or hydronium) ion enters the central pore of the ExbB pentamer from the periplasm and interacts with ExbD (ExbD chain Y, white), with its free aspartic acid residue D25. The binding and subsequent release by ExbD D25 is hypothesized to rotate ExbB in a clockwise manner (relative to ExbD) by 36°. The rotation of ExbB allows for the D25 of ExbD chain Z (black) to be exposed to the cytoplasm and release its bound proton. The cycle can continue with a new proton binding to the D25 of ExbD chain Z. Throughout the cycle proposed by Santiveri et al., MotB is predicted to be stably anchored to the peptidoglycan as MotA rotates around the dimer, which ultimately turns the complete rotor in a clockwise or counterclockwise direction to propel the flagellum. Deme et al. (73) proposed a somewhat similar model, but more data are needed to support these mechanisms of action. The direction of rotation has not been established.

As previously discussed, other homologous motor complexes to ExbB ExbD exist, including TolQ TolR and PomA PomB. These complexes share the highest level of conservation in the transmembrane helices that form the central hydrophobic pore. A complete structure of the Tol complex has yet to be determined, but the C-terminal domain of TolR has been determined via NMR, small X-ray light scattering (SAXS), and X-ray crystallography (82). Haemophilus influenzae and E. coli TolR were both found to be dimers via multi-angle light scattering (MALS) and analytical ultracentrifugation (AUC), respectively (82, 83). Homology models of TolQ-TolR have been created, based on the 5:2 subunit stoichiometry of ExbB ExbD (PDB: 6TYI), which appears promising (Fig. 2) (84). The TolQ pentameric model forms a large cytoplasmic chamber that is similar to that of ExbB but has some differences in electrostatics within the chamber.

In 2020, two low resolution cryoEM maps were also obtained for the Vibrio alginolyticus and Vibrio mimicus PomA PomB (73, 81). Although the structures have limited densities and no atomic coordinates, they suggest a 5:2 ratio of PomA to PomB, as well. For all of the determined structures of ExbBD, PomAB, and MotAB, the periplasmic domains of ExbD and MotB were not visible because of their high flexibility.

Due to the increase of evidence across homologous species, native mass spectrometry data, DEER measurements, and numerous cryoEM structures, it is reasonable to conclude that the ExbB ExbD subcomplex has a stoichiometry ratio of 5:2 (71). Still, the density for TonB has not yet been determined in complex with ExbB ExbD via cryoEM or X-ray crystallography. The single transmembrane spanning domain of TonB is predicted to interact with ExbB in the cytoplasmic membrane, whereas the ordered, periplasmic domain of TonB would interact with the periplasmic domain of ExbD and the TonB box of TonB-dependent transporters, as described earlier (36). Although the full TonB structure and stoichiometry are unknown, an extensive number of structures have been determined for the ordered, C-terminal domain of TonB interacting with various TBDTs across many species of Gram-negative bacteria (85, 86). Multiple structures of just the C-terminal folded domain of TonB have been determined, with all of them having a three-stranded antiparallel β-sheet and two α-helices in an α1β1β2α2β3 pattern, whereas the fourth beta-strand has a more controversial role. The C-terminal folded domain of TonB has been reported both as a dimer and as a monomer, and this was shown to depend on the length of the truncated construct that was used for the determination of the structure. In the dimer structures, the fourth beta strand is predicted to aid in dimerization, whereas in the monomer, it interacts with the third beta strand. In crystal structures in which TonB is bound to a TBDT, TonB is a monomer in which β3 forms a parallel β-strand interaction with the TonB box of the TBDT and β4 is disordered (85–88).

SUMMARIZING CONSIDERATIONS

TBDTs recognize multiple, distinct substrates, and their number reflects the wide variety of nutrients. Multiple chemical interactions between the nutrients and the TBDTs select the compounds that are transported. Selection is based on the stereochemistry of several amino acid side chains in the TBDTs and a few side chains in the substrates. For example, the chemical structures of ferrichrome and the synthetic antibiotic rifamycin CGP 4832 have little in common, but some side chains of the antibiotic are oriented such that they bind to FhuA (89). This results in the FhuA-dependent import of CGP 4832 and cell death. However, binding is not sufficient for transport, as exemplified by FecA, which transports iron liganded by citrate but does not transport citrate, although the binding site of citrate overlaps with the binding site of iron citrate (2, 3). The number of various tonB, exbB, and exbD genes is much lower than the number of genes encoding TBDTs, suggesting ExbBD specificity for certain TBDTs. However, the limited data in the literature reveal that only a single TonB ExbB ExbD combination energizes the import of various substrates through different TBDTs, whereas the additional combinations were inactive. Presumably, substrates of the inactive combinations were not yet uncovered, and they may be quite different, compared to the current known import systems, as exemplified by the POC secretion and OPH export. The future assignment of functions to TonB ExbB and ExbD combinations faces the problem that the ExbB activity must be separated from the TolQ activity as well as that of ExbD from that of TolR.

A bioinformatic survey revealed that in most genomes of Gram-negative bacteria and in genomes of Archaea, genes are assigned to exbB/tolQ/motA and exbD/tolR/motB, based on sequence similarity. motA motB encode proteins that form the stator of the bacterial motility apparatus (73). They use the pmf to generate torque that drives the rotation of the flagellum. The basic design of MotA₅ MotB₂ is similar to that of ExbB₅ ExbD₂. A lack of functional studies prevents an exact classification of these genes.

The genomes frequently encode a second exbD gene that is close to the first exbD gene, for example, tonB exbB exbD1 exbD2. The formation of the ExbB₅ ExbD₂ complex requires less ExbD than ExbB, and the lower transcription of the exbD gene, relative to the exbB gene, in E. coli takes account of this imbalance (21, 70, 90). The additional exbD genes may encode ExbD proteins that perform yet unknown functions.

Despite the great recent achievements in the elucidation of the mode of action of TonB ExbB and ExbD there are major questions that must still be answered. (i) How is the pmf harvested by ExbBD and then transmitted to a TBDT? (ii) How does the structure of “energized” ExbBD differ from that of ExbBD in the basic state? (iii) How does TonB react to the interaction with energized ExbBD? (iv) How do TBDTs react to energized TonB? (v) How are substrates that are bound to TBDTs vectorially transferred through energized TBDTs? (vi) Do TBDTs react differently to energized TonB, depending on whether they function in transport or signaling? (vii) How does the interaction of the TBDT energized signaling domain with the anti (pro) sigma factor differ from the interaction of the unenergized TBDT? (viii) How does energized TBDT induce the proteolytic fragmentation of the anti (pro) sigma factor? (ix) Finally, how is the protein’s activity of the entire system regulated?