|Home | About | Journals | Submit | Contact Us | Français|
Gram-negative bacteria can detect environmental iron using outer membrane transporters, and then regulate certain transport genes to take advantage of a readily available iron source. This process begins with an iron complex being bound by an outer membrane transporter, and results in a signal being sent across the outer membrane, the periplasmic space, and the inner membrane, to a sigma factor that interacts with RNA polymerase and initiates transcription of relevant genes. Many of the interactions contributing to signalling have been observed by genetic and biochemical studies, but structural studies, which potentially show these interactions in molecular detail, have been limited. In this issue, Garcia-Herrero and Vogel describe an NMR structure of the periplasmic domain of an outer membrane transporter, which had not been seen in previous X-ray crystal structures. This domain transmits the ‘iron availability’ signal to the next protein in the signal transduction cascade, which sits in the inner membrane and extends into the periplasm. The new structure extends our knowledge of transporter architecture and suggests how signalling may occur across the outer membrane.
Iron is a required nutrient for most life forms, from bacteria to humans. Because obtaining enough iron is so important, Gram-negative bacteria detect environmental iron levels and respond in a variety of ways. For example, many genes involved in iron uptake and metabolism are regulated by the ferric-uptake regulator called Fur, which is an iron-responsive DNA binding repressor (Hantke, 2001). When iron levels are high, a Fe2+-Fur complex binds to target DNA sequences in the promoter regions of certain genes, inhibiting transcription. When iron levels fall, Fe2+ dissociates from Fur, enabling RNA polymerase to access the cognate promoters and transcribe iron-responsive genes. In this manner, a bacterium can regulate gene expression, making higher amounts of certain proteins when it senses that iron supplies are scarce, and less when iron is plentiful.
The major iron-transport systems made by Gram-negative bacteria are regulated by the ferric-uptake regulator, Fur (a general mechanism for iron uptake is shown in Figure 1a). Transport of ferric (Fe3+) iron is initiated by an integral outer membrane protein transporter that specifically binds a ferric siderophore or other chelator (such as citrate). Siderophores are small organic molecules synthesized by bacteria or fungi that chelate Fe3+. Many bacteria are able to capture and internalise both endogenous and non-endogenous ferric siderophores, which are transported across the outer membrane with the help of an integral inner membrane protein complex, TonB-ExbB-ExbD. Although many of the details of the transport process are not understood, TonB appears to span the periplasm and interact physically with the outer membrane transporter when a ferric siderophore is bound (Postle and Kadner, 2003). A region at the N-terminus of the transporter called the ‘TonB box’ is a major contributor to this interaction. Both functional TonB protein and proton motive force across the inner membrane are required for transport of the ferric siderophore into the periplasm. Once there, the ferric siderophore is captured by a periplasmic binding protein (PBP) and transferred to an ATP-binding cassette (ABC) transporter, for subsequent transport across the inner membrane.
A TonB-dependent transport system consists of multiple proteins: an outer membrane transporter, a PBP, and several proteins constituting an ABC transporter (Ferguson and Deisenhofer, 2004; Wiener, 2005), each specific for a single ferric siderophore or family of siderophores. Bacteria use considerable metabolic energy to synthesize these systems and their own siderophores, so it makes sense that they can regulate the expression levels of certain transporters for non-endogenous chelators depending on whether they are detected in the environment. For example, Escherichia coli synthesizes and secretes a single siderophore, enterochelin, but makes specific TonB-dependent transport systems for ferric enterochelin and six other potential siderophores, including ferric citrate. When environmental citrate levels are high, E. coli makes more transporters for ferric citrate (Frost and Rosenberg, 1973; Lin et al., 1999). The ferric citrate transport genes are regulated by Fur, as are genes for all other TonB-dependent transport systems, but the increase in expression of ferric citrate transporters in response to environmental citrate suggests a second layer of regulation.
Sequence alignments of TonB-dependent outer membrane transporters have shown that a subset of transporters (OMTN) contain an additional domain at the N-terminus (N-domain), upstream from the TonB box (Schalk et al., 2004; Braun and Mahren, 2005; Koebnik, 2005). In E. coli, only the ferric citrate transporter, FecA out of the seven TonB-dependent iron transporters contains an N-domain. This region of FecA interacts with an integral inner membrane protein, FecR, which is positioned in the inner membrane by a single transmembrane span, with a large domain extending into the periplasm and just a few residues extending into the cytoplasm. FecR acts as a sigma regulator (Enz et al., 2003). The regulated sigma factor in this system is FecI, which interacts with RNA polymerase to initiate transcription of the fec transport genes (Figure 1b). Transcription is initiated without transport of ferric citrate into the cell; binding of ferric citrate to FecA suffices to initiate transcription (Kim et al., 1997; Enz et al., 2000). Because transport of a ferric siderophore is not required, FecA must undergo conformational changes that signal the presence of ferric citrate in the environment to FecR. This transmembrane signal is propagated across the outer membrane by FecA, and then across the inner membrane by FecR, to indicate to the cytoplasmic FecI that ferric citrate is bound at the cell surface.
To begin to understand transmembrane signalling at the molecular level, crystal structures of FecA were solved in the ligand-free (FecAApo) and diferric dicitrate-bound forms (FecAFeCit) (Ferguson et al., 2002; Yue et al., 2003), and in the dicitrate-bound form (FecACit) (Yue et al., 2003). Since all OMTN family members appear to bind iron-free siderophores (Schalk et al., 2004), we can compare the FecACit and FecAFeCit structures to look for conformational changes occurring upon diferric dicitrate binding. All TonB-dependent transporters whose structures have been solved share a common fold (Wiener, 2005). A 22-stranded beta barrel spans the outer membrane, displaying long extracellular loops that participate in ligand recognition and binding, and short periplasmic turns that barely protrude into the periplasm (Figure 2). A second domain is formed from the first 150 or so residues of the protein, which is inserted into the barrel lumen and plugs the pore that the barrel would otherwise form. The ‘plug’ domain contains two or three loops that protrude beyond the outer membrane, and residues from these loops form the floor of the siderophore-binding pocket. The walls of the binding pocket are formed from residues in the extracellular loops of the barrel. In addition, the TonB box is located near the beginning of the plug domain, extending into the periplasm to await interaction with the TonB protein. Upon binding diferric dicitrate, FecA undergoes several conformational changes. The largest changes are seen in extracellular barrel loops L7 and L8, which move on average 7 Å and 8 Å, respectively, to close the binding pocket and sequester diferric dicitrate (but not dicitrate alone). These loop movements also change the chemical properties of the siderophore binding pocket, with different residues binding dicitrate versus diferric dicitrate (Yue et al., 2003). The other significant conformational change is seen in the TonB box, which is visible in an extended conformation in the FecACit structure, but becomes completely disordered, and therefore not visible by X-ray crystallography, in the FecAFeCit structure. Together, these conformational changes send a transmembrane signal to TonB, announcing ferrichelate binding and readiness for transport. The conformational changes seen in the FecAFeCit structure might also form part of the signal to initiate transcription of the fec transport genes through the FecIR pathway.
The 79-residue domain of FecA that binds to FecR (N-domain) was not seen in any of the FecA crystal structures. This domain is absolutely required for FecIR induced transcription, although it is not required for siderophore transport (Kim et al., 1997). Since the N-domain immediately precedes the TonB box of FecA, and since the TonB box undergoes significant conformational changes upon siderophore binding, the unseen N-domain is also likely to move and/or change conformations during the signalling process. Clearly, a structure of the N-domain would help us to understand this signalling event, and now one is available.
In this issue, Garcia-Herrero and Vogel describe the NMR structure of a 96-residue portion of FecA, consisting of the entire N-domain plus the first few residues of the plug domain, including the TonB box (residues 80 to 84). The N-terminal 74 residues fold into a compact domain, with two alpha helices sandwiched between two beta sheets. This domain appears to represent a unique fold, as no other protein structures showed significant structural similarity. Sequence comparison of OMTN family members suggests that the fold will be conserved for this domain, implying also that interactions between TonB-dependent transporters and sigma regulators will be conserved on structural and functional levels (Schalk et al., 2004; Koebnik, 2005). Now that the fold for the N-domain is known, we can speculate on which regions participate in FecR recognition. Two mutations in the N-domain have been shown to suppress randomly generated FecR mutations (Enz et al., 2003) and these two residues in FecA, glycine 39 and aspartate 43, lie on the same face of the N-domain. This finding immediately indicates the N-domain binding surface that interacts with FecR.
The other portion of the NMR structure, consisting of residues 75-96 (which includes the TonB box), is completely unstructured and adopts many different conformations, suggesting flexibility and movement. This flexibility is further supported by NMR dynamics measurements conducted by Garcia-Herrero and Vogel. Superposition of three of the lowest energy NMR structures with the FecA crystal structure, using the TonB box as a reference point, suggests that large concerted N-domain movements are possible (Figure 3). The new structure does not indicate whether the folded portion of the N-domain undergoes conformational rearrangements upon ferric citrate binding to FecA, or upon subsequent interaction with FecR, but movement of the domain within the periplasm is very likely. We now know what a bacterial metal detector looks like, and how it responds when iron (in the form of diferric dicitrate) is bound.
How does FecR, the next protein in the signal transduction pathway, respond to these conformational changes in FecA? We need the structures of full-length FecA in complex with FecR, and of related transporters, in the presence and absence of ferric siderophores, to characterize further the mechanism of siderophore-induced transcription regulation. Such structures, consisting of intact outer and inner membrane proteins in complex, will be difficult to obtain, but the next step could be to look at a complex of intact FecA with the periplasmic domain of FecR. In addition, full-length FecR, or the cytoplasmic portion of it, might be studied in complex with FecI. Now that the structure of the N-domain is known, mutagenesis experiments could map the binding interaction with FecR, even in the absence of new structures. As is generally the case, a complete understanding of bacterial signalling across three compartments will require complimentary experiments from genetics, biochemistry, and structural biology.
I thank Travis Barnard, Harris Bernstein, and Reinhard Grisshammer for critically reading the manuscript. This research was supported by the Intramural Research Program of the NIH, National Institute of Diabetes and Digestive and Kidney Diseases.