MMBR Try AEM online
Home Help [Feedback] [For Subscribers] [Archive] [Search] [Contents]
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow Copyright Information
Right arrow Books from ASM Press
Right arrow MicrobeWorld
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Mascher, T.
Right arrow Articles by Unden, G.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Mascher, T.
Right arrow Articles by Unden, G.
Microbiology and Molecular Biology Reviews, December 2006, p. 910-938, Vol. 70, No. 4
1092-2172/06/$08.00+0     doi:10.1128/MMBR.00020-06
Copyright © 2006, American Society for Microbiology. All Rights Reserved.

Stimulus Perception in Bacterial Signal-Transducing Histidine Kinases

Thorsten Mascher,1* John D. Helmann,2 and Gottfried Unden3*

Department of General Microbiology, Georg-August-University Göttingen, D-37077 Göttingen, Germany,1 Department of Microbiology, Cornell University, Ithaca, New York 14853-8101,2 Institute of Microbiology and Wine Research, Johannes-Gutenberg-University Mainz, D-55099 Mainz, Germany3

SUMMARY
INTRODUCTION
PERIPLASMIC-SENSING HISTIDINE KINASES
    Prototypical Periplasmic-Sensing Histidine Kinases
    NarX/NarQ-Like Sensors of Environmental Nitrate and Nitrite
    Citrate and C4-Dicarboxylate Sensor Proteins CitA and DcuS
        Proposed mode of signal perception and transduction by DcuS/CitA.
    Sensor Kinases with PBPb Sensing Domains
    Novel Conserved Periplasmic Sensing Domains: CACHE, CHASE, and Reg_prop
    General Roles of Extracytoplasmic Sensor Domains in Stimulus Perception
HISTIDINE KINASES WITH SENSING MECHANISMS LINKED TO THE TRANSMEMBRANE REGIONS
    Intramembrane-Sensing HKs: Cell Envelope Stress Sensors with Two TMR (LiaS/BceS-Like HKs)
        LiaS-like HKs.
        BceS: a two-component system-ABC transporter connection.
        Miscellaneous IMHKs.
        IMHK-like periplasmic-sensing HKs: VanS/PrmB-like proteins.
    DesK-Like Thermosensors with Four to Six TMR
    RegB/PrrB-Like Redox-Responding Global Sensor Kinases with Six TMR
    Peptide Quorum Sensors with 6 to 10 TMR (AgrC/ComD-, ComP-, and LuxN-Like HKs)
    CbrA- and PutP-Like Proteins with More than 10 TMR: Sensor Kinases with Fused Secondary Carrier Domains
    Novel Conserved Putative TMR-Associated Sensing Domains with Six to Eight TMR: MHYT, MASE1, 7TMR-DISM, and 5TMR-LYT
    Models for Stimulus Perception Associated with TMR
CYTOPLASMIC-SENSING HISTIDINE KINASES
    Membrane-Anchored HKs with N-Terminal Cytoplasmic Sensing Domains
    Membrane-Anchored HKs with C-Terminal Cytoplasmic Sensing Domains
    Soluble HKs Associated with Membrane-Integral Sensory Proteins
    Soluble, Cytoplasmic-Sensing HKs
    Soluble Cytoplasmic-Sensing HKs: the ''Missing Link'' between Two-Component and One-Component Signal Transduction?
COMBINATION OF SENSING DOMAINS
CONCLUSIONS AND OUTLOOK
ADDENDUM IN PROOF
ACKNOWLEDGMENTS
REFERENCES

   SUMMARY
 Top
 Next
 References
 
Two-component signal-transducing systems are ubiquitously distributed communication interfaces in bacteria. They consist of a histidine kinase that senses a specific environmental stimulus and a cognate response regulator that mediates the cellular response, mostly through differential expression of target genes. Histidine kinases are typically transmembrane proteins harboring at least two domains: an input (or sensor) domain and a cytoplasmic transmitter (or kinase) domain. They can be identified and classified by virtue of their conserved cytoplasmic kinase domains. In contrast, the sensor domains are highly variable, reflecting the plethora of different signals and modes of sensing. In order to gain insight into the mechanisms of stimulus perception by bacterial histidine kinases, we here survey sensor domain architecture and topology within the bacterial membrane, functional aspects related to this topology, and sequence and phylogenetic conservation. Based on these criteria, three groups of histidine kinases can be differentiated. (i) Periplasmic-sensing histidine kinases detect their stimuli (often small solutes) through an extracellular input domain. (ii) Histidine kinases with sensing mechanisms linked to the transmembrane regions detect stimuli (usually membrane-associated stimuli, such as ionic strength, osmolarity, turgor, or functional state of the cell envelope) via their membrane-spanning segments and sometimes via additional short extracellular loops. (iii) Cytoplasmic-sensing histidine kinases (either membrane anchored or soluble) detect cellular or diffusible signals reporting the metabolic or developmental state of the cell. This review provides an overview of mechanisms of stimulus perception for members of all three groups of bacterial signal-transducing histidine kinases.


   INTRODUCTION
 Top
 Previous
 Next
 References
 
Life in the microbial world is characterized by continuous interactions between the bacterial cell and its environment. The ability of a bacterium to monitor environmental parameters, including osmotic activity and ionic strength, pH, temperature, and the concentrations of nutrients and harmful compounds, is a prerequisite for survival. For that purpose, bacteria have evolved surface-exposed signal transduction systems, typically comprised of transmembrane (TM) proteins that channel the input from sensory modules to intracellular responses. These TM signaling systems include the chemotaxis receptors, anti-{sigma}:{sigma} factor pairs, Ser/Thr protein kinases, and histidine protein kinases together with their cognate response regulators. This review focuses on the classical two-component systems (TCS) consisting of a usually membrane-bound sensor histidine protein kinase (HK) and a response regulator (RR), most often mediating differential gene expression (103, 108).

TCS allow adaptational responses to a huge variety of environmental stimuli, based on a simple modular system. Both proteins consist of (at least) two distinct domains. The HK harbors an N-terminal input domain that senses a specific stimulus, e.g., by binding or reacting with a signaling molecule or by interaction with a physical stimulus. The information is transduced through intramolecular conformational changes, resulting in the activation of the cytoplasmic transmitter domain (280). The transmitter, in turn, activates its cognate receiver, encoded by the N-terminal domain of the RR. The RR gives rise to the appropriate cellular response, which is mediated by the C-terminal effector (or output) domain of the RR through protein-protein interaction (e.g., chemotaxis) or protein-DNA interactions leading to differential gene expression.

The functional state of these two components is controlled by three phosphotransfer reactions: (i) the autophosphorylation of a conserved histidine in the transmitter domain of the sensor, (ii) the phosphotransfer to a conserved aspartate in the receiver domain of the RR (by the activity of the RR), and (iii) dephosphorylation of the RR to set the system back to the prestimulus state (202, 246). The phosphatase can be an intrinsic property (autophosphatase) of the RR or a phosphoprotein phosphatase activity of the kinase towards the regulator. However, external phosphatases are also common (129, 205, 206, 246). Some HKs show significant autophosphatase activity towards their own His-phosphate group. Additionally, some RRs also catalyze significant back transfer of the phosphate group to the corresponding His kinase (52).

With a few exceptions (i.e., Mycoplasma species), all bacterial genomes sequenced so far harbor multiple copies of genes encoding TCS. Typically, the structural genes for the HK and the cognate RR are organized in operons. In some bacteria, however, the two-component systems show an orphan organization at the gene level which impedes assigning sensor/regulator pairs and the identification of stimuli. This problem is prominent in Myxococcus xanthus, which contains a large number of TCS (146 HKs, including hybrid kinases), many of which are important for the control of complex differentiation programs such as fruiting body morphogenesis. More than 50% of the structural genes for the HKs are orphans and are separated by two or more genes from those of the next RR (238).

While some systems have been studied in great detail (most notably the paradigmatic systems EnvZ/OmpR, CheA/CheY, and NtrB/NtrC in Escherichia coli) (25, 109, 187) and transcriptome approaches have allowed initial genome-wide investigations on some TCS, many are still uncharacterized. Genome-wide sequence analyses to identify TCS have been performed for many bacteria, including Bacillus subtilis (59), Escherichia coli (178), Pseudomonas aeruginosa (223), Corynebacterium glutamicum (146), Streptomcyes coelicolor (107), and cyanobacteria (16). These analyses have been complemented, in some cases, by mutational and/or microarray approaches, as reported for B. subtilis (145), Streptococcus pneumoniae (153, 255), and E. coli (196, 286). It is anticipated that ongoing functional genomics approaches will rapidly advance our understanding of these large suites of sensory systems.

Sequence comparisons of TCS have been used to identify a number of conserved subfamilies. A structural classification of bacterial response regulators based on the diversity of output domains, domain architecture, and domain combinations was recently published (66). So far, comparisons of HK proteins have focused on their highly conserved intracellular catalytic (transmitter) domains (53, 87, 140). They show a homogenous composition of subdomains (or "boxes") and are generally cytoplasmically located (87, 280). A classification based on the H box of the kinase domain (containing the conserved site of autophosphorylation) was proposed by Fabret et al. (59) and has found widespread use. The most comprehensive and detailed sequence analysis, based on all six conserved boxes (the H, X, N, D, F, and G boxes) (see Fig. 2) in the transmitter domain (87), allows an even more precise subgrouping of HKs. A more recent classification (140), based on the sequence, organization, and predicted secondary structure of the H box, allowed the classification of archaeal HKs for the first time. These classification schemes are still evolving, as evidenced by the recent identification of a new subfamily of HKs, the HWE family (133). Although these analyses have focused specifically on the conserved features of the HK catalytic domain, they likely reflect the evolution of the TCS as a whole: in many cases specific subfamilies of HK are preferentially associated with specific subfamilies of RR proteins (87, 149). However, these classifications neglect functional aspects of the sensing and signal transduction process.


Figure 2
View larger version (18K):
[in this window]
[in a new window]

 
FIG. 2. Features, domains, and boxes of histidine protein kinases. The protein is symbolized by the gray line. The major domains are indicated by boxes, and their names are given above or below the line. Conserved boxes or amino acid residues are given below the line in one-letter code, according to the standard nomenclature (87, 280). The drawing is not to scale. See the text and Table 1 for details.

 
The principal biological function of TCS-mediated signal transduction manifests itself in the input (signal perception) and output (e.g., differential gene expression), rather than in the communication between its two components. Therefore, grouping HKs according to their input domains would reflect the biological role they play in the communication between a cell and its environment. While a number of novel conserved input domains were identified in HKs in recent years (5-7, 65, 67, 68, 186, 203, 288, 289), an overall classification based on the N-terminal sensor domains remains problematic, since these domains vary greatly in sequence, membrane topology, composition, and domain arrangement. All of these features have profound effects on sensing and signal transduction to the kinase domain. Therefore, this variability presumably reflects different principles in stimulus perception and processing, which are related to the topology and type of sensory domains but do not necessarily reflect the phylogenetic relationship of the sensor kinases. Consequently, a functional classification of HKs cannot be based on sequence alignments alone but rather requires the identification of domains and transmembrane helices and the prediction of the topological arrangement of these structures.

In order to arrange the large number of HKs according to functional aspects, the sensors are grouped here on the basis of their domain architecture, i.e., membrane topology, number of TM helices, and sequential arrangement of the sensory domain(s) within their N-terminal input domains. The available data clearly indicate that grouping HKs by these criteria is functionally related to the mechanism of sensing and signal transduction by the corresponding sensors but does not necessarily take into account phylogenetic aspects. Based on our analysis of the domain architecture and membrane topology of ~ 4,500 sensor kinase sequences in the SMART database (229) (http://smart.embl-heidelberg.de/) and the published results on signal perception presented here, most (if not all) HKs fall into three major groups (Fig. 1).


Figure 1
View larger version (42K):
[in this window]
[in a new window]

 
FIG. 1. Schematic representation of the three different mechanisms of stimulus perception. (A) Periplasmic-sensing HKs. (B) HKs with sensing mechanisms linked to the transmembrane regions (stimulus perception can occur either with the membrane-spanning helices alone or by combination of the transmembrane regions and short extracellular loops). (C) Cytoplasmic-sensing HKs (either soluble or membrane-anchored proteins). The stimulus is represented by a red arrow or red star. The parts of the proteins involved in stimulus perception are highlighted in color.

 
The largest group, the periplasmic (or extracellular)-sensing HKs, includes proteins with an extracellular sensory domain which is framed by at least two TM helices (Fig. 1A). The kinase is localized in the cytoplasm (as for all other HKs). Thus, sensory and kinase domains are located in two different cellular compartments which are separated by a membrane, necessitating TM signal transduction. This type of membrane topology is typical for sensing solutes and nutrients.

The second group contains HKs with sensing mechanisms associated with the membrane-spanning helices. The unifying feature of this highly diverse group of sensor kinases is the presence of 2 to 20 transmembrane regions (TMR) implicated in signal perception. These TMR are connected by very short intra- or extracellular linkers; i.e., these sensors lack an obvious extracellular input domain (Fig. 1B). Therefore, the stimuli sensed either are membrane associated or occur directly within the membrane interface. Stimuli from within the membrane include the mechanical properties of the cell envelope (such as turgor or mechanical stress) or are derived from membrane-bound enzymes or other membrane-integral components. Other membrane-related stimuli include ion or electrochemical gradients, transport processes, and the presence of compounds that affect cell envelope integrity. Most quorum sensors from gram-positive bacteria also fall into this category. For the latter group, two of the TMR are connected by a short (20 to 50 amino acid residues) intra- or extracellular linker, which seems to be involved in stimulus perception. Signal transfer occurs from the membrane to the cytoplasmic kinase domain.

The third (and second-largest) group of sensor kinases, the cytoplasmic-sensing HKs, includes either membrane-anchored or soluble proteins with their input domains inside the cytoplasm (Fig. 1C). This class of sensor proteins detects the presence of cytoplasmic solutes or of proteins signaling the metabolic or developmental state of the cell or of the cell cycle. Other cytoplasmic TCS respond to diffusible or internal stimuli, such as O2 or H2, or stimuli transmitted by TM sensors.

The subgroups within these three principal groups of HKs and their compositions of different sensory domains, as well as the stimulus perception mechanisms for well-characterized representatives from each group, will be the subject of the following sections of this review. It should be pointed out that all classes of sensors can contain, in addition to the principal features (Fig. 2), additional sensory or linker domains in various combinations. We also address the mechanism of intramolecular TM signal transduction and the occurrence and biological role of HKs that harbor more than one putative stimulus perception domain.


   PERIPLASMIC-SENSING HISTIDINE KINASES
 Top
 Previous
 Next
 References
 
Periplasmic-sensing HKs represent the classical type and comprise the largest group of membrane-bound sensor kinases. As a definition and for reasons of simplicity, we use the term "periplasmic-sensing" HK throughout this review for all sensor kinases with a significantly large extracytoplasmic input domain, irrespective of their origin (i.e., HKs from both gram-negative and gram-positive bacteria). At present, databases contain about 2,500 members of this type. They consist of two regions: an N-terminal periplasmic sensing domain flanked by TM helices on either side (TMR1 and TMR2), followed by the C-terminal cytoplasmic transmitter domain (Fig. 2). The transmitter domain comprises a sequence with the conserved histidine residue for autophosphorylation (the H box) and ends with the highly conserved kinase (or catalytic) domain. The domain with the conserved His residue typically contains two {alpha}-helices (X box), which serve as a dimerization domain (DHp [dimerization and histidine phosphotransfer] or HisKA domain) (Table 1 and Fig. 2). The catalytic domain (HATPase) contains the conserved N, D, F, and G boxes with the respective highly conserved amino acid residues. This domain catalyzes autophosphorylation of the HKs. The RRs then catalyze their own phosphorylation, with the phosphoryl-HK as the phosphodonor (279, 280). In some cases, low-molecular-weight donors such as acetyl phosphate or, in vitro, also phosphoramidates can be used for RR phosphorylation (45, 158).


View this table:
[in this window]
[in a new window]

 
TABLE 1. Names, term definitions, and features of conserved signaling and sensory domains

 
This prototypic domain organization can be varied by inclusion of "linker" regions, such as the HAMP or PAS domain (12, 276), between TMR2 and the transmitter domain or by additional phosphorylation domains downstream of the transmitter domain (Fig. 2). The linker domains vary considerably in size, extent, and type. The additional phosphorylation domains comprise receiver domains typical for RRs, with a conserved Asp residue for phosphorylation and an additional transmitter (histidine-containing phosphotransfer [HPt]) domain (Fig. 2). "Hybrid" kinases of this type constitute a phosphorelay, predominantly in gram-negative bacteria (285), whereas phosphorelay systems of gram-positive bacteria, such as the regulatory cascade of sporulation initiation in B. subtilis (58), normally consist of individual proteins mediating the stepwise phosphotransfer.

Sensor kinases sharing the prototypical architecture form a large and highly diverse group with regard to composition and function of the sensory or input domain (Fig. 3). The HKs are grouped primarily based on features of the periplasmic sensing domains, to reflect functional and sensory principles. The linker domain will be used in specific cases as an additional criterion for subgrouping, whereas transmitter domains will not be considered.


Figure 3
View larger version (51K):
[in this window]
[in a new window]

 
FIG. 3. Domain architecture of periplasmic-sensing histidine kinases. The figure is based on the graphical output of the SMART web interface at http://smart.embl-heidelberg.de (229), with modifications. The scale bar is in amino acids. Blue vertical bars represent putative transmembrane helices. Sizes and positions of conserved domains are indicated by the labeled symbols. Note that the transmitter domains are simplified, and as a default, only the HisKA and HATPase_c domains are shown. Additional cytoplasmic domains are possible and widespread but were ignored in all but obligatory cases (i.e., PAS domain for CitA-like HKs and HAMP domain for NarXQ-like HKs). A diagonal bar at the C terminus of the transmitter domain indicates the possible occurrence of hybrid kinases in that subgroup of sensor kinases. The periplasmic PAS domain of CitA/DcuS (in parentheses) is conserved by three-dimensional structure only and not by sequence. It is therefore not detectable by sequence analysis. VanS/PrmB-like proteins are described in the "Intramembrane-Sensing HKs: Cell Envelope Stress Sensors with Two TMR (LiaS/BceS-Like HKs)" section. See the text for details.

 
Despite a great diversity in sequence and stimulus specificity, the input domains of many sensors can be grouped by sequence alignment into a few families (Fig. 3 and Table 2). Most of the conserved domains (5-7, 67, 68, 186, 203, 288) are found also in other signaling and sensing systems. The functions of only a limited number of such sensing domains of HKs has been elucidated, in particular those of periplasmic (CitA, DcuS, and PhoQ) and cytoplasmic (FixL) PAS domains (40, 79, 85, 137, 144, 201, 219). Others are defined only by means of sequence conservation (such as the CHASE domains), and their functions and stimuli remain to be identified (65). Additionally, many HKs, including paradigmatic HKs such as EnvZ, do not contain conserved features in their periplasmic domains. Corresponding to the diversity of sensory domains, many different mechanisms of stimulus perception and processing can be anticipated, although these are mostly unknown. There are, however, a few well characterized examples that can serve as models.


View this table:
[in this window]
[in a new window]

 
TABLE 2. Groups of bacterial histidine kinases according to the domain architectures of their sensing domains

 
Perhaps the simplest mechanism for signal detection by periplasmic-sensing HKs occurs by direct interaction between the sensor domain and a chemically defined small molecule, such as nitrate/nitrite for NarX or citrate for CitA (115, 242). For some of the periplasmic-sensing HKs, the binding site and structure of the periplasmic binding domain has been determined (see below). In other cases, chemical stimuli are sensed indirectly through interaction with a periplasmic solute-binding protein, such as glucose by the sugar-binding protein ChvE for interaction with the Agrobacterium tumefaciens VirA sensor kinase (234). Alternatively, signaling may be triggered by mechanical, electrochemical, or concentration gradient stimuli, resulting in a conformational change of the input domain, as has been hypothesized for osmolarity or turgor sensors. There is also growing evidence that other periplasmic-sensing TCS use additional proteins, which transmit a primary signal to the HK, thereby complicating the identification of the primary stimulus. For most TCS the exact type of stimulus is not known. In the following descriptions, we will concentrate on characterizing general properties of the families, which will be exemplified by a few prominent members.

Prototypical Periplasmic-Sensing Histidine Kinases

Prototypical periplasmic-sensing histidine kinases are composed of two TM helices with an intervening extracytoplasmic domain of 50 to 300 amino acids (aa), lacking large cytoplasmic linker regions. The sequences of the extracytoplasmic regions of most prototypical periplasmic-sensing HKs reveal no conserved or known sensing domains. Structural analysis of the PhoQ periplasmic domain revealed, however, a PAS domain, which was not recognized by sequence analysis (similar to the PAS domains of CitA and DcuS [40]). The other classes of periplasmic-sensing HKs are characterized by the presence of defined types of sensory domains in the periplasm and by the presence of extended linker domains. EnvZ, PhoQ, TorS, and VirA are important members of the prototypical HKs and will be discussed in more detail.

EnvZ, together with its cognate RR OmpR, plays a central role in the adaptation of E. coli to changes in extracellular osmolarity. EnvZ is one of the best-understood HKs, and studies of this protein have contributed enormously to our understanding of dimerization and phosphorylation reactions in membrane-bound HKs. The structures of the transmitter subdomains were solved by nuclear magnetic resonance (253, 257). The HisKA or DHp subdomain consists of two {alpha} helices that dimerize to form a four-helix bundle, which represents the core of the transmitter domain (53, 257). The phospho-accepting His residue protrudes from the helices and is accessible to phosphorylation from the surface by the C-terminal catalytic HATPase subdomain. Thus, EnvZ is a functional dimer, with both the cytoplasmic transmitter and the periplasmic domains contributing to dimerization. The structures of the HisKA/DHp and HATPase domains gave important insights into the functions of the individual domains. Recently, the structure of the entire cytoplasmic transmitter, including the HAMP linker domain, of an HK was solved (165), which will help us to understand how the signal from TM helix 2 is received by the cytoplasmic domain and transferred to the kinase domain.

Despite the wealth of knowledge gained over the years on the function of this archetypical TCS, little is known about the mechanism of osmosensing by EnvZ. A close homolog, EnvZ of Xenorhabdus nematophilus, completely lacks a periplasmic domain [see "Intramembrane-Sensing HKs: Cell Envelope Stress Sensors with Two TMR (LiaS/BceS-Like HKs)"] but is still able to complement an E. coli envZ null mutant (251). In E. coli EnvZ, partial deletions of the periplasmic domain, or even a complete replacement with the periplasmic region of the nonhomologous PhoR of B. subtilis, did not significantly alter the process of osmosensing (156). These results call into question a direct and essential role of the periplasmic domain of E. coli EnvZ in osmosensing. Work on the yeast osmosensor Sln1 suggests that these kinases sense turgor as the key input stimulus. A systematic deletion/replacement analysis of the periplasmic domain of Sln1 suggests that only the integrity of the periplasmic domain as a whole is necessary for osmosensing, rather than specific amino acids sequence or regions (220). This work complements and supports the results obtained for E. coli EnvZ.

The PhoQ/PhoP TCS is important for the control of pathogenesis of Salmonella and other gram-negative bacteria. Multicellular organisms inhibit invading bacteria by use of cationic antimicrobial peptides, which contain a positive net charge and an amphipathic structure for interaction with negatively charged biological membranes (43). The bacteria acquire resistance to the antimicrobial peptides by modifying the cell surface, in particular lipopolysaccharide and lipid A, which are modified in antimicrobial peptide-resistant strains (174). The PhoQP TCS regulates modification of lipid A and other virulence factors, including those for antimicrobial peptide resistance (61, 91, 174, 217).

The sensor kinase PhoQ is activated at low concentrations of cations, such as Mg2+, and by increasing concentrations of the antimicrobial peptides (i.e., during invasion of macrophages by the bacteria) but is repressed by high concentrations of divalent cations (70). Thus, these effectors have opposing effects on PhoQ function (Fig. 4). The crystal structure of the periplasmic PhoQ domain was determined in the Ca2+ bound state (40). The periplasmic domain belongs to the PAS domain family, despite an apparent lack of sequence similarity to PAS domain proteins. The structure matches that of CitA and DcuS (see below), but there is an insertion of two {alpha}-helices in the PAS fold. This insertion creates a flat and negatively charged surface on one side of the protein, which is derived from Glu and Asp residues. Unlike the PAS domains of DcuS and CitA, the PhoQ periplasmic domain contains no cavity or discrete binding pocket for ligand binding. It is proposed that binding of the antimicrobial peptides and Mg2+ occurs at the acidic surface at the membrane-proximal side of the protein, which is in close contact to the lipid surface of the cytoplasmic membrane (Fig. 4A). The acidic surface binds at least three Mg2+ ions, which are proposed to form cation bridges between the acidic region of PhoQ and the acidic membrane phospholipids (40) (Fig. 4A).


Figure 4
View larger version (64K):
[in this window]
[in a new window]

 
FIG. 4. Structure of the periplasmic input domain of PhoQ and model for the sensing mechanism of cations and antimicrobial peptides. (A) The crystal structure and charge profile of the surface facing the outer side of the cytoplasmic membrane are shown on the left. The residues important for coordinating the divalent metal ions are shown. The crystal structure of the dimeric PhoQ sensor domain (upper panel) forms a flat surface that comes in close contact to the membrane. The bottom part of this domain contains a highly negatively charged surface that participates in metal binding (lower panel, view from the membrane). Red represents negatively charged residues. NT, N terminus; CT, C terminus. (B) Working model for the competitive binding of Mg2+ and cationic antimicrobial peptides to PhoQ. Divalent cations, such as Ca2+ or Mg2+ (shown as green balls), bind to the acidic surface (red) and repress PhoQ activity by locking the PhoQ sensor domain in an inactive conformation (top panel). Cationic antimicrobial peptides interact with membrane phospholipids, thereby coming in close contact with the Ca2+ and Mg2+ binding sites of PhoQ. They compete with and displace divalent cations from PhoQ (middle panel). This provokes a conformational change of the input domain, which leads to autophosphorylation of the transmitter domain and thereby activation of PhoQ (lower panel). See the text for details. (Reprinted from reference 18 with permission from Elsevier.)

 
Ionic interactions tether the periplasmic domain of PhoQ to the membrane. In this state, the kinase domain of PhoQ is inactive (18, 40). The binding of antimicrobial peptides is suggested to displace the cations and to disrupt the interaction between PhoQ and the membrane. The antimicrobial peptides are suggested to function as a lever (Fig. 4B), lifting the periplasmic domain off the membrane. The structural distortion could be transmitted mechanically to the TM helices, with a resultant motion of the TM helices, resulting in the autophosphorylation of PhoQ. Thus, the stimuli of PhoQ (antimicrobial peptides and divalent cations) are not bound at a distinct binding pocket but function by resolving and forming interactions between PhoQ and the membrane surface (18).

Other prototypic periplasmic-sensing HKs, i.e., VirA, TorS, and BctE, sense stimuli by interaction of their periplasmic domains with other proteins in the same compartment. These proteins are often solute-binding proteins, which are part of binding-protein-dependent transport systems. Periplasmic solute-binding proteins are also used for sensing by methyl-accepting chemotaxis proteins (MCP) in bacterial chemotaxis (147, 252).

Expression of virulence (vir) genes in the gram-negative plant pathogen Agrobacterium tumefaciens is regulated by the VirAG TCS in response to acidic pH. The decrease in pH is caused by phenolic compounds (such as acetosyringone) that are released by wounded plant cells (reviewed in references 99 and 287). In addition, aldose monosaccharides (e.g., arabinose) exuded from wounded plant sites serve as strong enhancers of phenolic-induced HK activity (34, 235). The sugar is sensed by binding to the periplasmic sugar-binding protein ChvE, which in turn interacts with the 220-aa-long periplasmic input domain of VirA (36, 234). Direct sensing of phenolic compounds by VirA occurs in the cytoplasmic linker region between TMR2 and the HisKA domain, i.e., at a site different from that of ChvE interaction (20, 36, 49, 204, 277). Recently, it was established that the periplasmic domain is also involved in pH sensing (69).

The TorS/TorR two-component system of E. coli controls the expression of the torCAD operon, encoding the periplasmic trimethyl amine-N-oxide (TMAO) reductase (TorA), a TorA-specific chaperone (TorD), and the c-type cytochrome TorC (171). TorC is a membrane-bound protein and carries the catalytic domain containing pentaheme c on the periplasmic surface of the membrane. TorC interacts with the catalytic domain of TorA in the periplasm and forms a functional TorC-TorA respiratory complex of TMAO reductase (83). TorS detects TMAO presumably by its periplasmic region (123) and stimulates the expression of the torCAD operon. In addition, apoTorC lacking the heme C groups binds to the sensor TorS and negatively regulates kinase activity (i.e., when it is inactive and not able to form an active TorC-TorA respiratory complex). This direct protein-protein interaction involves the C-terminal part of TorC and the periplasmic domain of TorS (83).

Bordetella pertussis uses the BctDE TCS for controlling the expression of a citrate uptake system during growth on citrate (9). The citrate carrier is a tripartite tricarboxylate transporter, consisting of the BctAB membrane carrier proteins. BctC is an extracytoplasmic citrate-binding protein and represents the third component of the tripartite tricarboxylate transporter system. BctE is a prototypical HK with two TM helices and requires BctC for response to the citrate. Citrate-liganded BctC interacts with the periplasmic sensing domain of BctE and controls the functional state of the sensor.

NarX/NarQ-Like Sensors of Environmental Nitrate and Nitrite

A second, well-characterized group of periplasmic-sensing HKs is represented by the NarX-NarQ-like sensors, which contain defined "boxes" or domains in the periplasm and extended linker domains between TMR2 and the transmitter domain (Fig. 3). In proteobacteria, anaerobic respiratory gene expression is controlled by one of two nitrate reductase (Nar) TCS, NarXL or NarQP, in response to environmental nitrate and nitrite. E. coli and Salmonella enterica contain both paralogs. The concentration of the two respiratory oxidants is sensed through ligand binding by the periplasmic domain (length, 115 aa) of the two corresponding HKs, NarX and NarQ. Multiple sequence alignments reveal two conserved stretches of 18 amino acid residues in length that flank each periplasmic side of the two TM helices (P and P' boxes) (Fig. 3) (242). Alanine substitutions of highly conserved residues in the P box (but not in P') strongly affected signal detection and were able to render NarX and NarQ in a constitutively active ("locked-on") or inactive ("locked-off") form (35, 242, 275). In addition to the conserved extracellular boxes, both types of Nar sensors have an extended cytoplasmic linker region. In the linker region, "locked-on" mutations that are dominant over "locked-off" mutations in the P box were identified (35, 131). Therefore, it was proposed that signal processing requires nitrate (or nitrite) detection in the periplasm by the P box, followed by signal transfer across the membrane and to the kinase, with the latter depending on transmission by the linker region. The linker region consists of a HAMP domain (12) and an unusual "Y-Cys-Q" module in front of the HisKA domain. Similarly to other HAMP linkers, the sequence immediately follows TMR2 and is predicted to form two short amphipathic {alpha}-helices, which are joined by an unstructured connector. Mutations in the HAMP linker disturb the function of NarX and NarQ, and it was concluded that the HAMP linker is required for proper signal transduction (10, 11). The central "Y-Cys-Q" module consists of three parts: (i) the Y box, a leucine-zipper like domain of 32 amino acids in length; (ii) the central cysteine cluster, which is present in NarX but missing in NarQ-like HKs; and (iii) the Q linker, which is reminiscent of glutamine-rich flexible interdomain linkers (Fig. 3) (242). While deletion and alanine replacement mutagenesis could establish a role of the Y box and Q linker in intramolecular signal transduction, no phenotype was observed in mutants lacking one or more of the conserved cysteine residues of the central region (242). The response regulator NarL is a member of the FixJ/LuxR family and is structurally and functionally well characterized (19). Taken together, the NarXQ-like HKs are distinct from most other periplasmic-sensing kinases due to conserved periplasmic and cytoplasmic domains involved in signal detection and transduction.

Citrate and C4-Dicarboxylate Sensor Proteins CitA and DcuS

CitA/DcuS-like HKs respond to the environmental concentrations of citrate (CitA) or C4-dicarboxylates and citrate (DcuS) and, together with their cognate RRs CitB/DcuR, regulate genes encoding carriers and enzymes for the degradation of externally supplied citrate or C4-dicarboxylates, in particular under anaerobic growth conditions. Metabolism of endogenously produced citrate or C4-dicarboxylates is not regulated by CitA/DcuS. In binding studies, the periplasmic domains of CitA of Klebsiella pneumoniae and E. coli function as high-affinity citrate receptors with KD values in the micromolar range (136). The molecular interactions between the sensor domain and its ligand have been elucidated through the cocrystal structure of the periplasmic domain of CitA of K. pneumoniae and mutagenesis (Fig. 5) (75, 137, 219). The structure of the corresponding domain of the C4-dicarboxylate sensor DcuS of E. coli was determined in solution by nuclear magnetic resonance (Fig. 5) (201). The extracellular input domains of both sensors adopt a PAS-like fold with a core structure consisting of four or five ß-strands, which is flanked by three N-terminal {alpha}-helices and one C-terminal {alpha}-helix. There is only one small {alpha}-helix located within the ß-strands (201). The nearest structural neighbor is the PAS domain of the photoactive yellow protein from Halorhodospira halophila (28), which has a similar tertiary structure but shows no obvious sequence similarity. The ß-strand structure forms the bottom of the citrate/C4-dicarboxylate binding pocket, which is framed by the {alpha}-helices. Citrate is bound at three carboxylate-binding sites, C1 to C3 (Fig. 5). Each of the carboxylic groups is liganded by one basic amino acid residue (Lys, Arg, or His) and at least one further residue with a hydroxyl side chain (Ser and Thr). The residues form extensive hydrogen bonds to the carboxylic groups. The hydroxyl group of citrate is bound at the hydroxyl site by an Arg residue through hydrogen bonding. Replacement of most of the residues by mutation inactivates citrate binding and the function of CitA. The tight interaction and binding of citrate are reflected by high-affinity binding of H-citrate2– to the isolated periplasmic domain (75, 137).


Figure 5
View larger version (46K):
[in this window]
[in a new window]

 
FIG. 5. Structures of the periplasmic sensing domains of DcuS (A) and CitA (B). The structures for the periplasmic domains of CitA and DcuS are derived from http://www.rcsb.org. The residues required for C4-dicarboxylate sensing by DcuS (144, 201) and direct binding of citrate to CitA (75, 219) are shown. The corresponding sites are highlighted in the structure.

 
The C4-dicarboxylate binding site of DcuS is known at lower resolution but shows remarkable similarity to that of CitA (144, 201). In contrast to CitA, DcuS has a broad substrate specificity and accepts all C4-dicarboxylates; the dicarboxylic groups have to be present with a distance of 3.1 to 3.8 Å for efficient activation of DcuS. Small polar (i.e., hydroxyl, amino, or thiol) side groups at position C2 or C3 of the C4-dicarboxylates, such as in malate, aspartate, or thiosuccinate, have no negative effect. The carboxylate groups are essential but can be replaced by a nitro group, such as in nitropropionate. Residues Arg107, His110, Arg147, and Phe149 are essential for stimulation by the C4-dicarboxylates (144) and are homologous to the basic residues at C1 to C3 and the hydroxyl (H) site of CitA. This suggests that for binding of the dicarboxylates, similar sites as for binding of the C6-tricarboxylate citrate are required.

DcuS/CitA homologs are found in many different bacteria, mostly proteobacteria but also gram-positive bacteria (115). The DcuS homolog in B. subtilis (DctS or YdbF) represents an interesting variation (15). DctS has an extracytoplasmic domain (predicted to be 140 aa), which is similar in size to the domain of E. coli DcuS (135 aa). DctS, however, requires a periplasmic C4-dicarboxylate-binding protein for sensing C4-dicarboxylates (15), although the essential residues for C4-dicarboxylate binding by E. coli DcuS (Arg107, His110, Phe120, Arg147, and Phe149) are conserved (144).

As described above, the structures of subdomains have been determined for various membrane-bound HKs (CitA, DcuS, EnvZ, and PhoQ). However, for none of these sensors are high-resolution structures of all domains known, let alone the structure of the intact full-length protein. Therefore, detailed models for the mechanism of signal transduction in periplasmic-sensing HKs are lacking. Nevertheless, comparisons of ligand-bound and free periplasmic sensory domains enabled researchers to develop first ideas on the mode of ligand sensing and signal transduction across the membrane.

Proposed mode of signal perception and transduction by DcuS/CitA. Structural studies suggest that ligand binding may trigger conformational changes that are propagated into the TM helices, thereby presumably affecting kinase dimerization and/or activity. The N- and C-terminal ends of the periplasmic domains of DcuS and CitA consist of long {alpha}-helices, which protrude to the surface of the globular structure at the membrane-proximal side and presumably are directly continued in TM helices 1 and 2. In the CitA structure, determined with bound ligand (citrate), the N- and C-terminal helices have a nearly parallel arrangement. In contrast, the corresponding helices in DcuS have an inclined arrangement with respect to each other. This arrangement presumably does not reflect the situation in the intact protein and could be caused by the lack of the following TM helices 1 and 2 in the domain structure or by the lack of bound ligand. Conformational movements in the periplasmic domain upon binding of the stimulus may therefore be propagated into the TM helices, providing a mechanism for signal transduction into and across the membrane. As described above, the mechanism of signal transmission could be similar to the mode suggested for PhoQ, but the initial event is different. For CitA/DcuS, binding of the ligand is proposed to cause a conformational change in the periplasmic domain which is transmitted to the transmembrane helices. In contrast, PhoQ binding of the antimicrobial peptide is thought to lift the periplasmic domain from the membrane surface by displacing the Mg2+ ions. The movement could then be transmitted in a lever-like mechanism to the TM helices. Conformational movements in the TM helices could involve a piston stroke mechanism, as suggested for signal transfer in the MCP Tar (197). However, signal transfer by other mechanisms such as torsional movement, including rotation of the TM helices against each other, as suggested earlier for MCP (39), cannot be excluded.

Sensor Kinases with PBPb Sensing Domains

Periplasmic-sensing HKs with periplasmic binding-protein (PBPb) sensing domains are exemplified by BvgS and EvgS of Bordetella spp. and E. coli, respectively (21, 41, 148). The sensors harbor periplasmic domains reminiscent of high-affinity periplasmic solute-binding proteins that mediate substrate binding in ABC transporters. The PBPb domains (up to three can occur side by side in the periplasmic domain of a single HK of this type) are located between TMR1 and TMR2 (Fig. 3). On the basis of sequence similarities, the PBPb domains can be grouped in eight families that correlate with the nature of the substrate bound. PBPb domain family (or cluster) 3 is homologous to periplasmic binding proteins specific for polar amino acids and opines (252). Some HKs of this group show a complex architecture of their transmitter domain, including a PAS domain in the linker region and a receiver and HPt domain at the C-terminal end of the kinase. Other (albeit so-far-uncharacterized) members contain only the "standard" HisKA/HATPase_c domains in their cytoplasmic transmitter regions.

The model TCS of this class is BvgAS of Bordetella pertussis (148), which controls expression of virulence genes and biofilm development in this human-pathogenic bacterium, the etiological agent of whooping cough (148, 175). BvgS responds to environmental stimuli such as temperature, magnesium sulfate, or nicotinic acid, but the actual stimulus has not yet been chemically defined (148). The stimuli are thought to be perceived by the PBPb domains. The cytoplasmic PAS domains of BvgS and EvgS are sensitive to water-soluble quinone analogs in vitro (26). It has been suggested that BvgS and EvgS respond in this way to the redox state of the ubiquinone from the aerobic respiratory chain (30), reminiscent of the O2-sensing HK ArcB of E. coli (174) (see below). BvgS-homologous systems have been described for Bordetella species and other pathogenic bacteria, all of which regulate complex behavioral shifts to a pathogenic life style: EvgAS controls the expression of a drug efflux pump in E. coli (138); AstRS controls swarming motility, phase variation, and stationary-phase adaptation in Photorhabdus luminescens (46); and CblSRT regulates the expression of cable pili in Burkholderia cenocepacia (256).

Novel Conserved Periplasmic Sensing Domains: CACHE, CHASE, and Reg_prop

Recently, sequence analyses led to the identification of several new types of sensory domains found in TCS, MCP, and signal transducing adenylate cyclases (Table 1) (65). The CACHE domain (Ca2+ channels, chemotaxis receptors) was the first extracytoplasmic sensing domain in this group that was identified by multiple-sequence alignments of sensor proteins (6). Initial experimental data indicate that the CACHE domain is involved in small-molecule binding. The CACHE domains of McpB and McpC of B. subtilis are important for sensing amino acids and carbohydrates, respectively (71, 94). So far, only a few HKs with CACHE domains are found in the databases. The CACHE domain of the sensor kinase DctB is crucial for sensing dicarboxylates in Rhizobium leguminosarum (218).

Multiple-sequence alignments of extracellular regions from membrane-bound signaling proteins led to the identification of six different conserved extracellular sensing domains, termed CHASE to CHASE6 (for cyclase/histidine kinase-associated sensing extracellular) (7, 288). As their name indicates, these domains (length, 150 to 300 aa) are found in the N-terminal region of HKs, adenylate cyclases, and predicted diguanylate cyclases/phosphodiesterases. Additionally, CHASE2 domains occur in serine/threonine kinases, and CHASE3 domains occur in MCP. CHASE domains are flanked by predicted TMR and are often, but not necessarily, found in association with other conserved cytoplasmic signaling domains, such as PAS, GAF, or HAMP (Table 1). CHASE2, however, is always followed by three TM helices, but never a HAMP domain (Fig. 3). About 100 CHASE domain-containing HKs can be found in the databases. They are predominantly encoded in the genomes of cyanobacteria but can also be found in proteobacteria, gram-positive bacteria, and even in some archaea (7, 182, 288). So far, only one bacterial CHASE domain-containing HK has been analyzed with regard to its biological function: VsrA, a CHASE3 domain-containing HK, is required for the expression of virulence factors in Pseudomonas solanacearum (227). CHASE3-containing HKs can also be found in the genome sequences of other pathogenic bacteria, such as Legionella pneumophila and Bacillus anthracis (288). Analysis of the CyaA adenylate cyclase from Myxococcus xanthus indicates that the CHASE2 domain might function in osmosensing (288).

A novel type of periplasmic-sensing HK has been identified in the human gut symbiont Bacteroides thetaiotaomicron. The periplasmic input domain has a length of more than 1,000 amino acids and harbors 14 tandem repeats of a Reg_prop domain. Sequence homology indicates that these repeats are likely to form two seven-bladed ß-propellers (63, 213). The role of these structures for sensing (either directly or indirectly through protein-protein interaction) remains to be determined. These kinases are found in large numbers in this bacterium. It has been suggested that the Reg_prop HKs are crucial (together with a plethora of extracytoplasmic-function {sigma} factors) for the regulation of the organism's large repertoire of genes for the metabolism of complex polysaccharides in response to their availability in the gut environment (283). Similar HKs with Reg_prop domain containing input regions are so far only found in Xanthomonas campestris.

General Roles of Extracytoplasmic Sensor Domains in Stimulus Perception

As shown in this section, there is a remarkable flexibility and variation in the type of sensory domains of periplasmic-sensing HKs. In many cases, the role of the periplasmic domain in signal sensing is unproven, controversial, or even questionable, such as (and most notably) in the case of E. coli EnvZ (see above). Another prominent example is the phosphate sensor PhoR of B. subtilis. By domain architecture, this protein has all features of a classical periplasmic-sensing HK: the N terminus of this sensor kinase contains two TM helices flanking an extracytoplasmic domain of 120 aa. However, strains expressing PhoR derivatives in which the extracytoplasmic domain between the two TM domains, or even the complete N terminus up to the cytoplasmic side of TM helix 2, are deleted (PhoR is expressed as a soluble, cytoplasmic protein in the latter case) respond almost normally to PhoR-mediated phosphate limitation (233). The authors of this and another study on the homologous PhoR sensor kinase from E. coli could demonstrate that an extended cytoplasmic linker region, termed the C2 region, between TM helix 2 and the kinase domain is necessary and sufficient for sensing phosphate limitation (228, 233). Therefore, PhoR has to be regarded as a membrane-anchored, cytoplasmic-sensing HK rather than a periplasmic-sensing HK. These findings raise the possibility that the extracytoplasmic domains of some sensor proteins classified as periplasmic-sensing HKs are unnecessary for stimulus perception (102). Alternatively, some periplasmic-sensing HKs are mixed periplasmic/cytoplasmic-sensing HK that detect additional, so-far-unidentified stimuli.

Remarkably, different types of sensing domains have been adopted for the same or similar stimuli. This can be exemplified by the sensors for C4-dicarboxylates: the C4-dicarboxylate is perceived by a PAS-like domain in DcuS from E. coli, a four-helix bundle domain in Tar (aspartate) MCP of E. coli, and the CACHE domain in the succinate sensing DctB of Rhizobium leguminosarum (6, 114, 218, 219). Osmosensing is mediated by an even larger diversity of sensor proteins that even belong to completely different families: E. coli EnvZ is a prototypical periplasmic-sensing HK, while its functional homolog from Xenorhabdus nematophilus is a small intramembrane-sensing HK (see below). The osmosensors KdpD from E. coli and MtrB from Corynebacterium glutamicum are also HKs but are completely different from EnvZ (128, 179) (see below). In addition, the carriers BetP from C. glutamicum and ProU from E. coli and other organisms are also able to sense osmolarity (216, 226).

The recent identification of new conserved periplasmic domains by sequence analyses indicates that additional, so- far-uncharacterized periplasmic sensing domains might exist, which further extend the number and types of domains (and presumably stimulus perception mechanisms) used by periplasmic-sensing HKs. Thus, our understanding of signal perception and transmission by periplasmic-sensing HKs is still at the very beginning.


   HISTIDINE KINASES WITH SENSING MECHANISMS LINKED TO THE TRANSMEMBRANE REGIONS
 Top
 Previous
 Next
 References
 
In this highly diverse group of sensor kinases, the TM helices appear to play a central role in stimulus perception. The N-terminal input domains of all HKs listed in this section consist primarily of putative TMR, most of which have only short or no significant (extra)cytoplasmic linkers between these TMR (Fig. 6). Their grouping is based on their function and the number of TM helices but most likely also reflects their phylogenetic relationship, as indicated by the sequence conservation of their respective transmitter regions (Table 2). Six subgroups can be differentiated: (i) small sensor kinases with 2 TMR lacking any significant extracytoplasmic linker (LiaS/BceS-like intramembrane-sensing HKs), which are involved in sensing cell envelope stress or mediating ABC transporter-coupled detoxification processes in gram-positive bacteria; (ii) DesK-like thermosensors with 4 or 5 TMR that respond to membrane fluidity; (iii) RegB-like global sensor kinases with 6 TMR; (iv) quorum-sensing HKs with 6 to 10 TMR, again occurring primarily in gram-positive bacteria (AgrC/ComD-, ComP-, and LuxN-like HKs); (v) HKs with 12 to 20 TMR which show homology to transport proteins (CbrA- and PutP-like HKs); and (vi) HKs with "unknown conserved" input domains of 6 to 8 TMR (MHYT-, MASE1-, 7TMR-DISM-, and 5TMR-LYT-containing HKs). Proteins harboring these domains have been identified by bioinformatic analyses and are grouped according to their input domains.


Figure 6
View larger version (52K):
[in this window]
[in a new window]

 
FIG. 6. Domain architecture of histidine kinases with sensing mechanisms linked to the transmembrane regions. The figure is based on the graphical output of the SMART web interface at http://smart.embl-heidelberg.de (229), with modifications. The scale bar is in amino acids. Blue vertical bars represent putative transmembrane helices. Sizes and positions of conserved domains are indicated by the labeled symbols. Note that the transmitter domains are simplified, and as a default, only HisKA and HATPase_c are shown. See the text for details.

 
Intramembrane-Sensing HKs: Cell Envelope Stress Sensors with Two TMR (LiaS/BceS-Like HKs)

The term intramembrane-sensing histidine kinase (IMHK) was originally coined for sensor kinases of the cell envelope stress stimulon of B. subtilis (167). The HKs of the three TCS involved—LiaS, BceS, and YvcQ—share striking similarities in their domain organization that led to the initial definition: small sensor kinases of less than 400 aa in total length with a short sensing domain consisting of two deduced TM helices with a extracytoplasmic linker of less than 25 aa between them. It was proposed that these kinases sense their stimuli at or within the membrane interface, hence their name (167). Two major groups can be distinguished, based on genomic context and sequence conservation: (i) LiaS-like sensors as part of three-component systems and (ii) BceS-like kinases that are linked, by function and genomic context, to ABC transporters (166). Both groups occur almost exclusively in gram-positive bacteria with a low G+C content (Firmicutes).

LiaS-like HKs. So far, only two members of the LiaS-like subgroup of IMHKs have been described in detail: the eponymous sensor LiaS from B. subtilis and VraS from Staphylococcus aureus. The kinases of both TCS sense the presence of cell-wall-active antibiotics. LiaS is induced by antibiotics (including bacitracin, vancomycin, ramoplanin, and cationic antimicrobial peptides) that interfere with the lipid II cycle in cell wall biosynthesis (168, 210). The VraS sensor kinase responds to an even broader spectrum of cell wall antibiotics, such as glycopeptides, ß-lactams, bacitracin, and D-cycloserine (151). Two additional LiaRS homologs also respond to the presence of cell wall antibiotics: TCS03 of Streptococcus pneumoniae and YvqEC of Bacillus licheniformis are activated by the presence of sublethal concentrations of vancomycin and bacitracin, respectively (92, 271), indicative of a general role of all LiaS-homologous IMHKs in the detection of cell envelope stress. Recently, it was found that LiaF, a membrane protein, whose gene is topologically linked to liaSR in the genomes of all species harboring LiaS homologs, is crucial for the LiaS-mediated sensing mechanism: in a liaF deletion mutant, the LiaRS system is constitutively "on," thereby no longer necessitating a stimulus for full activity (120). Therefore, LiaF, together with LiaRS, constitutes a cell envelope stress-sensing three-component system. The corresponding HKs are phylogenetically conserved and belong to the HPK7 subfamily (166).

BceS: a two-component system-ABC transporter connection. The largest distinct group of IMHKs is characterized by the location of their structural genes adjacent to (usually upstream of) those encoding ABC transporters. About 70 proteins belong to this phylogenetically conserved subgroup (HPK3i) (Table 2), and almost all are from gram-positive bacteria with a low G+C content (166). The genes encoding the TCS and the ABC transporter are expressed as independent transcriptional units and together form specific and efficient detoxification units: It is thought that the HK senses the presence of sublethal concentrations of harmful compounds (such as the cell wall antibiotic bacitracin) and activates its cognate RR, which in turn strongly induces the expression of the adjacent ABC transporter, which subsequently facilitates removal. Such a topological and functional link of the TCS and ABC transporter operons was observed earlier in the Bacillus/Clostridium group of low-G+C gram-positive bacteria and is well documented for B. subtilis (121, 122, 166, 167, 185, 194).

Miscellaneous IMHKs. About 50 additional, phylogenetically unrelated proteins from Firmicutes, Actinobacteria, and Proteobacteria share the overall domain architecture of IMHKs. So far, only a few examples have been described in any detail. The GtcRS is located adjacent to the grsAB operon of Bacillus brevis, which encodes enzymes of the biosynthesis of the peptide antibiotic gramicidin S (260). The SaeRS TCS is part of a complex regulatory network that controls the expression of virulence determinants in S. aureus (81, 82). So far, only one IMHK from gram-negative bacteria has been investigated: EnvZ of Xenorhabdus nematophilus is homologous to the "classical" osmosensor EnvZ of E. coli in its cytoplasmic C-terminal domain, but it lacks an extracytoplasmic domain. While the periplasmic domains have diverged extensively, EnvZ from X. nematophilus was still able to complement a {Delta}envZ mutant of E. coli to sense changes in environmental osmolarity and properly regulate the phosphorylation levels of the cognate RR OmpR of E. coli (251).

IMHK-like periplasmic-sensing HKs: VanS/PrmB-like proteins. Two phylogenetically unrelated groups of cell envelope stress-sensing HKs show striking similarities to IMHKs with regard to their overall size and domain organization: (i) small VanS-like HKs of the VanB type, the sensors of vancomycin resistance in gram-positive bacteria, and (ii) PmrB/BasS-like HKs mediating resistance to cationic antimicrobial peptides in E. coli and Salmonella spp. (89, 90, 282). However, in both cases experimental evidence points towards a role of the short extracytoplasmic linker between the two TMR in signal perception. VanS-like HKs interact with vancomycin through their short extracellular sensing domain (25 to 30 aa), presumably by binding the drug directly (104, 106). The same is true for PmrB/BasS-like HKs, with a periplasmic linker of 30 to 35 amino acids between the TMR. It has been demonstrated that PmrB senses ferric iron through two conserved ExxE motifs in this short extracytoplasmic sensor domain (281). Therefore, both VanS- and PrmB-like proteins are periplasmic-sensing HKs (Fig. 3).

In that regard, the 25-residue cutoff value for the periplasmic linker might indeed be a critical threshold to differentiate "true" IMHKs from related but periplasmic-sensing HKs involved in cell envelope stress response (166).

DesK-Like Thermosensors with Four to Six TMR

The cytoplasmic membrane is the primary selective barrier between a bacterial cell and its environment. Its function critically depends on the physical state of the lipid bilayer, which in turn is strongly influenced by the temperature. Low temperature leads to a phase transition from a disordered liquid-crystalline phase to a more rigid and ordered gel-like phase. In order to maintain a functional membrane under such conditions, cells can lower the melting point of their membranes either by incorporation of de novo-synthesized branched-chain fatty acids or by desaturation of fatty acid moieties in the existing membrane. The mechanism of temperature-dependent adaptation of membrane fluidity and its regulation have been studied intensively in the two paradigmatic species E. coli and B. subtilis and have been reviewed recently (164).

In B. subtilis, regulation of membrane fluidity is mediated by the cold shock-inducible expression of the des gene, encoding a fatty acid desaturase (2, 4). Its expression is controlled by the DesRK TCS. At higher growth temperatures (membrane in the liquid state), the bifunctional sensor kinase DesK is predominantly in the phosphatase state, thereby repressing des expression. At lower temperatures, DesK switches to a dominant kinase activity, thereby activating its cognate RR DesR, which results in a strong induction of des expression (3). It has been suggested that DesK senses these temperature changes through its hydrophobic N-terminal region, which consists of four or five TM regions (105). Experimental evidence points towards membrane fluidity as the stimulus of DesK-mediated des expression, which requires the presence of the N-terminal TM region of the protein (42). However, the mechanism of temperature sensing by this N-terminal input domain has not yet been addressed experimentally.

CorS of Pseudomonas syringae is another thermosensor that is not homologous to DesK. It resembles DesK with regard to the N-terminal hydrophobic input domain, consisting of six TM regions, with only very short or no periplasmic linkers in between (DesK-like HK; four to five TM regions) (Fig. 6). In contrast to the case for DesK, which is conserved in the Firmicutes group of gram-positive bacteria, no CorS homologs can be found in the databases. CorS regulates the thermoresponsive production of the phytotoxin coronatine, a polyketide that is synthesized predominantly at lower temperatures (31). LacZ/PhoA fusions confirmed the presence of six TMR, and deletion analysis indicated an important role of the N-terminal TM helix for thermosensing (237). However, as for DesK, it remains to be investigated whether membrane fluidity (a physical property of the membrane) or its chemical composition serves as a trigger for CorS activation.

RegB/PrrB-Like Redox-Responding Global Sensor Kinases with Six TMR

RegBA and PrrBA were originally identified as TCS involved in the anaerobic induction of the photosystem in Rhodobacter capsulatus and Rhodobacter sphaeroides 2.4.1, respectively (56, 181). RegBA and PrrBA are global regulatory systems that are highly conserved in photosynthetic and nonphotosynthetic alpha- and gammaproteobacteria and regulate energy-generating and -utilizing processes such as photosynthesis, carbon fixation, hydrogen oxidation, denitrification, aerobic and anaerobic respiration, electron transport, and aerotaxis in response to redox energy or catabolic [H] supply of the cell (55). RegB and PrrB are bifunctional proteins, exerting both kinase and phosphatase activities. While the kinase activity is regulated by the redox state of the electron transfer chain, the phosphatase activity is assumed to be constitutive (215). The input domains of RegB/PrrB consist of six TM helices (38, 198), which are required for stimulus perception (215). A mutation in the cytoplasmic loop between TM helix 2 and TM helix 3 causes constitutive (i.e., oxygen-insensitive) kinase activity in vivo (56). The aerobic respiratory chain has been shown to supply the signal by interaction of the cytochrome cbb3 oxidase with PrrB/RegB (55, 191-193). In a defined in vitro system, cytochrome oxidase cbb3 stimulated the phosphatase/kinase activity ratio of PrrB in the presence of O2, resulting in a decrease of PrrA-phosphate and thus of the active state of PrrA (193). Sensing and optimal kinase activity required the TM sensing domain of PrrB.

For optimal signal transfer and sensing by RegB/PrrB, the gene products of the senC/prrC genes, which are cotranscribed with regA, are required. SenC/PrrC are predicted membrane-spanning copper-binding proteins, and their inactivation results in an oxygen-insensitive phenotype (57, 170). SenC is suggested to modulate RegB activity or to form a signaling link between cytochrome cbb3 oxidase and RegB (192, 249). RegB contains a redox-sensitive cysteine residue in the cytoplasmic domain which becomes oxidized in the presence of oxygen, resulting in the formation of an intermolecular disulfide and an inactive RegB tetramer (248, 249). The oxidation of the cysteine residue by oxygen might occur via cytochrome cbb3 oxidase or directly.

In R. sphaeroides, the expression of photosynthesis genes is regulated by light in addition to oxygen. The regulation by light is initiated by the photosynthetic reaction center and transmitted by the electron transport chain and the PrrBA TCS. In addition, there is a blue-light photoreceptor, AppA, which interferes with the PrrBA-dependent regulation (95). Finally, it has been suggested that RegB in R. capsulatus may respond to additional stimuli, such as the flow of intermediates through the bacteriochlorophyll biosynthetic pathway (1).

In summary, RegB is a well-characterized redox sensor, which senses the redox state of the aerobic electron transport chain by interaction with cytochrome oxidase, in contrast to other O2 sensors, which interact with molecular O2 (such as FixL) or with the respiratory quinones (i.e., ArcB [see below]). Still, many mechanistic details, such as the role of the TM helices in signal perception and in the interaction between PrrB/RegB and the oxidase, have to be elucidated.

Peptide Quorum Sensors with 6 to 10 TMR (AgrC/ComD-, ComP-, and LuxN-Like HKs)

AgrC/ComD- and ComP-like HKs are sensors of peptide-dependent quorum-sensing systems, a widespread mechanism of cell density-responsive regulation in gram-positive bacteria. ComP-like HKs differ in the architecture of the input domain from those of the AgrC/ComD group (Fig. 6): ComP-like HKs tend to have 8 to 10 TM regions (based on computer predictions), and some include a larger (about 50-aa) extracellular loop between TM helix 1 and TM helix 2 that is important for signal detection, while AgrC/ComD-like HKs seem to have six TM regions and lack the extracellular loop region. The two groups are also phylogenetically distinct: ComP-like HKs belong to the HPK7 subfamily, whereas AgrC/ComD-like HK exclusively comprise the very unique HPK10 subfamily, which has already been recognized as being restricted to peptide quorum-sensing HKs (98, 280). Within the latter group, AgrC- and ComD-like kinases can be differentiated by the chemical nature of their stimuli: AgrC-like HKs sense cyclic thiolactone-containing autoinducing peptides (AIPs), while ComD-like sensors bind unmodified GG leader-type peptides (98, 160). The precursors of these signaling peptides harbor a characteristic double-glycine leader that is removed concomitant with its export (184). A number of excellent reviews on this topic have been published (48, 51, 141, 142, 160, 232, 247). Therefore, we will concentrate our description on stimulus detection only. The systems share a common mechanism for intraspecies communication. The signaling peptides are expressed at a constant low level as inactive cytoplasmic precursors. They are actively secreted, usually after posttranslational modifications. Examples include cleavage of a leader peptide (S. pneumoniae CSP), introduction of lanthionine bridges (lantibiotics such as subtilin or nisin), or isoprenylation (B. subtilis ComX). Peptide processing and secretion are often coupled processes. The active signaling peptides accumulate in the environment, and when they reach a certain threshold concentration, their presence is detected by the corresponding HK. These proteins, in turn, activate their cognate RR, resulting ultimately in a differential expression (usually upregulation) of the target genes, including the regulatory system, the signaling peptide, and its export/modification/sensing system (positive feedback loop). The concentration of the stimulating molecule is therefore a measure of the cell density of organisms having the same pherotype (i.e., those that are able to excrete and sense the same peptide species). Activation of the corresponding TCS leads to a sudden burst in peptide release, thereby ensuring that the cells of a culture become activated at the same time to achieve synchronization. Examples include the development of competence for genetic transformation in the genera Bacillus (ComX-ComAP) and Streptococcus (ComC-ComDE), transition to the virulence state in Staphylococcus (AgrBCD), and bacteriocin production in various Firmicutes bacteria (48, 51, 141, 142, 160, 232, 247).

AgrC is the sensor kinase of one of the four known TCS regulating staphylococcal virulence. AgrC mediates quorum sensing and is activated upon binding of the self-encoded AIP. The interaction between the signaling peptide and the sensing domain of S. aureus AgrC, the sole receptor of AIPs in staphylococci (116), has been studied in great detail. Localization, membrane topology, and enzymatic kinase function demonstrate that AgrC is a membrane-bound HK with six TM helices (157). Extensive studies on AgrC architecture and AIP/AgrC interaction resulted in the model of a two-step process of stimulus perception. First, the AIP interacts with the sensor domain by entering a hydrophobic pocket formed by the TM helices in a non-sequence-specific manner. In a second step, sequence-specific hydrophilic interactions between the AIP and amino acid residues in the last extracellular loop of the AgrC sensing domain leads to activation of the kinase and autophosphorylation (160). The sequences of AIPs (agrD genes) and of the corresponding agrB and agrC genes, which encode proteins for the export and the TCS, are highly variable. They comprise a functional unit that defines the pherotype of an individual strain (50). A genetic and pherotypic polymorphism of all peptide-specific parts of the signal perception mechanism has also been described for the competence quorum-sensing systems in S. pneumoniae and other streptococci (97, 274) and B. subtilis (258, 259). The AgrC sensor kinases detect only peptides of their own pherotype, and this specificity is based on the TM-sensing domain of the kinase alone: engineered fusion kinases composed of heterologous N-terminal input and C-terminal transmitter domains responded only to the peptide that corresponded to the pherotype of the sensor domain (118, 161). The membrane topology and mechanism of stimulus perception of two additional quorum-sensing HKs of this class, PlnB and ComP, have been investigated to some extent.

PlnB is a sensor kinase regulating transcription of the pln bacteriocin biosynthesis locus of Lactobacillus plantarum. It responds to the extracellular presence of the inducing peptide pheromone IP-C11 and activates the two cognate RRs PlnC and PlnD, which regulate the transcription of the bacteriocin operons plnEFI and plnJKLR as well as the regulatory plnABCD operon itself (47). A membrane topology with six TM helices is predicted for PlnB, but experiments suggest the presence of seven TM helices with a large N-terminal extracytoplasmic loop of 24 amino acids, as has also been predicted for the competence sensors ComD from S. pneumoniae (98, 118). The peptide pheromone interacts with the membrane-integral N-terminal input domain (118), and three amino acid residues (Asp54, Ser58, and Leu61) in the extracellular loop between TM helix 2 and TM helix 3 that are important for peptide/sensor interaction have been identified (119). The corresponding residues are also critical for ComD activity, indicating a common mechanism of peptide sensing, which could be similar to that described for AgrC (119). Thus, peptide quorum sensing apparently occurs by a common mechanism involving nonspecific hydrophobic interaction at the TMR, as well as specific interactions of the inducing peptide with polar residues in the extracytoplasmic loop of the sensor domain (situated between TM helices 2 and 3 for PlnB and ComD and between TM helices 3 and 4 for AgrC).

The ComPA TCS is required for the development of competence for genetic transformation, a process embedded in the complex developmental program during the adaptation of B. subtilis to nutrient limitations (183). ComP senses the presence of the ComX pheromone, a modified decapeptide (8), and in turn activates its cognate RR ComA. This ultimately leads to the activation of the competence transcription factor ComK, which controls the expression of genes encoding the DNA uptake and recombination machinery (37, 93). Membrane topology studies based on PhoA/LacZ fusion studies suggest that the input domain of ComP consists of eight TM helices. Two periplasmic linkers in the first four TM regions are important for binding of the ComX pheromone (208). A domain-based sequence analysis predicts the presence of 10 TM helices for three ComP homologs, with only one large extracellular loop remaining (equivalent to the first extracellular loop [Fig. 6]). This large loop contains a PDZ domain (domain present in PSD-95, Dlg, and ZO-1/2) with a length of about 70 amino acids. PDZ domains occur primarily in higher eukaryotes but have also been identified in bacterial proteins such as the cell envelope stress protease HtrA (214). They are involved in binding of (poly)peptides, indicative of a role in ComX binding. Therefore, the domain architecture of some ComP proteins, such as the one analyzed by Piazza et al. (208), closely resembles that of LuxN-like kinases (Fig. 6). The role of the PDZ domain in peptide sensing in the ComP-like sensors has not been addressed so far.

Interestingly, LuxN/CqsS-like quorum sensors from gram-negative bacteria, which respond to homoserine lactone, show a very similar domain architecture: the input domains of both kinases consist of eight or nine putative TM regions (Fig. 6). While the corresponding quorum-sensing systems have been investigated intensively with respect to signal tr