Bacu.rial Strct~ Responses, 2nd cd. Edited by Gisela St<>rt. and ReAine llcngge t'> 2011 ASM Press, Woshmgton, T>C t

Chapter 4

The Role of Two-Component Signal Transduction Systems in Bacterial Stress Responses

MICHAEl. T. . LAUB

Two-component signal transduction systems represent (Fig. lA). Receipt of an input signal by the histidine one of the primary means by which bacteria sense and kinase stimulates its autophosphorylation whereby respond to change in their environments, both intra- and the gamma-phosphoryl group of an ATP molecule is extrace/htlar. These versatile signaling systems have been transferred to a conserved histidine residue. Histidine implicated in regulating a wide range ofstress responses, kinases are often integral membrane proteins with a from nutrient starvation to cell envelope stress and pro­ periplasmic domain that receives a signal and then tein misfolding to antibiotics and many more. Two-com­ transduces this signal across the membrane to trig­ ponent signaling proteins are found in nearly all bacte­ ger a change in autophosphorylation activity. Some ria, with most species encoding dozens and sometimes kinases are entirely cytoplasmic and use domains lo­ hundreds of these proteins. These molecules are crucial cated upstream of the catalytic domains to sense in­ players in allowing bacteria to adapt and survive in the put signals. An autophosphorylated kinase becomes a face of many different stresses and environments. The phosphodonor substrate for a cognate response regu­ sensor histidine kinase can respond to intra- or extracel­ lator that catalyzes transfer of the phosphoryl group lular cues by catalyzing the phosphorylation of cognate on the kinase to a conserved aspartic acid on itself. response regulators, which are then capable of modify­ Phosphorylation of the response regulator typically ing gene expression or cellular physiology to help an induces a conformational change leading to activation organism cope with and survive changes in its environ­ of an output domain, frequently a DNA-binding do­ ment. As two-component signaling proteins are conspic­ main. Two-component signaling pathways thus often uously absent in animals and humans, they have begun culminate in gene expression changes, although, as to garner significant attention as possible antibiotic tar­ discussed later, this is not always the case. When not gets. Since their discovery more than 20 years ago, we stimulated to autophosphorylate, the histidine kinases have learned how specific pathways work at the cellu­ can also act as phosphatases for their cognate response lar, molecular, and atomic levels. More recent work has regulators. Or, to be precise, histidine kinases acceler­ also begun to reveal general design principles governing ate the intrinsic dephosphorylation rate of the aspartyl two-component pathways and the means by which cells phosphate on their cognate response regulators, but, coordinate the activity ofmany highly related pathways. for simplicity, are referred to as phosphatases. Most This chapter summarizes the state of our understand­ histidine kinases are thus considered bifunctional, act­ ing of two-component pathways on all levels, beginning ing as both kinascs and phosphatases. with the initial input stimulus through the final output Although many two-component signaling path­ response. The goal, however, is not to comprehensively ways follow this general paradigm, there are numer­ review all two-component pathways- which is infeasi­ ous variations. One common variant is the so-called ble here-but, instead, to demonstrate the key principles phosphorclay (Fig. 1B), which initiates with the au­ and paradigms for how these signaling pathways work tophosphorylarion of a histidine kinase that then by drawing on specific, illustrative examples. phosphotransfers to a response regulator just like in a canonical pathway. The phosphoryl group is sub­ The prototypical two-component signal trans­ sequently shuttled to a histidine phosphotransferase duction system involves a sensor histidine kinase and and ·then to a second, terminal response regulator its cognate response regulator (Stock et al., 2000) that ultimately triggers an output or physiological

Michael 'T. Laub • Department of Biolo~;y, Massachuscrrs Institute of Technology, 31 Ames St., Cambridge, MA 02139.

45 46 LAUE

A ATP~ H-P D-P I input~ _..output

histidine kinase response regulator

B

H-P 0-P H-P 0-P I input~ ~output I ;

hybrid histidine kinase histid ine phosphotransferase response regulator Figure 1. (A) Schematic of the canonical rwo-component signaling system. In response to an input signal, the sensor histidine kinase uses ATP co autophosphorylate. The phosphoryl group is then transferred to a cognate response regulator ro trigger an ourput. The symbol -P indicates phosphorylation of the conserved histidine and aspartate in the histidine kinase and response regulator, respectively. (B) Schematic of a phosphorelay. After aurophosphorylarion, the hybrid histidine kinase will transfer the phosphoryl group inrramolecularly to a response regulator-like domain. A hisridine phosphotransferase then shuttles the phosphoryl group to a soluble response regulator that affecrs an output. ln some cases, the hybrid kinase is split into a canoni­ cal histidine kinase and soluble response regulator (sec, for example, Fig. 2A). response. In many cases, the first response regulator These proteins are also found in some eukaryotes, of a phosphorclay has only the phosphorylatablc re­ such as plants, slime molds, and yeasts (Loomis et al., ceiver domain and no output domain. And, frequently, 1998}, but they arc conspicuously absent from animals, this receiver domain is fused to the C-terminus of the including humans, suggesting they may be valuable initiating histidine kinase, forming a so-called hybrid new antibiotic targets (Stephenson and Hoch, 2002; histidine kinase, which phosphotransfers intramolec­ Watanabe et al., 2008}. In fact, many pathogenic bac­ ularly before serving as a phosphodonor for the histi­ teria rely heavily on two-component signaling path­ dine phosphotransferase. Unlike histidine kinases and ways to adapt to life inside a host and to upregulate response regulators, histidine phosphotransferases genes critical to their virulence and the avoidance of do not constitute a single paralogous protein famHy, host immune systems. but they arc structurally similar to one another and to the domain that is autophosphorylared in histidine kinases (Kato et al., 1997; Varughese et al., 1998; HISTORICAL BACKGROUND Tomomori et al., 1999; Xu and West, J 999; Ulrich et a!., 2005; Xu et a!., 2009). It is not clear whether Two-component signaling proteins were first phosphorelays offer any advantage relative to canoni­ characterized in diverse genetic screens for Escherichia cal two-component pathways, but it has been sug­ coli mutants that could not properly chemotax, adapt gested that phosphorelays offer additional points of to nitrogen starvation, or respond to cell envelope control, enabling the integration of multiple signals. stress and in screens for B. subtilis mutants that could Phosphorclays arc often associated with major cellu­ not sporulate. Each of these' screens identified genes lar differentiation processes such as cell cycle transi­ involved in executing a given response as well as two­ tions in Caulobacter crescentus (Biondi et a!., 2006}, component signaling genes that regulate that response. the decision to sporulate in Bacillus subtilis (Burbulys for instance, early studies of nitrogen starvation in ct al., 199'1 ), or the switch from individual tO collec­ F. coli identified ginA, which encodes glutamine syn­ tive, or quorum, behavior in Vibrio harveyi (Freeman thetase and is necessary for responding to changes in and Rassler, l 999a; Freeman and Rassler, 1999b}. nitrogen availability, as well as glnG (ntrC} and gln/, Two-component signaling proteins arc among (ntrB}, which comprise a prototypical two-component the most prevalent signaling molecules in the bacte­ signaling pathway rhat regulates the transcription of rial kingdom and represent a primary means by which ginA and other genes required for adapting to nitrogen bacteria sense and respond to a range of stresses and starvation (Ninfa and Magasanik, 1986; Keener and environments. Most bacterial species encode dozens, if Kustu, 1988; Ninfa eta!., 1988; Weiss and Magasanik, not hundreds, of histidine kinases and response regula­ 1988). lt was quickly recognized that the regulatory tors (Galperin, 2005), underscoring their importance. genes from these different geaetic screens encoded CHAPTER 4 • THE ROLE OF TWO-COMPONENT SIGNAL TRANSDUCTION SYSTEMS 47 homologous proteins (Ferrari et al., 1985; Stock et al., typically contain two domains: a receiver domain 1985; Nixon et al., 1986). Subsequent biochemical that contains the conserved aspartate phosphoryla­ studies revealed the general mechanisms underlying tion site and an output, or effector, domain (Fig. 3A). two-component signal transduction systems (Ninfa Although the most common output domain is a and Magasanik, 1986; Hess et a!., 1987; Hess et al., DNA-binding domain that allows response regu­ 1988; Keener and Kustu, 1988; Ninfa et al., 1988; lators to control gene expression, there are at least Oosawa et al., 1988; Weiss and Magasanik, 1988; 10 additional types of output domains (Galperin, Igo et al., 1989a; Igo eta!., 1989b) and, importantly, 2006). Whatever the output domain, structural and demonstrated that these proteins all utilize a common biochemical studies suggest a common mechanism phosphotransfer mechanism, consistent with their be­ by which phosphorylation activates all response reg­ ing homologous proteins. for example, at high con­ ulators (Stock et al., 2000; West and Stock, 2001). centrations, the histidine kinase NtrB was shown, both Receiver domains all adopt a conserved a/13 fold in in vivo and in vitro, to phosphorylate the chemotaxis which a five-stranded 13-sheet is surrounded by five regulator CheY as well as its cognate substrate NtrC amphipathic alpha helices. Phosphorylation of the (Nin.£a eta!., 1988). conserved aspartate triggers conformational changes Early studies of sporulation mutants in B. sub­ that propagate through a series of highly conserved tilis identified multiple two-component signaling residues stretching from the phosphorylation site to proteins required for the initiation of sporulation. the a4-135-a5 face of the molecule. Conformational Elegant biochemical studies with these proteins led changes in this region of the receiver domain then to the identification of the first phosphorelay. This drive homodimerization or changes in protein-protein four-step signallng pathway initiates with a histidine interactions, ultimately enabling response regulators kinase, KinA, which then triggers a cascade of phos­ to affect cellular physiology. photransfer events from KinA to the response regula­ A systematic bioinformatic analysis of more than tor SpoOF to the histidine phosphotransferasc SpoOB 4,500 response regulators classified effector domains and, finally, to the master regulator of sporulation, into two broad categories: DNA-binding and those SpoOA (Burbulys eta!., 1991) (Fig. 2A). harboring enzymatic activities (Galperin, 2006; Gao These early studies laid the foundation for our et al., 2007). Although, as noted, at least 10 distinct understanding of two-component signaling pathways effector domains have been identified. In addition, and, since then, additional histidine kinases and re­ many response regulators, such as the chemotaxis sponse regulators have been unearthed in countless regulatOr CheY, have only a receiver domain and genetic screens. We have also begun to identify and use conformational changes in this domain alone to characterize numerous factors that influence the ac­ modulate interactions with target proteins (Jenal and tivity of these key signaling molecules, as discussed Galperin, 2009). in this chapter. In addition, because two-component More than 60% of all response regulators con­ proteins share significant homology to one another, tain DNA-binding domains; two-component sig­ the sequencing of complete bacterial genomes has en­ naling pathways thus represent one of the primary abled their comprehensive cataloging in htmdreds of means by which bacteria can couple changes in their organisms. These cataloging and related bioinformatic environments or intracellular states to changes in gene efforts have shed new light on both the diversity and expression. The DNA-binding effector domains can commonalities of two-component pathways. They ibe further categorized based on sequence similarity have also enabled global analyses that are starting to into three major classes: OmpR-like, NarL-like, and reveal how organisms simultaneously coordinate so NtrC-like (Fig. 3A). There are diverse mechanisms by many highly related proteins. This chapter discusses which phosphorylation induces changes in activity of each aspect of two-component signaling pathways, these DNA-binding response regulators, even within working back from the myriad output responses tO a group (Gao et al., 2007). In some cases, such as E. the flow of phosphoryl groups within pathways and coli NarL and B. subtilis SpoOA, the receiver domain finally to the initial sensing of signals and inputs. plays an autoinhibitory role such that its phospho­ rylation relieves inhibition and then permits DNA­ binding (Ireton et al., 1993; Baikalov et al., 1996). OUTPUTS In other cases, such as Fix], phosphorylation of the receiver domain drives the dimerization of a response Response regulators control an enormous ar­ regulator, leading to tighter binding of target promot­ ray of physiological, metabolic, and morphological ers (Da Re et al., 1999). And, finally, in other cases, processes in bacteria, a testament to the versatility such as PhoB and NtrC, receiver domain phosphory­ of two-component signaling proteins. These proteins lation triggers rearrangement from an inactive dimer 48 LAUB

A starvation signals ? ? ?

replication stress ~Sda __, t. ~~-----,l.----~~-----'1-,. ' ' KinA KinB KinC KinD KinE nlt•ogen ovoll•bmty? ...,._Kipi _..-\ ~ ~ h-/

cell population density ~Ph rA ---t RapA ---i SpoOF 1---RapB

YnzD~ ...... -' SpoOA 1--- SpoOE Yisl ~

sporulation genes

B c quorum signals (autoinducers)

HAI-1 Al-2 CAI-1 ~ ~ ~ LuxN LuxCVLuxP CqsS "" ~ /

ChpT ~ t CpdR CtrA ...-/ I\, cell cycle genes origin of replication quorum-regulated genes Figure 2. Schematics of the phosphorelays controlling (A) sporulation in B. subtilis, (B) quorum-sensing in V. harveyi, and (C) the cell cycle in C. crescentus. The lwo-componcnl signaling proteins in each panel arc shaded in gray. Also shown arc the auxiliary proteins rhat regulate phosphate flow through the pathway (see text for details), as well as the inputs and outputs for each palhway.

or multimer stare to an alternative and active dimer sire. Most response regulators that activate transcrip­ or mulrimer stare (Wyman et al., 1997; Bachhawat tion arc thought to do so by stimulating recruitment et al., 2005). Response regulators can either activate of RNA polymerase. The NtrC subfamily, however, or repress transcription, usually depending on the contains an AAA + domain (Fig. 3A) that has i\TPasc promoter architecture and the location of the regula­ activity, enabling these regulators to catalyze closed tor's binding site relative to the transcriptional start to open complex formation in RNA polymerase that CHAPTER 4 • THE ROLE OF TWO-COMPONENT SIGNAL TRANSDUCTION SYSTEMS 49

A signaling (Fig. 3A). This includes so-called GGDEF domains, which are typically diguanylate cyclases receiver OmpR-Iike that synthesize c-di-GMP, and EAL and HD-GYP domains, which function as phosphodiesterases for c-di-GMP (Jenal and Malone, 2006). In most bacte­ receiver Narl-Jike ria, the production of c-di-GMl> plays a critical role in the transition from a motile or planktonic growth phase to a sessile or biofilm state. Precisely how c-di­ receiver NtrC-Iike GMP influences cellular physiology is not yet entirely clear, but it likely involves the binding of this small molecule to specific target proteins that then control receiver various metabolic, motility, or regulatory processes (.Jenal and Malone, 2006). The decision to transi­ tion from a planktonic to a biofilm state is influenced receiver by a wide range of signals, but is often coordinated by two-component signaling pathways that couple ( receiver ) changes in environmental conditions to the produc­ I tion of c-di-GMP by phosphorylating response regu­ ·- lators with GGDEF output domains. For instance, in Pseudomonas aeruginosa, a chemotaxis-like system B drives the phosphorylation of the response regulator WspR that, in turn, stimulates its diguanylate cyclase activity and the accumulation of intracellular c-di­ GMP (Hickman et al., 2005). Similarly, in C. cres­ centus, the differentiation of a motile, swarmer cell to a sessile, stalked cell requires a response regulator, called PleD, which contains a GGDEF output domain (Paul et al., 2004; Hecht and N ewton, 1995). The diguanylate cyclase domain of PleD functions as a di­ mer and phosphorylation of the receiver domain is thought to increase c-di-GMP production by driving protein dimerization (Wassmann et al., 2007). WspR and other GGDEF-containing response regulators Figure 3. (A) Domain architecture of response regulators. Six of likely work by a similar mechanism. the mosr common domains found·a djacent to the receiver domain that is phosphorylated arc shown (Galperin, 2006). Three are in­ Finally, as noted, many response regulators con­ volved in DNA-binding and the regulation of transcription: wHTH tain only the receiver domain with no additional or (winged helix-turn-helix), H'IH (helix-turn-helix), and AAA + discrete output domain (Fig. 3A). The prototypical ATPase/FIS. ·r\vo are associated with cyclic-di-GMP signaling­ member of this subfamily is the chemotaxis regula­ GGDEF and EAL domains-each named for a conserved patch of tor CheY where phosphorylation induces a confor­ residues that synthesize and degrade c-di-GMP, respectively. Many response regulators are also single-domain proteins and harbor mational change in the et4-135-et5 face of the receiver only a phosphorylatable receiver domain. (B) Domain architecture domain that, in turn, modulates its ability to inter­ of histidine kinascs. The four roost common domains found adja­ act with the flagellum and control swimming behav­ cent ro the DHp and CA domains are PAS, HAMP, GAF, and TM ior (Cho et al., 2000; Lee et al., 2001a; Lee et al., (transmembrane) domains (Galperin, 2006). In many cases, the 2001b). The diverse roles for other single-domain linker between the input and DHp domains is a so-called signaling, or S, helix. Note that the diagram only indicates the most proximal response regulators are just now beginning to be dis­ domain, bur many histidine kjnases contain multiple domains N­ sected but they are likely to regulate a diverse range terminal to the DHp domain. of processes beyond chemotaxis (Jenal and Galperin, 2009). In C. crescentus, two single-domain response regulators are critical to cell cycle progression. One is is associated with the alternative a factor a54 (Kustu the single-domain response regulator DivK (Fig. 2C), et a!., 1991 ). The NtrC subfamily of regulators can which is essential for viability and for cell cycle pro­ also often bind at a distance and are often referred to gression (Hecht et al., 1995; Hung and Shapiro, 2002; as enhancer-binding proteins. Biondi et a!., 2006). Phosphorylated DivK is required After DNA-binding domains, the most common for the G1-S transition by downregulating the phos­ effector domains are those involved in cyclic-di-GMP phorelay that initiates with the histidine kinase CckA 50 LAUB and culminates in activation of the master regula­ Another classic example of a many-to-one rela­ tor CtrA (Biondi et al., 2006) (Fig. 2C). DivK has tionship in two-component signaling is the multiple also been shown to allosterically stimulate the au­ hybrid histidine kinases in V. harveyi that converge tophosphorylation of its cognate kinase DivJ and to on the histidine phosphotransferase LuxU to regu­ switch its cognate phosphatase PleC into the kinase late the quorum sensing response of this organism state (Paul ct al., 2008). These data suggest that (Fig. 2B). Under conditions of low cell density, the DivK has two independent binding sites on DivJ and hybrid kinases LuxN, LuxQ, and CqsS can each au­ PleC, one for phosphotransfer and one for modulat­ tophosphorylate and drive the phosphorylation of ing autophosphorylation activity. Yeast two-hybrid LuxU, which in turn can phosphorylate the response screens indicate that DivK also binds to the histidine regulator LuxO (Freeman and Bassler, 1999a; Free­ kinase DivL, an interaction that may be crucial to man and Bassler, 1999b; Freeman et al., 2000; Henke the role DivK plays in controlling CtrA (Ohta and and Bassler, 2004). Phosphorylated LuxO ultin1ately Newton, 2003). Another important single-domain activates the expression of five small RNAs that in­ response regulator in Caulobacter, called CpdR, was hibit translation of LuxR, a key transcription fac­ first discovered in a systematic deletion study of two­ tor involved in quorum-dependent behavior. Under component genes (Skerker et al., 200.5). It was subse­ conditions of high cell density, three different auto­ quently shown to control the localization and activity inducers can accumulate and switch LuxN, LuxQ, of the ClpXP protease which degrades the master cell and CqsS from their kinase to phosphatase states. cycle regulator CtrA at the Gl-S transition (Biondi Because phosphorclays are reversible, LuxO will et al., 2006; Iniesta et al., 2006) (Fig. 2C). Precisely backtransfer phosphoryl groups to LuxU, which then how CpdR influences ClpXP activity or localization transfers to the receiver domain of a hybrid histidine remains unclear, but CpdR exemplifies the crucial kinase where the bifunctional kinase domain drives and ever-expanding role of single-domain response dephosphorylation. Hence, switching any of the regulators. quorum histidine kinases to the phosphatase state is likely sufficient to drive the complete dephosphoryla­ tion of LuxO, resulting in the synthesis of LuxR and PHOSPHOTRANSFER its density-dependent genes. There arc also examples of one-to-many relation­ A critical step in two-component signaling is the ships and histidine kinases that have multiple, bona transfer of a phosphoryl group from the histidine kinase fide targets. The best such example is tl1e chemotaxis to its cognate response regulator. In most cases, a histi­ kinase CheA_that phosphorylates the single-domain dine kinase will phosphorylate a single cognate response regulator CheY to effect changes in flagellar rota­ regulator, forming an exclusive one-to-one signaling tion as well as the methylesterase CheB, which plays pair. However, in other cases, cells exhibit many-to-one a critical role in signal adaptation by demethylating and one-to-many relationships, further enhancing the chemotaxis receptors such as Tar (Armitage, 1.999). information-processing capabilities of bacteria. The operation of two-component signaling path­ A prime example of a many-to-one relationship ways, whether branched or simply one-to-one, demands is the multiple histidine kinases that phosphorylate an exquisite specificity of information flow. Cells must SpoOF to drive the initiation of B. subtilis sporulation somehow ensure that histidine kinases interact with (Antoniewski et aL, 1990; Trach and Hoch, 1993; and phosphotransfer to rhe "correct," or cognate, re­ Jiang et al., 2000). The kinases KinA and KinB are sponse regulator(s) while avoiding detrimental cross essential for proper sporulation and drive most of the talk with noncognate regulators. This is a significant phosphorylation of SpoOF under standard labora­ cha:llenge for bacterial cells becat1se histidine kinases tory conditions. KinC, KinD, and KinE arc formally and response regulators are often very similar to one dispensable, but each is capable of phosphorylating another at the sequence and structural levels, and most SpoOF (Fig. 2A). These three kinases may simply sup­ organisms encode dozens, if not hundreds, of each type plement the activity of KinA and KinB, or they may of protein. be essential for phosphorylating SpoOF in response to The specificity of phosphotransfer is driven different environmental conditions. The convergence primarily by molecular recognition and an intrin­ of multiple histidine kinases on SpoOF may enable sic specificity of histidine kinases for their. cognate different stress and starvation signals to drive the substrates (Skerker et a!., 2005; Laub and Goulian, initiation of sporulation. Alternatively, such a topol­ 2007). In principle, other mechanisms could contrib­ ogy may be allowing the integration of signals such ute to or reinforce specificity. For instance, differen­ that the subthreshold activation of multiple pathways tial expression of signaling proteins .could prevent triggers sporulation. the interaction of noncognate ·partners, or scaffold CHAPTER 4 • THE ROLE OF TWO-COMPONF..NT SIGNAL TRANSDUCTION SYSTEMS 51 proteins that simultaneously bind components of sets of co-operonic histidine kinases and response a signaling pathway and enforce spatial proximity regularors; such pairs typically interact in a one­ of cognate proteins could help prevent cross talk to-one fashion and are thus expected to coevolve. between noncognate partners. However there is a Indeed, sequence analysis identified a relatively small wealth of data that points to molecular recognition set of amino acids that appear to coevolve in the two as the dominant force behind signaling fidelity. Early molecules. These putative specificity residues map studies with the histidine kinase VanS from Entero­ to the molecular interface formed when a response coccus faecium demonstrated that it preferentially regulator docks to a histidine kinase, but comprise phosphorylated VanR relative to the noncognate sub­ only a subset of the interfacial residues (Skerker strate PhoB from E. coli (Fisher et al., 1996). Kinetic et al., 2008). Importantly, mutating these specificity studies also demonstrated that KinA in B. subtilis residues in the histidine kinase EnvZ to match the can phosphorylate both SpoOF and SpoOA, but has corresponding residues found in other E. coli histi­ a nearly 57,000-fold preference for phosphotransfer dine kinases, such as RstB, was sufficient to repro­ to SpoOF (Grimshaw et al., ·1998). More recently, gram the substrate selectivity of EnvZ (Skerker et al., in £ . coli and C. crescentus, histidine kinases were 2008). These results have thus highlighted the key shown to exhibit a strong, global kinetic preference residues that dictate' specificity. ln addition, they open in vitro for their in vivo cognate substrates (Skerker the door to the rational rewiring of two-component et al., 2005). For example, autophosphorylated EnvZ signaling pathways and the possibility of develop­ was systematically examined for phosphotransfer in ing algorithms to predict kinase-regulator pairings vitro to each of the 32 response regulators encoded in in any bacterium. the E. coli genome. With extended incubation times, Similar analyses of amino acid coevolution may EnvZ phosphorylated several response regulators; also be possible to identify the residues that dictate but, with shorter incubations, EnvZ preferentially the specificity of histidine kinase and response regu­ phosphorylated OmpR relative to all other E. coli lator homodimerizarion. A recent systematic analysis response regulators. More detailed kinetic analyses of Forster resonance energy transfer (FRET) between revealed that EnvZ has an approximately 1,000-fold all possible pairs of response regulators in E. coli preference for phosphotransfer to OmpR relative to demonstrated that these molecules specifically ho­ the next best substrate, CpxR, and hence even higher modimerize (Gao er al., 2008), suggesting that a set preferences relative to all other response regulators of residues exists to mediate homodimerization and (Skerker et al., 2005). This preference corresponds to prevent all possible heterodimerizations. with numerous in vivo studies indicating that EnvZ's Although molecular recognition is the dominant primary target is OmpR. determinant of phosphotransfer specificity in two­ Efforts to understand the specificity of molecular component signaling pathways, other mechanisms recognition in phosphotransfer have been hampered do exist. One key mechanism involves the phos­ by the lack of structural data and, in particular, by the phatase activity of bifunctional histidine kinascs. For lack of a co-crystal structure of a histidine kinase in example, the phosphorylation of OmpR by the non­ complex with a response regulator. There is a struc­ cognate histidine kinase CpxA is normally offset by ture of the histidine phosphotransferase SpoOB from EnvZ-dependent dephosphorylation of OmpR; simi­ B. subtilis in complex with the response regulator larly, phosphorylation of CpxR by EnvZ is offset by SpoOF (Zap£ et al., 2000) and, more recently, a struc­ Cpx:A's dephosphorylation of CpxR (Siryaporn and ture has been solved of the Thermotoga maritima ki­ Goulian, 2008). Hence, CpxA phosphorylates OmpR nase TM0853 in complex with its cognate regulator in a 6.envZ mutant and EnvZ phosphorylates CpxR TM0468 (Casino et al., 2009). However, these two in a 6.cpxA mutant, but wild-type cells are effectively static structures still do not fully reveal bow cognate buffered against spurious cross talk. and noncognate substrates are distinguished from Response regulator competition seems to fur­ one another. To tackle this problem, alternative ap­ ther reinforce the fidelity of information flow in two­ proaches have been pursued-most notably, analyses component pathways . .For example, in E. coli, the of amino acid coevolution in kinase-regulator pairs relatively high abundance of the response regulator (White ct al., 2007; Skerker et al. 2008). If the opera­ OmpR can prevent cross talk from EnvZ to CpxR tion of a two-component signaling pathway depends and, conversely, CpxR can prevent cross talk from on molecular recognition, then mutations in the his­ . CpxA to OmpR (Siryaporn and Goulian, 2008). In tidine kinase may be accompanied by compensatory each case, the cognate response regulatOr (which also mutations in its cognate response regulator or vice has a higher affinity for its cognate kinase) competes versa. To identify patterns of amino acid coevolution, for, and effectively occupies, its cognate histidine kinase, recent bioinformatic analyses have examined large which is much lower in abundance. By occupying its 52 LAUB cognate kinase, a regulator prevents the kinase from be dedicated phosphatases, such as the phosphatase phosphorylating a noncognate regulator. These twb CheZ that specifica lly targets CheY. The chemotactic mechanisms-dephosphorylation by a cognate kinase behavior of E. coli cells is highly sensitive to changes and response regulator competition- have been char­ in CheZ concentration, such that either increasing or acterized in numerous systems, not just with EnvZ­ decreasing CheZ disrupts proper chemotaxis (Sanna OmpR and CpxA-CpxR, suggesting they are general and Simon, 1996), underscoring the critical role that design principles of two-component signaling path­ phosphatases play in precisely controlling the activ­ ways (Laub and Goulian, 2007). These mechanisms, ity of response regulators and the output of two­ in combination with the intrinsic specificity of phos­ component signaling pathways. pbotransfer, ensure the fidelity of two-component Some response regulators have multiple phos­ pathways and prevent significant cross talk at the pbatases. In B. subtilis, several phosphatases have level of phosphotransfer. Two-component signaling been identified that target the response regulators pathways are not, however, entirely insulated from SpoOF and SpoOA to block the initiation of sporula­ one another because there are numerous cases of tion (Perego et al., 1996; Pcrego and Hoch, 1996a; response regulators that regulate overlapping sets Perego and Hoch, 1996b; Perego, 2001) (Fig. 2A). In of genes (Laub and Goulian, 2007). Consequently, some cases, these phosphatases are subject to further two distinct pathways can effect both common and regulation themselves and hence can integrate addi­ pathway-specific transcription. tional signals into the decision to sporulation. For in­ stance, the activity of the phosphatase Ra pA, which dephosphorylates SpoOr~ is regulated by a gene imme­ AUXllJARY PROTEINS diately downstream of rapA called phrA (Perego and Hoch, 1996a). This gene encodes for a small protein Given the importance of two-component signal­ that is secreted and subsequently processed to form ing pathways to bacteria, it is perhaps not surprising a five amino acid peptide that binds tO and antago­ that cells have evolved numerous ways of modulating nizes RapA. The accumulation of the phrA-derived the output of these signaling pathways, including co­ peptide thus depends on cell population density­ factors, phosphatascs, and protein inhibitors. These meaning, the RapA phosphatase effectively integrates auxiliary components provide additional layers of a quorum signal into the decision to sporulate by regulation, in some cases fine-tuning the output of modulating flow through the SpoOA phosphorelay. a signaling pathway, but in other cases significantly Intriguingly, the transcription of rapA is stimulated changing signaling dynamics. by another two-component pathway, ComP-ComA, which controls the cellular decision to become com­ Regulators of Response Regulators petent (Perego and Hocb, 1996a). The RapA phos­ phatase thus serves not only to integrate information In many cases, histidine kinases are bifunctional: on cell density, but may help coordinate two differ­ acting as both kinases and phosphatases for their ent pathways, helping to prevent their simultaneous cognate substrates. However, many two-component activation. Another phosphatase, RapB, also targets signaling pathways are also regulated by exogenous, SpoOF, but the signal that RapB responds to remains dedicated phosphatases. To date, all of the phos­ unclear (Perego et al., 1994). Several phosphatases phatases characterized, with one possible exception appear to directly rarget SpoOA, including SpoOE and (discussed later), target the aspartyl phosphates on perhaps YnzD and Yisl (Perego, 2001). The presence response regulators. Unlike histidine kinases and of numerous phosphatases dedicated to the sporu­ response regulators, the aspartyl phosphatases do lation phosphorelay indicates that B. subtilis cells not constitute a single paralogous gene family and integrate a wide range of disparate signals into the are often structurally dissimilar from one another. critical decision to sporulate or not (Fig. 2A). This fact has made their identification difficult and, In addition to phosphatascs, there is a growing hence, the repertoire of known phosphatases is still list of proteins that modulate response regulators by somewhat limited. direct protein-protein interaction without driving de­ The aspartyl phosphate on most response reg­ phosphorylation. These proteins act in diverse ways, ulators is often labile and hydrolyzes at an appre­ but ultimately inhibit response regularor activity. ciable rate on its own. Nevertheless, many response Two of the Rap proteins from B. subtilis, RapC and regulators have cognate, dedicated phosphatases, or RapG, interact with the response regulators ComA cognate proteins that accelerate the intrinsic rate of and DegU (Solomon et al., 1996; Ogura eta!., 2003). dephosphorylation. Even for response regulators with Although RapC and RapG are homologous to the extremely short phosphoryl group half-lives there can RapA and RapB phosphatases described previously, CHAPTER 4 • THE ROLE OF TWO-COMPONE.t~T SIGNAL TRANSDUCTION SYSTEMS 53 they do not affect dephosphorylation of ComA changes in extracellular iron concentrations by driv­ and DegU, rcspccr.ively, and instead block binding ing the phosphorylation of PmrA, which then triggers of these regulators to tl:tJ.eir target promoters. In E. changes in gene expression. The presence of PmrD coli, the protein Tori binds to the response regula­ can thus reinforce these changes in transcription. The tor TorR and inhibits TorR-dependent gene expres­ synthesis of PrnrD itself is regulated by a different two­ sion (Ansaldi et al., 2004). However, in contrast to component signaling pathway, PhoQ-PhoP, which re­ RapC and RapG, Tori does not prevent TorR from sponds to changes in extracellular magnesium concen­ binding to DNA, suggesting it may instead disrupt tration. Consequently, activation of the PhoQ-PhoP activation of RNA polymerase. As a final, somewhat system and the induction of PrnrD can help stimulate unconventional, e}\amplc in B. subtilis, the protein PrnrA-dependent gene expression, even after PmrB is Spx has been shown to inhibit transcriptional activa­ no longer stimulated to phosphorylate PmrA. tion by muJtiple response regulators, including ResD and ComA (Nakano et al., 2003). However, Spx docs Regulators of Histidine Kinases not bind to the regulators themselves, but instead binds to a site on the alpha subunit of RNA poly­ To date, the only reported phosphatase target­ merase that the regulators must normally interact ing a histidyl phosphate is the protein SixA in E. coli with to drive gene expression. that drives the dephosphorylation of the histidine Another emerging class of molecules that regu­ phosphotransferase domain of ArcB in vitro (Garcia late response regulators is small inhibitory proteins, Vescovi et al., 1996). However, the role of SixA in or peptides. In E. coli, the response regulator RssB modulating anaerobiosis in vivo, as ArcB does, re­ controls the stability of RpoS, an alternative (]" fac­ mains unclear. Although histidyl phosphatases seem to tor and master regulator of stationary phase (Muffler be uncommon, many histidine kinases are modulated ct al., 1996; Pratt and Silhavy, 1996; Zhou et al., by small inhibitory proteins. In B. subtilis, the sporu­ 2001 ). During exponential phase, RpoS levels are lation initiation kinases KinA and KinB are inhibited kept low by RssB, which is a proteolytic targeting by Sda, a 46 amino acid protein that binds directly factor for RpoS, helping deliver the (j factor to its to a region near the conserved histidine within each protease ClpXP. When cells enter stationary phase or kinase (Ogino et al., 1998; Burkholder et al., 2001; arc starved of key nutrients, such as phosphate, Rpo$ Bick et al., 2009) (Fig. 2A). Immediately prior to spo­ is stabilized allowing it to drive gene expression; in rulating, B. subtilis cells must complete a final round many cases this stabilization occurs by inhibiting the of DNA replication to ensure that both the mother and response regulator RssB. Recent work has identified daughter cells inherit a complete chromosome. Any a family of small proteins, termed antiadaptors, that disruption of DNA replication or replication stress, are transcriptionally induced following starvation such as that caused by DNA damage, induces the ex­ and that bind directly to RssB to inhibit it (Bougdour pression of Sda, which then inhibits KinA, prevent­ et al., 2006; Bougdour et al., 2008; Merrikh et al., ing both autophosphorylation and phosphotransfcr 2009). The first amiadaptor characterized was IraP, to SpoOF (Cunningham and Burkholder, 2009). This which is induced following phosphate starvation, Sda-based circuitry represents a checkpoint system, but others such as lraM and IraD, which respond coupling DNA replication to sporulation, by modu­ to magnesium starvation and DNA damage, respec­ lating output from the sporulation phosphorelay as tively, have since been identified and there are likely also described previously for the Rap phosphatascs. many more. Precisely how these small proteins inhibit The autophosphorylation of KinA (but not KinB) can RssB from delivering Rpo$ to the protease ClpXP or also be inhibited by a small protein called Kipl, whose whether they bind similar regions of RssB is not yer synthesis depends on the availability of fixed nitro­ clear. IraP, IraM, and lraD do not share significant gen (Fig. 2A); however, the precise signal sensed by similarity to one another suggesting they CO!Jid bind Kipi and its role in sporulation remain unclear (Wang RssB in different ways to inhibit its activity as a pro­ et al., 1997). Other examples of histidine kinase in­ tease adaptor. hibitors include the PII protein GloB in E. coli and Proteins that bind and influence response regula­ FixT in Sinorhizobium meliloti. GlnB directly inhib­ tors may not always function as negative regulators its autophosphorylation of NtrB, a histidine kinase and can instead promote the activity of a response critical for responding to nitrogen starvation, and regulator. A case in point is the small protein PmrD in c:oncomitandy enhances the phosphatase activity of Salmonella enterica that binds to and stabilizes the as­ NtrB on NtrC-P Uiang and Ninfa, 1999). FixT acts partyl phosphate on the response regulator PrnrA fol­ similarly, by inhibiting autophosphorylation of the lowing phosphorylation by PrnrB, its cognate histidine oxygen-sensitive kinase FixL and promoting its phos­ kinase (Kato and Groisman, 2004). PmrB responds to phatase activity on FixJ- P (Garnerone et al., 1999). 54 LAUB

The examples mentioned include cytoplasmic the DHp and CA domains; as discussed later, these regulators of histidine kinases; there arc also exam­ adjacent domains arc often PAS or HAMP domains. ples of transmembrane and periplasmic proteins that Although a few direct signals are known, for the regulate histidine kinases.1wo examples are B1500, a vast majority of histidine kinases, their stimuli have small inner-membrane protein that directly stimulates only been characterized phenomenologically. For ex­ the histidine kinase PhoQ (Eguchi et al., 2007), and ample, envZ was initially implicated in responding to CpxP, a periplasmic protein that inhibits the kinase changes in osmolarity of E. coli growth media (Hall CpxA (Fleischer et al., 2007). Finally, in P. aeruginosa, and Silhavy, 1981), but precisely how osmolarity is the histidine kinase GacS, which activates a response sensed by EnvZ remains a mystery. And some stud­ regulator called GacA that is crucial to the switch ies have reported that rhe small lipophilic molecule from acute to chronic infection, is directly inhibited procaine is a more potenr inducer of the EnvZ-OmpR by another membrane-bound histidine kinase called system than changes in osmolarity (Batchelor and RetS (Goodman et al., 2009). This inhibition does not Goulian, 2006), leaving open the question of what ~equire the catalytic residues of RetS indicating 'that EnvZ responds to in E. coli's natural setting and un­ RetS inhibits via a protein-protein interaction that derscoring the gap in our understanding of the input somehow disrupts autophosphorylation of GacS, al­ signals for histidine kinases. As another example, the though the domains involved are not yet known. histidine kinase CpxA was initially implicated in re­ sponding to defects in F-pilus biogenesis (McEwan and Silverman, 1980) and then subsequently shown INPUTS to respond to unfolded periplasmic proteins or the overexpression of periplasmic proteins such as NlpE One of the least understood aspects of two­ (Danese et al., 1995). Even more recently it was re­ component signaling systems is the initial activation ported that the CpxA-CpxR pathway is activated by of histidine kinases. All histidine kinases have two exposure to aminoglycoside antibiotics such as gen­ conserved domains: the dimerization and phospho­ tamicin and kanamycin (Kohanski et al., 2008). Al­ transfer (DHp) domain and the catalytic and ATP­ though these three inducers of CpxA arc seemingly binding (CA) domain (Fig. 3B). The DHp domain is incongruous, they each likely lead to the misfolding of a four-helix bundle formed by the homodimerir.ation membrane proteins, providing a unifying notion for of two histidine kinases and contains the conserved CpxA's sensory function. These studies of CpxA also histidine in a solvent-exposed position within the first underscore the central role that two-component sig­ alpha helix. TheCA domain binds ATP and catalyzes naling proteins play in coordinating diverse bacterial transfer of the gamma phosphoryl group to the DHp stress responses. However, it is still unclear whether domain. Autophosphorylation activity is likely reg­ the activation of CpxA in response to these diverse ulated by controlling the interaction between these stressors proceeds via a common mechanism at the two domains. Unlike the DHp and CA domains, the molecular level or whether there are different means sensor domains of a histidine kinase arc often struc­ by which CpxA can be activated. turally and phylogenetically unrelated, and in only a Ultimately, periplasmic signals must be trans­ few cases are the precise ligands known (Bader et al., duced across the membrane to induce changes in the 2005; Cheung and Hendrickson, 2008; Cheung and autophosphorylationstate of the histidine kinase. This Hendrickson, 2009; Kerf£ et al., 2003). Many of the critical step of the pathway remains mostly mysterious, known ligands are small molecules, metabolittes, or but recent studies of nitrate-sensing by the histidine ions that bind directly to the periplasmic domain of kinase NarX suggest that a piston-type displacement a histidine kinase. However, in some cases, adaptor occurs between alpha helices' in the periplasmic proteins may bind ligands and then transduce this domain (Cheung and Hendrickson, 2009); similar information to their partner kinase. For example, in movements have been suggested in the activation of V. harveyi, the histidine kinase LuxQ is found con­ the chemoreceptor Tar which ultimately activates the stitutively associated with a periplasmic binding pro­ histidine kinase CheA (Chervitz and Falke, 1996; tein called LuxP that reversibly binds the autoinducer Ottemann et al., 1999). Precisely how such piston­ AI-2 to regu late LuxQ kinase activity and quorum like movements in the peri plasm would be propagated sensing (Neiditch et al., 2005; Neiditch et al., 2006). to the cytoplasmic domains is unresolved. Studies of It should also be noted that not all histidine kinases the histidine kinase LuxQ ha~e also provided intrigu­ respond to periplasmic or extracellular cues and, in ing hints about the mechanism of signal transduction fact, many histidine kinascs are entirely cytoplasmic. across a membrane by a histidine kinase (Nciditch et Signal recognition for these histidine kinases typically al., 2005; Neiditch et al., 2006). As noted previously, involves domains located immediately N-terminal to the periplasmic domain of LuxQ is constitutively CHAPTER 4 • TIIE ROLE OF nVO-COMPONENT SIGNAL TRANSDUCTION SYSTEMS 55 associated with the protein LuxP, which can directly sensor from B. subtilis YtvA (not a histidine kinase) bind the quorum signal auroinducer AI-2 (Fig. 2B). to make FixL sense and respond to light. This chi­ Under conditions of low cell density when LuxP's meric kinase, dubbed YFl, was light-regulatable; the binding site is unoccupied by ligand, the LuxP/LuxQ default, or dark, state of YFl was competent for au­ complex adopts a symmetric quaternary structure in tophosphorylation whereas exposure to light led to which the LuxQ monomers are positioned in a man­ a more than 1,000-fold decrease in activity. A sys­ ner that allows autoph'osphorylation. The binding tematic mutational study of the linker connecting the of AI-2 to LuxP switches the periplasmic domains LOV domain to the DHp domain demonstrated that into an asymmetric configuration that is presumably only insertions of seven amino acids were tolerated propagated to,the cytoplasmic domains and results in and continued to yield a kinase that could be shut off LuxQ transitioning from the kinase to phosphatase by blue light. And, intriguingly, insertions of just two state. The studies of LuxQ and NarX have only amino acids generated a protein whose default state scratched the surface and much remains to be under­ was off, but that could now be activated by blue light. stood about the mechanisms by which ligands and Together these observations provide strong evidence signals control their histidine kinases. that input signals ,are transmitted to the DHp domain Translating periplasmic signals into changes in via a coiled coil linker that effectively acts as a rotary autophosphorylation activity probably also involves switch. Subsequent bioinformatic analysis revealed cytoplasmic domains that lie between the last trans­ that the length of the linker between PAS and DHp membrane helix and the beginning of the DHp domain. domains in most histidine kinases is either a multiple A careful census of histidine kinase domain strucrure of seven or a multiple of seven plus two. These studies found that the two most common domains imme­ are thus beginning to unveil general design principles diately N-terrninal to the DHp domain are so-called of histidine kinases, while simultaneously enabling I lAMP and PAS domains (Anantharaman et al., 2006) the rational rewiring of pathways. Such rewiring may (Fig. 3B). HAMP domains are coiled coils that have provide new tools for probing regulatory systems been suggested to relay receipt of a signal by adopting that utilize two-component pathways and they may one of two conformations that are rotationally related furtl1er enhance efforts in synthetic biology to design (T Iulko et al., 2006). PAS domains adopt a compact, regulatory circuits de novo. globular fold and may relay signals by reversible dock­ ing and release of the last alpha helix in the domain (Harper et al., 2003; Taylor and Zhulin, 1999). Many OUTI.OOK PAS domains also participate directly in sensing various ligands or small molecules. For example, one subclass Most bacteria are faced with a constantly chang­ of PAS domains called LOY (light-oxygen-voltage) do­ ing environment and a multitude of stressors that mains use flavin nucleotide cofactors to sense changes challenge their survival. Two-component signaling in redox state, oxygen tension, or available light. proteins are one of the predominant means by which Precisely how PAS and HAMP domains regulate bacteria sense and respond to such challenges. Since autophosphorylation at a molecular and atomic level their initial discovery two decades ago, histidine ki­ is unknown. However, HAMP and PAS domains are nases and response regulators have been implicated in usually linked to the DHp domain by an $-helix, or countless stress responses. Early studies elucidated the signaling helix, which is thought to form a coiled coil general paradigms for how these signaling pathways linker (with each monomer contributing one helix) function. Work since then has begun to reveal the that connects the terminal alpha helices in the HAM.P complex regulatory networks that are built around or PAS domain to the first alpha helix of the DHp these pathways and the many mechanisms control­ domain. The $-helix likely plays a central role in ling phosphate flux through these pathways, some of. converting ligand binding or signal recognition into which arc summarized in this chapter. Despite these changes in autophosphorylation activity of the ki­ advances, there is much that remains to be under­ nase (Anantharaman et al., 2006). One recent study stood about two-component signaling. We still have has made significant progress in characterizing how a rudimentary understanding of precisely how signals this linker controls a kinase's activity by rationally are sensed and translated into changes in histidine ki­ reprogramming FixL, an oxygen-sensing histidine nase activity. We arc also just beginning to understand kinase from Bradyrhizobium japonicum (Moglich how cells orchestrate the activities of so many highly ct al., 2009). FixL normally contains a heme-binding related pathways that must function in parallel and in PAS domain that allows it to sense changes in oxy­ close proximity to one another. And, finally, it remains gen tension within the cell. This PAS domain was a major challenge to understand how cells evolve new replaced with a structurally related LOV blue-light signaling pathways to respond to new stressors. 56 LAUB

REFERENCES pathway of Escherichia coli regulates transcription of the gene specifying rhe stress-inducible peripla.smic protease, DegP. Anantharaman, V., S. Balaji, and L. Aravind. 2006. The signaling Genes Dev. 9:387-398. helix: a common functional theme in diverse signaling pro­ Da Re, S., J. Schumacher, P. Rousseau, J. .Fourment, C. Ebcl, and reins. Bioi. Direct. 1:25. D. Kahn. 1999. Phosphorylation-induced dimcri:tation of the Ansaldi, M., L. Theraulaz, and V. Mcjcan. 2004. "loci, a resp onse Fixj receiver domain. Mol. Micrabiol. 34:504-511. regulator inhibitor of phage origin in Escherichia coli. Proc. Egucb.i, Y., J. Itou, M. Yamane, R. Dernizu, F. Yamato, A. O kada, Nat/. Acad. Sci. USA 101:9423-9428. H. Mori, A. Kato, and R. Utsumi. 2007. B1500, a small mem­ Antoniewski, C., B. Savelli, and P. Stragier. 1990. The spollj gene, brane protein, connects the two-component systems EvgS/ which regulates early developmental steps in Bacillus Sllbtilis, EvgA and PhoQ!PhoP in Escherichia coli. Proc. Natl. Acad. belongs to a class of environmentally responsive genes. J- Bac­ Sci. USA 104:18712-18717. teriol. 172:86-93. Ferrari, F. A., K. Trach, D. J.eCoq, J. Spence, E. Ferrari, and J. A. Armitage, J. P. 1999. Bacterial tactic responses. Adv. Microb. Hoch. 1985. Characterization of the spoOA locus and its de­ Physiol. 41 :229- 289. duced product. Proc. Nat/. Acad. Sci. USA 82:2647-2651. Bachbawat, P., G. V. Swapna, G. T. Montelione, and A. M. Stock. Fisher, $. L., S. K. Kim, B. L. Wanner, and C. T. Walsh. 1996. 2005. Mechanism of activation for transcription factor PhoB Kinetic comparison of the specificity of the vancomycin re­ suggested by different modes of dimerization in rhe inactive sistance VanS for two response regulators, VanR and PhoB. and active states. Strncture 13:1353-1363. Biochemistry 35:4732-4740. Bader, M., $. Sanowar, M. Daley, A. Schneider, U. Cbo, \V. Xu, R. Fleischer, R., R. Hccnnaon, K. Jung, and S. Hunke. 2007. Purification, Klevit, H. Le Moual, and S. Miller. 2005. Recognition of antimi­ reconstitution, and characterization of the CpxRAP envelope crobial peptides by a bacterial sensor kinase. Ce/1122:461-472. stress system of Escherichia coli. ]. Bioi. Chern. 282:8583-8593. Baikalov, 1., I. Schroder, M. Kaczor-Grzeskowiak, K. Grzeskowiak, Freeman,]. A., and B. L. Bassler. 1999a. A genetic analysis of d1e func­ R. P. Gunsalus, and R. E. Dickerson. 1996. Structure of tion of LuxO, a two-component response regulator involved in the Escherichia coli response regulator NarL. Biochemistry quorum sensing in Vibrio harveyi. Mol. Microbial. 31:665-677. 35:11053-11061. Fre{:man, J. A., and B. L. Bassler. 1999b. Sequence and function of Batchelor, E., and M. Goulian. 2006. lmaging OmpR localization LuxU: a two-component phosphorelay protein that regulates in Escherichia coli. Mol. Microbial. 59:1767-1778. quorum sensing in Vibrio harveyi. ]. Bacteriol. 181:899- 906. Bick, M., V. Lamour, K. Rajasbankar, Y. Gordiyenko, C. Robinson, Freeman, J. A., B. N. Lilley, and B. L. Bassler. 2000. A genetic and S. Darst. 2009. How to switch off a histidine kinase: analysis of the functions of LuxN: a two-component hybrid crystal structure of Geobacillus stearothernzophilus KinE with sensor kinase that regulates quorum sensing in Vibrio harveyi. the inhibitor Sda. ]. Mol. Bioi. 386:163-177. Mol. Microbiol. 35:139- 149. Biondi, E. G., S. J . Reisinger, J. M. Skerker, M. Arif, B. $. Pcrchuk, K. Galpcriu, M. Y. 2005. A census of membrane-bound and intrac­ R. Ryan, and M. T. Laub. 2006. Regulation of the bacterial cell ellular signal transduction proteins in bacteria: bacterial IQ, cycle by an integrated genetic circuit. Nature 444:899-904. extroverts and introverts. BMC Microbial. 5:35. Bougdour, A., C. Cunning, P.]. Baptiste, T. FJliott, and S. Gottesman. Galperin, M. Y. 2006. Structural classification of bacterial response 2008. Multiple pathways for regulation of sigmaS (RpoS) stabil­ regulators: diversity of output domains and domain combina­ ity in F,scherichia coli via the action of multiple anti-adaptors. tions.]. Bacterial. 188:4169-4182. Mol. Microbiol. 68:298-313. Gao, R., T. R. Mack, and A.M. Stock. 2007. Bacterial response Bougdour, A., S. \Vidmer, and S. Gottesman. 2006. Modulating regulators: versatile regulatory strategies from common do­ RssB activity: lraP, a novel regulator of sigma($) stability in mains. Trends Biochem. Sci. 32:225-234. Escherichia coli. Genes Dev. 20:884-897. Gao, R., Y. Tao, and A.M. Srock. 2008. System-level mapping of Burbulys, D., K. A. Trach, and J. A. Hoch. 1991. Initiation of Escherichia coli response regulator dimerization with FRET sporulation in B. Sllbtilis is conrroUed by a multicomponenr hybrids. Mol. Microbial. 69:1358-1372. phosphorelay. Ce/l64:54.5-552. Garcia Vescovi, E., f. C. Soncini, and E. A. Groisman. 1996. Burkholder, \V. F., I. Kurtser, and A. D. Grossman. 2001. Replica­ Mg2+ as an extracellular signal: environmental regulation <>f tion initiation proteins regulate a developmental checkpoint in Salmonella virulence. Ce/184:165- 174. Bacill11s subtilis. Ce/1 104:269-279. Gamcrone, A.M., D. Cabancs, M. Foussard, P. Boistard, andJ . Batut. Casino, P., V. Rubio, and A. Marina. 2009. Structural insight into 1999.lnhibition of theFix.L sensor kinase by the FixT protein in partner specificity and phosphoryl transfer in two-component Sinorhizobimn meliloti. ]. Bioi. Chem. 274:32500-32506. signal transduction. Ce/1139:1-12. Goodman, A., M . Merighi, M . Hyodo, I. Ventre, A. Filloux, and S. Chervitz, $. A., and J. J. Falke. 1996. Molecular mechanism of Lory. 2009. Direct interaction bctwe~;n sensor kinase proteins transmembrane signaling by the aspartate receptor: a model. mediates acute and chronic disease phenotypes in a bacterial Proc. Nat/. Acad. Sci. USA 93:254.5-2550. pathogen. Genes Dev. 23:249-259. Cheung, J., and \V. Hendrickson. 2008. Crystal structures of C4-di· Grimshaw, C. E., S. Huang, C. G. Han.stein, M. A. Strauch, D. carboxylate ligand complexes with sensor domains of histidine Burbulys, L. Wang, J. A. Hoch, and J. M. Whiteley. 1998. kinascs DeuS and DctB.]. Bioi. Chem. 283:30256-30265. Synergistic kinetic interactions between components of the Cheung, J ., and W. A. Hendrickson. 2009. Structural analysis of ligand phosphorelay controlling sporulation in Bacillus subtilis. Bio­ stimulation of the histidine kinase NarX. Stmcture 17:190-201. chemistry 37:1365- 1375. Cho, H. S., S. Y. Lee, D. Yan, X. Pan, J. S. Parkinson, S. Kusru, D. Hall, M . N., and T. J. Silhary:1981. Genetic analysis of the ma­ E. Wemmer, and J. G. Pelton. 2000. NMR structure of acti­ jor outer membrane proteins of Escherichia coli. Annu. Rev. vated CheY. f. Mol. Bioi, 297:543-551. Genet. 15:91-142. · Cunningham, K., and W. Burkholder. 2009. The h.istidine kinase Harper, S. M., L. C. Neil, and K. H . Gardner. 2003. Structural iathibitor Sda binds near the sire of autopbosphorylation and basis of a phorotropin light switch. Science 301:1541-1544. may stericaUy binder autopbosphorylation and phosphotrans­ Hecht, G. B., T. Lane, N. Ohta, J. M. Sommer, and A. Newton. fer to SpoOF. Mol. Microbial. 71:659-677. 1995. An essential single domain response regulator required Danese, P. N., W. B. Snyder, C. L. Cosma, L. J. Davis, a.od T. J. for normal cell division and differentiation in Caulobacter Silbavy. 199.5. The Cpx two-component signal transduction crescerztlls. EMBO f. 14:3915-3924. CHAPTER 4 • THE ROLE OF TWO-COMPONENT SIGNAL TIV\NSDUCTION SYSTEMS 57

Hecht, G. B., and A. Newton. 1995. Identification of a novel re­ Kustu, S., A. K. North, and D. S. Weiss. 1991. Prokaryotic tran­ sponse regulator required for the swarmer-to-stalked-cell transi­ scriptional enhancers and enhancer-binding proteins. Trtmds tion in Caulobacter crescentus.]. Bacterial. 177:6223-6229. Biochern. Sci. 16:397-402. Henke, J. M., and B. L. Bassler. 2004. Three parallel quorum­ Laub, M. T., and M. Goulian. 2007. Specificity in two-component sensing systems regulate gene expression in Vibrio harveyi. signal transduction pathways. Annu. Rev. Genet. 41:121-145. ]. Bacterial. 186:6902-6914. Lee, S. Y., H. S. Cho, J. G. Pelton, D. Yan, E. A. Berry, and D. E. Hess, J. F., K. Oosawa, N. Kaplan, and M. I. Simon. 1988. Phos­ Wemmer. 2001a. Crystal structure of activated CheY. Com­ phorylation of three proteins in the signaling pathway of parison with other activated receiver domains.]. Bioi. Chem. bacterial chemotaxis. Cell 53:79-87. 276:1 6425-16431. Hess, J. F., K. Oosawa, P. Matsumura, and M. I. Simon. 1987. Lee, S. Y., H. S. Cho, J. G. Pelton, D. Yan, R. K. Henderson, D. S. Protein phosphorylation is involved in bacterial chemotaxis. King, L. H uang, S. Kusru, F.. A. Berry, and D. E. Wemmer. Proc. Nat/. Acad. Sci. USA 84:7609-7613. 2001 b. Crystal strucrure of an activated response regulator Hickman, J. W., D. F. lifrea, and C. S. Harwood. 2005. A bound to its target. Nat. Struct. Biol. 8:52-56. chemosensory system that regulates biofilm formation through Loomis, W. F., A. Kuspa, and G. ShauJsky. 1998. Two-component modulation of cyclic diguanylate levels. Proc. Nat/. Acad. Sci. signal transduction systems in eukaryotic microorganisms. USA 102:14422-14427. Curr. Opin. Microbiol. 1:643-648. Hulko, M., F. Berndt, M. Gruber, J. Linder, V. Truffault, A. McEwen, J., and P. Silverman. 1980. Chromosomal mutations of Schultz, J. Martin, J. Schultz, A. Lupas, and M. Coles. 2006. Escherichia coli that alter expression of conjugative plasmid The liAMP domain structure implies helix rotation in trans­ functions. Proc. Nat/. 'Acad. Sci. USA 77:513-517. membrane signaling. Ce/1 126:929- 940. Merrikh, H., A. E. Ferrazzoli, A. Bougdour, A. Olivier-Mason, and Hung, D. Y., and L. Shapiro. 2002. A signal transduction protein S. T. Loven. 2009. A DNA damage response in Escherichia cues proteolytic events critical to Caulobacter cell cycle pro­ coli involving the alternative sigma factor, RpoS. Proc. Nat/. gression. Proc. Natl. Acad. Sci. USA 99:"13160- 13165. Acad. Sci. USA 106:611- 616. Igo, M. M., A. J . Ninfa, and T. J . Silhavy. 1989a. A bacterial Moglich, A., R. A. Ayers, and K. Moffat. 2009. Design and sig­ environmental sensor that functions as a protein kinase and naling mechanism of light-regulated histidine kioases.]. Mol. stimulates transcriptional activation. Genes Dev. 3:598-605. Biol. 385:1433-1444. Igo, M. M., A. J. Ninfa, J. B. Stod<, and T. J. Silhavy. 1989b. Phospho­ Muffler, A., D. Fischer, S. Alruvia, G. Storz, and R. Hengge-Aronis. rylation and dephosphorylation of a bacterial transcriptional ac­ 1996. The response regulatOr RssB controls stability of the tivator by a transmembrane receptor. Genes Deu. 3:1725-1734. sigma(S) subunit of RNA polymerase in Escherichia coli. lniesta, A. A., P. T. McGrad1, A. Reisenauer, H. H. McAdams, and L. EMBO]. 15:1333-1339. Shapiro. 2006. A phospho-signaling pathway controls the local­ Nakano, S., M. M. Nakano, Y. Zhang, M. leelakriangsak, and ization and activity of a protease comple.x critical for bacterial cell P. Zuber. 2003. A regulatory protein that interferes with ac­ cycle progression. Proc. Nat/. Acad. Sci. USA 103:10935-10940. tivalOr-stimulared transcription in bacteria. Proc. Nat/. Acad. leeton, K., D. Z. Rudner, K. J. Siranosian, and A. D. Grossman. Sci. USA 100:4233-4238. 1993. Integration of multiple developmental signals in Ba­ Neidircb, M., M . Federle, A. Pompeani, R. Kelly, D. Swem, cilft~s subtilis through the SpoOA transcription factor. Genes P. j effrey, B. Bassler, and F. Hughson. 2006. Ligand-induced Dev. 7:283- 294. asymmetry in histidine sensor kinase complex regulates quo­ JenaJ, U., and M. Y. Galperin. 2009. Single domain response regu­ rum sensing. Ce/1 126:1095-1108. lators: molecular switches with emerging roles in cell organi­ Neiditcb, M. B., M. J. Federle, S. T. Miller, B. l. Bassler, and F. M. zation and dynamics. Curr. Opin. Microbiol. 12:152- 160. Hughson. 2005. Regulation of LuxPQ receptor activity by the Jenal, U ., and J. Malone. 2006. Mechanisms of cyclic-di-GMP sig­ quorun1-sensing signal autoinduccr-2. Mol. Cell18:507-518. naling in bacteria. Annu. Rev. Genet. 40:385-407. Ninfa, A. ]., and B. Magasanik. 1986. Covalent modification of Jiang, M., W. Shao, M. Pcrcgo, and J. A. Hocb. 2000. Multiple the glnG product, NRI, by the glnL product, NRII, regulates histidine kinases regulate entry into stationary phase and spo­ the transcription of the glnALG operon in Escherichia coli. rulation in Bacillus subtilis. Mol. Microbiol. 38:535-542. Proc. Nat/. Acad. Sci. USA 83:5909-5913. Jiang, P., and A. J. Ninfa. 1999. Regulation of autophosphoryla­ Ninfa, A. J., E. G. Ninfa, A. N. Lupas, A. Srock, B. Magasanik, and rion of }-;scherichia coli nitrogen regulator ll by the Pll signal j . Stock. 1988. Crosstalk between bacterial chemotaxis signal transduction protein.]. Racteriol. 181:1906-1911. transduction proteins and regulawrs of transcription of the Kato, A., and E. A. Groisman. 2004. Connecting two-component Nrr regulon: evidence that nitrogen assimilation and chemot­ regulatory systems by a protein that protects a response regu­ axis are controlled by a common phosphotransfer mechanism. lator from dephosphorylation by irs cognate sensor. Genes Proc. Nat/. Acad. Sci. USA 85:5492-5496. Dev. 18:2302-2313. N"JXon, B. T., C. W. Ronson, and EM Ausubcl. 1986. Two-component Kato, M., T. Mizuno, T. Shimizu, and T. Hakoshima. 1997. In­ regulatory ~-ysrems responsive to environmental stimuli share strongly sights into multistep phosphorelay from the crystal sfri!Cture conserved domains with the nitrogen assimilation regulatory genes of the C-terminal I IPt domain of ArcB. Cell 88:717- 723. ntrB and ntrC. Proc. Nat/. Acad. Sci. USA 83:7850-7854. Keener, j., and S. Kusru. 1988. Protein kinase and phosphoprotein Ogino, T., M. Matsubara, N. Karo, Y. Nakamura, and T. Mizuno. phosphatase activities of nitrogen regulatory proteins NTRB and 1998. An Escherichia coli protein that exhibits phosphohisti­ NfRC of enteric bacteria: roles of the conserved amino-terminal dine phosphatase activity towards the I IPt domain of the ArcB domain of NTRC. Proc. Nat/. Acad. Sci. USA 85:4976-4980. sensor involved in the multistep His-Asp phosphorelay. Mol. Kcrff, F., P. Charlier, M. L. Colombo, E. Sauvage, A. Brans, J. M. Microbial. 27:.573-585. Frere, B. Joris, and E. Fom.e. 2003. Crystal struct ure of the Ogura, M., K. Shimane, K. Asai, N. Ogasawara, and T. Tanaka. sensor domain of the BlaR penicillin receptor from Bacillus 2003. Binding of response regulator DegU to the a prE promoter licheniformis. Biochemistry 42:12835-12843. is inhibited by RapG, which is counteracted by extracellular Kohanski, M. A., D. J. D'vyer, J. Wierzbowski, G. Conarel, and PhrG in Bacillus Sttbtilis. Mol. Microbiol. 49:1685-1697. J. J. Collins. 2008. Mistranslation of membrane proteins and Ohta, N., and A. Newton. 2003. The core dimerization domains two-component system activation trigger antibiotic-mediated of histidine kinases contain recognition specificity for the cog­ cell death. Cel/135:679-690. nate response regulator.]. Bacteriol. 185:4424-4431. 58 LAUB

Oosawa, K., J. F. Hess, and M. I. Simon. 1988. Murants defective Tomomori, C., T. Tanaka, R. Durra, H. Pack, S. K. Saba, Y. Zhu, in bacterial chemotaxis show modified protein phosphoryla­ R. lshima, D. Liu, K. I. Tong, H. Kurokawa, H. Qian, M. tion. Cell 53:89-96. Inouye, and M. lkuca. 1999. Solution structure of the homodi· Ottemann, K. M., W. Xiao, Y. K. Shin, and D. E. Koshland, Jr. meric core domain of Escherichia coli histidine kinase EnvZ. 1999. A piston model for transmembrane signaling of the Nat. Stmct. Bioi. 6:729-734. aspartate receptor. Science 285:1751- 1754. Trach, K. A., a.ttd j . A. Hoch. 1993. Multisensory activation of the Paul, R., T. Jaeger, S. Abel, L Wiederkehr, M. Folcher, E. G. Biondi, phosphorclay initiating sporulation in Bacillus subtilis: iden­ M. T. Laub, and U. Jenal. 2008. Allosteric regulation of hisri­ tification and sequence of the protein kinase of the alternate dine kinases by their cognate response rcgularor determines cell pathway. Mol. MiCTobiol. 8:69-79. fare. Cei/ 133:452-461. Ulrich, D. L., D. Kojetin, B. L. Bassler, J. Cavanagh, and]. P. Loria. Paul, R., S. Weiser, N. C. Amiot, C. Chan, T. Schirmer, B. Giese, 2005. Solution Structure and dynamics of l.uxU from Vibrio and U. Jenal. 2004. Cell cycle-dependent dynamic localization harveyi, a phosphotransferase protein involved in bacterial quo· of a bacterial response regulator with a novel di-guanylate cy­ rum sensing.]. Mol. Bioi. 347:297-307. clase output domain. Genes Dev. 18:715- 727. Varughese, K. I., Madhusudan, X. Z. Zhou, J. M. Whiteley, and Perego, M. 2001. A new family of aspartyl phosphate phospharases J. A. Hocb. 1998. Formation of a novel four-helix bundle and targeting the sporulation transcription f3ctor SpoOA of Bacil­ molecular recognition sites by dimerization of a response regu­ lus subtilis. Mol. Microbiol. 42:133-143. lator phosphotransferase. Mol. Cell2:485-493. Perego, M ., P. Glaser, and J. A. Hoch. 1996. Aspartyl-phosphate Wang, L., R. Grau, M. Percgo, and J. A. Hoch. 1997. A novel phosphatases deactivate the response regulator components of histidine kinase inhibitor regulating development in Bacill11s the sporulation signal transduction system in Bacillus subtilis. subtilis. Genes Dev. 11:2569-2579. Mol. Microbiol. 19:1151-1157. Wassmann, P., C. Chan, R. Paul, A. Beck, H. Heerldotz, U. Jenal, Percgo, M., C. Ha.ttstein, K. M. Welsh, T. Djavakhisbvili, P. Glaser, and T. Schirmer. 2007. Structure of BcF3- -modified response and}. A. Hoch. 1994. Multiple protein-aspartate phosphatases regularor PleD: implications for diguanylate cyclase activation, provide a mechanism for the integration of diverse signals in catalysis, and feedback inhibition. Struct~tre 15:915-927. the control of development in B. subtilis. Cell 79:1047-1055. Watanabe, T., A. O kada, Y. Gotoh, and R. Utsumi. 2008. Inhibi­ Perego, M., and J. A. Hoch. 1996a. Cell-cell communication regu­ tors targeting two-component signal transduction. Adv. Exp. lates the effects of protein aspartate phosphatascs on the phos­ Mcd. Bioi. 631:229-236. phore lay controlling development in Bacillus subtilis. Proc. Weiss, V., and 8. Magasanik. 1988. Phosphorylation of nitrogen Natl. Acad. Sci. USA 93:1549-1553. regulator I (NRl) of Escherichia coli. I'roc. Natl. Acad. Sci. Percgo, M., andJ. A. Hoch. 1996b. Protein aspartate phosphatases USA 85:8919- 8923. control the output of two-component signal transduction ~>-ys ­ West, A. H., and A. M. Stock. 2001. Histidine kinases and response tems. Trends Genet. 12:97-101. regulator proteins in t\vo-component signaling systems. Trends Pratt, L. A., and T. J. Silhavy. 1996. The response regulator SprE Biochem. Sci. 26:369- 376. controls the stability of RpoS. Proc. Nat/. Acad. Sci. USA White, R. A., H. Szurmant, J. A. Hoch, and T. Hwa. 2007. Fea­ 93:2488- 2492. tures of protein-protein interactions in t\VO-component sig­ Sanna, M. G., and M. I. Simon. 1996. In vivo and in vitro charac­ naling deduced from genomic libraries. Methods Enzymol. terization of Escherichia coli protein CheZ gain- and loss-of­ 422:75- 101. function mutants.]. Bacteriol. 178:6275-6280. Wyman, C., I. Rombel, A. K. North, C. Bustamante, and S. Kustu. Siryapom, A., and M. Goulian. 2008. Cross-talk suppression be­ 1997. Unusual oligomerization required for activity ofNtrC, a tween the CpxA-CpxR and EnvZ-OmpR two-component sys­ bacterial enhancer-binding protein. Science 275:1658- 1661. tems in E. coli. Mol. Microbiol. 70:494-506. Xu, Q., D. Carlton, M. D. Miller, M . A. Elsligcr, S. Sri Krishna, Skcrker, J., M. J'rasol, B. Pcrchuk, E. Biondi, a.ttd M. Laub. 2005. P. Abdubek, T. Astakhova, P. Burra, H. J. Chiu, T. Clayton, Two-component signal transduction pathways regulating M. C. Deller, L. Duan, Y. Elias, J. Fcuerhelm, J. C. Grant, growth a.ttd cell cycle progression in a bacterium: a system- A. Grzechnik, S. K. Grzcchnik, G. Won Han, L. J aroszewski, level analysis. PloS Bioi. 3:e334. • K. K. Jin, H. E. Klock, M. W. Knuth, P. Kozbial, A. Kumar, Skerker, J. M., B.S. l'erchuk, A. Siryapom, E. A. Lubin, 0. Ash­ D. Marciano, D. McMullan, A. T. Morse, E. Nigogho.ssian, enberg, M. Goulia.tt, and M. T. Laub. 2008. Rewiring the L. Okach, S. Oommachcn, J. Paulsen, R. Reyes, C. L. Rife, specificity of two-component signal transduction systems. Cell N. Sefcovic, C. Trame, C. V. Trout, H. van den Bedem, D. 133:1043-1054. Weekes, K. 0. Hodgson, J. Wooley, A. M . Deacon, A. Godzik, Solomon, J. M., 8. A. Lazazzcra, and A. D. Grossman. 1996. Puri­ S. A. Lesley, and I. A. \Vtlson. 2009. Crystal strucruce of his­ fication and characterization of an extracellular peptide factor tidine phosphotransfer protein ShpA, an essential regularor that affects two different developmental pathways in Bacillus of stalk biogenesis in Caulobacter crescentus. ]. Mol. Bioi. subtilis. Genes Dev. 10:2014-2024. 390:686-698. Stephenson, K., and J. A. Hoch. 2002. Two-component and phos· Xu, Q., a.ttd A. H. West. 1999. Conservation of structure and phorelay signal - tran.~ducrion systems as therapeutic targets. function among histidine-containing phosphotra.ttsfcr (HPt) Cun. Opin. Pharrnacol. 2:507-512. domains as revealed by the crystal structure of YPDJ.]. Mol. Stock, A., D. E. Koshland, Jr., a.ttd j. Stock. 1985. Homologies Bioi. 292:1039- 1050. between the Salmonella typhimurium CheY protein and pro­ Zapf, J., U. Sen, Madhusudan, J. A.-Hoch, a.ttd K. I. Varughese. teins involved in the regulation of chemotaxis, membrane 2000. A transi ent interaction between two phosphorclay protein synrhesis, and sporulation. l'roc. N at/. A cad. Sci. USA proteins trapped in a crystal lattice reveals the mechanism of 82:7989- 7993. molecular recognition and phosphotr:msfer in signal transduc­ Stock, A.M., V. L. Robinson, and P. N. Goudreau. 2000. Two-com­ tion. Structure 8:851- 862. ponent signal transduction. Annu. Rev. Biochem. 69:183-215. Zhou, Y., S. Gottesman, J. R. Hoskins, M. R. Maurizi, and Taylor, 8. L., and I. B. Zhulin. 1999. PAS domains: internal sen­ S. Wickner. 2001. The RssB response regulator directly sors of oxygen, redox potential, and light. Microbiol. Mol. targets sigma($) for degradation by ClpXP. Genes Dev. Bioi. Rev. 63:479-506. 15:627- 637.