Several characteristic sequence motifs are conserved across kinase domains of HKs, termed the H, N, G1, F and G2 boxes, based on their amino acid sequences (Stock et al., 2000). The H box containing the primary phosphorylation site is in the DHp domain, and the rest are located in the CA domain. The CA domain contains the N box, G1, F and G2 boxes (Dutta et al., 1999) and alone exists as a monomer in solution, capable of binding ATP and
transferring phosphoryl groups to the dimeric DHp domain. Common to all classes of HK are the dimerisation domain, catalytic ATP binding domain, and the sensor domain.
Atomic structures have been reported for the isolated DHp, CA, or sensor domains of a number of HK sensors (Tomomori et al., 1999, Bilwes et al., 1999, Marina et al., 2005) and more recently a handful of structures of histidine kinases with bound response regulators (Casino et al., 2009, Yamada et al., 2009). Furthermore in 2013, the first full-length structure of a histidine kinase was published (Diensthuber et al., 2013). Taken together these
structures have improved understanding of two-component signal transduction systems and enable greater knowledge of ways to inhibit these systems in bacteria. The structures and mechanistic insights into the function of the domains are discussed below.
(i) Dimerisation and Histidine Phosphotransfer Domain (DHp)
Histidine kinases are active in the form of a dimer, which is controlled by the interaction of the dimerisation domains of each monomer HK subunit. One of the first high resolution structures of the dimerisation domain produced was that of EnvZ, a class I HK (hybrid) outer membrane protein (OMP) from E. coli (Tomomori et al., 1999), which is involved in
osmoregulation (see Figure 1.5.3.1). Dimerisation of the transmembrane protein is essential for its autophosphorylation and phosphorelay signal transduction to its response regulator OmpR, to control expression of porin proteins OmpC and OmpF, which form an efflux pump in Gram-negative bacteria (Hall & Silhavy, 1981; Nikaido, 2003).
[38]
Figure 1.5.3.1: The dimerisation and phosphotransfer domain of EnvZ HK. The His residue is shown in stick representation on helix α1 of each subunit (cyan or green) of the dimeric EnvZ DHp domain (residues 223-289, PDB: 1JOY), produced in PyMol.
The homodimeric DHp structure of EnvZ includes a solvent-exposed His-243 site of
autophosphorylation and comprises two antiparallel coiled-coils (or ‘helical hairpins’) (Helix I and II), which form a four-helix bundle. The two active His sites are located near the middle of Helix I (nearest N-terminus) within the dimeric kinase, which appears more mobile than Helix II, which could be important in signalling. Similar structures are also observed for Spo0B, a histidine phosphotransferase (without kinase activity), involved in phosphorelay with its response regulator, Spo0F in B. subtilis (Zapf et al., 2000).
(ii) Catalytic and ATP binding domain (CA)
The C-terminal catalytic domain proceeds after the dimerisation domain and contains the ATP binding site. A number of structures of the CA domain have been solved including EnvZ (Tanaka et al., 1998) and PhoQ (Marina et al., 2001) from E. coli, HK853 (Marina et al., 2005) and CheA (Bilwes et al., 1999) from Thermotoga maritima, and DesK
(Trajtenberg et al., 2010) from B. subtilis. All CA domains have a characteristic α/β sandwich fold, consisting of three alpha-helices covered with a five-stranded β-sheet (see Figure 1.5.3.2), which are distinct from any known Ser/Thr kinases (Dutta & Inouye, 2000).
N
C
His-243 α1
[39]
Figure 1.5.3.2: Ribbon diagram of the ATP binding domain of DesK HK from B. subtilis with bound ATP (PDB: 3EHG). Each domain is coloured in a rainbow spectrum with blue for the N-terminus and red for the C-terminus, adapted from Wang, (2012).
The highly conserved ATP binding cavity is defined by the conserved residues in the N, G1, F and G2 boxes (Gao & Stock, 2009). Between the F and G2 boxes is a flexible region named the ATP lid, which can adopt different conformations upon nucleotide binding. This domain binds ATP, and donates a γ-phosphate group to the His residue on the DHp domain.
The CA domain for DesK from B. subtilis has the shortest known sequence and represents the minimal core structure for this domain. Upon binding ATP and a divalent cation (e.g. Mg2+) structural changes occur at the ‘ATP lid’ (a highly mobile loop) (Yamada & Shiro, 2008), giving it an ‘open’ or ‘closed’ (ATP bound) state. In the absence of ATP, this loop is partially disordered in crystal structures. Even in the presence of ATP, the lid shows
flexibility, allowing ATP to bind and interact with the DHp domain for phosphotransfer. The ATP-lid also allows the CA domain to adopt multiple positions relative to the DHp domain, to function as a kinase, phosphotransferase or phosphatase in response to external stimuli.
ATP
ATP lid
α1
β1
α2
α4
β2
β3
β4
β5
[40]
(iii) Sensor Domain
Histidine kinases detect environmental input at their sensor domain and must adapt to changes in their environment by modifying gene expression levels, using the two-component system. Despite the great diversity of sequences found within sensor domains, to detect a specific ligand, there are patterns of domain organisation which be sorted into discrete structural classes. The prototypical HK is a homodimeric membrane protein in which the sensor domain is positioned between two transmembrane helices, as a periplasmic loop, and the DHp domain follows the last transmembrane helix and is located in the cytoplasm (Cheung & Hendrickson, 2010). However, some histidine kinases deviate from the model, and have sensor domains either within the membrane or in the cytoplasm.
Periplasmic sensory HKs include VanS (the subject of this thesis), NarX and TorS (Cheung & Hendrickson, (2009)), which have an all α-helical fold, and PhoQ, DcuS and CitA
(Cheung & Hendrickson, 2010) which are the first three members of the PDC mixed α/β-fold (see Figure 1.5.3.3).
Figure 1.5.3.3: X-ray crystal structures of monomeric sensor domains of histidine kinases. Cartoon representation of the PDC sensor domains of PhoQ (PDB: 3BQ8) and CitA (2J80) and all-helical NarX (3EZH) and TorS (3I9Y) were prepared in PyMol.
[41]
Cytoplasmic sensory HKs can have a PAS-type fold (present in PER, ARNT and SIM proteins) (Hefti et al., 2004) e.g. FixL (Ayers & Moffat, 2008), which senses oxygen through heme. The other cytoplasmic sensor structure is a GAF-type fold (present in cGMP
phosphodiesterases, Adenyl cyclases and FhlA) e.g. DosS and DosT (Podust et al., 2008), which detect changes in the redox state of a bound iron or in bound oxygen. The PAS domain structure has the same mixed α/β-fold as the PDC domain, but also has other distinctive features. The GAF domain structure consists of a six-stranded anti-parallel β- sheet core (Cheung & Hendrickson, 2010). The majority of kinase sensors have a PAS domain (~33%), which can sense changes in light, oxygen and redox potential in the cell. The GAF domain is present in ~9% of HKs, and can also sense a variety of ligands. Both PAS and GAF domains have high structural plasticity and sequence variability, which is important for stimuli recognition and signal transduction (Cheung & Hendrickson, 2010). HAMP domains are also present in ~31% of HKs, which are implicated in signalling and follow the last TM helix of the sensor domain, and are discussed in section 1.5.5.
In contrast, there is little information on membrane-embedded sensor domains, as no crystal structure exists for a transmembrane sensor domain (Cheung & Hendrickson, 2010). Structural data on these systems does however suggest that most transmembrane sensors exhibit an all-helical fold. For example, sensory rhodopsin II (SrII) (Gordeliy et al., 2002), which is involved in chemotaxis, forms a dimeric four-helix bundle (within the membrane), to transduce signals downstream.
[42]