Data processing revealed that the crystals belonged to space group C2221, with unit-cell parameters a = 103.5, b = 210.2, c = 93.9 Å for His6-ChpT-Native and a = 106.4, b = 210.0, c = 94.1 Å for His6-ChpT-Eu-DO3A. Phasing by molecular replacement using known structures of HKs was unsuccessful. Therefore, we solved the His6-ChpT structure using a heavy-atom derivative. 13 Eu-DO3A sites were found in the asymmetric unit of the derivative, with occupancies ranging from 20 to 100%.
There are three His
6-ChpT subunits in the asymmetric unit of the crystal, corresponding to a Matthews coefficient of 3.3 Å
3 Da
−1 and an estimated solvent content of 62.8%. Two subunits form a homodimer, while the third subunit exploits a twofold symmetry axis of the lattice to generate a similar homodimer but with exact twofold symmetry. The structure has been deposited in the Protein Data Bank with code
4fpp. The presence of metal ions was suggested in the electron-density map. A fluorescence spectrum measured on the BM30 beamline allowed us to show that the only metal ions present in the cooled sample were nickel ions, which are presumably derived from the purification procedure.
His6-ChpT adopts the overall domain architecture of the intracellular part of a class I histidine kinase (HK; Fig. 2). Each subunit consists of two distinct domains, an N-terminal helical hairpin domain and a C-terminal α/β domain, which are connected by a short linker (residues 84–91). The helical hairpin domain is comprised of residues 20–83 and its two antiparallel α-helices are connected by an eight-residue turn (residues 51–58). The N-terminal residues 1–19 are disordered and were not included in the structure.
The dimer interface is exclusively between the two helical hairpin domains and the twofold symmetry axis runs parallel to the helices such that the N-termini are adjacent, forming a four-helix bundle referred to as the dimerization and phosphotransfer (DHp) domain. The DHp domain contains the two H-boxes, with each catalytic His33 located on an opposite face of the four-helix bundle. Helix α1 extends for about 40 Å from the N-terminus to residue 50, displaying a kink induced by Pro38 at the end of the H-box (helix α1, 20–38; helix α2, 39–50). Helix α2 has a similar extension but without any pronounced kink. The four-helix bundle is stabilized mainly by hydrophobic interactions involving Leu27, Leu31, Phe35, Ala39, Ile42, Leu46, Met62, Ala69, Leu72 and Leu76.
A structure-similarity search using the
PDBeFold protein-structure comparison service at the European Bioinformatics Institute (
http://www.ebi.ac.uk/msd-srv/ssm; Krissinel & Henrick, 2004
![[triangle]](/corehtml/pmc/pmcents/rtrif.gif)
) revealed that the His
6-ChpT C-terminal (CCT) domain is a close structural homologue of the ATP-binding domain of DesK (r.m.s.d. of 2.15 Å for 109 structurally aligned Cα atoms) (Trajtenberg
et al., 2010
![[triangle]](/corehtml/pmc/pmcents/rtrif.gif)
), among many other homologous domains belonging to various sensor-type histidine kinases (Fig. 3). It adopts a Bergerat ATPase fold (Dutta & Inouye, 2000
![[triangle]](/corehtml/pmc/pmcents/rtrif.gif)
), which consists of an α/β sandwich with one layer made up of a mixed five-stranded β-sheet and the other layer consisting of three α-helices (α3–α5). In addition, this domain contains a pair of short antiparallel β-strands (βA and βB). It has been shown that ChpT has no capability to autophosphorylate His33 using ATP (Biondi, Reisinger
et al., 2006
![[triangle]](/corehtml/pmc/pmcents/rtrif.gif)
). The ChpT crystal structure reveals that the CCT domain differs structurally when compared with ATPase domains found in HKs. In HKs with autokinase activity each ATPase domain characteristically hosts one ATP molecule between the ATP lid (a loop between strand β3 and helix α5) and the central helix α4, while the γ-phosphate is exposed and can be attacked by the catalytic histidine of the DHp domain. The bottom of the ATP pocket consists of β-strands 3, 4 and 5. In His
6-ChpT the ATP lid is substituted by two additional turns in the N-terminal part of helix α5. These turns occupy the space in which the β- and γ-phosphates are usually located in the CA domain of HKs. Also, there is an additional short α-helix (α4′) between strand β5 and helix α5 in ChpT, which together with the latter helix closes the ATP-binding pocket found in the CA domain of HKs. We also obtained crystals of His
6-ChpT in the presence of 2.5 m
M ATP or ADP in the crystallization buffer. Analyses of these crystals did not reveal any ATP or ADP bound to His
6-ChpT, which was consistent with the different conformation of the CCT domain (data not shown).
Even if the His
6-ChpT homodimer shares the overall architecture of class I HK, with its four-helix bundle flanked by two ATP-binding-like domains, significant differences are observed in the relative orientation of these domains. In His
6-ChpT the two domains adopt a compact conformation, with helices α3, α4 and α5 from the CCT domain lying along helices α1 and α2, leaving the five-stranded β-sheet of the CCT domain roughly parallel to the helical axis of helix α1. The KinB kinase from
Geobacillus stearothermophilus (Bick
et al., 2009
![[triangle]](/corehtml/pmc/pmcents/rtrif.gif)
) is one of the closest structural homologues of ChpT. Although the ChpT and KinB DHP domains superimpose reasonably well (2.1 Å r.m.s.d. on 52 structurally aligned Cα atoms), the disposition of the C-terminal domains with respect to the DHp domains is very different. Structural alignment of the CCT domain with that of KinB would require an approximate rotation of 55° initiated at the interdomain linker and an outward translation of 20 Å away from the homodimer twofold axis. Structural studies have reported that the CA domains in HKs adopt different positions with respect to the phosphoacceptor His residue according to the step of the phosphotransfer process (Marina
et al., 2005
![[triangle]](/corehtml/pmc/pmcents/rtrif.gif)
). ChpT is devoid of autokinase activity and thus such specific movements between domains are not expected in the context of an autophosphorylation process. However, it cannot be excluded that domain movement still occurs, for example in the partner-recognition process.
A key characteristic of two-component systems is the high specificity of the HK–RR interaction (Laub & Goulian, 2007
![[triangle]](/corehtml/pmc/pmcents/rtrif.gif)
). Because of their structural similarities, we assume that the molecular basis of ChpT–target recognition should be similar to that of HK–RR. Structural insight into HK–RR interactions has recently been revealed by studies of the ThkA–TrrA and HK853–RR468 complexes of
Thermotoga maritima (Marina
et al., 2005
![[triangle]](/corehtml/pmc/pmcents/rtrif.gif)
; Casino
et al., 2009
![[triangle]](/corehtml/pmc/pmcents/rtrif.gif)
, 2010
![[triangle]](/corehtml/pmc/pmcents/rtrif.gif)
; Yamada
et al., 2009
![[triangle]](/corehtml/pmc/pmcents/rtrif.gif)
) and specificity rewiring of TCSs (Skerker
et al., 2008
![[triangle]](/corehtml/pmc/pmcents/rtrif.gif)
; Capra
et al., 2010
![[triangle]](/corehtml/pmc/pmcents/rtrif.gif)
; Ashenberg
et al., 2011
![[triangle]](/corehtml/pmc/pmcents/rtrif.gif)
). These studies showed that the recognition domain of the response regulator (RR) binds to its partner protein
via interactions with helix α1 of the DHp domain below the phosphodonor His residue and also parts of the ATP lid and the interdomain linker in the HKs. Since the ATP-lid region is severely affected in the ChpT structure by the N-terminal extension of helix α5 and the presence of an additional helix α4′, we anticipate that these differences may have implications for partner recognition of ChpT.
We have demonstrated in this work that His6-ChpT, an essential phosphotransferase that controls the phosphorylation of CtrA in C. crescentus, adopts the class I histidine kinase fold, with a four-helix bundle flanked by two domains displaying a structurally different ATPase fold. This structure paves the way for future biochemical investigations aiming at deciphering the functional aspects of ChpT regulation and function.