The cytosolic PAS domain (PASc)

Im Dokument Transmembrane Signalling: Structural and Functional Studies on Histidine Kinase CitA (Seite 51-57)

G. thermodenitricans CitA PASc was puried in the same way as PASp. In contrast to PASp, PASc is a dimer in solution based on the gel ltration elution prole (see gure 3.1). Crystallisation of PASc yielded crystals which diracted up to 1.78 Å. To solve the phase problem, PASc was also produced using selenomethionine labelling. The crystal structure was solved by Dr. Stefan Becker (Department of NMR-based Structural Biology, Max-Planck-Institute for Biophysical Chemistry, Göttingen, GER) using multi-wavelength anomalous dispersion (see tables 8.1 and 8.2) and reveals a canonical PAS-fold with ve β-strands sandwiched between an N -terminal α-helix and four shorter helices (see gure 3.7 A). The dimer found in the asymmetric unit reects the dimeric state seen in solution and reveals an exchange of N-terminal helices between monomers. The interaction between N -terminal helices from dierent monomers is supported by a hydrophobic zipper motif similar to the PAS domain of VicK, the rst HK with a PAS domain adjacent to the kinase core (Wang et al., 2013).

44 3. Results

Figure 3.8: Alignment of Geobacillus thermodenitricans CitA PASc and Es-cherichia coli DcuS PASc. Functional mutants in PASc of E. coli DcuS were trans-ferred to G. thermodenitricans CitA based on a sequence alignment. Transtrans-ferred ON-mutants in DcuS are highlighted green, OFF-ON-mutants are shown in red. For DcuS N304D, R289 neighbouring the aligned N288 in CitA was selected as a second potential candidate for mutation. Capital letters indicate high sequence homology.

In DcuS, another HK of the CitA family, PASc mutants were tested for eects on the activity and dimerisation state of full-length DcuS (Monzel et al., 2013). The PASc mutants associated with alterations in DcuS signalling can be divided into ON-mutants leaving the HK in a constitutive kinase-competent state and OFF-mutants which trap the DcuS kinase in the inactive state. The ON-mutants trigger signalling even without dicarboxylate ligands while OFF-mutant DcuS variants cannot be activated by adding dicarboxylates. While OFF-mutants conserve the dimeric HK state and binding capability to DctA, a DcuS co-receptor, ON-mutants can be subdivided further. The ON I-subtype destroys the HK dimer, which most likely does not correspond to biologically relevant states. ON II-subtype mutants on the other hand retain the dimeric state; ON IIb-mutants also still bind DctA while this interaction is lacking in ON IIa-IIb-mutants.

Based on the mutation analysis carried out in vivo on the PASc domain of homologous E. coli DcuS (Monzel et al., 2013), mutants of G. thermodenitricans CitA PASc were generated. Conserved residues in G. thermodenitricans CitA PASc that were mutated in DcuS PASc were identied based on a sequence alignment of E. coli DcuS PASc and G. thermodenitricans CitA PASc (see gure 3.8) generated with Dialign (Morgenstern, 2004). While selection of OFF-mutants was not restrained, ON-mutants were chosen from the ON IIb-subtype. In contrast to ON I- and ON IIa-mutants, DcuS dimer formation

3. Results 45

Figure 3.9: SEC proles of CitA PASc mutants. Like for wild-type PASp and PASc, the molecular weight can be calculated from a calibration and is depicted for PASc mutant monomers, assuming dimers in solution. The SEC prole of wild-type PASc is shown as a reference.

and DctA interaction is intact in ON IIb-mutants, which therefore most likely correspond to the ligand induced wild-type conformation of DcuS.

In the case of N288 in G. thermodenitricans CitA PASc the neighbouring R289, being a polar residue as well, was chosen as a second possible mutant to rule out errors in the sequence alignment. Like wild-type PASc, all selected point mutants are dimeric in solution based on SEC proles (see gure 3.9). The dimer is stable in solution; SEC was run at concentrations of 1.5 mM, 150 µM and 15 µM for PASc N288D. In all cases, the elution proles corresponded to dimeric protein. As SEC is limited by UV detection sensitivity, a concentration where monomeric protein could be observed was not reached.

For all G. thermodenitricans CitA point mutants, crystallisation trials were carried out (see table 3.1).

Table 3.1: Crystallisation of PASc mutants PASc mutant Crystallisation condition

R218A (SeMet) 0.4 M MgCl2, 23.5 % PEG 3350, 0.1 M Tris pH 8.5

E219G no crystals

V285A (native) 0.8 M Na2HPO4, 0.8 M KH2PO4, 0.1 M HEPES pH 7.5 N288D (SeMet) 0.4 M MgCl2, 19 % PEG 8000, 0.1 M Tris pH 8.5

R289D (SeMet) 0.2 M Li2SO4, 27 % PEG 1000, 0.1 M phosphate-citrate pH 4.2 R307A (native) 2.2 M NaCl, 2 % PEG 6000

46 3. Results

Figure 3.10: Crystal structures of proposed functional mutants of CitA PASc.

A: The structure of PASc N288D reveals an anti-parallel dimer (monomers coloured green and cyan). PASc R218A yields an identical fold. B: PASc R289D crystallises as an open dimer in which the N -terminal helix interaction between the monomers is missing. C:

Alignment of the PAS-core of one monomer for wild-type PASc (olive), PASc N288D (cyan) and PASc R289D (green). While the overall fold is retained, the position of the N -terminal helix varies signicantly.

Crystal structures could be solved for the proposed OFF-mutants CitA PASc R218A and V285A as well as for the proposed ON-mutants N288D, R289D and R307A (for details on structural data contact Dr. Stefan Becker, Department of NMR-based Structural Biology, Max-Planck-Institute for Biophysical Chemistry, Göttingen, GER). The PAS-fold excluding the N -terminal helix was conserved in all mutants with a maximum backbone rmsd of 0.42 Å compared to wild-type PASc, but the position of the N -terminal helix varied substantially in mutants R218A, N288D and R289D (see gure 3.10 C). PASc V285A and R307A crystal structures are identical to the wild-type structure. The re-orientation of the N -terminal helix in PASc R218A, N288D and R289D is related to dierent orientations of PASc monomers with respect to each other. In PASc R218A and N288D, the N -terminal helices form an anti-parallel dimer not connected to the PAS core.

In PASc R289D the helices of the two monomers do not interact with each other, but are contacting the central β-sheet scaold in trans, thus creating an open dimer in which the centralβ-scaolds of the two monomers are anti-parallel (see gure 3.10 A, B). In contrast to DcuS, in vivo HK activity measurements on CitA PASc mutants carried out by the group of Prof. Unden did not show any eect over wild-type G. thermodenitricans CitA for any of the PASc mutants.

3. Results 47

Figure 3.11: Cα-Cβ-secondary shift analysis of PASc in solution. The secondary chemical shifts reect secondary structure propensity, with positive values indicating α -helix, negative values corresponding to β-strands. The secondary structure elements of the PASc crystal structure are superimposed (red: α-helix, green: β-strand).

Wild-type G. thermodenitricans CitA PASc was assigned using liquid-state NMR experiments (see table 2.5). In total, 92 % of proton, 85 % of carbon and 69 % of nitrogen resonances were assigned. The backbone assignment reached 86 % completeness.

The missing assignments are mainly clustered in the N -terminal helix due to very weak or non-existent peaks in15N-HSQC-based experiments (see gure 3.12). Assignment of these residues was therefore mainly based on13C-HSQC-NOESY contacts and HCCH-TOCSY correlations. The Cα-Cβ-secondary chemical shifts of assigned residues correspond well to secondary structure elements found in the crystal structure of PASc (see gure 3.11).

Some residues in the second and third β-strand, expected to be negative, are found to display positive secondary shift values.

The PASc structure reveals twists in the twoβ-strands for which secondary shift values do not match secondary structure elements. Analysis of the backbone dihedral angles in the crystal structure demonstrates a deviation from ideal angles for anti-parallelβ-strands (ψ = -140; φ= 135). The positive secondary chemical shift values therefore correspond with strainedβ-strands in the crystal structure.

48 3. Results

Figure 3.12: Peak intensity of PASc in a 15N-HSQC. Visualisation of the inten-sity of well separated peaks in a 15N-HSQC experiment demonstrates the generally low intensity at the N -terminus of the domain. As assignment based on amide-correlation based spectra in this region was impossible, most resonances in the N -terminal helix were assigned based on 13C-HSQC-NOESY and HCCH-TOCSY experiments.

To conrm that citrate is only binding to PASp and not to PASc, a control 15N-HSQC of isolated PASc was set up with a twofold excess of citrate. As the spectrum is unchanged compared to citrate-free PASc (see gure 3.13) and no citrate was found in the crystal structure, eects on PASc in the citrate-bound form must be transmitted through binding of citrate to PASp alone. Additionally, the residues in the citrate-binding pocket of PASp are highly conserved (see gure 3.4) and not to be found in the sequence and structure of PASc (Gerharz et al., 2003).

3. Results 49

Figure 3.13: 15N-HSQC of PASc with excess citrate. To exclude citrate binding capacities of CitA PASc, a spectrum of isolated PASc with twofold excess of citrate (red) was compared with the spectrum of a citrate-free sample (blue). Spectra are shown at slightly dierent contour levels for visibility. No citrate binding eect is observed.

Im Dokument Transmembrane Signalling: Structural and Functional Studies on Histidine Kinase CitA (Seite 51-57)