Elucidation of the c-di-GMP mediated signaling mechanism in a bacterial transmembrane receptor

The LapA/LapG/LapD system is a common biofilm effector system found in over 1000 bacterial genomes and is mediated by the bacterial second messenger, cyclic-di-GMP or c-di-GMP. C-di-GMP along with the proteins that synthesize it (diguanylate cyclases or DGCs with GGDEF domains) and degrade it (phosphodiesterases or PDEs with EAL or HD-GYP domains) are important regulators of biofilm formation. Although some aspects of the biofilm formation process are fairly well understood, much still remains to be elucidated at the molecular level, especially the signaling mechanisms and the associated conformational changes in these proteins that are crucial for cell adhesion in a wide variety of bacteria, including several human pathogens. This study was undertaken to fill some of these gaps in our knowledge, focusing on the molecular mechanism of the transmembrane c-di-GMP receptor LapD, the main signaling switch in this pathway in Pseudomonas fluorescens. In this study, we introduced single amino acid cysteine mutations spanning the entire HAMP domain and the S-helix in a LapD-green fluorescent protein-fusion protein. As LapD exists as a homodimer as its smallest functional unit, a single amino acid substitution is represented once in each monomeric unit and the intra-dimer crosslinking of these residues was reported. Four distinct states/conformations of LapD were studied: 1) apo, 2) in the presence of c-di-GMP, 3) in the presence of LapG, and 4) in the presence of both c-di-GMP and LapG. Cysteine crosslinking experiments were performed on these four states of LapD by mildly oxidizing the protein in the presence of the catalyst copper phenanthroline [Cu(Phen)2]. The presence of disulfide dimer formation was detected using in-gel fluorescence by monitoring electrophoretic mobility shifts of the msfGFP-fusion protein on SDS-PAGE gels. Most of the 68-cysteine mutants generated spanning the HAMP domain and the S-helix displayed a significant level of intra-dimer crosslinking irrespective of whether they were proximally or distally located in the helices, or whether they were buried or surface-exposed. LapD thus appears to exist in multiple conformations in solution such that even more distal residues in the helices could come into close contact with each other. Periplasmic domain mutants located closer to the LapG binding site showed the highest levels of crosslinking for the partially activated LapG-LapD bound protein. Less crosslinking was observed whenever c-di-GMP was also available for binding. However, mutants located closer to the transmembrane domain showed higher crosslinking rates in the presence of c-di-GMP. Cysteine substitutions made on helix α1-H crosslinked readily in the absence of oxidant, whereas residues on helix α2-H showed little crosslinking under similar experimental conditions. In partially and fully activated states of LapD, consecutive residues on α2-H helix showed large differences in disulfide bond formation, whereas it tended to be similar for residues belonging to the α1-H. Cysteine mutations in residues located on α2-H also revealed the presence of an additional cross-linked species in conditions where LapD was able to form dimer-of-dimers. The fact that crosslinks were observed in different states of LapD activation and that the rates of disulfide bond formation differed across different activation states implies that LapD may operate by oscillating between different bundle conformations. Our results, therefore, suggest that LapD’s signaling mechanism appears to be more consistent with a dynamic bundle model rather than a conventional two-state signaling model such as the gearbox model