Insertion of a Divergent GAF-like Domain Defines a Novel Family of YcgR Homologues That Bind c-di-GMP in Leptospirales

Abstract

The Leptospiraceae family, which includes the genera Leptospira, Leptonema, and Turneriella, is an ecologically diverse group that includes saprophytic strains from soil and water as well as important pathogenic strains. Adaptation to these multiple environments relies strongly on signal transduction to adjust their morphology, motility, and metabolism to the changing environmental conditions. Members of the genus Leptospira distinguish themselves among spirochetes for having an elevated number of signal transduction genes. In this study, we describe a novel signal transduction protein that has gained multiple paralogues in the Leptospiraceae. These proteins are members of the YcgR/DgrA/MotI family, whose orthologs in several bacterial lineages have been shown to regulate the flagellar motor upon binding to c-di-GMP through their N-terminal PilZ domain. Unlike previously described versions of YcgR, the spirochetal proteins are characterized by the insertion of a divergent GAF domain within their N-terminal PilZ domain. We show that one member of this protein family from Leptospira interrogans is still a monomeric c-di-GMP binding protein and that these novel YcgR-like proteins have mostly replaced other members of the YcgR family in Leptospiraceae. Marked divergence among the paralogs suggests this family's expansion was accompanied by neofunctionalization, with the likely emergence of novel interactions in the signal transduction network controlling the flagellum rotor and other processes affected by changes in levels of c-di-GMP.

Figure 1

Representation of L. interrogans genes containing the PilZ domain. (a) Domain architecture of L. interrogans proteins containing the PilZ domain. The canonical residues RxxxR and DxSxxG are highlighted in yellow. LIC_20173, whose PilZ domain is not recognized by the UNIPROT and NCBI databases, is the only protein containing a PilZ domain in L. interrogans Copenhageni Fiocruz L1–130 that is fused to an intramembrane metalloprotease domain of the PrsW family. (b) AlphaFold2 model of the protein encoded by NCBI’s locus tag LIC_11920, showing the PilZN domain (colored in gray), GAZ (colored in purple), and PilZ (colored in blue). Conserved residues among the 6 paralogues of L. interrogans, shown in panel a, are shown as sticks. (c) WebLogo representation of the amino acid residue conservation pattern for the 6 paralogs of LIC_11920 from L. interrogans shown in panel a. The dotted gray line marks the conservation cutoff for the residues. In panel c, the upper gray line marks the residues of the PilZN domain; the blue line marks the PilZ domain; and the purple line identifies residues of the GAZ domain. The amino acid sequence below the blog corresponds to the LIC_11920 protein, with residues involved in c-di-GMP binding highlighted in red.

Figure 2

Structural analysis of the GAZ domain of YcgR_{LIC_11920}. (a) Structure superposition of the heterodimer Lcd1 GAF domain (PDB 5W10, colored in light and dark green). The missing secondary structures in the GAZ domain are highlighted in gray in one of the Lcd1 molecules. The structural model of the GAZ domain of YcgR_{LIC_11920} is colored in purple in panels a and b. In panel b, the putative ligand pocket site of the YcgR_{LIC_11920} GAZ domain is indicated. Due to the absence of several secondary structures, this site is exposed to the solvent, in contrast to the more shielded ligand pocket observed in the Lcd1 GAF domain structure (panel c). (d) The topologies of the GAF and GAZ domains are shown. The missing secondary structures in the GAZ domain are highlighted in gray on the GAF domain topology.

Figure 3

YcgR_{LIC_11920} is monomeric in the presence and in the absence of c-di-GMP by SEC-MALS assays. (a) 15% SDS-PAGE analysis of purified YcgR_{LIC_11920} following elution from a size-exclusion chromatography column. The theoretical molecular weight of YcgR_{LIC_11920} is 46 kDa. The first column in the SDS-PAGE gel corresponds to the Precision Plus Protein Standards kaleidoscope molecular weight marker (BIO-RAD). Circular dichroism profile of YcgR_{LIC_11920} in the presence and absence of cyclic dinucleotides: (b) YcgR_{LIC_11920} in the absence (black line) and presence of c-di-GMP (red line) and (c) YcgR_{LIC_11920} in the absence (red line) and presence of c-di-AMP (green line). In both cases, the protein is well folded. (d) Aliquots of ∼43 μM purified YcgR_{LIC_11920} in the absence (black line) and presence of 400 μM c-di-GMP (red line) were applied to a Superdex 200 increase 10/300 column. The lines represent the normalized refractive index (dRI), while the circles represent the molar weight distribution (kDa). YcgR_{LIC_11920} eluted as a monomer, showing molecular weights of 46 ± 2 and 45 ± 1.6 kDa in the absence (black) and presence of c-di-GMP (red), respectively. The replicates are listed in Figure S2.

Figure 4

Amino acid residue conservation profile among YcgR_{LIC_11920} orthologs. (a) WebLogo representation of the amino acid residue conservation profile in the YcgR_{LIC_11920} orthologs. The dotted gray line indicates that residues above this line were considered conserved. The upper gray line represents residues that compose the PilZN domain; in blue, those that compose the PilZ domain; and in purple, those that compose the GAZ domain. The amino acid sequence below the Weblog belongs to the LIC_11920 protein, and those residues colored in red and highlighted are involved in c-di-GMP binding. The WebLogo was performed using a multiple sequence alignment performed by the COBALT server (computes a multiple protein sequence alignment using conserved domain and local sequence similarity information) with sequences searched using NCBI’s locus_tag LIC_11920 with percent identity between 40 and 100% and query coverage between 80 and 100%, excluding Models (XM/XP), Nonredundant RefSeq proteins (WP), and uncultured/environmental sample, and on the representative of each Leptospira specie. (b) The conserved residues are mapped in the AlphaFold2 model of YcgR_{LIC_11920}, showing the PilZN domain (colored in gray), GAZ (colored in purple), and PilZ (colored in blue). Conserved residues are shown as sticks. We observed two conserved sites in the PilZN domain (Site I and Site II) and one hydrophobic core. In the PilZ domain, we observed a c-di-GMP binding site with more residues probably involved in the ligand binding and residues conserved in site III and site IV. In the GAZ domain, the conserved residues (colored in orange in the structure) are distributed in the structure, showing that this domain did not diverge in sequence enough to detect important residues for the functionality of the domain.

Figure 5

YcgR_{LIC_11920} selectively binds to c-di-GMP by ITC assay. Exothermic profile of interaction between YcgR_{LIC_11920} and c-di-GMP at 20 °C (a) and 10 °C (b). Two biological replicates were performed (Figure S5). (c) YcgR_{LIC_11920} mutant for residues R112A and R116A showed no interaction with c-di_GMP at 20 °C. (d) YcgR_{LIC_11920} did not interact with c-di-AMP at 10 °C as shown by an endothermic profile, consistent with the dilution of c-di-AMP.

Figure 6

Molecular docking and molecular dynamics of YcgR_{LIC_11920} in a complex with a monomer and dimer of c-di-GMP. (a) Molecular docking simulation of a monomer and dimer of c-di-GMP. The top pose presented a docking score of −10.6 and −9.7 kcal mol^–1 for a monomer and dimer of c-di-GMP, respectively. The arrows are vectors relative to the center of mass of each domain. In this regard, we analyzed the amino acid residues comprising these domains as follows: PILZ has residues 1–12, 110–149, and 288–389, PILZN has residues 13–109, and GAZ has residues 150–287. These vectors (for each frame) were calculated by considering a sum of difference between the center of mass of each domain (reference) and spatial coordinates of each atom of this domain (without normalization). (b) Backbone root-mean-square fluctuation (RMSF) of each residue of YcgR_{LIC_11920} with the monomer (blue) and dimer (black) of c-di-GMP. Note that high structural flexibility is localized in PilZN and PilZ, when compared with GAZ. However, the GAZ domain presents more flexibility when the c-di-GMP dimer interacts with YcgR_{LIC_11920} than with the c-di-GMP monomer. Angles between representative resultant vectors of each domain (PilZN, GAZ, and PilZ) along with time for YcgR to interact with the (c) c-di-GMP monomer and (d) c-di-GMP dimer. Enthalpy was calculated from MM-PBSA from the molecular dynamics trajectory when YcgR_{LIC_11920} interacts with the (e) c-di-GMP monomer and (f) c-di-GMP dimer.

Figure 7

Dynamic cross-correlation matrix (DCCM) projected onto the structure of the LIC_11920 interacting with the monomer and dimer of c-di-GMP. Correlation and anticorrelations are represented by red and blue, respectively. (a) Full correlation matrix without filtering. (b) Filtered correlation matrix, where correlations between −0.8 and 0.8 were removed. Projection of the correlations (r > 0.8, in red) and anticorrelations (r < −0.8, in blue) onto the 3D structure of the YcgR_{LIC_11920} while interacting with the c-di-GMP monomer. The anticorrelations were observed only between PilZN and PilZ colored in blue. (c) Full correlation matrix without filtering. (d) Filtered correlation matrix, where correlations between −0.8 and 0.8 were removed. Projection of the correlations (r > 0.8, in red) and anticorrelations (r < −0.8, in blue) onto the 3D structure of the YcgR_{LIC_11920} while interacting with the c-di-GMP dimer. The anticorrelations, indicated in blue, were observed between PilZN and PilZ, as well as between GAZ and PilZ.

Figure 8

Phylogeny and genomic context of YcgR homologues. The upper left and lower right panels show gene neighborhoods for YcgR homologues (colored in pink) collected from several bacterial lineages. Flagellar genes (yellow) are rare near members of the YcgR^GAZ subfamily. At least six events of gene duplication that led to the expansion of the YcgR^GAZ subfamily can be inferred from tree topology. The phylogeny was inferred using maximum likelihood methods implemented in IQtree, version 2.3.3.