Y2H protein-protein interaction network centered on c-di-GMP signaling.

A. Proteins are represented as nodes colored according to their functional annotations, and edges indicate interactions between nodes with good-to-high (solid line) and moderate (dotted line) confidence, as described in Table S2. B. Left: Distribution of the functional categories of the interacting partners for the ten c-di-GMP bait proteins used in this study (extracted from Table S3). Right: Statistical significance of functional category enrichment for each c-di-GMP bait protein, calculated as cumulative hypergeometric distribution (from data in Table S3). P-values less than 0.05 are highlighted in color.

PPI subnetworks connecting c-di-GMP proteins with interacting proteins that belong to the following functional categories.

(A) c-di-GMP signaling, (B) signal transduction, (C) transport and efflux, (D) pathogenicity, (E) transcriptional regulation, and (F) DNA repair. Note that the functional category “other” has been omitted. Nodes represent proteins, and edges indicate interactions between nodes with good-to-high (solid line) and moderate (dotted line) confidence.

CRISPRi Phenotypic Screening.

A selected number of genes encoding in-network CDG proteins and their interacting partners, as well as out-of-network CDGs were targeted by CRISPRi and subjected to biofilm and motility phenotypes. A - Swarming assay, imaging, and calculation of swarm areas. NTC, non-targeting control; TC, targeted positive control. B - Biofilm pellicle assay and imaging by confocal microscopy. Pellicles at the liquid-air interface were peeled out onto circular glass coverslips and mounted on a microscopy chamber containing transparent media. Biofilms were stained with FilmTracerTM green fluorescent dye prior to volume imaging using confocal microscopy. Parameters were extracted by image analysis using Imaris software. C - Formation of submerged biofilm at pegs (MBECTM). Cells were released from the pegs by sonication and counted using flow cytometry. D - Real-time monitoring of amyloid fibers and extracellular matrix production in the presence of optotracer using fluorometry (EbbaBiolight680).

Heatmap with dendrogram of hierarchical clustering of network proteins according to their phenotypic traits.

The analyzed phenotypes were biofilm biovolume (BV), biomass (BM) mean thickness (MS), maximum height (MX), roughness (R), submerged biofilm at pegs (SBP), and production of amyloids (AMF), and swarming motility (SW). The targeted proteins (rows on the heatmap) are split into six groups by k-means clustering. Groups are indicated by colors (left side) and distance is Pearson correlation. The extent of phenotype change is indicated by the side bar, blue for decrease and red for increase.

Summary of statistical analysis of phenotypic enrichment.

The analyzed phenotypes were biofilm biovolume (BV), biomass (BM), mean thickness (MS), maximum height (MX), roughness (R), submerged biofilm at pegs (SBP), production of amyloids (AMF), and swarming motility (SW). Statistical significance was calculated using cumulative hypergeometric distributions. All P-values are reported in Table S6. A-Phenotypic enrichment within interaction modules comprising a CDG bait and its direct interaction partners. 1 = significant enrichment for gene partners that increase the phenotype; −1 = significant enrichment for gene partners that decrease the phenotype; and 0 = no enrichment. Note that a value of ‘no enrichment’ does not indicate a lack of change in the phenotype, rather it indicates that the observed frequency of change is equivalent to random chance given the distribution of phenotypes across all knockdowns. B-Phenotypic enrichment by annotation category. C-Phenotypic effects of in-network versus out-of-network knockdowns. Split heatmaps illustrating -Top: phenotypic decrease or increase for knockdown of genes encoding CDG proteins from the PPI network (CDGs IN Network, orange) versus the CDGs not present in the network (CDGs OUT-Network, blue); - Middle: Gene knockdowns of DipA interacting partners (DipA-IP CDGs, orange) versus all other CDGs (blue); - Bottom: Knockdown of all genes present in the network (IN Network, orange) versus those outside of the network (OUT Network, blue).

Phenotypic correlation analysis between all CDG genes.

A-Principal component analysis plot with K-means clustering of biofilm pellicle and swarming motility phenotypes. Note: the less robust SBP phenotype was excluded from this analysis; WspR (PFLU1225) is here missing due to the lack of AMF data). Clusters are indicated by colors (1= green, 2= blue and 3= red). Purple arrows indicate the eigenvectors associated with the biofilm and swarming phenotypes. B-Pearson correlation analysis between Log2 fold change of the biofilm and motility phenotypes (BV, MX, MS, R, SW, and AMF). Positive and negative correlations are indicated by red and blue shades, respectively. The genes were ordered to group the most correlated. “In-network” vs “out-network CDG genes are indicated by orange and blue asterisks, respectively. The vertical bar represents the color legend of the correlation coefficients. The size of the dot also reflects the value of the correlation coefficient. White asterisks indicate the statistical significance levels (* p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001). Left: edge-weighted network of significant correlations. Each node represents a CDG gene from the PPI network (orange) or outside the network (blue).

Phenotypic correlation analysis of gene pairs.

A-dipA interaction module. Pearson correlation matrix for Log2FC in biofilm and motility phenotypes (BV, MX, MS, R, SW, and AMF). Positive and negative correlations are indicated in shades of red and blue respectively. Genes were ordered from the highest to lowest correlation with dipA. The color intensity and size of the circles are proportional to the correlation coefficients. The vertical bar shows the color legend of the correlation coefficients. Significantly correlated positive and negative clusters are indicated by red and blue shades, respectively. White asterisks indicate the levels of statistical significance (* p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001). Domain composition is indicated (GGDEF, open triangle; GGDEG/EAL, open square). B-Principal component analysis plot with K-means clustering of genes based on biofilm and motility phenotypes. Clusters are indicated by colors (1 red, 2 blue and 3 black). The eigenvectors associated with the biofilm and swarming phenotypes are shown in purple. CDG proteins containing the GGEEF domain (filled triangle) or dual GGDEF/EAL domain (filled square). Open circles are non-CDG partners. C-PFLU5127 interaction module. Pearson correlation matrix for Log2FC in biofilm and motility phenotypes (BV, MX, MS, R, SBP, and AMF). D-BifA (PFLU4858) interaction module. Pearson correlation matrix for Log2 fold change in biofilm and motility phenotypes (BV, MX, MS, R, and SW). E-Alg44 (PFLU0988) interaction module. Pearson correlation matrix for Log2 fold change in biofilm and motility phenotypes (BV, MX, MS, R, and SW). Partners with similar functional annotations are framed in red (correlated) or blue (anti-correlated).

PFL5608 plays a role in cell division.

A-Intracellular localization of the PFLU5608-mNeonGreen protein. 2D focal plane (FP) and 3D reconstructions showing PFLU5608 localizing at mid cell and forming a ring (yellow arrows) and at the cell poles (white arrows). Cell membranes were stained with the lipophilic dye FM 4-64 (red). B-Effect of PFLU5608 depletion of on cell length. A comparison was made between knockdown strain (g5608) and knockout strain (Δ5608). Cell membranes were stained with FM 4-64 (red). C-Cell counts and cell length measurements in knockdown strains expressing a guide targeting PFLU5608 (g5608) or a non-targeting guide sequence (gNT) were performed using Fiji with the MicrobeJ plug-in.

Functional validation of the role of PFLU3650 in DNA repair.

A-Interacting domains with UvrA-like protein partners PFLU3600 and PFLU5498. B-Survival following exposure to mitomycin C (0.5 µg/ml) after CRISPRi-mediated knockdown of PFLU3650 or its interacting partners. The control was an isogenic SBW25 strain that did not express a guide RNA. A second control comprised a strain expressing a guide RNA targeting the cytosolic DGC DgcP (PFLU0085).

Analysis of RimA subcellular dynamics.

A-Cells expressing a RimA-FP fusion protein from a plasmid exhibit a single dynamic fluorescent focus that scans the entire cell along its long axis. B-Average speed (µm/s) of RimA-GFP foci in individual cells (n=27, left). RimA dynamic behavior depends on the integrity of the EAL catalytic site (right). The RimA-FP protein containing an E47A substitution (EAL → AAL) was homogenously dispersed in the cell and unable to form discrete foci.