A complete twelve-gene deletion null mutant reveals that cyclic di-GMP is a global regulator of phase-transition and host colonization in Erwinia amylovora

Abstract

Cyclic-di-GMP (c-di-GMP) is an essential bacterial second messenger that regulates biofilm formation and pathogenicity. To study the global regulatory effect of individual components of the c-di-GMP metabolic system, we deleted all 12 diguanylate cyclase (dgc) and phosphodiesterase (pde)-encoding genes in E. amylovora Ea1189 (Ea1189Δ12). Ea1189Δ12 was impaired in surface attachment due to a transcriptional dysregulation of the type IV pilus and the flagellar filament. A transcriptomic analysis of surface-exposed WT Ea1189 and Ea1189Δ12 cells indicated that genes involved in metabolism, appendage generation and global transcriptional/post-transcriptional regulation were differentially regulated in Ea1189Δ12. Biofilm formation was regulated by all 5 Dgcs, whereas type III secretion and disease development were differentially regulated by specific Dgcs. A comparative transcriptomic analysis of Ea1189Δ8 (lacks all five enzymatically active dgc and 3 pde genes) against Ea1189Δ8 expressing specific dgcs, revealed the presence of a dual modality of spatial and global regulatory frameworks in the c-di-GMP signaling network.

Fig 1. C-di-GMP is essential for host colonization.

A) A representative protein domain architectural overview of all identified proteins in E. amylovora Ea1189 that contain a GGDEF and/or EAL domain. Filled black vertical bars represent transmembrane helices on the N-terminal domain. The image was created using Biorender. B) Images depicting disease progression in apple shoots infected with Ea189 and Ea1189Δ12 at 3dpi. While Ea1189 infected shoots show signs of infection the leaf and the petiole (red arrows), shoots infected with Ea1189Δ12 show minor signs of necrosis limited to the apoplast region in the leaf (red arrow) C) Scanning electron micrographs depicting sectional images of the i) apoplast and ii) xylem tissue of young apple shoot tips 3 dpi with E. amylovora Ea1189 and Ea1189Δ12. Widespread colonization was observed when shoot tips were inoculated with WT Ea1189 in both the apoplast and the xylem. However, shoots inoculated with Ea1189Δ12 showed less severe bacterial colonization in the apoplast and no evidence of any biofilm development within the xylem tissue in the petiole. D) Growth patterns in vitro for Ea1189 and Ea1189Δ12 don’t show any significant difference. E) Bacterial population counts over the course of 3 days post inoculation of young shoot tips with WT Ea1189 and Ea1189Δ12. Leaf and petiole samples were separately examined. Over a time span of 72 hrs, the bacterial population of Ea1189 increases within the petiole and declines within the apoplast, whereas populations of Ea1189Δ12 decline in the leaf tissue and are at undetectable levels in the petiole. Error bars represent standard errors of the means. Tukey’s HSD (honestly significant difference) (P < 0.05) test was used to determine statistical significance over the course of the experiment for each tissue type. F) Z-stacked confocal microscopy images (color inverted) showing the overall attachment occurring within the flow chamber one hour after the introduction of either Ea1189 or Ea1189Δ12 cells into the chamber, followed by the flushing of the chamber with 0.5X PBS. Ea1189 cells displayed widespread even attachment with interspersed patches of elevated fluorescence signal indicating potential multilayered attachment. Ea1189Δ12 cells failed to attach to the chamber surface.

Fig 2. Edcs differentially regulate biofilm formation and virulence.

A) Confocal images (color inverted) and B) relative GFP intensity of flow cells inoculated with GFP labelled WT Ea1189, Ea1189Δ8 Ea1189Δ12 and Ea1189Δ12 complemented with individual edc genes. Bacterial inoculum was introduced into the flow cells and allowed to incubate for 1 h before being flushed out, followed by incubation under flow for 5 h. The flow cells were then imaged along a z-plane to assess the volume of bacterial adhesion to within the chamber. Ea1189Δ12 and Ea1189Δ8 are impaired in biofilm formation relative to Ea1189. Complementation of Ea1189Δ12 with the individual edc genes restores the biofilm formation to levels similar to Ea1189. C) C-di-GMP formation was attenuated in Ea1189Δ8 and Ea1189Δ12, and the complementation of Ea1189Δ12 with edcA-E was able to individually restore c-di-GMP levels to WT Ea1189 levels with the highest increase recorded in Ea1189Δ12/edcE. D) hrpL transcript levels, relative to WT Ea1189 were significantly increased in Ea1189Δ12, and complementation with edcC and edcE was able to significantly reduce the transcript levels as compared to Ea1189Δ12. Ea1189Δ8 had significantly lower hrpL transcript levels compared to Ea1189 and Ea1189Δ12. E) Virulence in apple shoots was significantly reduced in Ea1189Δ8 and Ea1189Δ12 relative to Ea1189. Only complementation with edcE was able to restore WT levels of shoot blight in Ea1189Δ12. F) Flagellar motility was not significantly affected in Ea1189Δ12/ Ea1189Δ8 compared to WT Ea1189. Complementation of Ea1189Δ12 with edcD and edcE was able to significantly reduce motility as compared to Ea1189Δ12. Error bars represent standard errors of the means. Tukey’s HSD (honestly significant difference) (P < 0.05) test was used to determine statistical significance for all experiments.

Fig 3. Type IV pilus and the flagellum mediate surface attachment.

A) Transcript levels of fliC (flagellar filament) and hofC (type IV pilus assembly platform protein), were significantly reduced in Ea1189Δ12 relative to WT Ea1189. crl (curli fimbriae activator) and fimA (fimbrial subunit) transcript levels were not significantly different among the two strains. B) Relative bacterial adhesion GFP intensity representing the level and C) Confocal z-stacked images (color inverted) of attachment within flow cells 1hr after incubation with Ea1189 and Ea1189Δ12 lacking or overexpressing fliC and/or hofC. The overexpression of hofC could restore attachment in Ea1189Δ12 to WT Ea1189 levels, however, this impact was lost if the overexpression occurred in the absence of fliC. WT Ea1189 showed considerable levels of cellular surface attachment, which diminished upon the overexpression of fliC and the deletion of hofC. The deletion of fliC and/or the overexpression of hofC did alter but did not abolish attachment in Ea1189. Error bars represent standard errors of the means. Tukey’s HSD (honestly significant.

Fig 4. Global c-di-GMP dependent regulation during biofilm initiation.

A) A volcano plot highlighting critical differentially expressed genes (DEGs) within surface exposed WT Ea1189 vs. Ea1189Δ12 cells analyzed via RNA-seq analysis. A DESeq2 FDR cutoff p-value of 0.05 was used and all DEGs highlighted in red have a two fold change (log2) in expression. The comparison revealed a total of 320 positively and 235 negatively expressed DEGs with functions including metabolism, extracellular appendage regulation and overall transcriptional/post-transcriptional regulators. B) GO enrichment analysis showing the top 20 overrepresented categories for the positive and negative DEGs, along with the overall frequency of gene/target occurrence within the DEG list.

Fig 5. Regulatory divergence among the Edcs.

A) A venn diagram representing the distribution of the DEGs in Ea1189Δ8 strain overexpressing individual edc genes measured via RNAseq. A total of 121 DEGs both positively and negatively affected in expression were found after being filtered through a DESeq2 FDR cutoff of 0.05 with at least a two fold change (log2) individually for each comparative condition of Ea1189Δ8 vs Ea1189Δ8 overexpressing an individual edc gene. Ea1189Δ8 overexpressing edcB had the highest number of uniquely regulated DEGs. There were three DEGs that were downregulated upon the overexpression of every edc gene in Ea1189Δ8. Note that these results also include each of the edc genes themselves in each comparison if filtered through the statistical cutoff requirement. The venn diagram tool software (accessible at bioinformatics.psb.ugent.be/webtools/Venn/) was used to generate the venn diagram using RNAseq data.

Fig 6. C-di-GMP regulatory model in surface exposed E. amylovora cells.

Our study indicates that there is a dimorphism in the regulatory targets of the c-di-GMP generated by each of the five Edcs. While each Edc uniquely regulates the transcription of several genes, and virulence factors in vitro and in planta, attachment/biofilm formation (dependent on EPS production) is regulated by all the Edcs, thus, leading us to hypothesize about the potential presence of a localized and a diffused pool of c-di-GMP that can achieve these varied regulatory targets. Red inhibitor lines and green arrows indicate negative and positive regulation respectively, with arrow breaks indicating intermediate regulatory steps.