Transcriptome plasticity underlying plant root colonization and insect invasion by Pseudomonas protegens

Abstract

Pseudomonas protegens shows a high degree of lifestyle plasticity since it can establish both plant-beneficial and insect-pathogenic interactions. While P. protegens protects plants against soilborne pathogens, it can also invade insects when orally ingested leading to the death of susceptible pest insects. The mechanism whereby pseudomonads effectively switch between lifestyles, plant-beneficial or insecticidal, and the specific factors enabling plant or insect colonization are poorly understood. We generated a large-scale transcriptomics dataset of the model P. protegens strain CHA0 which includes data from the colonization of wheat roots, the gut of Plutella xylostella after oral uptake and the Galleria mellonella hemolymph after injection. We identified extensive plasticity in transcriptomic profiles depending on the environment and specific factors associated to different hosts or different stages of insect infection. Specifically, motor-activity and Reb toxin-related genes were highly expressed on wheat roots but showed low expression within insects, while certain antimicrobial compounds (pyoluteorin), exoenzymes (a chitinase and a polyphosphate kinase), and a transposase exhibited insect-specific expression. We further identified two-partner secretion systems as novel factors contributing to pest insect invasion. Finally, we use genus-wide comparative genomics to retrace the evolutionary origins of cross-kingdom colonization.

Fig. 1. Toxicity (a), cell density (b), and MDS transcriptome analysis (c) for Pseudomonas protegens CHA0 colonizing different hosts and media.

The “Wheat” samples correspond to wheat-roots 1 week after inoculation, “P. xylostella 24/36 h” to Plutella xylostella 24 and 36 h after oral infection; “G. mellonella” to Galleria mellonella hemolymph 24 h after hemocoel injection; “LB” to lysogeny broth, and “GIM” to Grace’s insect medium. a Survival of P. xylostella larvae after exposure to artificial diet pellets spiked with 4 × 10⁶ CHA0 cells (top) and of G. mellonella larvae upon injection of 2 × 10³ CHA0 cells (bottom). One representative experiment with 64 (Plutella) and 30 (Galleria) larvae is shown. Time points where insects were sampled for RNA extraction are indicated with arrows. b Bacterial densities at collection time points. Boxplots are created from four replicates per environment and show CFU per mg of root dry weight (wheat), CFU per mg of larvae (24/36 h), CFU per µl hemolymph (G. mellonella) and CFU per ml medium (GIM, LB). c Multidimensional Scaling (MDS) analysis was performed with four replicate CHA0 transcriptomes per environment. The different replicates of the different biological samples are numbered from 1 to 4.

Fig. 2. Transcription profiles of Pseudomonas protegens CHA0 during the colonization of different hosts or environments.

a Normalized counts per million (CPM) of P. protegens CHA0 transcriptomes obtained using the K-means clustering method. Genes clustered in six different groups according to different hosts/media. Standard errors for four biological replicates show small variations among the genes included in each cluster. The number of genes that belong to the main cluster of each transcription profile are indicated. From each cluster, those genes with a gene ontology (GO) annotation were used in the enrichment analyses presented in b. b Gene ontology (GO) enrichment analysis of the specific genes of each transcription profile main cluster. Significant GO terms for the given set of genes are shown. The number of total genes related to a GO term present in the CHA0 genome is given between brackets and the indicated percentage shows how many of those have higher expression in the specific transcription profile (p value < 0.001). BP biological process, MF molecular function. Redundant GO terms were merged. The “Wheat” sample corresponds to wheat-roots 1 week after inoculation, “P. xylostella 24/36 h” to Plutella xylostella 24 and 36 h after oral infection; “G. mellonella” to Galleria mellonella hemolymph 24 h after hemocoel injection, “LB” to lysogeny broth; “GIM” to Grace’s insect medium.

Fig. 3. Comparisons of transcriptomes of Pseudomonas protegens CHA0 between different hosts (a, b) or between different insect compartments (c).

CHA0 transcriptomes colonizing different hosts were analyzed by the general linear model pipeline of edgeR package in R. Transcriptomes derived from different hosts were compared pairwise; the reference is always the host shown in the right side of the figure. Total differentially expressed genes for each comparison for each condition were subjected to a GO enrichment analysis. Significant GO terms for the given set of genes are shown. Total genes related to a GO term present in the CHA0 genome are given between brackets and the indicated percentage shows how many of those are differentially expressed in each comparison (p value < 0.001). a P. xylostella vs. wheat roots, b G. mellonella hemolymph vs. wheat roots, c P. xylostella vs. G. mellonella hemolymph. BP biological process, MF molecular function. The “Wheat” sample corresponds to wheat-roots 1 week after inoculation, “P. xylostella 24/36 h” to Plutella xylostella 24 and 36 h after oral infection; “G. mellonella” to Galleria mellonella hemolymph 24 h after hemocoel injection.

Fig. 4. Heatmap showing the normalized reads (counts per million) for genes related to toxins, type VI secretion system (T6SS), specific enzymes, and iron acquisition in Pseudomonas protegens CHA0 colonizing different hosts.

Black indicates low expression (less than 10 counts per million reads) and yellow indicates high expression (more than 10³ counts per million reads). The “Wheat” sample corresponds to wheat-roots 1 week after inoculation, “P. xylostella 24/36 h” to Plutella xylostella 24 and 36 h after oral infection; “G. mellonella” to Galleria mellonella hemolymph 24 h after hemocoel injection.

Fig. 5. Two-partner secretion (TPS) systems in Pseudomonas protegens CHA0: domain analysis (a), expression profiles in relation to different hosts (b), and contribution to insecticidal activity (c).

a Domain analysis of the secreted protein with HMMER database comparing Bordetella pertussis protein FhaB, Pseudomonas aeruginosa PA7 protein ExlA and P. protegens CHA0 proteins TpsA1, TpsA2, TpsA3, and TpsA4 (PPRCHA0_0168-169, PPRCHA0_0625-0626, PPRCHA0_1574-1575, and PPRCHA0_4277-4278, respectively). The signal peptide and TPS domains are used to interact with the transporter protein for membrane translocation; the filamentous hemagglutinin 1 attaches to the host-cell and the filamentous hemagglutinin 2 translocates the PT-VENN domain into the host; DUF637 is common to hemagglutinins but its function is still unclear [83]. b Heatmap showing the normalized expression values for genes related to the four complete two-partner secretion systems in P. protegens CHA0 colonizing different hosts. The “Wheat” sample corresponds to wheat-roots 1 week after inoculation, “P. xylostella 24/36 h” to Plutella xylostella 24 and 36 h after oral infection, “G. mellonella” to Galleria mellonella hemolymph 24 h after hemocoel injection. c Survival of P. xylostella larvae after exposure to artificial diet pellets spiked with 4 × 10⁶ cells of CHA0 wild type, or its tpsA2 or tpsA4 deletion mutants. Thirty-two 2nd instar larvae were used per bacterial strain. d Survival of G. mellonella larvae after injection of 2 × 10³ cells into the hemocoel. Eighteen 7th instar larvae were used per bacterial strain. One experiment of each is shown and two more P. xylostella feeding and one G. mellonella injection are shown in Supplementary Fig. S5. Asterisks indicate significant differences of mutants to the wildtype (log-rank test, p value < 0.05).

Fig. 6. Orthologue protein analysis examining the presence of Pseudomonas protegens CHA0 factors associated with insect pathogenicity across different phylogenetic Pseudomonas groups.

A comparison of the full in silico proteomes of 97 pseudomonads belonging to phylogenetic groups harboring insect pathogenic, human pathogenic, plant pathogenic, and plant beneficial strains (groups and subgroups as defined by Hesse et al. [2]) was performed and is shown in Supplementary Fig. S7. Here, we show the distribution of selected CHA0 traits investigated in this study in ecologically different Pseudomonas strains. Strains with described activity are marked in: pink for insecticidal activity (oral or injectable); dark-blue for human pathogenic activity; orange for plant pathogenic activity; green for plant-beneficial activity (references in Supplementary Table S2). Am. amidases, Enz. enzymes, Trans. transposase.

Fig. 7. Proposed pathogenesis model of P. protegens CHA0 colonizing Lepidoptera insect pests after oral infection.

In the proposed pathogenesis model, the insect immune response is marked in blue, CHA0 factors emerging from this study in dark green and factors shown to be involved in P. protegens CHA0 insecticidal activity in previous studies in pink. As described previously by Engel and Moran [55], in the event of a pathogen invasion, the insect will detect the presence of proteoglycans or other bacterial components that will trigger the immune response. The gut epithelium activates the production of reactive oxygen and nitrogen species (ROS and RNS) through the DUOX membrane oxidases and antimicrobial peptides (AMPs) through the IMD pathway. If the signal reaches the insect fat body, the Toll pathway will activate the production of AMPs as well [55, 94]. We hypothesize the following infection process: CHA0 is taken up by an insect feeding on plant colonized by the bacterium. In the gut, the bacterium faces the first line of the insect defense and has to compete with the resident microflora. Amidase activity degrading proteoglycan residues from the cell wall helps P. protegens CHA0 to avoid recognition by the immune system. The bacterium further uses oxidoreductases and the Pap protein to protect itself against reactive oxygen and nitrogen species. Nitrogen transporters might capture nitrogenous compounds resulting from the interaction of RNS and ROS. In order to better survive this adverse and stressful environment, it is possible that P. protegens CHA0 activates transposases for genomic rearrangements in order to increase its genomic variability. The bacterial cells can resist AMPs thanks to the O-polysaccharide conformation of its surface [33]. CHA0 also produces antimicrobial compounds as shown here and in Flury et al. [29] and it uses the type VI secretion system (T6SS) to fight the microflora of the insect or other ingested bacteria. For breaching the gut epithelium, we propose the following scenario: to adhere to the surface of the peritrophic matrix, CHA0 uses the cyclic lipopeptide surfactant orfamide A [29]. Then, the chitinase disrupts the chitinous peritrophic matrix [16]. P. protegens CHA0 may use the phospholipase PlcN to release nutrients from the mucus layer or to damage the epithelial cells [16, 95] and exopolysaccharides to establish in the epithelium [76]. Subsequently, the production of different two-partner secretion proteins (TPS) triggers host cell death and disrupts the cadherin junctions between epithelial cells. This will allow the bacteria to transmigrate into the hemocoel. Here CHA0 has to resist phagocytosis by granulocytes, encapsulation by plasmatocytes, melanin coating by oenocytoids and AMPs produced by the fat body [21, 96]. To fight the immune cells, CHA0 might use the FitD toxin [33], hydrogen cyanide (HCN) [29], and the TpsA proteins which, in combination with the bacterial multiplication, will finally lead to the death of the insect.