Microbiome and infectivity studies reveal complex polyspecies tree disease in Acute Oak Decline

Abstract

Decline-diseases are complex and becoming increasingly problematic to tree health globally. Acute Oak Decline (AOD) is characterized by necrotic stem lesions and galleries of the bark-boring beetle, Agrilus biguttatus, and represents a serious threat to oak. Although multiple novel bacterial species and Agrilus galleries are associated with AOD lesions, the causative agent(s) are unknown. The AOD pathosystem therefore provides an ideal model for a systems-based research approach to address our hypothesis that AOD lesions are caused by a polymicrobial complex. Here we show that three bacterial species, Brenneria goodwinii, Gibbsiella quercinecans and Rahnella victoriana, are consistently abundant in the lesion microbiome and possess virulence genes used by canonical phytopathogens that are expressed in AOD lesions. Individual and polyspecies inoculations on oak logs and trees demonstrated that B. goodwinii and G. quercinecans cause tissue necrosis and, in combination with A. biguttatus, produce the diagnostic symptoms of AOD. We have proved a polybacterial cause of AOD lesions, providing new insights into polymicrobial interactions and tree disease. This work presents a novel conceptual and methodological template for adapting Koch's postulates to address the role of microbial communities in disease.

Figure 1

Symptoms of AOD observed in the field (a–d), reproducing the symptoms of AOD through log inoculations with combinations of bacteria and Agrilus larvae associated with AOD (e–l), and statistical analysis of lesion formation in log inoculations (m–n). (a) Stem bleeds symptomatic of AOD on a mature Quercus robur in the field. (b) Close up view of stem bleeds and bark cracks characteristic of AOD. (c) Cross-section through AOD stem bleeds, showing degradation of vascular tissue (red arrow). (d) Longitudinal section through an AOD stem bleed showing the association between bacterial lesions (blue arrow) and Agrilus biguttatus galleries (red arrow). (e) Wound response in the innerbark of the control treatment (inoculation of a wound with sterile water) of log inoculation trials. (f) Lesion formed in the innerbark of oak logs inoculated with Brenneria goodwinii in log inoculation trials. (g) Lesion formed in the innerbark of oak logs inoculated with Gibbsiella quercinecans in log inoculation trials. (h) Lesion formed in the innerbark of oak logs inoculated with a combination of B. goodwinii and G. quercinecans in log inoculation trials. (i) Wound response in the innerbark of the A. biguttatus treatment (inoculation of a wound with live eggs of A. biguttatus). Note the wound response (blue arrow) and clean gallery created by a larva (red arrow). (j) Lesion formed in the innerbark of oak logs inoculated with B. goodwinii and eggs of A. biguttatus. Note the lesion developing from the inoculation point (blue arrow) and from the galleries (red arrow). (k) Lesion formed in the innerbark of oak logs inoculated with G. quercinecans plus eggs of A. biguttatus. Note the lesion developing from the inoculation point (blue arrow) and from the galleries (red arrow). (l) Lesion formed in the innerbark of oak logs inoculated with B. goodwinii, G. quercinecans plus eggs of A. biguttatus. Note the lesion developing from the inoculation point (blue arrow) and from the galleries (red arrow). (m) Mean lesion area formed in the inner bark of logs inoculated with bacteria. Colour indicates statistical significance (see key). The bacterial species inoculated were back-isolated in fulfilment of Koch’s fourth postulate. Error bars=−s.e. (n) Mean lesion area formed in the inner bark of logs inoculated with bacteria and A. biguttatus eggs. Colour indicates statistical significance (see key in panel (m)). The bacterial species inoculated were back-isolated in fulfilment of Koch’s fourth postulate. Eb—Erwinia billingiae, Bg—Brenneria goodwinii, Gq—Gibbsiella quercinecans. Error bars=−s.e.

Figure 2

Comparative metagenomic analysis of the taxonomic composition and function of the oak microbiome in healthy tissue (from trees without AOD) and diseased tissue (from trees with AOD). (a) The plot depicts the genera represented in all metagenomes based on One Codex binning of raw reads, demonstrating a clear shift in the microbial community between healthy and diseased trees. (b) Brenneria and Gibbsiella are statistically correlated to diseased tissue, as shown by a principal coordinate ordination analysis based on statistics calculated by Primer v7 and PERMANOVA+ using One Codex binning data. The first two axes depict the plotted community composition. Correlation vectors in the graph are significant using an R²>0.05. A Welch’s t-test was performed to test significance of differences between key taxa (identified above) and between healthy and diseased trees (pooled abundances for each factor). Resultant P-values from Welch’s t-test are overlaid on the correlation biplot. (c) Gene groups involved in bacterial phytopathogenic activity are significantly increased in AOD diseased trees, as shown by comparative functional analysis of SEED subsystem categories as annotated by MG-RAST on assembled metagenome contigs. The analysis of SEED subsystem metagenome data was performed using Stamp. Statistically significant functional differences between diseased and healthy communities were calculated using G-test with Yates’ correction. The Newcombe–Wilson test was performed to calculate confidence intervals between two binomial population proportions. (d) The genomes of 17 species were found to be common across all AOD metagenomes. Visualization of Kraken metagenome analysis of stem samples demonstrates the shifts in bacterial microbiome compositions. Bubble sizes are categorized based on the relative percentage frequencies of Kraken species-level alignment of raw metagenome sequencing reads and are depicted in the figure as 1 (read frequency <0.01), 2 (read frequency 0.01–0.1), 3 (read frequency 0.1–1), 4 (read frequency 1–10) and 5 (read frequency >10). Red bubbles signify samples from diseased trees, while blue bubbles signify samples from healthy trees. The 17 common species were determined based on common occurrence across all diseased tissue samples. Additionally, six species more abundant among the healthy trees were included to provide a contrasting shift.

Figure 3

Functional genome analysis of Gibbsiella quercinecans FRB97 (T), Brenneria goodwinii FRB141 (T) and Rahnella victoriana BRK18a. (a) Circular representations of B. goodwinii FRB141, G. quercinecans FRB97 and R. victoriana BRK18a genomes. From outside to inside, circles represent: (1) Assembled bacterial genomes, outermost (orange) circle, with encoded secretion system annotated at their genomic loci. (2) Metatranscriptome heatmap. Alignment of two in silico combined metatranscriptomes recovered from two necrotic lesions of AOD-affected trees against the bacterial genomes. Blue saturation represents increasing transcript alignments. (3–9) Seven metagenomes from necrotic lesions on AOD-affected trees and one healthy tree (metagenomes were extracted from two sites, Attingham and Runs Wood) were aligned through their coding domains to homologous regions in the bacterial genomes. (3) Attingham healthy (AT1) aligned metagenome-coding domains (light purple). (4) Attingham diseased (AT7) aligned metagenome-coding domains (aqua). (5) Attingham diseased (AT8) aligned metagenome-coding domains (blue). (6) Attingham diseased (AT9) aligned metagenome-coding domains (orange). (7) Runs Wood diseased (RW1) aligned metagenome-coding domains (green). (8) Runs Wood diseased (RW2) aligned metagenome-coding domains (pink). (9) Runs Wood diseased (RW3) aligned metagenome-coding domains (grey). (10) G+C content across the bacterial genome. (11) G+C skew across the bacterial genome. (b) Schematic diagram of transporters and transported proteins recovered from lesion metatranscriptomes and aligned against each bacterial genome. The number of significantly expressed (>3 aligned transcripts, covering >20% of the gene) genes is shown in parentheses. (c) Number of transcripts aligned against selected virulence genes encoded within B. goodwinii FRB141, G. quercinecans FRB97 and R. victoriana BRK18a. Gene categories are represented by the following colours: red—PCWDE, purple—general secretory pathway (GSP), yellow—type II secretion system (T2SS), blue—type III secretion system (T3SS), pink—type III secretion system effectors (T3SS effectors), and green—global regulators (GR).