Antibiotic-producing symbionts dynamically transition between plant pathogenicity and insect-defensive mutualism

Abstract

Pathogenic and mutualistic bacteria associated with eukaryotic hosts often lack distinctive genomic features, suggesting regular transitions between these lifestyles. Here we present evidence supporting a dynamic transition from plant pathogenicity to insect-defensive mutualism in symbiotic Burkholderia gladioli bacteria. In a group of herbivorous beetles, these symbionts protect the vulnerable egg stage against detrimental microbes. The production of a blend of antibiotics by B. gladioli, including toxoflavin, caryoynencin and two new antimicrobial compounds, the macrolide lagriene and the isothiocyanate sinapigladioside, likely mediate this defensive role. In addition to vertical transmission, these insect symbionts can be exchanged via the host plant and retain the ability to initiate systemic plant infection at the expense of the plant's fitness. Our findings provide a paradigm for the transition between pathogenic and mutualistic lifestyles and shed light on the evolution and chemical ecology of this defensive mutualism.

Figure 1. Burkholderia gladioli symbionts are transmitted vertically via egg smearing in Lagria villosabeetles.

(a,d) Adult females carry the symbionts within two pairs of accessory glands associated with the reproductive system as confirmed by FISH on cross-sections of a female reproductive system (inset). (b,e) Host eggs are covered with a secretion containing Burkholderia bacteria, as revealed by FISH on an egg wash. (c,f) The symbionts colonize invaginations of the cuticle that result in three dorsal compartments represented in red in a three-dimensional (3D) reconstruction of an L. hirta larva. The Burkholderia symbionts were localized by FISH in a cross-section of an L. villosa larva (inset). FISH pictures show Burkholderia-specific staining in red (Burk16S_Cy3), general eubacterial staining in green (EUB338_Cy5), the overlap of these two in yellow and host cell nuclei in blue (DAPI). Scale bars, 20 μm (d) and 50 μm (e,f).

Figure 2. B. gladioli symbionts protect L. villosa> eggs from fungal infestation.

In the absence of the symbionts on L. villosa eggs, there is a higher probability of the following three fungi to grow: (a) Purpureocillium lilacinum (N=180 per treatment, Cox mixed effects model, P<0.001 compared with all controls), (b) Trichoderma harzianum (N=20 per treatment, Mantel–Cox log rank test, P<0.01 compared with untreated control and P<0.001 compared with reinfected controls) and (c) Beauveria bassiana (N=20 per treatment, Mantel–Cox log rank test, P<0.01 compared with reinfected controls). (d) Picture of a representative symbiotic and aposymbiotic egg after 4 days of exposure to P. lilacinum spores. Scale bar, 0.5 mm. (e) The growth of P. lilacinum on the egg has a negative effect on the survival of the larvae during the first days after hatching (N=180 per treatment, Cox mixed effects model, P<0.001). (f) In vitro co-cultivation of B. gladioli Lv-StA (left) and P. lilacinum (right) on potato dextrose agar showing inhibitory activity of B. gladioli Lv-StA. Statistically significant differences: **P<0.01 and ***P<0.001. (a–c,e) Estimated survival curves (Kaplan–Meier) and the corresponding standard error are shown.

Figure 3. The Burkholderia symbionts of L. villosa produce several antimicrobial compounds.

(a) Organization of biosynthetic gene clusters in B. gladioli Lv-StA underlying the production of (1) toxoflavin (tox), (2) caryoynencin (cay) and (3) lagriene (lag). (b) Chemical structures of toxoflavin (1), caryoynencin (2), lagriene (3) and sinapigladioside (4). (c) LC-HRESI-MS profiles (extracted mass traces) of toxoflavin (green, m/z=194.0669–194.0677), sinapigladioside (black, m/z=468.1314–468.1352) and lagriene (blue, m/z=779.5265–779.5343) produced by B. gladioli Lv-StA in vitro (top) and in vivo, that is, on L. villosa eggs reinfected with B. gladioli Lv-StA (bottom). Peak intensity represents relative abundance within each chromatogram, but is not comparable among the green, black and blue profiles. In the egg extracts, toxoflavin was detected in one of the three replicates, whereas lagriene and sinapigladioside were detected in all three replicates. Production of the compounds was independent of the exposure to P. lilacinum, and was not detected in the aposymbiotic controls in any of the replicates.

Figure 4. The symbionts of L. villosa likely evolved from plant-pathogenic B. gladioli and retain their ability to infect a plant host.

(a) L. villosa females transmit Burkholderia to soybean plants (control plants N=9, L. villosa-exposed plants N=13; Mann–Whitney U-test, P<0.01). (b) Soybean plants infected with the symbiotic Burkholderia from in vitro cultures show reduced seed output (N=18 for each treatment; Mann–Whitney U-test, P<0.01). (c) Cotyledon assays reveal recognition of microbial elicitors by soybean (red colouration of wounded tissue) upon exposure to symbiotic B. gladioli. (d) Phylogenetic reconstruction based on Bayesian and approximately maximum likelihood algorithms of selected Burkholderia using partial 16S rRNA gene sequences (1,148 bp) showing the placement of Lagriinae-associated Burkholderia clustering with plant-pathogenic B. gladioli. Posterior probabilities (Bayesian inference) and local support values (FastTree) above 70% are reported at the nodes. References to sequences extracted from public databases and their categorization are listed in Supplementary Table 1. In (a, b), the centre value of the boxplots represents the median, the boxes denote the interquartile range, and the whiskers represent minimum and maximum values. Statistically significant differences: **P<0.01.