Plant-associated symbiotic Burkholderia species lack hallmark strategies required in mammalian pathogenesis

Abstract

Burkholderia is a diverse and dynamic genus, containing pathogenic species as well as species that form complex interactions with plants. Pathogenic strains, such as B. pseudomallei and B. mallei, can cause serious disease in mammals, while other Burkholderia strains are opportunistic pathogens, infecting humans or animals with a compromised immune system. Although some of the opportunistic Burkholderia pathogens are known to promote plant growth and even fix nitrogen, the risk of infection to infants, the elderly, and people who are immunocompromised has not only resulted in a restriction on their use, but has also limited the application of non-pathogenic, symbiotic species, several of which nodulate legume roots or have positive effects on plant growth. However, recent phylogenetic analyses have demonstrated that Burkholderia species separate into distinct lineages, suggesting the possibility for safe use of certain symbiotic species in agricultural contexts. A number of environmental strains that promote plant growth or degrade xenobiotics are also included in the symbiotic lineage. Many of these species have the potential to enhance agriculture in areas where fertilizers are not readily available and may serve in the future as inocula for crops growing in soils impacted by climate change. Here we address the pathogenic potential of several of the symbiotic Burkholderia strains using bioinformatics and functional tests. A series of infection experiments using Caenorhabditis elegans and HeLa cells, as well as genomic characterization of pathogenic loci, show that the risk of opportunistic infection by symbiotic strains such as B. tuberum is extremely low.

Figure 1. Distribution of virulence and symbiotic loci across Burkholderia species.

Virulence-associated loci were identified using BLASTP against characteristic sequences (B. pseudomallei or B. cenocepacia ATPases for secretion systems, B. pseudomallei pilus genes for pilus-related clusters (red), and B. tuberum nifH and nodA sequences (green)). Non-canonical or truncated clusters are indicated in pink, and clusters highly associated with virulence for the Type 3 and Type 6 secretion systems are denoted with an asterisk.

Figure 2. Flagellar gene clusters in Burkholderia.

All Burkholderia strains share a highly similar chemotaxis and flagellar gene cluster (fla1) on chromosome 1. Although the A group and the pathogenic B group share high homology in gene sequence and chromosomal arrangement, a phylogenetic analysis of five concatenated protein sequences (FliC, FlgM, FlgE, FlhB, FlgJ) shows a distinct clustering of the two lineages [A]. Additionally, the A group cluster is arranged entirely sequentially [B] whereas the pathogenic B cluster is split into four different regions throughout chromosome 1 (the B. mallei cluster is split into 5 regions)[C]. The fla1 gene cluster is responsible for Burkholderia motility on soft agar, but not for intracellular motility or plaque formation in models of infection . A second flagellar gene cluster on chromosome 2 (fla2) is necessary for this intracellular motility. This second cluster is present only in the pathogenic strains, i.e. B. pseudomallei, B. thailandensis, and B. oklahomensis [D], and a similar cluster exists on chromosome 2 of the opportunistic pathogen B. cenocepacia [E].

Figure 3. Phylogenetic analysis of Type 4 secretion system clusters in Burkholderia species.

Three clusters with a unique gene organization have the characteristic VirB and VirD4 proteins found in the Agrobacterium tumefaciens cluster. A fourth, truncated cluster (yellow) is also found in many Burkholderia strains, including B. tuberum STM678^T, but is unlikely to contribute to secretion.

Figure 4. Relationship between the Burkholderia Type 6 secretion systems and arrangement of gene clusters.

Clusters were identified with BLASTP against the B. pseudomallei T6SS-2, and a concatenated neighbor-joining phylogenetic tree was generated in MEGA 5.1 using highly divergent protein sequences–VgrG, Hcp, ClpV, IcmF, and the lysozyme-like protein VC_A0109. The gene arrangements of the cluster were manually verified. Although B. oklahomensis EO147 was found to have five T6SS clusters (Fig. 1), it was excluded from the analysis because the draft quality of the genome sequence did not allow for full reconstruction of the gene arrangement of the operon. The four symbiotic strains emphasized in this study are highlighted in bold. The R. leguminosarum imp region clusters with the T6SS-4 group and is indicated with an asterisk. The T6SSa and TG6SSb clusters are found only in the environmental and symbiotic Burkholderia species.

Figure 5. Antibiotic-resistance profiles of pathogenic and symbiotic Burkholderia species.

Relative resistance to a panel of antibiotics for the four environmental and symbiotic strains emphasized in this study compared to the pathogenic B. thailandensis E264, the opportunistic pathogen B. vietnamiensis G4, and the plant pathogen B. gladioli BSR3. The average clearing of five replicate experiments at the highest antibiotic concentration are represented in the heat map. Full resistance (no clearing) is indicated with an asterisk.

Figure 6. Symbiotic Burkholderia species are not pathogenic in vitro to the nematode C. elegans using the slow-killing assay.

The edge of bacterial lawns of test Burkholderia strains were seeded with age-synchronized C. elegans N2 juvenile worms on nematode growth agar (NGM) plates. An initial count of live worms was made, and again after 24, 48, and 72 h. The percent of survivors were enumerated based on a comparison of live worms present after 72 h versus the initial count. Four symbiotic species of Burkholderia (B. tuberum STM678^T, B. silvatlantica PVA5, B. silvatlantica SRMrh20^T, B. unamae MTI641^T) and one pathogenic species (B. thailandensis E264) were compared against the control, E coli OP50. Only the pathogenic species showed a significant and dramatic reduction in the number of live worms over the course of the experiments. All other test strains showed nearly 100% survival. Data from 48 h after the start of the experiment are shown. Error bars indicate standard error.

Figure 7. Symbiotic Burkholderia species are the preferred nutritional source of C. elegans compared to a pathogenic species.

E. coli and symbiotic Burkholderia species were individually compared to B. thailandensis E264 to examine the response of the nematode C. elegans to a nutritional preference. Live worms were age-synchronized and seeded into the center of a nematode growth agar (NGM) plate equidistant from a lawn of B. thailandensis E264 and the aforementioned non-pathogenic strains. Counts of live worms were taken initially after seeding, and at 24, 48, and 72 h intervals. In each competition assay, the B. thailandensis lawn was either avoided or contained mostly dead worms, whereas live worms thrived within the lawns of E. coli OP50 or symbiotic Burkholderia species. Error bars indicate standard error.

Figure 8. Symbiotic Burkholderia species are not toxic to HeLa cells in culture.

Symbiotic Burkholderia species (B. tuberum STM678^T, B. silvatlantica PVA5, B. silvatlantica SRMrh20^T, B. unamae MTI641^T) were compared to B. thailandensis E264 to determine if they were toxic to mammalian cells grown in culture. HeLa cells were grown until confluent and inoculated with an MOI of 50. After 8 and 24 h, samples of the supernatant of the inoculated and sham-inoculated cells were examined for LDH release using the Cyto-Tox 96 assay kit and spectrophotometric reading at 490 nm. At 8 h, only the positive control cells treated with detergent showed significant cell lysis. B. thailandensis E264 caused cell lysis about three times greater than that of the negative control (medium only). The effect of the symbiotic species was not statistically different from that of the negative control. At 24 h, B. thailandensis E264-induced morbidity had surpassed the detergent control, while the cytotoxicity caused by the symbiotic species increased only slightly from the earlier time point. Error bars indicate standard error.