Experimental adaptation of Burkholderia cenocepacia to onion medium reduces host range

Abstract

It is unclear whether adaptation to a new host typically broadens or compromises host range, yet the answer bears on the fate of emergent pathogens and symbionts. We investigated this dynamic using a soil isolate of Burkholderia cenocepacia, a species that normally inhabits the rhizosphere, is related to the onion pathogen B. cepacia, and can infect the lungs of cystic fibrosis patients. We hypothesized that adaptation of B. cenocepacia to a novel host would compromise fitness and virulence in alternative hosts. We modeled adaptation to a specific host by experimentally evolving 12 populations of B. cenocepacia in liquid medium composed of macerated onion tissue for 1,000 generations. The mean fitness of all populations increased by 78% relative to the ancestor, but significant variation among lines was observed. Populations also varied in several phenotypes related to host association, including motility, biofilm formation, and quorum-sensing function. Together, these results suggest that each population adapted by fixing different sets of adaptive mutations. However, this adaptation was consistently accompanied by a loss of pathogenicity to the nematode Caenorhabditis elegans; by 500 generations most populations became unable to kill nematodes. In conclusion, we observed a narrowing of host range as a consequence of prolonged adaptation to an environment simulating a specific host, and we suggest that emergent pathogens may face similar consequences if they become host-restricted.

FIG. 1.

Adaptation by 12 B. cenocepacia populations during 1,000 generations in onion medium. Fitness was quantified by direct competition with the ancestor of the opposite marker; Lac⁺ populations (D1 to D6) and Lac⁻ populations (L1 to L6) are represented as indicated on the figure. Error bars are ± standard error (df = 4).

FIG. 2.

Nematode pathogenicity declines during adaptation to onion medium. Bacterial virulence was measured in monoxenic liquid culture as percent mortality over time. Lac⁺ populations (D1 to D6) and Lac⁻ populations (L1 to L6) evolved for 500 generations in 2% liquid onion medium were introduced to axenically raised nematodes and observed over 7 days. E. coli OP50 (open star) was used as a negative control, and the wild-type (WT) HI2424 Lac⁻ ancestor (filled star) was used as a positive control. Error bars are standard error (df = 2).

FIG. 3.

Differential colonization and pathogenicity of evolved population L1 and the ancestral strain. (A and B) A light microscopy image at a magnification of ×200 of the ancestor marked with red fluorescent protein colonizing C. elegans (A) and confocal fluorescent microscopy of the same nematode (B) are shown. We modified the confocal image to exclude air bubbles (upper right and lower left corners). Ingested bacteria are highly concentrated behind the grinder region of the worm (white arrows) and cause distension throughout the intestine. (C and D) A light microscopy image at a magnification of ×200 of evolved population L1 marked with red fluorescent protein colonizing C. elegans (C) and confocal fluorescent microscopy of the same nematode (D) are shown. Ingested bacteria are distributed throughout the nondistended intestine of the nematode at lower density.

FIG. 4.

Motility, acyl-homoserine lactone production (quorum sensing), and biofilm production of wild-type (WT) B. cenocepacia and 12 populations evolved in 2% onion medium for 500 generations. Swimming motility was measured using 0.3% swim agar (in mm). Biofilm production and C₈ acyl-homoserine lactone production were measured as previously described (25, 37). All values were standardized by the ancestral values for each phenotype (the horizontal line represents the ancestor clone; values greater than 1.0 indicate increases and values less than 1.0 indicate decreases). Genotype designations are as follows: L, Lac⁻ evolved populations; D, Lac⁺ evolved populations; WT, Lac⁺ ancestor clone. Error bars indicate standard error (motility, df = 3; biofilm, df = 5; quorum sensing, df = 5).

FIG. 5.

(A) Relative fitness and percent nematode mortality over time for a single population (L1) of B. cenocepacia evolved in 2% onion medium for 1,000 generations. Error bars are standard error (fitness, df = 4; nematode virulence, df = 2). Increased fitness is temporally correlated with decreased nematode pathogenicity. (B) Dynamics of three phenotypes associated with B. cenocepacia virulence in population L1 (error bars indicate standard errors): top, C₈ acyl-homoserine-lactone production (25) (df = 5); center, biofilm production (37) (df = 5); bottom, motility quantified as radii of movement through 0.3% agar (df = 3).