Microbes must adapt to the presence of other species, but it can be difficult to recreate the natural context for these interactions in the laboratory.  We describe a method for inferring the existence of symbiotic adaptations by experimentally evolving microbes that would normally interact in an artificial environment without access to other species.  By looking for changes in the fitness effects microbes adapted to isolation have on their partners, we can infer the existence of ancestral adaptations that were lost during experimental evolution.  The direction and magnitude of trait changes can offer useful insight as to whether the microbes have historically been selected to help or harm one another in nature.  We apply our method to the complex symbiosis between the social amoeba Dictyostelium discoideum and two intracellular bacterial endosymbionts, Paraburkholderia agricolaris and P. hayleyella.  Our results suggest P. hayleyella – but not P. agricolaris – has generally been selected to attenuate its virulence in nature, and that D. discoideum has evolved to antagonistically limit the growth of Paraburkholderia.  The approach demonstrated here can be a powerful tool for studying adaptations in microbes, particularly when the specific natural context in which the adaptations evolved is unknown or hard to reproduce.

Estimation of D. discoideum fitness via spore production assays – In order to observe how evolution in isolation affected P. agricolaris and P. hayleyella’s impact on host fitness, we compared total spore production of ancestral D. discoideum strains of interest in the presence or absence of ancestral or evolved Paraburkholderia.  We suspended D. discoideum spores in KK2 buffer to a concentration of 10⁶ spores/mL.  We suspended K. pneumoniae and suspended ancestral or evolved Paraburkholderia strains in KK2 buffer at an OD₆₀₀ of 2.0.  We inoculated SM/5 plates with 100uL of the D. discoideum suspension (~10⁵ spores) and 500uL of either 100% K. pneumoniae suspension or a 95:5 (vol:vol) mixture of K. pneumoniae and Paraburkholderia suspensions.  We incubated plates at room temperature under ambient light for 7 days, after which we harvested all spores by washing plates with 10mL of a detergent solution of 0.1% NP40 in KK2.  We diluted the resulting spore suspensions and counted on a hemocytometer to estimate total spore production across the entire plate.

Spore production assay data is contained in the 'sporecountdata.csv' file.  Each line represents a single data point - a total spore count from a single agar plate, obtained as described above - along with several categorical variables to specify which strains, lines, and species of Dictyostelium and Paraburkholderia it represents.

Estimation of Paraburkholderia fitness via growth rate assays – In order to observe how evolution in isolation affected D. discoideum’s impact on Paraburkholderia fitness, we compared the maximum specific growth rate of ancestral Paraburkholderia strains in liquid culture in the presence or absence of ancestral or evolved D. discoideum.  We inoculated Paraburkholderia strains into liquid SM/5 media and grew them to stationary phase overnight at 30°C in a shaking incubator.  We quantified bacterial suspensions using a spectrophotometer and diluted them to an OD₆₀₀ of 0.1.  We prepared D. discoideum amoeba suspensions by inoculating SM/5 plates, incubating them for ~48 hours, and harvesting amoebae by washing plates with 10mL KK2.  We centrifuged amoeba suspensions and washed them with KK2 3-6 times to remove weakly-associated K. pneumoniae, then resuspended them in KK2 buffer containing 300ug/mL tetracycline for 1 hour.  We then centrifuged and washed the antibiotic-treated suspensions 2 more times with fresh KK2 to remove residual tetracycline, counted them using a hemocytometer, and diluted them to 10⁷ amoebae/mL.  We obtained growth curves from a Tecan Infinite M200 PRO microplate reader.  We prepared wells of a 96 well plate by combining 100uL SM/5 broth with either 10uL OD₆₀₀ 0.1 Paraburkholderia suspension or 10uL KK2 and either 10uL 10⁷ amoebae/mL tetracycline-treated D. discoideum suspension or 10uL KK2.  We took OD₆₀₀ measurements every 15 minutes for 48 hours to produce a growth curve for each well.  We fit growth curves and calculated maximum specific growth rates using the fitr script (https://github.com/dcangst) in R.

Growth rate assay data is contained in the 'growthratedata.csv' file.  Each line represents a single data point - a maximum growth rate measured from a growth curve as described above - along with several categorical variables to specify which strains, lines, and species of Dictyostelium and Paraburkholderia it represents.  Also included is the 'growthratedata_unprocessed.csv' file, which includes the growth curves in their entirety.  Maximum growth rates were determined from these data using the fitr script as described in the accompanying R script file.

See comments in R scripts for usage instructions.

NOTE:  Data files make use of categorical variables 'Bclade', 'Bstrain', 'Bline', and 'Bstatus', but the published manuscript instead calls them 'Pclade', 'Pstrain', 'Pline', and 'Pstatus'.  This is the result of Paraburkholderia agricolaris and P. hayleyella being reassigned from the genus Burkholderia to Paraburkholderia during the study.  They are referring to the same organisms.