Inferring the composition of a mixed culture of natural microbial isolates by deep sequencing
Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Next generation sequencing has unlocked a wealth of genotype information for microbial populations, but phenotyping remains a bottleneck for exploiting this information, particularly for pathogens that are difficult to manipulate. Here, we establish a method for high-throughput phenotyping of mixed cultures, in which the pattern of naturally occurring single-nucleotide polymorphisms in each isolate is used as intrinsic barcodes which can be read out by sequencing. We demonstrate that our method can correctly deconvolute strain proportions in simulated mixed-strain pools. As an experimental test of our method, we perform whole genome sequencing of 66 natural isolates of the thermally dimorphic pathogenic fungus Coccidioides posadasii and infer the strain compositions for large mixed pools of these strains after competition at 37°C and room temperature. We validate the results of these selection experiments by recapitulating the temperature-specific enrichment results in smaller pools. Additionally, we demonstrate that strain fitness estimated by our method can be used as a quantitative trait for genome-wide association studies. We anticipate that our method will be broadly applicable to natural populations of microbes and allow high-throughput phenotyping to match the rate of genomic data acquisition.
Author summary
The diversity of the gene pool in natural populations encodes a wealth of information about its molecular biology. This is an especially valuable resource for non-model organisms, from humans to many microbial pathogens, lacking traditional genetic approaches. An effective method for reading out this population genetic information is a genome wide association study (GWAS) which searches for genotypes correlated with a phenotype of interest. With the advent of cheap genotyping, high throughput phenotyping is the primary bottleneck for GWAS, particularly for microbes that are difficult to manipulate. Here, we take advantage of the fact that the naturally occurring genetic variation within each individual strain can be used as an intrinsic barcode, which can be used to read out relative abundance of each strain as a quantitative phenotype from a mixed culture. Coccidioides posadasii , the causative agent of Valley Fever, is a fungal pathogen that must be manipulated under biosafety level 3 conditions, precluding many high-throughput phenotyping approaches. We apply our method to pooled competitions of C. posadasii strains at environmental and host temperatures. We identify robustly growing and temperature-sensitive strains, confirm these inferences in validation pooled growth experiments, and successfully demonstrate their use in GWAS.
