FREQ-Seq: a rapid, cost-effective, sequencing-based method to determine allele frequencies directly from mixed populations

Abstract

Understanding evolutionary dynamics within microbial populations requires the ability to accurately follow allele frequencies through time. Here we present a rapid, cost-effective method (FREQ-Seq) that leverages Illumina next-generation sequencing for localized, quantitative allele frequency detection. Analogous to RNA-Seq, FREQ-Seq relies upon counts from the >10(5) reads generated per locus per time-point to determine allele frequencies. Loci of interest are directly amplified from a mixed population via two rounds of PCR using inexpensive, user-designed oligonucleotides and a bar-coded bridging primer system that can be regenerated in-house. The resulting bar-coded PCR products contain the adapters needed for Illumina sequencing, eliminating further library preparation. We demonstrate the utility of FREQ-Seq by determining the order and dynamics of beneficial alleles that arose as a microbial population, founded with an engineered strain of Methylobacterium, evolved to grow on methanol. Quantifying allele frequencies with minimal bias down to 1% abundance allowed effective analysis of SNPs, small in-dels and insertions of transposable elements. Our data reveal large-scale clonal interference during the early stages of adaptation and illustrate the utility of FREQ-Seq as a cost-effective tool for tracking allele frequencies in populations.

Figure 1. Illustration of the FREQ-Seq method.

A.) The bridging primers can be produced in house using standard PCR and purification. Individual, uniquely bar-coded bridging primers are amplified from the FREQ-Seq plasmid library set using two primers ABC1 (AATGATACGGCGACCAC) and ABC2 (ACTGGCCGTCGTTTTAC). The resulting bridging primers are stored as dsDNA at −20°C. B.) Construction of FREQ-Seq Illumina libraries is conducted in two stages. A first round of PCR amplification is used to generate an allele-specific, representative fragment pool with product inserts approximately 250 bp in size. Each product at this stage contains an M13f sequence at the sequencing end and an Illumina B adapter sequence at the non-sequencing end. These products are then treated with exonuclease I or Ampure XP beads to remove remaining PCR primers. In a second stage of PCR amplification, individual samples are barcoded and enriched using three primers; a small amount of a FREQ-Seq bridging primer and two Illumina enrichment primers. Initially, amplification proceeds via the sample-specific bridging primer that has the M13f sequence at its 3′ end. The resulting products can then be amplified to higher quantity with the generic, enrichment primers A and B. Following sample pooling and standard PCR product purification, the resulting products constitute a complete FREQ-Seq library.

Figure 2. Calibration of FREQ-Seq data for three ancestral/evolved (WT/EVO) allele mixtures pntAB^EVO (A.), gshA^EVO (B.), and fghA^EVO (C.).

Strains bearing each evolved allele were compared to one with the wild-type version that also expressed mCherry to permit precise quantification of expected ratios by flow cytometry (Table S3). Presented in each column are sequence variation between the ancestral WT and EVO alleles, and subsequent quadratic fits of the data for each respective allele. Observed data are the frequencies of the evolved allele as determined by FREQ-Seq, whereas expected data are wild-type allele frequencies estimated using flow cytometry. Quadratic model fitting was performed in R.

Figure 3. Allele frequencies for four beneficial alleles in population F4 (A) with correlated changes in relative growth rate (B).

Alleles gshA^EVO (circles), fghA^EVO (triangles), pntAB^EVO (squares), and icuAB^EVO (diamonds) were determined by FREQ-Seq and corresponding values were corrected with calibrated data shown in Figure 2 . Growth rates (B.) were normalized to the growth rate of the ancestral strain.

Figure 4. Genotyping of 72 isolates from population F4 at generation 150.

Genotypes were determined by PCR amplification and RFLP analysis using enzymes HhaI and HpyAV for the gshA, and fghA alleles, respectively. trfA::ISMex25 alleles were assayed by standard PCR amplification.