The Utility of High-Resolution Melting Analysis of SNP Nucleated PCR Amplicons—An MLST Based Staphylococcus aureus Typing Scheme

Rachael A. Lilliebridge,Steven Y.C. Tong,Deborah C. Holt,Philip M. Giffard,Michael Otto

doi:10.1371/journal.pone.0019749

Rachael A. Lilliebridge, Steven Y.C. Tong + Show 3 more

Open Access

https://doi.org/10.1371/journal.pone.0019749

Copy DOI

Abstract

High resolution melting (HRM) analysis is gaining prominence as a method for discriminating DNA sequence variants. Its advantage is that it is performed in a real-time PCR device, and the PCR amplification and HRM analysis are closed tube, and effectively single step. We have developed an HRM-based method for Staphylococcus aureus genotyping. Eight single nucleotide polymorphisms (SNPs) were derived from the S. aureus multi-locus sequence typing (MLST) database on the basis of maximized Simpson's Index of Diversity. Only G↔A, G↔T, C↔A, C↔T SNPs were considered for inclusion, to facilitate allele discrimination by HRM. In silico experiments revealed that DNA fragments incorporating the SNPs give much higher resolving power than randomly selected fragments. It was shown that the predicted optimum fragment size for HRM analysis was 200 bp, and that other SNPs within the fragments contribute to the resolving power. Six DNA fragments ranging from 83 bp to 219 bp, incorporating the resolution optimized SNPs were designed. HRM analysis of these fragments using 94 diverse S. aureus isolates of known sequence type or clonal complex (CC) revealed that sequence variants are resolved largely in accordance with G+C content. A combination of experimental results and in silico prediction indicates that HRM analysis resolves S. aureus into 268 “melt types” (MelTs), and provides a Simpson's Index of Diversity of 0.978 with respect to MLST. There is a high concordance between HRM analysis and the MLST defined CCs. We have generated a Microsoft Excel key which facilitates data interpretation and translation between MelT and MLST data. The potential of this approach for genotyping other bacterial pathogens was investigated using a computerized approach to estimate the densities of SNPs with unlinked allelic states. The MLST databases for all species tested contained abundant unlinked SNPs, thus suggesting that high resolving power is not dependent upon large numbers of SNPs.

Highlights

The identification and monitoring of strains of pathogenic bacteria by genotyping is central to public health, infection control, and veterinary and food microbiology
We identified a set of single-nucleotide polymorphisms (SNPs) optimised for resolving S. aureus strains from each other, and designed High resolution melting (HRM) fragments of the appropriate size that contained these SNPs
We conservatively assumed that sequence variations that do not affect the G+C content are not revealed by HRM, even though this is not always the case

Summary

Introduction

The identification and monitoring of strains of pathogenic bacteria by genotyping is central to public health, infection control, and veterinary and food microbiology. A previously described approach to SNP-based bacterial genotyping involved the computerized derivation from sequence alignments of sets of SNPs that are optimized with respect to the Simpson’s Index of Diversity (D), and the interrogation of these SNPs using allele specific real-time PCR (kinetic PCR). In this context, D is the probability that two sequences in the alignment, selected at random without replacement, will be discriminated by the SNPs. In this context, D is the probability that two sequences in the alignment, selected at random without replacement, will be discriminated by the SNPs This approach has been applied to several species, and in general has been based upon SNPs derived from MLST datasets [1,2,3,4,5]

Methods

Results

Conclusion