High-resolution melting genotyping of Enterococcus faecium based on multilocus sequence typing derived single nucleotide polymorphisms

Abstract

We have developed a single nucleotide polymorphism (SNP) nucleated high-resolution melting (HRM) technique to genotype Enterococcus faecium. Eight SNPs were derived from the E. faecium multilocus sequence typing (MLST) database and amplified fragments containing these SNPs were interrogated by HRM. We tested the HRM genotyping scheme on 85 E. faecium bloodstream isolates and compared the results with MLST, pulsed-field gel electrophoresis (PFGE) and an allele specific real-time PCR (AS kinetic PCR) SNP typing method. In silico analysis based on predicted HRM curves according to the G+C content of each fragment for all 567 sequence types (STs) in the MLST database together with empiric data from the 85 isolates demonstrated that HRM analysis resolves E. faecium into 231 "melting types" (MelTs) and provides a Simpson's Index of Diversity (D) of 0.991 with respect to MLST. This is a significant improvement on the AS kinetic PCR SNP typing scheme that resolves 61 SNP types with D of 0.95. The MelTs were concordant with the known ST of the isolates. For the 85 isolates, there were 13 PFGE patterns, 17 STs, 14 MelTs and eight SNP types. There was excellent concordance between PFGE, MLST and MelTs with Adjusted Rand Indices of PFGE to MelT 0.936 and ST to MelT 0.973. In conclusion, this HRM based method appears rapid and reproducible. The results are concordant with MLST and the MLST based population structure.

Figure 1. High-resolution melting curves for the eight fragments.

The normalized fluoresence (y-axis) is plotted against temperature in °C (x-axis) and shows the curves for 85 E. faecium isolates. Each graph is labelled with the amplified SNP and combines two HRM runs that have been calibrated against each other; different curves are represented by a different color and accordingly labelled.

Figure 2. High-resolution melting curves for pstS95.

The graphs represent (A and B) HRM curves for two separate runs demonstrating discrimination of curves 53 (red) and 54 (blue) when each run is considered separately; (C) raw data from both runs combined without calibration, demonstrating an inability to discriminate curves 53 and 54; (D) raw data from two isolates included in both runs (solid red and dashed red lines for curve 53 isolate and solid blue and dashed blue for curve 54 isolate) demonstrating a shift in the curves when runs are compared without callibration; (E) data from both runs combined with calibration demonstrating discrimination of curves 53 and 54.

Figure 3. Comparison of HRM and SNP methods.

Cumulative Simpson's Index of Diversity on a log scale for high-resolution melting and single nucleotide polymorphism methods against all 567 s multilocus sequence types.

Figure 4. PFGE dendrogram of 85 isolates.

Dendrogram of pulsed-field gel electrophoresis clusters (major clusters highlighted by grey shading) with corresponding multilocus sequence types (MLST), melting types (MelTs) and in silico determined single nucleotide polymorphism types (SNP types). PFGE clusters are based on 80% similarity.

Figure 5. goeBURST diagram of the Enteroccocus faecium clonal complex 17 population.

goeBURST diagram of the Enteroccocus faecium clonal complex 17 population from the multilocus sequence type (ST) database with superimposed melting types (MelTs). The top 26 MelTs (each with ≥5 associated STs) are represented by different colors. Other STs in grey are represented by smaller MelTs that each have <5 associated STs. STs are indicated by numbers in the circles, the size of each circle represents the relative abundance of that ST, and lines indicate single locus variants.