NEW YORK (GenomeWeb News) – In a study published online today in Genome Research, an international team of investigators in the UK, Turkey, and elsewhere described the genome sequencing and genome-wide association study approach it used to assess virulence patterns in methicillin-resistant Staphylococcus aureus.
The University of Bath's Ruth Massey and colleagues sequenced the genomes of nearly 100 MRSA isolates — a collection that showed widely variable toxicity profiles in their cell- and mouse-based assays.
The resulting sequencing data made it possible to perform a GWAS of that toxicity variation, revealing large sets of loci linked to toxicity in the bugs. The team's analysis also led to a genetic signature that appeared to coincide with the most pronounced toxicity patterns in the MRSA isolates. Those snippets of sequence subsequently proved useful for predicting which isolates would have high or low toxicity in a mouse model of MRSA infection.
"[T]he standard approach [to assessing MRSA toxicity] has always been to focus on a single or small number of genes and proteins," Massey said in a statement. "As the cost and speed of genome sequencing decreases, it is becoming increasingly feasible to sequence the genome of an infecting organism."
In an effort to take a wide-reaching look at potential contributors to MRSA virulence and infection severity, she and her colleagues used Illumina's GAII instrument to sequence 90 MRSA isolates form the same sequence type, ST239.
The team also turned to multiple types of binding and toxicity assays to characterize the set of MRSA isolates. For the most part, those analyses revealed only subtle differences in adhesiveness from one isolate to the next. On the other hand, the bugs exhibited wide variations in their apparent toxicity.
When they compared and contrasted genome sequences, the researchers tracked down more than 3,000 SNP sites in the isolates, which were employed in a GWAS of MRSA toxicity that relied on machine learning steps.
That search uncovered 121 SNPs and nearly two-dozen small insertions or deletions with potential ties to toxicity, including a handful of loci showing particularly pronounced associations.
Isolates that were most toxic tended to share specific SNPs and indels, prompting the study's authors to propose that these SNPs might comprise a genetic signature for predicting toxicity.
Using information at 31 SNPs and 21 indels representing a broad range of the toxicity-associated linkage groups detected in the GWAS, the investigators classified the available ST239 MRSA isolates as low, medium, or high toxicity.
Results in MRSA-infected mice indicated that the genetic signature predicted toxicity with more than 85 percent accuracy. The signature was most reliable when clustering high- and low-toxicity bugs, researchers noted, showing poorer accuracy for isolates in the middle-of-the-road toxicity group.
Even so, if similar toxicity signatures exist in human infections, the researchers speculated that it may be possible to use targeted genotyping or more widespread sequencing to classify MRSA isolates in the clinic so that aggressive treatment and isolation can be applied more judiciously.
"In addition to improving and personalizing the care of patients infected with highly toxic bacteria," they noted, "it would also prevent the needless and deleterious administration of cocktails of potent and expensive antibiotics to patients with low toxicity infections."
Members of the team are now taking a crack at applying the same genome sequencing approach to assess MRSA isolates from the USA300 sequence type, a strain of epidemic MRSA isolates implicated in worrisome community-acquired infections. They also plan to use a similar approach for studying other bacterial culprits, such as the respiratory tract infection-causing pathogen Streptococcus pneumonia.