Bioinformatics and antimicrobial resistance genes

Bioinformatics and antimicrobial resistance genes

In my work as a DNA detective (well, an R&D Manager involved in microbial bioinformatics) I see hundreds of microbial genomes yearly, most of which are bacterial. For bioinformaticians, bacteria seem the easiest targets, in comparison to eukaryotic species with huge genomes and complexity. From my point of view, however, bacterial genomes are perfect: you can get a nearly complete genomes with quite little effort, and the data never lie. This provides an amazing view on the developments of microbial biotechnology and microbial diversity.

My life as a DNA detective

Microbial bioinformatics

My job is to analyze the safety aspects of microbial genomes for companies who develop microbial products into the food chain. They may be production organisms for amino acids, enzymes or vitamins, or live micro-organisms. Many of the products will be evaluated by the European Food Safety Authority (EFSA). Whatever the purpose, safety means the absence of acquired antimicrobial resistance genes, genes for toxins and virulence factors, in some cases also describing the genetic modifications in detail. First of all, however, the genome is used for unequivocal taxonomic identification at the species level, sometimes even at the strain level. This is essential in granting the QPS (qualified presumption of safety) status for a microorganism.

The small and complete microbial genomes make the bioinformatics part relatively easy and not time consuming. What does take time, is the interpretation of the outcome of the analysis. I wanted to start this blog to give you a glance at the scientific and technical issues that must be considered when analyzing a genome. In this first post, I will scratch the surface of my favorite and at the same time my least favorite subject, antimicrobial resistance (AMR) genes.
I have a copy of the “Guidance on the characterization of microorganisms used as feed additives or as production organisms”,, (subsequently simply “guidance”) always on my desk. We actually anticipated certain aspects of the guidance – such as the requirement of whole genome sequencing – well before it was published and adopted in 2017. A safety aspect that EFSA has always emphasized if the demonstrated absence of transmissible resistances to antibiotics with clinical or veterinary importance. EFSA states in the guidance that intrinsic AMR is not considered a safety concern, “intrinsic” meaning something that is typical of “all the strains of that species”.

On the surface it seems clear, but when do you know you have analyzed “all strains of that species”? Never, I suppose. And, AMR genes are not always necessary for the survival of the microorganism, making them dispensable. This means that they can be replaced by e.g. mobile elements in the genome, leading to susceptibility of that particular strain to the antimicrobial. Suddenly there you have one susceptible strain, although the rest of the strains still carry the AMR gene. Also, the phenotype does not necessarily tell anything about a trait being acquired or intrinsic – if there have been modifications to the regulation of the AMR gene in one strain, which then becomes susceptible, it does not mean that the rest of the strains carry an acquired AMR gene. The requirement “all strains” simply can’t be accepted – and I think this is also understood in EFSA. Perhaps the guidance could be written in a way that allows some room for interpretation.
We are advised to interpret the results of the genome investigation in the light of minimal inhibitory concentrations (MIC) for the strain. In my opinion, the interpretation is clear in only two cases: if the genome search reveals no AMR candidates, and the MIC values are below the cut-off (safe strain); and if the genome search reveals a clearly acquired AMR gene which can be directly linked to the MIC value above the cut-off (hazard). In the latter case, one of course has to identify the gene as acquired by experience or by comparing it to other genomes, if available. (So, perhaps only one clear case). In all other cases the situation is more complex, and almost all genomes carry genes that are known to contribute to antimicrobial resistance. For example, did you know that the E. coli K-12 strain (generally considered as safe lineage) has matches to more than 70 genes that contribute to AMR?
We have developed our “Biosafe” way of dealing with AMR genes and interpreting whether they are a concern or not. In practice, we have to have multidisciplinary know-how of bioinformatics and microbiology, and in-depth EFSA knowledge to survive in the regulatory jungle.

Pauliina Halimaa