Microbial community profiling of fresh basil and pitfalls in taxonomic assignment of enterobacterial pathogenic species based upon 16S rRNA amplicon sequencing

• Published in: International journal of food microbiology 2017-09, Vol.257, p.148-156
• Main Authors: Ceuppens, Siele, De Coninck, Dieter, Bottledoorn, Nadine, Van Nieuwerburgh, Filip, Uyttendaele, Mieke
• Format: Article
• Language: English
• Subjects: 16S rRNA database; Algorithms; Annotation algorithms; Annotations; Bacteria; Bioinformatics; Computational Biology; Databases, Factual; Ecological monitoring; Food; Food pathogen; Foodborne Diseases - microbiology; Gene sequencing; Genes; High-Throughput Nucleotide Sequencing; Humans; Microbiology; Microorganisms; Molecular Typing; Ocimum basilicum; Ocimum basilicum - microbiology; Parameter estimation; Pathogens; Pipelines; Real-Time Polymerase Chain Reaction; RNA, Ribosomal, 16S - genetics; rRNA 16S; Salmonella; Salmonella enterica - classification; Salmonella enterica - genetics; Salmonella enterica - isolation & purification; Species
• ISSN: 0168-1605; 1879-3460
• DOI: 10.1016/j.ijfoodmicro.2017.06.016

Application of 16S rRNA (gene) amplicon sequencing on food samples is increasingly applied for assessing microbial diversity but may as unintended advantage also enable simultaneous detection of any human pathogens without a priori definition. In the present study high-throughput next-generation sequencing (NGS) of the V1–V2–V3 regions of the 16S rRNA gene was applied to identify the bacteria present on fresh basil leaves. However, results were strongly impacted by variations in the bioinformatics analysis pipelines (MEGAN, SILVAngs, QIIME and MG-RAST), including the database choice (Greengenes, RDP and M5RNA) and the annotation algorithm (best hit, representative hit and lowest common ancestor). The use of pipelines with default parameters will lead to discrepancies. The estimate of microbial diversity of fresh basil using 16S rRNA (gene) amplicon sequencing is thus indicative but subject to biases. Salmonella enterica was detected at low frequencies, between 0.1% and 0.4% of bacterial sequences, corresponding with 37 to 166 reads. However, this result was dependent upon the pipeline used: Salmonella was detected by MEGAN, SILVAngs and MG-RAST, but not by QIIME. Confirmation of Salmonella sequences by real-time PCR was unsuccessful. It was shown that taxonomic resolution obtained from the short (500bp) sequence reads of the 16S rRNA gene containing the hypervariable regions V1–V3 cannot allow distinction of Salmonella with closely related enterobacterial species. In conclusion 16S amplicon sequencing, getting the status of standard method in microbial ecology studies of foods, needs expertise on both bioinformatics and microbiology for analysis of results. It is a powerful tool to estimate bacterial diversity but amenable to biases. Limitations concerning taxonomic resolution for some bacterial species or its inability to detect sub-dominant (pathogenic) species should be acknowledged in order to avoid overinterpretation of results. •The selection of bioinformatics pipeline may introduce bias on the estimate of microbial diversity.•The natural variability in short 16S rRNA fragments may not be sufficiently large to use conservative annotation methods•NGS 16S amplicon sequencing is not accurate or sensitive enough for detecting Salmonella in food.•Expertise on both bioinformatics and microbiology needs to be combined for analysis of results.