Comparing different molecular typing methods of Listeria monocytogenes isolates from human and food circulating in Italy

Listeria monocytogenes is a Gram-positive, facultative intracellular pathogen that can cause severe disease in humans, mammals and birds. Even if L. monocytogenes is nearly ubiquitous in the environment (water, vegetation, soil), human listeriosis is relatively rare. Moreover, human listeriosis is a food-borne disease and most common means of transmission are ready-to-eat food (seafood, dairy, meat) as well as fresh food contaminated by environmental sources. In addition, L. monocytogenes can persist in the environment forming biolms and it can grow in extreme conditions of both temperature and salinity [1]. Given the wide spreading of Listeria organisms and the intra genera variability (there are more than 13 serotypes, few of them responsible for human listeriosis), subtyping of clinical and food isolates is needed to establish epidemiologic links during outbreaks or routine surveillance programs. Subtyping and strain characterization methods include pulsed eld gel electrophoresis (PFGE), serotyping, ribotyping but also sequencing based methods as the multi locus sequence typing (MLST), multi-locus-tandem-repeat analysis (MLVA) and, recently, clustering methods based on whole genome sequencing (WGS) [2]. The aim of this work is to compare molecular typing results from dierent methods applied on human and food isolates, in order to highlight homologies for source attribution.To build a representative collection of circulating clinical Listeria monocytogenes strains a total of 114 isolates were chosen among clinical cases notied from 2006 to 2014 in Northern Italy on the basis of serotype and pulse-type. A list of isolates present in the collection at National Reference Laboratory for Listeria monocytogenes were selected among the strains matching the PFGE proles of the human strains. The isolates were characterized by PFGE [3] and 7-locus MLST [4], sequence types were determined according to the MLST Pasteur database (www.pasteur.fr/mlst) while WGS of the isolates was performed using a trasposome-based Nextera XT kit to generate genomic libraries and an Illumina NextSeq 500 NGS platform. Quality control and Trimming of obtained reads were performed by in-house developed python scripts while strains clustering was performed according to two procedure, both of them based on a SNP matrix. The rst one is based on a reference-based Genome-wide Single Nucleotide Polymorphism (SNP) matrix and was carried out using the FDA SNP-pipeline program [5]; the second one is based on an alignment-free approach based on a k-mer search algorithm and was performed with the kSNP program [6]. In both cases distance tree (NJ or UPGMA) were built using MEGA6 [7]. In addition MLST typing was derived, for comparison, directly from reads data by using the SRST2 software [8]. The collection consists of strains belonging to ve serotypes that were further subtyped by PFGE. The 1/2a serotype resulted as the most frequent followed by the 4b, 1/2b, 1/2c and the 3a respectively. The 124 food and environmental selected strains and the 114 human isolates were then deeply sequenced and compared. The total number of sequenced sample is reported in Table 1, while the number of dierent pulse-types and sequence types (STs) are reported in Table 2. Strains clustering was performed separately for each serotype. Overall results conrmed strains clustering based on PFGE and MLST analysis. Moreover, even if clinical and environmental strains shared the same typing prole, NGS analysis highlighted their genetic distance, that drove isolates linking; in few cases, as in the examples highlighted in the boxes within the gures, a substantial genetic relationship among clinical and food isolates is evident. Furthermore, the STs derived from NGS data were exactly the same obtained previously by PCR followed by DNA sequencing. This study conrmed the high discriminatory power of WGS in comparison to other molecular typing methods, as well as the possibility of deriving MLST proling directly from short reads data with a consistent saving of time and money. Anyway, PFGE and serotyping still remain primary methods for molecular typing of Listeria monocytogenes strains while WGS is still limited to a small group of laboratories. This is due mainly to the high cost of NGS equipment but also to a shared standardization of analysis procedures that is still lacking. Our study shows how NGS represents a powerful technique for deep characterization of isolates and it also supports a functional characterization for those strains showing convergent PFGE and MLST proles. In fact, WGS analysis can identify a common origin of strains, better than other molecular typing methods. Moreover, NGS can provide additional information about the physiology of clinical, food and environmental isolates, such as for investigating antibiotic resistance genes or virulence and persistence factors. In the future, large scale sampling and WGS analysis of food and human isolates could increase the identication of L. monocytogenes clusters, improving surveillance programs and genomic studies.