The burden of foodborne disease in Australia has been estimated at an annual cost of AUD$1.2 billion, comprising approximately 5.4 million cases of disease (The OzFoodNet Working Group, 2012). Australia is divided into eight different states or territories, each with its own health department coordinating surveillance for between 10 and 15 foodborne diseases, including listeriosis. As in many other countries, incidence rates of listeriosis are generally far lower than other common foodborne illness, such as campylobacteriosis or salmonellosis, with the rate varying between 0.3 and 0.4 cases per 100,000 since 2008, with most cases/highest rates occurring in age groups over the age of 60 (Rosen, 2002; The OzFoodNet Working Group, 2012). A similar association with high mortality rates is also shared by other countries, with 20–30% of hospitalized cases of invasive listeriosis resulting in fatality (Dalton et al., 2011; Scallan et al., 2011; The OzFoodNet Working Group, 2012). Recent evidence suggests that this burden may be higher, particularly for pregnant women or patients with neurolisteriosis (Charlier et al., 2017).

In addition to the public health burden, the associated costs to the food industry are high – Listeria monocytogenes is the main contaminant linked to food recalls in Australia due to microbial contamination, with 45% of these recalls from 2005 through to 2014 due to this organism (based on FSANZ¹ data); where ready-to-eat (RTE) meat products represented the largest group of recalled products.

In efforts to address these public health and economic costs, countries have employed complex surveillance systems designed to provide knowledge of the epidemiology and population dynamics of L. monocytogenes, and to improve detection and response to associated outbreaks of disease (Kirk et al., 2008; Wang et al., 2009; Fox et al., 2012). OzFoodNet was established in Australia in 2000, with government-funded epidemiologists appointed to states and territories to improve surveillance of foodborne disease (Kirk et al., 2008). From 2010 onward, OzFoodNet has enhanced national surveillance of listeriosis, and now collects additional data on invasive L. monocytogenes isolates, incorporating molecular sub-typing data to facilitate rapid and precise identification of disease clusters (The OzFoodNet Working Group, 2012). Although recent studies have yielded detailed insights into the population distribution of L. monocytogenes globally as well as source associations of important subgroups, such as the over-representation of sequence type (ST) 121 in food sources or of clonal complex (CC) 1 in clinical cases, this data is lacking in the context of Australia (Chenal-Francisque et al., 2011; Maury et al., 2016; Moura et al., 2016).

An important component of understanding the epidemiology of foodborne disease is knowledge of the occurrence and molecular ecology of strains isolated from food products, and indeed the food chain as a whole (Fox et al., 2012). Such information can facilitate insights into the distribution of certain strains within different food chains or environments, and enable relevant associations to be made. Temporal analysis of both clinical and non-clinical surveillance data can allow monitoring of the occurrence of individual strain sub-types or epidemic clones over time, and provides improved understanding of potential risk of disease, and where corrective efforts may be directed. Recent studies have provided insights into global, continental and/or national trends in this area, such as the association of ST121 to food sources, the predominance of CC2 and CC1 globally and association of CC1 with outbreaks of disease, or the dominance of the ST328 subgroup in India (Chenal-Francisque et al., 2011; Haase et al., 2013; Yin et al., 2015; Barbuddhe et al., 2016; Maury et al., 2016). In addition to this, genomic analysis can provide insights into characteristics such as strain virulence. The invasion protein InlA, for example, plays a key role in invasive listeriosis by mediating translocation across the intestinal epithelium (Schäferkordt and Chakraborty, 1997). Mutations in the inlA gene have been shown to impact pathogenesis, with premature stop codons (PMSCs) leading to reduced invasion of the infected host (Nightingale et al., 2005; Chen et al., 2011).

Previous studies have reported on the incidence of listeriosis across Australia, and identified risk groups, outbreaks, and risk factors associated with the disease (Dalton et al., 2011; The OzFoodNet Working Group, 2012). In this study, we present molecular typing analyses of L. monocytogenes isolated from clinical, environmental, and food sources. Data has been interrogated to identify the prevalent STs among the L. monocytogenes population in Australia, and dominant strains have been identified, including their associated food chains and links to clinical illness. This study provides knowledge about L. monocytogenes across clinical and non-clinical settings in Australia, and together with previous studies (Dalton et al., 2011; The OzFoodNet Working Group, 2012; Kwong et al., 2016) provides new insights into the epidemiology of listeriosis in Australia and the organism’s associated genetic traits.

Building a detailed understanding of pathogen population demographics, along with clinical and food chain association, is central to building a robust response and control system directed at minimizing the impact of foodborne pathogens to public health. The data presented in this study provides detailed insight into the epidemiology of L. monocytogenes in an Australian context by using application of molecular techniques including molecular serotyping, MLST, and analysis of the inlA virulence gene marker.

Serotyping was one of the first subtyping methods established to investigate L. monocytogenes epidemiology. It has been applied extensively as a rapid means to characterize isolates, with applications in both understanding the relevance of certain serotypes to illness in humans and animals, as well as being employed as a tool to assist in outbreak investigations (Jonquières et al., 1998). Examination of the serotypes of all isolates in this study identified the IIA, IIB, and IVB groupings as the major serotypes (41, 30, and 23%, respectively). This distribution is similar to that observed in other countries, where these groupings, along with the IIC grouping (1/2c and 3c) dominate the L. monocytogenes population (Bille and Rocourt, 1996; Wang et al., 2009). Analysis of clinical isolates from Portugal from 1997 to 2004 showed the same top three serogroups, with IVB the most common, followed by IIB and then IIA (72, 18, and 11%, respectively) (Almeida et al., 2010). Of particular note is the absence of serogroup IIC among the Portuguese clinical isolates which comprised 8% of Australian clinical cases in this study. A serogroup hierarchy was noted among clinical isolates in this study with IVB being the most common, followed by IIB then IIA (56, 19, and 17%, respectively). Although lineage I (which includes IVB and IIB serogroups) has previously been identified as over-represented among clinical cases of listeriosis (Sauders et al., 2006), as was seen in this study, it should be noted that lineage II (which includes the IIA and IIC serogroups) has been equally significant in causing infection in some countries (Leong et al., 2015). This demonstrates the importance of lineage II strains in human listeriosis, and although approximately 25% of clinical isolates in this study were lineage II suggesting they remain an important etiological agent of listeriosis in Australia, it is worth noting that serogroup IIA was significantly under-represented among clinical isolates relative to the overall population structure (Supplementary Table S2).

The application of MLST to characterize L. monocytogenes has yielded many insights into the overall structure of the species population. It has served to establish genetic lineages (I, II, III, and IV), as well as define CCs of highly related L. monocytogenes strains, which have been shown to have global distribution (Meinersmann et al., 2004; Ragon et al., 2008). Analysis of the ST distribution among the population (Figure 2) identified two main groupings, comprising either lineage I or lineage II isolates. Such definitive separation of lineage I and II has previously been observed using similar analysis of large L. monocytogenes ST datasets (Haase et al., 2013). The two largest ST groupings identified, ST3 and ST1 comprising 40% of all isolates (n = 90) are both lineage I. These have also been identified globally among the most common STs found in both clinical and non-clinical settings (Ragon et al., 2008; Haase et al., 2013). Clinical isolates were significantly over-represented among the CC1 sub-group, in contrast to an under-representation of CC3 isolates among lineage I clinical isolates (Supplementary Table S2). This strong association of CC1 clones with clinical illness has been identified in a recent large study which determined source associations and virulence of L. monocytogenes subpopulations (Maury et al., 2016), and it has been hypothesized that this subgroup may have increased virulence relative to others (Henri et al., 2016). That large study also noted ST121 and ST9 as significantly over-represented in food sources, and although all the six ST121 isolates in this study were similarly of food origin, no significant association of CC9 isolates in this study were observed with respect to clinical or food sources (Supplementary Table S2).

Many dominant subgroups identified in this study have also been identified as major subgroups in other studies, such as CC1, CC2, CC3, CC9, and CC155 (Chenal-Francisque et al., 2011). In addition to this geographical trends have previously been reported in relation to the L. monocytogenes population structure; for example in India a ST328 clone has been identified as widely disseminated among multiple sources including clinical cases (Barbuddhe et al., 2016). Interestingly in this study, ST204 was identified as the third most common ST with isolates from both clinical and a range of non-clinical sources. This is in contrast to other similar studies which have not identified, or recorded at lower abundance, ST204 isolates among L. monocytogenes populations from sources outside Australia (Chenal-Francisque et al., 2011; Haase et al., 2013; Unterholzner et al., 2013; Stessl et al., 2014). A notable exception being a recent study on L. monocytogenes isolated from various sources in Switzerland reporting ST204 isolates primarily from meat sources (Grohmann et al., 2003). ST204 isolates in this study were identified in a range of sources, including meat, dairy, environmental and clinical samples; in the case of ST204 isolates in this study meat sources were significantly over-represented (Supplementary Table S2). Further studies into the genomics of ST204 isolates identified a range of accessory genes involved in different stress response mechanisms among ST204 isolates, often harbored on mobile genetic elements (Fox et al., 2016). A high frequency of plasmid carriage has also been observed (Allnutt et al., 2016). Taken together, this may underlie the capacity for ST204 isolates to colonize a broad range of environmental niches.

Each of the five most common STs (with >10 isolates) identified all contained isolates from a clinical and a variety of non-clinical sources (Figure 2). The largest non-clinical ST identified was ST20, which was the 8th largest ST grouping overall, and only contained caprine animal isolates. In addition to ST20, ST19 also comprised isolates exclusively of animal origin, suggesting this may be a niche for these subgroups. Analysis of non-clinical categories identified in the study suggests that there may be a subset of STs linked with dairy sources (ST101, ST122, and ST325), however, a larger dataset is required to further investigate these associations.

Of the STs containing clinical isolates, 50% did not contain isolates from other sources (Figure 3). This is consistent with the difficulty in attributing sporadic cases of listeriosis to a food source (Lee et al., 2013). This difficulty is exacerbated by the ubiquitous nature of the organism, and the wide variety of environmental niches it can colonize. A high proportion of STs with raw or RTE meat isolates (80%) also contained isolates from human listeriosis cases. It is noteworthy that meat products comprised the majority of food recalls due to L. monocytogenes contamination across Australia during the 2005–2014 period, with Food Standards Australia New Zealand (FSANZ) recall data showing L. monocytogenes was responsible for 45% of all food recalls due to microbial contamination in this period³. This association is further evidenced in the risk assessment study which suggested processed RTE meats could be responsible for up to 40% of Australian listeriosis cases (Ross et al., 2009). A similar examination of STs with dairy isolates in this study indicated 39% also included clinical isolates, lower than that observed with meat isolates (Figure 3). It should be noted, however, that although non-clinical isolates included a diverse representation from multiple States of Australia, the clinical isolates in this study were dominated by those isolates from Queensland. The application of PFGE or whole genome sequencing to epidemiologically linked isolates and analysis with a broader representation of isolates from other Australian States may provide greater insight into strains shared by food and clinical sources. Continued surveillance should include quantitative microbial risk assessment to identify the contribution of different foods to the burden of human infection (Ross et al., 2009). In addition the application of source attribution modeling should be applied. This approach utilizes modeling, such as the Hald Bayesian risk attribution model, to combine highly discriminatory subtyping analysis (e.g., SNP analysis or Pulsed-Field Gel Electrophoresis) with data relating to human sporadic infection and the prevalence of those subtypes in different foods, to identify likely sources (Hald et al., 2004; Little et al., 2010). Such approaches can help identify key points to direct food safety strategies and public health interventions.

Internalin A (InlA) is a virulence protein associated with the invasive form of listeriosis (Schäferkordt and Chakraborty, 1997). This membrane-associated protein binds E-cadherin and facilitates the organism’s entry into intestinal epithelial cells, a key step in the invasive process. Analysis of the inlA sequence from isolates in this study identified a heterogeneous nucleotide sequence among isolates, with 17 inlA sequence variants showing higher conservation among lineage I isolates (Figure 4). This cluster of closely related lineage I isolates was noted with a lower sequence variation relative to other InlA groups (InlA groups 4, 14, 15, and 16; Supplementary Figure S1). Seven of the eight STs in this closely related group included clinical isolates. Along with the lineage I InlA group 3 (ST1, ST2, and ST328) these lineage I isolates also clustered separately from other lineages (Figure 4). In contrast to lineage I isolates, lineage II isolates showed higher heterogeneity in their InlA sequences with a higher number of alleles showing greater genetic variation. In addition to this only 36% of these alleles (4/11) contained clinical isolates, suggesting this higher divergence in InlA tends to be associated with isolates from non-clinical sources. Previous studies have identified PMSCs in inlA, leading to truncated or secreted forms of the protein (Autret et al., 2001; Nightingale et al., 2005). These mutations have been associated with reduced invasiveness of strains harboring them. Typically the same mutations are shared by strains among an individual ST. Of the STs identified in this study, two had PMSCs: all ST121 isolates had a mutation AA⁴⁹² and all ST325 isolates had a mutation at AA⁹³ (both of which are lineage II STs). Although the ST121 PMSC mutation has been previously reported, to the authors’ knowledge this is the first report of the novel AA⁹³ mutation. No ST121 or ST325 isolates in this study were from clinical sources. Although taken together this data suggests these STs may be less invasive in human patients, further study is required to confirm this.

Single nucleotide polymorphism typing analysis showed a genetically diverse serogroup IIA, with individual STs or CCs often sharing a large number of SNPs (up to 636 in the case of the most diverse ST, ST121). In addition, epidemiologically linked isolates often included a high number of SNP differences (ranging from 13 to 241). In contrast, a high degree of genetic conservation was observed among serogroup IIB isolates, particularly CC3 isolates (Figure 6). Serogroup IVB was also dominated by a single CC (CC1), which also included a clinical isolate linked to contaminated dairy products (differing by only a single SNP). The only other clinical case which was associated with a closely related food (also a dairy food) isolate was noted among ST5, although there were nine SNPs in this instance. This number is in agreement with observations in similar studies where epidemiologically linked isolates differed by less than 10 SNPs (Kwong et al., 2016). Among serogroup IIC, a clinical ST9 isolate shared only six SNPs with a food isolates, although no epidemiological link was known in this case to suggest an association. All other epidemiologically linked isolates were either exclusively from foods, or only isolated from clinical cases with no known associated food source. Although this study used SNP typing analysis to examine population structure, further investigations in our laboratory are ongoing into the loci of SNPs in relation to their impact on genetic functionality and associated phenotypes such as virulence and stress resistance.

Surveillance of foodborne disease in Australia has continued to evolve to maintain Australia’s high food safety standards and well-coordinated public health program, notably through the establishment of OzFoodNet in 2000, and the enhanced listeriosis surveillance program implemented in 2010 (The OzFoodNet Working Group, 2012). Integrating detailed analysis of epidemiological data to national, and international, surveillance programs can serve to improve the efficacy and response to threats to human health posed by foodborne pathogens such as L. monocytogenes. This study highlights the association of subpopulations of L. monocytogenes to different sources in Australia, including those associated with clinical illness, and identifies genetic mutations suggesting attenuated virulence of certain subgroups in humans. The ubiquitous ecology of L. monocytogenes presents a significant challenge for source tracking, which is further exacerbated by the long incubation time of disease in patients following consumption of contaminated foods. Surveillance data where epidemiological linkages are known will help further understanding of key transmission routes and high risk foods. Additional analysis also including isolates from samples not highly represented in this study (e.g., fruit, vegetables, and seafood) will increase understanding of the distribution of L. monocytogenes subtypes through these different sources. These insights can be utilized to help maintain Australian public health and food safety.

AJ, NF, KG, JB, and EF conceived and designed the experiments. N-XF, RG, MB, TG, and CG performed wet laboratory experiments. AJ, JM, JB, and EF conducted bioinformatics analyses. AJ and EF drafted manuscript. All authors read and approved final manuscript and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.