Predominance of Distinct Listeria Innocua and Listeria Monocytogenes in Recurrent Contamination Events at Dairy Processing Facilities

The genus Listeria now comprises up to now 21 recognized species and six subspecies, with L. monocytogenes and L. innocua as the most prevalent sensu stricto associated species. Reports focusing on the challenges in Listeria detection and confirmation are available, especially from food-associated environmental samples. L. innocua is more prevalent in the food processing environment (FPE) than L. monocytogenes and has been shown to have a growth advantage in selective enrichment and agar media. Until now, the adaptive nature of L. innocua in FPEs has not been fully elucidated and potential persistence in the FPE has not been observed. Therefore, the aim of this study is to characterize L. innocua (n = 139) and L. monocytogenes (n = 81) isolated from FPEs and cheese products collected at five dairy processing facilities (A–E) at geno- and phenotypic levels. Biochemical profiling was conducted for all L. monocytogenes and the majority of L. innocua (n = 124) isolates and included a rhamnose positive reaction. L. monocytogenes isolates were most frequently confirmed as PCR-serogroups 1/2a, 3a (95%). Pulsed-field gel electrophoresis (PFGE)-typing, applying the restriction enzymes AscI, revealed 33 distinct Listeria PFGE profiles with a Simpson’s Index of Diversity of 0.75. Multi-locus sequence typing (MLST) resulted in 27 STs with seven new L. innocua local STs (ST1595 to ST1601). L. innocua ST1597 and ST603 and L. monocytogenes ST121 and ST14 were the most abundant genotypes in dairy processing facilities A–E over time. Either SSI-1 (ST14) or SSI-2 (ST121, all L. innocua) were present in successfully FPE-adapted strains. We identified housekeeping genes common in Listeria isolates and L. monocytogenes genetic lineage III. Wherever there are long-term contamination events of L. monocytogenes and other Listeria species, subtyping methods are helpful tools to identify niches of high risk.

The PCR reaction for the detection of SSI-1 + (9.7 kbp) and SSI-1 + (1.1 kbp) differed from the latter mix by the following components: 2.5 U long range DNA polymerase and 2 µL DNA template in a final volume of 25 µL. The gel electrophoresis of PCR-reactions was determined in a 1.5% agarose gel containing 0.5 × Tris-borate-EDTA (TBE) buffer and 3.5 µL peqGREEN DNA gel stain (VWR International, Radnor, PA, USA). The DNA standard Thermo Scientific™ GeneRuler™ 100 bp and 1 kb plus (Thermo Fisher Scientific Inc.) were applied for fragment length comparison.

L. Monocytogenes and L. Innocua Minimum Inhibitory Concentration (MIC) towards Biocides
The minimum inhibitory concentrations (MIC) of five disinfectant compounds (peracetic acid, benzalkonium chloride, sodium hypochlorite, hydrogen peroxide, and isopropanol; all supplied by Sigma AldrichCorp) were determined for 10 recurrent L. monocytogenes and L. innocua The disinfectant components were tested at concentration ranges of 31.3-1000 mg/L for peracetic acid and hydrogen peroxide, 0.5-1000 mg/L for benzalkonium chloride and 125-10,000 mg/L for sodium hypochlorite. An agar dilution method was performed in duplicates to determine the minimal inhibitory concentration (MIC) of the disinfectants against L. innocua and L. monocytogenes strains. As previously described, 5 µL of bacterial culture was spotted onto Mueller-Hinton agar (Oxoid, Basingstoke, UK) containing the disinfectants to be tested [28,29]. Plates were incubated at 37 • C for 24 to 48 h. Following incubation, the lowest disinfectant concentration that showed no bacterial growth was recorded as MIC. Mean MIC values were calculated using Excel (Microsoft Corporation, Redmond, WA, USA).
The PFGE-typing, applying the restriction enzyme AscI, revealed 33 distinct Listeria PFGE profiles with a Simpson's Index of Diversity of 0.75. Thereof, 11 and 22 AscI were specific for L. monocytogenes and L. innocua, respectively (Simpson´s Index 0.519 and 0.533). The ApaI macrorestriction digest resulted in fewer PFGE profiles for L. innocua (n = 20) PFGE-types with identical AscI profiles assigned to dairy producers A-E and isolated during two or more sampling events were classified as recurrent, whereas Listeria isolates with unique AscI profiles and isolated once were defined as sporadic genotypes. Listeria spp. genotypes recurrently isolated over a period of six or more months were defined as persistent. Generally, L. monocytogenes clustered together in a distinct subcluster A and could be clearly distinguished at a similarity level of 25% from L. innocua subcluster B and C (similarity level 40%) (Table 1, Figure 1).
The MLST typing resulted in 27 STs (Simpson´s Index of Diversity of 0.742). In total, 9 and 3 clonal complexes (CCs) and 1 and 14 singletons were identified among L. monocytogenes and L. inncoua isolates, respectively. The discriminatory power of MLST analysis was comparable to PFGE-typing for L. monocytogenes (10 STs; Simpson´s Index 0.469). ST121 could be differentiated by applying PFGE typing into two L. monocytogenes distinct fingerprints. MLST analysis for L. innocua isolates was less discriminative in comparison to PFGE (AscI), resulting in 17 STs (Simpson´s Index 0.531).
Seven new L. innocua STs (ST1595 to ST1601) were defined by submitting the sequences to the Institute Pasteur MLST database (https://bigsdb.web.pasteur.fr/listeria/listeria.html; Table 1). The ST attribution to dairy processing facilities A-E is depicted in Figure 2.
Seven new L. innocua STs (ST1595 to ST1601) were defined by submitting the sequences to the Institute Pasteur MLST database (https://bigsdb.web.pasteur.fr/listeria/listeria.html; Table 1). The ST attribution to dairy processing facilities A-E is depicted in Figure 2.  (Table S1). Interestingly, some housekeeping genes were not specific for L. monocytogenes genetic lineage III or L. innocua.

Molecular Epidemiological Interpretation
The majority of L. innocua and L. monocytogenes PFGE types were isolated once (n = 20/33; 60.61%), but certain genotypes were recurrently isolated from process associated samples and cheese for a short period of time (n = 6/33; 18.18%). These genotypes were present in PAL (smear) and environmental samples and after contamination events in cheese for between one and six months and were successfully eliminated. Other Listeria spp. PFGE types were persistent in the dairy processing environment for a long period and somehow adapted to niches (smear, brine, drain  (Table S1). Interestingly, some housekeeping genes were not specific for L. monocytogenes genetic lineage III or L. innocua.

Molecular Epidemiological Interpretation
The majority of L. innocua and L. monocytogenes PFGE types were isolated once (n = 20/33; 60.61%), but certain genotypes were recurrently isolated from process associated samples and cheese for a short period of time (n = 6/33; 18.18%). These genotypes were present in PAL (smear) and environmental samples and after contamination events in cheese for between one and six months and were successfully eliminated. Other Listeria spp. PFGE types were persistent in the dairy processing environment for a long period and somehow adapted to niches (smear, brine, drain water). The latter L. monocytogenes and L. innocua PFGE types cross-contaminated the surface of cheeses (e.g., hard cheese). Almost all of the persistent L. innocua and L. monocytogenes isolates were rhamnose positive (API profiles 7510 and 6510), except for one genotype with a rhamnose negative profile (IN7[C], ST1085). Almost all of the persistent L. innocua and L. monocytogenes isolates were rhamnose positive (API profiles 7510 and 6510) except for one genotype with a rhamnose negative profile (IN7[C], ST1085). The association coefficient Cramer's V showed a weak association between L. innocua, L. monocytogenes, rhamnose positive, rhamnose negative sporadic and persistent occurrences, although the result "for persistence and rhamnose positive" was highly significant (p = 0.0065; Cramer's V rV = WERT, p < 0.01).
The most abundant L. monocytogenes genotypes related to persistence in cheese processing associated samples and environments were PFGE profiles M5  1/2a, 3a), which were present for 7 and 11 years at dairy processing facility B and A in smear and drain water, respectively. The latter profiles were also isolated once and after four months at producers D and E. PFGE profiles M10[C] (ST155, 1/2a, 3a), M3[E] (ST403, 1/2a, 3a) were present at producers C and E for one year and seven months. M1 (ST59, 1/2b, 3b) and M2 (ST121, 1/2a, 3a) were recurrently isolated during a shorter timeframe (one and two months), both at producer E. Other sporadically isolated L. monocytogenes genotypes (ST1, ST3, ST7, and ST398) were isolated once during the monitoring period. L. innocua ST1597, ST603 and L. monocytogenes ST121 and ST14 were the most abundant genotypes in dairy producing facilities A-E over time (n = 178/220 isolates). The highest genotype diversity was identified in dairy producing facilities A, C, and E (n = 11, 12, and 9 different genotypes; Figure 2).

Susceptibility to Biocides
The MIC towards biocides was determined for four recurrent L. monocytogenes and six L.

Discussion
Dairy and cheese processing environments are frequently colonized by Listeria spp., including pathogenic L. monocytogenes. Even newly established dairy processing facilities become colonized after a short period of time [29,30].
Generally, prevalence and concentrations of L. monocytogenes in cheeses and cheese processing environments are low. Its growth is supported by the presence of fresh, ripened, veined, and smear cheeses (0.8%-5.1% prevalence in cheese lots). Brined cheeses are most often contaminated by L. monocytogenes (11.8%), according to a meta-analysis-based literature review [31]. This suggests that product-associated liquids (smear, brine) contribute to L. monocytogenes contamination of cheese lots [21,32]. In our study, smear and brine samples were indeed most often associated with L. monocytogenes and L. innocua surface contamination and supported the persistence of certain genotypes (ST121, ST14, ST603, and ST1597) in the dairy environment.
Floor drains are further niches for efficient Listeria spp. colonization of the FPE and hot-spots for cross and recontamination events [33,34]. These niches were also identified in our study. To a certain extent, drain waters harbored recurrent and persistent Listeria spp. (ST14, ST637, ST1595, ST1597; Table 1). Despite having a cleaning potential, the introduction of high-pressure water from hoses into a contaminated drain can cause the airborne spread of Listeria and further contribute to the successful establishment of persistent Listeria spp. strains in a facility [35].
As reports about dairy processing environment contamination scenarios in the literature are sparse, our goal was to identify any potential long-term contamination with certain L. innocua and L. monocytogenes genotypes.
Some studies indeed allude to wider contamination of dairy facilities. Parisi et al. [36] isolated Listeria spp. at 19/34 cheese factories (55.8%). Occasionally, L. innocua and L. monocytogenes were detected at the same sampling site (2/19 plants) and persisted in floor drains, which were identified as ideal sampling sites to be included in a monitoring system.
Relevantly, we clearly identified a higher fluctuation of L. innocua and L. monocytogenes genotypes in parallel in 4/5 dairy processing facilities ( Figure 1, Table 1). One dairy plant (B) harbored a persistent L. monocytogenes genotype (ST121) for seven years without further introduction of other Listeria spp. genotypes.
Lomonaco et al. [37] reported two persistent L. monocytogenes genotypes in the Gorgonzola processing chain. About 88% of the L. monocytogenes strains were serotype 1/2a, which is consistent with our findings (95% of the isolates were serotype 1/2a). However, genotypes were not comparable due to the lack of common nomenclature in Listeria spp. subtyping, which is urgently in need of rectification [38]. In this respect, Jagadeesan et al. [39] highlighted the need to include metadata for genotypic approaches, which should be sufficiently cleaned with the removal of replicates and unintended information. Actual studies indicate that whole-genome sequencing (WGS) and core genome (cg) MLST approaches already contribute to the real-time exchange of information on the emergence and geographic dispersal of clones [40][41][42]. Maury et al. [43] reported that L. monocytogenes CC1 are strongly associated with dairy products, whereas hypovirulent clones, CC9 and CC121, are related to meat products.
In the presented study, several disease-related genotypes that are globally distributed were introduced into the dairy processing environment (ST1, ST3, ST7, ST59, ST155, ST398, and ST403). However, ST14 and ST121 established themselves for a longer time (7 and 11 years, respectively) in the FPE and tended to be persistent. Almost all of the persistent L. innocua and L. monocytogenes isolates that we identified were rhamnose-positive (API profiles 7510 and 6510; Table 1). Rhamnose is a naturally occurring monosaccharide present in plant material and important for saprophytes such as Listeria spp.
Atypical L. innocua and L. monocytogenes lacking the ability to ferment rhamnose (where the pdu operon for propanediol utilization is missing) are potentially less capable of exploiting nutritional sources important for adaptation to the FPE [44].
Further, atypical strains with deficient rhamnose fermentation have been reported to be attenuated in virulence and have reduced resistance to temperature changes [45]. In contrast, L. monocytogenes serotype 1/2a mutants confer phage-resistance due to a loss of rhamnose. This is important when there is a field application of lytic phage cocktails as biocontrol measures [46]. Therefore, detailed characterizations of L. monocytogenes and L. innocua rhamnose-positive and -negative field strains in respect of persistence in the FPE and resistance to biocontrol measures are important.
Pasquali et al. [47] also identified ST14 and ST121 as persisters in a rabbit meat processing plant. Several strain-specific features, such as a stronger biofilm-forming potential (ST14) and the presence of the qacH gene associated with adaptation to BAC in ST121, contributed to successful establishment in the FPE.
The stress survival islet (SSI-1) inserted into the intergenic region lmo0443 to lmo0449 in L. monocytogenes is present in long-term persister ST14 and is related to acid tolerance [26]. L. monocytogenes ST121 and all L. innocua isolates that we identified harbored the SSI-2 (2.2 kbp fragment), which is related to elevated tolerance to oxidative and alkaline stress (Table 1) [48].
The entrance and establishment of certain successful genotypes in the FPE are influenced by several factors. These include strain properties (e.g., prophage diversification, transposons, plasmids), the possibility of reintroduced genotypes by raw material, over-diluted biocides on wet surfaces, and a permanent change in cleaning regimes by external cleaning companies and consultants [49]. Consequentially, Muhterem-Uyar et al. (2018) reported that in a heavily contaminated cheese processing environment, the strain variability (ST1, ST7, ST21, and ST37) was reduced to a persistent genotype (ST5) harboring a plasmid type (plM80 related), which is present in successful clones worldwide [20].
In In FPEs, resident Listeria are frequently exposed to sublethal concentrations of biocides due to the dilution effect of wet surfaces and the presence of food soil [50]. Therefore, testing the sensitivity of Listeria spp. to sublethal concentrations of biocides should be performed routinely to identify potential strain adaptations. Particularly, L. monocytogenes and L. innocua isolated from the pork processing chain have been shown to harbor efflux pumps and resistance genes (cadA1-cadA4, arsA1, arsA2) that confer resistance to benzalkonium chloride and heavy metals [51].
The exchange of genetic material between L. innocua and L. monocytogenes has been observed in a few studies. The possibility of horizontal gene transfer (HGT) of plasmids, including heavy metal resistance, enhanced tolerance to QACs and DNA intercalating dyes between L. welshimeri, L. innocua and L. monocytogenes, has been described by sequence analysis in experimental settings and by comparison of FPE isolates [52][53][54].
The presence of atypical hemolytic L. innocua strains in the food chain might also have been introduced by HGT, and this constitutes a reservoir of virulence genes transferable to other species [12,55]. This could be the same for genetic exchange between L. innocua and L. monocytogenes related to environmental adaptation. More research focusing on the uptake of genetic material by Listeria spp. in the FPE is warranted.
In fact, the draft genome of Listeria innocua UAM003-1A, available from NCBI, is also related to highly abundant CC140 [14].
We did not identify any atypical hemolytic ST188 or ST437 L. innocua strains (LIP-1 positive, hly positive), which have recently been isolated from wild bird feces in Finland [12] and previously described by Volokhov [56]. The L. innocua strains included in our study showed no relationship to the newly announced MEZLIS26 genome, due to their different housekeeping genes. The latter L. innocua was assigned to the highly abundant CC537 [15].
L. innocua has been reported to be more commonly found in the FPE than L. monocytogenes [57], which is supported by our findings. We fully agree with Jemmi and Stephan [58] who suggest that L. innocua is a good hygiene indicator and also a marker for unrecognized L. monocytogenes contamination events in the FPE. What should now be considered in retrospect is the demanding nature of the microbiological reference method, for example, ISO 11290-1, concerning the differentiation of L. innocua and L. monocytogenes [59]. Reports focusing on the challenges associated with Listeria detection and confirmation are available, including atypical strains or L. monocytogenes present in lower concentrations due to competitive L. innocua strains during enrichment or that mask detection of L. monocytogenes on selective agar plates such as ALOA medium [60,61] (https://www.fda.gov/food/laboratory-methodsfood/guidelines-bam-users-identification-atypical-hemolytic-listeria-isolates). This might also have contributed to a higher isolation rate of L. innocua compared to L. monocytogenes in our study.
In respect of long-term contamination events with L. monocytogenes and other Listeria species, subtyping methods are helpful tools to identify the true nature of persister candidates.

Conclusions
L. monocytogenes is a foodborne pathogen of significance to human health, and it is able to co-survive in the dairy FPE in microbial communities with other Listeria species and with other bacteria (e.g., Proteobacteria, lactic acid bacteria) [33,62]. Our study identified for the first time the recurrent isolation and persistence of L. innocua in L. monocytogenes-colonized habitats. Novel local L. innocua sequence types (ST1595 to ST1601) were identified, which shared, to a certain extent, the housekeeping genes that are also common in L. monocytogenes genetic lineage III. Either SSI-1 (ST14) or SSI-2 (ST121, all L. innocua) were present in strains successfully adapted to the FPE. There is a great need for further insight into the processes of FPE adaptation and exchange of genetic information between Listeria species so that appropriate food safety control measures can be designed.