Listeria innocua isolated from diseased ruminants harbour minor virulence genes of L. monocytogenes

Abstract Listeriosis is one of the most common nervous diseases in ruminants, and is caused almost exclusively by the Gram‐positive bacterium, Listeria monocytogenes. However, there are few reports of listeriosis associated with L. innocua, which is genetically closely related to L. monocytogenes, but considered non‐pathogenic. In this work, we report two cases of suppurative meningoencephalitis in apparently previously healthy ruminants from different farms, in which two strains of L. innocua were recovered. The whole genomes from both isolates were sequenced, allowing phylogenetic analyses to be performed, which indicated that the two strains were very closely related. Virulence determinants were searched, especially genes coding for the main L. monocytogenes virulence factors which have been previously described in L. innocua. Surprisingly, the two isolates do not possess such virulence determinants. Instead, both strains carried a set of genes that encode for other virulence factors of the genus Listeria detected using the Virulence Factor Database (VFDB): iap (division and invasion of host cells), lpeA (entry into non‐professional phagocytes cells), fbpA (multifunctional virulence factor, including adherence to host cells), lspA (surface protein anchoring), lap (adhesion to enterocytes and trans epithelial translocation), pdgA (resistance to lysozyme), oatA (resistance to different antimicrobial compounds and also required for growth inside macrophages), lplA1 (use of host‐metabolites for in vivo growth), gtcA (catalyses teichoic acid of bacterial wall), prsA2 (cell invasion, vacuole lysis and intracellular growth), clpC, clpE and clpP (survival under several stress conditions). These genes among others detected, could be involved in the ability of L. innocua to produce damage in animal and human hosts. These results highlight the multifactorial profile of Listeria pathogenesis and the need for comprehensive scientific research that address microbiological, environmental and veterinary aspects of listeriosis.


INTRODUCTION
Listeriosis is one of the most common nervous diseases reported in ruminants. The etiological agent responsible for the disease is Listeria monocytogenes, a Gram-positive, facultative intracellular bacterium, which enters the host through contaminated feed (Walland et al., 2015).
L. innocua is genetically close to L. monocytogenes, but classically considered non-pathogenic (Buchrieser et al., 2003). It has a wide distribution in the environment, including in ruminant farms as well as in the food industries (Matto et al., 2018;Moreno et al., 2012). Contradicting the classical idea, it has been shown that haemolytic strains of L.
innocua are capable of infecting eukaryotic cells in experimental models like mouse and zebrafish (Moura et al., 2019). However scarce, there are reports of this agent in humans (Favaro et al., 2014;Perrin et al., 2003) and animal infections (Rocha et al., 2013;Walker et al., 1994).
Here we describe two cases of nervous listeriosis in ruminants associated with non-haemolytic L. innocua isolates, as well as the phylogenetic analysis and the virulence profile based on the complete genomic sequences.  (Carlin et al., 2021;Hitchins et al., 2020). We also used the API Listeria system (BioMérieux®) following the manufacturer's instructions. L. monocytogenes ATCC 19111 was included as control. PCR was also performed to detect the inlA gene, according to Liu et al. (2007). for phylogenetic and virulence profile analysis (see Supplementary   Table S2). Whole genome phylogeny was performed from the assemblies using the Enterobase Tool Kit (EToKi) pipeline (Zhou et al., 2020).

MATERIALS AND METHODS
First, the align module was run using the genome sequence of strain Clip11262 as the reference with the following options: -a-and -c 1.
Then, the alignment output file obtained in the previous step was used to perform the phylogeny using the RAxML-NG software (version 0.9.0) (Kozlov et al., 2019) with the following options: -model GTR+G, -seed 3 and -bs-metric fbp. Using this setup the bootstrapping converged after 550 replicates. The tree support values were drawn on the best-scoring tree (best value for estimated likelihood).
To complement this information, a comparative test was performed to count the difference in single nucleotide polymorphisms (SNPs) between the two isolates. To obtain specific SNPs between pairs of genomes, the nucmer module of the MUMmer software (Marçais et al., 2018) (version 4.0.0) was used with the settings -maxmatch and -l 12, followed by the show-snps module with the -CHlrT setting.
Finally, genome sequence analysis was focused on the presence of genes encoding putative virulence factors. The software ABRicate version 1.0.1 (https://github.com/tseemann/abricate) was run with default parameters (≥80% sequence identity and ≥80% sequence coverage) using the core dataset from the Virulence Factor Database 2.0 (VFDB) (last update in April 2020). The core dataset of VFDB only contains genes from virulence factors experimentally verified from the same genus as the query genome (Liu et al., 2019). For all the genes found using this approach, the presence of premature stop codons and partial sequencing coverage was checked by manual curation. To search for orthologs of these putative virulence genes, the genomes were first annotated using Prokka (Seeman, 2014), to then retrieve target genes' protein sequences from the .faa files, together with their genome feature metadata (.gff files). Automatic annotations were con- A whole genome phylogenetic analysis was performed, including the two isolates reported here and 12 additional genomes that span L. innocua diversity. The obtained tree showed that the local isolates obtained from both animals were closely related to each other ( Figure 1). This was further studied by a SNPs comparative analysis, which resulted in a difference of 86 SNPs using the cattle isolate as ref- erence, and of 115 SNPs if the reference was the sheep strain.
Virulence factor profiling using the VFDB core dataset identified the presence of 13 genes conserved in both isolates as shown in Figure 1. Manual curation of these results showed that none of these genes had premature stop codons or failed to be covered by the sequencing process. Through this analysis we also found that the virulence factors harboured in these two isolates were also present in most of L. innocua reference strains (Figure 1). A more extensive profiling using the BIGSdb-Lm confirmed these results as well as the presence of other genes potentially involved in Listeria pathogenesis, which were also present in the other L. innocua genomes used in this study (see Supplementary Table S3).

DISCUSSION
In this work L. innocua was isolated from the central nervous system of two animals with encephalitis. Along with the observed histopathologic lesion in the CNS, this supports the hypothesis that L. innocua was the agent responsible for the nervous symptoms of these animals. It has been previously described that cattle are asymptomatic carriers of L.
innocua, shedding this microorganism in the faeces (Hofer & Reis, 2005;Matto et al., 2018). However, there are only two reports of L. innocua as the cause of meningoencephalitis in ruminants (Rocha et al., 2013;Walker et al., 1994), to be contrasted to the numerous reports of animal listeriosis due to L. monocytogenes strains (Walland et al., 2015). This is the first report of nervous listeriosis in ruminants due to L. innocua in which their genomes were sequenced, and the presence of virulence factors was studied.
The narrow phylogenetic difference found between the genomes of the two isolates is quite interesting (Figure 1), as they were isolated from different animals in two unrelated farms, distanced more than 150 km apart, and without exchange of animals between them. This result suggests a common source of contamination or, a common circulation pathway for these bacteria in ruminants throughout the country or the region. Also, our results encourage the need to maintain the surveillance of cases of nervous diseases in ruminants, and to sequence more Listeria isolates, in order to confirm or refute these hypotheses.
This finding is consistent with the absence of the hly gene, coding for Taking into account our results, we wonder if the genes identified in VFDB (Figure 1) could provide the molecular basis to explain the pathogenic behaviour of the isolates of this work. Some of these genes have been reported in L. monocytogenes and other Gram-positive bacteria, individually associated with roles in virulence and/or pathogenicity (Burkholder & Bhunia, 2010;Forster et al., 2011;Keeney et al., 2007;Meireles et al., 2020;Osanai et al., 2013;Rae et al., 2011;Réglier-Poupet et al., 2003a,b;Vázquez-Boland et al., 2001). At least 5 of the 13 virulence genes found encode proteins related to bacterial adhesion to, and/or invasion into mammalian cells. protein to host cells, especially hepatocytes (Osanai et al., 2013).
Among the virulence-encoding genes, these L. innocua isolates also include genes coding for enzymes that protect bacteria against host defences, or that enhance their survival within the cytosol of infected cells. For example, the pdgA and oatA genes (peptidoglycan Ndeacetylase and O-acetylase, respectively) may be essential to resist the host's lysozyme. Mutants in these two genes result in increase of peptidoglycan's sensitivity to lysozyme inducing L. monocytogenes virulence attenuation (Rae et al., 2011). We also found the lplA1 gene, which encodes a lipoate-ligase, an enzyme that promotes Listeria cytosolic replication within host cells (Keeney et al., 2007).
Other two genes found in these two genomes encode for enzymes that likely play important roles in maintaining the integrity and stability of the bacterial wall in Listeria. The gtcA gene encodes an enzyme that catalyses teichoic acid glycosylation on L. monocytogenes wall. Proper glycosylation mediates key pathogenicity features: the correct anchoring of major surface virulence factors (Ami e InlB); resistance to antimicrobial peptides and decreased susceptibility to antibiotics (Meireles et al., 2020). The second gene present is prsA2, which encodes a peptidyl prolyl cis-trans isomerase that assists in correct protein folding. As such, PrsA2 regulates the maturation and secretion of some proprotein virulence factors (such as phospholipase C PC-PLC) of L. monocytogenes (Forster et al., 2011).
Finally, three genes that encode proteases, clpC, clpE and clpP, were also identified; these are proposed to act as stress response mediators and to assist with intracellular replication (Vázquez-Boland et al., 2001).
This genetic background might be consistent with virulence retention in both L. innocua isolates from clinical cases of listeriosis. The genes identified using VFDB and BIGSdb-Lm databases, code for virulence factors called minor or accessory but it has been proven that many are capable of promote cell invasion and/or intracellular replication (Burkholder & Bhunia, 2010;Forster et al., 2011;Keeney et al., 2007;Meireles et al., 2020;Osanai et al., 2013;Rae et al., 2011;Réglier-Poupet et al., 2003a,b;Vázquez-Boland et al., 2001). It is noteworthy that genes identified in both isolates are present in most of the reference strains of L. innocua from the NCBI genome database. However, none of them were obtained from diseased ruminants. As we mentioned above, there are only two cases of animal listeriosis due to L.
innocua reported previously, but their genome sequences are not available to compare with those described here.
Regarding that just one of several animals per farm was affected, other issues to be considered in order to attempt to explain the situation are (i) the previous animal health condition and (ii) bacterial exposure load to which these animals were subjected. Disease could be due to unequal exposure to L. innocua present in feed or farm environment, presence of debilitating factors in these animals, or both simultaneously. However, until now, individual risk factors for ruminants are poorly understood (Walland et al., 2015). The findings reported in this work highlight the multifactorial nature of the Listeria pathogenesis and reinforce the need for detailed scientific research that include microbiological, environmental and veterinary aspects.