Plants as a realized niche for Listeria monocytogenes

Abstract Listeria monocytogenes is a human pathogen. It is the causative agent of listeriosis, the leading cause of bacterial‐linked foodborne mortality in Europe and elsewhere. Outbreaks of listeriosis have been associated with the consumption of fresh produce including vegetables and fruits. In this review we summarize current data providing direct or indirect evidence that plants can serve as habitat for L. monocytogenes, enabling this human pathogen to survive and grow. The current knowledge of the mechanisms involved in the interaction of this bacterium with plants is addressed, and whether this foodborne pathogen elicits an immune response in plants is discussed.


| INTRODUCTION
Understanding the ecology of pathogenic microorganisms requires a thorough knowledge of their habitats and their routes of transmission. Listeria monocytogenes (Lm) is a foodborne pathogen that is the causative agent of listeriosis, a serious foodborne disease that affects primarily at-risk people (pregnant women, elderly, immunocompromised individuals) after consumption of contaminated food. High intraspecific diversity is observed and the species is structured in well-defined genetic lineages and clonal complexes. Plants interact with microorganisms in their close vicinity and can offer habitats for commensal and human pathogens. Indeed, listeriosis outbreaks have been traced back to preharvest contamination of fresh produce due to the presence of Lm in the farm environment. In that sense, plants must be considered as habitats that are potentially colonized by the human pathogen, and as possible vectors of contamination.
To colonize plants bacteria must be able: (i) to utilize available nutrients, (ii) to sense the plant and develop a chemotactic response; (iii) to outcompete other microorganisms and occupy available microniches. In addition, for successful colonization of the rhizoplane or root tissue, microbes must be able to attach to the surface and/or enter root tissue while evading immune responses.
In this review, we discuss the current reports on the occurrence of Lm on plants and the experimental evidence that demonstrates the ability of Lm to colonize plants. We then address the current understanding of the intrinsic and extrinsic factors that underlie plant colonization. Finally, we discuss the current understanding of the contribution of plant biology in providing habitats for Lm and on the interplay between the plant and the human pathogen in light of plant immunity. factors (commensal/beneficial microorganisms and pathogen pressure) (Figure 1).
These reports confirm the preharvest transfer of Lm to growing plants.
Furthermore, the occurrence of contaminated raw vegetables and fresh produce at retail has been reported from several countries (Table 1). Lm prevalence on vegetables, herbs, and mushrooms is variable among countries. Although contamination may occur anywhere along the food chain and depends on many factors (Alegbeleye et al., 2018;K. Honjoh et al., 2018;Miceli & Settanni, 2019;Smith et al., 2018), these data give indications on the type of fresh produce and vegetables potentially contaminated in the field.
Overall, contamination is generally low. Indeed, based on prevalence data available in the literature, mathematical modeling suggested that the probability of contamination of unprocessed fresh vegetables with more than 10 Lm/g was 1.44% and it dropped to 0.17% for rates of contamination over 1000 Lm/g (Crepet et al., 2007).
| 5 of 16 3.1 | Lm attachment to plants The contribution of flagella to attachment and colonization of alfalfa, radish, and broccoli sprouts has been investigated in three genotypes of Lm (Gorski et al., 2009). Colonization was impaired in deletion mutants affected in flagella synthesis but results depended on the type of sprout and the genetic background of Lm strains (Gorski et al., 2009). Thus, the absence of flagellum affects the colonization of some plants but this is strain-dependent. Among the genes required for the synthesis of the flagellar rotor, disruption of motAB had a significant effect on surface attachment to radish tissues. However, deletion of motAB did not impact root attachment on sprouts but the fitness of the mutants was significantly lower than the parental strains during co-inoculation experiments. This suggests that motility improves colonization fitness. Conversely, colonization of cut cabbage was not affected by motility (Palumbo et al., 2005).
The lectin-mediated attachment mechanism is likely to be active during bacteria-root interactions (Danhorn & Fuqua, 2007;Wheatley & Poole, 2018). Indeed, agglutination assays showed that Lm reacts to different plant lectins in a strain-specific manner (Facinelli et al., 1998;Slifkin & Doyle, 1990). However, lectins of Canavalia ensiformis and Punica granatum have antibiofilm activities against Lm and other bacteria (Jin et al., 2019;Silva et al., 2021). This suggests that lectins of some plant species may limit adhesion to their surface.
Xyloglucan and pectins are plant cell wall components that affect Lm attachment (Tan et al., 2015). Moreover, a cellulose-binding protein enables Lm attachment to lettuce (Bae et al., 2013). Altogether, these reports highlight the importance of the structures and components of plant cell walls in the attachment of Lm.
Information on transcriptome variations triggered by plant colonization is limited, and genes whose expression is required during plant colonization remain to be duly identified. In one study, a differential display approach was undertaken to compare the Lm gene expression profile under two conditions. In the first, Lm was inoculated on cut cabbage. In the second Lm was cultivated in standard laboratory conditions (Palumbo et al., 2005). Although several genes were transcribed differentially, including genes contributing to cell surface characteristics, disruption of some of these genes did not impede attachment and growth on cabbage.

| Nutrient utilization during colonization/ proliferation of Lm on plants
The growth of Lm on plants relies on its ability to utilize plant-derived nutrients (Palumbo et al., 2005). Indeed plants release to their environment a blend of compounds produced constitutively or in response to environmental cues, including abiotic and biotic stressors (Bais et al., 2006;Chaudhry et al., 2021;Jacoby et al., 2020;Sasse et al., 2018). The composition of these nutrient-rich exudates depends on the plant species, age, nutrition, and physiology (Bais et al., 2006).
Exudates are mixtures of low molecular weight (organic acids, amino acids, sugars, secondary metabolites) and high molecular weight T A B L E 2 (Continued)

Plant species Culture condition inoculation doses
Detection methods

Basil
O. basilicum. • Greenhouse spray of above-ground parts at 10 6 CFU/ml with 3 ml
(mucilage, proteins) C-rich molecules. Leakage of nutrients at root junction sites, after tissue wounding or phytopathogen infection, can be another source of nutrients available for the development of Lm (Brandl, 2006). The increase in numbers of Lm on seeds germinating on sterile dampened filter papers confirms that Lm can make use of the plant compounds for growth, attaining levels of 5.5-6.9 log CFU/g (Jablasone et al., 2005). Furthermore, when Lm was inoculated on fresh-cut cabbage, higher transcription of genes associated with transport, carbohydrate metabolism, amino acid, vitamin, and nucleotide biosynthesis suggests that Lm can transport and metabolize a wide range of plant-derived resources (Palumbo et al., 2005).
Though leaf surfaces are oligotrophic environments, limited amounts of exudates can be released in the phyllosphere. The presence of nitrogen in leaf exudates was a critical factor promoting the growth of human pathogens on lettuce leaves (Brandl & Amundson, 2008), and bacterial multiplication on leaves is supported locally by discrete zones providing higher concentrations of sugars (Leveau & Lindow, 2001). Still, the leaf habitat displays harsher conditions than roots . For example, microscopic examination of germinated sprouts confirmed that Lm was preferentially localized on root hairs rather than on leaves (Gorski et al., 2004(Gorski et al., , 2009. However, these studies were performed with axenic sprouts and the absence of other microorganisms is a major bias in comparison to field conditions.

| Stress response
Although plants provide habitats for microorganisms, the production of specific molecules can induce stressful conditions for bacteria (Foreman et al., 2003). Coping with harsh conditions is a prerequisite for plant colonization. For example, intrinsic resistance to cumene hydroperoxide in a collection of Lm strains was correlated with higher colonization of sprouts, regardless of the type of sprout used in the study, but the results were to some extent strain-dependent (Gorski et al., 2008). The authors proposed that resistance to oxidative stress was one of the many factors contributing to the success of root colonization. The general stress response plays indeed a key role in the process of habitat colonization. Sigma B is the essential factor in the response of Lm to stressors (low pH, oxidizing conditions, starvation, and osmotic variations); it coordinates the transcription of approximately 10% of the genome (Ferreira et al., 2001, Fraser et al., 2003. Deletion of the gene encoding Sigma B (sigB) did not obliterate growth and survival in commercial potting soil nor on radish but the mutant population was 1-2 orders of magnitude lower than the parental strain (Gorski et al., 2011). These results were confirmed in another genetic background during in vitro root colonization of F. arundinacea and survival in agricultural soil microcosms (Marinho et al., 2020). These data suggest that regulation of transcription by Sigma B is required for optimal adaptation and survival in Further root colonization defects were observed with a strain (ΔagrAΔsigB) with a double mutation that affected both the general stress response and cell to cell communication (Marinho et al., 2020); this suggests that both, cell to cell communication and general stress response contribute to success during root colonization.
A variety of plant secondary metabolites act as defense compounds. Several volatiles produced by plant leaves or roots display antimicrobial properties against Lm (Kawacka et al., 2021). These include benzenoids, phenylpropanoids, phenolics, and terpenoids released by essential oils (Farré-Armengol et al., 2016). Interference with adherence ability, biofilm formation, and bacterial cell membrane disruption appear to be the mechanisms of action of some of these plant-derived antimicrobial compounds (Kawacka et al., 2021).
As the experiments were generally performed with concentrated extracts or purified compounds, how these data relate to plant/Lm interaction in vivo remains to be assessed.  (Devarajan et al., 2021), protection from pathogens (Ritpitakphong et al., 2016) and they are essential to carbon and nitrogen cycles (Abadi et al., 2021;Reed et al., 2010). Phyllosphere microorganisms are mainly bacteria (Alphaproteobacteria, Gammaproteobacteria, and the phyla Bacteroidetes and Actinobacteria).

| Biotic interactions with plant microbiome
Fungi are also detected in the phyllosphere and appear to be highly diverse (Kembel et al., 2014;Vorholt, 2012). Recent studies suggest that the soil contributes to phyllosphere microbes in addition to parental material and the atmosphere (Grady et al., 2019;Zheng & Lin, 2020;Zhou et al., 2021).
Experiments in unplanted soil microcosms clearly showed that soil microbiomes can act as efficient barriers preventing invasion by Lm (Dowe et al., 1997;Locatelli et al., 2013;Mclaughlin et al., 2011;Moynihan et al., 2015). Although the overall diversity of soil microbiomes plays a key role in generating hostile conditions for Lm, the phylogenetic composition has to be considered as well (Spor et al., 2020;Vivant et al., 2013). Experiments carried out on soil However, unlike in vitro, no growth could be observed and the population of Lm in the rhizosphere gradually declined. Therefore it is likely that, compared to bare soil, the rhizosphere environment is favorable for the survival and maintenance of Lm. The relationship between the characteristics of plant microbiome and the settlement of Lm in the rhizosphere or leaves has yet to be documented. Similar trends are expected in the rhizosphere as in unplanted soil. For example, specific strains of Azotobacter chroococcum, Bacillus megaterium, and Pseudomonas fluorescens can control Lm in the rhizosphere possibly through a combination of competition and antibiosis (Sharma et al., 2020). In conclusion, the plant microbiome is the major factor limiting Lm niche breadth. In the future, implementing farming practices favoring microbiome diversity is an exciting field of investigation to limit preharvest contamination and improve food safety.

| Conflicting information on Lm internalization in plant tissues
Internalization of human pathogens in plant tissues raises further food safety issues. Indeed, internalized bacteria, whether present in the extracellular space or intracellular compartments are protected from removal by washing and surface disinfection, and therefore may threaten consumers' health when fresh produce is eaten raw (Erickson, 2012). Whether or not Lm colonizes plants internally is still a matter of debate and conflicting reports are available (Table 2, Chitarra, Balestrini, et al., 2014;Koiv et al., 2019;Kutter et al., 2006;Shenoy et al., 2017). Detection of Lm in major plant tissues including vasculature supports its possible transport and dissemination within the plant (Shenoy et al., 2017). Fluorescence in situ hybridization with Lm-specific oligonucleotides and confocal imaging coupled with immunocytochemistry of a Green Fluorescence Protein-expressing Lm strain provided evidence of the presence of Lm in plant organs or intercellular spaces of A. thaliana leaves (Milillo et al., 2008), carrot, parsley, and celery (Kljujev et al., 2018). The occurrence of Lm in both extracellular and intracellular spaces of lettuce (Shenoy et al., 2017) and sweet corn (Kljujev et al., 2018) was also reported. Surface disinfection followed by enumeration confirmed the endophytic localization of Lm in lettuce and other plants (Chitarra, Decastelli, et al., 2014;Standing et al., 2013).
However, no internalization of Lm was evidenced in other plant species such as barley and basil (Table 2, Chitarra, Decastelli, et al., 2014;Jablasone et al., 2005;Kutter et al., 2006). These plant species-dependent differences in endophytic colonization by Lm could be linked to the presence or absence of plant metabolites that can either favor or prevent Lm growth. The production of antimicrobial compounds such as essential oils was proposed to limit the colonization of basil by human pathogens (Dorman & Deans, 2000).
In summary, conflicting data on Lm internalization requires further comprehensive investigations taking into account factors such as the concentration of inoculum, the method used to detect internalization, the plant genotype/species, which are all known to affect interactions with human pathogenic bacteria (Hirneisen et al., 2012).

| WHY IS THERE SO LITTLE INFORMATION ON PLANT/Lm INTERACTIONS IN LIGHT OF IMMUNITY?
Evolution has shaped defense mechanisms enabling plants to limit the growth of invading microorganisms. The plant immune system relies on the recognition of specific patterns (called Microbe-Associated Molecular Patterns, MAMPs) on the surface of microorganisms (Jones & Dangl, 2006). Detection of these patterns by pattern recognition receptors (PRRs) localized on the plasma membrane triggers the onset of signaling cascades including a rapid efflux of Ca 2+ , the activation of mitogen-activated protein (MAP) kinases, and the generation of ROS leading to Pattern Triggered Immunity (PTI) (Pitzschke et al., 2009).

| MAMPs and plant immunity
The 22-amino-acid flagellin epitope flg22 is one of the most studied

MAMPs. It triggers plant responses such as hypersensitive cell death in
A. thaliana through the binding to the PRR FLAGELLIN SENSING2 (FLS2) (Gomez-Gomez & Boller, 2000). The second epitope of flagellin, flgII-28, is sufficient to trigger immunity in Solanaceae (Clarke et al., 2013). Flagellin proteins from different bacterial species, pathovars, and strains can display variations in amino acid sequences, and studies have suggested that some phytopathogens can modify their MAMPs to avoid inducing PTI. For example, a single amino acid change in flg22 is sufficient to attenuate or even to block its interaction with FLS2 (W. Sun et al., 2006), and posttranslational modifications of flagellin, including glycosylation, can counteract elicitation (Rossez et al., 2015). Interestingly, MAMPs from commensal, beneficial microbes, and zoonotic human pathogens can be detected by PRRs. As reviewed by Trdá et al., the flagellin and flg22 of the plant growthpromoting rhizobacteria P. fluorescens (WCS374 and WCS417) and the endophytic Burkholderia phytofirmans induce an innate immune response in plant cells (Trda et al., 2015). Strategies to evade or suppress plant immunity such as MAMP divergence by sequence variation, MAMP degradation, sequestration, or MAMP modification seem to be similar among commensal, beneficial, and pathogenic microorganisms (Teixeira et al., 2019). Additional MAMPs include elongation factor Tu (EF-Tu), cold shock proteins, peptidoglycans, and lipopolysaccharides from bacteria, glucans, arachidonic acid, and ergosterol from oomycetes, and chitin from fungi (Boller & Felix, 2009). Interestingly, EF-Tu, one of the most abundant proteins found in bacteria, triggers an immune response in mammals as well as in plants where PRRs specific to EF-Tu have been characterized in monocots and dicots (Zipfel et al., 2006). Interaction of PRRs with EF-Tu involves specific amino acid patterns and is plant-dependent. The amino acid pattern EFa50 F I G U R E 3 The amino acid sequence of (a) epitopes flg22 and Fl-II-28 of Fla and (b) elf18 and EFa50 of EF-Tu in a selection of bacterial species. Listeria sp. sequences of FlaA and EF-Tu proteins were compared with plant and human bacteria. *Key amino acids of flg22-eliciting activity in tomato cells (Felix et al., 1999). **No significant homology with fl-II-28 of Rhizobium leguminosarum. D1 and D2 are sequence divergences (%) calculated with respect to Pseudomonas syringae. Plant phytopathogenic bacteria are indicated in red and plant beneficial bacteria in green. Nucleotides are numbered according to the P. syringae sequence. Sequence alignment and estimation of sequence divergence (p distance) were performed using MegaX (Kumar et al., 2018) (position 175-225 of EF-Tu) of Acidovorax avenae is recognized by rice PRRs (Furukawa et al., 2014), whereas A. thaliana recognizes the pattern composed of the first 18 aa (Kunze et al., 2004).

| A contribution of Lm flagellin and EF-Tu to plant immunity?
Perception of zoonotic human pathogens by plants is supported by several studies on Salmonella enterica and Escherichia coli O157:H7 (Schikora et al., 2008;Teplitski et al., 2012). Indeed the flg22 epitope of these bacteria appears to be perceived by plants and leads to growth restriction of these human pathogens. For example, flg22 St of S. enterica was found to be an effective MAMP triggering PTI (Garcia et al., 2014), and higher colonization of A. thaliana was observed with the flagellum-defective mutants of S. enterica and E. coli O157:H7 than with their isogenic parental strain (Melotto et al., 2014). In the case of Lm, however, experimental evidence of a plant immune response triggered by this bacterium is lacking. Therefore we analyzed in silico the available sequences of flaA and tuf, the Listeria genes encoding respectively flagellin and EF-Tu. The two plant im- The protein structure was predicted by structure homology using the Swiss-Model utility on the ExPasy server [1] (available online at https:// swissmodel.expasy.org/). The four models were built on the top-ranking template predicted by the software. For comparison, the structures were aligned on a reference (PDB accession: 6PWB.2, in blue) using the TM-align online tool [2] (available at https://zhanglab.ccmb.med.umich. edu/TM-align/). The model presented a Global Model Quality Estimate (GMQE) of 0.72 for L. monocytogenes, 0.91 for B. subtilis, 0.74 for P. syringae, and 0.53 for A. vinelandii observed within the variable region spanning Gln-130 to Asn-185, as expected (Nempont et al., 2008). Interestingly the location of flg22 within a conserved domain at 30-51 aa is common to the four models but variations are observed in the regions surrounding this MAMP. Further biochemical characterization of the flagellin of Lm is required to properly assess protein/protein interactions with the plant receptor FLS2 and the subsequent induction of PTI.
The two plant immunogenic epitopes elf18 and EFa50 of EF-tu are also conserved in Lm (Figure 3b). They display 83% and 58% identity with the respective sequences from P. syringae. At the DNA level, tuf nucleotide divergence between Lm lineages is 2.9% (0% and 0.1% for the two EF-Tu epitopes, respectively). Although this in silico analysis suggests that Lm could trigger PTI after the interaction of these MAMPs with their cognate plant receptors, this has yet to be

CONFLICT OF INTERESTS
None declared.

ETHICS STATEMENT
None required.

DATA AVAILABILITY STATEMENT
All data generated or analyzed during this study are included in this published article.