|Home | About | Journals | Submit | Contact Us | Français|
We determined mammalian cell invasion and virulence gene (inlA, inlB, and actA) sequences of Listeria monocytogenes strains belonging to a molecular subtype (RAPD 9) that often persists in Danish fish-processing plants. These strains invaded human placental trophoblasts less efficiently than other L. monocytogenes strains, including clinical strains, and they carry a premature stop codon in inlA. Eight of 15 strains, including the RAPD 9 and maternofetal strains, had a 105-nucleotide deletion in actA that did not affect cell-to-cell spread in mouse fibroblasts. The RAPD 9 strains may still be regarded as of low virulence with respect to human listeriosis.
Listeria monocytogenes is a Gram-positive pathogenic bacterium that can cause foodborne listeriosis, which affects immunocompromised individuals, causing septicemia and meningitis, and pregnant women, causing preterm delivery, miscarriage, or stillbirth. It is a ubiquitous environmental bacterium, and it is therefore continuously introduced to food-processing plants, where some molecular subtypes are able to persist despite thorough cleaning and disinfection procedures (1, 27, 32, 42). Such persistent strains are likely to contaminate the food products and may be the cause of foodborne infections (30).
We have shown that specific molecular subtypes of L. monocytogenes can persist for years in the seafood-processing environment (45), and strains representing a particularly prevalent, persistent molecular subtype, RAPD type 9 ([RAPD 9] random amplified polymorphic DNA), had a lower virulence potential than clinical strains in simple eukaryotic models (12). However, in a more complex biological model using oral dosing of pregnant guinea pigs, the tested RAPD 9 strain (strain La111) surprisingly infected the placentas and fetuses just as efficiently as a clinical strain (13). We therefore hypothesized that this specific subtype may have an altered (enhanced) ability to invade placental cells (e.g., trophoblasts) or an enhanced ability to spread intracellularly.
Several virulence factors are important for L. monocytogenes, and mutations in key virulence genes lead to less-virulent L. monocytogenes strains (8). The surface proteins InlA and InlB interact with their respective receptors, E-cadherin (23) and Met, gC1qR, or proteoglycans (4, 15, 34) and mediate internalization of L. monocytogenes into nonphagocytic cells. L. monocytogenes strains with mutations in inlA (premature stop codons [PMSCs]) have been found in France (29), the United States (25), and Japan (9), and these mutations lead to attenuation in the invasion of intestinal epithelial cells (25, 28, 33), but it is not known if invasion into trophoblasts is affected. ActA is important for cell-to-cell spread (5) and is involved in invasion of epithelial cells (39), and ActA-mediated cell-to-cell spread plays a major role in crossing the fetoplacental barrier in both a guinea pig and a mouse model (3, 22).
The purpose of this study was to determine if the high level of prevalence of a RAPD 9 strain in guinea pig fetuses after oral dosing could be explained by increased invasion into and spread between trophoblastic and fibroblastic cells, respectively. Subsequently, we sequenced selected virulence genes to determine if strain variations in cell invasion and spread could be explained by differences in sequences.
Fifteen L. monocytogenes strains representing different origins, RAPD types, serotypes, and lineages were used (Table (Table1).1). Four strains (N53-1, La111, H13-1, and M103-1) represent a unique, persistent subtype (RAPD 9) of L. monocytogenes. The bacteria were grown in brain heart infusion (BHI) broth (catalog no. CM0225; Oxoid) and enumerated on BHI agar. For all cell assays, bacteria were grown at 37°C with aeration for 20 h. Serogrouping was done using Listeria O antisera types 1 and 4 (catalog nos. 223001 and 223011; Becton Dickinson) according to the manufacturer's instructions, and lineage separation was done as in Wiedmann et al. (44). RAPD typing was done according to Vogel et al. (43). Caco-2 cell invasion was done according to Jensen et al. (12).
Invasion and intracellular growth were studied in the human choriocarcinoma cell line JAR (LGC Standards no. HTB-144), and invasion and plaque formation were studied in the mouse fibroblast cell line L929 (European Collection of Animal Cell Cultures [ECACC] no. 85011425). All cell assay experiments were carried out in duplicate in three independent trials. The JAR cells were propagated in F12 (catalog no. BE12-615; Lonza) and the L929 cell line in Dulbecco's modified Eagle's medium (DMEM) (catalog no. BE12-604F; Lonza), both supplemented with 10% fetal bovine serum (FBS) (catalog no. DE14-830; Lonza) and 25 μg/ml gentamicin (catalog no. 15750-037; Gibco). Cells were incubated at 37°C with 5% CO2. Invasion and intracellular growth were studied with a gentamicin protection assay as previously described (12). JAR cells were adjusted to 5 × 104 cells/ml and L929 cells to 3 × 105 cells/ml; 1 ml was added to 24-well tissue culture plates (catalog no. 92024; TPP) and incubated for 24 h. The monolayers of JAR and L929 cells were infected with 1 × 106 CFU/ml or 5 × 104 CFU/ml of bacteria, respectively. The cells were washed in saline water and overlaid with medium containing 50 μg/ml gentamicin. The cells were washed after 1 h with saline water and either lysed immediately with 0.1% Triton X-100 or overlaid with medium containing 25 μg/ml gentamicin and incubated for 2 h, 3.5 h, or 5 h before being lysed. The number of intracellular bacteria was determined by plating appropriate dilutions on BHI agar plates. For the plaque formation assay, 4 ml adjusted L929 cells was added to 6-well tissue culture plates (catalog no. 92006; TPP), incubated for 24 h to reach a monolayer, and then washed twice in saline water before 3 ml of bacteria culture (5 × 104 CFU/ml in cell culture medium) was added. The plates were centrifuged for 5 min at 150 × g. After 1 h, the L929 cells were washed in saline water, followed by the addition of a 2-ml overlay consisting of DMEM, 0.5% agarose, and 10 μg/ml gentamicin. After 4 days of incubation, the plaques were visualized by the addition of 2 ml DMEM containing 0.5% agarose and 0.01% neutral red (catalog no. N2889; Sigma). L. monocytogenes strains 10403S and 10403S with actA deleted (kindly provided by D. Portnoy, University of California) were included as positive and negative controls, respectively, for plaque formation (data not shown). The plaque areas were measured, using ImageJ software, and plaque size was normalized to the size of the well. The measurement included 10 plaques per strain from one experiment. The Mann-Whitney test was used to compare groups of L. monocytogenes strains with full-length sequences and strains with an actA deletion with respect to either number of plaques or plaque size. The Kruskal-Wallis test was used to test for associations between invasive ability, number of plaques, and plaque size among groups of strains of different origins. All analyses were performed using GraphPad Prism statistical software. P values of <0.05 were considered significant.
The group of persistent RAPD 9 strains invaded the trophoblasts at a significantly lower level than the maternofetal strains did (P = 0.0195), but EGD and LO28 also displayed a low level of invasion in JAR cells (Fig. (Fig.1).1). Since L. monocytogenes invades placental trophoblasts in an InlA-E-cadherin-dependent manner (21), the observed lower level of invasion of the RAPD 9 strains and LO28 in JAR cells is supported by the presence of a premature stop codon in inlA in these strains (see below). Lineage I strains invaded trophoblastic cells at a significantly higher level than lineage II strains (P = 0.0343). The two strains derived from food samples, 7418 and La22, also invaded the trophoblasts at a very high level, comparable to that of the maternofetal strains. The high level of invasion of the maternofetal and human clinical strains was also seen in Caco-2 cells (Table (Table1).1). There was no difference in intracellular growth in JAR cells between the RAPD 9 strains and the other strains (results not shown).
All strains invaded the L929 cells at a similar level (Table (Table1),1), and all strains were able to form plaques in L929 cells (Table (Table11 and Fig. Fig.2).2). One maternofetal strain (3272-03) formed a very high number of plaques, almost double the numbers of the other strains. We did not find differences between the number of plaques formed when strains were grouped according to origin (P = 0.8326), lineage (P = 0.2721), or presence of the actA deletion (P = 0.6126) (see below); however, lineage I strains formed significantly larger plaques than lineage II strains (P = 0.0160).
The inlA, inlB, and actA genes were amplified by PCR from total isolated DNA and sequenced in both directions by DNA Technology (Århus, Denmark). The inlA and inlB genes were fully sequenced, and the actA gene was partly sequenced. The primers (DNA Technology, Århus, Denmark) used for PCR amplification and sequencing are shown in Table Table2.2. The actA region was chosen, as previous studies proposed a link between deletions of repeats in this region of proline-rich repeats and reduced in vitro motility and plaque formation (36, 38).
We have hypothesized that single-point mutations in inlA could cause a lower level of invasion by RAPD 9 strains in Caco-2 cells (11, 12), and here we identified a premature stop codon in all four RAPD 9 strains. A nucleotide substitution from cytosine to thymidine at position 1474 results in a stop codon at position 492 (Fig. (Fig.3A),3A), reported to lead to export of InlA instead of its incorporation into the cell membrane (16, 28). LO28 was the only other strain containing PMSC, which was already known. The maternofetal strain 4810-98 had a small deletion of 9 nucleotides starting at position 2214 (data not shown). This deletion did not influence the ability of this strain to invade trophoblasts. The deletion was present at the membrane-anchoring region of InlA and therefore might not influence the activity of InlA.
Three of 15 strains contained a complete inlB sequence, but 12 strains had two nucleotide substitutions, one from guanine to adenine at position 350, resulting in a substitution from alanine to threonine (position 117), and one from guanine to adenine (position 395), leading to a substitution from valine to isoleucine (position 132) (Fig. (Fig.3B).3B). The first mutation was seen in the four RAPD 9 strains, two maternofetal strains (12443 and 3272), and one food strain (La22). The other type of mutation was seen in all of the strains except EGD, LO28, and the maternofetal strain 3495-04. None of the strains had any nonsense mutations in inlB.
Eight of 15 strains contained a deletion of 105 nucleotides (35 amino acids) in actA (Fig. (Fig.3C).3C). The deletion encodes amino acid 305 to amino acid 340, which is the central region of proline-rich repeats required for binding to the vasodilator-stimulated phosphoprotein (VASP) (6, 19). The deletion was seen not only in the four RAPD 9 strains, but also in four of the clinical strains, of which two have caused fetal infection. This suggests that the deletion in the actA repeat region does not influence the ability to cause fetal infection. Also, this deletion did not influence the number of plaques formed (P = 0.6126) (Table (Table11).
L. monocytogenes subtypes that persist in the food industry have caused listeriosis outbreaks, in the United States, for example (30, 31), and although the subtype RAPD 9, which is common in Danish fish-processing plants, has not been linked to outbreaks (Birgitte Smith, unpublished data), the apparent ability of a strain (La111) of this subtype to invade guinea pig fetuses (13) led us to study invasion of this subtype in trophoblastic cells and the sequences of its virulence genes.
The InlA receptor E-cadherin is present on the cell wall of trophoblastic cells and is responsible for InlA-E-cadherin-dependent entry of L. monocytogenes into trophoblasts (21). RAPD 9 strains invade Caco-2 cells, which are of another InlA-E-cadherin-dependent cell line, at a lower level than strains of other origins and RAPD types (11), and we found in the present study that they were poor invaders of trophoblastic cells as well (Fig. (Fig.1).1). This was likely caused by a premature stop codon in inlA in the four RAPD 9 strains. A strain lacking inlA is attenuated in its ability to invade cells from the placental barrier (8, 21), and our study is the first to demonstrate that strains with PMSCs in inlA are affected in their ability to invade placental cells. Strains with truncated InlA cannot be regarded as nonvirulent per se, as cases of maternofetal listeriosis or bacteremia have been caused by strains expressing truncated InlA (8, 10).
InlB is important for the interaction with several cell line receptors; however, full expression of InlB may not be a prerequisite for virulence, as strain F2365, isolated during the 1985 listeriosis epidemic in California, contains a stop codon in inlB at position 100 (26). We expected that InlB was important for invasion of L. monocytogenes into L929 fibroblast cells, as it influences invasion into Vero cells (40). We did find the same single-point mutations (SPMs) at amino acid positions 117 and 132, but these did not affect fibroblast invasion. The SPMs are located in the leucine-rich region of InlB, which is important for the interaction with the Met receptor (34). The contradiction between our results and the results seen by Temoin et al. (40) could be explained by the use of two different cell lines (Vero cells and L929 cells) or could be due to the presence of other mutations in other genes.
ActA is responsible for the formation of the actin tail, and strains lacking ActA are less virulent and unable to form plaques (5). ActA is also required for crossing of the fetoplacental barrier in guinea pigs and mice (3, 22). The actA deletion found in several strains in our study has been described for other L. monocytogenes strains (14, 24, 38, 44). Jiang et al. (14) found that a strain containing the deletion is unable to form plaques in the mouse fibroblastic cell line L929, but the control strain and the deletion strain in that study were not isogenic, and therefore other genetic differences could be involved. Neither Sokolovic et al. (38), Chen et al. (7), nor Moriishi et al. (24) found any correlation between plaque formation and actA deletion in either the rat epithelial cell line L2 or in L929.
All four tested RAPD 9 strains had exactly the same sequence (and mutations) for all three sequenced virulence genes, indicating that these strains are genetically highly similar. They were isolated from four different processing environments over a period of 8 years (45). By amplified fragment length polymorphism (AFLP) subtyping, we found that N53-1 (isolated in 2002), H13-1 (isolated in 2003), and La111 (isolated in 1996) were identical, whereas M103-1 (isolated in 2003) was marginally different.
The paradox that a strain (La111) with an otherwise low level of virulence that harbors a mutation in inlA is able to infect guinea pig fetuses indicates that InlA does not play a crucial role in placental infection of guinea pigs. This finding is also supported by a study with pregnant guinea pigs (2). In contrast, full-length inlA is important for human placental infection (2, 8, 10, 21), which indicates that InlA and InlB are species specific. The discrepancy between in vivo (guinea pig) and in vitro (cell lines) results could therefore be explained by the species specificity of InlA and InlB. InlA interacts with the human and guinea pig E-cadherin but not with the mouse E-cadherin (20), whereas InlB interacts with the human and mouse Met receptor but not with the guinea pig Met receptor (17). Like humans, the gerbil is permissive to both the InlA-E-cadherin and the InlB-Met pathways, and it has recently been established that InlA and InlB have interdependent roles in fetoplacental invasion in this species. Consequently, L. monocytogenes targets only the placenta in vivo if both InlA and InlB pathways are functional (8). Hence, we believe that the RAPD 9 subtype may be regarded as a group of L. monocytogenes with a low level of virulence.
The inlA, inlB, and actA gene sequences have been deposited in GenBank under accession numbers GU735663 to GU735675, GU079614 to GU079626, and GU060665 to GU060678, respectively.
We thank Vi Nguyen for excellent handling of the cell lines.
A.H. was supported by Microbial Opportunistic Pathogens (grant no. 2052-03-0013) and by the Danish Research Council for Technology and Production Sciences (project 274-08-042). C.T.G. was funded by a Ph.D. grant from the Technical University of Denmark.
Published ahead of print on 26 March 2010.