Growth Temperature-Dependent Contributions of Response Regulators,σ B,PrfA,and Motility Factors to Listeria monocytogenes Invasion of Caco-2 Cells

Bacterial strains and growth conditions

Bacterial strains used in this study are listed in Table 1. For invasion assays, L. monocytogenes 10403S (Bishop and Hinrichs, 1987) and associated mutant strains were grown to early stationary phase as previously described (McGann et al., 2007a). Briefly, a 12 to 18 h culture grown at 37°C with aeration (i.e., shaking at 220 rpm) in brain heart infusion broth (BHI) was diluted 1:100 into 5 mL of fresh BHI and grown at 37°C with aeration to OD₆₀₀ = 0.4. This culture was diluted 1:100 into another 5 mL of fresh BHI and grown at 30°C or 37°C with aeration to early stationary phase (defined as growth to OD₆₀₀ = 1.0, followed by an additional 3 h incubation). For temperature shift experiments, a 1 mL aliquot of the early stationary phase culture was centrifuged at 12,000 rpm for 10 min, resuspended in phosphate-buffered saline (PBS; 0.14 M NaCl, 2.7 mM KCl, 10 mM Na₂PO₄, and 1.8 mM KH₂PO₄; pH 7.2), and incubated statically at 30°C or 37°C for 0, 2.5, or 5 h.

Mutant construction

Internal, in-frame, nonpolar deletion mutant alleles in 14 RR genes (Table 1) were created using splicing overlap extension polymerase chain reaction (PCR) (see Supplemental Table S1, available online at www.liebertonline.com for primers; all supplemental material is located at http://author.cals.cornell.edu/cals/foodsci/research/labs/wiedmann/links/ivy2010-1.cfm) and cloned into the temperature-sensitive suicide shuttle vector pKSV7 (Smith and Youngman, 1992) as previously detailed by our group (Wiedmann et al., 1998). Mutant alleles were introduced into L. monocytogenes 10403S using standard allelic exchange mutagenesis procedures (Camilli et al., 1993). Mutant strains were confirmed with sequencing. lmo0287 is likely to be essential, as null mutants in this gene could not be constructed here, consistent with similar studies by others (Kallipolitis and Ingmer, 2001; Williams et al., 2005a).

Caco-2 invasion assays

The Caco-2 cell-line (ATCC HTB-37) was maintained at 37°C (4%–6% CO₂ and 85% humidity) in a Caco-2 medium, which was Dulbecco's minimal essential medium with Earle's salts, 1% sodium pyruvate, 20% fetal bovine serum, 1.0% nonessential amino acids, 1.5 g/L sodium bicarbonate, and, when appropriate, penicillin G and streptomycin (each at 100 μg/mL) (all reagents were obtained from Gibco, a subsidiary of Invitrogen). All invasion assays were performed at 37°C as previously described (Garner et al., 2006b) with minor modifications. Briefly, 48 h before the assay, Caco-2 cells were seeded (from a culture passaged no >60 times) into 24-well tissue culture plates (Corning Inc.) at a density of 5 × 10⁴ cells/well in the Caco-2 medium without antibiotics. For infection, ∼2 × 10⁷ CFU L. monocytogenes were added to each well (representing a multiplicity of infection of ∼200). All inocula were enumerated on BHI agar plates. Thirty minutes postinfection, the Caco-2 monolayers were washed three times with PBS to remove any unassociated L. monocytogenes, and the medium was replaced with a fresh Caco-2 medium. Forty-five minutes after infection, the medium was replaced with the Caco-2 medium plus 150 μg/mL gentamycin to kill any extracellular L. monocytogenes. At 90 min postinfection, Caco-2 cells were washed three times with PBS and lysed with ice-cold distilled water. Intracellular L. monocytogenes were enumerated by plating the appropriate dilutions of the Caco-2 lysate on BHI agar, using a spiral plater (Spiral Biotech). At least three independent trials of the invasion assays were performed with duplicate wells tested for each treatment in each replicate.

Quantitative reverse transcriptase-PCR (qRT-PCR)

Transcript levels of inlA, flaA, plcA, gadA, rpoB, sigB, and prfA were quantified for select strains grown to early stationary phase at 30°C or 37°C using TaqMan probes and primers and the ABI Prism 7000 Sequence Detection System as previously described (Sue et al., 2003; Chaturongakul and Boor, 2006; McGann et al., 2007a) with one exception: copy numbers for each gene were normalized to rpoB levels. Primers and probes for inlA, plcA, gadA, sigB, rpoB, and prfA have been previously described (Sue et al., 2004; Kim et al., 2005; Kazmierczak et al., 2006). flaA Taqman primers ( flaA-F: 5′-TCGTAAAAATAACGAAGGCATGAC-3′; flaA-R: AGAACTGTTAATACGTTTACCAGATGCT-3′) and the flaA MGB probe (FAM-5′-CAAGCGCAAGAAC-3′NFQ) were designed using Primer Express 1.0 (Applied Biosystems).

Statistical analyses

All statistical analyses were performed in JMP 7.0 (SAS Institute Inc.). Invasion efficiencies were initially analyzed using a one-way analysis of variance (ANOVA; α = 0.05). Strain and, where appropriate, date of experiment were included as variables in the model. For each ANOVA, data were log-transformed to ensure that the data set satisfied ANOVA assumptions of normality of residuals and equality of variances. To determine whether a particular mutant strain differed from the parent strain (e.g., in invasion efficiency), a post hoc Dunnett's many to one test was used (Shun et al., 2003). To compare invasion efficiencies for a given strain exposed to multiple conditions (i.e., as shown in Table 2) or among strains (in cases where a strain with a double mutation was included in the comparison), a post hoc Tukey honestly significantly different (HSD) test was used. For qRT-PCR, normalized log copy numbers for a given gene were compared among strains by using one-way ANOVA, followed by Tukey HSD. To measure whether two gene deletions showed an effect on invasion efficiency or transcript level that is more than additive (which indicates synergism), the parent strain, single mutants, and double mutants were assigned unique allelic states by coding dummy variables (e.g., “gene1” and “gene2”) for each allelic state with 1 meaning the gene is present and 0 meaning the gene is absent. A two-way ANOVA was performed to determine the effect of each allelic state on invasion efficiency or transcript level in these analyses. “Date of experiment,” “gene1,” “gene2,” and “gene1*gene2” were included as effects in the two-way ANOVA model. A significant effect of gene1*gene2 in the model (i.e., p < 0.05) indicates a more than additive effect of the double mutation (i.e., deletion of both gene1 and gene2 in the same genetic background has a greater effect than the sum of the effects of deleting gene1 alone and gene2 alone).

Contributions of RRs to Caco-2 invasion

Overall ANOVA analysis showed a highly significant effect of temperature on invasion efficiency (p = 0.0051), and all RR mutants (except ΔdegU and ΔcheY) and 10403S had numerically higher mean Caco-2 invasion efficiencies after growth to early stationary phase in BHI at 30°C than growth at 37°C. For several strains, including 10403S, ΔresD, Δlmo1022, ΔphoP, ΔcesR, and ΔkdpE, this difference was statistically significant (p < 0.05; t-test; Fig. 1).

OPEN IN VIEWER

FIG. 1.

Caco-2 invasion efficiencies of Listeria monocytogenes 10403S and 14 mutant strains with deletions in genes encoding RRs. Bacteria were grown to early stationary phase (growth to OD₆₀₀ = 1.0, followed by 3 h of incubation) with aeration (i.e., shaking at 220 rpm) at 30°C or 37°C. Invasion assays were performed at 37°C. Data represent the mean of at least three biological replicates. Error bars represent SD. A ΔinlA strain was included as a control. Overall ANOVA analyses of the 14 RR mutants and the parent strain showed a significant effect of temperature (p = 0.0051); all strains (except ΔcheY and ΔdegU) showed numerically higher invasion efficiencies when grown at 30°C. Strains showing statistically higher invasion after growth at 30°C than at 37°C (p < 0.05; t-test) are denoted by ∧. For 10403S and RR mutants, ANOVA showed a significant effect of strain on invasion efficiency for bacteria grown at 30°C (p < 0.0001). For bacteria grown at 37°C this effect was not significant (p = 0.3900). Among RR mutants grown at 30°C, ΔcheY and ΔdegU showed significantly lower invasion efficiencies than the parent strain (p < 0.0001; post hoc Dunnett's; denoted by *). RR, response regulator; ANOVA, analysis of variance; SD, standard deviation.

For the L. monocytogenes parent strain and the 14 RR mutants grown at 30°C, ANOVA showed a significant effect of strain on Caco-2 invasion efficiency (p < 0.0001). Specifically, ΔdegU and ΔcheY both had ∼50-fold lower mean invasion efficiencies than the parent strain (p < 0.0001 for each; Dunett's test) when grown at 30°C (Fig. 1). These results confirm the findings of other studies that have shown a role for CheY in host cell invasion (Dons et al., 2004). Unlike all other strains tested, the ΔcheY and ΔdegU strains showed numerically higher invasion efficiency after growth at 37°C than at 30°C (Fig. 1).

For L. monocytogenes parent strain 10403S and the 14 RR mutants grown to early stationary phase at 37°C, the effect of the factor “strain” on Caco-2 invasion efficiencies was not statistically significant (p = 0.3900; ANOVA). Therefore, the appropriate conservative statistical test (i.e., ANOVA) found no statistically significant reduction in Caco-2 invasion for any of the 14 RR mutants grown at 37°C.

As L. monocytogenes motility has been reported in other L. monocytogenes strains to be regulated by DegU and CheY at temperatures ≤30°C (Dons et al., 2004; Knudsen et al., 2004; Williams et al., 2005a; Mauder et al., 2008), we conducted swarming assays to evaluate motility of the L. monocytogenes 10403S, ΔdegU, ΔcheY, and ΔflaA strains. While 10403S was clearly motile when grown at 30°C, ΔdegU, ΔcheY, and ΔflaA showed reduced swarming (6%, 6%, and 1%, respectively, of the swarming area for 10403S grown at 30°C; see Supplemental Fig. S1, available online at www.liebertonline.com). When grown at 37°C, the parent strain showed limited and considerably reduced swarming compared to 30°C (Supplemental Fig. S1), and the three mutant strains showed no detectable swarming at this temperature. The finding that 10403S exhibited some swarming at 37°C is consistent with previous reports that L. monocytogenes 10403S (the parent strain used here) shows limited flaA expression even at 37°C (Grundling et al., 2004; Way et al., 2004).

Effects of temperature shifts and holds on growth temperature-dependent phenotypes

On the basis of the data reported above, we evaluated whether the increased invasion efficiency of 10403S at 30°C is maintained after a shift to 37°C. While L. monocytogenes 10403S grown at 30°C and held (in PBS) at 37°C for 2.5 h (30°C–37°C) showed numerically (about twofold) reduced invasion efficiency compared to bacteria either before the shift or bacteria shifted to 30°C for 2.5 h (Table 2), the invasion efficiency of the “30°C–37°C” treatment group was not significantly different from the control groups grown and held at 30°C (30°C–30°C in Table 2) (p > 0.05; Tukey HSD) but was significantly higher than control groups grown and held at 37°C (37°C–37°C) (p < 0.05; Tukey HSD). These results suggest that increased invasion efficiency of L. monocytogenes grown at 30°C is largely maintained during a 2.5 h hold at 37°C in PBS. When L. monocytogenes 10403S was grown at 30°C, and shifted to and held at 37°C for 5 h (in PBS), the invasion efficiency was reduced (0.15% invasion efficiency) and virtually identical to that of L. monocytogenes grown at 37°C (0.14% invasion efficiency) with no significant difference (p > 0.05; Tukey HSD). This indicates that prolonged exposure of L. monocytogenes grown at 30°C to 37°C (in PBS) reduces invasion efficiencies to values typical for bacteria grown at 37°C. Growth and de novo protein synthesis (which are unlikely to occur in PBS) thus do not seem to be required for reduced invasiveness observed at 37°C.

When the L. monocytogenes parent strain grown at 37°C was switched to 30°C and held at this temperature for 2.5 h, bacteria showed no changes in invasion efficiency and maintained lower invasion efficiency typical for bacteria grown at 37°C (p < 0.05; Tukey HSD; Table 2), suggesting that the increased invasion phenotype cannot be induced by simply switching the organism to 30°C but requires growth at 30°C. Controls included in these experiments showed that exposure to PBS for 2.5 h had no effect on invasion efficiency; bacteria grown at 30°C and exposed to PBS at 30°C for 2.5 h as well as bacteria grown at 37°C and exposed to PBS at 37°C for 2.5 h did not differ in their invasion efficiencies from bacteria grown at 30°C or 37°C, respectively, without subsequent exposure to PBS (p > 0.05; Tukey HSD; Table 2).

As L. monocytogenes ΔdegU, ΔcheY, and ΔflaA showed lower invasion than the parent strain after growth to early stationary phase at 30°C, we also tested whether this invasion defect was maintained after a shift from 30°C to 37°C, followed by a 2.5 h hold at this temperature (in PBS). All three strains (ΔdegU, ΔcheY, and ΔflaA) maintained lower invasion efficiencies than the parent strain even after bacteria grown at 30°C were held for 2.5 h at 37°C (Table 2). Invasion efficiencies for these strains grown at 30°C and shifted to 37°C for 2.5 h were not significantly different (p > 0.05; Tukey HSD; Table 2) from invasion efficiencies for (i) strains grown at 30°C (before shift) or (ii) strains grown at 30°C and subsequently held at 30°C in PBS for 2.5 h (Table 2). The invasion efficiencies for ΔdegU, ΔcheY, and ΔflaA grown at 30°C and held for 2.5 h at 37°C also continued to be significantly lower than the invasion efficiencies for the same strains grown at 37°C (p < 0.05; Tukey HSD); invasion efficiencies for these three strains in the “30°C–37°C” treatment group were ∼20 to 30-fold lower than the same strains grown at 37°C (Table 2). These three mutant strains thus maintained their invasion deficient phenotype even after 2.5 h exposure to 37°C. To test the effect of prolonged exposure to 37°C on the reduced invasion efficiency of motility-deficient mutant strains, invasion efficiency of the ΔflaA strain was also evaluated for bacteria grown at 30°C and shifted to 37°C with a hold in PBS at 37°C for 5 h. Even after a hold at 37°C for 5 h, the ΔflaA strain maintained reduced invasion efficiency (0.006%) compared to the invasion efficiency of either the parent strain grown at 37°C (0.14%) or ΔflaA grown at 37°C (0.14%; both comparisons had p < 0.05; Tukey HSD), suggesting that increased invasion efficiency at 37°C, of a ΔflaA mutant, requires growth at 37°C and probably de novo protein synthesis (which is unlikely to occur in PBS). When the ΔdegU, ΔcheY, and ΔflaA strains were grown at 37°C and shifted to 30°C with a hold at 30°C for 2.5 h, all three strains maintained the higher invasion efficiency as they displayed when grown at 37°C (Table 2).

Temperature-dependent contributions of InlA and FlaA to Caco-2 invasion

In initial experiments, the L. monocytogenes ΔinlA strain showed 4.4-fold higher invasion efficiency when grown at 30°C than at 37°C (Fig. 1). In subsequent experiments, when grown at 37°C, the ΔinlA strain showed, on average, greater than 100-fold lower invasion efficiency than either the parent strain or the ΔdegU, ΔcheY, or ΔflaA strains grown at 37°C (Table 2). However, when grown at 30°C, the ΔinlA strain showed only a 26-fold lower average invasion efficiency than the parent strain (Table 2). These experiments also confirmed a significantly higher invasion efficiency for the ΔinlA grown at 30°C than at 37°C (p < 0.05; Tukey HSD; Table 2). Temperature shifts (i.e., 30°C–37°C, 2.5 h or 37°C–30°C, 2.5 h) did not significantly affect these statistical differences (Table 2). These results indicate an InlA-independent increase in Caco-2 invasion for L. monocytogenes grown at 30°C (as compared to 37°C), which is maintained after a shift to 37°C.

To further investigate the contributions of InlA and flagellin to Caco-2 invasion after growth at 30°C and 37°C and to specifically determine whether there is synergism between InlA and flagellin in facilitating invasion, 10403S, ΔflaA, ΔinlA, and ΔinlAΔflaA were grown at 30°C or 37°C and tested for their Caco-2 invasion efficiencies (Fig. 2). After growth at 30°C, ΔflaA showed ∼30-fold lower invasion efficiency than the parent strain (Fig. 2). At both 30°C and 37°C, the invasion efficiency of the ΔinlAΔflaA strain was lower (p < 0.05; Tukey HSD) than that of the parent strain (0.007% vs. 0.680% for 37°C and 0.003% vs. 2.064% for 30°C, respectively). Further, two-way ANOVA analysis showed a significant “flaA*inlA” interaction effect on Caco-2 invasion for L. monocytogenes grown at both 30°C (p = 0.0015) and 37°C (p = 0.0269). Interestingly, the F-ratio of the interaction effect for bacteria grown at 30°C (F = 20.15) was higher than that of bacteria grown at 37°C (F = 6.97), suggesting a greater contribution of the interaction effect to the observed variance at 30°C than at 37°C. These statistical findings indicate that the effects of deleting inlA and flaA on Caco-2 invasion are more than additive, indicating a synergism between InlA and FlaA in facilitating Caco-2 invasion for bacteria grown at either 30°C or 37°C.

OPEN IN VIEWER

FIG. 2.

Caco-2 invasion efficiencies of L. monocytogenes 10403S, ΔinlA, ΔflaA, and ΔinlAΔflaA strains grown to early stationary phase with aeration (i.e., shaking at 220 rpm) at either 30°C or 37°C. Data represent the mean of four biological replicates. Error bars represent SDs. Strains with different letters had significantly different invasion efficiencies at a given growth temperature (p < 0.05; post hoc Tukey HSD); lower invasion for the ΔflaA strain than the parent strain (for growth at 37°C) was only borderline significant (p = 0.0279; Tukey HSD). Two-way ANOVA was performed within each temperature to determine effects of single and double mutations on invasion efficiency (see Table 3 for p-values). HSD, honestly significantly different.

Temperature-dependent contributions of transcriptional regulators σ^B and PrfA to Caco-2 invasion

To determine growth temperature effects on the contributions of σ^B and PrfA to Caco-2 invasion, ΔprfA, ΔsigB, and ΔsigBΔprfA strains grown to early stationary phase at 30°C or 37°C were used for Caco-2 invasion assays (performed at 37°C). After growth at 30°C, the invasion efficiency of ΔprfA was 2.737%, which was not significantly different (p > 0.05; Tukey HSD) than the parent strain (2.064% invasion efficiency); the ΔsigB (0.376% invasion) and ΔsigBΔprfA (0.231% invasion) strains were significantly less invasive (p < 0.05; Tukey HSD) than the parent strain (Fig. 3). These results indicate that only σ^B, not PrfA, is involved in invasion when L. monocytogenes is grown at 30°C. When the bacteria were grown at 37°C, ΔprfA (0.322%), ΔsigB (0.060%), and ΔsigBΔprfA (0.018%) had significantly lower invasion efficiencies (p < 0.05; Tukey HSD) than the parent strain (0.680%) (Fig. 3); the ΔsigBΔprfA strain also showed lower invasion efficiency than the parent strain and the ΔprfA strain (p < 0.05; Tukey HSD). A two-way ANOVA also showed a significant “sigB*prfA” interaction effect, indicating a synergism between σ^B and PrfA in regulating Caco-2 invasion for L. monocytogenes grown at 37°C (Table 3).

OPEN IN VIEWER

FIG. 3.

Caco-2 invasion efficiencies of L. monocytogenes 10403S, ΔprfA, ΔsigB, and ΔsigBΔprfA strains grown to early stationary phase with aeration (i.e., shaking at 220 rpm) at either 30°C or 37°C. Data for the parent strain (10403S) are the same as those shown in Figure 2. Data represent the mean of four biological replicates. Error bars represent SD. Strains with different letters have significantly different invasion efficiencies at a given growth temperature (p < 0.05; post hoc Tukey HSD). Two-way ANOVA was performed within each temperature to determine effects of single and double mutations on invasion efficiency (see Table 3 for p-values).

Temperature-dependent regulation of genes involved in Caco-2 invasion

To determine the effects of growth temperature on σ^B and PrfA-dependent regulation of L. monocytogenes genes involved in invasion, we determined transcript levels for inlA, flaA, plcA, gadA, sigB, prfA, and rpoB in 10403S, and ΔsigB, ΔprfA, and ΔsigBΔprfA grown to early stationary phase at 30°C or 37°C. Neither prfA nor sigB transcript levels differed between the parent strain grown at 30°C and 37°C (p > 0.05, t-test). The transcript levels for the σ^B-dependent gadA and the PrfA-dependent plcA (measured as indicators of σ^B and PrfA activity, respectively) (Fig. 4) also did not differ significantly between bacteria grown at 30°C and 37°C (p > 0.05; t-test), suggesting no differences in PrfA and σ^B activity between L. monocytogenes 10403S grown to early stationary phase in BHI at these two temperatures.

OPEN IN VIEWER

FIG. 4.

Normalized transcript levels of four genes in L. monocytogenes 10403S, ΔsigB, ΔprfA, and ΔsigBΔprfA grown to early stationary phase with aeration (i.e., shaking at 220 rpm) at 30°C or 37°C. Data presented as log₁₀ (target gene mRNA level/rpoB mRNA level). Bars represent the average of four biological replicates and error bars represent SD. Strains with different letters (e.g., A, B) had significantly different transcript levels (p < 0.05; post hoc Tukey HSD). Two-way ANOVA was performed for each gene at each temperature to determine effects of single and double mutations on transcript levels (see Supplemental Table S2 for p-value summary). For each gene, a t-test was used to compare transcript levels in 10403S between 30°C and 37°C. Only flaA levels were significantly different between temperatures (p = 0.0014; t-test; indicated by *).

For bacteria grown to early stationary phase at 30°C, inlA transcript levels were 10 times lower in the ΔsigB strain than the parent strain (p < 0.05; Tukey HSD; Fig. 4), whereas there was no significant effect of the prfA deletion on inlA transcript levels (p = 0.1282; two-way ANOVA) (Supplemental Table S2, available online at www.liebertonline.com). These data indicate that inlA transcription is σ^B dependent, but PrfA independent under these conditions. For L. monocytogenes grown at 37°C, inlA transcript levels were significantly lower in ΔsigB than in the parent strain (p < 0.05; ANOVA; Tukey HSD; Fig. 4), whereas inlA transcript levels were not significantly different in ΔprfA compared to the parent strain (p > 0.05; Tukey HSD; Fig. 4). However, the sigB*prfA interaction effect on inlA transcript levels was borderline significant (p = 0.0799; two-way ANOVA; Supplemental Table S2), suggesting contributions of both σ^B and PrfA to regulation of inlA transcription in L. monocytogenes grown at 37°C.

In the L. monocytogenes parent strain, flaA transcript levels were higher after growth at 30°C (1.26; standard deviation = 0.07) than at 37°C (1.104; standard deviation = 0.05) (p = 0.0014, t-test; data shown in Fig. 4). Although the difference is small, these results indicate at least some temperature-dependent regulation of flaA transcription in 10403S. flaA transcript levels were not significantly different between ΔprfA and the parent strain for either growth temperature (p > 0.05; Tukey HSD; Fig. 4). In L. monocytogenes grown at 37°C, flaA transcript levels were not different between the ΔsigB strain and the parent strain (p > 0.05; Tukey HSD; Fig. 4). However, at 30°C flaA transcription was significantly higher in ΔsigB and ΔsigBΔprfA than in the parent strain (p < 0.05; ANOVA, Tukey HSD; Fig. 4), and two-way ANOVA showed a significant “sigB effect” on flaA transcript levels (p = 0.0003; Supplemental Table S2). These results are consistent with other studies that suggest a role of σ^B in downregulating chemotaxis genes (Raengpradub et al., 2008; Toledo-Arana et al., 2009).

L. monocytogenes grown at 30 °C shows a higher Caco-2 invasion efficiency than bacteria grown at 37 °C

Out data showed that, with the exception of mutants with deletions of motility-related genes, all L. monocytogenes strains showed higher Caco-2 invasion efficiency when grown at 30°C than at 37°C. This finding supports previous work (Dons et al., 2004) reporting that a different L. monocytogenes strain (12067) showed about 10-fold increased association with Caco-2 cells and twofold increased invasion of Caco-2 cells for bacteria grown at 24°C (where 12067 was motile) compared to 37°C (where this strain was not motile); this previous study did not include a statistical evaluation of these differences in invasion efficiencies (Dons et al., 2004). As most L. monocytogenes, including strain 10403S, show increased flagellar motility when grown at 30°C, a trait that contributes to Caco-2 cell invasion (Dons et al., 2004) and intestinal colonization in mice (O'Neil and Marquis, 2006), expression of motility genes at ≤30°C may explain the increased Caco-2 invasion of L. monocytogenes grown at ≤30°C compared to bacteria grown at 37°C.

Interestingly, other bacterial pathogens have also been shown to differ in their invasiveness and virulence depending on growth temperature (Konkel and Tilly, 2000). While in a number of bacterial pathogens virulence genes have been found to be more highly expressed at 37°C than at lower temperatures (Maurelli, 1989), some pathogens have been shown to express invasion factors at higher levels when grown at temperatures below 37°C. For example, Yersinia pseudotuberculosis invA, which encodes an invasin required for host cell invasion, appears to be expressed at higher levels in bacteria grown at 28°C than at 37°C (based on visual examination of Western blot data) (Isberg et al., 1988). Yersinia enterocolitica has also been shown to express a motility phenotype when grown below 37°C, whereas motility, which may play a role in initiation of host cell invasion, is downregulated in bacteria grown at 37°C (Young et al., 2000). Therefore, adaptation to environments outside the host may increase the virulence potential of L. monocytogenes and other bacterial pathogens.

Enteric pathogens transmitted to humans from food and environmental sources often experience a sudden change in environmental temperature when they are ingested. Our results show that invasion phenotypes of the parent strain and the motility mutants grown at 30°C persisted for 2.5 h after the bacteria were switched to 37°C. As 70% to 90% of human stomach contents are emptied after 2 h (Bennink et al., 1999), our findings, in combination with other studies (Dons et al., 2004; O'Neil and Marquis, 2006), suggest that L. monocytogenes grown at temperatures that permit motility [i.e., 12°C–30°C (Di Bonaventura et al., 2008)] may have increased invasion potential that could be maintained during gastric passage. As regulation of motility appears to differ considerably between L. monocytogenes strains (Grundling et al., 2004; Way et al., 2004), further studies will need to use different strains to validate our findings on growth temperature dependence of Caco-2 invasion efficiency.

In addition to growth temperature, other studies (Garner et al., 2006a; Andersen et al., 2007) have shown that exposure to other environmental conditions (e.g., organic acids, anaerobic conditions) appear to affect the virulence potential of L. monocytogenes. For example L. monocytogenes grown in an oxygen-limited environment displayed 100-fold increased invasion efficiency in Caco-2 cells as well as increased virulence in a guinea pig model, including 10 to 100-fold higher L. monocytogenes fecal shedding levels, compared to numbers for bacteria grown under aerobic conditions (Andersen et al., 2007). Further, L. monocytogenes grown in the presence of sodium lactate or NaCl showed an about 10-fold higher invasion efficiency than bacteria grown without these compounds (Garner et al., 2006a). Overall, growth temperatures as well as other preinvasion environmental conditions (e.g., anaerobiosis) thus appear to affect the regulation of genes with roles in host attachment and invasion across different environmentally transmitted pathogens, including L. monocytogenes.

CheY and DegU significantly contribute to invasion in L. monocytogenes grown ≤30°C

Our data, in conjunction with other reports (e.g., Dons et al., 2004), clearly show that DegU and CheY play a temperature-dependent role in enhancing invasion of host cells across different L. monocytogenes strain backgrounds. For example, ΔcheA, ΔcheY, and ΔcheYA mutants in an L. monocytogenes 12067 background all showed about 100-fold reduced Caco-2 cell invasion as well as reduced motility as compared to their parent strain after growth at 24°C (Dons et al., 2004), whereas Williams et al. (2005a) found that L. monocytogenes EGD ΔdegU and ΔcheY strains, grown at 37°C, did not show reduced invasion efficiency of Cos-1 fibroblast cells. Previous studies have shown increased swarming and flagellar motility of L. monocytogenes grown at 24°C compared to L. monocytogenes grown at 37°C (Dons et al., 2004; Knudsen et al., 2004; Shen and Higgins, 2006), due to repression of L. monocytogenes motility genes at 37°C (Grundling et al., 2004; Shen and Higgins, 2006). Therefore, contributions of motility genes to virulence phenotypes are generally apparent only when L. monocytogenes is grown at temperatures where the organism is typically motile (i.e., ≤30°) (Dons et al., 2004; Knudsen et al., 2004; Williams et al., 2005b; O'Neil and Marquis, 2006). While DegU has also been shown to contribute to motility in L. monocytogenes grown at 24°C and to virulence in a mouse model (Williams et al., 2005b), we are not aware of any previous studies reporting a specific effect of a degU null mutation on invasion of human intestinal epithelial cells.

While we found no contributions of other RRs, besides DegU and CheY, to invasion of Caco-2 cells, some studies have shown contributions of other RRs to virulence phenotypes in L. monocytogenes strains other than 10403S and in other invasion models. For example, Williams et al. (2005a) reported that RRs Lmo1507 and LisR are involved in invasion of Cos-1 cells. A deletion of lisK (encoding the LisR-associated sensory kinase) in L. monocytogenes LO28 was also found to affect virulence in mice (Cotter et al., 1999), further supporting contributions of LisRK to L. monocytogenes virulence in some models. Similarly, L. monocytogenes EGD with a deletion of the RR VirR, grown at 37°C, has also been reported to be deficient in Caco-2 cell invasion (Mandin et al., 2005), even though the L. monocytogenes 10403S ΔvirR strain used in our current study did not show evidence for reduced invasion, again possibly reflecting strain differences. Overall, different TCS thus appears to contribute to L. monocytogenes virulence and virulence-associated characteristics, even though contributions appear to differ based on growth conditions, strain backgrounds, and assays used.

While InlA and FlaA show significant synergism in their contributions to Caco-2 invasion, synergism is less pronounced after growth at 37 °C

While flagellin has been identified as a critical part of the motility machinery involved in Caco-2 invasion in L. monocytogenes grown at temperatures that allow motility gene expression (i.e., ≤30°C) (Dons et al., 2004; O'Neil and Marquis, 2006), we found flagellin and InlA contribute synergistically to Caco-2 cell invasion in L. monocytogenes 10403S grown at 30°C and 37°C. While studies in other L. monocytogenes strains (e.g., 12067, EDGe) have shown that contributions of motility factors to host cell invasion typically are only apparent in L. monocytogenes grown at 30°C or less (Dons et al., 2004; Shen and Higgins, 2006), some studies (Grundling et al., 2004; Way et al., 2004) have shown increased flagellar motility at 37°C for strain 10403S as compared to other strains. For example, flaA repression by MogR was shown to be less stringent in 10403S than in EGDe (Grundling et al., 2004), and 10403S activated a flagellum-dependent innate immune response even after growth at 37°C (Way et al., 2004). Critical contributions of InlA to invasion of human intestinal epithelial cells (Dramsi et al., 1993; Lingnau et al., 1995) and virulence after oral infection (Lecuit et al., 1999; Garner et al., 2006b) have been well established for host species carrying the E-cadherin allotype that allows for InlA binding (e.g., humans and guinea pigs). Although we observed a synergism between the contributions of FlaA and InlA to Caco-2 invasion in L. monocytogenes grown at both 30°C and 37°C, we also found that the ΔinlA strain shows increased invasion efficiency when grown at 30°C compared to 37°C, suggesting InlA-independent contributions of flagellar motility to Caco-2 cell invasion. This observation further supports the importance of flagellar motility in L. monocytogenes virulence.

σ^B, but not PrfA, contributes to invasion of bacteria grown at 30 °C, whereas σ^B and PrfA show synergistic contributions to invasion if bacteria are grown at 37 °C

Our data showed that (i) σ^B contributes to Caco-2 invasion for L. monocytogenes grown at both 30°C and 37°C and (ii) σ^B positively regulates inlA transcription at both of these temperatures. The observation that σ^B contributes to Caco-2 invasion in bacteria grown at both temperatures is consistent with a number of studies that have shown that σ^B is critical for invasion of intestinal cells in vitro (Kim et al., 2004; Garner et al., 2006b) and in a guinea pig model of listeriosis (Garner et al., 2006b). While McGann et al. (2007a) found higher transcript levels in L. monocytogenes grown at 30°C than bacteria grown at 37°C, for certain σ^B-dependent internalin genes (i.e., inlC2, inlD, lmo331, and lmo0610) and opuCA, which has been shown to be regulated by both σ^A and σ^B (Cetin et al., 2004; Chan et al., 2007a), the same study also found that inlA transcript levels were similar in L. monocytogenes grown at 30°C and 37°C (McGann et al., 2007a). Another study also reported that genes found to be σ^B dependent in the intestine were generally not differentially expressed at 30°C and 37°C (Toledo-Arana et al., 2009). Overall, these data suggest that σ^B activity and σ^B-dependent regulation of invasion is similar in L. monocytogenes grown at 30°C and 37°C, even though some σ^B-dependent genes (e.g., opuCA and inlC2D) may be regulated by additional temperature-dependent mechanisms (McGann et al., 2007a; Chan et al., 2007a). We did find though that σ^B-dependent negative regulation of flaA transcript levels, which was previously described by two studies (Raengpradub et al., 2008; Toledo-Arana et al., 2009), was only apparent here in bacteria grown at 30°C, even though this regulation has previously been reported in bacteria exposed to salt stress at 37°C (Raengpradub et al., 2008). Temperature effects on σ^B-dependent regulation of some genes may thus be dependent on other environmental conditions (e.g., osmotic stress).

Contributions of PrfA to Caco-2 cell invasion in L. monocytogenes grown at 37°C, but not in bacteria grown at 30°C, are consistent with previous findings of low PrfA activity in L. monocytogenes grown at temperatures <37°C (Leimeister-Wachter et al., 1992) as well as data showing that PrfA-dependent inlA expression is significantly higher at 37°C than at 25°C (Dramsi et al., 1993). Interestingly, in our study here, prfA transcript levels and PrfA activity (as measured by plcA transcription levels) were not significantly different in the parent strain between 30°C and 37°C growth conditions; this observation may reflect low baseline PrfA activity in L. monocytogenes grown at 37°C in BHI (i.e., the conditions used here) as low levels of easily catabolized sugar and/or presence of other compounds (e.g., charcoal) seem to be required to induce PrfA activity at 37°C (Ripio et al., 1996; Milenbachs et al., 1997; Gilbreth et al., 2004). While PrfA activity and PrfA-dependent phenotypes are thus clearly temperature dependent, with maximum PrfA activity in bacteria grown under certain conditions at 37°C, it is increasingly clear that transcriptional patterns and phenotypic characteristics of L. monocytogenes are governed by complex, environmental-condition-dependent interactions between multiple regulators.

Synergisms between PrfA and σ^B were confirmed here through formal statistical analyses that showed (i) a statistically significant interaction effect between sigB and prfA deletions on Caco-2 invasion in bacteria grown at 37°C, but not in bacteria grown at 30°C and (ii) a borderline significant interaction effect between sigB and prfA deletions on inlA transcript levels in bacteria grown at 37°C, but not at 30°C. While this type of temperature-dependent synergism has not previously been described, contributions of both σ^B and PrfA to Caco-2 cell invasion and co-regulation of L. monocytogenes virulence genes, including inlA, have been reported previously (Lingnau et al., 1995; Kazmierczak et al., 2003; Sue et al., 2004; Kim et al., 2005; McGann et al., 2007b). The observation that flaA also appears to be downregulated by σ^B, possibly with an antisense RNA type mechanism (Toledo-Arana et al., 2009), further supports a temperature-dependent regulatory network involving PrfA and σ^B that affects multiple effector proteins contributing to L. monocytogenes invasion and virulence, even though possible (temperature-dependent) contributions of PrfA itself to transcription of flaA and other motility regulated genes (Michel et al., 1998; Milohanic et al., 2003) will require further confirmation. For example, while others (Ripio et al., 1997) previously reported (based on visual examination of an RNA slot blot) negative regulation of flaA transcription by PrfA* (i.e., a PrfA protein that is constitutively active) in L. monocytogenes grown at 20°C, we did not find any effect of the prfA deletion on flaA transcript levels.