Listeria monocytogenes σ^B Contributes to Invasion of Human Intestinal Epithelial Cells

Abstract

The role of σ^B in Listeria monocytogenes infection of human intestinal epithelial cells was investigated. Invasion defects associated with loss of σ^B paralleled those of a ΔinlA strain independently of the σ^B-dependent P2_prfA promoter. Concomitantly, amounts of inlA transcript and InlA protein were significantly decreased in the ΔsigB strain.

The gram-positive facultative intracellular food-borne pathogen Listeria monocytogenes is associated with serious invasive infections in humans and animals (10). To establish a food-borne bacterial infection, a pathogen must survive rapidly changing environmental conditions encountered in the gastrointestinal tract, such as exposure to bile salts, organic acids, and changing osmolarity. Survival of extreme and rapidly changing conditions requires timely and appropriate alterations in bacterial gene expression and protein activity. At the transcriptional level, these alterations are often controlled by alternative sigma factors and the catalytic core of RNA polymerase (24). Alternative sigma factors differentially associate with the core polymerase and essentially reprogram promoter recognition specificities of the enzyme, thus allowing expression of new sets of target genes. For L. monocytogenes, the alternative sigma factor σ^B (encoded by sigB) contributes to bacterial resistance to environmental stress conditions, such as reduced temperature, oxidative stress, carbon starvation, and low pH (2, 3, 12, 13, 39), as well as to virulence in the mouse (29, 39).

L. monocytogenes is capable of survival and growth under widely varying environmental conditions (11). Further, this organism is able to invade and multiply in a wide range of professional and nonprofessional phagocytic mammalian cells (38). Two cell wall-anchored proteins, internalin A (InlA) and internalin B (InlB), are necessary for efficient invasion of human nonprofessional phagocytic cells (7, 16). The intracellular infectious cycle is characterized by escape from the primary vacuole, intracytosolic replication, actin-based motility, and direct spread to neighboring cells, where a new cycle initiates (17, 36). Several virulence factors responsible for these cellular events have been identified. A pore-forming hemolysin (listeriolysin O) (17), along with two phospholipases (PI-PLC and PC-PLC) (6, 37), is involved in bacterial escape from vacuoles, and the ActA surface protein is responsible for F-actin polymerization and intracellular bacterial motility (21). A DNA binding protein, PrfA, controls expression of these virulence genes (5, 8, 22, 23, 25, 33), but the precise mechanism of regulation is not well understood. prfA is transcribed from three independent promoters, P_plcA, P1_prfA, and P2_prfA (15). PrfA has been suggested to exist in functionally active or inactive forms, depending on environmental conditions (5, 19, 27, 31, 32), and to interact differently at various target promoters (15, 40). The presence of σ^B-dependent promoters upstream of prfA (P2_prfA [29]) and inlA (P4_inlA [20]) suggests that σ^B contributes to L. monocytogenes virulence. Further, Milohanic et al. (28) have reported that a number of PrfA-regulated genes have putative σ^B-dependent promoters. In the present study, the role of σ^B in infection of human intestinal epithelial cells was investigated.

σ^B contributes to early stages of infection in human intestinal epithelial cells.

Intracellular growth of the wild-type L. monocytogenes strain 10403S (4) as well as ΔsigB (FSL A1-254) (39) and ΔP2_prfA (DP-L1957) (14) isogenic mutant strains was examined in two intestinal epithelial cell lines, Henle-407 and Caco-2. The assay was performed as previously described (30). Briefly, bacteria were grown at 30°C overnight in brain heart infusion (BHI) broth (with shaking at 250 rpm) and washed in phosphate-buffered saline (PBS) before infection. Host cell infections were performed at 37°C in a 5% CO₂ atmosphere. Monolayers of host cells on glass coverslips were infected with a multiplicity of infection (MOI) of approximately 30 and 3 for Henle-407 and Caco-2 cells, respectively. Cells were washed at 1 h postinfection, and gentamicin was added at 1.5 h postinfection. Gentamicin concentrations were experimentally optimized to 150 μg/ml (Henle-407) or 50 μg/ml (Caco-2) to achieve effective killing of extracellular bacteria within 30 min. The number of intracellular bacteria per coverslip was determined at various time points during infection by plating appropriate serial dilutions of infected host cell lysates onto Luria-Bertani (LB) agar plates. Intracellular doubling times were evaluated by one-way analysis of variance (ANOVA), and individual comparisons were made with Bonferroni's multiple comparison test to determine whether a mean doubling time for one strain was significantly different from those for other strains. Intracellular growth rates were not significantly different among the various strains tested (Fig. 1; Table 1). However, at 2.5 h postinfection intracellular ΔsigB numbers were consistently 2.0- to 2.5-fold lower than those of the wild-type strain in both cell lines. Intracellular ΔP2_prfA numbers were not different than those of the wild-type strain in either cell line. These results suggest that σ^B contributes to infection independently of its activity at P2_prfA, at a stage prior to intracellular bacterial multiplication in human intestinal epithelial cells.

FIG. 1.

Intracellular growth of L. monocytogenes in human epithelial cells. Viable bacterial counts were determined as CFU per coverslip at specific time points postinfection in (A) Henle-407 and (B) Caco-2 cells. Data represent mean CFU from three independent experiments. Error bars reflect standard deviations from each mean. WT, wild type.

TABLE 1.

Intracellular doubling times^a

Strain	Doubling time (min) for Henle-407 at:		Doubling time (min) for Caco-2 at:
	2.5-5.5 h	5.5-8.5 h	2.5-5.5 h	5.5-8.5 h
Wild type	59 ± 5	119 ± 61	58 ± 3	67 ± 5
ΔsigB	59 ± 11	163 ± 87	59 ± 7	70 ± 8
ΔP2prfA	59 ± 8	116 ± 35	54 ± 5	77 ± 12

^aThe means and standard deviations from three independent experiments are shown for each strain. At 2.5 to 5.5 h and 5.5 to 8.5 h, P values were 0.9934 and 0.6299 for Henle-407, respectively, and they were 0.5380 and 0.8503, respectively, for Caco-2.

σ^B promotes invasion of human epithelial cells.

To define the role of σ^B in early infection, we evaluated the relative abilities of the wild-type, ΔsigB, and ΔP2_prfA strains to adhere to Henle-407 and Caco-2 human epithelial cells. At 30 min prior to infection, cytochalasin D (1 μg/ml) was added to host cell monolayers to disrupt actin polymerization and prevent bacterial internalization. Bacteria were grown as described above, and cells were infected to an MOI of approximately 5. After either 1 h (Henle-407) or 30 min (Caco-2) of incubation, cells were washed six times with PBS and the number of adherent bacteria per coverslip was determined as described above. The fraction of inoculum recovered was calculated for each individual sample. To control for inhibition of bacterial internalization, a sample was treated with gentamicin after the wash and bacterial counts were determined after 1 h. No bacteria were recovered under these conditions. Results were evaluated by one-sample t test to determine if a given mutant strain phenotype was significantly different from that of the wild-type strain. The three bacterial strains tested did not differ in their abilities to adhere to either host cell line (Table 2).

TABLE 2.

Adhesion and invasion characteristics of L. monocytogenes wild-type and mutant strains^a

Strain	Adhesion		Invasion
	Henle-407	Caco-2	Henle-407	Caco-2
Wild type	1.00	1.00	1.00	1.00
ΔsigB	0.81 ± 0.17	0.68 ± 0.20	0.54 ± 0.11***b	0.23 ± 0.10***
ΔP2prfA	0.95 ± 0.18	1.28 ± 0.33	1.08 ± 0.11	1.07 ± 0.11
ΔinlA	NDc	ND	0.52 ± 0.20**	0.21 ± 0.10**
ΔinlB	ND	ND	0.79 ± 0.27	0.93 ± 0.35

^aResults are reported as ratios of mutant strain numbers relative to wild-type strain numbers, with wild-type numbers arbitrarily set to 1. The means and standard deviations from at least three independent experiments are shown for each strain.

^bP values (*, <0.05; **, <0.01; ***, <0.001) for comparison between respective mutant strain and wild-type invasion by one-sample t test are indicated by asterisks.

^cND, not determined.

Next we evaluated σ^B's contribution to invasion of human epithelial cells. L. monocytogenes invasion of nonphagocytic cells is predominantly mediated by two cell surface-anchored proteins, InlA and InlB (7, 16). The inlA and inlB genes comprise an operon. Transcriptional analysis has identified four promoters upstream of inlA and one upstream of inlB (23). P3_inlA is a PrfA-regulated promoter (23), whereas P4_inlA is σ^B dependent (20). A sequence resembling a σ^B-dependent promoter also has been reported upstream of inlB, but promoter activity at this site has not been confirmed (20). Therefore, we hypothesized that σ^B's contribution to inlA expression, and possibly to inlB expression, would influence the ability of L. monocytogenes to invade nonphagocytic cells. To assess the respective contributions of σ^B, P2_prfA, InlA, and InlB to host cell invasion, invasion capabilities were compared among wild-type 10403S, ΔsigB (FSL A1-254) (39), ΔP2_prfA (DP-L1957) (14), ΔinlA (EJL12) (provided by Jeff Miller) (1), and ΔinlB (HEL-137; from this study). For this purpose, we created an inlB internal in-frame deletion mutant. The deletion was generated by site-directed mutagenesis by overlap extension (18), using primers Marq 37 (5′-GCG GAT CCT ACG GCA CCA ACA AAA G-3′), Marq 38 (5′-GTG GAA CTA GTC CTT ATT CGC TTC CTT CTT GGG TTG TGC-3′), Marq 39 (5′-GCA CAA CCC AAG AAG GAA GCG AAT AAG GAC TAG TTC CAC-3′), and Marq 41 (5′-GCT AGC ATG CAG TGA AAT TAT TGC TGG T-3′) (7). The inlB deletion was introduced into the chromosome of L. monocytogenes strain 10403S by allelic exchange as previously described (6), creating HEL-137. As for the intracellular growth assay, bacteria were grown at 30°C overnight (with shaking at 250 rpm) and host cell infections were performed at 37°C in a 5% CO₂ atmosphere. Host cells were infected to an MOI of approximately 30, and infection times were optimized for each cell type. For invasion of Henle-407 cells, cells were washed at 1 h postinfection and were treated with gentamicin at 1.5 h postinfection. For invasion of Caco-2 cells, cells were washed at 30 min postinfection and were treated with gentamicin at 45 min postinfection. Numbers of internalized bacteria per coverslip were determined 30 min after gentamicin treatment as described above. The fraction of inoculum recovered was calculated for each individual sample. Results were statistically evaluated by one-sample t test. The ΔsigB strain was less effective at invading Henle-407 (P < 0.001) and Caco-2 cells (P < 0.001) than the wild-type strain (Table 2). The invasion defect of the ΔsigB strain was comparable to that of the ΔinlA strain (P = 0.791 for Henle-407, P = 0.784 for Caco-2), but it was significantly more pronounced than that of the ΔinlB strain (P = 0.029 for Henle-407, P = 0.001 for Caco-2). Specifically, relative invasion capabilities of the ΔsigB, ΔinlA, and ΔinlB strains were reduced by 1.9-, 1.9-, and 1.3-fold, respectively, in Henle-407 cells and by 4.3-, 4.8-, and 1.1-fold, respectively, in Caco-2 cells. Invasion phenotypes of ΔP2_prfA and wild-type strains did not differ. Overall, these results show that σ^B significantly increases the efficacy of L. monocytogenes to invade human epithelial cells, independent of P2_prfA transcriptional activity and of InlB. It is possible that σ^B influences InlB-dependent invasion in other cell types, such as hepatocytes or fibroblasts cells (7).

σ^B contributes to inlA expression.

The above observations led us to focus on InlA as an important factor affecting L. monocytogenes virulence in a σ^B-dependent manner. One of the key early events in human listeriosis is bacterial entry into the intestinal epithelium. A number of studies have shown that InlA is a major bacterial ligand involved in this process (9, 16, 23, 26). Therefore, levels of inlA expression in wild-type, ΔsigB, and ΔP2_prfA strains were determined by semiquantitative reverse transcription-PCR (RT-PCR). Bacteria were grown with shaking (250 rpm) in BHI broth overnight at 30°C. Total RNA was isolated with the RNeasy Midi kit (Qiagen, Valencia, Calif.) according to the manufacturer's protocol. Purified RNA was quantified by absorbance at 260 nm immediately before use, and RT was performed with the Superscript First-Strand Synthesis RT-PCR system (Invitrogen, Carlsbad, Calif.) with 50 ng of total RNA as template (35). The rpoB gene was used as an internal control. Primers used in this assay (InlATqMnF1 [5′-GGT CTC ACA AAC AGA TCT AGA CCA AGT-3′], InlATqMnR1 [5′-TCA AGT ATT CCA CTC CAT CGA TAG ATT-3′], RpoBTqMnF1, and RpoBTqMnR1 [35]) were designed with PrimerExpress software (Applied Biosystems, Foster City, Calif.). Briefly, cDNA was synthesized with oligonucleotides specific for inlA (InlATqMnR1) and rpoB (RpoBTqMnR1) and then amplified by PCR using the AmpliTaq Gold DNA polymerase system (Applied Biosystems). Thirty PCR cycles were selected for cDNA amplification, as no visible products were observed after either 15 or 22 cycles. RT-PCR products were run on 3% agarose gels and quantified with LabImage software (Kapelan, Halle, Germany). Results were statistically analyzed by one-sample t test. RT-PCR results clearly showed decreased levels of inlA transcript in the ΔsigB strain relative to that in the wild-type strain (Fig. 2A). In the absence of σ^B, inlA expression was decreased by 3.8-fold (P = 0.001), which is consistent with the relative invasion defect observed for the ΔsigB strain in human epithelial cells. Quantities of the inlA transcript in the ΔP2_prfA strain were essentially identical to those of the wild-type strain (P = 0.212), which is also consistent with the invasion phenotype of this strain in human epithelial cells.

FIG. 2.

Detection of inlA transcript and InlA protein. (A) Total RNA was isolated from wild-type L. monocytogenes as well as from ΔsigB and ΔP2_prfA strains, and inlA transcript levels were assayed by RT-PCR. rpoB, encoding the β subunit of the DNA-dependent RNA polymerase, was used as an internal control. No PCR products were obtained in the absence of RNA (data not shown). Lane 1, DNA markers in base pairs (bp); lane 2, wild type; lane 3, ΔsigB; lane 4, ΔP2_prfA. (B) Cell wall proteins were isolated from wild-type L. monocytogenes as well as from ΔsigB, ΔP2_prfA, and ΔinlA strains, and InlA was detected by Western immunoblotting. Samples were normalized for equal CFU (7.5 × 10⁸) from each original culture. Lane 1, prestained protein markers; lane 2, wild type; lane 3, ΔsigB; lane 4, ΔP2_prfA; lane 5, ΔinlA. kD, kilodaltons.

Levels of cell wall-associated InlA in wild-type, ΔsigB, and ΔP2_prfA strains were estimated by Western immunoblotting. For isolation of L. monocytogenes cell wall proteins, bacterial cells in protoplast buffer (0.5 M sucrose, 50 mM Tris-HCl [pH 8.0], 10 mM MgCl₂, 0.02% sodium azide) were treated with mutanolysin to a final concentration of 100 U/ml of original bacterial culture at an OD₆₀₀ of 1.0. Protoplast formation was monitored by microscopy. Cell wall extracts were processed for detection of InlA by Western immunoblotting as previously described (34). InlA was virtually undetectable in the ΔsigB strain (Fig. 2B). On the other hand, equivalent amounts of InlA were present in the wild-type and ΔP2_prfA strains (Fig. 2B). Based on the observations that inlA transcription is regulated in a growth-phase-dependent manner and that inlA expression peaks upon entry into stationary phase in the absence of PrfA, Sheehan et al. (33) suggested the existence of a PrfA-independent mechanism that contributes to regulation of inlA expression in stationary-phase L. monocytogenes. Our data suggest that this proposed stationary-phase mechanism is mediated by σ^B. σ^B activity is also regulated in a growth-phase-dependent manner, with maximal activity upon entry into stationary phase (13, 35). Most importantly, we have demonstrated that σ^B is a major contributor to stationary-phase inlA expression. Taken together, our observations highlight the critical contribution of σ^B to expression of inlA and consequently to invasion of human intestinal epithelial cells.

In conclusion, we have identified a specific mechanism linking the general stress-responsive sigma factor, σ^B, with the ability of L. monocytogenes to invade human epithelial cells. Specifically, we have demonstrated that the loss of σ^B significantly impairs the ability of L. monocytogenes to invade human epithelial cells in a manner that is independent of the σ^B-dependent promoter P2_prfA. Moreover, our data clearly demonstrate that the importance of σ^B in bacterial invasion is related, at least in part, to its role in the regulation of inlA expression, presumably by controlling activity at P4_inlA.