A gene-expression program reflecting the innate immune response of cultured intestinal epithelial cells to infection by Listeria monocytogenes

Abstract

Background: Listeria monocytogenes is a Gram-positive, facultative, intracellular bacterial pathogen found in soil, which occasionally causes serious food-borne disease in humans. The outcome of an infection is dependent on the state of the infected individual's immune system, neutrophils being key players in clearing the microorganism from the body. The first line of host defense, however, is the intestinal epithelium.

Results: We have examined the transcriptional response of cultured human intestinal epithelial cells to infection by L. monocytogenes, which replicates in the host cell cytoplasm and spreads from cell to cell using a form of actin-based motility. We found that the predominant host response to infection was mediated by NFkappaB. To determine whether any host responses were due to recognition of specific virulence factors during infection, we also examined the transcriptional response to two bacterial mutants; actA which is defective in actin-based motility, and prfA, which is defective in the expression of all L. monocytogenes virulence genes. Remarkably, we found no detectable difference in the host transcriptional response to the wild-type and mutant bacteria.

Conclusions: These results suggest that cultured intestinal epithelial cells are capable of mounting and recruiting a powerful innate immune response to L. monocytogenes infection. Our results imply that L. monocytogenes is not specifically detected in the host cytoplasm of Caco-2 cells by intracellular signals. This suggests that entry of bacteria is mediated in the host cell post-translationally, and that these bacteria seek the cytosol not only for the nutrient-rich environment, but also for protection from detection by the immune system.

Figure 1

Monitoring of infections. Fluorescence and phase microscopy to monitor infections at 4 h post-infection for (a) mock infection, (b) wild-type L. monocytogenes 10403S, and isogenic mutants deficient in (c) actA or (d) prfA. Filamentous actin was stained with rhodamine phalloidin. Yellow arrows in (b) show comet tails behind wild-type L. monocytogenes (10403S). Purple arrows in (c) indicate intracellular colonies formed by actA (DP-L1942). prfA (DP-L1075) bacteria (d) are seen individually as they do not escape the endosome or replicate after being internalized. Scale bar is 10 μm.

Figure 2

Schematic representation of time-zero transformation. In this process, the logR of the zero time point is subtracted from each time point in the time course. Zero transformation enables visual comparison of trends in gene expression for different genes in the same time course, and also for trends in gene expression from different experiments. (a) Time course of three different genes, plotted as the logR relative to the reference pool of mRNA. (b) Zero transformation of the time courses in (a). (c) Visual representation of trends in gene expression for untransformed data (a) as seen in TreeView. The Color scheme depicts LogR values with gray equalling zero, red > 0, and green < 0. (d) TreeView image of zero-transformed data.

Figure 3

Rank average of two 8-h wild-type time courses. The 50 most strongly induced named genes are shown. Individual time courses were averaged, and transformed to the zero time point measurement (first column: black equals zero). The five time points shown (in min) are 30, 60, 120, 240 and 480. The scale of induction after transformation is shown at the bottom, where the brightest red indicates an induction of at least 12-fold. NFκB-responsive genes are indicated by arrows.

Figure 4

Megacluster of all time-course infections. The data across all time courses was averaged, ranked, and then clustered. All genes that were induced in the mock time course at levels higher than an average log₂(ratio) of 0.5 were omitted. The four parallel time courses are shown on the left, with trends in gene expression appearing very similar for all time courses. Four individual 8-h time courses are on the right, and show the day-to-day variability of the response (which includes dependence on the time zero measurement).

Figure 5

Comparison of gene expression in uninfected Caco-2 cells. Scatterplots were made of log-transformed ratios log₂(T_n/T₀) of (Caco-2 mRNA/reference pool mRNA). As the reference is relatively constant for each gene, scatter in the data represents biological variability in the Caco-2 cells. The scale for each plot is identical, with the axes ranging from log₂(T_n/T₀) -4 to 4. (a) Comparison of uninfected cells harvested several months apart. Correlation is poor (Pearson correlation coefficient 0.74). (b) Comparison of the same uninfected Caco-2 mRNA, hybridized on different days. Correlation is excellent (correlation coefficient 0.97), verifying reproducibility with respect to the reference pool. (c) Comparison of uninfected cells harvested from different plates on the same day. Correlation is good (correlation coefficient 0.94).

Figure 6

TreeView image of named genes induced during parallel time course. Genes were ranked by average log₂(ratio) over all time courses to look at similarity of induction (the mock time course was filtered for genes induced by the infection protocol), and then clustered. Many NFκB-related and responsive genes form a single cluster, representing 40% (20/50) of the most highly induced genes. Outside the node representing these genes, there are a number of other immune-response-related genes, including small inducible cytokines (SCY) which are also regulated by NFκB. Image contrast is set at log₂(ratio) = 3.5, indicating a minimum of 12-fold induction for the brightest red spots. Gray spots indicate missing data.

Figure 7

The NFκB and proinflammatory response. (a) The response of NFκB family members and related kinases that are present on the 24,000-spot (24K) arrays. (b) Cytokines, chemokines and their receptors that are induced during the time courses. Image contrast is set at 2, such that the brightest red is greater than a fourfold change. Gray spots indicate missing data.

Figure 8

Induced cytoskeletal genes and the genes required for actin-based motility. (a) Cytoskeleton-related genes present on 24K arrays which respond to the infection with at least a threefold change. (b) Named genes known to be involved in actin-based motility of L. monocytogenes. Image contrast is set at 2, such that the brightest red is greater than a fourfold change. Gray spots indicate missing data. Note that genes previously known to be involved in actin-based motility are not specifically induced by wild-type bacteria.

Figure 9

ELISA for the presence of secreted Gro-1 chemokine in supernatants of subconfluent and polarized Caco-2 monolayers. (a) Subconfluent monolayers infected with wild-type, actA, and prfA bacteria under conditions used for microarray analysis. (b) ELISA assay for Gro-1 production by polarized Caco-2 cells. Wild-type L. monocytogenes (WT) was used to infect either apically (A-inf) or basolaterally (BL-inf). Supernatants from both apical (A) and basolateral (BL) membranes were analyzed for secreted Gro-1.