As sentinels of host defense, intestinal epithelial cells respond to the viral pathogen rotavirus by activating a gene expression that promotes immune cell recruitment and activation. We hypothesized that epithelial sensing of rotavirus might target dsRNA, which can be detected by TLR3 or protein kinase R (PKR). Accordingly, we observed that synthetic dsRNA, polyinosinic acid:cytidylic acid (poly(I:C)), potently induced gene remodeling in model intestinal epithelia with the specific pattern of expressed genes, including both classic proinflammatory genes (e.g., IL-8), as well as genes that are classically activated in virus-infected cells (e.g., IFN-responsive genes). Poly(I:C)-induced IL-8 was concentration dependent (2–100 μg/ml) and displayed slower kinetics compared with IL-8 induced by bacterial flagellin (ET50 ∼24 vs 8 h poly(I:C) vs flagellin, respectively). Although model epithelia expressed detectable TLR3 mRNA, neither TLR3-neutralizing Abs nor chloroquine, which blocks activation of intracellular TLR3, attenuated epithelial responses to poly(I:C). Conversely, poly(I:C)-induced phosphorylation of PKR and inhibitors of PKR, 2-aminopurine and adenine, ablated poly(I:C)-induced gene expression but had no effect on gene expression induced by flagellin, thus suggesting that intestinal epithelial cell detection of dsRNA relies on PKR. Consistent with poly(I:C) detection by an intracellular molecule such as PKR, we observed that both uptake of and responses to poly(I:C) were polarized to the basolateral side. Lastly, we observed that the pattern of pharmacologic inhibition of responses to poly(I:C) was identical to that seen in response to infection by live rotavirus, indicating a potentially important role for PKR in activating intestinal epithelial gene expression in rotavirus infection.

Rotaviruses are the single most important etiologic agents of severe dehydrating diarrheal disease in young children, accounting for as many as 100 million cases and >440,000 deaths annually (1, 2, 3, 4). Rotavirus has a rather limited tissue tropism, primarily infecting only the epithelial cells lining the villi of the small intestine consistent with gastroenteritis being its major clinical manifestation (5). Rotaviral infection of such epithelial cells induces a substantial induction of epithelial gene expression, including the activation of a panel of chemokines that promote the recruitment and activation of immune cells (6, 7, 8). Although such immune cell recruitment in response to rotavirus occurs on a much smaller scale than that seen in response to bacterial pathogens (e.g., Salmonella, Shigella) (9), the fact that rotavirus infection is localized generally to the gastrointestinal tract suggests that, nonetheless, rotaviral-induced immune cell recruitment may be important for preventing viral spread throughout the host. Rotavirus indeed has the potential to infect extraintestinal sites as rotaviral RNA has been found in cerebral spinal fluid and serum of some rotavirus-infected children, possibly associated antigenemia and viremia (10). In addition to a potential role in impeding viral dissemination, epithelial gene expression in response to rotavirus may also be important for regulating the adaptive immune response analogous to processes thought to occur in response to bacterial colonization of epithelia. Such adaptive responses to rotavirus result in lasting protection against reinfection (11). Thus, in light of its potential importance, we sought to determine the mechanism by which gut epithelia might detect rotavirus and, subsequently, regulate remodeling of gene expression.

While a number of studies have observed rotaviral-induced activation of epithelial chemokines secretion in well-defined model systems and have carefully examined the roles of host transcription factors in regulating these responses (6, 7, 8, 12), little is known in regarding the primary host receptor(s) or other sensing mechanism that initiate these responses. Based on the emerging paradigm that epithelial sensing of bacteria is largely based on a series of intracellular and extracellular pattern recognition receptors (13, 14), we reasoned that epithelia may also have receptors capable of directly recognizing viral products. As rotavirus is a nonenveloped, 11-segmented dsRNA virus (15, 16) and several epithelial cell types (e.g., retinal pigment, lung) have been observed to respond to dsRNA (17, 18, 19), this seemed an especially likely molecule to be promoting activation of intestinal epithelial gene expression in response to rotavirus. Thus, we sought to define the effects of dsRNA on gut epithelial gene expression using a well-defined polarized model system.

Two distinct mechanisms by which mammalian cells can recognize dsRNA have been described. One mechanism is the activation of dsRNA-dependent protein kinase R (PKR)3. PKR isactivated upon binding dsRNA to undergo dimerization and autophosphorylation. This 68-kDa, cytoplasmic serine/threonine kinase phosphorylates its physiological substrate eukaryotic initiation factor 2-α (eIF2-α) and inhibits translation and perhaps other substrates that results in activation of a panel of genes that ultimately leads to cessation of virus replication in cells (20, 21, 22). It has also been shown that PKR regulates other pathways, including those activating p53, p38, IFN regulatory factor-1, and NF-κB (23, 24, 25). Induction of NF-κB has a relevant role in mediating PKR functions, and NF-κB activation by PKR is involved in IFN-β induction in response to dsRNA (26). More recently, it has been shown that TLR3 is also a receptor for dsRNA (27), supported by demonstration that expression of TLR3 confers cells with the ability to respond to polyinosinic acid:cytidylic acid (poly(I:C)) and that TLR3 null mice exhibit substantially reduced responses to poly(I:C). Although there appears to be substantial overlap in some of the genes activated by TLR3 and PKR, these pathways are nonetheless independent in that both PKR-null and TLR3-null mice still gave clear, albeit reduced, responses to poly(I:C) (27, 28). As the relative importance of these receptors in response to different viral pathogens is only beginning to be elucidated (29), we investigated whether either of these pathways were important for intestinal epithelial detection of dsRNA and rotavirus. As we anticipated, similar to other epithelial cells, intestinal epithelia cells displayed robust responses to synthetic dsRNA. However, in contrast to the case for other epithelia, intestinal epithelial responses to dsRNA and rotavirus were mediated by PKR.

Poly(I:C), poly(dI:dC), and poly(C) and anti-rabbit and mouse IgG-HRP were purchased from Amersham Biosciences. Bafilomycin A1 (BFA), chloroquine, 2-aminopurine (2-AP), adenine, mouse anti-human β-actin, and TRIzol reagent were purchased from Sigma-Aldrich. Affinity purified and functional grade mAbs (clone 3.7) to human TLR3 were obtained from eBioscience. Matrix metalloproteinase (MMP)-7 Abs were procured from Chemicon International. Rabbit anti-human STAT-1, STAT-1-phos-Y701 eIF2-α, eIF2-α-phos-S51, PKR, and PKR-phos-T446 were obtained from Cell Signaling Technology. IFN-α, IFN-β, and IFN-γ were obtained from National Institute of Allergy and Infectious Diseases Reference Reagent Laboratory, Braton Biotech. Rabbit anti-human inducible NO synthase (iNOS), TNF-α, and IL-1β were purchased from R&D Systems. Human TLR3 primers (upstream, 5′-GATCTGTCTCATAATGGCTTG-3′, and downstream, 5′-GACAGATTCCGAATGCTTGTG-3′) were obtained from Invitrogen Life Technologies (17). Flagellin was purified from Salmonella typhimurium-conditioned medium by anion/cation exchange chromatography and purity verified as described previously (30). In brief, such flagellin does not activate any TLR other than TLR5 (13) and has <0.5 pg of LPS/μg of flagellin (31). Mouse mAbs to human neutrophil-gelatinase-associated lipocalin (NGAL) was clone 211.1 (32).

Model human intestinal epithelia were prepared by culturing the colonic epithelial cell line T84 (passages 63–72) grown in DMEM:F-12 medium supplemented with 10% FBS and HT-29 (passages 130–144), and Caco-2 and Caco-2 brush border enhanced (passages 30–35) were grown in DMEM supplemented with 10% FBS, 1% nonessential amino acids, l-glutamine (2 mM), and d-glucose (4.5 g/L). Cells were grown in the presence of 100 U/ml penicillin and 100 μg/ml streptomycin in 5% CO2 at 37°C. All the tissue culture reagents were purchased from Invitrogen Life Technologies or Mediatech.

Lyophilized poly(I:C) was a potassium salt having a sedimentation coefficient (S20.W) of 15.4. Based on the S20.W, the approximate m.w. and average nucleotide length will be 3.5 × 106 Da and 4,655 bp, respectively, although analysis of this poly(I:C) by agarose gel electrophoresis indicates that it is substantially larger than our highest DNA standard (10,000 bp). It was dissolved in sterile PBS at 2 mg/ml and heated at 50°C water bath until solubilized and slowly cooled to room temperature for proper annealing. Similarly, poly(C) and poly(dI:dC) were reconstituted as described for poly(I:C). Throughout this study, we used poly(I:C) from the same lot (no. 3074729011).

On the day of stimulation, confluent cells were washed twice with serum-free medium (SFM) and stimulated as described in figure legends. For inhibition studies, TLR3 mAbs (affinity purified and functional grade), BFA, 2-AP, and adenine were added 1 h before addition of the stimuli. After stimulation, supernatants were collected and centrifuged at 15,000 × g at 4°C for 10 min and stored at −80°C for IL-8 analysis.

Confluent T84 cells on permeable support with transepithelial resistance of >1000 Ω cm2 were washed twice with SFM and stimulated with 100 μg/ml poly(I:C) either basolaterally, or apically or basolaterally pretreated (1 h before) with BFA on both sides. Supernatants were collected at different times and centrifuged at 15,000 × g at 4°C for 10 min and subjected to agarose gel (0.9%) electrophoresis.

Rhesus rotavirus (RRV) was cultivated in MA104 cells, then frozen and thawed three times, clarified by low-speed centrifugation, and stored in aliquots at −80°C until used (33). Trypsin-activated RRV was titrated by the fluorescent Ab method (34). Monolayers of T84 cells grown on 24-well plates were infected with RRV activated immediately before use with 10 μg/ml trypsin at 37°C for 30 min at different multiplicities of infection (MOI) ranging from 2.5 to 20. Trypsin was added to convert noninfectious virus into infectious virus via cleavage of virus spike protein VP4 into VP5 and VP8 (35). Before infection, cells were washed three times with SFM, and the desired amount of RRV, diluted in 1 ml of SFM, was added to the cells. Mock-infected cells (control) were treated with an equivalent amount of trypsin-treated SFM. Cells were incubated for 1 h at 37°C in 5% CO2, followed by removal of the virus containing medium. Cells were washed three times with SFM, and the infection was continued for the indicated times in a 37°C incubator with 5% CO2. Viral infection was monitored and documented by viewing under a light microscope. At designated times, supernatants were removed and stored at −80°C.

Model intestinal epithelial cells (T84) were prepared on 5-cm2 permeable filters and were used 8 days after plating and achieving a stable transepithelial resistance of >1000 Ω cm2. Cells were washed with SFM and stimulated with 100 μg/ml poly(I:C) on both sides or 100 ng/ml flagellin applied basolaterally (1 ml apically and 2 ml basolaterally). After 2 and 48 h of stimulation, supernatants were collected for IL-8 assay, and RNA was isolated and subjected to microarray analysis as described previously (36). Briefly, RNA was isolated from treated and untreated T84 cells via TRIzol, subjected to DNase I digestion, and purified by using commercially available kit from Qiagen. Total cellular RNA (30 μg) was used to synthesize cDNA labeled with fluorlink Cy5 dCTP. Universal human reference RNA (20 μg) was used to synthesize reference cDNA labeled with fluorlink Cy3 dCTP. Sample and reference cDNA were cohybridized on to the RG Human 11K gene chip purchased from the Vanderbilt MicroArray Shared Resources (Vanderbilt University). Washed and dried chips were scanned using the GenePix 4100A scanner. Images were captured and analyzed using the GenePix Pro 5.0 software. Ratio-based normalization was performed so that ratio of medians in the Cy5 and Cy3 channels equals one. Ratio of Cy5:Cy3 of untreated and treated samples was calculated to assess relative changes in gene expression.

Confluent monolayers of T84 cells in six-well plates were washed with SFM and treated with 100 μg/ml poly(I:C). Total RNA was prepared using the TRIzol reagent. RT-PCR analysis of TLR3 gene expression was conducted using 0.5 μg of total RNA/reaction (30 cycles) using Applied Biosystems kit (Applied Biosystems). TLR3 primers yielded a product of 304 bp (17). Ten microliters of the PCR product were separated on a 1.5% agarose gel electrophoresis and viewed under UV light. For quantitative real-time PCR, the following primers were used: PKR forward, 5′-TGGCGGTCTTCAGAATCAACATC-3′, and reverse, 5′-CAGCCATTTCTTCTTCCCGTATCC-3′; and GAPDH forward, 5′-ACCCAGAAGACTGTGGATCG-3′, and reverse 5′-GGATGCAGGGATGATGTTCT-3′. Reactions were performed in triplicate and values for PKR normalized to GAPDH RNA using SYBR Green Super Mix purchased from Bio-Rad (36).

T84 cells grown in six-well plates were stimulated as described in figure legends, rinsed in ice-cold HBSS, lysed in radioimmunoprecipitation assay II buffer (20 mM Tris-HCl, 2.5 mM EDTA, 1% Triton X-100, 10% glycerol, 1% deoxycholate, 0.1% SDS, 50 mM NaF, 10 mM Na2P2O7, and 2 mM NaVO4 plus protease inhibitor mixture) at a concentration of 107 cells/ml cleared by centrifugation (10 min at 15,000 × g at 4°C), and equal amounts of protein assayed for iNOS, total STAT-1, PKR, eIF2-α, and phospho-STAT-1, PKR, and eIF2-α by 12% SDS-PAGE immunoblotting. Membranes were stripped and probed for β -actin (control). T84 cell supernatants were used for SDS-PAGE immunoblotting NGAL and MMP-7 using 4–20% gels. Briefly, cells were stimulated as in figure legends, and supernatants were collected, centrifuged at (10 min at 15,000 × g at 4°C), and used for SDS-PAGE immunoblotting. NGAL SDS-PAGE immunoblotting was conducted in both reducing (with 2-ME) and nonreducing (without 2-ME) conditions to distinguish homo- and heterodimers of NGAL. The immunoblots were visualized with the ECL system (Amersham Biosciences).

Human IL-8 was measured as described previously (37). Human IFN-β enzyme immunoassay kit (sensitivity 250–10,000 pg/ml) from R&D Systems and used for the quantification of IFN-β in the cell-free supernatants, according to manufacturer’s instructions.

Nitrite level was quantified in the supernatants using Griess Reagent System (sensitivity 1.5–100 μM) (Promega), according to the manufacturer’s instructions.

Although intestinal viruses such as rotavirus have long been appreciated as activators of intestinal epithelial cell proinflammatory gene expression, little is known regarding either the viral or host molecules that mediate this response. Because rotavirus is a dsRNA virus and dsRNA activates cytokine expression in retinal pigment and lung epithelia (17, 18, 19, 38), dsRNA seemed a strong candidate to be a viral activator of gut epithelial cells. To investigate this notion, we stimulated T84 intestinal epithelial cells with a range of doses of synthetic dsRNA (poly(I:C) 2–100 μg/ml) and measured changes in epithelial gene expression. First, we measured secretion of IL-8 because this gene is known to be induced by rotavirus (6, 7, 8) and is commonly used as a general indicator of epithelial proinflammatory gene expression. As shown in Fig. 1,A, poly(I:C) induced IL-8 secretion in a concentration-dependent manner with 100 μg/ml (a common dose used in other studies) eliciting similar levels of IL-8 secretion as a maximal (100 ng/ml) concentration of flagellin. Poly(C) and poly(dI:dC) did not induce IL-8 secretion, indicating that the cells were responding to dsRNA rather than DNA or simply a bolus of nucleotides. However, as shown in Fig. 1 B, poly(I:C) was a relatively slow-acting inducer of IL-8 expression compared with flagellin. Specifically, poly(I:C)-induced IL-8 secretion was not detectable until 12 h (vs 3 h for flagellin) and with half-maximal induction requiring over 24 h (vs 8 h for flagellin).

FIGURE 1.

Poly(I:C) induces IL-8 in intestinal epithelial cells. A, Confluent monolayers of T84 cells were stimulated with increasing doses of poly(I:C) (100 μg/ml) or poly C (100 μg/ml) or poly(dI:dC) (100 μg/ml) or flagellin (Flag) (100 ng/ml) in SFM for 48 h. B, T84 cells were stimulated with either poly(I:C) (100 μg/ml) or Flag (100 ng/ml). Supernatants were taken at various time points and assayed for IL-8 by ELISA. Values are mean ± SD obtained from duplicate samples of three representative experiments.

FIGURE 1.

Poly(I:C) induces IL-8 in intestinal epithelial cells. A, Confluent monolayers of T84 cells were stimulated with increasing doses of poly(I:C) (100 μg/ml) or poly C (100 μg/ml) or poly(dI:dC) (100 μg/ml) or flagellin (Flag) (100 ng/ml) in SFM for 48 h. B, T84 cells were stimulated with either poly(I:C) (100 μg/ml) or Flag (100 ng/ml). Supernatants were taken at various time points and assayed for IL-8 by ELISA. Values are mean ± SD obtained from duplicate samples of three representative experiments.

Close modal

We next used cDNA microarray analysis to globally assess the changes in epithelial gene expression induced by synthetic dsRNA again comparing responses to those induced by flagellin, the best characterized microbial activator of gut epithelial cells. Specifically, model epithelia were treated with buffer only (control) or the above-defined maximal doses of poly(I:C) or flagellin. RNA was isolated 2 and 48 h later to assess short-term and long-term effects on gene expression. RNA was subjected to cDNA microarray analysis using RG Human 11K gene chips. Table I lists the 30 genes most up-regulated in response to poly(I:C) in comparison to control. Consistent with its slow induction of IL-8, poly(I:C) did not induce substantial changes at 2 h but at 48 h potently up-regulated a panel of genes known to play roles in host defense in general (e.g., lipocalin-2/NGAL, hereafter referred to as NGAL and MMP-7) and in particular a number genes associated with antiviral/IFN responses. Although a couple of these general host-defense genes (e.g., NGAL) were also moderately induced by flagellin, the antiviral associated genes were not induced substantially by flagellin at either time point. Conversely, consistent with its fast action, flagellin potently up-regulated a panel of genes at 2 h, whose expression had returned toward baseline by 48 h (Table II). Most of the genes exhibiting the highest levels of activation in response to flagellin were not induced by poly(I:C) at either time point. Thus, while flagellin and poly(I:C) are both potent activators of epithelial gene expression in general, the specific patterns of genes activated by these agonists are very different, suggesting intestinal epithelial cells can appropriately tailor their responses to specific classes of microbes.

Table I.

Poly(I:C) (100 μg/ml) up-regulated genes in T84 cellsa

Accession No.Gene Name2 h/Control48 h/Control
1. N33920 Ubiquitin D 1.00 (ND)b 12.73 (ND) 
2. AA400973 Lipocalin 2 (oncogene 24p3) 1.08 (1.22) 11.61 (3.12) 
3. NM_002423 Matrix metalloproteinase 7 (matrilysin, uterine) 4.11 (1.24) 11.46 (1.27) 
4. R95691 Unknown 0.89 6.50 
5. AA457042 IFN-inducible protein p78 3.04 (0.66) 6.43 (0.98) 
6. AA410188 IFN-induced protein 1.58 (1.03) 6.40 (1.01) 
7. N55205 β-globin pseudogene 1.00 (1.48)b 5.86 (0.12) 
8. AA401441 B-factor, properdin 1.17 (1.12) 5.70 (1.68) 
9. AA458965 NK cell transcript 4 1.18 (1.10) 5.40 (0.72) 
10. AA286908 Myxovirus (influenza virus) resistance 2 (mouse) 1.00 (0.93)b 5.34 (1.06) 
11. AA464246 Major histocompatibility complex, class I, C 0.90 (0.81) 5.13 (1.04) 
12. H48533 Apoptosis inhibitor 2 3.68 (7.10) 4.85 (2.05) 
13. AA448478 IFN, α-inducible protein (clone IFI-6–16) 1.00 (0.79) 4.79 (1.18) 
14. AA292074 Ubiquitin-conjugating enzyme E2L 6 1.02 (0.70) 4.54 (1.09) 
15. AA489640 IFN-induced protein 56 1.00 (1.19)b 4.49 (1.00) 
16. AA644657 Major histocompatibility complex, class I, A 0.90 (0.80) 4.44 (1.04) 
17. AA443090 Interferon regulatory factor 7 0.61 (0.75) 4.03 (1.04) 
18. AA677534 Laminin, γ2 0.77 (2.19) 4.01 (1.00) 
19. AA459401 Kallikrein 10 0.58 (1.49) 3.97 (0.72) 
20. AA487921 KIAA0152 gene product 1.36 (0.68) 3.89 (1.57) 
21. AA476272 TNF, α-induced protein 3 1.00 (1.65)b 3.66 (1.10) 
22. H69561 Mannosidase, α class 2A, member 1 1.70 (16.35) 3.54 (2.24) 
23. T64469 P8 protein (candidate of metastasis 1) 1.00 (0.83)b 3.50 (1.01) 
24. AA479882 Keratin 10 2.37 (1.58) 3.40 (0.85) 
25. AA058323 IFN-induced transmembrane protein 1 1.45 (1.66) 3.34 (0.33) 
26. R44417 Apoptosis inhibitor 1 0.87 (3.05) 3.34 (1.22) 
27. AA487893 Transmembrane 4 superfamily member 1 1.19 (3.41) 3.22 (0.48) 
28. NM_013671 Mn superoxide dismutase 1.07 (2.56) 3.22 (1.08) 
29. AA406020 IFN, α-inducible protein 0.85 (0.91) 3.09 (0.93) 
30. AA464250 Keratin 19 1.07 (0.82) 3.08 (0.64) 
Accession No.Gene Name2 h/Control48 h/Control
1. N33920 Ubiquitin D 1.00 (ND)b 12.73 (ND) 
2. AA400973 Lipocalin 2 (oncogene 24p3) 1.08 (1.22) 11.61 (3.12) 
3. NM_002423 Matrix metalloproteinase 7 (matrilysin, uterine) 4.11 (1.24) 11.46 (1.27) 
4. R95691 Unknown 0.89 6.50 
5. AA457042 IFN-inducible protein p78 3.04 (0.66) 6.43 (0.98) 
6. AA410188 IFN-induced protein 1.58 (1.03) 6.40 (1.01) 
7. N55205 β-globin pseudogene 1.00 (1.48)b 5.86 (0.12) 
8. AA401441 B-factor, properdin 1.17 (1.12) 5.70 (1.68) 
9. AA458965 NK cell transcript 4 1.18 (1.10) 5.40 (0.72) 
10. AA286908 Myxovirus (influenza virus) resistance 2 (mouse) 1.00 (0.93)b 5.34 (1.06) 
11. AA464246 Major histocompatibility complex, class I, C 0.90 (0.81) 5.13 (1.04) 
12. H48533 Apoptosis inhibitor 2 3.68 (7.10) 4.85 (2.05) 
13. AA448478 IFN, α-inducible protein (clone IFI-6–16) 1.00 (0.79) 4.79 (1.18) 
14. AA292074 Ubiquitin-conjugating enzyme E2L 6 1.02 (0.70) 4.54 (1.09) 
15. AA489640 IFN-induced protein 56 1.00 (1.19)b 4.49 (1.00) 
16. AA644657 Major histocompatibility complex, class I, A 0.90 (0.80) 4.44 (1.04) 
17. AA443090 Interferon regulatory factor 7 0.61 (0.75) 4.03 (1.04) 
18. AA677534 Laminin, γ2 0.77 (2.19) 4.01 (1.00) 
19. AA459401 Kallikrein 10 0.58 (1.49) 3.97 (0.72) 
20. AA487921 KIAA0152 gene product 1.36 (0.68) 3.89 (1.57) 
21. AA476272 TNF, α-induced protein 3 1.00 (1.65)b 3.66 (1.10) 
22. H69561 Mannosidase, α class 2A, member 1 1.70 (16.35) 3.54 (2.24) 
23. T64469 P8 protein (candidate of metastasis 1) 1.00 (0.83)b 3.50 (1.01) 
24. AA479882 Keratin 10 2.37 (1.58) 3.40 (0.85) 
25. AA058323 IFN-induced transmembrane protein 1 1.45 (1.66) 3.34 (0.33) 
26. R44417 Apoptosis inhibitor 1 0.87 (3.05) 3.34 (1.22) 
27. AA487893 Transmembrane 4 superfamily member 1 1.19 (3.41) 3.22 (0.48) 
28. NM_013671 Mn superoxide dismutase 1.07 (2.56) 3.22 (1.08) 
29. AA406020 IFN, α-inducible protein 0.85 (0.91) 3.09 (0.93) 
30. AA464250 Keratin 19 1.07 (0.82) 3.08 (0.64) 
a

Values are ratios of expression from poly(I:C)-treated to untreated (control) state. The numbers in parentheses corresponds to induction by flagellin; ND, not detected. Results are representative of two independent experiments. Genes shown in bold are associated with IFN/antiviral responses.

b

Gene expression was basally undetectable; thus, data are given as fold induction above values of minimum reliable detection.

Table II.

Flagellin (100 ng/ml) up-regulated genes in T84 cellsa

Accession No.Gene Name2 h/Control48 h/Control
1. AA043551 UDP-GlcNAc:βGal β-1,3-N-acetylglucosaminyltransferase 5 18.52 (1.16) 1.31 (1.12) 
2. H69561 Mannosidaseclass 2A, member 1 16.34 (1.70) 2.23 (3.53) 
3. W56300 IκBα 10.40 (2.16) 2.02 (2.53) 
4. W46900 Gro-α 7.84 (1.33) 1.63 (1.00) 
5. AA002126 Apoptosis inhibitor 2 7.1 (3.68) 2.05 (4.85) 
6. H69683 Chromosome 13 open reading frame 18 5.45 (0.81)b 1.00 (1.00) 
7. AA148737 Syndecan 4 5.39 (1.39)b 1.00 (2.34) 
8. AA449440 IFNreceptor 2 4.97 (0.92) 0.92 (0.53) 
9. R44417 Apoptosis inhibitor 1 4.62 (0.87) 1.16 (3.33) 
10. AA490466 Gap junction protein β2 26kDa 4.62 (0.67) 1.21 (1.51) 
11. T61649 SOD 2, mitochondrial 4.57 (ND) 1.69 (ND) 
12. AA443688 GTP cyclohydrolase 1 (dopa-responsive dystonia) 4.21 (0.65)b 1.00 (0.98) 
13. N32768 Pregnancy specific β-1-glycoprotein 3 4.13 (ND) 0.81 (ND) 
14. AA436152 Semaphorin 5A 4.05 (ND) 0.89 (ND) 
15. T94279 Fibrinogen, γ polypeptide 3.78 (0.44) 0.75 (0.49) 
16. R70685 Jagged - 1 (Alagille syndrome) 3.75 (0.34) 0.79 (0.88) 
17. R64600 Oxysterol binding protein-like 3 3.73 (1.00) 0.99 (0.85) 
18. AA495790 Ras homologue gene family, member B 3.72 (0.88) 0.29 (1.51) 
19. R67336 Homo sapiens LOC158525 (LOC158525), mRNA 3.71 (ND)b 1.00 (ND) 
20. AA017383 EBNA-2 coactivator (100-KD) 3.67 (1.86) 1.06 (1.85) 
21. AA458884 S100 calcium binding protein A2 3.66 (ND)b 1.00 (ND) 
22. AA457705 Immediate early response 3 3.64 (1.64) 0.80 (2.97) 
23. AA284669 Plasminogen activator, urokinase 3.51 (1.26) 0.70 (1.26) 
24. AA487893 Transmembrane 4-super family member 1 3.41 (ND) 0.48 (ND) 
25. N92502 Unknown 3.41 (1.98) 0.84 (0.56) 
26. T72596 Homo sapiens cDNA clone IMAGE:5288160, partial cds 3.41 (ND) 1.15 (ND) 
27. T95748 Pregnancy specific β-1-glycoprotein 1 3.34 (0.86) 1.01 (1.34) 
28. T89996 Fos-related antigen 1 3.21 (1.41) 1.15 (2.02) 
29. N33214 Matrix metalloproteinase 14 (membrane-inserted) 3.19 (1.28)b 1.00 (1.00) 
30. W42723 Chemokine (C-X-C motif) ligand 1 3.19 (ND)b 1.00 (ND) 
Accession No.Gene Name2 h/Control48 h/Control
1. AA043551 UDP-GlcNAc:βGal β-1,3-N-acetylglucosaminyltransferase 5 18.52 (1.16) 1.31 (1.12) 
2. H69561 Mannosidaseclass 2A, member 1 16.34 (1.70) 2.23 (3.53) 
3. W56300 IκBα 10.40 (2.16) 2.02 (2.53) 
4. W46900 Gro-α 7.84 (1.33) 1.63 (1.00) 
5. AA002126 Apoptosis inhibitor 2 7.1 (3.68) 2.05 (4.85) 
6. H69683 Chromosome 13 open reading frame 18 5.45 (0.81)b 1.00 (1.00) 
7. AA148737 Syndecan 4 5.39 (1.39)b 1.00 (2.34) 
8. AA449440 IFNreceptor 2 4.97 (0.92) 0.92 (0.53) 
9. R44417 Apoptosis inhibitor 1 4.62 (0.87) 1.16 (3.33) 
10. AA490466 Gap junction protein β2 26kDa 4.62 (0.67) 1.21 (1.51) 
11. T61649 SOD 2, mitochondrial 4.57 (ND) 1.69 (ND) 
12. AA443688 GTP cyclohydrolase 1 (dopa-responsive dystonia) 4.21 (0.65)b 1.00 (0.98) 
13. N32768 Pregnancy specific β-1-glycoprotein 3 4.13 (ND) 0.81 (ND) 
14. AA436152 Semaphorin 5A 4.05 (ND) 0.89 (ND) 
15. T94279 Fibrinogen, γ polypeptide 3.78 (0.44) 0.75 (0.49) 
16. R70685 Jagged - 1 (Alagille syndrome) 3.75 (0.34) 0.79 (0.88) 
17. R64600 Oxysterol binding protein-like 3 3.73 (1.00) 0.99 (0.85) 
18. AA495790 Ras homologue gene family, member B 3.72 (0.88) 0.29 (1.51) 
19. R67336 Homo sapiens LOC158525 (LOC158525), mRNA 3.71 (ND)b 1.00 (ND) 
20. AA017383 EBNA-2 coactivator (100-KD) 3.67 (1.86) 1.06 (1.85) 
21. AA458884 S100 calcium binding protein A2 3.66 (ND)b 1.00 (ND) 
22. AA457705 Immediate early response 3 3.64 (1.64) 0.80 (2.97) 
23. AA284669 Plasminogen activator, urokinase 3.51 (1.26) 0.70 (1.26) 
24. AA487893 Transmembrane 4-super family member 1 3.41 (ND) 0.48 (ND) 
25. N92502 Unknown 3.41 (1.98) 0.84 (0.56) 
26. T72596 Homo sapiens cDNA clone IMAGE:5288160, partial cds 3.41 (ND) 1.15 (ND) 
27. T95748 Pregnancy specific β-1-glycoprotein 1 3.34 (0.86) 1.01 (1.34) 
28. T89996 Fos-related antigen 1 3.21 (1.41) 1.15 (2.02) 
29. N33214 Matrix metalloproteinase 14 (membrane-inserted) 3.19 (1.28)b 1.00 (1.00) 
30. W42723 Chemokine (C-X-C motif) ligand 1 3.19 (ND)b 1.00 (ND) 
a

Values are ratios of expression from flagellin-treated to untreated (control) state. The numbers in parentheses correspond to induction by poly(I:C), ND, not detected. Results are representative of two independent experiments.

b

Gene expression was basally undetectable; thus, data are given as fold induction above values of minimum reliable detection.

We next sought to determine whether changes measured in mRNA would in fact be manifested at the level of protein expression. In consistent with the microarray data, poly(I:C) induced relatively slow changes in expression of both NGAL and MMP-7 (Fig. 2,A, i–iii). Such NGAL induction contained only 25-kDa monomer rather than the homo- (46 kDa) or heterodimers (135 kDa) observed in human neutrophils by nonreducing SDS-PAGE immunoblotting as described in Ref. 32 . In light of poly(I:C)-activating genes associated with both NF-κB and IFN regulatory factor pathways, e.g., IL-8 and MHC class I, respectively, we also measured whether poly(I:C) might regulate expression of iNOS because this gene is known to play a central role in host defense and a variety of immunopathologic events (39). Poly(I:C) induced expression iNOS at a level sufficient to be detected by both immunoblot and by a functional assay (Fig. 2, Aiv and B). Although the concentration dependence of iNOS induction was similar to that of IL-8 and NGAL, activation of iNOS expression was somewhat faster appearing to be maximal by 12 h and almost returned to baseline by 24 h (Fig. 2 Aiv). Lastly, we measured whether poly(I:C) had similar effects on model epithelia made from other intestinal epithelial cells lines. Although some cell lines tested (Caco-2 and its more differentiated subclone Caco-2 brush border enhanced) did not respond to poly(I:C) (data not shown), such as T84 cells, HT29 cells made copious amounts of IL-8 and NGAL with similar kinetics and concentration dependence in response to poly(I:C), indicating the response to this dsRNA analogue was not a cell line-specific phenomenon (data not shown).

FIGURE 2.

Poly(I:C) up-regulates NGAL, MMP-7, and iNOS expression. Confluent monolayers of T84 cells were stimulated with NGAL protein expression (Ai) at various doses of 10–100 μg/ml poly(I:C) (Aii). Supernatants from wells treated with 100 μg/ml poly(I:C) collected at different times and subjected to SDS-PAGE immunoblotting. NGAL (Aii), MMP-7 (Aiii), and T84 (Aiv) cell lysates were prepared from six-well plates treated with 100 μg/ml poly(I:C) and subjected to SDS-PAGE immunoblotting for iNOS. B, Nitrite levels were measured in supernatants from cells stimulated with 100 μg/ml poly(I:C) at different time points (left panel) and at various doses of poly(I:C) (right panel). The figure shown is a representative of two experiments.

FIGURE 2.

Poly(I:C) up-regulates NGAL, MMP-7, and iNOS expression. Confluent monolayers of T84 cells were stimulated with NGAL protein expression (Ai) at various doses of 10–100 μg/ml poly(I:C) (Aii). Supernatants from wells treated with 100 μg/ml poly(I:C) collected at different times and subjected to SDS-PAGE immunoblotting. NGAL (Aii), MMP-7 (Aiii), and T84 (Aiv) cell lysates were prepared from six-well plates treated with 100 μg/ml poly(I:C) and subjected to SDS-PAGE immunoblotting for iNOS. B, Nitrite levels were measured in supernatants from cells stimulated with 100 μg/ml poly(I:C) at different time points (left panel) and at various doses of poly(I:C) (right panel). The figure shown is a representative of two experiments.

Close modal

In light of the relatively slow time course of poly(I:C)-induced gene expression and the observation that flagellin was activating some genes thought to be regulated by IFN/STAT pathways, we next investigated the involvement of these molecules in epithelial responses to poly(I:C). Poly(I:C) induced a modest (2-fold) increase in expression of total STAT-1 (Fig. 3,A) with a time course of response paralleling the above-described changes in induced gene expression. Preceding increases in total STAT-1 were a concomitantly slow activation of STAT-1 phosphorylation of tyrosine 701. Specifically, while such poly(I:C)-induced activation of STAT-1 required at least 5 h, IFNs α, β, and γ activated STAT-1 within 15 min of stimulation (Fig. 3,B). Similar kinetics was observed for total STAT-3 and its tyrosine phosphorylation (data not shown). This time course of poly(I:C)-induced STAT-1 activation suggested poly(I:C) might induce the synthesis of IFN or other proteins that might mediate STAT activation as occurs in LPS-treated macrophages or in epithelial cells in response to flagellin (40, 41). However, sensitive ELISA kits failed to detect IFN-β expression, and moreover, global blockade of protein synthesis indicated that poly(I:C)-induced STAT activation did not require new protein synthesis. Specifically, we observed that treating T84 cells with cycloheximide under conditions that block protein synthesis and subsequently STAT activation in response to flagellin (Ref. 41 and data not shown) had no effect on poly(I:C)-induced STAT-1 tyrosine phosphorylation (Fig. 3 C), indicating that poly(I:C) activation of STAT signaling does not require new protein synthesis.

FIGURE 3.

Poly(I:C) up-regulates STAT-1 and induces tyrosine phosphorylation. Epithelial cells were stimulated with 100 μg/ml poly(I:C) or 200 IU/ml IFN-α, IFN-β, or IFN-γ as indicated times. Cells were lysed and assayed for total and phosphorylated STAT-1 by SDS-PAGE immunoblotting. Total STAT-1 (A), STAT-1 tyrosine phosphorylation (B), and STAT-1 tyrosine phosphorylation (C) in the presence of 10 μg/ml cycloheximide (CHX).

FIGURE 3.

Poly(I:C) up-regulates STAT-1 and induces tyrosine phosphorylation. Epithelial cells were stimulated with 100 μg/ml poly(I:C) or 200 IU/ml IFN-α, IFN-β, or IFN-γ as indicated times. Cells were lysed and assayed for total and phosphorylated STAT-1 by SDS-PAGE immunoblotting. Total STAT-1 (A), STAT-1 tyrosine phosphorylation (B), and STAT-1 tyrosine phosphorylation (C) in the presence of 10 μg/ml cycloheximide (CHX).

Close modal

Epithelial responses to bacterial products and cytokines are often highly polarized, consistent with the role of these cells as an interface between lumenal microbes and immune cells (13, 30). Defining the polarity of such responses can provide insights into both the mechanism of response and its physiological role. While unlike microbial surface component products, such as flagellin, dsRNA would not be expected to be readily available for detection, “unpackaged” dsRNA can be envisioned to be released upon host cell lysis. Epithelial responses were indeed polarized to poly(I:C) in that epithelia exhibited greater secretion of both IL-8 and NGAL in response to basolateral than apical application of poly(I:C) (Fig. 4, A and B). Such greater responses to basolateral poly(I:C) could be observed whether IL-8 or NGAL were measured in the apical or basolateral reservoir. Epithelia exhibiting greater responses to basolateral poly(I:C) could result from either preferential basolateral expression of a poly(I:C) receptor, analogous to the basolaterally polarized expression of TLR5 that mediates the polarity of response to flagellin (13) or, alternatively, might result from a basolaterally preferential uptake of poly(I:C) and subsequent delivery to an intracellular receptor. Thus, the role for internalization of poly(I:C) was investigated via BFA, which inhibits vacuolar H+-ATPases and subsequently blocks endocytosis (35, 42, 43). Although BFA alone induced a modest level of IL-8 secretion, it nonetheless completely blocked subsequent IL-8 expression induced by poly(I:C). In contrast, BFA had no effect on flagellin-induced IL-8 expression (IL-8 secretion by flagellin and BFA was additive) (Fig. 4,C). Similarly, BFA ablated poly(I:C)-induced expression of NGAL (Fig. 4 D). These results suggests that, in contrast to flagellin, which signals through a surface receptor, poly(I:C)-induced activation of epithelial gene expression may require internalization of this viral mimetic.

FIGURE 4.

Enhanced basolateral response to poly(I:C). Polarized epithelial cells cultured on 5-cm2 filters were stimulated with 100 μg/ml poly(I:C) either basolaterally or apically. After 48 h, basolateral and apical supernatants were collected separately and assayed for IL-8 secretion (A) and NGAL secretion (B). Epithelial cells grown on plastic were stimulated with 20 μg/ml poly(I:C) or 10 ng/ml flagellin with or without 100 nM BFA for 48 h. Supernatants were assayed for IL-8 (C) and NGAL (D). Data in this figure are mean ± SD obtained from duplicate samples of three representative experiments.

FIGURE 4.

Enhanced basolateral response to poly(I:C). Polarized epithelial cells cultured on 5-cm2 filters were stimulated with 100 μg/ml poly(I:C) either basolaterally or apically. After 48 h, basolateral and apical supernatants were collected separately and assayed for IL-8 secretion (A) and NGAL secretion (B). Epithelial cells grown on plastic were stimulated with 20 μg/ml poly(I:C) or 10 ng/ml flagellin with or without 100 nM BFA for 48 h. Supernatants were assayed for IL-8 (C) and NGAL (D). Data in this figure are mean ± SD obtained from duplicate samples of three representative experiments.

Close modal

Although approaches to directly measure uptake of poly(I:C) have not yet been successfully developed, we examined levels of poly(I:C) in epithelial supernatants to allow an indirect assessment of poly(I:C) uptake by epithelia. The commercially available poly(I:C) we used is a large polymer whose m.w. is not well defined. We observed that it does not consistently penetrate agarose gels (even following ethanol precipitation to desalt) and thus appears on the edges of the wells where it was loaded (Fig. 5). Following 12 h of incubation with the basolateral surface of epithelia, this high-m.w. poly(I:C) completely disappeared from the epithelial supernatant, indicating that the poly(I:C) had been taken up by the cells and/or degraded it to a lower m.w. species (Fig. 5,A). Indeed, the disappearance of the high-m.w. poly(I:C) corresponded with appearance of a smear of lower m.w. nucleic acids, suggesting degradation had occurred. This same pattern of events was exhibited by poly(I:C) added to the apical reservoir but occurred with considerably less efficiency (i.e., slower), still being incomplete by 48 h (Fig. 5,B). As both the disappearance of the high-m.w. species and appearance of the low-m.w. species were both potently blocked by BFA (Fig. 5 C), it is likely that such disappearance of the high-m.w. poly(I:C) and appearance of low-m.w. nucleic acids reflect uptake of the poly(I:C) by epithelial cells, possibly followed by release of intracellularly degraded poly(I:C). Accordingly, the much slower disappearance of apical than basolateral intact poly(I:C) likely represents a less efficient uptake of poly(I:C) from this surface, thus suggesting that the polarity of response to poly(I:C) (as assessed by secretion of IL-8 and NGAL) may be underlied by the polarity of uptake of this viral mimetic. In contrast, flagellin (100 ng/ml) applied basolaterally remained intact even after 48 h (assessed by immunoblotting; data not shown), consistent with it acting on surface receptors. While the more rapid degradation of basolateral poly(I:C) may yet possibly occur via a BFA-sensitive mechanism that does not directly involve poly(I:C) internalization, these results nonetheless demonstrate that a greater response to basolateral poly(I:C) does not necessarily indicate the presence of a specific receptor.

FIGURE 5.

Enhanced basolateral uptake of poly(I:C). Polarized epithelial cells grown on 5-cm2 filters were stimulated with 100 μg/ml poly(I:C) either basolaterally (A) or apically (B) or basolaterally (C) in the presence of BFA (100 nM) on both sides. Supernatants were taken at various times and separated on 0.9% agarose gel containing ethidium bromide and viewed under UV. Data are from representative of two independent experiments.

FIGURE 5.

Enhanced basolateral uptake of poly(I:C). Polarized epithelial cells grown on 5-cm2 filters were stimulated with 100 μg/ml poly(I:C) either basolaterally (A) or apically (B) or basolaterally (C) in the presence of BFA (100 nM) on both sides. Supernatants were taken at various times and separated on 0.9% agarose gel containing ethidium bromide and viewed under UV. Data are from representative of two independent experiments.

Close modal

We next investigated whether epithelial recognition of poly(I:C) involved either TLR3 or PKR because both of these molecules are capable of independently (of each other) mediating responses to dsRNA. Microarray experiments did not observe detectable levels of TLR3 mRNA (with or without poly(I:C) treatment) in model epithelia, suggesting that TLR3 may not be abundantly expressed by these cells. However, there is likely at least some expression of TLR3 by these cells as using the more sensitive assay of RT-PCR we did observe detectable levels of TLR3 transcript (inset of Fig. 6,A). Such detectable expression of TLR3 is consistent with the possibility that TLR3 might mediate poly(I:C) detection in these cells. Thus, a potential role for TLR3 in mediating intestinal epithelial responses to poly(I:C) was investigated via a commercially available mAb to TLR3 that has been demonstrated to inhibit (by 60%) surface-expressed, TLR3-mediated signaling (44) but not intracellularly expressed TLR3 signaling (45). This Ab did not inhibit poly(I:C)-induced IL-8 secretion from model epithelia (Fig. 6,A). Because such an Ab might not be able to inhibit intracellular TLR3, we measured the effect of chloroquine on poly(I:C)-induced responses because this compound is known to inhibit endosomal acidification and block signaling of TLR expressed intracellularly (45, 46, 47). Chloroquine had no effect on poly(I:C)-induced epithelial IL-8 production, suggesting intracellular TLR are not required for generating this response (Fig. 6 B).

FIGURE 6.

TLR3-independent IL-8 secretion. Confluent monolayers of epithelial cells were stimulated with 20 μg/ml poly(I:C) in the absence or presence of 20 μg/ml TLR3 Ab (A) or 50 μM chloroquine (B) as described in Materials and Methods. Supernatants assayed for IL-8 at indicated time periods. (Inset: A, Constitutive expression of TLR3 transcript.) Data in A and B are mean ± SD obtained from duplicate samples of three representative experiments.

FIGURE 6.

TLR3-independent IL-8 secretion. Confluent monolayers of epithelial cells were stimulated with 20 μg/ml poly(I:C) in the absence or presence of 20 μg/ml TLR3 Ab (A) or 50 μM chloroquine (B) as described in Materials and Methods. Supernatants assayed for IL-8 at indicated time periods. (Inset: A, Constitutive expression of TLR3 transcript.) Data in A and B are mean ± SD obtained from duplicate samples of three representative experiments.

Close modal

We next investigated the possibility that the other known detector of dsRNA, PKR, might mediate intestinal epithelial responses to poly(I:C). Our microarray experiments indicated that PKR is expressed and up-regulated 2-fold in response to (48 h) poly(I:C). In accordance with this result, quantitative real-time PCR indicated that PKR mRNA levels were up-regulated 3.5-fold (± 0.5). Furthermore, consistent with such changes in PKR mRNA levels, we observed easily detectable expression of PKR by immunoblotting (Fig. 7). Use of a phospho-specific and total PKR Abs further revealed an increase in levels of both phospho-PKR and total PKR within 18 h of poly(I:C) treatment and persisting for at least 48 h (Fig. 7). As eIF2-α is known to serve as a downstream effector of PKR, we also measured phospho- and total eIF2-α. While basal levels of phospho-eIF2-α were easily detectable, we nonetheless observed a consistent, albeit modest, increase in levels of this phosphoprotein, consistent with the activation of PKR, while levels of total eIF2-α did not appear to change in response to poly(I:C) (Fig. 7). We next examined the requirement for PKR in mediating responses to poly(I:C) using two well-studied inhibitors of this kinase, namely 2-AP and adenine (48, 49). Both 2-AP and adenine potently inhibited poly(I:C)-induced IL-8 secretion in a concentration-dependent manner (Fig. 8, A and B). Neither compound had a significant effect on IL-8 secretion induced by flagellin, indicating these PKR inhibitors were not acting nonspecifically. A similar result was obtained using NGAL as a readout further suggesting the importance of PKR for generating epithelial responses to dsRNA (Fig. 8 C). Taken together, these results indicate that PKR is activated by poly(I:C) and is required for activation of epithelial gene expression in response to this mimetic of viral dsRNA.

FIGURE 7.

Poly(I:C) induces PKR activation. Confluent monolayers of T84 cells were stimulated with 100 μg/ml poly(I:C) for different time periods, and lysates were prepared as described in Materials and Methods and subjected to SDS-PAGE immunoblotting. Total PKR (i), phospho-PKR (ii), total eIF2-α (iii), phospho-eIF2α (iv), and β-actin (v).

FIGURE 7.

Poly(I:C) induces PKR activation. Confluent monolayers of T84 cells were stimulated with 100 μg/ml poly(I:C) for different time periods, and lysates were prepared as described in Materials and Methods and subjected to SDS-PAGE immunoblotting. Total PKR (i), phospho-PKR (ii), total eIF2-α (iii), phospho-eIF2α (iv), and β-actin (v).

Close modal
FIGURE 8.

PKR inhibitors block IL-8 secretion. Confluent epithelial cells were stimulated with 20 μg/ml poly(I:C) or 10 ng/ml flagellin in the presence of 2-AP (A) or adenine (B) (1 h prior treated) for 48 h. Supernatants were collected and assayed for IL-8. C, Poly(I:C) induced NGAL expression in the presence of inhibitors. Values are mean ± SD obtained from duplicate samples of three representative experiments.

FIGURE 8.

PKR inhibitors block IL-8 secretion. Confluent epithelial cells were stimulated with 20 μg/ml poly(I:C) or 10 ng/ml flagellin in the presence of 2-AP (A) or adenine (B) (1 h prior treated) for 48 h. Supernatants were collected and assayed for IL-8. C, Poly(I:C) induced NGAL expression in the presence of inhibitors. Values are mean ± SD obtained from duplicate samples of three representative experiments.

Close modal

In light of the apparent role for PKR in mediating epithelial responses to poly(I:C), a synthetic mimetic of viral dsRNA, we next investigated whether PKR was important for epithelial responses to intact live rotavirus. First, we defined how our polarized model epithelia would respond to RRV, a common laboratory adapted strain of rotavirus. Epithelia were mock infected or infected with live trypsin-activated rotavirus at MOI of 2.5–20, and supernatants were collected at 48 h. Although all MOI tested elicited substantial induction of both IL-8 (Fig. 9,A) and NGAL (Fig. 9,B, i and ii), an MOI of 10 was chosen for further study as a potent but not saturating MOI. Furthermore, at this MOI, the response required the RRV to be preactivated by trypsin, indicating the response requires epithelial interaction with infectious virus rather than the presence of only virions with uncleaved VP4 protein that are poorly infectious (data not shown). In addition, BFA had no effect on RRV-elicited IL-8 secretion (data not shown), consistent with knowledge that RRV does not enter the cells by endocytosis (35, 50, 51). Having defined conditions for measuring rotaviral-induced activation of epithelial cells, we next measured the effect of PKR inhibition on RRV-induced IL-8 and NGAL expression. As described above, verification that these compounds had no effect on IL-8 expression induced by flagellin served as a negative control for these inhibitors. Paralleling their ablation of responses to poly(I:C), both 2-AP and adenine significantly attenuated both IL-8 expression (Fig. 9,C) and NGAL expression (Fig. 9 D) induced by RRV, although the absolute level of inhibition was somewhat less for live rotavirus than for poly(I:C), suggesting that rotavirus uses both activation of PKR and other pathways in activating epithelial gene expression.

FIGURE 9.

RRV induces IL-8 and NGAL. Epithelial cells were infected with RRV or mock as described in Materials and Methods in the presence or absence of 2-AP or adenine. After 48 h, supernatants were collected and assayed for IL-8 (A) and NGAL (B, i and ii) secretion in MOI-dependent secretion and IL-8 (C) and NGAL (D) secretion in the presence of PKR inhibitors, respectively.

FIGURE 9.

RRV induces IL-8 and NGAL. Epithelial cells were infected with RRV or mock as described in Materials and Methods in the presence or absence of 2-AP or adenine. After 48 h, supernatants were collected and assayed for IL-8 (A) and NGAL (B, i and ii) secretion in MOI-dependent secretion and IL-8 (C) and NGAL (D) secretion in the presence of PKR inhibitors, respectively.

Close modal

Most viruses first encounter their hosts at mucosal surfaces and thus the first cells contacted and infected are often epithelial cells. Although such epithelial cells serve as an important barrier to microbes, the epithelium is also a vigorous participant in generating immune responses, especially via their activation of proinflammatory and immunomodulatory gene expression. Although a number of mechanistic studies of how epithelia sense bacteria have been investigated, relatively little is known regarding the primary mechanisms by which epithelia might sense viruses. Based on the emerging paradigms governing bacterial-epithelial interactions, especially the predominant role for bacterial flagellin in mediating activation of epithelial gene expression in response to motile bacteria, we hypothesized that intestinal epithelial cells might sense the products of the intestinal pathogen rotavirus by detection of the one known microbial pattern displayed by rotavirus, namely its dsRNA. Herein, we observed that, indeed, intestinal epithelial cells are capable of sensing a synthetic analogue of viral RNA. Such sensing of dsRNA occurs intracellularly, uses the dsRNA-dependent kinase PKR, and was required for optimal epithelial sensing of live activated rotavirus. Consistent with poly(I:C) serving as a major mediator of rotaviral activation of epithelial cells, we observed that one intestinal epithelial cell line (Caco-2) that was previously shown not to respond to rotavirus (6, 7) also does not respond to poly(I:C) despite responding robustly to the bacterial product flagellin. We have observed that Caco-2 cells express similar levels of PKR as T84 cells (data not shown) and are currently investigating whether this pathway is operable in this cell line.

Although some genes were activated by both synthetic dsRNA (i.e., poly(I:C)) and the bacterial stimulus flagellin, the overall patterns of gene expression activated by these agonists as assayed by microarray analysis were markedly different. Specifically, only one gene (apoptosis inhibitor 1) was among the 30 most up-regulated genes induced by both poly(I:C) and flagellin at either time point assayed. This is in marked contrast to our expression profiling studies comparing flagellin to TNF-α, in which these stimuli induced very similar patterns of epithelial gene expression with induction of only a few genes, showing differential activation in response to these agonists (36). Although the precise role of each up-regulated gene in responding to pathogens is largely unknown, the fact that poly(I:C)-induced genes included genes known to be associated with IFN/antiviral responses validates the use of this well-defined model to mechanistically understand rotaviral interaction with gut epithelia. Consistent with poly(I:C) activating an antiviral gene program, poly(I:C) potently activated the STAT-signaling pathway, and unlike STAT activation described in response to TLR agonists (40, 41), poly(I:C)-induced STAT activation was independent of new protein synthesis. Conversely, the relatively delayed course of poly(I:C)-induced IL-8 secretion compared with both STAT activation and iNOS induction suggests that while both flagellin and poly(I:C) induce IL-8 expression, that induced by poly(I:C) might be a secondary response (i.e., IL-8 transcription dependent on new protein synthesis).

Although the role of the presumed antiviral genes in response to poly(I:C) makes clear physiologic sense, the role of some poly(I:C)-induced genes is less clear. For example, while several studies have shown intestinal epithelial cells make IL-8 in response to RRV and, herein, we show poly(I:C) recapitulates this response, the role of induction of this gene is not clear in the pathophysiology of RRV-induced diarrhea in that a characteristic feature of RRV-induced diarrhea is the absence of intestinal inflammation. Another potently up-regulated gene in response to both poly(I:C) and RRV for which an obvious role in response to RRV is not clear is NGAL. Specifically, the best defined role for NGAL in host defense is to bind bacterial siderophores, thus preventing their use of iron (52, 53). Although it is certainly possible that IL-8 and/or NGAL may serve an as yet unappreciated role in retarding RRV, alternatively, epithelial activation of antibacterial genes and genes that would promote neutrophil infiltration may be a nonspecific means to reduce the possibility of opportunistic infection that might otherwise occur following immune-mediated destruction of virally infected epithelial cells.

In light of the recent attention focused on TLR and observations that epithelia express some TLR and exhibit TLR-mediated responses, we hypothesized that intestinal epithelial recognition of poly(I:C) would be mediated by TLR3. Intestinal epithelial cells had detectable TLR3 mRNA (by RT-PCR), suggesting some level of expression. However, we were not able to demonstrate a role for TLR3 in mediating responses to poly(I:C). Rather, the failure of both an established TLR3-neutralizing Ab and chloroquine, which inhibits activation of intracellular TLR3, to attenuate epithelial responses to poly(I:C) suggests TLR3 is not involved in our model system, at least as a primary sensor of poly(I:C). However, it certainly remains possible that TLR3 may be expressed in inflammatory conditions and/or in response to stimulation by poly(I:C) as has been demonstrated recently in endothelial cells (54, 55) and analogous to the observation that gut epithelial cells display functional TLR4 only under inflammatory conditions (56). The reason for lack of apparent TLR3-mediated detection of poly(I:C) is not clear. Although lack of available specific antiserum prevents us from determining the precise level of TLR3 expression and microarray studies herein observed no detection of TLR3, previous microarray analysis using our “homemade” chip suggests our epithelial cells have similar mRNA levels for TLR3, TLR4, and TLR5 (36) but yet are exquisitely sensitive to flagellin, exhibit undetectable responses to LPS, and appear not to use TLR3 in response to poly(I:C). Nonetheless, such lack of TLR3-mediated responses could result from insufficient expression of TLR3 and/or any potential coreceptors as has been shown for TLR4 in gut epithelial cells (57, 58) or perhaps TLR3 could signal in a nonclassical manner, analogous to the way TLR2 ligands alter status of epithelial tight junctions rather than activates proinflammatory gene expression (59). However, as little is known regarding levels of nonpathogenic (i.e., commensal) viruses in the gut, at present, a precise role for TLR3 in regulating intestinal homeostasis is difficult to envision.

In contrast to bacterial flagellin, which is an overriding determinant for intestinal epithelial responses to S. typhimurium (13), blocking recognition of dsRNA only partially reduced activation of gene expression in response to challenge with live rotavirus. This suggests epithelial cells have alternate mechanisms of recognizing viral pathogens. Both the delayed kinetics of responses to RRV and that RRV-induced response requires activated virus suggest such responses, like responses to poly(I:C), also require viral entry into epithelial cells. Although poly(I:C)-induced responses required endocytosis as indicated by their inhibition with BFA, RRV-induced responses were not reduced by BFA likely due to RRV entry of epithelial cells using lipid rafts but yet not being mediated endocytosis (35, 50, 51). That PKR-independent activation of epithelial gene expression also occurs intracellularly suggests that such viral recognition is also not occurring through a cell surface pattern recognition receptor and rather occurs intracellularly. Such PKR-independent recognition of RRV could perhaps occur through another TLR or an as yet unidentified intracellular pattern recognition receptor. Regardless, considering that the epithelial response to RRV required viral entry and its active replication in cytoplasm, which generates abundant dsRNA, it would seem difficult for RRV to avert detection by PKR. Thus, such epithelial PKR-mediated recognition of dsRNA likely plays a broad role in recognizing intestinal viruses.

We thank R. I. Glass for the helpful discussion. We greatly appreciate S. Malkapuram for microarray analysis and Susan Voss and Sean Lyons for technical assistance. We acknowledge National Institute of Allergy and Infectious Diseases Reference Reagent Laboratory, Braton Biotech (Gaithersburg, MD) for providing World Health Organization standards IFN-α, IFN-β, and IFN-γ.

The authors have no financial conflict of interest.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1

This work was supported by National Institutes of Health Grants DK061417 and R24 DK064399.

3

Abbreviations used in this paper: PKR, protein kinase R; poly(I:C), polyinosinic acid:cytidylic acid; eIF2-α, eukaryotic initiation factor 2 α; BFA, bafilomycin A1; 2-AP, 2-aminopurine; MMP, matrix metalloproteinase; iNOS, inducible NO synthase; NGAL, neutrophil-gelatinase-associated lipocalin; SFM, serum-free medium; RRV, Rhesus rotavirus; MOI, multiplicity of infection.

1
Bresee, J. S., R. I. Glass, B. Ivanoff, J. R. Gentsch.
1999
. Current status and future priorities for rotavirus vaccine development, evaluation and implementation in developing countries.
Vaccine
17
:
2207
-2222.
2
Ramig, R. F..
2004
. Pathogenesis of intestinal and systemic rotavirus infection.
J. Virol.
78
:
10213
-10220.
3
Parashar, U. D., E. G. Hummelman, J. S. Bresee, M. A. Miller, R. I. Glass.
2003
. Global illness and deaths caused by rotavirus disease in children.
Emerg. Infect. Dis.
9
:
565
-572.
4
Glass, R. I., J. R. Gentsch, B. Ivanoff.
1996
. New lessons for rotavirus vaccines.
Science
272
:
46
-48.
5
Greenberg, H. B., H. F. Clark, P. A. Offit.
1994
. Rotavirus pathology and pathophysiology.
Curr. Top. Microbiol. Immunol.
185
:
255
-283.
6
Sheth, R., J. Anderson, T. Sato, B. Oh, S. J. Hempson, E. Rollo, E. R. Mackow, R. D. Shaw.
1996
. Rotavirus stimulates IL-8 secretion from cultured epithelial cells.
Virology
221
:
251
-259.
7
Casola, A., M. K. Estes, S. E. Crawford, P. L. Ogra, P. B. Ernst, R. P. Garofalo, S. E. Crowe.
1998
. Rotavirus infection of cultured intestinal epithelial cells induces secretion of CXC and CC chemokines.
Gastroenterology
114
:
947
-955.
8
Casola, A., R. P. Garofalo, S. E. Crawford, M. K. Estes, F. Mercurio, S. E. Crowe, A. R. Brasier.
2002
. Interleukin-8 gene regulation in intestinal epithelial cells infected with rotavirus: role of viral-induced IκB kinase activation.
Virology
298
:
8
-19.
9
Gewirtz, A. T., Y. Liu, S. V. Sitaraman, J. L. Madara.
2002
. Intestinal epithelial pathobiology: past, present and future.
Best Pract. Res. Clin. Gastroenterol.
16
:
851
-867.
10
Blutt, S. E., C. D. Kirkwood, V. Parreno, K. L. Warfield, M. Ciarlet, M. K. Estes, K. Bok, R. F. Bishop, M. E. Conner.
2003
. Rotavirus antigenaemia and viraemia: a common event?.
Lancet
362
:
1445
-1449.
11
Blutt, S. E., K. L. Warfield, D. E. Lewis, M. E. Conner.
2002
. Early response to rotavirus infection involves massive B cell activation.
J. Immunol.
168
:
5716
-5721.
12
Rollo, E. E., K. P. Kumar, N. C. Reich, J. Cohen, J. Angel, H. B. Greenberg, R. Sheth, J. Anderson, B. Oh, S. J. Hempson, E. R. Mackow, R. D. Shaw.
1999
. The epithelial cell response to rotavirus infection.
J. Immunol.
163
:
4442
-4452.
13
Gewirtz, A. T., T. A. Navas, S. Lyons, P. J. Godowski, J. L. Madara.
2001
. Cutting edge: bacterial flagellin activates basolaterally expressed tlr5 to induce epithelial proinflammatory gene expression.
J. Immunol.
167
:
1882
-1885.
14
Chamaillard, M., N. Inohara, G. Nunez.
2004
. Battling enteroinvasive bacteria: Nod1 comes to the rescue.
Trends Microbiol.
12
:
529
-532.
15
Arias, C. F., P. Isa, C. A. Guerrero, E. Mendez, S. Zarate, T. Lopez, R. Espinosa, P. Romero, S. Lopez.
2002
. Molecular biology of rotavirus cell entry.
Arch. Med. Res.
33
:
356
-361.
16
Estes, M..
2001
. Rotaviruses and their replication. D. M. K. B. N. Fields, and P. M. Howley, eds. 4th Ed.In
Fields Virology
Vol. 2
:
1747
-1786 Lippincott, Williams & Wilkins Co., Philadelphia. .
17
Kumar, M. V., C. N. Nagineni, M. S. Chin, J. J. Hooks, B. Detrick.
2004
. Innate immunity in the retina: Toll-like receptor (TLR) signaling in human retinal pigment epithelial cells.
J. Neuroimmunol.
153
:
7
-15.
18
Gern, J. E., D. A. French, K. A. Grindle, R. A. Brockman-Schneider, S. Konno, W. W. Busse.
2003
. Double-stranded RNA induces the synthesis of specific chemokines by bronchial epithelial cells.
Am. J. Respir. Cell Mol. Biol.
28
:
731
-737.
19
Meusel, T. R., K. E. Kehoe, F. Imani.
2002
. Protein kinase R regulates double-stranded RNA induction of TNF-α but not IL-1 β mRNA in human epithelial cells.
J. Immunol.
168
:
6429
-6435.
20
Levin, D., I. M. London.
1978
. Regulation of protein synthesis: activation by double-stranded RNA of a protein kinase that phosphorylates eukaryotic initiation factor 2.
Proc. Natl. Acad. Sci. USA
75
:
1121
-1125.
21
Clemens, M. J., A. Elia.
1997
. The double-stranded RNA-dependent protein kinase PKR: structure and function.
J. Interferon Cytokine Res.
17
:
503
-524.
22
Lemaire, P. A., J. Lary, J. L. Cole.
2005
. Mechanism of PKR activation: dimerization and kinase activation in the absence of double-stranded RNA.
J. Mol. Biol.
345
:
81
-90.
23
Cuddihy, A. R., S. Li, N. W. Tam, A. H. Wong, Y. Taya, N. Abraham, J. C. Bell, A. E. Koromilas.
1999
. Double-stranded RNA-activated protein kinase PKR enhances transcriptional activation by tumor suppressor p53.
Mol. Cell. Biol.
19
:
2475
-2484.
24
Goh, K. C., M. J. deVeer, B. R. Williams.
2000
. The protein kinase PKR is required for p38 MAPK activation and the innate immune response to bacterial endotoxin.
EMBO J.
19
:
4292
-4297.
25
Kumar, A., Y. L. Yang, V. Flati, S. Der, S. Kadereit, A. Deb, J. Haque, L. Reis, C. Weissmann, B. R. Williams.
1997
. Deficient cytokine signaling in mouse embryo fibroblasts with a targeted deletion in the PKR gene: role of IRF-1 and NF-κB.
EMBO J.
16
:
406
-416.
26
Chu, W. M., D. Ostertag, Z. W. Li, L. Chang, Y. Chen, Y. Hu, B. Williams, J. Perrault, M. Karin.
1999
. JNK2 and IKKβ are required for activating the innate response to viral infection.
Immunity
11
:
721
-731.
27
Alexopoulou, L., A. C. Holt, R. Medzhitov, R. A. Flavell.
2001
. Recognition of double-stranded RNA and activation of NF-κB by Toll-like receptor 3.
Nature
413
:
732
-738.
28
Edelmann, K. H., S. Richardson-Burns, L. Alexopoulou, K. L. Tyler, R. A. Flavell, M. B. Oldstone.
2004
. Does Toll-like receptor 3 play a biological role in virus infections?.
Virology
322
:
231
-238.
29
Wang, T., T. Town, L. Alexopoulou, J. F. Anderson, E. Fikrig, R. A. Flavell.
2004
. Toll-like receptor 3 mediates West Nile virus entry into the brain causing lethal encephalitis.
Nat. Med.
10
:
1366
-1373.
30
Gewirtz, A. T., P. O. Simon, Jr, C. K. Schmitt, L. J. Taylor, C. H. Hagedorn, A. D. O’Brien, A. S. Neish, J. L. Madara.
2001
. Salmonella typhimurium translocates flagellin across intestinal epithelia, inducing a proinflammatory response.
J. Clin. Invest.
107
:
99
-109.
31
McSorley, S. J., B. D. Ehst, Y. Yu, A. T. Gewirtz.
2002
. Bacterial flagellin is an effective adjuvant for CD4+ T cells in vivo.
J. Immunol.
169
:
3914
-3919.
32
Kjeldsen, L., C. Koch, K. Arnljots, N. Borregaard.
1996
. Characterization of two ELISAs for NGAL, a newly described lipocalin in human neutrophils.
J. Immunol. Methods
198
:
155
-164.
33
Tatti, K. M., J. Gentsch, W. J. Shieh, T. Ferebee-Harris, M. Lynch, J. Bresee, B. Jiang, S. R. Zaki, R. Glass.
2002
. Molecular and immunological methods to detect rotavirus in formalin-fixed tissue.
J. Virol. Methods
105
:
305
-319.
34
Coulson, B. S., K. J. Fowler, R. F. Bishop, R. G. Cotton.
1985
. Neutralizing monoclonal antibodies to human rotavirus and indications of antigenic drift among strains from neonates.
J. Virol.
54
:
14
-20.
35
Kaljot, K. T., R. D. Shaw, D. H. Rubin, H. B. Greenberg.
1988
. Infectious rotavirus enters cells by direct cell membrane penetration, not by endocytosis.
J. Virol.
62
:
1136
-1144.
36
Zeng, H., A. Q. Carlson, Y. Guo, Y. Yu, L. S. Collier-Hyams, J. L. Madara, A. T. Gewirtz, A. S. Neish.
2003
. Flagellin is the major proinflammatory determinant of enteropathogenic Salmonella.
J. Immunol.
171
:
3668
-3674.
37
Gewirtz, A. T., B. McCormick, A. S. Neish, N. A. Petasis, K. Gronert, C. N. Serhan, J. L. Madara.
1998
. Pathogen-induced chemokine secretion from model intestinal epithelium is inhibited by lipoxin A4 analogs.
J. Clin. Invest.
101
:
1860
-1869.
38
Cuadras, M. A., D. A. Feigelstock, S. An, H. B. Greenberg.
2002
. Gene expression pattern in Caco-2 cells following rotavirus infection.
J. Virol.
76
:
4467
-4482.
39
Bogdan, C., M. Rollinghoff, A. Diefenbach.
2000
. The role of nitric oxide in innate immunity.
Immunol. Rev.
173
:
17
-26.
40
Toshchakov, V., B. W. Jones, P. Y. Perera, K. Thomas, M. J. Cody, S. Zhang, B. R. Williams, J. Major, T. A. Hamilton, M. J. Fenton, S. N. Vogel.
2002
. TLR4, but not TLR2, mediates IFN-β-induced STAT1α/β-dependent gene expression in macrophages.
Nat. Immunol.
3
:
392
-398.
41
Yu, Y., H. Zeng, M. Vijay-Kumar, A. S. Neish, D. Merlin, S. V. Sitaraman, A. T. Gewirtz.
2004
. STAT signaling underlies difference between flagellin-induced and tumor necrosis factor-α-induced epithelial gene expression.
J. Biol. Chem.
279
:
35210
-35218.
42
Chemello, M. E., O. C. Aristimuno, F. Michelangeli, M. C. Ruiz.
2002
. Requirement for vacuolar H+-ATPase activity and Ca2+ gradient during entry of rotavirus into MA104 cells.
J. Virol.
76
:
13083
-13087.
43
Suzuki, T., M. Yamaya, K. Sekizawa, M. Hosoda, N. Yamada, S. Ishizuka, K. Nakayama, M. Yanai, Y. Numazaki, H. Sasaki.
2001
. Bafilomycin A1 inhibits rhinovirus infection in human airway epithelium: effects on endosome and ICAM-1.
Am. J. Physiol.
280
:
L1115
-L1127.
44
Matsumoto, M., S. Kikkawa, M. Kohase, K. Miyake, T. Seya.
2002
. Establishment of a monoclonal antibody against human Toll-like receptor 3 that blocks double-stranded RNA-mediated signaling.
Biochem. Biophys. Res. Commun.
293
:
1364
-1369.
45
Matsumoto, M., K. Funami, M. Tanabe, H. Oshiumi, M. Shingai, Y. Seto, A. Yamamoto, T. Seya.
2003
. Subcellular localization of Toll-like receptor 3 in human dendritic cells. [Published erratum appears in 2003 J. Immunol. 171: 4934.].
J. Immunol.
171
:
3154
-3162.
46
Yi, A. K., R. Tuetken, T. Redford, M. Waldschmidt, J. Kirsch, A. M. Krieg.
1998
. CpG motifs in bacterial DNA activate leukocytes through the pH-dependent generation of reactive oxygen species.
J. Immunol.
160
:
4755
-4761.
47
Ahmad-Nejad, P., H. Hacker, M. Rutz, S. Bauer, R. M. Vabulas, H. Wagner.
2002
. Bacterial CpG-DNA and lipopolysaccharides activate Toll-like receptors at distinct cellular compartments.
Eur. J. Immunol.
32
:
1958
-1968.
48
Farrell, P. J., K. Balkow, T. Hunt, R. J. Jackson, H. Trachsel.
1977
. Phosphorylation of initiation factor elF-2 and the control of reticulocyte protein synthesis.
Cell
11
:
187
-200.
49
Wong, M. L., Y. R. Yen.
1998
. Protein synthesis in pseudorabies virus-infected cells: decreased expression of protein kinase PKR, and effects of 2-aminopurine and adenine.
Virus Res.
56
:
199
-206.
50
Cuadras, M. A., H. B. Greenberg.
2003
. Rotavirus infectious particles use lipid rafts during replication for transport to the cell surface in vitro and in vivo.
Virology
313
:
308
-321.
51
Lopez, S., C. F. Arias.
2004
. Multistep entry of rotavirus into cells: a Versaillesque dance.
Trends Microbiol.
12
:
271
-278.
52
Goetz, D. H., M. A. Holmes, N. Borregaard, M. E. Bluhm, K. N. Raymond, R. K. Strong.
2002
. The neutrophil lipocalin NGAL is a bacteriostatic agent that interferes with siderophore-mediated iron acquisition.
Mol. Cell
10
:
1033
-1043.
53
Yang, J., K. Mori, J. Y. Li, J. Barasch.
2003
. Iron, lipocalin, and kidney epithelia.
Am. J. Physiol.
285
:
F9
-F18.
54
Cario, E., D. K. Podolsky.
2000
. Differential alteration in intestinal epithelial cell expression of toll-like receptor 3 (TLR3) and TLR4 in inflammatory bowel disease.
Infect. Immun.
68
:
7010
-7017.
55
Kaiser, W. J., J. L. Kaufman, M. K. Offermann.
2004
. IFN-α sensitizes human umbilical vein endothelial cells to apoptosis induced by double-stranded RNA.
J. Immunol.
172
:
1699
-1710.
56
Abreu, M. T., E. T. Arnold, L. S. Thomas, R. Gonsky, Y. Zhou, B. Hu, M. Arditi.
2002
. TLR4 and MD-2 expression is regulated by immune-mediated signals in human intestinal epithelial cells.
J. Biol. Chem.
277
:
20431
-20437.
57
Abreu, M. T., P. Vora, E. Faure, L. S. Thomas, E. T. Arnold, M. Arditi.
2001
. Decreased expression of Toll-like receptor-4 and MD-2 correlates with intestinal epithelial cell protection against dysregulated proinflammatory gene expression in response to bacterial lipopolysaccharide.
J. Immunol.
167
:
1609
-1616.
58
Naik, S., E. J. Kelly, L. Meijer, S. Pettersson, I. R. Sanderson.
2001
. Absence of Toll-like receptor 4 explains endotoxin hyporesponsiveness in human intestinal epithelium.
J. Pediatr. Gastroenterol. Nutr.
32
:
449
-453.
59
Cario, E., G. Gerken, D. K. Podolsky.
2004
. Toll-like receptor 2 enhances ZO-1-associated intestinal epithelial barrier integrity via protein kinase C.
Gastroenterology
127
:
224
-238.