Toll-Like Receptor 3 Mediates a More Potent Antiviral Response Than Toll-Like Receptor 4

Cell culture and reagents

Murine bone marrow-derived macrophages (BMMs) were differentiated from marrow as previously described (15). A129 (IFNAR-1^−/−) (17) and B6129SF2/J wild-type control mice were obtained from B&K Universal (Grimston, Aldbrough Hull, U.K.) and The Jackson Laboratory (Bar Harbor, ME), respectively. C57/B6 mice were used for all experiments not involving the A129 mice (The Jackson Laboratory). Specific TLR activation was achieved using F-583 (Rd mutant) Escherichia coli lipid A for TLR4 (Sigma-Aldrich, St. Louis, MO), CpG oligonucleotides for TLR9 (Invitrogen, San Diego, CA), and poly(I:C) for TLR3 (Pharmacia, Peapack, NJ). For experiments using the TIRAP/MAL inhibitory peptide (CN Biosciences, San Diego, CA), cells were pretreated for 1 h with 20 μM peptide or DMSO alone. Cells were then stimulated with PAMPs in the presence of the inhibitory peptide. For experiments using murine rIFN-β (R&D Systems, Minneapolis, MN), wild-type macrophage cells were stimulated with 10, 100, or 1000 U. Viral infection and harvest were performed using MHV68 at a multiplicity of infection of five, as previously described (15).

mRNA quantification

RNA was isolated by standard guanidium isothiocyanate methods. cDNA template for quantitative real-time PCR analysis was then synthesized, and PCR was performed using the iCycler thermocycler (Bio-Rad, Hercules, CA) as previously described (15). IFN-β, IFN-inducible protein 10 (IP10), IκBα, and IFI-204 primers were the same as those previously described (15). For other genes, the following primers were used: IL-6, forward, CACAGAGGATACCACTCCCAACA, and reverse, TCCACGATTTCCCAGAGAACA; TLR3, forward, TCTGGAAACGCGCAAACC, and reverse, GCCGTTGGACTCTAAATTCAAGAT; TLR4, forward, AGAAATTCCTGCAGTGGGTCA, and reverse, TCTCTACAGGTGTTGCACATGTCA; TIRAP, forward, CAGGCAGGCTCTGTTGAAGAA, and reverse, TGTGTGGCTGTCTGTGAACCA; MyD88, forward, CATGGTGGTGGTTGTTTCTGAC, and reverse, TGGAGACAGGCTGAGTGCAA; ICAM1, forward, TGTCAGCCACTGCCTTGGTA, and reverse, CAGGATCTGGTCCGCTAGCT; and L32, forward, AAGCGAAACTGGCGGAAAC, and reverse, TAACCGATGTTGGGCATCAG.

Plasmids and GST pull-down assays

A human TLR4 construct was generously provided by Dr. R. Modlin (University of California, Los Angeles). Expressed sequence tags containing the intracellular domain of human TLR3 and full-length human MyD88 were obtained from Research Genetics (Huntsville, AL). Each of the two constructs was used as a PCR template for amplification of the sequence corresponding to their respective intracellular domains. EcoRI and XhoI sites were engineered into the forward and reverse primer sequences, respectively, and used to ligate the PCR products into pGEX1λT. The recombinant constructs were then transformed into Topp10 cells by electroporation. Following isopropyl β-d-thiogalactoside-induced expression, the cells were lysed in a Sarkosyl buffer (1% Sarkosyl, 100 mM EDTA, and 1 mM DTT, in PBS) followed by sonication. The fusion proteins were then immobilized on glutathione beads (Sigma-Aldrich). The pCDNA3–2xFlag-mTIRAP/MAL construct was donated by T. Roni from the laboratory of Dr. S. Smale (University of California, Los Angeles). The TIRAP/MAL and MyD88 constructs were overexpressed in 293T cells and lysed in immunoprecipitation lysis buffer (1% Triton X-100, 400 μM EDTA, 150 mM NaCl, 20 mM HEPES (pH 7.2), 10 mM NaF, and a protease inhibitor mixture). The lysate was then incubated with the immobilized GST-TLR fusion proteins, and interactions were detected by immunoblotting with an anti-flag monoclonal or anti-MyD88 polyclonal Ab.

Immunoblotting

For STAT1 immunoblotting, cells were lysed in modified radioimmunoprecipitation assay buffer and 20 μg of protein were loaded per lane and separated by SDS-PAGE. Gels were transferred to nitrocellulose filters and immunoblotted using the Ab manufacturers’ recommended instructions. Abs specific to the STAT1 or the phosphorylated forms of STAT1 were obtained from Cell Signaling Technologies (Beverly, MA) and Santa Cruz Biotechnology (Santa Cruz, CA), respectively. The anti-MyD88 Ab was purchased from ProSci (Poway, CA). For detection of murine gammaherpesvirus 68 (MHV68), equal amounts were loaded in each lane and analyzed by Western blotting techniques using rabbit anti-M9. Blots were stripped and reprobed with anti-actin (Sigma-Aldrich) to verify equal loading.

TLR3 is a more potent inducer of antiviral gene expression than TLR4

We have previously shown that both TLR3 and TLR4 can induce a number of antiviral/IFN-β-inducible genes (15). To compare the intensity and duration of expression of these genes following TLR3 vs TLR4 stimulation over an extended time course, we stimulated BMM cells for up to 12 h with either the TLR3 agonist poly(I:C) or the TLR4 agonist lipid A. Using quantitative real-time PCR technologies, we next assessed the expression levels of a number of antiviral genes throughout the time course. As seen in Fig. 1⇓, IFN-β, IP10, and IFI-204 were all induced to higher levels and for extended periods of time by TLR3 compared with TLR4. Despite the increased induction of antiviral gene expression mediated by TLR3 relative to TLR4, both receptors induced IκBα mRNA to similar levels, albeit with slightly different kinetics (Fig. 1⇓). The constitutively expressed ribosomal protein L32 was assayed to ensure equal cDNA loading. In Fig. 1⇓, we used 10 μg/ml poly(I:C) for stimulations because it gave us comparable IκBα levels between both TLR3 and TLR4 stimulated cells. We also use 1 μg/ml poly(I:C), as shown in later figures, which still results in higher TLR3-mediated antiviral gene induction compared with TLR4 (see Fig. 5⇓), because it is less toxic to the cells.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

TLR3 is a more potent inducer of antiviral gene expression than TLR4. Murine BMMs were stimulated with poly(I:C) (10 μg/ml) or lipid A (1 ng/ml) for the indicated times. Total RNA was isolated and converted to cDNA for quantitative real-time PCR analysis using primers specific for IFN-β, IFI-204, IP10, IκBα, or L32. Experiments were repeated three times, and the data are presented in relative expression units on a log scale.

TLR3 can directly interact with MyD88 but not with TIRAP/MAL

The receptor-proximal signaling complexes used by TLR3 and TLR4 to activate the antiviral gene program are relatively uncharacterized. We hypothesized that these receptors interact with distinct adaptor molecule-containing complexes which may contribute to the differences in signaling output observed in Fig. 1⇑. MyD88 and TIRAP/MAL are both TIR-domain containing adaptor molecules that have been shown to directly bind to the cytoplasmic tail of TLR4 (11, 12, 16). However, these same experiments have not been performed with TLR3. To see whether MyD88 or TIRAP/MAL is able to interact with TLR3, we performed GST pull-down assays. To conduct these experiments, we fused the complete intracellular domains of TLR3 and TLR4 to GST and immobilized the fusion proteins on glutathione agarose beads. Next, we attempted to pull down overexpressed MyD88 or flag-TIRAP/MAL with the GST-TLR3 or GST-TLR4 beads. As expected, MyD88 and TIRAP/MAL were both able to bind to TLR4. The TLR3 intracellular domain was also able to associate with MyD88. However, we found that TLR3 was incapable of interacting with TIRAP/MAL (Fig. 2⇓). These data strongly suggest that the receptor-proximal signaling complex directly engaged by TLR3 differs compositionally from the complex engaged by TLR4.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

MyD88, but not TIRAP/MAL, directly interacts with TLR3. 293T lysate-containing MyD88 or flag-TIRAP/MAL was incubated with the intracellular domains of TLR3 and TLR4 fused to GST and immobilized on glutathione beads. TLR-MyD88 interaction was determined by Western blotting using a polyclonal anti-MyD88 Ab (upper panel). TLR-TIRAP/MAL interaction was determined by Western blotting using an anti-flag Ab to detect flag-TIRAP/MAL (middle panel). Equal amounts of beads containing GST-TLR3 or -TLR4 intracellular domains were boiled and the eluted proteins were size-fractionated by SDS-PAGE. Coomassie blue staining (lower panel) was used to ensure that comparable amounts of GST-TLR protein were loaded on the beads. The data represent three independent experiments.

The TIRAP/MAL inhibitory peptide is able to block TLR4 but not TLR3 signaling

Although knockout studies have suggested that both MyD88 and TIRAP/MAL are dispensable for induction of IFN-β by TLR3 and TLR4, dominant-negative TIRAP/MAL has been shown to prevent TLR4 but not TLR3 signaling through IRF3 (3, 8, 14, 16). In addition, a cell-permeable TIRAP/MAL-inhibitory peptide has been shown to block TLR4-mediated induction of an IFN-β-specific reporter construct in RAW 264.7 cells (10). However, this inhibitory peptide has not been used to study TLR4 signaling in primary macrophage cells, nor have its effects on TLR3 signaling been addressed.

Because knockout studies leave open the possibility for redundancy, we decided to assess whether the TIRAP/MAL peptide could block TLR3 or TLR4 induction of IFN-β gene expression and activation of STAT1 following primary macrophage treatment with either TLR3 or TLR4 ligands. Results from these experiments show that the TIRAP/MAL peptide abrogated TLR4-mediated expression of IFN-β and STAT1 activation in primary macrophage cells. These findings corroborate peptide studies using macrophage cell lines, but disagree with TIRAP/MAL knockout results (10, 13, 14) (Fig. 3⇓, A and B). In contrast to its effect on TLR4 signaling, the peptide was completely unable to block TLR3-induced expression of IFN-β and IL-6, which is in complete agreement with all previous studies (Fig. 3⇓A). Likewise, STAT1 was still activated in cells stimulated with poly(I:C) in the presence of the TIRAP/MAL peptide (Fig. 3⇓B). Even at lower concentrations of poly(I:C) stimulation (Fig. 3⇓B) or higher concentrations of the TIRAP/MAL peptide (data not shown), the inhibitor was still incapable of blocking STAT1 activation via TLR3 signaling. Poly(I:C) has been shown to weakly induce proinflammatory cytokine production in TLR3-deficient mice, while poly(I:C)-induced IFN-β induction appears to be TLR3 dependent (18). Although we cannot exclude the possibility that poly(I:C) may signal through alternative receptors other than TLR3, our studies show that the peptide is incapable of reducing poly(I:C)-mediated gene expression and that TIRAP/MAL cannot bind to the cytoplasmic tail of TLR3. These data strongly suggest that TLR3 does not use TIRAP/MAL for signaling.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

The TIRAP/MAL inhibitory peptide is able to block TLR4 but not TLR3 transactivation of IFN-β and IL-6, as well as IFN-β-mediated activation of STAT1. BMMs were pretreated with the TIRAP/MAL peptide (20 μM) or DMSO for 1 h and then stimulated with lipid A (1 ng/ml), poly I:C (1 μg/ml), or medium alone for 2 h. A, IFN-β, IL-6, and L32 mRNA levels were assayed by quantitative real-time PCR. All samples were run in duplicate or triplicate, and data are presented in relative expression units. B, STAT1 activation was determined by Western blotting analysis to detect phosphorylated STAT1. For STAT1 experiments, lipid A was used at 10 ng/ml, and poly I:C was administered at 100 and 10 ng/ml. Total STAT1 was also assayed to ensure equal loading. The data are representative of three independent experiments.

We found that the TIRAP/MAL peptide was able to inhibit IL-6 expression following TLR4 ligation, which is consistent with TIRAP/MAL knockout data (13, 14). One possible explanation for this would be that the TIRAP/MAL peptide not only affected TIRAP/MAL but also interfered with MyD88 function. TLR9 signaling has previously been shown to be completely dependent on MyD88 (18). Therefore, we used TLR9-mediated IL-6 activation as a readout to determine whether the TIRAP/MAL peptide could specifically interrupt MyD88 signaling. Consistent with previous reports using dendritic cells, we found that CpG-induced IL-6 expression in macrophages was not affected by the TIRAP/MAL peptide (Fig. 3⇑A) (11). Thus, the TIRAP/MAL peptide does not specifically interfere with MyD88 signaling. These results show that the TIRAP/MAL inhibitory peptide can disrupt TLR4, but not TLR3 or TLR9, signaling in primary macrophage cells. Furthermore, because the peptide appears to inhibit both MyD88-dependent and -independent signaling events following TLR4 stimulation, the peptide may disrupt the entire TLR4-proximal signaling complex.

TLR3 and TLR4 ligands induce expression of TLR3, MyD88, and TIRAP/MAL

Thus far, our data suggest that TLR3 and TLR4 signal via unique adaptor molecule-containing complexes. We next wanted to see whether either TLR3 or TLR4 was capable of transcriptionally inducing molecules involved in eliciting early signaling events, which may explain why TLR3 can sustain and enhance antiviral gene expression to a greater degree than TLR4. To address this issue, we stimulated primary macrophage cells with poly(I:C) or lipid A for up to 12 h. By 4 h, treatment of macrophages with TLR3 or TLR4 agonists caused the induction of TLR3, but not TLR4, mRNA production (Fig. 4⇓A). We also observed that TLR9 signaling, which does not induce IFN-β in primary macrophage cells, was incapable of inducing TLR3 expression (E. K. Chow and G. Cheng, unpublished data). In Fig. 4⇓A, we show that both TLR3 and TLR4 agonists can induce the expression of MyD88 as well as TIRAP/MAL. Taken together, these data suggest that TLR3 is able to prolong and enhance its induction of antiviral genes by rapidly up-regulating the expression of additional TLR3.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

Both TLR3 and TLR4 ligands can induce expression of TLR3 through IFN-β. Primary macrophage cells derived from bone marrow cells were stimulated with poly I:C (1 μg/ml) or lipid A (1 ng/ml) for the indicated times. A, Quantitative real-time PCR was used to assay the expression levels of TLR3, TLR4, MyD88, and TIRAP/MAL. B, TLR3 and TLR4 mRNA levels were also assessed in cells deficient in IFNAR, and cells stimulated with rIFN-β (10, 100, and 1000 U). Experiments were repeated at least two separate times, and data are presented in relative expression units. L32 was used to normalize all samples.

TLR3 and TLR4 induce TLR3 expression through IFN-β

We have shown that both TLR3 and TLR4 use the IRF3 transcription factor to induce IFN-β gene expression (15). Once secreted from the cell, IFN-β is believed to act in an autocrine/paracrine manner leading to STAT1 activation and secondary antiviral gene induction (10, 15, 19). Due to the fact that both TLR3 and TLR4 can induce TLR3 expression, and because this induction takes place after IFN-β production has begun, we investigated whether TLR3 expression was induced by IFN-β. Using cells deficient in the IFNAR, we show in Fig. 4⇑B that treatment of cells with TLR3 or TLR4 agonists does not cause the induction of TLR3 expression in the absence of IFNAR. Furthermore, stimulating primary macrophage cells with rIFN-β resulted in a dose-dependent up-regulation of TLR3 mRNA (Fig. 4⇑B). TLR4 mRNA levels were relatively unaltered in the IFNAR-deficient cells or by induction of wild-type cells with rIFN-β (Fig. 4⇑B). These data indicate that TLR3 and TLR4 can potently induce TLR3, but not TLR4, expression through IFN-β production.

TLR3 and TLR4 induce both IFN-β-enhanced and IFN-β-dependent antiviral genes

We have previously characterized TLR3 and TLR4 antiviral gene induction as either primary or secondary based upon sensitivity to cycloheximide (15). Our previous data suggested that primary genes are induced in the absence of novel protein synthesis, whereas secondary genes require the initial expression of IFN-β. To further characterize these primary and secondary genes, we induced both wild-type and IFNAR-deficient cells with either poly(I:C) or lipid A and assessed antiviral gene expression at 1 and 4 h. As seen in Fig. 5⇓, IFN-β and IP10 (both primary genes) are induced by 1 h in both wild-type and IFNAR knockout macrophage cells, while IFI-204 (a secondary gene) remained at basal levels. By 4 h, IP10 expression was significantly enhanced in the wild-type cells, but remained relatively unchanged in the IFNAR knockout cells. These data indicate that the primary expression of IP10 is enhanced by the IFN-β positive feedback loop. The secondary gene, IFI-204, was induced to high levels by 4 h, yet was not detectable in the IFNAR knockout cells at the same time point. Similar results were also obtained for the primary and secondary genes RANTES and Mx1, respectively. As an induction control, ICAM1 mRNA was elevated by 1 h and remained high by 4 h in both wild-type and knockout cells stimulated with poly(I:C) or lipid A. As shown in Fig. 4⇑B, TLR3 is part of the secondary gene subset.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5.

TLR3 and TLR4 induce both IFN-β-enhanced and IFN-β-dependent antiviral genes. Both wild-type cells and cells deficient in IFNAR were stimulated with poly I:C (1 μg/ml) or lipid A (1 ng/ml) for either 1 or 4 h. IFN-β, IP10, IFI-204, ICAM1, and L32 mRNA levels were assessed. Data are representative of three independent experiments and presented in relative expression units.

The IFNAR is required for both TLR3 and TLR4 activation of STAT1 and resistance to MHV68 infection

In mice deficient in the IFNAR, STAT1 activation has been shown to be blocked in macrophage cells stimulated with LPS (19). Although blocking Ab studies have suggested that IFN-β is also essential for poly(I:C)-induced STAT1 activation, these studies have not been conducted to date in primary macrophage cells deficient in IFNAR (15). To address this issue, we stimulated IFNAR^−/− BMMs with lipid A and poly(I:C) and assayed for STAT1 phosphorylation. In Fig. 6⇓A, we show that, like that of TLR4, TLR3-mediated STAT1 activation was also abolished in IFNAR^−/− macrophage cells. The TLR9 agonist CpG, which fails to induce IFN-β in primary macrophage cells, was used as a negative control.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 6.

Both TLR3 and TLR4 fail to activate STAT1 and induce the antiviral gene program in IFNAR^−/− primary macrophage cells. BMMs from both wild-type and IFNAR^−/− mice were stimulated with lipid A (1 ng/ml), poly I:C (1 μg/ml), CpG (100 nM), or fresh medium (M) for the indicated time periods. A, Cell lysates were subjected to western blotting analyses to detect phosphorylated STAT1 (P-STAT1) and total STAT1. B, For viral replication assays, BMMs were simultaneously stimulated with PAMPs (10, 1, or 0.1 ng/ml lipid A, or l, 0.1, or 0.01 μg/ml poly(I:C)) and infected with MHV68 using a multiplicity of infection of five. Cell lysates were harvested at 48 h postinfection and subjected to Western blotting analysis using an Ab specific to the MHV68 protein M9. Actin levels were also assayed to ensure equal loading. The data represent two independent experiments.

Blocking Ab and conditioned medium experiments have suggested that TLR3- and TLR4-mediated viral resistance is IFN-β dependent (15). Using the MHV68 protein M9 as a readout for viral load, we show in Fig. 6⇑B that the antiviral activity of macrophage cells infected with MHV68 was abolished in the absence of IFNAR despite cotreatment with either lipid A or poly(I:C). Thus, IFNAR-mediated up-regulation of secondary response genes, such as IFI-204, and enhancement of primary genes, such as IP10, is essential for antiviral activity. In fact, MHV68 protein synthesis was enhanced in IFNAR^−/− vs wild-type cells under all conditions tested. These data provide genetic evidence that the IFN-β autocrine/paracrine loop is essential for induction of the TLR3- and TLR4-specific antiviral gene program.