dsRNA-activated protein kinase
Natural interferon-producing cells (IPC) secrete type I IFN (IFN-α and -β) in response to influenza virus. This process is independent of viral replication and is mediated by Toll-like receptor 7 (TLR7), which recognizes single-stranded RNA (ssRNA). DC also express TLR7 but its function in DC response to influenza virus is unknown. To address this, we compared the DC and IPC responses to influenza virus and ssRNA oligoribonucleotides (ORN) that activate TLR7. When stimulated by ORN in vitro and in vivo, DC matured and produced inflammatory cytokines but not IFN-α. DC did secrete IFN-α in response to influenza virus. However, this response was independent of TLR7 signaling and required viral replication but not dsRNA-activated protein kinase (PKR). We conclude that DC and IPC are hard-wired to secrete IFN-α via different pathways, reflecting their complementary but distinct roles in anti-viral immunity.
One fundamental aspect of the innate response against viruses is the secretion of type I interferons (IFN-α and -β), which induce cellular mechanisms of resistance to the cytopathic effect of the virus and activate host immune responses 1–3. Natural interferon-producing cells (IPC, also called plasmacytoid DC) specialize in the secretion of large amounts of type I IFN in spleen, lymph nodes and liver in response to viral infection 4–6. In these sites, type I IFN secreted by IPC amplify DC, NK cell and T cell responses against viruses 7, 8. Another important purpose of the innate response is the generation, processing and presentation of viral antigens primarily by DC, which can be infected by the virus or acquire viral antigens by continuous sampling of the environment via endocytosis 9, 10. Infected DC also produce type I IFN, which sustain their antigen-presenting function by protecting DC from the cytopathic effect of the virus 11, 12.
The ability of IPC, DC and other cell types to launch type I IFN responses is critically dependent on cellular sensors that promptly detect the presence of virus. Recently, attention has been focused on the function of Toll-like receptor (TLR) 7 and, in human, TLR8, which recognize single-stranded RNA (ssRNA) derived from viruses such as influenza virus, HIV or vesicular stomatitis virus (VSV) 13–15. TLR7 and TLR8 are expressed in the endosomal compartment of IPC, DC and other cell types 16. They recognize ssRNA sequences rich in uridines 13, 14 and trigger cytokine secretion through the adapter MyD88, which recruits signaling mediators that activate NF-κB 17. Of note, recognition of ssRNA viruses through TLR7 does not require active viral replication; after binding and/or fusing to cells, ssRNA viruses may be internalized and delivered to the lysosomal compartment, where ssRNA interacts with TLR7, triggering production of IFN-β and IFN-α. Although the TLR7/MyD88 pathway is necessary for type I IFN responses of IPC to ssRNA viruses 13–15, its impact on anti-viral responses of DC is not known.
Another mechanism for detection of RNA viruses relies on recognition of double-stranded RNA (dsRNA) by TLR3 on the cell surface 18 and intracellular receptors, such as the dsRNA-activated protein kinase (PKR) 19. Although the function of TLR3 in RNA virus infections is still unclear 19, 20, PKR has been extensively investigated 21. PKR is a cytoplasmic serine-threonine kinase that is ubiquitously expressed and activated by binding dsRNA. Active PKR catalyzes phosphorylation of the eukaryotic initiation factor 2 (eIF-2), which effectively shuts off host protein synthesis, thus protecting cells from viral replication and cytopathic effects 21. Activation of PKR also induces type I IFN secretion, probably through phosphorylation of I-κB and subsequent activation of NF-κB 22. The importance of dsRNA recognition in anti-viral responses is underscored by the fact that many viruses have evolved countermeasures directed at dsRNA-induced pathways. Influenza virus encodes the NS1 protein, which binds dsRNA and inhibits PKR activation 23–25. Moreover, NS1 interferes with the processing, and thereby translation, of cellular pre-mRNA, thus inhibiting the expression of host cell genes that have been transcriptionally up-regulated in response to virus infection 26, 27.
TLR7/MyD88 and PKR are expressed in DC, but the relative contribution of these pathways to the type I IFN response of DC to viruses is not clear. To address this, we compared DC responses to oligoribonucleotides (ORN) that mimic influenza virus RNA and efficiently trigger TLR7/MyD88-dependent signals, as well as influenza viruses expressing a WT or a dsRNA-binding-defective NS1 protein. The data demonstrate that ORN activate TLR7 in DC in vitro and in vivo, eliciting secretion of proinflammatory cytokines but not IFN-α. In contrast, influenza virus triggers IFN-α in DC through a pathway that is independent of MyD88 and requires viral replication but not dsRNA-activated PKR.
Results and discussion
TLR7 agonists elicit maturation and proinflammatory cytokine responses in DC but not secretion of IFN-α in vitro
To investigate the function of TLR7 in DC, we synthesized a panel of single-strand ORN that specifically activate TLR7. This panel included ORN spanning the 5′ and 3′ untranslated regions (UTR) of influenza A virus, as well as arbitrary ORN containing UU or GU repeats (see Fig. 1 of Supporting Information). The TLR7-agonistic function of these ORN was entirely dependent on the presence of uridines (U) and their delivery by encapsulation with lipids, consistent with previous studies with HIV-derived ORN and synthetic poly-uridine 13, 14 (see Supporting Fig. 1).
We incubated murine bone-marrow-derived DC and IPC with ORN and measured expression of CD86 as a marker of activation, as well as secretion of IFN-α, IL-12, IL-6 and the inflammatory chemokine MCP-1. TLR7 agonists induced up-regulation of CD86 in both DC and IPC as much as whole influenza virus did (data not shown). DC secreted high amounts of IL-6 and MCP-1 and significant levels of IL-12, but no detectable IFN-α (Fig. 1). IPC secreted high quantities of IFN-α and IL-12 p70, and low amounts of IL-6 and MCP-1 (Fig. 1). The specialized pattern of cytokine response triggered by TLR7 in DC resembled that previously described for macrophages cultured from bone marrow in the presence of M-CSF 14. All cytokine responses to ORN were abrogated in MyD88-deficient cells (data not shown). Moreover, TLR7 agonists also induced vigorous up-regulation of activation markers, such as CD40, in human monocyte-derived DC, but IFN-α secretion was triggered only in human IPC (see Supporting Fig. 2). Together, these experiments demonstrate that ORN activate DC, promoting maturation and secretion of pro-inflammatory cytokines in vitro. Yet, ORN induce IFN-α responses in IPC only.
The IFN-α response to TLR7 agonists in vivo is exclusively mediated by IPC
To assess effects of TLR7 activation on IFN-α secretion in vivo, we injected WT and MyD88–/– mice intravenously with lipid-encapsulated ORN. In accordance with in vitro experiments, uridine-rich ORN led to a strong cytokine response (Fig. 2a). IFN-α and MCP-1 peaked at around 5 h after injection. IL-6 and TNF-α showed more rapid induction. Serum levels of all cytokines tended towards baseline levels approximately 10 h after injection. These cytokine responses were completely absent in MyD88–/– mice, indicating that the TLR7/MyD88 pathway accounts for most cytokine responses to ssRNA in vivo (Fig. 2a). To distinguish the contribution of IPC and non-IPC to cytokine responses elicited by ORN in vivo, we depleted IPC in vivo before stimulation with ORN, using the IPC-specific monoclonal antibody anti-mPDCA-1 that has been described previously 8. Remarkably, IPC depletion abrogated the IFN-α response to ORN, while serum levels of other inflammatory cytokines were either unchanged (TNF-α and IL-6) or reduced only partially (MCP-1) (Fig. 2b). Thus, consistent with our in vitro data, IPC account for the entire IFN-α response to ORN in vivo whereas non-IPC cell types respond to ORN by secreting inflammatory cytokines.
Both MyD88-dependent and -independent pathways contribute to the IFN-α response to influenza virus in vivo
To investigate if the MyD88-dependent pathway also accounts for IFN-α responses to influenza virus in vivo, we challenged WT and MyD88–/– mice by intranasal infection with various doses of WSN. Although this route of infection did not yield detectable amounts of IFN-α or other inflammatory cytokines in the serum (data not shown), viral titers in the lung and ultimately the survival of infected WT and MyD88–/– mice were surprisingly similar (see Supporting Fig. 3).
To be able to evaluate inflammatory cytokines in the serum of infected mice, we injected influenza virus WSN i.v. In this settings, WT mice produced high levels of IFN-α that peaked 12 h after injection and waned within 24 h (Fig. 3). Although MyD88-deficiency caused a significant reduction of serum levels of IFN-α, a substantial IFN-α response that followed kinetics similar to those observed in WT mice was still detected (Fig. 3) and survival of infected WT and MyD88–/– mice was similar (data not shown). These results indicate that the IFN-α response to systemic influenza virus infection is only partially mediated by the TLR7/MyD88-dependent pathway and strongly suggest the involvement of MyD88-independent pathways in IPC, DC or other cell types stimulated by influenza virus. These pathways can compensate for lack of TLR7/MyD88, ensuring control of viral titers and host survival.
The IFN-α response of DC to influenza virus in vitro requires viral replication and is independent of MyD88 and PKR
DC express TLR7/MyD88 and PKR, which has been reported to promote IFN-α secretion in a MyD88-independent dsRNA-dependent fashion 19. Therefore we investigated the contribution of each pathway to IFN-α responses of DC to influenza virus. We measured in vitro responses of WT and MyD88–/– DC and IPC to WT influenza virus or the recombinant influenza virus WSN-R38A, which expresses a mutated form of NS1 that does not bind dsRNA and therefore does not inhibit PKR activation.
Incubation with WT WSN influenza virus elicited an IFN-α response in both DC and IPC, the DC response being at least 100-fold weaker than that of IPC. Incubation with WSN-R38A led to a 10-fold increase of IFN-α secretion by DC, but not by IPC. Despite this increase, secretion of IFN-α by DC remained significantly lower than that of IPC. MyD88 deficiency had no effect on IFN-α production by DC, whereas it abrogated IFN-α secretion by IPC (Fig. 4a). Similar to mouse DC, human DC secreted more IFN-α in response to WSN-R38A than to WT virus (see Supporting Fig. 4). These results indicate that DC secrete IFN-α in response to influenza virus through a TLR7-independent pathway, which is sensitive to NS1 inhibition and yields less IFN-α than that produced by IPC. In contrast, IPC secretion of IFN-α is entirely MyD88-dependent and insensitive to NS1 inhibition.
Since NS1-deletion mutants 19 and WSN-R38A are more effective than WT influenza virus is in triggering the IFN-α response of DC, it is conceivable that DC respond to influenza virus through PKR or other intracellular sensors of dsRNA, such as RIG-I 28. To test this possibility, we measured in vitro responses of WT and PKR–/– DC to WSN or WSN-R38A influenza virus. WSN-R38A elicited a stronger response in both WT and PKR–/– DC than the WSN virus did (Fig. 4b). Thus, IFN-α response of DC to influenza virus occurs through a pathway that is partially inhibited by NS1, but does not require PKR. Finally, we asked if stimulation of DC with influenza virus requires active viral replication or can occur with inactivated virus, as observed in IPC 6, 13. Analysis of in vitro responses of WT and MyD88–/– DC to UV- or heat-inactivated WT or WSN-R38A viruses showed that viral inactivation completely abrogates the stimulatory activity of both WT and WSN-R38A viruses (Fig. 4c). Thus, in contrast to IPC, DC activation by influenza virus requires viral replication.
The cytokine response of DC to influenza virus is resistant to inhibition by NS1
Since DC secrete not only IFN-α but also proinflammatory cytokines, we investigated the contribution of MyD88-dependent and MyD88-independent pathways to this response to influenza virus. WT and MyD88–/– DC stimulated with WSN secreted similar amounts of TNF-α, MCP-1 and IL-6 and low amounts of IL-12 (Fig. 5). Thus, DC secretion of proinflammatory cytokines is substantially MyD88-independent, as is secretion of IFN-α. Interestingly, stimulation of DC with WSN-R38A did not significantly increase secretion of cytokines, in particular TNF-α and IL-6, indicating that the production of these cytokines is resistant to NS1 inhibition. In contrast, IPC stimulated by WSN secreted IL-12 p70 in a MyD88-dependent fashion and mutation of NS1 did not increase IPC stimulation (Fig. 5).
In summary, our study demonstrates that although the TLR7/MyD88 pathway is functional in DC and can be triggered by ssORN, it is not essential for DC responses to influenza virus. Instead, this response occurs through a TLR7/MyD88-independent pathway that requires viral replication. This pathway is partially sensitive to inhibition by NS1, but does not require PKR, indicating the potential involvement of additional intracellular sensors of dsRNA, such as the recently identified RIG-I 28. This observation does not exclude a previously proposed role of PKR in DC responses to influenza virus 19, as different intracellular sensors of dsRNA may be redundant. Moreover, our data show that influenza virus NS1 has little or no effect on DC secretion of some proinflammatory cytokines, suggesting the involvement of additional mechanisms of viral recognition. In contrast to what was observed in DC, NS1 did not inhibit IFN-α secretion of IPC, indicating that the TLR7/MyD88 pathway provides the main mechanism for IPC responses to influenza virus.
The use of differing mechanisms for detection and response to influenza virus between DC and IPC is surprising, as DC and IPC express both TLR7/MyD88, PKR and RIG-I (data not shown). This difference may be due to a different fate of the virus in the two cell types. After binding to IPC, ssRNA viruses enter the cell but do not extensively replicate, probably because of the constitutive expression of anti-viral genes, such as Mx, induced by type I IFN 5, 12. Virus particles are delivered to an intracellular acidic compartment, where viral RNA interacts with TLR7, triggering the MyD88 signaling cascade. Thus, IPC can provide a prompt IFN-α response to viral infection that precedes extensive viral replication. In contrast to IPC, influenza virus replicates in DC, generating dsRNA that is detected by PKR, RIG-I or other cellular sensors for dsRNA, thereby activating DC responses.
Although viral replication induces sufficient IFN-α to protect DC from the cytopathic effect of the virus, it allows the intracellular synthesis of viral antigens that can be presented to T cells, triggering adaptive responses. We predict that the relative contribution of MyD88-dependent and -independent pathways in response to RNA viruses other than influenza virus may vary depending on their mechanisms of entry, their ability to replicate and reach different cellular compartments. Thus, the availability of multiple pathways for detection of RNA viruses provides the innate system with remarkable flexibility in anti-viral immunity.
Materials and methods
Influenza virus A/WSN/33 was propagated in Madin-Darby canine kidney (MDCK) cells. Infected-cell supernatants were used for infections as such [multiplicity of infection (MOI) of 1–5 pfu per cell] or after purification by ultracentrifugation over a 20% sucrose cushion. The dsRNA-binding activity of the NS1 protein was abrogated by introducing an Arg (R) to Ala (A) substitution at amino acid 38 of the NS1 ORF into the A/WSN/33-derived RNA segment 8.
Cell culture and stimulation in vivo
Mouse bone-marrow-derived IPC and DC, human IPC and monocyte-derived DC were generated as previously described 5, 8, 11, 29. Cells were stimulated for 18–24 h with influenza virus or synthetic ORN (Dharmacon) complexed with DOTAP (Roche). ORN sequences are: 5′UTR 5′-AGUAGAAACAAGGUAGUUU-3′; 5′UTRU→C 5′-AGCAGAAACAAGGCAGCCC-3′; 4x GU 5′-GGCAGUCGUAGUCGUACGG-3′ and 4x UU 5′-GGCAUUCUUAUUCUUACGG-3′. Additional ORN sequences are indicated in Supporting Fig. 1. Other TLR ligands included poly I:C (Sigma) and R-848 (Invivogen). Mice were injected with 25 µg of ORN complexed with 125 µg DOTAP. Where indicated, C57BL/6 mice were depleted of IPC by i.v. injection of 500 µg of the IPC-specific mAb anti-mPDCA-1 (JF05–1C2.4.1; Miltenyi Biotec, Bergisch Gladbach, Germany) 8 twice (48 h and 24 h) before stimulation.
Cytokine analysis and flow cytometry
IFN-α was detected in cell-free supernatants and mice sera by ELISA (PBL Biomedical Laboratories, New Brunswick, NJ, USA). IL-12 p70, TNF-α, MCP-1 and IL-6 were measured by cytometric bead array (BD Biosciences). Mouse CD86 and human CD40 were detected by flow cytometry using specific mAb (BD Biosciences).