Our study reveals novel information concerning the interactions between the dsRNA virus BTV and primary dendritic cell subsets for the induction of IFN-α/β, a key innate immunity cytokine that is instrumental for optimal CD8 T cell and antibody responses (39
) and for direct antiviral defense. We showed that BTV displays specific interactions with primary pDCs and cDCs: while it infects both cell types, it induces IFN-α/β only in pDCs without requiring viral replication, via a mechanism involving endocytosis and signaling through the MyD88 adaptor without TLR7/8 engagement. Our data show for the first time the involvement of MyD88 in dsRNA virus signaling in pDCs. Thus, MyD88 remains the only known transducing molecule in pDCs for type I IFN induction. However, the evidences presented here from our BTV study demonstrate clearly that viral sensors other than TLR7/8 can be used in pDCs, as was found for TLR9 with some DNA viruses that can alternatively use the DHX36 helicase in pDCs (36
Although IFN-α/β was detected in blood and lymph early after in vivo
BTV infection, it was no longer detected in lymph and blood by day 10. BTV has disappeared in lymph at that time, but relatively high viral loads were still detectable in blood, as previously reported by several authors (49
). The lack of IFN-α/β detection in blood at day 10 could be related to the decreased number of pDCs in blood that has been observed in the course of many viral infections, due to selective apoptosis (65
). Among other hypotheses, it is also possible that the association of BTV with erythrocytes during viral clearance at day 10 prevents effective interactions between BTV and pDCs (44
The specific induction of IFN-α/β in pDCs and not in cDCs, even with live virus, could be explained by cell type-specific molecular mechanisms, such as the expression of proper sensors, subcellular trafficking, and transducing cascades. Regarding cell type-specific sensors for IFN-α/β induction by viruses, sensing of Newcastle disease virus in cDCs largely relies on RIG-I helicase, whereas it relies on TLR7 in pDCs (35
). The lack of IFN-α/β synthesis by cDCs that permit replication of BTV may also involve the expression of nonstructural proteins that block IFN-α/β synthesis. Indeed, BTV NS1 and NS2 appear to interfere with the interferon regulatory factor (IRF-3 and -7) response in HeLa cells (62
). Following that scenario, the pDC fraction that expresses BTV may not be producing IFN-α/β, as shown for rotavirus in pDCs (11
). However, we could not test whether viral expression and IFN-α/β were exclusive phenomena in BTV-infected pDCs cultures as no anti-sheep IFN-α/β antibody exists to label intracellular IFN-α/β.
Our mechanistic investigations show that endo-/lysosomal acidification and maturation are required for IFN-α/β induction by UV-BTV in pDCs, similar to what is found with rotavirus IFN-α/β induction in human pDCs (11
). This finding indicates that an intracellular vesicular processing of BTV is required for appropriate sensing and signaling. Notably, early endosomal low pH was also shown to be essential for BTV uncoating in mammalian cells (17
). The UV-BTV elements that are sensed for triggering IFN-α/β synthesis within pDCs could be dsRNA structures from the core and/or proteins of the capsid. In order to test the role of protein capsid components, we stimulated sheep pDCs with sucrose gradient-purified virus-like particles produced from recombinant baculoviruses (not shown); however, the control mock fractions of the empty baculovirus-infected insect cell cultures also induced large amounts of IFN-α/β, probably due to the presence of baculovirus remnants (28
). However, the protein structures may more likely be involved in the entry mechanism of the virus and in the proper addressing to subcellular compartments rather than in the sensing by pathogen recognition receptors. Indeed only enveloped viruses were found capable of activating pDCs for IFN-α/β production via non-nucleic acid structures, although this pathway is thought to be marginal compared to the nucleic acid one (7
). Furthermore, in the case of rotavirus in human pDCs, the rotaviral dsRNA encapsidated in intact virus particles was found to be the likely signal for IFN activation (11
). However, the integrity of the capsid proteins and/or of the dsRNA structures appeared to be very important for leading to optimal IFN-α/β production, as prolonged UV irradiation of BTV reduced the level of IFN production in sheep lymph cells (see Fig. S2 in the supplemental material).
The requirement for endo-/lysosomal acidification may also indicate that BTV was sensed in pDCs via a TLR-dependent mechanism. Both TLR3 (1
) and TLR7 (29
) can sense dsRNA. However, TLR3 mainly plays a role in epithelial cells for IFN-α/β induction (69
) and it is not expressed by pDCs in mice and humans (42
). In sheep pDCs, poly(I·C) does not induce type I IFN unless it is introduced into the cytosol by lipofection (data not shown), thus excluding TLR3 as an endosomal dsRNA sensor. In addition, we demonstrated that the A151 TLR7/9 inhibitor was not efficient at blocking UV-BTV-induced IFN-α/β, whereas it was efficient at blocking the IFN-α/β production induced by influenza virus. This finding excludes the possibility of both TLR7 and TLR8 involvement in UV-BTV signaling in pDCs.
The lack of TLR7 involvement in IFN-α/β induction in pDCs by an RNA virus as described here is a very rare event. Independence of TLR7 has also been reported for the respiratory syncytial paramyxovirus, which enters and signals via an unknown mechanism in human pDCs after plasma membrane fusion, and not via an endosomal pathway (30
). The other known viral sensors in pDCs are DNA sensors, i.e., TLR9 and the DHX36 helicase, which both signal via MyD88. TLR9 is endosomal, while DHX36 is cytosolic. It could be possible that UV-BTV dsRNA reaches the pDC cytosol after uncoating and triggers (a) cytosolic helicase(s) that remain(s) to be identified. The known RIG-1/MDA-5/LGP2/DDX1 helicases (40
) can all bind various forms of dsRNA, and they all signal via the mitochondrial adaptor MAVS (also named IPS-1/Cardif/VISA) (59
). However, mouse pDCs appear not to rely on MAVS for IFN-α/β induction by viruses (63
). Thus, our results suggest the hypothesis that dsRNA sensors, possibly novel helicases linked to MyD88, are implicated in Reoviridae
-induced IFN-α/β in pDCs.
PKR is the first dsRNA cell sensor that has been described. Its autophosphorylation and dimerization upon dsRNA binding primarily lead to the establishment of an antiviral state, but it can also trigger IFN-α/β synthesis in different cell types, including cDCs (13
), via multiple, complex, and sometimes controversial mechanisms. In human pDCs, the PKR inhibitor 2-AP was found to prevent the IFN-α/β secretion induced by CpG-A (30
), indicating that PKR can integrate signal transduction from TLR in this cell type. Some published reports indicate that PKR can control the expression and activation of IRF3 and IRF1, as well as the activation of NF-κB (18
), thus having an impact on IFN-α/β gene transcription. In the case of rotavirus-infected embryonic fibroblasts, PKR was shown to promote IFN-α/β secretion not by an increase of transcriptional activity but at the posttranscriptional level via a yet-to-be-defined mechanism (60
). Significantly, PKR was recently found to induce IFN-α/β production in coordination with MDA-5 activation by some viruses via the stabilization of IFN-α/β mRNA poly(A) (56
). Thus, PKR can promote IFN-α/β production by many mechanisms, although it has not been studied in pDCs. In the case of BTV in pDCs, PKR activation could be direct, via intracytosolic sensing of dsRNA, or indirect, as for TLR signaling. Our results indicate that UV-BTV increases IFN-α/β synthesis and IFN-α mRNA production by a mechanism independent of stabilization of mRNA polyadenylation. Finally, possibly downstream or independently of PKR, we found that signaling via JNK was implicated in IFN-α/β production induced by UV-BTV in primary pDCs, whereas ERK1/2 signaling was not involved. Adenovirus that triggered IFN-α/β production via TLR and MAVS-independent mechanisms in cDCs also used JNK-dependent and ERK1/2-independent signaling related to sensors that were not identified (15
Altogether, our data show that a dsRNA virus triggers IFN-α/β in primary host pDCs via a novel mechanism that is independent of TLR7/8 but dependent on the MyD88 adaptor. We bring some indications that the PKR and JNK signaling pathways may also be involved. Other members of the Reoviridae
family, such as members of the genera Rotavirus
, may use similar pathways, associated with a restriction of IFN-α/β production in pDCs (47
). Our findings aid in the understanding of BTV pathogenicity and, importantly, will have impacts on the improvement of vaccine development against dsRNA viruses. For example, in order to induce optimal adaptive immune responses, industrial processes of viral inactivation must maintain the ability of BTV and/or other reovirus particles to trigger IFN-α/β production by pDCs.