Poxvirus host tropism is linked to the ability of the host to mount an early and vigorous innate immune response, including the induction of antiviral effectors TNF and type I IFN that can restrict the replication of poxviruses like myxoma virus in a nonpermissive host 1
. Accordingly, successful virus infection and dissemination in a permissive host would rely on either a compromised viral sensing mechanism or a viral strategy to antagonize the host's innate responses. pDCs are potent producers of type I IFN and other early response cytokines like TNF, and play an important role in mediating the antiviral immune responses. The present study shows that human pDCs respond differently to infections by a potentially pathogenic poxvirus (vaccinia) compared to a non-pathogenic poxvirus (myxoma). We report that myxoma virus infection of human pDCs induced IFN-α and TNF production, whereas live vaccinia did not. It has been reported that myxoma virus infection also induces type I IFN and TNF in primary human macrophages 23
. Strikingly, WT vaccinia infection blocks type I IFN/TNF induction in response to myxoma, TLR9 agonist CpG, or TLR7 agonist imiquimod. Heat-VAC, however, gained an ability to induce IFN-α and TNF secretion by pDCs, underscoring the conclusion that untreated live vaccinia introduces inhibitor(s) of poxvirus sensing in human pDCs. Furthermore, genetic studies revealed that Heat-VAC-induced type I IFN induction requires TLR7/MyD88, IRF7 and IFNAR1 in murine pDCs, implying that Heat-VAC infection produces novel RNA species detected by the endosomal RNA sensor TLR7.
Human pDCs express a variety of innate immune sensors, including TLR7 and TLR9. TLR7 is required for the recognition of ssRNA viruses, such as vesicular stomatitis virus and influenza virus 21
. TLR9 is required for detecting herpes simplex, a dsDNA virus 18
. TLR7 and TLR9 play overlapping roles in immunity to herpes virus infection in vivo
. We observed that chloroquine, which blocks endosomal acidification, inhibits IFN-α and TNF induction by myxoma virus or Heat-VAC, which is consistent with our findings that type I IFN induction in murine pDCs by myxoma virus or Heat-VAC is dependent on TLR9/MyD88 or TLR7/MyD88, respectively 15
. A similar genetic analysis is not feasible in human pDCs, because MyD88-deficient human pDCs are not available and transient knockdowns are difficult to achieve in primary pDCs. We suspect that poxvirus nucleic acids, either RNA or DNA, might be sensed by an endosome-localized pathway component. Lee et al. 43
reported that ssRNA virus infection triggers type I IFN production in pDCs via TLR7, which requires the transport of cytosolic viral replication intermediates into the endosome/lysome compartment through autophagy 43
. It is possible that myxoma virus and Heat-VAC can also trigger autophagy upon entry into pDCs, which would make poxvirus nucleic acids more accessible to TLR7 and/or TLR9.
Harper et al. 44
examined the effects of heat-treatment (55°C for 1 h) on vaccinia virion transcription. They found that (i) vaccinia capping enzyme, which is also required for transcription termination, was more sensitive to heat-inactivation than RNA polymerase; (ii) RNA transcripts made by the heat-treated virion cores were longer, suggesting a defect in transcription termination. It is likely that Heat-VAC infection of pDCs produces long, uncapped and partially double-stranded viral RNA transcripts that are sensed by the endosomal RNA sensor TLR7, which utilizes its adaptor MyD88 to activate transcription factor IRF7, resulting in the induction of type I IFN. Such uncapped, partially double-stranded, aberrant RNA transcripts are unlikely to be translated as evidenced by the lack of GFP signal in pDCs infected with Heat-VAC. We have observed that infection of murine primary keratinocytes (KCs) with Heat-VAC induced the production of IFN-β and CCL5 that is dependent on the cytosolic dsRNA sensing pathway mediated by MDA5/MAVS and transcription factor IRF3 (Dai and Deng, unpublished), supporting the viral RNA transcripts might be partially double-stranded.
Using PI3K inhibitor LY294002 and two Akt inhibitors, we also show that PI3K/Akt activation is important for IFN-α and TNF induction in human pDCs by CpG, myxoma virus, and Heat-VAC. This result is consistent with a recent report that PI3K is required for type I IFN production by pDCs in response to TLR stimulation by CpG 27
. Their study did not test whether Akt kinase activity was required, however. We hypothesize that viral RNA or DNA binding by endosomal TLRs leads to activation of PI3K, which subsequently activates Akt through PIP3. How Akt activation leads to IFN-α production is still unclear. It was reported recently that mTOR (a downstream target of Akt) is also involved in the induction of type I IFN by TLR ligands in pDCs 45
Poxviruses employ multiple mechanisms to evade the host antiviral immune systems, including antagonizing the actions of IFN 12
; however, these inhibitory mechanisms can be species-specific, depending on the poxvirus-host pairing. For example, vaccinia produces soluble secreted IFN-binding proteins that prevent type I IFNs from engaging their receptors on target cells 46
. Vaccinia E3 blocks multiple intracellular pathways to attenuate IFN production by immune cells and its effect on target cells 35
. The myxoma M029 protein, a truncated ortholog of E3, possesses the C-terminal dsRBD but lacks the N-terminal ZBD 40
. We observed that the induction of IFN-α and TNF by myxoma virus or Heat-VAC is inhibited by co-infection with untreated WT vaccinia, but only partially attenuated when E3 is absent, or only the E3 dsRBD is produced, thus implicating the N-terminal ZBD of E3 in masking poxvirus infection from sensing by human pDCs.
This cellular response scenario in primary pDCs is different from what we observed in primary keratinocytes. Infection with ΔE3L, but not WT vaccinia or E3LΔ83N, induced a vigorous antiviral innate immune response in murine keratinocytes via MAVS (mitochondrial antiviral signaling protein, an adaptor for cytosolic RNA sensors RIG-I and MDA-5) and transcription factor IRF3 37
. These results indicated that murine keratinocytes sense dsRNAs produced during ΔE3L virus infection via a MAVS/IRF3-dependent signaling pathway that is normally inhibited by the E3 C-terminal dsRBD.
By contrast, this E3 C-terminal dsRBD does not suffice to inhibit poxvirus sensing in human pDCs, whereas the E3 N-terminal ZBD is required. Similar ZBD domains are present in various cellular members of the Zα family of Z-DNA and Z-RNA binding proteins 49
, including dsRNA adenosine deaminase (ADAR1) and mammalian ZBP1, recently re-identified as a cytosolic DNA sensor called DNA-dependent activator of IFN-regulatory factor (DAI) 50
. Both ADAR1 and ZBP1/DAI are interferon-inducible. The crystal structures of the Zα domains of ADAR1, ZBP1/DAI, and Yatapox E3 bound to Z-DNA or Z-RNA revealed similar folds and Z-nucleic acid-binding modes 51
. Indeed, mutant vaccinia viruses in which the E3 ZBD was swapped for the Zα domains of ADAR1 or ZBP1/DAI were as pathogenic as wild-type vaccinia, indicating that the cellular and poxvirus ZBDs are functionally interchangeable 39
. We propose that the N-terminal ZBD domain of E3 might interfere with endosomal TLR sensing of viral nucleic acids possibly through interactions with components of that pathway or through inhibition of the induction of autophagy that allows the transport of viral nucleic acids to the endosomes.
We observed that infection of pDCs with ΔE3L vaccinia virus fails to induce IFN-α and TNF secretion, however, implying that additional inhibitors are produced by the ΔE3L vaccinia virus in human pDCs. For example, vaccinia A46 is a Toll/interleukin-1 receptor (TIR) domain-containing protein that modulates host immune responses. Over-expression of A46 partially blocks IL-1 induced NF-κB activation 54
. A46 interacts with MyD88 and blocks MyD88 signaling 55
. Vaccinia A52 interacts with interleukin-1 receptor-associated kinase 2 (IRAK2) and TNF receptor-associated factor 6 (TRAF6) 56
. Over-expression of A52 inhibits NF-κB activation by IL-1, IL-18, TLR3 and TLR4 54
. We observed that infection with ΔA46R, ΔA52R or ΔA46R ΔA52R alone did not induce the production of IFN-α or TNF (data not shown). Co-infection with these deletion mutants blocked IFN-α or TNF induction in pDCs infected with Heat-VAC to the same extent as co-infection with WT vaccinia (data not shown). We conclude that neither A46 nor A52 is involved in masking the innate cytokine response of human pDCs to vaccinia infection. Other potential inhibitors include vaccinia K7, N1, and B14. Vaccinia K7 is a viral immune modulator that has significant homology to A52 57
. K7 inhibits TLR-mediated NF-κB activation via its interactions with IRAK2 and TRAF6. In addition, it blocks IRF3 and IRF7 activation and IFN-β promoter induction through targeting DEAD box protein 3 (DDX3), an RNA helicase 57
. Vaccinia N1 is another intracellular immunomodulatory protein. N1 inhibits apoptosis, NF-κB and IRF3 activation 59
. Deletion of N1L gene from vaccinia or N1L ortholog from ectromelia virus causes attenuation of the virus 61
. Vaccinia B14 is another virulence factor that targets NF-κB activation through targeting IKKβ 63
. Interestingly, recent structural studies have shown that A52, K7, N1 and B14 have Bcl-2-like folds that might underscore their biological functions 60
In summary, we report a striking difference between myxoma virus and vaccinia in their induction of type I IFN and TNF responses in virus-infected human pDCs, which is likely pertinent to their permissive and restrictive behavior in human hosts. This distinction between the two viruses merits consideration in ongoing efforts to optimize myxoma virus and vaccinia as oncolytic agents for the treatment of human cancer 67
. The novel finding that non-replicating Heat-VAC or live myxoma virus are both potent inducers of an innate immune response in human pDCs has implications for their potential use as immune adjuvants as part of vaccination strategies.