The inflammasome as a target of modulation by DNA viruses

Abstract

The cellular innate immune response represents the initial reaction of a host against infecting pathogens. Host cells detect incoming microbes by way of a large and expanding array of receptors that react with evolutionarily conserved molecular patterns exhibited by microbial intruders. These receptors are responsible for initiating signaling that leads to both transcriptional activation of immunologically important genes as well as protease-dependent processing of cellular proteins. The inflammasome refers to a protein complex that functions as an activation platform for the cysteine protease caspase-1, which then processes inflammatory molecules such as IL-1β and IL-18 into functional forms. Assembly of this complex is triggered following receptor-mediated detection of pathogen-associated molecules. Receptors have been identified that are essential to inflammasome activation in response to numerous molecular patterns including virus-associated molecules such as DNA. In fact, the importance of cytoplasmic DNA as an immune stimulus is exemplified by the existence of at least nine distinct cellular receptors capable of initiating innate reactivity in response to this molecule. Viruses that employ DNA as genomic material include herpesviruses, poxviruses and adenoviruses. Each has been described as capable of inducing inflammasome-mediated activity. Interestingly, however, the cellular molecules responsible for these responses appear to vary according to host species, cell type and even viral strain. Secretion of IL-1β and IL-18 are important components of antimicrobial immunity and, as a result, pathogens have evolved factors to evade or counteract this response. This includes DNA-based viruses, many of which encode multiple redundant counteractive molecules. However, it is clear that such phenotypes are only beginning to be uncovered. The purpose of this review is to describe what is known regarding the activation of inflammasome-mediated processes in response to infection with well-examined families of DNA viruses and to discuss characterized mechanisms of manipulation and neutralization of inflammasome-dependent activity. This review aims to shed light on the biologically important phenomena regarding this virus–host interaction and to highlight key areas where important information is lacking.

Infection of host cells by viruses and other microbial pathogens stimulates rapid intracellular innate immune responses. These are largely initiated by pattern recognition receptors (PRRs) that detect evolutionarily conserved pathogen-associated molecular patterns (PAMPs). Innate immune reactivity leads to initiation of cellular signaling that culminates in the expression or processing of directly antiviral or otherwise immunologically active molecules. These include antiviral effector molecules, interferons, and proinflammatory cytokines and chemokines. Such proteins are essential to the antiviral immune response and are involved directly by creating cellular and tissue environments refractory to virus replication, and indirectly by coordinating adaptive cell-mediated immune processes.

PRR-dependent signaling can lead to the activation of transcription factors including IRF 3 and 7, NF-κB, ATF2 and Jun. These proteins are involved in transcriptional upregulation of numerous immunologically important genes including type I IFNs (IFN-β and IFN-α sub-types), antiviral effectors (e.g., PKR, Viperin and ISG56) and proinflammatory molecules (e.g., TNF-α, IL-6 and IL-8). Stimulation of specific PRRs can also lead to activation of the cysteine protease caspase-1 (previously known as IL-1β-converting enzyme) by way of assembly of a protein complex termed the `inflammasome' [1]. The inflammasome is commonly composed of an individual PRR, an adaptor molecule termed ASC, as well as caspase-1 that, following assembly, is cleaved from an inactive zymogen to an active protease. However, an ASC-independent inflammasome has also been described, in which the PRR is able to directly bind and activate procaspase-1 [2–5]. Active caspase-1 processes IL-1β and IL-18 from their immature forms (pro-IL-1β and pro-IL-18) into secretable molecules. IL-33 has also been shown to be a target of caspase-1 [6], although its function in an antiviral context remains unclear. In some cell types (e.g., undifferentiated monocytes), IL-1β processing first requires an initial `priming' event involving PRR-triggered, NF-κB-dependent expression of pro-IL-1β [7,8], although pro-IL-18 is constitutively expressed [9]. In addition to IL-1β and IL-18 processing, inflammasome-mediated caspase-1 activation is also required for cleavage of other, less well-defined substrates that ultimately lead to `pyro ptosis', a process of caspase-1-mediated apoptosis [10]. As such, inflammasome-associated activities constitute a microbe-triggered innate immune response involved in cell-intrinsic and paracrine-directed antimicrobial activity.

The immunological effects of IL-1β and IL-18 secretion are diverse and entail both innate and adaptive antiviral functions. Both cytokines are involved in activation of immune cells, such as monocytes, macrophages and neutrophils, and control development of CD4⁺ T-cell activity toward Th17 (IL-1β) or Th1 (IL-18) responses [11–13]. IL-1β induces strong pyrogenic and inflammatory effects in numerous tissues [14], and the protein can affect nearly every cell type due to the constitutive expression of the IL-1 receptor (IL-1R) with which it interacts. Once stimulated, IL-1R leads to the activation of NF-κB, p38 and JNK transcription factors that trigger expression of numerous proinflammatory genes and endothelial adhesion molecules facilitating the migration of immune cells to sites of infection [15]. IL-18, by contrast, does not confer pyrogenic activity and signals via a receptor (IL-18R) that exhibits a more restricted expression pattern on Th1 cells, some myeloid cells and intestinal epithelial cells [16,17]. IL-18R signaling triggers activation of transcription factors that are similar to those triggered by IL-1R signaling due to functionally related intra cellular domains of the two receptors. IL-18 strongly contributes to the induction of the Th1 response involving production and secretion of IFN-γ [12]. Organismal regulation of the effects of IL-1β and IL-18 involves at least five levels: control of transcription, mRNA stability, protein processing, secretion and receptor antagonism [17]. As will be discussed below, viral evasion of these cytokines' effects can also involve multiple, redundant mechanisms perhaps acquired evolutionarily from host cells.

The PRRs involved in microbial detection comprise a large and expanding group of molecules [18]. These include cell surface and endosomal proteins, such as Toll-like receptors (TLRs), of which there are ten human and 12 mouse members [19]. TLRs react with diverse ligands including lipopolysaccharide, nucleic acids and glycoproteins. Their stimulation induces activation of IRF3/7 or NF-κB and subsequent expression of antiviral effectors, type I IFNs or proinflammatory cytokines (including pro-IL-1β). Cytoplasmic PRRs include the RIG-I-like receptors, such as LGP2, RIG-I and MDA5, which are involved in sensing dsRNA species and signaling to IRF3 and NF-κB. RIG-I has also been shown to be capable of triggering virus-induced caspase-1 activation [20,21]. Another class of cytosolic receptors, termed nucleotide oligomerization domain-like receptors (NLR), consists of 23 human and at least 34 mouse proteins [22]. These are mostly categorized by their effector domains, which include PYRIN domains (NLRP) and caspase activation and recruitment domains, among others [23]. NLRP1 and NLRP3 (cryopyrin) are known to stimulate inflammasome assembly and caspase-1 activation in response to microbial molecules including anthrax lethal toxin (NLRP1) and muramyl dipeptide, bacterial RNA, poly(I:C), lipopolysaccharide and microbial lipopeptide, as well as numerous nonmicrobial stimuli (NLRP3) [24]. The broad diversity of NLRP3-stimulatory molecular structures suggests that the protein is not a ligand-interacting receptor, but rather a requisite intermediary for upstream sensors.

In recent years a surprising number of PRRs required for the detection of cytoplasmic dsDNA have emerged [25]. These include many that are involved in stimulating production of type-I IFNs such as ZBP1/DAI [26], DDX41 [27], RNA polymerase III [28,29], DHX9 and DHX36 [30] and LRRFIP1 [31]. Two other proteins, AIM2 and γ-IFI16, are dsDNA-sensing hematopoietic interferon-inducible nuclear proteins with a 200-amino acid repeat. These have both been shown to stimulate inflammasome assembly and caspase-1 activation in response to cytoplasmic DNA and infection with large DNA viruses [32–34]. Interestingly, IFI16 appears unique in that it has also been shown to be important for IRF3-mediated induction of type I IFN in response to infection with DNA viruses [35]. The large number of cellular receptors found to be capable of and required for innate immune activation in response to dsDNA overtly highlights the emerging importance of this molecule as a PAMP. Importantly, while DNA-based genomes are universal among cell-based microbial parasites (e.g., bacteria, fungi and eukaryotes), they are employed by only a subset of viral pathogens that exhibit highly diverse morphology and replication strategies themselves.

The focus of this review is a presentation of what is currently known regarding the molecular basis of activation of inflammasome-mediated processes by infection with DNA-based viruses, and the mechanisms used by these agents to impair or otherwise tolerate these responses. We chose to examine DNA viruses for two key reasons. First, as exemplified by the long and expanding list of DNA-associated PRRs, it is clear that detection of microbial DNA, whether genomic or otherwise, is an extremely important sentinel mechanism for innate immune activation and subsequent antiviral responses. Second, DNA viruses (especially large DNA viruses such as herpesviruses and poxviruses) exhibit potent, often redundant strategies for modulating inflammasome-dependent processes. These strategies are only beginning to come to light and, thus, numerous virus-encoded phenotypes directed at inflammasome-mediated activity remain to be characterized. Ultimately, it is our hope that this work provides a useful condensing of the key phenomena, both known and unknown, related to the interaction between the inflammasome and DNA viruses that facilitate and direct future hypothesis formulation and experimental endeavors. We present summaries of our current understanding of the molecular bases of induction and modulation of inflammasome-mediated processes phylogenetically, according to virus family, type and strain to allow evolutionary contrasts to be made evident.

Herpesviruses

The Herpesviridae comprise a highly diverse family of large, enveloped, dsDNA viruses that utilize the host cell nucleus for viral gene expression [36]. Herpesviruses are categorized into α, β and γ subfamilies according to evolutionary history and key aspects of tissue tropism and persistence strategy. Herpesviral infection triggers numerous and diverse innate and adaptive immune responses [37–40], yet despite these processes, all herpesviruses are able to persist in an infected host for life through the use of sophisticated mechanisms of immune evasion [41–43]. Activation and inhibition of the inflammasome by herpesviral species is still evolving as a field of research, yet important discoveries have been made, with implications for understanding viral pathogenesis and immune evasion (Table 1). The following sections will discuss what is currently known regarding the inflammasome-terminal molecular sensors involved in herpesvirus detection and strategies employed to block these responses.

Table 1.

List of virus-specific inflammasome-activating pattern recognition receptors and corresponding virus-encoded inhibitors.

Virus type	Detecting pattern recognition receptor	Inhibitor(s)	Ref.
Herpesvirus
HSV-1	Unknown	Unknown	[44,45,49]
VZV	NLRP3?	Unknown	[45]
MCMV	AIM2	Unknown	[44]
KSHV	IFI16	Orf63	[32,56]
EBV	RIG-I, AIM2	miR-BART15	[64,71]
Poxvirus
VV (WR)	AIM2	B13R (CrmA/SPI-2), B15R (vlL-1βR), C12L (vlL-18BP)	[33,93,105,115]
VV(MVA)	NLRP3	Loss of CrmA/SPI-2; retained vlL-1βR, vlL-18BP	[78]
CPXV	Unknown	CrmA/SPI-2, vlL-1βR,vlL-18BP	[80.106.115]
ECTV	Unknown	CrmA/SPI-2, vlL-1βR,vlL-18BP	[97,107,130]
MCV	Unknown	MC54L(vlL-18BP)	[115]
MYXV	NLRP3	Serp-2 (CrmA/SPI-2), M13L(ASC inhibitor)	[79,98,100]
SFV	Unknown	gp013L(ASC inhibitor)	[101]
Adenovirus
Ad 5	NLRP3	Unknown	[135,136,138]

Ad5: Adenovirus 5; CPXV: Cowpox virus; ECTV: Ectromelia virus; KSHV: Kaposi's sarcoma-associated herpesvirus; MCMV: Mouse CMV; MCV: Molluscum contagiosum virus; MVA: Modified vaccinia virus Ankara; MYXV: Myxoma virus; SFV: Shope fibroma virus; VV: Vaccinia virus strain; VZV: Varicella-zoster virus; WR: Western reserve.

α-herpesviruses

HSV-1 is the prototypic α-herpesvirus member and is responsible for mild herpetic lesions (cold sores), but can also lead to more serious conditions in immunocompromised hosts. Despite its prominence in herpesvirus research efforts, surprisingly little is known regarding HSV-1-mediated activation or inhibition of inflamma-some-associated processes. Two separate groups have investigated HSV-1 in this context and concluded that the virus induces IL-1β processing and secretion in mouse thioglycollate-elicited macrophages [44] or human promonocytic THP-1 cells [45], and that this is an AIM2-independent process. Intriguingly, while IFI16 is known to be important for inflammasome activation by Kaposi's sarcoma-associated herpesvirus (KSHV) [32] (see below), to be essential for HSV-1-induced IFN synthesis [35,46,47] and even to promote resistance to HSV-1 infection in epithelial cells [48], the receptor has yet to be examined for its importance in inflammasome activation by HSV-1. Contrasting with these reports, global secretome analysis of HSV-1-infected human primary macrophages found no caspase-1, IL-1β or IL-18 release, even when cells were primed to express all necessary inflammasome components [49]. These results suggest that HSV-1 infection either does not activate inflammasome function in these cells or does so, but encodes a potent inhibitor(s) that subsequently blocks activation. It is worth noting that the reports mentioned here employ disparate multiplicities of infection (MOI) during HSV-1-induced inflammasome activation analysis ranging from MOI = 40 [44] to MOI = 1 [49]. Thus, the discrepancy in observed phenotype between species, cell lines and experimental procedures implies that further exploration including standardization of HSV-1-associated inflammasome activity is needed.

Varicella-zoster virus (VZV; human herpes virus 3) is also a member of the α-herpesvirus subfamily and is responsible for varicella (chicken-pox) during primary infection and herpes zoster (shingles) following reactivation from viral latency later in life [50]. Nour et al. demonstrated that VZV infection is capable of stimulating caspase-1 activation and IL-1β secretion in three permissive human cell lines (promonocytic THP-1, primary lung fibroblasts and melanoma cells) [45]. Using a cross-linked anticaspase-1 antibody column, the authors showed that VZV infection led to the formation of an inflammasome complex that includes caspase-1, ASC and NLRP3. Furthermore, VZV-mediated inflammasome activation did not involve free radical reactive oxygen species (ROS), described as a microbe-induced trigger for other NLRP3-dependent responses [51]. Increased NLRP3 expression was also demonstrated in VZV-infected human skin xenografts in a SCID mouse model. Importantly, as with HSV-1, VZV-induced IL-1β secretion occurred independently of AIM2, a receptor shown to be necessary for inflammasome activation by other DNA viruses, such as the herpesvirus MCMV and vaccinia virus (VV) (discussed below) [33,44,45]. These results further illustrate the lack of consistency in host cell molecules involved in innate reactions to evolutionarily related viruses, a phenomenon likely associated with patterns of PRR expression across cell types and differences in tissue/cell tropism between related virus types (Table 1).

β-herpesviruses

The β-herpesvirus subfamily includes various CMVs, as well as human herpesvirus (HHV) 6A, 6B and 7. Human CMV (HCMV) is a genomically large (235 kb; ~170 open reading frames) pathogen infecting up to 80% of adults by age 40 in the USA [201]. The virus establishes a latent infection in myeloid progenitors and is generally asymptomatic in immunologically healthy individuals. However, it is a leading infectious cause of developmental abnormalities and can cause serious disease under conditions of immunocompromise [52]. CMVs are highly species-specific and, thus, animal models require infection with CMVs that have coevolved with the model species – a disadvantage when considering highly divergent mammals such as mice and humans. While IL-1β secretion has been observed during monocyte infection with HCMV [53], mouse CMV (MCMV) represents the most studied β-herpesvirus in the context of the molecular basis of inflammasome activation. Infection of macrophages harvested from wild-type mice with MCMV was found to activate caspase-1, as well as IL-1β processing and release [44]. By contrast, inflammasome activation is abolished during MCMV infection of macrophages from AIM2^−/− mice. During MCMV infection in vivo, IL-18 plays an important role since it induces NK cells to produce IFN-γ [54,55]. In mice infected with the virus, the absence of AIM2 significantly diminished the serum concentration of IL-18 and similar results were seen using mice lacking ASC. Moreover, absence of AIM2 and ASC led to a decrease in IFN-γ-producing NK cells in the spleen, determined ex vivo by intracellular staining, as well as a significant increase in splenic MCMV titers [44]. Thus, MCMV-induced inflammasome activity confers important downstream effects on the adaptive antiviral immune response.

To date it has been shown that HCMV induces IL-1β secretion during infection of primary myeloid cells [53]. However, unlike other human herpesviruses [32,45,56], little is known about the molecular basis of this secretion. Interestingly, the inflammasome receptor IFI16 has been investigated in the context of HCMV replication and viral gene expression and somewhat conflicting results are evident. Gariano et al. [57] recently showed that the receptor acts as a restriction factor for viral replication by sequestering a necessary host transcription factor. In contrast, another report demonstrated that IFI16 actually promotes viral replication in association with the viral protein pp65 [58]. While these reports imply different roles for IFI16, they suggest that the receptor might be important for the HCMV life cycle and its effect on inflammasome activation invites further investigation.

γ-herpesviruses

Members of the γ-herpesvirus subfamily are grouped in their potential to transform infected host cells [59]. The most clinically prominent γ-herpesviruses include EBV (HHV-4) and KSHV (HHV-8). EBV primarily infects B cells and is associated with Burkitt's lymphoma, Hodgkin lymphoma and nasopharyngeal carcinoma (NPC) [59]. KSHV is associated with Kaposi's sarcoma, primary effusion lymphoma and multicentric Castleman's disease – conditions that are most prevalent under conditions of immunosuppression such as with AIDS [59]. Animal-associated γ-herpesviruses that are employed as models for human disease and immunology include murine herpesvirus strain 68 and rhesus macaque rhadinovirus.

KSHV was recently shown to activate the inflammasome via the DNA receptor IFI16 [32]. KSHV infection of human endothelial cells was found to trigger the formation of a protein complex containing IFI16, ASC and caspase-1 in the nucleus. This result is the first demonstration of nuclear operation of inflammasome activity, a particularly relevant finding in light of the importance of nuclear localization of herpesvirus replication events. In addition, KSHV genomic DNA was found to colocalize with IFI16 and depletion of this receptor (but not AIM2) eliminated KSHV-induced caspase-1 activation. These results strongly argue that KSHV-associated DNA is responsible for triggering inflammasome-dependent responses via its detection by nuclear IFI16 and is consistent with the high significance of viral DNA as an immunostimulatory molecule.

KSHV exhibits a well-defined latent state involving the expression of few viral proteins [60,61], but during this period the virus downregulates the expression of immune response genes, including proinflammatory cytokines such as IL-1β [62]. Moreover, among herpesviruses, KSHV was the first for which an inflammasome inhibitor was described [56]. Orf63 protein was shown to interact directly with NLRP1 and inhibit its association with and activation of pro-caspase-1. Importantly, when primary human monocytes are infected with KSHV in the presence of shRNA directed against Orf63, an increase in IL-1β secretion and a decrease in lytic viral gene expression (Orf49, Orf50 and Orf57) is observed. In KSHV-infected BCBL-1 primary effusion lymphoma cells, the absence of Orf63 also decreased lytic viral gene expression and virus production. These data indicate that Orf63 is able to inhibit NLRP1-dependent inflammasome function, and that this is critical for viral gene expression and reactivation from latency. Interestingly, the authors also demonstrate that Orf63 can interact with NLRP3 and inhibit inflammasome activation induced by different NLRP3 agonists in THP-1 cells. This observation, along with bioinformatics analysis suggesting that Orf63 is likely not a structural NLR homolog [63], indicates that the protein evolved to behave as an inhibitor of multiple NLR-dependent responses. Importantly, the relationship between IFI16-dependent inflammasome induction by KSHV-and ORF63-mediated inhibition has not been examined [56]. While Orf63 was shown to inhibit NLRP3 inflammasome activation in THP-1 cells [56], NLRP3 was not found to be relevant for KSHV-induced inflammasome activity in endothelial cells [32], although its activity against KSHV in other cell types is not currently known. Thus, many questions remain regarding the exact basis and physiological relevance of Orf63's inhibitory ability. Furthermore, the relevance of NLR inflammasomes to KSHV infection appears to differ between cell types, a phenomenon that remains to be appropriately addressed.

EBV has also been studied in the context of inflammasome modulation. Extensive work in EBV-associated NPC cells has shown that they present upregulation of the inflammasome-related components ASC, caspase-1, IL-1β, AIM2, RIG-1 and NLRP3, and that this correlates with better survival in NPC patients [64]. Importantly, by silencing the different receptors with specific siRNAs, investigators demonstrated that AIM2 and RIG-I activate the inflammasome in NPC cells upon detection of transfected EBV dsDNA and small RNAs, respectively. The involvement of a RIG-I inflammasome is not surprising, since it has been shown that EBV also induces IRF3 activation via RIG-I-mediated detection of virus-encoded RNAs [65,66]. Interestingly, activation of NLRP3 inflammasome by factors from the tumor microenvironment has also been demonstrated [63]. This is the first report implying a role for specific inflammasomes in an EBV-relevant context.

EBV also illustrates a potentially novel mechanism of inflammasome inhibition. Recently, it was shown that cellular miR-223 is capable of controlling NLRP3 levels in a differentiation-dependent manner through post-transcriptional effects targeting the 3′-untranslated region (UTR) of NLRP3 mRNA, and that this can subsequently control inflammasome-dependent processes [67]. The virus is known to express multiple miRNA species [68], many of which are actually released from infected cells in exosomes [69,70]. Recently Haneklaus et al. demonstrated that the EBV-encoded miR-BART15 could target the NLRP3 3′-UTR using the same binding site as miR-223 [71]. In addition, exosomes containing miR-BART15 released from EBV-infected B cells were found to be capable of targeting 3′-UTRs in proximal uninfected cells and even diminishing NLRP3-dependent inflammasome activity in those cells. While additional studies will be needed to characterize the biological importance of this mechanism, exosome-associated miRNA represents a potentially paradigm-shifting viral immune-evasion strategy. This also points to a role for NLRP3 in EBV-induced inflammasome activation.

Poxviruses

Poxviruses comprise an ancient family of closely related viruses that include a linear, dsDNA genome and that infect both vertebrate and invertebrate hosts [72]. The poxvirus genome is generally 150–200 kb, but can be >300 kb, and encodes >200 open reading frames [73]. Within the vertebrate-infecting subfamily Chordopoxvirinae, 90 genes are centrally conserved and comprise those most essential to virus replication [74]. The remaining genes, especially in the 5′ and 3′ flanking ends, encode numerous immunomodulating proteins that facilitate species-specific replication and transmission [75]. Compared with other DNA viruses, poxvirus replication is unusual in that no nuclear localization or processes occur. As such, poxvirus-sensing PRRs trigger signaling that initiates exclusively in the cytoplasm. The most thoroughly examined poxviruses, with respect to innate immune detection and inflammasome interaction, include VV (the agent used as a vaccine to eradicate smallpox), as well as the attenuated modified VV Ankara (MVA; a strain of passaged VV lacking many genes involved in immunomodulation [76,77]), cowpox virus (CPXV) and myxoma virus (MYXV), causative agent of myxomatosis in rabbits.

Inflammasome induction by poxviruses is represented by highly curious observations. While ASC dependence during poxvirus-mediated induction is clearly based on multiple studies, a role for NLRP3 in IL-1β processing and secretion was uncovered in a study using the human promonocytic cell line THP-1 during infection with MVA [78]. This detection of NLRP3-mediated inflammasome activation was likely facilitated by the absence of many immunomodulatory proteins in the MVA vaccine strain compared with other VV isolates (Table 1). NLRP3 was also shown to be necessary for inflammasome activation in these cells by a MYXV mutant lacking the ASC inhibitor M13L and to be dependent on infection-associated ROS and cathepsin B [79]. Unfortunately, these studies did not examine potential roles for DNA-associated PRRs such as AIM2 and IFI16. Conversely, using the virulent VV strain Western Reserve (WR), Hornung and colleagues observed normal caspase-1 activation following infection of macrophages harvested from NLRP3^−/− mice [33]. However, AIM2 and ASC were found to be necessary for IL-1β secretion and caspase-1 activation by this strain. Thus, it is possible that the relative contributions of inflammasome components are influenced by poxvirus strain (especially regarding the presence or absence of immunomodulatory genes, for instance when comparing the highly attenuated VV strain MVA to a virulent VV isolate like WR), infected cell type or species, and this requires more thorough investigation.

Poxviral serine protease inhibitors impair caspase-1 activity

The first virally encoded protein described to interfere with constituents of the inflammasome was actually discovered in CPXV. CrmA is an abundant 38-kDa protein expressed early in infection with sequence homology to cellular serine protease inhibitors (serpins) [80]. CrmA directly inhibits the enzymatic activity of caspase-1 (as well as other caspases) [81–84] by binding and behaving as a pseudosubstrate for protease activity [85]. Interestingly, as a cysteine proteinase, caspase-1 inhibition by CrmA represents one of the first described cross-class inhibitors within the serpin family [85,86]. CrmA-mediated caspase-1 inhibition has also been shown to be effective at preventing pyroptosis [87–90]. Viral serpin-like inhibitors were also identified in other poxvirus species such as VV and ectromelia virus and were subsequently termed serpin-like protease inhibitors (SPIs). Interestingly, CPXV and VV encode multiple SPIs exhibiting different functions during infection [91]. In CPXV, only the deletion of CrmA (SPI-2) led to attenuation in mice, whereas deletion of any other serpin gene resulted in a wild-type phenotype [92]. This suggests that CrmA/SPI-2 is the dominant serpin involved in immune evasion in vivo. Moreover, the CPXV model demonstrates a connection between the efficiency of inflammasome inhibition in relation to the absence of an immune evasion gene and loss of virulence, underscoring the in vivo importance of inflammation control by the virus. By contrast, deletion of either SPI-1 or SPI-2 in VV both led to wild-type phenotypes in vitro and in vivo, showing no signs of attenuation in the mouse [93]. In recent years, multiple studies have been conducted to elucidate the effects of CrmA deletion in VV, CPXV or rabbitpox virus in vivo using mice, and most of the resulting data showed attenuation of the deletion mutants compared with the wild-type [92,94,95], although other studies indicated no in vivo effect of the deletion [93] or even increased virulence in the absence of the viral serpin [96]. These substantial differences in outcome can most likely be explained by the different routes of infection used, ranging from subcutaneous and intraperitoneal to intratracheal infections, and the different mouse strains used by the separate groups. In cases where in vivo attenuation was observed, this was probably the result of enhanced viral clearance via increased inflammatory responses and an influx of inflammatory cells at the site of replication [92]. An aspect all these studies have in common is that the virus species probably does not use the mouse as its natural host. When ectromelia virus (also known as mousepox virus and a natural pathogen of mice) was deleted of its CrmA homolog, subcutaneous infection of BALB/c mice showed severe attenuation of the mutant [97]. In the absence of the viral serpin, more effective inflammasome signaling could be observed, leading to increased secretion of IL-18 and, subsequently, to increased NK-cell activation. Additionally, in MYXV, deletion of Serp-2 (MYXV homolog of CrmA) resulted in severe attenuation in vivo, as indicated by lack of secondary lesions in infected rabbits and a substantially higher survival rate. This also led to increased inflammation, with an enhanced influx of mononuclear cells into infected primary lesions resulting in an improved virus control and clearance [98]. In addition, rabbits infected with the deletion mutant showed a high degree of lymphodepletion due to increased apoptosis of lymphocytes in lymph nodes in the infected animals, underscoring a secondary function of the viral serpin in blocking the initiation of pyroptosis through caspase-1.

PYD-protein family member, M13L, acts as an ASC-1 inhibitor

MYXV also encodes the first viral protein identified as an inhibitor of ASC-dependent processes. The M13L gene product was first described when the complete MYXV genome was sequenced as a viral protein with low homology to the mouse interferon-inducible protein 203 [99], but was later shown to be more closely related to PYRIN domain-containing proteins [100]. M13L was found to interact with ASC and, thereby, inhibit its function and prevent virus-induced caspase-1 activation as well as IL-1β and IL-18 secretion [100]. In vivo experiments in rabbits infected with a mutant lacking M13L-mediated ASC inhibition demonstrated strong attenuation compared with wild-type virus [100]. Here, the mutant virus was no longer able to control the inflammation induced by infection and the increased secretion of proinflammatory cytokines led to enhanced killing of virus-infected cells and viral control. Dissemination of virus to secondary sites likely involves lymphatic mobility of infected leukocytes, and M13L deletion led to diminished viral spread due to low productive infection of monocytes and lymphocytes. Interestingly, PYRIN domain-containing proteins have only been described in the genera Leporipoxvirus, Yatapoxvirus and Suipoxvirus [101]. The poxvirus shope fibroma virus encodes gp013L, a PYRIN domain-only protein capable of blocking ASC-dependent activity likely through competitive PYD–PYD binding [101]. Intriguingly, cellular PYRIN domain-only proteins exist and are capable of regulating ASC function [102], perhaps representing the evolutionary source of virus-encoded ASC effectors. It is also highly likely that additional PYD-containing immunomodulators will be characterized.

A virally encoded soluble IL-1β receptor abrogates IL-1β signaling

It is worth noting that genes encoding secreted neutralizing inhibitors of IL-1β and IL-18 have been characterized in poxviral genomes. While not exhibiting anti-inflammasome effects in the technical sense via direct inhibition of inflammasome components, these molecules effectively impair inflammasome-dependent physiological responses by impairing the downstream effects of inflammasome activation and, thus, likely evolved to counteract IL-1β- and IL-18- mediated antiviral functions. VV WR genes B15R and B18R encode immunoglobulin superfamily members with high homology to human and murine IL-1R [103,104]. When supernatants from VV-infected cells were examined for the ability to bind murine IL-1β, a secreted 33-kDa protein corresponding to B15R was identified [105]. These supernatants were also capable of inhibiting the proliferative response of B and T lymphocytes to IL-1β in vitro. Functional homologs of this protein have so far been described for CPXV and ectromelia virus [105–107]. Intracranial infection of weanling mice showed substantially increased LD₅₀ for the B15R-knockout mutant, thus implicating this immune evasion function as important for virus-induced pathogenesis [105]. Contrastingly, intranasal infection of mice with VV lacking B15R resulted in significantly increased virulence [106]. This result might indicate a relationship between inflammation (otherwise controlled by the virus) and pathological effects mediated by excessive IL-1β production. In fact, IL-1β was shown to be a key inducer of fever, and the appearance of fever as well as increased weight loss are major differences between infection with VV strains that encode or lack the viral IL-1R [108].

Soluble poxvirus IL-18-binding proteins mimic host proteins in modulating the host antiviral response

In addition to IL-1β, poxviruses also counteract secreted IL-18 [109], which is important for NK- and T-cell stimulation, as well as Th1 development [12,110]. When the full genome sequence of the molluscum contagiosum virus (MCV; genus Molluscipoxvirus) was determined and analyzed, three viral homologs (MC51L, MC53L and MC54L) to cellular IL-18-binding proteins (IL-18BPs) were discovered [111]. Host IL-18BPs regulate IL-18 activity by efficiently competing with the IL-18 receptor for cytokine binding [112,113], although they show no significant homology to either IL-18 receptor subunit [112]. The isolated MCV IL-18BPs, MC53L and MC54L, showed murine IL-18 binding affinities similar to their cellular homologs [114], yet further experiments demonstrated that only MC54L-bound human IL-18, indicating that it was the only functional viral IL-18BP [115]. In spite of low overall levels of protein homology, MC54L and cellular IL-18BPs share similar IL-18 binding sites [115–117], and experiments with VV and ectromelia virus IL-18BPs revealed competition between the viral protein and the cellular IL-18 receptor due to an overlap of IL-18 binding sites [118–120]. Viral IL-18BPs are maintained in many pox viruses [121–124] and functionality has been shown for several of them [125–129]. In vivo, an ectromelia virus IL18BP deletion mutant was clearly attenuated and virus titers in tissues of infected mice were decreased relative to wild-type due to NK-cell- and cytotoxic T-lymphocyte-mediated control of virus infection [130]. Similarly, infection of BALB/c mice by a VV mutant lacking the IL-18BP C12L led to a generally weaker infection compared with the wild-type, with milder signs of illness, loss of body weight and reduced mortality [128]. The basis of this attenuation was found to be elevated levels of secreted IFN-γ due to abundant functional IL-18 and, as a direct consequence, heightened NK-cell and VV-specific cytotoxic T-lymphocyte killing during viral infection [131]. In light of these results, it is possible that viral proteins directed against the secreted products of inflammasome activation actually confer the majority of anti-inflammatory activity. Overall, these results indicate the importance of inhibiting IL-18-dependent effects for poxviral in vivo replication and pathogenesis.

Adenoviruses

Adenoviruses (Ads) are nonenveloped, dsDNA viruses that consist of seven subgroups, A–G, depending on their hemagglutination properties and sequence homology [132]. These viruses possess characteristics, such as manipulability and ability to infect dividing and nondividing cells, that make them suitable as vectors for ectopic gene therapy strategies, and which are a key reason they have been broadly studied with respect to their innate immune properties [133,134]. While no studies to date have described the inhibition of inflammasome-mediated activities by Ad, activation of the inflammasome by Ad molecules and virus-induced processes has been well-studied.

Several reports support a role for NLRP3 during Ad-induced inflammasome activation [135–138]. Muruve et al. first demonstrated that infection with Ad serotypes 3 and 5 activates inflammasome-mediated IL-1β processing in human THP-1 cells, and that this response also occurs when Ad DNA is transfected into these cells, but not when DNA-free viral capsids are used as inoculum – implicating virus-associated DNA as a pivotal PAMP [138]. Using NLRP3^−/− mice, it was shown that this molecule has an essential role in IL-1β secretion during Ad infection in vitro and in vivo. However, the absence of NLRP3 had no effect on general DNA-induced IL-1β processing (although Ad DNA specifically was not investigated in this context). Ad-induced NLRP3 inflammasome activation was then hypothesized to be mediated by endosomal membrane disruption generated by the virus during cytoplasmic entry. The NLRP3 inflammasome can be activated by endogenous danger signals, including lysosomal membrane damage and associated release of cathepsin B and ROS [139–142]. Barlan and colleagues demonstrated that Ad-induced IL-1β secretion was absent when cells were infected with an Ad mutant incapable of rupturing endosomal membranes [136]. Further more, chemical inhibition of cathepsin B or ROS also prevented Ad-induced IL-1β secretion. Surprisingly, the NF-κβ-terminal DNA receptor TLR9 was found to be important to prime expression of inflammasome constituents NLRP3 and pro-IL-1β for Ad-induced activation in human but not mouse macrophage cells [136,138]. These studies highlight important realities regarding investigation into innate viral immune responses. First, the role of specific PAMPs (e.g., virus-associated DNA) may not be as relevant to innate immune induction as cell-derived danger-associated molecular patterns that appear as a physiological outcome of infection. This phenomenon is exemplified by the predominance of NLRP3 inflammasome activation following destabilization of cell membranes during Ad infection. Second, the molecules detected by specific PRRs may differ between species and cell types, and this should be taken into account when studying PRR-mediated processes.

Conclusion

Caspase-1-mediated processing and secretion of inflammatory cytokines represents an essential component of the antiviral immune response. Infection with diverse DNA virus types leads to inflammasome-mediated activities, such as IL-1β and IL-18 secretion, as well as pyroptosis. Intriguingly, evolutionarily and physiologically related virus types appear to trigger inflammasome activation via divergent receptors and processes, as in the case of herpesviral species (Table 1). Curiously, the importance of DNA sensors for inflammasome activation by related viruses (e.g., KSHV and MCMV) can vary significantly and is likely to be affected by infected cell type and PRR expression patterns therein. Moreover, the importance of NLRP3 to inflammasome activation owing to numerous and varied stimuli is perhaps an indication that this molecule is less likely to represent a PRR, but rather an adaptor for as-yet unidentified PRRs. The importance of inflammasome-mediated antiviral activity is clearly underscored by the diversity of viral phenotypes that have evolved to counteract the activation or physiological effects of the inflammasome (Table 1). This includes proteins directed at inflammasome assembly (KSHV Orf63; MYXV M13L), caspase-1 activity (poxvirus CrmA/SPI-2), IL-1β/IL-18 function (poxvirus MC54L, B15R), or even PRR mRNA stability (EBV miR-BART15). Given the diversity of viral systems and genes yet to be examined, it remains probable that additional, likely novel inhibitory genes will be identified.

Future perspective

Numerous questions remain regarding virus-associated stimulation and manipulation of inflammasome-mediated processes. For example, in the event that ligand-binding properties continue to be undetected for NLRP3, it will be necessary to identify the PRRs that specifically detect viruses such as Ad. In addition, the respective roles of NLRP3 and DNA sensors in innate immune stimulation by VV strains require greater elucidation, including the exploration of potentially contradictory observations, as described here. In addition, the importance of virus strain, host species and cell type to virus-induced inflammasome activation and function remains an important area with multiple questions yet to be answered. Ultimately, fundamental knowledge regarding the importance of viral infection in chronic inflammatory diseases, such as atherosclerosis, diabetes, cancers and auto-immune disorders, remains a largely unexplored area of research. Given the ability of viruses that exhibit chronic infection (e.g., herpesviruses) to stimulate inflammasome-mediated responses, it appears likely that this will play a role in diseases that involve inflammatory dysfunction. Characterizing the molecular basis of virus-induced inflammasome activity and the viral factors capable of impairing or manipulating these responses could lead to the identification of new therapeutic targets for multiple, seemingly unrelated diseases.

Executive summary

• ■The inflammasome is an intracellular, innate immune apparatus that is activated by receptor-mediated detection of pathogen-associated molecular patterns.
• ■The inflammasome is responsible for caspase-1-dependent processing of proinflammatory cytokines IL-1β and IL-18, in addition to less-well-characterized substrates that trigger pyroptosis.
• ■Inflammasome activity directly affects the adaptive immune response to viral infection and is important for viral clearance.
• ■Multiple pattern recognition receptors are known that are involved in infection-triggered initiation of innate immune responses, such as inflammasome activation.
• ■A large and expanding repertoire of DNA-sensing pattern recognition receptors exist that are involved in both transcriptional activation and processing of antiviral cytokines.
• ■Inflammasome activation in response to herpesvirus species is dependent on multiple virus-specific receptors, including IFI16, AIM2 and NLRP3.
• ■Thus far, the described inhibition of inflammasome-dependent activity by herpesviruses includes targeting of the inflammasome receptors NLRP1 and NLRP3 (Kaposi's sarcoma-associated herpesvirus Orf63) and reduced expression of the NLRP3 mRNA (EBV miBART15). It is highly likely that other herpesvirus-encoded inhibitory phenotypes will be identified.
• ■Vaccinia virus (a poxvirus)-mediated inflammasome activation requires either AIM2 or NLRP3, depending on the viral strain or host species. The molecular bases of these disparities require elucidation.
• ■Poxviruses encode multiple, redundant inflammasome-directed inhibitory phenotypes, including those targeting caspase-1 function (CrmA, serpin-like protease inhibitors), ASC-mediated assembly (PYRIN domain-only proteins) or soluble binding proteins of IL-1β and IL-18 (B15R and MC54L, respectively).
• ■Redundant inhibition of inflammasome-mediated physiological effects are essential for virus replication, but in some cases can lead to increased pathogenesis, as exemplified in animal models using deletion mutant viruses.
• ■Adenoviruses induce NLRP3-dependent inflammasome activation.
• ■While adenoviral DNA is capable of triggering inflammasome activation, infection-associated activation involves reactive oxygen species and cathepsin B release in response to endosomal rupture.
• ■The role of viral DNA in inflammasome activation requires more thorough investigation, with consideration of the potential role of infection-associated cellular stress responses.
• ■The importance of infection-dependent inflammasome activity in chronic inflammatory disease remains an important area of inquiry.