Role of the DNA Binding Activity of Herpes Simplex Virus 1 VP22 in Evading AIM2-Dependent Inflammasome Activation Induced by the Virus

VP22, a major component of the HSV-1 virion tegument, is conserved in alphaherpesviruses and has structural similarity to ORF52, a component of the virion tegument that is well conserved in gammaherpesviruses. Although the potential DNA binding activity of VP22 was discovered decades ago, its significance in the HSV-1 life cycle is poorly understood.

ABSTRACT

AIM2 is a cytosolic DNA sensor of the inflammasome, which induces critical innate immune responses against various invading pathogens. Earlier biochemical studies showed that the binding of AIM2 to DNA triggered the self-oligomerization of AIM2, which is essential for AIM2 inflammasome activation. We recently reported that VP22, a virion tegument protein of herpes simplex virus 1 (HSV-1), inhibited activation of the AIM2 inflammasome in HSV-1-infected cells by preventing AIM2 oligomerization. VP22 binds nonspecifically to DNA; however, its role in HSV-1 replication is unclear. We investigated the role of VP22 DNA binding activity in the VP22-mediated inhibition of AIM2 inflammasome activation. We identified a VP22 domain comprising amino acids 227 to 258 as the minimal domain required for its binding to DNA in vitro. Consecutive alanine substitutions in this domain substantially impaired the DNA binding activity of VP22 in vitro and attenuated the inhibitory effect of VP22 on AIM2 inflammasome activation in an AIM2 inflammasome reconstitution system. The inhibitory effect of VP22 on AIM2 inflammasome activation was completely abolished in macrophages infected with a recombinant virus harboring VP22 with one of the consecutive alanine substitutions, similar to the effect of a VP22-null mutant virus. These results suggest that the DNA binding activity of VP22 is critical for VP22-mediated AIM2 inflammasome activation in HSV-1-infected cells.

IMPORTANCE VP22, a major component of the HSV-1 virion tegument, is conserved in alphaherpesviruses and has structural similarity to ORF52, a component of the virion tegument that is well conserved in gammaherpesviruses. Although the potential DNA binding activity of VP22 was discovered decades ago, its significance in the HSV-1 life cycle is poorly understood. Here, we show that the DNA binding activity of VP22 is critical for the inhibition of AIM2 inflammasome activation induced in HSV-1-infected cells. This is the first report to show a role for the DNA binding activity of VP22 in the HSV-1 life cycle, allowing the virus to evade AIM2 inflammasome activation, which is critical for its replication in vivo.

INTRODUCTION

Herpes simplex virus 1 (HSV-1) is a large DNA virus and a member of the Alphaherpesvirinae subfamily of the Herpesviridae family (1). This virus causes a variety of human diseases, such as herpes labialis, genital herpes, encephalitis, herpetic whitlow, and keratitis (1). The HSV-1 virion contains a linear double-stranded DNA (dsDNA) viral genome in an icosahedral capsid, which is enclosed by a viral envelope (1). A unique structure termed the tegument, a proteinaceous layer consisting of at least 18 different viral proteins, is present between the envelope and nucleocapsid (1). Immediately upon HSV-1 entry into a host cell, tegument proteins are released into the cytoplasm to establish an environment suitable for effective viral infection, including the initiation of viral transcription, preclusion of host protein synthesis, and evasion of innate immune responses (1–5).

HSV-1 is recognized by various pattern recognition receptors (PRRs) in infected cells that induce innate immune responses, including the induction of type I interferon (IFN-I) and the activation of inflammasomes (5–7). However, HSV-1 has evolved various strategies to evade the PRR-mediated induction of host innate immune responses (7), many of which are thought to involve the tegument proteins. To date, more than 10 HSV-1 tegument proteins have been reported to inhibit the activation of various host innate immune responses against HSV-1 (7). PRRs sense components of incoming virions upon HSV-1 infection; therefore, successful infection requires that HSV-1 inhibits host innate immune responses induced by PRRs prior to the de novo synthesis of viral proteins that antagonize these responses. This is thought to involve the immediate release of tegument proteins into the cytoplasm upon viral entry. However, to date, only two HSV-1 proteins in the virion tegument—virion host shutoff (VHS) and VP22—have been demonstrated to inhibit the induction of innate immune responses against HSV-1. VHS inhibits the induction of proinflammatory cytokines interleukin-6 (IL-6) and tumor necrosis factor alpha via NF-κB in dendritic cells (4), whereas VP22 inhibits the activation of AIM2 inflammasomes in macrophages (5).

VP22 is a major HSV-1 protein in the virion tegument—each virion contains approximately 2,000 copies—and its amino acid sequence is conserved only in the Alphaherpesvirinae subfamily (8). VP22 forms multimers, is thought to be nucleotidylylated by casein kinase II (CKII) (9–11), and is phosphorylated by CKII and UL13, an HSV-1 protein kinase (12, 13). VP22 release from virions is dependent on its phosphorylation status (14). Phenotypical studies using HSV-1 VP22 mutant viruses demonstrated that VP22 promotes viral replication, cell-to-cell spread in cell cultures, and viral replication and pathogenicity in murine models of HSV-1 infection (15–17). Furthermore, VP22 interacts with chromatin, cellular membranes, microtubules, DNA, and various viral and cellular proteins, suggesting that it has multiple functions related to the virus life cycle, including gene expression, protein synthesis, virion assembly, evasion of innate immune responses, and cell cycle regulation (5, 15, 18–28). However, information on the direct linkage between these potential VP22 functions and VP22-mediated promotion of viral replication and pathogenicity is limited.

The proinflammatory cytokines IL-1β and IL-18, secreted from cells infected with pathogens via the activation of cytosolic multiprotein complexes, termed inflammasomes, mediate innate immune responses against pathogens, including HSV-1 (29–32). Inflammasomes are activated by pathogenic factors recognized by sensor proteins, which undergo oligomerization to recruit pro-caspase-1 via an adaptor protein, ASC, leading to its autocleavage to yield an active caspase-1 p10/p20 tetramer. Activated caspase-1 then cleaves the inactive pro forms of IL-1β and IL-18 to their active forms (33). In this study, we focused on AIM2, a cytosolic DNA sensor of the AIM2 inflammasome. AIM2 consists of an amino-terminal pyrin domain (PYD) that recruits ASC and a carboxy-terminal hematopoietic, interferon-inducible, nuclear localization (HIN) domain responsible for the binding of AIM2 to DNA (34). Previous biochemical studies suggested that the binding of AIM2, via its HIN domain, to DNA induced self-oligomerization, which is an essential step for AIM2 inflammasome activation (5, 35).

Recently, we reported that immune evasion in a murine model of HSV-1 infection was mediated by HSV-1 VP22 in the virion tegument and, possibly, de novo synthesized VP22, which inhibited AIM2 inflammasome activation in infected cells, allowing efficient viral replication (5). Mechanistically, HSV-1 VP22 interacted with the HIN domain of AIM2 to prevent its oligomerization (5). However, the specific mechanism by which VP22 inhibits AIM2 inflammasome activation, i.e., how VP22-AIM2 interactions and/or DNA binding activity of VP22 are involved in the inhibition, remains to be elucidated. To gain further insights into the mechanism of the VP22-mediated evasion of AIM2 inflammasome activation, we mapped the domains of VP22 required for its interaction with AIM2 and/or DNA and examined the effects of a mutation(s) in each of the identified domains on VP22-mediated inhibition of AIM2 inflammasome activation.

RESULTS

Fine mapping of VP22 regions required for interactions with AIM2.

We previously reported that the VP22 domain comprising amino acids 227 to 267 (VP22 domain:227-267) was required for its interaction with the HIN domain of AIM2 (5). To fine map the VP22 amino acid(s) required for interactions with the HIN domain of AIM2, we constructed a series of mutants in a plasmid expressing Flag-tagged VP22 (Flag-VP22) where successive blocks of 3 or 4 amino acids in the VP22 domain:227-267 were replaced with alanines (Fig. 1A and B). These were then tested in glutathione S-transferase (GST) pulldown experiments. 293FT cells were transfected with each plasmid expressing Flag-VP22 and its mutants, lysed, and reacted with the purified HIN domain of AIM2 fused to GST (GST-AIM2-HIN) immobilized on glutathione-Sepharose beads (5). After extensive washing, the beads were subjected to immunoblotting with anti-Flag antibody. In agreement with our earlier study (5), wild-type Flag-VP22 was efficiently pulled down by GST-AIM2-HIN but not by GST alone (Fig. 1C). The expression levels of VP22 mutants varied, and all Flag-VP22 mutants except Flag-VP22:243-246A were pulled down by GST-AIM2-HIN at levels comparable to those of the inputs of the mutants, as observed for wild-type Flag-VP22 (Fig. 1C). In contrast, the Flag-VP22:243-246A mutant was pulled down by GST-AIM2-HIN at a level lower than that of the input of the mutant (Fig. 1C). These results indicate that VP22 amino acids 243 to 246 were required for the efficient interaction of VP22 with the AIM2 HIN domain in the GST pulldown experiments. We should note that Flag-VP22:243-246A and Flag-VP22:247-250A mutants showed apparently lower solubility and/or stability than wild-type Flag-VP22 when transiently expressed in 293FT cells (Fig. 1C). We constructed other mutants in the plasmid encoding Flag-VP22 in which each of the VP22 amino acids 243 to 246 was replaced with alanine (Fig. 2A) and tested them in the GST pulldown experiments (Fig. 2B). The expression levels of wild-type Flag-VP22 and its mutants were comparable, and the three Flag-VP22 mutants (Flag-VP22:243A, Flag-VP22:244A, and Flag-VP22:246A) were pulled down by GST-AIM2-HIN with an efficiency similar to that of wild-type Flag-VP22 (Fig. 2B). In contrast, the Flag-VP22:245A mutant was pulled down by GST-AIM2-HIN less efficiently than wild-type Flag-VP22 (Fig. 2B). These results indicate that a valine at position 245 (Val-245) in VP22 was required for the efficient interaction of VP22 with the HIN domain of AIM2 in the GST pulldown experiments.

FIG 1.

Mapping of VP22 domains required for VP22 interactions with AIM2-HIN. (A) Top panel, crystal structure of the conserved domain of VP22 consisting of three α-helices (α1 to α3) and one β-strand (β1) (39). Bottom panel, schematic diagrams showing the alanine substitution mutants of Flag-VP22 used in Fig. 1C, 3D, and 4B. (B) Mutation sites in the crystal structure of the conserved domain of VP22 are shown in red. (C) GST or GST-AIM2-HIN immobilized on glutathione-Sepharose beads were reacted with lysates of 293FT cells transfected with an expression plasmid encoding each of the indicated alanine substitution mutants of Flag-VP22 described in panel A. The beads were analyzed by immunoblotting, electrophoretically separated in a denaturing gel, and stained with Coomassie brilliant blue (CBB). The data are representative of three independent experiments.

FIG 2.

Mapping of VP22 amino acids required for VP22 interactions with AIM2-HIN. (A) Top panel, schematic diagrams showing the alanine substitution mutants of Flag-VP22 used in panel B. Bottom panel, mutation sites in the crystal structure of the conserved domain of VP22 (39) shown in red. (B) GST or GST-AIM2-HIN immobilized on glutathione-Sepharose beads was reacted with lysates of 293FT cells transfected with an expression plasmid encoding each of the alanine substitution mutants of Flag-VP22. The beads were then processed as described in the legend to Fig. 1C. (C) Lysates of 293FT cells transfected with an expression plasmid encoding wild-type (WT) or the 245A mutant of Flag-VP22 were reacted with Benzonase at 25°C for 1 h. The cell lysates were then reacted with GST or GST-AIM2-HIN immobilized on glutathione-Sepharose beads. The beads were processed as described in the legend to Fig. 1C. (D) The amount of WT or 245A mutant of Flag-VP22 pulled down by GST or GST-AIM2-HIN shown in the top of panel C relative to those of the input signal. In panels B and C, the data are representative of three independent experiments. In panel D, each value is the mean ± standard error of the results of three independent experiments. Statistical analysis was performed by one-way analysis of variance (ANOVA) and Tukey’s test. ns, not statistically significant; *, P < 0.05; ***, P < 0.001.

Effect of DNA on interactions between VP22 and AIM2.

As described above, the HIN domain of AIM2 and VP22 showed DNA binding activity (21, 34). Therefore, VP22 might bind to the HIN domain of AIM2 by a DNA-dependent mechanism. To test this, we treated lysates of transfected 293FT cells with Benzonase, a nonspecific endonuclease for single- and double-stranded DNA and RNA, in GST pulldown experiments. In agreement with the result shown in Fig. 2B, the amounts of Flag-VP22:245A mutant pulled down by GST-AIM2-HIN were significantly lower than those of wild-type Flag-VP22 pulled down by GST-AIM2-HIN without Benzonase treatment (Fig. 2C and D). As shown in Fig. 2C and D, treatment with Benzonase had little effect on the pulldown efficiency of wild-type Flag-VP22 and the Flag-VP22:245A mutant by GST-AIM2-HIN. In contrast, Benzonase treatment abolished the binding of wild-type Flag-VP22 to DNA in the DNA pulldown experiments described below (see Fig. 3B). These results suggested that VP22 interacts with the HIN domain of AIM2 independent of any nucleic acids.

FIG 3.

Mapping of VP22 regions responsible for DNA binding. (A) Schematic diagrams showing the truncated mutants of Flag-VP22 used in panel C. (B) Lysates of 293FT cells transfected with an expression plasmid Flag-VP22 (wt) were reacted with 250 nM biotinylated dsDNA in the presence or absence of Benzonase. Streptavidin-Sepharose beads were then added and rotated at 4°C for 1.5 h. After extensive washing, the beads were analyzed by immunoblotting. (C, D, and E) A 250 nM concentration of biotinylated dsDNA (bio-dsDNA) was reacted with lysates of 293FT cells transfected with an expression plasmid encoding each of the indicated truncated mutants (C), alanine substitution mutants (D), or the WT and the 245A mutant (E) of Flag-VP22. Streptavidin-Sepharose beads were then added and rotated at 4°C for 1.5 h. After extensive washing, the beads were analyzed by immunoblotting. In panels B, C, D, and E, the data are representative of three independent experiments.

Mapping of VP22 regions required for interactions with DNA.

To map potential VP22 domains required for the binding of VP22 to DNA, we established a DNA pulldown system. Cell lysates of 293FT cells transfected with a plasmid expressing Flag-VP22 were incubated with biotinylated 80-mer DNA with a randomized sequence, which was then pulled down by streptavidin Sepharose beads. After extensive washing, the beads were subjected to immunoblotting with anti-Flag antibody. As shown in Fig. 3B and C, Flag-VP22 was efficiently pulled down by streptavidin Sepharose beads only in the presence of the biotinylated DNA. Furthermore, Benzonase treatment abolished the binding of Flag-VP22 to DNA. These results indicate that this system can be used to evaluate the binding of VP22 to DNA. We then tested plasmids encoding a series of 5′ and 3′ sequential deletion mutants of Flag-VP22 (Fig. 3A) (5) in the DNA pulldown experiments. As shown in Fig. 3C, all 5′ deletion mutants of Flag-VP22 and a Flag-VP22 mutant with a deletion of 34 residues from the carboxyl terminus of VP22 were pulled down by biotinylated DNA immobilized on streptavidin Sepharose beads at levels comparable to those of wild-type Flag-VP22. In contrast, Flag-VP22 mutants with deletions of 75, 109, and 141 residues from the carboxyl terminus of VP22 were barely pulled down by biotinylated DNA immobilized on streptavidin Sepharose beads (Fig. 3C). These results indicate that VP22 domain:227-267 was required for the binding of VP22 to DNA.

To further map VP22 amino acids in VP22 domain:227-267 required for VP22 binding to DNA, we tested the series of plasmids encoding Flag-VP22 mutants shown in Fig. 1A in the DNA pulldown experiments. Among the 10 Flag-VP22 mutants tested, only Flag-VP22:259-262A and Flag-VP22:263-266A were pulled down at levels comparable to wild-type Flag-VP22 (Fig. 3D). In contrast, other mutants of Flag-VP22 (Flag-VP22:227-230A, :231-234A, :235-238A, :239-242A, :243-246A, :247-250A, :251-254A, and :256-258A) were barely pulled down (Fig. 3D). We also tested the Flag-VP22:245A mutant, which had impaired binding ability to GST-AIM2-HIN (Fig. 2B to D), in the DNA pulldown experiments. As shown in Fig. 3E, the Flag-VP22:245A mutant was not pulled down. These results suggest that the integrity of the higher-order structure in the VP22 domain comprising amino acids 227 to 258 was required for the VP22 binding to DNA.

Effect of successive alanine substitutions in the VP22 domain required for VP22 binding to DNA on VP22-mediated inhibition of AIM2 inflammasome activation.

We next investigated the relationship between the inhibitory effect of VP22 on AIM2-dependent inflammasome activation and the DNA-binding activity of VP22 or its interaction with AIM2. To this end, we examined the effect of each of the successive alanine substitutions in VP22 domain:227-267 on VP22-mediated inhibition of AIM2 inflammasome activation in an AIM2 reconstitution system. In this system, the AIM2 inflammasome is activated and IL-1β is secreted from 293FT cells transfected with plasmids expressing AIM2, pro-IL-1β, and pro-caspase-1 (5). As shown in Fig. 4, Flag-VP22:259-262A and Flag-VP22:263-266A inhibited AIM2-dependent IL-1β secretion, pro-caspase-1 cleavage, and IL-1β maturation as efficiently as wild-type Flag-VP22. In contrast, other Flag-VP22 mutants (Flag-VP22:227-230A, :231-234A, :235-238A, :239-242A, 243-246A, :247-250A, :251-254A, and :256-258A) showed an impaired ability to inhibit AIM2-dependent IL-1β secretion, pro-caspase-1 cleavage, and IL-1β maturation compared with wild-type Flag-VP22 (Fig. 4). Of note, only Flag-VP22 mutants that could still bind to DNA retained the wild-type ability to inhibit AIM2 inflammasome activation. Furthermore, most Flag-VP22 mutants (Flag-VP22:227-230A, :231-234A, :235-238A, :239-242A, :247-250A, :251-254A, and :256-258A) showed binding to the HIN domain of AIM2 similar to that of wild-type VP22 in the GST pulldown experiments (Fig. 1C). These results suggest that the DNA binding ability of VP22 was required for the efficient inhibition of AIM2-dependent inflammasome activation in the AIM2 inflammasome reconstitution system.

FIG 4.

Inhibitory effect of VP22 mutants on AIM2-dependent inflammasome reconstitution system. (A and B) 293FT cells were transfected with expression plasmids encoding AIM2, ASC, pro-caspase-1, pro-IL-1β, and each of the alanine substitution mutants of VP22 or enhanced green fluorescent protein (EGFP). (A) Cell-free supernatants were collected at 24 h posttransfection, and IL-1β release into supernatants was measured by ELISA. (B) Cell lysates were analyzed by immunoblotting with the indicated antibodies. In panel A, each value is the mean ± standard error of results of triplicate experiments. Statistical analysis was performed by one-way ANOVA and Dunnett’s test. ns, not statistically significant; **, P < 0.01; ***, P < 0.001. In panels A and B, the data are representative of three independent experiments.

Construction and characterization of a series of recombinant viruses with mutations in VP22 domain:227-267.

To investigate the effect of the DNA binding activity of VP22 on VP22-mediated inhibition of the AIM2 inflammasome induced in HSV-1-infected cells, we constructed and characterized a series of recombinant viruses with successive alanine substitutions in VP22 domain:227-267 (Fig. 5). As shown in Fig. 6A, Vero cells infected with recombinant viruses encoding VP22:227-230A, VP22:231-234A, VP22:235-238A, VP22:239-242A, VP22:243-246A, VP22:247-250A, or VP22:251-254A accumulated VP22 at levels lower than those infected with wild-type HSV-1(F). These results were in agreement with previous reports (16) showing that cells infected with each of the recombinant viruses carrying alanine substitutions for the dileucine motifs in VP22 at amino acids 235 and 236, or at amino acids 251 and 252, accumulated at lower levels than those infected with wild-type HSV-1(F). In contrast, cells infected with recombinant viruses encoding VP22:256-258A, VP22:259-262A, or VP22:263-266A accumulated VP22 at levels comparable to those infected with wild-type HSV-1(F) (Fig. 6A). As described above (Fig. 3D), VP22:256-258A had impaired binding to DNA in the DNA pulldown experiments compared with wild-type VP22. These results suggest that the DNA binding activity of VP22 was not required for the proper accumulation of VP22 in HSV-1-infected cells. We next focused on the recombinant virus YK470 (VP22:256-258A) encoding VP22:256-258A, because only this recombinant virus produced VP22 in infected cells at the level comparable to wild-type HSV-1(F) among the VP22 mutant viruses which harbor mutations that impair VP22-mediated inhibition of AIM2 inflammasome activation in the inflammasome reconstitution system and the DNA binding activity of VP22 in the DNA pulldown experiments (Fig. 3D, 4, and 6A). We also generated recombinant virus YK473 (VP22:256-258A-repair) in which the YK472 mutations were repaired (Fig. 5). To examine the growth properties of these recombinant viruses, Vero cells were infected with wild-type HSV-1(F), YK470 (VP22:256-258A), YK473 (VP22:256-258A-repair), a VP22-null mutant virus YK461 (VP22ΔΜ) (36), or its repaired virus YK462 (VP22ΔM-repair) (36) at a multiplicity of infection (MOI) of 5 or 0.05, and virus titers were assayed at 24 or 48 h after infection. As reported previously (36), progeny virus yields of YK461 (VP22ΔM) were significantly lower than those of wild-type HSV-1(F) or YK462 (VP22ΔM-repair) at MOIs of 5 and 0.05 (Fig. 6B and C). In contrast, progeny virus yields of YK470 (VP22:256-258A) were similar to those of wild-type HSV-1(F) or YK473 (VP22:256-258A-repair) at MOIs of 5 and 0.05 (Fig. 6B and C). These results suggest that the DNA binding activity of VP22 was not required for HSV-1 replication in cell cultures.

FIG 5.

Schematic diagrams of the genomic structure of wild-type HSV-1(F) and the relevant domains of the recombinant viruses used in this study. Line 1, wild-type HSV-1(F) genome; line 2, domain of the UL48 gene to the UL50 gene; line 3, domains of the UL49 gene; lines 5 to 16, recombinant viruses used in this study.

FIG 6.

Effect of VP22 mutations on the accumulation of VP22 and viral growth in HSV-1-infected cells. (A) Vero cells were mock infected or infected with wild-type HSV-1(F), YK463 (VP22:227-230A), YK464 (VP22:231-234A), YK465 (VP22:235-238A), YK466 (VP22:239-242A), YK467 (VP22:243-246A), YK468 (VP22:247-250A), YK469 (VP22:251-254A), YK470 (VP22:256-258A), YK471 (VP22:259-262A), and YK472 (VP22:263-266A) at an MOI of 5 for 18 h. These cells were then analyzed by immunoblotting with antibodies to the indicated proteins. (B and C) Vero cells were infected with wild-type HSV-1(F), YK470 (VP22:256-258A), YK473 (VP22:256-258A-repair), YK461 (VP22ΔM), and YK462 (VP22ΔM-repair) at an MOI of 5 (B) or 0.05 (C). The infected cells were harvested at 24 h (B) or 48 h (C) postinfection, and progeny viruses were assayed on Vero cells. In panel A, the data are representative of three independent experiments. In panels B and C, each value is the mean ± standard error of the results of three independent experiments. Statistical analysis was performed by one-way ANOVA and Tukey’s test. ns, not statistically significant; *, P < 0.05; **, P < 0.01.

To examine the effect of VP22 DNA binding activity on the VP22-mediated inhibition of AIM2 inflammasome induced in HSV-1-infected cells, bone marrow macrophages derived from wild-type mice (WT BMMs) were mock infected or infected with wild-type HSV-1(F), YK470 (VP22:256-258A), YK473 (VP22:256-258A-repair), YK461 (VP22ΔM), or YK462 (VP22ΔM-repair), and pro-caspase-1 processing in these infected cells was analyzed by immunoblotting. In agreement with our previous report (5), the levels of pro-caspase-1 accumulation in mock-infected cells and cells infected with wild-type HSV-1(F), YK461 (VP22ΔM), and YK462 (VP22ΔM-repair) were similar. However, caspase-1 (p10) accumulation was barely detectable in mock-infected cells and cells infected with wild-type HSV-1(F) and YK462 (VP22ΔM-repair) compared with cells infected with YK461 (VP22ΔM) (Fig. 7A). Caspase 1 (p10) accumulation in cells infected with YK470 (VP22:256-258A) was similar to that in cells infected with YK461 (VP22ΔM) but was barely detectable in cells infected with YK473 (VP22:256-258A-repair), similar to that in mock-infected cells and cells infected with wild-type HSV-1(F) and YK762 (VP22ΔM-repair) (Fig. 7A). To examine whether the processing of caspase-1 in BMMs infected with YK470 (VP22:256-258A) and YK461 (VP22ΔM) was AIM2 dependent, WT BMMs (AIM2^+/+ WT BMMs) and BMMs derived from AIM2^−/− mice (AIM2^−/− BMMs) were mock infected or infected with YK470 (VP22-256-258A) or YK461 (VP22ΔM), and pro-caspase-1 processing in these infected cells was analyzed. Consistent with the results in Fig. 7A, pro-caspase-1 processing was detected in AIM2^+/+ WT BMMs infected with YK470 (VP22:256-258A) and YK461 (VP22ΔM) compared with very low levels in AIM2^−/− BMMs infected with YK470 (VP22:256-258A) and YK461 (VP22ΔM) (Fig. 7B). Caspase 1 (p10) accumulations in AIM2^−/− BMMs infected with YK470 (VP22:256-258A) and YK461 (VP22ΔM) were similar to those in mock-infected AIM2^+/+ WT and AIM2^−/− BMMs (Fig. 7B). Finally, we analyzed IL-1β secretion in AIM2^+/+ WT or AIM2^−/− BMMs mock infected or infected with wild-type HSV-1(F), YK470 (VP22:256-268A), YK473 (VP22:256-258A-repair), YK461 (VP22ΔM), or YK462 (ΔVP22-repair). In agreement with the results for pro-caspase-1 processing (Fig. 7A and B), significant IL-1β secretion was detected in AIM2^+/+ WT BMMs infected with YK470 (VP22:256-258A) and YK461 (VP22ΔM) (Fig. 7C). In contrast, IL-1β secretion was barely detectable in AIM2^+/+ WT BMMs mock infected or infected with wild-type HSV-1(F), YK473 (VP22-256-258A-repair), or YK462 (VP22ΔM-repair), similar to the results with AIM2^−/− BMMs mock infected and infected with each of all the viruses tested (Fig. 7C). These results suggest that the DNA binding activity of VP22 was required for VP22-mediated inhibition of AIM2 inflammasome induced in HSV-1-infected BMMs.

FIG 7.

Effect of VP22 mutations on the inhibitory effect of the AIM2-dependent inflammasome in HSV-1-infected cells. (A) BMMs were mock infected or infected with wild-type HSV-1(F), YK470 (VP22:256-258A), YK473 (VP22:256-258A-repair), YK461 (VP22ΔM), and YK462 (VP22ΔM-repair) at an MOI of 3 for 12 h. These cells were then analyzed by immunoblotting with antibodies to the indicated proteins. (B) WT (AIM2^+/+) and AIM2^−/− BMMs were mock infected or infected with YK470 (VP22:256-258A) and YK461 (VP22ΔM) at an MOI of 3 for 12 h and analyzed as described for panel A. (C) WT (AIM2^+/+) and AIM2^−/− BMMs were mock infected or infected with wild-type HSV-1(F), YK470 (VP22:256-258A), YK473 (VP22:256-258A-repair), YK461 (VP22ΔM), and YK462 (VP22ΔM-repair) at an MOI of 3 for 12 h. IL-1β secretion was measured by ELISA. The dotted line indicates the detection limit (7.8 pg/ml) of the ELISAs. In panels A and B, the data are representative of three independent experiments. In panel C, each value is the mean ± standard error of the results of eight independent experiments. Statistical analysis was performed by one-way ANOVA and Tukey’s test. ns, not statistically significant; ***, P < 0.001.

DISCUSSION

Under physiological conditions, the self-oligomerization of AIM2 is triggered by the recognition of target DNA by the HIN domain of AIM2 (35). Therefore, DNA binding and self-oligomerization are coupled in AIM2, similar to other AIM2-like receptor family (ALRs) members such as IFI16 (37). Although VP22 can bind nonspecifically to DNA (21), its role in the HSV-1 life cycle is unclear. These observations prompted us to test the hypothesis that the DNA binding activity of VP22 affects the DNA-mediated oligomerization of AIM2, thereby inhibiting AIM2 inflammasome activation in HSV-1-infected cells. To this end, we mapped which domain(s) of VP22 was required for its DNA binding activity. We found that VP22 domain:227-267 was required for VP22 binding to DNA in DNA pulldown experiments. We also identified the minimal amino acids in VP22 domain:227-267 required for the VP22 binding to DNA: successive alanine substitutions in VP22 domain:227-258 substantially reduced the ability of VP22 to bind to DNA. These results are in agreement with previous reports showing that VP22 domains comprising amino acids 159 to 301 or 174 to 301 were sufficient for VP22 DNA binding in vitro (21) and its association with chromatin during mitosis of cells ectopically expressing VP22 (20), respectively. In general, DNA binding proteins use a combination of α helices, β strands, and/or loops to recognize and bind DNA (38). In agreement with this, the structure of VP22 domain:227-258 contains two α helices and one β strand (Fig. 1A) (39). Taken together, these observations suggest that the integrity of these secondary structures in VP22 domain:227-258 is critical for VP22 binding to DNA, as observed with other DNA binding proteins (40–42). The successive alanine substitutions in VP22 domain:227-258 tested in this study may disrupt each of the secondary structures, thereby resulting in a loss of DNA binding activity. Furthermore, among nine successive alanine substitutions in VP22 domain:227-258, most mutations other than 243-246A and 245A had no effect on the binding of VP22 to AIM2, suggesting that these mutations did not affect the global structure of VP22.

Mutational analyses of VP22 in this study showed that the ability of the VP22 mutants to bind to DNA in the DNA pulldown experiments correlated with the ability of the mutants to inhibit AIM2 inflammasome activation in the AIM2 inflammasome reconstitution experiments. Furthermore, recombinant HSV-1 carrying the VP22 mutation 256-258A, which substantially impaired VP22 DNA binding activity, did not inhibit AIM2 inflammasome activation in HSV-1-infected cells. These results suggest that the DNA binding activity of VP22 was critical for the VP22-mediated inhibition of AIM2 inflammasome activation induced in HSV-1-infected cells. Although the mechanism involved is unclear, there are three possible mechanisms (Fig. 8). First, VP22 binds to AIM2 on target HSV-1 genome DNA and inhibits the oligomerization of AIM2 (Fig. 8A). In AIM2, the PYD, responsible for homophilic association in ALRs (34), and HIN domains contribute to its oligomerization (35). Taken together with our earlier observation that VP22 might inhibit the oligomerization of the AIM2 HIN domain (5), this suggests that VP22 on target viral DNA binds to the HIN domain of AIM2 to (i) inhibit conformational changes of AIM2, induced by DNA binding, for its oligomerization (43) [Fig. 8A(i)] or (ii) sterically hinder oligomerization of the HIN domain and maybe that of PYD [Fig. 8A(ii)]. Second, VP22 competitively inhibits AIM2 binding to DNA by masking the target viral DNA (Fig. 8B). However, this mechanism is less likely because 2,000 copies of VP22 in the virion tegument were insufficient to fully mask the HSV-1 genome DNA, which is approximately 150 kbp (1, 8). In general, the portion of DNA in contact with a single DNA binding domain spans 4 to10 bp (44), and therefore, more than 2,000 copies of VP22 might be required to fully mask the viral genome DNA. Third, VP22 on target viral DNA binds to AIM2 to inhibit its binding to target viral DNA (Fig. 8C). In the first and third mechanisms, VP22 needs to bind to AIM2. However, we were unable to identify a mutation in VP22 that specifically impaired the binding of VP22 to AIM2. Mutations 243-246A and 245A with impaired VP22 binding to AIM2 also had abolished DNA binding activity. Thus, the cell-based assays used in this study were unable to demonstrate the specific requirement for AIM2 binding of VP22 for VP22-mediated AIM2 inflammasome activation. Further studies such as structural analyses of VP22 interactions with AIM2 and/or DNA and biochemical analyses using purified VP22 and AIM2 will be important to clarify the underlying molecular mechanisms involved.

FIG 8.

Models of the inhibition of AIM2 inflammasome activation by VP22. (A) (i) According to one model (43), AIM2 exists in an autoinhibitory conformation in which the HIN domain sequesters PYD via intramolecular interactions under resting conditions. Once the HIN domain of AIM2 binds to target HSV-1 DNA, a conformational change occurs, leading to the oligomerization of AIM2. VP22 on target viral DNA binds to the HIN domain of AIM2 and inhibits the conformational change of AIM2, thereby preventing the oligomerization of AIM2. (ii) After the conformational change of AIM2 upon binding to DNA shown in panel i, VP22 on the target HSV-1 DNA binds to the HIN domain of AIM2 and sterically hinders the oligomerization of AIM2. Alternatively, in another model (35) wherein AIM2 is not autoinhibited, AIM2 binding to target viral DNA leads to its oligomerization without a conformational change and VP22 on the target viral DNA binds to the HIN domain of AIM2 to sterically hinder the oligomerization of AIM2. (B) VP22 competitively inhibits the DNA binding of AIM2 by masking target viral DNA. (C) VP22 on the target HSV-1 DNA sequesters AIM2 to inhibit its binding to target viral DNA.

The amino acid sequence of the minimal domain of VP22 (VP22 domain:227-258) required for its DNA binding activity and VP22-mediated AIM2 inflammasome activation is highly conserved in the VP22 homologs of alphaherpesviruses (45), suggesting that other alphaherpesvirus VP22 homologs bind to DNA to exert effects through these conserved domains. In agreement with this, we previously showed that the inhibitory function of HSV-1 VP22 on AIM2 inflammasome activation was conserved in other alphaherpesvirus VP22 homologs, such as HSV-2 and pseudorabies virus (5). Thus, it seems likely that the DNA binding activity of alphaherpesvirus VP22 homologs acts to inhibit AIM2 inflammasome activation. Furthermore, it has been reported that the conserved carboxy-terminal VP22 domain, including VP22 domain:227-258, shows structural similarity to ORF52, a well-conserved component of the virion tegument of murine herpesvirus 68, a member of the Gammaherpesvirinae subfamily (39, 46). Of note, an ORF52 homolog in Kaposi’s sarcoma-associated herpesvirus (KSHV), another member of the Gammaherpesvirinae subfamily, as well as HSV-1 VP22 antagonized antiviral innate immune responses mediated by cGAS, another critical cytosolic DNA sensor (47, 48). Mechanistically, KSHV ORF52 and HSV-1 VP22 directly inhibited cGAS enzymatic activity, which was activated by the binding of cGAS to DNA (47, 48). Similar to the effects of VP22 on AIM2, KSHV ORF52 inhibited cGAS by binding to DNA and cGAS (47). These observations raise an interesting possibility that herpesviral tegument proteins antagonize cytosolic DNA sensors via a common mechanism. If this is the case, among the proposed mechanisms for the VP22-mediated inhibition of AIM2 inflammasome described above, the third mechanism might be most likely, because the only commonality between the cascades mediated by these DNA sensors is that they are activated upon binding to DNA (7).

In conclusion, we showed that the DNA binding activity of VP22 is critical for the evasion of AIM2 inflammasome activation in HSV-1-infected cells. This is the first report demonstrating that the DNA binding activity of VP22 functions in the HSV-1 life cycle by evading AIM2 inflammasome activation, which is critical for HSV-1 replication in vivo (5).

MATERIALS AND METHODS

Cells and viruses.

Vero, rabbit skin, and 293FT cells were described previously (5, 49). BMMs were prepared as described previously (5). Briefly, bone marrow was flushed from the tibias and femurs of mice with Dulbecco modified Eagle medium (DMEM). BMMs were cultured with DMEM supplemented with 20% fetal calf serum and 30% L929 cell supernatant containing macrophage colony-stimulating factor at 37°C for 5 days. Wild-type HSV-1(F), YK461 (VP22ΔM), and YK462 (VP22ΔM-repair) were described previously (36, 49).

Mice.

Age- and sex-matched C57BL/6J mice obtained from Charles River were used as WT controls. AIM2^−/− mice were described previously (5).

Generation of recombinant HSV-1.

Recombinant viruses YK463 (VP22:227-230A), YK464 (VP22:231-234A), YK465 (VP22:235-238A), YK466 (VP22:239-242A), YK467 (VP22:243-246A), YK468 (VP22:247-250A), YK469 (VP22:251-254A), YK470 (VP22:256-258A), YK471 (VP22:259-262A), YK472 (VP22:263-266A), and YK473 (VP22:256-258A-repair) (Fig. 5) were generated by the two-step Red-mediated mutagenesis procedure using Escherichia coli GS1783 containing pYEbac102Cre as described previously (50–52), except that the primers listed in Table 1 were used. The viruses used in this study were propagated and titrated in Vero cells.

TABLE 1.

Primer sequences used for the construction of recombinant viruses in this study

Recombinant virus	Primer sequence (5′–3′)
YK463 (VP22:227-230A)	5′-CCGGATGGCGGCGGTCCAGCTCTGGGACATGTCGCGTCCGGCTGCCGCAGCGGACCTCAACGAACTCCTTGGAGGA TGACGACGATAAGTAGGG-3′
	5′-TCACGCGGATGGTGGTGATGCCAAGGAGTTCGTTGAGGTCCGCTGCGGCAGCCGGACGCGACATGTCCCAGACAAC CAATTAACCAATTCTGATTAG-3′
YK464 (VP22:231-234A)	5′-GGTCCAGCTCTGGGACATGTCGCGTCCGCGCACAGACGAAGCTGCCGCAGCGCTCCTTGGCATCACCACCATAGGAT GACGACGATAAGTAGGG-3′
	5′-CCTCGCAGACCGTCACGCGGATGGTGGTGATGCCAAGGAGCGCTGCGGCAGCTTCGTCTGTGCGCGGACGCGCAAC CAATTAACCAATTCTGATTAG-3′
YK465 (VP22:235-238A)	5′-GGACATGTCGCGTCCGCGCACAGACGAAGACCTCAACGAAGCTGCCGCAGCGACCACCATCCGCGTGACGGTAGGA TGACGACGATAAGTAGGG-3′
	5′-GCAGGTTTTTGCCCTCGCAGACCGTCACGCGGATGGTGGTCGCTGCGGCAGCTTCGTTGAGGTCTTCGTCTGCAACCA ATTAACCAATTCTGATTAG-3′
YK466 (VP22:239-242A)	5′-TCCGCGCACAGACGAAGACCTCAACGAACTCCTTGGCATCGCTGCCGCAGCGGTGACGGTCTGCGAGGGCAAAGGA TGACGACGATAAGTAGGG-3′
	5′-TGGCGCGCTGAAGCAGGTTTTTGCCCTCGCAGACCGTCACCGCTGCGGCAGCGATGCCAAGGAGTTCGTTGACAACC AATTAACCAATTCTGATTAG-3′
YK467 (VP22:243-246A)	5′-CGAAGACCTCAACGAACTCCTTGGCATCACCACCATCCGCGCCGCTGCGGCGGAGGGCAAAAACCTGCTTCAAGGA TGACGACGATAAGTAGGG-3′
	5′-TCACCAACTCGTTGGCGCGCTGAAGCAGGTTTTTGCCCTCCGCCGCAGCGGCGCGGATGGTGGTGATGCCAACAACC AATTAACCAATTCTGATTAG-3′
YK468 (VP22:247-250A)	5′-CGAACTCCTTGGCATCACCACCATCCGCGTGACGGTCTGCGCTGCCGCAGCGCTGCTTCAGCGCGCCAACGAAGGAT GACGACGATAAGTAGGG-3′
	5′-CCACGTCTGGATTCACCAACTCGTTGGCGCGCTGAAGCAGCGCTGCGGCAGCGCAGACCGTCACGCGGATGGCAACC AATTAACCAATTCTGATTAG-3′
YK469 (VP22:251-254A)	5′-CATCACCACCATCCGCGTGACGGTCTGCGAGGGCAAAAACGCTGCCGCAGCGGCCAACGAGTTGGTGAATCCAGGAT GACGACGATAAGTAGGG-3′
	5′-CGACGTCCTGCACCACGTCTGGATTCACCAACTCGTTGGCCGCTGCGGCAGCGTTTTTGCCCTCGCAGACCGCAACCA ATTAACCAATTCTGATTAG-3′
YK470 (VP22:256-258A)	5′-CCGCGTGACGGTCTGCGAGGGCAAAAACCTGCTTCAGCGCGCTGCCGCAGCGGTGAATCCAGACGTGGTGCAAGGAT GACGACGATAAGTAGGG-3′
	5′-CCGTGGCCGCGTCGACGTCCTGCACCACGTCTGGATTCACCGCTGCGGCAGCGCGCTGAAGCAGGTTTTTGCCAACCA ATTAACCAATTCTGATTAG-3′
YK471 (VP22:259-262A)	5′-CTGCGAGGGCAAAAACCTGCTTCAGCGCGCCAACGAGTTGGCTGCCGCAGCGGTGGTGCAGGACGTCGACGCAGGAT GACGACGATAAGTAGGG-3′
	5′-GCCCTCGAGTCGCCGTGGCCGCGTCGACGTCCTGCACCACCGCTGCGGCAGCCAACTCGTTGGCGCGCTGAACAACC AATTAACCAATTCTGATTAG-3′
YK472 (VP22:263-266A)	5′-AAACCTGCTTCAGCGCGCCAACGAGTTGGTGAATCCAGACGCTGCCGCAGCGGTCGACGCGGCCACGGCGACAGGAT GACGACGATAAGTAGGG-3′
	5′-ACGCCGCAGAACGCCCTCGAGTCGCCGTGGCCGCGTCGACCGCTGCGGCAGCGTCTGGATTCACCAACTCGTCAACC AATTAACCAATTCTGATTAG-3′
YK473 (VP22:256-258A-repair)	5′-CCGCGTGACGGTCTGCGAGGGCAAAAACCTGCTTCAGCGCGCCAACGAGTTGGTGAATCCAGACGTGGTGCAAGGAT GACGACGATAAGTAGGG-3′
	5′-CCGTGGCCGCGTCGACGTCCTGCACCACGTCTGGATTCACCAACTCGTTGGCGCGCTGAAGCAGGTTTTTGCCAACCA ATTAACCAATTCTGATTAG-3′

Plasmids.

pCA7-NLRP3, pCA7-ASC, pCA7-pro-IL-1b, pCA7-pro-caspase-1, CA7-Flag-NLRP3, pCA7-AIM2, pFlag-CMV2-HSV-1/VP22, pFlag-CMV2-HSV-1/VP22(40-301), pFlag-CMV2-HSV-1/VP22(80-301), pFlag-CMV2-HSV-1/VP22(120-301), pFlag-CMV2-HSV-1/VP22(1-267), pFlag-CMV2-HSV-1/VP22(1-226), pFlag-CMV2-HSV-1/VP22(1-192), and pFlag-CMV2-HSV-1/VP22(1-160) were described previously (5) (Table 2). pFlag-CMV2-HSV-1/VP22(227-230A), pFlag-CMV2-HSV-1/VP22(231-234A), pFlag-CMV2-HSV-1/VP22(235-238A), pFlag-CMV2-HSV-1/VP22(239-242A), pFlag-CMV2-HSV-1/VP22(243-246A), pFlag-CMV2-HSV-1/VP22(247-250A), pFlag-CMV2-HSV-1/VP22(251-254A), pFlag-CMV2-HSV-1/VP22(256-258A), pFlag-CMV2-HSV-1/VP22(259-262A), pFlag-CMV2-HSV-1/VP22(263-266A), pFlag-CMV2-HSV-1/VP22(243A), pFlag-CMV2-HSV-1/VP22(244A), pFlag-CMV2-HSV-1/VP22(245A), and pFlag-CMV2-HSV-1/VP22(246A) (Table 2) were constructed by site-directed PCR mutagenesis as described previously (53).

TABLE 2.

Plasmids used in this study

Plasmid	Description	Source
pCA7-ASC	pCA7 harboring human ASC	5
pCA7-pro-IL-1β	pCA7 harboring human pro-IL-1β	5
pCA7-pro-caspase-1	pCA7 harboring human pro-caspase-1	5
pCA7-AIM2	pCA7 harboring human AIM2	5
pGEX4T-AIM2-HIN	pGEX4T harboring HIN domain of human AIM2	5
pFlag-CMV2-HSV-1/VP22	pFlag-CMV2 harboring HSV-1 VP22	5
pFlag-CMV2-HSV-1/VP22(40-301)	pFlag-CMV2 harboring HSV-1 VP22 codons 40–301	5
pFlag-CMV2-HSV-1/VP22(80-301)	pFlag-CMV2 harboring HSV-1 VP22 codons 80–301	5
pFlag-CMV2-HSV-1/VP22(120-301)	pFlag-CMV2 harboring HSV-1VP22 codons 120–301	5
pFlag-CMV2-HSV-1/VP22(1-267)	pFlag-CMV2 harboring HSV-1 VP22 codons 1–267	5
pFlag-CMV2-HSV-1/VP22(1-226)	pFlag-CMV2 harboring HSV-1 VP22 codons 1–226	5
pFlag-CMV2-HSV-1/VP22(1-192)	pFlag-CMV2 harboring HSV-1 VP22 codons 1–192	5
pFlag-CMV2-HSV-1/VP22(1-160)	pFlag-CMV2 harboring HSV-1 VP22 codons 1–160	5
pFlag-CMV2-HSV-1/VP22(227-230A)	pFlag-CMV2 harboring HSV-1 VP22 with alanine replacements of R227, T228, D229, and E230	This study
pFlag-CMV2-HSV-1/VP22(231-234A)	pFlag-CMV2 harboring HSV-1 VP22 with alanine replacements of D231, L232, N233, and E234	This study
pFlag-CMV2-HSV-1/VP22(235-238A)	pFlag-CMV2 harboring HSV-1 VP22 with alanine replacements of L235, L236, G237, and I238	This study
pFlag-CMV2-HSV-1/VP22(239-242A)	pFlag-CMV2 harboring HSV-1 VP22 with alanine replacements of T239, T240, I241, and R242	This study
pFlag-CMV2-HSV-1/VP22(243-246A)	pFlag-CMV2 harboring HSV-1 VP22 with alanine replacements of V243, T244, V245, and C246	This study
pFlag-CMV2-HSV-1/VP22(247-250A)	pFlag-CMV2 harboring HSV-1 VP22 with alanine replacements of E247, G248, K249, and N250	This study
pFlag-CMV2-HSV-1/VP22(251-254A)	pFlag-CMV2 harboring HSV-1 VP22 with alanine replacements of L251, L252, Q253, and R254	This study
pFlag-CMV2-HSV-1/VP22(256-258A)	pFlag-CMV2 harboring HSV-1 VP22 with alanine replacements of 256N, 257E, and 258L	This study
pFlag-CMV2-HSV-1/VP22(259-262A)	pFlag-CMV2 harboring HSV-1 VP22 with alanine replacements of V259, N260, P261, and D262	This study
pFlag-CMV2-HSV-1/VP22(263-266A)	pFlag-CMV2 harboring HSV-1 VP22 with alanine replacements of V263, V264, Q265, and D266	This study
pFlag-CMV2-HSV-1/VP22(243A)	pFlag-CMV2 harboring HSV-1 VP22 with alanine replacement of V243	This study
pFlag-CMV2-HSV-1/VP22(244A)	pFlag-CMV2 harboring HSV-1 VP22 with alanine replacement of T244	This study
pFlag-CMV2-HSV-1/VP22(245A)	pFlag-CMV2 harboring HSV-1 VP22 with alanine replacement of V245	This study
pFlag-CMV2-HSV-1/VP22(246A)	pFlag-CMV2 harboring HSV-1 VP22 with alanine replacement of C246	This study

Antibodies.

Commercial antibodies used in this study were mouse monoclonal antibodies to Flag (M2; Sigma), ICP4 (58S; ATCC), α-tubulin (DM1A; Sigma), and AIM2 (3B10; Adipogen), rabbit monoclonal antibody to human caspase-1 (3866S; Cell Signaling Technology), and rabbit polyclonal antibodies to human IL-1β (ab2105; Abcam), ASC (sc-22514-R; Santa Cruz Biotechnology), and mouse caspase-1 (sc-514; Santa Cruz Biotechnology). Rabbit polyclonal antibodies to VP22 were described previously (16).

ELISA.

Enzyme-linked immunosorbent assay (ELISA) was performed as described previously (5).

Immunoblotting.

Immunoblotting was performed as described previously (54). The amount of protein in immunoblot bands was quantitated using Image Lab 6.0.1 software (Bio-Rad) and normalized using the sum of all data points in each replicate (55).

GST pulldown assays.

GST pulldown assays were performed as described previously. Briefly, 293FT cells were transfected with pFLAG-CMV2-VP22 or each of its mutants using PEI Max and harvested 48 h posttransfection. Cells were then lysed in NP-40 buffer (120 mM NaCl, 50 mM Tris-HCL [pH 8.0], 0.5% NP-40, 50 mM NaF) containing a proteinase inhibitor cocktail (Nacalai Tesque). GST and GST-AIM2-HIN were expressed in Escherichia coli as described previously (5). The transfected cell lysates were reacted with purified GST and GST-AIM2-HIN immobilized on glutathione-Sepharose beads (GE Healthcare) for 1.5 h at 4°C. After extensive washing of the beads with NP-40 buffer, the beads were divided into two parts. One part was analyzed by immunoblotting with anti-Flag antibody, and the other was electrophoretically separated in a denaturing gel and stained with Coomassie brilliant blue (CBB). For Benzonase treatment, Benzonase (Sigma; E1014) was added to the cell lysates at a final concentration of ≥125 U/ml and incubated at 25°C for 1 h. After incubation, purified GST and GST-AIM2-HIN immobilized on glutathione-Sepharose beads were added and the same procedure was performed.

DNA pulldown assays.

293FT cells were transfected with pFLAG-CMV2-VP22 or each of its mutants using PEI Max and harvested 48 h posttransfection. Cells were then lysed in NP-40 buffer (120 mM NaCl, 50 mM Tris-HCL [pH 8.0], 0.5% NP-40, 50 mM NaF) containing a proteinase inhibitor cocktail (Nacalai Tesque). A biotinylated 80-mer DNA was added to the transfected cell lysate to a final concentration of 250 nM and incubated at 25°C for 1 h. Streptavidin-Sepharose beads (Novagen) were then added and rotated at 4°C for 1.5 h. After the beads were washed extensively with NP-40 buffer, they were analyzed by immunoblotting with the indicated antibodies. For Benzonase treatment, Benzonase (Sigma; E1014) was added to a final concentration of ≥125 U/ml, and 80-mer DNA was added simultaneously to the cell lysate. The 80-mer double-stranded DNA with a biotinylated 3′ end on one side was prepared by annealing 5′-AGCTCTACGCATGGCAATCTACCCGCGAGGTCAGACGTGAGACTCGGAAACTAAATATTGGTAGCTTCCTTAGCTGAAGT-3′ and 3′-end-biotinylated 5′-ACTTCAGCTAAGGAAGCTACCAATATTTAGTTTCCGAGTCTCACGTCTGACCTCGCGGGTAGATTGCCATGCGTAGAGCT-3′. These sequences were produced using the Random DNA Generator (http://www.faculty.ucr.edu/∼mmaduro/random.htm), with a random GC content of 50%.

Inflammasome reconstitution assays.

Inflammasome reconstitution assays were performed as described previously (5).