Vaccinia Virus F1L blocks the ribotoxic stress response to subvert ZAKα-dependent NLRP1 inflammasome activation

Abstract

Inflammasomes are essential for host defense, recognizing foreign or stress signals to trigger immune responses, including maturation of IL-1 family cytokines and pyroptosis. Here, NLRP1 is emerging as an important sensor of viral infection in barrier tissues. NLRP1 is activated by various stimuli, including viral double-stranded (ds) RNA, ribotoxic stress, and inhibition of dipeptidyl peptidases 8 and 9 (DPP8/9). However, certain viruses, most notably the vaccinia virus, have evolved strategies to subvert inflammasome activation or effector functions. Using the modified vaccinia virus Ankara (MVA) as a model, we investigated how the vaccinia virus inhibits inflammasome activation. We confirmed that the early gene F1L plays a critical role in inhibiting NLRP1 inflammasome activation. Interestingly, it blocks dsRNA and ribotoxic stress-dependent NLRP1 activation without affecting its DPP9-inhibition-mediated activation. Complementation and loss-of-function experiments demonstrated the sufficiency and necessity of F1L in blocking NLRP1 activation. Furthermore, we found that F1L-deficient, but not wild-type MVA, induced ZAKα activation. Indeed, an F1L-deficient virus was found to disrupt protein translation more prominently than an unmodified virus, suggesting that F1L acts in part upstream of ZAKα. These findings underscore the inhibitory role of F1L on NLRP1 inflammasome activation and provide insight into viral evasion of host defenses and the intricate mechanisms of inflammasome activation.

Introduction

Inflammasomes are large cytoplasmic complexes that assemble in response to various pathogen- or damage-associated molecular patterns. They comprise a sensor molecule, adaptor protein ASC, and the cysteine protease caspase-1. Inflammasome assembly results in autocleavage and activation of caspase-1, which in turn matures the proinflammatory cytokines, IL-1β and IL-18, and cleaves the pore-forming protein gasdermin D (GSDMD), leading to a form of cell death called pyroptosis [1].

NLRP1 is one of the inflammasome-forming sensor proteins belonging to the nucleotide-binding domain leucine-rich repeat (NLR) family. It is expressed primarily in epithelial cells, such as skin keratinocytes or airway epithelia [2, 3]. Under steady-state conditions, NLRP1 forms a complex with dipeptidyl peptidase 9 (DPP9), and the DPP8/9 inhibitor Val-boroPro (VbP) was the first identified activator of endogenous human NLRP1 [4]. NLRP1 plays a role in antiviral defense by responding to a variety of stimuli; it is activated by enteroviral proteases [2], dsRNA produced during viral replication [5], and it responds to perturbations that affect translation and thus activate the ribotoxic stress response (RSR), such as UVB exposure or certain compounds. This latter pathway leads to the activation of the kinase ZAKα, which directly - or via engagement of mitogen-activated protein kinases (MAPKs) - phosphorylates NLRP1 [6, 7].

The importance of inflammasomes in antiviral defense is indirectly underscored by reports that some viruses have evolved mechanisms to inhibit inflammasome signaling. Examples include the inhibition of the AIM2 inflammasome by the HSV-1-encoded protein VP22 [8], Kaposi’s sarcoma-associated virus-encoded ORF37 [9], and Murine cytomegalovirus-encoded M84 [10]. Vaccinia virus (VACV)-encoded protein F1L was reported to specifically inhibit NLRP1 activation by muramyl dipeptide [11]. However, since that discovery, there has been much progress in understanding NLRP1 biology, and muramyl dipeptide has not been substantiated as an NLRP1 activator. However, VACV was confirmed to inhibit NLRP1 activation in keratinocytes, and RNA isolated from VACV-infected cells was shown to activate NLRP1 [12], whereas the mechanism of VACV-mediated NLRP1 inhibition and F1L involvement remained unclear. Here, we show that infection with an attenuated strain of VACV, modified vaccinia virus Ankara (MVA), inhibits NLRP1 activation in an F1L-dependent manner. Notably, the F1L-mediated inhibition is not direct, as the MVA virus failed to inhibit NLRP1 upon stimulation with the DPP8/9 inhibitor VbP. Instead, we found that infection with F1L-deficient MVA virus disrupts protein translation and causes ZAKα phosphorylation leading to NLRP1 inflammasome activation in a ZAKα-dependent fashion.

Results

MVA virus inhibits the NLRP1 inflammasome in keratinocytes

VACV interferes with inflammasome signaling at several levels; for example, it expresses the protein B13R, which inhibits caspase-1 activation [13], B15R, which scavenges IL-1β and thus inhibits its signaling [14], and F1L, which has been reported to prevent NLRP1 activation in THP-1 cells and in vitro [11]. To determine the effect of VACV infection on endogenous NLRP1 inflammasome signaling in human keratinocytes, we turned to the immortalized keratinocytes, N/TERT1, often used for NLRP1 studies. We activated the NLRP1 inflammasome with anisomycin (ANS), which induces RSR, the double-stranded RNA analog poly(I:C), and the DPP8/9 inhibitor VbP. To explore the impact of VACV infection, we additionally pre-infected the cells with the replication-deficient MVA at increasing multiplicities of infection (MOI). Here, we found that infection with MVA virus inhibited ANS- and poly(I:C)-induced IL-18 release in a dose-dependent manner, but did not affect VbP stimulation (Fig. 1A–C). We used IL-18 as a readout of inflammasome activation instead of IL-1β, as we found that infection with MVA virus substantially decreased pro-IL-1β expression in keratinocytes (Fig. 1D), which is consistent with its NF-kB inhibitory function [15, 16]. Importantly, MVA virus infection did not consistently decrease the levels of any other inflammasome component (Fig. 1D). To explore at what level in the signaling cascade MVA virus blocks inflammasome activation, we examined ASC speck formation in ASC^KO N/TERT1 keratinocytes complemented with ASC-mCherry (Fig. 1E). We found that MVA virus also potently inhibited ANS- and poly(I:C)-induced ASC speck formation while reducing the number of pyroptotic cells induced by these stimuli. Taken together, these results suggest that MVA virus infection inhibits inflammasome activation at the level of the ASC or further upstream.

Figure 1. MVA virus infection prevents NLRP1 activation in keratinocytes.

(A–C) N/TERT1 cells were infected with the MVA virus at MOI 1–5 for 4 h and then stimulated with ANS for 2 h (A), or infected for 2 h and then stimulated with poly(I:C) for 6 h (B), or infected for 1 h and then stimulated with VbP for 7 h (C). IL-18 release was measured by ELISA. Data from three independent experiments were summarized and are presented as mean values ± SEM. Data are analyzed by using a one-way ANOVA with Dunnett’s multiple comparisons test. (D) N/TERT1 cells were infected with the MVA virus at MOI 5 for 8 h and the protein expression was analyzed by immunoblotting. Data are representative of three independent experiments with similar results. (E) ASC^KON/TERT1 cells expressing ASC-mCherry were infected and stimulated as in (A) and (B), and ASC speck formation was imaged at the end of the stimulation. Bars represent 100 µm. Data are representative of three independent experiments with similar results. (F) NLRP1^KO N/TERT1 cells complemented with NLRP1 or NLRP3 were infected with the MVA virus at MOI 5 for 2–8 h and stimulated with ANS or Nigericin, respectively, for 2 h. Data from five independent experiments were summarized and are presented as mean values ± SEM. Data were analyzed by two-way ANOVA with Šídák’s multiple comparisons test. (G) N/TERT1 cells complemented as in (F) were infected with the MVA virus at MOI 5 for 4 h, stimulated with ANS or Nigericin for 2 h, and treated with DSS. Protein expression was analyzed by immunoblotting. Data are representative of two independent experiments with similar results. (H) HEK293T cells complemented with ASC-mCherry and NLRP1 or NLRP3 were infected with the MVA virus for 6 h and stimulated with ANS or Nigericin for 2 h. ASC specking was analyzed by flow cytometry. Data from three independent experiments were summarized and are presented as mean values ± SEM. Data were analyzed by two-way ANOVA with Šídák’s multiple comparisons test. (I) NLRP1^KO N/TERT1 cells complemented with mmNLRP1b were infected with the MVA virus at MOI 5 for 6 h and stimulated with VbP or lethal factor with PA for 2 h. Data from three independent experiments were summarized and are presented as mean values ± SEM. Data were analyzed using a one-way ANOVA with Šídák’s multiple comparisons test.

To determine the latter scenario, we used NLRP1^KO N/TERT1 cells complemented with NLRP1 or the unrelated inflammasome sensor NLRP3. NLRP3 is not expressed and thus not functional in N/TERT1 cells but can be stimulated by heterologous overexpression with potassium efflux-inducing agonists such as the ionophore Nigericin (Fig. S1A and B). Pre-infection with MVA virus significantly decreased NLRP1-mediated IL-18 release, but had no effect on Nigericin-mediated NLRP3 activation (Fig. 1F). However, MVA virus pretreatment prevented IL-1β release from NLRP3-expressing cells (Fig. S1C), consistent with infection decreasing the levels of pro-IL-1β (Fig. 1D). The lack of NLRP3 inhibition by MVA virus suggested that the inhibition of NLRP1 occurred at the level of or upstream of the sensor molecule. To address this, we examined the effect of MVA virus infection on inflammasome-mediated ASC polymerization using a cross-linking approach. We examined dimerization of endogenous ASC in NLRP1- or NLRP3-expressing NLRP1^KO N/TERT1 cells stimulated with either ANS or Nigericin in the absence or presence of the protein cross-linker disuccinimidyl suberate (DSS). Both NLRP1- and NLRP3-stimulated cells exhibited ASC dimerization in the presence of DSS, but MVA virus infection specifically blocked NLRP1-mediated inflammasome activation, whereas it had no effect on NLRP3-stimulated cells (Fig. 1G). Furthermore, the MVA virus also prevented ASC specking in HEK cells complemented with ASC-mCherry and NLRP1 stimulated with ANS, while it had little effect on cells expressing NLRP3 stimulated with Nigericin (Fig. 1H).

The inability of the MVA virus to inhibit NLRP1 upon VbP stimulation suggests that it interferes with a signaling step upstream of NLRP1. To explore this possibility, we turned to NLRP1^KO N/TERT1 cells complemented with FKBP-NLRP1 fusion protein consisting of PYD-deficient NLRP1 with two dimerization domains (DmrB) fused to the N-terminus; this allowed NLRP1 activation using B/B homodimerizer, which causes dimerization of FKBP-containing proteins. B/B homodimerizer treatment induced IL-18 release from cells expressing the FKBP-NLRP1 fusion protein, but not from cells expressing unmodified NLRP1 (Fig. S1D). Of note, infection of FKBP-NLRP1 cells with MVA virus did not prevent VbP- or B/B homodimerizer-induced IL-18 release (Fig. S1E). F1L was reported to bind murine Nlrp1b and deletion of F1L was reported to decrease the virulence of VACV infection in mice [11]. To determine if this effect could be mediated by the Nlrp1 inflammasome, we infected NLRP1-deficient N/TERT1 cells complemented with murine Nlrp1b with MVA virus and stimulated them with VbP or anthrax lethal factor (Fig. 1I). We found that MVA virus had little effect on Nlrp1b-mediated IL-18 release but considerably decreased IL-1β release (Fig. S1F).

To further determine specificity, we examined the effect of the MVA virus on CARD8, an inflammasome sensor related to NLRP1 that is also activated in response to DPP9 inhibition [17]. Analogous to NLRP3, CARD8 is not expressed and functional in N/TERT1 cells but can be stimulated by VbP when overexpressed. Consistent with our previous observations, the MVA virus did not affect VbP-induced NLRP1 activation (Fig. S1G). Surprisingly, however, it strongly inhibited CARD8-dependent inflammasome activation, suggesting differences in the signaling of the two inflammasomes induced by DPP9 inhibition. Taken together, our data suggest that the MVA virus inhibits ANS- and poly(I:C)-induced NLRP1 inflammasome activation upstream or at the level of the sensor molecule.

MVA virus-encoded protein F1L inhibits NLRP1 inflammasome

VACV was previously reported to encode the protein F1L, which interferes with NLRP1 inflammasome signaling [11]. F1L has been studied primarily in the context of apoptosis inhibition and is known to localize to mitochondria via its C-terminus where it binds to proapoptotic proteins, thereby preventing the induction of apoptosis [18–20]. It contains a disordered N-terminus, which has been reported to interfere with caspase-9 and NLRP1 inflammasome signaling [11, 21, 22]. To investigate the effect of F1L on NLRP1 signaling in keratinocytes, we infected N/TERT1 cells with wild-type (WT), F1L-deficient (ΔF1L), or revertant (rF1L) MVA virus [23] prior to stimulation with ANS, poly(I:C), and VbP. As expected, the WT virus inhibited ANS- and poly(I:C)-, but not VbP-induced IL-18 release. In contrast, infection with ΔF1L MVA virus had little effect on ANS and poly(I:C) stimulation and strikingly enhanced VbP-induced IL-18 release (Fig. 2A). Interestingly, ΔF1L MVA virus still inhibited the CARD8 inflammasome (Fig. S2A), further strengthening the notion that the CARD8-antagonistic property of MVA virus is distinct from its NLRP1 inhibitory function.

Figure 2. NLRP1 inhibition during MVA virus infection is mediated by F1L.

(A) N/TERT1 cells were infected with WT, ΔF1L, or ΔF1L + rF1L MVA virus at MOI 5 for 4, 1, or 2 h, and then stimulated with ANS for 2 h, VbP for 7 h, and poly(I:C) for 6 h, respectively. IL-18 release was measured by ELISA. Data from three independent experiments were summarized and are presented as mean values ± SEM. Data are analyzed by using a two-way ANOVA with Šídák’s multiple comparisons test. (B) N/TERT1 cells were transfected with in vitrotranscribed mRNA encoding mScarlet or F1L for 6 h and stimulated with ANS for 2 h. IL-1β release was measured by ELISA. Data from three independent experiments were summarized and are presented as mean values ± SEM. Data are analyzed by using a two-way ANOVA with Šídák’s multiple comparisons test. (C) NLRP1^KO N/TERT1 cells complemented with NLRP1 were transfected with mRNA encoding mScarlet or F1L for 6 h and then stimulated with ANS for 2 h or transfected for 4 h and then stimulated with VbP for 4 h. IL-1β release was measured by ELISA. Data from seven independent experiments were summarized and are presented as mean values ± SEM. Data are analyzed by using a two-way ANOVA with Šídák’s multiple comparisons test. (D) NLRP1^KO N/TERT1 cells complemented with NLRP1 or NLRP3 were transfected with mRNA encoding mScarlet or F1L for 6 h and then stimulated with ANS or Nigericin for 2 h. IL-1β release was measured by ELISA. Data from six independent experiments were summarized and are presented as mean values ± SEM. Data shown in (D) include data from (C) (Control and ANS stimulation for NLRP1-expressing cells). Data are analyzed by using a two-way ANOVA with Šídák’s multiple comparisons test. (E, F) N/TERT1 cells were transfected with mRNA encoding mScarlet or F1L and its indicated variants for 6 h and then stimulated with ANS for 2 h. IL-1β release was measured by ELISA. Data from three independent experiments were summarized and are presented as mean values ± SEM. Data are analyzed by using a two-way ANOVA with Dunnett’s multiple comparisons test. (G) HEK293T cells expressing ASC-mCherry and NLRP1 or NLRP3 were transfected with plasmids encoding YFP or YFP-F1L and stimulated the next day with ANS or Nigericin for 2 h. ASC specking was analyzed by flow cytometry. The graph represents the percentage of specking cells in the YFP-positive population. Data from four independent experiments were summarized and are presented as mean values ± SEM. Data are analyzed by using a two-way ANOVA with Šídák’s multiple comparisons test. (H, I) HEK293T cells expressing ASC-mCherry and NLRP1 were transfected with the indicated constructs and treated and analyzed as in (G). Data from four (H) or three (I) independent experiments were summarized and are presented as mean values ± SEM. Data are analyzed by using a two-way ANOVA with Dunnett’s multiple comparisons test. (J) N/TERT1 cells were treated with Z-VAD-FMK for 6 h (Control), with Z-VAD-FMK immediately followed by infection with WT, ΔF1L, or ΔF1L + rF1L MVA virus at MOI 5 for 4 h and then stimulated with ANS for 2 h, or treated with Z-VAD-FMK for 4 h, followed by stimulation with ANS for 2 h. Data are representative of three independent experiments with similar results.

Next, we tested the role of F1L expressed in keratinocytes. Since we were unable to express the protein by lentiviral transduction, we transfected cells with in vitro transcribed mRNA encoding F1L from the MVA virus or with mRNA encoding for the fluorescent protein mScarlet as a control. Control transfection reduced IL-1β release compared with cells treated with transfection media alone, and F1L transfection further reduced it despite much lower transcript levels (Fig. 2B; Fig. S2B). However, this inhibition was not entirely specific; F1L transfection also partially decreased VbP-induced cytokine release, which is not inhibited by the virus (Fig. 2C and 1C). F1L transfection also decreased NLRP3-mediated IL-1β release, although to a lesser extent than NLRP1-mediated IL-1β release (Fig. 2D). F1L was previously reported to inhibit NLRP1 through direct interaction [11], while several point mutations in the N-terminal region of F1L abolished this inhibition. To test whether these residues were required for endogenous NLRP1 inhibition, we expressed two F1L mutants, H31A (corresponding to H35A in the Western Reserve strain of VACV) and N28A (corresponding to N32A) in N/TERT1 keratinocytes. However, neither mutant affected NLRP1 inhibition (Fig. 2E). To determine if the N-terminal portion of F1L is necessary for NLRP1 inhibition, we expressed truncated F1L lacking residues 1–54 or 1–81. Moreover, since F1L has been described as an anti-apoptotic protein, we investigated whether its anti-apoptotic function, attributed to the C-terminus that tethers it to the mitochondria (residues 215–222), is required for inflammasome inhibition [19]. However, we found that all deletion mutants inhibited IL-1β release to a similar extent as the WT protein (Fig. 2F), suggesting that the N-terminal residues and inhibition of apoptosis do not contribute to blocking NLRP1 signaling. Because transfection of mRNA into keratinocytes unspecifically decreased inflammasome signaling (Fig. 2C), we went on to study F1L in the HEK overexpression system. To this end, we transiently expressed F1L in HEK cells complemented with NLRP1, NLRP3, and ASC. In contrast to mRNA-transfected keratinocytes, F1L in HEK cells had a specific effect on NLRP1; it reduced specking only in NLRP1- but not in NLRP3-expressing cells (Fig. 2G). Of note, for these experiments double the amount of F1L-encoding plasmid compared with the control YFP plasmid was used, because the expression of F1L was very low. However, transfection of equal amounts of DNA had a comparable effect on inflammasome signaling (Fig. S2C). Using this system, we also addressed the impact of the different F1L mutants on the inhibition of the NLRP1 inflammasome. As observed for the keratinocyte system, all of these mutants still blocked NLRP1 inflammasome activation (Fig. 2H and I).

Our data suggest that F1L does not directly inhibit NLRP1 (Fig. 1C; Fig. S1E), at least not by modulating its DPP9 inhibition-dependent activation. Since the MVA virus inhibited ANS- and poly(I:C)-induced activation, both of which depend on ZAKα activation [6, 7], we examined the effect of MVA virus infection on ZAKα phosphorylation. When studying lysates from ANS-stimulated cells, we observed a small, yet reproducible shift of the ZAKα band on the immunoblot, indicative of phosphorylation. This shift was not reduced by pretreatment with the MVA virus, suggesting that F1L does not interfere with ZAKα phosphorylation (Fig. 2J). In addition, MVA virus infection did not consistently decrease p38 phosphorylation, which was suggested to play a role in NLRP1 activation [6, 7]. MVA virus also did not inhibit TNFα-induced p38 phosphorylation, confirming that this is not the signaling step inhibited by F1L (Fig. S2D). Interestingly, infection with ΔF1L MVA virus itself caused ZAKα and p38 phosphorylation, suggesting that the virus could activate NLRP1 through this pathway.

F1L-deficient MVA virus activates NLRP1 inflammasome

Since ΔF1L MVA virus potentiated VbP-induced IL-18 release (Fig. 2A) and caused ZAKα and p38 phosphorylation (Fig. 2J), we went on to explore if this virus variant could activate NLRP1 on its own. In line with this notion, we found that F1L-deficient but not WT or revertant virus caused IL-18 release from N/TERT1 cells, which was dependent on NLRP1 and ZAKα (Fig. 3A). In addition, ΔF1L MVA-virus-induced NLRP1-dependent LDH release and GSDMD cleavage in N/TERT-1 keratinocytes (Fig. 3B and C). In our previous experiments, ΔF1L MVA virus enhanced IL-18 release only after treatment with VbP, but not with ANS or poly(I:C). Its dependence on ZAKα suggests that the mechanism of MVA virus-induced NLRP1 activation is similar to that of ANS or dsRNA. Thus, it appears conceivable that ANS, poly(I:C), and MVA virus converge on the same activation pathway, making further enhancement impossible. VACV was previously reported to generate dsRNA during infection [12]; however, we did not detect dsRNA in MVA virus-infected keratinocytes (Fig. S3A). Another possibility we considered was that the MVA virus caused translation stress. Compounds that disrupt different translation stages activate NLRP1 inflammasome [6, 7]. VACV employs multiple mechanisms to disrupt host–protein translation (reviewed by [24]); therefore, we examined the effect of MVA virus infection on keratinocyte protein synthesis by studying the incorporation of puromycin into the nascent polypeptide chain. We found that the MVA virus decreased puromycin incorporation at increasing MOI, and the decrease was markedly more pronounced in ΔF1L MVA virus-infected cells (Fig. 3D). Furthermore, we detected a ZAKα shift on immunoblot and ZAKα-dependent p38 phosphorylation in N/TERT1 cells infected with ΔF1L MVA virus (Fig. 3E). Ribosome collisions trigger eIF2α phosphorylation in a ZAKα-dependent manner [25]. We observed eIF2α phosphorylation upon ΔF1L MVA virus infection, which negatively correlated with puromycin incorporation in ΔF1L MVA virus-infected cells (Fig. 3D and E); however, this response was not dependent on ZAKα. Consistent with this, the decrease in puromycin incorporation was also independent of ZAKα (Fig. S3B). Notably, eIF2α phosphorylation was also independent of PKR, as it was similar between WT and PKR-deficient keratinocytes upon ΔF1L MVA virus infection (Fig. S3C). To further characterize MVA virus-mediated NLRP1 activation, we co-treated the cells with NEDD8-cullin inhibitor MLN4924 and cytosine arabinoside (Ara C, Fig. S3D). MLN4924 inhibits cullin neddylation, which prevents N-terminal fragment degradation and NLRP1 activation [2]. Ara C inhibits MVA virus DNA replication, and, in consequence, the expression of replication-dependent intermediate and late genes. MLN4924 completely abolished ΔF1L MVA virus-induced IL-18 release, whereas Ara C considerably enhanced it, suggesting that the activation is not mediated by a protein encoded by an intermediate or late gene. Since Ara C potently enhanced NLRP1 activation, we examined if it increased cellular stress manifested by increased MAPK activation or eIF2α phosphorylation. However, Ara C did not affect ZAKα or p38 phosphorylation (Fig. S3E). In fact, it decreased eIF2α phosphorylation, suggesting that this pathway is independent of NLRP1 activation.

Figure 3. F1L-deficient MVA virus activates NLRP1 inflammasome.

(A) WT, NLRP1^KO, and ZAKα^KO N/TERT1 cells were stimulated with ANS for 2 h or infected with WT, ΔF1L, or ΔF1L + rF1L MVA virus at MOI 5 for 8 h. IL-18 release was measured by ELISA. Data from three (ZAKα^KO) to six (WT and NLRP1^KO) independent experiments were summarized and are presented as mean values ± SEM. Data were analyzed using a two-way ANOVA with Šídák’s multiple comparisons test. (B) NLRP1^KO N/TERT1 cells complemented with NLRP1 were stimulated with ANS for 2 h or infected with WT, ΔF1L, or ΔF1L + rF1L MVA virus at MOI 5 for 8 h and LDH release was measured. Data from three independent experiments were summarized and are presented as mean values ± SEM. Data are analyzed using a two-way ANOVA with Dunnett’s multiple comparisons test. (C) NLRP1^KO N/TERT1 cells complemented with NLRP1 were stimulated as in (B) and GSDMD cleavage was analyzed by immunoblotting. Data are representative of three independent experiments with similar results. (D) N/TERT1 cells were treated with Z-VAD-FMK for 8 h (Ctrl), treated with Z-VAD-FMK for 1 h and stimulated with ANS for 2 h, or treated with Z-VAD-FMK immediately followed by infection with WT, ΔF1L or ΔF1L + rF1L MVA virus at the indicated MOI. Puromycin was added 30 min before the end of the stimulation and protein expression was analyzed by immunoblotting. Data are representative of three independent experiments with similar results. (E) WT and ZAKα^KO N/TERT1 cells were treated with Z-VAD-FMK for 8 h (Control), pre-treated with Z-VAD-FMK for 1 h and then stimulated with ANS for 2 h, or treated with Z-VAD-FMK immediately followed by infection with WT, ΔF1L, or ΔF1L + rF1L MVA virus at MOI 5 for 8 h. Protein expression was analyzed by immunoblotting. Data are representative of three independent experiments with similar results. (F) N/TERT1 cells were stimulated with Z-VAD-FMK, ANS, and ΔF1L MVA virus as in (E). 5 min before the end of the experiment, the cells were treated with Harringtonine as indicated, followed by a 10 min puromycin pulse. Protein expression was analyzed by immunoblotting. Data are representative of three independent experiments with similar results.

To determine if the decrease in translation was caused by impaired initiation or elongation, we conducted a ribosome run-off assay [26, 27]. In this assay, infected cells are treated with Harringtonine, which inhibits translation initiation, followed by a puromycin pulse. In untreated cells, puromycin is incorporated into the nascent peptide chain. In cells treated with Harringtonine, actively translating ribosomes finish their transcripts, and since no new translation is initiated, the levels of polysomes are decreased, and puromycin incorporation decreases [28]. Elongation inhibition globally blunts puromycin incorporation, and because the translation rate is strongly reduced additional Harringtonine treatment does not decrease polysome levels and thereby does not affect puromycin incorporation [28]. Five minutes after Harringtonine treatment, untreated cells finished most of their transcripts, with mostly larger proteins remaining (Fig. 3F). In contrast, in ANS-treated cells, the amount of incorporated puromycin remained low, but steady throughout the conditions. Cells infected with ΔF1L MVA virus had lower overall puromycin incorporation levels, which decreased after 5 min of Harringtonine treatment, although to a lesser extent than in control cells. This data suggested that the effect of ΔF1L MVA virus on elongation is not as strong as that of ANS, although it could be the potential cause of ZAKα phosphorylation. Lastly, infection with F1L-deficient VACV was reported to decrease virulence and increase caspase-1 cleavage in the lungs of infected mice [11]. To determine if this effect could be mediated by Nlrp1b, we stimulated Nlrp1b-expressing N/TERT1 cells with ΔF1L MVA (Fig. S3F). However, we did not detect Nlrp1b-dependent IL-18 release. Overall, we showed that the F1L-deficient MVA virus caused ZAKα-dependent NLRP1 activation, possibly by inducing translational stress on keratinocytes.

Discussion

Our work demonstrated that in the absence of the F1L protein, the MVA virus activates NLRP1 in a ZAKα-dependent fashion. We showed that infection with MVA virus, similar to infection with VACV [12], inhibits endogenous NLRP1 inflammasome in keratinocytes. Unexpectedly, the MVA virus inhibited only ZAKα-dependent activation by RSR-induction and dsRNA transfection but not by DPP9 inhibition. In addition, the MVA virus did not inhibit another NLR family inflammasome sensor, NLRP3, suggesting that the inhibition is specific to NLRP1 and happens upstream of the sensor molecule. VACV has been reported to inhibit VbP-induced IL-1β release in keratinocytes [12]; however, we found that the MVA virus decreased the levels of pro-IL-1β, which could also be the case for VACV. In support of this hypothesis, we found that MVA virus decreased NLRP3-mediated IL-1β but not IL-18 release. We also found that MVA did not inhibit murine Nlrp1b, suggesting that in studies where F1L-deficient VACV or MVA were used for mice infection or immunization, any effects are independent of the Nlrp1 inflammasome [11, 23].

We confirmed that F1L expression during MVA virus infection was necessary for NLRP1 inhibition. Expression of F1L in N/TERT1 and HEK293T cells proved sufficient for NLRP1 inhibition. However, in contrast to MVA virus infection, in vitro transcribed F1L also inhibited NLRP3 and VbP-stimulated NLRP1 inflammasome. Control mRNA transfection caused a decrease in IL-1β release in ANS-stimulated cells and induced low amounts of IL-1β release from unstimulated NLRP1-expressing cells. These data suggest that mRNA transfection could induce low-level activation of the NLRP1 inflammasome while non-specifically blunting inflammasome signaling at the same time. On the other hand, F1L transiently expressed in HEK293T cells was specific for NLRP1.

Furthermore, we found that infection with F1L-deficient MVA virus activated NLRP1 inflammasome in a ZAKα-dependent manner. We demonstrated that MVA virus infection disrupted translation, which was more prominent for the ΔF1L virus. We also showed by ribosome runoff assays that ΔF1L MVA virus infection decreased the elongation rate, albeit not as strongly as ANS. This translation disruption could be the stimulus that causes NLRP1 inflammasome activation, as compounds that inhibit translation initiation and elongation have been shown to activate NLRP1 in ZAKα- or MAPK-dependent fashion [6, 7, 26, 27]. However, it is not clear how F1L prevents translation disruption and RSR induction during MVA virus infection. Further, our data suggest that it may act on multiple levels of the signaling pathway, preventing translational stress and inhibiting a signaling step downstream of ZAKα.

Data Limitations and Perspectives

In the present study, we confirmed the role of MVA virus-encoded protein F1L in NLRP1 inhibition. We showed that MVA virus infection and F1L expression prevented NLRP1 inflammasome activation in N/TERT1 and HEK cells. We demonstrated that the MVA virus inhibited ANS- and poly(I:C)- but not VbP-induced NLRP1 activation. These results suggest that the inhibition is indirect since a direct interaction would presumably inhibit NLRP1 activation in response to all stimuli. However, we have not identified which step in the signaling pathway is inhibited. We have not excluded that F1L binds to NLRP1 in a region that is redundant for activation by DPP9 inhibition but necessary for ZAKα-mediated activation. In addition, we demonstrated that the F1L-deficient MVA virus perturbs translation and activates NLRP1 in a ZAKα-dependent manner. These results suggest that F1L expression prevents RSR induction by MVA virus, acting upstream of ZAKα. However, the MVA virus also inhibited ANS without decreasing ZAKα and p38 phosphorylation, indicating that F1L could act on two levels, upstream and downstream of ZAKα. The mechanism by which F1L could prevent RSR remains unclear, be related to its function in preventing apoptosis.

Materials and methods

Cell culture

N/TERT1 immortalized human keratinocytes (a gift from J. Rheinwald) were cultured in a 1:1 mix of DMEM (high glucose, no glutamine, no calcium) and Ham's F12 (both Thermo Fisher Scientific) supplemented with 1% penicillin/streptomycin (Thermo Fisher Scientific), 1% nonessential amino acids (Thermo Fisher Scientific), 2 mM GlutaMAX (Thermo Fisher Scientific), 10 mM HEPES (Sigma Aldrich), 0.5% EpiLife defined growth supplement (Thermo Fisher Scientific), 25 µg/mL bovine pituitary extract (Thermo Fisher Scientific), 20 ng/mL epidermal growth factor (MPI of Biochemistry, CoFa), and 0.1 mM CaCl₂ (Sigma Aldrich). Knockout and transgenic cell lines were previously described [5]. HEK293T cells were cultured in DMEM (high glucose) supplemented with 1% penicillin/streptomycin, 1% sodium pyruvate, and 10% fetal calf serum (FCS, all Thermo Fisher Scientific).

Stimulations and viral infections

N/TERT1 cells were plated at 5 × 10⁴ cells/96-well plate for cytokine release experiments and at 2.5 × 10⁵, 5 × 10⁵, or at 1 × 10⁶ cells per 24-, 12-, or 6-well plate, respectively. The next day, the cells were treated as indicated with 2 µM Val-boroPro (ApexBio and Biozol), 2 µM ANS (Biomol), 100 ng/mL TNFα (PeproTech), 6.5 µM Nigericin (Sigma Aldrich), 0.1 µM B/B homodimerizer (AP20187, MedChem Express), 20 µM Z-VAD-FMK (Peptanova), 10 µM cytarabine (Ara C, Sigma Aldrich), 1 µM MLN4924 (MLN), 1 µg/mL anthrax lethal factor, and 1 µg/mL protective antigen (both List Biological Labs). For poly(I:C) transfection, a transfection mix of poly(I:C) (Invivogen) and Lipofectamine 2000 (Thermo Fisher Scientific) in Opti-MEM (Thermo Fisher Scientific) was prepared according to the manufacturer’s instructions. For 96- and 12-well plates, 0.2 and 1.6 µg of poly(I:C) with 0.3 and 2.4 µL of lipofectamine were used, respectively. WT, ΔF1L, and rF1L MVA viruses kindly provided by Prof. Gerd Sutter (LMU München) were added as indicated for individual experiments.

In vitro transcription

For in vitro transcription of FLAG-tagged F1L from strain Ankara (codon-optimized) and FLAG-tagged mScarlet, Invitrogen mMESSAGE mMACHINE SP6 Transcription Kit was used. The gene was amplified from a plasmid by PCR, adding the SP6 promoter and poly(A) tail, and the product was in vitro transcribed according to the manufacturer’s instructions. For N/TERT1-cell transfection, a transfection mix of mRNA and Lipofectamine 2000 (Thermo Fisher Scientific) in Opti-MEM (Thermo Fisher Scientific) was prepared according to the manufacturer’s instructions. For 96- and 12-well plates, 0.2 and 1.6 µg of poly(I:C) with 0.5 and 4 µL of lipofectamine were used, respectively.

HEK293T cell transfection and ASC specking analysis

For analyzing ASC specking in MVA virus-infected cells, HEK293T cells expressing ASC and inflammasome sensors were plated at 5 × 10⁵/12-well plates and infected and stimulated as indicated the next day. At the end of the stimulation, the cells were detached, and ASC specking was analyzed using a BD FACS Melody flow cytometer. For analyzing specking after F1L expression, the cells were transfected by reverse transfection. In short, a transfection mix of 0.1 or 0.2 µg of pEYFP-C3 (encoding mEYFP) or pEYFP-MVA029L-C3 (codon-optimized F1L from MVA strain labeled with N-terminal mEYFP), respectively, and 0.25 or 0.5 µL GeneJuice (Merck) or 1 µg of plasmid (0.5 µg for pEYFP-C3) and 3 µL GeneJuice (1.5 µL for pEYFP-C3) were prepared according to the manufacturer’s instructions. The cells were plated at 5 × 10⁴ or 5 × 10⁵ per 96- or 12-well, transfected, and stimulated the next day as indicated. At the end of the stimulation, the cells were detached, and ASC specking in YFP-positive cells was analyzed using a BD LSRFortessa or BD FACS Melody flow cytometer.

Ribosome runoff assay

A ribosome runoff assay was conducted as previously described [27]. In short, cells were treated with Z-VAD-FMK and stimulated as indicated. Five to one minute before the end of the stimulation, the cells were treated with 2 µg/mL of Harringtonine, and then 10 µg/mL of puromycin was added for 10 min. The supernatant was removed, the cells were lysed in Laemmli buffer, and puromycin incorporation was determined by immunoblotting.

Puromycin incorporation

Puromycin incorporation assay was conducted as previously described [27]. In brief, 30 min before the end of the stimulation, the cells were treated with 1 µg/mL of puromycin. The supernatant was removed, the cells were lysed in Laemmli buffer, and puromycin incorporation was determined by immunoblotting.

Retro- and lentiviral transduction

HEK cells constitutively expressing ASC-mCherry, NLRP1, and NLRP3 were generated by retro- and lentiviral transduction. For retro- and lentivirus production, HEK293T cells were used as previously described [5]. The cells were transfected by reverse transfection of a mix of transfer plasmid, helper plasmids (pCMV_Gag-Pol, pCMV_VSV-G, pMDLg/pRRE, pRSV-rev), and PEI. After 72 h incubation, the supernatant was harvested, and HEK293T cells were transduced. After retroviral transduction of ASC-mCherry (pRP-PYCARD-mCherry), mCherry-positive, nonspecking cells were sorted into a 96-well plate to generate monoclonal cell lines. One monoclonal cell line was expanded, and NLRP1 and NLRP3 transgenic cells were generated by lentiviral transduction (both pFUGW_Blast). For the selection of NLRP1- and NLRP3-expressing cells, 10 µg/mL of Blasticidin was added to the media. Transgenic N/TERT1 cells were previously described [5, 29].

DSS crosslinking

The DSS crosslinking procedure was performed as described [5]. In short, at the end of stimulation, the cells were detached, washed with PBS pH 8, and resuspended in PBS pH 8. Half of the cells were incubated with 4.4 mM DSS for 10–30 min at room temperature. Next, all cells were lysed in Laemmli buffer, and ASC dimerization was examined by immunoblotting.

Immunoblotting

For immunoblotting, the cells were lysed in Laemmli buffer and denatured for 5–10 min at 95°C. Supernatant proteins were precipitated using methanol and chloroform [30], resuspended in Laemmli buffer, and denatured for 5–10 min at 95°C. Proteins were separated by SDS PAGE using Tris-Glycine Mini Gels (Thermo Fisher Scientific) and blotted on a 0.2 or 0.45 µm nitrocellulose membrane (GE Healthcare #10600004). The membranes were blocked in 3% milk in TBST and incubated with primary and secondary antibodies. If possible, membranes were re-probed for a subsequent antibody incubation. A chemiluminescent signal was detected with a CCD camera. The following antibodies were used: anti-ASC (Adipogen #AL177), anti-caspase-1 (Adipogen #Bally-1), anti-GSDMD (Novusbio #NBP2-33422), anti-IL-1β (R&D Systems #AF-201-NA), anti-NLRP1 (BioLegend #9F9B12), anti-p38α (Cell Signaling Technology #9217S), anti-phospho-p38 (Cell Signaling Technology #9216S), anti-ZAKα (Biomol #A301-993A-T), anti-phospho-eIF2α (Cell Signaling Technology #9721), anti-puromycin (Merck #MABE343), anti-β-actin-HRP (Santa Cruz Biotechnology #sc-47778), donkey anti-goat IgG-HRP (Santa Cruz Biotechnology #sc-74405), goat anti-mouse IgG-HRP (Santa Cruz Biotechnology #sc-2005), and goat anti-rabbit IgG-HRP (Santa Cruz Biotechnology #sc-2004). For size reference, PageRuler Prestained Protein Ladder, 10–180 kDa (Thermo Fisher Scientific #26616), was used.

dsRNA staining

dsRNA staining was performed as described [5]. In detail, the N/TERT1 cells were plated at 5 × 10⁴ cells/8-well µ-slide (ibidi) and infected as indicated the next day. After stimulation, the cells were washed with PBS and fixed in 4% formaldehyde in PBS for 10 min at room temperature. Next, the cells were washed twice with PBS, permeabilized with 0.5% Triton X-100 in PBS for 10 min, washed twice with PBS, and blocked with 10% FCS in PBS overnight at 4°C. The next day, the cells were incubated with J2 antibody (SCICONS #10010200) diluted 1:400 in the blocking buffer for at least one hour at room temperature, washed three times with PBS, incubated with secondary goat anti-mouse IgG-Alexa Fluor 488 antibody (BioLegend #405319) diluted 1:500 in the blocking buffer for at least 1 h at room temperature. Before imaging, the cells were washed three times with PBS. dsRNA was imaged with a Leica DMi8 inverted microscope using HC PL FLUOTAR L 20X/o.40 DRY objective and ORCA-Flash4.0 LT+ Digital CMOS camera.

Elisa

IL-18 release was measured using Human Total IL-18 DuoSet ELISA (R&D #DY318-05) according to the manufacturer’s instructions. IL-1β release was measured using a homemade ELISA kit (Gevokizumab as capture antibody and biotinylated Canakinumab as detection antibody) according to the protocol from BD OptEIA Human IL-1β ELISA Set II (BD Biosciences #557953).

LDH release

LDH release was measured using CyQUANT LDH cytotoxicity assay (Thermo Fisher Scientific #C20300) according to the manufacturer’s instructions. LDH release was calculated using the following formula: % LDH release = (sample − unstimulated control)/(full lysis control − unstimulated control) × 100

ASC speck imaging

For imaging of ASC specking in ASC^KO N/TERT1 keratinocytes complemented with ASC-mCherry, the cells were plated at 5 × 10⁴ cells/8-well µ-slide (ibidi) and stimulated as indicated the next day. At the end of the stimulation, the cells were imaged with Leica DMi8 inverted microscope using HC PL FLUOTAR L 20X/o.40 DRY objective and ORCA-Flash4.0 LT+ Digital CMOS camera.

Statistics

The specific number of replicates, denoted as (n), is detailed in the legends accompanying each figure. GraphPad Prism 10 was used for data visualization and statistical analysis. In cases where a comparison bar is used to illustrate multiple comparisons, the extended line on the bar indicates the reference data set to which the significance level information pertains.

Data availability statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.