Poxvirus-encoded TNF receptor homolog dampens inflammation and protects from uncontrolled lung pathology during respiratory infection

Absence of CrmD Increases Susceptibility of C57BL/6 Mice to Infection with ECTV.

A mutant virus lacking both copies of CrmD (ECTV^ΔCrmD) and wild-type ECTV (ECTV^WT) were used. To dissect the contribution of the TNF- vs. the chemokine-binding domains, a virus expressing a mutant version of CrmD (point mutation in CRD) unable to bind TNF but retaining chemokine binding activity of the SECRET domain (ECTV^Rev.SECRET) was also used (22). The ability of CRD or the SECRET domain to bind TNF or chemokines, respectively, by each of these viruses is shown in SI Appendix, Table S1.

C57BL/6 wild-type (WT) mice, TNF-deficient (TNF^−/−) mice, and triple-mutant (TM) mice, engineered to express only noncleavable mTNF but not TNFRI or TNFRII (mTNF^Δ/Δ.TNFRI^−/−.TNFRII^−/−) were used. TNF^−/− and TM mice were used to determine whether the presence or absence of vTNFR impacted on the pathogenesis of ECTV infection in the absence of TNF or presence of only mTNF, respectively. The TM mice are devoid of an endogenous TNF response but can respond to exogenous TNFR or CrmD. The characteristics of these mouse strains with respect to expression of sTNF, mTNF, TNFRI, or TNFRII are shown in SI Appendix, Table S2. The TM mice are marginally more susceptible than mTNF^Δ/Δ mice, which express both TNFRI and TNFRII, to ECTV^WT infection (SI Appendix, Fig. S1). We predicted that CrmD binding to sTNF in WT mice can reduce the bioavailability of the cytokine, whereas its binding to mTNF may potentially modulate the inflammatory response further through reverse signaling (15, 16) in ECTV^WT-infected WT and TM mice but not in TNF^−/− mice.

Mice were inoculated intranasally (i.n.) with virus and monitored for changes in body weight and disease signs, which are presented as clinical scores using a scoring system (SI Appendix, Table S3) (41).

WT mice infected with the mutant viruses exhibited significant weight loss and higher clinical scores compared with ECTV^WT-infected animals from day 7 postinfection (p.i.) (Fig. 1 A and B). Clinical scores between groups of WT mice infected with either of the mutant viruses were comparable. However, ECTV^Rev.SECRET-infected mice exhibited significantly higher weight loss at days 7 to 9 compared to ECTV^WT-infected mice. TNF^−/− mice fared poorly regardless of the type of virus used for infection, and the presence or absence of CrmD and/or SECRET did not influence weight loss or clinical scores (Fig. 1 C and D). TM mice infected with ECTV^WT lost less weight than those infected with mutant viruses at days 8 and 9, but they were not statistically significant (Fig. 1E). The clinical scores in TM mice infected with any of the viruses were also similar until day 8, but by day 9, the clinical scores of ECTV^WT-infected mice were significantly lower than ECTV^ΔCrmD-infected animals (Fig. 1 E and F). WT and TM mice infected with ECTV^WT fared much better than all three strains infected with mutant viruses. Although no mortality was recorded in this experiment, all animals were killed at day 9 p.i. due to ethical and animal welfare reasons.

The results indicated that CrmD plays an important role in reducing weight loss and clinical symptoms during respiratory ECTV infection and that TNF is critical for CrmD to manifest its biological effects.

CrmD Deficiency Exacerbates Lung Pathology But Does Not Affect Viral Load in C57BL/6 Mice.

We investigated whether the significant morbidity among WT and TM animals infected with mutant viruses was due to exacerbated lung pathology, increased viral load, or both.

Fig. 2 shows hematoxylin and eosin (H&E)-stained lung sections from uninfected (Fig. 2 A–C) and virus-infected (Fig. 2 D–L) mice. WT mice infected with ECTV^WT exhibited minimal parenchymal and perivascular edema with minor damage to the epithelial lining of the bronchioles, and minimal diffusion of cellular infiltrates (Fig. 2D). In contrast, ECTV^WT-infected TNF^−/− mice exhibited severe lung injury with total destruction of most alveoli due to fluid accumulation, diffuse to multifocal infiltration of leukocytes, and extensive damage to the entire epithelial lining of the bronchioles (Fig. 2E). Lung pathology in TM mice infected with ECTV^WT was more severe than WT mice but less severe than TNF^−/− mice (Fig. 2F).

ECTV^ΔCrmD infection exacerbated lung pathology in WT mice with massive fluid accumulation, destruction of most alveoli, diffuse to multifocal infiltration of leukocytes, and significant damage to the epithelial lining of the bronchioles (Fig. 2G). The absence of CrmD did not impact on lung pathology in TNF^−/− (Fig. 2H) or TM (Fig. 2I) mice.

In ECTV^Rev.SECRET-infected WT (Fig. 2J), TNF^−/− (Fig. 2K), and TM lungs (Fig. 2L), most of the bronchial walls exhibited complete or near-complete epithelial necrosis, interstitial infiltration of leukocytes, alveolar septal wall thickening, and collapse of alveolar septal walls.

We semiquantified the histopathological changes in the lungs using a scoring system (SI Appendix, Table S4) (41). Five distinct observations were made from the analysis. First, WT lung sections had the lowest scores, TM lung sections had intermediate scores, and TNF^−/− lung sections had the highest scores in response to ECTV^WT infection (Fig. 3A). Second, WT mice infected with either mutant virus exhibited substantially greater lung histopathological scores. Third, in TNF^−/− mice, all three viruses caused considerable levels of lung damage with no significant differences in the histopathological scores. Fourth, TM lung sections from mice infected with mutant either virus had significant increases in their histopathological scores compared to those infected with ECTV^WT. Finally, the mutant viruses caused comparable levels of lung damage in all three strains of mice.

The lung viral load in all three strains of mice infected with ECTV^WT, ECTV^ΔCrmD, or ECTV^Rev.SECRET were comparable as the titers of viruses within each strain or across strains were not statistically different. This finding indicated that the TNF-binding or SECRET domains had no effect on viral replication in vivo (Fig. 3B). In vitro, the replication kinetics of ECTV^WT, ECTV^ΔCrmD, or ECTV^Rev.SECRET in bone marrow-derived macrophages (BMDMs) from WT (Fig. 3C), TNF^−/− (Fig. 3D), or TM (Fig. 3E) mice over a 48-h period were comparable, consistent with the in vivo results.

CrmD Regulates Leukocyte Recruitment/Infiltration into Lungs of ECTV-Infected Mice.

We used flow cytometry of collagenase-digested lungs from uninfected and virus-infected mice to ascertain whether the TNF-binding domain and/or the SECRET domain affected leukocyte recruitment into the lungs, which might have contributed to pathology. We enumerated the total number of leukocytes and leukocyte subsets at early (day 4) and later (day 8) stages of infection.

At day 4 p.i., numbers of leukocytes or leukocyte subsets did not increase significantly in WT animals following infection with ECTV^WT when compared with naive animals (Fig. 4 A–K). However, in the lungs of ECTV^ΔCrmD-infected animals, there were significant reductions in numbers of leukocytes (CD45.2⁺) (Fig. 4A) and leukocyte subsets [T cells (CD3⁺)] (Fig. 4B), T cell subsets (CD3⁺CD8⁺, CD3⁺CD4⁺, CD3⁺TCR γδ⁺) (Fig. 4 C–E), B cells (CD45R⁺) (Fig. 4F), NK cells (NK1.1⁺) (Fig. 4H), neutrophils (CD11b^hiLy6c^low) (Fig. 4J)], and inflammatory monocytes (CD11b^hiLy6c^hi) (Fig. 4K) compared to those infected with ECTV^WT or ECTV^Rev.SECRET. Although the absence of CrmD did not affect numbers of macrophages (F4/80⁺) (Fig. 4G) or dendritic cells (CD11c⁺) (Fig. 4I), the presence of only the SECRET domain in ECTV^Rev.SECRET-infected animals significantly increased numbers of these subsets. The SECRET domain appeared to be critical for leukocyte recruitment into the lungs of infected WT mice.

In contrast to WT mice, TNF^−/− animals did not exhibit any statistically significant increases or decreases in leukocyte subsets present in the lungs regardless of the type of infecting virus. This finding further indicated that TNF is critical for the function of CrmD and that any biological consequences of CrmD interacting with LT-α or LT-β had minimal impact on leukocyte recruitment.

In TM mice, the total infiltrating leukocytes (CD45.2⁺) (Fig. 4A) and most subsets examined were significantly higher in numbers in lungs of ECTV^WT-infected animals compared to uninfected or mutant virus-infected animals. The absence of CrmD reduced leukocyte subset numbers in ECTV^ΔCrmD-infected mice, similar to the outcome in WT mice. However, unlike in WT mice, the presence of only the SECRET domain in ECTV^Rev.SECRET-infected mice did not contribute to any increases in numbers of leukocyte subsets. As TM mice do not produce sTNF, TNFRI, and TNFRII, the lack of any effect suggested that endogenous TNF signaling was likely necessary for SECRET to mediate its activity in relation to leukocyte recruitment into the lungs.

At day 8 p.i., leukocyte numbers in WT lungs were comparable irrespective of the infecting virus as was the case in TNF^−/− mice (SI Appendix, Fig. S2). The one exception was a significant reduction in the number of B cells (CD45R⁺) in TNF^−/− mice infected with ECTV^Rev.SECRET compared with those infected with ECTV^ΔCrmD or ECTV^WT. TM mice were not included in the day 8 p.i. analysis of lung leukocytes as sufficient numbers of age- and sex-matched animals were unavailable. Results from the day 8 time point suggested that the exacerbated lung pathology seen at this time cannot be attributed to leukocyte numbers.

vTNFRs Dampen Inflammation In Vitro through Reverse Signaling.

The binding of TNFR to mTNF on a cell can induce reverse signaling in vitro and in vivo to dampen inflammatory gene expression (15, 16). We reasoned that CrmD and other OPV-encoded vTNFRs, which are known to bind mTNF (22), could also potentially reverse signal through mTNF and dampen inflammation in the lungs of infected hosts. Treatment with lipopolysaccharide (LPS), a potent inducer of TNF, stimulated the production of sTNF and mTNF in the murine macrophage cell line RAW264.7 and expectedly, it only induced mTNF in BMDMs from TM mice (SI Appendix, Fig. S3). To demonstrate reverse signaling, we used BMDMs from TM mice stimulated with LPS to induce mTNF and inflammatory gene expression, and subsequently treated the cells with vTNFR. We used the Mouse Signal Transduction Pathway Finder PCR array that screened for 84 genes (SI Appendix, Fig. S4 and Table S5; (42)). If vTNFRs dampened inflammation, we expected to see the down-regulation of mRNA transcripts for some LPS-induced inflammatory genes. We used recombinant CrmD from ECTV, CrmB, CrmC, and CrmD from cowpox virus (CPXV) and CrmB from VARV, expressed in the baculovirus system and purified as described (25).

LPS stimulation up-regulated the mRNA for 22 of the 84 genes that were screened (SI Appendix, Fig. S4). Treatment with Crm proteins down-regulated 18 of the 22 mRNA up-regulated by LPS (Table 1). Individual Crm proteins had varying effects on reducing mRNA levels, with CPXV CrmD downregulating the greatest number of mRNA transcripts induced by LPS. All of the Crm proteins significantly reduced IL-1α, CCL2, and IL-2Rα mRNA levels following reverse signaling (Table 1). The pathways that were most affected by reverse signaling such as NF-κB and JAK-STAT cytokine signaling pathways are involved in inflammatory processes. Crm proteins also modulated those genes that were down-regulated by LPS stimulation (SI Appendix, Table S6) and these are represented by several signaling pathways including stress, NF-κB, CREB, Wnt, Hedgehog, and PIA3K/AKT pathways.

Absence of CrmD Results in Dysregulated Expression of Proinflammatory Mediators.

LPS stimulation of BMDMs followed by vTNFR treatment may not be a true reflection of changes in mRNA expression that can occur in the various cell types present in an organ as complex as the lung during viral infection, where many cell types produce TNF (41) and can thus be subjected to reverse signaling. We therefore determined changes in mRNA transcripts in lung tissue of WT, TNF^−/−, and TM mice at day 9 p.i. using quantitative reverse transcription real-time PCR (qRT-PCR). We also measured the protein levels of some of the cytokines in separate groups of infected WT mice.

WT and TNF^−/− mice do not express detectable levels of IFN-γ, IL-6, IL-10, TGF-β, and TNF (WT mice only) (41). The cytokine responses to ECTV^WT infection in WT and TM lungs were similar, whereas the levels of most mRNA transcripts were significantly higher in TNF^−/− lungs compared to WT or TM lungs (Fig. 5 A–J). The absence of CrmD in ECTV^ΔCrmD-infected WT mice led to significant increases in levels of mRNA transcripts for all mediators compared with ECTV^WT infection. In the presence of the SECRET domain in ECTV^Rev.SECRET-infected WT mice, three types of responses were apparent when compared with ECTV^ΔCrmD infection. First, IL-1β, IL-12p40, and TGF-β mRNA levels were reduced. Second, mRNA levels of IL-6 and IL-10 were increased significantly. Third, mRNA levels of TNF, IL-1α, IFN-γ, CCL2, and NOS2 were similar but significantly higher than levels induced by ECTV^WT.

Expectedly, TNF^−/− mice did not express mRNA for TNF, but infection with ECTV^WT induced significantly higher levels of several inflammatory mediators compared to levels in WT or TM mice. The only exception was NOS2 mRNA, whose levels were similar to those in WT and TM mice. IL-1β, IL-6, and NOS2 mRNA levels in TNF^−/− mice infected with any one of the three viruses were comparable. Some mRNA transcripts in TNF^−/− mice infected with either of the mutant viruses changed significantly. IL-10, IFN-γ, and CCL2 levels increased, whereas IL-1α, IL-12p40, and TGF-β levels decreased regardless of whether both domains were absent or only the SECRET domain was present. Thus, any changes in mRNA expression in TNF^−/− mice infected with ECTV^ΔCrmD or ECTV^Rev.SECRET were clearly not due to interactions of LT-α and/or LT-β with the CRD as the both the mutant viruses do not express the TNF-binding domain.

In TM mice, CrmD deficiency led to increases in levels of expression of a number of mediators, except IL-1α, IL-12p40, and TGF-β. TNF and IL-1β levels were significantly higher in the ECTV^ΔCrmD-infected group compared to ECTV^WT- or ECTV^Rev.SECRET-infected mice. Similarly, IL-6 and IL-10 levels were higher in the absence of CrmD, reduced when only the SECRET domain was present, but were clearly higher than in ECTV^WT-infected animals. NOS2 mRNA transcripts were only increased in the lungs of mice infected with ECTV^ΔCrmD.

The CRD and SECRET domain differentially regulated mRNA transcripts for a number of inflammatory genes in the three different strains of mice. In WT animals, the CRD was particularly important for dampening TNF, IL-1α, IL-1β, TGF-β, CCL2, IFN-γ, and NOS2, whereas the SECRET domain appeared critical for induction of IL-6 and IL-10. In TM mice, the presence of the SECRET domain alone reduced levels of CCL-2, IFN-γ, IL-1β, IL-6, IL-10, and TNF compared to levels in the absence of both the TNF-binding and SECRET domains.

We measured the levels of some cytokines by ELISA in lung homogenates of WT mice infected with ECTV^WT, ECTV^ΔCrmD, or ECTV^Rev.SECRET (SI Appendix, Fig. S5). The absence of CrmD in ECTV^ΔCrmD-infected mice increased levels of TNF, IL-10, and CCL2, whereas the presence of only the SECRET domain in ECTV^Rev.SECRET-infected mice significantly increased levels of IL-6 and IL-10, consistent with the gene expression data shown in Fig. 5. CrmD deficiency induced the highest levels of IL-1β and IL-12p40 mRNA transcripts; however, the SECRET domain alone was sufficient to induce significantly higher levels of these proteins. Finally, while the absence of CrmD significantly increased levels of mRNA transcripts for IFN-γ, the protein levels were comparable in mice infected with ECTV^WT, ECTV^ΔCrmD, or ECTV^Rev.SECRET. Expression of mRNA transcripts and protein for some of the cytokines were incongruent (Fig. 5 and SI Appendix, Fig. S5). This may have been due to distinct expression kinetics, posttranscriptional regulation of cytokine gene expression, and/or the cytokine milieu in lungs of the different mouse strains infected with different viruses, which may have had an impact on protein translation.

Blockade of IFN-γ, IL-6, IL-10R or TNF Reduces Weight Loss and Disease and Increases Survival Rates.

The increased production of a number of inflammatory cytokines in mice infected with mutant ECTV possibly contributed to the increased morbidity and mortality, as shown previously in ECTV-infected TNF^−/− mice (41). We postulated that the in vivo blockade of some of these cytokines might reduce the severity of the clinical symptoms and lung pathology and possibly allow the recovery of mice infected with ECTV^ΔCrmD or ECTV^Rev.SECRET. Groups of WT mice infected with either mutant virus were treated with mAb against IFN-γ, IL-6, IL-10R, or TNF, beginning at day 7 p.i. and treatment continued every alternate day until day 20 p.i. and surviving animals were killed on day 21. We selected day 7 p.i. to start cytokine blockade therapy as this is the time when animals start losing weight, have relatively higher clinical scores, and lung pathology starts to develop. Control groups were treated with an isotype-matched rat IgG mAb, and the data are presented in Fig. 6.

In ECTV^ΔCrmD-infected WT mice, blockade of IFN-γ, IL-6, TNF, or IL-10R with specific mAb significantly increased survival rates compared to the rat IgG isotype control mAb-treated group (Fig. 6A). The control mAb-treated group exhibited 100% mortality by day 8 p.i., whereas 60% of animals treated with mAb against TNF or IL-10R were alive at day 21. Anti–IL-6 mAb treatment resulted in the survival of 80% of animals in the group, whereas IFN-γ blockade rescued 100% of the mice. Individually, IFN-γ, IL-6, TNF, and IL-10 were necessary and sufficient to cause significant morbidity and mortality due to lung pathology, but the blockade of just one of the cytokines was sufficient to protect 60 to 100% of the mice.

ECTV^Rev.SECRET-infected mice also succumbed to infection by day 8 p.i. (Fig. 6B). Survival rates significantly increased from 0 to 60% and from 0 to 80% after treatment with anti-TNF or anti–IL-10R mAb, respectively. There was 100% survival in the group treated with anti–IL-6 (Fig. 6B). Thus, blockade of IL-6, TNF, or IL-10R resulted in similar outcomes in animals infected with ECTV^ΔCrmD or ECTV^Rev.SECRET. The only exception was anti–IFN-γ treatment of ECTV^Rev.SECRET-infected mice, which had minimal effects on the outcome and all animals succumbed to infection. This finding is in complete contrast to ECTV^ΔCrmD-infected mice, which were fully protected from anti–IFN-γ treatment. Thus, IFN-γ, IL-6, TNF, and IL-10 are responsible for the pathology in the absence of the CRD and the SECRET domain in ECTV^ΔCrmD-infected mice. In contrast, in the presence of only the SECRET domain, IFN-γ was not sufficient to cause pathology.

Weight loss (Fig. 6 C and D) and clinical scores (Fig. 6 E and F) were also markedly lower after anti-cytokine treatment compared to the control mAb-treated groups. Mice treated with anti-cytokine mAb regained weight beginning at day 12 and those mice that recovered had similar clinical scores from this time. Improvements in weight loss and clinical scores correlated with the survival rates, as evidenced by body weight recovery and the absence of any clinical signs and symptoms in mice that were alive at day 21 p.i.

The histopathological scores (Fig. 6 G and H) generated from lung histology sections (SI Appendix, Fig. S6) were highest in those animals that had the highest clinical scores, lost significant body weights, and succumbed to mousepox compared to those animals that had survived. The viral load was very high in those animals that died early between days 8 and 12, but was at or below the limit of detection in animals that recovered and killed at day 21 (Fig. 6 I and J). Of particular interest is the outcome of IFN-γ blockade in mice infected with ECTV^ΔCrmD or ECTV^Rev.SECRET. In mice infected with ECTV^ΔCrmD and treated with anti–IFN-γ, viral load was below the limit of detection and the histopathological scores were substantially reduced. In contrast, in mice infected with ECTV^Rev.SECRET and treated with anti–IFN-γ, the viral load and histopathological scores were comparable with those of control mAb-treated mice. The results indicate that excessive IFN-γ production in mice infected with ECTV^ΔCrmD, when both CRD and SECRET domain are absent, results in significant lung pathology. In contrast, excessive IFN-γ production in mice infected with ECTV^Rev.SECRET, when only the SECRET domain is present, is not sufficient to cause pathology. It is possible that the significantly higher levels of IL-1β, IL-6, and IL-10 in ECTV^Rev.SECRET-infected mice, compared to levels induced by ECTV^ΔCrmD infection, could have overcome any protective effects that might have been afforded by IFN-γ blockade.

Animal Studies.

Animal experiments were performed in accordance with protocols approved by the Animal Ethics and Experimentation Committee of the ANU (protocol numbers A2011/011 and A2014/018) and the Animal Ethics Committee of the University of Tasmania (protocol number A0016372). Six- to 12-wk-old female WT, Tnf^tm1Jods (TNF^−/−) (2), Tnf^tm2Jods (mTNF^Δ/Δ) (52), and Tnf^tm2JodsTnfrsf1a^tm1ImxTnfrsf1b^tm1Imx (designated triple mutant, TM) mice on a C57BL/6 background were bred under specific pathogen-free conditions at the Australian Phenomics Facility, Australian National University, Canberra, Australia. We generated the TM mice by crossing the mTNF^Δ/Δ with TNFRI- and TNFRII-deficient mice (53). Animal experimental setup, virus inoculation, clinical scoring, weighing mice, and treatment with mAb were carried out as previously described (41) and outlined in SI Appendix. Details of sTNF, mTNF, TNFRI, and/or TNFRII expression by each mouse strain are presented in SI Appendix, Table S2.

Cell Lines and Viruses.

BS-C-1 (ATCC CCL-26) cells were cultured in Eagle’s minimum essential medium (EMEM) supplemented with 2 mM l-glutamine (Sigma-Aldrich), antibiotics (penicillin, 120 µg/mL; streptomycin, 200 µg/mL; and neomycin sulfate), 1 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (Hepes), and 10% heat-inactivated fetal calf serum. This medium referred to as complete EMEM10. RAW264.7 (RAW 264.7; ATCC TIB-71) and NCTC clone 929 (L-929 cells; ATCC CCL-1) cells were cultured in complete DMEM (DMEM10) with additional supplements of 1 mM sodium pyruvate and 1 mM nonessential amino acids.

The Naval (22, 31) and Moscow (35) ECTV strains were propagated and quantified as described (54). The CrmD mutant viruses were generated from the Naval strain of ECTV^WT (22, 31). Both strains of virus encode CrmD, with a single amino acid mutation (F263L) in a 320-amino acid polypeptide, have minor genetic differences (mainly due to different ways to annotate the viral genes) and exhibit similar degrees of virulence in BALB/c and C57BL/6 mice (55). A schematic overview of the three viruses with respect to their ability to bind TNF and/or chemokines is presented in SI Appendix, Table S1. For more information, see SI Appendix, Materials and Methods.

Generation of BMDMs and Virus Replication Kinetics.

BMDMs from WT, TNF^−/− and TM mice were generated as described elsewhere (56) with some modifications and used to determine whether the CRD or SECRET domain influence ECTV replication in vitro. L929-cell conditioned medium was used as a source of macrophage-colony stimulating factor to generate BMDMs. For more details, see SI Appendix, Materials and Methods.

Viral Plaque Assay for Quantification of ECTV.

To enumerate the number of virus plaque-forming units (PFUs) present in samples, a viral plaque assay using BS-C-1 cells was used as previously described (54).

Reverse Signaling through mTNF in BMDMs from TM Mice.

A total of 5 × 10⁶ mature BMDMs generated from TM mice was resuspended in 10 mL of complete DMEM10 and added to 25-cm² tissue culture flasks. Cells were allowed to adhere for 2 h, prior to addition of LPS at 50 ng/mL, and incubated for 6 h. To induce reverse signaling through mTNF, recombinant Crm proteins were added to individual flasks at 10 μg/mL for 3 h. Cells were then harvested for isolation of RNA, generation of cDNA, and qRT-PCR analysis.

Histology and Microscopic Assessment of Lung Pathology.

Histological assessment of H&E-stained lung sections was undertaken as described previously (41). For more details, see SI Appendix, Materials and Methods.

Flow-Cytometric Analysis of Lung Leukocytes.

Lung tissue samples were digested with DNase I and collagenase, and flow cytometry of lung leukocytes was undertaken as described (41). Data were acquired on a BD LSR Fortessa flow cytometer with BD FACS Diva software and analyzed using FlowJo software, version 9.5 (Tree Star). For more details, see SI Appendix, Materials and Methods.

Detection of Cytokine Proteins in Lung Homogenates.

The levels of cytokines and chemokines in the lung homogenates were measured using capture ELISA kits (Biolegend), performed according to the manufacturer’s protocol. Optical density was measured at 450 nm with SOFTmax Pro software (Molecular Devices Corporation). For details on lung homogenate preparation, see SI Appendix, Materials and Methods.

RNA Extraction, cDNA Generation, and qRT-PCR.

RNA extraction, cDNA synthesis, and qRT-PCR were carried out as described (41). For more details, see SI Appendix, Materials and Methods.

Viral TNFR Expression and Purification.

Recombinant vTNFRs from ECTV (CrmD), CPXV (CrmB, CrmC, and CrmD), and VARV (CrmB) were expressed in the baculovirus system, fused to a C-terminal V5-6xHis tag as described elsewhere (25). Proteins were purified in metal chelate affinity columns from supernatants of infected Hi5 cell cultures. Permission from the World Health Organization was granted to hold VARV DNA encoding CrmB, and its manipulation was performed in accordance with the established rules.

Statistical Analysis.

Statistical analyses of experimental data, as indicated in the legend to each figure, were performed using appropriate tests to compare results using GraphPad Prism 8 (GraphPad Software). A value of P < 0.05 was taken to be significant: *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001.