Cutting Edge: DNA in the Lung Microenvironment during Influenza Virus Infection Tempers Inflammation by Engaging the DNA Sensor AIM2 Free

Nucleic acid recognition is a primary mechanism driving protective host defenses during infection with a variety of microbial pathogens. In contrast to microbial products, such as LPS, which are unique to bacteria and not found in mammalian hosts, nucleic acids are common to both the pathogen and the host cell they infect. During the course of infection when viral pathogens replicate, nucleic acids accumulate within cells, and distinct families of pattern recognition receptors, including RIG-I–like receptors, Nod-like receptors, TLRs, and PYHIN proteins, detect these molecules and mobilize intracellular signaling pathways, leading to antiviral gene expression (1, 2).

Under normal circumstances, host nucleic acids are segregated from endosomal or cytosolic pattern recognition receptors, and DNA located in the nucleus and mitochondria of healthy cells fails to be recognized by the innate immune system. During viral infection, endogenous danger signals, including DNA, can also be released from damaged or dying cells. It is unclear whether these danger signals are detected and whether they, in turn, alter the host’s response to the pathogen. We set out to address this question using influenza A virus (IAV). IAV is an orthomyxovirus with a (-) ssRNA genome, and the viral life cycle involves the creation of RNA species only. IAV is sensed by RIG-I, TLR7, and TLR3 (3), all of which mobilize antiviral defenses that curb viral replication, prevent viral spread, and activate adaptive immunity.

We report the abundant release of endogenous DNA into the lung microenvironment, which is protective during infection with IAV. We identified an important role for the DNA sensor absent in melanoma 2 (AIM2) in sensing this host DNA. Mice lacking AIM2 were hypersusceptible to IAV infection. Although AIM2 contributes to the production of IL-1β early during IAV infection, the enhanced susceptibility of AIM2-deficient mice was not due to a failure in IAV-specific T cell responses. Rather, AIM2 appears to dampen inflammatory responses that would otherwise lead to excessive immunopathology. Thus, our study uncovers previously unappreciated cross-talk between pathogen- and damage-associated molecular pattern sensing in controlling the magnitude and duration of the host response to virus infection.

Aim2^−/− mice were generated as described (4, 5) and backcrossed to C57BL/6 mice. Wild-type (WT) and fully backcrossed Aim2^−/− mice were maintained and caged separately. All procedures used in this study complied with federal guidelines and were approved by the University of Massachusetts Medical School Institutional Animal Care and Use Committee.

Influenza virus A/Puerto Rico/8/1934 H1N1 (PR8) grown in chicken eggs was purchased from Charles River Laboratories (Wilmington, MA). Mice were anesthetized with isoflurane and inoculated via the intranasal route with 4 × 10⁴ PFU in 30 μl PBS. Viral titers in lung homogenates were measured by immunoplaque assay with MDCK cells. Monolayers were fixed, permeabilized, and stained with anti-IAV nucleoprotein (NP), followed by secondary goat anti-mouse HRP. Plaques were visualized using DAB substrate.

Bronchoalveolar lavage fluid (BALF) was harvested using 1 ml PBS and spun at 5000 × g for 10 min, and supernatants were kept as cell-depleted BALF. dsDNA in BALF supernatants was quantitated by PicoGreen assay (Life Technologies). Albumin levels in BALF were quantified by ELISA (GenWay).

Mice were perfused with PBS in the right ventricle. Cell suspensions from total lung tissue were stained for TCRβ, CD19, B220, CD11b, CD11c, Ly6G, Ly6C, NK1.1 (all from eBioscience), CD45, CD4, CD8α (all from BD Biosciences), and LD Blue (Life Technologies). For tetramer staining, cells were stained for TCRβ, CD8α, CD4, CD44, and allophycocyanin-labeled tetramers (K^b;NP 366, PA 244, PB1 703) or I-A^b/NP 311 (Trudeau Institute, Saranac Lake, NY). Live cells were gated based on forward and side scatter and Live/Dead Blue negative staining prior to subsequent gating. Data acquisition was performed on an LSR II (BD Biosciences), and data were analyzed using FlowJo (TreeStar).

Lung tissues were homogenized in 500 μl PBS and spun, and levels of IL-1β, IL-6, and TNF-α in supernatants were quantified by ELISA (eBioscience).

Briefly, mouse DNase I was cloned into a recombinant adeno-associated virus (AAV) vector plasmid carrying a vector genome, with the expression cassette driven by CMV-enhanced chicken β-actin promoter and flanked by AAV2 inverted terminal repeats (6). The AAV-DNase I plasmid was cotransfected into HEK 293 cells with an AAV9 packaging plasmid and adenovirus helper plasmid. The recombinant virus was purified by the standard CsCl gradient sedimentation method and desalted by dialysis. Mice were anesthetized with isoflurane and inoculated via the intranasal route with 10¹⁰ PFU in 30 μl PBS ≥2 wk prior to experiments.

Data were analyzed using the two-tailed Student t test, comparing means between groups. Kaplan–Meier survival curves were analyzed by the Mantel–Cox log-rank test. Graphing and statistical analyses were done using GraphPad Prism.

While examining the lungs of influenza virus–infected mice, we noticed what appeared to be extracellular DNA extruded from necrotic cells primarily localized to the bronchi (Fig. 1A). The source of this DNA is likely necrotic bronchiolar epithelial cells or neutrophil extracellular traps, both of which were identified previously in IAV-infected animals (7, 8). We formally quantified the levels of extracellular DNA released within tissues by obtaining BALF that we depleted of host cells. BALF collected from uninfected mice had low quantities of dsDNA (Fig. 1B). Levels of dsDNA increased significantly as early as 1 d postinfection (dpi) and continued to increase in step with the spread of infection at 2, 3, and 6 dpi (Fig. 1B). Given that IAV is an orthomyxovirus with a negative-sense ssRNA genome that is known to only produce RNA intermediates during replication, host-derived DNA is the only plausible source of extracellular DNA in this model.

It is unknown what effect host signals, such as endogenous DNA, have on the quality or duration of the immune response to viruses. To test the role of DNA in this context, we generated a recombinant AAV to ectopically express mouse DNase I in the lungs of WT mice (Fig. 2A). DNase I is a secreted protein, normally absent from the lung with respect to its enzymatic activity and expression (9) (Fig. 2B). Intranasal infection with AAV-DNase I resulted in robust and stable expression of the transgene in the lung ≥2 wk after transduction (Fig. 2B). Importantly, the levels of DNA in the BALF of IAV-infected mice were significantly reduced in mice treated with AAV-DNase I compared with GFP-expressing control virus, confirming that the product of the DNase I transgene was secreted and enzymatically active (Fig. 2C). In response to lethal PR8 challenge, mice treated with AAV-DNase I showed a significant decrease in survival (40%) compared with controls (79.6%) out to 14 dpi following challenge with an LD₂₅ of PR8 (Fig. 2D). Comparison of viral loads in the lungs of AAV-GFP– and AAV-DNase I–treated mice at 3, 6, and 9 dpi revealed similar viral loads (Fig. 2E). FACS analysis of infiltrating leukocyte populations in the lungs of AAV-DNase I–treated mice revealed a significant increase in the number of CD4⁺ and CD8⁺ T cells compared with controls at 5 dpi (Fig. 2F–H). Combined, these data suggest that DNA present within the lung microenvironment during IAV infection is protective and may serve to limit the number of infiltrating T cells able to drive the development of immune pathology.

Upon finding that DNA accumulates in the lungs following IAV infection and that sensing of this DNA leads to protective responses, we thought it plausible that one or more DNA-sensing receptors might be engaged in the lungs of IAV-infected mice. To this end, we tested mice lacking AIM2, a well-described cytosolic DNA receptor required for activation of the inflammasome complex in response to dsDNA (5, 10, 11). Previous studies identified roles for inflammasomes in IAV infection. IAV infection was shown to regulate IL-1β via the NLRP3 inflammasome (12, 13). Each component of the NLRP3 inflammasome, including ASC and caspase-1, was shown to be required to drive the inflammatory response and leukocyte recruitment during IAV infection (13–15). To test the contribution of AIM2 during IAV infection, mice lacking AIM2 were infected with a lethal IAV challenge. Although 75% of WT mice survived IAV infection, only 28% of AIM2^−/− mice were alive at 14 dpi (Fig. 3A). There was no difference in viral load in the lungs of WT and Aim2^−/− mice at 3, 6, and 9 dpi; this suggested that, although AIM2 was required for protection during lethal IAV challenge, AIM2 did not directly impact IAV replication (Fig. 3B).

We next assessed whether the susceptibility of Aim2^−/− mice correlated with a defect in IL-1β production. Analysis of IL-1β, IL-1α, and IL-1Ra mRNA showed no defect in their induction between Aim2^−/− and WT mice at 3 and 5 dpi (data not shown). However, the levels of IL-1β protein in lung homogenates were significantly lower at 3 dpi in AIM2-deficient mice, but these levels increased by 5 dpi to match those of WT mice (Fig. 3C). Similar to the effects that we observed early for IL-1β levels, we found that IL-6 and TNF-α were also decreased in lung homogenates of Aim2^−/− mice at 3 dpi compared with WT controls (Fig. 3D, 3E). However, by 5 dpi, the levels of TNF-α and IL-6 were significantly elevated in Aim2^−/− mice (Fig. 3D, 3E). The early effects of AIM2 deficiency on these cytokines may result from decreased IL-1β production and IL-1R–dependent induction of these cytokines, or they may reflect a greater number of leukocytes able to produce cytokines in the lungs of Aim2^−/− mice. We observed a slight increase in albumin levels in the BALF of Aim2^−/− mice at 3 dpi compared with WT controls, suggesting that there was more extensive lung damage in the absence of AIM2 (Fig. 3F). We also performed bone marrow transplants to elucidate whether expression of AIM2 in hematopoietic cells or stromal cells was required for protection from lethal IAV infection. WT mice engrafted with Aim2^−/− bone marrow were significantly more susceptible to IAV infection compared with those receiving WT bone marrow (Fig. 3G). These data indicate that AIM2 in hematopoietic cells is responsible for the protective effects that we observed.

IL-1R signaling during IAV infection is required for eliciting a protective flu-specific T cell response (16). AIM2 is required for IL-1β production upon sensing DNA in the cytosol; thus, we sought to determine whether AIM2 deficiency impaired the generation of flu-specific T cells. Postinfection with a sublethal dose of PR8, we performed tetramer staining in lung and spleen cell suspensions with four class I and one class II IAV epitopes to determine the frequency and number of flu-specific CD8⁺ and CD4⁺ T cells, respectively. We found no significant difference in the frequency (Fig. 4A) of NP 366-tetramer⁺ CD8⁺ T cells in the lungs at 9 dpi. Similarly, the numbers of CD8⁺ T cells positive for tetramers bearing three different IAV epitopes (NP 366, PA 224, PB1 703) were unaffected by AIM2 deficiency (Fig. 4B). Similar results were found in both the lung and spleen for all class I and class II epitopes tested (data not shown). These data indicate that AIM2 plays a protective role during IAV infection and that the protective effect is unlikely to be due to the limited effects of AIM2 on early IL-1β production or the formation of flu-specific T cell responses, both of which are unaffected by AIM2 deficiency.

We next used flow cytometric analysis to examine whether the absence of AIM2 altered leukocyte recruitment to the lungs of IAV-infected mice. This analysis revealed a significant increase in CD45⁺ leukocytes in the lungs of Aim2^−/− mice compared with WT controls in single-cell suspensions prepared from total lung at 5 dpi (Fig. 4C). Looking at individual populations of immune cells within the CD45⁺ population, we found increased numbers of CD4⁺ and CD8⁺ T cells and immature macrophages in the lungs of Aim2^−/− mice (Fig. 4D–F). We observed no significant difference in either the frequency or number of neutrophils (Fig. 4G), B cells, monocytes, or NK cells (data not shown).

Overall, our studies demonstrate that host-derived DNA accumulates in lung tissue damaged by IAV infection and has the capacity to modulate the intensity of the host response. Using a mouse influenza model, we found that host-derived DNA released from necrotic cells accumulates in the lung, and the levels continue to rise as the infection spreads throughout the tissue. Ablation of extracellular DNA using DNase I (delivered via AAV) and genetic deletion of AIM2, a cytosolic sensor primarily known for its role in controlling inflammasome activation in response to cytosolic DNA, increased the susceptibility to IAV infection. Although AIM2 affected the magnitude of the IAV-induced IL-1β response in vivo early during infection, this response was indistinguishable from that of WT mice at later time points. The increases in cellular infiltration and inflammatory cytokines observed in Aim2^−/− mice were not observed in prior studies that characterized the roles of NLRP3, ASC, and caspase-1 during IAV infection (13–15). Our findings suggest that the levels of IL-1β in Aim2^−/− mice are sufficient for mounting an immune response, and the observations reported in this article reflect an inflammasome-independent role for AIM2 in this context. Consistent with a largely intact IL-1β response, Aim2^−/− mice were able to mount normal flu-specific adaptive responses, which are known to require IL-1R signaling (16). AIM2-deficient mice had similar viral loads as those seen in WT mice, indicating that the effect of AIM2 in vivo also was not due to a failure to curb viral replication. However, the numbers of infiltrating leukocytes were increased in the lungs of Aim2^−/− mice. Moreover, the levels of TNF-α and IL-6, although slightly decreased early during infection, were also greater in Aim2^−/− mice than in WT mice at 5 dpi, likely as a result of the increased number of leukocytes in the lungs of the former. Consistent with our findings that a DNA sensor is able to regulate inflammation and morbidity during infection with an RNA virus, we also found that ablation of DNA in the lung microenvironment by ectopic DNase I expression yielded a similar effect to that observed in Aim2^−/− mice. Collectively, these findings suggest that accumulation of DNA in tissues damaged by infection may provide a mechanism for alerting the immune system to the extent of tissue damage and functions as a signal to limit excessive immune pathology. However, it remains to be seen whether the mechanism that we have identified in this study is specific to IAV infection or is more generalizable. We predict that similar mechanisms may be at play in other viral infections.

The identification of AIM2 as a regulator of tissue inflammation adds to our understanding of the cross-talk that exists between innate immune pathways. Further study of endogenous damage-associated molecular patters and their role in shaping antiviral immunity, repair, and regenerative responses will also lead to improved understanding of viral pathogenesis and could yield new insights for vaccine development.

We thank Dr. Zhaozhao Jiang and Kelly Army for help with animal husbandry, as well as Dr. Vijay Rathinam, Dr. Kai McInstry, and Dr. Gabriel Hendricks for technical assistance.

The authors have no financial conflicts of interest.