Acute infection is implicated as a trigger for chronic inflammatory disease, but the full basis for this switch is uncertain. In this study, we examine this issue using a mouse model of chronic lung disease that develops after respiratory infection with a natural pathogen (Sendai virus). We investigate this model using a combination of TLR3-deficient mice and adoptive transfer of immune cells into these mice versus the comparable responses in wild-type mice. We found that acute and transient expression of TLR3 on monocyte-derived dendritic cells (moDCs) was selectively required to induce long-term expression of IL-33 and consequent type 2 immune-driven lung disease. Unexpectedly, moDC participation was not based on canonical TLR3 signaling and relied instead on a trophic effect to expand the alveolar epithelial type 2 cell population beyond repair of tissue injury and thereby provide an enriched and persistent cell source of IL-33 required for progression to a disease phenotype that includes lung inflammation, hyperreactivity, excess mucus production, and remodeling. The findings thereby provide a framework wherein viral infection activates TLR3 in moDCs as a front-line immune cell niche upstream of lung epithelial cells to drive the type 2 immune response, leading to chronic inflammatory diseases of the lung (such as asthma and chronic obstructive pulmonary disease in humans) and perhaps progressive and long-term postviral disease in general.
This article is featured in Top Reads, p.1115
To prevent the development of chronic inflammatory disease, it would seem critical to determine the early events that initiate the disease process. It is possible that infection might trigger this process, but a clear connection between acute infection and chronic disease as well as a corresponding immune mechanism remain uncertain. A useful paradigm for understanding this issue is represented by one of the most common types of chronic inflammatory disease (i.e., the spectrum of chronic lung disease that includes asthma and chronic obstructive pulmonary disease [COPD]) (1–3). Moreover, these types of diseases are subject to initiation, exacerbation, and/or progression due to infection, particularly with respiratory viruses (4). A critical challenge then is to determine the nature of the immune response that is responsible for connecting acute respiratory viral infection to chronic inflammatory lung disease.
In that regard, the development of a mouse model of chronic lung disease that develops after virus infection has implicated a central role for the innate immune system in the disease process. In this model, infection with a natural pathogen known as Sendai virus (SeV) leads to expansion and activation of a subset of lung epithelial cells as a source of cytokine IL-33 and, in turn, immune cell production of IL-13 and consequent airway inflammation, hyperreactivity, mucus production, and structural remodeling (5–11). This innate immune axis is relevant to human disease because the same immune components are implicated in the initiation, exacerbation, and/or progression of chronic and often progressive lung disease in the form of asthma or COPD (6, 7, 12, 13). Moreover, a similar progression from illness to disease after clearance of infectious virus might also occur after severe respiratory viral infections in general, including influenza virus and coronavirus, as found in Covid-19 (14, 15).
In the present work, we aimed to better define the initial events that trigger the downstream activation of the IL-33/IL-13–driven type 2 immune response for inflammatory lung disease after respiratory viral infection. We recognized that respiratory viruses (including SeV) rapidly activate several types of pathogen-recognition receptors (PRRs), including TLRs (notably TLR3, TLR7, and TLR8), RIG-I–like receptors (RLRs; such as RIG-I and melanoma differentiation–associated protein 5 [MDA5]), and NOD-like receptors (NLRs; such as NALP3) (4). This early recognition system is actuated in concert with the recruitment of immune cells such as tissue macrophages that are presumably helpful for repair of cell and tissue injury (9, 16–20). However, any link between these acute illness events and the subsequent development of chronic disease was uncertain. Indeed, it was difficult to understand why this type of disease would develop long after infectious virus is cleared and injury is repaired. Initial insight into an upstream immune mechanism was provided from our recent report of the response to SeV infection in Csf1-deficient wt/opT mice that are missing the population of myeloid–macrophage lineage cells. This deficiency resulted in a marked attenuation in the usual expansion of alveolar epithelial type 2 (AT2) cells (11) that are generally thought to be responsible for at least some aspects of postviral lung repair (21). However, this population is also the primary source of IL-33 in the mouse lung (11, 22–24), so loss of the myeloid–macrophage lineage cells also blocked the IL-33–driven type 2 immune response responsible for postviral disease. In this study, we extend this work to show that monocyte-derived DCs (moDCs) provide this trophic signal to a progressive and excessive AT2 cell expansion and further that moDCs require early activation of TLR3 signaling to function as a front-line immune cell niche for expansion of AT2 cells. This finding results in a revision of the conventional view of the epithelial–immune cell interface in the development of inflammatory disease after viral infection, as proposed in the past (4). We thereby define an early and upstream immune cell event that is required for long-term inflammatory disease and appears relevant to virus-linked lung diseases such as asthma, COPD, and the progressive dysfunction that can develop after any severe respiratory viral infection.
Materials and Methods
Mouse generation and infection
Wild-type (WT) C57BL/6J mice (no. 000664) and Trif−/− mice (no. 005037) were purchased from The Jackson Laboratory. The Tlr3−/− mice were generated originally by R. Flavell (Yale University, New Haven, CT) (25) and were obtained from M. Diamond and M. Colonna (Washington University, St. Louis, MO). All mouse strains were fully backcrossed onto the C57BL/6J genetic background. Mice were infected with SeV (Sendai/52, Fushimi strain) given intranasally (2 × 105 PFU per mouse) or an equivalent amount of UV light–inactivated virus or PBS as described previously (26). Viral titers for stock solutions and lung infections were monitored by PCR-based assay as described previously (10). The animal studies committee of the university approved all experimental protocols. Male and female mice (6–10 wk of age) were used for all experiments; however, no significant sex differences were observed in response to viral infection as reported previously (10).
RNA was purified from homogenized lung tissue using TRIzol (Invitrogen) or from isolated cells with the RNeasy Mini Kit (QIAGEN) and was used to generate cDNA with the High-Capacity cDNA Archive Kit (Life Technologies). We quantified target mRNA and viral RNA levels using real-time quantitative PCR assay with specific fluorogenic probe-primer combinations and Fast Universal PCR Master Mix systems (Applied Biosystems) with forward and reverse primers and probes as described previously (11) in addition to Tlr3 for the present work using assay Mm.PT.58.8085919 with forward primer 5′-CCTGTATCATATTCTACTCCTTGCT-3′, reverse primer 5′-GACGCACCTGTTCTCTATCTG-3′, and probe 5′/56-FAM/AACTGCCTG/ZEN/AATCACAATCGCGC/3IABkFQ/-3′ from Integrated DNA Technologies. Samples were assayed using the 7300HT or QuantStudio 6 Fast Real-Time PCR System and analyzed using Fast System Software (Applied Biosystems). All real-time PCR data were normalized to the level of GAPDH mRNA. Values were expressed as fold change based on the Δ-Δ Ct method as described previously (27), with the exception of viral RNA, which was quantified by copy number using an SeV nucleoprotein (NP)–expressing plasmid as an internal standard (10).
Lung tissue was fixed with 10% formalin, embedded in paraffin, cut into 5-μm sections, and placed on charged slides. Prior to staining, sections were deparaffinized in Fisherbrand CitroSolv (Thermo Fisher Scientific), hydrated, and heat-treated with Ag unmasking solution (Vector Laboratories). Sections were stained with periodic acid–Schiff (PAS) and hematoxylin as described previously (10, 14). In addition, immunostaining was performed with the following primary Abs: anti–mucin 5AC (MUC5AC; clone 45M1; Thermo Fisher Scientific), rabbit anti-mouse F4/80 (clone D2S9R; Cell Signaling Technology), chicken anti-GFP (Abcam), goat anti-mouse IL-33 (R&D Systems), and rabbit anti–surfactant protein C (Sftpc; Abcam). For detection of primary Ab binding, sections were incubated in Alexa Fluor 488– or 594–conjugated secondary Abs and counterstained with DAPI-containing mounting media (Vector Laboratories). Slides were imaged by light microscopy using a Leica DM5000 B microscope and by immunofluorescent microscopy using an Olympus BX51 microscope, and staining was quantified in whole lung sections using a NanoZoomer S60 slide scanner (Hamamatsu) and ImageJ software as described previously (10, 14).
Lung cell suspensions were prepared from lung tissue by mincing, followed by digestion in digestion buffer containing collagenase (Liberase Blendzyme III; Roche Diagnostics) and hyaluronidase (Sigma-Aldrich) and DNAse I (grade II; Roche Diagnostics) for 45 min at 37°C. For epithelial cell isolation, Dispase II (Roche Diagnostics) was added to this buffer. Cells were briefly incubated in ACK Lysing Buffer (Lonza) to lyse RBCs and washed before Fc receptor blockade using anti-mouse CD16/CD32 (clone 2.4G2; BD Biosciences). The isolated cells were purified using FACS with an iCyt Synergy high-speed flow cytometer (Sony Biotechnology) or FACSAria IIu (BD Biosciences). We used the following anti-mouse Abs: BV421 anti-CD11b (clone M1/70; BD Biosciences), PE-Cy7 anti–Ly-6G (clone 1A8; BD Biosciences), FITC anti-CD11c (clone HL3; BD Biosciences), APC-Cy7 anti–Siglec F (clone E50-2440; BD Biosciences), BV421 anti-CD45 (clone 30-F11; BD Biosciences), Alexa Fluor 647 anti-CD64 (clone ×54-5/7.1; BD Biosciences), APC-Cy7 anti-EpCAM/CD326 (clone G8.8; BioLegend), PE-Cy7 anti-CD31 (clone MEC 13.3; BD Biosciences), and PerCP-Cy5.5 anti-CD103 (clone M290; BD Biosciences). Specific combinations of mAbs were chosen to identify lung tissue macrophages (Ly-6G–, CD11c–, Siglec-F–, CD64+, and CD11b+), alveolar macrophages (Ly-6G–, CD11c+, Siglec F+, CD64+, and CD11b–), epithelial cells (CD45–, CD31–, and EpCAM/CD326+), and endothelial cells (CD45–, CD31+, and EpCAM/CD326–). Dendritic cell (DC) subsets were identified as CD11b+ DCs (Ly-6G–, CD11c+, Siglec F–, CD64–, and CD11b+), moDCs (Ly-6G–, CD11c+, Siglec F–, CD64+, and CD11b+), and CD103+ DCs (Ly-6G–, CD11c+, CD64+, Siglec F–, CD11b–, and CD103+). For analysis of TLR3 expression, cells were fixed and permeabilized with fixation/permeabilization solution (BD Biosciences) and stained with rat anti-mouse PE TLR3 IgG2a mAb (clone 11F8; BioLegend). Cells were counted using FACS and fluorescent counting beads (CountBright; Life Technologies). Analysis of flow cytometry data were performed using FlowJo software (Tree Star).
Lung tissues were homogenized in T-PER Tissue Protein Extraction Reagent (Thermo Fisher Scientific) with protease inhibitor mixture (cOmplete; Roche Diagnostics) and Halt phosphatase inhibitor mixture (Thermo Fisher Scientific) using a rotor homogenizer (Tissue-Tearor; Biospec Products). Levels of Ccl2, Ccl7, and IL-33 were measured using ELISA kits (R&D Systems and Abcam) and normalized to total protein.
Adoptive cell transfer
For adoptive cell transfer, lung-derived moDCs, tissue macrophages, or bone marrow–derived DCs (BMDCs) were delivered intranasally (1 × 106 cells in 30 μl of PBS) to recipient mice as described previously (28, 29). Mice treated identically with intranasal PBS served as controls. All mice were then infected with SeV or PBS control at 1 d after adoptive cell transfer or PBS treatments. The moDCs and tissue macrophages were obtained using FACS-based isolation from lung tissue of mice at 12 d after SeV infection. The BMDCs were generated from cultures derived from bone marrow aspirates maintained in RPMI 1640 containing 10% FBS and recombinant mouse GM-CSF (20 ng/ml; PeproTech) for 8 d at 37°C changed at 3 and 6 d as described previously (30). The BMDC-enriched nonadherent cells were harvested for adoptive transfer.
All data are presented as mean and SEM and are representative of at least three experiments with at least five data points per experiment. Unpaired Student t test with Bonferroni correction as well as mixed-model repeated measures ANOVA with Tukey correction for multiple comparisons were used to assess statistical significance between means. In all cases, significance threshold was set at p < 0.05.
TLR3 influence on acute illness after viral infection
Based on the role of TLR3 as a primary sensor for dsRNA derived from replication of RNA viruses, our initial experiments defined phenotype for Tlr3−/− mice during acute illness after SeV infection. We found that the usual decrease in body weight during acute illness was significantly, albeit only slightly, attenuated in Tlr3−/− mice compared with WT control mice (Fig. 1A). In contrast, we observed no significant differences in viral RNA levels during acute illness in Tlr3−/− mice compared with WT mice (Fig. 1B), thereby raising the possibility that there might be differences in other aspects of the host immune response. In fact, PAS–hematoxylin staining of lung sections showed a decrease in immune cell accumulation in lung sections from Tlr3−/− mice compared with WT mice at 5 d after SeV infection (Fig. 1C) that was significant based on image analysis of hematoxylin+ staining (Fig. 1D, 1E). In addition, we used a flow cytometry scheme for immune cells (Fig. 1F) to show that the increased levels of lung cells and in particular myeloid cells (i.e., tissue macrophages, moDCs, CD11b+ DCs, and CD103+ DCs) at 5 and/or 12 d after SeV infection were significantly attenuated in Tlr3−/− mice compared with WT mice (Fig. 1G). In contrast, levels of alveolar macrophages were increased at baseline and after SeV infection in Tlr3−/− compared with WT mice (Fig. 1G), consistent with distinct development and regulation of this cell population (9, 31). Together, the results suggested protection against myeloid cell accumulation and morbidity during acute viral illness in Tlr3−/− mice.
TLR3 influence on chronic disease after viral infection
To determine whether TLR3 function might also be linked to the development of chronic postviral disease, we tested the response of Tlr3−/− mice at 49 d after SeV infection when disease phenotype (in particular, immune cell accumulation and mucus production) reach maximal level (6, 8, 10, 11, 32). In contrast to the relatively small effect of Tlr3 deficiency on acute viral illness, we found marked attenuation of hematoxylin staining (reflecting immune and epithelial cell accumulation in this model) and PAS staining (reflecting mucus production) in lung sections from Tlr3−/− mice compared with WT mice at 49 d after SeV infection (Fig. 2A, 2B). Similarly, we observed marked attenuation of the usual increase in Muc5ac immunostaining in lung sections from Tlr3−/− mice compared with WT mice at 49 d after SeV infection (Fig. 2C). Each of these effects was significant based on image analysis of hematoxylin+, PAS+, and Muc5ac+ staining levels in lung sections from Tlr3−/− mice compared with WT mice at 49 d after SeV infection (Fig. 2D). In concert with histologic markers of lung disease, we also found that the usual induction of Il13, Muc5ac, and Arg1 mRNA in lungs from WT mice and this response was also markedly decreased in Tlr3−/−mice at 49 d after SeV infection (Fig. 2E). Similarly, monocyte/macrophage accumulation in lung tissue (marked by immunostaining for F4/80) was markedly attenuated in Tlr3−/− mice compared with WT mice at 49 d after SeV infection (Fig. 2F, 2G). Together, these findings indicate that TLR3 might serve a specific functional role in the type 2 immune response that drives chronic postviral lung disease.
TLR3 function does not rely on canonical signal transduction
TLR3 (like other TLR) signal transduction is generally thought to be mediated by activation of the downstream adapter protein known as TIR domain–containing adapter-inducing IFN-β (TRIF) (33, 34). Accordingly, we determined the phenotype of Trif−/− mice during acute illness and chronic lung disease after SeV infection.
During acute illness, we found that the usual decrease in body weight was not significantly different in Trif−/− mice compared with WT control mice (Fig. 3A). Similarly, we observed no significant differences in viral RNA levels during acute illness in Trif−/− mice compared with WT mice (Fig. 3B). Consistent with these findings, PAS–hematoxylin staining of lung sections showed no difference in immune or stromal cell accumulation in lung sections from Trif−/− mice compared with WT mice at 5 d after SeV infection (Fig. 3C). This finding was confirmed with quantitative image analysis of hematoxylin+ staining (Fig. 3E). In addition, we found that the increased levels of myeloid cells (i.e., tissue macrophages, moDCs, CD11b+ DCs, and CD103+ DCs) at 5 and 12 d after SeV infection were often not attenuated (particularly for DCs) in Trif−/− mice compared with WT mice (Fig. 3F). Similarly, levels of alveolar macrophages were not increased at baseline and were increased only slightly after SeV infection in Trif−/− compared with WT mice (Fig. 3F). Thus, the virus-induced changes in immune cell levels in Trif−/− mice were not as pronounced as found in Tlr3−/− mice (Fig. 1G). In particular, the results suggested little protection against the accumulation of myeloid cells (especially DCs) and associated morbidity during acute viral illness in Trif−/− mice as we had observed in Tlr3−/− mice (Fig. 1A–F). In contrast, we found similar induction of monocyte–macrophage chemokine (Ccl2 and Ccl7) production (Fig. 3G) that is dependent on canonical Tlr3 and Trif functions for activating NF-κB–dependent gene expression (35). These data demonstrate that Tlr3–Trif signal transduction is active in this setting but is not sufficient to influence long-term postviral lung disease.
During chronic disease, we found no significant attenuation of hematoxylin or PAS staining in lung sections from Trif−/−compared with WT mice at 49 d after SeV infection (Fig. 4A). Similarly, we detected no significant changes in these end points based on image analysis of hematoxylin+ or PAS+ staining levels in lung sections from Trif−/− mice compared with WT mice at 49 d after SeV infection (Fig. 4B, 4C). In concert with histologic markers of lung disease, we also found that the usual induction of Il13 and Muc5ac mRNA was no different in Trif−/− mice compared with WT mice at 49 d after SeV infection (Fig. 4D). Together, these findings indicate that TRIF function was not required for the observed changes in either acute illness or chronic disease after SeV infection. These results further suggested TLR3 function in this context depended on an alternative signal transduction pathway to influence acute and chronic effects of SeV infection. Given the initial observation that some, but not all, subsets of macrophages might require a TRIF signal (34), we considered whether a specific subset of TLR3-expressing cells might be responsible for the development of postviral lung disease, recognizing that DCs might be a key target cell type.
moDCs are the most abundant TLR3-high myeloid cells after viral infection
As introduced above, to better understand how TLR3 might influence acute illness and chronic disease after viral infection, we next defined the cellular site of TLR3 induction and expression. This approach used the same FACS scheme for myeloid–macrophage and DC lineage cells as described above (Fig. 1) combined with analysis of lung epithelial cells based on a CD45– CD31–EpCAM+ gating scheme as described previously (11). Initial analysis indicated that SeV RNA was localized to each of the myeloid and epithelial cell types with peak viral titers at 5 d that decrease markedly by 12 d postinfection (Fig. 5A). Remnant viral RNA at 49 d postinfection was detectable primarily in lung epithelial cells, thereby defining the cellular site for persistence found in our analysis of whole lung RNA reported previously (6). In concert with viral RNA levels, we also found that induction of Tlr3 mRNA expression was also maximal at 5 d after viral SeV infection in each cell subset (Fig. 5B) with the exception of CD103+ DCs that are reported to exhibit constitutive expression of Tlr3 mRNA (36). Similarly, low level Tlr3 mRNA expression persisted in concert with detection of SeV RNA, particularly in lung epithelial cells (Fig. 5B). Concomitant flow cytometry analysis showed induction of TLR3 protein primarily in moDCs, alveolar macrophages, and epithelial cells at 5 d after SeV infection that returned to near control levels by 12 d postinfection (Fig. 5C). Similar to Tlr3 mRNA levels, TLR3 was expressed on CD103+ DCs at similar levels at 0–49 d after SeV infection (Fig. 5B, 5C). These alterations in histograms were confirmed by mean fluorescence index for TLR3 signal above background for each cell population (Fig. 5D). Together, the findings indicated that SeV infection induces transient expression of TLR3 at 5 d postinfection in moDCs, alveolar macrophages, and lung epithelial cells and leaves constitutive expression on CD103+ DCs unchanged as a starting point for assigning TLR3 function in postviral lung disease.
In that regard, we also found that the increases in TLR3 expression were accompanied by accumulation of TLR3-expressing myeloid–macrophage and DC lineage cells in the lung after viral infection. Thus, we used our flow cytometry scheme (Fig. 1F) to define the full time course for increases in tissue macrophages, moDCs, alveolar macrophages, and CD11b+ and CD103+ DCs at 0–49 d after SeV infection (Fig. 5E). This analysis indicated that moDCs were the most abundant TLR3-high myeloid cell population early in the course of illness with increases by 5 d after viral infection and the subsequent course with peak levels at 12–21 d postinfection (Fig. 5E). These results placed the activation of TLR3-expressing moDCs ahead of the subsequent type 2 immune response that is marked by increases in M2 macrophages, eosinophils, and associated disease traits that are detectable at 21 d and maximal at 49 d postinfection (5, 6, 8–11, 13, 14). In contrast, alveolar macrophages exhibit increased TLR3 expression but decreased numbers of cells at 5–12 d after viral infection. Together, these findings, like those for TRIF deficiency, suggested a role for TLR3+ myeloid cells, particularly moDCs, in the development of postviral lung disease.
TLR3 function can be localized to moDCs versus tissue macrophages
Based on the findings that moDCs were the most abundant subset of TLR3-expressing myeloid cells that accumulate in the lung after SeV infection (Figs. 1G, 5B–D), we examined the possible contribution of this cell population to the development of postviral lung disease. To define this issue, we isolated moDCs from the lungs of WT and Tlr3−/− mice using FACS at 12 d after SeV infection (the peak of moDC levels) and transferred these cells into Tlr3−/− mice that were then subjected to SeV infection 1 d later and assessment of lung disease at 49 d postinfection (as diagrammed in Fig. 6A). Remarkably, we found that transfer of moDCs from WT, but not from Tlr3−/−, mice fully reconstituted postviral lung disease at 49 d after SeV infection. In particular, adoptive transfer of WT moDCs restored histopathology (marked by hematoxylin and PAS staining) in lung sections of Tlr3−/−mice to levels ordinarily found in WT mice at 49 d after SeV infection (Fig. 6B, 6C). Similarly, we detected the usual increase in Muc5ac immunostaining in lung sections from Tlr3−/− mice after WT moDC transfer (Fig. 6D). Each of these effects was significant based on image analysis of hematoxylin+, PAS+, and Muc5ac+ staining levels in lung sections from Tlr3−/− mice that received WT versus Tlr3−/− moDC transfer or control PBS treatment (Fig. 6E). Together, these findings indicate that TLR3 function in moDCs was critical to trigger the progression from acute illness to chronic disease in the lung after SeV infection.
To define the specificity and timing of TLR3–moDC action, we next compared moDC to tissue macrophage function and checked an earlier time point in the disease process. Accordingly, we modified our protocol to include adoptive transfer with both moDCs and tissue macrophages (defined as tissue monocytes plus interstitial macrophages). In addition, we assessed the development of postviral lung disease at 21 d after SeV infection (when disease is significant but not yet maximal). In this set of experiments (as diagrammed in Fig. 7A), we found that transfer of moDCs, but not tissue macrophages, from mice reconstituted postviral lung disease after SeV infection. Thus, adoptive transfer of WT moDCs restored histopathology (marked by hematoxylin and PAS staining) in lung sections of Tlr3−/−mice to levels found in WT mice at 21 d after SeV infection (Fig. 7B, 7C). In contrast, the disease was not reconstituted with comparable transfer of WT tissue macrophages (Fig. 7B, 7C). Each of these effects was significant based on image analysis of hematoxylin+ and PAS+ staining levels in lung sections from Tlr3−/− mice that received WT moDC or tissue macrophage transfer versus control PBS treatment into WT mice (positive control) or Tlr3−/− mice (negative control) (Fig. 7D). Together, these findings indicate that TLR3+ moDCs, but not tissue macrophages, are critical to trigger the development of postviral lung disease in the time window from TLR3 expression and presumably activation (at 5 d after viral infection) to disease onset (at 21 d postinfection).
To further define the specificity of moDC participation and any requirement for viral conditioning in postviral disease, we also studied the capacity of BMDCs to reconstitute postviral lung disease. For this approach, we cultured BMDCs from WT or Tlr3−/− donor mice before transfer into Tlr3−/− recipient mice that then undergo PBS challenge and disease phenotyping (as diagrammed in Fig. 8A). The cells harvested under these culture conditions were significantly enriched in BMDCs (Fig. 8B) as reported previously (30, 37) and thereby allowed for better assignment of TLR3 function to DCs versus other TLR3+ immune cell populations. Under these cell culture conditions, BMDCs from WT, but not Tlr3−/−, mice exhibit expression of intracellular TLR3 based on flow cytometry analysis (Fig. 8C). To validate our adoptive transfer, we also checked for arrival of adoptive transfer of moDCs in the lung using BMDCs isolated from CAG-gfp reporter mice. This approach showed that BMDCs obtained from CAG-gfp mice and transferred into Tlr3−/− mice can be detected in lung sections generally at parenchymal sites adjacent to alveolar epithelial cells with the morphology of AT2 cells (Fig. 8D). In contrast, transfer with unlabeled cells did not generate a detectable GFP signal as a marker of specific immunostaining (Fig. 8D). Under these conditions, we found that adoptive transfer of WT BMDCs into Tlr3−/− mice resulted in barely detectable airway inflammation (marked by immune cell accumulation) and mucus production (marked by PAS staining) that was nonetheless significantly increased when quantified in comparison with control mice (Fig. 8E–G).
Given the minor nature of the observed lung disease after BMDC transfer without viral infection, we next applied the same approach with BMDC transfer into mice, followed by SeV infection. In these experiments, we also tracked the time course for detection of transferred BMDCs obtained from CAG-GFP reporter mice, based on serial photomicrographs and quantitation (Fig. 9B, 9C). We found levels of GFP+ BMDCs in lung sections that are maximal at the time of transfer and are markedly decreased even by 5 d after transfer. This time course resembles the findings that Tlr3 expression and consequent function are also transient after viral infection (e.g., Fig. 3B–D). Indeed, we found that adoptive transfer of these TLR3+ BMDCs results in reconstitution of postviral lung disease in Tlr3−/− mice at 49 d after SeV infection. Thus, transfer of WT BMDCs fully restored histopathology (marked by hematoxylin and PAS staining) in lung sections of Tlr3−/−mice to levels found in WT mice at 49 d after SeV infection (Fig. 9D). These effects were significant based on image analysis of hematoxylin+ and PAS+ staining levels in lung sections from Tlr3−/− mice that received WT BMDCs versus control PBS treatment into WT mice (positive control) or Tlr3−/− mice (negative control) (Fig. 9E, 9F). As noted above, these data remain consistent with transient Tlr3 expression and function after viral infection and the possible requirement for viral infection of moDCs themselves or neighboring cells to fully manifest postviral disease.
Together, the three sets of adoptive transfer experiments (using moDCs, tissue macrophages, and BMDCs) establish a role for TLR3-activated moDCs in the development of chronic lung disease after SeV infection. In all cases, histopathology showed immune cell infiltration in peribronchial locations with extension to adjacent alveolar tissue in concert with excess mucus production. This bronchoalveolar pattern of disease resembles the histopathology of respiratory viral infection and subsequent consequences of virus-induced lung disease found in our previous studies of mouse models of postviral disease (10, 11, 14). Further, the deposition of adoptively transferred moDCs near alveolar epithelial cells suggested this site as at least one possible mechanism for moDC function in driving postviral lung disease.
TLR3+ moDCs drive IL33+ AT2 cell expansion
As introduced above, we next aimed to better define the underlying mechanism for moDC function in postviral lung disease. Given the accumulation of moDCs and increases in TLR3 expression during the acute illness after viral infection, we first examined type 2 immune events upstream of IL-13 production and consequent mucus production. Indeed, we found significant attenuation of induction of Il33 mRNA expression in the lungs of Tlr3−/− (but not Trif1−/−) mice compared with WT mice at 49 d after SeV infection (Fig. 10A). The downregulation of Il33 mRNA with TLR3 deficiency was accompanied by a similar decrease in induction of IL-33 protein in the lungs of Tlr3−/− mice (Fig. 10B). In addition, IL-33 production was restored with moDC transfer from WT (but not Tlr3−/−) mice into Tlr3−/− mice at 49 d after SeV infection (Fig. 10C).
Based on the adjacency of adoptively transferred moDCs to AT2 cells and the role of AT2 cells in IL-33 production during postviral lung disease (11), we next examined the effect of moDCs on AT2 cells during postviral disease. In the first set of experiments, we verified that the primary source of Il33 mRNA after viral infection was the lung epithelial cell population. Thus, FACS separation of lung endothelial, immune, and epithelial cells and subsequent analysis of Il33 mRNA was able to localize expression to the epithelial cell population (Fig. 10D, 10E). In addition, we found that the usual expansion of the AT2 cell population at 49 d postinfection in WT mice was blocked in Tlr3−/− mice based on immunostaining for the AT2 cell marker Sftpc (Fig. 9F) and quantitation of Sftpc+ cells (Fig. 10G). In addition, we found that expansion of the Sftpc+ AT2 cell population at 49 d after SeV infection was restored with moDC transfer from WT (but not Tlr3−/−) mice into Tlr3−/− mice, again based on Sftpc immunostaining and tissue and cell morphology (Fig. 10H) confirmed by quantitation of Sftpc+ cell levels (Fig. 10I). These findings continued to suggest an early effect of TLR3+ moDCs in the disease process that drives Sftpc+ AT2 cell expansion. Indeed, we also found that the attenuation of Sftpc+ AT2 cell expansion in Tlr3−/− mice was detected as early as 5 d after SeV infection and then continued to 49 d postinfection (Fig. 10J). In addition, we showed that IL-33 expression was primarily localized to Sftpc+ AT2 cells based on immunostaining (Fig. 10K) and that IL-33+Sftpc+ AT2 cells were significantly decreased in Tlr3−/− compared with WT mice (Fig. 10K, 10L). Quantitation showed that the number of IL-33+Sftpc+ AT2 cells were decreased as a percentage of total lung cells without a change in the ratio of IL-33+ to Sftpc+ AT2 cells (Fig. 10L). Thus, TLR3 deficiency resulted in attenuation of AT2 cell expansion without a change in the level of IL-33 expression within AT2 cells. Together, the findings indicated that the early arrival of TLR3+ moDCs is essential for the subsequent and progressive expansion of IL-33–expressing AT2 cells as a mechanism to drive the downstream type 2 immune response responsible for the development of long-term lung disease after SeV infection (as summarized in Fig. 11).
This study provides evidence that the development of chronic inflammatory lung disease after respiratory viral infection depends on seminal input from TLR3+ moDCs. This conclusion is supported by findings that 1) respiratory infection (in this study using the SeV mouse model) causes inducible expression of TLR3 in moDCs in concert with transient viral replication; 2) mice deficient in Tlr3, but not Trif, gene expression exhibit decreased acute illness (base on weight loss and inflammation) and marked attenuation of long-term disease that develops and progresses after infectious virus is cleared; 3) adoptive transfer of WT moDCs or BMDCs (but not tissue macrophages) into Tlr3 gene–deficient mice restores the type 2 immune response and consequent disease after viral infection; and 4) moDC transfer also restores the excessive expansion of the AT2 cell population that is responsible for IL-33 production and consequent postviral lung disease. Taken together, the findings provide for a pathway from viral infection to activation of TLR3+ moDCs to drive downstream expansion of IL-33+ AT2 cells and consequent IL-13 production for a type 2 immune response leading to inflammatory lung disease (as depicted in Fig. 11). In this study, we discuss how these findings provide unexpected insights for a revised paradigm for postviral lung disease and the implications for a similar process in humans with lung disease.
The first unconventional aspect of our findings is the role of TLR3 in promoting a progressive, long-term inflammatory disease process. Thus, PRRs such as the TLRs are generally reported to provide a rapidly coordinated response that provides for initial host defense against infection and/or injury. Consistent with this idea, in the present model, TLR3 function contributed to acute illness manifested by weight loss and inflammation. These findings are similar to reports of short-term inflammation in a mouse model of human rhinovirus infection (38), worsened survival after influenza A virus infection (39), and excessive airway inflammation and mucus production in respiratory syncytial virus infection (40). Presumably, this response contributes to the call for immune cell influx as a means to halt viral spread (e.g., by recruitment of phagocytic macrophages that control inflammatory debris and Ag-seeking DCs that will arm the adaptive immune response). However, we show in this study that TLR3 also triggers a long-term inflammatory response that leads to chronic disease even after infectious virus is cleared. The biological rationale for the prolonged type 2 immune response is uncertain but is likely linked at least in part to primordial defense against pathogens via mucociliary clearance (4). In any case, the TLR3 phenotype seems distinct from other PRRs, such as MDA5, that dampen the inflammatory response to SeV (41). Similarly, TLR3, but not MDA5, disrupts the epithelial barrier in response to dsRNA (42). Further, TLR3 deficiency had no effect on viral RNA levels, likely reflecting a redundant role in the IFN response that is key to control of SeV replication and consequent severity of infection (10, 43). Indeed, stromal cell RIG-I and immune cell TLR7 and TLR8 are each reported to manage the control of SeV replication independent of TLR3 (44). Nonetheless, TLR3 deficiency in humans is marked by HSV type 1 (HSV-1) infection as a sign of specificity and complexity for TLR3 function based on tissue and virus type (45). A preliminary report suggests that TLR3 (and TLR4 and TRIF) signals are also protective in a mouse model of SARS-CoV-2 infection based perhaps on controlling viral replication (46), but any understanding of mechanism or translation to humans still needs to be defined.
In that regard, the second unexpected result of the present work is related to the nature of the TLR3 signal transduction pathway for postviral lung disease. In particular, TLR3 signaling is generally thought to proceed via the TRIF adapter protein, as introduced above in the 10Results section. Instead, TRIF-deficient mice showed no significant change in the acute illness or the development of chronic lung disease after SeV infection. Non-TRIF–dependent signaling for TLR3 function has been reported previously. Examples include a TRIF-independent branch of TLR3 signaling to c-Src kinase (47). However, this signal actually downregulates cell motility, adhesion, and proliferation and, therefore, appears inconsistent with the present phenotype. Moreover, this pathway was not yet studied in immune cells or in vivo. TLR3 has also been linked to ligand induction for EGFR signaling (48), but in this case, the mouse model was performed in a genetic background (BALB/cJ) that does not develop chronic disease after SeV infection (32, 49) and was also not used to address mechanism in vivo. In that regard, others reported that TRIF deficiency attenuates the response to dsRNA-induced exacerbation of Ag (OVA) challenge (50), but no comparison with TLR3 deficiency was reported to define a difference for allergic versus viral inflammation. Further study of TLR3 signal transduction will also need to account for the complexity of intracellular localization for TLR3 and the N-terminal–TLR3–C-terminal–TLR3 complex that might be required for signaling (51, 52) as well as TLR3 traffic inside endosomal vesicles (53) and stimulation of inflammatory vesicle traffic (54).
The third result of significance is the identification of moDC participation in long-term postviral lung disease. As mentioned in the 1Introduction, our previous work indicated that SeV infection can activate lung macrophages during acute illness via an antiapoptotic chemokine (CCR5) signal that is key to initial host defense (16). In addition, we found that lung macrophages contribute to IL-13 production and consequent type 2 inflammation based on at least two mechanisms: 1) soluble TREM-2–dependent autoamplification (9); and 2) synergy with group 2 innate lymphoid cells (ILC2s) (11). The effect on ILC2s coincided with an action of myeloid–macrophage lineage cells on AT2 cell expansion for IL-33 production that in turn is required for ILC2 activation (11). However, the precise nature of the myeloid–macrophage cell type that acted on AT2 cells was not yet defined. In this study, we demonstrate that moDCs are at least in part responsible for driving increases in AT2 cell and consequent IL-33 levels linked to long-term postviral lung disease. Others have also described a role for monocyte-derived CD11b+ DCs in an allergen-challenge mouse model of short-term allergic lung disease (55). We also found a similar effect of FcεRIα+ DCs on activation of T cells (in this case, CD4+ Th2 cells) after SeV infection that also provides a link to allergic disease (29, 56). However, we also established that CD4+ (and CD8+) conventional T cells are not required for long-term postviral disease in the present SeV mouse model (6). Thus, the connection of moDC activation to long-term disease sparks a set of new questions for pathogenesis and the consequent paradigm for chronic inflammatory lung disease.
In that regard, the fourth consequence of our findings is the relatively upstream and presumably early site of action for TLR3-moDC participation in the inflammatory cascade to disease. Previously, we and subsequently many others proposed that airway inflammatory disease was initiated by a primary response from sentinel epithelial cells that are the first to respond to inhaled stimuli with endogenous danger signals such as CCL5, IL-33, and MAPK13 (7, 8, 16). Activation of TLR3 has proven difficult to track; however, the tight link between viral replication (providing dsRNA) and TLR3 expression suggests that expression can be a measure of functional activation in vivo. In any case, the transient nature of TLR3 expression makes it unlikely that TLR3 manifests chronic activation to explain a persistent drive for inflammatory disease. Relevant to this issue, endogenous RNA from ongoing cell turnover and/or necrosis might also activate TLR3 signaling (57–59), but this issue still needs to be established in the context of viral injury and subsequent repair. In support of another long-lasting mechanism, others have shown that SeV vectors can also provide a persistent signal for moDC maturation in the setting of long-term immunotherapy (60), perhaps similar to the present observations. In either case, the nature of endosomal TLR3 activation suggests that immune cell participation depends on a transient signal that is endogenous to virus-infected moDCs. However, it is also possible that other virus-infected cells (e.g., lung epithelial cells) could provide a signal to moDCs for accumulation (via cell recruitment or survival) and activation at the site of disease (9, 16). Others have separated mouse lung DCs into conventional type 1 DC and conventional type 2 DC populations that correspond to CD11b+ DC and CD103+ DC populations, as well as an inflammatory conventional type 2 DC population that corresponds to a subset of our CD11b+ DC population (61). Thus, the present approach excludes these subsets from the moDC population that drives postviral lung disease in the present model. These findings are consistent with the development of postviral disease without a detectable contribution from conventional T cells that might be activated by non-moDC populations (6).
These issues raise the fifth aspect of our findings and the consequent proposal for reprogramming of lung epithelial cells, and in particular epithelial stem cells, that might explain persistent progression and susceptibility to lung disease. In that regard, we found that moDC-TLR3 function was required for the expansion of the AT2 cell population, which in the mouse, is likely the primary site of IL-33 expression versus airway basal cells as the epithelial site in humans (8, 11, 22–24). In both cases, this epithelial cell expansion could therefore drive a downstream type 2 immune response characterized by increased IL-13 production and IL-13–stimulated inflammation and mucus production. Notably, the same epithelial cell expansion is also found after infection with other respiratory viruses including SARS-CoV-2 in animal models (mice, hamsters, and monkeys) and human autopsies (15, 62–64) as a potential pathway to postviral disease. Similarly, as introduced above, others have shown that TLR3 function is required for maintenance of epithelial barrier function in the airway (42) and repair of the epithelial barrier after UV damage in the skin (65) and irradiation-induced damage in the gut (66). Other studies suggest that TLR3 could contribute to this epithelial effect via cytokine signaling such as activation of IL-6–STAT3 (59) and/or an epigenetic alteration such as the gene-specific chromatin modification as found for activation of TLR4 (67). However, these observations derive from studies of stromal cells, and the mechanism requiring immune cells, such as moDCs, still needs to be defined. In that regard, one report found defective wound healing in TLR3-deficient mice that was associated with poor recruitment of myeloid cells, but this effect was short term and TRIF dependent (68). Thus, further study will need to define the pathway for the TLR3-moDC combination to drive a persistent and excessive epithelial cell response after viral infection that is independent of TRIF signaling.
As introduced above, the sixth impact of our study is the implication for human diseases. Thus, the same bronchoalveolar pattern of disease is found after infections with other common respiratory viruses, including SARS-CoV-1 (69) and SARS-CoV-2 (62), when the viral receptor is endogenous (using species-adapted virus or naturally susceptible species). In cases in which the receptor is expressed via vector or transgenic methods, histopathology manifests instead primarily in the site of expression (e.g., alveolar epithelial cells) (70). In all cases, however, there is an alveolar epithelial repair process that depends on AT2 cell expansion derived from proliferation and differentiation of epithelial stem cells, as exemplified by infections with influenza A virus in mice (21, 71) and SARS-CoV-2 in humans (15). The present results indicate that the AT2 stem cell expansion might depend on viral activation of the TLR3-moDC component of the immune response and, therefore, might trigger a self-renewing process that allows for progression to long-term lung disease. Human studies already suggest that this type of disease pathway is also activated in chronic lung diseases, such as asthma and COPD, that are subject to initiation, exacerbation, and progression after respiratory viral infection (4). The present identification of the TLR3-moDC immune component in the epithelial response to injury therefore provides another diagnostic and therapeutic target for correcting an excessive inflammatory response and thereby attenuating postviral lung disease. Indeed, TLR3 itself has been targeted with small molecular-weight molecules (66, 72) and Abs to block TLR3-dependent signal transduction (52, 58, 73, 74). However, the small molecule that was selected suffers from uncertain pharmacokinetic properties in vivo, and the anti-TLR3 Ab might not be effective if it does not reach the preferred intracellular (often endosomal) location of TLR3 in DCs versus stromal cells (51). Indeed, our analysis of TLR3 expression using flow cytometry (as shown in Fig. 5) required cell permeabilization for TLR3 detection (as described in 2Materials and Methods). Therefore, intracellular localization of TLR3 might contribute to the poor efficacy found in the clinical trial of anti-TLR3 mAb in human rhinovirus challenge in asthma patients (74). Similarly, there also needs to be precise assays of TLR3 expression/activation to follow target behavior. These pursuits and related strategies to limit epithelial stem cell activation to the initial repair phase will be productive based on the present model of postviral lung disease.
Together, the present results provide for several general and specific concepts for understanding and controlling long-term and often progressive inflammatory lung disease after respiratory viral infection. General concepts include 1) a front-line immune cell niche instead of the generally proposed sentinel epithelial cell population as the initial trigger for virus-induced disease; 2) the selective nature of TLR3 activation and moDC participation in this niche instead of the usual downstream effector role for immune cells in inflammation; 3) the trophic action of TLR3+ moDCs to expand the lung epithelial cell population as a source of IL-33 that is required to drive a downstream type 2 immune response versus previous reliance on other types of immune cells such as ILC2s, M2-macrophages, and eosinophils. These elements provide a revised scheme (as depicted in Fig. 11) that translates to the likelihood of early and transient activation of the innate immune response can switch the host to a long-lasting disease pathway. The progressive and chronic nature of the disease process is distinct from short-term role of TLR3 (or other PRRs) in inflammation (38–40, 75–78) and DC maturation after viral infection (60, 79). The alternative dendritic–epithelial cell interface still needs to be fully defined, but the present evidence already offers the likelihood that this site can reprogram an epithelial stem cell subset from useful repair of injury to an excessive proliferative/survival response with consequences for downstream inflammation, mucus production, and remodeling in the lung. The present results thereby provide an additional checkpoint for correcting the progressive disease that can develop after respiratory viral infections.
We thank Lora Benoit, Di Wu, Xiaohua Jin, Chaeho Park, Rose Tidwell, Yingjian You, and the staff in the Siteman Flow Cytometry Core and Pulmonary Morphology Core at Washington University for outstanding advice and technical support.
This work was supported by National Institute of Allergy and Infectious Diseases Grant R01 AI130591, National Heart, Lung, and Blood Institute Grant R35 HL145242, Cystic Fibrosis Foundation Grant P19-06099, and the Hardy Trust and Schaeffer Funds.
Abbreviations used in this article:
alveolar epithelial type 2
bone marrow–derived DC
chronic obstructive pulmonary disease
group 2 innate lymphoid cell
melanoma differentiation–associated protein 5
surfactant protein C
TIR domain–containing adapter-inducing IFN-β
M.J.H. is a member of the Data Safety Monitoring Board for AstraZeneca and is the founder for NuPeak Therapeutics, Inc. The other authors have no financial conflicts of interest.