Critical Adverse Impact of IL-6 in Acute Pneumovirus Infection

Severe respiratory virus infections are notable for robust local host inflammatory responses that contribute to disease severity (1, 2). Immunomodulatory strategies designed to limit virus‐induced inflammation may be of substantial therapeutic importance, particularly when no vaccines or broadly effective agents are available (3).

Our studies on inflammation and immunomodulatory strategies in vivo have used the natural rodent pathogen pneumonia virus of mice (PVM). PVM is a virus of the family Pneumoviridae, genus Orthopneumovirus, and is closely related to the human pathogen respiratory syncytial virus (RSV) (4, 5). RSV does not replicate effectively in inbred mouse strains (6). By contrast, PVM undergoes robust replication in mouse lung tissue and generates severe inflammatory disease in association with cytokine production and prominent leukocyte recruitment to the airways. Acute PVM infection in mice phenocopies the inflammatory responses characteristic of the most severe forms of RSV infection in human subjects and serves as a clinically relevant in vivo model for evaluating anti‐inflammatory and immunomodulatory therapeutic strategies (6, 7).

Among the individual cytokines with an impact on the pathogenesis of pneumovirus infection, CCL3 was shown to promote neutrophil recruitment; blockade of CCL3 or its receptor, CCR1, contributes improved survival in response to acute PVM infection (8, 9). Other cytokine ligands and receptors implicated in promoting inflammatory pathology in PVM infection include cysteinyl leukotrienes, ChemR23, Fas/FasL, and CD8⁺ T cell–derived mediators (10–13).

Development of immunomodulatory strategies for the prevention of lethal sequelae of respiratory virus infection is currently an area of active exploration (reviewed in Ref. 14). Toward this end, we and others have found that administration of otherwise benign Lactobacillus species directly to the respiratory mucosa of virus-infected mice has resulted in protection against lethal disease (15–19). The mechanisms underlying this response and, more broadly, the ways in which bacteria, bacterial components, and pattern recognition receptor–activating ligands administered to the respiratory tract promote protection against lethal virus infection remain to be fully elucidated (20–22). We have found that Lactobacillus-mediated protection at the respiratory mucosa persists in the absence of B lymphocytes or Abs and likewise does not rely on IL-10, type I, or type II IFN signaling (23, 24). Nonetheless, Lactobacillus at the respiratory mucosa does result in suppression of a distinct cohort of virus-induced proinflammatory cytokines, most profoundly, IL-6 (25). IL-6 is a pleiotropic cytokine produced by monocytes and macrophages, lymphocytes, endothelial cells, and fibroblasts (26); it is capable of cis- and trans-signaling, the latter via soluble signaling receptor gp130, and has both proinflammatory and anti-inflammatory activities (reviewed in Refs. 27–29).

The role of IL-6 in chronic virus infections has been considered at length (30). Among the most recent of these findings, Harker and colleagues (31) reported that IL-6 generated late in chronic murine lymphocytic choriomeningitis virus infection generated T follicular helper cell responses that were critical and protective in nature. With respect to acute virus infections, IL-6 has been linked to severe RSV infection in adults and infants [(32, 33); see 21Discussion] as a biomarker in severe pandemic H1N1 influenza A infection (34) and in association with fatal infection with influenza A H5N1 (35). IL-6 was also among the cytokines expressed prominently in the lungs and airways of mice subjected to acute infection with influenza and found at levels correlating with clinical symptoms and lung pathology (reviewed in Ref. 36). Interestingly, IL-6 deficiency had either no impact on survival (34) or resulted in poor outcomes, suggesting that this cytokine maintained a protective role. Specifically, Dienz and colleagues (37) reported that IL-6^−/− mice succumbed to sublethal doses of influenza A H1N1 in association with apoptosis of antiviral neutrophils in the lung, whereas Lauder and colleagues (38) found that infection of IL-6^−/− mice with influenza A H1N1 resulted in infiltration of the lungs with monocytes in association with progression to pulmonary edema and respiratory distress due to an inefficient antiviral T cell response. Subsequent studies that focused on the roles of CD4⁺T cells, cytotoxic CD8⁺T cells, and IL-27, IL-21, and Ab production as well as lung repair mechanisms (39–42) all identified IL-6 as a protective factor in acute influenza infection.

By contrast, the studies presented in this paper indicate that IL-6 plays an adverse role in acute respiratory infection with PVM. In direct contrast to what has been reported for acute influenza infection, PVM infection is less lethal in IL-6–deficient compared with wild-type mice, an observation associated with diminished neutrophil recruitment and reduced fluid accumulation in lung tissue. A specific cellular source of IL-6 in PVM infection is identified as contributing to disease pathogenesis.

Wild-type C57BL/6 mice (6–10 wk old) were from the Jackson Laboratory (Bar Harbor, ME) (JAX 000664) and Charles River Laboratories (Frederick, MD). IL-6^−/− (JAX 002650) and Nr4a1^−/− (JAX 006187) mice were from The Jackson Laboratory and maintained on site (14BS vivarium; National Institute of Allergy and Infectious Diseases [NIAID]/National Institutes of Health [NIH]). Ccr2^−/− mice were from stocks maintained by NIAID/Taconic consortium. The NIAID Division of Intramural Research Animal Care and Use Committee, as part of the NIH Intramural Research Program, approved all the experimental procedures as per protocol LAD8E.

PVM strain J3666 was prepared from mouse-passaged stocks stored in liquid nitrogen. In experiments described in this paper, mice were inoculated via the intranasal route while under isoflurane anesthesia with 50 μl of PVM diluted 1:8000 or 1:10,000 to 0.25 or 0.20 tissue culture infectious dose (TCID)₅₀ units/mouse, respectively, followed by evaluation of serial weights and survival as well as specific outcomes at time points as indicated below. Recombinant PVM (rk2-PVM) encoding the fluorescent marker mKatushka2 (mKATE) was prepared and characterized as previously described (43). Influenza A/HK/1/68 (H3N2; gift from J. Keicher, Symmune Therapeutics, Raleigh, NC) was provided as an egg-passaged stock; this was passaged three times in wild-type–specific pathogen-free mice in our high-barrier facility prior to use in in vivo experiments. Mouse-passaged stocks were maintained at −80°C. Virus was administered intranasally in a total volume of 2.5 μl/nare (5 μl/mouse, diluted 1:2000, to 75 TCID₅₀ units/mouse). Mice were evaluated by serial weights and survival.

Lactobacillus plantarum NCIMB 8826 (BAA-793; ATCC) was grown overnight in Mann Rogosa Sharp medium and administered live or as heat-inactivated (70°C for 30 min). Live cells were washed once in sterile PBS and stored at 10¹¹/ml in sterile PBS with 0.1% BSA at 4°C for no longer than 24 h prior to administration; heat-inactivated cells were washed once in sterile PBS and stored at −20°C at 10¹¹/ml in PBS with 0.1% BSA. Mice inoculated with PVM on day 0 as described above received 50 μl of inocula of live or heat-inactivated L. plantarum at 10⁸ cells or cell equivalents per dose or diluent control (PBS with 0.1% BSA) on days 1 and 2. In some experiments, PVM-infected mice received a reduced dose of live L. plantarum at 10⁷ cells in 50 μl on days 1 and 2 as indicated. In some experiments, PVM-infected mice treated with live L. plantarum as described also received recombinant murine IL-6 (R&D Systems) 25 μg/mouse or diluent (PBS with 0.1% BSA) in a 50 μl intranasal inoculation on days 4, 5, 6, and 7.

Mice were euthanized by cervical dislocation under isoflurane inhalation and subjected to bronchoalveolar lavage (BAL) with PBS with 0.1% BSA. BAL fluid was clarified by centrifugation and stored at −80°C prior to analysis. Whole lung tissue isolated from mice was immersed in 1 ml of PBS with 0.1% BSA and subjected to blade homogenization and clarification by centrifugation at 4°C. Clarified homogenates were stored at −80°C prior to analysis. For histology, lungs were perfused with 10 ml of PBS, inflated with 0.8 ml of ice-cold 10% phosphate-buffered formalin, and fixed in 10 ml of cold 10% phosphate-buffered formalin. H&E-stained sections were prepared by Histoserv (Germantown, MD). Photographs of the microscopic images were obtained with a DMI 4000 light microscope (Leica Microsystems, Wetzlar, Germany) equipped with a Retiga 2000R camera and analyzed using QCapture software (both from QImaging, Surrey, BC, Canada).

Fluid accumulation in the lungs was evaluated quantitatively essentially as described (44, 45). Briefly, whole lung tissue was removed from PVM-infected wild-type mice (both untreated and those treated with L. plantarum as described above) and also from PVM-infected IL-6^−/− mice at sequential time points. Lung tissue weights (significant to 0.05 g, Denver Instruments top-loading balance, Model SI-403) were determined immediately after removal and again after fluid evaporation in a drying oven at 37°C for 4 d or more. The ratio of the wet lung weight to the dry lung weight at each time point was determined.

Quantitative analysis of cytokine mediators in BAL fluid, lung tissue homogenates, and serum was performed by ELISA (DuoSet, Quantikine; R&D Systems) as per manufacturers’ instructions. Mediator levels detected in lung tissue homogenates were normalized for total protein by BCA assay (Pierce).

For preparation of single-cell suspensions, PVM-infected wild-type and IL-6^−/− mice were euthanized as noted above prior to perfusion with 8 ml of PBS with 10 mM EDTA; lungs were removed from the body cavity and minced manually. Minced tissue was incubated in digestion buffer (RPMI 1640 with 5% FBS, 20 μg/ml DNase I, and 40 μg/ml collagenase D) for 60 min at 37°C; tissue digests were filtered through 70-μm nylon mesh, and erythrocytes were lysed with ACK buffer (Lonza). Cells were counted manually on a hemocytometer; live cells were identified by trypan blue exclusion (Lonza). For evaluation of myeloid subsets, cells (10⁶ cells per 100 μl) were stained with LIVE/DEAD Aqua or Blue viability dyes (20 min at 4°C; Invitrogen) followed by anti-CD16/32 (BD Bioscience) and fluorochrome-conjugated Abs, including: from eBioscience: anti-CD45 (clone 30-F11), anti-CD11c (clone N418), anti–MHC class II (MHC II)–1-A/1-K (clone M5/114.15.2); from BD Biosciences: anti–Siglec F (clone E50-2440), anti-Gr1 (clone RB6-8C5). In studies that included identification of Ly6C^hi and Ly6C^lo monocytes, fluorochrome and biotin-conjugated Abs included: from eBioscience: anti-CD45 (clone 30-F11), anti–MHC II (I-A/I-E; clone M5/114.15.2); from BD Biosciences: anti-CD45 (clone 30-F11), anti-CD19 (clone 1D3; biotinylated); anti-NK1.1 (clone PK136; biotinylated); anti–Siglec F (clone E50-2440); from BioLegend: anti-CD11b (clone M1/70); anti-Ly6G (clone 1A8); anti-Ly6C (clone HK1.4); anti-F4/80 (clone BM8); also from BD Bioscience, fluor-conjugated Streptavidin. Ab staining was performed for 20 min at 4°C. Cell suspensions were then fixed in 4% paraformaldehyde in PBS prior to analysis on an LSR II flow cytometer (BD Bioscience). For evaluation of T lymphocytes in lung tissue of PVM-infected wild-type and IL-6^−/− mice, single-cell suspensions generated as described above were stained with LIVE/DEAD Aqua (Invitrogen) followed by anti-CD16/32, anti-CD45 (clone 30-F11), anti-CD3 (clone eBIO500A2), anti-CD4 (clone RM4-5), and anti-CD8a (clone 53-6.7). After staining, cells were fixed and permeabilized (Thermo Fisher Scientific), and intracellular IL-17 was evaluated (clone TC11-18H10). For all analyses, compensation and analysis of flow cytometry data were performed using FlowJo 10.2 software (Tree Star, Ashland, OR). Gates were set using relevant “fluorescence minus one” or isotype controls.

Mice were infected with rK2-PVM (50 μl with 400 TCID₅₀ units/mouse) on day 0; lung tissue was collected for preparation of single-cell suspensions on days 5, 6, and 7. Single-cell suspensions were prepared as described above; all manipulations were carried out in the dark to limit quenching of the fluorescent tag. Cells were suspended at 10⁶ per 100 μl and stained with LIVE/DEAD Aqua and anti-CD16/32 as above. Monocytes and macrophage populations were identified using the protocol developed by Misharin et al. (46) with fluorochrome-conjugated Abs, including anti-CD24 (FITC; clone M1/69), anti-CD11b (PerCP/Cy5.5; clone M1/70), anti–Siglec F (BV421; clone E50-2440); anti-Ly6C (BV785; clone HK1.4); anti–MHC II (BV711; clone M5/114.15.2); anti-CD45 (BV605; clone 30-F11); anti-CD64 (allophycocyanin; clone ×54-5/7.1); and anti-CD11c (PE-Cy7; clone HL3); the virus-tag, mKATE is detected on the PE–Texas Red channel. Cells were fixed as described above, and analysis was performed on an LSR Fortessa flow cytometer. Compensation and analysis of flow cytometry data were performed using FlowJo 10.2 software (Tree Star, Ashland, OR). Gates were set using relevant fluorescence minus one or isotype controls.

Clodronate-filled liposomes (60 μl per mouse; Encapsula Nanosciences) were administered on day −1 prior to intranasal inoculation with PVM (day 0) and once again on day 4. Depletion of alveolar macrophages (AMs; live CD45⁺CD11c⁺SiglecF⁺ cells) from whole lung tissue after administration of clodronate-filled liposomes was determined by flow cytometry through day 15 as described above.

Virus was evaluated in whole lung and spleen tissue using a quantitative PCR (qPCR) assay that targets the PVM small hydrophobic (SH) protein as previously described (15, 47). Briefly, RNA was prepared from mouse lung tissue that had been immersed and stored in RNAlater (Ambion, Austin, TX). Isolated RNA (RNAzol, Tel-test, Friendswood, TX) was treated with DNase I to remove genomic DNA contaminants. Reverse transcription was performed using a first-strand cDNA synthesis kit (Roche) with random primers; a no reverse transcriptase control was included. The qPCR reactions were amplified in triplicate, with the ABI 2× TaqMan reagent, primer-probe mixes, and cDNA or plasmid standard in a 25-μl final volume (Applied Biosystems). Thermal cycling parameters for the ABI7500 absolute quantitation program (Applied Biosystems) include 50°C for 2 min, 95°C for 10 min, and 40 amplification cycles alternating 95°C for 15 s and 60°C for 1 min. Custom design primer-probes include primers including: primer 1 5′-GCC GTC ATC AAC ACA GTG TGT-3′; primer 2 5′-GCC TGA TGT GGC AGT GCT T-3′; probe 6FAM -CGC TGA TAA TGG CCT GCA GCA-TAMRA. The PVM SH gene (GenBank accession no. AY573815 http://www.ncbi.nlm.nih.gov/genbank) cloned into pBACpak8 was used to generate a standard curve for absolute quantification. Experimental triplicate data points were interpolated to linear standard curves. A sample calculation from data generated by this method is included in reference (47). The data from lung tissue RNA are normalized to absolute copies of GAPDH. This value is generated using commercially available mouse GAPDH primer-probes (catalog no. 4308313; Applied Biosystems); values obtained are interpolated to a standard curve generated using a mouse GAPDH plasmid in pCMV pSport 6 (ATCC catalog no. 10539385), also as previously described (15, 47).

All quantitative findings were from multiple datasets. Findings were analyzed via algorithms (log-rank, ANOVA, Mann–Whitney U test, Student t test) within GraphPad Prism 7.0. All error bars represent SD unless otherwise indicated.

PVM is a natural rodent pathogen that replicates in mouse lung tissue and generates a severe lethal respiratory infection (5, 7). Following the protocol shown in Fig. 1A, we showed that administration of L. plantarum directly to the respiratory tract on days 1 and 2 immediately after virus inoculation generated substantial protection against the lethal sequelae of PVM infection. Comparable to findings reported in earlier studies (25), survival was associated with profound suppression of virus-induced cytokines, notably reduction of local levels of IL-6 as well as CXCL1 and CCL2 shown in this study at 18-, 7-, and 6-fold respectively at day 8 of infection [Fig. 1B]). Survival was also associated with a significant reduction in IL-6 in systemic circulation at this time point [Fig. 1C] and a small but statistically significant reduction in virus recovery from lung tissue [Fig. 1D]. Of note, PVM infection remains localized to the lung and does not result in viremia; no virus was detected in peripheral blood (data not shown) or spleen of untreated or L. plantarum–treated mice.

Microscopic images of lung tissue from PVM-infected mice are featured in Fig. 1E. Compared with lung tissue from uninfected mice (i), PVM infection resulted in profound alveolitis with inflammatory cells within the airspaces evident throughout (ii). The region within the box in (ii) is shown at higher magnification in (iii), in which neutrophil recruitment (white arrows) and early onset pulmonary edema within the airspaces was delineated. By contrast, lung tissue from PVM-infected mice that were treated with L. plantarum as in Fig. 1A display markedly reduced pathology, with alveoli that were largely clear of neutrophils and edema fluid (iv). Consistent with the tissue histology, BAL fluid leukocyte differential indicated a profound reduction in airway neutrophils among PVM-infected mice treated with L. plantarum– versus PVM-infected mice that were otherwise untreated (Fig. 1F). Likewise, lungs from L. plantarum–treated PVM-infected mice maintained constant wet-to-dry ratios over time, consistent with minimal accumulation of edema fluid. This is in profound contrast to the wet-to-dry ratios determined for lungs from the otherwise untreated PVM-infected mice; these ratios increased over time with ratios on days 8 and 9 significantly higher than those determined for lungs from PVM-infected mice treated with L. plantarum at these same time points (***p < 0.001; Fig. 1G).

To explore the role of IL-6 in promoting inflammatory responses in acute PVM infection alone, in the absence of L. plantarum treatment, IL-6 gene-deleted (IL-6^−/−) mice were inoculated at day 0 with PVM; weights and survival were evaluated at serial time points thereafter. As shown in Fig. 2A, 73% of the IL-6^−/− mice survived longterm as compared with only 18% of wild-type mice inoculated with a reduced inoculum of 0.2 TCID₅₀/mouse (***p < 0.001). The corresponding weights for this group (days 0 through 16) are shown in Fig. 2B. Lung virus titer on days 5–10 of infection is shown in Fig. 2C. In both strains, virus titer reached a maximum at day 7 after inoculation, immediately prior to the onset of mortality in wild-type mice. Although there were some small differences noted between strains, both wild-type and IL-6–deficient mice cleared virus from the lungs, with PVM detected at only 6–7% respective peak levels by day 10.

Interestingly, the responses to PVM infection are qualitatively different from those reported for influenza virus infection in mice, the latter defining a protective role for IL-6 (37–42). To explore this further, we inoculated wild-type and IL-6^−/− mice with Influenza A/HK/1/68 (H3N2) (Fig. 2D). We found that, in contrast to PVM infection, IL-6^−/− mice responded to infection with influenza A with significantly more weight loss and with delayed recovery from an acute infection compared with their wild-type counterparts. This observation, that these two respiratory virus infections exhibited distinct and opposing responses to IL-6 deficiency, is considered further in the 21Discussion.

IL-6 has notable impact on priming and activation of neutrophilic leukocytes (48, 49). We show in this study that PVM infection resulted in local production of IL-6, which was detected in the airways as early as day 6 of infection (Fig. 3A). IL-6 levels in the airways reached a maximum at day 7–8 and were in decline by day 10. As to be expected, no IL-6 was detected in the airways of PVM-infected IL-6^−/− mice. Other cytokines detected in the airways in response to PVM infection include CXCL10, TNF-α, CXCL1, and CCL2, with levels of CXCL10, TNF-α, and, to a lesser extent, CCL2 in the airways of IL-6^−/− mice, exceeding those detected in the wild-type at days 7, 8, and/or 9 of infection (Fig. 3B, 3C, Supplemental Fig. 1). By contrast, relatively low levels of IL-17 were detected in the airways of wild-type and IL-6^−/− mice, but no modulation in response to gene deletion or to infection was observed (Supplemental Fig. 1).

Flow cytometric evaluation of myeloid leukocyte subsets in whole lung tissue revealed a substantial >5-fold reduction in neutrophil recruitment to the lungs of PVM-infected IL-6^−/− mice, from 2.2 ± 0.95 × 10⁵ to 0.40 ± 0.16 × 10⁵ total cells (***p < 0.001; Fig. 4A). Likewise, the wet-to-dry weight ratio determined for lungs from wild-type mice was significantly higher than the value obtained for their IL-6–deficient counterparts (*p < 0.05; Fig. 4B).

T lymphocytes (CD3⁺ cells) were also recruited to the lungs in response to PVM infection (13, 24, 50); in contrast to neutrophils, the numbers of T lymphocytes detected in lung single-cell suspensions from wild-type and IL-6^−/− mice were comparable to one another (Fig. 4C) as were the fractions of CD4⁺ and CD8⁺ T cells (Fig. 4D); few to no IL-17⁺CD4⁺ (Th17) cells were detected in the lungs of PVM-infected wild-type or IL-6^−/− mice (Fig. 4E), consistent with the aforementioned findings on immunoreactive IL-17 (Supplemental Fig. 1).

In previous studies, we found that AMs were targets of PVM infection and that AMs isolated from PVM-infected mice generated IL-6 in ex vivo culture (51). However, we had not previously explored the role of AMs as a source of specific cytokines in PVM infection in vivo. As shown in Fig. 5A, PVM-infected AMs were isolated from mice inoculated with fluorescent-tagged virus; detection of mKATE⁺ AMs reached a maximum (38 ± 6% of mKATE⁺CD45⁺CD11c⁺ CD11b⁻CD64⁺CD24⁻ cells) at day 6 after inoculation (Fig. 5B). AMs were depleted by administration of clodronate liposomes 1 d prior to and again 4 d after inoculation with PVM. AM depletion resulted in improved survival in response to PVM infection in wild-type mice (0 versus 58%, **p < 0.01; Fig. 5C). Prolonged survival was associated with a substantial reduction in levels of IL-6 in the airways (Fig. 5D) and a small but significant reduction in virus titer (Fig. 5E).

Inflammatory (Ly6C^hiCCR2⁺) monocytes are typically recruited from circulation in response to signals from pathogen infection and have also been characterized as a prominent source of proinflammatory cytokines (52–54). As shown in Fig. 6A, Ly6C^hi inflammatory monocytes were recruited to the lungs in response to PVM infection in wild-type mice but were not detected in the lungs in mice devoid of the critical chemokine receptor CCR2. Interestingly, the absence of proinflammatory monocytes had no impact on virus titer (data not shown) and no impact on the level of virus-induced IL-6 in lung tissue (Fig. 6B). In contrast to Ly6C^hi cells, nonclassical or “patrolling” Ly6C^lo monocytes (55, 56) are significant targets of PVM infection; at peak, 17.5 ± 10% of these cells were mKATE⁺ (Fig. 6C). Nonetheless, PVM infection in Nr4a1^−/− mice, which are devoid of patrolling monocytes (57), display similar virus titers in lung tissue (Fig. 6D), and the levels of virus-induced IL-6 in the airways were indistinguishable from those detected in PVM-infected wild-type mice (Fig. 6E). Survival rates in PVM-infected Ccr2^−/− and Nr4a1^−/− strains were indistinguishable from one another (9.2 ± 0.84 versus 8.8 ± 1.8 d, respectively) and were indistinguishable from that determined for PVM-infected wild-type mice (9.9 ± 1.3 d; Fig. 6F).

We have shown clearly that administration of immunobiotic Lactobacillus at the respiratory tract results in survival in response to lethal PVM infection in association with profound suppression of virus-induced IL-6, among other proinflammatory cytokines [Fig. 1 (25)]. We have also shown in this study in Fig. 2A that PVM infection is substantially less lethal in mice devoid of IL-6 than it is in IL-6–sufficient, wild-type mice, a finding that suggests that IL-6 is a central mediator of lethal inflammation in this infection. Finally, we show in this study that suppression of virus-induced IL-6 is a critical element of the mechanism of Lactobacillus-mediated protection at the respiratory tract. As shown in Fig. 7A, wild-type mice were fully protected in response to administration of live L. plantarum to the respiratory tract at the standard dose of 10⁸ cfu/mouse. However, PVM-infected wild-type mice did not respond to treatment with a 10-fold-lower inoculum, at 10⁷ cfu L. plantarum/mouse; weight loss was indistinguishable from that of wild-type mice that received diluent alone (PBS). By contrast, IL-6^−/− mice were partially protected, with limited weight loss in response to the lower inoculum of L. plantarum (10⁷ cfu/mouse, *p < 0.05; Fig. 7B). As shown in Fig. 7C, the lower inoculum of L. plantarum (10⁷ cfu/mouse) resulted in minimal suppression of IL-6 in PVM-infected wild-type mice compared with the standard (10⁸ cfu/mouse) dose. Similarly, in Fig. 7D and 7E, the lower inoculum of L. plantarum resulted in more effective suppression of CXCL1 and CCL2, respectively, in IL-6^−/− mice compared with the wild-type. The mechanism(s) linking IL-6 to these cytokines directly or indirectly remains unclear, although Ahuja and colleagues have reported on cross-talk between IL-6 and CXCL1 in acute kidney injury in mice (49). Of note, no differences in virus titer were determined at any L. plantarum inoculum (Fig. 7F). In a parallel experiment, PVM-infected mice were treated with L. plantarum (days 1 and 2) and then treated with recombinant IL-6 or diluent control (days 4–7). Although recombinant IL-6 did not fully overcome the protective effects of L. plantarum, significant weight loss among the L. plantarum–treated mice was observed (*p < 0.05; Fig. 7G).

In our ongoing work, we have focused on a role for immunomodulatory therapy for treatment and prophylaxis of acute respiratory virus infection. Specifically, we and others found that the benign or lethal nature of a given respiratory virus infection relates directly to the nature and extent of the virus-induced cytokine storm (5, 7, 58, 59). Using the endogenous rodent pathogen PVM, we have found that virus-induced inflammation, once initiated, is not readily suppressed, even upon timely addition of agents capable of halting virus replication (8, 9).

Identifying the appropriate pathways for inhibition is not a straightforward matter. Nonetheless, we have shown that full and sustained protection against lethal inflammatory sequelae of PVM infection can be generated using preparations of the benign probiotic bacterium L. plantarum administered directly to the respiratory tract (15, 23–25, 60). The impact of Lactobacillus species resident in and introduced into the lung microenvironment is an area of ongoing and active exploration (61). The impact of L. plantarum administered to the respiratory mucosa may be similar that described for inhaled immunostimulants such as OM-85 BV (Broncho-Vaxom [62, 63]); however, unlike the single species L. plantarum administered in this study, OM-85 BV is an alkaline extract prepared from 21 distinct bacterial strains, including Gram-negative pathogens such as Haemophilus influenzae and Bordetella pertussis and the mycoplasma Mycoplasma pneumoniae (58). With respect to mechanism, we have shown that L. plantarum exerts its impact largely via coordinate engagement of the pattern recognition receptors TLR2 and NOD2 (25, 60). Furthermore, an extensive microarray analysis of lung tissue from PVM-infected mice has shown clearly that L. plantarum administered directly to the respiratory tract results in profound suppression of a distinct cohort of virus-induced proinflammatory mediators, including, among others, IL-6 (25), although until now, no distinct role for any single cytokine had emerged. With this study, we have uncovered both a central role for IL-6 in promoting lethal inflammation in acute pneumovirus infection and clarified the molecular mechanism of L. plantarum in circumventing this response.

As noted earlier, PVM is a virus of the same family (Pneumoviridae) and same genus (Orthopneumovirus) as the human pathogen (RSV; 4, 5, 7). The inflammatory nature of human RSV has been documented in numerous studies (2, 64, 65). Of specific note, Walsh and colleagues (32) found that adults requiring hospitalization for severe RSV infection exhibited extensive virus shedding in association with elevated levels of mucosal IL-6. Likewise, Tabarani and colleagues (33) identified elevated levels of nasopharyngeal IL-6 as a factor correlating with severe disease in a study of >800 RSV-infected pediatric subjects. Yet, and perhaps paradoxically, both infants and adults with the -174C/C promoter polymorphism (a genotype that predicts reduced IL-6 production) are at higher risk for severe RSV disease (66, 67). IL-6 has also been identified as an immunomodulatory cytokine in the lungs of RSV-challenged BALB/c mice. Among these findings, Pribul and colleagues (68) reported diminished release of IL-6 along with other proinflammatory cytokines (CCL3, TNF, IFN-α) in AM-depleted mice subjected to RSV challenge. More recently, Pyle and colleagues (40) reported that IL-6 was critical for resolution of RSV-induced immunopathology in BALB/c mice, a feature that was dependent on induction of IL-27 and maturation of regulatory T cells.

Moving forward, one hopes that these findings may ultimately be translated into workable therapeutic strategies, particularly in settings where no effective vaccines are available. Anti-cytokine and anti-cytokine–signaling therapeutics are now used for a wide variety of inflammatory conditions in which blockade of single pathways has been proven to be effective (69–71). Inflammation associated with acute respiratory virus infection has been perceived as the result of “cytokine storm,” with cross-amplification and redundancy at multiple levels (58, 59). Although it may not be immediately clear whether cytokine blockade will be useful in this setting, the contributions of individual cytokines have already been examined in complex acute inflammatory conditions, notably acute respiratory distress syndrome (reviewed in Ref. 72), including a specific antagonist to TNFR1, which is currently undergoing clinical evaluation (73).

In this paper, we have focused on IL-6 and show that the actions of this cytokine alone may be critical in promoting lethal sequelae of acute pneumovirus infection in vivo. IL-6 is produced by fibroblasts, monocytes, and macrophages and interacts with numerous target cells largely due to its capacity for trans-signaling by soluble gp130 (reviewed in Refs. 74, 75). It is critical to note that elimination of IL-6 may be remarkably effective in PVM infection due to the ability of this cytokine to prime and activate neutrophils (48, 49). We note that Cortjens and colleagues (76) have presented data suggesting that the lethal sequelae of PVM infection are not directly dependent on neutrophils in the airways. However, our findings indicate that neutrophils are critical intermediaries in acute pneumovirus infection (8, 9, 15).

Monocytes and macrophages are major sources of proinflammatory mediators in respiratory virus infection (77–79). We show in this study that classical Ly6C^hi monocytes were recruited to the lungs in response to PVM infection and that this response is absent in mice devoid of the chemokine receptor CCR2. Interestingly, there were no differences in the level of total IL-6 nor do we observe any differential survival in PVM-infected Ccr2^−/− mice compared with their wild-type counterparts. These results suggest that Ly6C^hi monocytes recruited in response to PVM have limited direct impact in this infection and/or there is an as yet-uncharacterized immediate and direct compensation in their absence. By contrast, we showed that AMs are not only a target of PVM infection but a substantial source of IL-6 in vivo; elimination of AMs results in prolonged survival in response to PVM infection in association with a reduction in total IL-6 in lung tissue and airways. However, as noted earlier, AMs are only one of many potential sources of this cytokine. Although CD8⁺ T lymphocytes have been implicated as a source of critical inflammatory mediators in acute PVM infection (13), we showed that PVM-infected lymphocyte-deficient Rag1^−/− mice generated no TNF-α but maintained IL-6 in the airways at levels that were indistinguishable from their PVM-infected wild-type counterparts (24); these findings indicate that lymphocytes are not a significant source of IL-6 in acute PVM infection. Nonetheless, we need to consider the possibility that IL-6 may be released directly or indirectly from epithelial cells and/or fibroblasts, respectively, in conjunction with acute infection. Toward this end, we are exploring respiratory virus infection and the impact of L. plantarum in mouse tracheal epithelial cell cultures (80), as important interactions linking virus and L. plantarum to IL-6 signaling may be revealed in this setting (E. Mai, C.M. Percopo, and H.F. Rosenberg, unpublished observations).

Likewise, although we have shown previously that AMs produce IL-6 in response to PVM infection ex vivo (51), we cannot rule out the possibility that ablation of AMs may have indirect effects on other sources of this proinflammatory cytokine in vivo. We were intrigued to note that the role of IL-6 in PVM infection is strikingly different from that observed in response to acute infection with influenza A. Specifically, Dienz and colleagues (37) and Lauder and colleagues (38) found that influenza A H1N1 A/PR/8/34 infection in IL-6^−/− mice resulted in more severe outcomes, with mechanisms including loss of neutrophil-mediated virus clearance and failure to mount an antiviral T cell response, respectively; of note, Paquette and colleagues (34) found that IL-6 gene deletion had no impact on disease severity in mice infected with influenza A H1N1pdm. In our hands, IL-6^−/− mice infected with Influenza A/HK/1/68 H3N2 also lost more weight overall and experienced delayed recovery from a sublethal virus inoculum. When considering both influenza A and PVM infections in mice, we note that the former results in acute depletion of AMs (81, 82), whereas loss of AMs is not observed in PVM infection. A closer evaluation of the role of AMs as both an independent source and a target of IL-6 in both pneumovirus and influenza infections might help to explain some of these divergent responses. It is also intriguing to note that the lethal sequelae influenza A can be limited by Lactobacillus at the respiratory mucosa (16–19), results that suggest that AMs and IL-6 are not unique or universal targets of this immunomodulatory strategy.

In conclusion, we have examined the role of IL-6 in promoting pathology in acute respiratory virus infection with PVM, a natural rodent pathogen that generates local inflammation in lung tissue. We showed in this study that IL-6 has a critical negative impact in PVM infection. Among our results, we found that PVM was substantially less lethal in IL-6–deficient than in wild-type mice and was associated with reduced neutrophil recruitment to the lungs. Likewise, administration of immunobiotic L. plantarum to the respiratory mucosa results in profound suppression of PVM-induced IL-6, and L. plantarum is significantly more effective in protecting against weight loss and virus-induced inflammation in IL-6–deficient mice. Ly6C^hi proinflammatory monocytes were recruited in response to PVM, but they were not a significant source of IL-6 and did not contribute to the lethal sequelae of infection. By contrast, AMs were readily infected with PVM in vivo; ablation of AMs resulted in prolonged survival in association with a reduction in virus-induced IL-6 in lung tissue. The results of this work connect the actions of IL-6 to the immunomodulatory sequelae of L. plantarum at the respiratory tract and suggest that cytokine blockade may be a strategy to consider in severe pneumovirus infections.

We thank the caretakers and technicians of the 14BS Animal Facility, NIAID/NIH, for their ongoing assistance with our research program.

The authors have no financial conflicts of interest.