All animal experiments were performed in accordance to the national guidelines provided by “The Guide for Care and Use of Laboratory Animals” and the University of Georgia Institutional Animal Care and Use Committee (IACUC). The IACUC approved all animal experiments.

Wild type (WT) C57BL/6 mice were purchased from the Jackson Laboratory. Tpl2^-/- mice backcrossed 10 generations onto the C57BL/6 strain were kindly provided by Dr. Philip Tsichlis (32). Animals were housed in micro-isolator cages in the Coverdell Rodent Vivarium, UGA. Both male and female mice were used in experiments to evaluate sex as a biological variable.

Embryonated, specific pathogen-free chicken eggs were purchased from Poultry Diagnostics and Research Center, UGA. Influenza virus A/HKx31 (H3N2; hereafter x31) stocks were propagated in the allantoic cavity of 9- to 11-day-old, embryonated, specific pathogen-free (SPF) chicken eggs at 37°C for 72 hours, and viral titers were enumerated by plaque assays. Madin Darby Canine Kidney (MDCK) cells were cultured and plated on a 12-well plate at a concentration of 5x10⁵ cells/well. After 24 hours the well is generally confluent, and 100 μl of serially diluted sample was added to the well, along with 200 μl of the Infection Media (Minimal Essential Media containing 1 µg/ml TPCK treated typsin and lacking serum). The sample was allowed to incubate with the cells for an hour at 37°C to promote infection of the monolayer, followed by the addition of 2.4% Avicel in Overlay Media (Infection Media with 40 mM HEPES, 4 mM L-gutamine, 200 U/ml penicillin, 200 U/ml streptomycin and 0.15% Sodium Bicarbonate) to facilitate localized viral infection and plaque formation. After 72 hours, wells were washed with PBS, cells were fixed with 60% acetone:40% methanol, and plaques were stained with crystal violet (made by mixing one volume of 0.0012 w/v of crystal violet powder in 5% Methanol, 11.1% Formaldehyde, 60% H₂O with one volume of PBS) for visualization and enumeration.

Age-matched, 6- to 8-week-old, WT and Tpl2^-/- mice were anesthetized with approximately 250 mg/kg of 2% weight/volume Avertin (2,2,2- Tribromoethanol, Sigma) followed by intranasal instillation of 50 µl PBS containing 10⁴ pfu of influenza A/HKX31 (H3N2, hereafter referred to as x31). The mice were studied for their susceptibility to infection by measuring daily weight loss and clinical scores according to the following index: piloerection, 1 point; hunched posture, 2 points; rapid breathing, 3 points. Mice with a cumulative score of 5 or that had lost 30% of their initial weight were humanely euthanized.

Mice were sacrificed at 7 to 9 dpi. Blood was collected from the heart by cardiac puncture into serum collection tubes, centrifuged at 9000 x g for 5 min, and the sera were stored at -80°C until cytokine analysis. Bronchoalveolar lavage fluid (BALF) was obtained from the lungs prior to harvest using 1 ml of PBS instilled twice into the lungs. The BALF was centrifuged at 500 x g for 5 min, and the cell-free BALF was stored at -80°C until cytokine analysis; the cellular pellet was lysed in TRK lysis buffer (E.Z.N.A Omega Bio-Tek, Inc. Norcross, GA, USA) for quantitation of gene expression. The lungs were perfused with 10 ml of PBS injected directly into the right ventricle of the heart. Lungs were harvested into 1 ml of PBS and homogenized in a bead mill homogenizer (Qiagen Tissue Lyser II) at 25 hz for 2-4 min. The homogenate was centrifuged at 500 x g for 5 min, and the pre-cleared homogenate was either: (1) directly aliquoted for viral titer assessment, (2) lysed in TRK tissue lysis buffer for RNA extraction, or (3) centrifuged at 5000 x g for 5 min to clarify the homogenate for cytokine analysis by ELISA. For mice sacrificed at 9 dpi, whole lungs were processed without perfusion or BALF harvesting.

Cytokine quantitation in the blood, BALF, and clarified lung homogenates was performed using the Mouse Inflammation Cytometric Bead Array (CBA) (IL-6, IFN-γ, MCP-1, TNF, IL-10 and IL-12p70, Becton Dickenson), Mouse ProcartaPlex 9-plex (RANTES, IP-10, MIP-1α, MIP-1β, IL-1α, IL-1β, IL-28, GM-CSF, Invitrogen), Standard murine ABTS ELISA Development kit (CXCL1 & IFN-γ, Peprotech) and Lumikine express kits (IFN-β, Invivogen).

At 4 and 7 dpi, the following protocol was used to assess the BALF and lung cellular composition after the BALF harvesting and lung perfusion as noted above. The lungs were harvested into Hyclone RPMI media (15-040-CV, Corning, Manassas, VA) containing 10% FBS and 2 mM L-glutamine (Invitrogen, Grand Island, NY). Lungs were minced with razor blades, and incubated in EDTA solution [RPMI 1640 containing 0.01 M HEPES (Lonza, Walkersville, MD), 1.25 mM EDTA (Fisher Bioreagents, Fair Lawn, NJ),] for 1 hour at 37°C in an incubator shaking at 250 RPM. The tissue was centrifuged at 350 x g for 10 min and then digested with 10 mL of collagenase solution [RPMI 1640 containing 1 mM CaCl_2, 0.01 M HEPES (Lonza, Walkersville, MD), 2 mM L-glutamine (Invitrogen, Grand Island), 100 U/ml penicillin, 100 U/ml streptomycin, 5% FBS, 0.2 µg/mL Gentamicin, and 150 U/ml collagenase (Sigma-Aldrich C2139)] for 30 min at 37°C in an incubator shaking at 350 RPM. The digested tissue was passed through a 70 µm cell strainer, and the cell suspension was centrifuged at 350 x g for 10 min, resuspended in 44% Percoll, and layered on top of 67% Percoll. The gradients were spun at 900 x g for 20 min (without brake), and the enriched leukocytes were recovered from the interface. The cells were washed with PBS at 350 x g for 10 min and enumerated using an automated cell counter (Cell Countess, Life Technologies).

Cells were stained at 4°C for 20 min with fluorescently-labeled antibodies against the following cell surface markers in the presence of Fc blocker (eBioscience, San Diego, CA location): Siglec F, CD11b, CD11c, Ly6C, Ly6G, CD45.2 (Stain 1); TCRαβ, TCRγδ, CD4, CD8, DX5, CD45.2 (Stain 2). The cells were fixed with 1% formalin and analyzed on the LSR II flow cytometer (BD Biosciences). CD45.2-gated hematopoietic-derived leukocyte populations were characterized as follows: inflammatory monocytes (Siglec F^-, CD11b^high, CD11c^low, Ly6C⁺), neutrophils (Siglec F^-, CD11b^high, CD11c^low, Ly6G⁺), alveolar macrophages (Siglec F^high, CD11b^int), eosinophils (Siglec F^high, CD11b^high), NK cells (αβTCR^-, DX5⁺), CD4 T cells (αβTCR⁺, CD4⁺), CD8 T cells (αβTCR⁺, CD8⁺), and γδ T cells (αβ TCR^-, γδ TCR⁺). The gating strategy is shown in Supplementary Figure 3.

Messenger RNA was extracted from lung homogenates using the E.Z.N.A. Total RNA kit (Omega Bio-Tek, Inc. Norcross, GA, USA) and converted into cDNA using a High Capacity RNA-to-cDNA kit (Thermo Fisher, Waltham, MA) according to the manufacturer’s protocol. Relative expression of various genes was assessed using Sensifast Probe Hi-ROX kit (BIO-82020 Bioline, Taunton, MA) and probes sourced from Applied Biosystems (Beverly, MA) using a StepOne Plus instrument (Applied Biosystems, Beverly, MA). Results are expressed relative to the actin internal control and the WT or untreated sample using the ΔΔC_T method. The probes used are as follows: IFNβ1(Mm00439552), IFNα1(Mm03030145), IFNα4(Mm00833969), IFNγ(Mm0116813), IL-6(Mm00446190), IL1β(Mm00434228), CCL2(Mm00441242), CXCL1(Mm04207460), CCL5(Mm01302427), TNFSF10(Mm01283606), NOS2(Mm00440502), MPO(Mm01298424), STAT1(Mm00439518), SOCS1(Mm01342740), IL-10(Mm01288386), SOCS3(Mm01249143), SOCS4(Mm00439518) and STAT4(Mm00448890)

WT and Tpl2^-/- mice were irradiated at 1100 Rads after reaching adulthood (> 6 weeks of age) and then injected with bone marrow from C57BL/6 mice at 3 x 10⁶ cells in 200 µl PBS. The mice were then maintained on acidified water (pH 2.5) for 2 months to allow for reconstitution of the hematopoietic compartment with WT cells. The resulting chimeras were infected with 10⁴ pfu x31 virus, and body weights were measured over a period of 8 to 10 days, at which times the mice were euthanized to assess the cytokine profiles on days with varying pathologies.

P values were calculated with GraphPad PRISM software version 9.2.0(332) using (one-way ANOVA with Tukey’s multiple comparisons test. Differences were considered statistically significant if p ≤ 0.05. Data represent means ± SEM. Survival data are graphed as Kaplan-Meier plots using GraphPad PRISM software, and p values were determined by Mantel-Cox test. Gaussian Correlation was performed to calculate the coefficient based on Pearson’s Correlation test with a two-tailed test. Simple Linear Regression analysis was also performed to analyze the best fit value or the slope and intercept to see if the two variables being compared for a particular genotype correlated or not (as seen by the straight dashed line). Additionally, the confidence interval was set at 95% (as seen by the curved dashed line).

We have previously observed that Tpl2^-/- mice are more susceptible than WT mice to influenza A virus infection using a low pathogenicity strain (x31; H3N2) (22). We reasoned that increased morbidity in the Tpl2^-/- mice was likely due to impaired or delayed viral clearance or to an excessive anti-viral immune response. To distinguish between these possibilities, we infected WT and Tpl2^-/- mice with x31 and monitored viral titers and inflammatory cytokine production as functions of morbidity and mortality late during the disease course. WT mice showed signs of disease from 1 to 7 dpi, at which time they began to recover as evidenced by weight gain and decreasing clinical scores (Figures 1A–C). In contrast, Tpl2^-/- mice displayed progressive weight loss and increasing clinical symptoms from 7 to 9 dpi (Figures 1A–C). Notably, the weight loss and clinical symptoms were not different between male and female mice after 7 dpi with influenza (Supplementary Figures 1A–D). Importantly, despite severe clinical symptoms in Tpl2^-/- mice, no virus was observed in either WT or Tpl2^-/- mice at peak morbidity and mortality (9 dpi; Figure 1D), demonstrating that both strains had successfully cleared the virus by this time point. As expected from these findings, there was no correlation between morbidity as measured by weight loss and viral titers (Figure 1E), confirming that the morbidity observed in influenza-infected Tpl2^-/- mice was not due to increased viral loads.

Hypercytokinemia is widely reported in severe influenza-infected patients that eventually succumb to disease (9, 13). It is characterized by significantly increased levels of interferons (IFNs), IL-6, TNF, IL-12, IL-1β and various chemokines such as CCL2, MIP-1α, MIP-1β, RANTES, and IP-10 (9, 13). Therefore, the pro-inflammatory cytokine profile of lung homogenates from influenza-infected WT or Tpl2^-/- mice was assessed using a multiplex protein assay. Significantly increased levels of IFN-β, CCL2, IFN-γ, CCL3, CCL4, CCL5 and CXCL10 were observed in the lung tissue of influenza-infected Tpl2^-/- mice compared to WT (Figures 2A–C, I–L); increased levels of IFN-β, IL-6 and IL-10 were observed in the air spaces of Tpl2^-/- mice (Supplementary Figures 2H, K, N); and increased levels of IFN-γ were observed in the blood of Tpl2^-/- mice at 7 dpi (Supplementary Figure 2C). Notably, the levels of some cytokines associated with influenza infection were unaltered (Figures 2D–H, M–P) and Supplementary Figures 2A, B, D–G, I, J, L, M). These data demonstrate that Tpl2^-/- mice display increased levels of pro-inflammatory cytokines and chemokines typically observed in human patients with influenza-induced hypercytokinemia (8–11). Notably, increased weight loss at 7 dpi correlated with high levels of IFN-β in the lungs of the Tpl2^-/- mice (Figure 2Q). However, there was no correlation between weight loss and viral load in the tissue at 7 dpi (Figure 2R), as was the case at 9 dpi (Figure 1E). These data demonstrate that the morbidity in Tpl2^-/- mice is due to the overexuberant immune response in Tpl2^-/- mice at the late stage of influenza infection rather than impaired viral control.

We next assessed the cellular composition of the lung tissue and alveolar air spaces to identify cells that would be consequently recruited due to the hypercytokinemia and could contribute to tissue damage in the lungs and mortality. Mice were infected with influenza, and bronchoalveolar lavage fluid (BALF) and lung tissue were harvested at 7 dpi for analysis of cellular composition by flow cytometry (Supplementary Figure 3). The total cellularity of the lungs was significantly increased in Tpl2^-/- mice. Higher numbers of cells were present in the perfused and lavaged lung tissue of the Tpl2^-/- mice (Figure 3A). Furthermore, Tpl2^-/- mice had significantly increased absolute numbers of inflammatory monocytes and neutrophils compared to WT mice at 7 dpi (Figures 3B, C). An increase in frequency of inflammatory monocytes and neutrophils was also noted (Supplementary Figures 4A, B), however no differences in total alveolar macrophages, NK cells, CD4 or CD8 αβ T cells or even γδ T cells were observed (Figures 3D–H). Therefore, it is the numerical increase in inflammatory monocytes and neutrophils that account for the higher total cellular infiltrates observed in infected Tpl2^-/- mice. Typically, innate immune cells are recruited early during influenza infection to phagocytose or endocytose infected cells (35); therefore, it was unexpected to see such high numbers of them late in the infection in Tpl2^-/- mice. Because Tpl2 deficiency has been demonstrated to impair monocyte, macrophage and neutrophil recruitment in response to inflammatory stimuli (21, 36–39), we further assessed the kinetics for the paradoxically increased monocytes and neutrophils in the lung tissue of influenza-infected Tpl2^-/- mice. Therefore, we characterized the cellular composition of lungs at an earlier time point (4 dpi) with uninfected mice as negative controls. Although we noted influenza infection-induced recruitment of both inflammatory monocytes and neutrophils by 4 dpi, there was no difference between the WT and Tpl2^-/- mice (Figures 3I, J). This was also true of all the other cell types examined (Supplementary Figures 4E, F). These data suggest a late acting effect of Tpl2, possibly in limiting the amplitude of the response or in promoting resolution of inflammation.

Highly pathogenic influenza viruses induce exaggerated immune responses that cause immunopathology via damage to the pulmonary epithelium by the recruited immune cells and their effector molecules (17, 40). A study of juvenile mice that exhibit severe disease in response to influenza infection revealed recruitment of inflammatory monocytes with high expression of inducible nitric oxide synthase (NOS2) (41), which induces apoptosis of epithelial cells. Another mediator of influenza-associated lung injury is myeloperoxidase (MPO), which is predominantly secreted by neutrophils during cases of severely pathogenic influenza infections (42). Neutrophil Elastase (ELANE) is an inflammatory mediator of neutrophils that is predictive of development of acute lung injury (ALI) or acute respiratory distress syndrome (ARDS), however its role in influenza infections is debatable (43–46). TNF-related apoptosis-inducing ligand (TRAIL), released by NK cells and inflammatory monocytes, interacts with death receptors on the surface of epithelial cells to induce apoptosis (47). Given that Tpl2 ablation leads to increased recruitment of inflammatory monocytes and neutrophils to the lungs, we next assessed the lung expression of pro-inflammatory cytokines and chemokines that recruit these cells as well as their effector molecules that could potentially damage the pulmonary epithelium and compromise lung function. Consistent with protein data, we noted overexpression of various pro-inflammatory cytokine mRNAs in lung tissue from influenza-infected Tpl2^-/- mice at 7 dpi, including IFN-β, IFN-γ, IL-6 and the anti-inflammatory cytokine IL-10, as well as overexpression of chemokines CCL2, CXCL1, CCL5 and CXCL10 which are collectively involved in recruitment of inflammatory monocytes and neutrophils (Figures 4A–L). We also examined the level of CXCL2, another neutrophil recruiting chemokine (48) known to be active in bacterial infection models and found no difference between WT and Tpl2^-/- lung tissue (Figure 4K). On testing for various inflammatory mediators, including NOS2, MPO, ELANE and TNFRSF10 (the gene encoding TRAIL), we observed upregulation of only NOS2 in the lungs of Tpl2^-/- mice at 7 dpi (Figures 4M–P), suggesting that elevated NOS2 secretion by the increased numbers of inflammatory monocytes or neutrophils may contribute to morbidity. Notably, we observe that the IFN-β mRNA expression correlated with CCL2 mRNA expression (Figure 4S), and NOS2 mRNA expression in the lungs of the Tpl2^-/- mice correlated with both CCL2 and IFN-β expression (Figures 4T, U), supporting the hypothesis that over-expression of IFN-β in Tpl2^-/- mice stimulates increased production of the IFN-inducible gene, CCL2, which is responsible for the recruitment of NOS2-expressing monocytes that contribute to the damage of the pulmonary epithelium. Additionally, CXCL1 mRNA expression positively correlated with both IFN-β and NOS2 mRNA expression in the lungs of Tpl2^-/- mice (Figures 4V, W), even though CXCL1 was not upregulated by protein expression at 7 dpi.

Because IFN-γ protein and mRNA expression were very high in influenza-infected Tpl2^-/- mice (Figures 2C and 4D) and have been reported to induce damage via NOS2 (49), we also examined whether IFN-γ correlated with NOS2 mRNA expression and did not find any correlation (Figure 4X). To further address the source of the high levels of IFN-γ, Tpl2^-/- mice and Tpl2^-/-Rag1^-/- mice were infected with influenza, and IFN-γ protein levels in lung homogenates were assessed at 7 dpi. Both Tpl2^-/- and Tpl2^-/-/Rag^-/- mice that lack T cells, produced similar levels of IFN-γ protein (Supplementary Figure 5A), suggesting that T cells are not the source of IFN-γ overproduction. Collectively, these significant correlations support the hypothesis that the expression of NOS2 is linked to the recruitment of inflammatory monocytes and neutrophils under the influence of IFN-β overexpression and the most likely cause of the morbidity seen in the Tpl2^-/- mice.

The IFNs signal primarily via activation of the JAK/STAT pathway, with STAT1 playing a central role for both Type I and II interferons (50). Furthermore, interferons participate in a feed-forward loop with IFN-β amplifying the signal through Interferon Alpha Receptor 1 (IFNAR1) by inducing multiple IFNαs and other interferon-stimulated genes (ISGs), such as CCL2 (50, 51). Finally, resolution of this pathway is mediated in large part by the IFN-mediated induction of suppressors of cytokine signaling 1 (SOCS1), which downregulates interferon expression and signaling via STAT1 (52). Because Tpl2^-/- mice show higher recruitment of inflammatory monocytes and neutrophils as the infection progresses, we hypothesized that Tpl2 either limits the amplitude or promotes resolution of the antiviral IFN response. To address the regulation of the IFN pathway by Tpl2 in response to influenza infection, we first measured the expression of both STAT1 and SOCS1, which serve as positive and negative regulators of the IFN pathway, respectively. At the peak of morbidity in Tpl2^-/- mice at 9 dpi, STAT1 was markedly downregulated in the lungs of Tpl2^-/- mice (Figure 5A), but not earlier at 7 dpi (Figure 4Q). In order to determine the cause of STAT1 downregulation, we assessed expression of the various SOCS genes and found that SOCS1 was overexpressed by 7 dpi (Figure 4R) and remained elevated through 9 dpi (Figure 5B). This upregulation was specific to SOCS1 and not seen with the other SOCS or STAT proteins (Figures 5C–E). Notably, the IFNs and ISGs were no longer upregulated at a transcriptional level in Tpl2^-/- mice at 9 dpi, suggesting that elevated SOCS1 is suppressing the IFN response (Figures 5F–P). Additionally, in Tpl2^-/- mice, SOCS1 is upregulated at 9 dpi while all IFNs and ISGs decline to WT levels by 9 dpi, providing further evidence of SOCS1-mediated transcriptional repression of the interferon response. Overexpression of SOCS1 in Tpl2^-/- lung tissue was associated with increased levels of IL-10 protein in the lungs (Figure 5R), indicative of a reparative response. However, despite elevated SOCS1 and decreased NOS2 mRNA expression in Tpl2^-/- mice at 9 dpi (Figures 5B, N), CCL2 protein levels remained elevated (Figure 5Q) and were accompanied by trending higher levels of other pro-inflammatory cytokines, including IFN-γ and IL-6 (Figures 5S–V). Cxcl1 levels were not different (Figure 5W), consistent with protein levels observed at 7 dpi (Figure 2H). Therefore, we conclude that inefficient regulation of the IFN/STAT1 pathway via SOCS1 permitted persistently elevated levels of CCL2, the chemokine recruitment signal for monocytes, in Tpl2^-/- mice at the peak of morbidity (9 dpi). Collectively, these findings suggest that dysregulation of the IFN pathway in Tpl2^-/- mice promotes excessive and prolonged influx of inflammatory cells that contribute to lung damage and morbidity observed in influenza-infected Tpl2^-/- mice.

In an effort to localize Tpl2 functions that regulate hypercytokinemia in late stages of influenza infection, we generated chimeras using WT or Tpl2^-/- recipient mice that were given WT donor bone marrow post irradiation, ensuring that hematopoietic cells would be of WT origin post recovery (as outlined in Figure 6A). Differential weight loss was observed in the Tpl2^-/- chimeras from 7 to 8 dpi (Figure 6B), after which the mice recovered. Upon examination of the cytokines in the lungs, CCL2, IFN-γ and IL-6 were all upregulated in Tpl2^-/- chimeras (Figures 6C–E) at 8 dpi, whereas the levels of other pro-inflammatory cytokines such as TNF, IL-12, IL-10 and CXCL1 were not affected (Figures 6F–I). Upregulation of CCL2, IFN-γ and IL-6 at 8 dpi in Tpl2^-/- chimeras suggests that the cytokine dysregulation is partially attributed to Tpl2 deficiency in radioresistant cells, such as epithelial (53, 54) or stromal cells (55). Importantly, the regulation of these cytokines normalized in Tpl2^-/- chimeras by 10 dpi, at which time the Tpl2 chimeras fully recovered. This overall phenotype is unlike the prolonged cytokine dysregulation and progressive weight loss seen in germline Tpl2^-/- mice at 9 dpi (Figure 1A), suggesting that Tpl2^-/- chimera recovery is mediated by suppression of hypercytokinemia by WT hematopoietic cells. As alveolar macrophages are also part of the lung resident immune cell population and are radioresistant (56), we considered the possibility that the source of the dysregulated cytokines at 7 dpi was the alveolar macrophages rather than the epithelial or stromal cells. However, we found no differences in the gene expression for CCL2, IL-6, IFN-γ or IFN-β at 7 dpi in sorted alveolar macrophages from WT and Tpl2^-/- lungs (Supplementary Figures 6A–D).