The signaling adaptor insulin receptor substrate 2 (IRS2) limits lethal host responses induced by influenza infection

Abstract

Influenza is a highly contagious virus with most individuals showing recovery within a week after symptom onset. However, children, the elderly, and those with chronic health conditions are at high risk of developing serious complications. Extensive immune cell infiltration of the lung and massive production of pro-inflammatory cytokines can result in lung tissue damage and disruption, resulting in hypoxia and death. During the 2024-2025 influenza season in the USA, influenza caused an estimated 27,000-130,000 deaths. Our previous studies have shown that agents known to drive polarization of macrophages into the M2a phenotype mitigated inflammatory responses to influenza H1N1 A/PR/8/34 (PR8). Since IRS2 dampens the IL-4-induced differentiation to the M2a subtype, we sought to determine the impact of IRS2 on PR8 infection. Contrary to expectations, IRS2^-/- mice exhibited enhanced susceptibility to PR8 when compared to IRS2^+/+ mice, which was associated with a significant drop in blood oxygen saturation. Additionally, PR8 infection induced significantly greater inflammation in the lungs of IRS2^-/- mice than IRS2^+/+ mice. Conversely, viral replication was not significantly different. Significantly increased levels of the inflammatory mediator HMGB1 and airway epithelial cell denuding were observed in the lungs of IRS2^-/- mice, with a concomitant decrease in M2a gene expression. Mice with myeloid-specific knockout of IRS2 showed no enhanced sensitivity to PR8, supporting the hypothesis that the major impact of IRS2-deficiency on host response to PR8 infection is not myeloid cell-intrinsic. Overall, these results suggest that in the absence of IRS2, airway epithelial cells are especially sensitive to influenza-induced damage through dysregulated inflammation, resulting in increased susceptibility during infection.

Introduction

Influenza is a highly contagious respiratory illness. The WHO estimate of annual influenza-associated deaths globally is ∼300,000-650,000, ¹ and during the 2024-2025 influenza season in the USA, influenza caused an estimated 47-82 million illnesses and 27,000-130,000 deaths.^2,3 Influenza targets the nose, throat, bronchi, and lower respiratory tract. Most people recover within a week after exhibiting headache, sore throat, rhinitis, fever, malaise, and coughing; however, others are at high risk of developing serious complications including children, the elderly, and individuals with chronic conditions such as asthma, diabetes, heart disease, and obesity.^4–9

Some strains of influenza, such as influenza A H1N1 pandemic strains, or H5N1 and H7N9 zoonotic strains, can trigger an overzealous host response leading to extensive immune cell infiltration of the lung, and massive production of pro-inflammatory cytokines (“cytokine storm”) resulting in lung tissue damage such that the lungs cannot deliver enough oxygen to the blood resulting in hypoxia and death.^10,11 Seasonal influenza A viruses that currently circulate in humans, including H1N1 and H3N2, are often self-limited but still infect the respiratory epithelium and can trigger inflammatory responses that contribute to airway injury, pneumonia, and lung damage in some patients.^12–15 Shirey et al. previously demonstrated that infection of mice with the mouse-adapted strain of influenza H1N1 A/PR/8/34 (PR8) mediated a pro-inflammatory cytokine storm and death by releasing High Mobility Group Box 1 (HMGB1), the tissue damage-associated molecular pattern (DAMP). HMGB1 stimulated responses via Toll-like receptor (TLR) 4 and its co-receptor, MD-2,^11,16 as evidenced by the efficacy of the TLR4 antagonist, Eritoran, and the small molecule HMGB1 antagonist, P5779, to protect mice from lethal infection.^11,16 In cotton rats (Sigmodon hispidus), a rodent model that is uniquely sensitive to infection by non-adapted human respiratory virus infections, ¹⁷ the severity of human influenza strains was shown to correlate well with the induction of HMGB1, and in this same model, blocking TLR4-mediated signaling (including by HMGB1) with Eritoran or P5779 blunted lung inflammation and proinflammatory cytokine production.^16,18

Severe responses to PR8 were also mitigated by agents that influence tissue repair and macrophage (Mϕ) differentiation phenotypes. ¹⁶ Mϕs can adopt a spectrum of phenotypes in response to different environmental stimuli.^19–22 Mϕs exposed to strong inflammatory stimuli, like LPS and IFN-γ, become highly microbicidal, “classically activated” (M1) Mϕs, producing high amounts of many pro-inflammatory cytokines including IL-1β, TNF-α, and IL-6. Conversely, exposure of Mϕs to IL-4 or IL-13 leads to the development of “alternatively activated” (M2) Mϕs that counter the strong pro-inflammatory M1 phenotype and are associated with tissue repair and healing.^19,20 However, it is now appreciated that in a spectrum of Mϕ phenotypes may develop in response to complex environmental stimuli in vivo.^21–24 M2 Mϕs have been subdivided into M2a, M2b, M2c, M2d and regulatory subtypes based on their inducers and expression of surface markers and cytokines produced. ²² Our previous studies have shown that cytokines known to drive M2a Mϕs, e.g., IL-4 and IL-13, that signal through a shared signaling protein, IL-4Rα, ²⁵ or by agonists of Peroxisome Proliferator-Activated Receptor gamma (PPARγ), a transcription factor important for M2a differentiation,^26–30 led to induction of M2a Mϕs with concomitant mitigation of inflammatory responses to viral infections.^31,32

Previous studies by the Keegan laboratory reported that M2a Mϕs participate in enhancement of allergic lung inflammation and increased muscularization of small pulmonary vessels and pulmonary hypertension in response to hypoxic conditions.^33–35 Importantly, using global IRS2^-/- mice, ³⁴ insulin receptor substrate 2 (IRS2), a signaling molecule important for IL-4 signaling and insulin resistance and metabolism, was shown to have an anti-inflammatory role in Th2-skewed lung inflammation and pulmonary vascular remodeling in vivo. This was due, in part, to the observation that IRS2 dampens the IL-4-induced differentiation of Mϕs to the M2a subtype. However, the role of IRS2 in controlling responses in a strongly pro-inflammatory model such as PR8 infection has not been evaluated. ³² Herein, we provide evidence that global IRS2^-/- mice were highly susceptible to a dose of PR8 that is sublethal in IRS2^+/+ littermate controls, with 100% of mice succumbing to infection by day 9 post-infection (p.i.). These mice also showed a significant decrease in blood oxygen saturation 5 days p.i., markedly increased pro-inflammatory gene expression and decreased tissue repair gene expression. Furthermore, using mice with a deletion of IRS2 in myeloid cells, we show that cells of myeloid lineage do not contribute significantly to the increased sensitivity of the global IRS2^-/- mice to PR8. Taken together, these results demonstrate that the IRS2 adaptor limits host inflammatory responses to influenza in vivo via a mechanism that is not myeloid cell-intrinsic.

Results

We previously reported that infection of wild-type C57BL/6J mice with a low dose of influenza A/PR/8/34 (PR8) induced early, robust expression of pro-inflammatory genes in the lungs that was followed by induction of tissue repair genes known to be produced by M2a Mϕs including Arg1, Retnla, and Chi3L3. ³⁶ We also reported that BALB/c mice lacking the IL-4Rα receptor, required for IL-4- and IL-13-driven M2a Mϕ differentiation, were more susceptible to PR8-induced lethality, while mice that lacked the M1 gene, Ptgs2 (encoding cyclooxygenase 2, COX2), were more resistant. ³² In addition, LysMCre-mediated conditional knockout of PPARγ, a nuclear receptor that plays a role in immune cell function and inflammation, in myeloid cells, significantly enhanced PR8-induced lethality when compared to WT mice. ³² The PPARγ-induced effect was correlated with its role in M2a Mϕ differentiation. ³² Taken together, these results suggest that M2a-tissue repair Mϕs mediate lung damaging host responses activated by PR8 infection. These host responses include production of HMGB1 by dying PR8-infected lung epithelial cells that stimulates TLR4 signaling, induction of inflammatory cytokines, and increases COX2 expression that has been shown to play a role in the influenza infection-induced pathology.^{16,32,36–38}

Since host responses to influenza are influenced by the M1/M2a Mϕ balance and include tissue and blood hypoxia, we first examined the responses of IRS2-deficient mice to a dose of PR8 (LD₁₀) that is sublethal in WT mice.^39,40 The absence of IRS2 in lungs of PR8-infected IRS2^-/- mice was confirmed by Western blot analysis (Figure 1(a)). We initially hypothesized that IRS2^-/- mice would be relatively protected from PR8-induced pathology compared to IRS2^+/+ mice because enhanced M2a Mϕ differentiation had been reported previously in allergen- or hypoxia-induced models of inflammation.^34,35 Unexpectedly, global IRS2^-/- mice were highly susceptible to a sub-lethal challenge with PR8 compared to IRS2^+/+ littermate controls with 100% of mice succumbing to infection by day 9 p.i. (Figure 1(b)). Heterozygotic IRS2^+/- mice demonstrated an intermediate susceptibility to PR8 infection. The IRS2^-/- and IRS2^+/- mice also showed a significant drop in blood oxygen saturation (SpO₂) measured 5 days p.i. as compared to IRS2^+/+ mice (Figure 1(c)). Lungs were harvested from PR8-infected mice 5 days p.i. and stained with H&E (Figure 1(d), representative images). Histopathology was scored blindly for each slide for four distinct parameters of lung inflammation: peribronchiolitis (inflammatory cell infiltration around bronchioles), perivasculitis (inflammatory cell infiltration around the small blood vessels), interstitial pneumonitis (inflammatory cell infiltration; thickening of alveolar walls), and alveolitis (cells within alveoli) that were combined to create a cumulative pathology score. ⁴¹ Figure 1(e) illustrates that PR8 infection induced significantly greater inflammation in the lungs of IRS2^-/- mice than IRS2^+/+ mice, with IRS2^+/- exhibiting an intermediate inflammatory response.

Figure 1.

IRS2-deficient mice are more susceptible to PR8-induced lethality. (a) Confirmation of global IRS2 deletion in IRS2^-/- mice. IRS2^+/+ and IRS2^-/- mice were infected with an LD₁₀ of PR8 (∼1500 TCID₅₀). Lungs were extracted on day 5 post-infection (p.i.) for protein measurements. Lung lysates from 4 IRS2 ^+/+ and 4 IRS2 ^-/- mice either uninfected (M) or infected with PR8 (P) were prepared as detailed in the Materials and Methods. These lysates were separated by SDS-PAGE, transferred to a PVDF membrane, and probed with antibodies specific for IRS2 and β-actin. Band intensity was calculated by densitometry using ImageJ and IRS2 expression was normalized to beta-actin for each sample and listed below the gel images. (b) IRS2^+/+, IRS2^+/-, and IRS2^-/- mice were infected with an LD₁₀ of PR8 (∼1500 TCID₅₀). Mice were monitored for survival for 14 days p.i. N = 7-10 mice/strain. * p =0.0104 (IRS2^+/- vs IRS2^-/-); *** p = 0.0007 (IRS2^+/+ vs IRS2^-/-). Survival data were analyzed by the Mantel-Cox log rank test. (c) IRS2^+/+, IRS2^+/-, and IRS2^-/- mice were infected with an LD₁₀ of PR8 (∼1500 TCID₅₀). Blood Oxygen saturation (SpO₂) was measured at day 5 post-infection. N = 2-4 mice/group (mock); N = 7-13/group (PR8). ** p = 0.0014; *** p = 0.0005; ****; p < 0.0001. (d) Mice were infected as in C and euthanized on day 5 p.i. Lungs were harvested, sectioned, and H&E stained for histopathology and examined for tissue damage and inflammatory cellular infiltration. Representative micrographs are shown. (e) Combined mean histopathology scores for each group were determined blindly as previously described ³⁷ N = 2-4 mice/group (mock); N = 7-13/group (PR8). ** p < 0.01; *** p < 0.0006; **** p < 0.0001.

The fact that increased susceptibility to PR8 was observed in IRS2^+/- mice that do not develop overt diabetes ⁴² suggests that the observed increase in sensitivity to PR8 is not directly due to high blood glucose levels. To examine the relationship of blood glucose to PR8 sensitivity phenotype, blood glucose levels in the infected mice fed ad libitum were measured and compared to both SpO₂ levels ¹¹ and histopathology scores (Figure 2). Figure 2(a) illustrates that the blood glucose levels of infected IRS2^-/- mice were modestly elevated (350 mg/dL) compared to those of the IRS2^+/+ or IRS^+/- mice (185 and 192 mg/dL, respectively). ⁴² To evaluate the relationship of blood glucose levels to PR8 susceptibility, we plotted the SpO₂ of individual mice after PR8 infection against their corresponding blood glucose levels (Figure 2(b)). There was no correlation between the two outcomes, indicating that they are unrelated. When SpO₂ versus histopathology score was plotted in mice after PR8 infection, we observed clustering of IRS2^+/+ mice with high oxygen levels and low histopathology scores while the IRS2^-/- mice clustered with low oxygen levels, high histopathology scores (Figure 2(c)). The IRS2^+/- mice failed to show tight clustering in this analysis. Taken together, these results indicate that the sensitivity of the mice to PR8 infection is not related to the blood glucose levels, while lower SpO₂ levels correlate with worsened histopathology.

Figure 2.

Sensitivity to PR8 is not related to blood glucose levels or viral replication. (a) IRS2^+/+, IRS2^+/-, and IRS2^-/- mice were infected with an LD₁₀ of PR8 (∼1500 TCID₅₀) as in Figure 1(c). Blood glucose levels were measured at day 5.i. N = 8-13 mice/strain. ** p = 0.0012; *** p = 0.0006. (b) and (c) Percent blood O₂ saturation levels measured in Figure 1(c) were compared with blood glucose levels (b) and the combined histology scores (c) at day 5 post-infection. Each point represents a single mouse. (d) IRS2^+/+, IRS2^+/-, and IRS2^-/- mice were infected with an LD₁₀ of PR8 (∼1500 TCID₅₀) as in Figure 1(c). On day 5 p.i., mice were euthanized and lungs were extracted and processed for total RNA and gene expression of the influenza matrix protein (M1) analyzed by qRT-PCR.

To dissect the mechanism by which IRS2 influences the severity of influenza infection, we first evaluated viral load prior to the loss of the IRS2^-/- mice. No significant differences in viral replication were observed among the IRS2 genotypes as measured by the relative abundance of the viral matrix protein, M1, mRNA 5 days after infection (Figure 2(d)). We further evaluated the abundance of pro- and anti-inflammatory genes induced by PR8 in the lungs of mice. Contrary to our initial expectation that IRS2^-/- mice would be less sensitive to infection, we observed significant increases in PR8-induced proinflammatory molecules characteristic of M1 Mϕs (Il1b, Tnf, Il6, Ptgs2; top 4 panels) in the lungs of IRS2^-/- mice, with a concomitant decrease in M2a “tissue repair” genes (Arg1, Chil3, Retnla, Mrc1; bottom 4 panels) compared to IRS2^+/+ mice, with heterozygotic mice exhibiting intermediate levels of both pro- and anti-inflammatory gene expression (Figure 3(a)). This result is consistent with enhanced lung pathology in IRS2^-/- and IRS2^+/- mice (Figure 1), and is in contrast to the impact of IRS2-deficiency on repair genes in the lungs of mice experiencing allergic inflammation or hypoxia.^34,35

Figure 3.

Gene induction in lungs of PR8-infected, IRS2-deficient mice. IRS2^+/+, IRS2^+/-, and IRS2^-/- mice were infected with an LD₁₀ of PR8 (∼1500 TCID₅₀) as in Figure 1(c). On day 5 p.i., mice were euthanized and lungs were extracted and processed for total RNA and gene expression analyzed by qRT-PCR (a) and HMGB1 protein levels (b). N = 2-5 mice/strain (mock); N = 7-13 mice/strain (PR8). The data shown is combined from 3 separate experiments. * p < 0.05; ** p < 0.005; *** p < 0. 0.001; **** p < 0.0001.

Since extracellular release of the protein HMGB1, an endogenous DAMP produced by dying lung epithelial cells, mediates influenza-induced pathogenesis by stimulating TLR4/MD-2,^16,43 we also evaluated the impact of IRS2 on HMGB1 levels in lung homogenates. Consistent with increased susceptibility of IRS2^+/- and IRS2^-/- mice to PR8, increased amounts of HMGB1 in the lung homogenates of the IRS2^+/- and IRS2^-/- mice were detected compared to the IRS2^+/+ mice (Figure 3(b)). These results suggest that the IRS2 adaptor controls the relative balance of pro- and anti-inflammatory genes in the lungs of influenza-infected mice by modulating HMGB1 levels.

Given that IRS2 influences Mϕ polarization ³⁴ and that 5-lipoxygenase (5-LO) contributes to the production of both pro-inflammatory (leukotrienes) and anti-inflammatory (lipoxins) mediators,^44,45 it is possible these pathways may interact to regulate the overall inflammatory and tissue repair processes. Using Western blot analysis (Figure 4(a)) and ImageJ analysis of those blots (Figure 4(b)), we observed a significant decrease in 5-LO protein expression in the lungs of PR8-infected IRS2^-/- mice compared to the lungs of IRS2^+/+ mice (Figure 4(a), top panel; Figure 4(b), left panel). IRS2 has been shown to play a role in the insulin signaling pathway that is downstream of PPARγ ⁴⁶ as well as the activation of Akt through PI3K-dependent phosphorylation of Ser473. ⁴⁷ PR8-infected mice exhibited reduced phosphorylation of PPARγ on Ser112 in IRS2^-/- lungs compared to IRS2^+/+ lungs, as well as reduced levels of total PPARγ protein (Figure 4(a) and (b), second and third panels). In addition to the role IRS2 plays in Akt activation, PPARγ has also been shown to increase the phosphorylation of Akt. ⁴⁸ However, there was no statistical difference in the phosphorylation of Akt on Ser473 between the PR8-infected or total Akt levels in IRS2^-/- compared to IRS2^+/+ mice (Figure 4(a), fourth and fifth panels). These data demonstrate that IRS2 modulates the abundance of 5-LO and phosphorylated PPARγ in the lung during influenza infection.

Figure 4.

Changes in protein expression in the lungs of PR8- infected, IRS2-deficient mice. Lung lysates from 7 IRS2 ^+/+ and 7 IRS2 ^-/- mice infected with PR8 were prepared as detailed in the Materials and Methods. (a) Lysates were separated by SDS-PAGE, transferred to a PVDF membrane, and probed with antibodies specific for 5-LO, phospho-PPAR-γ (S112), total PPAR-γ, phospho-Akt (S473), total Akt, and β-actin. (b) Band intensity was calculated by densitometry using ImageJ and the ratio of the indicated two groups was compared.

Several tissue repair genes characteristic of M2a Mϕs are also produced by lung epithelial cells as well as other cell types. ⁴⁹ To directly test the contribution of IRS2 in myeloid cells to the inflammatory response induced by PR8, we utilized mice with LysMCre-mediated conditional knockout of IRS2^50,51 (Figure 5). In contrast to the IRS2^+/+ and IRS2^-/- mice, there was no difference in the survival of IRS2^fl/flLysMCre^+/- compared to IRS2^fl/flLysMCre^-/- littermates after infection with PR8 (LD₁₀). Similar results were obtained using a higher dose of PR8 (LD₄₀) or using IRS2^fl/flLysMCre^+/+ mice (data not shown). These data suggest that loss of IRS2 expression in the myeloid compartment (e.g., Mϕs, monocytes, neutrophils) is not sufficient to alter influenza sensitivity.

Figure 5.

Conditional deletion of IRS2 in myeloid cells via LysM-Cre has no effect on sensitivity to PR8. IRS2^fl/fl/LysMCre^+/- and IRS2^fl/flLysMCre^-/- littermates were infected with an LD₁₀ of PR8 (∼1500 TCID₅₀). Mice were monitored daily for survival for 14 days p.i. N = 5-7 mice/strain. Combined data from two separate experiments.

To understand the responses of myeloid Mϕs in an environment of pro-inflammatory signals, IRS2 ^+/+ and IRS2 ^-/- peritoneal exudate Mϕs were cultured and stimulated with medium alone or IL-4 (20 ng/mL) for 4, 8, or 24 h and analyzed for M2a markers by qRT-PCR. No statistical difference was observed between IL-4-stimulated IRS2^+/+ and IRS2^-/- Mϕs for Arg1 and Chil3 gene expression at any time point (Figure 6). However, a ∼2-fold increase was observed for Retnla gene expression in the IRS2^-/- Mϕs (Figure 6). We also observed a significant increase in total PPARγ protein in both IL-4-stimulated IRS2^+/+ and IRS2^-/- Mϕs at 4, 8, and 24 h (Figure 7(a)–(c), respectively) with a significant increase seen at 8 h in the IRS2^-/- compared with the IRS2^+/+ Mϕs (Figure 7(b)). Additionally, a decrease in the phosphorylation of Akt was observed in the IRS2^-/- compared to the IRS2^+/+ Mϕs (Figure 7(a)–(c), respectively). Unlike the PR8-infected lungs (Figure 4), we observed no significant changes in 5-LO protein, even with rIL-4 stimulation, in either IRS2^+/+ or IRS2^-/- Mϕs. This pattern of responses contrasts with the decreased abundance of M2a genes including Retnla and PPARγ and 5-LO protein in the lungs of IRS2^-/- compared to IRS2^+/+ mice infected with PR8 (Figures 3 and 4), supporting the hypothesis that the major impact of IRS2-deficiency on host response to PR8 infection is not myeloid cell-intrinsic.

Figure 6.

Induction of M2a Mϕ markers in IL-4-treated IRS2 ^+/+ and IRS2 ^-/- Mϕ. Thioglycollate-elicited peritoneal macrophages from IRS2 ^+/+ and IRS2 ^-/- mice were left untreated (-) or stimulated (+) with rIL-4 (20 ng/mL) for 4, 8, or 24 h. Expression of Arg1 (a), Retnla (b), and Chil3 (c) mRNAs was quantified by qRT-PCR at the three indicated time points. Data in (a)–(c) represent the compilation of the “fold-induction” values compared to untreated IRS2 ^+/+ cells from 3 independent experiments with each symbol representing the mean value +/- SEM calculated from these individual experiments. * p < 0.05.

Figure 7.

Changes in protein expression in rIL-4-treated IRS2 ^+/+ AND IRS2 ^-/- Mϕ. IRS2 ^+/+ and IRS2 ^-/- peritoneal macrophages were left untreated (-) or stimulated (+) with IL-4 (20 ng/mL) for 4, 8, or 24 h. Whole-cell lysates were separated by SDS-PAGE, transferred to a PVDF membrane, and probed with antibodies specific for PPAR-γ, phospho-Akt (S473), total Akt, and β-actin. Band intensity was calculated by densitometry using ImageJ and the two groups were compared. A representative blot is shown for each time point. Data in bar graphs (a)–(c) represents the compilation of the relative intensities from three independent experiments with each symbol representing the mean value calculated from these individual experiments. * p < 0.05. ** p < 0.01. *** p < 0.001.

Since lung epithelial cells express IRS2, which has been shown to regulate epithelial cell repair, survival, and migration,^52–55 we carefully re-examined lung sections from PR8-infected mice, previously scored for histopathology in Figure 1(d), at high magnification to assess airway epithelial integrity. We found that in the IRS2^+/+ mice infected with a low dose of PR8 (LD₁₀), approximately 98.5% of airways demonstrate intact epithelial layers, while in marked contrast, in the PR8-infected IRS2^-/- mice, most (∼96%) of the airways were denuded of epithelial cells; representative low power micrographs of IRS2^+/+ and IRS2^-/- lungs at day 5 post-infection are shown in Figure 8. Overall, these results suggest that in the absence of IRS2, airway epithelial cells are especially sensitive to influenza-induced damage, resulting in increased susceptibility to influenza infection.

Figure 8.

IRS2^-/- infected lungs demonstrate enhanced denuding of airway epithelia. Lung sections of IRS2^+/+ and IRS2^-/- sections from mice described in Figure 2 were examined under high power to evaluate lung epithelial cell damage. Representative images are shown. Lower magnification images to show the overall tissue pathology are presented in the left panels, with higher magnification images in the right panels. Arrows point to airways. Asterisk in lower magnification of IRS2^-/- mice shows a partially denuded airway.

Methods

Mice

IRS2^-/- global knockouts (B6; 129-Irs2 ^tm1Mfw /J; 004421 JAX) on the C57BL/6J background were generated by heterozygote matings as previously described.^34,42 C57BL/6 IRS2^fl/fl mice, were obtained from Dr. Morris White (Harvard Medical School) and crossed with C57BL/6J LysMCre^+/+ (B6.129P2-Lyz2 ^tm1(cre)Ifo /J; 004781 JAX) to generate the C57BL/6 IRS2^fl/flLysM-Cre^+/- line as described. ⁵¹ The mice were bred at the University of Maryland School of Medicine by brother sister mating. Heterozygous mice are bred and genotyped to determine WT littermate controls and IRS2^-/- mice. Mouse genotypes from tail biopsies were determined using real time PCR with specific probes designed for each gene by Transnetyx (Cordova, TN). Both male and female littermates were randomized to treatment groups in all studies. At the end of each study, all mice were humanely euthanized. CO₂ was dispensed with a fixed presseure regulator and inline restrictor controlling gas flow within 30-70% of the chamber volume per minute. CO₂ flow was maintained for over 60 seconds following respiratory arrest, followed by cervical dislocation or bilatoral thoracotomy to assure euthanasia. All animal husbandry and experimental procedures were approved by and complied with the guidelines issued by the Institutional Animal Care and Use Committee at the University of Maryland School of Medicine.

Virus and in vivo challenges

Mouse-adapted H1N1 influenza A/PR/8/34 virus (“PR8”) (ATCC, Manassas, VA) was grown in the allantoic fluid of 10-day old embryonated chicken eggs as described ⁵⁶ and was kindly provided by Dr. Donna Farber (Columbia University). For inoculation, mice are first lightly anesthetized with inhaled isoflurane (to effect in dessiccator jar) saturated on a sterile guaze pad under a separating platform in a jar. The mice are anesthetized in approximately 10 seconds. A sterile pipette tip is used for each inoculation.

For survival experiments, IRS2^+/+, IRS2^+/-, IRS2^-/- global knockouts, and LysMCre-mediated conditional knockout of IRS2 (IRS2^fl/fl Cre^-/+, IRS2^fl/fl Cre^-/-) were infected with mouse-adapted influenza virus, strain A/PR/8/34 (PR8; ∼1500 TCID₅₀ i.n., 25 μl/nares), a dose of PR8 that kills ∼10% of infected WT C57BL/6J mice ⁴⁰ and monitored 14 days post-infection for survival. Additional groups of IRS2^+/+, IRS2^+/-, IRS2^-/- global knockout mice were infected in the same way, and on day 5 p.i., peripheral blood oxygen saturation (SpO₂) ¹¹ was measured using the PulseOx System (Starr Life Sciences). Hair was removed from the neck of mice and the oxygen sensor collar was placed around the neck. Readings were acquired over several minutes and an average value was calculated by the PulseOx software. These same mice were euthanized on day 5, and their lungs were harvested for histology, gene expression, blood glucose and protein levels.

In vitro Mϕ studies

Inflammatory peritoneal Mϕs were elicited from global IRS2^+/+ and IRS2^-/- mice by i.p. injection of sterile 3% thioglycolate. Cells were isolated 4 days later and were cultured as described previously. ⁵⁷ Mϕs were stimulated with medium alone or recombinant IL-4 (R&D Systems, 20 ng/mL) for 4, 8, or 24 h. Lysates were harvested in TriPure (Roche Diagnostics, North America, Indianapolis, IN, USA) for gene expression studies and lysis buffer (20 mM HEPES, pH 6.8, 1.0% Triton X-100, 0.1% SDS, 150 mM NaCl, 10 mM sodium fluoride (NaF), 1 mM sodium orthovanadate and 1 mM phenylmethylsulfonyl fluoride (PMSF) for protein analysis and densitometry by Western Blot.

Histology and staining

Lungs were inflated and perfused and fixed with 4% PFA. Paraffin-embedded sections (5 μm) of lungs were stained with hematoxylin and eosin (H&E). Slides were randomized and scored for four inflammatory parameters by an investigator blinded to the experimental groups. The scores were from 0 to 4 for each section based on the severity and distribution of the following lesions: peribronchiolitis (inflammatory cells, primarily lymphocytes, surrounding a bronchiole), perivasculitis (inflammatory cells, primarily lymphocytes, surrounding a blood vessel), alveolitis (inflammatory cells within alveolar spaces), and interstitial pneumonitis (increased thickness of alveolar walls associated with inflammatory cells). Data is shown as the average of the sums of the four parameters measured. ⁴¹ Quantification of airways that were epithelialized or not were carried out by counting the airways on low power magnification in 37 low power images forIRS2^+/+ and IRS2^-/- mice derived from multiple independent experiments.

Western blots

Western blots were prepared using Mini-PROTEAN® TGX™ Precast Protein Gels (Bio-Rad #4561046 and #4561034). Sample lysates were diluted with 2x Laemmli Sample Buffer (Bio-Rad #1610737) containing 5% (v/v) 2-mercaptoethanol, then heated at 95°C for 5 minutes before being cooled to 4°C. The molecular weight markers used were Precision Plus Protein Dual Color Standards (Bio-Rad #1610374). The running buffer consisted of Tris/Glycine/SDS Buffer (Bio-Rad #1610772) diluted with distilled water to concentrations of 25mM Tris, 192 mM glycine, and 0.1% (w/v) SDS. The transfer buffer contained Tris/Glycine Buffer (Bio-Rad #1610771) diluted with distilled water and methanol to concentrations of 25 mM Tris, 192 mM Glycine, and 20% (v/v) methanol. The wash buffer contained Tris-Buffered Saline (Bio-Rad #1706435) diluted with distilled water to concentrations of 20 mM Tris and 500 mM sodium chloride with an added 0.1% (v/v) Tween-20. Following transfer to PVDF, membranes were blocked for 1 hr at room temperature in a solution containing 5% (w/w) Blotting-Grade Blocker nonfat dry milk (Bio-Rad #1706404) dissolved in wash buffer.

Membranes were incubated with rocking overnight at 4°C with the primary antibody at 1:1,000 (v/v) in blocking solution. Primary antibodies were purchased from Cell Signaling Technology (IRS-2 #4502, Phospho-Akt [Ser473] #4060, PPAR-gamma #2443, Beta-Actin #4967), MyBioSource (Phospho-PPAR-gamma [Ser112] #MBS9402552) and Cayman Chemical (5-LO # 160402).

Membranes were then washed and incubated with the secondary antibody (Jackson ImmunoResearch Laboratories, Inc. #111-035-003) at 1:10,000 (v/v) in blocking solution at room temperature for 1 hr with rocking. Signal was developed using Pierce™ ECL Plus Western Blotting Substrate (Thermo Fisher Scientific #32132) with Hyperfilm™ ECL™ (Cytiva #28906838). For sequential re-probing of the membrane, Restore™ Western Blot Stripping Buffer (Thermo Fisher Scientific #21059) was used for stripping according to manufacturer’s instructions. Finally, densitometry was performed using ImageJ as previously described ⁵⁸ to quantify band intensities.

Quantitative real-time PCR (qRT-PCR)

Total RNA isolation and qRT-PCR were performed as previously described.^57,59 Levels of mRNA for specific genes were normalized to the level of the housekeeping gene, Hprt, in the same samples and are expressed as “fold-increase” over the relative gene expression measured in mock-infected lungs. For in vitro experiments involving Mϕs, results are expressed as “fold-increase” over the relative gene expression measured in untreated IRS2^+/+ cells.

HMGB1 protein levels in lung homogenates

The inferior right lobe of each mouse was removed and homogenized in 1 mL of viral buffer with an Omni homogenizer probe. The homogenized lung was centrifuged at 4500 rpm for 10 minutes at 4°C. The supernatant was transferred to newly labeled tubes and stored at -80°C until use. HMGB1 protein levels in lung homogenates were measured by ELISA according to the manufacturer’s protocol (IBL International; Catalog # ST51011; Toronto, Ontario, Canada).

Statistics

Statistical differences between two groups were determined using an unpaired, one-tailed Student’s t test with significance set at p < 0.05. For comparisons between ≥3 groups, analysis was done by one-way ANOVA followed by a Tukey’s multiple comparison post-hoc test with significance determined at p < 0.05. For survival studies, a Log-Rank (Mantel-Cox) test was used. If not indicated, differences between groups were not significantly different.

Discussion

IRS2 is a key adaptor protein that facilitates signal transduction from the insulin receptor and the IL-4 receptor to regulate gene expression and glucose transport.^{34,54,60–62} This report describes the striking finding that the IRS2 signaling adaptor limits the severity of the host response to influenza A PR8 infection, protecting mice from lethality. Global IRS2^-/- mice were highly susceptible to a sublethal dose of PR8 compared to IRS2^+/+ littermate controls with 100% of IRS2^-/- mice dying by day 9 after infection. IRS2^-/- mice showed significant increases in lung HMGB1 levels and histopathology, and a significant decrease in blood oxygen saturation levels measured before death (5 days p.i.), while the abundance of the influenza gene matrix protein-1 was unaffected at this time. Although IRS2 does not directly influence viral replication (as evidenced by expression of viral M1 RNA at Day 5 p.i.), like viral polymerase components or canonical host restriction factors,^63–66 it plays a critical indirect role in the host response to influenza infection by modulating immune cell function and tissue repair.

Host responses to influenza are known to be influenced by the M1/M2a Mϕ balance that regulates the proinflammatory versus tissue repair phases of the inflammatory response.^36,39 We observed significant increases in PR8-induced proinflammatory molecules in the lungs of IRS2^-/- mice (Il6, Ptgs2, Il1b, Tnfa), which can be produced by M1 Mϕs, with a concomitant decrease in tissue repair genes characteristic of M2a Mϕs (Arg1, Chil3, Retnla, Mrc1). We also observed reduced expression of 5-LO and phospho- and total PPARγ in lungs of infected IRS2^-/- mice, which has been shown to promote anti-inflammatory/tissue repair pathways in the lung during influenza and Respiratory Syncytial Virus infections.^32,67 Taken together, increased HMGB1, increased proinflammatory molecules, and decreased tissue repair mediators in the lungs of IRS2^-/- mice likely contribute to their enhanced sensitivity to influenza. However, the molecular mechanisms by which IRS2 mediates these effects is unclear.

One mechanism of IRS2 action may be to influence a PPARγ-5-LO regulatory loop. Gopalakrishnan et al. reported that agonists of PPARγ mitigated responses to PR8 infection. ³² PPARγ is a transcription factor known to bind to the promoter of the gene encoding 5-LO (Alox5) and increase its expression; 5-LO is an enzyme that produces both proinflammatory (leukotrienes) and anti-inflammatory (lipoxins) mediators.^68,69 Lipoxin A4 is a PPARγ ligand that enhances its DNA-binding function to regulate genes containing its response element including the Alox5 and Irs2 genes.^68–70 PPARγ agonist ligands have been shown to increase IRS2 expression in adipocytes and pancreatic β cells.^46,68,71 Thus, it is possible that the absence of IRS2 disrupts a positive regulatory loop that would normally enhance 5-LO production and PPARγ-gene regulating activity, including tissue repair, during influenza-induced inflammation. There are numerous regulatory serine/threonine kinase pathways that are activated downstream of the IRS2 adaptor that could be involved in such a regulatory pathway.^54,72,73 Cyclin-dependent kinase 5 (CDK5) and extracellular signal-regulated kinase (ERK) both can phosphorylate PPARγ, resulting in altered functional outcomes such as insulin sensitivity and inflammatory responses.^74,75

Because of the shift in M1/M2a-associated gene expression in the lungs of IRS2^-/- mice after influenza expression and the important role of myeloid-PPARγ in M2a differentiation and protection from influenza-induced lethality, ³² we expected the IRS2-mediated effect would be myeloid cell-intrinsic. In support of this expectation, basal phosphorylation of Akt was partially mediated by IRS2 in resting Mϕs in vitro, consistent with a previously characterized role for IRS2 in Akt activation in conjunction with IRS1. ⁴⁷ Although this in vitro data indicates that IRS2 can regulate signaling pathways within Mϕs, signaling studies comparing the IL-4-induced responses of inflammatory Mϕs isolated from IRS2^+/+ and IRS2^-/- mice in vitro showed response patterns that were markedly different from the in vivo lung responses to PR8 infection. Additionally, we observed that mice with myeloid-specific (LysM) deletion of IRS2^51,52 failed to exhibit significantly enhanced sensitivity to PR8 in multiple experiments. Overall, the in vivo results strongly suggest that non-myeloid cells are more likely to be central in the IRS2-dependent events that regulate the severity of host response to infection. Indeed, lung epithelial cells are the major target for influenza and release HMGB1 after infection^11,16,43 suggesting the hypothesis that IRS2 is modulating epithelial cell responses to infection. Because IRS2 is expressed in multiple cell types and acts downstream of IL-4, IL-13, insulin, and IGF1 receptors, the precise cell-based mechanism by which it acts to limit lung inflammation and remodeling remains unclear.

Lung airway epithelial cells (AEC) express IRS2 that has been shown to regulate epithelial cell survival, migration, and tissue repair. ⁵³ We observed extensive denuding of the airway epithelium, in addition to increased HMGB1 levels in lung homogenates of influenza infected IRS2^-/- mice, which was markedly greater than that observed in IRS2^+/+ mice. This suggests that AEC-intrinsic IRS2 may be important in epithelial cell survival and/or repair after influenza. IRS2 is a critical adaptor linking growth factor and cytokine receptors to phosphatidylinositol 3-kinase (PI3K).^54,76 PI3K activation then leads to the phosphorylation and activation of the protein kinase Akt. ⁷³ Activated Akt phosphorylates various downstream targets to inhibit apoptosis. It has been shown that decreasing IRS2 expression promotes apoptosis of pancreatic β cells, ⁶² while increasing its expression protects against fatty acid-induced cell death. ⁷⁷ IRS2 signaling also protects kidney epithelial podocytes from damage and cell death. ⁷⁸ Furthermore, IRS2 was recently shown to maintain specific airway basal stem cells (basal cell adhesion molecule hi cells) in chronic rhinosinusitis with nasal polyps. ⁵⁵ These finding suggest that IRS2 may be protective during influenza because it enhances AEC survival, resulting in a decrease in HMGB1 release. However, we did not observe any differences in the abundance of phospho- or total Akt in lung homogenates of IRS2^+/+ or IRS2^-/- mice. Interestingly, AEC also express PPARγ which has been shown to suppress their release of proinflammatory cytokines by interferring with NFκB. ⁶⁸ Thus, it is possible that AEC-IRS2 maintains PPARγ levels which could suppress inflammatory cytokine production in response to influenza.

A recent study defined cross-talk between alveolar Mϕs and AEC to activate a specific epithelial repair program during influenza A infection (PR8). ⁷⁹ This program was mediated by bone-marrow-derived, tissue resident alveolar Mϕs with an M2a-like phenotype producing the protein Placenta-expressed transcript 1 (Plet1). Plet 1 acts directly on AEC to induce lung repair by inducing proliferation of epithelial cells and re-sealing of the epithelial barrier. These Plet1⁺ Mϕ also expressed tissue regenerating growth factor mRNA including Pdgfa and b, Vegfb, and Igf1, a potent activator of the IRS-pathway. Thus, it is possible that AEC expression of IRS2 enhances their responsiveness to myeloid-cell derived growth factors to accelerate tissue repair after injury by the influenza infection. This would be consistent with previous observations of AEC-myeloid cell cooperation in models of allergic lung inflammation and helminth infection.^40,49,80

The currently available mouse models to study IRS2 have certain caveats. Global IRS2-deficiency from conception leads to the eventual loss of pancreatic β cell mass, insulin production, and elevation of blood glucose and type-2 diabetes.^42,52 In this study, IRS2^+/- mice, that do not develop overt diabetes, demonstrated intermediate sensitivity to influenza infection and there was no correlation between measured blood glucose levels at time of infection and drop in SpO₂. These results suggest that the observed sensitivity of IRS2^-/- mice to PR8 is not primarily due to elevated blood glucose. ⁴²

IRS2-flox lines with conditional deletion in myeloid cells have been developed to negate the high blood glucose problem found in global deficient mice.^51,52 These lines have been used to study obesity, diabetes, and cancer. However, the deletion of IRS2 is constitutive, rather than inducible, and these mice lack an indicator to track deletion efficiency at the single cell level. It is now clear that deletion efficiency should be evaluated to interpret experimental outcomes especially in complex organs containing multiple subpopulations of cells such as the lung.^81–83 While LysMCre has been used extensively to delete genes in myeloid cells and was recently shown to be the most specific deleter strain available for Mϕs, ⁸⁴ others have shown that not all myeloid or Mϕ subpopulations in the lungs demonstrate efficient Cre-mediated deletion ⁸¹ in response to specific inflammatory stimuli, such as parasite egg antigen. Overcoming these obstacles to definitively characterize the cellular and molecular role of IRS2 in limiting the deleterious host responses to influenza awaits the development of additional genetic models.