Influenza virus, an enveloped, negative-sense, single-stranded RNA virus, is an important human pathogen. Influenza infection predisposes the host to secondary bacterial infection. This superinfection is a clinically significant problem and a major cause of mortality and morbidity. Superinfection with Staphylococcus aureus following influenza leads to severe disease with approximately 41% mortality (Hageman et al., 2006). S. aureus is a Gram-positive bacterium estimated to be carried by 20% of the population (Kluytmans et al., 1997). The emergence and prevalence of MRSA (methicillin-resistant S. aureus) and VRSA (vancomycin-resistant S. aureus) have significantly increased the threat posed by these bacteria (Kobayashi et al., 2012).

Secondary bacterial infection occurs as the immune system is resolving the influenza-induced inflammation. While the induction of inflammation has been the subject of active research, the mechanisms underlying the resolution of inflammation have remained elusive. Induction of inflammatory response ensures successful pathogen clearance. Resolution of inflammation, on the other hand, returns the immune system to homeostasis thus avoiding excessive tissue damage (Serhan et al., 2015). Eicosanoids are bioactive lipids that play critical roles in both the induction and resolution of inflammation (Dennis and Norris, 2015). During microbial insult or cellular injuries, arachidonic acids and other related polyunsaturated fatty acids, like eicosapentaenoic acids (EPA) and docosahexaenoic acids (DHA), are metabolized via three major metabolic pathways, Cyclooxygenase (COX), Lipoxygenase (COX), and CYP450, to produce hundreds of lipid species with diverse physiological activities.

We have previously characterized the lipidomic landscape during the induction and resolution of inflammation in mice and human patients who were infected with influenza (Tam et al., 2013). Eicosanoids are bioactive lipids acting as signaling molecules that play a major role in both the induction and resolution of inflammation (Tam, 2013; Dennis and Norris, 2015). Eicosanoid metabolism pathways have provided highly successful targets for pharmaceutical interventions: non-steroidal anti-inflammatory drugs (NSAIDs) inhibit the cyclooxygenase pathway (COX) (Simmons et al., 2004), while asthma and COPD drugs inhibit the lipoxygenase pathway (LOX) (Scow et al., 2007; Bruno et al., 2018). Subsequently, we have applied systems biology approaches to define the transcriptional and lipidomic responses in a mouse model of influenza/S. aureus superinfection (Tam et al., 2020). We identified an anti-inflammatory eicosanoid response (CYP450 lipid mediators) that was highly induced during superinfection. CYP450 lipid mediators activate the nuclear receptor and transcription factor PPARα which can affect the regulatory networks of other transcription factors via protein-protein interactions. During influenza single infection, a moderate induction of CYP450 during the resolution phase which may allow for an appropriate anti-inflammatory response to promote the return to homeostasis. In contrast, during S. aureus single infection, a minimal level of CYP450 metabolites was produced. Therefore, transcription factors mediating pro-inflammatory signaling ensure successful pathogen clearance. However, excessive induction of CYP450 during superinfection leads to the suppression of innate immune response thus inhibiting efficient bacterial clearance. As S. aureus persists, the lipidomic response amplifies the infiltration of inflammatory cells, which eventually causes excessive tissue damage and increased mortality. Interestingly, excessive CYP450 lipid mediators have been observed in COVID patients with severe disease (Schwarz et al., 2021). The pathological production of these lipid mediators may dysregulate the physiological process of resolving inflammation and exacerbate morbidity and mortality during microbial infections.

Macrophages play an essential role in both immunity lipid homeostasis through their scavenger ability to phagocytose microbes or lipids in their resident tissues (Rigamonti et al., 2008). When macrophages are exposed to specific lipids, activated receptors can change the pathological states associated with the local environment (Rigamonti et al., 2008). Some of the common nuclear receptors (NR) include glucocorticoid receptors (GR) or estrogen receptors (ER), but macrophages also have a retinoid-x receptor (RXR) called the peroxisome proliferator activated receptor (PPAR) (Rigamonti et al., 2008). PPAR has three isoforms- PPARα, PPARδ/β, and PPARγ, which are ligand dependent transcription factors that bind to peroxisome proliferator response elements (PPRE) that are in enhancer sites of specific genes (Berger and Moller, 2002). Among immune cells, PPARα is specifically present and highly expressed in peripheral mononuclear immune cells like macrophages (Rakhshandehroo et al., 2010).

PPARα has been shown to play a critical role during microbial infections. In a mouse Mycobacterium abscessus infection model, Ppara ^–/– knockout mice show notably higher bacterial loads and increased cytokine expression of pro-inflammatory genes, including Il6, Il1b, and Cxcl10 (Kim et al., 2020). Furthermore, macrophage polarization profiles can be influenced by PPAR activation. Infection of macrophages with the obligate intracellular parasite, Trypanosoma cruzi, increased classically activated (M1) markers (e.g. NOS2) and increased proinflammatory cytokine signaling (Penas et al., 2015). Activation of PPARγ with 15dPGJ2 or PPARα with WY14643 showed increased Arginase-1 (M2 marker) and decreased pro-inflammatory cytokine expression (Penas et al., 2015). These studies demonstrate that activation of PPAR promote an anti-inflammatory phenotype.

While we demonstrated the role of CYP450 during superinfection in vivo, the mechanisms by which the lipid mediators affect the molecular signaling and cellular function on inflammation and bacterial clearance are not well understood. In this study we investigated the impact of CYP450-PPARα axis on the inflammatory signaling in macrophages since they are the dominant cell types in the broncho-alveolar lavage during superinfection. Using Nanostring nCounter Technology, we determined that the CYP450 lipid metabolites dampened the TLR inflammatory transcriptional responses in macrophages. We demonstrated that the activation of PPARα inhibits NFκB, hinders antibacterial activities and modulates macrophage polarization. Lastly, using liquid chromatography-Mass spectrometry (LC-MS), we determined the lipidomic profiles in wild type and Ppara ^–/– mice during superinfection. The increased eicosanoid metabolism in Ppara ^–/– mice may contribute to increased survival in during superinfection.

Influenza is an important human pathogen causing serious clinical complications. Prevalence and mortality vary greatly depending on the circulating seasonal or pandemic virus strain. Using a non-biased global systems biology approach and focusing on eicosanoids, we have identified a subset of lipid mediators that are produced during the resolution of inflammation (Tam et al., 2020). These eicosanoids are natural ligands for the nuclear receptors/transcription factors PPAR, which play critical roles in regulating macrophage polarization, inflammation, and lipid metabolism (Ng et al., 2007). To understand the mechanisms by which the CYP450-PPARα axis exacerbates mortality and morbidity during superinfection, we investigated how activation of PPARα affects the inflammatory signaling in macrophages. Using targeted transcriptional profiling with Nanostring nCounter Technology, we determined that CYP450 metabolites dampens the TLR signaling pathway via PPARα activation. We assessed the transcriptional activities of NFκB with lentiviral luciferase reporters. Activation of PPARα with CYP450 metabolite or synthetic ligand decreased NFκB activities. Moreover, activation of PPARα inhibits NFκB p65 phosphorylation and nuclear translocation.

PPAR has been shown to modulate the transcriptional networks by multiple mechanisms (Ricote and Glass, 2007; Pawlak et al., 2015). PPAR can directly interact with other transcription factors (i.e. p65) and prevent binding to NFκB response element or activating transcription. Activated PPAR can also induce IκBα which inhibits NFκB in the cytoplasm. Additionally, activated PPAR can regulate kinase activity, compete for coactivator complex or even inhibit co-repressor clearance.

The precise mechanism by which activated PPARα inhibits NFκB will be elucidated in future studies. We used both poly:IC and LPS to highlight that regardless of signaling through TLR3-TRIF or TLR4-TRIF and MyD88, the effects of CYP450 lipid-PPARα axis acts further downstream of the signaling cascade which may be significant due to the difference in PAMPs detected in multi-pathogenic interactions with the host immunity in superinfection.

High levels of CYP450 lipid mediators were detected in influenza superinfection as well as in serum from patients with severe COVID disease (Schwarz et al., 2021). Besides hindering the anti-bacterial activities, CYP450 lipid mediators decimated the anti-viral response, including numerous Type I interferon response and regulated genes (Ifnb1, Mx1, Mx2, Oasl1, Ifit1, Ifit2, Ifit3) ( Figures 1A, B ). The blunted anti-viral response may allow the unrestrained replications of viral pathogens which lead to an eventual cytokine storm (Ragab et al., 2020). Concordantly, immunosuppressed adaptive and innate immune cells and early-stage immune suppression have been observed from COVID patients (Remy et al., 2020; Tian et al., 2020). In contrast, the production of CYP450 lipid mediators occur during the resolution phase of influenza (7-10 days post infection), the dampening of antiviral response would allow for the return to homeostasis while adaptive immunity continues to eradicate the virus.

While we observed a persistence of S. aureus in the lungs of influenza-infected mice (Tam et al., 2020), using an in vitro system, we determined that the anti-bacterial function of macrophages was greatly hindered due to PPARα activation ( Figure 3 ). By utilizing CFU assays and digital microscopy, we were able to quantify a diminished phagocytic capability during PPAR activation. Digital microscopy using fluorescent markers to distinguish killed bacteria further showed that bacterial killing functions were also hampered during infection. Macrophages greatly affect the mediation of inflammation from the onset to resolution of infection. Interestingly, activation of PPARα is protective against Mycobacterium tuberculosis and Pseudomonas aeruginosa infections (Kim et al., 2017; Gugliandolo et al., 2019). While Ppara ^–/– mice have increased cytokine expression of pro-inflammatory genes during bacterial infection, as similarly observed during influenza/S. aureus superinfection, Ppara ^–/– mice have increased bacterial loads during M. tuberculosis (MTB) and P. aeruginosa infections. While MTB and P. aeruginosa are restricted and controlled by autophagy, S. aureus induces and exploits autophagy for its survival and growth (Schnaith et al., 2007; Huang and Brumell, 2014). Since PPARα plays a role in modulating autophagy (Jiao et al., 2014; Lee et al., 2014), the difference in phenotypes in Ppara ^–/– mice may be explained by the pathogenic mechanisms of the pathogens.

Since the elucidation of a specific transcription factor (GATA-3) promoting CD4 Th2 subset differentiation, mirroring M1 and M2 macrophage polarization have been defined to play diverse roles in health and disease (Zheng and Flavell, 1997; Mantovani et al., 2004). During superinfection, we have determined that there was a significant increase in alternatively activated M2 macrophages during the resolution of superinfection ( Figure 4A ). Using an in vitro polarization to the M2b subtype, enhancement of M2b (induced by immune complex/TLR ligand) polarization by PPARα agonist is intriguing because generation of anti-influenza IgG antibodies begins around day 7 post primary infection (Doherty et al., 2006). Increased polarization to M2b may play a critical role for effective dampening of the immune response via IL10 to prevent tissue damage. During influenza infection, the production of anti-influenza antibodies occurs during the resolution phase of inflammation which is also when the host is vulnerable to superinfection ( Figure 5 ). While influenza infection has been shown to promote alveolar macrophages to M1/M2b phenotype (Zhao et al., 2014), M2 macrophages have been shown to play diverse roles during superinfection: STAT2 deficiency increased M1, M2, and M1/M2 macrophages which promoted bacterial clearance (Gopal et al., 2018); SHP2 deficiency increased M2 macrophages and hindered antibacterial activities (Ouyang et al., 2020).

The effects of CYP450-PPARα axis on macrophage polarization, anti-bacterial activities, and abilities to recruit other immune cells (particularly neutrophils), may be exploited by the bacterial pathogens during superinfection ( Supplementary Figure 4 ). When PPARα is activated during superinfection, macrophages have reduced immune response involving cytokines, such as Cxcl3 ( Figure 1 ), and eicosanoids, such as LTB4 ( Figure 4C ). These signaling molecules play critical roles in recruitment of neutrophils (Zhang et al., 2001; Afonso et al., 2012; Wu et al., 2012; Lämmermann et al., 2013). While neutrophils are recruited in knockout mice during superinfection, the dampening of Cxcl3 and LTB4 correlates with decreased infiltration into the lungs of wildtype mice. This is problematic for the host immune system that has decreased macrophage functions in both immune response and function (namely phagocytosis and bacterial killing). Neutrophils and macrophages both play an essential role during influenza pneumonia infections as the main immune cell infiltrates during infection (Rudd et al., 2019). When uncontrolled recruitment and activation of neutrophils occur during influenza infection, the exacerbated outcomes are altered due to acute lung injury caused by excessive infiltration and generation of NETs (Rudd et al., 2019). However, the regulatory network to prevent exacerbated immune response may benefit the opportunistic secondary bacterial infection. Besides LTB4, other eicosanoids, such as PGE2, which has both pro- and anti-inflammatory activities (Sander et al., 2017), and 13-HODE (anti-inflammatory lipid mediator) (Vangaveti et al., 2010), were significantly increased during superinfection in Ppara ^–/– mice. Moreover, while 14, 15 EET (precursor to 14,15 DHET) was produced at similar levels, 14,15 DHET was produced at a significantly higher level in Ppara^–/– mice, compared to wild type. The increased lipidomic response of Ppara^–/– mice may be due to the lack of anti-inflammatory signals via CYP450-PPARα axis. The increased lipid mediators produced in Ppara ^–/– mice suggests that the CYP450-PPARα axis limits the eicosanoid metabolism. This negative feedback loop may be signaled via the transcriptional changes, immune cell activation status, or indirectly via altered fatty acid metabolism which generates the precursors for eicosanoids. The precise mechanism by which PPARα affects eicosanoid metabolism will be investigated in future studies.

The shift in anti-bacterial activities, macrophage polarization, and lipidomic responses may explain the difference in mortality and morbidity during superinfection, where Ppara ^–/– mice are partially protected. When individuals are infected with influenza, resolution of inflammation occurs approximately around 7-10 days ( Figure 5 ). The activation of PPARα promotes a systemic dampening of inflammatory response, propagating a cascade that affects macrophage function and immune cell recruitment to assure successful resolution. The resultant dampening of the immune response during resolution is sometimes exploited by a secondary bacterial infection post-influenza. Due to this anti-inflammatory immune environment, the macrophages are not activated to properly respond to the infection, nor recruit the proper immune cell response (i.e., neutrophils). The failure to control the bacterial infection ultimately leads to an immune shock that will severely impact the pathology of the lung, possibly causing death ( Figure 5 ).

PPARα has been demonstrated in other infection and disease models to exert an anti-inflammatory or repair activity. For example, activation of PPARα can reduce the inflammatory effects of LPS-induced acute lung injury by ameliorating vascular leakage and release of cytokine and eicosanoids into the alveolar space (Schaefer et al., 2008). Activation of PPARα can also restore the mitochondrial structure and promote gut epithelial repair during SIV infection in a nonhuman primate (Crakes et al., 2019). Moreover, PPARα is a master regulator for lipid metabolism (Bougarne et al., 2018). The activation of PPARα may modulate the metabolism which alters the immune function and activity of the cells. Surprisingly, we observed a decreased NFκB activity during TLR stimulation and an absence of increased bacterial killing in Ppara ^–/– macrophages. The lack of PPARα in the Ppara ^–/– macrophages may alter the threshold for immune response. The mechanisms by which CYP-PPARα affect the immunometabolism will be elucidated in future studies. Interestingly, while fenofibrates (PPARα agonist) are prescribed for abnormal blood lipid levels, infection and pneumonia are cited as possible adverse side effects (FDA, 2008; FDA, 2010) and fenofibrate has been associated with higher risk of mortality in MTB infections (Liu et al., 2020). In contrast, in a mouse influenza model, administration of oseltamivir (antiviral) and fibrates (PPARα agonist) prolong survival time during lethal H7N9 infection (Xu et al., 2015). Similarly, administration of gemfibrozil (fibrates) also increased survival during H2N2 infection (Budd et al., 2007). From these data, fibrates had been proposed as an inexpensive treatment against severe influenza infections or influenza pandemic (Fedson, 2009). Our data may explain the seemingly contradictory concepts. PPARα agonists may be useful for successful resolution during influenza infection ( Figure 5 ). However, during superinfection, the pathological production of DHET excessively activates PPARα, compromising the immune system’s ability to control secondary bacterial infections ( Figure 3 ). If fibrates therapy drastically increase the activation of PPAR, patients may be at risk for superinfection. However, the use of fibrates is confounded by the fact that it also inhibits CYP2C enzymes (Gong et al., 2016) which may prevent dangerous levels of DHET. Furthermore, data is not available to assess PPAR activation in the lungs of patients under fibrates therapy. And the level of PPAR activation may not be increased at a pathological level for bacteria to cause opportunistic infections.

The complexity of the superinfection stems from a triad of interacting players: the host immune response, influenza virus, and the bacterium. Successful pathogen clearance and resolution of inflammation ensure physiological return to homeostasis. In contrast, dysregulated functions resulting from aberrant PPAR activation will hinder pathogen control and eventually amplify inflammation. Combining systems approaches, targeted molecular methods, and high throughput cell imaging, we have determined how the CYP450-PPARα axis potentiates the increase of morbidity and mortality during superinfection. Understanding the molecular mechanism by which resolution of inflammation affects our immune response will yield therapeutic targets for sever microbial infections and inflammation-mediated diseases.

The data presented in the study are deposited in the MetaboLights repository, accession number MTBLS2927.

The animal study was reviewed and approved by IACUC Temple University.

VT, RL, BB, and JB contributed to conception and design of the study. RL, JT, and NG-E performed the experiments. BB provided expertise on confocal microscopy. JB performed mass spectrometry analysis. RL and VT wrote the first draft of the manuscript. JB and BB wrote sections of the manuscript. All authors contributed to manuscript revision, read, and approved the submitted version.

This work was supported by National Institute of Allergy and Infectious Disease R21 AI142278 and R01 AI168550 (VT).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.