High expression of oleoyl-ACP hydrolase underpins life-threatening respiratory viral diseases

inﬁltration, and inﬂammation. This was underpinned by differential lipid droplet dynamics as well as reduced viral replication and virus-induced inﬂammation in macrophages. Supplementation of oleic acid, the main product of OLAH, increased inﬂuenza replication in macrophages and their inﬂammatory potential. Our ﬁndings deﬁne how the expression of OLAH drives life-threatening viral disease

disease. 4 Hospital admissions can also occur in previously healthy individuals with no prior risk factors.Among patients hospitalized with influenza-like illness during 2015-2016, >60% were previously healthy with no underlying conditions. 5Factors driving severe influenza disease are of key importance if we are to design effective vaccines and therapies to control respiratory disease.
Increased susceptibility to severe influenza disease is multifactorial, and apart from impaired or overactive immune responses and/or minimal preexisting immunity, [6][7][8][9] host factors contribute to disease outcomes.Interferon-induced transmembrane protein-3 (IFITM3) can restrict virus replication by preventing endocytosed virus from entering the cytoplasm.1][12] Mutations of genes associated with Toll-like receptor 3 13 and regulatory elements mediating interferon type I and type III production 14,15 increase susceptibility to severe influenza.Regulation of host lipid factors and their metabolism also impact influenza virus replication, inflammation, and disease outcome.][19] Following the 2013 outbreak of the avian A(H7N9) influenza virus in China, with $35% mortality rates, we defined immunity in fatal A(H7N9) disease. 11,20,21Patients hospitalized with fatal A(H7N9) had dysregulated immune responses and hypercytokinemia in contrast to patients who recovered. 11,20Here, we investigated mechanisms associated with hypercytokinemia and fatal disease.We identified a gene encoding for an enzyme involved in endogenous fatty acid production, oleoyl-acyl-carrier-protein (ACP) hydrolase (OLAH), as an early driver of fatal outcomes during A(H7N9).OLAH levels were associated with life-threatening seasonal influenza virus, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), respiratory syncytial virus (RSV) infection, and multisystem inflammatory syndrome in children (MIS-C).OLAH regulates lipogenesis by catalyzing the release of fatty acids from an acyl-carrier protein following carbon chain elongation within the fatty acid synthase complex. 22While OLAH liberates fatty acids of varying chain lengths, it has a high catalytic efficiency for oleoyl-ACP 23 to produce oleic acid.
To date, the role of OLAH in modulating immunity and driving disease outcomes has not been investigated for any disease.We found that OLAH gene expression in A(H7N9) patients with fatal disease is substantially elevated during early disease and remains high until death.High OLAH levels were found in four additional cohorts of patients hospitalized with life-threatening seasonal influenza, COVID-19, RSV infection, and in pediatric patients hospitalized with MIS-C.A CRISPR-Cas9-mediated gene knockout of olah in mice unraveled mechanisms underpinning OLAH-mediated disease severity.Infection of olah À/À mice with a lethal dose of influenza virus led to survival, mild disease, reduced lung viral loads, tissue damage, infection-driven pulmonary innate cell infiltration, and inflammation.This was associated with differential lipid droplet dynamics, reduced viral infection in macrophages, and inflammatory milieu.Inhibition of lipid droplet formation reduced viral infection in macrophages, while supplementation of oleic acid, OLAH's main product, increased influenza virus infection in macrophages and inflammation.Our findings provide mechanistic insights into how the expression of OLAH drives life-threatening respiratory disease.

RESULTS
High expression levels of OLAH during early avian A(H7N9) influenza predict fatal disease outcomes As avian influenza A(H7N9) virus leads to severe influenza disease with $35% morbidity rates, 24,25 we investigated early biomarkers of fatal outcomes.To identify molecular signatures associated with fatal disease, we performed blood transcriptomics using our cohort of hospitalized H7N9 patients at Shanghai Public Health Clinical Centre. 11,20Affymetrix Human Gene ST-2.0 arrays were performed on 8 H7N9 patients, 4 who recovered (mean age of 69) and 4 with fatal disease (mean age of 70) (Table S1).Patients who recovered (a9, a20, a73, and a134) were discharged within 14-23 days after disease onset, while fatal outcomes occurred at 19 days (patient a118) or after prolonged hospital stays on days 64, 70, and 76 (patients a131, a33, and a22) (Figure 1A).Plasma cytokine levels at 1-2 days following hospital admission showed that patients who died had higher inflammation compared with patients who recovered (p = 0.028; Figure 1B; data for the entire cohort in Wang et al. 11 ), indicative of early hypercytokinemia preceding severe outcomes.Transcriptomics, performed on patients' blood collected within 6 days of hospital admission, identified differential expression of 10 genes defined by 16 probe sets, with 7 probe sets specific for one gene, OLAH, expressed at $82-fold higher levels in patients who died compared with patients who recovered (p < 0.001; Figure 1C).High OLAH levels in fatal H7N9 patients were detected early after hospital admission and persisted until patients died, while patients who recovered exhibited low OLAH expression throughout hospital stays (Figure 1D).These analyses defined OLAH as the main differentially expressed gene during fatal A(H7N9) disease.
High OLAH expression is linked to life-threatening seasonal influenza, SARS-CoV-2, and RSV infections Given scarce data on OLAH expression in humans, we analyzed mRNA levels across tissues and peripheral blood mononuclear cells (PBMCs) in healthy adults.High OLAH levels were found in lungs, spleens, T cells, monocytes, and remaining PBMC populations (Figure 1E).Detectable OLAH levels were detected in thymus and tonsil tissues.Within lungs, the site of influenza virus infection, OLAH expression was predominantly detected in CD14 + monocytes (Figure 1F).
To investigate OLAH expression in patients hospitalized with seasonal influenza, we utilized a patient cohort with single-cell RNA sequencing (scRNA-seq) data performed using PBMCs collected at 1-14 days after hospital admission. 26This cohort consisted of 2 healthy participants and 3 intensive care unit (ICU) patients hospitalized with influenza A(H1N1) or influenza B virus requiring mechanical ventilation.Our findings showed elevated OLAH expression in ICU influenza patients but not among healthy individuals (Figure 1G), predominantly within monocytes/macrophages.Accordingly, in vitro infection of PBMCs from healthy individuals with A/California/07/2009 led to increased intracellular nucleoprotein (NP) levels in monocytes, demonstrating their susceptibility to influenza virus infection (Figure 1H).
To determine whether the observed association between OLAH levels and acute viral infection-related illness severity extended beyond influenza, we evaluated gene expression profiles in patients hospitalized with COVID-19 or pediatric acute respiratory disease (pARDS).Bulk RNA-seq analysis of blood from infants, children, or young adults (aged 0-21) with SARS-CoV-2 (Tables S2 and S3) identified OLAH as the single most differentially expressed gene increased in patients hospitalized with life-threatening respiratory dysfunction compared with those hospitalized with no to minimal respiratory dysfunction (Figure 2A).When compared with healthy individuals, OLAH expression was minimally elevated in hospitalized COVID-19 patients with no to minimal respiratory dysfunction (Figure 2A; p = 0.08).Conversely, increased OLAH expression was detected in patients with moderate-to-severe (p = 0.01) and life-threatening respiratory dysfunction (p < 0.001).Patients hospitalized with MIS-C, a post-infection severe hyperinflammatory complication temporally related to prior SARS-CoV-2 infection, showed even further elevation in OLAH levels compared with patients with moderate-to-severe disease (p < 0.05).We found no significant associations between OLAH and BMI, age, or sex (Figures S1A-S1D).Visualizing OLAH expression as a function of time (days between symptom onset and sample collection) demonstrated an increase in OLAH expression from early time points after infection ($day 5) among patients with severe disease, with a peak in OLAH at $9 days after symptom onset, conversely to minimal OLAH expression in patients with minimal disease severity (Figure 2B).
To investigate downstream consequences of differential OLAH expression in patients with severe respiratory disease, we performed a priori analysis of oleic acid abundance in an independent cohort of patients diagnosed with COVID-19 (Table S4).HPLC was used to assay plasma lipid abundances.Oleic acid (C18:1) was identified based on retention time, m/z, and fragment data.After controlling for potential effects of sex, age, and race, oleic acid abundance was significantly higher among hospitalized patients than those who remained ambulatory (p < 0.05; Figure 2C).
In our fourth disease cohort, we interrogated scRNA-seq data of tracheal aspirate samples obtained from children hospitalized with acute respiratory failure requiring endotracheal intubation secondary to known or suspected lower respiratory tract infection (LRTI) or for an etiology unrelated to lung injury (e.g., head injury neurologic failure; control patients). 28As with severe influenza, OLAH expression was elevated among monocytes/macrophages and neutrophils in patients with mild and moderate/severe pARDS, particularly in confirmed RSV infections (Figure 2D).When grouped by respiratory disease severity, OLAH expression was increased in proportion to the severity of pARDS experienced (Figure 2E).Elevated OLAH expression was detected in patients without pARDS who were intubated due to severe disease caused by an LRTI and acute lung failure.OLAH levels in patients without LRTI and control patients intubated because of neurologic failure without lung disease were negligible, indicating a strong link between OLAH expression and respiratory viral infection but not to unrelated non-respiratory injuries.To understand OLAH expression among healthy individuals and those with mild respiratory infections, we analyzed blood microarray expression data from human challenge models of infections with H1N1 (2 studies), H3N2 (2 studies), human rhinovirus virus (HRV) (2 studies), and RSV (1 study) spanning 38 h prior to infection to 170 h post-infection. 29We found no differences in OLAH expression following sham infection and mild infection across these datasets (Figures 2F and S1E).Significant effects of time on OLAH expression were observed in three of human challenge studies, but these slight changes in OLAH were consistent regardless of infection status, suggesting technical rather than biological variation.OLAH expression from human samples obtained up to 7 days after challenge exhibited no differences between sham and mild infection.Additionally, bulk RNA-seq from nasal curettage cells obtained after controlled human RSV infection 30 showed no variation in OLAH expression across participants who became actively infected with or without symptoms or those not exhibiting active viral replication (Figure S1F).
Overall, our data introduce the concept that highly elevated OLAH expression in adults and children underpins severity of human viral infections caused by influenza viruses, RSV and SARS-CoV-2, and life-threatening post-infectious hyperinflammatory MIS-C.The consistently low OLAH levels observed among healthy individuals and during mild infections from human challenge suggest that OLAH is specific to severe disease and merits further investigation as a potential target for novel therapeutics.
OLAH-mediated lipogenesis is attenuated in olah À/À mice As our findings linked OLAH expression with severe influenza disease in humans, we investigated OLAH's role in driving severe influenza in a mouse model.We analyzed olah mRNA expression across organs and tissues in naive wild-type (WT) C57BL/6 mice.The highest levels of olah mRNA were detected in testis, placenta, brain, lung, and thymus, and, to a lesser extent, in liver, spleen, and lymph nodes (Figure 3A).Within lungs, higher expression of olah was detected in non-hematopoietic CD45 À cells compared with hematopoietic CD45 + immune cells (p = 0.014; Figure 3B).To generate olah-deficient (olah À/À ) C57BL/6 mice, we performed CRISPR-Cas9-targeted deletion of exons 1 and 2 of the olah allele on chromosome 2 (Figure 3C).Successful deletion was confirmed by undetectable expression of the olah gene (Figure 3C).Comparative histological analysis of 28 tissue and organ sections from naive olah À/À and WT mice did not reveal any differences aside from early indications of urethral obstructive syndrome and mild hyperplasia in spleens of olah À/À mice (Table S5).Both olah À/À and WT mice were similar in outward appearance, body weights (Figure S2A), and immune cell frequencies within the lung, spleen, and thymus (Figure S2B).
Lipidomic analysis of lung tissues from naive olah À/À and WT mice showed differences in abundance of cardiolipins and lysophosphatidylethanolamines at the class level (Figure S2C).At the species level, 69 distinct lipids were decreased in olah À/À mice; 13 were upregulated (Figures 3D and 3E).Of those downregulated, $25% were cardiolipins, containing abundant OLAH catalytic products and their derivatives (Figure 3F). 31,32Additionally, $37% were lysophosphatidylethanolamines, phospholipids implicated in intracellular signaling and innate response induction. 33xamining fatty acids produced by OLAH, we detected reduced amounts of oleic (p = 0.033), palmitoleic (p = 0.035), palmitic (p = 0.019), and myristic acid (p % 0.001) in olah À/À mice, while levels of stearic acid and lauric acid, the latter a fatty acid not produced by OLAH, remained unaltered (Figure 3G).The lack of difference in stearic acid levels between WT and olah À/À mice may be attributed to the preferential cleavage of oleoyl-ACP intermediates by OLAH at the expense of stearic acid production.Our data indicate that the genetic deletion of olah results in functional impacts on OLAH-mediated lipogenesis, shown by decreased fatty acid synthesis and levels of related lipid species.

Olah deficiency protects from severe influenza
To determine the impact of OLAH on influenza outcomes, we infected olah À/À and WT mice intranasally (i.n.) with 2 different doses of A/HKx31 (H3N2; X31) influenza strain corresponding to lethal (10 5 plaque-forming units [PFU]) and non-lethal (10 4 PFU) infection (Figure 4A).To elucidate effects of OLAH on disease severity, we examined the body weight loss and survival Figure 2. OLAH expression is elevated during severe respiratory infection and MIS-C but not among healthy participants or mild infections (A) Bulk RNA-seq was performed on blood from hospitalized COVID-19 patients <21 years (n = 143).MD plot of differentially expressed genes between hospitalized patients with no/minimal respiratory dysfunction (blue) and hospitalized patients experiencing life-threatening fatal respiratory failure (red).OLAH expression in healthy, non-infected individuals (n = 22) and hospitalized SARS-CoV-2 patients grouped by disease severity (hospitalized with no/minimal respiratory dysfunction, n = 43; hospitalized with moderate/severe respiratory dysfunction, n = 25; hospitalized with life-threatening respiratory dysfunction, n = 23; and hospitalized/diagnosed with MIS-C 27 (n = 30).p values were obtained from a model controlling for days since symptoms onset, sex, whether a patient was previously healthy, steroid administration prior to sampling, bacterial co-infection, age, race, and ethnicity; adjusted for multiple comparisons.(B) OLAH expression from (A) plotted across time.Curves were fit using the LOESS function with default parameters.(C) Log 2 oleic acid abundances from SARS-CoV-2-positive patients as outpatients (n = 38) or hospitalized (n = 35).Significance was assessed using analysis of variance while controlling for age, sex, and race.(D) OLAH expression in single-cell RNA transcriptomic data acquired from tracheal aspirate samples of RSV-infected children with no/mild (n = 5) or moderate/ severe pediatric ARDS (pARDS) (n = 7), non-RSV-related infection but with moderate/severe pARDS (n = 5), and control children without acute lower respiratory tract infection or lung injury (n = 6).(E) OLAH expression in children grouped by pARDS severity and lower respiratory tract infection (LRTI) status.(F) OLAH expression across time for four human challenge models of mild respiratory infections (H1N1-DEE3, n = 477; H3N2-DEE2, n = 355; HRV-Duke, n = 471; and RSV-DEE4, n = 420).Smooth curves are fit using a locally estimated scatterplot smoothing (LOESS) model, with red depicting samples from infected participants; blue showing samples from sham-infected participants.Linear mixed models were used to test for effects of time, infection status, and interaction of time with infection status, with participants included as a random effect.Infection status was not significant for any study.Time was statistically significant for studies H1N1-DEE3 (p < 0.001) and H3N2-DEE2 (p < 0.001).See also Figure S1. of olah À/À and WT mice following infection.Animals infected with the 10 4 PFU dose survived (Figure 4B); however, olah À/À mice lost significantly less body weight comparing to WT mice over 10 days (Figure 4C) (p < 0.05 on days 4, 6, and 8); confirmed by area under the curve (AUC) profiles (p = 0.009).Differences in disease outcomes were striking following infection with the lethal 10 5 PFU dose.While influenza-infected olah À/À mice survived, $50% of WT mice succumbed to disease (p = 0.021; referred to as LD 50 infection; Figure 4B).Infection in WT mice exhibited more body weight losses during disease, particularly at days 5-8 post-infection (p < 0.05); shown by cumulative AUC comparisons (p = 0.008; Figure 4C).Pooled data from 5 experiments (n = 42 WT; 28 olah À/À mice) showed that protection against lethality in olah À/À mice was not sex-specific, with survival rates of male and female olah À/À mice being equivalent (>93%), compared with $50% of sex-matched WT mice (males p = 0.005; females p = 0.016; Figure 4D).
Histopathological analysis of lung sections from LD 50 -infected WT mice indicated the presence of moderate-to-severe pulmonary pathology as early as day 1 post-infection (Figure 4E; Table S6).Observations included acute necrotizing bronchitis, respiratory epithelium disruption accompanied by multifocal lymphocytic infiltration of the interstitium and vasculitis, and areas of high-protein edema suggestive of viral pneumonia.Conversely, these were milder and less consistent in olah À/À mice, reflected in lower percentages of damage and lesions in tissue sections from olah ÀlÀ mice across 5 days post-infection (p = 0.013; days 3 and 5 p < 0.001; Figure 4F), indicating less severe pulmonary pathology in olah ÀlÀ mice.
Our results provide evidence that OLAH drives severe influenza disease in mice, both with respect to morbidity and mortality, supporting our observations in H7N9 patients linking high OLAH levels with fatal outcomes.
LD 50 infection of WT and olah ÀlÀ mice resulted in >74 differentially expressed lipid species in lungs on day 1 post-infection (Figure S2D).Higher amounts of phospholipids containing polyunsaturated fatty acids, known eicosanoid precursors, were upregulated in infected WT mice, including those associated with arachidonic acid (20:4), dihomo-g-linolenic acid (20:3), and adrenic acid (22:4) (Figure S2E), indicating a role for lipid mediators in driving OLAH-mediated inflammation during infection.Hence, we provide evidence that severe OLAH-mediated disease is underpinned by inflammatory responses reminiscent of early hypercytokinemia observed during fatal H7N9 infection in humans during early infection (Figure 1B).Without OLAH, the early onset of hypercytokinemia was averted.
To determine the role of immune and non-immune cells in driving OLAH-mediated disease, we generated bone marrow (BM) chimeras using olah ÀlÀ and WT mice.Adoptive transfer of donor WT BM cells into irradiated olah ÀlÀ recipients (WT / olah ÀlÀ ) provided a model with only hematopoietic-derived immune cells expressing olah.In the olah ÀlÀ / WT scenario, only non-immune cells express olah (Figures 5B and S4A).We also established controls where donor WT or olah ÀlÀ BM cells were transferred into the same recipient strains (WT / WT; olah ÀlÀ / olah ÀlÀ ).Following LD 50 infection, 90% of recipients with WT BM cells (WT / olah ÀlÀ and WT / WT) succumbed to disease within 7 days, whereas 60% of mice with olah ÀlÀ BM (olah ÀlÀ / WT and olah ÀlÀ / olah À/À ; p % 0.001) survived (Figure 5C).WT BM-recipients also experienced more body weight loss (from day 3, p < 0.05, Figure 5C), implicating that OLAH-expressing immune cells exacerbate disease severity.Analysis of donor-derived immune cells in lungs of influenza-infected recipients revealed lower numbers of olah ÀlÀ macrophages in olah ÀlÀ / WT mice compared with WT macrophages in WT / olah ÀlÀ mice (Figure 5D), linking olah expression in macrophages with disease outcomes.
Since lipid droplets play important roles in viral replication cytokine production, 17,35 and given the role of OLAH in lipogenesis, we investigated how OLAH affected lipid droplet formation in lungs following LD 50 infection.Although higher numbers of lipid droplets were initially detected in lungs from naive olah ÀlÀ (p < 0.001) (Figures 5E, 5F, and S4B), they decreased at day 1 post-infection (p < 0.001) and were reduced for at least 3 days.Conversely, lipid droplet accumulation in WT mice increased after infection comparing to olah ÀlÀ mice (days 1-3; p < 0.001).CD36, the fatty acid scavenger receptor, and IAV-NP co-localized with concentrated regions of lipid droplets (Figure 5E).Olah ÀlÀ mice exhibited reduced lung viral titers compared with WT mice, especially at day 1 post-infection (p < 0.001) (Figure 5G), demonstrating that expression of OLAH promotes virus-induced lipid droplet formation in infected lungs.

Epitope-specific CD8 + T cells are not affected by olah
To determine whether olah affected adaptive immunity, we analyzed influenza-specific T cells at day 10 post-LD 50 infection (Figure S5A).Equivalent numbers of CD8 + and CD4 + T cells were detected in lungs and BAL of infected WT and olah ÀlÀ mice (Figure S5B).Although the majority of CD8 + T cells exhibited effector phenotypes (CD62L À CD44 + ) (Figure S5C), their numbers did not differ between mice.The same occurred for T cell populations with naive (CD62L + CD44 À ) and memory (CD62L + CD44 + ) phenotypes.olah ÀlÀ and WT mice induced similar levels of influenza-specific CD8 + T cells directed at immunodominant D b NP 366 and D b PA 224 epitopes (Figure S5D), with similar capacity of CD8 + T cells from WT and olah ÀlÀ mice secreting cytokines (Figure S5E), indicating that OLAH does not impact T cell-mediated immunity.

Macrophages play an important role in driving OLAHmediated disease severity
As our data in humans and mice demonstrated that macrophages were associated with severe influenza, we investigated how olah expression impacted virus infection of macrophages.We analyzed mitochondrial abundance and membrane potential, and glucose uptake in WT and olah ÀlÀ macrophages in vivo following LD 50 infection using fluorescent metabolic dyes, Mito-Tracker, tetramethylrhodamine (TMRM), and 2-(7-nitro-2,1,3-benzoxadiazol-4-yl)-D-glucosamine (2-NBDG).Although mitochondrial abundance and glucose uptake increased over 3 days post-infection, olah expression had minimal effects on metabolic features (Figure S6A).Both WT and olah ÀlÀ macrophages exhibited similar M1-like phenotypes (iNOS high , Arg-1 low , CD206 low ) early after infection and acquired M2-like features (iNOS low , Arg-1 high , CD206 high ) at day 3 (Figure S6B), indicating unchanged phenotypic polarization.
We quantified viral mRNA encoding for NP and matrix (M) antigens in CD64 + F4/80 + peritoneal macrophages from WT and olah ÀlÀ mice (Figure 6A) after in vitro infection. 36In Madin Darby canine kidney (MDCK) cells, viral mRNA expression increased over time (Figure 6B).Although increases were observed in WT and olah ÀlÀ macrophages, expression levels were similar between groups, suggesting that OLAH may not modulate early stages of viral infection, viral entry, or viral mRNA production.
Increased viral mRNA in influenza-infected MDCKs was concomitant with intracellular expression of viral proteins and detected in $65%-70% of cells at 8 h post-infection (Figure 6C).Despite similar viral mRNA levels, fewer olah ÀlÀ macrophages expressed viral proteins compared with WT macrophages ($2-fold; NP: p = 0.022; M: p = 0.028).Reduced expression of viral proteins in olah ÀlÀ macrophages was confirmed by proteomic analysis (Table S7).Differences in antigen expression occurred from 4 h post-infection (Figure 6D), reflected in the amount of viral antigen expressed (Figure S6C), indicating that olah expression in macrophages increased viral antigen production in male-and femalederived macrophages (Figure 6E).Infection of WT macrophages resulted in production of higher levels of inflammatory mediators TNF, MIP-1a, and MIP-1b (Figure 6F), reminiscent of the profiles observed in lungs and BAL of infected WT mice (Figure 4G).
To understand whether olah's association with increased viral infection was limited to respiratory viruses, we investigated replication of West Nile Virus (WNV) and murine norovirus (MNV) in olah ÀlÀ macrophages.Viral titers of WNV, Kunjin virus (WNV KUN ) were significantly reduced ($2-fold) in culture supernatants of infected olah ÀlÀ compared with WT macrophages (p = 0.047; Figure 6G).No differences were observed following infection with MNV (Figure 6G), but this suggests that olah expression could also be involved in promoting flavivirus infections.

Oleic acid drives aberrant virus-induced responses in macrophages and exacerbates disease severity following influenza virus infection
As OLAH promoted virus-induced lipid droplets in lungs (Figures 5E and 5F), we analyzed lipid droplet usage in macrophages.Higher numbers of lipid droplets were initially detected in olah ÀlÀ compared with WT macrophages prior and during early infection (Figures 7A and S4C).However, while viral infection increased lipid droplets in olah ÀlÀ macrophages over 8 h by 3-fold, a more substantial increase of >20-fold occurred in WT macrophages.Thus, expression of olah in macrophages increased lipid droplet formation following viral infection, leading to elevated cytokine production.
Since oleic acid, the major catalytic product of OLAH-mediated lipogenesis, is increased in patients hospitalized with respiratory infections (Figure 2C), we investigated its impact on infection in vivo.We infected WT mice fed on an oleic acid-enriched diet (WT OA diet mice) and age-matched WT mice fed on a normal diet, with a non-lethal X31 dose.In WT OA diet mice, we detected higher lung viral titers and greater body weight loss, particularly during early infection (days 2-5; Figure 7B), demonstrating that increased dietary intake of oleic acid leads to more severe influenza.We examined how oleic acid treatment of macrophages affected viral protein production and cytokine responses following infection.Oleic acid treatment of infected olah ÀlÀ and WT macrophages increased NP + and M + cell frequencies compared with untreated macrophages (p < 0.001 for both NP and M) (Figures 7C  and S6D).Frequencies of viral protein-expressing cells in oleic acid-treated WT macrophages were higher than treated olah ÀlÀ macrophages (NP: p < 0.001; M: p = 0.036), presumably due to the presence of endogenous oleic acid.
Although oleic acid treatment increased cytokine production by uninfected WT and olah ÀlÀ macrophages (compared with untreated controls; olah ÀlÀ : p = 0.016; WT: p = 0.006), exposure of these cells to infection further increased cytokine responses (both p < 0.001) and again with higher levels in WT compared with olah ÀlÀ macrophages (p = 0.005; Figure 7D).In olah ÀlÀ macrophages, differences between infected and uninfected cells were not evident without oleic acid and only observed between infected and uninfected WT macrophages (p = 0.010).However, the addition of oleic acid to olah ÀlÀ macrophages overcame this indifference to heighten not only virus-induced inflammatory responses but also antigen production.
As oleic acid can inhibit anti-viral action of IFITMs, 37 we investigated the impact of oleic acid treatment on influenza-infected macrophages from ifitm3 À/À mice.Compared to influenza-infected WT macrophages, infection of ifitm3 À/À macrophages increased infection (8 h p % 0.001; 24 h, p % 0.003; Figure 7E). 38equency of infected macrophages between 8 and 24 h was similar, indicating that intracellular NP peaks early at 8 h postinfection.However, while oleic acid treatment of WT macrophages increased viral protein expression (8 h, p = 0.014; 24 h, p < 0.001), no significant effects were observed in ifitm3 À/À macrophages, implicating that inhibition of IFITM3 by oleic acid may contribute to OLAH-mediated infection in these cells.

DISCUSSION
Although respiratory viral infections cause profound disease, underlying mechanisms are unclear, especially why some individuals present with mild infections while others succumb to respiratory disease. 5When avian A(H7N9) emerged in humans with mortality of $35%, we revealed that elevated inflammatory cytokines/chemokines early in infection predicted fatal H7N9 outcomes.Here, we investigated mechanisms associated with hypercytokinemia and fatal influenza disease.We identified OLAH as an early driver of severe A(H7N9) disease, seasonal influenza, COVID-19, RSV, and MIS-C.We demonstrated that influenza virus infection in olah À/À mice protected from lethal disease, stemming from reduced lung viral titers, tissue damage, pulmonary hypercytokinemia, innate cell infiltration, and lipid droplet formation, associated with differential lipid droplet dynamics, reduced viral infection, and inflammation in macrophages.Oleic acid, the main product of OLAH, exacerbated influenza infection and inflammatory potential of macrophages.Oleic acid also systemically increased in patients hospitalized with moderate-to-severe COVID-19.Elevated OLAH levels  (legend continued on next page) were previously reported in children experiencing neurological symptoms and pneumonia during H1N1pdm2009 infection. 43hese findings provide mechanistic insights into how elevated OLAH drives immunopathology and severe disease outcomes following respiratory infections.
OLAH-mediated inflammatory responses and severe disease outcomes were attributed to elevated cellular levels of fatty acids produced.Our observations agree with reports showing that viral infection can be exacerbated by oleic or palmitic acids. 16,37,44,457][48][49] We demonstrate that exuberant pro-inflammatory cytokine responses observed during OLAH-mediated disease are associated with effects observed in influenza-infected macrophages.We found that olah expression did not affect susceptibility of macrophages to infection as viral mRNA remained unimpacted, suggesting that mechanisms facilitating virion attachment and internalization are not impacted by fatty acid availability.Viral protein levels were significantly reduced in olah ÀlÀ macrophages, indicating OLAH's role in promoting viral infection, suggesting a block in viral mRNA translation and/or in processes further upstream.As palmitoylation is a common modification of influenza virus glycoproteins and membrane proteins, 50 reduced amounts of fatty acids might affect the stability of viral proteins produced and their trafficking/ degradation. 51eutrophils and NK cells were also detected at higher levels in infected WT compared with olah ÀlÀ mice and might potentially contribute to OLAH-mediated disease.How OLAH is involved in recruitment of immune populations can be gleaned from oleic acid-induced lung injury animal models to study ARDS, where cell infiltration in response to tissue damage and inflammation are prominent hallmarks. 52,53Our findings suggest a role for OLAH during infection of lung epithelial cells.Studies highlighted importance of palmitolylation of IFITMs for their anti-viral activity 54,55 and direct action of oleic acid in altering endosomal membrane fluidity and overcoming the ability of IFITMs to block HAmediated fusion. 37We observed negligible effects of oleic acid treatment on infectivity of ifitm3 À/À macrophages; thus, any OLAH-mediated impact on IFITM3 activity in influenza-virus-infected cells might enhance early stages of viral infection at entry, 56 leading to enhanced viral production.
Given the function of OLAH in fatty acid biosynthesis, olah increases LDs in lungs and macrophages following infection, akin to cells infected with flaviviruses, 57,58 hepatitis C virus, 59 herpes simplex virus-1, 17 influenza. 17,60Influenza-infected macrophages in WT mice were reduced by inhibiting lipid droplet formation, indicating that disease outcomes may depend on lipid droplet formation.Regarded as cellular energy sources to fuel viral life cycle events, these organelles are key in early anti-viral responses by mediating cytokine production. 17,35Changes in fatty acids, including oleic and palmitic acids, impact macrophage functionality by direct signaling and/or reprogramming metabolic processes involved in cellular activation and polarization. 61,62Lacking OLAH can establish a cellular state primed for viral control and regulated cytokine responses.
We defined the role of OLAH in driving severe influenza disease, underpinning impacts of fatty acids on immune dysfunction and viral infection.Having well-known host factors defining patients at risk of dying at hospital admission and the ability to predict the severity of disease early during respiratory viral infection is critical for alleviating disease and preventing deaths.Elevated OLAH expression could serve as a biomarker to predict disease outcome and inform how to manage/prioritize early interventions, including artificial ventilation or steroid administration, more effectively in patients at risk of dying.Our study proposes that OLAH could constitute a target for development of novel therapeutics to protect against severe respiratory diseases, including pandemic, avian/seasonal influenza, SARS-CoV-2, RSV, and MIS-C.

Limitations of the study
Although we found no associations between olah expression and BMI, age, or sex in COVID-19 patients, this cohort was young (0-21 years).While we provide mechanistic insights into how macrophages drive OLAH-mediated disease, involvement of other cells remains unknown.Due to lack of OLAH-specific antibodies, we could not determine OLAH protein levels.Future studies should explore how OLAH levels in humans are regulated and mechanisms of upregulation during infections.OLAH could play a role in exacerbation of non-respiratory infections.Analyses of virus-induced lipid droplet compositions may inform on how OLAH-mediated products regulate cytokine responses.(legend continued on next page) An olah-deficient (olah À/À ) mouse strain on a C57BL/6 background was generated by constitutive knock out of the olah allele on chromosome 2 via CRISPR/Cas9-mediated gene editing.All embryonic procedures and generation of founder mice were performed at the Melbourne Advanced Genome Editing Center (Walter Eliza Hall Institute, Australia).Deletion of exons 1 and 2 of the olah allele using sgRNAs to target sites (with the sequence TGACAGCCATCTATACACGG and GCGAGATAAAGCCTACTGAC) flanking these exons resulted in the deletion of 4467bp.To detect the resulting sequence following deletion, forward (GGCTGCATTTCTCTCTTTA TGG) and reverse (GGGATGTGAGAAGGAAAAGG) primers were used to confirm the presence of a 312bp PCR product.Presence or absence of the wildtype allele was also performed using forward and reverse primer sequences, CTCAGACTCCGGTTCTGGAC and GAGGTCCCTTGTGTGTCTCC, respectively, to produce a product of 311bp.DNA samples from founder mice positive for the deleted allele were sequenced and confirmed by alignment to the predicted excised sequence prior to selection for breeding to produce homozygous knockouts.The genotyping of all breeding pairs and their first 2 generations of offspring was performed by Transnetyx (Tennessee, USA) using real-time PCR.

Microarray-based transcriptomic analysis
Peripheral blood (2 ml) was collected from each patient, preserved in PAXgene Blood RNA Tubes (QIAGEN) and stored at À80 C until use.Total RNA was extracted using a PAXgene Blood RNA Kit (QIAGEN) and transcriptomic analysis performed by Gene Tech (Shanghai) using a Affymetrix GeneChip Human Gene 2.0 ST Array.All analyses used the R statistical programming environment (v3.03).Raw CEL intensity files underwent 'Perfect Match' probe set-level normalization including background correction using robust multiarray averaging (RMA), quantile normalization and summarization of gene expression using the median published algorithm. 65Probe set intensity distributions between arrays (before and after normalization) and MA plots of the log intensity ratio versus the average log intensity were assessed for excessive technical variability.The early phase sample from patient a20 was excluded from analysis as the interquartile range of the expression deviation was substantially outside that exhibited by all other samples.Unexpressed probe sets, exhibiting less fluorescence than background probes and w a PSDABG P value > 0.05 were removed from subsequent analyses.The LIMMA package was used to identify differentially expressed probe sets. 66''Fatal'' and ''Recovered'' cohorts were compared using an empirical Bayes procedure to moderate standard error, and a t test was used to test for difference in expression levels.Corresponding p values were adjusted for multiple testing using the Benjamini & Hochberg approach.Differentially expressed genes were defined by p value <0.05 and an absolute log 2 fold-change of R2.0.For human challenge studies, microarray data were obtained from accession GEO: GSE73072, processed as described above, and OLAH expression was analysed using mixed effects models with the lme4 package in R to test for effects of time and infection status, with subject coded as a random effect.

Single-cell gene expression
For previously published single-cell datasets, data were processed and analysed exactly as indicated in the corresponding manuscript, with new figures generated using Seurat. 67For example, for healthy donor lung single-cell data, samples from a non-smoker were obtained from the NDRI, viable CD45 + and CD45 À fractions were sorted into distinct populations, processed using the 5 0 gene expression kit following manufacturer instructions (10X Genomics), and sequenced on an Illumina HiSeq 2000. 63Data from the CD45 + and CD45 À libraries were processed using CellRanger (v3.0.2; 10X Genomics) and aggregated with CellRanger aggr using default parameters, and downstream analyses were conducted in Seurat. 63

RNA sequencing
Whole blood from consented participants was collected directly into PAXgene Blood RNA Tubes using standard phlebotomy practices and immediately stabilized for storage at À80 C. Total RNA was isolated from PAXgene whole blood collection tubes on a QIAcube Connect MDx system using the PAXgene Blood RNA Kit following the manufacturer's instructions.Briefly, PAXgene tube contents were thawed at RT for 2 h to ensure RBC lysis then centrifuged for 10 min at 4,000 xg to pellet, washed, and resuspended.The resulting nucleic acid-containing pellets were transferred into 2 mL processing tubes, treated with proteinase K, and placed into the QIAcube worktable.The QIAcube automated further sample processing through a PAXgene Shredder spin column ensuring homogenization, binding total RNA (>18 nt) to a PAXgene RNA spin column, treatment with DNase-I (15 min at 30 C) to eliminate genomic DNA (gDNA), and final elution.Purified RNA samples were denatured for 5 min at 65 C in preparation for downstream applications.For FLU09 healthy control samples, total RNA was extracted from PBMCs by lysis and separation of aqueous phase using Trizol Reagent (Invitrogen) and the RNeasy Mini Kit (Qiagen) with DNase-I digestion.To ensure quality RNA sequencing, purified RNA samples were screened and selected for high total RNA integrity (RIN e R 7) and low genomic DNA contamination using Agilent High Sensitivity ScreenTape run on a 2200 TapeStation.RNA concentration was determined by Qubit HS RNA Assay Kit and a minimum of 300 ng total RNA was submitted for RNA sequencing.Sequencing libraries were generated with the Illumina TruSeq Stranded Total RNA with Ribo-Zero Globin and rRNA depletion kit and sequenced at 1003100bp on an Illumina NovaSeq 6000.
Raw sequencing reads were trimmed using the Trim Galore wrapper for CutAdapt 68 with default parameters before alignment to the GRCh38 reference using STAR (v2.7.9a) 69 with GENCODE annotations (v31).RSEM (v1.3.1) 70was used to quantify gene counts, and edgeR 71 and limma 72 were used for CPM calculation and normalization and modelling differential expression across severity groups while controlling for differences in days between symptom onset and sample collection, sex, whether a patient was previously healthy, steroid administration prior to sample collection, identification of a bacterial coinfection, age category, race, and ethnicity.p values were adjusted for multiple testing using the Benjamini & Hochberg FDR method.Glimma 73 was used for data exploration.For analysis of published bulk RNAseq from nasal curettage cells obtained after controlled human RSV infection, data were obtained from accession GEO: GSE155237 and analysed in the same manner but using annotated counts provided by the original study. 30fluenza virus infection of PBMCs Thawed PBMCs (1.5310 6 ) were infected with A/California/07/2009 (H1N1) at a MOI of 4 for 1 hour at 37 C and 5% CO 2 in RPMI-1640 media supplemented 2mM L-glutamine, 1mM MEM sodium pyruvate, 100mM MEM non-essential amino acids, 5mM HEPES buffer solution, 55mM 2-mercaptoethanol, 100 U/ml penicillin and 100 mg/ml streptomycin.Fetal calf serum was then added to achieve a 10% concentration (v/v) and incubated for a further 3 hours followed by the addition of Golgi Plug (BD Biosciences; final concentration of 1mg/ml).After 18 hours, cells were harvested and analysed for intracellular IAV nucleoprotein (NP) expression (Figure S7A).

Quantitative polymerase chain reaction (qPCR)
Tissues were collected in RNAlater (Thermo Fischer) or processed into single cell suspensions.RNA was extracted from tissue or cell samples using an RNeasy Mini Kit (QIAGEN).Gene expression levels were determined using a OneStep RT-PCR Kit (QIAGEN), with probes specific for genes encoding for OLAH (human: Hs00217864_m1, mouse: Mm00521437_m1, Thermo Fisher) and actin (human: Hs01060665_g1, mouse: Mm00607939_s1, Thermo Fisher).To detect influenza gene expression levels encoding for viral antigens, the following primers were used; forward primer 5 0 CCGGAACCCAGGAATG; and reverse primer 5 0 GAGTGCAAGACCGTG CTAGAA for NP, and forward primer 5 0 GACCRATCCTGTCAC CTCTGAC; and reverse primer 5 0 GGGCATTYTGGACAAKC GTCTACG for M. Standards for both NP and M were run alongside the tested samples to generate a standard and copy numbers per mg RNA were interpolated.qPCR was performed and analyzed on QuantStudio 6 and 7 Pro Real-Time PCR Systems (Applied Biosystems).

Cytometric bead array
Cytokines and chemokines in human plasma were previously quantified using the human inflammatory cytokine kit or a human soluble protein flex-set system (BD Biosciences) as previously reported. 11Mouse tissue homogenates and macrophage culture supernatants were assayed using a murine soluble protein flex-set system (BD Biosciences) according to the manufacturer's instructions.All samples were analysed using a BD FACSAria II or BD FACS Canto II (BD Biosciences).Data were analysed using FCAP Array software (Soft Flow Inc., Hungary).

Influenza virus infection of mice
Influenza virus infection was performed under light anaesthesia with isofluorane and intranasal instillation (30ml) with 10 4 to 10 5 plaque forming units (PFU) of A/HK/x31 (X31; H3N2).Mice were culled when a humane endpoint was reached; R25% of original body weight lost.For lethal LD 50 -infection, mice were infected with a viral titre that results in 50% mortality rate.All virus stocks were dose-titrated in mice to determine the LD 50 dose prior to experimental use.An LD 50 dose of 1 3 10 5 PFU was used for experiments in Figures 4B-4F and 5E-5G, and 2 3 10 4 PFU was used in Figures 4G, 4H, and 5A-5D.All harvested tissues were passed through 70 mm cell sieves to obtain single-cell suspensions.For the lungs, enzymatic digestion in collagenase III (Worthington Biochemical Corporation, USA; 1 mg/mL) and DNase I (Sigma-Aldrich, Germany; 0.5 mg/mL) for 30 min at 37 C was performed prior to this.Bronchoalveolar lavages were collected by flushing the lungs in situ with 1ml of PBS by means of a 23G needle inserted into the trachea and cells obtained by centrifugation.Where necessary, cell suspensions were incubated with 0.15 M NH4Cl and 17 mM Tris-HCI at pH 7.2 for 5 min at 37 C to lyse red blood cells.To prepare samples for histopathology, whole tissues and organs were collected into PBS containing 10% formalin and sent for analysis at the Phenomics Australia Histopathology Slide Scanning Service (The University of Melbourne, Australia).

Bone marrow chimeras
Bone marrow chimeras were established using a method previously described. 74,75Briefly, CD45.1-expressingC57BL/6 WT and olah À/À mice were lethally irradiated with 2 doses of 5.5 Gy 3 hours apart and reconstituted with 5 3 10 6 T cell-depleted bone marrow cells from CD45.2-or CD45.1-expressingWT or olah À/À mice.Donor and recipient cells were discriminated by isoform expression (CD45.1 or CD45.2) in vivo.Mice were depleted of radioresistant T cells on the following day by i.p. injection with anti-Thy1 antibody (clone T24; in the form of 100mL of cell culture supernatant manufactured in-house).Animals were maintained on antibiotics (neomycin and polymyxin B, Millipore Sigma) contained in drinking water for 6 weeks.Chimeric mice were allowed to reconstitute for at least 8 weeks.Using flow cytometry staining with CD45.1 and CD45.2-specific antibodies, we have validated each of our chimeras has high reconstitution rates (>90%) of cells of the proper phenotype in vivo (Figure S4A).

Determination of viral titres
Clarified supernatants from centrifuged homogenized lungs of influenza-infected mice were assayed for titres of infectious virus in plaque assays using Madin Darby canine kidney (MDCK) cells cultured in six-well plates. 75Briefly, supernatants were added to confluent MDCK layers and held for 1 hour at 37 C. Monolayers were then overlaid with Leibovitz L15 medium containing 0.9% agarose and 2mg/ml of Trypsin (Worthington Biochemical, NJ, USA).Plates were incubated at 37 C and 5% CO 2 for 3 days and virus-mediated cell lysis counted as plaques on the monolayer and expressed as plaque forming units (PFU).
Clarified supernatants from macrophage cultures in vitro infected with murine norovirus (MNV) were assayed for titres as previously described. 76Briefly, supernatants were added to confluent monolayers of RAW264.7 cells and for 1 hour at 37 C. Cells were overlaid with 2 ml of medium containing 70% DMEM, 2.5% FCS, 15 mM NaHC0 3 , 5 U of penicillin-streptomycin, 25 mM HEPES, 2 mM Glutamax, and 0.35% low-melting-point agarose/well and incubated at 37 C and 5% CO 2 for 2 days.Cells were fixed by adding 1 ml of 4% formaldehyde/well directly onto the overlay and held for 30 min at room temperature before rinsing with water and staining with 0.2% crystal violet for 20 min, to visualise and count plaques.Titres of West Nile Virus, Kunjin strain (WNV KUN ), was performed as previously described 77 where supernatants were added to confluent monolayers of VERO cells instead and held for 2 hours before overlay with 1ml of the same media as above, except 44% DMEM and 0.8% carboxymethylcellulose was used (instead of agarose), and incubated at 37 C and 5% CO 2 for 3 days before fixing and staining as above.

Intracellular protein and lipid droplets
Lungs were inflated with optimum cutting temperature fluid and snapped frozen prior to slicing into 14mm sections with a Leica CM 3050 S cryostat and mounting on microscope slides for storage at -80 C. For analysis, sections were fixed with 4% paraformaldehyde (Sigma) for 15 min at room temperature, permeabilised with 0.1% Triton X-100 (Sigma T9284-500ml) for 10 min and blocked with 1% BSA for 30 mins.Sections were stained with anti-IAV nucleoprotein (IAV-NP) (Prof Weisan Chen, La Trobe University) for 1 hour then washed and incubated with Alexa Fluor 555 (Life technologies) secondary antibody for 1 hour.BODIPY 493/503 (Life technologies) was used to stain for lipid droplets (LDs) at 1ng/ml for 1 hour at room temperature and nuclei were stained with DAPI (Sigma) for 5 min.Vectashield Antifade Mounting Medium (Vector Laboratories) was added prior to imaging on a ZEISS 800 confocal microscope.Droplet numbers were analysed from acquired z-stack maximum intensity projection images (Zeiss ZEN blue edition software) in ImageJ software (National Institutes of Health, USA) using the particle analysis tool and normalised to average LDs per cell.For each sample, at least 9 fields of view from different locations across the coverslip were imaged at 633 magnification.For enumeration of LDs in influenza-infected peritoneal macrophages (see below), cells were seeded into 8-well glass chamber slides (Nunc, USA) and fixed and permeabilised as indicated above.At least 4 fields of view for each replicate were imaged.
Tetramer and antibody staining Cells were stained with Fixable Live/Dead AquaBlue viability dye (Life Technologies) at room temperature for 10 mins.For peptide MHC (pMHC) tetramer staining, cells were incubated D b NP 366-374 -APC and D b PA 224-233 -PE for 60 mins in FACS Buffer (PBS containing 0.5% BSA and 2mM EDTA) at room temperature in the dark.Surface antibody staining was performed in 50 ml FACs Buffer for 30 mins at 4 C. Various combinations of fluorochrome-conjugated antibodies were used (key resources table) with specific gating strategies to define distinct cell populations (Figures S7B-S7E).For intracellular antibody staining, cells were fixed and permeabilised following surface antibody staining using a Cytofix/Cytoperm Plus kit (BD Biosciences) as per the manufacturer's instructions unless specified below.Following washing with Perm Wash buffer (BD Biosciences), cells were stained with fluorochrome-conjugated antibodies in 50 ml of the same buffer for 30 mins at 4 C. To detect intracellular IAV-NP, cells were fixed and permeabilised using an eBioscience FOXP3 staining kit (Thermo Fisher) instead as per the manufacturer's instructions and stained using FITC-anti-IAV NP antibody (Sapphire Biosciences).For detection of intracellular M, cells were first fixed with 4% PFA for 30 mins, then permeabilized with 0.1%Triton X-100 for 15 mins at 4 C, followed by staining with anti-M antibody (Bio-Rad) and AF568-anti mouse Ig.To detect surface HA expression, cells were stained with polyclonal anti-H3 HA (A/Hong Kong/1/68) in the form of goat antiserum (BEI resources) and an AF633-anti-goat Ig (Invitrogen).All staining using metabolic markers Mitotracker-APC, 2-DNBG-FITC and TMRM-PE were performed at 37 C following surface staining with antibodies.Samples were acquired on a BD FACSAria II or BD LSR Fortessa flow cytometer and data was analyzed by Flowjo Software version 10 (FlowJo LLC, Becton, Dickinson).

Stimulation of T cells
To measure influenza-specific T cell responses in mice, cells from lungs, BAL and spleen were harvested at day 10 post-infection and seeded onto U-bottom 96-well plates (Sigma-Aldrich) and incubated at 37 C and 5% CO 2 for 5 hours in cRPMI (RPMI-1640 media supplemented with 2mM L-glutamine, 1mM MEM sodium pyruvate, 100mM MEM non-essential amino acids, 5mM HEPES buffer solution, 55mM 2-mercaptoethanol, 100 U/ml penicillin, 100 mg/ml streptomycin and 10% fetal calf serum [all Gibco]) containing 1mg/ml Golgi-Plug (BD Biosciences) and 10 U/ml recombinant human IL-2 (Roche Diagnostics, Basel, Switzerland) in the absence or presence of 1mM NP 366-374 or PA 224-233 -peptide.Cells were subsequently stained for surface markers and intracellular cytokines and analysed by flow cytometry as described above.
Infection of murine peritoneal macrophages Peritoneal cells were obtained by in situ lavage of the peritoneal cavity three times with 5ml of FACS Buffer.Exudate cells from WT or olah À/À mice were pooled, centrifuged and red blood cells lysed before seeding in 24-well plates (10 6 cells/ml) for 4 hours at 37 C and 5% CO 2 in cDMEM (DMEM media supplemented with 2mM L-glutamine, 1mM MEM sodium pyruvate, 100 U/ml penicillin, 100 mg/ml streptomycin and 10% fetal calf serum [all Gibco]).Following removal of non-adherent cells, 5ml of FACS Buffer was added for 5 min and adherent cells detached by gentle pipetting.Collected cells were then re-seeded in cDMEM at 2-5 x 10 5 cells/250 ml in a flatbottom 96-well plate overnight and infected the following day with X31 influenza virus in serum-free DMEM at an MOI of 4 for 1 hour.Infections using WNV KUN and MNV was performed at an MOI of 1 and 5, respectively, for 2 hours.Virus was then aspirated, and media replaced with fresh serum-free DMEM until harvest for analysis.For treatment of macrophages with fatty acid synthase or lipid droplet inhibitors, the replaced media was supplemented with TOFA (Abcam; 10 mg/ml), C75 (Sigma; 10 mg/ml), Beauveriolide I (Cayman Chemical; 10 mM) or ML262 (Cayman Chemical; 1 mM).For treatment of macrophages with oleic acid, exudate cells were cultured overnight in cDMEM supplemented with oleic acid (500 mM; Sigma-Aldrich, Australia) prior to infection the next day and maintained in fresh serum-free DMEM containing oleic acid until harvest for analysis.

Monophasic lipid extraction from mouse lungs
Whole lungs were harvested, weighed, minced and freeze dried.Prior to lipid extraction, freeze dried lungs were incubated with 200 ml of ice-cold 60% methanol containing 0.01% (w/v) butylated hydroxytoluene (BHT) and incubated overnight at À20 C. SPLASH LIPIDOMIX Mass Spec Standard (10ml; Cat.330707, Avanti Polar Lipids, Birmingham, AL, USA) and deuterated saturated/monounsaturated fatty acid standards (10ml; Cayman Chemical, USA) were added to each sample and tissue homogenised with a Bullet Blender (Next Advance, Troy, NY, USA) and sonicated for 20 minutes.Monophasic lipid extraction was performed according to Lydic et al. 78 wherein 120 ml of water, 420 ml of methanol with 0.01% (w/v) BHT, and 270 ml of chloroform were added to all samples prior to thorough vortexing and incubated on a shaker at 1400 rpm for 30 mins.Following centrifugation, supernatants containing lipids were transferred to new tubes and the remaining pellets re-extracted by adding 100ml of water and 400ml of chloroform:methanol (1:2 v/v) containing 0.01% (w/v) BHT and incubation on a shaker at 1000 rpm for 15 mins.Supernatants from each extraction round were pooled, dried by evaporation under vacuum and freeze-dried.Lyophilised samples were then resuspended in chloroform:methanol (1:9 v/v) containing 0.01% BHT and centrifuged to remove precipitates.

Tandem mass spectrometry (MS/MS) analysis
Samples were analyzed by ultrahigh performance liquid chromatography (UHPLC) coupled to tandem mass spectrometry (MS/MS) employing a Vanquish UHPLC linked to an Orbitrap Fusion Lumos mass spectrometer (Thermo Fisher Scientific, San Jose, CA, USA), with separate runs in positive and negative ion polarities.Solvent A was 6/4 (v/v) acetonitrile/water with 5 mM medronic acid and solvent B was 9/1 (v/v) isopropanol/acetonitrile.Both solvents A and B contained 10 mM ammonium acetate.10 uL of each sample was injected into an Acquity UPLC HSS T3 C18 column (1 x 150 mm, 1.8 mm; Waters, Milford, MA, USA) at 50 C at a flow rate of 60 ml/ min for 3 min using 3% solvent B. During separation, the percentage of solvent B was increased from 3% to 70% in 5 min and from 70% to 99% in 16 min.Subsequently, the percentage of solvent B was maintained at 99% for 3 min.Finally, the percentage of solvent B was decreased to 3% in 0.1 min and maintained for 3.9 min.All MS analysis were performed using an electrospray ionization source (3.5 kV in positive-and 3.0 kV in negative ionization-mode).In both polarities, the flow rates of sheath, auxiliary and sweep gases were 25 and 5 and 0 arbitrary unit(s), respectively.The ion transfer tube and vaporizer temperatures were maintained at 300 C and 150 C, respectively, and the ion funnel RF level was set at 50%.From 3 to 24 min, top speed data-dependent scan with a cycle time of 1 s was used.Within each cycle, a full-scan MS-spectra were acquired firstly in the Orbitrap at a mass resolving power of 120,000 (at m/z 200) across an m/z range of 300-2,000 using quadrupole isolation, an automatic gain control (AGC) target of 4 5 and a maximum injection time of 50 milliseconds, followed by higher-energy collisional dissociation (HCD)-MS/MS at a mass resolving power of 15,000 (at m/z 200), a normalized collision energy (NCE) of 27% at positive mode and 30% at negative mode, an m/z isolation window of 1, a maximum injection time of 35 milliseconds and an AGC target of 5e4.For the improved structural characterization of glycerophosphocholine (PC) lipid cations, a data-dependent product ion (m/z 184.0733)-triggered collision-induced dissociation (CID)-MS/MS scan was performed in the cycle using a q-value of 0.25 and a NCE of 30%, with other settings being the same as that for HCD-MS/MS.For the improved structural characterization of triacylglycerol (TG) lipid cations, the fatty acid + NH3 neutral loss product ions observed by HCD-MS/MS were used to trigger the acquisition of the top-3 data-dependent ion trap CID-MS 3 scans in the cycle using a q-value of 0.25 and a NCE of 30%, with other settings being the same as that for HCD-MS/MS.Dynamic exclusion of 15 seconds was enabled and only ions with charge state of 1-3 were selected for fragmentation.

Lipidomics analysis of mouse lungs
LC-MS/MS data was searched through MS Dial 4.90.The mass accuracy settings are 0.005 Da and 0.025 Da for MS1 and MS2.The minimum peak height is 50000 and mass slice width is 0.05 Da.The identification score cut off is 80%.Post identification was done with a text file containing name and m/z of each standard.In positive mode, [M+H]+, [M+NH4]+ and [M+H-H2O]+ were selected as ion forms.In negative mode, [MÀH]À and [M+CH3COO]À were selected as ion forms.All lipid classes available were selected for the search.PC, LPC, DG, TG, CE, SM were identified and quantified at positive mode while FA, PE, LPE, PS, LPS, PG, LPG, PI, LPI, PA, LPA, Cer, CL were identified and quantified at negative mode.The retention time tolerance for alignment is 0.1 min.Lipids with maximum intensity less than 5-fold of average intensity in blank were removed.All other settings were default.All lipid LC-MS features were manually inspected and re-integrated when needed.These four types of lipids; 1) lipids with only sum composition except SM and PC, 2) lipid identification due to peak tailing, 3) retention time outliner within each lipid class, 4) LPA and PA artifacts generated by in-source fragmentation of LPS and PS were also removed.The shorthand notation used for lipid classification and structural representation follows the nomenclature proposed previously. 79Quantification of lipid species in the unit of pmol/100 mg of lung tissue from each sample was achieved by normalizing the LC peak areas of identified lipids against those of the corresponding internal lipid standards in the same lipid class, the quantity of each lipid standard at pmol, and finally the weight of lung tissue.For the lipid classes without correspondent stable isotope-labelled lipid standards, the LC peak areas of individual molecular species within these classes were normalized as follows: the MG species against the DG (18:1D7_15:0); the LPG against the PG(18:1D7_15:0), the LPA against the PA(18:1D7_15:0), the LPS against the PS (18:1D7_15:0), and the Cer against the SM(d18:1_18:1D9).LipidSigR (version 0.99.0)R package was used to log transform normalised lipid concentrations prior to comparison by multiple t-test (package rstatix version 0.7.2) with p values < 0.01 considered significant.Volcano plots and heatmaps were generated using EnhancedVolcano (version 1.14.0) and ComplexHeatmap (version 2.12.1) respectively.

Proteomic sample preparation
Isolated macrophages were lysed in 4% SDS, 100 mM Tris pH 8.5 by boiling at 90 C for 10 minutes with shaking at 2000 rpm before being prepared using the SP3 sample preparation approach. 80,81Samples were reduced with Dithiothreitol (20 mM) for 10 minutes at 90 C and alkylated with Iodoacetamide (40 mM) for 40 minutes at room temperature in the dark.500 mg of Milli-Q water prewashed Sera-mag speed magnetic beads (GE Healthcare) were added to samples and proteins precipitated onto beads by the addition of ethanol to a final concentration of 80%.Samples were then shaken for 5 minutes at 1000 rpm and beads washed twice with 80% ethanol using a magnetic rack.Beads were then resuspended in 100mM tetraethylammonium bromide, pH 8.5 containing 0.5 mg of Trypsin/Lys-C (Promega, protein:protease ratio 1:100) and samples incubated at 37 C for 18 hours at 1000 rpm.3][84] SDB-RPS material packed into 200 ml tips were washed with 30% methanol, 1% TFA before being equilibrated with 90% isopropanol, 1% TFA and samples loaded onto SDB-RPS columns by centrifugation.Columns were washed with 90% isopropanol, 1% TFA, followed by 90% ethyl acetate, 1% TFA and then 1% TFA.Samples were then eluted twice with 5% ammonium hydroxide, 80% acetonitrile before being dried by vacuum centrifugation.

Data independent acquisition (DIA) analysis
Peptide samples were re-suspended in Buffer A (2% acetonitrile, 0.1% TFA) and separated using a two-column chromatography set up composed of a PepMap100 C18 (20 mm 3 75 mm) trap and a PepMap C18 (500 mm 3 75 mm) analytical column (Thermo Fisher Scientific) coupled to a Orbitrap Eclipse Tribrid Mass Spectrometer (Thermo Fisher Scientific).Samples were loaded onto the trap column for 5 minutes at 5 ml/min with Buffer A (0.1% formic acid, 2% DMSO) and then separated over 125-minutes using the analytical column at 300 nl/min by altering the Buffer B composition (0.1% formic acid, 77.9% acetonitrile, 2% DMSO) concentration from 2% to 28% over 106 minutes, then to 40% over 9 minutes, and to 80% over 3 minutes.The composition was held at 80% B for 2 minutes, dropped to 2% over 2 minutes and held at this concentration for 3 minutes.The Orbitrap Eclipse Mass Spectrometer was operated in a data-independent mode automatically switching between the acquisition of a single Orbitrap MS scan (350-951 m/z, maximal injection time of 50 ms, an Automatic Gain Control (AGC) of 250% and a resolution of 120k) and the collection of 16 m/z DIA windows between 350 and 951 m/z (200À2,000 m/z, NCE 28%, maximal injection time of 54 ms, a AGC set to 2000% and a resolution of 30k).The proteomic data was searched with Spectronaut (Biognosys, version 17.1.221229)using the DIA library free analysis workflow allowing oxidation of Methionine and N-terminal acetylation as variable modifications as well as Carbamidomethyl as a fixed modification of cysteine with Trypsin as the digestion specificity.The protein label-free quantitation (LFQ) method was set to MaxLFQ with single hit proteins excluded, and the precursor PEP cut-off altered to 0.05 from the default 0.2 to improve quantitative accuracy.Searches were performed against the Mouse proteome (Uniprot Accession: UP000000589) supplemented with the Influenza A virus strain AX-31 H3N2_I000X Proteome (UniProtKB ID: 132504).Protein outputs from Spectronaut were processed using Perseus (version 1.6.0.7) 85 with missing values imputed based on the total observed protein intensities with a range of 0.3 s and a downshift of 1.8 s.Statistical analysis was undertaken within Perseus by grouping biological replicates together which were assessed using student t-tests with multiple hypothesis correction undertaken using a permutation-based FDR approach allowing an FDR of 5%.

Human plasma preparation and lipidomics
Day 1 plasma from SARS-CoV-2 PCR positive outpatient (n = 38) and hospitalized (n = 35) participants in the CIViC-19 cohort were carefully thawed on ice.Lipid extraction from 25 mL plasma was achieved by using a modified version of the Bligh and Dyer extraction protocol. 86Aliquots of stable internal standard mixtures-SPLASH Lipidomix Mass Spec Standard were spiked into all samples.Lipid extract (4 mL) was separated using a Thermo Accucore C30 column on a Thermo Vanquish Ultimate 3000 UPLC coupled to a Thermo Orbitrap Exploris 240 mass spectrometer (Thermo Fisher Scientific).Mass spectrometric (MS) data were operated in negative ionization electrospray modes.Data were acquired in full-scan mode (m/z 70 to 1,000) at a resolution of 120,000 for all samples.MS/MS data was collected in a sequence of separate runs operated by the intelligence-driven software AcquireX (Thermo Scientific, Waltham, MA) for FA(18:1) identification.The raw data were processed using LipidSearch (vers.5.0; Thermo Scientific, Waltham, MA).

QUANTIFICATION AND STATISTICAL ANALYSIS
Where appropriate, data were tested for normality using a Shapiro-Wilk test.For nonparametric analysis between 2 groups, an unpaired two-tailed Mann-Whitney U test was used.For parametric analysis between 2 groups, an unpaired two-tailed Welch test was used.When 3 or more data sets were compared, a 2-way ANOVA with a Holms-Sidak post-hoc test was used.Analysis of survival data was performed using the Kaplan Meier method.A P value less than 0.05 was considered statistically significant.Specific details of statistical tests used in each experiment, including dispersion measures, are indicated within the figure legends and unless otherwise stated, were performed using Graph-Pad Prism, version 9.0.

Figure 1 .
Figure 1.Elevated OLAH expression in humans is associated with life-threatening influenza disease outcomes (A) Timeline of hospital stay of H7N9-infected patients who recovered or died.Circles denote admission and discharge/death days post-disease onset.Arrows indicate early and late blood samples for microarray-based transcriptomic analysis.(B) Plasma cytokines (mean ± SD) were assessed within 1-2 days of admission, and cytokine data reported as a larger cohort in Wang et al. 11 ).(C) Volcano plot showing differential gene probeset expression between recovery and fatal patients at early disease time points.OLAH expression levels (mean ± SD) in early (circles) and late (squares) samples with paired data.(D) OLAH transcript expression in paired early and late samples.(E) qPCR was performed on RNA from single-cell suspensions of healthy human tissues and PBMCs for OLAH transcript levels (n = 2-3/sample).Bar graphs depict average expression levels (±SD) relative to house-keeping gene actin.(F) t-distributed stochastic neighbor embedding representation of single-cell RNA expression from healthy human lungs showing OLAH expression and association with genes used for cell-type annotations.(G) OLAH expression in single-cell RNA transcriptomic data acquired from PBMCs of healthy individuals (n = 2) or patients hospitalized with influenza (n = 3) at 1-14 days post-symptom onset. 26Expression levels are segregated by disease state and cell population.(H) PBMCs from healthy donors (n = 11) were uninfected or infected with A/California/07/2009 at MOI = 4; intracellular nucleoprotein (NP) expression analyzed.Representative concatenated dot plots showing percentages of cells expressing NP from infected or uninfected cultures.Bar graphs show mean frequencies (±SD) of cells expressing NP (n = 11) and representative MFI (±SD) from 4 donors.Statistical analysis was performed using Mann-Whitney test.See also Figure S7.

Figure 3 .
Figure 3. Deletion of olah in mice reduces de novo fatty acid synthesis (A) qPCR measuring olah was performed on RNA from tissues or single-cell suspensions derived from organs (n = 1-6).(B) Sorted CD45 + and CD45 À cells from lungs of naive C57BL/6 (WT) mice (n = 3).Bar graphs depict average expression (±SD) relative to expression of housekeeping actin gene.(C) Schematics of the olah gene on mouse chromosome 2 and the region within exon 1 and 2 targeted for deletion.qPCR on RNA from WT and olah À/À mice confirm knockout of olah.(D) Volcano plot showing differentially expressed lipid species in lungs of naive mice.Dark gray and red dots represent species upregulated (false discovery rate [FDR] < 0.01) in WT and olah À/À mice.(E) Heatmap depicting levels of differentially expressed lipid species (p < 0.01) in lungs of naive WT and olah À/À mice.LPE 16:1(a) and LPE 16:1(b) are sn substitution isomers.(F) Proportion of lipid species, typed by lipid class, downregulated in olah À/À mice.(G) Fatty acids levels in lungs of naive mice (n = 5/group) catalyzed by OLAH, except lauric acid.Statistical analysis was performed by Welch t test.See also Figure S2.

Figure 5 .
Figure 5. Olah promotes influenza-induced lipid droplet formation and viral infection in the lungs (A) Intracellular NP in CD64 + CD11b + macrophages and CD31 À EpCAM + epithelial cells from lungs of mice at day 1 after LD 50 infection with X31 (n = 8/group).(B) Contribution of immune and non-immune cells in driving OLAH-mediated disease severity.Bone marrow chimeras were established by adoptive transfer of donor bone marrow cells from WT or olah À/À mice into irradiated recipient mice (n = 10/group).(C) After 8 weeks, mice were infected with a severe dose of X31; survival rates and body weights (mean ± SEM) monitored.(D) Numbers of donor-derived WT (CD45.1 + ) or olah ÀlÀ (CD45.2+ ) macrophages and neutrophils (mean ± SEM) in lungs of LD 50 -infected olah ÀlÀ (CD45.2+ ) or WT (CD45.1 + ) recipients, respectively, at day 1 post-infection.(E) Lungs from mice following severe X31-infection (n = 3/group) were sliced (14 mm), fixed, and permeabilized prior to staining with antibodies to CD36 and influenza-NP, BODIPY for lipid droplets, and visualization by confocal microscopy.Representative single and merged (incorporating DAPI nuclei staining) images (34 resolution) at each time point.Scale bars, 200 mm.(F) Quantitation of lipid droplets from >150 cells in each image (at 363 resolution) was performed from R9 different fields of view across the coverslip.LOESS regression lines with 95% confidence interval (CI) are shown.(G) Viral titers (mean ± SD) in lungs following severe X31 infection (n = 5-9/group).Statistical analysis was performed by (A, C, right, and D) Mann-Whitney tests, (F and G) two-way ANOVA with Holms-Sidak post hoc test, and (C, left) Kaplan-Meier method.See also Figures S4 and S7.

Figure 6 .
Figure 6.Increased viral protein expression and cytokine production in macrophages are driven by olah expression (A) Co-expression of CD64 and F4/80 on peritoneal macrophages isolated from WT and olah À/À mice.(B) Pooled macrophages or MDCK cells (n = 4/group) were infected with MOI = 4 X31.qPCR was performed for viral RNA encoding for NP or M (mean ± SD). (C) Intracellular expression of viral protein in infected/uninfected macrophages and MDCK cells at 8 h post-infection (n = 4-5/group).Representative concatenated dot plots depicting frequencies of NP-or M-expressing F/480 + macrophages and MDCK cells from infected/uninfected cultures, with bar graphs showing mean ± SD within each group.(D) Frequencies of intracellular NP-and M-expressing cells (mean ± SD) over 8 h post-infection (n = 4/group/time point).(E) Segregation of NP-expressing macrophages into male or female mice (mean ± SD).

Figure 7 .
Figure 7. OLAH-mediated infection in macrophages is driven by oleic acid and lipid droplet usage(A) Lipid droplets in influenza-infected macrophages were measured by infecting cells in chamber slides followed by BODIPY staining and visualization by confocal microscopy.Quantitation of lipid droplets was performed from >150 cells per image (363 resolution) from 4 different fields of view/coverslip.LOESS regression line with 95% CI across time is shown.Fold-change in lipid droplet numbers post-infection was calculated relative to average numbers prior to infection.(B) Male mice fed on an oleic acid rich diet (WT OA diet ) or normal chow (WT) were infected with X31 (10 4 PFU; n = 6/group) and monitored for body weight loss; lung viral titers (mean ± SD) measured at 5 days post-infection.(C) Viral protein expression in macrophages (n = 4/group) cultured with 500 mM oleic acid prior and during infection.Frequencies of intracellular NP-or M-expressing cells (mean ± SD). (D) Cytokine levels in supernatants (mean ± SD) at 24 h post-infection.(E) Frequencies of intracellular NP-expressing WT and ifitm3 À/À macrophages (n = 4-5/group) at 8 and 24 h post-infection following culture with oleic acid prior and during infection.

Figure S2 .
Figure S2.Phenotype and lung lipidomic profiles of WT and olah À/À mice, related to Figures3C and 4 (A) Images of naive 7-week-old male WT and olah ÀlÀ mice and body weight profiles.(B) Numbers of immune cell populations (mean ± SD) in organs of naive 7-week-old male WT and olah ÀlÀ mice.(C) Levels of lipid classes in lungs from naive WT and olah ÀlÀ mice.Multiple t tests on log 10 transformed data with FDR adjustment.(D) Volcano plot of differentially expressed lipid species in lungs of infected WT and olah ÀlÀ mice at 1 day post-infection.Dark gray and red dots represent species that are upregulated (FDR < 0.01) in WT and olah ÀlÀ mice, respectively.The heatmap below depicts levels of differentially expressed lipid species (p <0.01).(E) Levels of lipid species containing arachidonic acid (20:4), dihomo-g-linolenic acid (20:3), and adrenic acid(22:4) in lungs of naive and infected WT and olah ÀlÀ mice.Two-way ANOVA with a Holms-Sidak post hoc test.

Figure S3 .
Figure S3.Innate immune responses in lungs of infected WT and olah ÀlÀ mice, related to Figures 4G and 4H

(
A) Cytokine levels (mean ± SD) in BAL fluid and lungs of LD 50 -infected mice (n = 5-6/group) at 1 day post-infection.Mann-Whitney test.(B) Total cell numbers (mean ± SD) across 3 days of infection.Two-way ANOVA with a Holms-Sidak post hoc test.(C) Numbers (mean ± SD) of innate immune cell populations on days 2 and 3 post-infection.Mann-Whitney test.(D) Frequencies of lung epithelial cells, neutrophils, and macrophages (mean ± SEM) expressing viral antigens (intracellular NP and surface HA) in LD 50 -infected mice (n = 8/group).

Figure S4 .
Figure S4.Reconstitution rates of bone marrow chimeras and representative microscopy images used to quantify lipid droplet numbers, related to Figures 5B-5F and 7A (A) Expression of congenic markers (CD45.1 or CD45.2) on cells in blood of recipient mice following establishment of bone marrow chimeras.The congenic marker for each donor and recipient pairing is indicated in brackets.(B) Representative merged images (incorporating DAPI nuclei staining) used to quantify lipid droplets in frozen lung sections of mice following LD50 infection.White framed boxes in each top image depict a field of view used for analysis of corresponding bottom images.Colours within each image represent DAPI (blue), BODIPY for lipid droplets (green), influenza NP (purple), and CD36 expression (red).(C) Representative images used to quantify lipid droplets in macrophages following X31-infection.The white scale bar represents 50 mm.

Figure S6 .
Figure S6.Expression of metabolic and polarization markers and NP expression in macrophages, related to Figure6

Figure S7 .
Figure S7.Gating strategy to define cell populations, related to Figures 1H, 4H, 5A, and S5 (A) Gating strategy to define PBMC populations for Figure 1H.(B and C) Gating strategy to define innate immune cell populations and respiratory epithelial cells for Figures 4H and 5A.(D) Gating strategy to define adaptive immune populations for Figure S5.(E) Gating strategy to define tetramer-specific, cytokine-producing CD8 + T cells and subsets for Figure S5.
Detailed methods are provided in the online version of this paper and include the following: d KEY RESOURCES TABLE d RESOURCE AVAILABILITY B Lead contact B Materials availability B Data and code availability d EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS B Human study participants B Experimental mice d METHOD DETAILS B Microarray-based transcriptomic analysis B Single-cell gene expression B RNA sequencing B Influenza virus infection of PBMCs B Quantitative polymerase chain reaction (qPCR) B Cytometric bead array (F) Cytokine levels (mean ± SD) in supernatants of macrophages at 24 h post-infection (top).Crossed-out boxes indicate cytokines not detected.Depicted are individual cytokines (mean ± SD). (G) West Nile Virus Kunjin (WNV KUN ) and murine norovirus (MNV) titers (mean ± SD) in supernatants of macrophages infected at MOI = 1 or MOI = 4, respectively, harvested at 24 h post-infection.Statistical analysis was performed by (B and D) two-way ANOVA with Holms-Sidak post hoc test, (C, E, and F) Mann-Whitney test, and (G) Student's t test.See also Figure S6.
Experimental mice C57BL/6 mice were bred and maintained in the Biological Research Facility in the Department of Microbiology and Immunology at The University of Melbourne.Following weaning ($3 weeks old), mice were placed on a normal maintenance diet (Barastoc, Ridley Corporation, Australia).In some cases, mice were fed with a semi-pure high fat rodent diet containing 7.7% oleic acid (93.4% of total monounsaturated fat content) and 5.8% palmitic acid (57.8% of total saturated fat content) (Diet SF04-001, Specialty Feeds, Australia) for at least 12 weeks before use.All animal experimentation was conducted in accordance with the Australian National Health and Medical Research Council Code of Practice for the Care and Use of Animals for Scientific Purposes Guidelines and institutional regulations following approval (permit numbers 1714304, 1614073 and 20532) by The University of Melbourne Animal Ethics Committee.Both female and male mice 7-15 weeks of age were used for experiments.