Impact of aging on gut-lung-adipose tissue interactions and lipid metabolism during influenza infection in mice
Article Open access27 October 2025
Main
The immune system is a critical determinant of host survival, enabling resistance to infections and maintenance of tissue integrity. However, the activation of immune responses is inherently associated with costs, necessitating precise regulation to achieve the optimal balance between the benefits and costs of mounting an immune response. The theory of antagonistic pleiotropy, first proposed in the context of evolutionary biology, suggests that traits beneficial to early-life fitness can incur costs that manifest later in life, after the period of strongest natural selection4. Traits that enhance immune resistance, such as heightened inflammatory responses, may provide survival and fitness advantages during reproductive years but contribute to chronic inflammation, autoimmunity and tissue degeneration in organisms later in life5,6. As a consequence, the immune system demonstrates how evolutionary pressures for early-life survival can embed long-term liabilities within host defence programmes. Antagonistic pleiotropy provides a compelling evolutionary framework for understanding why defence responses that once protected us may contribute to declining health later in life.
Beyond resistance, host defences also include cooperative strategies that mitigate infection-induced damage while having a neutral to positive impact on pathogen fitness. These include anti-virulence responses, which limit harmful pathogen and host-derived signals, and disease tolerance mechanisms, which protect from physiological damage by changing how the host responds to damage signals1,2,7. Whether cooperative defence alleles impose costs for the host and are themselves subject to antagonistic pleiotropy is unknown.
A consideration of the specificity of disease tolerance mechanisms shows the further complexities ageing introduces for cooperative defences. The specificity of disease tolerance is defined by the physiological perturbations or pathology that may occur in response to the infection, which can change with age1,8,9,10,11. These differences stem from age-related structural and functional changes in host physiology, which can alter both the nature and severity of pathology during infection. As a result, hosts of different ages may show distinct disease courses despite infection with the same pathogen. It would then follow that the disease tolerance mechanisms required for survival hosts may become maladaptive or incompatible in older individuals, contributing to increased susceptibility to infection-related pathologies, supporting the possibility that disease tolerance alleles show antagonistic pleiotropy.
Here, we use a polymicrobial sepsis model and a 50% lethal dose (LD50)-based framework to define how ageing alters host–pathogen cooperation and disease tolerance. We find that the Foxo1–Trim63 axis mediates a protective cardiac tolerance programme in young hosts but drives pathogenic remodelling in aged hosts. These findings reveal an age-dependent shift in the evolutionary function of disease tolerance mechanisms, highlighting antagonistic pleiotropy as a key principle linking ageing to infectious disease outcomes.
LD50 approach to disease tolerance
To investigate the role of disease tolerance at different life stages, we applied the phenomenon of LD50 to our study. LD50 describes the dose of a pathogen that kills 50% of a genetically identical host population. We previously demonstrated that this phenomenon can be used to elucidate mechanisms of host–pathogen cooperation3. For the present study, we used a polymicrobial sepsis model consisting of a 1:1 mixture of the gram-negative bacterium Escherichia coli and the gram-positive bacterium Staphylococcus aureus in 12-week-old (20–30 years equivalent for humans) and 75-week-old (56–69 years equivalent for humans) mice12, which we call ‘young’ and ’aged’, respectively, for the purposes of this study. We used this infection model because these two bacteria represent some of the most common gram-negative and gram-positive agents that cause sepsis in humans13. We used these two age groups because they allowed us to capture age-associated physiological changes (Extended Data Fig. 1a–q) while avoiding the increased frailty and non-specific mortality that can confound infection studies in more advanced age. We found the LD50 dose for polymicrobial sepsis in both young and aged mice to be roughly 1 × 108 total colony forming units (CFU) (Fig. 1a,b and Extended Data Fig. 1r), suggesting that lethality does not increase with age for this infection at the ages examined.
a,b, Young (a) and aged (b) LD50 survival. c,d, Young (c) and aged (d) LD50 health trajectories. Individual data points in Extended Data Fig. 2c,f. n (in parentheses) shows biologically independent animals from one out of three independent experiments (a–d). e, Representative heart images at the interventricular septum. Prominent myocardial vessels in dying. Increased red blood cells (congestion, large arrows), leukocytes (small arrows), mildly increased hypereosinophilic cardiomyocytes, oedema (asterisks). f, Representative images of spleen (scale bar, 100 μm); lung (scale bar, 50 μm); liver (scale bar, 50 μm) and kidney (scale bar, 100 μm). Dying mice have expanded splenic red pulp (congestion, asterisks); increased congestion, leukocyte infiltration in lung interstitium (arrows); increased sinusoidal congestion in liver (arrows); increased congestion in kidney medulla (asterisks). Some aged lungs, have perivascular and peribronchiolar lymphoid aggregates (asterisk) as incidental findings. Two independent experiments were performed (e,f). g, BUN. h, Creatinine. i, AST. j, ALT. For g–j, n shows biologically independent animals from three out of three (aged) or four out of four (young) independent experiments. k, Troponin I. l, BNP. For k,l, n shows biologically independent animals from two out of two independent experiments. m, Galectin-3. n shows biologically independent animals from two out of two (young) and three out of three (aged) independent experiments. n, Representative images of hearts. Original images in Supplementary Fig. 1. Three independent experiments performed. o, Heart weights normalized to body weight. p, Values of infected mice from o normalized to average of uninfected values from o. For o,p, n shows biologically independent weight-matched animals from two out of three independent experiments. Mean ± s.e.m. Two-way ANOVA with Sidak’s (c,d) or Tukey test (g–m). One-way ANOVA with Tukey test (o). Two-tailed unpaired t-test (p). Number of biologically independent mice shown in each panel. Scale bars, 50 μm (e); 3 mm (n).
Body composition parameters before the infection were not useful for predicting the infection trajectories of LD50-challenged animals (Extended Data Fig. 1s–y). However, those that succumbed to the LD50-challenge in both age groups developed quantifiable clinical signs of disease during infection, including hypothermia and morbidity, that were predictive of fate (Extended Data Figs. 1z–dd and 2a–f). Housing mice at thermal neutrality did not protect mice in either age group (Extended Data Fig. 2g–r). We calculated health scores of individual animals over the course of the infection to generate infection trajectories. We found a bifurcation in the trajectories of young mice that was apparent by 6 h postinfection, whereas the bifurcation in the aged mice was apparent by roughly 10 h postinfection (Fig. 1c,d and Extended Data Fig. 2c,f). The survivors in each age group did not differ in their ability to control the infection levels, nor did we detect differences in pathogen burdens between those that succumb to the LD50 (Extended Data Fig. 3a–o). Thus, infection outcomes in both age groups are driven by differences in the ability of the host to adapt to the infection, rather than differences in resistance mechanisms1,2,7. Taken together, our data show that the infectious inoculum is equivalent across age groups, and there are no apparent age-related differences in resistance mechanisms needed to control pathogen levels. instead, differences in survival reflect variation in cooperative defences. Thus, our LD50 polymicrobial sepsis model is an ideal system to investigate how ageing affects disease tolerance and the potential role of antagonistic pleiotropy.
Age shapes sepsis disease trajectories
Sepsis is defined as a life-threatening organ dysfunction caused by a dysregulated host response to infection14. Despite receiving the same inoculum challenge and showing comparable pathogen burdens (Extended Data Fig. 3a–o), the differences in physiologies before infection (Extended Data Fig. 1a–q) caused young and aged dying mice to have distinct responses that yielded different disease trajectories. Although dying mice of both age groups showed hypothermia and morbidity of comparable severity (Extended Data Figs. 1z–dd and 2a–f), the kinetics of disease and death were delayed in aged mice (Fig. 1a–d and Extended Data Figs. 2a–f and 4a). Young dying mice showed severe congestion in all organs analysed including the liver, spleen, kidney, lung and heart demonstrating impaired circulatory integrity (Fig. 1e,f and Extended Data Fig. 4b–f). Aged mice showed congestion in the liver, spleen and kidney, albeit congestion severity was less than what we observed in these same organs from young mice (Fig. 1e,f and Extended Data Fig. 4b–f). Compared with age-matched uninfected and surviving mice, dying mice from both age groups also showed significantly higher amounts of the renal damage marker blood urea nitrogen (BUN); however, concentrations were higher in aged mice compared with young dying mice (Fig. 1g). Consistent with this, aged dying mice had elevated amounts of circulating creatinine compared with young dying mice (Fig. 1h). Both young and aged dying mice showed comparable decline in the concentrations of albumin and comparable elevated concentrations of liver damage markers including alanine aminotransaminase (ALT) and aspartate transaminase (AST) (Fig. 1i,j and Extended Data Fig. 4g). Young but not aged dying, mice showed elevated amounts of troponin I, a marker of cardiac damage (Fig. 1k). Dying mice of both age groups showed elevated amounts of B-type natriuretic peptide (BNP), which is released on ventricular wall stretch and is a biomarker for heart failure15,16, although concentrations were significantly higher in the young dying mice compared with the aged dying mice (Fig. 1l). Galectin-3, which is implicated in cardiac fibrosis and remodelling and is clinically used as a biomarker for myocardial injury and the diagnosis and prognostication of heart failure17, was also slightly elevated at comparable amounts in dying mice of both age groups (Fig. 1m).
Our gross examination of organs and histopathology analyses, revealed further cardiac differences. Young dying mice showed enlarged hearts with changes in geometric shape, and histological evidence of oedema, changes in cardiomyocyte appearance and hypereosinophilia with increased cardiomyocyte score and leukocyte infiltration (Fig. 1e,n–p, Extended Data Fig. 4h–j and Supplementary Fig. 1). Aged dying mice showed severe cardiac atrophy with changes in geometric shape (Fig. 1n–p, Extended Data Fig. 4h–j and Supplementary Fig. 1). The rapid kinetics of cardiac remodelling we observed in our mouse models is comparable to what has been reported in humans18,19. Histopathological analysis revealed that hearts from aged dying mice showed similar, although less pronounced, histological changes in cardiomyocyte appearance and leucocyte infiltration indicating that similar acute microscopic changes are associated with different cardiac remodelling at the macroscopic level in young and aged dying septic hosts, perhaps consistent with a limited spectrum of histologic changes in the acute time frame of the disease (Fig. 1e and Extended Data Fig. 4h–j). RNA sequencing (RNA-seq) analysis revealed distinct molecular signatures indicative of cardiac remodelling (Extended Data Fig. 4k–t). Young dying mice had enrichment of genes involved in apoptosis, negative regulation of protein catabolic process, response to muscle stretch, regulation of cell shape and cardiac muscle tissue morphogenesis, which is consistent with cardiac remodelling and cardiomegaly (Extended Data Fig. 4n–t). The cardiac transcriptome of aged dying mice had enrichment of genes involved in ubiquitination, autophagic cell death, negative regulation of cell growth, cardiac muscle development and cell proliferation, which is consistent with their cardiac atrophic remodelling phenotype (Extended Data Fig. 4n–t). Gene ontology analysis also revealed that aged and young dying mice had distinct gene depletion profiles (Extended Data Fig. 4n–t). Taken together, our data demonstrate that genetically identical mice at different life stages, though infected with the same pathogen and dose, and showing comparable pathogen burdens, experience distinct disease trajectories leading to death. These divergent courses are marked by opposing patterns of cardiac remodelling, differences in cardiac injury and ventricular stretch (possibly suggestive of cardiac failure), evidence of differential renal damage as well as differences in sickness and death kinetics.
Age-specific sepsis survival paths
The distinct routes to death suggested that what is important for survival of the LD50 would be different for aged and young mice. LD50-challenged young survivors were able to maintain their body temperature and did not show clinical signs of disease over the course of the infection, which is indicative of an endurance phenotype1 (Fig. 1c and Extended Data Figs. 1z–dd and 2a–c). By contrast, aged LD50-challenged survivors showed a resilience health trajectory1, with a sickness phase characterized by morbidity and hypothermia that plateaued at roughly 6–8 h postinfection, and a return to baseline health between 24 h and 48 h postinfection (Fig. 1d and Extended Data Figs. 1z–dd and 2d–f). These differences in health trajectories were independent of pathogen burdens (Extended Data Fig. 3a–o), indicating that the presence of a sickness phase in aged survivors was not due to impaired resistance defences.
Aged surviving mice were protected from both kidney and liver damage, whereas young survivors were protected from kidney damage, but showed elevated amounts of AST and a trend towards elevated ALT suggesting they experienced liver damage (Fig. 1g–j). Young survivors were protected from cardiomegaly, oedema and leucocyte infiltration, and instead showed a trend towards mild cardiac atrophy compared with uninfected young mice (Fig. 1n–p, Extended Data Fig. 4h–j and Supplementary Fig. 1). Aged survivors showed cardiomegaly with histological changes in cardiomyocyte appearance, and without histological changes in oedema or a significant increase in leucocyte infiltration (Fig. 1n–p, Extended Data Fig. 4h–j and Supplementary Fig. 1). RNA-seq analysis revealed the cardiac transcriptomes of the aged and young survivors of the LD50 were distinct. Furthermore, they were different from age-matched LD50-infected dying and uninfected control mice (Extended Data Fig. 4k–m). Gene ontology analysis showed that young survivors had an enrichment of genes involved in the unfolded protein response, amino acid transport, retinoid and/or retinol process and sodium ion processes (Extended Data Fig. 4n,o). By contrast, the aged survivors of the LD50 had enrichment of genes involved in the regulation of collagen fibril organization and biosynthetic processes, wound healing, cellular response to amino acid and response to endoplasmic reticulum stress (Extended Data Fig. 4p). Finally, both aged and young survivors were protected from cardiac damage (Fig. 1k–m). Thus, despite identical pathogen exposure, surviving animals follow distinct, age-dependent survival trajectories.
Foxo1 supports tolerance in young hosts
From our RNA-seq analysis of the heart, we revealed 316 genes that were significantly upregulated in the young LD50-challenged survivor hearts compared with the other 5 animal groups (Fig. 2a,b). Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis of this expression signature showed the enrichment of various metabolic, cell cycle or survival, muscle and hormone signalling processes, with FoxO signalling pathways being the most significantly enriched pathway (Fig. 2c). Whereas cardiac Foxo1 expression and protein concentrations were not different between aged and young animals under uninfected fed and fasted conditions (Supplementary Fig. 1 and Extended Data Fig. 5a–g), we found cardiac Foxo1 was differentially regulated in aged and young hosts when challenged with sepsis and in a fate specific manner. Cardiac Foxo1 expression was significantly higher in young LD50-challenged survivors compared with young uninfected and LD50-challenged dying mice, as well as aged uninfected, dying and surviving mice (Fig. 2d and Extended Data Fig. 5h). Protein concentrations of FoxO1 were also elevated without an increase in the amounts of Thr24 or Ser256 phosphorylation (phosphorylated FoxO1, inactive; unphosphorylated FoxO1, active), indicating there was increased cardiac FoxO1 activity in young surviving mice compared with all other conditions (Supplementary Fig. 1 and Extended Data Fig. 5i–t). Examination of other organs revealed the association between Foxo1 induction and survival in young mice to be specific to the heart (Extended Data Fig. 5h). In aged mice, survival of sepsis did not correlate with increased expression or activation of FoxO1 in the heart or any other organ examined (Fig. 2d, Extended Data Fig. 5h–t and Supplementary Fig. 1).
a, Venn diagram of cardiac genes induced in young survivors. b,c, Heat map (b) and top ten enriched KEGG pathways (c) of overlapping genes across all five comparisons in a. One experiment was performed. d, Cardiac Foxo1 expression (10 h young, 24 h aged). Data also in Extended Data Fig. 5h. n shows biologically independent animals from one out of two independent experiments. e, Young mice ±FoxO1 inhibitor and ±infection survival. f, Young infected Foxo1 Myh6cre− and Foxo1 Myh6cre+ littermate survival. n shows biologically independent animals from two out of two independent experiments. g, Representative heart images at 10 h postinfection. Original images in Supplementary Fig. 1. Four independent experiments performed. h, Heart/body weight ratios at 10 h postinfection. i, Infected mice from h normalized to average uninfected. n shows biologically independent weight-matched animals from four out of four independent experiments (h,i). j, Troponin I. k, BNP. l, Galectin-3. m, AST. n, ALT. o, BUN in young mice ±FoxO1 inhibitor and ±infection at 10 h postinfection. n shows biologically independent animals from two out of two (k,l) or three out of three (j,m–o) independent experiments. p, Representative heart images at 10 h postinfection. Original images in Supplementary Fig. 1. Four independent experiments performed. q, Heart/body weight ratios at 10 h postinfection. r, Infected values from q normalized to average uninfected. n shows biologically independent weight-matched animals from four out of four independent experiments (q,r). s, Troponin I. t, BNP. u, Galectin-3. v, AST. w, ALT. x, BUN in Foxo1 Myh6cre− and Foxo1 Myh6cre+ littermates ±infection at 10 h postinfection. n shows biologically independent animals from four out of four independent experiments. Mean ± s.e.m. Two-way ANOVA with Tukey test (d,h,j–n,o,q,s–w), log-rank analysis (e,f), two-tailed unpaired t-test (i,r). Number of biologically independent mice shown in panels. Scale bars, 3 mm (g,p). a.u., arbitrary units; D, dying; I, infected; S, surviving; U, uninfected.
To test the importance of FoxO1 for infection defence in young hosts, we infected young mice with a low dose of polymicrobial sepsis (LD10) and treated with a cell permeable inhibitor of FoxO1 that blocks its transcriptional activity20. Inhibition of FoxO1 activity rendered young mice significantly more susceptible to infection-induced sickness and lethality (Fig. 2e, Extended Data Fig. 6a–c). This increase in susceptibility was specific to the infected state, as uninfected mice treated with the FoxO1 inhibitor survived and showed no changes in their health (Fig. 2e and Extended Data Fig. 6a–c). We next tested the necessity of cardiac Foxo1 for defence against sepsis in young mice using Foxo1 Myh6cre+ that lack Foxo1 in cardiomyocytes. Similar to our results with the inhibitor, Foxo1 Myh6cre+ were more susceptible to a low dose of polymicrobial sepsis compared with Foxo1 Myh6cre− littermates (Fig. 2f and Extended Data Fig. 6d–f). Furthermore, Foxo1 Mckcre+ that lack Foxo1 in cardiomyocytes and myocytes were more susceptible to a low dose of polymicrobial sepsis compared with Foxo1 Mckcre− littermates (Extended Data Fig. 6g–l). We found no differences in S. aureus, E. coli or total pathogen burdens in any organs examined in young infected mice treated with the FoxO1 inhibitor or vehicle, or in Foxo1 Myh6cre+ and Foxo1 Myh6cre− littermates (Extended Data Fig. 6m–r). This demonstrates that cardiac Foxo1 is necessary to promote host–pathogen cooperation rather than host resistance defences.
We found no differences in the circulating cytokine and/or chemokine profile between vehicle mice and FoxO1 inhibitor-treated mice (Extended Data Fig. 6s,t), suggesting that FoxO1 mediated protection is largely independent of the magnitude of the inflammatory response and instead promotes disease tolerance defences. To begin to understand how cardiac Foxo1 in young mice promotes disease tolerance in response to polymicrobial sepsis infection, we asked whether deletion of cardiomyocyte Foxo1 or inhibition of FoxO1 activity rendered mice more susceptible to organ damage. We found that the inhibition of FoxO1 activity in young infected mice led to cardiomegaly characterized by a change in geometrical shape, increased heart weight and histological changes in cardiomyocyte appearance in the myocardium, oedema, cardiac congestion and leucocyte infiltration, as well elevated concentrations of the cardiac markers troponin I, BNP and galectin-3 (Fig. 2g–l, Extended Data Fig. 7a–e and Supplementary Fig. 1). Young FoxO1 inhibitor-treated septic mice did not show elevated amounts of the liver enzymes AST and ALT, but did show elevated concentrations of the renal damage marker BUN; however, this appears to be independent of the infection as uninfected FoxO1 inhibitor-treated mice also showed signs of renal damage (Fig. 2m–o). Finally, young infected mice treated with the FoxO1 inhibitor showed congestion in many organs (Extended Data Fig. 7d–j). Foxo1 Myh6cre+ mice showed similar patterns to infected or inhibitor-treated mice including cardiomegaly, and elevated circulating amounts of cardiac and renal damage markers compared with Foxo1 Myh6cre− littermates (Fig. 2p–x, Extended Data Fig. 7k and Supplementary Fig. 1). Taken together, cardiac FoxO1 is necessary to promote disease tolerance in young hosts challenged with systemic bacterial infection, mediating protection against cardiomegaly, cardiac and renal damage, as well as sickness and death.
Trim63 supports tolerance in young hosts
We examined the cardiac expression profile of FoxO1 target genes in young LD50-challenged mice and age-matched uninfected controls and revealed four classes of genes that show distinct regulation in the survivors, dying and uninfected young mice (Fig. 3a). Within Class 2 were the atrogenes Trim63 and Fbxo32, which encode the muscle specific E3 ubiquitin ligases MuRF1 and Atrogin1, respectively, and are transcriptionally upregulated during muscle remodelling21 (Fig. 3a and Extended Data Fig. 8a,b). We found cardiac Trim63 and Fbxo32 to be differentially regulated in aged and young hosts when challenged with sepsis and in a fate specific manner. Cardiac Trim63 and Fbxo32 expression were significantly induced in the hearts of young LD50-challenged survivors compared with age-matched uninfected mice and LD50-challenged dying mice (Fig. 3b,c and Extended Data Fig. 8c,d). In aged mice, there was no correlation with sepsis outcome and cardiac Trim63 and Fbxo32 expression in LD50-challenged mice (Fig. 3b,c and Extended Data Fig. 8c,d). Treatment of young infected mice with a FoxO1 inhibitor prevented the induction of both cardiac Trim63 and Fbxo32 expression compared with infected vehicle-treated controls, demonstrating that FoxO1 activity is necessary for sepsis-induced expression of these atrogenes in the heart of young hosts (Fig. 3d,e).
a, Heat map of FoxO1 targets in hearts from young uninfected and LD50-challenged mice from RNA-seq analysis presented in Fig. 2. One experiment performed. b,c, Young and aged mice were infected with the LD50 dose of polymicrobial sepsis. Hearts were harvested when dying animals in each age group each maximal morbidity (10 h for young and 24 h for aged); cardiac expression of Trim63 (b) and Fbxo32 (c). Data are also presented in Extended Data Fig. 8c,d. n shows biologically independent animals from one out of two independent experiments. d,e, Cardiac expression of Trim63 (d) and Fbxo32 (e) in young mice ±FoxO1 inhibitor and ±infection. n shows biologically independent animals from two out of two independent experiments. f, Survival of young wild-type and Trim63+/− littermates infected with polymicrobial sepsis (roughly LD25). n shows biologically independent animals from one out of two independent experiments. g–o, Young Trim63+/+ and Trim63+−,−/− littermates were infected with a low dose of polymicrobial sepsis (roughly LD25). Serum and hearts were harvested roughly 10 h postinfection. g, Representative hearts images. Original images are shown in Supplementary Fig. 1. Four independent experiments performed. h, Weights of hearts normalized to body weight. i, Infected from h normalized to uninfected average in h. n shows biologically independent animals from two out of four independent experiments. j–o, Serum concentrations of troponin I (j), BNP (k), galectin-3 (l), BUN (m), ALT (n) and AST (o). For j–o, n shows biologically independent animals from two out of two independent experiments. Mean ± s.e.m. One-way ANOVA with post hoc Tukey test (b,c). Two-way ANOVA with post hoc Tukey test (d,e,h,j–o), log-rank analysis (f). Two-tailed unpaired t-test (i). Number of biologically independent mice shown in panels. Scale bar, 3 mm.
To test the importance of atrogenes for infection defence in young hosts, we generated mice deficient for Trim63. Young Trim63-deficient mice were highly susceptible to morbidity and mortality when challenged with a low dose of polymicrobial sepsis (Fig. 3f and Extended Data Fig. 8e–g). The increased susceptibility of young Trim63-deficient mice was associated with the development of cardiac remodelling characterized by macroscopic and microscopic changes including geometric shape, weight and oedema, as well as elevated concentrations of cardiac, liver and kidney damage markers (Fig. 3g–o, Extended Data Fig. 8h–q and Supplementary Fig. 1). Infected young Trim63 mutant mice also showed more severe congestion in all organs examined (Extended Data Fig. 8h,l–p). As Trim63 is also expressed in skeletal muscle21 (Extended Data Fig. 8a), we examined how Trim63 in skeletal muscle was affected by our sepsis model. Trim63 expression in skeletal muscle did not correlate with survival in young infected mice nor was it regulated by FoxO1 during infection (Extended Data Fig. 9a–d). Consistent with this, the absence of Trim63 had no effect on skeletal muscle size in young infected mice and skeletal muscle wasting did not correlate with fate in LD50-challenged young mice (Extended Data Fig. 9e–l). Thus sepsis-induced FoxO1 regulation of Trim63 and muscle size is specific to the cardiac muscle. Together, these findings identify cardiac FoxO1 as a regulator of Trim63 during sepsis and establish Trim63 as a critical effector of disease tolerance in the host response to systemic bacterial infection.
Foxo1–Trim63 drive sepsis in ageing
We next investigated whether ageing alters the function of Foxo1 and Trim63 in a manner consistent with antagonistic pleiotropy by contributing to infection-related pathology in aged hosts. We found no association between cardiac Foxo1 expression or activity and infection outcome in aged mice using our LD50 sepsis model (Fig. 2d, Extended Data Fig. 5h–t and Supplementary Fig. 1). Genetic deletion of Foxo1 in cardiomyocytes and myocytes (Foxo1 MckCre+) and pharmacological inhibition of FoxO1 activity significantly improved survival and reduced morbidity in aged hosts following sepsis challenge (Fig. 4a and Extended Data Fig. 9m–u). Consistent with our LD50 model, this protection was associated with cardiac remodelling characterized by enlargement of the heart (Fig. 4b–d, Extended Data Fig. 9v–w and Supplementary Fig. 1). Furthermore, Foxo1 deletion in aged mice blunted the infection-induced rise in circulating BNP amounts while having no effect on troponin I and mild effect on galectin-3 amounts during infection (Fig. 4e–g). Foxo1 MckCre+ and Cre− aged littermates showed similar elevated amounts of BUN and creatinine amounts during infection, although the increase in creatinine did not reach statistical significance (Fig. 4h,i), suggesting that renal injury was not substantially mitigated by Foxo1 deletion. In addition, neither genotype showed elevated amounts of ALT or AST, indicating limited hepatic injury under these condition despite systemic infection (Fig. 4j,k).
a, Survival of aged Foxo1 Mckcre− and Foxo1 Mckcre+ littermates infected with polymicrobial sepsis. n shows biologically independent animals from one out of two independent experiments. b, Representative heart images of aged wild-type mice with or without FoxO1 inhibitor and with or without infection. Original images shown in Supplementary Fig. 1. Three independent experiments performed. c, Heart weights of aged wild-type mice ±FoxO1 inhibitor and ±infection. d, Infected heart weights from c normalized to uninfected average from c. n shows biologically independent animals from three out of three independent experiments. e–k, Serum concentrations of troponin I (e), BNP (f), galectin-3 (g), BUN (h), Creatinine (i), ALT (j) and AST (k). n shows biologically independent animals from three out of three experiments for e–k. l, Cardiac Trim63 expression in uninfected and infected aged wild-type mice treated with a FoxO1 inhibitor or vehicle. n shows biologically independent animals from one out of two independent experiments. m, Survival of aged Trim63−/− and Trim63+/+ mice infected with polymicrobial sepsis. n biologically independent animals from five out of five independent experiments. n, Representative heart pictures from uninfected and infected aged Trim63+/+ and Trim63−/− mice. Original images shown in Supplementary Fig. 1. Three independent experiments performed. o, Heart weights of aged uninfected and infected Trim63−/− and Trim63+/+ mice. p, Infected hearts weights from o normalized to uninfected wild-type average in o. n shows biologically independent animals from three out of three independent experiments for o,p. q–w, Serum concentrations of troponin I (q), BNP (r), galectin-3 (s), BUN (t), creatinine (u), ALT (v) and AST (w). n shows biologically independent animals from three out of three independent for q–w. x, Proposed model. Mean ± s.e.m., log-rank analysis (a,m). Two-way ANOVA post hoc Tukey test (c,e–l,o,q–w). Two-tailed unpaired t-test (d). Two-tailed Mann–Whitney test (p). Number of biologically independent mice shown in panels. Scale bar, 3 mm (b,n). Panel x was created using BioRender (https://biorender.com).
Cardiac Trim63 expression was induced by 10 h postinfection in both aged surviving and dying LD50-challenged wild-type mice (Extended Data Fig. 8c) and in a FoxO1-dependent manner (Fig. 4l). Notably, genetic deletion of Trim63 and pharmacological inhibition of MuRF1 activity protected aged mice from sepsis-induced morbidity and mortality (Fig. 4m and Extended Data Fig. 10a–g) opposite to what we observed in young animals. This protection was accompanied by enlarged hearts (Fig. 4n–p, Extended Data Figs. 9v,w and 10h and Supplementary Fig. 1), reduced cardiac leucocyte infiltration, oedema and preservation of cardiomyocyte morphology and appearance (Extended Data Fig. 10i–l). Aged Trim63−/− mice and MuRF1-inhibited wild-type animals did not show changes in troponin I amounts, were protected from elevated amounts of BNP but not from elevated amounts of galectin-3 (Fig. 4q–s and Extended Data Fig. 10m–o). Aged Trim63−/− mice were also protected from renal and hepatic damage during infection indicated by lower BUN, creatinine and ALT, but not AST concentration, whereas inhibition of MuRF1 was not sufficient to protect aged wild-type mice from elevated BUN concentrations (Fig. 4t–w and Extended Data Fig. 10p–s). Inhibition of MuRF1 in aged hosts also protected from spleen and liver congestion caused by infection (Extended Data Fig. 10t–y). Finally, although MuRF1 inhibition protected aged mice from skeletal muscle wasting during infection, this effect was independent of FoxO1 and did not correlate with survival in LD50-challenged mice (Extended Data Figs. 9i–l and 10z–ee). Together, these data demonstrate that, in contrast to their protective roles in young hosts, Foxo1 and Trim63 promote sepsis pathogenesis in aged mice by exacerbating multi-organ damage and mortality suggesting they show antagonistic pleiotropy (Fig. 4x).
Discussion
Host determinants of infectious diseases include resistance defences, collateral damage from the host response, baseline physiological vigour and disease tolerance or anti-virulence mechanisms. The ageing process will influence each of these host factors. Thus, age not only changes whether a host becomes ill but can alter the trajectory of illness. We used an LD50 polymicrobial sepsis model to ask how ageing affects disease tolerance. Although the LD50 and tissue pathogen burdens were comparable between young and aged mice, the dying animals followed distinct clinical courses, indicating that age-dependent differences in host physiology and responses can drive divergent outcomes. These results show that ageing does not simply compromise immune resistance but also reshapes the host’s ability to tolerate infection, and that physiological programmes that once protected the young host can lead to maladaptive outcomes in the aged.
Sepsis is defined as a dysregulated host response to infection that leads to multi-organ failure and death14. Although multi-organ failure leads to fatal sepsis, the precise cause of death remains difficult to define due to disease heterogeneity and the complexity of the host responses14. Using a polymicrobial LD50 sepsis model, we found distinct age-dependent disease trajectories. In young mice, mortality was associated with cardiomegaly and elevated cardiac stress and injury markers, consistent with cardiac involvement in human sepsis22,23,24. Young dying mice also showed elevated BUN without elevated creatinine, indicating mild renal involvement and transient hepatic dysfunction. By contrast, aged mice showed the opposite pattern: dying animals developed cardiac atrophy and more pronounced renal failure, whereas survivors showed cardiomegaly and milder cardiac injury. Although we cannot yet determine the precise cause of death in either age group, the divergent organ phenotypes provide insight into age-specific responses to infection. Echocardiographic assessment of cardiac function could help clarify the contribution of heart dysfunction to mortality in future studies.
Cardiac plasticity is the ability of the heart to remodel by either hypertrophy or atrophy, enabling adaptation to physiological and environmental stressors but can also become pathogenic. In our sepsis model, young and aged mice showed opposing, fate specific remodelling phenotypes: young dying mice developed cardiomegaly, whereas aged dying mice showed cardiac atrophy, conversely aged survivors showed cardiomegaly. Although the contribution of these remodelling patterns to sepsis outcomes remains unclear, their divergence suggests there may be age-specific adaptations in cardiac physiology. Hypertrophy can maintain output and contractility under stress but can predispose to fibrosis, inflammation and arrhythmia25,26, whereas atrophy can help preserve metabolic balance can also impair contractility and oxygen delivery27,28,29,30,31. Because cardiovascular structure and haemodynamics change markedly with age32,33, the functional impact of hypertrophy or atrophy during sepsis probably differs across life stages. Indeed, histologic abnormalities including myofibre changes and inflammatory infiltration appeared in hearts from both young and aged dying mice, consistent with previous reports of sepsis-induced myocardial damage. Yet despite, similar microscopic lesions, the macroscopic cardiac responses diverged by age, similar to findings from caecal ligation and puncture and genetic models of cardiac hypertrophy34,35,36,37. Future studies should define how changes in myocyte length and ventricular chamber size contribute to age-specific remodelling and cardiac outcomes during sepsis.
We discovered an age-specific, cardiac Foxo1-dependent disease tolerance mechanism that protects young hosts during systemic bacterial infection. Foxo1 regulates key aspects of heart physiology that defend against stress and prevent pathological hypertrophy through the ubiquitin ligase system38. Protection from sepsis-induced cardiac remodelling, multi-organ damage and mortality in young mice required Trim63, encoding the E3 ubiquitin ligase MuRF1, revealing a role for Foxo1–Trim63 signalling in physiological defence during infection. Whether cardiac Foxo1 and Trim63 are necessary for defence in young septic mice by protecting against cardiac muscle hypertrophy or some other function such as metabolic function remains to be determined39. Also, at present, we do not know whether all protective effects mediated by cardiac Foxo1 are through Trim63. For instance, Adipoq, which encodes Adiponectin was upregulated in the hearts of young survivors and has known cardioprotective, anti-hypertrophic and anti-fibrotic effects40,41,42. Although Trim63-deficient mice showed increased susceptibility, cardiomegaly and elevated cardiac stress markers, they also showed liver injury not seen with Foxo1 deletion or inhibition, possibly reflecting broader systemic roles for Trim63 or higher sepsis severity (LD25 versus LD10).
Therapeutic strategies for infectious diseases have largely focused on resistance-based strategies and blocking pathogenic responses that lead to physiological damage. Our study challenges this view by showing that inhibitors of FoxO1 and MuRF1, although protective in aged hosts, were detrimental in young mice exposed to the same infectious challenge. Although we did not perform the inverse experiment and test whether activation or gain of function of FoxO1 in young mice would be beneficial and possibly detrimental in aged hosts, our data indicate this is a likely outcome. Our findings demonstrate the importance of charting the disease course of different age stages and understanding the health mechanisms that shape those courses. Our findings have important implications for the tailoring of therapy to the age of an infected individual.
Methods
Mice
For LD50 and inhibitor experiments, C57BL/6 female mice from Jackson Laboratories (000664) were used. Young mice were used at 12 weeks of age or roughly 12 weeks of age (12–14 weeks). Aged mice were used at 72–76 weeks of age. Trim63 mutant mice were generated by Cyagen using CRISPR–Cas mediated genome engineering to delete exons 1–8. The mutant line was maintained as heterozygous and Trim63+/− and Trim63−/− were used as Trim63 knockout in our study. The Trim63+/+ generated from the heterozygous breeding scheme were used as wild-type controls. Foxo1 mckcre mice were a generous gift from M. Febbraio (Monash University) and were bred in our colony with the following breeding scheme: Foxo1 mckcre+ × Foxo1 mckcre−. cre+ and cre− littermates were used for experiments. Foxo1 Myh6cre mice were generated from a series of crosses using our floxed Foxo1 mice to B6/FVB-Tg(MyH6-cre_2182Mds/J) from Jackson Laboratories (011038). All experiments were performed in our AAALAC-certified vivarium, with approval from The Salk Institute Animal Care and Use Committee (IACUC). In accordance with our IACUC guidelines, in an effort to reduce the number of animals used for our experiments, where possible and appropriate, experiments were done in parallel and shared controls were used. This is indicated in the figure legends where relevant. Mice were housed with a 12-h light/dark cycle, humidity 30–70% and temperature of 18–26 °C (64–79 °F).
Bacteria
Bacteria used were E. coli O21:H+ (ref. 43) and S. aureus (American Type Culture Collection strain 12600). For a list of complete key resources used in this study, refer to Supplementary Table 1 in the Supplementary Information.
Method details
Culturing E. coli O21:H+ and S. aureus for mouse infection
E. coli O21:H+ was incubated overnight at 37 °C on an eosin methylene blue plate treated with ampicillin sodium salt (1 mg ml−1), vancomycin hydrochloride (0.5 mg ml−1), neomycin sulfate (1 mg ml−1) and metronidazole (1 mg ml−1) antibiotics to grow single colonies. S. aureus was incubated overnight on a Luria-Bertani (LB) plate at 37 °C without antibiotics for colony growth. The next day, a single colony of E. coli O21:H+ was inoculated into LB-AVNM media. A single colony of S. aureus was inoculated into LB without antibiotics. Both cultures were shaken overnight at 37 °C (250 rpm). The following morning, the optical density was measured and an inoculum with a 1:1 mixture of the bacteria was prepped with the appropriate amount of both bacteria in sterile 1× PBS that was used directly for mouse infections.
Mouse infection models
Mice were infected intraperitoneally with the appropriate dose of bacteria and put back into their home cage. Mice were group housed for the experiments. For LD50 experiments the dose of total bacteria used was 1 × 108 CFU. This dose was titrated up or down for low dose and high dose models. Inoculums were serially diluted and plated to confirm the infectious doses. Immediately after infection, food was removed for the first 10–12 h postinfection to control for any potential variations in the sickness-induced anorexic response. Mice were clinically monitored as described below every 2 h postinfection. For some experiments, mice were clinically monitored every 2 h for the first 10–12 h postinfection and then again at 24 h. For experiments involving inhibitors, the details are provided below. Mice that reached clinical endpoints were euthanized according to our animal protocol.
Survival
Mice were clinically monitored as described below every 2 h postinfection. For some experiments, mice were clinically monitored every 2 h for the first 10–12 h postinfection and then again at 24 h. Mice that had to be euthanized because they reached clinical endpoints during the infection, in addition to those that succumb to the infection, were included in our survival counts.
Rectal temperature
Rectal temperatures were taken every 2 h postinfection for the first 10–12 h, and then every 24 h as noted using the Digisense Type J/K/T thermocouple meter. Temperatures are shown as temperatures over time, temperatures at a defined time point or the minimal temperature shown by the animals over the course of the infection. This is indicated in the figure legends. Rectal temperatures were also used to calculate health scores to generate health trajectories.
Grading system for monitoring morbidity
We use the following morbidity scale to quantify the morbidity of mice. Infected mice are clinically assessed using this morbidity scale every 2 h postinfection. For some experiments, mice were clinically monitored every 2 h for the first 10–12 h postinfection and then again at 24 h. Morbidity scores are shown as scores over time, scores at a defined time point or the maximal morbidity shown during the infection (the lower the score, the greater the morbidity). This is indicated in the figure legends. Morbidity scores were also used to calculate health scores to generate health trajectories.
(5) Normal. Normal exploratory behaviour, rearing on bind limbs and grooming.
(4) Mild. Reduced exploratory behaviour, rearing on bind limbs and grooming. Slower and/or less stead gait, but free ambulation throughout the cage.
(3) Moderate. Limited voluntary movement. Slow, unsteady gait for less than 5 s.
(2) Severe. No voluntary movement, but mouse can generate slow, unsteady gait for more than 5 s.
(1) Moribund. Mouse does not move away from stimulation by research but can still right itself.
(0) Deceased.
Generating health trajectories
Rectal temperatures were assigned bin scores based on the following strategy. The sum of the temperature bin score (below) and morbidity score (above) for each mouse at each time point was determined to generate health scores. Mice that were deceased or were euthanized because they reached clinical endpoint were assigned a score of 0. Health scores were then plotted against time to generate the health trajectories. Temperature bin scores: (5) 35–>38 °C, (4) 31–34.9 °C; (3) 28–30.9 °C, (2) 25–27.9 °C, (1) 22–24.9 °C and (0) lower than 22 °C.
MRI
An Echo magnetic resonance imaging (MRI) machine and Echo MRI body composition software v.2008.01.18M were used for MRI measurements. The MRI produces a total amount of fat (g), total lean muscle (g) and total water (g) per mouse. Both the total fat and lean were normalized to the total body weight (g) of the mouse.
Heart weights
Weight-matched mice were used for heart weight analyses for young mice and when possible for aged mice. The body cavity was opened, hearts were removed from the body cavity and blood was drained from the chambers. The heart was then placed on a scale and the weight was recorded. Heart weights were normalized to body weights.
Heart pictures
For heart pictures, the body cavity was opened, hearts were removed from the body cavity, blood was drained from the chambers and a picture was taken of the heart with a reference ruler using an iPhone SE 2020 or a rose gold iPhone 11 Promax. For presentation purposes, heart images were scaled, cropped and placed on a uniform background using Photopea as follows: using the reference ruler, each image was scaled to the same size. Hearts were then cropped using a polygonal lasso select tool with 1 pixel feathering and placed on top of an aquamarine background. A 3-mm scale bar is provided in each figure based on the reference ruler in the images. Uncropped images are included in Supplementary Fig. 1.
Histology
Heart, spleen, lungs, liver and kidneys were harvested and fixed in 10% neutral buffered formalin. Lungs were inflated by injecting the lobes with 1× PBS before fixation. Then samples were routinely processed, paraffin embedded, sectioned at 4–5 μm and haematoxylin and eosin stained. Tissues were evaluated by a board certified veterinary pathologist who was blinded to mouse genotype, age and experimental manipulation, and scored semi-quantitatively for the following parameters: oedema (defined as separation of cardiomyocytes by clear to pale eosinophilic material); heart congestion or haemorrhage (defined as blood vessels expanded by erythrocytes in the myocardium and/or extravascular red blood cells); changes to the cardiomyocytes of the myocardium including pallor and/or cytoplasmic vacuolation, disorganization and/or hypereosinophilia with or without increased size, loss of nuclear detail or loss of cross striations; increased leukocytes within the vessels and parenchyma of the heart; spleen congestion (expansion of the red pulp sinuses by erythrocytes); liver congestion or haemorrhage (expanded sinusoids and/or extravascular red blood cells); increased leukocytes within the sinusoids and parenchyma of the liver; congestion and leucocyte infiltration of the lung interstitium; and kidney congestion (expanded vascular spaces in the cortex and medulla). These parameters were scored on a scale of 0–4 with 0 representing normal tissue; 1 representing minimal changes; 2 representing mild changes; 3 representing moderate changes and 4 representing severe changes relative to a score of 1. Representative images were obtained from glass slides using NIS-Elements BR 3.2 64-bit and plated in Adobe Photoshop CC 2015 and/or Adobe Photoshop 2019. Image white balance, lighting and/or contrast was adjusted using corrections applied to the entire image.
Quantification of E. coli O21:H+ and S. aureus in mouse tissues
For quantification of pathogen in tissues, CFUs were quantified. Spleen, kidney, liver, lung and heart were harvested and homogenized in sterile 1× PBS with 1% Triton X-100 using a BeadMill 24 bench-top bead-based homogenizer (Fisher Scientific). Homogenates were serially diluted and plated on LB agar and EMB-AVNM agar and incubated at 37 °C. Colonies were quantified the following day. The limit of detection for the liver was 50 CFU and 100 CFU for all other tissues examined. Any sample with values below the limit of detection are indicated on the plots as ‘X BLD’ where X indicates the number of mice that had values below the limit of detection for that tissue. Data are plotted as geometric mean ± standard deviation as indicated in the figure legends.
Thermal neutrality experiments
The experiment was performed in a temperature-controlled housing unit (thermal cabinet), purchased from Columbus instruments (model no. ENC52). Mice were housed in the thermal cabinet set to 30 °C, 4 days before infection. At the time of infection, mice were fasted and food was given back 24 h later. Mouse weight, temperature and morbidity were measured every 2 h during the first day of infection. Mice were tracked for survival on subsequent days. A set of control mice were infected at room temperature with the same dose, fasted for 24 h, measured every 2 h during the first day and tracked for survival on subsequent days.
FoxO1 inhibitor infection model
The FoxO1 inhibitor AS1842856 (End Millipore) was injected into mice intraperitoneally at a dose of 40–60 mg kg−1 body weight 36 h before infection and immediately after infection on the opposite side of bacterial injection. Control mice were injected with vehicle. Mice were infected, clinically monitored and used for downstream analyses as described throughout the Methods.
MuRF1 inhibitor infection model
Old mice were orally gavaged with 30 mg kg−1 body weight of the MuRF1 inhibitor EMBL (European Molecular Biology Laboratory) CAS 445222-91-3 (Glixx Laboratories) or vehicle as control 12 h before infection. Mice were then infected with polymicrobial sepsis as described above and 2 h postinfection were gavaged with a second dose of the inhibitor or vehicle. Mice were infected, clinically monitored and used for downstream analyses as described throughout methods.
Leg muscle measurements
Mice were euthanized, and the quadricep, tibialis anterior, extensor digitorum longus, soleus and the gastrocnemius were harvested and weighed to determine the mass of each muscle. Muscle weights were then normalized to the body weight and are shown in the figures.
Homeostatic mouse monitoring
For young and old mouse homeostatic mouse data, mice were weighed daily over the course of ten consecutive days for total body weight tracking. Lean mass, fat mass, free water and total water were determined using Echo MRI. For food consumption, mice were single caged for 48 h before food was weighed every 24 h for 2 consecutive days at the same time of day. Food consumption over the 2 days was averaged to report a grams per day average. For rectal temperature, temperature was measured 3 times per day over a consecutive 5-day period at the same time of day using the Digisense Type J/K/T thermocouple meter. The averages of the three time points per mouse per day were reported. For dissections, mice were euthanized, blood was collected by cardiac puncture and serum stored at −80 °C for future analysis as previously described. Liver, kidney, lung, heart and spleen were harvested and weighed to determine the mass of each organ.
Blood pressure
The non-invasive CODA monitor was used to quantify mean blood pressure, systolic blood pressure, and diastolic blood pressure. Mice were placed in a restrainer and warmed on a far infrared warming platform while recording. The average of 10–15 recordings per mouse per parameter was reported.
Troponin ELISA
Serum from mice infected with polymicrobial sepsis was harvested by cardiac puncture. Troponin I amounts were measured using an ultra-sensitive mouse cardiac troponin I enzyme-linked immunosorbent assay (ELISA) kit (Life Diagnostics) according to the manufacturer’s protocol. Plates were read on a VERSAmax microplate reader manufactured by Molecular Devices and data analysis was done using SoftMax Pro v.5.4.
In vivo metabolic measurements
For the glucose tolerance test, mice were fasted overnight (12 h). Tail tips were cut using a razor blade for blood glucose monitoring using the Nova Max Plus glucose monitoring system. Mice were injected intraperitoneally with 2 g kg−1 of glucose (in PBS) based on total body weight. Blood glucose was measured at 0 (before glucose treatment), 15 min, 30 min, 45 min, 60 min, 90 min and 120 min postinjection. Blood glucose was normalized to percent of time 0, and plotted in Prism where the area under the curve was calculated.
Quantification of whole body metabolic parameters
For quantification of VO2, mice were single caged in metabolic cages within a comprehensive laboratory animal monitoring system Oxymax for Windows v.5.64 24 h before metabolic parameter data collection. Mice were left untouched for roughly 3 days for data collection.
qPCR
For hearts, livers, spleens, kidneys and lungs, organs were harvested and then frozen at −80 °C. They were then homogenized into a tissue powder in liquid nitrogen. The powder was used for subsequent RNA extraction, using the Allprep DNA/RNA Mini Kit (Qiagen) as per the manufacturer’s protocol. For leg muscles, the tissues were harvested and then frozen at −80 °C. They were then homogenized into a tissue powder in liquid nitrogen. Ground muscle tissue powder was added to cold screw cap tube with bead in liquid nitrogen and then homogenized with 800 μl of TRIzol LS Reagent (Invitrogen). Then 160 μl of chloroform (TCI) was added and shaken by hand for 15 s and incubated at room temperature for 3 min. The TRIzol/chloroform lysate was transferred to 1.5-ml tubes and spun in a fume hood centrifuge for 15 min at maximum speed (14,000 rpm). The upper aqueous phase was transferred to a new 1.5-ml tube and 400 μl of isopropanol was added and mixed. The samples were held in a −20 °C freezer for a minimum of 2 h. The samples were added to an Allprep RNeasy spin column (Qiagen) and spun at max speed for 1 min, the flow through was discarded. The samples were then processed according to the Allprep RNA (Qiagen) extraction protocol resulting in 60 μl of diluted muscle RNA. Complementary DNA was generated using SuperScript IV Reverse Transcriptase (Invitrogen) as per the manufacturer’s protocol. Two-step quantitative PCR with reverse transcription (RT–qPCR) was performed using a QuantStudio5 Real-Time PCR instrument (Applied Biosystems) and the QuantStudio5 Design and Analysis software v.1.5.0. Primers sequences are listed in Supplementary Table 1 in the Supplementary Information. The annealing temperature used was 60 °C for Trim63, Fbxo32 and Rps17. A temperature of 58 °C was used for Foxo1.
BNP and GAL3 quantification
Blood was collected by cardiac puncture and serum was stored at −80 °C as described above. For analysis, serum was defrosted and BNP and GAL3 were quantified by a BNP (Cusabio) and GAL3 (Invitrogen) ELISA kit. Samples were diluted at 1:2 for BNP, and at 1:100 to 1:300 for GAL3, and the ELISA was run as specified by both manufacturer’s protocols. Plates were read on a VERSAmax microplate reader manufactured by Molecular Devices and data analysis was done using SoftMax Pro v.5.4.
Western blot analysis
Heart tissue powder were homogenized in 700 μl of Tissue Extraction Reagent II supplemented with Protease Inhibitor Cocktail (100:1) using a BeadMill 24 bench-top bead-based homogenizer (Fisher Scientific). Lysates were centrifuged at 4 °C for 15 min at 27,000 rpm and transferred to a new tube. Samples were quantified with a bicinchoninic acid reaction and subjected to western blot analysis of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (14C10) tabbit monoclonal antibody Cell Signaling 21185 (1:1,000), Phospho-FoxOl (Thr24)/FoxO3a (Thr32) antibody Cell Signaling 94645 (1:1,000), Phospho-FoxOI (Ser256) antibody Cell Signaling 9461 (1:1,000), FoxOl (C29H4) rabbit monoclonal antibody Cell Signaling 28805 (1:1,000), anti-rabbit lgG HRP-linked antibody Cell Signaling 70745 (1:3,000). Samples were mixed with 15 μl of a 1:10 mixture of 2-mercaptoethanol to NuPAGE LDS Sample Buffer (4×) (Invitrogen), then incubated at 70 °C for 10 min and sonicated in a water bath for 10 min. Samples were loaded equally in 7% NuPAGE 1.0 mm × 12 well Tris-acetate gels with 4 °C Tris-acetate SDS running buffer (50 ml 20× Tris-acetate SDS in 950 ml deionized H2O) for 60 min at 150 V. Gels were placed on Trans-Blot Turbo Midi 0.2 µm Nitrocellulose Transfer Packs (Bio-Rad) and inserted into Trans-Blot Turbo Transfer System (Bio-Rad) for 10 min. Membranes were stained with Ponceau Total Protein Stain (Prometheus) to be marked, washed with 1× TBST to destain and placed into blocking solution (5% BSA in 1× TBST) on a shaker platform overnight at 4 °C. Blocking solution was removed and primary antibody solutions were added to membranes (1:1,000 antibody in Prometheus OneBlock Western-CL Blocking Buffer) and placed on a shaker platform overnight at 4 °C. Primary antibody solutions were removed and membranes were washed on a shaker platform for 5 min with 1× TBST, which was repeated 4 times at room temperature. Secondary antibody solution was added containing 1:3,000 anti-rabbit IgG HRP-linked antibody in blocking buffer (5% BSA in 1× TBST) and membranes were placed on a shaker platform for 1 h at room temperature. Secondary antibody solutions were removed and membranes were washed again on a shaker platform for 5 min with 1× TBST, which was repeated 4 times at room temperature. Nitrocellulose blots were developed using a mixture of Femto and Dura chemiluminescent reaction and visualized with a Bio-Rad Gel Doc XR+ Gel Documentation System and Image Lab software v.5.2.1. For total FoxO1 and pFoxO1 blots, the same lysate was run on two different gels at the same time in the same gel rig. The gels were then transferred to the same membrane. After staining of the membrane with Ponceau, the membranes were cut and destained. The cut membranes were then probed for the relevant protein. pFoxO1 and total FoxO1 blots had their own GAPDH loading control that was run on the same gel. All uncropped blots are provided in Supplementary Fig. 1. All primary antibodies were obtained from Cell Signaling Technology and have been validated by the manufacturer for the indicated applications and species. Specifically, the following antibodies are validated for immunoblotting in mouse and rat samples according to the manufacturer’s website: GAPDH (14C10) rabbit monoclonal antibody (no. 21185) validated for western blot, mouse, rat and human. Phospho-FoxO1 (Thr24/FoxO3a (Thr32) antibody (no. 94645) validated for western blot, mouse, rat and human. Phospho-FoxO1 (Ser256) antibody (no. 9461) validated for western blot, mouse, rat and human. FoxO1 (C29H4) rabbit monoclonal antibody (no. 28805) validated for western blot, mouse, rat and human. Anti-rabbit IgG, HRP-linked antibody (no. 70745) validated for use as a secondary antibody in western blot applications. All antibodies were used at the dilutions recommended by the manufacturer and produced bands at the expected molecular weights in our samples, consistent with the manufacturer’s validation data.
Serum analysis
Sera harvested by cardiac puncture were analysed by IDEXX Bioanalytics for total bicarbonate, potassium, anion gap, cytokines, AST, ALT, BUN, albumin and creatinine.
RNA-seq and data processing
Total RNA was extracted from heart tissue using the Allprep DNA/RNA Mini Kit (Qiagen) per manufacturer protocol. Libraries were generated using the Illumina TruSeq Stranded messenger RNA Sample Prep Kit following the manufacturer’s instructions (Illumina). Then 75-base pair single-end sequencing was preformed using the Illumina HiSeq 2500 platform. Read quality was assessed using FastQC, version 0.11.5 (Babraham Bioinformatics). Reads were mapped to the mm10 genome using STAR v.2.5.3a (ref. 44). Gene expression levels were quantified across all exons using HOMER v.4.10 (ref. 45). Differential gene expression analysis was carried out by using edgeR v.3.26.7 (refs. 46,47). Results corrected for multiple hypotheses testing using the Benjamini–Hochberg method48. The false discovery date threshold for significance was set at 0.05 or lower and a log2 fold-change of 1 or greater. Heat maps and clustering were performed using R v.3.6.1 (R Core Team) using variance stabilizing transformed counts from DESeq2 v.1.24.0. Graphical packages (pheatmap v.1.0.12, ggplot2 v.3.3.2, gplots v.3.0.1.1, RColorBrewer v.1.1-2, VennDiagram v.1.6.20) were used to visualize data by hierarchical clustering, and generating heat maps, Venn diagrams and principal components analysis plots. Gene Ontology and KEGG analyses were done using DAVID49,50.
Statistics and data presentation
GraphPad Prism 10 for Mac OS X version 10.1.1 and Microsoft Excel v.16.78 were used for graphing of data and statistical analyses. For survival, log-rank analysis was used. D’Agostino and Shapiro–Wilk normality tests were performed on data sets to determine the distribution of the data sets. For pairwise comparisons, unpaired t-test, Mann–Whitney test, one-way analysis of variance (ANOVA) or Kruskal–Wallis test with Tukey or Dunn’s test, or two-way ANOVA with Tukey or Sidak’s multiple comparison were performed. Statistical tests used are noted in the figure legends. In accordance with our IACUC guidelines, in an effort to reduce the number of animals used for our experiments, where possible and appropriate, experiments were done in parallel and shared controls were used. This is indicated in the figure legends where relevant. For some experiments, data are shown in several panels to simplify viewing of different comparisons within the data sets. This is indicated in the figure legends where relevant. Power analyses showed the appropriate sample sizes to use for 80% power to detect an estimated detectable effect size, assuming a 5% significance level and a two-sided test. Allocation was random within sex-matched and age-matched cohorts. For data exclusions, see Fig. 1 for the troponin I measurements. We had to estimate the appropriate amount to dilute samples for the ELISA. Four uninfected young serum samples and one young surviving infected serum sample were diluted below the limit of detection. We did not have enough serum for these samples to rerun at a greater concentration, so we omitted these samples from the analysis. In Extended Data Fig. 4b–f,h–j, a pathologist flagged one young surviving and three young dying slides as having artefacts present that can compromise interpretations and one young surviving and two young dying slides as having severe artefacts present and slides of poor quality, therefore, these were omitted from data plots. In Extended Data Fig. 7h a pathologist flagged one lung from a vehicle surviving mouse as a possible artefact. This sample was omitted. In Extended Data Fig. 8i–p, a pathologist flagged one wild-type infected slide as having artefacts, and it was not scored. We still had sufficient sample sizes even with these samples omitted. Experiments were performed with biologically independent animals for each condition and experiments were repeated independently as indicated in the figure legends. Allocation was random within sex-matched and age-matched cohorts. Within each genotype, mice of the same sex and age were randomly assigned to experimental groups. Littermates were used whenever possible. Independent experimental replicates included mice from different litters and cages and were performed on different days, allowing potential litter, cage or day effects to be identified. Potential confounding factors were controlled through experimental design and replication. Blinding was not possible for animal infection experiments because disease outcomes (for example, survival and visible morbidity) were readily apparent to investigators. For quantitative assays including ELISAs, qPCR, western blots, blood chemistry, histopathology, tissue weights, the individuals performing the assays and/or analyses were blinded to group allocation. Data for ELISAs, qPCRs and blood chemistry were collected by machine-based, unbiased, quantitative means. A single representative experiment or many independent experiments combined are shown in the figures. This is noted in the figure legends.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
