Immunometabolic T cell failure induces multimorbidity and aging

The emerging field of immunometabolism (1–5) is opening up therapeutic opportunities for the treatment not only of inflammatory and autoimmune disorders (6), but also of metabolic diseases and cancer (7–10). However, the utility of targeting immunometabolism to prevent the onset of age-associated diseases and multimorbidity has not been previously addressed.

An age-related decline in mitochondrial function has been observed in various cells and tissues, including T cells (11). To investigate the consequences of T cell metabolic decline in healthy aging, we used Tfam^fl/fl Cd4^Cre mice. Tfam is a nuclear gene encoding mitochondrial transcription factor A (TFAM), which both stabilizes mtDNA and initiates mtDNA replication (12). In Tfam^fl/fl Cd4^Cre mice, Tfam is depleted in both CD4 and CD8 T lymphocytes (13). Lack of Tfam in T cells reduced the numbers of total circulating CD4 and CD8 T cells (fig. S1A). Tfam depletion induced a sharp decrease in their mtDNA content (fig. S1B) and the loss of expression of key electron transport chain components, forcing a metabolic reprogramming towards glycolysis (13) (Fig. 1A-C). T cells from young (2 months old) Tfam^fl/fl Cd4^Cre mice recapitulated features of the mitochondrial dysfunction that appears in aged (22 months old) wild-type mice (Fig. 1A-C). This mitochondrial decline was associated with Th1 cell skewing, characterized by higher secretion of inflammatory type 1 cytokines IFN-γ and TNF-α, and the increased expression of the Th1 cell master regulator T-bet (Fig. 1D-E, fig. S1C). In addition to this pro-inflammatory phenotype, young (2 months old) Tfam^fl/fl Cd4^Cre mice were immunocompromised to a similar extent as old (22 months old) wild-types. We infected Tfam^fl/fl Cd4^Cre mice with ectromelia virus (ECTV), a highly virulent mouse poxvirus that causes a disease similar to human smallpox, and compared them with young or old control mice. Infection with ECTV killed old wild type mice and young Tfam^fl/fl Cd4^Cre mice in the first 10 days of infection, whereas all young controls were able to survive this acute infection (Fig. 1F). Thus, Tfam-deficient T cells recapitulate metabolic, phenotypic and functional features of aged T cells. We observed that Tfam^fl/fl Cd4^Cre mice presented premature inflammaging, with 7-month-old Tfam^fl/fl Cd4^Cre mice having similar serum levels of the inflammaging-associated cytokines IL-6, IFN-γ, and TNF-α to 22-month-old wild types (Fig. 1G, fig. S1D-E). In humans, inflammaging predicts susceptibility to cardiovascular diseases, neurodegeneration, frailty, and multimorbidity (14–16). Tfam^fl/fl Cd4^Cre mice displayed a premature aged appearance from the age of 7 months (Fig. 1H) and developed anemia, kyphosis, and low body weight (Fig. 1I-K, fig. S1F). An additional indicator of aging in Tfam^fl/fl Cd4^Cre mice was a significant thinning of hypodermal fat (Fig. 1L). Consistent with a premature aging phenotype, metabolic cage experiments revealed that Tfam^fl/fl Cd4^Cre mice were less active and slower than controls, despite higher energy expenditure (Fig. 1M, fig. S1G). Importantly, mean lifespan in Tfam^fl/fl Cd4^Cre mice was half that of controls (483 days versus 984 days) (Fig. 1N). Although Tfam deletion in regulatory T cells induces lethal autoimmunity (17), Tfam^fl/fl Cd4^Cre mice showed no differences in the level of serum autoantibodies (fig. S1H).

To evaluate the potential of Tfam^fl/fl Cd4^Cre mice as a model of age-related multimorbidity, we analyzed muscular, cardiovascular, and cognitive fitness. Histological analyses of Tfam^fl/fl Cd4^Cre muscle revealed a reduced fiber diameter (Fig. 2A). Imaging analysis of gastrocnemius muscle after injection of ¹⁸F-fluorodeoxyglucose (¹⁸F-FDG) revealed significantly lower glucose uptake in Tfam^fl/fl Cd4^Cre mice (Fig. 2B). Lower skeletal muscle strength was confirmed by the grip strength test (Fig. 2C). Moreover, Tfam^fl/fl Cd4^Cre muscle upregulated the expression of genes encoding the ubiquitin ligases MuRF-1 and Atrogin-1, as well as inflammatory markers (Fig. 2D, fig. S2A). Tfam^fl/fl Cd4^Cre mice displayed a marked loss of gWAT mass, and smaller adipocytes (Fig. 2E,F). These differences were accompanied by elevated expression levels of the adipose triglyceride lipase (ATGL) and plasma levels of non-esterified fatty acids in Tfam^fl/fl Cd4^Cre, indicating lipolysis induction (Fig. 2G,H). Thus, the loss of T cell immunometabolic control induces sarcopenia and lipolysis. Tfam^fl/fl Cd4^Cre mice showed evidence of cardiac atrophy (Fig. 2I, fig. S2B). Histological and echocardiographic analysis revealed relative reduction in left ventricular thickness, reduced left ventricular diameter and volume, and smaller cardiomyocyte size (Fig. 2J, fig. S2C–F). These features were accompanied by a higher heart rate (Fig. 2K) and the induction of cardiac stress markers Foxo3a and Nppa (fig. S2G). Tfam^fl/fl Cd4^Cre mice displayed diastolic failure characterized by reduced cardiac output, as well as elevated normalized lung weight together with signs of lung congestion on CT scan (Fig. 2L,M, and fig. S2H). Diastolic dysfunction was also evident from left ventricular (LV) relaxation defects detected in the mitral flow pattern (Fig. 2N). Tfam^fl/fl Cd4^Cre mice developed an age-dependent aortic dilation (Fig. 2O) correlated with increased mRNA expression of inducible nitric oxide synthase (Nos2), decreased mRNA expression of the contractile markers as well as decreased blood pressure (Fig. 2P,Q). Consistent with these findings, 77% of Tfam^fl/fl Cd4^Cre mice had aortic regurgitation, compared with 16% of controls. Histological analysis of the aortas allowed us to identify aortic dissections associated with inflammatory foci in 50% of Tfam^fl/fl Cd4^Cre animals (fig. S2I). Thus Tfam^fl/fl Cd4^Cre mice appear to develop cardiac atrophy and overt heart failure, and severe cardiovascular alterations, which can precipitate death. Tfam^fl/fl Cd4^Cre mice also showed signs of neurological disability, including an influx of T cells in the fornix area of the brain and defects in motor coordination in both the rotarod and the tail suspension tests (Fig. 2R-T). Together, these data support a role for T cells beyond host defense (18, 19), and indicate that the metabolic fitness of T cells is critical for organismal homeostasis.

To verify whether this multimorbidity phenotype was due to a mitochondrial failure in T cells, we used an alternative to delete Tfam in T cells. We generated Tfam^fl/fl Lck^Cre mice, in which Cre recombinase is expressed under the control of the lymphocyte protein tyrosine kinase (Lck) promoter. Tfam^fl/fl Lck^Cre mice presented comparable metabolic and functional alterations in T cells, and similar premature age-associated multimorbidity to Tfam^fl/fl Cd4^Cre mice (fig. S3). To identify the molecular mechanism by which T cell metabolic failure drives organismal frailty and multimorbidity, we performed liver transcriptomics. Tfam^fl/fl Cd4^Cre liver showed significant upregulation of genes associated with senescence (fig. S4A). One of the most upregulated transcripts related to senescence was Cdkn1a, which encodes the cyclin inhibitor p21 (q-value = 0.003) (Fig. 3A). Tfam^fl/fl Cd4^Cre mice showed elevated protein levels of senescence markers p21^Waf/Cip1 and p53 in the liver. This elevation of p21 was also observed in the heart, gWAT, and pancreas (Fig. 3B). In addition, activity of senescence-associated β-galactosidase was higher in Tfam^fl/fl Cd4^Cre gWAT and kidney than in controls (Fig. 3C,D, fig. S4B,C). Analysis of Tfam^fl/fl Lck^Cre mice confirmed senescence induction in various tissues (fig. S3M).

Because in vitro incubation of cancer cells with the type 1 cytokines induces senescence (20), we hypothesized that the type 1 cytokines present in Tfam^fl/fl Cd4^Cre mice drive systemic senescence. Incubation of mouse cells with Tfam^fl/fl Cd4^Cre serum or TNF-α was sufficient to increase p21^Waf/Cip1 levels in hepatocytes, and pre-adipocytes, supporting the argument that inflammatory mediators induce senescence and premature aging (Fig. 3E, fig. S4D). To dissect the contribution of TNF-α to the multimorbidity phenotype, we treated Tfam^fl/fl Cd4^Cre mice with the TNF-α inhibitor etanercept. Blocking TNF-α prevented systemic senescence (Fig. 3F,G) and muscle, cardiovascular and cognitive alterations (Fig. 3H-K). Bone marrow (BM) transplantation from Tfam^fl/fl Cd4^Cre mice into T cell-deficient mice (Cd3e^-/-) (fig. S5A,B) recapitulated the type 1 cytokine storm observed in Tfam^fl/fl Cd4^Cre mice (fig. S5C-E) and showed sings of senescence in multiple tissues (Fig. 3L-N). Cd3e^-/- mice reconstituted with Tfam^fl/fl Cd4^Cre BM, but not with control BM, developed a premature aged appearance after 16 weeks (fig. S5F), and reproduced the multimorbidity phenotype (fig. S5G-J). Thus, T cells with mitochondrial dysfunction are able to induce systemic senescence in peripheral tissues via type 1 cytokine storm.

Notably, the senescence observed in Tfam^fl/fl Cd4^Cre mice was accompanied by a low NAD⁺/NADH ratio in peripheral tissues (Fig. 4A, fig. S6A). This was consistent with the reported decline in NAD⁺ levels during aging (21). NAD⁺ is a metabolic cofactor with a critical role in mitochondrial function, and restoring NAD⁺ levels confers protection against age-associated diseases (21–25). Recent data support that supplementation with the NAD⁺ precursor nicotinamide riboside (NR) reduces inflammaging in the elderly (26). Accordingly, we observed that NR decreased the serum levels of TNF-α in aged mice (fig. S6B). To test whether NAD⁺ boosting strategies could prevent tissue damage caused by inflammaging, we treated Tfam^fl/fl Cd4^Cre mice with NR for 10 weeks to restore the NAD⁺/NADH ratio (Fig. 4A,B). Although NR neither restored weight loss nor reversed the Th1 cell phenotype (Fig. 4C), it successfully reversed skeletal muscle wasting and anemia. Furthermore, most of the cardiovascular alterations found in Tfam^fl/fl Cd4^Cre mice were also reversed (Fig. 4D-G). NR treatment also increased physical activity levels of Tfam^fl/fl Cd4^Cre mice and reversed tissue senescence and inflammation (Fig. 4H-K, and fig. S6C). Finally, a principal component analysis of liver transcriptomics classified Tfam^fl/fl Cd4^Cre mice treated with NR closer to controls than to untreated Tfam^fl/fl Cd4^Cre mice (Fig. 4L). IPA analysis linked NR treatment to of pathways associated with lipid metabolism, inflammation, morbidity, and mortality (Fig. 4M and table S1). Thus, NR can reverse transcriptional changes related to aging in Tfam^fl/fl Cd4^Cre mice.

With the continuous extension of life expectancy, there is an urgent need to understand the common molecular pathways by which aging results in a progressively higher susceptibility to diseases (27). Our results indicate that metabolic changes in the immune system promote age-related deterioration in other tissues, leading to multimorbidity and premature death. Dysregulated T cell metabolism triggers a type 1 cytokine storm that induces senescence in several tissues. Furthermore, we have used this novel model of multimorbidity and premature aging to test for drugs that can delay aging signs. We found that NAD⁺ precursors can prevent tissue damage associated with sustained inflammation, supporting their potential for preventing age-associated multimorbidity. Whether targeting senescence or boosting NAD⁺ levels would have beneficial effects beyond age-associated diseases in patients with cachexia or cytokine-release syndrome will require further investigation. Our results place immunometabolism at the crossroads of inflammation, senescence and aging, highlighting its potential as a therapeutic target for delaying aging and aging-associated diseases.