T-cell Metabolism as Interpreted in Obesity-associated Inflammation

Abstract

The appreciation of metabolic regulation of T-cell function has exploded over the past decade, as has our understanding of how inflammation fuels comorbidities of obesity, including type 2 diabetes. The likelihood that obesity fundamentally alters T-cell metabolism and thus chronic obesity-associated inflammation is high, but studies testing causal relationships remain underrepresented. We searched PubMed for key words including mitochondria, obesity, T cell, type 2 diabetes, cristae, fission, fusion, redox, and reactive oxygen species to identify foundational and more recent studies that address these topics or cite foundational work. We investigated primary papers cited by reviews found in these searches and highlighted recent work with >100 citations to illustrate the state of the art in understanding mechanisms that control metabolism and thus function of various T-cell subsets in obesity. However, “popularity” of a paper over the first 5 years after publication cannot assess long-term impact; thus, some likely important work with fewer citations is also highlighted. We feature studies of human cells, supplementing with studies from animal models that suggest future directions for human cell research. This approach identified gaps in the literature that will need to be filled before we can estimate efficacy of mitochondria-targeted drugs in clinical trials to alleviate pathogenesis of obesity-associated inflammation.

Entire volumes have been written on the impact of obesity and type 2 diabetes (T2D) on mitochondrial function in skeletal muscle, adipose tissue, liver, pancreatic β cell, and brain function (1), with more focused reviews highlighting changes in mitochondrial redox balance, autophagy (mitophagy), and calcium flux (2-4) as drivers of widespread pathology. Numerous lines of evidence indicate that chronic inflammation is a factor in many sequelae of obesity, including T2D and other comorbidities that are the ultimate cause of death (5-8), but potential causal relationships between altered mitochondrial function and metabolic-associated inflammation, or “meta-flammation,” remain underexplored. Given the recent explosion in our understanding of how mitochondria control the function of immune cells and thus inflammation outside the context of obesity/T2D, the stage is set for exploring the possibility that obesity-associated changes in mitochondria regulate meta-flammation and thus obesity pathogenesis. Understanding the impacts of obesity on immune cell function and thereby meta-flammation will be critical for designing and/or targeting drugs aimed at slowing, halting, or reversing obesity-associated pathogenesis. We summarize key foundational and late-breaking work aimed at establishing the possibility that ongoing clinical trials of drugs that manipulate T cell mitochondria to treat cancer and autoimmune disease (9, 10) will be useful for normalizing T-cell metabolism towards alleviating obesity/T2D-associated inflammation.

Methods

Searches were done in PubMed English titles/abstracts were assessed for relevance and reference sections were screened for related publications. Search terms included: (“obesity”[MeSH terms] OR “type 2 diabetes”[ALL FIELDS]) AND “T cell”[ALL FIELDS] (1062 results);(“obesity”[MeSH terms] OR “type 2 diabetes”[ALL FIELDS]) AND (“mitochondria”[ALL FIELDS] AND (“reactive oxygen species”[ALL FIELDS] OR “redox”[ALL FIELDS]) (817 results); (“mitochondria”[MeSH terms] or “cristae”[ALL FIELDS] AND (“T cell”[ALL FIELDS] AND “redox”[ALL FIELDS]) (49 results); (“mitochondria”[MeSH terms] or “cristae”[ALL FIELDS]) AND (“obesity”[MeSH terms] OR “type 2 diabetes”[ALL FIELDS]) AND (“Redox”[ALL FIELDS] OR “reactive oxygen species”[ALL FIELDS]) (672 results); (“mitochondria”[MeSH terms] or “cristae”[ALL FIELDS]) AND (“obesity”[MeSH terms] OR “type 2 diabetes”[ALL FIELDS]) AND (“fission”[ALL FIELDS] OR “fusion”[ALL FIELDS]) (147 results). Papers on type 1 diabetes, polycystic ovarian syndrome, case reports, and most cancers were excluded.

Mitochondria Fuel T-cell Function in Health

The importance of metabolism in immune cell function has been appreciated for more than half a century, but the concepts of metabolic reprogramming/exhaustion and their promise for regulating any number of inflammatory processes has skyrocketed. In addition to generating energy, metabolic reprogramming produces intermediates for nucleic acid and protein synthesis, as well as signal transduction and epigenetic regulation (reviewed in (11)). Newer data have refined the groundbreaking discovery that T cells switch from mitochondrial oxidative phosphorylation (OXPHOS) to nonoxidative metabolism upon stimulation (12). The field now posits that more subtle shifts in T-cell mitochondrial function combine with quantitatively large changes in nonmitochondrial glucose metabolism, which may in part be overemphasized by in vitro culture conditions and/or biological traits of the cell donor (13, 14). The understanding that specific fuels guide the balance among possible contributions of T cells to inflammation was developed more recently and has been refined by indications that designing in vitro studies to mimic physiological (in vivo) conditions will be critical for the field moving forward (13). Examples include the requirement of lipids for IL-17⁺Th17 cells (15, 16) and γδ T cells (17), and glucose as essential for murine Th1 and interferon γ ⁺ γδ (IFNγ ⁺ γδ) T cells (12, 17). Ketone bodies, added to cells from healthy humans in vitro or through a ketogenic diet, increased production of cytokines predominantly made by CD4+ and CD8+ (rather than myeloid) cells, but only increased mitochondrial function, as measured by oxygen consumption, in CD8+ (not CD4⁺) T cells (18). These data suggest unsurprisingly complex relationships between differences in mitochondrial function and inflammatory outcomes from a T-cell perspective, consistent with studies showing bioinformatics was required to identify small differences in mitochondrial mass of human CD4⁺T-cell subsets, which in turn associated with functional differences (15). Experimentally blocking mitochondrial glutamine metabolism prevented Th17-associated inflammatory disease, and increased Tbet expression and downstream function of Tbet-positive Th1 cells, although the latter became exhausted (eg, unable to respond to stimuli) in the longer term (19). Arguably the most comprehensive analysis of the role T-cell metabolism plays in T-cell function, as indicated by “effector” vs “memory” fate, used CD8⁺T cells from lean/healthy mice. This landmark work set the bar for similar analyses in T cells from people by showing that metabolic reprogramming, as characterized by mitochondrial fission/fusion, cristae remodeling, and aerobic glycolysis vs fatty acid oxidation determines naïve CD8⁺T-cell fate (20). Independent work showed that deleting protein atrophy 1p (Opa1), the inner mitochondrial membrane protein that regulates cristae structure and OXPHOS, in developing T cells affected mature T-cell responses that are temporally far downstream. Bioenergetic failure of memory T cells and less long-term immune protection were noted in T cell-specific Opa1-deficient mice with CD4⁺and CD8⁺mature T cells similarly impacted (21). Blocking mitochondrial pyruvate uptake by CD8⁺T cells promoted memory cell over effector cell differentiation (22), adding evidence that T-cell mitochondria regulate function over both acute and extended periods.

Independent analyses showed that differences in immune cell physiology can also be regulated by feeding electrons into the electron transport chain of mitochondria through complex I vs complex II, as reported for the CD4⁺Th1 subset from mice. Complex I is essential for proliferation and histone modifications during an acute response to stimulation. In contrast, succinate dehydrogenase, a component of both the tricarboxylic acid cycle and complex II, promotes IFNγ production while blocking proliferation and histone acetylation, which in turn activated T cell-preferential genes (23). IFNγ production by Th1 or CD4⁺regulatory T cells (Tregs) required nonmitochondrial or mitochondrial metabolism, respectively, indicating that different fuel sources can culminate in activation of the same gene based on an undefined partnership between CD4⁺subset-specific factors and the metabolic machinery (24, 25).

The field has known for a decade that Tregs are fueled at least in part by mitochondrial OXPHOS (12, 26), as bolstered by the demonstration that oleic acid improved suppressive function of Tregs taken from patients with autoimmune multiple sclerosis (27). Detailed analysis showed that mitochondrial complex III is specifically required for suppressive function, though not proliferation or survival, of mouse Tregs (28). Perhaps paradoxically, reactive oxygen species (ROS) from complex III is also required for non-Treg CD4⁺activation in the absence of glucose, with mitochondria-targeted antioxidants attenuating IL-2 production (29). Regardless, tissue location may furthermore determine whether Tregs preferentially generate ATP from mitochondrial or nonmitochondrial metabolic pathways (30). The Treg-associated transcription factor Forkhead box protein p3 (Foxp3), which plays important roles in Treg suppression and anti-inflammatory cytokine production, is also required for Tregs to oxidize fatty acids, in part because of its ability to increase production of multiple electron transport chain (ETC) proteins, although outcomes differed somewhat between Foxp3-expressing T-cell lines and bona fide Tregs. Notably, fatty acid oxidation by Tregs was important to prevent lipotoxicity, albeit outside the context of obesity (31). Independent work supported these findings by demonstrating that nutrient-induced activation of mechanistic target of rapamycin complex 1 promoted Foxp3 expression and mitochondrial OXPHOS in parallel. However, inappropriate mechanistic target of rapamycin complex 1 activation and unrestrained glycolysis decreased Foxp3 and lowered suppressive activity (32), consistent with the concept of signal strength regulating T-cell function. Treg metabolism is also under the regulation of acyl-CoA synthetase bubblegum family member 1 (Acsbg1), an acyl-CoA synthetase preferentially expressed by Tregs. Deletion of Acsbg1 decreased Treg mitochondrial membrane potential, OXPHOS, and biogenesis, and was essential for Tregs to proliferate in the lung. Although Acsbg1-dependent regulation of the fatty-acyl-CoA biosynthetic pathway in airway inflammation was also reported by these authors (33), the involvement of Acsbg1 in obesity-associated inflammation remains unknown.

Taken together, these data support the conclusion that differences in metabolism, including differences in mitochondrial fuels and ETC function, regulate T-cell biology and thus inflammation through subset-preferential mechanisms (Fig. 1).

Figure 1.

Multiple mechanisms that change in obesity regulate T-cell mitochondrial function and thus T cell contributions to peripheral inflammation (bottom left). Although T cells have functions specific to their resident tissue, recirculation of many T-cell subsets through AT and direct comparison of human blood and AT indicates overlaps between tissue-associated and circulating T cells. (A) Different components of the mitochondrial electron transport chain (ETC; complexes I-V) shuttle electrons into complex IV, whose activity is measured by oxygen consumption rate in extracellular flux (++), to preferentially fuel T-cell proliferation vs cytokine production, as outlined in the main text. ROS is generated by activity of complexes I and III of the ETC. ROS from complex III can also occur from reverse electron flow (RET) as indicated by the dotted/double-headed arrow between complexes I and III. Evidence for roles of complex III in regulatory T cell (Treg) function is indicated by *. The ETC translates changes in (B) mitochondrial fission/fusion, as exemplified by the fission protein Drp1; (C) macroautophagy, as indicated by a mitochondria-loaded autophagosome fusing with a lysosome; and (D) mitochondrial/endoplasmic reticulum contacts (MAMs, which together with protein kinase C/protein kinase C signaling regulate calcium signaling) into changes in T-cell function that control meta-inflammation. Although multiple studies indicate differences in mitochondrial biology discriminate Tregs and non-Tregs (indicated by Th1/2/17), few studies compare these populations from the same individuals in parallel analyses.

Mitochondrial Changes in Obesity

Numerous experimental systems indicate that obesity, in the presence or absence of confounding T2D, profoundly changes mitochondrial structure/function through many mechanisms. These include changes in OXPHOS, fission/fusion, ETC function, mitochondrial and perhaps nonmitochondrial ROS production, mitophagy, mutations in mitochondrial DNA, reductions in enzymatic activities, and calcium handling, among others that have been the subject of numerous recent volumes/reviews (1-4). Adipocyte mitochondria are amongst the best understood in the context of obesity as characterized by decreases in expression of mitochondrial outer and inner membrane proteins, amidst an increase in matrix metabolic pathway proteins in subcutaneous adipose tissue (AT) of obese mice. These changes cause imbalance among pathways that are functionally coupled in the inner membrane and the matrix to fundamentally change mitochondrial function (34). To make analysis even more challenging, many of the proteins that regulate mitochondrial biology are multifunctional. One notable example is dynamin-related protein 1 (Drp1), which functions in fission/fusion, mitochondrial/ER contact sites termed mitochondrial-associated membranes (MAMs), mitophagy, mitochondrial calcium flux, and cristae structure (20, 35-37). Thus, obesity-triggered compromise of a single protein may impact numerous molecular mechanisms that together regulate mitochondrial OXPHOS, tricarboxylic acid metabolite production, and signal transduction pathways through interconnected mechanisms. Obesity-associated alterations of mitochondrial structure, function and turnover are outlined in the 4 subsections that follow and in Fig. 2, with insights derived from multiple tissue types. Although some of these characteristics have been queried in immune cells, much continues to be unknown. Relevance of the following to mechanisms driving meta-inflammation remain a work in progress.

Figure 2.

Prominent mitochondrial changes during obesity: Several processes are altered in cellular mitochondria exposed to obesity. Dysregulation and decline in the function of the electron transfer chain (ETC; complexes I-V in the inner mitochondrial membrane) promotes accumulation of reactive oxygen species (ROS) such as superoxide (O^2-) and hydrogen peroxide (H₂O₂). A decline in the ROS regulatory enzymes such as super oxide dismutase 1 and 2 (SOD1/2) is also responsible for the oxidative insult to the mitochondria that generates mitochondrial ROS (mtROS). Impaired proteostasis promoted by uncoupling of inner membrane and matrix proteins because of structural alterations in the mitochondria is often observed during obesity. ROS also contributes to impaired proteostasis by causing oxidative damage to proteins. Damage to mitochondrial DNA and release of mitochondrial DNA (mt DNA), among other molecules, can lead to inflammasome activation in cells such as monocytes and macrophages. Untoward changes in mitochondrial membrane potential (ΔΨM) and dysregulation of mitochondrial cristae (through Opa1 among others) and recycling mechanism such as mitophagy, mediated by PINK-1 and ubiquitination (Ub) of Parkin compound the damages concurred during obesity. Mitochondrial associated membranes (MAMs), along with proteins like VDAC and MCU contribute through regulating calcium flux.

OXPHOS/ROS

Based on decades of research, ROS is recognized as the common denominator in many diseases, including obesity and T2D. The sources and the kinds of the molecules collectively termed ROS may vary at different stages of obesity, and include mitochondrial and cytosolic ROS, peroxides, and superoxides. It has long been known that mitochondrial OXPHOS is a key source of ROS. OXPHOS is decreased, whereas ROS is increased in AT and beyond to support obesity-associated metabolic decline (38, 39). However, higher OXPHOS amidst similar peroxide (1 species of ROS) was demonstrated in adipocytes from high body mass index (>40) people regardless of concomitant hyperglycemia/hyperinsulinemia (40). Proposed mechanisms that explain this seeming disconnect include parallel reduction in mitochondrial mass and antioxidant enzymes in obese people (41, 42). Lower expression of a subset of ETC genes, some of which play critical roles in oxygen consumption and ROS generation, was noted in visceral adipose tissue of women with T2D (but not obese women without T2D), compared with nonobese/normoglycemic women. Downregulation of ETC components could be recapitulated by treating cells with the T2D-associated cytokine TNFα (43). Preventive administration of ROS scavengers throughout a 12-week high-fat diet reduced weight gain in mice, while concomitantly preventing liver complications and enhancing insulin sensitivity/glucose tolerance. This result requires interpretation in light of the in vitro ability of ROS scavengers to limit superoxide production only modestly in a hepatocyte cell line (44), highlighting tissue-specific outcomes of attempts to manipulate ROS.

Fission/fusion

Compromising mitochondrial fission proteins like Drp1 and mitochondrial fission 1 protein, or fusion proteins like Opa 1, mitofusins (mfns), or membrane associated ring-CH type 5, either accelerate or protect against obesity in mouse models, but in no simple pattern (reviewed in (45)). Discrepancies are likely explained by loss of function of fission/fusion mediators tested in different tissues, including pro-opiomelanocortin neurons, brown AT, and white AT. Drp-1 is perhaps the best understood fission/fusion mediator in obesity. Mdivi-1, a Drp-1 inhibitor, improved metabolism diet-induced obese mice using a treatment (rather than prevention) protocol, and lowered skeletal muscle peroxide amid no changes in OXPHOS or ETC protein expression (46). Given that Mdivi also inhibits complex I (47), direct measures of mitochondrial structure in this approach will be critical to interpret relationships among the many possible actions of this chemical, let alone the multiple functions of Drp1.

Although the animal model studies provide mixed signals regarding mechanisms linking mitochondrial fission/fusion and obesity, findings in people like (1) low mfn2 transcription in obese skeletal muscle; (2) inverse correlation of mfn2 expression with body mass index; and (3) mfn2 mutations associate with AT dysregulation as indicated by lipomatosis (48-50), indicate that relatively universal increases in fused mitochondria may defend leanness, regardless of tissue-specific effects that may become apparent only during clinical trials aimed at fission/fusion protein manipulation.

Mitophagy

Compromised mitophagy has long been linked to obesity pathogenesis, with more recent work emphasizing mechanistic underpinnings of mitophagy defects in liver function (51). However, mitophagy modestly changes in obese compared with lean human muscle (52), and failed to clarify data paradoxically showing increases or decreases in mitophagy markers in skeletal muscle from animal models of obesity/insulin resistance, depending on the study (summarized in (53)). Connection between mitophagy and production of ROS and reactive nitrogen species is strong (recently reviewed in (54, 55)) as are links between mitophagy and mitochondrial fission/fusion (51, 56). However, direct connections between mitophagy and obesity sequalae are not possible given that mitophagy manipulation simultaneously alters nonmitochondrial autophagy and/or other fundamental energy regulators like mTOR. Specific manipulation of mitophagy to establish cause/effect between mitophagy and meta-flammation remains a future goal that will require new reagents and/or more careful analysis of multiple overlapping mediators.

Despite these challenges, a few studies have supported cause/effect relationships between mitophagy and inflammation, raising the possibility that obesity-associated defects in mitophagy impact meta-flammation. Deficiency of the autophagy related protein Atg5 in murine embryonic fibroblasts and macrophages led to accumulation of both healthy and dysfunctional mitochondria, elevated mitochondrial ROS, and increased IFNγ secretion (57). The relationship between inflammation and mitophagy can be further ascertained, albeit indirectly, by observing the effects of inflammasome activators that promote release of mitochondrial DNA and ROS, and in turn induce mitophagy. Parkin (mitophagy)-deficient macrophages tested through this approach accumulated damaged mitochondria and secreted more IL-1β than wild-type cells (58). Comparison of peripheral blood mononuclear cells among subjects that were metabolically healthy/nonobese, metabolically healthy/obese, or metabolically abnormal/diabetic/obese showed that obesity induced mitochondrial ROS but did not further influence mitophagy in metabolically healthy obesity. On the contrary, attenuated mitophagy, abnormalities in mitochondria structure, and oxidative stress were observed in peripheral blood mononuclear cells from metabolically abnormal/diabetes subjects with obesity. These data indicate that maintenance of mitophagy in obesity may be important for restricting oxidative stress and metabolic damage. The authors hypothesized that accumulating obesity-induced oxidative damage to proteins involved in mitophagy explained the development of mitophagy-resistant mitochondria in metabolically abnormal/diabetic obesity (59). These findings suggest the existence of a vicious cycle of obesity-induced oxidative damage, impaired mitophagy, and inflammation, albeit not specific to T cells.

Calcium flux

The mitochondrial calcium (Ca²⁺) uniporter complexes (MCUs) and MAMs are 2 structures that control Ca²⁺flux and therefore play roles in obesity-associated changes in bioenergetics. These complexes coordinate with signaling pathways downstream of surface receptors like the T-cell receptor to activate protein kinase C (Fig. 1D). MCU supports the store-operated calcium entry response following release of Ca²⁺from the endoplasmic reticulum to control multiple inflammatory mediators including the inflammasome, Th17 differentiation, and NF-κB activation (60-62). Ca²⁺also moves into the mitochondria through MAMs, which coordinate essential crosstalk between mitochondria and the endoplasmic reticulum. MAMs are involved in a myriad of inflammatory and cellular regulation mechanisms (eg, inflammasome activation and ROS production), among others (63). Individuals with obesity have more MAM area, likely because obesity causes an increase in the steady-state levels of MAM proteins (64). This increase in MAM formation in obesity has been established as a significant factor in harmful metabolic changes. MAMs also contribute to glucose homeostasis and hepatic insulin resistance, and thus to the development of T2D (65, 66). Although the mechanistic details of MAM alteration during obesity remain in development (64), the intra-MAM distance between the mitochondria and the endoplasmic reticulum increases following chronic high glucose, with protein complexes such as inositol 1,4,5-triphosphate receptors and mitochondria-shaping mitofusins contributing to interorganelle distance and Ca^2 +transport. MAM dysregulation in obesity and T2D not only affects Ca²⁺signaling, but also impacts insulin signaling: some critical signaling proteins for insulin usage are located in the MAM interface or are in close physical approximation with MAMs (64, 67-69). MAMs are important for rapid reprogramming and effector functions of memory CD8⁺T cells through activation-induced engagement of the mTORC2-protein kinase B-glycogen synthase kinase 3β pathway, which was critical for recruitment of hexokinase 1 to the mitochondria via VDAC and downstream IFNγ production. These pathways have not been investigated in CD4⁺T cells, nor the context of obesity (70). Anti-obesity roles for MAMs have also been reported (reviewed in (67)).

In addition to the cross-talk among mitochondrial regulatory mechanisms outlined previously, the important concept of mitochondrial heterogeneity remains largely unexplored in obesity (reviewed in (71, 72)). This area of future work will be especially challenging for analysis of primary cells from people with obesity and/or T2D apart from mitochondrially rich tissues like muscle or neurons.

Roles for T Cells in Meta-flammation

Despite overwhelming evidence that inflammation promotes obesity-associated T2D, the modest efficacy of anti-inflammatory drugs in clinical trials for T2D (73-76) exposes a critical gap in knowledge that prevents translation of our understanding of meta-flammation into clinic practice. This gap may be, in part, an outcome of the prevailing focus on meta-flammation in obese mice, and legitimate disagreements over the differences between murine and human meta-inflammation (77, 78). Mouse models of obesity have revealed critical concepts and actionable information on metabolic sequalae that more closely recapitulate the insulin resistance of obesity-associated prediabetes (rather than T2D) in people, although obese mice have more M1 macrophages and crown-like structures in their AT, absent the extensive AT fibrosis and beta cell failure of human obesity/T2D (79-84). Producing mice more similar to human T2D requires putting 100 mice on an obesogenic diet, then analyzing only the ~10% most metabolically unhealthy mice, a strategy beyond the resources of most researchers that still fails to recapitulate the beta-cell failure of human T2D (79). Recent work unexpectedly concluded that human AT macrophages do not associate with AT insulin resistance in people (85), raising the likelihood that other cellular sources play important roles in meta-flammation in people. We conclude that work supplementing the fundamental concepts generated by macrophage-dominated obesity models is essential for greater clinical impact than has been realized thus far.

Characterization of inflammation in human prediabetes, as a clinically relevant extension of the copious animal model literature, has been historically limited to plasma cytokines (86, 87). This approach has been unable to leverage the rich understanding of peripheral and tissue-associated meta-flammation in mice into strategies that delay T2D in people. Circulating T cells outnumber macrophages and their close cousins, monocytes, as dominant sources of systemic cytokines suggesting that, regardless of the controversy over relationships between human AT inflammation and insulin resistance, T cells fuel systemic inflammation in obesity. Our work, in press at Obesity (Pugh et al), provides the most comprehensive demonstration to date that human T cells are at least as important as monocytes in systemic T2D inflammation. These findings complement work showing that prototypic (CD4⁺) T-cell cytokines are more important than CD8⁺T-cell cytokines for defining “inflammation” in T2D (15, 88), consistent with seminal papers showing T cells are important in mouse meta-flammation (89-91). Other evidence for the importance of T cells in systemic meta-flammation include (1) a similar distribution of T-cell subsets in human blood and AT (92) and (2) evidence that many T-cell subsets recirculate from AT into blood (93). Developing studies suggest roles for each of the regulatory mechanisms that control mitochondria function outlined previously (OXPHOS/ROS; fission/fusion; mitophagy; Ca^2 +flux/MAMs) in T cell-supported meta-flammation in the CD4⁺subset and beyond.

T-cell OXPHOS/ROS/ETC in Obesity

Prediabetes-associated changes in CD4⁺T-cell mitochondria may fuel a unique cytokine profile to support meta-flammation: cells from subjects with prediabetes compared to either leans or subjects with T2D had much higher mitochondrial OXPHOS, amidst similar nonmitochondrial glycolysis. Methods that lowered CD4⁺T-cell OXPHOS also lowered prediabetes-associated inflammation (94). High OXPHOS, which can lead to high ROS in the absence of compensatory antioxidants, can lead to T cells that do not respond to stimulation (eg, exhausted cells), at least in chronic viral infection and tumor models (95). On the flip side, high OXPHOS, in combination with high ROS, can promote age-related inflammation during a phase of human life also characterized by AT redistribution, which together support insulin resistance (14, 96). T cells in human visceral AT have the exhaustion marker programmed cell death protein 1 (PD-1), although PD-1 was not linked to measures of glycemic control or T2D status, and PD-1 blockade failed to impact glycemic control or AT inflammation in parallel mouse studies. Obesity-induced expression of a less understood exhaustion marker B- and T-lymphocyte attenuator in mice was notable, but the possibility that exhausted T cells have changes in OXPHOS and redox balance will require further mining of the single-cell data generated by these investigators, followed by experimental validation (97). Mitochondrial ROS increased in murine CD4⁺T cells in brown but not white AT in response to obesity in somewhat less comprehensive studies (98), consistent with pathogenic/pro-inflammatory changes in T-cell mitochondria in obesity (Fig. 3).

Figure 3.

Mechanisms that regulate T-cell mitochondrial ROS in obesity. Obesity, perhaps through the ability to promote T-cell senescence, promotes mitochondrial changes in T cells that, in combination with noncompensatory upregulation of antioxidants, results in excess reactive oxygen species (ROS). The functional impact of obesity-associated changes outlined in Fig. 2 remain largely untested in T cells from people or animals with obesity. However, multiple approaches indicate molecules like KLF10, DsbA-L, and metformin reinforce “healthy” mitochondrial ROS dynamics, that we speculate may in part be due to the availability of sufficient amounts of antioxidants. The potential of T-cell subset-specific functions of these molecules is represented by KLF10, with obesity-associated actions predominantly in regulatory T cells (Tregs). A more nuanced view is represented by activation of the AMPK-downstream mediator ACC1, which supports (potentially inflammatory) Th17 cells in obesity by regulating fatty acid oxidation, despite the general paradigm that AMPK activation supports metabolic health and thus would be expected to support a less inflammatory milieu.

Other T cell-targeted work showed that Kruppel-like transcription factor 10 (KLF10), which regulates mitochondrial function of Tregs, decreases in murine CD4⁺T cells in response to obesity. KLF10 knockdown in murine CD4⁺T cells reduced energy expenditure and exacerbated weight gain in response to obesity, with the expected glucose intolerance and insulin resistance in these higher weight mice. The concomitant reduction in mitochondrial respiration, nonmitochondrial glycolysis, and expression of ETC complexes I through V was limited to Tregs, with other CD4⁺T cells having nonsignificant changes. Adoptive transfer of WT Tregs into CD4⁺KLF10 KO mice improved circulating lipid profiles, insulin, and C-peptide, and promoted Treg-indicative Foxp3 expression while lowering liver inflammation (99). These data demonstrate that KLF10-mediated changes in T-cell mitochondria can impact responses to obesity through subset-specific mechanisms.

Obesity has also been shown to affect the main evolutionary function of the immune system, protecting the host from pathogens. Obesity-associated changes in CD4⁺T cells from mice included more glucose uptake and mitochondrial OXPHOS, yet reduced pathogen responses (100). These changes contrasted with lower OXPHOS, nonmitochondrial glycolysis and IFNγ production by pulmonary CD8⁺T cells from obese mice (101). Perhaps unexpectedly, obesity-associated changes in CD4⁺T cells were refractory to weight loss that restored leanness and whole-body glycemic control. Metformin treatment lowered T-cell OXPHOS in vitro, and in contrast to weight loss, improved animal survival following influenza infection (100). The mechanism of action for metformin in this context remains a future direction.

As in other cell types, obesity-associated changes in OXPHOS alter T-cell ROS. Metabolic deterioration and systemic inflammation specifically occur as the result of ROS produced during the metabolic switch of T cells that accompanies terminal differentiation/senescence, as evidenced in liver CD8⁺CD28⁻ T cells from aged and high-fat fed mice. CD8⁺CD28^– cells from people with abnormal glucose homeostasis also produced more ROS and resorted to a more glycolytic metabolism (102). This result differed from our demonstration that CD4⁺T cells, either from older subjects or people with prediabetes, have much higher OXPHOS than cells from younger or lean/healthy subjects, respectively (14, 94), perhaps from cell-type specific differences. The authors of the CD8⁺T-cell analysis concluded that T cells with markers of senescence (not measured in our work) associated with exaggerated ROS and glucose dysregulation (102).

Additional molecules link shifts in T-cell mitochondrial function to obesity pathogenesis in mice. These include the mitochondrial chaperone protein disulfide-bond A oxidoreductase-like protein (part of the glutathione S-transferase antioxidant family), which when knocked out in mouse CD3⁺(eg, total) T cells, protects mice from obesity and insulin resistance while reducing IFNγ production. OXPHOS (but not extracellular acidification, an indicator of nonmitochondrial glycolysis), and mitochondrial Ca²⁺are also reduced by this manipulation, whereas mitochondrial fission is increased. The end result of T-cell mitochondrial perturbation through disulfide-bond A oxidoreductase-like protein manipulation was more BAT thermogenesis (98), with detailed analysis of inflammatory outcomes pending.

Metabolic fuel regulators, like the de novo fatty acid synthesis enzyme acetyl-CoA carboxylase 1 (ACC1) are also implicated in T cell-generated meta-flammation. ACC1 supports CD4⁺Th17 cells at the expense of Tregs (103) and is upregulated to increase Th17 inflammation in obese mice (16). These findings raise the possibility that, as seen for CD8⁺memory T cells (104), fatty acid OXPHOS may play important roles in Th17-associated meta-flammation. Our demonstration that experimentally induced defects in mitochondrial import of long-chain fatty acids only support Th17 inflammation when combined with physiologically high amounts of long-chain acyl-carnitine suggests that simply feeding mitochondria an altered lipid balance alone cannot generate a cytokine profile from healthy cells that recapitulates meta-flammation (15).

Mechanisms explained by obesity/T2D-associated changes in mitochondrial function within the (generally) anti-inflammatory Treg population, which decreases in people and animals with meta-flammation (89, 105), are highlighted by seminal work demonstrating that the transcription factor peroxisome proliferator-activated receptor γ plays roles in both AT Tregs and mitochondrial function/turnover/fatty acid oxidation in a variety of cell types (106, 107). New findings expand previous work by showing Tregs influence obesity and insulin resistance via the transcription factor B-lymphocyte induced maturation protein 1 (Blimp-1), which promotes mitochondrial reprogramming and activation of the IL-10 axis in Tregs. These authors concluded that IL-10 from Tregs unexpectedly promoted obesity-associated insulin resistance by lowering energy expenditure and thermogenesis of adipocytes. Knockout approaches showed that Blimp-1-regulated IL-10 prevented beiging of white AT, and Treg-specific deletion of IL-10 or Blimp-1 improved insulin sensitivity. These genetically manipulated mice were protected from insulin resistance, glucose intolerance, and obesity in response to a high-fat diet (108). Clearly, such newer work adds significant nuance to the early models showing fewer Tregs in obese AT promote metabolic decline (89), and may in part explain age-related changes in Tregs that support insulin resistance (109).

Little is known about T-cell mitochondria fission/fusion or Ca flux in obesity. However, hints from findings outlined previously suggest similar queries in T cells from obese/T2D subjects may reveal actionable differences in the mechanistic underpinnings of T-cell contributions to meta-flammation. More comprehensive analyses following genetic manipulations will be critical to generate models that explain seemingly contradictory findings on pro- vs anti-inflammatory T-cell fuels throughout the literature.

Roles of CD4⁺and CD8⁺T-cell cousins in obesity

T-cell subsets beyond plentiful CD4⁺or CD8⁺T cells have been implicated in obesity-associated metabolic decline. These include the more “innate-like” invariant natural killer cells (iNKTs) and MAIT subsets, along with γδ T cells, the latter of which use different gene segments to encode the T-cell receptor compared with their αβ cousins like CD4, CD8, MAIT, and iNKT cells. Although the particulars of metabolic regulation of many of these T-cell types is incomplete, detailed work showed that glycolysis vs OXPHOS preferentially fuels IFNγ- or IL-17-producing γδ T cells, respectively. Furthermore, γδ T cells in the AT of obese mice contained more lipid if they were IL-17⁺, consistent with causality between lipid metabolism and IL-17 production that has been indicated for αβT cells (16, 17). Work on MAIT cells from adults with obesity indicated defects in glucose metabolism and less mTOR1 signaling, culminating in less IFNγ production, which is not mirrored by CD8⁺T cells from the same subjects and thus interpreted as MAIT-cell specific (110). Obesity-associated changes in MAIT function similarly contributed to increased IL-17 production in children (111). However, AT from subjects with obesity had fewer MAIT cells compared with subjects without obesity, and these MAITs produced less IL-10 but more IL-17. Increased MAIT IL-17 production in childhood obesity correlated positively with insulin resistance. Although the mechanisms underlying these changes remain under investigation, evidence that mitochondrial ROS is an important driver of IL-17⁺MAITs in obesity is published (112). Finally, despite significant understanding of roles for iNKT cells in adipose tissue homeostasis (93, 113), hypothesized shifts of iNKT cells from oxidative to nonmitochondrial glycolytic metabolism in obesity, which triggers an increase in IFNγ production (114), await independent confirmation.

Overall, examples cited above indicate the time is ripe for merging our detailed understanding of T-cell metabolism and T-cell mitochondrial function to examine the role mitochondria play in T cell-mediated meta-flammation, to complement ongoing work on monocyte/macrophage metabolism in obesity/T2D. Current and proposed tests for efficacy of mitochondrial-targeted therapies in cancers, autoimmune disease, and traumatic brain injury (9, 10, 22, 115) will set baselines for estimating whether such treatments might provide relief for obesity-associated inflammation and thus insulin resistance/T2D. Testing paradigms established in other cell types that have more mitochondria than primary human T cells will undoubtedly uncover T cell-specific traits that might be exploited to blunt meta-flammation in the endocrinology clinic. The appreciation of overarching metabolic “fingerprints” in cells purified from various tissues amidst stark differences in detailed metabolism of individual cell types within the same tissue provides calls for new investigations that will be essential to predict safety and efficacy of mitochondria or metabolically targeted drugs in clinical trials as the ultimate goal of basic scientific inquiries.

Glossary

Abbreviations

ACC1

acetyl-CoA carboxylase 1

Acsbg1

acetyl-CoA synthetase bubblegum family member 1

AT

adipose tissue

Blimp-1

B-lymphocyte induced maturation protein 1

Drp1

dynamin-related protein 1

Foxp3

forkhead box P3

IFN

interferon

iNKT

invariant natural killer T cell

KFL10

Kruppel-like transcription factor 10

March5

membrane associated ring-CH type 5

MCU

mitochondrial calcium uniporter complex

MAM

mitochondrial-associated membrane

MAIT

mucosal-associated invariant T cells

OXPHOS

oxidative phosphorylation

PD-1

programmed cell death protein 1

Opa 1

protein optic atrophy 1 p

ROS

reactive oxygen species

T2D

type 2 diabetes

Treg

regulatory T cell

Funding

This work was supported by R56AG069685, R01AG079525, UL1TR001998, The Barnstable Brown Diabetes and Obesity Center, and the University of Kentucky College of Medicine to B.S.N. R15AG068957 provided support for L.P.B. and B.S.N. The Pasini Fellowship and Merrimack College provide support for L.P.B. Figures were created in BioRender.

Disclosures

The authors have no disclosures.

Data Availability Statement

No data were generated for this review.