Obesity has become a major public health issue, reaching epidemic proportions worldwide. In 2022, more than 2.5 billion adults were classified as being overweight, including 890 million who had obesity (1). Obesity is a chronic multifactorial disease defined by an abnormal and/or excessive accumulation of adipose tissue that can impair health, and predisposes to numerous other non-communicable diseases, such as type 2 diabetes (T2D), cardiovascular diseases and some cancers (2).

A key feature of obesity is the presence of a chronic low-grade inflammation, which plays a key role in the pathogenesis of T2D (3) and the development of cardiovascular diseases and some cancers (4). Paradoxically, obesity has also been linked to impaired immune responses, as evidenced by higher rates of vaccine failure and complications from infection (5) as well as reduced effectiveness of cancer immuno-surveillance and immunotherapy (6). It has become increasingly clear that obesity-associated systemic metabolic modulations contribute to immune dysfunction by disrupting both intrinsic metabolism and functional capacities of immune cells (6, 7). This concept has been illustrated by the work of L. Lynch’s team, which demonstrated the impact of obesity on peripheral Natural Killer (NK) cells (8). Obesity has also been shown to disrupt several populations of human unconventional T cells, like Mucosal-Associated Invariant T (MAIT), γδ T and invariant Natural Killer T (iNKT) cells (9–15).

iNKT cells are a unique subset of T lymphocytes that co-express receptors of the NK lineage along with an invariant T cell receptor (TCR). In mice, invariant NKT TCR combines the canonical Vα14-Jα18 α-chain paired with a limited set of Vβ-chains (Vβ8.1, Vβ8.2, Vβ8.3, Vβ7, or Vβ2). In humans, TCR variability is even more restricted, as this TCR is composed of a Vα24Jα18 α-chain paired with the unique Vβ11 chain (16, 17). Their TCR recognizes lipid antigens presented through CD1d, a non-polymorphic homolog of class I antigen presenting proteins (16). Upon stimulation with lipid antigens, iNKT cells can rapidly trigger effector functions, orchestrating various immune responses against infections, cellular stresses, or tumors (18). New advances support the existence of an ever-increasing repertoire of lipid antigens that are available for CD1d presentation (19). The prototypical antigen for iNKT cells is the α-galactosylceramide (α-GalCer), a glycosphingolipid (GSL) initially extracted from marine sponges (20). Since then, several glycolipid antigens capable of activating iNKT cells have been identified in environmental bacteria, bacteria in the microbiome and pathogens (21). The fact that iNKT cells undergo thymic selection and respond to non-infectious insults implies that endogenous ligands may also control iNKT cell activation. Although several endogenous lipid antigens have been proposed, their antigenicity remains low (22). Interestingly, an α-GalCer variant, with a slightly modified ceramide lipid structure but similar potency to the canonical α-GalCer antigen, was very recently identified in fetal bovine serum, as well as in other mammalian fluids and tissues (23, 24). Due to their thymic development and positive selection by double-positive thymocytes, iNKT cells exit the thymus as functionally mature cells (25). They respond rapidly to antigenic and/or cytokine stimulation by releasing large amounts of Th1, Th2 or Th17 cytokines (26). In addition, upon TCR or NK receptors activation, iNKT cells can also mediate cytotoxicity via perforin and granzymes release (27). The expression of CD4 on human iNKT cells has been used as a predictor of their functional profile, with CD4⁺ iNKT cells being more prone to produce Th2-type cytokines, whereas CD4⁻ (CD8⁺ or double negative) iNKT cells are more associated with pro-inflammatory cytokines production and cytotoxicity (28).

Some studies have reported reduced frequency of peripheral iNKT cells in individuals with obesity, although some discrepancies remain (11–15). Moreover, the phenotype and activity of peripheral iNKT cells in obesity remain poorly characterized and the origin of their decrease is still unknown. In this context, we conducted this study to fully characterize the profile of peripheral blood iNKT cells from individuals with obesity in comparison to lean individuals, and following bariatric-surgery-induced weight loss. Since iNKT cells TCR recognizes lipid antigens presented through CD1d, we also assessed CD1d expression on peripheral blood antigen presenting cells (APCs).

We highlighted, for the first time, an increase in apoptosis of peripheral blood iNKT cells in individuals with obesity, predominantly affecting the CD4⁻ subset. This phenomenon could contribute to the decrease in absolute counts of iNKT cells observed in obesity. Residual iNKT cells in individuals with obesity show an activated, but not exhausted, profile with impaired Th1 cytokines production in response to PMA/Ionomycin stimulation. Interestingly, the activated profile of iNKT cells is correlated with their functional hypo-responsiveness and these disruptions are significantly improved following bariatric surgery-induced weight loss. To further investigate mechanisms underlying peripheral blood iNKT cells activation in obesity, we assessed CD1d expression on peripheral blood APCs. Our results highlighted a significant upregulation of CD1d on both intermediate and non-classical monocyte subsets in individuals with obesity, which was also reversed after bariatric surgery-induced weight loss. These findings suggest that increased CD1d expression may enhance the presentation of endogenous lipid antigens, thereby contributing to iNKT cell activation in the context of obesity.

Obesity is associated with impaired immune response, as supported by the increased susceptibility to various cancers and a higher risk of complications from infections (5, 6). This immune dysfunction may, in part, results from a negative impact of obesity on immune cell function. In this study, we demonstrated that peripheral blood iNKT cells from individuals with obesity display features of activation, increased apoptosis and functional impairment.

We did not observe a significant decrease in the frequency of iNKT cells among total peripheral blood lymphocytes, but only a trend, in contrast to previous reports (12). This discrepancy could be explained by the significantly higher BMI of the participants included in those studies (> 50 kg/m²) compared to ours (> 30 kg/m²). Absolute iNKT cell counts are also not significantly different between groups; however, a significant decrease was observed specifically within the CD4⁻ iNKT cell subset, whereas CD4⁺ iNKT cells remained unaffected. Consequently, the distribution of these subsets shifted with an increased proportion of CD4⁺ iNKT cells, at the expense of CD4⁻ iNKT cells, in individuals with obesity. The reduction of CD4⁻ iNKT cells in the peripheral blood of individuals with obesity could result from several mechanisms: increased apoptosis, tissue infiltration, or impaired thymic development. However, studies in high fat diet (HFD)-fed mice reported a decrease in iNKT cell frequencies not only in peripheral blood but also at the level of some tissues (e.g., in adipose tissue, spleen and liver) (12), suggesting that tissue infiltration is unlikely to explain our findings. Moreover, it is well-established that some endogenous glycosphingolipids are critical for iNKT cell thymic development (37). Accordingly, elevated circulating lipid levels could theoretically impair iNKT cell development and reduce their peripheral abundance. However, this hypothesis is challenged by some findings showing that both the numbers and maturation markers of thymic iNKT cells remain largely unaltered in hyper-lipidemic (ApoE-/-) mice (38). Instead, our data support the hypothesis that the iNKT cell defect results from changes in the periphery. Indeed, in our study, we demonstrated for the first time that, in individuals with obesity, peripheral blood iNKT cells, particularly CD4⁻ cells, exhibit increased apoptosis, suggesting that apoptotic death may contribute to their loss. Further, residual iNKT cells from individuals with obesity display an activated profile characterized by increased expression of early (CD69) and late (CD25) activation markers, correlated with the TCR co-receptor, NKG2D. Interestingly, the expression of CD25, CD69 and NKG2D inversely correlate with surface TCR levels, further confirming their activated status. Although we assessed CD25 and CD69 expression across the total iNKT cell population, we have not determined whether activation was uniform across all subsets. Nevertheless, NKG2D expression was significantly and selectively upregulated on CD4⁻ iNKT cells, suggesting that TCR activation in this subset might be reinforced through NKG2D engagement. Basal cytokine production (IFN-γ, TNF-α, IL-4) was also significantly and specifically increased in CD4⁻ (CD8⁺) iNKT cells from individuals with obesity compared to lean individuals. Despite this moderate yet significant basal cytokines production, iNKT cells from individuals with obesity display profound functional defect, characterized by a hypo-responsiveness to ex vivo stimulation with PMA/ionomycin, as demonstrated by the marked reduction in Th1 cytokines (IFN-γ, TNF-α) and perforin production. This dysfunction correlates with both increased expression of activation markers and TCR downregulation.

Initially, we hypothesized that iNKT cell activation might lead to exhaustion, characterized by a lack of responsiveness. However, the absence of PD-1 and CTLA-4 upregulation on iNKT cells from individuals with obesity argues against a classical exhaustion phenotype. Previous studies have shown that repeated activation of conventional T cells, such as in chronic viral infections, induces exhaustion (39), while Parekh et al. reported that a single administration of α-GalCer in mice induces a long-term unresponsive anergic iNKT cell phenotype (40). However, this concept of glycolipid-induced anergy was later challenged by Sag et al. (41), who demonstrated that α-GalCer-pretreated iNKT cells actually acquire an active immunoregulatory phenotype characterized by the IL-10 production upon restimulation with α-GalCer, along with a reduced secretion of IFN-γ and IL-4. These IL-10-producing iNKT cells, known as NKT10 cells, represent a distinct regulatory subset, enriched in mouse adipose tissue and also detectable in human PBMCs (42). We did not assess IL-10 production by peripheral blood iNKT cells in our study. However, the phenotype of iNKT cells from C57BL/6 mice intravenously injected with α-GalCer one month earlier appears quite different from that of peripheral blood iNKT cells from obese patients. Notably, these murine iNKT cells show upregulation of CTLA-4 and PD-1 and a downregulation of CD69, CD25 and NKG2D compared to iNKT cells from control mice (41). Further experiments are needed to characterize the phenotype of peripheral blood iNKT cells from patients with obesity, including analyses of proliferation, cytotoxic function, downstream TCR signaling pathways (e.g., Mitogen-Activated Protein Kinase, MAPK) and the expression of anergy-associated genes following α-GalCer stimulation (41). As both activation and exhaustion markers have been defined based on conventional T cells, it is possible that the “dysfunctional” phenotype of peripheral blood iNKT cells from patients with obesity does not align precisely with classical definitions of anergy or exhaustion.

What could drive this activation of peripheral blood iNKT cells in obesity? Direct effects of obesity-related factors (e.g., cytokines, glucose and free fatty acids (FFAs)) or indirect effects via altered self-lipid antigen presentation by CD1d-expressing APCs are both plausible. Lynch’s team previously demonstrated a direct impact of obesity on peripheral blood natural killer (NK) cells involving an intracellular accumulation of lipids impairing NK cells metabolism and therefore their function (8). However, we did not detect increased levels of lipid droplets within iNKT cells of individuals with obesity (data not shown). Furthermore, no correlation was found between blood glucose, FFAs, insulin or CRP levels and iNKT cells parameters. However, BMI correlated positively with CD1d on monocytes (data not shown). Furthermore, blood lipids levels (i.e., total cholesterol, LDL-cholesterol and triglycerides) were associated with basal cytokines production by iNKT cells. In turn, CD1d expression on both NCM and IM correlated with Th1 cytokine levels in iNKT cells, either positively or negatively, depending on whether production was constitutive or induced by PMA/Ionomycin treatment. Based on these preliminary results, we support the hypothesis of an impact of obesity on the presentation of CD1d-loaded lipid antigens by APCs rather than a direct effect on iNKT cells, even if this possibility cannot be excluded.

In peripheral blood, DCs, monocytes and B cells express CD1d and are therefore potential drivers of iNKT cell activation. While no change in CD1d expression was detected on DCs or B cells, a downregulation was observed on CM, whereas IM and NCM showed increased CD1d expression in individuals with obesity. To our knowledge, it is the first study showing CD1d modulation on peripheral monocytes in the obesity context. Indeed, CD1d expression is commonly studied in adipose tissue in the context of obesity. Accordingly, previous studies also revealed an increase in CD1d expression on visceral adipocytes from patients with obesity (43). CM have a very short circulating lifespan (1.0 ± 0.26 days) compared to IM (4.3 ± 0.36 days) and NCM (7.4 ± 0.53 days) (44). Although IM and NCM are much less abundant than CM, their markedly longer lifespan in the bloodstream, their functional characteristics (35, 36), together with CD1d upregulation, could promote their interactions with peripheral blood iNKT cells in obesity.

Like class I and class II MHC, CD1d is constitutively expressed by many cell types and is subject to additional regulation by cytokines (e.g., IFN-γ, TNF-α) or microbial products (45). Interestingly, CD1D gene transcription can also be upregulated under sterile conditions. Indeed, activation of the lipid-activated transcription factor, peroxisome proliferator activated receptor gamma (PPARγ), indirectly regulates CD1d expression via the intracellular production of all-trans retinoic acid (ATRA) and activation of the retinoic acid receptor alpha (RARα) (46). Serum lipids such as lysophosphatidic acid and cardiolipin are PPARγ ligands and upregulate CD1d expression on DCs and their ability to activate iNKT cells (47). Oxidized low-density lipoprotein (Ox-LDL) can also elicit retinoid signaling leading to CD1d upregulation (46). Since circulating Ox-LDL levels are increased in obesity (48), they may represent a relevant source of CD1d regulation in this context. Moreover, when whole blood from healthy donors is incubated with various forms of LDL, Ox-LDL is preferentially taken up by CD16⁺ monocytes (49). This is consistent with our results showing CD1d upregulation specifically on IM and NCM. Finally, the correlation observed in this study between LDL cholesterol levels and basal cytokine production by CD4⁻CD8⁺ iNKT cells, although weak, further supports this hypothesis. Nevertheless, this hypothesis remains theoretical and requires further investigation to be confirmed.

The nature and origin of the self-lipid antigen(s) involved remain unclear. These antigens could already be present in the serum of lean individuals but may be more abundantly loaded onto monocytes CD1d in individuals with obesity, thereby promoting iNKT cells activation. Mammalian cells do not synthesize α-GalCer but rather glycosphingolipids (GSLs) with β-linked sugars, which are weak stimulators of iNKT cells. Over the years, several different self-ligands or autoantigens for iNKT cells have been proposed, including phospholipids. Very recently, using supercritical CO₂ chromatography combined with high-resolution MS/MS, Hosono et al. (24) found α-GalCer compounds (d18:0/16:0) in fetal bovine serum, bovine bile and murine lymphoid organs. These lipids can activate both mouse and human iNKT cell TCR with similar strength to the canonical α-GalCer antigen, despite slight differences in ceramide lipid. Other dihydrosphingosine-based α-GalCer (d18:0) species with various lengths of saturated side chains were also detected in mammals, including in human serum (24). Although antigenic α-GalCer were found in mammalian tissues, their precise origin remains unknown, as no α-specific ceramide galactosyl-transferases have been reported in mammals. Importantly, antigenic α-GalCer detected in this study differ from recently reported antagonistic/weak branched and hydroxyl α-GalCer variants derived from symbiotic bacteria (50, 51). Alternatively, other α-GalCer variants with odd-numbered acyl chains are common in microbes. Accordingly, iNKT cell-activating antigens in the absence of infection could derive from the diet, the microbiome, or others that are truly self. In addition to modulate CD1d expression on peripheral blood monocytes, obesity could also impact the composition and/or levels of these lipid antigens in tissues and serum, contributing to peripheral iNKT cell activation.

In this study, bariatric surgery-induced weight loss and associated improvement in patients’ metabolic and inflammatory status were accompanied by a significant reduction in iNKT cell activation and partial restoration of their responsiveness to PMA/Ionomycin stimulation. The increase in CD1d expression on intermediate monocytes was also reversed. Bariatric surgery significantly decreased circulating lipid levels (triglycerides, total cholesterol, LDL cholesterol) ( Table 2 ) and Ox-LDL levels are also known to decrease significantly after bariatric surgery (48). Our hypothesis implicating Ox-LDL in CD1d upregulation on intermediate monocytes and subsequent iNKT cell activation is consistent with our findings after bariatric surgery. These results, however, do not exclude a role for native LDL in facilitating lipid antigen uptake by APCs, as recently described (52). It was surprising to see that the lean individuals’ data looked more like obese individuals pre-bariatric surgery rather than post-bariatric surgery. Indeed, the expression levels of activation markers (i.e., CD25, CD69, NKG2D) appear lower in obese patients after surgery than in lean individuals. This could be explained by the fact that the lean population included individuals with moderate hypercholesterolemia. Since a positive correlation was found between LDL cholesterol and basal Th1 cytokine production by iNKT cells, we can suggest that the expression of activation markers appears lower in obese patients after surgery because they recovered a better LDL cholesterol level (84.7 mg/dL) than in lean individuals (114 mg/dL) ( Tables 1 , 2 ).

Obesity is also associated with gut dysbiosis, characterized by reduced microbial diversity and increased abundance of both classically pathogenic species and short-chain fatty acid (SCFA)-producing bacteria (53). The gut microbiome is known to influence iNKT cell biology, potentially through producing glycolipid ligands or microbial metabolites such as SCFAs, amino acids, and bile acids (54). Composition of the gut microbiome is altered in patients after bariatric surgery (55–57). These changes persist beyond the initial weight loss period of 12 months (56). Interestingly, bariatric surgery induces an ‘intermediate’ microbiome phenotype with notably increased microbial diversity but not a complete return to a lean-like gut microbiome composition (57). Thus, if dysbiosis contributes to iNKT cell dysfunction in obesity, it is plausible that this alteration may not be fully reversed, or may even be accentuated, after bariatric surgery.

In conclusion, this study demonstrates for the first time that iNKT cell disruption in obesity is characterized by increased apoptosis and functional impairment, predominantly affecting the CD4⁻ iNKT cell subset. This dysfunction could contribute to the loss of immunosurveillance observed in obesity. Further studies are required to confirm the link between iNKT cell disruption and altered lipid antigen presentation by monocytes, and to elucidate the underlying molecular mechanisms. Such insights could help to identify potential therapeutic strategies targeting iNKT cell dysfunction in obesity.