NLRX1 Regulates Effector and Metabolic Functions of CD4⁺ T Cells

Nucleotide oligomerization domain–like receptor X1 (NLRX1) is a pattern recognition receptor and a member of the negative-regulatory subclass of Nod-like receptors (1–3). NLRX1 was linked to the viral response, innate immunity, and downstream effects on NF-κB signaling (4–7). Located on the outer mitochondrial membrane, NLRX1 was shown to influence reactive oxygen species (ROS) production and the proliferation of epithelial cells in the context of colorectal cancer disease models (8–10). As such, NLRX1 is a potential immunoregulatory molecule that integrates immune and metabolic function. However, its role in the adaptive mucosal immune response, specifically CD4⁺ T cells, has not been described.

The differentiation of T cells is an important determinant in the progression of immune-mediated disease and response to infectious disease contributing to exacerbated and sustained inflammatory responses (11–14). As amplifiers or controllers of the immune response, CD4⁺ T cells are highly sensitive and responsive to their environment, with activation occurring through multiple mechanisms, including dendritic cell contact or the cytokine microenvironment (15–17). The activation is paired with a phenotype commitment, albeit a plastic one, with distinct effector (Th1, Th2, Th9, Th17, Th22, T follicular helper [Tfh]) and regulatory (natural regulatory T cells [nTregs], induced Tregs [iTregs], type 1 Tregs [Tr1], T follicular regulatory) behavior (12, 18). Recently, the effect of metabolic pathways on the differentiation and proliferation of T cells arose as a potential factor in these cell phenotype decisions (19–21). In particular, a divide exists in the preferred metabolic activity of effector cells and Tregs (22). Effector T cells display a preference for glycolysis, even in the presence of sufficient oxygen, similar to the Warburg effect described within cancer cells (21). In contrast, Tregs have a lower metabolic rate and produce energy via oxidation (22).

Also contributing to the homeostasis of T cells is signaling derived from immune checkpoint pathways. PD-1 and CTLA-4 signaling are the most prominent and the earliest described of these pathways, although more recently discovered pathways, such as LAG3, TIM3, and TIGIT, displayed similar functionality in terms of suppressive effects on the proliferation, metabolism, and levels of cytokine production (23–25). Although blockade of the immune checkpoint pathways is an emerging cancer therapy, impaired immune checkpoint responses were implicated in many inflammatory and immune-mediated diseases (26–28). PD-1 and PD-L1 polymorphisms were associated with systemic lupus erythematosus, arthritis, multiple sclerosis, and Crohn’s disease (24). Crohn’s disease and ulcerative colitis, the two main clinical manifestations of inflammatory bowel disease (IBD), were mechanistically linked to overly exuberant T cell responses within the gastrointestinal mucosa (29).

In this article, we describe the integrative effects of NLRX1 on immunity and metabolism through the greater proliferation, inflammatory bias, and decreased sensitivity in response to immune checkpoint pathways of CD4⁺ T cells. We use in vitro and in vivo findings to illustrate the implications of NLRX1 deficiency in T cells using three mouse models of IBD.

C57BL/6 wild-type (WT) and NLRX1^−/− mice, ranging from 8 to 10 wk of age, were administered dextran sodium sulfate (DSS) in drinking water for 7 d. Control mice received standard drinking water. All mice were weighed and scored daily. Disease activity index was scored holistically by physical appearance, fecal consistency, rectal bleeding, and weight loss. Mice were euthanized by CO₂ narcosis for sample collection on days 3, 7, and 10 post-DSS initiation. All experimental procedures were approved by the Institutional Animal Care and Use Committee.

CD4⁺ T cells were enriched from WT and NLRX1^−/− donor spleens using an IMag cell separation system. For FACS, cells were labeled with CD45RB, CD4, and CD25 and separated into CD4⁺CD45RB^hiCD62L⁺CD25⁻ cells (naive T cells), CD4⁺CD45RB^hiCD25⁻ cells (effector T cells), and CD4⁺CD45RB^loCD25⁺ cells (Tregs) in a FACSAria cell sorter. Rag2^−/− mice were transferred with 4 × 10⁵ sorted naive CD4⁺ cells from WT or NLRX1^−/− mice for initial studies and with 4 × 10⁵ effector T cells and 1 × 10⁵ sorted Tregs for secondary studies. Mice were weighed and scored on a weekly basis until the development of clinical signs of disease. After development of clinical signs, mice were weighed and scored daily until necropsy. Mice were euthanized for sample collection 8 wk posttransfer.

C. rodentium (strain DBS100) was grown aerobically, at 37°C with 200 rpm shaking, in LB broth within a capped Erlenmeyer flask overnight. An aliquot of bacterial suspension was added to fresh LB broth 5 h prior to challenge to provide bacteria in log-phase growth. Bacterial concentration was adjusted to 5 × 10¹⁰ CFU/ml by spectrophotometer. Mice were given 1 × 10⁹ CFU in LB broth by orogastric gavage. Control mice were given an equal volume of sterile LB broth. Mice were weighed and scored daily postchallenge. Colonization was monitored by feces collection every other day. Mice were euthanized for sample collection at day 12 and week 4 postchallenge.

Naive CD4⁺ T cells were obtained by magnetic sorting of WT and NLRX1^−/− splenocytes. Twelve-well plates were prepared through incubation with anti-CD3 and anti-CD28 Abs for 1 h at 37°C. Cells were prepared in complete IMDM containing 10% FBS, 20 mM HEPES, 55 μM 2-ME, and penicillin/streptomycin. Cells were plated at 500,000 cells per well. Th17 cytokine differentiation mixture was composed of 2.5 ng/ml TGF-β, 25 ng/ml IL-6, 10 μg/ml anti–IL-4, 10 μg/ml anti–IFN-γ, and 50 ng/ml IL-23. iTreg cytokine differentiation mixture was composed of 5 ng/ml TGF-β, 5 ng/ml IL-10, 10 μg/ml anti–IL-4, and 10 μg/ml anti–IFN-γ. Cells were collected for assay after 60 h of culture. Five hours prior to collection, cells were stimulated with PMA/ionomycin and incubated with GolgiStop. Cell proliferation was measured by CFSE staining. A lactate dehydrogenase (LDH) activity kit was used to measure LDH activity. Briefly, cells were homogenized, and the homogenate was assayed using colorimetric detection of the reduction of NAD to NADH. Fatty acid oxidation rates were obtained from measures of ¹⁴CO₂ production and acid soluble metabolites from the oxidation of [1-¹⁴C]palmitic acid, as previously described (30). For measurement of extracellular acidification rate, cells were incubated under CO₂-deficient conditions and mixed with a cell-impermeable pH-sensitive fluorophore. Glucose, carbonyl cyanide-4-phenylhydrazone (FCCP), and 2-deoxy-d-glucose (2DG) were injected into each well at marked intervals and allowed to normalize for 5 min prior to fluorescence lifetime measurement. For measurement of oxygen consumption rate, cells were mixed with MitoXpress Xtra solution, containing a phosphorescent oxygen probe, and overlaid with mineral oil. Glucose oxidase, FCCP, and 2DG were injected into each well at marked intervals and allowed to normalize for 5 min prior to fluorescence lifetime measurement.

Mesenteric lymph nodes and spleen were excised, and single-cell suspensions were prepared and resuspended in PBS with 5% BSA and GolgiStop. Colon sections were incubated in RPMI/FBS buffer (87.5% RPMI 1640, 10% FBS, 2.5% HEPES) containing collagenase (300 U/ml) and DNase (50 U/ml) for 1 h with stirring at 37°C. Immune cells were purified from the cell suspension by Percoll gradient. Cells were labeled with extracellular and intracellular Abs in 96-well plates. A BD LSRII flow cytometer, in combination with FACSDiva software, was used to acquire and analyze flow cytometry data.

Colonic samples were fixed in 10% buffered formalin, embedded in paraffin, processed routinely, and sectioned at 5 μm. Sections were stained with H&E and examined and graded by a board-certified veterinary pathologist using an Olympus microscope and Image-Pro software. Sections were graded (from 0 to 3) for leukocyte infiltration, epithelial erosion, and mucosal thickness.

Total RNA was isolated from mouse colons using a QIAGEN RNA Isolation Mini Kit and used to generate cDNA via the iScript cDNA Synthesis Kit. Standards were produced through PCR on the cDNA with Taq DNA Polymerase (Invitrogen). The amplicon was purified using a MinElute PCR Purification Kit (QIAGEN). Expression levels were obtained through quantitative real-time PCR using a CFX 96 Thermal Cycler and SYBR Green Supermix (both from Bio-Rad). For analysis, the starting amount of target gene cDNA was compared with that of β-actin as a control.

Nuclear and cytoplasmic protein extracts were quantified with the Pierce BCA protein assay. Standardized samples were run on 10% gels, transferred to nitrocellulose membranes, and incubated with target or control Abs overnight at 4°C. Secondary Abs (anti–rabbit-HRP and anti–mouse HRP) were added for 1 h at room temperature, after which the membrane was visualized using Image Lab software.

Data are expressed as mean and SEM. Parametric data were analyzed using ANOVA, followed by the Scheffé multiple-comparisons test. ANOVA was performed using the general linear model procedure of SAS (SAS Institute, Cary, NC). A 2 × 2 factorial arrangement comparing genotype and treatment was used. Statistical significance was determined at p ≤ 0.05.

WT and NLRX1^−/− mice were administered DSS in drinking water over a 7-d period. Significant increases in disease activity index were observed within NLRX1^−/− mice as soon as day 3 of DSS challenge and continued throughout the remainder of the administration period (Fig. 1A). Increased disease severity was paired with increased Th1 and Th17 populations within the colonic lamina propria (Fig. 1B, 1C) and spleen (Fig. 1E, 1F) at days 3 and 7 of DSS challenge. No differences were observed in splenic or lamina propria Tregs throughout the experimental period (Fig. 1D, 1G). A significant increase in TNF-producing CD4⁺ T cells was also observed, but no differences in IL-22–producing or IL-4–producing CD4⁺ T cells were observed between the genotypes (Supplemental Fig. 1). Broad increases in the expression of inflammatory cytokines (IL-17, IFN-γ, and TNF-α) were observed in whole-colon RNA (Fig. 1H–J). Using the cre-lox recombination system, NLRX1^fl/fl; CD4^cre+ (NLRX1^ΔT) mice were generated (Supplemental Fig. 2) and challenged with DSS to validate whether NLRX1 has an intrinsic role within T cells during colitis. In a similar manner to NLRX1^−/− mice, NLRX1^ΔT mice displayed worsened disease activity (Fig. 2A) throughout the time course, culminating in increased colonic lamina propria (Fig. 2B, 2C) and splenic (Fig. 2E, 2F) Th17 and Th1 cells at day 7 of DSS challenge. However, no differences in Tregs were observed at the colonic (Fig. 2D) or splenic (Fig. 2G) level.

To determine the direct effect of the loss of NLRX1 on the differentiation of CD4⁺ T cells, we sorted naive CD4⁺ T cells from the spleens of WT and NLRX1^−/− mice. When exposed to a Th17-promoting cytokine environment (IL-6, TGF-β, IL-23, anti–IL-4, anti–IFN-γ), NLRX1^−/− T cells exhibited a higher rate of differentiation into Th17 cells than did WT T cells (Fig. 3A). In contrast, when exposed to an iTreg-promoting cytokine environment (TGF-β, IL-10, anti–IL-4, anti–IFN-γ), NLRX1^−/− T cells did not behave significantly differently than WT cells with regard to iTreg differentiation (Fig. 3B). However, the rate of iTreg differentiation within Th17-promoting media was significantly lower for NLRX1^−/− T cells (Fig. 3C). The difference was exacerbated by the addition of PD-L1 to the culture media, which significantly increased the population in WT samples but induced no change in NLRX1^−/− samples (Fig. 3C). The proliferation of CD4⁺ T cells was measured in vitro by CFSE staining. NLRX1^−/− Th17-differentiated cells displayed a greater proliferative index compared with WT Th17 cells (Fig. 3D). The administration of PD-L1 decreased the proliferative index to a greater degree in WT cells (Fig. 3D). Increased PD-L1 expression was observed within WT and NLRX1^−/− mice on day 7 of DSS challenge compared with non-DSS controls (Supplemental Fig. 1).

Based on the in vitro proliferation results and the in vivo RNA-sequencing metabolic differences at the colonic level (Supplemental Fig. 3), we sought to determine the effect of NLRX1 on the metabolic processes of T cells. NLRX1 deficiency results in increased expression of Cpt1a, Fabp4, and Glut1 genes that are responsible for the uptake and utilization of glucose and fatty acids (Fig. 4A–C). Similarly, the loss of NLRX1 increases the activity of LDH and the rate of incomplete fatty acid oxidation, a measure of acid soluble metabolites following the oxidation of labeled palmitate, as opposed to complete oxidation to CO₂ (Fig. 4D, 4E). Administration of PD-L1 decreased LDH activity and increased total fatty acid oxidation in WT Th17 cells but it did not result in significant changes in NLRX1^−/− cells (Fig. 4D). Additionally, the rate of total fatty acid oxidation failed to increase following PD-L1 stimulation within NLRX1^−/− cells (Fig. 4F). The increased glycolytic flux in NLRX1^−/− Th17 cells was further confirmed by an increased extracellular acidification rate (Fig. 4G). No consistent differences existed in the oxygen consumption rate in Th17 cells between genotypes (Fig. 4H). Similar differences were observed in CD4⁺ T cells isolated from the colons of mice on day 3 of DSS challenge; however, no differences were observed with regard to extracellular acidification or oxygen-consumption rates within differentiated Tregs (Supplemental Fig. 1). NLRX1^−/− differentiated Th17 cells also displayed increased glycolytic flux compared with WT, as measured by CO₂ production from the metabolism of radiolabeled glucose (Fig. 4I). Direct inhibition of LDH activity, through treatment with sodium oxamate, normalized NLRX1^−/− Th17 differentiation rates and proliferative index to WT levels (Fig. 4J, 4K). The administration of sodium oxamate in vivo resulted in similar decreases in NLRX1^−/− Th17 cell numbers, in addition to Th1 cell numbers, during DSS colitis (Supplemental Fig. 4). Additionally, treatment with metformin, an activator of AMP-activated protein kinase (AMPK), rescued the responsiveness to PD-L1 signaling and decreased LDH activity (Fig. 4L, 4M). Notably, culture within microaerophilic conditions further increased the differences between NLRX1^−/− and WT cells with regard to measures of Th17 differentiation rate and LDH activity (Fig. 4N, 4O). The lack of NLRX1 increased the expression of Hif1a in normoxic and microaerophilic conditions (Fig. 4P).

To determine the impact of effector and regulatory CD4⁺ T cells on experimental IBD, Rag2^−/− mice were administered WT or NLRX1^−/− sorted naive CD4⁺ T cells via i.p. injection (Fig. 5A). Postadministration, increased leukocytic infiltration and epithelial erosion were observed in mice given NLRX1^−/− cells (Fig. 5D, 5G–I). Spleen and colonic lamina propria samples from mice administered NLRX1^−/− cells contained greater percentages and absolute numbers of Th1 and Th17 cells (Fig. 5B, 5C, 5E, 5F). To determine whether the effect of NLRX1 was through effects on effector or regulatory cells, effector cell and Treg subsets were sorted from WT and NLRX1^−/− spleens. Rag2^−/− mice were administered either WT effectors, WT effectors and regulatory cells, WT effectors and NLRX1^−/− regulatory cells, NLRX1^−/− effectors, NLRX1^−/− effectors and regulatory cells, or NLRX1^−/− effectors and WT regulatory cells. Transfers and sorting of WT and NLRX1^−/− occurred concurrently. The administration of NLRX1^−/− effectors accelerated disease onset and worsened overall severity (Fig. 6A). Groups given NLRX1^−/− effector T cells possessed increased numbers of Th1 and Th17 subsets compared with groups given WT effector T cells, suggesting an effector-intrinsic role of NLRX1 (Fig. 6B–G). Groups given NLRX1^−/− Tregs displayed no significant differences in Th1 and Th17 populations compared with corresponding groups given WT Tregs (Fig. 6B–G).

To confirm the findings within an infectious model of colitis, NLRX1^ΔT mice were challenged with C. rodentium. At the peak inflammatory response, NLRX1^ΔT mice had greater numbers of Th1 and Tfh cells in the colonic lamina propria and spleen (Fig. 7A–D). A significant increase in Th17 cell number was observed in the colonic lamina propria (Fig. 7E). In addition, NLRX1^ΔT mice had lower levels of IL-10–producing cells in the colonic lamina propria, including iTreg and Tr1 populations (Fig. 7G, 7H). NLRX1^ΔT samples possessed higher LDH activity than WT samples, as determined from sorted colonic T cells at peak inflammation (Fig. 7I). The expression of Glut1 was also increased in NLRX1^ΔT mice (Fig. 7J). As a result of increased inflammation, NLRX1^ΔT mice had lower C. rodentium burdens at day 12 postinfection, as measured by reisolation from fecal samples (Fig. 7K). Meanwhile, NLRX1^ΔT mice scored higher on histological assessment of leukocytic infiltration and mucosal thickness (Fig. 7L).

NLRX1 is important in controlling the proliferation and differentiation of CD4⁺ T cells, suggesting that it is implicated in T cell–mediated diseases. We present evidence that NLRX1 in T cells may provide protection from IBD, as demonstrated in three disease models spanning chemical-, infectious-, and cellular-induced disease. The loss of NLRX1 leads to the expansion of inflammatory T cell subsets, including Th1, Th17, and Tfh, locally in the colonic lamina propria and systemically within the spleen. Our in vivo and in vitro observations suggest that NLRX1^−/− T cells are biased toward inflammatory phenotypes, increased proliferation, and decreased sensitivity to immune checkpoint pathways.

Notably, the increased proliferation and Th17 differentiation in vitro is paired with an increase in aerobic glycolysis and expression of glucose transporters. The receptor PD-1 was reported to promote immunosuppressive effects through multiple pathways, including switching the metabolic pathways of a cell from a glycolysis-dominated state to a fatty acid utilization state (31, 32). In colonic biopsies from IBD patients, the expression of PD-L1 on intestinal epithelial cells was upregulated to suppress inflammation (33). Meanwhile, the presence of PD-1⁺ T cells was negatively correlated with disease severity (34), and initial testing of PD-L1 treatment showed efficacy in suppressing disease (35). However, the stimulation of PD-1 by PD-L1 does not activate this switch or attenuate the inflammatory phenotype within NLRX1^−/− cells. The activation of AMPK is among the pathways affected by PD-1 signaling and is a well-described factor in the control of cellular metabolic flexibility (32, 36, 37). When AMPK is directly stimulated by metformin, the metabolic and immunologic differences between WT and NLRX1^−/− T cells are abrogated, suggesting that NLRX1 may be a critical element in the linkage between PD-1 and AMPK signaling. In addition to metabolic contributions, PD-1 and other immune checkpoint signaling contributes to the development of regulatory and memory cell types (38–40). Although the deficiency of NLRX1 in Tregs does not impair their function or abundance, the impaired responsiveness to PD-L1 treatment leads to lower levels of iTregs in Th17-differentiating media. We propose that PD-1 signaling may be important in the known plasticity between Th17 cells and iTregs, and the lack of NLRX1 could contribute to the flexibility of cell phenotype, as well as metabolism. The effect of NLRX1 on memory T cell development was not explored but could contribute to the increased reactivity to native commensal bacteria, a known factor in the progression of IBD.

The loss of NLRX1 selectively influences the differentiation and behavior of effector CD4⁺ T cells. When transferred, NLRX1^−/− Tregs retained equal regulatory effects on disease severity, histopathological markers, and expansion of effector populations as their WT counterparts. Importantly, the regulation of metabolism differs in thymus-derived or nTregs and peripheral or iTregs (22). iTregs are dependent on a switch to lipid metabolism, whereas nTregs are balanced in their substrate utilization. Due to a decreased need for metabolic reprogramming, the development of nTregs may be unaffected metabolically by the loss of NLRX1. However, the metabolic switch–dependent iTregs are more likely affected by the loss of NLRX1, particularly with regard to plasticity compared with the Th17 phenotype. In the mixed adoptive transfer, the sorting of already committed Tregs may negate some of the assumed differences in dynamic and flexible Treg commitment resulting from the deficiency of NLRX1. Similarly, in vitro, the differences between WT and NLRX1^−/− Tregs do not result from a de novo differentiation but rather as a result of the persistence and greater abundance of the Treg phenotype in inflammatory and PD-1–induced conditions.

In contrast to Tregs, effector CD4⁺ T cells have an established preference for energy production through the LDH pathway (21). Therefore, the increase in LDH activity in in vitro–differentiated T cells and colonic isolated T cells during peak disease in the absence of NLRX1 is notable. Importantly, the inhibition of LDH abrogates the increased proliferation and Th17 differentiation differences in NLRX1^−/− cells, implying that the activation of LDH activity may not only be a characteristic of effector CD4⁺ T cells but also a contributing factor in their differentiation. Additionally, NLRX1^−/− T cells had a higher rate of incomplete fatty acid oxidation. Rather than being an indicator of oxidative phosphorylation as complete fatty acid oxidation would be, higher rates of incomplete fatty acid oxidation were linked to mitochondrial overload (41) and dysregulated metabolic pathways (42). Although the effect of NLRX1 on LDH may be through multiple mechanisms, a potential contributor is the effect of NLRX1 on ROS production. Previously, NLRX1 was shown to be crucial in the production of ROS (9, 43). The loss of NLRX1 may not allow a cell to correctly sense the oxygen environment or the ROS concentration within the cell, promoting the utilization of both oxidative and anaerobic metabolic pathways in all conditions. The inability to switch metabolic pathways to influence ROS levels may also affect the ability of a cell to respond to canonical death pathways, such as TNF-mediated death, to which NLRX1 was shown to contribute (43).

With the suggestion that an alteration in oxygen sensing may be involved, HIF1α may be implicated downstream of NLRX1 signaling (44). Activation of HIF1α was linked to an increase in LDH activity, as well as to an increase in Th17 differentiation through the activation of IL-23R and RORγt and the inhibition of FOXP3 (45–47). HIF1α is an important transcription factor controlling effector CD4⁺ T cell differentiation and cytokine production. The loss of NLRX1 leads to increased HIF1α expression, and culture of NLRX1^−/− CD4⁺ T cells in microaerophilic conditions further increases the magnitude of the difference between WT and NLRX1^−/− Th17 differentiation. HIF1α is activated by MITF, whereas sumoylation by RanBP2 decreases its transcriptional activity (48, 49). Regulated by ROS production, RanBP2 is also important for the entry of virus into the nucleus, potentially an additional mechanism by which NLRX1 contributes to viral susceptibility (5, 50).

NLRX1 was linked to viral susceptibility and other disease models, such as experimental autoimmune encephalomyelitis and cancer (1, 10, 51). Although these previous studies focused largely on the role of NLRX1 in innate immune cells, neurons, and tumor cells, we provide evidence that NLRX1 is important in the metabolic and proliferative control of effector CD4⁺ T cells. The involvement of effector CD4⁺ T cells in IBD has been well characterized from the increased number, the resistance to apoptosis, and the spontaneous reaction to the commensal microflora (52–54). The evidence provided from multiple models of IBD suggests that NLRX1 joins several other members of the NLR family in the pathogenesis of IBD. Further, the contribution of NLRX1 to effector T cell differentiation merits study in other T cell immune-mediated diseases.

In conclusion, the loss of NLRX1 in T cells promotes increased metabolic activity and a preference for LDH activity. The metabolic preferences are combined with a decreased sensitivity to immunosuppressive checkpoint pathways to provide greater proliferative capabilities and an inflammatory phenotype bias. The immunometabolic dysregulation caused by the loss of NLRX1 may exacerbate disease severity and gut pathology in DSS-induced, adoptive-transfer, and C. rodentium models of IBD. The contribution of NLRX1 deficiency to worsened disease suggests that further work on mechanisms of NLRX1 activation may provide useful therapeutic advances in IBD and other T cell immune-mediated diseases.

The authors have no financial conflicts of interest.