CD38-mediated metabolic reprogramming promotes the stability and suppressive function of regulatory T cells in tumor

Abstract

In the tumor microenvironment (TME), regulatory T cells (T_regs) adapt their metabolism to thrive in low-glucose, high-lactate conditions, but the mechanisms remain unclear. Our study identifies CD38 as a key regulator of this adaptation by depleting nicotinamide adenine dinucleotide (oxidized form) (NAD⁺), redirecting lactate-derived pyruvate toward phosphoenolpyruvate and bypassing the tricarboxylic acid (TCA) cycle. This prevents accumulation of α-ketoglutarate, which destabilizes T_regs by inducing hypermethylation at the Foxp3 locus. Restoring NAD⁺ with nicotinamide mononucleotide reverses this adaptation, pushing T_regs back to the TCA cycle and reducing their suppressive function. In YUMM1.7 melanoma-bearing mice, small-molecule CD38 inhibition selectively destabilizes intratumoral T_regs, sparking robust antitumor immunity. These findings reveal that targeting the CD38-NAD⁺ axis disrupts T_regs metabolic adaptation and offers a strategy to enhance antitumor responses.

Targeting the CD38-NAD⁺ axis disrupts T_regs metabolism in the TME, destabilizes intratumoral T_regs, and boosts antitumor immunity.

INTRODUCTION

Regulatory T cells (T_regs) are crucial for maintaining peripheral tolerance but pose a major challenge in cancer immunotherapy (1, 2). In the tumor microenvironment (TME), T_regs use various suppressive mechanisms to hinder T cell functions essential for tumor control (3, 4). T_regs have been shown to promote hyper-progression of cancer after programmed cell death protein 1 (PD1) therapy in a fraction of patients with gastric cancer due to the amplification of PD1⁺ T_regs, which are highly suppressive (1). While T_regs-depleting approaches lead to bolstering the effector functions of T cells and have generated promising antitumor response, the clinical implication is limited because of the development of severe autoimmunity caused by systemic loss of T_regs (5). Therefore, identifying strategies to selectively target intratumoral T_regs is paramount for successful cancer immunotherapy.

The metabolic adaptability of T cells, enabling cells to adjust and switch between different metabolic pathways in response to changes in the environment or functional requirements, is imperative to evoke the tissue-specific immune response (6). Studies have revealed that, within the complex and often hostile TME, T_regs undergo specific metabolic adaptations to support their survival and function (7–11). It has been shown that, while a low-glucose, high-lactate TME impairs the functionality of effector T cells (T_effs), T_regs adjust their metabolic machinery and nutrient-sensing pathways to use lactate, meeting their bioenergetic demands (7, 8, 12, 13). These adaptations are also found to be crucial for maintaining the cellular identity and suppressive functions of T_regs within the tumor (8). However, the cellular mechanisms that reprogram the metabolic machinery of T_regs, enabling them to adopt metabolic traits that support their functionality and survival in the TME, remain unclear.

Efforts to understand lactate metabolism in intratumoral T_regs using ¹³C₃ l-lactate have revealed an alternative pathway that promotes phosphoenolpyruvate (PEP) production, indicating a shift toward gluconeogenesis (8). PEP, essential for gluconeogenesis, can be derived from oxaloacetate (OAA) either through cytosolic malate dehydrogenase, converting TCA cycle–derived malate to OAA and then to PEP via cytosolic PEP carboxykinase–C (PEPCK-C), or through pyruvate carboxylase (PC) converting mitochondrial pyruvate to OAA, followed by PEP production via mitochondrial PEPCK-M (14, 15). This raises the question of whether intratumoral T_regs prefer one pathway over another in lactate utilization to support their metabolic needs. Such a preference could have major consequences, as promoting the TCA cycle might increase the levels of key intermediate metabolites, including α-ketoglutarate (α-KG), impinging on their suppressive potential and stability (11, 16). A recent study has shown that T_regs within the TME exhibit reduced mitochondrial mass and respiration compared to splenic T_regs, which primarily use oxidative phosphorylation (OXPHOS) to support their suppressive activity (9). Moreover, reduced mitochondrial OXPHOS in intratumoral T_regs has been identified as critical for their suppressive function (9, 11). These findings suggest that intratumoral T_regs in the TME must adapt mechanisms to ensure efficient lactate utilization through the pathway that meets their bioenergetic demands while maintaining stability and functionality. However, the regulatory mechanisms governing this decision-making process in lactate metabolism in T_regs remain poorly understood.

CD38, an ectonucleotidase, has emerged as an important metabolic checkpoint in T cells due to its NADase activity, which regulates intracellular nicotinamide adenine dinucleotide (oxidized form) (NAD⁺) pools (17–19). Studies have shown that elevated CD38 expression on T_effs affects mitochondrial metabolism and, thereby, their functionality at the tumor site (19, 20). Recently, elevated expression of CD38 has been reported in T_regs isolated from patients with different malignancies (21, 22). However, the role of CD38 in the metabolic adaptation and regulation of T_regs functionality at the tumor site remains incompletely understood. Our findings demonstrate that CD38 expression regulates the stability and suppressive activity of T_regs by modulating intracellular NAD⁺ levels. CD38-mediated reduction in NAD⁺ content in T_regs restricts TCA cycle–mediated generation of α-KG, thus preserving the suppressive methylation pattern of the Foxp3 gene. Our observations reveal that the CD38-NAD⁺ axis dictates the lactate flux in T_regs by redirecting its utilization toward the generation of PEP rather than fueling the TCA cycle, thereby supporting T_regs function at the tumor site.

RESULTS

Intratumoral T_regs express CD38 and exhibit distinct transcriptional features

To elucidate the functional relevance of CD38 in intratumoral T_regs, we first examined their abundance at the tumor site. Analysis of tumor tissues and paired peripheral blood samples from patients with breast and muscle-invasive bladder cancer (MIBC) revealed that T_regs infiltration, marked by CD4⁺ and FoxP3⁺ expression, was significantly higher in tumor tissue than in peripheral blood (fig. S1, A and B), as reported (23, 24). However, upon analysis of CD4⁺FoxP3⁺ cells, we noted that the expression of CD38 was more frequent on intratumoral T_regs compared to that on T_regs from peripheral blood (Fig. 1, A and B).

Fig. 1. Intratumoral T_regs display elevated CD38 expression.

(A to E) Frequency of FoxP3⁺CD38⁺ T cells within the T_regs compartment (CD4⁺FoxP3⁺ T cells) was assessed in [(A) and (B)] tumor tissue and paired blood samples from patients bearing (A) breast cancer (Ca) and (B) muscle-invasive bladder cancer (MIBC); [(C) to (E)] nonlymphoid tissues, lymphoid tissues, and tumor tissue from B6 mice bearing 15 days subcutaneously established (C) B16-F10 melanoma, (D) YUMM1.7 melanoma, and (E) EL-4 thymoma. Adjacent bar plots represent cumulative data from 9 (A), 12 (B), 4 (C), 4 (D), and 4 (E) samples. (F to H) The fluorescence intensity of CD38 was assessed in T_regs from (F) B16-F10, (G) YUMM1.7, and (H) EL-4 thymoma bearing mice. The bar diagrams represent cumulative data from four [(F) to (H)] independent experiments. (I) t–Distributed stochastic neighbor embedding (t-SNE) representation highlights the five identified T_regs (CD3⁺CD4⁺FOXP3⁺) subclusters derived from the single-cell RNA sequencing (scRNA-seq) clusters of CD3G⁺ T cells. (J) t-SNE plot showing single-cell transcription levels of key genes, with colors indicating their expression: blue for CD3G and CD4, green for FOXP3, and red for CD38. (K) Pie chart depicting the relative abundance of five identified T_regs clusters across the five breast cancer samples. (L) The heatmap displays the top 100 differentially expressed genes, ranked by log₂ fold change, across all identified T_regs clusters. *P < 0.05; ****P < 0.0001. FMO, fluorescence minus one.

To confirm this observation, we used various preclinical models in which mice were subcutaneously implanted with B16-F10 melanoma, YUMM1.7 melanoma, or EL-4 thymoma. As anticipated, the abundance of T_regs was markedly higher at tumor sites compared to the spleen and draining lymph nodes (DLNs) (fig. S1, C to E). Notably, across all tumor types assessed, CD38-expressing T_regs were significantly enriched at tumor sites compared to those at the spleen and DLN (Fig. 1, C to E). Furthermore, CD38⁺ T_regs were scarce in nonlymphoid tissues such as the skin and liver (fig. S1, F to H, and Fig. 1, C to E). When evaluating CD38 expression levels, we found that intratumoral T_regs exhibited markedly higher CD38 expression than their lymphoid and nonlymphoid counterparts (Fig. 1, F to H). Together, these findings suggest that intratumoral T_regs have a strong propensity to express CD38, in stark contrast to T_regs residing in peripheral tissues, where CD38 expression is markedly lower.

Given the recent studies highlighting high heterogeneity among intratumoral T_regs, we sought to determine the expression pattern of CD38 in these cells and explore the associated transcriptional signature. To achieve this, we analyzed publicly available single-cell RNA sequencing (scRNA-seq) data of tumor-infiltrating lymphocytes from patients with breast cancer (GSE 114727) (25). After applying stringent quality control and filtering, a total of 17,645 CD3⁺ T cells were subjected to unsupervised clustering analysis, identifying 12 distinct T cell clusters (fig. S1I). Among these clusters, clusters 1 and 8 were characterized as T_regs distinguished by their enriched expression of CD3G, CD4, and the subset determining master transcription factor FOXP3 (fig. S1, J and K). To further investigate the transcriptional diversity, we re-clustered 3127 T_regs from clusters 1 and 8, revealing five distinct subclusters of intratumoral T_regs (Fig. 1I). Clusters 0 and 3, which accounted for 36.65 and 12.98% of total T_regs, respectively, were highly enriched in CD38 expression (Fig. 1, J and K). Moreover, genes associated with superior suppressive function and stability in T_regs, including ICOS, ID3, PIM2, TIGIT, MCL1, IL2RA, ISG15, and CREM, were predominant in clusters 0 and 3, which exhibited high CD38 expression (Fig. 1L). Cluster 0 further differed from cluster 3 by expressing additional T cell suppression-associated genes such as RGS1, IFIT1, and CRIP1. In contrast to clusters 0 and 3, the other three clusters (1, 2, and 4) displayed very low CD38 expression. Cluster 2 was characterized as “lymph node–homing T_regs” due to their expression of SELL, CCR7, and KLF2. Clusters 1 and 4 remain undefined but were enriched in the expression of PRMD1, FKBP5, GPX (cluster 1), and BTG2 (cluster 4), genes associated with T_regs function and stability (Fig. 1L). Collectively, these data suggest that a substantial fraction of intratumoral T_regs expresses CD38 and that CD38-expressing T_regs harbor genes that are associated with heightened suppressive function and stability of T_regs.

CD38^Hi T_regs exhibit superior suppressive activity

Having found enriched expression of genes associated with heightened immunosuppression in CD38-expressing T_regs, we next interrogated their suppressive potential. Because of the technical challenges in retrieving sufficient CD38⁺ T_regs from the tumor site for functional studies, we first checked whether in vitro–differentiated T_regs (iT_regs) also expressed CD38. We observed that naïve CD4⁺ T cells differentiated to T_eff barely express CD38, while the frequency of CD38⁺ cells was significantly higher in iT_regs, comprising ~70 to 75% of the iT_regs population (Fig. 2A and fig. S2A). Notably, the frequency of CD38⁺ iT_regs could not be further increased upon iT_regs differentiation in the tumor-conditioned medium (TME) (fig. S2B) or under hypoxic conditions (fig. S2, C and D), suggesting that transforming growth factor–β (TGF-β) is sufficient to induce CD38 expression on iT_regs, as reported earlier (20, 26).

Fig. 2. CD38 endows T_regs with superior suppressive function.

(A) Naïve CD4⁺ T cells from B6 mice were differentiated into T_eff or T_regs, and CD38 expression was assessed. The bar diagram shows cumulative data from 11 experiments. (B) Quantitative polymerase chain reaction (qPCR) analysis of various genes in CD38^Hi and CD38^Lo iT_regs. Data are representative of five independent experiments. (C) CellTrace Violet (CTV)–labeled naïve T cells were activated with anti-CD3/CD28 (1 μg/ml each) in the presence of fluorescence-activated cell sorting (FACS)–sorted CD38^Hi and CD38^Lo iT_regs for 3 days. CD8⁺ T cell proliferation was assessed via CTV dilution. Bar diagram shows data from four experiments. (D) Supernatant from (C) was used to measure interferon-γ (IFN-γ) levels. (E) CD4⁺FoxP3⁺ T cell frequency on day 3 of iT_regs differentiation (left) and after sorting into CD38^Hi and CD38^Lo subsets, followed by culturing in interleukin-2 (IL-2) for 2 days (right). The bar plot represents data from four experiments. (F) FoxP3 expression in CD38^Hi and CD38^Lo iT_regs after culturing in IL-2 for 2 days. The bar plot shows data from four independent experiments. (G) CTV-labeled naïve T cells were activated with anti-CD3/CD28 in the presence of wild-type (WT) and CD38^−/− iT_regs for 3 days. CD8⁺ T cell proliferation was assessed via CTV dilution. Adjacent bar diagram represents data from five independent experiments. n.s., not significant. (H) Supernatant from (G) was used to assess IFN-γ levels. (I) Human iT_regs transduced with either control shRNA or CD38 shRNA were cocultured with CTV-labeled naïve T cells in the presence of anti-CD3/CD28 for 3 days. CD8⁺ T cell proliferation was assessed via CTV dilution. The adjacent bar diagrams represent cumulative data from five independent experiments. *P < 0.05; **P < 0.01; ***P < 0.005; ****P < 0.0001.

Given the expression of CD38 on iT_regs, we examined whether the CD38-expressing iT_regs subset also harbors the suppressive gene signature as observed in intratumoral CD38⁺ T_regs. We sorted iT_regs based on the CD38 expression and analyzed immunosuppressive gene expression in CD38^Hi and CD38^Lo subsets. We found that the expression of Ctla4, Icos, IL2ra, Ikzf2, Nrp1, Entpd1 (gene encodes CD39), and Lag3 genes associated with the immunosuppressive potential of T_regs was markedly up-regulated in CD38^Hi iT_regs compared to CD38^Lo that in iT_regs (Fig. 2B). Next, we subjected the sorted CD38^Hi and CD38^Lo iT_regs to the suppression assay with naïve T cells purified from wild-type (WT) B6 mice. We observed that CD38^Hi iT_regs were superior to CD38^Lo iT_regs in suppressing the proliferation and effector cytokine production of T cells at all the tested ratios (Fig. 2, C and D). Notably, this differential suppressive capacity was not due to differences in viability, as both subsets remained equally viable after the assay (fig. S2E). Instead, the superior function of CD38^Hi iT_regs could partly be attributed to their enhanced retention of FoxP3 expression, as CD38^Lo iT_regs were more prone to losing FoxP3 expression when cultured for extended periods (Fig. 2, E and F).

To further confirm the role of CD38 in regulating the suppressive potential of iT_regs, we differentiated naïve CD4⁺ T cells from both CD38^−/− and WT B6 mice into iT_regs and assessed their suppressive potential. We noted that, although there were no overt differences between CD38^−/− and WT CD4⁺ T cells in their differentiation into iT_regs (fig. S2F), CD38^−/− iT_regs were less effective than WT iT_regs in suppressing the proliferation and effector cytokine production of T cells (Fig. 2, G and H).

Last, we sought to determine whether iT_regs differentiated from human CD4⁺ T cells also express CD38 and whether CD38 contributes to their suppressive function. CD4⁺ T cells purified from the peripheral blood mononuclear cells (PBMCs) of healthy volunteers were differentiated into T_effs or iT_regs (fig. S2G), and CD38 expression was evaluated. Consistent with findings in murine iT_regs, CD38 expression was predominant on human iT_regs compared to that on T_effs (fig. S2H). Short hairpin RNA (shRNA)–mediated knockdown of CD38 in human iT_regs significantly reduced their suppressive function (Fig. 2I and fig. S2I), underscoring the critical role of CD38 in regulating human iT_regs functionality, mirroring the observations in murine iT_regs. Together, these findings suggest that CD38 expression on iT_regs is pivotal for their immunosuppressive potential and aids in maintaining FoxP3 expression, highlighting its critical role in T_regs stability and function.

CD38^Hi iT_regs display restricted mitochondrial activity, essential for their suppressive function

Given the metabolic heterogeneity in iT_regs and its impact on their functional traits, we investigated whether CD38^Hi and CD38^Lo iT_regs are metabolically distinct. We found that glucose metabolism, assessed by measuring the extracellular acidification rate (ECAR), a measure of glycolysis, was significantly higher in CD38^Lo iT_regs compared to that in CD38^Hi iT_regs (Fig. 3, A to C). However, the expression of genes encoding various glycolytic enzymes was comparable between CD38^Lo and CD38^Hi iT_regs, except for Ldha, which showed significant up-regulation in CD38^Hi iT_regs (fig. S3A). The data suggest that glycolytic enzyme levels were less likely to account for differential glucose metabolism in these cells, as reported recently in T_regs subsets having discrete proclivities for glucose metabolism (8). CD38^Hi iT_regs are highly reminiscent of low-glucose–avid T_regs that have been reported to have superior suppressive function and FoxP3 stability (8).

Fig. 3. CD38 expression restricts the mitochondrial metabolism in Treg.

(A to G) iT_regs were sorted into CD38^Hi and CD38^Lo subsets and were used to determine: (A) extracellular acidification rate (ECAR) in response to glucose, oligomycin, and 2DG; (B) glycolysis rate after glucose addition; (C) glycolytic capacity; (D) oxygen consumption rate (OCR) under basal condition and in response to different inhibitors as indicated; (E) basal respiratory capacity; (F) maximal respiratory capacity; and (G) spare respiratory capacity. Data are representative of four [(A) to (G)] independent experiments. (H to K) WT and CD38^−/− iT_regs were assessed for (H) OCR under basal condition and in response to the indicated inhibitors, (I) basal respiratory capacity, (J) maximal respiratory capacity, and (K) spare respiratory capacity. Data are representative of four [(H) to (K)] independent experiments. (L and M) iT_regs were differentiated in the presence or absence of rotenone (Rote; 1 μM, added at 48 hours) and antimycin A (Anti; 1 μM, added at 48 hours). After washing, they were cocultured with CTV-labeled T cells in the presence of a T cell activation cocktail (anti-CD3/CD28) for 3 days. (L) CD8⁺ T cell proliferation was assessed via CTV dilution. The adjacent bar diagram shows data from four independent experiments. (M) Supernatant from (L) was used to measure IFN-γ levels. (N to R) Mass spectrometry (MS)–based determination of indicated intracellular metabolites in iT_regs differentiated in the presence or absence of rotenone (Rote; 1 μM, added at 48 hours). Data are representative of four independent experiments. *P < 0.05; **P < 0.01; ***P < 0.005; ****P < 0.0001.

Because oxidative metabolism was found to be critical in regulating T_regs function (27, 28), we next determined the OXPHOS in CD38^Hi and CD38^Lo iT_regs by measuring their oxygen consumption rate (OCR). Contrary to earlier studies (9, 10, 27), we noted that CD38^Hi iT_regs, despite having high suppressive potential, exhibited decreased mitochondrial respiration compared to CD38^Lo iT_regs, indicating reduced mitochondrial activity in CD38^Hi iT_regs (Fig. 3, D to G). Further analysis revealed that the observed reduction in mitochondrial OXPHOS in CD38^Hi iT_regs was not due to the decreased expression of the electron transport chain (ETC) complexes (fig. S3B). Next, to confirm the role of CD38 in affecting mitochondrial metabolism in CD38^Hi iT_regs, we measured OXPHOS in iT_regs differentiated from either WT or CD38^−/− B6 mice. We observed that iT_regs differentiated from CD38^−/− CD4⁺ T cells had higher mitochondrial respiration compared to WT iT_regs (Fig. 3, H to K), suggesting that CD38 restricts mitochondrial activity in iT_regs. These observations highlight a complex metabolic regulation in iT_regs exerted by CD38 that limits both glycolysis and oxidative metabolism. Yet, the relevance of these metabolic features to the suppressive function of CD38^Hi iT_regs was unclear.

Glycolysis and OXPHOS are coupled through the oxidation of pyruvate and its subsequent incorporation into the TCA cycle, generating reducing equivalents [reduced form of NAD⁺ (NADH) and flavin adenine dinucleotide (FADH2)] for oxidation-reduction reactions in the ETC. Any impediment restricting the utilization of pyruvate in the TCA cycle and subsequently generating reducing equivalents for ETC could impinge upon the whole metabolic circuit. We argued that reduced metabolic activity in CD38^Hi iT_regs could be due to their restricted mitochondrial activity, which might be important for their heightened suppressive capacity. To test this possibility, we took an indirect approach where different complexes of ETC were inhibited using pharmacological inhibitors, and the relevance of mitochondrial metabolism in regulating the suppressive capacity of T_regs was assessed. Inhibition of either complex I or complex III of the ETC using rotenone or antimycin A, respectively, at 48 hours of iT_regs differentiation, did not affect the viability or FoxP3 expression in iT_regs (fig. S3, C and D). Intriguingly, iT_regs differentiated in the presence of rotenone were highly immunosuppressive, as both the proliferation and interferon-γ (IFN-γ) production by T cells cocultured with rotenone-treated iT_regs were significantly reduced. Conversely, inhibition of complex III during iT_regs differentiation impaired their suppressive potential (Fig. 3, L and M). Complex III inhibition–mediated iT_regs dysfunction was contemplated, as the ETC was partially operative through complex I, leading to the accumulation of TCA cycle intermediates α-KG, succinate, and 2-hydroxyglutarate that promote T_regs dysfunction, as reported (28). To further confirm that the CD38-NAD⁺ axis–mediated restriction of mitochondrial activity is crucial for the superior suppressive function of CD38^Hi iT_regs, we repeated the experiments using WT and CD38^−/− iT_regs. While CD38^−/− iT_regs exhibited impaired suppressive potential, their suppressive function was significantly restored to levels comparable to that of WT iT_regs when differentiated in the presence of a complex I inhibitor (rotenone) but not a complex III inhibitor (antimycin A) (fig. S3E). These findings strongly suggest that the heightened suppressive function of CD38^Hi iT_regs is driven by restricted mitochondrial activity regulated by the CD38-NAD⁺ axis.

To elucidate whether rotenone-mediated hyper-immunosuppression was, in part, due to the reduced generation of TCA cycle intermediates, we performed metabolomics analysis. Although citrate levels were high, the downstream metabolites α-KG and succinate were significantly low in T_regs differentiated in the presence of rotenone (Fig. 3, N to R). Correspondingly, the OCR levels were markedly low in the rotenone-treated T_regs, while a modest increase in their ECAR levels was evident, likely to compensate for their bioenergetic needs (fig. S3, F and G).

Because mitochondrial impairment has been reported in CD4⁺ T cells expressing CD38 (20), we investigated whether CD38-mediated mitochondrial dysfunction alone was sufficient to confer suppressive function on CD4⁺ T cells, independent of FoxP3 expression. However, CD4⁺CD38⁺FoxP3⁻ T cells did not exhibit any substantial suppressive function compared to iT_regs (fig. S3, H to J). Overall, these data indicate that mitochondrial oxidative metabolism is pivotal in regulating the suppressive function of T_regs, and CD38 limits oxidative metabolism, endowing T_regs with heightened immunosuppressive capacity.

CD38-mediated depletion of intracellular NAD⁺ pool regulates the function and stability of T_regs through regulating mitochondrial metabolism

CD38 is a NADase and has been reported to maintain the intracellular NAD⁺ pool (29). Intracellular NAD⁺ content regulates diverse metabolic processes, including the TCA cycle, by acting as a coenzyme for various dehydrogenases, which catalyze the reduction of NAD⁺ to NADH that is subsequently used in the oxidation-reduction reactions in the ETC (30). To elucidate whether CD38 affected mitochondrial metabolism by regulating the intracellular NAD⁺ pool, we first checked NAD⁺ content in CD38^Hi and CD38^Lo iT_regs. We found that NAD⁺ levels were substantially low in CD38^Hi iT_regs, albeit the comparable expression of genes involved in the biosynthesis of NAD⁺ between CD38^Hi and CD38^Lo iT_regs, indicating that the reduced NAD⁺ levels in CD38^Hi iT_regs were a result of CD38-mediated utilization of NAD⁺ (Fig. 4A and fig. S4A). Moreover, NAD⁺ levels were also found to be low in T_regs compared to those in T_effs (fig. S4B). Next, we checked whether low NAD⁺ content in CD38^Hi iT_regs was commensurate with reduced mitochondrial metabolism. Metabolomics analysis revealed that the mitochondrial metabolites, including citrate, α-KG, fumarate, and malate levels, were significantly low CD38^Hi in iT_regs compared to those in CD38^Lo iT_regs (Fig. 4B).

Fig. 4. CD38-NAD⁺ axis regulates the mitochondrial metabolism and suppressive function in T_regs.

(A and B) iT_regs sorted into CD38^Hi and CD38^Lo subsets were used to determine: (A) intracellular NAD⁺ levels and (B) indicated intracellular metabolites using MS. (C to G) FACS-sorted CD38^Hi iT_regs were cultured overnight in the presence or absence of NMN (500 μM) and were used for (C) determining the indicated intracellular metabolites using MS, (D) OCR under basal condition and in response to the indicated inhibitors, (E) basal respiratory capacity, (F) maximal respiratory capacity, and (G) spare respiratory capacity. (H to I) CD38^Hi iT_regs cultured overnight in the presence or absence of NMN were cocultured with CTV-labeled naïve T cells in the presence of a T cell activation cocktail (anti-CD3/CD28) for 3 days and assessed for (H) responder CD8⁺ T cell proliferation via CTV dilution and (I) measuring IFN-γ levels from the supernatant of the experiment shown in (H). (J) qPCR-based analysis of CpG methylation at CNS regions of the Foxp3 locus was performed on CD38^Hi and CD38^Lo iT_regs. (K and L) CD38^Hi iT_regs cultured overnight in the presence or absence of cell-permeable α-KG (100 μM) were used to assess (K) CpG methylation at CNS regions of the Foxp3 locus and (L) suppressive potential by measuring responder CD8⁺ T cell proliferation after 3 days of coculture with naïve T cells under T cell activation conditions. (M) qPCR-based analysis of CpG methylation at CNS regions of the Foxp3 locus was performed in CD38^Hi iT_regs supplemented with NMN. The data are representative of three (A), three (B), six (C), four [(D) to (I)], five [(J) and (K)], and three [(L) and (M)] independent experiments. *P < 0.05; **P < 0.01; ***P < 0.005; ****P < 0.0001.

To ascertain the role of NAD⁺ in regulating mitochondrial activity and, subsequently, the suppressive function of iT_regs expressing high CD38, we replenished intracellular NAD⁺ levels by supplementing NAD⁺ precursor nicotinamide mononucleotide (NMN) during iT_regs differentiation. NMN supplementation did increase the intracellular NAD⁺ levels in CD38^Hi iT_regs but had no adverse effect on the viability or the differentiation of T_regs (fig. S4, C to E). NMN supplementation restored mitochondrial metabolism in CD38^Hi iT_regs, as evidenced by the increased levels of various TCA cycle metabolites and improved mitochondrial respiration; however, it had no impact on the glycolysis of CD38^Hi iT_regs (Fig. 4, C to G, and fig. S4F).

Next, we sought to determine whether restoring mitochondrial activity in CD38^Hi iT_regs by replenishing their intracellular NAD⁺ content would affect their suppressive function. NMN treatment markedly diminished the suppressive function of CD38^Hi iT_regs, and the effect was found to be dose dependent (Fig. 4, H and I, and fig. S4, G and H). We found no reduction in cell viability or defect in T_regs differentiation even at a higher dose of NMN (fig. S4, I and J). Similar to NMN, the supplementation of nicotinamide riboside (NR), another precursor of NAD⁺, also affected the suppressive potential of CD38^Hi iT_regs (fig. S4, K and L). These data collectively indicate that CD38 expression in T_regs is essential for maintaining their intracellular NAD⁺ levels, which, in turn, supports their mitochondrial metabolism. This process is likely crucial for the stability and functionality of T_regs.

Next, we aimed to elucidate the mechanism by which reduced mitochondrial activity facilitates the superior functionality and stability of CD38^Hi iT_regs. Several studies have identified the demethylation of the Foxp3 locus, particularly at three well-characterized conserved noncoding regions (CNS1 to CNS3), as critical epigenetic modifications that govern the differentiation, stability, and functionality of T_regs (31–34). Therefore, we first assessed the CpG methylation of the Foxp3 locus. CNS2, along with CNS1 and CNS3, was hypomethylated in iT_regs compared to that in T_effs (fig. S4M). When comparing CD38^Hi and CD38^Lo iT_regs, we found that all three CNS regions were highly hypomethylated in CD38^Hi iT_regs compared to those in CD38^Lo iT_regs, underscoring the mechanistic basis for the superior stability and functionality of CD38^Hi iT_regs (Fig. 4J).

Because the TCA cycle metabolite α-KG has been shown to regulate DNA methylation, including regions in the Foxp3 gene locus (11), we sought to determine whether the maintenance of the hypomethylated state at the Foxp3 gene locus in CD38^Hi iT_regs was due to their reduced α-KG levels. To this end, we supplemented CD38^Hi iT_regs cultures with cell-permeable α-KG and subsequently assessed methylation at the CNS regions and suppressive function. We observed that α-KG supplementation significantly increased methylation at CNS2 and CNS3 of the Foxp3 locus and rendered CD38^Hi iT_regs functionally impaired (Fig. 4, K and L). Notably, NMN-supplemented CD38^Hi iT_regs, with increased levels of α-KG, also exhibited increased methylation at the CNS regions of the Foxp3 locus (Fig. 4M). To further investigate whether hypermethylation induced by NMN or α-KG destabilizes FoxP3 expression, we assessed FoxP3 levels in CD38^Hi iT_regs following supplementation. While FoxP3 expression remained unaffected at an early time point (24 hours posttreatment), a significant reduction was observed after an additional 2 days of culture, reinforcing the idea that NMN- or α-KG-mediated hypermethylation of the CNS regions predisposes CD38^Hi iT_regs to FoxP3 loss and functional instability over time (fig. S4, N and O). Collectively, these data suggest that CD38 in iT_regs acts as a metabolic checkpoint, regulating intracellular NAD⁺ content to ensure restricted mitochondrial metabolism and the generation of TCA cycle metabolites, particularly α-KG, which adversely affect T_regs stability and functionality.

CD38^Hi-NAD^Lo axis redirects lactate metabolism for the generation of PEP in T_regs

Uptake and metabolism of lactate is a hallmark of intratumoral T_regs (7, 8). Given that CD38 expression plays a distinctive role in rewiring the metabolic traits of iT_regs, we explored whether this axis contributes to the efficient metabolism of lactate, enabling iT_regs to retain their identity and functionality. Consistent with the previous study (8), we observed that iT_regs had a greater propensity to uptake lactate compared to T_effs (fig. S5A). The lactate uptake was comparable between CD38^Hi and CD38^Lo iT_regs (fig. S5B). Because lactate metabolism has been shown to profoundly influence T_reg functionality, we hypothesized that despite comparable lactate uptake, CD38^Hi and CD38^Lo iT_regs might be differentially equipped to metabolize lactate, leading to distinct functional outcomes.

The metabolic fate of lactate is usually determined by two enzymes: pyruvate dehydrogenase (PDH) and PC (15). PDH converts lactate-derived pyruvate to acetyl coenzyme A (CoA) for its further incorporation into the TCA cycle, whereas PC promotes the conversion of lactate-derived pyruvate to OAA, which is subsequently converted to PEP by PEPCK to be used in gluconeogenesis (15). To assess the contribution of these enzymes in lactate metabolism in CD38^Hi and CD38^Lo iT_regs, we first traced the metabolic fate of lactate by incubating the cells with uniformly labeled ¹³C₃ l-lactate for 18 hours in glucose-free medium. We tracked ¹³C incorporation into acetyl CoA and PEP, reflecting lactate’s commitment to entering the TCA cycle or gluconeogenesis, respectively. Notably, CD38^Hi iT_regs exhibited significantly higher allocation of ¹³C to PEP (M+3), while CD38^Lo iT_regs showed an increase, albeit nonsignificant, in ¹³C labeling in acetyl-CoA (M+2) (Fig. 5, A and B). These findings suggest a differential metabolic routing of lactate, with CD38^Hi iT_regs favoring gluconeogenesis, a feature also observed in intratumoral T_regs (8). In contrast, CD38^Lo iT_regs appear to preferentially engage the TCA cycle. Additionally, consistent with previous reports linking low NAD⁺ levels to reduced lactate oxidation into pyruvate (7), we observed a modest, although nonsignificant, decrease in ¹³C incorporation into pyruvate in CD38^Hi iT_regs compared to that in CD38^Lo iT_regs (Fig. 5C).

Fig. 5. CD38 is crucial for the metabolic adaptability of T_regs to use lactate and maintain suppressive function.

(A to C) Isotopologue distribution for (A) PEP, (B) acetyl CoA, and (C) pyruvate in CD38^Hi and CD38^Lo iT_regs overnight cultured in glucose-free medium containing ¹³C₃ l-lactate. (D and E) iT_regs were differentiated in glucose-free, l-lactate–containing medium and sorted into the CD38^Hi population. These were then treated overnight with the indicated inhibitors in the same l-lactate–containing medium and were used for coculturing with CTV-labeled naïve T cells in the presence of a T cell activation cocktail (anti-CD3/CD28) for 3 days and assessed for (D) CD8⁺ T cell proliferation via CTV dilution and (E) measuring IFN-γ levels from the supernatant of the experiment shown in (D). (F) T cell proliferation was assessed via CTV dilution after 3 days of coculturing naïve T cells with CD38^Hi iT_regs that were differentiated in glucose-free, l-lactate–containing medium, following the indicated RNA interference–mediated knockdown. (G to J) iT_regs were differentiated in glucose-free, l-lactate–containing medium and sorted into the CD38^Hi population before being treated overnight with NMN and indicated inhibitors. Cells were then assessed for (G) OCR under basal condition and in response to the indicated inhibitors, (H) CpG methylation at CNS1, (I) CpG methylation at CNS2, and (J) CpG methylation at CNS3 regions of the Foxp3 locus. (K and L) NMN-treated and untreated CD38^Hi iT_regs differentiated in glucose-free, l-lactate–containing medium were cocultured with CTV-labeled naïve T cells for 3 days in the presence of anti-CD3/CD28. (K) Proliferation of CD8⁺ T cells was assessed by CTV dilution, and (L) supernatant from (K) was used to measure IFN-γ levels. The data are representative of three [(A) to (C)], three [(F) to (J)], four [(D) and (E)], and four [(K) and (L)] independent experiments. *P < 0.05; **P < 0.01; ***P < 0.005; ****P < 0.0001.

Next, to assess whether PC-mediated metabolic routing of lactate in gluconeogenesis was key for maintaining the functionality of CD38^Hi iT_regs, we blocked the enzyme and checked the functionality. Notably, inhibition of PC in CD38^Hi iT_regs cultured in lactate medium or even in complete medium (CM) significantly curbed their function, while no such impairment in the functionality was evident upon PDH inhibitor treatment (Fig. 5, D and E, and fig. S5, C and D). The pivotal role of gluconeogenesis in contributing to the functionality of CD38^Hi iT_regs was further confirmed by using RNA interference targeting PC and PEPCK (Fig. 5F and fig. S5E).

To explore the mechanism underpinning the routing of lactate into gluconeogenesis by PC in CD38^Hi iT_regs, we examined whether the intracellular NAD⁺ pool plays a decisive role. Because PDH is an NAD⁺-dependent enzyme, we postulated that CD38, through depleting the NAD⁺ pool, enforces lactate utilization through PC rather than PDH. To delineate this, we supplemented NMN while culturing CD38^Hi iT_regs in lactate medium and checked their functionality. We observed that NMN supplementation in CD38^Hi iT_regs cultured in lactate-rich medium improved their mitochondrial OXPHOS and reversed DNA hypomethylation at the Foxp3 locus (Fig. 5, G to J). A similar effect was seen when PC, but not PDH, was inhibited in CD38^Hi iT_regs cultured in lactate medium (Fig. 5, G to J). Moreover, commensurate with increased OXPHOS and CNS methylation, NMN supplementation alleviated the suppressive function of CD38^Hi iT_regs (Fig. 5, K and L). Last, we assessed OXPHOS in CD38^−/− iT_regs, which have intrinsically elevated intracellular NAD⁺ levels, under lactate-rich conditions. Our findings revealed that CD38 deficiency enhanced OXPHOS in the presence of lactate (fig. S5, F to I), highlighting the critical role of the CD38-NAD⁺ axis in regulating lactate metabolism in T_regs. Together, these findings establish the CD38-NAD⁺ axis as a crucial determinant of T_reg metabolic adaptability, enabling them to thrive in the lactate-rich TME.

Small-molecule targeting of CD38 alleviates T_regs function and improves the antitumor response in mice

After identifying CD38 as essential for maintaining iT_regs function and stability in a lactate-rich environment, we next sought to extend our findings to the TME. To this end, gp100-reactive T cell receptor (TCR) transgenic CD8⁺ T cells (Pmel T cells) were activated with their cognate antigen for 3 days and adoptively cotransferred (at a 1:1 ratio) with either WT or CD38^−/− iT_regs, which were CellTrace Violet (CTV)–labeled, into B16-F10 melanoma-bearing mice with tumors established subcutaneously for 10 days. Upon retrieval on 48 hours after adoptive transfer, Pmel T cells, when transferred without iT_regs, were notably abundant at the tumor site, more so than in the spleen and DLN, likely due to antigen-driven expansion. Cotransferring these cells with WT iT_regs significantly reduced their frequency in the TME. However, when Pmel T cells were cotransferred with CD38^−/− iT_regs, only a modest reduction was observed (Fig. 6A). Consistent with these findings, analysis of the transferred T_regs population revealed a substantial reduction in the frequency of CD38^−/− iT_regs at the tumor site compared to that of WT iT_regs, while their frequencies in the spleen and DLN remained comparable (Fig. 6B). These findings indicate that CD38 deficiency impairs T_regs function in the TME, thereby permitting greater proliferation of Pmel T cells.

Fig. 6. Small-molecule targeting of CD38 impairs intratumoral T_regs and elicits antitumor response.

(A) Frequency of adoptively transferred Pmel (CD8⁺vβ13⁺) T cells from mice bearing 10 days subcutaneously established B16F10 melanoma. The adjacent bar represents data from n = 4 mice. (B) The bar diagrams depict the frequency of donor T_regs population (Foxp3⁺ CTV⁺) from the tumor-bearing mice shown in (A). (C and D) iT_regs were differentiated either in complete medium (CM) or in 30% TDS in the presence or absence of CD38i and were assessed for (C) evaluating suppressive potential by checking the proliferation of CTV-labeled CD8⁺ T cells from the coculture and (D) CpG methylation at CNS1, CNS2, and CNS3 regions of the Foxp3 locus. The data are representative of four (C) and three (D) independent experiments. (E to J) C57BL/6 mice (n = 5 mice per group) with subcutaneously established YUMM1.7 melanoma tumor were treated either with vehicle control or CD38i (15 mg/kg, intraperitoneal, once a week) and evaluated for (E) tumor growth; (F) T_regs frequency at the spleen, DLN, and tumor site; (G) assessment of T_eff (CD8⁺ T cells) to T_regs ratio at the tumor site; and [(H) to (J)] the ability of CD8⁺ T cells from the tumor site to coproduce different effector cytokines upon restimulation in vitro. TNF-α, tumor necrosis factor–α. (K and L) BrdU incorporation in (K) CD38^Lo and CD38^Hi intratumoral T_regs and (L) intratumoral T_regs from Ctrl and CD38i-treated mice. The representative data are from five [(E) to (L)] independent samples. *P < 0.05; **P < 0.01; ***P < 0.005; ****P < 0.0001. FSC-A, forward scatter area.

Given the critical role of CD38 in T_regs function and stability, we next explored whether small-molecule inhibition of CD38 could enhance the antitumor response. To investigate this, we first tested this approach on iT_regs differentiated in vitro in the presence of tumor-derived supernatant (TDS). We observed that iT_regs differentiated with TDS were modestly but significantly more suppressive than those differentiated in CM. Inhibition of CD38 with 78c (referred to as CD38i hereafter) reduced the suppressive function of iT_regs differentiated in both TDS and CM (Fig. 6C). The functional impairment was more pronounced in iT_regs differentiated in TDS compared to that in CM, suggesting that T_regs undergo distinct metabolic adaptations depending on carbon sources, which renders them differentially susceptible to CD38 inhibition–mediated functional impairment. Consistent with these findings, CD38^−/− iT_regs differentiated in 30% TDS exhibited profound functional impairment compared to their counterparts differentiated in CM (fig. S6A).

Next, we assessed the methylation status of the Foxp3 locus in CD38i-treated iT_regs. CD38i treatment induced hypermethylation in the CNS2 region of the Foxp3 locus in iT_regs differentiated under both CM and 30% TDS conditions. The hypermethylation of CNS2 was more pronounced in iT_regs differentiated in 30% TDS, further supporting the observed severe functional impairment in these cells following CD38 inhibition (Fig. 6D).

Last, considering the established role of hypoxia in promoting T_regs functionality within the TME (35, 36), we evaluated the effect of CD38 inhibition on iT_regs differentiated under hypoxic conditions. Similar to our observations with TDS-cultured iT_regs, CD38 inhibition significantly impaired the suppressive function of iT_regs differentiated in hypoxic conditions (fig. S6B).

To evaluate the therapeutic efficacy of CD38i, we subcutaneously engrafted the YUMM1.7 mouse melanoma tumor, which carries human-relevant mutations (BrafV600E; Cdkn2a^−/−; Pten^−/−; Tyrosinase:CreERT2), in B6 mice. We treated the mice with CD38i once a week until the tumor volume in vehicle control–treated mice reached the humane endpoint. We observed that CD38i treatment significantly delayed tumor growth (Fig. 6E). Analysis of the T_regs population revealed that CD38i treatment substantially reduced the frequency of intratumoral T_regs without affecting T_regs frequencies in the spleen or DLN (Fig. 6F). The reduction of intratumoral T_regs improved the CD8⁺ T_eff-to-T_regs ratio at the tumor site (Fig. 6G) and substantially augmented the prevalence of multifunctional CD8⁺ T cells at the tumor site (Fig. 6, H to J).

To further investigate whether CD38i directly affected the stability of intratumoral T_regs, we repeated the experiment in another cohort of YUMM1.7 tumor-bearing mice. The day before euthanasia, we administered 5-bromo-2′-deoxyuridine (BrdU) intraperitoneally. BrdU incorporation analysis revealed that CD38^Hi T_regs in the TME were more proliferative than CD38^Lo T_regs (Fig. 6K). CD38i treatment markedly diminished their proliferation (Fig. 6L). These data suggest that CD38 inhibition renders intratumoral T_regs metabolically unfit to thrive in the TME, leading to a loss of stability and reduced proliferation.

DISCUSSION

Understanding how immune cells metabolically adapt to their local microenvironment, which influences their function and persistence, is becoming crucial for devising strategies to modulate the host immune response to various pathological conditions. Previous studies have demonstrated the metabolic plasticity of T_regs, enabling them to reorient their metabolic traits in a tissue-specific manner (7, 13, 37). Notably, intratumoral T_regs have shown metabolic reliance to use lactic acid, thereby retaining their cellular identity and functionality (7, 8). However, the precise mechanisms that enable intratumoral T_regs to rewire their metabolism to use lactic acid, supporting their functionality in the TME, remain unclear. The present study reveals that the CD38-NAD⁺ axis in T_regs plays a pivotal role in directing the metabolic routing of lactate, which appears critical for ensuring stable function and survival of T_regs in the TME.

While T_regs exhibit tissue-specific metabolic adaptations, the process within the TME appears to be more complex, largely due to the unique metabolic characteristics observed in intratumoral T_regs (7, 9, 10). Emerging evidence suggests that lactate serves as the primary carbon source for intratumoral T_regs (8). Therefore, to effectively metabolize lactate and support their function and survival in the TME, these cells must have adopted mechanisms distinct from those of peripheral T_regs, which usually rely on carbon sources other than lactate (27). A recent study investigating the metabolic flux of sodium lactate in intratumoral T_regs uncovered an intriguing observation: Intratumoral T_regs tend to generate PEP from TCA cycle–derived malate and fuel gluconeogenesis (8). This finding suggests that the steady incorporation of lactate-derived pyruvate in the mitochondrial TCA cycle may be constrained in intratumoral T_regs, as a fraction of malate is transported out of mitochondria to fuel gluconeogenesis. Congruently, studies from other groups have also shown compromised mitochondrial function and fitness in intratumoral T_regs, with mitochondrial aberrations found to enhance their suppressive function (9, 11). Our data align with the previous studies by placing CD38 as a critical regulator of mitochondrial activity in intratumoral T_regs. We found that the restricted mitochondrial activity in T_regs was, in part, due to the CD38-mediated depletion of the intracellular NAD⁺ content. We posited that the reduced NAD⁺ content in CD38-expressing T_regs impeded their ability to efficiently drive the TCA cycle, as the essential dehydrogenase reactions were constrained by the limited supply of the critical coenzyme, NAD⁺. We observed that T_regs with high CD38 expression exhibited reduced levels of various TCA cycle metabolites, which could be ameliorated by replenishing the intracellular NAD⁺ levels by supplementing NAD⁺ precursors NMN and NR. CD38 has long been considered an ectonucleotidase due to its type II orientation, where its catalytic domains face extracellularly. However, recent findings suggest a type III orientation, with the catalytic domain facing the cytoplasm (38, 39). This dual orientation provides a mechanism for CD38’s regulation of intracellular NAD⁺ levels, as observed in T_regs and other cell types (40). In the present study, we observed that CD38-mediated regulation of intracellular NAD⁺ levels affects the suppressive function of T_regs by modulating TCA cycle–derived α-KG production, which, in turn, affects iT_regs stability by influencing methylation in the CNS regions of the Foxp3 locus. The notion was ascertained by assessing the stability of iT_regs differentiated in the presence of NMN and α-KG, which rendered iT_regs more prone to losing FoxP3 expression over extended culture periods.

Our data shed light on a very crucial aspect of T_regs metabolism in tumor. We found that restricted mitochondrial metabolism was critical for T_regs to keep the TCA cycle generated α-KG levels in check, which has been reported to affect the stability and functionality of T_regs by promoting methylation at the Foxp3 locus (11). When T_regs were cultured in lactate medium or tumor-conditioned media (TCM) to mimic the carbon source available in TME, CD38 primarily directed the metabolic routing of lactate-derived pyruvate toward PEP rather than their complete oxidation in the TCA cycle, due to maintaining low NAD⁺ levels that affect the TCA cycle. Moreover, we noted that PC-mediated conversion of pyruvate to OAA was more favored in the T_regs cultured in lactate, indicating that CD38-mediated low NAD⁺ content rewires the metabolic features of T_regs to undergo alternative metabolic pathway supporting gluconeogenesis. This metabolic diversion appeared extremely critical for T_regs fed on lactate, as increasing the intracellular NAD⁺ contents reoriented the lactate-derived pyruvate entry into the TCA cycle, leading to markedly enhanced methylation at the FoxP3 locus and subsequently affecting the functionality of T_regs. It is important to note that, NAD⁺ replenishment–mediated metabolic modulation, leading to the functional impairment of CD38-expressing T_regs, was highly apparent in T_regs cultured in the lactate or TCM compared to that in T_regs cultured in the CM. These discrepancies could be due to the discrete metabolic features adopted by T_regs based on the nutrients available in the milieu, as reported earlier (8, 41, 42).

Last, the study investigated the therapeutic potential of targeting CD38 to enhance antitumor immunity. Pharmacological inhibition of CD38 in tumor-bearing mice resulted in a significant reduction of intratumoral T_regs, an improved T_eff-to-T_regs ratio, and delayed tumor growth. These findings suggest that CD38 inhibition could be a viable strategy to disrupt T_regs-mediated immunosuppression in the TME by affecting their metabolic adaptability. Notably, CD38 inhibition selectively disrupted the proliferation of CD38-expressing intratumoral T_regs while leaving peripheral T_regs unaffected. Similar to our observations, targeting CD38⁺ T_regs with the CD38 monoclonal antibody isatuximab in patients with multiple myeloma has been shown to alleviate T_regs-mediated suppression and concurrently improve the effector response of natural killer (NK) and CD8⁺ T cells, without any systemic toxicity (22). Beyond T_regs, CD38 has also been implicated in regulating the function of myeloid-derived suppressor cells and terminally exhausted T cells, both of which are critical in dampening the durable antitumor response (17, 43). Therefore, we propose that small-molecule targeting of CD38 could mitigate the immunosuppressive milieu at the tumor site, fostering a favorable environment for T_effs and NK cells to mount an effective antitumor response.

Overall, the results underscore the critical role of CD38 in modulating the function and metabolic adaptation of T_regs within tumors. By regulating NAD⁺ levels and mitochondrial activity, CD38 ensures the stability and suppressive capacity of T_regs, which, in turn, supports tumor growth. The study’s findings pave the way for potential therapeutic interventions targeting CD38 to modulate T_regs function and improve antitumor immunity.

MATERIALS AND METHODS

Mice

C57BL/6 mice were bred and housed in a pathogen-free facility at the CSIR–Indian Institute of Chemical Biology (IICB). CD38^−/− mice were obtained from the Jackson Laboratory (Bar Harbor, MA). All experimental protocols were approved by the Institutional Animal Ethics Committee, CSIR-IICB, India (ref no. IICB/AEC/Meeting/Sep/2019/13). Both male and female mice, aged 6 to 8 weeks, were randomly selected for experimentation. In tumor control experiments, an equal number of male and female mice of the same age group were included.

Human samples

Surgically resected tumor tissue (1 to 2 g in normal saline) and peripheral blood (3 to 4 ml in heparin tube) from chemo naïve patients with MIBC and breast cancer [both triple-negative breast cancer (TNBC) and non-TNBC] were collected by medical personnel on obtaining written informed consent from all human patients, with ethical approval from the Institutional Review Board (IRB) of IPGME&R and SSKM Hospitals, Kolkata India (ref no. IPGME&R/IEC/2020/624) and R.G. Kar Medical College and Hospital, Kolkata, India (memo no. RKC/892 dated 15 June 2023), respectively. The use of human samples complied with the safety regulations of CSIR-IICB, Kolkata, India.

Cell lines

B16-F10 and YUMM1.7 melanoma cell lines were obtained from American Type Culture Collection. EL4 was obtained from the National Center for Cell Science, Pune, India. All cell lines were confirmed to be free of mycoplasma contamination using the MycoAlert Mycoplasma Detection Kit (Lonza) before use. Cells between the four to six passages were used for all the in vivo experiments.

Preparation of TDS

YUMM1.7 melanoma cells (0.75 × 10⁶ per mouse) were injected subcutaneously into the flanks of C57BL/6 WT mice. Mice were euthanized once the tumor volume reached 2000 mm³. Tumors were dissociated into a single-cell suspension and plated at a concentration of 1 × 10⁷ cells/ml in complete RPMI 1640 medium. After overnight incubation in a CO² incubator, the culture medium was centrifuged, filtered, and stored at −80°C for future use.

T_regs differentiation and culture conditions

Naïve CD4⁺ T cells were isolated from splenocytes of 6- to 8-week-old mice using the Dynabeads Untouched Mouse CD4⁺ T Cell Isolation Kit (Thermo Fisher Scientific, USA). These cells were differentiated into iT_regs [using TGF-β (5 ng/ml), interleukin-2 (IL-2; 300 U/ml), anti-mouse IFN-γ (10 μg/ml), and anti-mouse IL-4 (10 μg/ml)] or T_effs [with IL-2 (100 U/ml) and IL-12 (10 ng/ml)] in the presence of plate-bound mouse anti-CD3 (5 μg/ml) and anti-CD28 (5 μg/ml). Differentiation occurred in complete RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS; Thermo Fisher Scientific, USA), 1% penicillin-streptomycin (HiMedia Laboratories, India), and 55 μM β-mercaptoethanol (β-ME) (Thermo Fisher Scientific, USA), under 5% CO₂ at 37°C in a humidified incubator for 3 days. In some experiments, on day 2 of iT_reg differentiation, cells were treated with specific inhibitors (1 μM rotenone and 1 μM antimycin) and harvested on day 3 for downstream assays. In other experiments, iT_regs were fluorescence-activated cell sorting (FACS)–sorted on day 3 into CD38^Hi and CD38^Lo populations and maintained in IL-2 (100 U/ml) until use. In some cases, FACS-sorted CD38^Hi iT_regs were treated overnight with NMN (500 μM), NR (400 μM), or dimethyl α-KG (100 μM), then thoroughly washed with phosphate-buffered saline (PBS), and used in subsequent assays. To assess Foxp3 stability, CD38^Hi iT_regs were treated overnight with either NMN (500 μM) or dimethyl α-KG (100 μM) and then cultured for an additional 2 days in the presence of IL-2 (100 U/ml).

For lactate-fed iT_regs studies, cells were differentiated in either CM or glucose-free medium supplemented with 10 mM l-lactate. On day 3, sorted CD38^Hi iT_regs were cultured overnight in their respective medium with IL-2 (100 U/ml), with or without inhibitors (100 μM PCi or 100 μM PDHi). In some cases, 3-day activated iT_regs were further cultured for two additional days in either 30% TDS-containing medium or CM, along with CD38i (5 μM), NMN (500 μM), or vehicle control.

For some studies, naïve CD4⁺ T cells were differentiated into iT_regs under hypoxic (1% O₂) or normoxic (20% O₂) conditions for 3 days and subsequently used in assays.

In other experiments, purified naïve CD4⁺ T cells were activated using plate-bound anti-CD3 (5 μg/ml) and anti-CD28 (2 μg/ml) in the presence of IL-2 (100 U/ml) in complete RPMI 1640 medium supplemented with 10% FBS for 72 hours. After activation, sustained TCR stimulation was provided using plate-bound anti-CD3 (5 μg/ml) for up to 7 days. Cells were passaged every 48 hours and maintained in the presence of IL-2 (25 U/ml). Chronically expanded T cells were used in subsequent assays.

For human T_eff or iT_regs differentiation, PBMCs were isolated from buffy coats of healthy human donors using Ficoll-Hypaque gradient centrifugation. The buffy coats were anonymized before use, and all samples were handled in accordance with IRB approval (no. IICB/IRB/2020/2P). Purified naïve CD4⁺ T cells were differentiated into iT_regs [in the presence of TGF-β (5 ng/ml), IL-2 (300 U/ml), anti-human IFN-γ (10 μg/ml), and anti-human IL-4 (10 μg/ml)] or T_effs [in the presence of IL-2 (100 U/ml)]. Differentiation was performed with plate-bound human anti-CD3 (5 μg/ml) and anti-CD28 (5 μg/ml) in complete RPMI 1640 medium supplemented with 10% FBS (Thermo Fisher Scientific, USA) and 1% penicillin-streptomycin (HiMedia Laboratories, India) under 5% CO₂ at 37°C in a humidified incubator for 3 days. In all experiments, iT_regs were thoroughly washed with PBS before being used for coculture.

T cell suppression assay

Naïve T cells were isolated from the spleens of WT C57BL/6 mice and labeled with CTV using the CellTrace Violet Cell Proliferation Kit (Thermo Fisher Scientific, USA) following the manufacturer’s instructions. The CTV-labeled T cells were then plated at a density of 0.2 × 10⁶ cells per well in a 96-well plate, along with iT_regs at varying ratios: 0.2 × 10⁶ cells per well for a 1:1 ratio, 0.1 × 10⁶ cells per well for 1:0.5, 0.05 × 10⁶ cells per well for 1:0.25, and 0.02 × 10⁶ cells per well for 1:0.10. The cells were cocultured for 3 days in the presence of soluble mouse anti-CD3 and anti-CD28 antibodies (1 μg/ml each), except for the negative control group. A positive control group was included to assess T cell proliferation in the absence of iT_regs. On day 3 of coculture, the supernatant was collected by centrifuging the 96-well plate at 300g for 10 min and stored at −80°C until used for measuring IFN-γ levels by enzyme-linked immunosorbent assay (ELISA). The cell pellet was dissolved in PBS and subsequently assessed for CD8⁺ T cell proliferation by measuring CTV dilution using flow cytometry. For suppressive assay with human iT_regs, freshly isolated naïve T cells were labeled with CTV and cocultured with scrambled or CD38 shRNA transduced iT_regs.

siRNA-mediated knockdown

Freshly isolated murine naïve CD4⁺ T cells were differentiated to iT_regs, and CD38^Hi iT_regs were sorted using FACS. To achieve the knockdown of PC/PEPCK, CD38^Hi iT_regs were electroporated with 150 nM PC/PEPCK-targeting small interfering RNA (siRNA) (MISSION esiRNA; Sigma-Aldrich). Following electroporation, cells were cultured for 48 hours before functional studies were conducted. Statistical validation of functional differences was performed in experiments that achieved at least a 40% reduction in PC/PEPCK expression at the transcript level.

Lentiviral transduction of T cells

Human embryonic kidney (HEK) 293T cells were cultured in T75 flasks before transfection. HEK293T cells were transfected with lentiviral plasmid DNA encoding CD38 shRNA (Origene) or scrambled negative shRNA (Origene) along with structural plasmid ps-PAX2 and p-MD2G using CaCl₂ and HBS method. After 24 hours, the medium was replaced with low-serum–containing medium. Four hours after medium change, sodium butyrate was added at 5 mM concentration, and the cells were incubated for additional 48 to 72 hours, after which lentivirus-containing supernatant was collected and filtered. The viral content was concentrated using LentiX Concentrator (Takara Bio, Shinga, Japan) following the manufacturer’s protocol and stored as aliquotes at −80°C until use. After 3 days of differentiation, human iT_regs were collected, washed, and transferred onto non–tissue culture–treated 24-well plates (Corning, NY, USA) coated overnight with RetroNectin (30 μg/ml) (Takara Bio). Cells were transduced with concentrated CD38 shRNA lentiviral/control shRNA lentivirus particle and spinoculated at 1000g for 2 hours at 32°C, and IL-2 was added post-spinoculation, followed by a 48-hour culture at 37°C in a CO₂ incubator.

Flow cytometry

Surface staining was performed by incubating cells with the antibody at a 1:200 dilution in FACS buffer (0.5% bovine serum albumin and 0.1% sodium azide in PBS) for 20 min on ice. For intracellular cytokine staining, T cells were restimulated for 4 hours with phorbol 12-myristate 13-acetate (100 ng/ml), ionomycin (1 μg/ml), and GolgiPlug (BD Biosciences, San Jose, CA), followed by staining of surface markers. Cells were then fixed and permeabilized using the BD Cytofix/Cytoperm Kit (BD Biosciences, San Jose, CA), following the manufacturer’s protocol. Intracellular cytokine staining was done in the fixed/permeabilized cells. For FoxP3 staining, cells were first stained with surface markers, then fixed and permeabilized using a FoxP3 staining buffer set (Thermo Fisher Scientific, Waltham, MA), and subsequently used for FoxP3 staining. To assess BrdU incorporation, cells were fixed with the FoxP3 staining buffer set, and BrdU staining was performed following deoxyribonuclease treatment.

For all samples, cells were incubated with a fixable live/dead stain (Live/Dead fixable yellow dead cell stain kit from Thermo Fisher Scientific, Waltham, MA) after surface staining, before proceeding with further analysis. All stained samples were acquired using the BD LSRFortessa (BD Biosciences, San Jose, CA) and analyzed with FlowJo software (BD Biosciences). Details of antibodies are provided in table S1.

qPCR analysis

For real-time quantitative polymerase chain reaction (qPCR) analysis, cells were lysed using TRIzol reagent (Thermo Fisher Scientific, Waltham, MA), and RNA was isolated following the TRIzol extraction protocol. The isolated RNA was resuspended in UltraPure water (Thermo Fisher Scientific, Waltham, MA) and quantified using a Biotek Synergy HT Plate Reader. From 1 μg of RNA, cDNA was synthesized using the iScript cDNA Synthesis Kit (Bio-Rad, USA) and diluted 1:3 in nuclease-free water for qPCR reactions. qPCR was performed with the iTaq Universal SYBR Green Supermix (Bio-Rad, USA) on a CFX96 Real-Time System (Bio-Rad, USA) for a total of 40 cycles at a standard reaction speed. The primer details are provided in table S2.

Lactate uptake assay and intracellular NAD⁺ assay

Lactate uptake was measured using the Lactate-Glo assay kit (Promega, USA) in 1.5 × 10⁶ cells per group following the manufacturer’s protocol. Luminescence was quantified using the Varioskan LUX Multimode Reader (Thermo Fisher Scientific, USA).

Intracellular NAD⁺ content was determined in 1 × 10⁶ cells per group using the NAD/NADH Cell-Based Assay Kit (Cayman Chemical, USA). Absorbance at 450 nm was measured using MULTISKAN SkyHigh (Thermo Fisher Scientific, USA).

Enzyme-linked immunosorbent assay

IFN-γ levels in the culture supernatant were measured using the mouse IFN-γ–uncoated ELISA kit (Thermo Fisher Scientific, USA) according to the manufacturer’s protocol.

Metabolic flux assay

Metabolic flux analysis was performed using the Seahorse XFe24 analyzer (Agilent Technologies). Briefly, 0.3 × 10⁶ cells were plated on a Seahorse culture plate coated with Cell-Tak (Corning) overnight and incubated at 37°C in a CO₂-free incubator for 30 min to allow the cells to adhere. To assess glycolysis, the ECAR was measured under basal conditions and in response to glucose (10 mM; Sigma Aldrich), oligomycin (1.0 μM; Cayman), and 2-DG (100 mM; Sigma Aldrich). To assess OXPHOS, OCR was measured under basal conditions and in response to oligomycin (1 μM), carbonyl cyanide p-trifluoromethoxyphenylhydrazone (FCCP; 1 μM), and rotenone plus antimycin A (2 μM and 100 nM, respectively). Some experiments were conducted with minor modifications; OCR/ECAR analysis was performed under basal conditions or in response to NMN/vehicle control, followed by oligomycin (1 μM), FCCP (1 μM), and rotenone and antimycin A (2 μM and 100 nM).

Metabolomics and stable isotope tracing

For the metabolomics study, 3 × 10⁶ cells were taken, and, before statistical analysis, samples were normalized to Bradford values. Metabolic quenching and extraction of metabolites were performed by using ice-cold 80% methanol, followed by four freeze-thaw cycles in dry ice. Cellular debris was removed by centrifugation at 14,000 rpm for 15 min at 4°C. The supernatant was then collected and stored at −80°C. All the samples were evaporated in nitrogen stream within a hood. Evaporated samples were dissolved in mass spectrometry (MS)–grade water and placed in an autosampler of Ultimate 3000 UHPLC by Thermo Fisher Scientific interfaced to LTQ-Orbitrap XL Mass Spectrometer and the injection amount was 10 μl. The electrospray ionization (ESI)–MS analysis was performed in both negative and positive modes. Full-scan mass spectra were acquired over a mass range of mass/charge ratio of 80 to 1000 at 60,000 resolution, source voltage of 4.5 KV, capillary temperature of 320, and sheath gas flow of 25. Liquid chromatography separation was done in a reverse-phase C18 column (hypersil gold, diameter of 100 mm by 2.1 mm, and particle size of 1.9 μ), maintained at 400°C. Integrated pick areas were then extracted manually using Qual Browser (Thermo Fisher Scientific Xcalibur 2.1).

For isotopic flux analysis, iT_regs was generated and, on day 3, FACS sorted into CD38^Hi and CD38^Lo iT_regs. Cells were washed with PBS twice to remove the remaining medium and cultured (2 × 10⁶ cells per well) with glucose-free RPMI 1640 medium supplemented with uniformly labeled ¹³C l-lactate (Cambridge Isotope Laboratories) for 18 hours. Cells were then washed with cold PBS before metabolic quenching and extraction using the method described above. Metabolic flux analysis was performed through comprehensive isotopic targeted MS as previously described (44). Samples were analyzed on a triple quadrupole hybrid ion trap mass spectrometer (QTRAP 6500+, SCIEX) coupled with a SCIEX ExionLC UHPLC system. Optimized source and gas parameters were for the mass spectrometer, and the data were acquired through Analyst 1.6.3 software in both negative and positive modes. Metabolites (10 μl) were loaded and resolved on an Acquity UPLC BEH HILIC (1.7 μm, 2.1 mm by 100 mm) column where the autosampler temperature was set at 4°C, and the column compartment was set to 40°C. Mobile phases of solvents A (10 mM ammonium acetate and 10 mM ammonium hydroxide in 95% H₂O/5% ACN, pH 9) and B (10 mM ammonium acetate and 10 mM ammonium hydroxide in 5% H₂O/95% ACN, pH 7) were used for separation with a flow rate was 0.3 ml/min. The gradient program used is as follows: 90% buffer B for an initial 1 min, and buffer B was brought to 40% in the next 10 min and held constant for the next 4 min. Buffer B was brought to the initial 90% buffer B concentration in 1 min, and the chromatography system was re-equilibrated for 2 min at this concentration before the next injection. Peak review and area extraction were performed using MultiQuantTM software v.3.0 (SCIEX). A pooled QC from samples was used to analyze the technical variability, and only metabolites with <30% coefficient of variation were quantified.

Immunoblotting

To evaluate the expression of mitochondrial complexes, CD38^Hi iT_regs and CD38^Lo iT_regs were lysed using radioimmunoprecipitation assay buffer containing a protease inhibitor cocktail (Thermo Fisher Scientific). Total protein was isolated and quantified using the Bradford assay. Equal amounts of protein (60 μg) were separated by 12% SDS–polyacrylamide gel electrophoresis and transferred to a polyvinylidene difluoride membrane. The membranes were probed overnight at 4°C with an anti-OXPHOS complex antibody (Abcam) at a 1:2000 dilution, followed by incubation with a secondary antibody (horseradish peroxidase–conjugated goat anti-mouse immunoglobulin G at a 1:2000 dilution; Jackson ImmunoResearch Laboratories) for 2 hours at room temperature. Pre-stained protein markers (Invitrogen) were run in parallel to determine the molecular weight of the proteins. For chemiluminescent detection, the membranes were treated with Clarity Western ECL substrate (Bio-Rad), and the signal was monitored using the Bio-Rad Versa Doc Imaging System (Bio-Rad). The same membrane was then stripped with stripping buffer and re-probed with an antibody specific to heat shock protein 60 (HSP 60) (BioBharati LifeScience) as a reference control.

Single-cell sequencing data analyses

In the scRNA-seq analysis using Seurat (v4.3.0), the gene expression matrices for each sample were imported and quality controlled by filtering cells on the basis of mitochondrial gene expression and feature counts. The data were then normalized using the “LogNormalize” method that normalizes feature expression by scaling each cell’s measurements on the basis of the total expression and then applies a natural logarithm transformation. Next, we identified most variable genes per sample with “FindVariableFeatures” using the “vst” method. Further, the gene expression matrices were merged into a single matrix and linearly transformed with “ScaleData” function, and, subsequently, dimensionality reduction was done with “RunPCA” function in Seurat, capturing the main sources of variation in the dataset. To address batch effects, the scaled principal components analysis matrix was processed with the “RunHarmony” function (v1.0) in Seurat, integrating the data to minimize technical variation and enable unsupervised clustering. t–Distributed stochastic neighbor embedding plots of the harmony-corrected embeddings visualized cell clusters. Cluster analysis used a shared nearest neighbor graph and the Louvain algorithm to group cells by gene expression patterns. Marker genes for each cluster were identified with “FindAllMarkers” and visualized using heatmaps and feature plots created with ggplot2, providing insights into the gene expression profiles of different cell types or states.

DNA methylation assay

For the methylation study, 1 × 10⁶ cells per replicate were used to examine specific methylation in the CNS1, CNS2, and CNS3 regions of the Foxp3 gene. The study involved bisulfite conversion of genomic DNA followed by PCR amplification. Bisulfite treatment converts cytosine residues to uracil while leaving 5-methylcytosine intact, using the Methylamp DNA Modification Kit. Approximately 60 ng of DNA was used for methylation analysis, which was performed by qPCR with the Methylamp MS q-PCR Fast Kit (Epigentec). The study was conducted according to the manufacturer’s protocol.

Mouse tumor experiment

B16-F10 melanoma (0.5 × 10⁶ cells per mouse), YUMM1.7 melanoma (0.75 × 10⁶ cells per mouse), or EL4 thymoma (1 × 10⁶ cells per mouse) was injected subcutaneously into the flanks of 6- to 8-week-old C57BL/6 mice. Tumor volumes were measured every 2 days following tumor engraftment. For the CD38i experiment, YUMM1.7 melanoma-bearing mice were administered either CD38i (15 mg/kg) or vehicle control intraperitoneally on day 6 post–tumor inoculation. CD38i was then given intraperitoneally once a week until the tumor volume in control mice reached a humane endpoint. To assess intratumoral T_regs proliferation, 150 μl (150 mg) of BrdU solution (BioLegend) was injected intraperitoneally into both control and CD38i-treated YUMM1.7 melanoma-bearing mice 1 day before euthanasia.

Statistical analysis

For statistical analysis between two groups, the nonparametric unpaired Wilcoxon-Mann-Whitney U test was applied. Comparisons involving more than two groups were performed using one-way or two-way analysis of variance (ANOVA). Data analysis was conducted using GraphPad Prism 9.

Supplementary Materials

This PDF file includes:

Figs. S1 to S6

Tables S1 and S2