The research team, led by Vaishnavi Kaipilyawar, Samantha Leong, and senior author Padmini Salgame, set out to characterize the four major subsets of memory CD4+ T cells that circulate in the blood of individuals with remote latent tuberculosis infection. These subsets form a differentiation gradient: stem cell memory T cells (T-SCM), the most primitive and self-renewing population; central memory T cells (T-CM), which retain proliferative capacity and tissue-homing receptors; transitional memory T cells (T-TM), an effector-like intermediate; and effector memory T cells (T-EM), the most differentiated short-lived killers. Each subset was identified by a precise combination of surface markers, including CD45RA, CD45RO, CCR7, CD27, and CD95, and purified by fluorescence-activated cell sorting from donors in Brazil and North America.

To stimulate a realistic recall response, the researchers cultured peripheral blood mononuclear cells for sixteen hours with two immunodominant Mtb antigens, ESAT-6 and CFP-10, alongside co-stimulatory antibodies targeting CD28 and CD49d. They then sequenced the RNA of each sorted subset, generating roughly fifty million reads per sample and identifying more than four thousand differentially expressed genes. Principal component analysis revealed a clear segregation by stimulation condition, and the memory subsets clustered in a graded manner, with T-SCM most distinct and T-TM and T-EM grouping closely together. This graded transcriptional architecture set the stage for the study’s central question: what metabolic programs underlie each differentiation state?

Pathway analysis of the RNA sequencing data, performed using Ingenuity Pathway Analysis software, uncovered a striking metabolic split along the differentiation gradient. T-SCM and T-CM subsets exhibited elevated mitochondrial activity and oxidative metabolism, particularly fatty acid oxidation, a profile consistent with their identity as long-lived, self-renewing reservoirs rather than immediate effectors. The enhanced fatty acid oxidation in T-SCM was not accompanied by increased oxidative phosphorylation, suggesting a diversion of citrate to the cytosol through the carnitine-shuttle pathway, a process that may sustain self-renewal while reducing circulating fatty acids. In contrast, T-TM and T-EM subsets underwent glycolytic reprogramming and engaged the pentose phosphate pathway, which produces ribose-5-phosphate for nucleotide biosynthesis and NADPH to buffer oxidative stress during ongoing infection.

To validate these transcriptional findings at the single-cell protein level, the team employed a technique called Met-Flow cytometry, which tracks metabolic enzymes and rate-limiting proteins directly in living cells over time. Using a panel comprising CPT1A for fatty acid oxidation, G6PD for the pentose phosphate pathway, GLUT1 for glucose transport, HK1 for glycolysis, and MitoTracker for mitochondrial activation, they monitored metabolic shifts across five timepoints following Mtb stimulation. The results confirmed the early transcriptional modules with remarkable temporal precision: T-SCM cells showed high mitochondrial activity and CPT1A expression as early as six hours post-stimulation, with a gradual shift toward G6PD and GLUT1 by forty-eight and seventy-two hours, while T-TM and T-EM cells rapidly co-expressed G6PD, HK1, and GLUT1, indicating swift engagement of glycolysis and the pentose phosphate pathway.

The metabolic differences translated directly into functional specialization. When the researchers quantified cytokine production by flow cytometry, T-TM and T-EM subsets exhibited significantly higher magnitude of Mtb-specific interferon-gamma expression compared with T-CM and T-SCM. T-TM cells displayed the highest frequency and magnitude of interferon-gamma production among all four subsets, indicating enhanced effector function in more differentiated memory populations. Although T-SCM and T-CM had higher frequencies of IL-17-producing cells, their geometric mean fluorescence intensity was lower, suggesting a more restrained cytokine output consistent with their less differentiated state.

Beyond cytokines, the team performed a direct antimycobacterial assay to test whether metabolic state predicted the ability to restrict bacterial growth. Monocyte-derived macrophages from donors with latent infection were infected with luciferase-expressing Mtb Erdman and then co-cultured with each of the four memory T cell subsets at a one-to-one ratio. Bacterial burden, quantified as relative luminescence units, was measured at days one, three, and five post-infection. By days three and five, macrophages alone had significantly higher bacterial burden in the absence of any memory T cell subsets. However, the co-culture showed a graded effector response, with T-TM and T-EM subsets demonstrating greater Mtb growth restriction capability than T-SCM and T-CM. These results were independently confirmed by colony-forming unit assays on 7H11 agar plates, providing robust evidence that glycolytic, pentose phosphate pathway-engaged effector subsets are the primary mediators of direct bacterial control in vitro.

The most consequential finding emerged when the researchers extended their analysis to a longitudinal cohort of Mtb-exposed household contacts from Brazil, using publicly available bulk RNA sequencing data from the GSE112104 dataset. This cohort comprised sixteen individuals who eventually progressed to active tuberculosis and twenty-one who did not. Using CIBERSORTx cellular deconvolution, the team inferred the composition and gene expression of memory CD4+ T cells at baseline, before any clinical signs of disease. The results were concordant with the experimental findings: non-progressors showed the highest stem-cell-like and fatty acid oxidation signatures, with reduced glycolytic activity and lower interferon-gamma levels, while progressors and active tuberculosis cases exhibited elevated glycolysis, pentose phosphate pathway activity, exhaustion markers, and pro-inflammatory profiles aligned with disease progression.

Progressors showed increased expression of HK1, ATP5A1, GLUT1, and IFNG, whereas non-progressors expressed higher levels of CPT1A and IL17C. Comparison analysis of other pathways further revealed that progressors upregulated cytokine, integrin, eicosanoid, retinoic acid receptor, and apoptosis signaling, while non-progressors favored mitochondrial metabolism and amino acid signaling, including proline biosynthesis, choline catabolism, and glutamine metabolism. This metabolically adaptive Th17-like phenotype in non-progressors, favoring survival, homeostasis, and long-term persistence, mirrors recent reports linking higher IL-17 levels and Th17-like T cells to contained tuberculosis infection. The convergence across independent datasets suggests that immunometabolic specialization within the memory compartment, rather than the absolute presence of a single subset, may determine the durability of immune control versus progression to disease.

The authors frame their work as a proof-of-concept study that identifies metabolic state as a key axis connecting Mtb antigen-induced memory T cell differentiation, restimulation-induced transcriptional programming, and the durability of immune control. They acknowledge limitations, including the use of peripheral blood samples from a limited number of donors, which may not capture tissue-resident memory populations or account for confounders such as Mtb re-exposure and comorbidities. Nevertheless, the findings provide a technical foundation for future longitudinal studies examining dynamic metabolic modulation in Mtb antigen-specific memory T cell subsets from contained infection to disease progression. They also suggest therapeutic potential: modulating metabolic checkpoints to promote fatty acid oxidation could restore or sustain the functional capacity of memory T cells in tuberculosis progressors, particularly in T-SCM and T-CM subsets, while the transitional metabolic features observed in late progressors hint at a window of plasticity amenable to such immunomodulation. Future vaccine strategies, the authors propose, might preferentially expand the ratio of long-lived, self-renewing CD4+ memory T cells rather than short-lived effector populations, thereby engineering durable protection at the metabolic level.

Subject of Research: Immunometabolism of tuberculosis memory CD4+ T cells

Article Title: Immunometabolic reprogramming of Mycobacterium tuberculosis-responsive memory CD4+ T cell subsets is linked to long-term protective immunity

Article References: Kaipilyawar, V., Leong, S., Lovey, A., Stringari, L. L., Dietze, R., Ellner, J. J., Ribeiro-Rodrigues, R., & Salgame, P. (2026). Immunometabolic reprogramming of Mycobacterium tuberculosis-responsive memory CD4+ T cell subsets is linked to long-term protective immunity. Frontiers in Tuberculosis, 4, Article 1783887. https://doi.org/10.3389/ftubr.2026.1783887

Image Credits: AI Generated

DOI: 10.3389/ftubr.2026.1783887

Keywords: Immunometabolic, reprogramming, Mycobacterium, tuberculosis-responsive, memory, cell, subsets, linked, long-term, protective, immunity, scientific research