Patients with type 2 diabetes (T2D) have a lower risk of Mycobacterium tuberculosis infection, progression from infection to tuberculosis (TB) disease, TB morality and TB recurrence, when being treated with metformin. However, a detailed mechanistic understanding of these protective effects is lacking. Here, we use mass cytometry to show that metformin treatment expands a population of memory-like antigen-inexperienced CD8+CXCR3+ T cells in naive mice, and in healthy individuals and patients with T2D. Metformin-educated CD8+ T cells have increased (i) mitochondrial mass, oxidative phosphorylation, and fatty acid oxidation; (ii) survival capacity; and (iii) anti-mycobacterial properties. CD8+ T cells from Cxcr3−/− mice do not exhibit this metformin-mediated metabolic programming. In BCG-vaccinated mice and guinea pigs, metformin enhances immunogenicity and protective efficacy against M. tuberculosis challenge. Collectively, these results demonstrate an important function of CD8+ T cells in metformin-derived host metabolic-fitness towards M. tuberculosis infection.
Tuberculosis (TB) is a major threat to global health owing to the inadequacies of antimicrobial treatment, the lack of a highly effective preventive or therapeutic vaccine, and the rise of drug-resistant Mycobacterium tuberculosis (Mtb) infection. Interest is now focused on a new therapeutic paradigm, host-directed therapies (HDT) for TB, to boost the antimicrobial efficacy of innate and adaptive immunity to TB while limiting excessive inflammation and tissue damage1,2.
Patients with type 2 diabetes mellitus (T2D) have reduced risk of severe TB disease3 and Mtb infection when treated with metformin4. Metformin enhances the efficacy of anti-TB drugs, ameliorates lung pathology and reduces inflammation in Mtb-infected mice3. Retrospective data from eight clinical cohorts comprising a combined >250,000 individuals treated with metformin for T2D showed that metformin reduces the risk for progression to TB disease, mortality, lung cavitation, and relapse, and that it accelerates sputum conversion independently of glycemic control3,4,5,6,7,8,9. Metformin is a top candidate in the TB-HDT pipeline, but a detailed mechanistic understanding of its protective immunometabolic benefits is lacking.
Although CD4+ T cells are generally recognized as the predominant effectors of adaptive immunity to TB, CD8+ T cells are also recognized to play a critical role in determining whether the Mtb infection is contained or progresses to active disease10. Upon infection, naive CD8+ T cells (TN) differentiate into antigen (Ag)-specific effector T cells (TE) and central memory T cells (TCM). The latter have the ability to survive, expand, and generate cytotoxic CD8+ TE cells upon encounter with a cognate Ag later in life11. These TCM are the major memory-like T cells in an infected or an immunized host. In contrast, in naive mice housed under specific-pathogen free conditions, Ag-inexperienced memory-like CD8+ T-cell (TM) population, sometimes known as virtual-memory (TVM) or innate-memory CD8+ T cells, have been described12,13,14. These Ag-inexperienced CD8+ TM cells (i) display unique and similar phenotypes to Ag-experienced CD8+ TCM cells13,14,15,16, (ii) rapidly respond to primary antigenic stimuli14, and (iii) mediate the protective immunity via non-canonical effector functions17.
Survival, activation, and effector function of T cells is fundamentally linked to cellular metabolic programming18. TE cells predominantly use glycolysis, whereas TCM cells use oxidative phosphorylation (OXPHOS) to meet energy demands18,19. CD8+ TCM exhibit increased mitochondrial fatty-acid oxidation (FAO), spare respiratory capacity (SRC), and biogenesis20. SRC is the reserve capacity to produce energy in the mitochondria beyond the basal state. The increase in FAO required for optimal CD8+ TCM generation and survival depends on the mitochondrial import of long chain fatty acids20,21. Of note, CD8+ TCM cells have been recently demonstrated to have lower SRC than CD8+ TVM cells22.
In this study, we identify that metabolic reprograming by metformin empowers CD8+ T cells to contain Mtb infection. We show that metformin treatment (i) expands memory-like CD8+CXCR3+ T cells in naive mice and in healthy and diabetic humans, (ii) induces mitochondrial SRC and FAO in CD8+ T cells, and (iii) enhances Bacillus Calmette–Guérin (BCG) vaccine-elicited CD8+ T-cell responses and efficacy in mouse and guinea pig TB models. Metformin-educated Ag-inexperienced CD8+ TM cells have gene expression signatures similar to an activated T cells, and restrict Mtb growth in T-cell-deficient mice.
Metformin-educated CD8+ T cells restrict Mtb replication
We previously reported that metformin attenuates immunopathology and reprograms CD4+ and CD8+ T-cell responses in tissues of Mtb-infected wild-type C57BL/6 (WT) mice3. To dissect the relative importance of metformin-educated CD4+ vs CD8+ T cells in TB defense, we compared their capacity to protect against Mtb infection in an adoptive transfer model. Splenic CD4+ or CD8+ T cells from WT mice treated with metformin or not (control group) were isolated and adoptively transferred into irradiated recipients, which were then infected with Mtb (Fig. 1a). Irradiated naive recipients that did not receive any T cells were also infected (no transfer group). In two independent experiments we found that mice which received metformin-educated CD8+ T cells had 0.5 log10 reduced lung bacterial load at 21 days post infection (p.i.) compared with the mice that did not receive any T cells (Fig. 1b, c). At the same time, we noticed a 0.3 log10 reduction in the lungs of mice receiving metformin-educated CD8+ T cells compared to those receiving CD8+ T cells from untreated mice (Fig. 1b, c). In these experiments, CD4+ T cells from metformin-treated and untreated donors failed to mediate the protection against Mtb (Fig. 1b, c).
The reduction in the lung bacterial load in recipients of metformin-educated (or metformin-primed) CD8+ T cells was paralleled by an increased frequency and total numbers of splenic memory CD8+CD44+CD62L+ T cells and naive CD8+CD44−CD62L+ T cells, and decreased effector CD8+CD44+CD62L− T cells compared to the mice that did not receive CD8+ T cells (Fig. 1d, e and Supplementary Fig. 1a). Of note, there was no change in the total number of CD3+, CD4+, and CD8+ T cells in the recipient mice among different groups (Supplementary Fig. 1b). These results suggested that the predominant effect of metformin treatment in donor mice was on the memory compartment of CD8+ T cells over effector compartment, as more memory cells were observed within the total CD8+ splenic T cells from Mtb-infected recipients receiving metformin-educated CD8+ T cells. In contrast, the proportions of memory to effector CD8+ T cells were equal in the spleens of Mtb-infected mice that received control CD8+ T cells (Fig. 1e). Mice receiving either metformin-educated or control CD8+ T cells had increased frequencies of perforin-expressing (Perforin+IFNγ−TNF−) memory CD8+ T cells compared with the mice that did not receive any T cells (Fig. 1f). In congenic recipients (receiving either metformin-educated or control CD8+ T cells) we saw similar increased numbers of Perforin+IFNγ− donor memory CD45.1+CD8+ T cells compared with the mice that did not receive any T cells (Fig. 1g and Supplementary Fig. 1c). Surprisingly, mice receiving metformin-educated CD8+ T cells had increased numbers of double- or triple-negative memory CD8+ T cells (Fig. 1f, g), indicating perforin-, IFNγ- and TNF-independent Mtb killing property of metformin-educated CD8+ T cells. CD4+ T cells have been reported to have such non-canonical anti-Mtb property23. Overall, these results demonstrated the anti-Mtb activity of metformin-educated CD8+ T cells and suggested the potential of metformin to induce the expansion of memory CD8+ T cells with a unconventional effector phenotype.
Metformin expands CD8+CXCR3+ TM cell population in naive mice
To understand the effects of metformin exposure on CD8+ T cells in naive mice, we investigated the phenotype of metformin-educated splenic CD8+ T cells by mass cytometry (CyTOF)24,25 using a panel of 40 markers (Fig. 2a and Supplementary Table 1). Analysis of live CD45+CD19−TCRβ+CD90+CD4−CD8+ spleen cells (Supplementary Fig. 2a) using the nonlinear dimensionality reduction technique t-SNE with a machine-learning clustering algorithm (Phenograph) identified 20 distinct cell clusters with different expression characteristics of surface markers and intracellular transcription factors (Fig. 2b, c and Supplementary Fig. 2b). Based on the expression of CD62L and CD44, 20 clusters of CD8+ T cells were broadly classified into TN, TE, and TM cells (Fig. 2b and Supplementary Fig. 2b). Automated machine-learning gating showed an expansion in the proportion of cells in clusters 9, 12 and 18, which comprised all TM cells, within the total metformin-educated CD8+ T-cell population (Supplementary Fig. 2c). These clusters were strongly positive for conventional and unconventional memory T-cell markers including CD62L, CD44, CD127, CD27, CD122, and Eomes (Fig. 2c and Supplementary Fig. 2b). They also expressed Tbet, CXCR3 and CD49d (Fig. 2c and Supplementary Fig. 2b), indicating an effector phenotype and differentiating them from CD8+ TVM cells15,26. Tbet directly transactivates CXCR327 and retroviral-mediated CXCR3 expression in Tbet−/− CD8+ T cells reconstitutes their ability to infiltrate inflamed tissues28. CXCR3 plays an important role in T-cell trafficking and function26 and expression of CXCR3 on CD8+ TM cells is critical for populating the airways under steady state conditions29.
Using a manual gating strategy, we validated the enrichment of CD8+ TM cells identified by machine-learning strategy, confirming an increased frequency of memory-like CD8+CXCR3+ cells after metformin treatment (Fig. 2d). The enhanced CD8+ TM frequency induced by metformin was further confirmed by flow cytometry of splenocytes from naive mice (Fig. 2e, f). Flow cytometric analysis also showed an enhanced frequency of total CD8+CXCR3+ cells in the spleen of metformin-treated naive mice (Supplementary Fig. 2d). Interestingly, OVA protein-stimulated CD8+ OT-I cells cultured with IL-15 (a cytokine known to stimulate TM cell differentiation30) showed increased TCM/TE ratio in the presence of metformin (Supplementary Fig. 3a, b), further demonstrating effectiveness of metformin in expanding the CD8+ TCM compartment as well.
We next used CyTOF to decipher the heterogeneity in the lung CD8+ T cells from metformin-treated naive WT mice (Supplementary Fig. 4a). Machine-learning analysis showed metformin-mediated increased expression of CXCR3 on lung (i) total CD8+ T cells (Supplementary Fig. 4b), (ii) CD8+ TE cells (Supplementary Fig. 4c) and CD8+ TM cells (Fig. 2g). This was further confirmed by manual gating of lung CyTOF data (Supplementary Fig. 4d) and flow cytometric analysis of lung cells (Supplementary Fig. 4e) from metformin-treated naive mice. To test if CXCR3 was involved in the metformin-mediated expansion of CD8+ TM cells in the spleen and lung, we compared untreated and metformin-treated Cxcr3−/− mice. The Cxcr3−/− mice had increased CD8+ TE and reduced CD8+ TM cells compared with WT mice in spleen (Fig. 2h) and lung (Supplementary Fig. 4f), confirming an earlier report31, while in the present study metformin treatment did not alter the percentage of TE, TM, or TN cells in Cxcr3−/− mice.
To determine why metformin failed to expand CD8+ TM cells in Cxcr3−/− mice, we performed RNA sequencing (RNAseq) on splenic CD8+ T cells from Cxcr3−/− and WT mice. We found 202 differentially expressed genes (DEGs) in CD8+ T cells from Cxcr3−/− compared with WT mice (Fig. 2i, left panel and Supplementary Data 1). Ingenuity pathway analysis (IPA) of these DEGs revealed an enrichment of genes associated with communication between innate and adaptive immune cells and IL-7 signaling pathway (Fig. 2i, right panel and Supplementary Table 2). For optimal CD8+ TM cell homeostasis a discontinuous IL-7 signaling has been proposed32. Among DEGs, we found downregulation of Foxo1 (Forkhead Box O1), Gsk3b (Glycogen synthase kinase 3 beta) and Nkg7 (Natural killer cell granule protein 7, a cytotoxic marker), and upregulation of Myc, Jun, and Il1b (Interleukin 1 beta; Supplementary Data 1). Importantly, FOXO1, GSK3B, and Myc are required in the maintenance of memory state of CD8+ T cells33,34,35. Further analysis of DEGs revealed a significant enrichment of FOXO1 targets (n = 35)36 among 202 DEGs (Fig. 2j, P = 1.2 × 106, hypergeometric test, see “Methods”), indicating a perturbation in FOXO1 signaling in CD8+ T cells from Cxcr3−/− mice. Taken together, our data suggested that metformin reprograms the CD8+ T cells phenotype and shapes the generation of the CD8+CXCR3+ TM population. This generation of CD8+CXCR3+ TM cells could be mediated by FOXO1, which can also control CD8+ TM responses during infection37.
Metformin promotes CD8+ T-cell SRC and mitochondrial health
Enrichment of CD8+ TM cells by metformin suggested an effect dependent on the regulation of cellular metabolism. Therefore, we measured the bioenergetic profiles of spleen CD8+ T cells from untreated and metformin-treated naive WT mice. Oxygen consumption rate (OCR), an indicator of OXPHOS, was higher in metformin-treated CD8+ T cells compared to untreated CD8+ T cells (Fig. 3a). Importantly, metformin-treated CD8+ T cells demonstrated a larger mitochondrial SRC when compared with CD8+ T cells from untreated mice (Fig. 3a). This increase in SRC was found to be dependent on CXCR3 expression (Fig. 3b). The absence of metformin-mediated increased OXPHOS in CD8+ T cells from Cxcr3−/− mice could be owing to the inherent reduced expression of Foxo1 in these cells (Fig. 2i). Of note, FOXO1 regulates oxidative stress responses, proliferation, cell survival, and apoptosis, and connects metabolic activity with immune response38,39,40.
Enhanced SRC in cells might be explained by increased cellular mitochondrial mass20. CD8+ T cells from metformin-treated WT mice showed increased Mitotracker Green median fluorescence intensity (MFI) compared with CD8+ T cells from untreated mice, indicating increased mitochondrial mass (Fig. 3c and Supplementary Fig. 5a). Metformin-educated CD8+ T cells from WT mice also showed a trend of increased mRNA expression for Sell (CD62L), Cpt1a and Tfam (mitochondrial transcription factor A) genes (Supplementary Fig. 5b). In addition, metformin-educated CD8+ T cells showed increased accumulation of tetramethylrhodamine, methyl ester (TMRM), indicating healthy and functional mitochondria (Fig. 3d).
Metformin promotes survival of CD8+ T cells
To further assess the impact of metformin on the CD8+ TM compartment, we characterized the survival capacity of CD8+ T cells in vivo by adoptively transferring metformin-treated or untreated cells into congenic mice (Fig. 3e). After 3 days, we found significantly more CD45.1+ donor metformin-treated CD8+ T cells in spleen and lymph node of the recipient mice (Fig. 3f). Together, our data suggested that metformin enhanced mitochondrial SRC and health of CD8+ T cells and promoted their survival, which are characteristic features of TM cells.
Metformin-educated CD8+ T cells use glucose for FAO and OXPHOS
As metformin promoted the enrichment of the CD8+ TM cells (Fig. 2) with increased SRC (Fig. 3a), we reasoned that metformin-educated CD8+ T cells might utilize extracellular glucose to fuel mitochondrial FAO and OXPHOS, a property of IL-15-induced CD8+ TM cells20. We evaluated this by measuring OCR in CD8+ T cells cultured in low glucose (LG) medium. The increased OCR and SRC seen in CD8+ T cells from metformin-treated mice cultured in glucose-sufficient medium (25 mM, Fig. 3a) was not observed under LG conditions (0.5 mM, Fig. 3g). This suggested that glucose was driving FAO and OXPHOS in metformin-educated CD8+ T cells. In addition, under LG conditions, supplementation with fatty acid (palmitate) rescued the SRC of metformin-educated CD8+ T cells (Fig. 3g), indicating that increased SRC of metformin-educated CD8+ T cells in glucose-sufficient media (Fig. 3a) was derived from FAO.
Next, we used the CPT1a inhibitor etomoxir (Eto), which blocks mitochondrial FAO, to confirm that the enhanced SRC in CD8+ T cells from metformin-treated mice depended on FAO. Treatment with Eto, either before or after carbonylcyanide-p-trifluoromethoxyphenylhydrazone (FCCP) injection, impaired basal OCR and SRC, respectively, in metformin-educated CD8+ T cells (Fig. 3h, i). As CPT1a-independent effects of Eto on mitochondrial respiration have been observed41 and we used higher concentration of Eto (viz. 200 μM), FAO independent effect of metformin on OXPHOS cannot be ruled out. To further support the hypothesis that metformin-educated CD8+ TM cells used glucose for FAO and OXPHOS, we performed glucose uptake assay using the fluorescent glucose analog 2-NBDG. Lung and spleen CD8+ TM cells from metformin-treated mice acquired more 2-NBDG than CD8+ TM cells from untreated mice (Fig. 3j). We did not find an increased uptake of 2-NBDG by CD8+ TE cells from metformin-treated mice (Supplementary Fig. 5c), indicating minimal effect of metformin on the metabolic reprogramming of those cells. Together, our data suggested that glucose fuels mitochondrial FAO, which might be responsible for the induction of OXPHOS in metformin-educated CD8+ T cells.
Distinct gene expression in metformin-educated CD8+ TM cells
We next used RNAseq to characterize the effect of metformin on splenic CD8+ TM cells (Fig. 4a). A total of 267 genes were found to be differentially expressed in metformin-educated CD8+ TM cells (Fig. 4a and Supplementary Data 2). IPA-based molecular function enrichment revealed an effect of metformin on the expression of 86 genes associated with cell survival (Fig. 4b and Supplementary Table 3). Among these 86 genes we found an upregulation of Cdk1 (cyclin-dependent kinase 1) and Agap2 (Arf-GAP with GTPase, ANK repeat and PH domain-containing protein 2) genes, consistent with a role of metformin in modulating CD8+ TM cell fate (Fig. 4c). In addition, metformin reduced the expression of Mtor (mammalian target of rapamycin), Immt (inner membrane mitochondrial protein, mitofilin), and Casp1 (caspase-1) genes (Fig. 4c). Metformin is known to inhibit mTOR42, whereas IMMT deficiency has been associated with perturbed mitochondrial function43. As caspase-1 is required for pyroptosis, a form of programmed cell death44, we reasoned that the enhanced survival of metformin-educated CD8+ T cells (Fig. 3f) might be explained by the reduced caspase-1 activity. As predicted, metformin-treated Mtb-infected mice had reduced numbers of CD8+caspase-1+ cells in the spleen compared to the untreated Mtb-infected mice (Fig. 4d).
To test whether metformin-educated CD8+ TM cells survived better in vivo under infection settings, we adoptively transferred flow-sorted CD8+ TM cells into congenic TCRβδ−/− mice that were subsequently infected with Mtb or BCG (Fig. 4e; total three experiments). We found significantly more metformin-educated donor cells in the lungs (Fig. 4f) and spleens (Fig. 4g) of the recipient Mtb-infected TCRβδ−/− mice. This paralleled with a trend for lower bacterial load in the lungs of Mtb-infected TCRβδ−/− mice that received metformin-educated CD8+ TM cells compared with those receiving control CD8+ TM cells (1.4 log10 reduction), although this did not reach statistical significance (Fig. 4h). In other experiments, TCRβδ−/− mice receiving metformin-educated CD8+ TM cells had significantly lower lung and spleen bacterial load compared with the TCRβδ−/− mice that did not receive any T cells (0.9-1.3 log10 reduction; Fig. 4i, j). In this experiment, CD8+ TE cells from control and metformin-treated mice equally restricted lung bacterial loads in the TCRβδ−/− recipient mice to a comparable level (Fig. 4i). Collectively, our data indicated that metformin reprogramed gene expression of CD8+ TM cells, increasing their survival and anti-mycobacterial properties.
Metformin enhances BCG vaccine immunogenicity and efficacy
The generation of TM cells has important implications for vaccine efficacy. As metformin-educated CD8+ TM cells display distinct phenotypes and function associated with host protection, we evaluated the effect of metformin on the response to vaccination with BCG (Supplementary Fig. 6a). In BCG-vaccinated mice, metformin treatment was associated with an increased frequency of PPD (purified protein derivative of Mtb)-specific IFNγ-secreting splenic CD8+ T cells (Fig. 5a) but not CD4+ T cells (Supplementary Fig. 6b), compared with untreated BCG-vaccinated animals. Similar trend was observed in PPD-specific CD8+ TM and TE cells from metformin-treated BCG-vaccinated mice (Fig. 5a). We also observed an increased frequency of IFNγ+ total CD8+ cells (unstimulated) from metformin-treated BCG-vaccinated mice (Fig. 5a). CD8+ T cells from metformin-treated BCG-vaccinated mice also demonstrated an increased mitochondrial mass (Fig. 5b) and health (Fig. 5c) similar to that seen in unvaccinated animals (Fig. 3b, c). We next performed RNAseq on flow-sorted CD8+ TM cells from metformin-treated and untreated BCG-vaccinated mice. Comparative analysis revealed 607 DEGs that were enriched for genes associated with OXPHOS (Fig. 5d, Supplementary Data 3 and Supplementary Table 4). A heat map of OXPHOS genes showed an increased expression of electron transport chain genes in CD8+ TM cells from metformin-treated BCG-vaccinated mice (Supplementary Fig. 6c).
We next examined whether metformin treatment would enhance BCG vaccine-mediated Mtb control in mice. A trend for reduced Mtb bacillary load in the lungs of metformin-treated vaccinated mice compared to untreated vaccinated mice was observed (Fig. 5e). BCG-vaccinated metformin-treated mice showed an increased frequency of memory CD8+ T cells, and decreased effector CD8+ T cells upon Mtb infection compared to the BCG-vaccinated mice alone (Supplementary Fig. 6d, upper panel). When stimulated with class I-restricted peptide (TB10.4) of the Mtb EsxH gene, memory CD8+ T cells from BCG-vaccinated metformin-treated mice demonstrated increased TNF production compared to the BCG-vaccinated mice alone (Supplementary Fig. 6d, lower panel). We next used the guinea pig TB model to further evaluate metformin-mediated enhancement of BCG efficacy (Fig. 5f)45. Lung Mtb bacillary load in metformin-treated BCG-vaccinated guinea pigs was significantly lower compared to unvaccinated animals or animals which received only metformin, reduction of 3.4 log10 and 2.9 log10, respectively (Fig. 5g). In this experiment, bacillary load in untreated vaccinated animals was not significantly different from unvaccinated animals (reduction of 0.4 log10, Fig. 5g). Histopathological evaluation of the infected lungs showed reduced infiltrating leukocytes and lesions in BCG-vaccinated guinea pigs compared with unvaccinated animals (4.3-fold reduction in lung lesion burden, Fig. 5h). Importantly, lungs of metformin-treated BCG-vaccinated guinea pigs showed 11.2- and 2.6-fold reduced lesion burden compared with the control and untreated BCG-vaccinated guinea pigs, respectively (Fig. 5h, i). Taken together, our data suggested that metformin enhances the BCG vaccine response and efficacy against Mtb.
Metformin expands CD8+CXCR3+ TM cell population in humans
To investigate the effect of metformin in humans, we performed CyTOF on peripheral blood mononuclear cells (PBMCs) from healthy human participants who took a standard dose of metformin for five days (Fig. 6a, See Methods)46. The PBMCs were stained with a panel of 38 markers (Supplementary Table 5). Analysis of CD8+ T cells (CD45+CD14−CD19−HLA-DR+CD3+CD4− γδTCR−TCRVα7.2-CD161−CD56−CD8+) using t-SNE (Supplementary Fig. 7a) identified TN cells (CCR7+CD45RO−CD45RA+), TE cells (CCR7−CD45RO+CD45RA+) and TM cells (CCR7+CD45RO+CD45RA−) (Fig. 6b, c; Supplementary Fig. 7a)47. Two distinct TM and TE subgroups were observed (Fig. 6b). Interestingly, the TM2 subset was positive for CD127 and CXCR3 (Fig. 6d), similar to what we observed in mouse CD8+ TM cells (Fig. 2c). This CD8+ TM2 (CD8+CD127+CXCR3+) population expanded upon metformin intake in healthy subjects (Fig. 6e), and also showed increased expression of anti-apoptotic BCL2 (Fig. 6f), which was previously shown to correlate with CD8+ T-cell survival48. We next performed flow cytometric analysis on the PBMCs from T2D patients who were being treated with either metformin or with other anti-diabetic drugs as monotherapy (Supplementary Fig. 7b and Supplementary Table 6). These two groups had similar fasting plasma glucose levels (Supplementary Table 6). Metformin-receiving T2D patients showed an increased frequency of CD8+CXCR3+ TM cells compared with T2D patients treated with other anti-diabetic drugs (Fig. 6g). Overall, our data indicated that metformin induced the expansion of memory-like CD8+CXCR3+ T cells both in mice and humans.
Effect of metformin on Mtb clearance and reactivation
CD8+ T cells play an important role in long-term protective immune responses against Mtb infection and reactivation of latent TB10,49. To assess the persistence of metformin effects, we first measured the bioenergetic profiles of splenic CD8+ T cells from naive mice that received metformin for 6 months. Similar to our previous finding (Fig. 3a), OCR was higher in CD8+ T cells from mice chronically treated with metformin compared with those from untreated mice (Supplementary Fig. 8a). This was associated with an increased number of splenic CD8+ TM cells in metformin-treated mice (Supplementary Fig. 8b). We next used a modified version of the TB Cornell model50 to evaluate the long-term effects of metformin on Mtb clearance and reactivation. In this model, Mtb-infected mice were treated with pyrazinamide (PZA) and isoniazid (INH) ± metformin for 70 days starting on day 40 p.i., followed by a rest period of 40 days, and then challenged with dexamethasone to reactivate any persisting bacilli (Supplementary Fig. 8c). In the animals which received metformin along with PZA and INH, bacilli was recovered from the lungs of 40–50% mice sacrificed 40 days after the drugs were stopped (day 150 p.i, Table 1 and Supplementary Fig. 8c). However for mice which received only PZA and INH, bacilli were recovered from 75–80% of animals at day 150 p.i. After 15 days of dexamethasone treatment bacilli were recovered from the lungs of 20–50% of mice treated with metformin+PZA + INH compared with 75–80% mice treated with PZA + INH (Table 1). Taken together, these data suggest that metformin enhances the sterilizing activity of suboptimal antimicrobial treatment for Mtb infection by boosting intrinsic host immune function.
The goals of HDT for TB are to accelerate lesion sterilization and mitigate immune pathology by educating host anti-Mtb responses2,51. Previously, we proposed metformin as an adjunctive TB treatment, reporting that it lowers risk for cavitary TB and mortality in adults with T2D3. In the present study, we showed that metformin treatment (i) expands CD8+CXCR3+ TM cells in mice and humans, (ii) reprograms CD8+ T cells metabolic and transcriptional circuits, (iii) enhances BCG vaccine protective efficacy, and (iv) improves sterilizing ability of antibiotics, thus preventing Mtb reactivation.
Metformin-educated Ag-inexperienced CD8+CXCR3+ TM cells share phenotypic and functional properties with Ag-specific CD8+ TCM, Ag-inexperienced TVM, innate-memory CD8+ T, and/or IL-15-derived CD8+ TM cells52, and have enhanced survival and anti-Mtb properties. Cytokine-activated CD8+ TM cells are known to contribute to host protection against diverse pathogens17. The differential transcriptomic and phenotypic properties of metformin-educated CD8+CXCR3+ TM cells suggest that these cells may represent another type of “innate CD8+ T-cell”17. The CXCR3/CXCL10 axis can shape the distribution of CD8+ TM cells in the tissues of non-immunized mice53 and CXCR3-expressing innate CD8+ TM cells are known to express activation markers, respond to IL-2 and IL-15, and possess antibacterial effector function54,55. In addition, CXCR3 is required for the migration and establishment of CD8+ TCM cell population in the respiratory tract, where they protect against influenza A virus infection29.
CD8+ TM cells protect the host against infectious diseases by virtue of their recall ability to expand and execute effector functions10,56. These responses are promoted by AMP-activated protein kinase (AMPK)57, a crucial energy sensor and an indirect target of metformin. AMPK signaling promotes OXPHOS during CD8+ TM generation58 and maintains mitochondrial health in TM cells59. This indicates the importance of AMPK and OXPHOS in rewiring of metformin-educated Ag-inexperienced CD8+ TM cells in our experiments. We demonstrated that CXCR3 signaling is required for metformin-mediated induction of OXPHOS in CD8+ T cells, suggesting a possible CXCR3- and AMPK-dependent effect(s). How CXCR3 regulates metformin-mediated induction of OXPHOS, mitochondrial health and expansion of CD8 memory remains to be determined. Interestingly, mice devoid of AMPK in CD4+ T cells shows deficiency in the CXCR3 expression60. Our transcriptomic analysis of CD8+ T cells from Cxcr3−/− mice suggested dependency of metformin-mediated reprograming of mitochondrial function on Foxo1 network (Fig. 2i, j), which is an important metabolic check point40. Of note, TB immunopathology could be modulated by targeting T cells mitochondrial metabolism61.
The recall ability of TM cells is also the basis for vaccinations against numerous diseases including TB. The current TB vaccine, BCG, is unable to protect against pulmonary TB in adults, hence there is an urgent need to improve BCG or to develop new TB vaccine. Here we showed that metformin treatment resulted in an increased (i) elicitation of Ag-specific CD8+ TCM in the tissues of BCG-vaccinated mice, (ii) TNF secreting Ag-specific CD8+ TCM cells in the tissues of BCG-vaccinated Mtb-challenged mice, and (iii) protection of BCG-immunized mice and guinea pig against Mtb challenge. Importantly, expression of CXCR3 by TB vaccine-specific CD8+ T cells is needed for their homing to the lung mucosa62. This raises the possibility of the presence of metformin-educated CD8+ T cells in different anatomical location within lung parenchyma, which is responsible for enhanced protective immunity. We did not see enhanced immunogenicity among Ag-specific CD4+ TCM cells from metformin-treated BCG-vaccinated mice, which goes in line with no benefit of metformin-educated Ag-inexperienced CD4+ T cells on Mtb growth in adoptive transfer experiments (Fig. 1b, c). The differential effect of metformin on the CD4+ and CD8+ cells in naive or vaccinated host is not clear at present. However, as metformin normalizes CD4+ T-cell mitochondrial function to alleviate lupus in B6.Sle1.Sle2.Sle3 mice63 and aging-associated inflammation64 in humans, an effect of metformin on CD4+ T-cell metabolism in Mtb-infected hosts cannot be excluded. Nevertheless, our results support the concept of enhancing the efficacy of BCG vaccine by small molecule drugs65.
In conclusion, the data presented here show that metformin shifts CD8+ T-cell metabolism, leading to the expansion of a memory-like CD8+CXCR3+ phenotype with functional properties that contribute to host control of Mtb infection and disease. Our findings are in line with evidence that metabolic reprograming by metformin reverses CD8+ T-cell dysfunction during chronic Mtb infection66. Our current and previously reported studies show that metformin ameliorates TB immune pathology, whereas others reported that it reverses established pulmonary fibrosis in the bleomycin model67. That evidence, together with our demonstration that metformin enhances BCG vaccine immunogenicity and protective efficacy adds to the basis in evidence supporting metformin as a candidate for TB-HDT and as a TB vaccine adjunct.
Healthy human samples
PBMCs were isolated from eight male healthy non-obese volunteers without medication, kidney function loss, or metabolic disorders, who had received metformin for 5 days in increasing doses starting at 500 mg once a day and ending with 1000 mg twice a day46. Blood was drawn at baseline time point before metformin intake (T0) and immediately after metformin intake on day 6 (T6).
Diabetic human patient cohort
Patients with T2D included in the study are part of the Singapore Longitudinal Aging study 2 (SLAS-2), which is a population-based cohort to study the biology of aging among Singaporean elderly individuals >55 years old68. PBMCs from the blood samples collected from overnight fasting individuals were washed with PBS and cryopreserved in liquid nitrogen. PBMCs were thawed and used for flow cytometric staining. Fasting blood glucose, blood count, hematocrit level, albumin, creatinine, estimated glomerular filtration rate was performed on the blood samples.
CD45.1, CD45.2, Cxcr3−/−, TCRβδ−/−, and OT-I mice were obtained from SIgN mouse core. All mice strains were on C57BL/6 J background. CD45.1 and CD45.2 mice for chimeric in vivo experiments were obtained from Jackson Labs (Bar Harbor, USA), and performed at UMMS. For all in vivo experiments female mice aged between 8 and 12 weeks were used. Hartley strain female guinea pigs aged between 7 and 8 weeks (350–400 g) were obtained from Charles Rivers (Wilmington, MA), maintained on a normal chow diet, acclimated for 1 week then randomly assigned to four groups. Animals were housed in the SPF and BSL3 animal facilities on a 12/12 h light/dark cycle, with food and water ad libitum.
Metformin treatment for mice and guinea pigs
Metformin (1,1-Dimethylbiguanide hydrochloride, Sigma Aldrich) was provided to mice in drinking water (1.83 mg/ml which is equivalent to 270 mg/kg). Details of metformin treatment for different experimental models is provided in the figures. Guinea pigs were treated with 25 mg/kg of metformin suspended in carrier (Ora-Sweet, Paddock Laboratories, Minneapolis, MN), or carrier alone, and administered orally once daily, for 30 days, prior to Mtb infection. The metformin dose was determined in guinea pigs based on PK/PD analysis in which the peak plasma concentration reached 1.1 μg/mL (peak plasma concentration in human 1.1 μg/mL).
Female mice aged 8–10 weeks were vaccinated subcutaneous with a single dose BCG Pasteur strain (1 × 106 colony-forming units (CFU) in 100 µl PBS). In BCG-vaccinated mice metformin treatment was initiated 1 day post vaccination. Guinea pigs were vaccinated intra-dermally (i.d) with 1 × 105 BCG Pasteur strain. In vaccinated guinea pigs metformin treatment was initiated 14 days post vaccination. Sham vaccinated guinea pigs received an equal volume of saline i.d.
Female mice were infected via aerosol or intra tracheal (i.t)69 route with Mtb H37Rv or Mtb Erdman to deliver 100–200 viable bacilli per lung. For BCG infection experiments, 10 × 106 BCG bacilli were used. The implanted bacterial loads in the lungs of infected mice were quantified 1 day post infection. Guinea pigs were infected with the Mtb H37Rv by aerosol to deliver 10–20 bacilli per animal and confirmed by culture.
Modified Cornell model
Starting 40 days p.i Mtb-infected female mice were given INH (10 µg/mL) and PZA (15 mg/mL) in drinking water ad libitum70 for 70 days. This was followed by a 40 day rest for all treatments. Subsequently, at day 150 p.i mice were treated with 1 mg/mL of dexamethasone in drinking water for 15 days.
Culturing of mycobacterial strains
Mtb H37Rv, Erdman, and M. bovis BCG were grown in Middlebrook 7H9 broth (BBL Microbiology Systems) supplemented with 10% ADS (Albumin, Dextrose, Saline, Sigma) enrichment medium with 140 mM NaCl, 0.5% glycerol and 0.5% Tween-80 at 37 °C for 5–10 days to an OD600 of 0.4–0.5. Bacterial cultures were harvested, re-suspended in 7H9 broth with 20% glycerol, and aliquots were frozen at −80 °C until later use. Before infection, cells were thawed, washed, and sonicated. Viable cell counts per mL in thawed aliquots were determined by plating serial dilutions on to Middlebrook 7H11 agar plates supplemented with 10% OADC (Oleic acid, Albumin, Dextrose, Catalase, BBL Microbiology Systems) and 0.5% glycerol followed by incubation at 37 °C for three weeks.
Enumeration of Mtb load in the tissues of infected animals
Mycobacterial loads in lung and spleen of euthanized mice were evaluated at different time points post infection by homogenizing the organs in PBS or PBS with 0.05% Tween-80 using MACS tissue dissociator (Miltenyi Biotec). Guinea pigs were killed with an intra-peritoneal overdose of pentobarbital (>120 mg/kg), and representative lung tissue samples were collected sterilely and homogenized in 9 ml of sterile saline. Tenfold serial dilutions of tissue homogenates in sterile PBS were plated in triplicates on Middlebrook 7H11 agar plates supplemented with 10% OADC and 0.5% glycerol, incubated for 3–6 weeks at 37 °C and CFUs counted.
Cell isolation from lungs and spleens
Lungs were aseptically isolated from mice and incubated at 37 °C in 10% FBS RPMI media containing 1.4 mg/ml collagenase A (Roche) and 30 µg/ml DNase I for 60 min. The treated lung tissue and spleen was dissociated over the 70 μm cell strainer (Fisherbrand). Strainer was washed to collect single-cell suspension. Red blood cells were lysed with ACK lysing buffer (Lonza) for 5 min following by washing of cells with culture media. Cells were counted and adjusted to 5 × 106 cells/ml.
Magnetic sorting of murine T cells
Spleen cells were washed twice in fresh magnetic-activated cell sorting (MACS) buffer. CD4+ and CD8+ T cells were magnetically purified via negative selection using Dynabeads or MACS cell separation system (Supplementary Table 7) according to the manufacturer’s protocols, followed by purity check which was around 90-95%.
FACS sorting of murine memory CD8+ T cells
MACS-sorted splenic CD8+ T cells were stained with live dead marker, blocked with Fc block and surface stained with anti-CD3-FITC, anti-CD4-Pacific blue, anti-CD8-APC-Cy7, anti-CD44-BV650, and anti-CD62L-PE-Cy7 (Supplementary Table 8). Stained cells were sorted on fluorescence-activated cell sorting (FACS) Aria II, FACS Aria III, or BD Influx (Becton Dickinson, San Jose, CA, USA), followed by purity check which was ~90–95%.
Adoptive T-cell transfers
Purified splenic CD4+ or CD8+ T cells from untreated or metformin-treated CD45.1 WT or CD45.2 WT mice were transferred via tail vein injection into sub lethally irradiated (600 rads) CD45.2 WT mice (Fig. 1). MACS-sorted splenic CD8+ T cells or TM or TE from untreated or metformin-treated CD45.1 WT mice were transferred via intra orbital injection into CD45.2 WT or Tcrβδ−/− (Fig. 3d, e). The details of number of T cells transferred are in Supplementary Table 9.
Real-time mitochondrial OCR and ECAR of splenic CD8+ T cells was measured in media (Dulbecco’s modified eagle medium; DMEM containing 25 mM glucose and 2 mM l-glutamine), in response to 1 µM oligomycin, 5 µM FCCP, and 0.5 µM rotenone/antimycin A, using Seahorse XF Cell Mito Stress and Glyco Stress Test Kit and XFe96 Extracellular Flux Analyzer (Agilent Technologies, USA, Supplemental Table 7). FAO assay were carried out using Seahorse XF Palmitate-BSA FAO Substrate kit in media (KHB containing 2.5 mM glucose and 0.5 mM Carnitine) in response to 1 µM oligomycin, 1 µM FCCP, and 0.5 µM rotenone/antimycin A. For FAO assay, cells were prior starved in Seahorse DMEM medium (DMEM containing 0.5 mM glucose, 1 mM l-glutamine, and 0.5 mM Carnitine) for 2 hrs. Also for FAO assay, BSA or Palmitate:BSA was added into the 96-well plate and 200 µM etomoxir were used additionally as drug injection. SRC was determined as the absolute increase in OCR after FCCP injection compared with basal OCR. Data was analyzed with Wave software, version 2.4 (Agilent Technologies).
Single-cell suspensions of spleens, lungs, and lymph nodes (from mice) were prepared at the indicated time points. Live dead staining was performed for 15 minutes at room temperature or for 20 minutes at 37 °C. Non-specific antibody binding was blocked by adding anti-CD16/32 (Fc block) (BD Biosciences). Following this surface markers were stained for 20–25 minutes at 4 °C with the following anti-mouse antibodies: anti-CD45-Pacific blue/BV605/ FITC, CD45.1-BV711, CD45.2-APC, anti-CD3-FITC/ Pacific blue/ BUV395/ PerCP-Cy5.5, anti-CD4-Pacific blue/Alexa Fluor 700/PerCP-Cy5.5/PE, anti-CD8-APC-Cy7/ FITC, anti-CD44-BV650/ PE-Cy7/ PE-Cy5, anti-CD62L- PE-Cy7/ PE-CF594 and anti-CXCR3-BV605 (Supplementary Table 8). In some experiments, mitochondrial staining was performed with 200 nM Mitotracker green and/or 20 µM TMRM in RPMI media without FCS for 30 minutes at 37 °C prior to staining of surface markers.
For intracellular cytokine staining, spleen cells were ex vivo stimulated for 20 h with 40 μg/ml PPD (purified protein derivative, Statens Serum Institute, Denmark) in the presence of Golgi StopTM (BD Biosciences) and Monensin (1000×, BioLegend) for the last 4 hours, at 37 °C and 5% CO2. In some experiment’s spleen cells were ex vivo stimulated with IL-2 and TB10.4 peptides for 5 hours at 37 °C. Brefeldin A was added to the stimulation for the last 4 hours. Live dead staining and blocking was performed. Cells were then stained with anti-CD45-BV605, anti-CD3-Pacific blue/PerCP-Cy5.5, anti-CD4-PE/Alexa Fluor 700, anti-CD8a-APC-Cy7/FITC, anti-CD44-PE-Cy7/PE-Cy5 and anti-CD62L-AlexaFluor 700/PE-Cy7 antibodies for 25 min at 4 °C (Supplementary Table 8). After washing, cells were fixed and permeabilized for 20 minutes with Cytofix/CytopermTM (BD Biosciences). Permeabilized cells were washed with Perm/Wash bufferTM (BD Biosciences) and stained with anti-IFN-γ-APC/BV 421, TNF-APC or Perforin-PE antibody for 25 minutes at 4 °C. Subsequently, cells were fixed in 2% paraformaldehyde before acquisition.
Human PBMCs were stained with live dead marker and following anti-human antibodies: anti-CD34-PE, anti-CD4-BV750, anti-CD8-BV605, anti-CD14-BV605, anti-TCRγ/δ-APCR700, anti-CD27-BV510, anti-CD45-BUV805, anti-CxCR3-BV650, anti-TCRVδ2-BV711, anti-CD3-BV570, anti-Vα7.2-APC-Cy7, anti-CD45RO-PerCP/Cy5.5, and anti-CD161-BV785 (Supplementary Table 10). Dilution of different antibodies used was either 1:50, 1:100, or 1:200. Each antibody was validated for flow cytometry staining as per manufacturer’s description. Fluorescence-minus-one and isotype controls were used during validation. Stained cells were acquired using a LSRII cytometer with five lasers (BD, San Jose, CA, USA) or Fortessa cytometer with four lasers (Becton Dickinson, San Jose, CA, USA) or Symphony cytometer (BD Biosciences, California, USA) and data was collected using FACSDivaTM version 7 (BD Biosciences). Flow cytometry data were analyzed with FlowJoTM version 10.3 (Tree Star, Inc., USA).
Assessment of caspase-1 activity
The activity of caspase-1 was determined using fluorescent inhibitor of caspase-1 (FAM-Flica Caspase-1 Assay kit, ImmunoChemistry Technologies). Staining procedure was performed according to manufactures protocol. In brief after live/dead and surface antibody staining of lung and spleen single cells, FLICA reagent FAM-YVAD-FMK was provided to single cells and binding to activated caspase-1 was allowed for 1 h at 37 °C in the dark. Stained cells were washed with apoptosis wash buffer, followed by fixation. The green fluorescent was detected in FITC channel and samples were acquired using LSRII cytometer.
Glucose uptake assay
2-(N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino)-2-deoxyglucose (2-NBDG) uptake was performed on glucose-deprived cultures of spleen and lung single cells. Cells were incubated in media containing 200 µM 2-NBDG for 20 minutes at 37 °C. Followed this live dead and surface marker staining was performed. Fluorescent of 2-NBDG was detected in FITC channel using Fortessa cytometer.
CyTOF staining and data analysis
Spleen and lung cells from mice, and human PBMCs were first stained with 200 µM cisplatin (Sigma) for the discrimination of dead cells. Upon this, cells were washed and stained with a primary antibodies cocktail for 30 minutes at 4 °C (Supplementary Table 1 (for human antibodies) and Supplementary Table 5 (for mouse antibodies)). Dilution of different antibodies used was either 1:100 or 1:200 and was standardized by our CyTOF platform. After washing, cells were stained with the heavy metal-labeled antibodies cocktail for 30 minutes at 4 °C. Cells were washed, fixed and permeabilised using fixation FoxP3 buffer (eBioscience) for 45 minutes at 4 °C. Upon this cells were stained with intranuclear antibodies cocktail for 30 minutes at 4 °C. Labeled cells were washed twice with PBS and fixed in 2% or 4% PFA overnight. Subsequently, cells were washed and stained for barcodes (30 minutes at 4 °C) and DNA (10 minutes at 4 °C). Samples were acquired on the Helios CyTOF (Fluidigm), and analyzed as mentioned before24,25,71. For details, see supplementary methods.
Ex vivo culture of primary T cells
OT-1 splenocytes were activated with 20 µg/ml OVA protein (Invivogen) and 100 U/ml hIL-2 (Miltenyi Biotec) and cultured in IMDM media (supplemented with 10% FBS, 2 mM L-glutamine, 1 mM Sodium pyruvate, 0.1 mM NEAA, 1% Kanamycin) for 3 days. After washing, cells were subsequently cultured with 0.4, 2, 10 ng/ml of hIL-15 (R&D) and/or 2 mM MET for 4 days. IL-15-derived antigen-specific CD8+ T cells were analyzed for memory markers by flow cytometry.
Real-time quantitative RT-PCR
Cells or tissues were lysed in TRIzol Reagent (Invitrogen), and total RNA was extracted using the RNeasy Mini Kit (Qiagen). RNA was reverse-transcribed using the iScript cDNA synthesis Kit (Bio-Rad). Complementary DNA (cDNA) was used for real-time quantitative PCR using 2x iQ SYBR Green supermix and C1000 Thermal Cycler (Bio-Rad) using specific primers (see supplemental methods). The relative mRNA expression level of each gene was normalized to Gapdh housekeeping gene expression in the untreated condition, and fold induction was calculated by the ΔΔCT method relative to those in untreated cells per tissue.
RNAseq and analysis
Total RNA was extracted following the double extraction protocol: RNA isolation by TRIzol followed by a Qiagen RNeasy Micro clean-up procedure (Qiagen, Hilden, Germany). The extracted RNA samples were inspected on Agilent Bioanalyzer (Agilent Technologies, Santa Clara, CA) for quality assessment. All RNA samples were analyzed on Perkin Elmer Labchip GX system (Perkin Elmer, Waltham, MA, USA) for quality assessment with a median RNA Quality Score of 9. cDNA libraries were prepared using 500 pg of total RNA and 1ul of a 1:200,000 dilution of ERCC RNA Spike in Controls (Ambion, Thermo Fisher Scientific, Waltham, MA, USA) using SMARTSeq v2 protocol72 except for the following modifications: 1. Use of 20 µM TSO; 2. Use of 250 pg of cDNA with 1/5 reaction of Illumina Nextera XT kit (Illumina, San Diego, CA, USA). The length distribution of the cDNA libraries was monitored using DNA High Sensitivity Reagent Kit on the Perkin Elmer Labchip GX system (PerkinElmer, Waltham, MA, USA). The libraries were qPCR quantified (Kapa Biosystems, Wilmington, MA) to ascertain the loading concentration. RNAseq libraries of FACS sorted CD8 TM cells subjected to an indexed PE sequencing run of 2 × 51 cycles on an Illumina HiSeq 2000. FASTQ files were checked for quality control issues using fastQC and none were found. The reads in the FASTQ files were mapped onto the mouse genome build mm10 using STAR and the gene counts counted using featureCounts (part of the Subread package). DEGs were detected using DESeq273 running in R version 3.3.3. DEGs having a p value <0.05 (calculated by DESeq273 using t test) were considered to be statistically significant. DEGs were used in IPA for a core analysis. IPA uses Fisher’s exact test to determine the probability of the association of genes with the pathway/molecular function. Heat maps were generated using the package ComplexHeatmap in R version 3.3.3.
Enrichment of FOXO1 targets
The overlap between the DEGs from the RNAseq comparison of CD8+ T cells from Cxcr3−/− vs WT mice and the FOXO1 target genes from Webb et al.36 was determined (Supplementary Data 1). The hypergeometric distribution was used to determine the significance of this overlap/enrichment using all the RNAseq genes tested and the genes analyzed in Webb et al.36 as the background. A p value of <0.05 was considered to be significant. Analysis conducted in R version 3.3.3.
Histology and morphometry of guinea pigs lung samples
Left caudal lung lobe of guinea pigs was collected for histopathology and was fixed in 4% paraformaldehyde for >48 hours, processed and embedded in paraffin, sectioned and stained with hematoxylin and eosin. The Mtb lesion burden was determined morphometrically using the area fraction fractionator method with Stereo investigator software (MBF Biosciences, Williston, VT). The data was expressed as the percent of lung area affected by characteristic Mtb lesions. Photomicrographs were obtained from tissue sections representative of the mean values within individual animal treatment groups.
All values are mean ± SD or SEM of individual samples. Data analysis was performed with GraphPad Prism Software (GraphPad Software Inc., version 7.01). The statistical tests utilized are two-tailed and respective details have been indicated in figure legends.
The study was approved by the Institutional Biosafety Committee (IBC) and Institutional Animal Care and Use Committee (IACUC) of the (i) Biological Resource Council (BRC), A*STAR, Singapore; (ii) Defence Science Organization (DSO) national laboratories, Singapore, (iii) Colorado State University, USA, and (iv) University of Massachusetts Medical School (UMMS), USA. The study in human healthy volunteers and diabetic patients was approved by the Arnhem-Nijmegen Ethical Committee (NL47793.091.14) and National University of Singapore Institutional Review Board (NUS-IRB #04-140), respectively. Informed consent from all participants was obtained.
Further information on research design is available in the Nature Research Reporting Summary linked to this article.
RNAseq data that support the findings of this study have been deposited in GeneBank GEO, accession GSE139948. Other data are available from the corresponding author upon request.
Wallis, R. S. & Hafner, R. Advancing host-directed therapy for tuberculosis. Nat. Rev. Immunol. 15, 255–263 (2015).
Kaufmann, S. H. E., Dorhoi, A., Hotchkiss, R. S. & Bartenschlager, R. Host-directed therapies for bacterial and viral infections. Nat. Rev. Drug Disco. 17, 35–56 (2018).
Singhal, A. et al. Metformin as adjunct antituberculosis therapy. Sci. Transl. Med. 6, 263ra159 (2014).
Magee, M. J., Salindri, A. D., Kornfeld, H. & Singhal, A. Reduced prevalence of latent tuberculosis infection in diabetes patients using metformin and statins. Eur. Respir. J. 53, 1801695 (2019).
Pan, S. W. et al. The risk of TB in patients with type 2 diabetes initiating metformin vs sulfonylurea treatment. Chest 153, 1347–1357 (2018).
Lin, S. Y. et al. Metformin is associated with a lower risk of active tuberculosis in patients with type 2 diabetes. Respirology 23, 1063–1073 (2018).
Tseng, C. H. Metformin decreases risk of tuberculosis infection in type 2 diabetes patients. J. Clin. Med. 7, 264 (2018).
Degner, N. R., Wang, J. Y., Golub, J. E. & Karakousis, P. C. Metformin use reverses the increased mortality associated with diabetes mellitus during tuberculosis treatment. Clin. Infect. Dis. 66, 198–205 (2018).
Lee, Y. J. et al. The effect of metformin on culture conversion in tuberculosis patients with diabetes mellitus. Korean J. Intern. Med. 33, 933–940 (2018).
Lin, P. L. & Flynn, J. L. CD8 T cells and Mycobacterium tuberculosis infection. Semin. Immunopathol. 37, 239–249 (2015).
Graef, P. et al. Serial transfer of single-cell-derived immunocompetence reveals stemness of CD8(+) central memory T cells. Immunity 41, 116–126 (2014).
Haluszczak, C. et al. The antigen-specific CD8+ T cell repertoire in unimmunized mice includes memory phenotype cells bearing markers of homeostatic expansion. J. Exp. Med. 206, 435–448 (2009).
Lee, Y. J., Jameson, S. C. & Hogquist, K. A. Alternative memory in the CD8 T cell lineage. Trends Immunol. 32, 50–56 (2011).
Huang, W., Hu, J. & August, A. Cutting edge: innate memory CD8+ T cells are distinct from homeostatic expanded CD8+ T cells and rapidly respond to primary antigenic stimuli. J. Immunol. 190, 2490–2494 (2013).
Lee, J. Y., Hamilton, S. E., Akue, A. D., Hogquist, K. A. & Jameson, S. C. Virtual memory CD8 T cells display unique functional properties. Proc. Natl Acad. Sci. USA 110, 13498–13503 (2013).
Lin, J. S. et al. Virtual memory CD8 T cells expanded by helminth infection confer broad protection against bacterial infection. Mucosal Immunol. 12, 258–264 (2019).
Lauvau, G., Boutet, M., Williams, T. M., Chin, S. S. & Chorro, L. Memory CD8(+) T cells: innate-like sensors and orchestrators of protection. Trends Immunol. 37, 375–385 (2016).
Geltink, R. I. K., Kyle, R. L. & Pearce, E. L. Unraveling the complex interplay between T cell metabolism and function. Annu. Rev. Immunol. 36, 461–488 (2018).
Krauss, S., Brand, M. D. & Buttgereit, F. Signaling takes a breath–new quantitative perspectives on bioenergetics and signal transduction. Immunity 15, 497–502 (2001).
van der Windt, G. J. et al. Mitochondrial respiratory capacity is a critical regulator of CD8+ T cell memory development. Immunity 36, 68–78 (2012).
Pearce, E. L. et al. Enhancing CD8 T-cell memory by modulating fatty acid metabolism. Nature 460, 103–107 (2009).
Quinn, K. M. et al. Metabolic characteristics of CD8(+) T cell subsets in young and aged individuals are not predictive of functionality. Nat. Commun. 11, 2857 (2020).
Gallegos, A. M. et al. A gamma interferon independent mechanism of CD4 T cell mediated control of M. tuberculosis infection in vivo. PLoS Pathog. 7, e1002052 (2011).
Cheng, C. Y. et al. Host sirtuin 1 regulates mycobacterial immunopathogenesis and represents a therapeutic target against tuberculosis. Sci. Immunol. 2, eaaj1789 (2017).
Fehlings, M. et al. Checkpoint blockade immunotherapy reshapes the high-dimensional phenotypic heterogeneity of murine intratumoural neoantigen-specific CD8(+) T cells. Nat. Commun. 8, 562 (2017).
Groom, J. R. & Luster, A. D. CXCR3 in T cell function. Exp. Cell Res. 317, 620–631 (2011).
Koch, M. A. et al. The transcription factor T-bet controls regulatory T cell homeostasis and function during type 1 inflammation. Nat. Immunol. 10, 595–602 (2009).
Hu, J. K., Kagari, T., Clingan, J. M. & Matloubian, M. Expression of chemokine receptor CXCR3 on T cells affects the balance between effector and memory CD8 T-cell generation. Proc. Natl Acad. Sci. USA 108, E118–E127 (2011).
Slutter, B., Pewe, L. L., Kaech, S. M. & Harty, J. T. Lung airway-surveilling CXCR3(hi) memory CD8(+) T cells are critical for protection against influenza A virus. Immunity 39, 939–948 (2013).
Schluns, K. S., Williams, K., Ma, A., Zheng, X. X. & Lefrancois, L. Cutting edge: requirement for IL-15 in the generation of primary and memory antigen-specific CD8 T cells. J. Immunol. 168, 4827–4831 (2002).
Kurachi, M. et al. Chemokine receptor CXCR3 facilitates CD8(+) T cell differentiation into short-lived effector cells leading to memory degeneration. J. Exp. Med. 208, 1605–1620 (2011).
Kimura, M. Y. et al. IL-7 signaling must be intermittent, not continuous, during CD8(+) T cell homeostasis to promote cell survival instead of cell death. Nat. Immunol. 14, 143–151 (2013).
Gattinoni, L. et al. Wnt signaling arrests effector T cell differentiation and generates CD8+ memory stem cells. Nat. Med. 15, 808–813 (2009).
Haque, M. et al. C-Myc regulation by costimulatory signals modulates the generation of CD8+ memory T cells during viral infection. Open Biol. 6, 150208 (2016).
Delpoux, A. et al. Continuous activity of Foxo1 is required to prevent anergy and maintain the memory state of CD8(+) T cells. J. Exp. Med 215, 575–594 (2018).
Webb, A. E., Kundaje, A. & Brunet, A. Characterization of the direct targets of FOXO transcription factors throughout evolution. Aging Cell 15, 673–685 (2016).
Kim, M. V., Ouyang, W., Liao, W., Zhang, M. Q. & Li, M. O. The transcription factor Foxo1 controls central-memory CD8+ T cell responses to infection. Immunity 39, 286–297 (2013).
Link, W. & Fernandez-Marcos, P. J. FOXO transcription factors at the interface of metabolism and cancer. Int. J. Cancer 141, 2379–2391 (2017).
Cabrera-Ortega, A. A., Feinberg, D., Liang, Y., Rossa, C. Jr. & Graves, D. T. The role of forkhead box 1 (FOXO1) in the immune system: dendritic cells, T cells, B cells, and hematopoietic stem cells. Crit. Rev. Immunol. 37, 1–13 (2017).
Wilhelm, K. et al. FOXO1 couples metabolic activity and growth state in the vascular endothelium. Nature 529, 216–220 (2016).
Raud, B. et al. Etomoxir actions on regulatory and memory T cells are independent of Cpt1a-mediated fatty acid oxidation. Cell Metab. 28, 504–515 e507 (2018).
Nair, V. et al. Mechanism of metformin-dependent inhibition of mammalian target of rapamycin (mTOR) and Ras activity in pancreatic cancer: role of specificity protein (Sp) transcription factors. J. Biol. Chem. 289, 27692–27701 (2014).
John, G. B. et al. The mitochondrial inner membrane protein mitofilin controls cristae morphology. Mol. Biol. Cell 16, 1543–1554 (2005).
Man, S. M. & Kanneganti, T. D. Converging roles of caspases in inflammasome activation, cell death and innate immunity. Nat. Rev. Immunol. 16, 7–21 (2016).
Ordway, D. et al. The cellular immune response to Mycobacterium tuberculosis infection in the guinea pig. J. Immunol. 179, 2532–2541 (2007).
Lachmandas, E. et al. Metformin alters human host responses to Mycobacterium tuberculosis in healthy subjects. J. Infect. Dis. 220, 139–150 (2019).
Romero, P. et al. Four functionally distinct populations of human effector-memory CD8+ T lymphocytes. J. Immunol. 178, 4112–4119 (2007).
Grayson, J. M., Zajac, A. J., Altman, J. D. & Ahmed, R. Cutting edge: increased expression of Bcl-2 in antigen-specific memory CD8+ T cells. J. Immunol. 164, 3950–3954 (2000).
Foreman, T. W. et al. CD4+ T-cell-independent mechanisms suppress reactivation of latent tuberculosis in a macaque model of HIV coinfection. Proc. Natl Acad. Sci. USA 113, E5636–E5644 (2016).
Scanga, C. A. et al. Reactivation of latent tuberculosis: variations on the Cornell murine model. Infect. Immun. 67, 4531–4538 (1999).
Young, C., Walzl, G. & Du Plessis, N. Therapeutic host-directed strategies to improve outcome in tuberculosis. Mucosal Immunol. 13, 190–204 (2019).
White, J. T., Cross, E. W. & Kedl, R. M. Antigen-inexperienced memory CD8(+) T cells: where they come from and why we need them. Nat. Rev. Immunol. 17, 391–400 (2017).
Alanio, C. et al. CXCR3/CXCL10 axis shapes tissue distribution of memory phenotype CD8(+) T cells in nonimmunized mice. J. Immunol. 200, 139–146 (2018).
Oghumu, S. et al. Distinct populations of innate CD8+ T cells revealed in a CXCR3 reporter mouse. J. Immunol. 190, 2229–2240 (2013).
Oghumu, S. et al. CXCR3 expression defines a novel subset of innate CD8+ T cells that enhance immunity against bacterial infection and cancer upon stimulation with IL-15. FASEB J. 29, 1019–1028 (2015).
van der Windt, G. J. et al. CD8 memory T cells have a bioenergetic advantage that underlies their rapid recall ability. Proc. Natl Acad. Sci. USA 110, 14336–14341 (2013).
Rolf, J. et al. AMPKalpha1: a glucose sensor that controls CD8 T-cell memory. Eur. J. Immunol. 43, 889–896 (2013).
Blagih, J. et al. The energy sensor AMPK regulates T cell metabolic adaptation and effector responses in vivo. Immunity 42, 41–54 (2015).
Adams, W. C. et al. Anabolism-associated mitochondrial stasis driving lymphocyte differentiation over self-renewal. Cell Rep. 17, 3142–3152 (2016).
Lepez, A. P. et al. Long-term homeostatic T cell proliferation is driven by AMPK-dependent regulation of oxygen reactive species. Biorxiv (2020) https://doi.org/10.1101/2020.03.06.978791.
Tzelepis, F. et al. Mitochondrial cyclophilin D regulates T cell metabolic responses and disease tolerance to tuberculosis. Sci. Immunol. 3, eaar4135 (2018).
Jeyanathan, M. et al. CXCR3 signaling is required for restricted homing of parenteral tuberculosis vaccine-induced t cells to both the lung parenchyma and airway. J. Immunol. 199, 2555–2569 (2017).
Yin, Y. et al. Normalization of CD4+ T cell metabolism reverses lupus. Sci. Transl. Med. 7, 274ra218 (2015).
Bharath, L. P. et al. Metformin enhances autophagy and normalizes mitochondrial function to alleviate aging-associated inflammation. Cell Metab. 32, 44–55 e46 (2020).
Bhattacharya, D. et al. Simultaneous inhibition of T helper 2 and T regulatory cell differentiation by small molecules enhances Bacillus Calmette-Guerin vaccine efficacy against tuberculosis. J. Biol. Chem. 289, 33404–33411 (2014).
Russell, S. L. et al. Compromised metabolic reprogramming is an early indicator of CD8(+) T cell dysfunction during chronic mycobacterium tuberculosis infection. Cell Rep. 29, 3564–3579 e3565 (2019).
Rangarajan, S. et al. Metformin reverses established lung fibrosis in a bleomycin model. Nat. Med. 24, 1121–1127 (2018).
Tizazu, A. M. et al. Metformin monotherapy downregulates diabetes-associated inflammatory status and impacts on mortality. Front Physiol. 10, 572 (2019).
Singhal, A. et al. Experimental tuberculosis in the Wistar rat: a model for protective immunity and control of infection. PLoS ONE 6, e18632 (2011).
Cheigh, C. I. et al. Posttreatment reactivation of tuberculosis in mice caused by Mycobacterium tuberculosis disrupted in mce1R. J. Infect. Dis. 202, 752–759 (2010).
Levine, J. H. et al. Data-driven phenotypic dissection of aml reveals progenitor-like cells that correlate with prognosis. Cell 162, 184–197 (2015).
Picelli, S. et al. Full-length RNA-seq from single cells using Smart-seq2. Nat. Protoc. 9, 171–181 (2014).
Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550 (2014).
We thank SIgN Flow Facility, ABSL3, BRC laboratory personnel and Yannick Simoni, Nicholas Ang, and Kandhadayar G. Srinivasan for assistance. This research was supported by SIgN A*STAR; Immunomonitoring platform grant to SIgN (#BMRC/IAF/311006, #H16/99/b0/011, and #NRF2017_SISFP09); NIH Grant #R01HL081149 and #U19AI111224 (to H.K.) and #R01HL152078 (to H.K. and A.S.), ERC Advanced Grant (#833247) and a Spinoza grant of the Netherlands Organization for Scientific Research to M.G.N.
The authors declare no competing interests.
Peer review information Nature Communications thanks Maziar Divangahi, Thomas Lindenstrøm, and the other, anonymous reviewer(s) for their contribution to the peer review of this work.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
About this article
Cite this article
Böhme, J., Martinez, N., Li, S. et al. Metformin enhances anti-mycobacterial responses by educating CD8+ T-cell immunometabolic circuits. Nat Commun 11, 5225 (2020). https://doi.org/10.1038/s41467-020-19095-z
This article is cited by
Respiratory Research (2023)
Nature Reviews Endocrinology (2023)
Nature Reviews Microbiology (2022)
Nature Immunology (2021)
Applied Microbiology and Biotechnology (2021)