Metformin and 2-Deoxyglucose Collaboratively Suppress Human CD4⁺ T Cell Effector Functions and Activation-Induced Metabolic Reprogramming

During inflammation, T cells activated by cognate Ags undergo global cellular rewiring to produce proinflammatory cytokines and prepare for clonal expansion, increasing their metabolic demands for biosynthetic precursors and reducing equivalents necessary for cell proliferation, cytokine production, and other effector functions (1). Following antigenic ligation of the TCR and inputs from costimulatory molecules, T cells rapidly increase their rates of glucose uptake and aerobic glycolysis. Uptake of amino acids such as glutamine and anaplerotic processes such as glutaminolysis are also upregulated to replenish rapidly consumed TCA cycle intermediates (2–4). Furthermore, the importance of metabolic homeostasis in immune cells is evidenced, at least in part, by the dysregulated metabolic profiles that have been detected in immune cells from patients with autoimmune diseases. CD4⁺ T cells from systemic lupus erythematosus patients display hallmarks of oxidative stress and mitochondrial hyperpolarization (5, 6), and CD4⁺ T cells from rheumatoid arthritis patients exhibit reduced oxidation of glucose for ATP generation and increased shunting of glucose into the pentose phosphate pathway (7).

Given the central role of metabolic reprogramming in T cell activation and differentiation (8) and the enhanced glycolytic signatures of T_H1 and T_H17 subsets implicated in the pathogenesis of inflammatory diseases such as psoriasis and inflammatory bowel disease (9), the inhibition of key metabolic pathways that are upregulated during T cell inflammation presents a credible therapeutic strategy for the treatment of autoimmune diseases. Indeed, an increasing number of studies have demonstrated that the inhibition of anabolic or anaplerotic processes ameliorates disease severity in murine models of autoimmune disease (10, 11). Of note, the antidiabetic agent metformin, best studied in the context of its suppression of hepatic gluconeogenesis (12–14), has more recently been ascribed anti-inflammatory properties in mouse models of autoimmune disease (15). Metformin inhibits complex I in the electron transport chain (16–18) and oxidative phosphorylation signatures in immune cells from human cohorts (19). Both metformin and 2-deoxyglucose (2-DG), an inhibitor of hexokinase and phosphoglucoisomerase (20, 21), have been used to study the inhibition of oxidative phosphorylation and glycolysis, respectively, in immune cells. In a mouse skin allograft model of graft versus host disease, simultaneous inhibition with metformin, 2-DG, and the glutaminase inhibitor 6-diazo-5-oxo-l-norleucine suppressed allograft rejection, whereas metformin plus 2-DG treatment effectively suppressed mouse T cell cytokine production and proliferation in vitro (22). Furthermore, in murine models of systemic lupus erythematosus–like disease, dual treatment with metformin and 2-DG reversed lupus pathologies, but treatment with either metformin or 2-DG alone did not (11).

These studies demonstrate that the efficacy of inhibiting single metabolic pathways may be undermined by the plasticities and redundancies that exist within metabolic networks, allowing other pathways to compensate for the blockade of any one (11, 22, 23). However, although the reported mouse studies shed light on the enhanced suppression of T cell effector functions upon inhibition of multiple metabolic pathways, the mechanisms underlying the enhanced efficacy of the metformin + 2-DG combination treatment regimen have not been elucidated, and the extent to which these effects are preserved in human cells remains unclear.

To determine if the enhanced efficacy of metformin + 2-DG combination treatment is observed in human immune cells, we investigated how metformin + 2-DG treatment would alter primary human CD4⁺ T cell effector functions. We report in this study that metformin + 2-DG treatment more potently suppressed primary human CD4⁺ T cell IFN-γ production and cell proliferation following TCR stimulation than either metformin or 2-DG treatment alone. Transcriptomic analyses of human CD4⁺ T cells revealed that metformin + 2-DG treatment attenuated remodeling of metabolic transcriptional programs induced by TCR activation. Among key transcriptional factors implicated in metabolic reprogramming, MYC and HIF-1A expression were more strongly downregulated by metformin + 2-DG treatment compared with either treatment alone. Accordingly, metformin + 2-DG significantly suppressed metabolic genes and processes that are MYC dependent. In addition, we show that MYC, but not HIF-1A, is indispensable for TCR-induced proliferation and glutamine uptake in human CD4⁺ T cells. Our results provide significant insights into the effects of metabolic inhibition by metformin + 2-DG treatment of primary human T cells.

Primary human CD4⁺ T cells were isolated from whole blood with >90% postisolation purity using the RosetteSep Human CD4⁺ T Cell Enrichment mixture (STEMCELL Technologies, Vancouver, BC, Canada), and cultured at 37°C at 5% CO₂ in complete RPMI 1640 (RPMI 1640 supplemented with 10% FBS, 10 mM HEPES, 1 mM sodium pyruvate, 1× nonessential amino acids, and 500 μM GlutaMAX; Thermo Fisher Scientific, Waltham, MA). Whole blood from human donors was collected onsite into sodium heparin–containing tubes at Pfizer according to protocols approved by the Pfizer Institutional Review Board (protocol no. GOHW RDP-01) or was obtained from STEMCELL Technologies.

Primary human CD4⁺ T cells were activated with soluble anti-CD3 and anti-CD28 tetrameric complexes (ImmunoCult; STEMCELL Technologies) for 24 h in the presence of indicated compounds. Cells were collected, washed, resuspended in XF RPMI 1640 media (Agilent Technologies, Santa Clara, CA), and plated onto Seahorse XF cell plates coated with Cell-Tak (Corning Life Sciences, Corning, NY). A total of 0.4 × 10⁶ cells per well were analyzed with the Seahorse XF96 instrument (Agilent Technologies) using the glycolytic rate assay or the mitochondrial stress test according to manufacturer’s directions for glycolytic proton efflux rate (GlycoPER) or oxygen consumption rate (OCR) measurements, respectively. Data were analyzed using Wave software (version 2.6).

Supernatants were collected from cells activated with anti-CD3/anti-CD28 for 24 h. Cytokines in the T cell culture media were measured using the Meso Scale Discovery 10 plex V-plex human proinflammatory cytokine panel according to manufacturer’s instructions and analyzed using the Meso Scale Discovery MESO SECTOR 600 instrument (Meso Scale Discovery, Rockville, MD).

Cell proliferation was measured by flow cytometric analysis of Cell Trace Violet (CTV) dilution. Primary human CD4⁺ T cells were washed with PBS and stained with 0.5 μl CTV (Thermo Fisher Scientific) per 1 × 10⁶ cells, according to the manufacturer’s instructions. Stained cells were resuspended in complete RPMI 1640 for treatment and reactivation, then allowed to proliferate for 96 h before staining for viability with Near-IR LIVE/DEAD (Thermo Fisher Scientific). Cell surface markers were assessed on primary human CD4⁺ T cells activated with anti-CD3/anti-CD28 for 24 h, stained with near-infrared fluorescent dye, then incubated with Abs in FACS buffer (PBS containing 2% FBS, 1 mM EDTA, and 0.1% sodium azide), and fixed in FACS buffer containing 2% paraformaldehyde prior to data collection on a BD LSR II Fortessa (BD Biosciences, San Jose, CA). The following Abs conjugated to BUV395, APC, BV421, A647, and BV786 were from BD Biosciences: CD4 (catalog no. 563550), CCR7 (catalog no. 562555), CD27 (catalog no. 563327), CD45RA (catalog no. 550855), and CD98 (catalog no. 744502). Abs against SLC7A5 were from Novus Biologicals (catalog no. NBP2-50465AF647). Flow cytometry data were analyzed using and proliferation index calculated using FlowJo software (version 10) with the following gating strategy: side scatter versus forward scatter > singlets > live cells > CD4.

Activation-induced pS6 protein in human T cells was measured in cells activated with anti-CD3/anti-CD28 for 24 h, stained with near-infrared fluorescent dye, then fixed with FACS buffer containing 2% paraformaldehyde. Fixed cells were permeabilized with 90% methanol for 1 h at −20°C then stained with an anti-pS6 Ab S235/S236 (catalog no. 560435; BD Biosciences) overnight at 4°C (24). Cells were washed and resuspended in FACS buffer prior to data collection on a BD LSR II Fortessa (BD Biosciences). Flow cytometry data were analyzed using FlowJo software (version 10) using the same gating strategy described above.

RNA was isolated and purified (RNeasy kit; QIAGEN) from primary human CD4⁺ T cells following treatment, pelleting, and washing with PBS. For untargeted RNA sequencing (RNA-Seq), 400 ng of RNA was used for cDNA library generation using the Illumina TruSeq Stranded mRNA Library Prep (catalog no. 20020594; Illumina). Libraries were sequenced on NextSeq 500 High Output Kit v2 with 76 cycles to read one and six cycles for the index read (catalog no. FC-404-2005, 75 cycles). For targeted RNA-Seq, cDNA libraries were prepared using a customized panel of 547 genes consisting predominantly of metabolic genes and constructed with the QIAseq Targeted RNA Panel (QIAGEN). Two hundred nanograms of RNA was used for library construction with 18 cycles for PCR amplification in the QIAseq Targeted Panel workflow, according to manufacturer’s instructions, before preparation with the Illumina TruSeq Stranded mRNA Library Prep. Targeted RNA-Seq libraries were sequenced on MiniSeq High Output Kit with 151 cycles to read one and eight cycles each for index reads i5 and i7 (catalog no. FC-420-1003, 300 cycles). The data discussed in this publication have been deposited in National Center for Biotechnology Information’s Gene Expression Omnibus (25) and are accessible through Gene Expression Omnibus series accession number GSE144354 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE144354).

Sequencing reads from untargeted RNA-Seq were processed using a suite of bioinformatics programs. Reads were mapped to human genome assembly GRCh38 using Spliced Transcripts Alignment to a Reference aligner (26) and gene-specific read counts were obtained using featureCounts (27) and genome annotations from the University of California, Santa Cruz, Genome Browser. Read counts were analyzed in the R/Bioconductor environment (version 3.5) using the DESeq2 package (version 1.20.0) (28) to identify differentially expressed genes. Principal component analysis indicated that in addition to activation state of the cells, donor identity contributed substantially to expression variation among the samples. Following normalization by the DESeq method, gene expression was modeled as a function of cell activation state, adjusting for the donor. A Benjamini–Hochberg false discovery rate cutoff of 0.05 was used to identify genes that are differentially expressed. The p values for differential expression of genes displayed in Figs. 2–4 are provided in Supplemental Table I.

Human CD4⁺ T cells were pelleted and washed with PBS before lysis in T-PER tissue protein extraction buffer (Thermo Fisher Scientific) containing protease and phosphatase inhibitors. Lysate protein concentration was quantified by bicinchoninic acid then denatured by addition of 6× Laemmli SDS sample buffer (Alfa Aesar, Haverhill, MA) to a final concentration of 1× and heated at 95°C for 10 min. Forty to one hundred micrograms of denatured protein sample was run on 4–12% Bis-Tris Invitrogen NuPage gels (Thermo Fisher Scientific) and transferred onto nitrocellulose membranes before blocking with Odyssey TBS blocking buffer (LI-COR Biosciences, Lincoln, NE) and incubating with primary and secondary Abs. Abs against MYC (catalog no. 5605), HIF-1A (catalog no. 14179), TSC1 (catalog no. 6935), TSC2 (catalog no. 4308), phospho-S6 ribosomal protein (catalog no. 4857), S6 ribosomal protein (catalog no. 2317), phospho-p70 S6 kinase (catalog no. 9205), p70 S6 kinase (catalog no. 2708), phospho-4E-BP1 (catalog no. 2855), 4E-BP1 (catalog no. 9644), and β-tubulin (catalog no. 86298) were from Cell Signaling Technology (Danvers, MA). The same sets of samples were run in duplicate to probe for loading controls. Imaging of immunoblot membranes was performed with the LI-COR Odyssey CLx (LI-COR Biosciences).

Primary human CD4⁺ T cells were treated as indicated for 24 h, and then 0.15 × 10⁶ cells per well were plated in 96-well CytoStar-T scintillation plates (PerkinElmer, Waltham, MA) coated with Cell-Tak (Corning Life Sciences). For glutamine uptake, cells were labeled with 0.1 μCi ¹⁴C-glutamine (PerkinElmer) per well in the presence of 1 μM unlabeled l-glutamine in uptake buffer (25 mM HEPES, 2 mM KCl, 5 mM glucose, 1 mM MgCl₂, 1 mM CaCl₂, and 100 mM NaCl [pH 7.4]). For leucine uptake, cells were labeled with 0.05 μCi ¹⁴C-leucine (PerkinElmer) in the presence of 0.5 μM unlabeled l-leucine in uptake buffer. Scintillation was measured at various time points with a MicroBeta2 2450 microplate scintillation counter (PerkinElmer).

Gene deletion in primary human T cells was performed using the Alt-R CRISPR-Cas9 system, according to manufacturer’s instructions (Integrated DNA Technologies, Coralville, IA). Ribonucleoprotein complexes containing trans-activating CRISPR RNA and CRISPR RNA to each gene target were delivered into 2 × 10⁶ CD4⁺ T cells activated for 48 h with anti-CD3/anti-CD28 by electroporation at 1550 V for three 100-ms pulses using the Invitrogen Neon Transfection System (Thermo Fisher Scientific). Electroporated cells were reintroduced into media saved from the same cells prior to electroporation, and human rIL-2 was added to a final concentration of 5 ng/ml.

All graphs were generated with the GraphPad Prism 8 software. Statistical analyses were performed using GraphPad Prism 8 unless otherwise stated. For comparisons between two groups, an unpaired Student t test with Welch correction was performed, and comparisons between groups over time were analyzed using two-way ANOVA.

To determine if the enhanced efficacy of the metformin + 2-DG combination treatment reported in mouse models of autoimmune disease (11, 22) is also observed in human immune cells, we investigated the effects of combined metformin + 2-DG on primary human CD4⁺ T cell effector functions. As TCR-induced changes in metabolic pathways can occur proximally to TCR activation, we determined the kinetics of metformin + 2-DG–induced alterations to metabolic reprogramming during human T cell activation. Targeted transcriptomic analyses were performed on a subset of metabolic genes in primary human CD4⁺ T cells at various time points following stimulation with anti-CD3/anti-CD28 Abs. We observed the most pronounced transcriptional changes between vehicle treated and metformin + 2-DG–treated cells at 24 h following TCR stimulation (Fig. 1A), consistent with a previously reported time frame during which murine T cells accumulate metabolites and increase cell size prior to proliferation (29). Dose-dependent inhibition of oxidative metabolism in primary human CD4⁺ T cells by metformin, and of glycolytic metabolism by 2-DG, were confirmed by extracellular metabolic flux analysis using the Mito Stress Test or the Glycolytic Rate Assay, respectively (data not shown). By calculating the proton efflux rate, the Glycolytic Rate Assay allows determination of extracellular acidification attributable to glycolysis (GlycoPER) while removing the contribution of mitochondrial acidification due to bicarbonate generation via the TCA cycle (30). Compared with either metformin or 2-DG treatment alone, dual metformin + 2-DG treatment caused enhanced suppression of GlycoPER in human CD4⁺ T cells activated for 24 h with anti-CD3/anti-CD28 (Fig. 1B). Although metformin treatment reduced the OCR, as expected, this effect was not further potentiated by metformin + 2-DG treatment (Fig. 1C). Consistent with previous studies in murine T cells (11, 22), metformin + 2-DG treatment more effectively suppressed IFN-γ production in human CD4⁺ T cells than either metformin or 2-DG treatment alone (Fig. 1D). No significant changes in cell viability were observed with any treatments, so this reduction in cytokine production could not be ascribed to an increase in cellular toxicity of the dual treatment (Fig. 1E). In addition, metformin + 2-DG treatment resulted in greater suppression of human CD4⁺ T cell proliferation compared with either treatment alone (Fig. 1F). Taken together, combined metabolic inhibition by metformin + 2-DG was more effective in suppressing cytokine production and proliferation in primary human CD4⁺ T cells than either metformin or 2-DG treatment alone. This can be attributed, at least in part, to a greater inhibition of glycolysis resulting from the metformin + 2-DG combination treatment.

To better understand the changes in the use of metabolic programs following activation of primary human T cells, and how this metabolic reprogramming might be altered by metformin + 2-DG combination treatment, we examined the transcriptional profiles of primary human CD4⁺ T cells stimulated with anti-CD3/anti-CD28 Abs for 24 h in the presence or absence of metformin + 2-DG. Consistent with the enhanced inhibition of IFN-γ production noted above, transcriptomic analyses using RNA-Seq revealed that metformin + 2-DG treatment resulted in an enhanced downregulation of expression of TBX21, the master transcriptional regulator governing T_H1 differentiation, compared with vehicle treatment alone (Fig. 2A). Expression of the transcriptional regulators GATA3 and RORC, central to the differentiation of T_H2 and T_H17 lineages, respectively, were unaltered by metformin + 2-DG treatment, relative to controls (Fig. 2A). Taken together, these results align with published findings with murine T cells, which demonstrate that metabolic inhibitors can differentially modulate individual murine T cell lineages (31, 32).

Given that metformin + 2-DG efficiently suppresses glycolysis in human CD4⁺ T cells (Fig. 1B), we conjectured that metformin + 2-DG treatment might promote T cell populations that are less reliant on glycolytic metabolism. Memory T cells (T_mem) have been reported to be less dependent on glycolytic metabolism and preferentially fueled by fatty acid oxidation (32–34). Our transcriptomic analyses revealed that characteristic T_mem markers, such as CD27 and CCR7, were significantly downregulated following TCR activation but that those reductions were reversed in cells treated with metformin + 2-DG (Fig. 2A). Flow cytometry analyses confirmed that metformin + 2-DG treatment significantly increased T_mem subpopulations in activated human CD4⁺ T cells, as characterized by the absence of CD45RA and increased surface expression of CCR7 (Fig. 2B) and CD27 (Fig. 2C). Together, these results indicate that by suppressing TCR-induced glycolytic upregulation, metformin + 2-DG treatment promotes a shift toward a memory-associated phenotype in activated primary human T cells.

To gain further insights into the alterations in human T cell metabolic pathways caused by treatment with metformin + 2-DG, we mined our transcriptomic data for TCR-induced metabolic programs that were most significantly altered in human CD4⁺ T cells activated for 24 h in the presence of metformin + 2-DG or vehicle alone. Consistent with the suppression of glycolytic metabolism, as shown by its suppression of the glycolytic rate (Fig. 1B), metformin + 2-DG treatment significantly inhibited activation-induced genes, such as PKM, ALDOA, and SLC2A1, that are involved in glycolysis and glucose transport (Fig. 3A).

The polyamine synthesis pathway is highly induced upon activation of mouse T cells (22) and is required for mouse T cell proliferation (35). Our transcriptomic data revealed that in primary human CD4⁺ T cells, anti-CD3/anti-CD28 stimulation upregulated the expression of genes involved in polyamine synthesis, including SRM and SMS, and, importantly, that these increases were markedly reduced by metformin + 2-DG treatment (Fig. 3B).

The two transcription factors MYC and HIF-1A have been shown to be critical for the induction of glycolytic transcriptional programs in mouse T cells and human cancer cells (29, 36), and MYC has been reported to be a transcriptional activator of genes in the polyamine synthesis pathway (37, 38). We observed that MYC and HIF-1A were transcriptionally induced in activated primary human CD4⁺ T cells and that this effect was significantly reduced by metformin + 2-DG treatment (Fig. 3C). As MYC and HIF-1A are subjected to posttranslational regulation (39, 40), we examined the effects of metformin + 2-DG treatment on protein levels of these transcription factors in activated primary human T cells. Immunoblot analyses of human CD4⁺ T cells confirmed that protein expression of both MYC and HIF-1A were dramatically upregulated by anti-CD3/anti-CD28 activation (Fig. 3D) and that metformin + 2-DG treatment suppressed activation-induced MYC and HIF-1A protein levels to levels observed in resting human T cells (Fig. 3D). Importantly, the metformin + 2-DG combination treatment effected a more pronounced inhibition of MYC and HIF-1A protein levels in activated human T cells than either metformin or 2-DG treatment alone (Fig. 3D).

In addition to its role in the induction of glycolysis and polyamine synthesis pathways, studies in tumor cells and activated mouse T cells have revealed that MYC drives transcriptional reprogramming required for the high rates of glutamine uptake and glutaminolysis in these cells (29, 41). We observed that in primary human CD4⁺ T cells, TCR activation by anti-CD3/anti-CD28 Abs induced transcriptional programs required for glutamine uptake, glutaminolysis, and anaplerosis (Fig. 4A), and the activation-induced expression of these genes was attenuated by the metformin + 2-DG combination treatment (Fig. 4A). Building from these transcriptional profiling results, we observed that ¹⁴C-glutamine uptake by primary human CD4⁺ T cells was markedly increased following anti-CD3/anti-CD28 stimulation and that this increase was significantly inhibited by treatment with metformin + 2-DG (Fig. 4B).

Notably, the activation-induced transcriptional expression of the glutamine transporter SLC1A5 and glutamine–leucine antiporter SLC7A5 were downregulated by metformin + 2-DG treatment (Fig. 4A, Supplemental Table I). SLC7A5, a gene transcriptionally controlled by MYC (42), heterodimerizes with the H chain of CD98, commonly known as CD98hc or SLC3A2, another target gene under the direct transcriptional control of MYC (43, 44). Both SLC7A5 and CD98 are critical for the induction of TCR-mediated mouse T cell effector processes (4). Surface expression of SLC7A5 and CD98 proteins were upregulated upon anti-CD3/anti-CD28 stimulation of primary human CD4⁺ T cells, but these increases were significantly attenuated by treatment with metformin + 2-DG (Fig. 4C, 4D). Moreover, downregulation of activation-induced SLC7A5 and CD98 by metformin + 2-DG corresponded with significantly reduced ¹⁴C-leucine uptake in T cells treated with metformin + 2-DG (Fig. 4E). These results demonstrate that metformin + 2-DG downregulates MYC-dependent metabolic programs required for glutamine and leucine uptake and catabolism in primary human T cells.

To distill the respective contributions of MYC and HIF-1A to human T cell effector functions and metabolic reprogramming, we deleted MYC and HIF-1A expression in primary human CD4⁺ T cells using the CRISPR-Cas9 gene editing system. Robust knockout (KO) of both transcription factors was confirmed by immunoblot analysis (Fig. 5A). Deletion of MYC, but not HIF-1A, significantly suppressed human T cell proliferation compared with controls (i.e., activated cells that were electroporated but did not receive guide RNA) (Fig. 5B). Although MYC deficiency in human T cells significantly impaired cell proliferation, cytokine production was unchanged (data not shown). This profile aligns with previously published results with Myc-deficient mouse T cells (29). Proliferation was attenuated by a similar amount in human CD4⁺ T cells depleted of both MYC and HIF-1A (MYC/HIF-1A double KO [dKO]) and human CD4⁺ T cells depleted of MYC alone (Fig. 5B). Unexpectedly, we observed that proliferation was slightly enhanced in HIF-1A KO human CD4⁺ T cells compared with activated wild-type control cells (Fig. 5B). Based on these results, we conclude that MYC is required for anti-CD3/anti-CD28–induced proliferation of primary human CD4⁺ T cells, whereas HIF-1A is dispensable for this process.

Surface expression of the MYC target CD98 was reduced in MYC KO and MYC/HIF-1A dKO, but not HIF-1A KO, human CD4⁺ T cells (Fig. 5C). Unexpectedly, although SLC7A5 has been reported to be an MYC target (42), activation-induced SLC7A5 expression was unaltered by MYC or HIF-1A deficiency (Fig. 5D). Glutamine uptake was significantly impaired in MYC KO T cells but was unaffected by HIF-1A deletion (Fig. 5E). Consistent with reduced glutamine uptake and glutaminolysis, OCR was reduced, albeit slightly, in MYC KO human CD4⁺ T cells compared with wild-type control cells (Fig. 5F). HIF-1A KO T cells exhibited significantly increased OCR, likely because of a compensatory increase in oxidative metabolism in the absence of HIF-1A–coordinated glycolytic metabolism (Fig. 5F). Overall, these results demonstrate that MYC is required for the full induction of effector functions in primary human CD4⁺ T cells, including anti-CD3/anti-CD28–induced proliferation, the upregulation of glutamine and leucine transporters, and increased glutamine uptake and glutaminolysis. In contrast, our data demonstrate a dispensable role for HIF-1A in coordinating human T cell effector functions, consistent with previously reported characterizations of mouse T cells (29, 45).

SLC7A5 and CD98 are crucial for mediating leucine influx, and intracellular leucine activates mechanistic target of rapamycin complex 1 (mTORC1) in mouse T cells (29, 46, 47). One consequence of metformin treatment is activation of AMPK, which phosphorylates and activates the tuberous sclerosis proteins TSC1 and TSC2. TSC1 and TSC2 suppress mTOR signaling by inhibiting the GTP-binding protein RHEB, which is required for mTORC1 activation (48–50). We investigated whether metformin + 2-DG–mediated repression of mTORC1 signaling could be responsible for mediating the effects of the metformin + 2-DG combination treatment on human CD4⁺ T cells. Activated mTORC1 phosphorylates 4E-BP1 and thereby inhibits the interaction of 4E-BP with eIF4E, promoting a derepression of translation. Activated mTORC1 also phosphorylates and activates p70S6 kinase (p70S6K), which subsequently phosphorylates the ribosomal protein S6, a prerequisite for increased cell size and cell growth (51, 52). We observed that compared with either metformin or 2-DG treatment alone, metformin + 2-DG treatment exerts enhanced suppression of the mTORC1 signaling pathway, as demonstrated by reduced activation-induced phosphorylation of rapamycin-sensitive sites on p70S6K, S6, and 4E-BP1 (Fig. 6A).

As mTORC1 is negatively regulated by TSC1 and TSC2, inhibition of TSC1 or TSC2 causes hyperactivation of the mTORC1 signaling pathway (53). To investigate if the suppression of human T effector functions by metformin + 2-DG occurred through the inhibition of mTORC1 signaling, we deleted TSC1, TSC2, or both, in primary human CD4⁺ T cells (Fig. 6B). As expected, deletion of either or both TSC1 and TSC2 led to dramatically enhanced mTORC1 signaling as measured by increased phosphorylation of S6 (Fig. 6C). Furthermore, deletion of TSC1, TSC2, or both in primary human CD4⁺ T cells resulted in significantly enhanced IFN-γ production and proliferation (data not shown). However, TSC1 KO, TSC2 KO, and TSC1/2 dKO T cells could not overcome the suppression of IFN-γ, cell proliferation, and CD98 surface expression caused by metformin + 2-DG treatment (Fig. 6D, 6E, and data not shown). Thus, we conclude that the inhibition of activation-induced IFN-γ production, cell proliferation, and CD98 transporter expression by metformin + 2-DG combination treatment cannot be attributed to its inhibition of mTORC1 pathway signaling.

Current understanding of T cell metabolism draws heavily from immunometabolism studies performed on mouse models and mouse T cells (22, 29, 36, 54). In this paper, we present what is, to our knowledge, the first reported characterization of the effects of metabolic inhibition by metformin and 2-DG on primary human CD4⁺ T cells and provide a dissection of the metabolic pathways modulated by this combination treatment regimen. Using primary human T cells, we recapitulated the results of prior mouse studies, demonstrating the enhanced efficacy of combined metformin and 2-DG treatment, versus either drug treatment alone, for suppression of T cell proliferation and proinflammatory cytokine production (22). Furthermore, our results with human immune cells reinforce the concept of plasticity and compensatory upregulation of orthogonal metabolic pathways that can follow blockade of one pathway. For example, in human T cells, the inhibition of oxidative phosphorylation with metformin led to elevated rates of glycolysis (Fig. 1B) and enhanced expression of HIF-1A (Fig. 3D).

Metabolic inhibition by metformin + 2-DG treatment promoted a memory-like phenotype in activated human CD4⁺ T cells, lending further credence to the concept that T cell subsets have different metabolic profiles that can be preferentially targeted to promote the emergence of certain T cell populations (33, 55). Strikingly, metformin + 2-DG treatment synergistically suppressed activation-induced upregulation of MYC, a key regulator of glycolytic and amino acid metabolic programs. MYC-dependent markers of T cell activation CD98 and SLC7A5 that coordinate glutamine and leucine flux were also downregulated by metformin + 2-DG treatment. In addition to heterodimerizing with l-system amino acid transporters such as SLC7A5 to facilitate amino acid transport, CD98 interacts with β1 integrins to mediate integrin-dependent signaling, a role shown to be central to cell proliferation (56, 57). Given the pleiotropic roles of CD98, the suppression of activation-induced amino acid transport and cell proliferation in metformin + 2-DG–treated or MYC-deficient human T cells may be mediated by the reduction in CD98 expression. Further investigations are warranted that examine the mechanism by which the MYC–CD98 axis coordinates the metabolic plasticity and effector functions of activated human T cells.

Given that HIF-1A has been implicated in coordinating the aerobic glycolysis program in cancer cells and activated immune cells (58), it is intriguing that HIF-1A appears to be dispensable for the induction of human T cell effector responses. Nonetheless, our observations are consistent with published observations using Hif-1α–deficient mouse T cells (29). We speculate that HIF-1A may have an indispensable role in specific T cell subsets or perhaps in immune cell types other than T lymphocytes. Indeed, current studies in mice indicate that HIF-1A deletion may ameliorate inflammation in the context of T_H17-dependent diseases by dampening T_H17 responses and promoting T_reg numbers (31, 36). Moreover, other HIF family members with overlapping functions, such as HIF-2A (59), may mask the phenotype in HIF-1A–deficient cells. Comparison of HIF-1A deletion in specific human T cell subsets in the context of HIF-2A sufficiency or deficiency will clarify some of these outstanding questions.

The precise signaling events that metformin + 2-DG impact to block the MYC and HIF-1A transcriptional regulators and, concomitantly, activation-induced metabolic programs in human T cells remain to be elucidated. Based on published reports describing MYC and HIF-1A as downstream of mTORC1 (60), we speculated that metformin + 2-DG combination treatment exerts its effects on MYC- and HIF-1A–dependent metabolic programs via downregulation of mTORC1 signaling. Although metformin + 2-DG did synergistically suppress mTORC1 signaling in activated human T cells, enhanced mTOR signaling in TSC1 and TSC2 KO cells did little to abrogate the suppression of human T effector cell responses by metformin + 2-DG treatment. Thus, the metformin + 2-DG–dependent downregulation of MYC and HIF-1A, and the metabolic pathways they control, appear to be mediated by mTORC1-independent mechanisms. MYC mRNA and protein levels were dramatically reduced by metformin + 2-DG treatment; however, the effects of MYC deficiency on suppressing human T cell effector functions were much milder than those exerted by metformin + 2-DG treatment. Taken together, these observations implicate the involvement of additional, currently unidentified pathways that are responsible for mediating the effects of metformin + 2-DG treatment on human T cells.

It is conceivable that the primary mechanism of action of metformin + 2-DG combination treatment in activated primary human T cells occurs through inhibition of processes proximal to the TCR, such as SLC7A5-mediated amino acid transport or signaling. Amino acid and glucose transport are both required for the hexosamine biosynthetic pathway (HBP) that provides substrates for O-GlcNAcylation, a process that is induced upon T cell activation and is indispensable for murine T cell clonal expansion (40). Studies in cancer cells treated with metformin + 2-DG demonstrate a synergistic inactivation of GFAT1, the rate limiting enzyme in the HBP, alongside an induction of ER stress markers (61). It is possible that direct or indirect inhibition of the HBP by metformin + 2-DG is responsible for the effects of metformin + 2-DG treatment on human T cell proliferation. Further work that addresses the effects of metformin + 2-DG treatment on the TCR–SLC7A5–MYC signaling axis in primary human T cells and the HBP should increase our understanding of the sequence of events and mechanism of action of metformin + 2-DG treatment.

Overall, our study demonstrates that metformin + 2-DG combination treatment dampens effector functions in primary human immune cells by collaboratively suppressing expression of the key transcriptional regulators MYC and HIF-1A and the metabolic programs of aerobic glycolysis, amino acid uptake, glutaminolysis, and anaplerosis. Nevertheless, stimulation of T cells with anti-CD3/anti-CD28, although widely used for in vitro activation of T cells, may not fully recapitulate the milieu of downstream signals following physiological APC–T cell engagement. Indeed, in vivo models in which these interactions between APC and human T cells can be reconstituted would provide a more robust system to examine the effects of metformin + 2-DG treatment. We are currently attempting to confirm and extend the results presented in this study on the effects of metformin + 2-DG combination treatment in an in vivo model of human T cell engraftment in immune-deficient mice. Future studies that identify the multiple metabolic pathways in primary human immune cells that can be simultaneously targeted to enhance efficacy in reducing inflammatory processes will help progress the development of combination therapies for treating autoimmune diseases.

We thank Erick Moy for assistance with amino acid uptake experiments and Drs. Rafael de Queiroz Prado, Mingcan Xia, and Russell Miller for helpful discussions.

The authors have no financial conflicts of interest.