Mannose Metabolism Is Essential for Th1 Cell Differentiation and IFN-γ Production

Glycolysis is one of the crucial metabolic pathways in the cell. Apart from its well-known role as an energy source, it provides living cells with intermediates and substrates for other metabolic pathways (1). Its first intermediate, glucose-6-phosphate, is the substrate of the pentose phosphate pathway. Furthermore, glycolysis steps provide intermediates for the biosynthetic pathways of many amino acids. Its end product, pyruvate, is an important source of acetyl-CoA used for the synthesis of carboxylic acids, whereas earlier steps of glycolysis provide the glycerol backbone for lipid synthesis. Glucose metabolism is also an important source of sugar residues for protein glycosylation.

One of the common components of glycoproteins is mannose, a C-2 epimer of glucose (2). Cells obtain mannose mostly from glucose (3). It is generated by phosphomannose isomerase from a glycolytic intermediate, fructose-6-phosphate, and therefore, its availability is dependent on glycolytic flux. Mannose is essential for proper protein glycosylation. Inhibition of its synthesis leads to not only disorders of glycosylation but also inhibition of cell growth and proliferation (4). Free mannose of dietary source is also present in mammalian plasma. The importance of the contribution of free mannose in protein glycosylation is still under discussion (4). However, it is clear that the negative effects of inhibition of mannose synthesis can be annulled by free mannose supplementation (4). Free mannose is transported into cells via unspecific hexose transporters (3). It is subsequently phosphorylated by hexokinase to produce mannose-6-phosphate. Mannose-6-phosphate is further directed by phosphomannomutase into the GDP–mannose biosynthesis pathway and subsequently used in glycosylation.

2-Deoxy-d-glucose (2DG) is a glycolysis inhibitor commonly used in research studies (5, 6). 2DG is a phosphoglucose isomerase competitive inhibitor, which causes phospho-hexose accumulation and subsequently inhibition of hexokinase (7, 8). Because of its structural similarity with mannose, 2DG also exerts a direct, glycolysis-independent effect on protein glycosylation (9). It inhibits several enzymes taking part in glycoprotein synthesis. Competitive inhibition can be overcome by increasing the mannose concentration (10). The mechanism leading to inhibition of cell proliferation and growth by 2DG is highly context dependent, as it has been shown in several tumor cell lines (11–13). 2DG may exert its effect by inhibiting glycolysis or direct protein glycosylation, as it was proved to do in NK cells (14).

One characteristic of immune cells is the inconstancy of their metabolic profiles, which is a consequence of changes in the immune cell’s activation (15). Glycolysis has been demonstrated to play an essential role in the maturation and differentiation of many immune cells. Upregulation of glycolytic flux seems to be essential for proper Th1, Th2, and Th17 differentiation (16). Inhibition of glycolysis has been shown to result in anergy in activated T cells (17) or induce differentiation of regulatory T cells (16) and memory cells (18). Glucose metabolism also controls cytokine expression by macrophages (19) and is essential for dendritic cell (DC) maturation upon LPS stimulation (20). Further studies on this topic have shown that glycolysis supports de novo synthesis of fatty acids for the expansion of immune cell membranes, which is crucial for their immune functions (20, 21). Another report showed the importance of glucose metabolism for protein O-GlcNAcylation (22).

In Th1 cells, glycolysis has been shown to be essential for proliferation and expression of its signature cytokine IFN-γ (16). It was shown that naive CD4⁺ T cells activated in medium depleted of glucose do not proliferate and produce no IFN-γ (23). Partial inhibition of glycolytic flux by 2DG supplementation or exchange of glucose with galactose limits IFN-γ expression and proliferation (16, 23). Limited glucose availability was also shown to be responsible for inhibition of Th1 cell immune functions in a tumor environment (24, 25). Different molecular mechanisms have been proposed to link Th1 cell differentiation with the glycolytic flux. A direct role of the glycolytic enzyme GAPDH in the control of IFN-γ translation has been shown (23). Under conditions of diminished glycolytic flux, GAPDH was shown to bind IFN-γ mRNA and block its translation. Another study indicated that the glycolytic metabolite phosphoenolpyruvate is necessary for proper calcium signaling and, as a result, for Th1 cell differentiation (24). Phosphoenolpyruvate was shown to block reuptake of calcium ions from the cytoplasm to endoplasmic reticulum and thus enhance NFAT-dependent gene translation and IFN-γ expression. Because the currently available data on the influence of glycolytic flux on Th1 cell differentiation are conflicting, further research on this topic is required.

Aware of the crucial role of glycolysis in mannose metabolism, as well as the significance of glycolysis in Th1 cell differentiation, we decided to assess if mannose metabolism is important for Th1 cell development. Our results demonstrated that although the activation conditions determine sensibility of Th1 cells to glycolysis restriction, mannose metabolism is the limiting factor for IFN-γ expression (Fig. 1). The data presented in this manuscript shed new light on the role of glycolysis in Th1 cell differentiation, which needs to be considered in the broad context of interconnected metabolic pathways.

Naive C57BL/6 animals were purchased from Harlan or from the Mossakowski Medical Research Centre Polish Academy of Sciences and were used at the age of 8–14 wk. OTII animals were bred at the animal facilities of the Helmholtz Centre for Infection Research under specific pathogen-free conditions.

Lymphocytes were isolated from spleens of C57BL/6 and OTII mice. Naive CD4 cells were defined as CD4⁺, CD25⁻, CD62L^high, and CD44^low and sorted using FACSAria. The purity of the cells was >95%. Cells were cultured in DMEM without glucose (Life Technologies), and DMEM was supplemented with 10% FBS, 100 U/ml of penicillin–streptomycin, 2 mM l-glutamine, 50 nM 2-ME, 25 mM HEPES, 1 mM sodium pyruvate, and MEM nonessential amino acids solution (Life Technologies). If not stated otherwise, the cultures were supplemented with 10 mM glucose (Sigma-Aldrich), 10 mM galactose (Sigma-Aldrich), 0.5 mM 2DG (Sigma-Aldrich), 1 mM mannose (Sigma-Aldrich), or their combinations. FBS used in experiments was extensively dialyzed to remove hexoses. In some experiments, IMDM medium supplemented with 10% FBS, 100 U/ml of penicillin–streptomycin, 2 mM l-glutamine, and 50 nM 2-ME was used. The following compounds were used: thapsigargin (0.5 nM; Sigma-Aldrich), ionomycin (40 ng/ml; Sigma-Aldrich), and PMA (2 ng/ml; Sigma-Aldrich). Naive CD4 T cells were activated with plate-bound αCD3 Ab (2 μg/ml, clone 145-2C11; eBioscience) and αCD28 Ab (2 μg/ml, clone 37.51; eBioscience). The flat or round-bottom 96-well plates were coated with Abs for 4 h at 37°C. Naive OTII cells were activated with bone marrow–derived DC. Prior to this, DC were loaded for 24 h with 1 μg/ml of the OVA 323–339 peptide (Sigma-Aldrich) in the presence of LPS (500 ng/ml; Sigma-Aldrich) and extensively washed with DMEM without glucose. Th1 cells were differentiated with IL-12 (5 ng/ml; PeproTech) in the presence of αIL-4 Ab (clone 11B11; Bio X Cell) for 4 or 5 d. In time course experiments, cells were washed three times with DMEM without glucose to remove hexoses or PMA present in the culture. Subsequently, cells were provided with fresh medium supplemented with IL-12, αIL-4 Ab, and a new mix of hexoses and PMA. For proliferation studies, cells were stained prior to the culture with CFSE (1 μM).

For surface staining, cells were incubated with Abs against surface markers and after washing with PBS acquired by FACS. For intracellular cytokine staining, 1 μg/ml ionomycin and 10 ng/ml PMA (Sigma-Aldrich) were added to cultures for 4 h. For the last 2 h of culture, Brefeldin A (Sigma-Aldrich) was added to a final concentration of 5 μg/ml. After washing with PBS, cells were fixed for 20 min in 2% paraformaldehyde. Intracellular cytokine staining was performed using Permeabilization Buffer (eBioscience) according to the manufacturer’s protocol. For T-bet staining, Transcription Factor Staining Buffer Set (eBioscience) was used according to the manufacturer’s protocol. During acquisition, samples were first gated on lymphocytes by their forward scatter and side scatter characteristics, and dead cells were excluded by using LIVE/DEAD Fixable Dead Cell Stain from Life Technologies. Cell proliferation was assessed by means of CFSE dilution. FACS data were analyzed using the FlowJo software (Tristar).

mRNA was isolated using TRIzol (Invitrogen) according to the manufacturer’s protocol. cDNA was synthetized using RNA to cDNA EcoDry Premix (Takara) according to the manufacturer’s protocol. The quantitative PCR (qPCR) reaction was performed using LightCycler 480 (Roche). The data were analyzed using the LightCycler 480 software. Hprt was used as a housekeeping gene to normalize the data.

IFN-γ content in culture supernatants was measured using Mouse IFN-γ ELISA MAX Deluxe (BioLegend) according to the manufacturer’s protocol.

Cells cultured with different hexoses were harvested and transferred to a new 96-well plate filled with Matrigel Basement Membrane Matrix (50 μl/well; BD Pharmingen) by gently placing them on top of the gel (50 μl of cell suspension in respective hexose-containing medium). After 1 h incubation at 37°C in 5% CO₂, the cells were imaged on a Zeiss Cell Observer SD microscope equipped with an environmental chamber and a 96-well plate heating insert to maintain the same culture conditions. Imaging was performed in the bright field mode using 10× objective (NA 0.3) and Rolera EM-C2 EMCCD camera (QImaging). Cell migration over the Matrigel was recorded for 5 min every 5 s in the central field of the well where the gel surface was flat. Cell tracking was performed using the Imaris 8.0 software (Bitplane), and the instantaneous speeds of detected cells were calculated.

Lactate production was measured using Lactate Assay Kit (Abcam) according to the manufacturer’s protocol.

Statistical analysis was performed using the GraphPad Prism software (Version 6). Data are represented as mean ± SD of independent biological replicates. Statistical comparisons were performed by Student paired t test or repeated measure one-way ANOVA with Tukey correction. Differences were considered statistically significant with p < 0.05. Significance is indicated as follows: *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. On the figures, statistical significance is indicated for the comparisons that are mentioned in the manuscript text. The full list of adjusted p values is presented in Supplemental Table I. In the experiments in which variances between biological replicates were introduced by differences in staining (proliferation measurement by means of CFSE dilution) or experimental setup (measurement of IFN-γ production by ELISA and lactate measurement), the median fluorescence intensities of samples within a single experiment were normalized to median fluorescence intensity of all the samples within a single experiment.

To answer the question regarding the importance of mannose metabolism in Th1 cell differentiation, we generated these cells under conditions of diminished glycolytic flux. Naive lymphocytes were activated in medium supplemented with 2DG or in glucose-free medium supplemented with galactose. To enter glycolysis, galactose needs to be transformed into glucose-1-phosphate in the Leloir pathway (26). Because of limited availability of Leloir pathway enzymes, substitution of glucose with galactose considerably decreases glycolytic flux, making cells dependent on respiration (26, 27). It was observed that such a treatment blocks IFN-γ expression (23). It is well established that the mannose metabolism, which is altered by diminished glycolytic flux, may be restored by supplementation of culture medium with free mannose (4). We found that external mannose restored IFN-γ expression in Th1 cells differentiated under the conditions of diminished glycolytic flux (Figs. 1, 2A, 2B), which provides direct evidence for the importance of mannose metabolism for the expression of this cytokine. Mannose did not influence IFN-γ production when the glycolytic flux was not inhibited (Fig. 2C). It was shown that glycolysis inhibition influences several parameters of differentiating T cells. Mannose supplementation enhanced the proliferative capacity of Th1 cells cultured in the presence of galactose. When the glycolytic flux was blocked with 2DG, mannose supplementation restored the full proliferative capacity of Th1 cells (Fig. 2D, Supplemental Fig. 1A).

We also observed that decreased glycolytic flux is associated with substantial inhibition of cell motility (Fig. 2E, Supplemental Video 1). This is in agreement with observations that cell migration is particularly dependent on glycolysis-derived ATP (28, 29). Therefore, limited availability of glycolysis-derived ATP could have a negative effect on cell migration. In contrast, mannose supplementation did not have an effect on cell motility.

To exclude the possibility that the effect that mannose had on Th1 cells cultured under conditions of diminished glycolytic flux is a consequence of glycolytic flux restoration, we measured glycolysis by means of lactate production. The samples were collected 24 h after culture onset. At this time point, cells did not divide yet; hence, the proliferation rate does not influence the results. Under these experimental conditions, we observed that mannose supplementation did not influence lactate production (Fig. 2F).

To test whether mannose metabolism alone is sufficient to support IFN-γ production, we generated Th1 cells in medium containing mannose as the only sugar. There were no statistically significant differences in the percentage of IFN-γ–producing cells in cultures grown on glucose or mannose (Fig. 3A, Supplemental Fig. 1B). This shows that mannose is sufficient to support expression of IFN-γ. Mannose did not support full proliferative capacity in the presence of galactose; however, Th1 cells underwent more cell divisions in the presence of mannose only, as compared with those cultivated with galactose alone (Fig. 3B, Supplemental Fig. 1C). Because mannose blocked the inhibitory effect of 2DG on Th1 cell differentiation in medium containing glucose (Fig. 2A, 2B), we tested whether mannose alone could support Th1 cell development if 2DG was present. We observed that both glucose and mannose were necessary for IFN-γ production in the presence of 2DG (Fig. 3C, Supplemental Fig. 1D). Similarly, glucose and mannose were required to support cell proliferation in the presence of 2DG (Fig. 3D, Supplemental Fig. 1E).

It was shown that galactose, as the only sugar in the culture medium, restricts glycolysis because its flux is much lower than the glucose flux (27). Therefore, we tested whether the decrease in glycolytic flux caused by a low glucose concentration has a similar effect on the differentiation of Th1 cells. Standard blood glucose concentration is 5 mM (30), whereas the culture medium contained 10 mM glucose or 25 mM in the high-glucose format. At a low glucose concentration (1 mM), IFN-γ expression was restricted (Fig. 4A). Similarly, a low galactose concentration (1 mM) was less efficient in supporting IFN-γ production than a medium dose (10 mM) or a high dose (25 mM). This suggests that, indeed, there is a limited dose of galactose that lymphocytes could metabolize and that a further increase in its concentration above a certain threshold did not influence cells. A low mannose dose (1 mM) was sufficient to support Th1 cell differentiation, whereas a high mannose concentration (25 mM) inhibited the differentiation of Th1 cells (Fig. 4B). The inhibitory effect of a high mannose concentration was also observed when glucose was present in the culture medium (Fig. 4B). It is known that mannose is a highly reactive sugar, and its inhibitory effect on Th1 cell differentiation may be caused by nonenzymatic glycation occurring at a high mannose concentration (31). Cell proliferation was much less influenced by sugar availability (Fig. 4C), which is consistent with other reports (23, 27). In fact, cells cultivated at a low concentration of glucose underwent less cell divisions (Fig. 4C). The same holds true for galactose (Fig. 4C). In contrast, a high concentration of mannose had a limited inhibitory effect on cell proliferation (Fig. 4C).

It is well established that inhibition of glycolytic flux blocks Th1 cell differentiation and IFN-γ expression (16, 23, 24). However, different mechanisms have been proposed to be responsible for this phenomenon. There is a report showing that glycolysis controls IFN-γ expression at the translational level (23), whereas others show that alterations in calcium signaling are responsible for inhibition of Th1 cell differentiation under conditions of diminished glycolytic flux (24). Thus, we measured IFN-γ mRNA levels (Fig. 5A) to determine if mRNA was blocked in our experimental model translation of IFN-γ. The obtained results showed that glycolysis inhibition lowered IFN-γ mRNA content, which excluded solely translational control of its expression by glycolysis. In fact, if this were the case, we would expect no differences in IFN-γ mRNA abundance. There was no statistically significant difference in IFN-γ mRNA content in cells cultured in the presence of glucose, mannose alone, mannose and galactose, or 2DG. T-bet is a transcription factor that directs the Th1 cell differentiation program and controls IFN-γ expression (32). Thus, we measured T-bet expression (Fig. 5B) to assess if diminished glycolytic flux blocked Th1 cell lineage commitment. We observed lower T-bet expression in cells differentiated in medium with galactose or in the presence of 2DG. Similar results were obtained when we measured phosphorylation of the STAT1 and STAT4 transcription factors (Fig. 5C), which regulate differentiation of Th1 cells (32). This further excluded translation control as the sole mechanism regulating IFN-γ expression because T-bet, STAT1, and STAT4 control expression of this cytokine on a transcriptional level (32). Mannose supplementation restored the Th1 differentiation program (Fig. 5B, 5C), which indicated modulation of Th1 differentiation at the transcriptional level. To assess whether calcium signaling was responsible for the inhibition of Th1 cell differentiation, we supplemented cell cultures with thapsigargin or ionomycin. Thapsigargin increases the cytoplasmic calcium level by blocking its reuptake to the endoplasmic reticulum, whereas ionomycin is an ionophore. It was shown that thapsigargin restores Th1 cell differentiation under conditions of diminished glycolytic flux (24). However, this treatment did not restore IFN-γ expression, showing that in our experimental conditions, increased intracellular calcium concentration did not rescue Th1 cell differentiation during glycolysis restriction (Fig. 5D). PMA is an analog of diacylglycerol (DAG), the important second messenger taking part in TCR signaling. To our surprise, Th1 cultures treated with PMA were resistant to the inhibitory effect of diminished glycolytic flux on IFN-γ production (Fig. 5D). To assess if PMA and mannose had an additive effect, we differentiated Th1 cells at different concentrations of 2DG (Fig. 5E). The 2DG is a competitive inhibitor of phosphoglucose isomerase and other glucose transforming enzymes; therefore, the strength of its effect depends on the ratio between the concentration of 2DG and glucose. PMA and mannose restored IFN-γ expression at a 2DG dose high enough to block the effect of each of these two compounds alone. This result showed that the effect of PMA and mannose on Th1 cell differentiation in the presence of 2DG is additive and complementary.

To better understand the mechanism responsible for the dependence of Th1 on glycolysis, we performed time course studies. To this end, 2DG was added or removed from culture medium at different time points. We observed that the presence of 2DG only during the first 24 h of activation did not block IFN-γ production (Fig. 6A). In contrast, 2DG added 24 h after activation exerted a similar effect as 2DG present during the whole course of the experiment. Removal of 2DG from culture medium after 48 h of stimulation resulted in a partial blockage of IFN-γ production. This suggests that glycolysis was important for IFN-γ expression starting from the second day after activation throughout the rest of the process. Similar effects were observed when glucose was exchanged with galactose and vice versa (Fig. 6B). Cells grown during the first 24 h on galactose and, later on, glucose differentiated into Th1 cells as well as cells cultured in medium supplemented with glucose throughout the whole experiment. Exchange of glucose by galactose 24 h after cell activation exerted a similar inhibitory effect as cultivation on galactose only. Analogous results were obtained in mannose time course experiments (Fig. 6C). Its presence or absence in culture medium during the first 24 h after T cell activation did not influence differentiation of Th1 cells under conditions of diminished glycolytic flux. When mannose was added or removed from cultures treated with 2DG 48 or 72 h after activation, no statistically significant differences in IFN-γ expression were observed. If galactose was used to block glycolytic flux, the presence of mannose during the first 48 h of culture was sufficient to restore IFN-γ production. Different results were obtained in PMA time course experiments. PMA rendered Th1 cells resistant to glycolysis inhibition mostly during the first 24 h after activation (Fig. 6D), and it did not restore IFN-γ production when added 24 h after activation. In contrast, PMA had a partial effect on IFN-γ expression when removed 24 h after activation, whereas its presence for 48 h was sufficient for restoration of IFN-γ expression. These results show that the PMA and mannose modes of action on IFN-γ expression under diminished glycolytic flux are different because these two compounds acted at different time points. PMA made Th1 cells more resistant to glycolysis inhibition, whereas mannose blocked the inhibitory effect resulting from glycolysis restriction.

PMA is an analog of DAG, which is synthetized from glycolytic intermediates (33). Therefore, we tested whether glycolysis inhibition influences cellular DAG content (Fig. 6E). The obtained results showed no statistically significant differences, thereby suggesting that de novo DAG synthesis is not important in T cell activation.

Because the time of exposure to mannose or PMA is crucial for their effect on IFN-γ production under conditions of diminished glycolytic flux, we performed time course experiments. In these experiments, cells were treated with calcium signaling modulators at different time points after activation (Fig. 6F). We did not observe any influence of ionomycin or thapsigargin on IFN-γ expression, irrespective of the time when these compounds were added or removed. To assess if IFN-γ expression is controlled at the translational level in our experimental conditions, we measured IFN-γ and T-bet mRNA levels 24, 48, and 72 h after activation (Fig. 6G, 6H). The pattern of IFN-γ mRNA expression fully resembled the one observed for T-bet, suggesting that the production of these two proteins is regulated in a similar way. Inhibition of the glycolytic flux by the exchange of glucose to galactose affected IFN-γ transcription only 48 h after activation, and the same holds true for T-bet. This further supports the hypothesis that IFN-γ expression under conditions of diminished glycolytic flux is controlled at the transcriptional level. We showed in the above-mentioned experiments that the inhibition of glycolysis during the first 24 h after activation has no effect on IFN-γ expression if the glycolytic flux is restored at later time points. Our data showed that there is no difference in IFN-γ or T-bet mRNA expression during the first 24 h if the glycolytic flux is restricted by the exchange of glucose to galactose, which supports our earlier observation. There was a statistically significant decrease in IFN-γ or T-bet mRNA levels if glycolysis was blocked by 2DG; however, this decrease was much smaller than that which we observed at later time points. Also, in agreement with previous observations, mannose did not influence IFN-γ or T-bet mRNA expression during the first 24 h of culture (Fig. 6G, 6H).

Both calcium and DAG are important second messengers taking part in TCR signal transduction (34). The αCD3/αCD28 stimulation model used in our and other studies mimics T cell activation only to a limited extent. It misses many costimulatory molecules or cytokines produced by DC. Taking that into account, we tested whether the effect of PMA observed by us and the effect of thapsigargin observed by others can be reproduced under more physiological conditions. To this end, we activated naive OTII cells with DC in a Th1 cell–inducing milieu. To minimize the effect of glycolysis inhibition on DC activation (35), we loaded DC with an activating peptide in the presence of LPS 24 h earlier. Unexpectedly, glucose substitution with galactose did not have a negative effect on IFN-γ expression (Fig. 7A, Supplemental Fig. 2A). Like in previous experiments, the presence of 2DG in the culture medium resulted in a blockage of IFN-γ expression. Mannose supplementation restored IFN-γ production at a low 2DG dose, whereas it had no effect at a high 2DG concentration, which is in turn consistent with previous results. A similar effect was observed when T-bet expression was analyzed (Fig. 7B, Supplemental Fig. 2B). There was no difference in the IFN-γ mRNA level in cells cultured in medium containing glucose or galactose (Fig. 7C). The 2DG blocked IFN-γ transcription at a low and a high dose. Mannose seems to block this effect, although the observed differences in the IFN-γ mRNA level in cultures treated with 2DG or 2DG with mannose were statistically significant only at a high 2DG dose. Replacement of glucose with galactose resulted in a moderate inhibition of cell proliferation (Fig. 7D, Supplemental Fig. 2C), which was similar to the data obtained using the αCD3/αCD28 activation model. In contrast, the effect of 2DG on cell proliferation was dose dependent. Mannose supplementation increased the proliferative capacity of Th1 cells differentiated in the presence of 2DG. The effect of mannose on cell proliferation if the cells were grown in medium containing galactose was not statistically significant. We also observed that neither thapsigargin nor PMA restored IFN-γ expression in OTII cells treated with 2DG (Fig. 7E, Supplemental Fig. 2D), which showed that their effector functions are highly dependent on the activation conditions.

In this work, we showed that mannose can regulate IFN-γ expression by Th1 cells. Limited mannose availability is responsible for the inhibitory effect of glycolysis restriction on IFN-γ production by Th1 cells. Glycolysis is the substrate source for many metabolic pathways, and its inhibition influences numerous metabolic processes. Restoration of mannose metabolism blocks the negative effect of glycolysis restriction on IFN-γ expression, indicating its important role in glycolysis-dependent IFN-γ production. Limited glucose metabolism influences several parameters of differentiating Th1 cells, although cytokine expression is the one disturbed the most. It was shown that proliferation of Th1 cells depends mostly on respiration; however, restriction of glycolysis also decreases their proliferative capacity (23). We also showed that the migratory properties of Th1 cells were compromised under conditions of diminished glycolytic flux. Mannose supplementation had only a limited effect on proliferation and no effect on the migratory capacity of Th1 cells, which is in striking contrast to its effect on IFN-γ expression. This demonstrates the complexity of the interrelationships between metabolism and cell functions. We showed that inhibition of glycolysis causes several changes in Th1 cell functions; however, these changes are underlined by different metabolic processes. It was shown in DC (20) and B cells (21) that cell membrane synthesis relies on lipids originated from glycolysis products. Because proliferation requires new cell membrane biogenesis, it is possible that limited lipid availability results in a restriction of Th1 cell proliferation under conditions of diminished glycolytic flux. In contrast, cell migration is a highly energy-dependent process (29). Thus, limitation in ATP supply may be responsible for the observed restricted motility of Th1 cells under reduced glycolytic flux. This is further supported by the fact that migration of tumor cells and macrophages depends on glycolysis-derived ATP (28, 29).

TCR signaling is influenced by factors such as Ag dose or spatiotemporal organization of the immunological synapse (34, 36). We showed here that PMA restored IFN-γ production under conditions of diminished glycolytic flux when Th1 cells were activated with αCD3 and αCD28 Abs, whereas thapsigargin had no effect. These data are in contrast to a report showing that thapsigargin restores IFN-γ production by Th1 cells under conditions of diminished glycolytic flux (24). The authors suggested that diminished glycolytic flux alters calcium signaling, thereby leading to inhibition of Th1 cell differentiation. The protocols used in different laboratories to generate Th1 cells with αCD3/αCD28 Abs differ significantly regarding Ab concentration, plate coating conditions, or plate type. Thus, we believe that the observed discrepancies in the results obtained by different groups might be at least in part related to variations in the experimental settings. The fact that PMA influences Th1 cell differentiation early after activation, whereas 2DG exerts its effect later, supports the hypothesis that T cell activation conditions determine susceptibility to glycolytic flux inhibition. The experiments performed in this study using Ag-specific OTII cells activated with DC, which is a more physiological setting for T cell activation, showed that OTII cells did not respond to PMA or thapsigargin treatment. Effective T cell activation depends on costimulatory molecules (36). Among the many costimulatory molecules present on DC or T cells, OX40 (37), 4-1BB (38), and ICOS (39) were shown to influence calcium or DAG signaling. The αCD3/αCD28–mediated stimulation model is missing these elements. Thus, it is possible that during Ag-specific OTII TCR stimulation, the cells receive the optimal calcium and DAG signal, whereas they are not provided upon αCD3/αCD28–mediated stimulation. This would explain why PMA and thapsigargin did not influence DC-activated OTII Th1 cells with regard to IFN-γ production, whereas they restore expression of this cytokine under conditions of diminished glycolytic flux in αCD3/αCD28–activated Th1 cells. However, regardless of the molecular mechanism behind it, the data presented in this article show the limitations of the αCD3/αCD28 activation model in studies of glucose metabolism in lymphocytes. In this regard, it is particularly important to take into account that many studies on lymphocyte metabolism are performed using this model (16, 23, 24, 40), and the obtained results are rarely controlled using Ag-specific stimulation systems (41).

Differences in activation conditions may also be responsible for inconsistencies among published results. Glycolysis was shown to control IFN-γ expression in Th1 cells solely at the translational level (23). We cannot exclude that this control mechanism exists; however, in our experimental conditions it is definitely not the only mechanism. We clearly observed that the inhibition of glycolysis blocked a Th1 differentiation program, as manifested by decreased STAT1 and STAT4 phosphorylation, diminished T-bet expression, and impaired transcription of the IFN-γ gene. The studies carried out using OTII Th1 cells confirmed the transcriptional control of IFN-γ expression because cells differentiated under conditions of diminished glycolytic flux had a decreased IFN-γ mRNA level and T-bet expression. In contrast, mannose supplementation increased IFN-γ mRNA transcription in OTII cells at a high 2DG dose without restoration of IFN-γ protein production, which in turn suggests a translational expression control.

Another unexpected observation was that when OTII cells were grown in media containing galactose as the only hexose, Th1 polarization and expression of IFN-γ was preserved. As mentioned earlier, 2DG block glycolysis, but it is also a direct inhibitor of protein glycosylation, whereas galactose, as the only hexose in the culture medium, has only an indirect effect on glycosylation (9). It is possible that the optimal activation conditions provided by DC make Th1 OTII cells resistant to limited availability of glycolytic intermediates. To overcome the stronger effect of simultaneous inhibition of glycolysis and glycosylation, cells need to be provided with external mannose. There are several reports indicating that 2DG disturbs cell functions because of direct inhibition of glycosylation rather than glycolysis (9, 11, 13, 14, 42), and our observations further support this hypothesis. Our data showed that mannose alone restores IFN-γ production if cells are cultured on galactose, whereas if 2DG is present in the culture medium, glucose and mannose are necessary for IFN-γ expression. This further proves that 2DG and galactose influence Th1 cells in a different way.

Our data and those reported by others demonstrate that cell signaling regulates Th1 cell sensitivity to glycolysis inhibition (24). The prospect of rendering immune cells resistant to limited glucose availability is extremely attractive and physiologically relevant under the framework of clinical applications. There are recent reports showing that limited glucose availability in a tumor environment limits antitumor immune response (23–25). Overcoming this limitation could be crucial for effective cancer treatment. The work on chimeric Ag receptor T cells showed that, indeed, this approach may be useful. The CD28 domain in chimeric Ag receptor T cells was shown to enhance glycolysis (43). Therefore, further studies aimed at unraveling the underlying mechanisms determining lymphocyte susceptibility to glycolysis inhibition are of great importance to identify potential intervention targets amenable to manipulation by immune interventions.

In contrast, dietary mannose supplementation was shown to be an efficient treatment of congenital glycosylation disorders (44). Our data showed that IFN-γ expression by Th1 cells depends on mannose metabolism and not glycolysis per se and that this phenomenon is not dependent on the experimental model used. Therefore, it is justified to speculate that even an easy-to-implement approach such as mannose supplementation could have a positive effect on IFN-γ expression in, for example, a tumor environment. Because IFN-γ expression is important for tumoricidal activity of lymphocytes, such a treatment could have beneficial effects on cancer treatment (24). However, further studies on this topic using appropriate preclinical validation models will be required.

We thank Blair Prochnow for critical reading of the manuscript. We thank Maciej Sobczynski and Pawel Bulkowski for statistical analysis.

The authors have no financial conflicts of interest.