Research ArticleImmunology

Tumor-derived TGF-β inhibits mitochondrial respiration to suppress IFN-γ production by human CD4+ T cells

See allHide authors and affiliations

Science Signaling  17 Sep 2019:
Vol. 12, Issue 599, eaav3334
DOI: 10.1126/scisignal.aav3334

Suppressing antitumor immunity

The cytokine TGF-β has both immune-suppressive and tumor-suppressive functions; thus, a better understanding of the cell-type specificity of the effects of TGF-β might improve therapeutic strategies that target it. Dimeloe et al. found that TGF-β from tumor effusions suppressed the antitumor activity of CD4+ T cells by inhibiting their production of the inflammatory cytokine IFN-γ. The effects of TGF-β were mediated by Smad proteins in the mitochondria, rather than in the nucleus, and led to decreased mitochondrial respiration. Indeed, direct inhibition of a mitochondrial electron transport chain complex in CD4+ T cells was sufficient to inhibit IFN-γ production. Thus, these data suggest that TGF-β targets T cell metabolism to suppress antitumor immunity.

Abstract

Transforming growth factor–β (TGF-β) is produced by tumors, and increased amounts of this cytokine in the tumor microenvironment and serum are associated with poor patient survival. TGF-β–mediated suppression of antitumor T cell responses contributes to tumor growth and survival. However, TGF-β also has tumor-suppressive activity; thus, dissecting cell type–specific molecular effects may inform therapeutic strategies targeting this cytokine. Here, using human peripheral and tumor-associated lymphocytes, we investigated how tumor-derived TGF-β suppresses a key antitumor function of CD4+ T cells, interferon-γ (IFN-γ) production. Suppression required the expression and phosphorylation of Smad proteins in the TGF-β signaling pathway, but not their nuclear translocation, and depended on oxygen availability, suggesting a metabolic basis for these effects. Smad proteins were detected in the mitochondria of CD4+ T cells, where they were phosphorylated upon treatment with TGF-β. Phosphorylated Smad proteins were also detected in the mitochondria of isolated tumor-associated lymphocytes. TGF-β substantially impaired the ATP-coupled respiration of CD4+ T cells and specifically inhibited mitochondrial complex V (ATP synthase) activity. Last, inhibition of ATP synthase alone was sufficient to impair IFN-γ production by CD4+ T cells. These results, which have implications for human antitumor immunity, suggest that TGF-β targets T cell metabolism directly, thus diminishing T cell function through metabolic paralysis.

INTRODUCTION

Transforming growth factor–β1 (TGF-β) plays a complex role in tumor initiation and progression, demonstrating both tumor-suppressive and tumor-promoting activities. The tumor-suppressive activities of TGF-β are linked to its capacity to inhibit cellular proliferation, induce apoptosis, and suppress growth factor production (1). However, many tumor types produce TGF-β in large quantities, and this is associated with metastasis and poor patient prognosis. Tumor-promoting activities of TGF-β include dysregulation of the cell cycle, increased extracellular matrix formation, angiogenesis, and, importantly, inhibition of antitumor T cell immunity (1).

A critical immune-regulatory role for TGF-β was established by the lethal inflammatory phenotype of TGF-β–deficient mice (2). T cells are critical targets of TGF-β, because specific deletion of TGF-β receptor II (TGF-βRII) on T cells phenocopied this disease (3, 4). TGF-β influences thymic selection and differentiation of mature T cells to promote central and peripheral immune tolerance (57). TGF-β inhibits T cell proliferation and suppresses the differentiation and function of cytotoxic CD8+ T cells, as well as proinflammatory T helper 1 (TH1) and TH2 CD4+ T cell subsets. Conversely, TGF-β promotes expression of the transcription factor FoxP3, driving regulatory T (Treg) cell development. However, reconstitution with Treg cells only partially rescues disease in mice with TGF-βRII–deficient T cells (4, 5), indicating the importance of direct T cell inhibitory activities of TGF-β.

Because of its tumor-promoting and immune-regulatory roles, TGF-β may present a valuable therapeutic target in cancer. Small-molecule inhibitors of TGF-βR signaling, TGF-β–neutralizing antibodies, and oligonucleotides targeting TGF-β are currently under investigation in clinical trials (8). However, completely inhibiting TGF-β activity may also impair its beneficial tumor-suppressive activity. A deeper understanding of TGF-β signaling and its cellular effects is therefore required to inform precise and context-appropriate therapeutic approaches.

TGF-β binds to its tetrameric receptor complex of TGF-βRI and TGF-βRII to stimulate receptor serine and threonine kinase activity and phosphorylation of downstream targets. In the canonical model of TGF-β signaling, phosphorylated Smad2 and Smad3 interact with Smad4. Subsequent nuclear translocation of this trimeric complex leads to transcriptional gene regulation (9). Smad-independent TGF-β signaling pathways have also been described, which involve phosphoinositide 3-kinase (PI3K), the mitogen-activated protein kinase (MAPK) p38, the small guanosine triphosphatase (GTPase) RhoA, and the kinase ROCK (1). Furthermore, a TGF-β–independent role for Smad4 in T cells was identified (10). In addition to their role in transcriptional regulation, Smad proteins also directly interact with other proteins (11). These include p53 or p63 complexes, inhibiting their tumor-suppressive function (1), as well as subunits of the mitochondrial electron transport chain (ETC), which results in increased reactive oxygen species production and apoptosis (12). Here, we assessed the effect of tumor-derived TGF-β on human CD4+ T cell effector function and identified TGF-β–mediated cellular mechanisms that may inform targeted therapeutic interventions to rescue the suppression of antitumor immunity by TGF-β.

RESULTS

Tumor-derived TGF-β inhibits effector memory CD4+ T cell interferon-γ production, which requires Smad phosphorylation but not nuclear translocation

The tumor microenvironment can be rich in TGF-β, which is implicated in preventing effective antitumor T cell responses. We quantified TGF-β in a panel of 11 effusion fluids from a range of human metastatic tumors (Fig. 1A and table S1) and assessed their effect on the production of the cytokine interferon-γ (IFN-γ) by effector memory CD4+ T cells, which constitute most of the tumor-infiltrating CD4+ T lymphocytes (13). Specifically, activated human effector memory CD4+ T cells (characterized as CD127+CD25CD45RACCR7 cells) from healthy donors were stimulated for 16 hours in the presence of tumor effusion fluid (50% with 50% serum-free medium to buffer pH) and, additionally, either an isotype control antibody (Ab.) or a specific antibody to neutralize TGF-β. These assays showed that neutralization of TGF-β in the effusion fluids increased the percentage of effector memory CD4+ T cells that produced IFN-γ and showed a positive correlation between the concentration of TGF-β in the respective effusions and the capacity to rescue IFN-γ production by blocking this cytokine (Fig. 1, B and C).

Fig. 1 Tumor-derived TGF-β inhibits effector memory CD4+ T cell IFN-γ production, which requires Smad phosphorylation but not nuclear translocation.

(A) Quantification of the amounts of TGF-β in 11 tumor effusions (see table S1 for patient details). (B to F) Activated effector memory CD4+ T cells were restimulated for 16 hours in tumor supernatant (mixed 50:50 with serum-free medium) together with either an isotype control antibody (Ab.) or a TGF-β–neutralizing mAb (α–TGF-β). (B) For each tumor effusion, comparisons were made of the mean fold change in IFN-γ+ cells (α–TGF-β/isotype control mAb) as assessed by flow cytometry, corrected for dilution (factor of 2) of tumor effusion fluid, with the concentration of TGF-β present in the sample [as described in (A) and table S1]; n = 11 pairs, each effusion tested on three to five independent healthy donors. Further measurements were made of the percentage of IFN-γ+ cells (C) and the corrected fold increase in IFN-γ+ cells (D), as well as the abundance (E) and the corrected fold change in the amount of IFN-γ secreted into the culture medium (F). For (C) to (F), n = 22 independent biological replicates; combined data for eight tumor effusions (178, 222, 225, 280, 238, 164, 193, and 167), each tested on cells from three to six independent donors. (G and H) Calculation of the percentage of IFN-γ–producing (+) cells (G) and the corrected fold change in IFN-γ+ cells [α–TGF-β/isotype control mAb; (H)] among activated effector memory CD4+ T cells that were pretreated for 30 min with 10 μM SB-431542 or 5 μM ivermectin, as indicated, and then were restimulated for 16 hours in the presence of tumor effusion, as described, with isotype control (red dots/bars) or α–TGF-β mAb (black dots/bars) in 21% O2. ns, not significant. (I and J) Experiments were performed as described for (G) and (H) but were performed in 1% O2. Tumor effusions 178, 222, 225, and 167 were used (n = 14/15 for at least three independent donors for each effusion). P values were calculated by Pearson correlation for (B); paired t test for (C) and (E); Wilcoxon test for (D) and (F); and two-way analysis of variance (ANOVA) for (G) to (J). *P < 0.05, **P < 0.01, and ***P < 0.0001.

TGF-β signals by canonical pathways involving Smad phosphorylation, nuclear translocation, and transcriptional regulation, as well as noncanonical Smad-dependent and Smad-independent pathways (1, 9). Understanding the relative importance of these respective signaling mechanisms in driving a given outcome should inform therapeutic strategies that aim at targeting TGF-β signaling to restore T cell function. To investigate the cellular mechanisms involved in the TGF-β–mediated suppression of IFN-γ production by CD4+ T cells, we further characterized effusion fluids with TGF-β–dependent inhibitory capacity. We then analyzed the effect of neutralizing TGF-β in these effusions on the percentage of IFN-γ–producing cells and on the total amounts of IFN-γ secreted into the cell culture medium and confirmed that neutralization of TGF-β in these effusion fluids achieved a statistically significant increase in the percentage of IFN-γ–producing effector memory CD4+ T cells and in the total amounts of IFN-γ that they secreted (Fig. 1, C to F). Using these fluids, we then interrogated the requirement for Smad phosphorylation and nuclear translocation in our system in experiments with pharmacological inhibitors, namely, the activin receptor-like kinase (ALK) inhibitor SB-431542, which prevents Smad phosphorylation (14), and ivermectin, which inhibits the importin α/β nuclear transporters involved in Smad nuclear translocation (fig. S1) (15,16). These experiments showed that the increase in IFN-γ expression achieved by neutralization of TGF-β in the effusion fluid was no longer achieved in the presence of SB-431542 but did still occur in the presence of ivermectin (Fig. 1, G and H). Therefore, there was no additional effect of neutralizing TGF-β in the effusion fluid when Smad phosphorylation was already blocked, which was not the case for Smad nuclear translocation, implying that Smad phosphorylation, but not nuclear translocation, was required for the immunosuppressive effect of tumor-derived TGF-β on effector memory CD4+ T cells. Increased IFN-γ expression upon TGF-β neutralization was also not observed under hypoxic conditions (Fig. 1, I and J), which suggested that the observed nuclear translocation–independent mechanism may have a metabolic basis.

TGF-β inhibits the respiratory capacity of activated effector memory CD4+ T cells

To directly investigate whether TGF-β affected the metabolism of CD4+ T cells, we next exposed activated effector memory CD4+ T cells to TGF-β for 16 hours and then measured their respiratory and glycolytic capacity (fig. S2; see Materials and Methods for details). We found that TGF-β statistically significantly decreased the basal and adenosine triphosphate (ATP)–coupled oxygen consumption rate (OCR) (by about 30 and 40%, respectively) in these cells but had no effect on spare respiratory capacity (SRC) (Fig. 2, A to D). Comparison of ATP-coupled OCR with FCCP (carbonyl cyanide p-trifluoromethoxyphenylhydrazone)–stimulated maximal OCR showed that the TGF-β–mediated suppression of ATP-coupled OCR was partially rescued by FCCP (with 40% suppression decreasing to ~20%) (Fig. 2E), implying that a functional inhibition of the ETC may be at least partly responsible for the observed decrease in basal and ATP-coupled respiration. A modest, yet consistent, decrease in the basal and maximal rates of glycolysis was also observed in TGF-β–treated cells (Fig. 2, F to H). The effect of TGF-β on mitochondrial respiration was, however, greater than that on glycolysis, as evidenced by a decreased OCR/ECAR ratio (Fig. 2I). Note that the inhibition of basal and ATP-coupled OCR by TGF-β was unhindered in the presence of ivermectin but did not occur when the cells were treated with SB-431542, which suggests that Smad phosphorylation, but not nuclear translocation, was required also for these metabolic effects of TGF-β (fig. S3, A and B). The role of Smad2 in TGF-β–mediated metabolic suppression was further confirmed by genetic manipulation of its expression. Smad2 protein abundance was efficiently reduced in primary CD4+ T cells and Jurkat T cells (by about 75 and 60%, respectively) by transfection with Smad2-specific small interfering RNA (siRNA) (fig. S4). Subsequent metabolic analyses of these cells revealed that the extent to which TGF-β suppressed ATP-coupled respiration in cells transfected with Smad2-specific siRNA was statistically significantly less than that of cells transfected with control siRNA (Fig. 2J). Similar to previous results, glycolysis was only modestly affected by TGF-β, and knockdown of Smad2 had no statistically significant effect (Fig. 2K). Furthermore, TGF-β did not affect the viability of activated effector memory CD4+ T cells after 16 hours of incubation (Fig. 2L).

Fig. 2 TGF-β inhibits the respiratory capacity of activated effector memory CD4+ T cells.

(A to I) Activated human effector memory CD4+ T cells were treated for 16 hours with either control medium (black line) or TGF-β (5 ng/ml; red line), as indicated, and then were assessed for the following metabolic parameters as described in fig. S2: (A) OCR profile, (B) basal OCR, (C) ATP-coupled OCR, (D) SRC, (E) percentage of control ATP-coupled and maximal OCR, (F) extracellular acidification rate (ECAR), (G) basal ECAR, (H) maximal ECAR, and (I) ATP-coupled OCR/basal ECAR (OCR/ECAR) ratio. Data are means ± SEM of four independent donors. (J and K) ATP-coupled OCR and basal ECAR of (J) total primary CD4+ T cells or (K) Jurkat cells, which were transfected with control scrambled siRNA or Smad2-specific siRNA and then treated with TGF-β as described for (A) to (I). Data are expressed as a percentage of the rates under control treatment in the absence of TGF-β. (L) Percentage of annexin V+ cells among activated effector memory CD4+ T cells that were treated as described for (A) to (F). Data are means ± SEM of (J) eight independent donors and (K) three biological replicates. All P values were calculated by paired t test. *P < 0.05.

Activation of naïve murine CD4+ T cells in the presence of TGF-β for prolonged periods of time, which induces the differentiation of Treg cells, increases, rather than decreases, mitochondrial respiratory capacity, which is further associated with increased fatty acid oxidation (FAO) (17). Contrary to the reduction in ATP-coupled respiration observed when activated effector memory CD4+ T cells were treated with TGF-β for only 16 hours (Fig. 2), we also observed that the activation of bulk human CD4+ T cells for 72 hours with antibodies against CD3 and CD28 in the presence of TGF-β (2 ng/ml), which increased the percentage of FoxP3+ Treg cells (fig. S5A), increased mitochondrial membrane potential (ΔΨm) and SRC (fig. S5, B to D); however, the rates of glycolysis were not statistically significantly changed (fig. S5, E and F). Thus, TGF-β differentially affected mitochondrial function, depending on the duration of exposure.

One mechanism by which TGF-β might inhibit ATP-coupled respiration in CD4+ T cells is through modulation of the activity of mechanistic target of rapamycin (mTOR) complex 1 (mTORC1), a master regulator of cellular metabolism. To interrogate this, we first assessed mTORC1 activity by examining the phosphorylation of a key target protein, p70S6 kinase (p70S6K). These experiments revealed that treatment of activated total CD4+ T cells with TGF-β statistically significantly suppressed mTORC1 activity, which was dependent on Smad2 phosphorylation (fig. S6, A and B). Therefore, we next interrogated the extent to which mTORC1 inhibition was responsible for the observed effects of TGF-β on CD4+ T cell metabolism using the mTORC1 inhibitor rapamycin. These assays showed that TGF-β and rapamycin both independently reduced the ATP-coupled oxygen consumption of activated CD4+ T cells and, furthermore, that there was an additive effect in cells treated with both compounds, indicating that there were mTORC1-independent effects of TGF-β on mitochondrial function (fig. S6, C and D), consistent with our earlier data (Fig. 2E), which suggested functional inhibition of the ETC. The effects of rapamycin and TGF-β on glycolysis were modest, and there was no additive effect of both compounds (fig. S6, C and D).

Smad proteins are present in the mitochondria of CD4+ T cells

To further dissect these mTORC1-independent effects, we first investigated whether TGF-β affected the transcription of a panel of master regulators of mitochondrial biogenesis. After 5 or 24 hours of treatment with TGF-β, the abundances of these mRNAs in total, activated CD4+ T cells were unaffected (Fig. 3A). Consistent with this, exposure to TGF-β for 24 hours also did not affect the overall abundances of ETC complexes I to V (Fig. 3B). These data therefore also pointed to a potential direct inhibition of mitochondrial function, rather than a transcriptional regulation of mitochondrial component abundance. On the basis of their structure, the TGF-β signaling proteins Smad2, Smad3, and Smad4 are predicted to localize to mitochondria (18), and they have been detected in mitochondrial fractions of nonlymphoid cells (12). We therefore examined the subcellular localization of pSmad2/3 by imaging flow cytometry (Fig. 3C). In addition to being stained for pSmad2/3, CD4+ T cells were stained with MTR and 4′,6-diamidino-2-phenylindole (DAPI) to identify mitochondria and nuclei, respectively. Analysis of fluorochrome abundance in these cellular compartments confirmed a statistically significant enrichment of MTR in the mitochondria compared to the nuclear marker DAPI and furthermore identified that pSmad2/3 was enriched in the mitochondria compared to DAPI (Fig. 3D). Comparison of control and TGF-β–treated cells revealed an increased abundance of pSmad2/3 in both nuclear and mitochondrial compartments in response to TGF-β, which was comparable in terms of its fold increase (Fig. 3, E and F). These data suggest that phosphorylated Smad proteins were present in the mitochondria of CD4+ T cells and that the extent of their phosphorylation was increased upon TGF-β treatment. Evidence for in vivo relevance of mitochondrial pSmad2/3 was obtained by studying tumor-associated CD4+ T cells (recovered from the effusion fluids used earlier). Direct ex vivo staining of pSmad2/3 in these cells and analysis by flow cytometry revealed that the abundance of pSmad2/3 in tumor-associated CD4+ T cells correlated with the concentration of TGF-β in the respective effusion fluid from which the cells were recovered (Fig. 3G; see Fig. 1A and table S1). In addition, imaging analysis identified a similar enrichment of pSmad2/3 in the mitochondria of tumor-associated CD4+ T cells compared to DAPI, as was observed for peripheral counterparts in vitro (Fig. 3, H and I).

Fig. 3 Smad proteins are present in the mitochondria of CD4+ T cells.

(A) Activated total CD4+ T cells were treated with vehicle (control; black bars) or TGF-β (5 ng/ml; red bars) for 5 or 24 hours, as indicated. The abundances of the indicated mRNAs were then quantified relative to that of 18S ribosomal RNA and expressed as relative quantities (RQ) compared to the control treatment. Data are means ± SEM of six independent donors. (B) Activated total CD4+ T cells were treated with vehicle (control; black bars) or TGF-β (5 ng/ml; red bars) for 24 hours, as indicated. Cells were then analyzed by Western blotting with antibodies against the indicated mitochondrial ETC complexes. A representative Western blot (noncontiguous lanes) is shown. Graphs show the relative abundance (normalized to that of actin) of the indicated ETC components. Data are means ± SEM of five independent donors. (C) Representative images of activated total CD4+ T cells that were treated for 1 hour with TGF-β (5 ng/ml; top) or with vehicle as a control (bottom) and then were stained with MitoTracker Red (MTR; red), DAPI (blue), and anti-pSmad2/3 (turquoise) before being analyzed with an ImageStream X. (D) Summary data of the mean fluorescence intensity (MFI) in mitochondria/MFI in nuclei of the indicated fluorochromes in total CD4+ T cells treated with TGF-β as described for (C). Data are from four independent donors, with MFIs of >2000 cells per condition and per donor. (E) Summary data of the MFI of pSmad2/3 in defined mitochondrial and nuclear regions in control (control; black bars) and TGF-β–treated (5 ng/ml; red bars) CD4+ T cells. Data are means ± SEM of four independent donors. (F) Measurement of the fold increase in pSmad2/3 MFI in the mitochondrial and nuclear regions of total CD4+ T cells treated as described for (E). Data are from four independent donors. (G) Flow cytometric determination of pSmad2/3 MFIs (black bars, left y axis) of tumor-associated CD4+ T cells (from effusions 178, 225, and 201) stained ex vivo, compared with the TGF-β concentration in each effusion (red bars, right y axis). (H) Representative images (from sample 201) of CD4+ T cells (gated on CD4) among tumor-associated cells, which were stained ex vivo with MTR (red), DAPI (blue), and anti-pSmad2/3 (turquoise) and analyzed with an ImageStream X. (I) Summary data of the MFI in mitochondria/MFI in nuclei of the indicated fluorochromes in cells treated as described for (H). Data are from three independent donors, with MFIs of >2000 cells per donor. P values were calculated by Wilcoxon test for (A), paired t test for (B) and (F), one-way ANOVA for (D), and two-way ANOVA for (E). *P < 0.05 and **P < 0.01.

TGF-β specifically inhibits ATP synthase activity

To define the molecular basis through which mitochondrial Smad proteins affected CD4+ T cell respiration, we assessed what aspect of mitochondrial function was impaired. The functions of ETC complexes I, II, III, and IV were assessed by measuring the OCRs of isolated mitochondria in response to the provision of specific substrates or electron donors for each complex (fig. S7, A to D). Performing such analyses on mitochondria isolated from activated effector memory CD4+ T cells precultured for 5 hours in control or TGF-β–containing medium demonstrated decreased oxygen consumption in response to all four substrates or electron donors (Fig. 4A and fig. S7, A to D). These findings were suggestive of (i) the decreased functionality of all ETC complexes; (ii) the decreased function of complex IV (which consumes the oxygen that is measured); or (iii) the decreased function of complex V, ATP synthase, which may subsequently affect the efficiency of the entire ETC and cannot be directly measured in these experiments. To answer this question, we next undertook biochemical studies to selectively measure the activities of complex I, II, IV, or V using cell lysates of control and TGF-β–treated, activated total CD4+ T cells. With TGF-β–treated cells, these assays revealed no inhibition of the functions of complex I, II, or IV, but showed statistically significant inhibition of complex V activity (Fig. 4B). Consistent with the inhibition of complex V activity, TGF-β also caused a statistically significant increase in T cell ΔΨm (Fig. 4C).

Fig. 4 TGF-β specifically inhibits ATP synthase activity.

(A) Mean OCR of activated, effector memory CD4+ T cells that were treated for 5 hours with either control medium (black bars) or TGF-β (5 ng/ml; red bars) and then permeabilized to assess mitochondrial respiration of specific substrates or in response to electron donors provided for complexes I, II, III, and IV. Data are means ± SEM of eight independent donors. (B) Mean activities of complexes I, II, IV, and V as assessed by biochemical assays in activated, total CD4+ T cells treated for 5 hours with control medium (black bars) or TGF-β (5 ng/ml; red bars). Data are means ± SEM of 10 independent donors. mOD, milli optical density. (C) ΔΨm of activated, total CD4+ T cells expressed as a ratio of MitoSpy Orange MFI in untreated versus FCCP-treated cells previously treated with control medium or TGF-β as described for (B). Data are means ± SEM of four independent donors. P values were calculated by Wilcoxon test for (A) and paired t test for (B) and (C). *P < 0.05 and ***P < 0.0001.

TGF-β inhibits effector memory CD4+ T cell IFN-γ production independently of nuclear Smad import but requires mitochondrial respiration

Both the glycolytic and mitochondrial functionality of T cells critically underpin key immune functions (19). We next examined the effect of metabolic inhibition by TGF-β on the immune function of effector memory CD4+ T cells—specifically on IFN-γ production—which we had observed to be statistically significantly impaired by the TGF-β present in tumor effusions (Fig. 1). By performing multiplex cytokine analyses, we confirmed that TGF-β markedly inhibited the secretion of IFN-γ by effector memory CD4+ T cells compared to that of other cytokines, including interleukin-10 (IL-10) and IL-17 (Fig. 5A). Inhibition of IFN-γ secretion by TGF-β in other subsets of activated CD4+ T cells (naïve and central memory) was less substantial and did not reach statistical significance (fig. S8A), which was consistent with the fact that these cells have reduced amounts of TGF-βRII and less TGF-β–induced Smad2/3 phosphorylation (fig S8, B to E).

Fig. 5 TGF-β inhibits effector memory CD4+ T cell IFN-γ production independently of nuclear Smad import but requires mitochondrial respiration.

(A) Measurement of the amounts of IFN-γ, tumor necrosis factor–α (TNF-α), IL-10, IL-17A, and IL-4 secreted by activated effector memory CD4+ T cells that were restimulated for 16 hours in control medium (black bars) or with TGF-β (5 ng/ml; red bars). The cytokines in the cell culture medium were measured by cytometric bead array. Data are means ± SEM of six independent donors. (B) MFIs of isotype control antibody staining or anti-pSmad2/3 antibody staining in activated, total CD4+ T cells that were pretreated with 5 μM ivermectin or 10 μM SB-431542 for 30 min and then treated for 1 hour with control medium (black bars) or TGF-β (5 ng/ml; red bars). Data are means ± SEM of three independent donors. (C) Fold increase in colocalization of fluorescence between DAPI (nuclear marker) and p-Smad2/3 in cells treated as described for (B). Data are means ± SEM of three independent donors. (D) Activated CD4+ T cells, pretreated for 30 min with ivermectin or SB-431542, as indicated, were then cultured for 16 hours with either control medium (black bars) or TGF-β (5 ng/ml; red bars). The abundances of the indicated mRNAs were quantified relative to that of 18S rRNA and expressed as relative quantities compared to control cells. Data are means ± SEM of three independent donors. (E) Percentage of IFN-γ+ cells among activated effector memory CD4+ T cells that were pretreated for 30 min with ivermectin or SB-431542, as indicated, and then restimulated overnight in the presence of control medium (black bars) or TGF-β (5 ng/ml; red bars). Data are means ± SEM of five independent donors. (F) Measurement of the concentration of IFN-γ in culture medium from tumor-associated cells that were stimulated overnight with anti-CD3/anti-CD28 mAb in the presence of TGF-β (same concentration as per effusion; see Fig. 1A and table S1) together with either an isotype control antibody (red bars) or a TGF-β–neutralizing mAb (black bars), performed either under control conditions (n = 3 independent donors; left) or in the presence of 5 μM ivermectin (n = 2 independent donors; right). <Det. indicates below the assay detection limit. (G) Percentage of IFN-γ+ cells among activated effector memory CD4+ T cells that were restimulated for 16 hours in control medium (black bar) or in the presence of 1 μM oligomycin (blue bar). Data are means ± SEM of five independent donors. (H) Measurement of the amount of IFN-γ in the culture medium of effector memory CD4+ T cells that were treated as described for (H). Data are means ± SEM of eight independent donors. (I) Measurement of the amount of IFN-γ in the culture medium of effector memory CD4+ T cells that were restimulated for 16 hours in control medium or in the presence of TGF-β with or without oligomycin, as indicated. Data are means ± SEM of four independent donors. (J) Percentage of IFN-γ+ cells among activated effector memory CD4+ T cells that were restimulated for 16 hours at 1% atmospheric O2 in control medium (black bar), in the presence of TGF-β (5 ng/ml; red bar), or 1 μM oligomycin (blue bar). Data are means ± SEM of five independent donors. (K and L) Measurement of the amount of IFN-γ in the culture medium of effector memory CD4+ T cells that were treated as described for (J). Data are means ± SEM of six to eight independent donors. P values were calculated by paired t test for (A), (H), (I), (K), and (L); two-way ANOVA for (B), (D), (E), and (F); and one-way ANOVA for (C) and (J). *P < 0.05.

TGF-β can inhibit IFN-γ production by transcriptional repression of the genes TBX21 (which encodes T-bet) and IFNG (which encodes IFN-γ) (9, 20). However, we observed that nuclear translocation of pSmad2/3 was not required for the TGF-β–dependent inhibition of IFN-γ production by tumor effusions (Fig. 1). To further define the TGF-β–mediated transcriptional versus metabolic regulation of IFN-γ production, we again blocked the nuclear import of Smad proteins using the nuclear import inhibitor, ivermectin, and compared its effects to those of inhibition of Smad phosphorylation with SB-431542. We first confirmed, by flow cytometric analysis of pSmad2/3 abundance, that SB-431542, but not ivermectin, prevented the TGF-β–mediated phosphorylation of Smad2/3 (Fig. 5B). Next, having verified by imaging flow cytometry that ivermectin and SB-431542 both inhibited the nuclear import of pSmad2/3 (Fig. 5C and fig. S1), we furthermore confirmed that both inhibitors prevented the transcriptional repression of TBX21 and IFNG expression by TGF-β (Fig. 5D). We then found that TGF-β still inhibited IFN-γ production even in the presence of ivermectin, which blocked the nuclear import of phosphorylated Smad proteins and thus the transcriptional effects of TGF-β. However, inhibiting Smad phosphorylation with SB-431542 abolished the inhibitory effects of TGF-β on IFN-γ production (Fig. 5E), suggesting that there were critical roles for phosphorylated Smad proteins in inhibiting IFN-γ production by effector memory CD4+ T cells by transcription-independent mechanisms. The findings from these experiments are consistent with our initial observations from experiments with tumor effusions.

We then aimed to verify the biological relevance of TGF-β–mediated suppression of IFN-γ production by tumor-associated T cells. To do so, we assessed IFN-γ production by tumor-associated T cells activated in the presence of TGF-β at the concentration that was present in their respective effusion fluids and either an isotype control antibody or a TGF-β–neutralizing monoclonal antibody (mAb), which would inactivate any TGF-β produced by tumor cells in the culture. Consistent with the in vitro functional and mechanistic experiments, these assays confirmed that TGF-β at the amounts present in the effusions also inhibited tumor-associated T cells directly ex vivo in a concentration-dependent manner compared to cultures in which TGF-β was neutralized (Fig. 5F, left; see Fig. 1A and table S1). Furthermore, with two donors for which sufficient cells were available, we additionally observed that the TGF-β–mediated suppression of IFN-γ production occurred despite the blockade of Smad nuclear import by ivermectin (Fig. 5F, right).

Next, we assessed whether inhibition of complex V and other ETC complexes per se could affect IFN-γ production by effector memory CD4+ T cells. We found that inhibition of complex V with oligomycin was sufficient to decrease both the percentage of effector memory CD4+ T cells that produced IFN-γ and the total amount of IFN-γ secreted by these cells (Fig. 5, G and H), which was also true for the inhibition of complexes I, II, III, and IV (fig. S9, A to D). Unhindered mitochondrial ETC function was thus required for IFN-γ production by effector memory CD4+ T cells and seems to be specifically targeted, at the level of complex V, by TGF-β signaling. Notably, oligomycin did not achieve the same degree of suppression of IFN-γ production as did TGF-β, suggesting that TGF-β acts by additional pathways in the cytoplasm to suppress IFN-γ production (for example, through the inhibition of mTORC1). However, when cells were treated with TGF-β, no additional suppression of IFN-γ production was achieved through the addition of oligomycin, suggesting that the ETC-linked IFN-γ production was already maximally suppressed by TGF-β (Fig. 5I). Further confirmation that inhibition of mitochondrial function played an important role in the suppression by TGF-β of IFN-γ production by effector memory CD4+ T cells came from observations that IFN-γ production was not inhibited by TGF-β under hypoxic conditions, which was also true for the complex V inhibitor, oligomycin (Fig. 5, J to L), and for tumor effusions (Fig. 1, I and J).

Tumor-derived TGF-β inhibits effector memory CD4+ T cell metabolism

To assess whether direct metabolic inhibition, driven by TGF-β, was recapitulated using clinical samples rich in this cytokine, we assessed the capacity of two tumor effusions (see table S1 and Fig. 1) to affect the metabolic parameters of activated effector memory CD4+ T cells. These assays confirmed that activated effector memory cells, cultured in the presence of effusion fluid with an isotype control antibody, had reduced ATP-coupled respiration compared to that of cells cultured identically, yet in the presence of a specific TGF-β–neutralizing mAb (Fig. 6, A and B). Basal rates of glycolysis were also slightly reduced in these cells; however, calculation of OCR/ECAR ratios again indicated that the effect of TGF-β in effusion fluids on respiration was greater than that on glycolysis (Fig. 6C).

Fig. 6 Tumor-derived TGF-β inhibits effector memory CD4+ T cell metabolism.

(A) OCR and ECAR of activated effector memory CD4+ T cells that were preincubated for 5 hours in tumor supernatant (mixed 50:50 with serum-free medium) together with isotype control antibody (red line) or TGF-β–neutralizing mAb (α–TGF-β; black line) and additionally treated with 1 μM oligomycin, 2 μM FCCP, and 1 μM rotenone, as indicated, to measure metabolic parameters as described in fig. S1. Data are from a representative example with effusion 178. (B) Fold change in ATP-coupled OCR and basal ECAR in cells treated as described for (A) with isotype control antibody (red circles) or TGF-β–neutralizing mAb (α–TGF-β; black circles). Data are from seven independent donors; combined data are for experiments with effusion 178 and 225. (C) ATP-coupled OCR/basal ECAR ratios of cells treated as described for (A). P values were calculated by Wilcoxon test for (B) and paired t test for (C). *P < 0.05.

DISCUSSION

Production of TGF-β by tumors is associated with metastasis and poor patient prognosis. This is related to both the tumor-promoting and immune inhibitory effects of TGF-β; however, tumor-suppressive activities of TGF-β have also been described (1). Thus, although TGF-β is a promising therapeutic target for cancer, fully exploiting its potential will require a precise and context-specific understanding of its cellular effects.

Here, we investigated the cellular mechanisms involved in the TGF-β–mediated inhibition of an important CD4+ T cell antitumor function: IFN-γ production (21). We found that effusions from metastatic human tumors impaired IFN-γ production in a TGF-β–dependent manner and, furthermore, that TGF-β used at concentrations observed ex vivo suppressed IFN-γ production by human tumor–associated T cells. This immunosuppression occurred through a mechanism that required the phosphorylation, but not the nuclear translocation, of Smad proteins. Rather, the inhibitory effect of TGF-β depended on environmental oxygen abundance. We identified that Smad proteins were present and phosphorylated within the mitochondria of T cells exposed to TGF-β both in vitro and in vivo and that TGF-β statistically significantly impaired both complex V activity and ATP-coupled respiration, the latter in a Smad2-dependent manner. In addition, the inhibition of complex V per se was sufficient to impair IFN-γ production by CD4+ T cells.

TGF-β signaling affects metabolism at both the organismal and cellular level. Increased hypothalamic TGF-β abundance in obesity and aging has been linked to systemic glucose intolerance (22). More directly, mice deficient in Smad3 or treated with TGF-β–neutralizing antibodies are protected from diet-induced obesity and diabetes, which is associated with the “browning” of white adipocytes, heightened mitochondrial biogenesis, and the augmented respiratory capacity of these cells (23). We found no transcriptional regulation of a panel of transcription factors that orchestrate mitochondrial biogenesis after 5 or 24 hours of treatment with TGF-β. However, activation of total CD4+ T cells for 72 hours in the presence of TGF-β (to promote the generation of Treg cells) increased mitochondrial respiratory capacity and FAO, consistent with previous reports (17), which is suggestive of duration-dependent effects of TGF-β on CD4+ T cells. At the cellular level, TGF-β has been detected in the mitochondria of cardiomyocytes, hepatocytes, and T cells (2426). T cells lacking TGF-β have abnormal mitochondrial morphology, indicating a possible homeostatic role for TGF-β in mitochondrial biogenesis, dynamics, or both (26). To date, cell-intrinsic versus autocrine signaling-dependent effects of TGF-β have, however, not been assessed. Functionally, TGF-β decreases ΔΨm in human fibroblasts (27), ETC complex IV activity in murine lung epithelial cells (28, 29), and the function of complex IV in murine renal podocytes (30). TGF-β also inhibits cellular metabolism through effects on mTORC1 activity. We confirmed that TGF-β inhibited mTORC1 in human CD4+ T cells and, furthermore, that direct mTORC1 inactivation statistically significantly repressed CD4+ T cell respiration. However, we additionally showed, by combining TGF-β treatment with direct mTORC1 inhibition, that TGF-β has substantial effects on CD4+ T cell mitochondrial respiration independently of mTORC1 regulation. TGF-β also impairs mitochondrial Ca2+ uptake in renal arteriolar cells (31). Because Ca2+ flux is critical for T cell signaling, this observation is of particular interest. Smad signaling proteins, based on their structure, are predicted to localize to mitochondria (18); Smad3, Smad4, and Smad5 were detected in the mitochondria of various cell types, but not immune cells (12, 18). Our study now links complex V–associated Smad and TGF-β–driven phosphorylation thereof with mitochondrial suppression and impaired immune functionality.

The activation of T cells drives important changes in cellular metabolism that are critical for their effector function, through bioenergetic and nonbioenergetic mechanisms (32, 33). The signaling pathways that drive metabolic reprogramming during T cell activation are quite well characterized and involve the PI3K-Akt-mTOR axis, as well as the transcription factors c-Myc, estrogen-related receptor α (ERR-α), and AP4 (19). Signaling from inhibitory receptors, such as CTLA-4 and PD-1, antagonizes these pathways to limit metabolic reprogramming (34). T cell effector function is also limited in the tumor microenvironment by competition for metabolic precursors (35, 36). To date, cell-extrinsic signals that directly modulate T cell metabolism to mediate an immune-regulatory effect have not been described. For TGF-β, such a role is consistent with both its immune-regulatory function and abundance in malignantly transformed tissue sites. Selectively alleviating T cell metabolic paralysis induced by TGF-β, while maintaining transcriptional activity with postulated antitumor activity, might prove to be an effective strategy for cancer therapy.

MATERIALS AND METHODS

Study design

The objective of this research was to investigate the mechanism by which human tumor–derived TGF-β suppresses the immune function of human CD4+ T cells. This research was performed with biological material (tumor effusions and T cells) from patients with cancer, as well as T cells isolated from the peripheral blood of healthy donors. The experimental design consisted of controlled laboratory in vitro studies and ex vivo analyses of patient cells as described. No data were excluded from the analyses. Experiments were performed at least three times using cells from independent human donors. Exact numbers of replicates are stated in the figure legends.

Peripheral blood CD4+ T cell isolation, sorting, and culture

Bulk CD4+ T cells were isolated as described previously (37). Where indicated, flow cytometry–sorted cell populations were used. T cell populations were sorted according to the cell surface expression of CD127, CD25, CD45RA, and CCR7 to purify effector memory cells (CD127+CD25CD45RACCR7). Purity was typically >90%. The following mAbs were used: anti-CD45RA mAb–Pacific Blue (clone 2H4; Beckman Coulter), anti-CD25 mAb–Brilliant Violet 605 (clone 2A3; BD Biosciences), anti-CCR7 mAb–phycoerythrin (PE) (clone FABP197; R&D Systems Europe), and anti-CD127 mAb-allophycocyanin (clone eBioRDR5; eBioscience). Unless otherwise indicated in the figure legend, cells were resuspended at 1 × 106/ml in RPMI 1640 containing 10% AB+ human serum, penicillin (50 U/ml), streptomycin (50 mg/ml; Invitrogen; RPMI/10% AB), and recombinant IL-2 (rIL-2) (50 IU/ml; PeproTech). Where serum-free medium is indicated in the figure legend, AIM-V serum-free medium (Invitrogen) containing rIL-2 (50 IU/ml) was used. Cells were activated where indicated by being placed in cell culture plates precoated with anti-CD3 and anti-CD28 (1 and 5 μg/ml; clones OKT-3 and CD28.2, respectively; BioLegend) for 72 hours. Activated cells were then washed and resuspended at 1 × 106/ml in RPMI/10% AB or AIM-V medium (as indicated) under the experimental conditions described in the figure legend. Additions to the cell cultures included TGF-β (2 to 20 ng/ml; PeproTech), 5 μM ivermectin (Sigma-Aldrich), 10 μM SB-431542 (SelleckChem), rapamycin (20 ng/ml; Merck), 1 μM oligomycin (Sigma-Aldrich), 0.1 μM rotenone (Sigma-Aldrich), 200 μM 2-thenoyltrifluoroacetone (Sigma-Aldrich), 2 μM antimycin A (Sigma-Aldrich), 10 μM sodium azide (Sigma-Aldrich), mouse immunoglobulin G1 (IgG1) isotype control antibody (5 μg/ml; clone 11711, R&D Systems Europe), and anti–TGF-β mouse IgG1 mAb (5 μg/ml; clone 1D11, R&D Systems Europe). The effects of SB-431542 and ivermectin on CD4+ T cell IFN-γ expression are shown in fig. S10.

Tumor effusion samples

Tumor effusion fluids were collected from patients with cancer (see table S1 for details) presenting with malignant effusions at the University Hospital Basel after obtaining written informed consent. After collection, the cellular components of the malignant effusions were separated by centrifugation, and the cells and fluids were frozen at −80°C until further use. Where indicated in the legends, total effusion cells were cultured or analyzed, and CD4+ T cells were identified by staining with CD4–fluorescein isothiocyanate (FITC) (clone RPA-T4, BD Biosciences).

RNA-mediated interference

Activated primary human CD4+ T cells or Jurkat cells were transfected with pools of Smad2-specific siRNA or control scrambled siRNA (Qiagen) for 72 hours using the AMAXA T cell nucleofection kit (Lonza). Knockdown efficiency was verified by Western blotting analysis.

Extracellular metabolic flux analysis of intact cells and isolated mitochondria

For analysis of the OCR (in pmol/min) and ECAR (in mpH/min), the Seahorse XFe96 metabolic extracellular flux analyzer was used (Seahorse Bioscience). Total or sorted CD4+ T cells were resuspended in serum-free, unbuffered RPMI 1640 (Sigma-Aldrich) and were plated onto Seahorse cell plates (2.5 × 105 cells per well) coated with Cell-Tak (BD Biosciences) to enhance T cell attachment. Perturbation profiling of the use of metabolic pathways was done by the addition of oligomycin (1 μM), FCCP (2 μM), and rotenone (1 μM; all are given as final concentrations, all from Sigma-Aldrich). For a description of how the various metabolic parameters were calculated, see fig. S1. The contribution of FAO to OCR was assessed by the addition of etomoxir (40 μM; Sigma-Aldrich). Assessment of the OCR of isolated mitochondria was also performed with the Seahorse XFe96 metabolic extracellular flux analyzer and followed a protocol described previously (38). Sorted effector memory CD4+ T cells were plated onto Seahorse cell plates, as described earlier, in RPMI/10% AB containing TGF-β (5 ng/ml) where indicated in the figure legends and cultured at 37°C and 5% CO2 for 5 hours. The medium was then exchanged for mannitol and sucrose buffer [70 mM sucrose, 220 mM mannitol, 10 mM KH2PO4, 5 mM MgCl2, 2 mM Hepes, and 1 mM EGTA (pH 7.2); all reagents from Sigma-Aldrich] containing fatty acid–free bovine serum albumin (4 mg/ml; Sigma-Aldrich). During the assay, first plasma membrane permeabilizer (1 nM; Seahorse Bioscience) and then the following substrates (final concentrations given) together with adenosine diphosphate (1 mM) and FCCP (2 μM; both from Sigma-Aldrich) were added: complex I: pyruvate (5 mM)/malate (2.5 mM; both from Sigma-Aldrich); complex II: succinate (10 mM; Sigma-Aldrich); complex III: duroquinol (0.5 mM; TCI America); and complex IV: tetramethyl-p-phenylenediamine (0.5 mM)/ascorbate (2 mM; both from Sigma-Aldrich). Last, the following specific inhibitors were added: complex I: rotenone (1 μM); complex II: malonate (40 μM); complex III: antimycin A (20 μM); and complex IV: sodium azide (20 mM; all from Sigma-Aldrich).

Flow cytometric analysis of protein abundance and phosphorylation

Phosphorylation of Smad2/3 was assessed in cells (0.2 × 106) previously fixed for 20 min at 37°C with fixation/permeabilization solution (BD Biosciences), permeabilized for 30 min at 4°C with Phosflow Perm Buffer III (BD Biosciences), and washed twice with phosphate-buffered saline (PBS) by staining for 30 min with Alexa Fluor 647–conjugated anti-pSmad2/3 (clone O72-670, BD Biosciences), further washing, and analysis. For assessment of IFN-γ production by intracellular cytokine staining, cells (0.2 × 106) were restimulated for 16 hours with soluble anti-CD3 and anti-CD28 (1 and 5 μg/ml, respectively) under culture conditions as indicated in the figure legends and then treated for 4 hours with phorbol 12-myristate 13-acetate (10 ng/ml; Sigma-Aldrich) and ionomycin (500 ng/ml; Sigma-Aldrich). During the final 2 hours of activation, the cells were treated with monensin (BioLegend) to block cytokine secretion. The cells were then washed and fixed for 20 min at 37°C with fixation/permeabilization solution, washed with permeabilization buffer before being stained for 45 min with PE-conjugated anti–IFN-γ (clone B27, ImmunoTools), and undergoing further washing and analysis. FoxP3 abundance was assessed in cells (0.2 × 106) previously fixed for 20 min at 37°C in FoxP3 fixation/permeabilization solution (eBioscience) and washed with FoxP3 permeabilization buffer (eBioscience) by staining for 45 min with anti-human FoxP3-PE (clone PCH101, eBioscience), which was followed by further washing and analysis.

Detection of apoptosis

Annexin V staining was performed to identify apoptotic cells by flow cytometry. Staining was performed using annexin V binding buffer (BD Biosciences) and FITC-conjugated annexin V (ImmunoTools).

ΔΨm measurement

MitoTracker staining was performed to assess differences in ΔΨm. Cells (0.2 × 106) were incubated in RPMI 1640/10% AB with 100 nM MitoTracker Green or MTR (Invitrogen) for 20 min at 37°C and 5% CO2 before undergoing washing and analysis by flow cytometry. ΔΨm was calculated as follows: MTR MFI/MitoTracker Green MFI.

Quantification of mRNA

The relative abundances of mRNAs of interest were quantified by real-time reverse transcription polymerase chain reaction (PCR) (qPCR). The mRNA was extracted with Qiagen spin columns (Qiagen), and cDNA was transcribed with the GoScript Reverse Transcription System (Promega) according to the manufacturer’s instructions. SYBR Green or TaqMan primers (Applied Biosystems) were used for qPCR analysis (see table S1 for further details).

Western blotting analysis

Cell lysates for Western blotting analysis were prepared in radioimmunoprecipitation assay buffer (Thermo Fisher Scientific), and protein concentrations were determined with a BCA protein assay kit (Thermo Fisher Scientific). Whole-cell lysates were resolved by 4 to 15% SDS–polyacrylamide gel electrophoresis and were transferred onto nitrocellulose membranes. The membranes were then incubated with the Total OXPHOS Human WB Antibody Cocktail (ab110411, Abcam). The membranes were then stained with the appropriate secondary antibody IRDye 800CW–conjugated goat polyclonal antibody to rabbit IgG (926-32211) and IRDye 800CW– or IRDye 680RD–conjugated goat polyclonal antibody to mouse IgG (926-68070; both from LI-COR). The Odyssey imaging system (LI-COR) was used for band detection.

Imaging flow cytometry

MTR staining was performed as described earlier. Subsequently, cells (0.5 × 106) were fixed and stained for pSmad2/3 as described earlier. Last, cells were resuspended in PBS with 300 nM DAPI (Molecular Probes) to stain nuclei and analyzed using the ImageStream X imaging flow cytometer. Nuclear and mitochondrial masks were defined based on regions of DAPI/MTR intensity, respectively, and colocalization was calculated using analysis tools in IDEAS software.

Biochemical assays of ETC function

Complex I (ab109721), II (ab109908), IV (ab109909), and V (ab109714) function was assessed in lysates from 5 × 106 to 10 × 106 activated CD4+ T cells using the indicated kits from MitoSciences (Abcam) according to the manufacturer’s instructions.

Cytokine measurements in culture medium

Cytokine production by T cells was measured using the TH1/TH2/TH17 Cytokine Bead Array (560484, BD Biosciences) or by enzyme-linked immunosorbent assay (ELISA) for IFN-γ (eBiosciences) according to the manufacturers’ instructions. The amount of IFN-γ present in total tumor-associated cell cultures was normalized (based on flow cytometry phenotyping) for 50,000 CD4+ T cells.

TGF-β ELISA

TGF-β concentrations in cell culture medium were measured using the human TGF-β ELISA set (BD Biosciences) according to the manufacturer’s instructions.

Statistical analysis

Data were tested for normality with the Shapiro-Wilk test. Data with normal distribution were assessed by paired Student’s two-sided t test. Multiple groups were compared by one- or two-way ANOVA and a Bonferroni posttest for multiple comparisons. Non-normally distributed data were compared using a Wilcoxon test. Error bars represent SEM of the data presented.

Study approval

Blood samples were obtained from healthy donors after written informed consent. The study was approved by the Swiss Red Cross (blood transfusion service) and Institutional Review Board.

SUPPLEMENTARY MATERIALS

stke.sciencemag.org/cgi/content/full/12/599/eaav3334/DC1

Fig. S1. Analysis of pSmad2/3 abundance in CD4+ T cells treated with TGF-β, ivermectin, and SB-431542.

Fig. S2. Representative profiles of OCR and ECAR measurements of quiescent and activated T cells as assessed with an extracellular flux analyzer.

Fig. S3. Inhibition of CD4+ T cell mitochondrial respiration by TGF-β requires Smad phosphorylation, but not nuclear localization.

Fig. S4. Knockdown efficiency of Smad2-specific siRNA in human primary CD4+ T cells and Jurkat cells.

Fig. S5. CD4+ T cells differentiated with TGF-β to induce FoxP3+ Treg cells demonstrate increased mitochondrial SRC.

Fig. S6. TGF-β decreases mTORC1 activity in CD4+ T cells, and the TGF-β–mediated decreases in CD4+ T cell mitochondrial respiration are partly mTORC1 dependent.

Fig. S7. Treatment of effector memory CD4+ T cells with TGF-β decreases oxygen consumption in response to provision of specific substrates for complexes I, II, III, and IV.

Fig. S8. Effector memory CD4+ T cells are more susceptible to TGF-β–mediated IFN-γ suppression than are naïve or central memory cells, consistent with increased TGF-βRII abundance and signaling.

Fig. S9. Inhibition of ETC complexes I, II, III, and IV suppresses IFN-γ production by effector memory CD4+ T cells.

Fig. S10. Effects of SB-431542 and ivermectin on CD4+ T cell IFN-γ production as measured by intracellular staining or ELISA.

Table S1. Clinical details for tumor effusion samples.

Table S2. Primer sequences used for the quantification of mRNA by PCR.

REFERENCES AND NOTES

Acknowledgments: We thank T. Krebs, E. Traunecker, and D. Labes for technical support with cell sorting and A. Buser (University Hospital Basel) for buffy coats. Funding: This work was supported by the Roche postdoctoral fellowship program and University of Basel research funds (to S.D.) and by Swiss National Science Foundation grants 310030_153059 (to C.H.) and 323530-139181 (to M.F.). Author contributions: S.D. designed, performed, and analyzed most experiments and wrote the manuscript. J.L., C.F., L.D., M.F., F.M., G.R.B., and Y.L. performed experiments. P.G. designed, performed, and analyzed experiments. D.T. and A.Z. obtained and prepared tumor effusion samples and provided intellectual input regarding experimental design and interpretation. A.L. and C.H. oversaw the study and experiments. C.H. wrote the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper or the Supplementary Materials.
View Abstract

Navigate This Article