TGF-β Regulates CD8+ T Cell Memory by Triggering mTORC1Weak-Mediated Activation of the Transcriptional FOXO1-TCF1-Eomes and Metabolic AMPK-ULK1-ATG7 Pathways

TL;DR

TGF-β helps create long-lasting CD8+ memory T cells by activating specific transcription and metabolic pathways, which could improve vaccines for cancer and infectious diseases.

Contribution

This study reveals a novel mechanism by which TGF-β regulates CD8+ T cell memory through mTORC1Weak-mediated activation of FOXO1-TCF1-Eomes and AMPK-ULK1-ATG7 pathways.

Findings

TGF-β triggers weak mTORC1 signaling, which activates transcriptional factors like FOXO1, TCF1, and Eomes.

TGF-β promotes metabolic changes involving AMPK-ULK1-ATG7 pathways and fatty acid oxidation in CD8+ memory T cells.

TGF-β-induced CD8+ memory T cells show enhanced survival and mitochondrial mass.

Abstract

What are the main findings?
TGF-β functions as an antagonist of mTORC1 activation.TGF-β regulates CD8+ T cell memory via the transcriptional FOXO1-TCF1-Eomes and metabolic AMPK-ULK1-ATG7 pathways.

TGF-β functions as an antagonist of mTORC1 activation.

TGF-β regulates CD8+ T cell memory via the transcriptional FOXO1-TCF1-Eomes and metabolic AMPK-ULK1-ATG7 pathways.

What are the implications of the main findings?
The Smad-independent TGF-β signaling pathway plays an important role in controlling T cell immunity.TGF-β-related signaling is a critical node to target for the development of novel vaccines for the treatment of cancer and infectious diseases.

The Smad-independent TGF-β signaling pathway plays an important role in controlling T cell immunity.

TGF-β-related signaling is a critical node to target for the development of novel vaccines for the treatment of cancer and infectious diseases.

CD8+ memory T (TM) cells are essential for vaccine-induced protective immunity. While transforming growth factor beta (TGF-β) triggers CD8+ TM cell differentiation, the underlying molecular mechanism(s) has yet to be uncovered. We therefore used a well-established cell culture protocol to prepare TGF-β-triggered CD8+ TM cells derived from chicken ovalbumin (OVA)-specific T cell receptor (TCR) transgenic OTI mice, and systematically characterized them using Western blotting, confocal microscopy, flow cytometry and Seahorse assay analyses. We found that TGF-β/T cells exhibit a TM cell phenotype (CD62L+KLRG1−) and display long-term survival upon adoptive transfer into mice. To elucidate the signaling circuitry underpinning the observed transcriptional and metabolic changes required to promote CD8+ TM cell differentiation, we measured the expression of several critical factors and found that TGF-β triggered weak mTORC1 (mTORC1Weak) signaling. mTORC1Weak signaling in turn led to an increase in the abundance of key transcriptional (TCF1, FOXO1 and Eomes) and metabolic (AMPK-α1, ATG7, ULK1, SIRT1, OPA1 and LAL) factors and an elevation in mitochondrial mass and reliance on fatty acid oxidation (FAO). Our data thus reveal for the first time that TGF-β regulates CD8+ T cell memory by triggering mTORC1Weak-mediated activation of the transcriptional FOXO1-TCF1-Eomes and metabolic AMPK-ULK1-ATG7 pathways. Given that induction of more qualified CD8+ TM cells is one of the ultimate goals of vaccination, our findings identify additional targets critical to TGF-β-induced T cell memory, which may greatly impact future vaccine development for the treatment of cancer and infectious diseases.

Figures

Full text

1. Introduction

CD8^+^ T cells are essential for combatting cancer and infectious diseases [1]. CD8^+^ T cell responses proceed through three programmed phases: T cell proliferation, contraction and memory [2]. In the T cell proliferation phase, naïve T cells differentiate into effector T (T_E_) cells, which rely on glycolysis to meet cellular ATP demands. This is followed by a contraction phase, during which the majority (90–95%) of T_E_ cells die of activation-induced cell apoptosis. The surviving 5–10% of T cells then form a long-term memory T (T_M_) cell population, which uses fatty acid oxidation (FAO) to generate ATP during the memory phase, and it is this T_M_ cell population that promptly responds to the same pathogen upon re-invasion to induce a potent and protective CD8^+^ T cell recall response [2]. One of the classic phenotypic markers used to distinguish long-lived CD8^+^ T_M_ cells from short-lived T_E_ cells is the expression of the T_M_ cell marker CD62L relative to the T_E_ cell marker killer cell lectin-like receptor subfamily G member-1 (KLRG1) [2]. The T_M_ cell population further consists of CD44^+^CD62L^−^ effector memory T (T_EM_), CD44^+^CD62L^+^ central memory T (T_CM_) and CD44^+^CD62L^+^CD45RA^+^ stem cell-like memory T (T_SCM_) cell subsets [2]. Among these CD8^+^ T cell subsets, T_CM_ cells circulate in the blood and patrol lymphoid tissues, while T_EM_ cells traffic through the blood and non-lymphoid tissues to protect against pathogen invasion [3]. Recently, it has been reported that the T_E_ cell marker KLRG1 is also expressed on T_EM_ cells [4]. Unlike T_CM_ cells, which exhibit lower glycolytic activity and utilize FAO to preserve energy homeostasis, T_EM_ cells rely on both FAO and glycolysis but maintain a higher rate of aerobic glycolysis to support their immediate effector functions and their capacity to rapidly respond to secondary infection [5].

Mammalian target of rapamycin complex-1 (mTORC1) is a major, evolutionally conserved environmental and energy sensor that regulates immune cell proliferation, differentiation and metabolism [6]. We previously used modern state-of-the-art biotechnology and genetic and pharmacological tools to elucidate the core molecular pathways that govern distinct mTORC1 signaling strengths critical to T cell memory formation [7,8,9]. We demonstrated that the inflammatory cytokine IL-2 stimulated strong mTORC1 signaling (mTORC1^Strong^) and induced short-term CD8^+^ T_E_ (IL-2/T_E_) cell differentiation by activating the transcriptional T-bet and metabolic cMyc-HIF-1α (hypoxia-inducible factor-1α) pathways to promote the maturation of the CD8^+^ T_E_ phenotype and glycolytic metabolism, respectively [8,9]. In contrast, the pro-survival cytokines IL-7 and IL-15 stimulated weak mTORC1 signaling (mTORC1^Weak^) to promote long-lived CD8^+^ T_M_ (IL-7/T_M_ and IL-15/T_M_) cell formation by coordinately activating the transcriptional FOXO1 (forkhead box-O-1)-TCF1 (T cell factor-1)-ID3 (inhibitor of DNA binding-3)-Eomes and metabolic AMPK-α1 (AMP-activated protein kinase-α1)-ULK1 (Unc-51-like autophagy-activating kinase-1)-ATG7 (autophagy-related gene-7) pathways controlling the CD8^+^ T_M_ phenotype and FAO metabolism, respectively [8,9].

Transforming growth factor beta (TGF-β) is an immunosuppressive cytokine that directly inhibits both innate and adaptive T cell immune responses [10] and indirectly suppresses T cell responses by TGF-β-induced immunosuppressive CD4^+^ regulatory T (Treg) cells [11]. In addition, TGF-β signaling has been reported to promote CD8^+^ T_M_ cell differentiation in vitro and in vivo in a mouse infection model [12,13]. TGF-β exposure has also been found to upregulate the levels of T_M_ cell transcription factor signaling ID3 [12], reduce the abundance of the T_E_ cell phenotypic marker KLRG1 [14] and suppress the metabolic regulator c-Myc essential for glycolysis metabolism [15]. However, the underlying transcriptional and metabolic pathways by which TGF-β signaling induces CD8^+^ T cell memory formation are largely unknown.

In this study, we systematically characterized TGF-β-stimulated T (TGF-β/T) cells and IL-2-stimulated T (IL-2/T) cells (used as a control) prepared by using well-established protocols [8,9,16]. We found that IL-2/T cells became short-term CD62L^−^KLRG1^+^ T_E_ (IL-2/T_E_) cells, while TGF-β/T cells differentiated into long-lived CD62L^+^KLRG1^−^ T_M_ (TGF-β/T_M_) cells. To identify the molecular pathways that control the formation of TGF-β/T_M_ cells, we used flow cytometry, molecular and biochemical analyses [8,9] to demonstrate that TGF-β indeed triggered mTORC1^Weak^ signaling, induced upregulation of transcription regulators FOXO1, TCF1 and Eomes and metabolic regulators AMPK-α1, ULK1, ATG7, SIRT1 (silent information regulator of transcription-1), OPA1 (optic atrophy-1) and LAL (lysosomal acid lipase), and promoted mitochondrial biogenesis and FAO. Thus, our findings provide the first evidence that TGF-β indeed regulates CD8^+^ T cell memory by triggering our previously reported mTORC1^Weak^ signaling-mediated activation of the transcriptional FOXO1-TCF1-Eomes and metabolic AMPK-α1-ULK1-ATG7 pathways [8,9].

2. Materials and Methods

2.1. Mice, Antibodies and Reagents

In this study, the animal protocol (AUP#20180065) was approved by the Animal Use and Care Committee, University of Saskatchewan. C57BL/6 mice (B6, CD45.2^+^) were purchased from Charles River (Wilmington, MA, USA). B6.SJL-Ptprca Pepcb/BoyJ mice (B6.1, CD45.1^+^) and ovalbumin (OVA)-specific T cell receptor transgenic OTI mice on a B6 background were purchased from Jackson Laboratory (Bar Harbor, MA, USA). B6.1 mice were intercrossed with OTI mice to generate CD45.1^+^ B6.1/OTI mice in house at the University of Saskatchewan [8,9]. IL-2 and TGF-β recombinant proteins were purchased from Peprotech (Rocky Hill, NJ, USA). The following antibodies (Abs) were purchased from BioLegend (San Diego, CA, USA): Allophycocyanin (APC)-KLRG1, Brilliant Violet 421-CD62L, PE-CD44, PE-Cy5-CD8, FITC-CD45.1, APC-CD8, FITC-FOXO1 and FITC-TCF1. A PE-conjugated H-2K^b^/OVA_257–264_ tetramer (PE-tetramer) was obtained from the Fred Hutchinson Cancer Research Center (Seattle, WA, USA). The following Abs were obtained from Cell Signaling Technology (Danvers, MA, USA): pS6 (S_235/236_), FOXO1, T-bet, TCF1, Eomes, cMyc, HIF-1α, pAMPK-α1 (T_172_), pULK1 (S_555_), ATG7, SIRT1, LAL, OPA1, β-actin and HRP-conjugated anti-rabbit IgG. The Easysep CD8^+^ T cell purification kit was acquired from StemCell Technologies (Vancouver, BC, Canada). DAPI (4′,6-diamidino-2-phenylindole) was purchased from ThermoFisher Scientific (Waltham, MA, USA). Fixation/permeabilization buffer was procured from BD (Franklin Lakes, NJ, USA). MitoTracker Green and Hoechst solution were purchased from Life Technologies Inc. (Carlsbad, CA, USA). Seahorse XF Mito Stress Test Kits were purchased from Agilent Technologies (Lexington, MA, USA).

2.2. Optimized Culture Protocol for IL-2/TE and TGF-β/TM Cell Preparation

CD8^+^ Tn cells were purified from the spleens of CD45.1^+^ B6.1/OTI mice using an Easysep CD8^+^ T cell purification kit. We then generated IL-2-stimulated effector T (IL-2/T_E_) cells by using our own established culture protocol [8,9] and TGF-β-stimulated memory T (TGF-β/T_M_) cells by making minor modifications to another culture protocol [16]. Briefly, CD8^+^ Tn cells were cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS), 2-mercaptoethanol (2-ME, 50 µM) and ovalbumin (OVA)257–264 peptide (OVAI, SIINFEKL, 0.1 nM) in the presence of either IL-2 (100 U/mL) or IL-2 and TGF-β (10 ng/mL) (Peprotech, Rocky Hill, NJ, USA) for 3 days to yield IL-2/T_E_ and TGF-β/T_M_ cells, respectively. The activated CD8^+^ T cells were then cultured for another 2 days in the same medium lacking OVAI peptide to generate CD45.1^+^ IL-2/T_E_ and TGF-β/T_M_ cells. To ensure that our culture protocol was optimal, we performed two related experiments examining the effect of culture duration (3, 4 or 5 days) and TGF-β concentration (1, 3 or 10 ng/mL) on the relative abundance of factors critical to TGF-β/T_M_ cell formation. Our results clearly indicate that culturing the TGF-β/T_M_ cells for 5 days (Figure 1A) in media containing 10 ng/mL TGF-β (Figure 1B) resulted in the most significant reduction in pS6 (S_235–236_) levels and most pronounced upregulation in the abundance of the transcriptional FOXO1 and metabolic pAMPKα1 (T_172_) regulators. Thus, this culture protocol is indeed optimal for stimulating TGF-β-induced T cell memory.

2.3. Flow Cytometry

CD8^+^ IL-2/T_E_ and TGF-β/T_M_ cell populations were incubated for 30 min on ice in the dark with a mixture of APC-KLRG1, PE-CD44 and Brilliant Violet 421-CD62L Abs (each at 1:100) in 200 µL of flow cytometry buffer (2% FBS and 0.1% sodium azide in PBS). Cell samples were then washed twice with the same buffer and analyzed by flow cytometry. The frequency of KLRG1^+^CD62L^−^ T_E_ cells was measured initially. Cell samples were then gated KLRG^−^ and KLRG1^+^CD62L^+^ T_M_ cell populations were isolated to quantitate CD44^+^CD62L^−^ T_EM_ and CD44^+^CD62L^+^ T_CM_ cell subsets, respectively. Flow cytometry analyses were performed on a Cytoflex Multicolour Flow cytometer (Beckman, San Diego, CA, USA). Data were analyzed with FlowJo 10 software (FlowJo, Ashland, OR, USA) [8,9].

2.4. T Cell Survival Analyses

To assess T cell survival, viable CD45.1^+^ IL-2/T_E_ and TGF-β/T_M_ cells (5 × 10^6^ cells/mouse) were intravenously (i.v.) injected into CD45.2^+^ B6 mice. Mouse peripheral blood samples were then harvested 1, 7 and 15 days post injection and stained with PE-Cy5-CD8, FITC-CD45.1, PE-CD44 and BV421-CD62L Abs, and CD8^+^CD45.1^+^ IL-2/T_E_ and TGF-β/T_M_ cell abundance was quantified by flow cytometric analysis [8,9]. The CD8^+^CD45.1^+^ TGF-β/T_M_ cells were then analyzed for T_EM_ and T_CM_ by measuring CD44 and CD62L expressions. To assess T_M_ cell recall responses, recipient mice were challenged one month post adoptive T cell transfer by i.v. injection with 2000 CFUs of recombinant Listeria monocytogenes OVA (rLmOVA), and the abundance of PE-tetramer^+^ FITC-CD8^+^ T cells was quantified by flow cytometry 4 days after the rLmOVA boost [8,9].

2.5. Western Blotting

CD8^+^ IL-2/T_E_ and TGF-β/T_M_ cells were lysed in ice-cold RIPA buffer containing a protease and phosphatase inhibitor cocktail (Thermo Fisher Scientific, Waltham, MA, USA). Equal amounts of cell lysate (20 µg) were separated by SDS-PAGE and transferred onto PVDF membrane. The membrane was blocked with 5% BSA in PBS containing 0.05% Tween-20 (PBST) and incubated with various antibodies overnight at 4 °C. After washing with PBST for 10 min/wash for a total of 30 min, the membrane was incubated in secondary antibody for 1 h at room temperature. The membrane was then washed again as described above and imaged using a BioRad Chemidoc MP (Bio-Rad, Hercules, CA, USA). Band intensities were analyzed with Image J software (ImageJ.JS (https://ij.imjoy.io/), TreeStar, Ashland, OR, USA), and β-actin intensity was used as an internal reference control to normalize for differences in loading across lanes [8,9].

2.6. Confocal Microscopy

To visualize the intracellular localization of TCF1 or FOXO1, CD8^+^ IL-2/T_E_ and TGF-β/T_M_ cells were incubated with an APC-CD8 Ab at 4 °C for 30 min. T cells were then fixed and permeabilized in fixation/permeabilization buffer at 4 °C for 20 min, washed once with 1× permeabilization wash buffer, and incubated at 4 °C for 30 min with FITC-Abs against FOXO1 or TCF1 (1:100) in 1× permeabilization buffer. After 3 washes with 1× permeabilization wash buffer, an antifade mountant with DAPI was added and the cells were then deposited on microscope slides and coverslipped. Microscope slides were imaged by using the Zeiss LSM700 confocal microscope (Carl Zeiss, Oberkochen, BW, Germany). Confocal imaging was analyzed using ZEN 3.8 imaging software, as previously described [8,9].

2.7. Mitochondrial Analysis

To measure mitochondrial mass, IL-2/T_E_ or TGF-β/T_M_ cells were incubated with 50 nM MitoTracker Green (Life Technologies Inc., Carlsbad, CA, USA) for 15 min at 37 °C in the dark, washed three times with PBS and analyzed by flow cytometry. To directly visualize mitochondrial content, after staining with MitoTracker Green, IL-2/T_E_ and TGF-β/T_M_ cells were then incubated with Hoechst 33342 solution (5 µg/mL) for 5 min at 37 °C in the dark. Cells were then deposited on microscope slides, coverslipped and imaged using the Zeiss LSM700 confocal microscope (Carl Zeiss, Oberkochen, BW, Germany) [8,9].

2.8. Seahorse Assay

Mitochondrial respiration and glycolytic activity of IL-2/T_E_ and TGF-β/T_M_ cells were assessed by measuring the oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) using a Seahorse XFp Analyzer (Agilent Technologies, Santa Clara, CA, USA), according to the manufacturer’s instructions. Cells were seeded onto XF8 cell culture microplates at 1.5 × 10^5^ cells per well in assay medium supplemented with glucose (10 mM), sodium pyruvate (1 mM) and L-glutamine (2 mM). A mitochondrial stress test was conducted by measuring OCR (pmol/min) under basal conditions and following sequential injections of 1.5 µM oligomycin (port A), 2.5 µM FCCP (port B), and 0.5 µM rotenone and antimycin A (port C) into their respective ports, as previously described [8,9]. Data were analyzed by Seahorse Wave Desktop Software (v2.6.3; Agilent Technologies, Santa Clara, CA, USA).

2.9. Statistical Analysis

Results are expressed as mean values with standard deviation (SD). Statistical significance between groups was determined using Student’s t-test. All statistical analyses were performed with Prism 8 software (GraphPad, La Jolla, CA, USA) [8,9]. Probability values of p < 0.05 were considered statistically significant, and p < 0.01 was considered very statistically significant.

3. Results and Discussion

3.1. TGF-β Stimulates the Formation of TGF-β/TM Cells with Long-Term Survival

We used our optimized cell culture protocol to prepare TGF-β-stimulated CD45.1^+^ TGF-β/T cells and IL-2-stimulated IL-2/T cells from naïve CD8^+^ T cells derived from CD45.1^+^ OTI mouse spleen (Figure 2A) [8,9]. To characterize their phenotypes, we first stained the IL-2/T and TGF-β/T cell populations with a cocktail of antibodies specific for the T_E_ cell marker KLRG1 and the T_M_ cell markers CD44 and CD62L and conducted flow cytometric analysis. We found that IL-2/T and TGF-β/T cell populations consisted of 97% and 10% KLRG1^+^CD62L^−^ T_E_ cells, respectively (Figure 2B). We next assessed the enrichment of T_M_ cell subsets within the IL-2/T and TGF-β/T cell populations. Since the T_E_ cell marker KLRG1 was found to be expressed on T_EM_ cells [4], we gated the 2% and 1% KLRG1^+^CD62L^+^ T_M_ cells found in each population for further analysis of CD44 expression and found that they were all KLRG1^+^CD44^+^CD62L^+^ T_CM_ cells (Figure 2B). We next gated the 89% KLRG1^−^ (28% KLRG^−^CD62L^−^ and 61% KLRG1^−^CD62L^+^) T cell population to measure CD44 expression and found that the KLRG^−^ T cell population consisted of a similar amount of 49% CD44^+^CD62L^−^ T_EM_ and 50% CD44^+^CD62L^+^ T_CM_ cells, respectively (Figure 2B). Taken together, our data indicate that the IL-2-stimulated IL-2/T cells consisted almost entirely of KLRG1^+^CD62L^−^ T_E_ cells and are indeed IL-2/T_E_ cells, while TGF-β-stimulated TGF-β/T cells mainly consisted of T_M_ cells (49% CD44^+^CD62L^−^ T_EM_, 50% CD44^+^CD62L^+^ T_CM_ and 1% KLRG1^+^CD44^+^CD62L^+^ T_CM_ cells) and may thus be classified as TGF-β/T_M_ cells.

To further assess their in vivo survival, we adoptively transferred an equal number of viable CD45.1^+^ IL-2/T_E_ and TGF-β/T_M_ cells (5 × 10^6^ cells/each mouse) into CD45.2^+^ B6 mice, and tracked their survival 1, 7 and 15 days post adoptive T cell injection by flow cytometric analysis of peripheral blood samples. Our data demonstrate that TGF-β/T_M_ and IL-2/T_E_ cells were found at similar percentages in peripheral blood 1 day post injection (Figure 2C), confirming that we had indeed injected a similar number of TGF-β/T_M_ and IL-2/T_E_ cells into B6 mice. However, we observed many more TGF-β/T_M_ cells (1.58% and 1.43%) in peripheral blood when compared to IL-2/T_E_ cells (0.20% and 0.12%) at the 7- and 15-day time points (Figure 2C), indicating that TGF-β/T_M_ cells survive much longer than IL-2/T_E_ cells in host B6 mice. Our data thus confirm that our TGF-β/T and IL-2/T cells are authentic, long- and short-lived CD62L^+^KLRG1^−^ T_M_ and CD62L^−^KLRG1^+^ T_E_ cells, respectively [8,9]. To further characterize the long-lived T_M_ cells, we also analyzed the frequencies of the CD44^+^CD62L^−^ T_EM_ and CD44^+^CD62L^+^ T_CM_ subsets within the CD8^+^CD45.1^+^ T cell population in mouse peripheral blood 15 days post adoptive transfer of TGF-β/T_M_ cells. Flow cytometry analyses demonstrated that T_EM_ and T_CM_ cells each represent roughly 50% of the total population (Figure 2C), indicating that both CD44^+^CD62L^−^ T_EM_ and CD44^+^CD62L^+^ T_CM_ cells exhibit similar survival profiles.

To further assess the competency of T_M_ cells, we assayed recall response one month after adoptive transfer of IL-2/T_E_ and TGF-β/T_M_ cells by boosting the mice with the same adjuvant and measuring the response 4 days later. These analyses demonstrated that mice harboring TGF-β/T_M_ cells exhibited an 8-fold-higher OVA-specific CD8^+^ T cell recall response than mice infused with IL-2/T_E_ cells (Figure 2C) and thus corroborate the functional integrity of infused T_M_ cells in recipient mice.

3.2. TGF-β Modulates mTORC1 Signaling

We recently demonstrated that IL-7 and IL-15 stimulate mTORC1^Weak^ signaling to promote the differentiation and formation of CD8^+^ IL-7/T_M_ and IL-15/T_M_ cells [8,9]. To assess the relative strength of mTORC1 signaling triggered by TGF-β, we measured the abundance of the downstream target substrate ribosomal S6 protein by Western blot analysis of total TGF-β/T_M_ and IL-2/T_E_ cell lysates. Consistent with previous reports [8,9,16], we found that the abundance of phosphorylated S6 protein (pS6; Ser_235/236_) was significantly downregulated in TGF-β/T_M_ cells relative to IL-2/T_E_ cells (Figure 3A), which are known to harbor mTORC1^Strong^ signaling [8,9]. Our data therefore indicate that TGF-β functions as an antagonist of mTORC1 activation, thereby triggering mTORC1^Weak^ signaling in TGF-β/T_M_ cells.

3.3. TGF-β Induces CD8+ TM Cell Formation via mTORC1Weak Signaling-Mediated Activation of the Transcriptional FOXO1-TCF1-Eomes Pathway

We next measured the abundance of transcription factors known to be downstream of mTORC1^Weak^ (FOXO1, TCF1 and Eomes) and mTORC1^Strong^ (T-bet) signaling. Consistent with our earlier findings of mTORC1^Weak^ signaling, Western blot data demonstrated that the abundance of FOXO1, TCF1 and Eomes was elevated in TGF-β/T_M_ cells while that of T-bet was reduced, with the reciprocal expression profile being observed in IL-2/T_E_ cells (Figure 3A). These data collectively indicate that TGF-β indeed triggered mTORC1^Weak^ signaling to promote TGF-β/T_M_ cell formation via activation of the transcriptional FOXO1-TCF1-Eomes pathway controlling the CD62L^+^KLRG1^−^ T_M_ cell phenotype, while IL-2-stimulated mTORC1^Strong^ signaling induces IL-2/T_E_ cell differentiation via activation of the transcriptional T-bet pathway regulating the CD62L^−^KLRG1^+^ T_E_ cell phenotype and functional cytotoxicity [8,9].

It has been reported that FOXO1 and TCF1 must localize to the nucleus to exert their functional effects on T_M_ cell formation [8,9]. Therefore, we also performed confocal microscopy analyses to visualize their subcellular localization and found that more FOXO1 and TCF1 protein was indeed localized to the nuclei of TGF-β/T_M_ cells when compared with IL-2/T_E_ cells (Figure 3B,C). Our results thus provide further evidence that the nuclear localization of FOXO1 and TCF1 is required to express the CD62L^+^KLRG1^−^ TGF-β/T_M_ cell phenotype.

3.4. TGF-β Stimulates CD8+ TM Cell Formation via mTORC1Weak Signaling-Mediated Activation of the Metabolic AMPK-α1-ULK1-ATG7 Pathway

AMPK-α1 is a major, evolutionally conserved energy sensor that controls more than one hundred autogenic and mitochondrial respiratory proteins involved in almost all branches of catabolic metabolism [17]. Phosphorylated AMPK-α1 (pAMPK-α1, T_172_) activates ULK1 (pULK1, S_555_) and ATG7 [18] along with various metabolic regulators including SIRT1, OPA1 and LAL [19,20,21], which act in concert to increase mitochondrial mass and reliance on FAO metabolism [22] in support of CD8^+^ T_M_ cell formation [8,9,16]. T_E_ cells exhibit strong mTORC1 signaling, which instead activates the metabolic master regulator hypoxia-inducible factor (HIF)-1α to augment glycolytic flux and generate the ATP required for CD8^+^ T_E_ cell functional activity [8,9]. To investigate which metabolic pathway TGF-β activates in TGF-β/T_M_ cells, we performed Western blot analysis and demonstrated that the autophagy regulators pAMPK-α1 (T_172_), pULK1 (S_555_) and ATG7 as well as their downstream metabolic regulators SIRT1, OPA1 and LAL were all upregulated in TGF-β/T_M_ cells, while cMyc and its downstream target HIF-1α were downregulated (Figure 4A). The reciprocal expression profile was observed for IL-2/T_E_ cells (Figure 4A). Our data collectively indicate that TGF-β triggered mTORC1^Weak^ signaling to activate the metabolic AMPK-α1-ULK1-ATG7 pathway and stimulate ATP production via FAO in TGF-β/T_M_ cells.

3.5. TGF-β Promotes an Increase in Mitochondrial Content and FAO Metabolism in TGF-β/TM Cells

Mitochondria serve as bioenergetic and signaling organelles that play a crucial role in support of T_M_ cell survival [23] by providing adequate oxidative phosphorylation (OXPHOS) potential to maintain spare respiratory capacity, which is essential for FAO [24,25]. Therefore, we used MitoTracker Green to specifically stain mitochondrial membranes in IL-2/T_E_ and TGF-β/T_M_ cells, and quantified and visualized organelle content by flow cytometry and confocal microscopy analyses, respectively [8,9]. Our data show that TGF-β promoted an increased mitochondrial mass in TGF-β/T_M_ cells when compared to IL-2/T_E_ cells (Figure 4B,C). To assess how TGF-β impacts energy metabolism in T_M_ cells, we used the Seahorse assay to analyze the bioenergetic profiles of IL-2/T_E_ and TGF-β/T_M_ cells. Measurements were taken under baseline conditions and after the sequential application of specific OXPHOS inhibitors, as previously described [8,9]. These analyses demonstrated that TGF-β/T_M_ cells displayed a low extracellular acidification rate (ECAR) (Figure 5A) and a high oxygen consumption rate (OCR) (Figure 5B,C), with the elevated OCR/ECAR indicating that TGF-β/T_M_ cells rely on FAO metabolism to maintain energy homeostasis [8,9]. In contrast, IL-2/T_E_ cells had a higher ECAR (Figure 5A) and a lower ratio of OCR/ECAR (Figure 5B), indicating that IL-2/T_E_ cells rely on glycolytic metabolism to meet cellular ATP demands [8,9].

The signaling pathway of the TGF-β superfamily is conserved in both immune and non-immune cells [26]. Active TGF-β binds to its specific receptor TGF-βRII. Then, two TGF-βRIIs and two TGF-βRIs form a tetrameric receptor complex. TGF-βRII phosphorylates and activates TGF-βRI to regulate the downstream activity of two distinct signaling pathways [26]. First, TGF-β activate r-Smad2 proteins, which in turn interact with co-Smad4 to translocate into the nucleus to control target gene expression by interacting with various transcription factors and co-regulators. For example, TGF-β signaling triggers CD4^+^Foxp3^+^ Treg cell differentiation via activation of a Smad-dependent Smad3-Foxp3 pathway (Figure 6A) [27,28]. Second, TGF-β binding to its receptor also induces Smad-independent inhibition of the PI3K-AKT-mTORC1 pathway (Figure 6A), consistent with previous reports [8,9,16].

The “linear cell differentiation (LCD)” or “distinct signaling strengths” model was originally proposed by Sallusto’s group in 2000 and posits that strong and weak strengths of stimulatory signals control T cell differentiation into either short-lived T_E_ or long-lived T_M_ cells [29]. In 2009, Ahmed’s group provided the first evidence that treatment with rapamycin (Rapa), an mTORC1 inhibitor, promotes CD8^+^ T_M_ cell differentiation [30]. This finding was further supported by evidence that Rapa induces T cell memory via a FOXO1-mediated transcriptional switch from T-bet to Eomes [31]. At the time, however, the molecular mechanism(s) underlying Rapa-dependent promotion of T cell memory was largely unknown. We previously demonstrated that treatment with the pro-survival cytokines IL-7 and IL-15 [8,9] or Rapa all induced mTORC1^Weak^ signaling [2] to promote CD8^+^ T_M_ cell formation via activation of the transcriptional FOXO1-TCF1-Eomes and metabolic AMPK-α1-ULK1-ATG7 pathways. In the current study, we further found that TGF-β also induced mTORC1^Weak^ signaling to stimulate T_M_ cell differentiation via activation of the above molecular pathways, thus providing another good example in support of the “distinct signaling strengths” model for T_M_ cell differentiation [7].

Consistent with our previous findings [2,8,9], we show in this study that IL-2/T_E_ cells exhibit strong mTORC1 signaling, which promotes CD8^+^ T_E_ cell formation by activating the transcriptional T-bet and metabolic cMyc-HIF-1α pathways controlling the CD62L^−^KLRG1^+^ phenotype and conferring the cytotoxic properties and reliance on glycolysis (Figure 6B). Interestingly, our study also demonstrates that TGF-β signaling functions as an antagonist of mTORC1 activation and induces its Smad-independent inhibition to trigger mTORC1^Weak^ signaling. We thus provide the first evidence that TGF-β signaling promotes CD8^+^ T_M_ cell formation by triggering our previously reported mTORC1^Weak^-mediated activation of the transcriptional FOXO1-TCF1-Eomes and the metabolic AMPK-α1-ULK1-ATG7 pathways, which act in concert to control the CD62L^+^KLRG1^−^ phenotype and increase mitochondrial content and reliance on FAO to support the longevity and functionality of TGF-β/T_M_ cells (Figure 6B) [2,8,9].

Exhausted CD8^+^ T (Tex) cells are often seen in response to chronic infection or in the tumor microenvironment under conditions of chronic antigenic stimulation. Tex cells are characterized by a dysfunctional phenotype and upregulation of inhibitory co-stimulatory molecules such as programmed death-1 (PD-1), lymphocyte activation gene-3 (LAG3) and T cell immunoglobulin- and mucin-containing protein-3 (TIM3). Immune checkpoint PD-1 blockade is known to reverse exhaustion in early progenitor exhausted T (Tpex) cells [32]. Similarly, TGF-β-dependent suppression of mTORC1 signaling in CD8^+^ Tpex cells has also been shown to reduce the formation of terminally exhausted Tex cells and prevent their death [16].

3.6. Conclusions

Taken together, our study uncovered that TGF-β triggers mTORC1^Weak^ signaling-mediated activation of the transcriptional FOXO1-TCF1-Eomes and metabolic AMPK-α1-ULK1-ATG7 pathways to induce CD8^+^ T cell memory. As such, our study identified additional targets, which may greatly impact vaccine development for the treatment of cancer and infectious diseases, given that one of the ultimate goals of vaccination is to induce more qualified CD8^+^ T_M_ cells.