Autophagy Is Essential for Mitochondrial Clearance in Mature T Lymphocytes¹ Free

Macroautophagy (hereafter referred to as autophagy) is a well-conserved catabolic process in eukaryotic cells characterized by the formation of double-membrane vesicles 0.5–1.5 μm in diameter in the cytoplasm of cells termed autophagosomes (1, 2). Autophagosomes arise through the elongation of cup-shaped isolation membranes that form spherical vesicles before fusing with lysosomes to become degradative compartments. Mature autophagosomes encase both cytosol as well as organelles, consistent with the early characterization of autophagy as a major pathway for protein degradation during periods of starvation (3). Autophagy induction is regulated by the activity of class III PI3K in complex with the essential autophagy gene Beclin-1 (yeast Atg6). Two ubiquitin-like conjugation pathways involving the autophagy genes Atg3, Atg5, Atg7, microtubule-associated protein light chain 3 (LC3, yeast Atg8), Atg10, and Atg12 are also required for the formation of autophagosomes (1, 4).

The induction of autophagy generally executes two complementary functions in eukaryotes that include the recycling of useful metabolic substrates as well as the removal of cytoplasmic material (1, 5, 6). The digestion of intracellular proteins to generate metabolic substrates is essential for cell survival during periods of starvation or growth factor deprivation. This is best demonstrated in mice lacking Atg5 or Atg7, which both succumb to starvation as neonates before robust suckling (7, 8). Importantly, it has also been shown that the survival of an IL-3-dependent hematopoietic cell line during cytokine deprivation depends on autophagosome formation to provide metabolic substrates (9). Additionally, autophagy-mediated clearance of toxic cytoplasmic materials is critical for neural cell survival, as conditional deletion of Atg5 or Atg7 in neurons leads to the accumulation of ubiquitin-positive inclusions and widespread neural cell death (10, 11).

Mounting evidence implicates autophagy in mitochondrial removal (often designated “mitophagy”) in both mammalian and yeast cells (12). Mitochondria have been found within vesicles that possess double or multiple membranes in rat hepatocytes (13), hamster erythroid cells (14), and yeast (15). Modulating autophagy by rapamycin treatment or expression of GAPDH affects mitochondrial mass within cells (16, 17). Mitochondrial permeability transition triggers autophagy-mediated mitochondrial degradation (18, 19). Mitophagy is also induced by cytotoxic agents in the presence of caspase inhibitors (20). In vivo, there has been much interest in the physiologic role of autophagy in mitochondrial clearance during lens and erythroid differentiation (14, 21, 22, 23, 24).

Within immunology, there is an expanding role for autophagy in both the innate and adaptive immune system (25, 26). Autophagy contributes to the clearance of intracellular pathogens (27, 28, 29, 30, 31) as well as the MHCII cross-presentation of endogenous Ags (32, 33). We and others have also demonstrated a role for autophagy in lymphocytes. In T lymphocytes, double-membrane autophagosomes form in both human and murine T cells and can be induced in TCR-stimulated proliferating cells in vitro (34, 35, 36, 37). Although the functional consequences of autophagosome formation in lymphocytes are not well understood, genetic and molecular studies have demonstrated a complex role for autophagy in T cell survival and function. Loss of the essential autophagy gene Atg5 impairs the survival and proliferation of mature T cells in vivo (35), while induction of autophagy by the HIV envelope glycoprotein induces death in target lymphocytes (36). In B lymphocytes, autophagy is required for survival of developing pre-B cells as well as mature B1 B cells (38).

In this paper, we demonstrate that the essential autophagy gene Atg7 is required for mature T cell survival. We show that mitochondrial content is developmentally regulated in T lymphocytes and that autophagy is critical for the clearance of mitochondrial content in mature T cells.

Atg7-floxed and Atg5-deficient mice were generated and characterized previously (7, 8). To generate mice with Atg7-deficient T lymphocytes, Atg7-floxed mice were crossed to Lck-Cre transgenic mice (39) (The Jackson Laboratory). Atg7 genomic deletion was assessed by PCR detection of the Atg7-floxed allele (forward, TGA GAC ATG GCC TGA AGA AAC CCA; reverse, ATG CTG CAG GAC AGA GAC CAT CA) and the wild-type genomic locus (forward, TTA CAG TCG GCC AGG CTG AC; reverse, CCT GGG CTG CCA GAA TTT CTC) in FACS-sorted thymocytes and mature T cells isolated with DNeasy spin kits (Qiagen). Atg5 chimeras were generated by transferring fetal liver cells into lethally irradiated congenic recipient mice as described (35). All animal usage has been approved by the Duke University Institutional Animal Care and Use Committee.

Total thymocytes, EasySep-enriched lymph node (LN)³ T cells (StemCell Technologies), and sorted TCRβ⁺CD4⁺ and TCRβ⁺CD8⁺ T cells from LNs were lysed in 50 mM Tris (pH 6.8), 10% glycerol, 2% SDS, and 100 mM DTT. Membranes were blotted with Abs recognizing Atg7 (ProSci), Atg5 (Proteintech Group), cytochrome C (BD Pharmingen), Tom20 (Santa Cruz Biotechnology), apoptosis-inducing factor (AIF; Cell Signaling Technology), mtHsp70 (Affinity BioReagents), Bcl-2 (BD Pharmingen), Bcl-x_L (BD Pharmingen), Mcl-1 (Rockland), Bak (Upstate), Bax (eBioscience), and actin (Santa Cruz Biotechnology). All blots except Bcl-2 were visualized with anti-rabbit Alexa 680 (eBioscience), anti-mouse Alexa680 (eBioscience), and anti-goat Alexa 800 (Rockland) secondary Abs and read using an infrared imaging system (LI-COR Bioscience). Bcl-2 was visualized with HRP-conjugated anti-Armenian hamster (Jackson ImmunoResearch Laboratories) and Pico chemiluminescent substrate (Pierce).

Single-cell suspensions of thymus, spleen, and LNs were lysed of RBC and incubated with FcR blocker (2.4G2; eBioscience). Cells were maintained in PBS with 2% FBS on ice throughout staining before analysis. Cells were stained with FITC, PE, PE-Cy5, allophycocyanin, and/or allophycocyanin-Cy7 anti-CD4, -CD8, -TCRβ, -CD44, -CD62L, -CD25, -B220, and -CD45.2 (eBioscience, BioLegend, and BD Pharmingen). Cell events were collected on a FACScan or FACStar, and data were analyzed using CellQuest (Becton Dickson) and FlowJo (Tree Star) software. Cell death was assayed using annexin V, 7-aminoactinomycin D, and propidium iodide staining (BD Pharmingen) or an activated caspase 9 detection kit (Immunochemistry Technologies). To stain mitochondria, lymphocytes were incubated for 30 min at 37°C with 100 nM MitoTracker Green (Molecular Probes) in RPMI 1640 complete medium before surface Ab staining. For intracellular stains, cells were fixed in 2% PFA and permeabilized with 0.2% saponin (Sigma-Aldrich) before staining with anti-Bcl-2 PE (BD Pharmingen) or anti-Mcl-1 (Rockland) followed by FITC anti-rabbit Abs (Jackson ImmunoResearch Laboratories).

Mature LN T cells after EasySep negative bead selection (StemCell Technologies) were Ab stained and lightly fixed in 1% PFA. One to two million TCRβ⁺CD8⁺ T cells were sorted and fixed in a 4% glutaraldehyde 0.1 M sodium cacodylate buffer overnight. The samples were rinsed in 0.1 M cacodylate buffer containing 7.5% sucrose three times for 15 min each and fixed in 1% osmium in cacodylate buffer for 1 h. After being washed three times in 0.11 M veronal acetate buffer for 15 min each, the samples were incubated with 0.5% uranyl acetate in veronal acetate buffer for 1 h at room temperature. Specimens were then dehydrated in an ascending series of ethanol (35%, 70%, 95%, and two changes of 100%) for 10 min each, followed by two changes of propylene oxide for 5 min each. The samples were incubated with a 1:1 mixture of 100% resin and propylene oxide for 1 h, followed by two changes of 100% resin, each for 30 min. Finally, the samples were embedded in resin and polymerized at 60°C overnight. Thick sections (0.5 μm) were cut and stained with toluidine blue for light microscopy selection of the appropriate area for ultrathin sections. Thin sections (60–90 nm) were cut, mounted on copper grids, and poststained with uranyl acetate and lead citrate. Micrographs were taken with a Philips LS 410 electron microscope. Images were analyzed using AxioVision software (Zeiss).

To assay for ROS production, lymphocytes were incubated in 5 μM dihydroethidium (Sigma-Aldrich) or CM-H₂DCFDA (Molecular Probes) for 1 h in RPMI 1640 complete medium at 37°C and subsequently analyzed by flow cytometry. ROS production was inhibited in vitro by the addition of 150 μM manganese (III) tetrakis 4-benzoic acid (MnTBAP; Calbiochem), 1 mM N-acetylcysteine (Sigma-Aldrich), 50 U/ml superoxide dismutase-polyethylene glycol (Sigma-Aldrich), or 50 U/ml catalase-polyethylene glycol (Sigma-Aldrich).

To determine mitochondrial DNA content in primary T cells, EasySep-enriched (StemCell Technologies) CD8⁺ cells were sorted and DNA was extracted using a DNeasy kit (Qiagen). Quantitative PCR was performed using a LightCycler FastStart DNA Master SYBR Green I kit on an iCycler iQ real-time PCR detection system (Roche). Products from two sets of mitochondrial DNA-specific primers (mtDNA F1, ACC ATT TGC AGA CGC CAT AA; mtDNA R1, TGA AAT TGT TTG GGC TAC GG (40); mtDNA F2, GCC CCA GAT ATA GCA TTC CC; mtDNA R2, GTT CAT CCT GTT CCT GCT CC (41); actin DNA F, TGT TCC CTT CCA CAG GGT GT; actin DNA R, TCC CAG TTG GTA ACA ATG CCA (41)) were normalized to a genomic actin cell loading control and analyzed using relative expression software tool (REST) v2 software (42, 43).

Previous work by our group (35) and others (34, 36, 37) has demonstrated that autophagosomes form in primary T lymphocytes. Additionally, we have also demonstrated that the autophagy gene Atg5 contributes critically to the homeostatic survival of mouse T lymphocytes in vivo (35). As this autophagy protein also interacts with FADD (Fas-associated death domain protein) and Bcl-x_L (44, 45), Atg5 may regulate T cell survival through its role in autophagosome formation or through interactions with extrinsic or intrinsic death pathways (46, 47). To determine whether the process of autophagy contributes to mature T cell survival, we examined the T cell compartment in a second autophagy-deficient genetic model system. We crossed Atg7^f/f mice (8) with mice expressing the T cell-specific Lck-Cre transgene (39). Consistent with the induction of Lck-Cre expression at the CD4⁻CD8⁻ double-negative (DN) stage of thymocyte development, we observed efficient genomic and protein deletion of Atg7 in double-positive (DP) and single-positive (SP) thymocytes as well as in mature CD4⁺ and CD8⁺ T lymphocytes in Atg7^f/fLck-Cre mice (Fig. 1,A). Additionally, we found impaired conjugation of Atg5 to Atg12 in both thymocytes and peripheral T cells lacking Atg7, an essential upstream enzyme for this autophagic pathway conjugation process (Fig. 1 B). These data demonstrate that Atg7 is efficiently deleted, and they are consistent with a functional deficiency in autophagy in our model system.

To determine the role of autophagy in T cell development and function, we examined thymocytes and mature T cells in Atg7^f/fLck-Cre and Atg7^f/f littermate controls. Although thymocyte development appeared grossly normal in Atg7^f/fLck-Cre mice (Fig. 1,C), pooled data from large numbers of mice showed a modest but statistically significant decrease in CD4⁺ and CD8⁺ SP thymocytes in these animals (Fig. 1,E). In the spleen and lymph nodes, there was a dramatic reduction in the percentage of T cells in Atg7^f/fLck-Cre mice (Fig. 1,C). Atg7^f/fLck-Cre mice also demonstrated an increase in the relative percentage of CD44^highCD62L^low peripheral T cells (Fig. 1,D). This phenotype is most consistent with lymphopenia, as markers of cell activation including CD69 and CD25 were not increased in these mice (Fig. 1,D). Ultimately, this resulted in a 75% decrease in total naive T cell numbers in peripheral Atg7-deficient T cells in secondary lymphoid tissues (Fig. 1,E). Finally, unlike Atg5^−/− fetal liver chimeras where total B lymphocyte numbers are decreased (35, 38), we observed no reduction in B cell numbers in our model system consistent with the selective elimination of Atg7 in T lineage cells (Fig. 1 E).

Although numerous factors including modestly reduced thymocyte numbers (Fig. 1,E) and alterations in the proliferative capacity of autophagy-deficient T cells (38) may contribute to the defect in Atg7-deficient peripheral T cell homeostasis, mature T cells lacking this essential autophagy gene undergo enhanced apoptotic death. Freshly isolated mature Atg7^f/fLck-Cre T cells displayed 3- to 4-fold enhanced rates of apoptosis ex vivo as measured by annexin V surface staining (Fig. 2,A). Consistent with enhanced apoptotic death, freshly isolated Atg7-deficient T cells also demonstrated enhanced caspase 9 activity by flow cytometry (Fig. 2 B). Taken together, these data demonstrate an important role for the essential autophagy gene Atg7 in maintaining peripheral T lymphocyte homeostasis. With previously published data from Atg5-deficient chimeric mice, our results support a role for autophagy in promoting mature T cell survival.

The above results suggest that autophagy may have cytoprotective functions in T lymphocytes. Given the central role of mitochondria in lymphocyte apoptotic cell death (48) and the emerging role for autophagy in maintaining proper homeostasis of this organelle (6), we investigated the mitochondrial compartment in autophagy-deficient T cells. We first examined the mitochondria in Atg7^f/f and Atg7^f/fLck-Cre mature T cells using the relatively potential-independent mitochondria-specific vital dye MitoTracker Green (49). MitoTracker Green is a lipophilic thiol-reactive dye that selectively labels mitochondrial inner membrane and matrix (50) and has been used for measuring mitochondrial content in hematopoietic cells (51, 52, 53). Analysis by flow cytometry revealed a 50–150% increase in the MitoTracker Green staining of Atg7-deficient CD4⁺ and CD8⁺ T lymphocytes in the spleen and lymph nodes (Fig. 3, A and B). As expected from this T cell-specific deletion model, we observed no difference in MitoTracker Green staining of B220⁺ B lymphocytes (Fig. 3 B).

Additionally, we confirmed this difference in the CD44^low naive T cell subset since Atg7^f/fLck-Cre mice have a relative increase in CD44^highCD62L^low T cells consistent with a homeostatic response to lymphopenia (Fig. 3, A and B). We also examined whether MitoTracker Green staining measured by flow cytometry was specific for mitochondria by immunofluorescent microscopy. In both Atg7^f/f and Atg7^f/fLck-Cre T cells, MitoTracker stained a networked cytoplasmic structure that also stained positive for the outer mitochondrial membrane marker Tom20 (Fig. 3,C). Finally, we observed an increase in staining by the mitochondrial potential- and volume-dependent dye TMRE in Atg7^f/fLck-Cre T cells when compared with Atg7^f/f controls, consistent with an expansion of the functional mitochondrial content in these cells (Fig. 3 D).

We next examined mitochondrial content in mature T cells lacking Atg5. In Atg5^−/− fetal liver chimeric mice, CD45.2⁺ donor-derived CD4⁺ and CD8⁺ T cells from the spleen and lymph nodes demonstrated dramatically enhanced MitoTracker Green staining (Fig. 3,E). No difference in MitoTracker Green staining in residual host T cells was observed in Atg5^+/+ and Atg5^−/− chimeras (Fig. 3,E). Interestingly, Atg5^−/− and control peripheral B lymphocytes had comparable MitoTracker Green staining, suggesting a T cell-specific role for autophagy in mitochondrial homeostasis within the lymphocyte compartment (Fig. 3 E).

To determine whether developing autophagy-deficient T cells also contain enhanced mitochondrial content, we analyzed MitoTracker Green fluorescence in Atg7^f/f and Atg7^f/fLck-Cre thymocytes. Although there was a significant increase in MitoTracker Green mean fluorescent intensity (MFI) in Atg7-deficient CD4⁺ and CD8⁺ SP thymocytes, the magnitude of this change (10–40%) was low when compared with mature peripheral Atg7-deficient CD4⁺ and CD8⁺ T cells (Fig. 4, A and B). Additionally, we did not observe an increase in MitoTracker Green staining in Atg5^−/− thymocytes when compared with Atg5^+/+ controls (Fig. 4 C). The reason for this discrepancy is unknown but may reflect subtle differences in the two model systems or the activity of Atg5 and Atg7 in T cell autophagy.

When MitoTracker staining values from multiple mice were normalized to DP cell mean fluorescence, we observed a consistent 50–75% reduction in the staining of wild-type T lymphocytes as they transitioned from the thymus to peripheral circulation (Fig. 4,D). This reduction was larger in CD8⁺ than CD4⁺ T cells, which may reflect the intermediate mitochondrial content phenotype in CD8⁺ SP thymocytes (Fig. 4,D). In contrast, mitochondrial content was not changed in developing B lymphocytes from CD43⁺B220⁺ B cell precursors to peripheral mature B cell populations (Fig. 4,E). When Atg7^f/fLck-Cre MitoTracker Green values were normalized to control DP fluorescence, it became clear that the relative MFI increase in autophagy-deficient peripheral T lymphocytes largely reflects an impaired reduction in mitochondrial content when compared with thymocytes (Fig. 4 D). Thus, autophagy-dependent regulated changes in mitochondrial content mark the important developmental transition from thymocyte to circulating peripheral T lymphocyte.

Given the dramatic changes in mitochondrial content, particularly in CD8⁺ Atg7^f/fLck-Cre peripheral T cells, we sorted autophagy-deficient and control CD8⁺ T cells to visually examined mitochondria by transmission electron microscopy. Although there was notable variation in mitochondrial morphology between individual cell cross-sections, the mitochondrial shape and appearance in Atg7^f/fLck-Cre T cells were largely comparable to controls (Fig. 5,A). The mitochondria in autophagy-deficient T cells exhibited clear outer membranes and cristae without specific evidence of dysfunction, including swelling or condensation (Fig. 5 A).

To determine whether differences observed in mitochondrial content from MitoTracker Green staining might reflect differences in mitochondrial volume in these cells, we calculated mitochondrial surface area in >40 cell cross-sections by manually outlining mitochondria using a quantification tool in AxioVision (Zeiss). Large variations between individual cross-sections were found in all samples analyzed (Fig. 5,B) and likely reflect the uneven distribution of mitochondria in three-dimensional space within the cytoplasm of T cells observed by immunofluorescent microscopy. Nevertheless, a statistically significant increase in mitochondrial surface in Atg7-deficient CD8⁺ T cells was observed when compared with control T cells (Fig. 5,B). To confirm this finding, we used a third measure of mitochondrial content in cells, mitochondrial DNA copy number (40). By quantitative PCR analysis, we found that sorted Atg7^f/fLck-Cre CD8⁺CD44^low T cells contained ∼2-fold more mitochondrial DNA than did control cells (Fig. 5,C). Finally, we performed Western blots for a range of mitochondrial proteins in Atg7^f/f and Atg7^f/fLck-Cre thymocytes and sorted T cells. These results demonstrated a large increase in the inner-membrane and intermembrane space proteins cytochrome C and AIF, with less dramatic changes in the outer mitochondrial transport protein Tom20 and inner matrix chaperone mtHsp70 (Fig. 5, D and E). Taken together, these results suggest that in the absence of autophagy increased numbers of grossly normal mitochondria are present in mature T cells and that autophagy is critical to clear mitochondria in these cells.

Mitochondria are responsible for numerous important metabolic pathways within eukaryotic cells. To determine whether increased mitochondrial content in autophagy-deficient T cells was associated with functional changes in these cells, we examined ROS production in cultured T lymphocytes. We observed a 2-fold increase in ROS production by Atg7-deficient T cells as measured by dihydroethidium and CM-H₂DCFDA staining (Fig. 6 A). ROS production has been shown to play a proapoptotic role in T cell hybridomas as well as in activated primary T cells (54, 55).

In addition to producing ROS, mitochondria can become toxic to cells through their role in the intrinsic death pathway. Death initiated through the mitochondria is largely regulated by Bcl-2 family proteins and classically results in the release of proapoptotic factors normally sequestered away from downstream effectors in the mitochondrial intermembrane space (48). To determine whether abnormalities in Bcl-2 family protein expression are associated with enhanced apoptosis in autophagy-deficient mature T lymphocytes, we first examined the expression of the two major antiapoptotic Bcl-2 family members expressed in naive T cells, Bcl-2 and Mcl-1 (56, 57). The expression of Bcl-2 as assessed by both flow cytometry and Western blot was increased in Atg7^f/fLck-Cre T cells (Fig. 6, B and C). Conversely, we found no significant increase in Mcl-1 expression in Atg7^f/fLck-Cre cells as measured by both flow cytometry and Western blot (Fig. 6, D and E). We observed low levels of Bcl-x_L expression as assessed by Western blot in both Atg7-deficient and control T cells, consistent with the low expression of this antiapoptotic protein in naive T lymphocytes (unpublished data). Thus, T cells lacking the essential autophagy gene Atg7 demonstrate a selective up-regulation of Bcl-2, but not other antiapoptotic family members.

We next examined the proapoptotic members of the Bcl-2 family, Bax and Bak. While Bax protein levels remained unchanged in autophagy-deficient T lymphocytes, Bak protein was increased ∼2-fold by Western blot analysis (Fig. 6,F). Additionally, consistent with the increase in mitochondrial content in our autophagy-deficient T cells, we found a significant increase in death-inducing mitochondrial proteins cytochrome c and AIF expression in Atg7^f/fLck-Cre T cells when compared with Atg7^f/f controls (Fig. 5, D and E). Given the central role of the BH3 only protein Bim in the induction of naive and activated T cell death (58), we also examined Bim expression but observed no difference between autophagy-deficient and control T cells (unpublished data). Taken together, these data suggest that perturbations in the balance of pro- and antiapoptotic proteins may contribute to cell death in primary T cells in the absence of autophagy.

Although autophagy has long been recognized, its functions in various physiological and pathological processes have only begun to be elucidated. Our studies have revealed several key findings regarding the role of autophagy in regulating T lymphocyte survival. First, our data strongly suggest that autophagy itself is essential for mature T cell survival. Although our previous results show that Atg5-deficient mature T cells undergo massive apoptosis (35), it was not clear whether the defective survival of Atg5-deficient T cells was caused by the role of Atg5 in the induction of autophagy or other survival pathways, as Atg5 interacts with FADD and Bcl-x_L (44, 45). The fact that T lymphocytes lacking a second essential autophagy gene, Atg7, exhibit a similar survival defect to that of Atg5-deficient T cells strongly argues for a role for autophagy in maintaining mature T cell survival. Second, we show that mitochondrial content in T lymphocytes is developmentally regulated. A reduction of mitochondrial content marks the transition from thymocytes to peripheral mature T cells. Third, the clearance of superfluous mitochondria in mature T cells critically depends on autophagy. This is supported by our measurement of mitochondrial content through FACS analysis of MitoTracker Green-stained mitochondria, direct survey of mitochondrial surface area under transmission electron microscopy, quantitative determination of mitochondrial DNA, as well as the expression levels of various proteins associated with mitochondria. Interestingly, this phenomenon appears to be limited to T lymphocytes, as mitochondrial content is not developmentally regulated in B lymphocytes. Fourth, in the absence of autophagy, we have observed not only enhanced mitochondrial content but also abnormality in mitochondria-associated functions, including ROS production and apoptosis.

The precise role of autophagy in cell survival and cell death remains controversial, and this controversy likely reflects the complex functional and molecular intersection of autophagy with pro-survival and pro-death pathways (46, 59, 60). In cells where apoptosis is inhibited, the degradation of cytoplasmic material in autophagosomes protects cells from death by providing essential metabolic support during periods of nutrient deprivation (9); however, the autophagic degradation of essential proteins, including catalase (61, 62), and organelles (20) can also commit theses cell to death. Although the outcome of autophagosome formation in cell survival may depend on cell type and context, autophagy clearly serves a critical role in the homeostatic survival of cells in vivo. This has been most clearly demonstrated in the CNS, where the conditional deletion of the essential autophagy genes Atg5 or Atg7 leads to neurodegeneration associated with an increase in TUNEL-positive apoptotic cells (10, 11). Our results here, together with our previous data (35), demonstrate that deletion of Atg5 or Atg7 in primary mouse T lymphocytes leads to a dramatic impairment of naive T cell peripheral survival with up to a 75% reduction in the number of circulating T cells. Thus autophagy constitutes a novel pro-survival pathway in mature T lymphocytes.

Our finding that mitochondrial content is developmentally regulated in T lymphocytes and the reduction of mitochondrial content in peripheral T cells critically depends on autophagy has identified a novel mechanism by which mature T lymphocyte homeostasis is maintained. Thymocytes from DN, DP, and CD4⁺ SP compartments contain high content of mitochondria. In contrast, mitochondrial content in CD8⁺ SP thymocytes is reduced to an intermediate level. Furthermore, although mitochondrial content in CD4⁺ mature T cells is reduced by ∼50% when compared with DP thymocytes, this reduction is ∼80% for CD8⁺ mature T cells. This more dramatic reduction of mitochondrial content in developing CD8⁺ T cells may reflect their lower tolerance for superfluous mitochondria than in CD4⁺ mature T lymphocytes. Consistent with this notion, CD8⁺ T cells lacking Atg5 (35) or Atg7 exhibit higher apoptosis rates and more dramatic reductions in cell numbers than do autophagy-deficient CD4⁺ T cells. Interestingly, although mitochondrial content is increased, morphology appears normal under transmission electron microscopy. This is in contrast to the deformed mitochondria in Atg7-deficient hepatocytes (8). This difference may suggest a role of autophagy in removing superfluous mitochondria in T cells and damaged mitochondria in liver cells. Alternatively, low levels of mitochondrial damage may result in the rapid apoptosis of T lymphocytes, making abnormal mitochondria more difficult to detect among live autophagy-deficient T cells.

In addition to T lymphocytes, two other cell types appear to clear mitochondria and other intracellular organelles during development. It has been observed that intracellular organelles in epithelial cells during lens development and in erythrocytes during erythroid cell maturation are rapidly eliminated, and autophagy has been proposed as a mechanism for the degradation of these organelles (14, 21, 22, 24). However, recent data using Atg5-deficient mice have demonstrated that organelle degradation during lens and erythroid differentiation is independent of this essential autophagy gene (23). Therefore, T lymphocytes provide an important example of cells that undergo autophagy-dependent organelle clearance in vivo.

Why does mitochondrial content need to be down-regulated as T lymphocytes undergo maturation from thymocytes to mature T cells? As the transition from thymocytes to mature T cells marks a dramatic change in their environment, factors in the periphery (blood and secondary lymphoid organs) may cause stress to mature T cells if mitochondrial content remains static. One such likely environmental factor is the oxygen level in the periphery. It has been shown that the oxygen tension in blood (5–13 kPa) (63) is dramatically higher than in the thymus (1.3 kPa) (64). High mitochondrial content in circulating mature T cells may lead to the production of excess amount of ROS, a byproduct of mitochondrial respiration. As ROS is toxic to T lymphocytes (54, 55), clearance of superfluous mitochondria by autophagy in mature T cells may be necessary to avoid ROS toxicity. Consistent with this notion, we observed higher ROS production in autophagy-deficient T cells. This idea is further supported by recent evidence showing that ROS induces autophagy through Atg4 (65). Thus, peripheral mature T cells may use a physiologically relevant stress, high oxygen tension and ROS, to induce autophagy to remove extra mitochondria.

The apoptosis of autophagy-deficient T cells is likely caused by multiple factors. While enhanced ROS may cause T cell death, the enhanced expression of proapoptotic Bak may further contribute to apoptosis of autophagy-deficient T cells. Although Bcl-2 expression is increased in autophagy-deficient T cells, this level of Bcl-2 may not be sufficient to prevent Bak-dependent apoptosis. A previous report has shown that Bak can be effectively sequestered by Mcl-1 and Bcl-x_L but not by Bcl-2 (66). Although Mcl-1 is also associated with mitochondrial membrane, its expression level in autophagy-deficient T cells is not obviously changed. The reason for this observation is not clear, but may be related to the extremely short half-life of Mcl-1 in cells as well as the tight regulation of this protein’s stability by upstream signaling pathways within the cell (67, 68). Additionally, the high expression level of cytochrome c in autophagy-deficient T cells may further prime these cells to caspase activation and cell death. Thus, it is likely that autophagy-deficient T cells die due to the activation of caspases by multiple factors derived from the superfluous mitochondria. Interestingly, increased mitochondrial content in HIV-specific CD8⁺ T cells in human patients also correlates with enhanced susceptibility of these cells to apoptosis, suggesting that mitochondrial homeostasis may play an important role in T cell survival in normal physiology as well as in pathologic situations (51).

We thank Dr. Jeffrey Rathmell for useful discussion and critical review of this manuscript, Dr. Tso-Pang Yao for discussion and sharing of reagents, Dr. Sara Miller and Philip Christopher for help in transmission electron microscopy, and Michael Forrester for ROS assays.

The authors have no financial conflicts of interest.