Impaired Autophagy, Defective T Cell Homeostasis, and a Wasting Syndrome in Mice with a T Cell–Specific Deletion of Vps34

Autophagy is a “self-eating” catabolic process used to break down and recycle long-lived proteins and organelles to maintain a homeostatic environment within the cell (1). This process is usually functional at a low level, but it is upregulated in response to nutrient starvation, stress, or cellular damage (2, 3). Autophagy is initiated by the formation of a cup-shaped membrane structure that extends around the cellular organelles, forming double membrane vesicles called autophagosomes (4–6). Autophagosomes fuse with lysosomes and late endosomes for degradation of the inner autophagosomal membrane and their cargo (4).

Genetic analyses in yeast have identified a large number of evolutionary conserved genes, termed autophagy-related (Atg) genes, that are required for autophagy (7–9). The class III PI3K vacuolar protein sorting (Vps)34 (also called PIK3C3) and its binding partner Atg6 (also called Beclin 1) have been reported to play important roles for the initiation of autophagy, including formation of the cup-shaped “omegasome” or isolation membrane (10, 11). Elongation of the isolation membrane is regulated by two ubiquitin-like systems, the Atg12 and the Atg8 (or its mammalian homolog LC3) conjugation pathways that are required for the generation of LC3-bound phosphatidylethanolamine as a building block to form double-membraned autophagosomes (12, 13). The class III PI3K Vps34 converts phosphatidylinositol (PtdIns) to PtdIns-3-phosphate (PtdIns3P), which recruits FYVE domain–containing proteins and members of the WD-repeat domain PtdIns-interacting family of proteins to the site of autophagosome formation. This, in turn, provides a scaffold for Atg proteins to initiate autophagy (11, 14). In yeast, Vps34 activity is critical for the recruitment of Atg proteins to the preautophagosomal structure and for autophagy initiation (15). However, in higher eukaryotes, the role of Vps34 and its product PtdIns3P in autophagy is less well understood. Recent studies have described the deletion of Vps34 in embryonic fibroblasts, heart, liver (16), sensory neurons (17), or T cells (18) in mice. Whereas the study with embryonic fibroblasts, heart, and liver argued for a critical role of Vps34 in regulating functional autophagy (16), the study in sensory neurons favored a predominant role of Vps34 in endocytosis but not autophagy (17), and the study in T cells concluded that Vps34 is critical for IL-7Rα–chain expression (18).

In this study, we have generated mice with a T cell–specific deletion in Vps34 to study the role of Vps34 in T cell homeostasis and function. We found that deletion of Vps34 in T cells results in severe defects in autophagic flux and accumulation of cellular organelles. This phenotype correlated with enhanced apoptosis in Vps34-deficient T cells. Mice with T cell–specific deletion of Vps34 exhibited normal intrathymic development of conventional CD4⁺ and CD8⁺ T cells but impaired intrathymic development of invariant NKT (iNKT) cells. In peripheral organs, T cell–specific ablation of Vps34 resulted in a profound loss of T cells. Furthermore, we found that Vps34 was required for the peripheral homeostasis and function of CD4⁺Foxp3⁺ regulatory T (Treg) cells. Consequently, aged animals developed an inflammatory wasting syndrome characterized by weight loss, intestinal inflammation, and anemia.

Vps34^flox/flox (Vps34^f/f) mice have been described (16). T cell–specific deletion of Vps34 was achieved by crossing the Vps34^f/f mice with CD4-Cre transgenic mice (Taconic). Mice were genotyped as described previously (16). Six- to 8-wk-old animals were used in this study. An inflammatory wasting phenotype in 18- to 25-wk-old Vps34^f/f;CD4-Cre mice was observed. These mice consistently developed a rectal prolapse and were carefully monitored. Mice were treated with an antibiotic ointment three times per week. Rag2^−/− mice were obtained from The Jackson Laboratory. All breeder and experimental mice were housed in specific pathogen-free conditions in compliance with guidelines from the Institutional Animal Care and Use Committee at Vanderbilt University.

T cells were purified from Vps34^f/+;CD4-Cre or Vps34^f/f;CD4-Cre splenocytes by negative selection using magnetic sorting (Miltenyi Biotec). Cell purity was routinely >96% as determined by flow cytometry. T cells were activated with plate-bound anti-CD3ε and anti-CD28 Abs in complete RPMI 1640 medium for 36 h and were used for electron microscopy, Western blot analyses, and measurements of glycolytic rate and β-oxidation.

Activated T cells were fixed with buffer (50 mM cacodylate [pH 7.2], 50 mM KCl, and 2.5 mM MgCl₂) containing 2.5% glutaraldehyde at 4°C overnight. For pre-embed immunolabeling, 2% paraformaldehyde-fixed cells were stained with anti-LC3B Abs, followed by goat anti-rabbit IgG conjugated to 10-nm gold particles (BBInternational). The cells were washed and fixed as above (19). The cells were washed three times (5 min) in 0.1 M cacodylate buffer followed by incubation for 1 h in 1% osmium tetraoxide at room temperature. The cells were then washed three times with 0.1 M cacodylate buffer and were dehydrated through a series of increasing ethanol washes followed by propylene oxide incubations then infiltration with Poly/Bed 812 (Polysciences) resin and finally embedded in Poly/Bed 812 resin. Sections (70 nm thick) were collected on copper grids and stained with uranyl acetate and lead citrate and subsequently analyzed with a Philips/FEI Tecnai T-12 or Philips CM-12 electron microscope equipped with a side-mounted 2k × 2k AMT CCD camera. Thirty to 35 individual cells in each group were imaged to measure the size of mitochondria. The size of mitochondria was determined using the ImageJ program (National Institutes of Health) by manually marking single mitochondria, measuring their area, and normalizing them against the scale bar as previously described (20). The data presented are the total size of all the mitochondria normalized against the size of the cell. A similar analysis was carried out with the nuclear size as a negative control.

Purified T cells were washed with PBS and adhered to polylysin-coated coverslips for 5 min and fixed with 2% paraformaldehyde. Cells were then treated with 0.1% saponin (Sigma-Aldrich) in PBS containing 0.5% BSA for 10 min on ice and intracellularly stained with an anti-LC3B Ab (Enzo) overnight in a humified chamber at 4°C. After washing, anti-rabbit Cy3 Ab (Jackson ImmunoResearch Laboratories) was added and incubated on ice for 30 min. Z-stack images were acquired at 1 μm spacing using a ×60 oil objective by a Zeiss LSM 510 Meta confocal microscope. Three-dimensional deconvolution of the images was performed and the LC3 punctae were quantified by MetaMorph 7.6 software (Universal Imaging) as described previously (21). The threshold was set to >4 pixels for LC3 puncta count, and a total of 30–35 cells from each group were analyzed.

Activated T cells or CD4⁺ thymocytes purified by magnetic sorting were washed three times with PBS before preparation of cellular proteins for Western blotting. Some T cells were treated with 3-methyladenine (3-MA) (5 mM; Sigma-Aldrich) for the last 5 h before cell lysate preparation for Western blot analysis. Cells were lysed with lysis buffer (Cell Signaling Technology) containing protease inhibitor mixture (Sigma-Aldrich) for 2 h at 4°C. The protein samples were separated by 12% SDS-PAGE (Bio-Rad Laboratories) and transferred to nitrocellulose membranes overnight. The membranes were blocked with 5% nonfat milk and incubated with primary Ab overnight at 4°C, followed by HRP-conjugated secondary Ab for 2 h at room temperature. The blots were developed with the ECL method (GE Healthcare). The protein bands were quantified by densitometric analysis using ImageJ software, and the presented data are normalized results against respective β-actin controls. LC3 Ab (5F10) was obtained from Enzo Life Sciences, p62 Ab was from Cell Signaling Technology, Bcl-2–interacting mediator of cell death (Bim) and Bcl-2–associated X protein (Bax) Abs were from eBioscience, and β-actin, Atg5, Atg7, Bcl-2–associated death protein (Bad), and Vps34 Abs were from Sigma-Aldrich. All secondary Abs were obtained from Promega.

Autophagic flux assay was performed by culture of CD4⁺ thymocytes or activated T cells in the presence of E64D (10 μg/ml; Sigma-Aldrich) plus pepstatin A (10 μg/ml; Sigma-Aldrich) for 90 min. After three washes with PBS, cell lysates were prepared for immunoblot analysis with anti-LC3B Ab.

Single-cell suspensions of the spleen, thymus, lymph nodes, and liver were prepared and stained with fluorescently labeled mAbs in FACS buffer (PBS containing 2% FBS and 0.05% sodium azide) as described previously (22). In all experiments, dead cells were excluded from the analysis by electronic gating. Fluorescently labeled Abs against mouse TCRβ, CD11b, CD11c, TCRγδ, CD4, CD8, Ki67, IL-2, IFN-γ, IL-13, IL-17A, CD24, CD25, Bcl-2, and isotype control Abs were obtained from BD Biosciences. Anti-human Bcl-x_L Ab and its isotype control were obtained from SouthernBiotech, and Foxp3 staining sets were obtained from eBioscience. Apoptosis was measured using an annexin V apoptosis detection kit from BD Biosciences. CD1d/α-galactosylceramide tetramers (CD1d-tetramers) were obtained from the National Institutes of Health Tetramer Core Facility (Emory University, Atlanta, GA). The iNKT cell population was identified as B220⁻TCR-β⁺CD1d-tetramer⁺ cells as described (22). For analysis of intracellular cytokines produced by CD4 and CD8 T cells, 18- to 25-wk-old mice with rectal prolapse were sacrificed and splenocytes and mesenteric lymph node cells were activated with or without ionomycin (1 μM) and PMA (100 ng/ml) in the presence of GolgiPlug (all from BD Biosciences-Pharmingen) for 5 h. Intracellular cytokine staining was performed with Cytofix/Cytoperm reagents (BD Biosciences Pharmingen) according to the manufacturer’s protocol. For measurement of endoplasmic reticulum and mitochondrial content in T cells by flow cytometry, splenocytes and lymph node cells were prepared and stained with MitoTracker Red (50 nM) or ER-Tracker Red (1 μM; Invitrogen) at 37°C for 30 min in complete medium. The cells were then washed and stained with anti-CD4 and anti-CD8 Abs at 4°C in FACS buffer. Flow cytometric analyses were performed using a FACSCalibur instrument (BD Biosciences), and studies with ER-Tracker Red were analyzed on an LSRFortessa instrument (BD Biosciences). The acquired data were analyzed using FlowJo software (Tree Star).

Total RNA was purified using TRIzol reagent (Invitrogen) according to the manufacturer’s instructions, including a DNase I digestion step. cDNA was prepared from 1–2 μg RNA using the ThermoScript reverse transcriptase system (Invitrogen). Real-time PCR was performed using SYBR Green Master mix (Bio-Rad) and 20 ng cDNA per well in duplicates for each sample on a CFX96 real-time PCR machine (Bio-Rad). The C_T values were collected for the housekeeping gene β-actin and the genes of interest during the log phase of the cycle. The level of the gene of interest was normalized to that of β-actin for each sample and compared with the values obtained for the test sample. Each gene was compared with every normalizer in succession and the ΔC_T was calculated (ΔC_T= C_{T Gene of interest} − C_{T Normalizer}). The normalized expression (ΔΔC_T) of the gene of interest was calculated using the CFX manager software (Bio-Rad). The primers used for PCR were as follows: Vps34, forward, 5′-CAGCCCTGGATGAGATGTTT-3′, reverse, 5′-GGATGGGTGACAGAACCAAG-3′; β-actin, forward, 5′-TACAGCTTCACCACCACAGC-3′, reverse, 5′-AAGGAAGGCTGGAAAAGAGC-3′.

Treg cells were purified on the basis of CD25 staining from negatively selected CD4⁺ T cells by magnetic sorting according to the manufacturer’s protocol (Miltenyi Biotec). CD4⁺CD25⁻ T effector (Teff) cells (2 × 10⁵) were activated with anti-CD3ε Ab (1 μg/ml) and 2 × 10⁴ dendritic cells (DCs) in the presence of Treg cells purified from splenic CD4⁺ T cells derived from Vps34^f/f;CD4-Cre or Vps34^f/+;CD4-Cre mice at various Teff /Treg cell ratios for 60 h with the addition of 1 μCi [³H]thymidine (PerkinElmer Life Sciences) for the last 12 h culture.

CD4⁺ T cells were enriched from pooled splenic and mesenteric lymph node cells by magnetic sorting using negative selection, followed by enrichment of CD25⁻ T cells and CD25⁺ Treg cells by positive selection. CD4⁺CD25⁻ cells (2 × 10⁶ cells) derived from Vps34^f/f;CD4-Cre or Vps34^f/+;CD4-Cre mice were adoptively transferred into Rag2^−/− mice to induce colitis as described (23). Groups of recipient mice also received 5 × 10⁵ CD4⁺CD25⁺ Treg cells derived from Vps34^f/f;CD4-Cre or Vps34^f/+;CD4-Cre mice. The mice were followed for signs of colitis twice a week for up to 12 wk. Animals with severe clinical signs of disease were sacrificed according to the guidelines from the Institutional Animal Care and Use Committee at Vanderbilt University. Mice were monitored for weight change and signs of colitis, including diarrhea, rectal bleeding, and scruffiness, each given a clinical score of 1. Histological sections of the colons were graded for signs of lymphocyte infiltration, loss of globlet cells, and ulcers as described (24). Each of these histological parameters was given a score ranging from 0 to 3 according to severity. The clinical and the histological scores were pooled to obtain the final disease score.

The glycolytic rate was measured as described (25). In brief, 1 million activated cells were suspended in 0.5 ml RPMI 1640 that was previously equilibrated at 37°C under 5% CO₂. Ten microcuries 5-[³H]glucose was added to each well and samples were incubated for 1 h at 37°C in a humidified chamber. The reaction was terminated by mixing 50 μl cells with an equal volume of 0.2 N HCl in a PCR vial and samples were then placed upright in 4-ml scintillation vials containing 0.5 ml water. The vials were sealed with parafilm and [³H]₂O generated during glycolysis in the PCR tubes was allowed to equilibrate with the water in the scintillation vials for 2 d. The contents of the PCR tubes were transferred into a new scintillation vial and scintillation fluid was added to the original vials for diffused counts and to the new vials for undiffused counts. The fraction of [³H]₂O that diffused in 2 d was determined with control PCR tubes containing 1 μCi [³H]₂O. The background diffusion rate was determined using tubes containing cell-free medium. The glycolytic rate was calculated as sample diffusion ratio minus the background diffusion ratio divided by the diffusion fraction from the [³H]₂O control. This number was multiplied by 5500 (the nanomoles of glucose in 0.5 ml RPMI 1640) to obtain the glycolytic rate as nanomoles of glucose consumed per million cells per hour.

β-Oxidation of lipids was measured in activated T cells as described (26, 27). In brief, T cells were activated as described above, washed twice with PBS, and cultured in 0.4 ml RPMI 1640 containing 2% fatty acid–free BSA (Sigma-Aldrich) for half an hour. After two washes with PBS, 4 × 10⁶ cells were cultured in 0.4 ml RPMI 1640 containing 2 μCi 9,10-[³H]palmitate (MP Biomedicals), 2% BSA (fatty acid–free), and 0.25 mM l-carnitine for 4 h under 5% CO₂ in an incubator. The oxidation of 9,10-[³H]palmitate was measured by release of [³H]₂O in the supernatant. The excess of radioactive lipid from the supernatant was removed twice by precipitation of lipids with an equal volume of 10% TCA followed by incubation at room temperature for 15 min and centrifugation at 13,000 × g for 10 min. The resultant supernatant was treated with 750 μl chloroform/methanol mixture (2:1) and 300 μl 2 M KCl/HCl and then centrifuged at 3000 × g for 5 min. Supernatant (500 μl) was collected and counted after the addition of 5 ml scintillation fluid (Ecolite; MP Biomedicals). The rate of β-oxidation was determined by subtracting cpm of cell-free supernatant (background) from [³H]₂O released in the sample supernatant and was expressed as cpm/4 × 10⁶ cells.

Vps34^f/f;CD4-Cre or Vps34^f/+;CD4-Cre mice were immunized with 200 μg OVA emulsified in CFA at day 0, and were boosted with the same amount of Ag emulsified in IFA at day 7. The mice were sacrificed at day 15 and splenic and draining lymph node cells were stimulated in vitro with 25–100 μg/ml OVA. The supernatants were collected at 72 h culture and assayed for IL-17A and IFN-γ by ELISA.

CFSE dilution analysis was performed as described in our previous study (22). In brief, splenocytes derived from Vps34^f/f;CD4-Cre or Vps34^f/+;CD4-Cre mice were labeled with 5 mM CFSE (Invitrogen) for 10 min at 37°C in PBS containing 0.1% BSA and washed twice with complete RPMI 1640 medium. Labeled cells (3 × 10⁵ cells/well) were then stimulated with anti-CD3ε Ab (1 μg/ml) with or without the apoptosis inhibitor z-VAD-fmk (50 μM; Sigma) for 65 h in complete RPMI 1640 medium. At the end of the culture, cells were harvested, stained with anti-CD4 and anti-CD8 Abs, and analyzed by flow cytometry.

DN32.D3 iNKT cell hybridoma cells (28) (obtained from Dr. A. Bendelac, University of Chicago, Chicago, IL) were used as responder cells in cultures with thymocytes as described (29). Briefly, 3 × 10⁴ DN32.D3 cells were cultured with 5 × 10⁵ or 10⁶ total thymocytes from Vps34^f/f;CD4-Cre or Vps34^f/+;CD4-Cre mice per well for 20 h. Supernatants were collected and IL-2 was measured by ELISA.

A standard sandwich ELISA was performed to measure mouse IL-2, IL-17A, and IFN-γ in the culture supernatants using a BD OptEIA kit (BD Biosciences). For detection, streptavidin-HRP conjugate (Zymed Laboratories) was used, and the color was developed with the substrate 3,3′,5,5′-tetramethylbenzidine (Dako) in the presence of H₂O₂. Total IgG Abs in the serum were measured by anti-mouse IgG (H+L) capture Ab and biotin-labeled anti-mouse IgG reagents (SouthernBiotech) according to the manufacturer’s instructions.

Complete small and large intestinal rolls were prepared and fixed in formalin. Sections (8 μm thick) from paraffin-embedded tissues were stained with H&E. Total blood counts were carried out in an automated analyzer at the Translational Pathology Shared Resource Core Facility at Vanderbilt Medical Center.

Statistical significance between the groups was determined by application of an unpaired two-tailed Student t test using Graphpad Prism software. A p value <0.05 was considered significant.

Vps34^f/f mice (16) were crossed with transgenic mice expressing Cre recombinase driven by the CD4 promoter, resulting in T cell–specific deletion of the Vps34 gene starting at the CD4⁺CD8⁺ (double-positive, DP) thymocyte stage (Supplemental Fig. 1A). Such deletion resulted in complete loss of Vps34 mRNA and protein expression in thymocytes and splenic CD4⁺ and CD8⁺ T cells (Supplemental Fig. 1B, 1C).

We assessed the steady-state levels of LC3 in Vps34-deficient T cells. In purified CD4⁺ thymocytes, we observed increased levels of both LC3B-I and LC3B-II in mutant T cells (Fig. 1A, 1B). In purified splenic T cells activated with plate-bound anti-CD3 and anti-CD28 Abs for 36 h, we detected little LC3B in wild-type cells, but increased levels of both forms of LC3B were observed in Vps34-deficient T cells (Fig. 1A, 1B).

p62, a protein that binds polyubiquitinated proteins and is degraded in lysosomes during autophagy (30), serves as an important indicator of defective autophagy (31). Although no significant change in p62 levels was observed in CD4⁺ thymocytes, profound accumulation of p62 was observed in activated Vps34-deficient T cells as compared with wild-type T cells (Fig. 1A, 1B), suggesting a defect in autophagic protein degradation.

Abnormally high levels of LC3B and accumulation of p62 in Vps34-deficient T cells suggested that these cells have a defect in autophagic flux. To test this possibility, we performed autophagic flux assays by using the lysosomal protease inhibitor E64D. In wild-type thymocytes or activated splenic T cells, such treatment increased the protein levels of LC3B-II as compared with cells treated with medium alone, whereas in Vps34-deficient cells, high levels of LC3B remained unchanged with E64D treatment (Fig. 1C). These results indicated a failure of autophagic protein turnover in Vps34-deficient T cells. Consistent with decreased LC3B turnover in Vps34-deficient T cells, an increased amount of large-sized LC3-positive structures was observed in these cells as compared with wild-type T cells (Fig. 1D, 1E). Immunostaining of LC3B on electron micrographs of activated T cells derived from Vps34^f/+;CD4-Cre mice revealed that LC3B gold staining was predominantly localized to autophagic membrane structures and partly to the cytoplasm. However, in activated T cells from Vps34^f/f;CD4-Cre mice, LC3B staining failed to associate with membranes; instead, increased staining was observed in the cytoplasm (Fig. 1G). Because Vps34 and autophagy are inhibited by 3-MA, we next treated activated T cells derived from Vps34^f/f;CD4-Cre or Vps34^f/+;CD4-Cre mice with 3-MA for 5 h and observed the accumulation of LC3B by Western blot analysis. The results showed that 3-MA treatment caused accumulation of LC3B in both Vps34-sufficient and -deficient T cells, suggesting that Vps34 is not the sole target of 3-MA (Fig. 1H). Collectively, these results indicated that functional autophagy is severely blocked in Vps34-deficient T cells.

Analysis of TCRαβ⁺ T cells in the peripheral lymphoid organs revealed that the percentage of both CD4⁺ and CD8⁺ T cells was significantly reduced in the spleen, lymph nodes, and peripheral blood of Vps34^f/f;CD4-Cre mice as compared with Vps34^f/+;CD4-Cre controls (Fig. 2A, 2B). In the liver, the percentage of CD8⁺ T cells in Vps34^f/f;CD4-Cre mice was slightly increased, whereas the percentage of CD4⁺ T cells was decreased. A similar trend was observed when absolute numbers of T cells were calculated based on total organ lymphoid cellularity (Fig. 2B, Supplemental Fig. 2A).

To understand the mechanism for the loss of peripheral T cells in Vps34-deficient organs, we analyzed thymic T cell development. We found that the percentages of CD4 single-positive (SP), CD8 SP, and CD4 CD8 DP cells were comparable between Vps34^f/f;CD4-Cre and Vps34^f/+;CD4-Cre mice (Fig. 2C, 2D). We found that the thymic cellularity in Vps34^f/f;CD4-Cre mice was slightly higher as compared with Vps34^f/+;CD4-Cre mice (Supplemental Fig. 2A). However, differences in the absolute numbers of CD4 and CD8 SP cells did not reach statistical significance (Fig. 2D). Analysis of double-negative (DN) cells, based on expression of CD44 and CD25 during various stages of thymocyte development, revealed that Vps34^f/f;CD4-Cre mice had no defect in transition from the DN1 to DN4 stages as compared with Vps34^f/+;CD4-Cre mice (Fig. 2C, 2D), which is consistent with initiation of CD4 transgene expression at the DP thymocyte stage (32). We further analyzed the mature T cell subset that was positively selected, based on the acquisition of TCRβ expression and loss of CD24 expression (33). Surprisingly, we found that Vps34^f/f;CD4-Cre mice had a higher percentage of TCRβ⁺CD24⁻ cells among the CD4 and CD8 SP fractions as compared with the controls (Fig. 2D). The DP population was used as a negative control, as they hardly displayed any positively selected cells (Fig. 2C). Such an increase in positively selected cells in the thymus may represent a compensatory mechanism for peripheral T cell loss. These results further supported the notion that thymic development of conventional T cells in T cell–specific Vps34-deficient mice was largely unperturbed.

iNKT cells are a small population of T cells that recognize glycolipid Ags bound with the MHC class I–related protein CD1d (34–37). iNKT cell development in the thymus requires expression of the Vα14-Jα18 invariant TCR and its recognition of endogenous glycolipid ligands presented by CD1d expressed on thymocytes (28, 29, 38). The positively selected iNKT cells then arise from the DP stage to mostly become CD4⁺ iNKT cells and a smaller population of CD4⁻CD8⁻ iNKT cells (39, 40). iNKT cell development progresses through various stages, based on their early expression of CD24 and CD44 and late expression of NK1.1 (41–43). After their development, these cells populate peripheral lymphoid organs, mainly liver and spleen. We found that Vps34^f/f;CD4-Cre mice exhibited a profound defect in intrathymic iNKT cell development, as hardly any TCRβ⁺CD1d-tetramer⁺ cells were observed as compared with the heterozygote controls (Fig. 3A, 3B). Consistent with these findings, the iNKT cell population was lacking in the spleen, liver, and lymph nodes (Fig. 3A, 3B). We next assessed the precise stage of iNKT cell developmental blockade. The developmental stages of iNKT cells are defined as CD1d-tetramer⁺ cells that are CD24⁺CD44⁻NK1.1⁻ (stage 0), CD24⁻CD44⁻NK1.1⁻ (stage 1), CD24⁻CD44⁺NK1.1⁻ (stage 2), and CD24⁻CD44⁺NK1.1⁺ (stage 3). We found that most iNKT cells in the thymus of Vps34^f/f;CD4-Cre mice were CD44⁻CD24⁺, with few cells of the other stages (Fig. 3C), suggesting that iNKT cells in these mice exhibited a developmental blockade at stage 0. Because loss of CD24 expression on thymocytes is a marker for positive selection, we considered the possibility that iNKT cells in these animals had a defect in positive selection due to defective expression of endogenous glycolipid ligands that mediate this process. We therefore measured activation of DN32.D3 iNKT cell hybridoma responder cells, which react with endogenous glycolipids displayed by CD1d (28), by thymocytes derived from Vps34^f/f;CD4-Cre or Vps34^f/+;CD4-Cre mice (29, 38). The results showed comparable DN32.D3 cell activation (data not shown), suggesting that Vps34-deficient thymocytes express normal levels of endogenous iNKT cell glycolipid ligands. Consistent with these findings, surface expression of CD1d on thymocytes derived from both groups of animals was comparable (data not shown). These results suggested that the blockade in iNKT cell development at stage 0 in Vps34^f/f;CD4-Cre mice was most likely due to a cell-intrinsic defect rather than a defect in the expression of endogenous glycolipid ligands on the selecting thymocytes.

Consistent with previous findings demonstrating that iNKT cells and NK cells share common requirements for homeostatic cytokines such as IL-15 and IL-7 (44), Vps34^f/f;CD4-Cre mice exhibited a reciprocal increase in NK cells in the spleen, liver, and lymph nodes (Supplemental Fig. 3A). Owing to partial loss of T cells, the overall percentages of other cells of the immune system, including γδ T cells, DCs, and B cells (Supplemental Fig. 3B–D), were increased in spleen, liver, and lymph nodes of mice with a T cell–specific deficit in Vps34, but the absolute numbers of these cells remained largely unaffected.

To understand the mechanism of peripheral T cell loss in Vps34^f/f;CD4-Cre mice, we first tested whether these cells were dying because of apoptosis. Staining of splenic and lymph node T cells with annexin V showed that Vps34-deficient T cells consistently exhibited a higher percentage of early apoptotic cells (annexin V⁺7-aminoactinomycin D [7-AAD]⁻) (Fig. 4A, Supplemental Fig. 2B). We also observed a higher percentage of late apoptotic cells (annexin-V⁺7-AAD⁺) in Vps34^f/f;CD4-Cre mice as compared with the heterozygote controls. However, this was not observed in the thymus of Vps34^f/f;CD4-Cre mice (Fig. 4B, Supplemental Fig. 2B), which is consistent with the normal cellularity of thymic SP T cells (Supplemental Fig. 2A).

Previous studies with autophagy-related genes such as Atg3 (21) and Atg5 (45) in T cells have shown that defective autophagy causes accumulation of cellular organelles, ultimately leading to apoptosis of T cells. Consistent with these findings, we similarly found, using organelle-specific dyes and flow cytometry, that Vps34^f/f;CD4-Cre mice have an increase in mitochondrial and endoplasmic reticulum size in their splenic and lymph node T cells but not in thymic T cells (Fig. 4C). Furthermore, electron microscopic images showed increased numbers and sizes of mitochondria in splenic Vps34-deficient T cells (Fig. 4D).

Next, we measured expression of the antiapoptotic factors Bcl-2 and Bcl-x_L in T cells. We found that expression of Bcl-2 was significantly increased and expression of Bcl-x_L was modestly increased in Vps34-deficient CD4⁺ and CD8⁺ T cells as compared with the controls (Fig. 4E), suggesting that the increase in apoptosis in Vps34-deficient T cells was not due to lack of induction of antiapoptotic factors in response to impaired growth factor signaling. We also measured the expression of proapoptotic molecules such as Bim, Bax, and Bad in control and Vps34-deficient T cells by Western blot analysis. The results showed that expression of Bim and Bad remained unchanged in thymocytes as well as in activated T cells derived from Vps34f^/+;CD4-Cre mice or Vps34^f/f;CD4-Cre mice. Interestingly, the expression of Bax was significantly elevated in activated T cells derived from Vps34^f/f;CD4-Cre mice as compared with Vps34^f/+;CD4-Cre mice, whereas its expression in thymocytes remained unchanged (Fig. 4F).

To further understand the mechanism for increased cellular apoptosis in Vps34-deficient T cells, we next measured parameters related to energy metabolism. We found that the rate of glycolysis in activated T cells derived from control and Vps34-deficient mice was similar. Interestingly, however, the rate of β-oxidation of fatty acids was significantly higher in Vps34-deficient cells than in wild-type T cells (Fig. 4G), suggesting dysregulated energy metabolism in autophagy-defective T cells.

Collectively, these results suggested that the defect in autophagy in Vps34-deficient T cells caused an accumulation of cellular organelles, leading to apoptosis, possibly due to dysregulated energy metabolism.

Previous studies with Atg3 conditional knockout mice have described lymphopenia-mediated T cell proliferation with acquisition of the CD44 activation marker and loss of the naive T cell marker CD62L (21). Phenotyping of T cells using Abs against markers such as CD44, CD69, CD25, and CD62L showed that Vps34^f/f;CD4-Cre mice have an increased frequency of CD44⁺CD62L⁻ T cells as compared with Vps34^f/+;CD4-Cre mice, predominantly reflected among splenic T cells (Fig. 5A, 5B). Whereas none of the T cells derived from Vps34 conditional knockout or control mice expressed CD69 under steady-state conditions, expression of CD25 was mostly confined to Foxp3⁺ Treg cells (see Fig. 7). This phenotype correlated well with the increased expression of intracellular Ki67 in Vps34-deficient T cells as compared with control T cells (Fig. 5C), suggesting a higher level of homeostatic proliferation among Vps34-deficient T cells, possibly due to their gradual loss of quiescence. Next, we tested how Vps34-deficient T cells respond to TCR-mediated stimulation. For this purpose, we stimulated CFSE-labeled splenocytes with anti-CD3 Abs and measured proliferation of CD4⁺ and CD8⁺ T cells using CFSE-dilution assay. We found that Vps34-deficient T cells proliferated less efficiently as compared with Vps34-sufficient T cells, suggesting a defect in the response to TCR-mediated stimulation (Fig. 5D). Such a defect was T cell intrinsic and was not due to higher levels of apoptosis observed in Vps34^f/f;CD4-Cre mice as compared with Vps34^f/+;CD4-Cre mice, as inclusion of the apoptosis inhibitor zVAD-fmk in these cultures failed to restore the T cell proliferative defect observed in Vps34-deficient T cells (Fig. 5E).

We next tested the capacity of Vps34-deficient T cells to induce an Ag-specific response in vivo. For this purpose Vps34^f/f;CD4-Cre mice and Vps34^f/+;CD4-Cre mice were immunized with OVA Ag emulsified in CFA and boosted at day 7. At day 15 after immunization, splenocytes and draining lymph node cells were restimulated with OVA in vitro, and IFN-γ and IL-17A responses were measured in the culture supernatant. The results showed that Vps34^f/f;CD4-Cre mice displayed a profound defect in the generation of OVA-specific IFN-γ and IL-17A responses (Fig. 5F). Consistent with these findings, we observed reduced levels of total IgG Abs in the serum of Vps34^f/f;CD4-Cre mice (Fig. 5G). Collectively, these results suggested a defect in TCR-mediated stimulation, resulting in defective Ag-specific immune responses in vivo.

We observed that aged (>18 wk old) Vps34^f/f;CD4-Cre mice developed a wasting syndrome with significant weight loss (Fig. 6A). This was often accompanied by a rectal prolapse, which suggested intestinal inflammation leading to the development of colitis. Histological examination of tissue sections showed significant but mild inflammation within both the small and large intestines (Fig. 6B–D). We found an increased frequency of lymphoid follicles that had a larger size, located in the mucosa of the colons of Vps34^f/f;CD4-Cre mice. In two of the Vps34^f/f;CD4-Cre mice we found unusually large lymphoid follicles in the small intestine, and this was never observed in control mice (Supplemental Fig. 4). The small and large intestinal lamina propria displayed infiltration of inflammatory cells such as neutrophils, macrophages, and plasma cells, as determined by their nuclear morphology and cytoplasmic stains (Fig. 6B, 6D). Other significant observations included fusion of the crypts (Fig. 6B), hyperplasia of epithelial cells as determined by nuclear crowding in the crypts, increased numbers of mitotic cells, and polyp formation in the small intestine (Fig. 6C). We further observed inflammatory lesions ranging from hyperplasia to adenoma in the proximal duodenum of Vps34^f/f;CD4-Cre mice (Fig. 6C). We next carried out full body necropsy to determine whether other organs of the body were affected. The results showed that inflammation was mostly confined to the intestine, whereas all the vital organs of the body such as heart, liver, stomach, kidney, lungs, and the brain appeared normal in Vps34^f/f;CD4-Cre mice as compared with the controls. Interestingly, pockets of erythropoiesis (but not myelopoiesis) were observed in the splenic sections (Fig. 6E), which could be suggestive of extramedullary erythropoiesis due to anemia. Consistent with this possibility, complete blood counts showed low hematocrit and a compensatory increase in the numbers of reticulocytes in the blood of Vps34^f/f;CD4-Cre mice (Table I).

We next measured cytokine responses of T cells in aged mice. The results showed that Vps34^f/f;CD4-Cre mice had an increased percentage of CD4⁺ T cells in spleen and lymph nodes that produced IL-17A and IL-13 cytokines, whereas CD8⁺ T cells expressed higher levels of IFN-γ as compared with cells from the heterozygous controls (Fig. 6F). These results suggested the development of T cell–mediated autoinflammatory disease in Vps34^f/f;CD4-Cre mice.

The combination of intestinal inflammation and anemia may provide an explanation for the wasting syndrome observed in Vps34^f/f;CD4-Cre mice.

To provide a possible explanation for the development of the inflammatory wasting syndrome in Vps34^f/f;CD4-Cre mice, we next assessed the development of the CD4⁺Foxp3⁺ Treg cell population of these animals. We found that Treg cell development within the thymus was evident among the CD4 SP cells in Vps34^f/f;CD4-Cre mice, but at reduced levels as compared with Vps34^f/+;CD4-Cre mice (Fig. 7A, 7B). In the peripheral lymphoid organs such as spleen and lymph nodes and in the liver, the percentage of Treg cells among CD4⁺ T cells was reduced by 20–30% in Vps34^f/f;CD4-Cre mice as compared with the heterozygote controls. We also found that, owing to reduced numbers of T cells and a decrease in overall cellularity of all peripheral lymphoid organs (Supplemental Fig. 2A), the absolute numbers of Treg cells were significantly reduced in Vps34^f/f;CD4-Cre mice (Fig. 7B). Next, we tested the capacity of these Treg cells to function as suppressor cells using an ex vivo Treg cell functional assay. For this purpose we purified CD4⁺CD25⁺ Treg cells from Vps34^f/f;CD4-Cre and Vps34^f/+;CD4-Cre mice on the basis of their CD25 expression and cultured these cells with CD4⁺CD25⁻ Teff cells from wild-type animals at various Teff/Treg cell ratios in the presence of DCs and anti-CD3 Abs. The results clearly showed that Treg cells derived from Vps34^f/f;CD4-Cre mice, compared with Treg cells derived from Vps34^f/+;CD4-Cre mice, had severely reduced functional capacity to suppress the Teff cell response (Fig. 7C). This reduced suppressive capacity of Treg cells derived from Vps34^f/f;CD4-Cre mice was not due to increased contamination with CD4⁺Foxp3⁻CD25⁺ T cells, as these cells were equally represented in both groups of mice (Fig. 7D).

We next tested the capacity of Treg cells derived from Vps34^f/+;CD4-Cre or Vps34^f/f;CD4-Cre mice to prevent the development of colitis induced by adoptive transfer of wild-type CD4⁺CD25⁻ T cells into Rag2^−/− mice as described (23). Adoptive transfer of wild-type CD4⁺CD25⁻ T cells into Rag2^−/− mice resulted in the development of colitis characterized by weight loss, moderate diarrhea, hunched posture, anal inflammation, infiltration of cells in the lamina propria, loss of goblet cells, and occasional ulceration between 2 and 3 wk after transfer. These clinical signs of colitis were significantly reversed by cotransfer of CD4⁺CD25⁺ Treg cells derived from Vps34^f/+;CD4-Cre mice. However, cotransfer of Treg cells derived from Vps34^f/f;CD4-Cre mice failed to prevent the development of colitis in recipient animals (Fig. 7E). These findings provide further evidence for a functional defect of Treg cells in Vps34^f/f;CD4-Cre mice. Next, we determined the effect of adoptive transfer of CD4⁺CD25⁻ T cells derived from Vps34^f/f;CD4-Cre mice into Rag2^−/− mice and found that these cells failed to cause a wasting disease between 2 and 3 wk after transfer. Instead, development of a wasting syndrome was observed 10 wk after adoptive transfer (Fig. 7E). This disease was characterized by loss of weight gain, diarrhea, infiltration of cells in the colonic lamina propria, loss of goblet cells, anal bleeding, and colonic ulcers, without signs of severe weight loss or hunched posture (Fig. 7E). These disease signs were prevented by cotransfer of Treg cells derived from Vps34^f/+;CD4-Cre mice. These results therefore suggested a cell-intrinsic capacity of T cells from Vps34^f/f;CD4-Cre mice to cause wasting disease, which could be counteracted by functional Treg cells from wild-type mice.

Collectively, our findings showed that intrathymic Treg cell development was evident in Vps34^f/f;CD4-Cre mice but that peripheral Treg cell homeostasis and acquisition of functional capacity were profoundly defective.

Autophagy has emerged as a key process that regulates many aspects of T cell function, including their development, survival, and homeostasis (6, 21, 45, 46). In this study, we describe the role of the autophagy-related class III PI3K Vps34 in the survival, peripheral homeostatic maintenance, and TCR-stimulated proliferation of conventional T cells. Whereas we were unable to identify a defect in the development of conventional T lymphocytes in the thymus of Vps34^f/f;CD4-Cre mice, the development of iNKT cells was profoundly blocked. Vps34-deficient thymocytes and peripheral T cells were defective in autophagic flux, making these cells susceptible to apoptosis, as reflected by their reduced numbers in the periphery. Our findings further suggested that the remaining T cells gradually lost quiescence due to the mild lymphopenic environment and/or due to reduced and ineffective Treg cells. Although these T cells had defects in TCR-induced proliferation, they retained the capacity to produce inflammatory cytokines, which might explain the development of an inflammatory wasting disease in these animals.

Vps34 has been suggested to play a critical role in the initiation of the autophagic process by generating PtdIns3P at the site of autophagosome formation, which recruits PtdIns3P-binding proteins (15). In yeast, the absence of Vps34 results in a severe blockade in autophagosome formation, but the role of Vps34 in autophagy in mammalian cells has been controversial. Recent studies have described the deletion of Vps34 in sensory neurons (17), embryonic fibroblasts, heart, liver (16), or T cells (18) from mice, indicating an essential role of Vps34 in cellular function and survival. Whereas the studies with neuron- or T cell–specific ablation of Vps34 failed to identify a defect in autophagy (17, 18), a study using the same Vps34^f/f mice as in the present study to ablate Vps34 in embryonic fibroblasts, liver, and heart showed that Vps34 is required for autophagy (16). Our results from this study provide evidence that ablation of Vps34 in T cells results in defective autophagic flux, as well as impaired turnover of autophagic cargo. In turn, these abnormalities in autophagy result in the accumulation of damaged proteins and subcellular organelles and subsequent T cell death. Immunostaining of LC3B in electron microscopic sections revealed that LC3B staining in Vps34-defcient T cells failed to localize to membrane structures but instead localized to the cytoplasm. This might be indicative of accumulating LC3B protein aggregates due to defective autophagic flux. The Vps34-independent conversion of LC3BI to LC3BII found in our study (Fig. 1A, 1C, 1H) is in agreement with previous reports describing the function of Vps34 in heart, liver, T cells, and neurons (16–18, 47). Consistent with these findings, we found that Vps34 deletion led to enhanced T cell apoptosis, similar to that caused by T cell–specific deletion of Atg3 (21), Atg5 (45), or Atg7 (21, 45, 48). Our findings in this study are consistent with a recent study reporting a critical role of Vps34 in autophagy in T cells (49).

The precise mechanism for the induction of apoptosis in Vps34-deficient T cells and the factors that mediate this process remain incompletely understood. A recent study with T cell–specific Vps34-deficient mice suggested a role for defective IL-7Rα expression, resulting in reduced induction of Bcl-2 and T cell loss (18). Although we similarly found decreased expression of IL-7Rα by T cells in our line of Vps34 conditional knockout mice (data not shown), we observed an increase in antiapoptotic factors such as Bcl-2 and Bcl-x_L. The reasons for such discrepancies could be due to differences in the Cre transgenes employed (Lck-Cre employed by McLeod et al. in Ref. 18 versus CD4-Cre employed by us and by Willinger and Flavell in Ref. 49) to selectively ablate Vps34 in T cells or due to differences in specific targeting of Vps34 exons (exons 17/18 employed by McLeod et al. in Ref. 18 versus exon 4 employed by us in Ref. 16 and by Willinger and Flavell in Ref. 49). Our results, nevertheless, are in line with high Bcl-2 expression found in T cells from T cell–specific Atg7 knockout mice (48), suggesting that defective cytokine signaling and lack of induction of antiapoptotic factors may not be the primary cause of T cell loss induced by Vps34 deficiency. Based on the similarities between the T cell phenotype observed in mice with a T cell–specific deletion of Vps34 and mice with a T cell–specific deletion in Atg3, Atg5, or Atg7 (50), as well as our results demonstrating an increase in fatty acid oxidation in Vps34-null T cells, we speculate that nutrient starvation in the face of reduced autophagy results in apoptotic death of Vps34-deficient T cells.

The most striking phenotype we observed in T cell–specific Vps34 mutant mice was the development of a wasting syndrome in aged animals. This was often accompanied by a rectal prolapse and anemia. Histological analysis showed mild inflammation throughout the intestine, and this was typically characterized by the presence of large lymphoid follicles in the mucosa, infiltration of inflammatory cells such as neutrophils, macrophages, and plasma cells in the lamina propria, crypt hyperplasia in the small intestine, and occasional polyp formation and lateral crypt fusion. Interestingly, crypt hyperplasia has often been observed during inflammatory diseases in the intestine and is typically induced by inflammatory cytokines such as IFN-γ and IFN-α (44, 45), whereas lateral crypt fusion has been observed during prolonged low-grade inflammation in the intestine of patients with inflamed appendices (51). Vps34-deficient T cells exhibited a proliferative defect in response to TCR engagement but retained the capacity to induce inflammatory cytokines. This capacity to produce cytokines was not dependent on the age of the animals, as young (6- to 8-wk-old) Vps34-deficient mice also displayed high cytokine production (Supplemental Fig. 4C). Furthermore, adoptive transfer of enriched Vps34-null CD4⁺CD25⁺ T cells was sufficient to cause colonic disease in recipient Rag2^−/− mice. This disease resembled the wasting syndrome observed in aged Vps34^f/f;CD4-Cre mice in terms of its timing. Although colitis induced by adoptive transfer of CD4⁺CD25⁻ T cells derived from either Vps34-sufficient or -deficient mice differed in its kinetics, disease in both groups of animals was ameliorated by cotransfer with Treg cells derived from wild-type mice (Fig. 7E). However, Treg cells derived from Vps34^f/f;CD4-Cre mice were defective in suppressing disease induced by adoptive transfer of CD4⁺CD25⁻ T cells derived from Vps34^f/+;CD4-Cre mice. Therefore, we conclude that the inflammatory disease observed in the intestine of T cell–specific Vps34 mutant mice is mediated by T cells in an environment that is devoid of fully competent Treg cells.

Another interesting phenotype observed in T cell–specific Vps34-deficient mice was anemia. These mice displayed a low hematocrit, low numbers of RBCs, and reduced hemoglobin in the blood (Table I). Thus, a combination of inflammation in the intestine and anemia may explain the wasting syndrome observed in these animals. Such a phenotype was not observed in conditional Vps34-deficient mice where the Cre transgene was driven by the Lck promoter (18), and we have confirmed the lack of such a wasting disease in an independently generated line of Vps34^f/f;Lck-Cre mice (V.V. Parekh, W.-X. Zong, J. Zhang, and L. Van Kaer, unpublished observations). Reasons for such divergent effects are unclear but might be due to the differential timing of Vps34 deletion induced by the distinct Cre transgenes during thymic development. Nevertheless, the wasting disease observed in Vps34^f/f;CD4-Cre mice resembles the phenotype of IL-2–deficient mice, which develop intestinal autoinflammatory disease and anemia (52–54). In these animals, IL-2 deficiency results in the generation of functionally defective Treg cells, which, in turn, causes colitis and the generation of hemolytic autoantibodies against RBCs. However, we did not observe any evidence of autoantibodies against RBCs in Vps34-null mice (data not shown). Consistent with reduced numbers of T cells and defective TCR stimulation, the total IgG Ab levels in the serum of Vps34-null mice were significantly reduced as compared with control mice (Fig. 5G). Whereas the phenotype observed in T cell–specific Vps34-deficient mice was milder and required aging of the animals, it was interesting that both of these strains developed intestinal inflammation and anemia that could be attributed to defective Treg cell function. We did not find any defect in IL-2 production by Vps34-deficient T cells, suggesting that the functional defect in Treg cells due to Vps34 deficiency is likely due to an intrinsic defect in autophagy in these cells. Further experiments will be required to precisely delineate the mechanism by which Vps34 and autophagy affect Treg cell function.

We thank the National Institutes of Health Tetramer Facility for providing CD1d tetramers and Dr. Albert Bendelac (University of Chicago, Chicago, IL) for providing the DN32.D3 hybridoma. We thank Drs. Sung Hoon Cho, Mark Boothby, and Masakazu Shiota for providing reagents related to glycolysis and β-oxidation assays. We thank the Vanderbilt University Medical Center Cell Imaging Shared Resource core for confocal microscopy and transmission electron microscopy.

The authors have no financial conflicts of interest.