The Development and Survival of Thymic Epithelial Cells Require TSC1-Dependent Negative Regulation of mTORC1 Activity Free

Thymus is a primary immune organ for the development and generation of functionally competent T cells that are critical components of the adaptive immune system (1). Thymic epithelial cells (TECs) are essential for T cell development and repertoire selection (1). Based on their function, phenotypes, and localization, TECs are divided into cortical TECs (cTECs) and medullary TECs (mTECs) (2). cTECs are responsible for the homing of T cell precursors through the expression of CXCL12 and CCL25, T cell lineage commitment via DL4-dependent activation of Notch1, and positive selection of thymocytes based on the proper affinity with MHC molecules (3–6). In contrast, mTECs provide proper microenvironment for the negative selection and maturation of thymocytes as well as the generation of thymic regulatory T (Treg) cells (7, 8).

The development, proliferation, and survival of TECs are regulated by various signaling pathways. It has been established that early TEC specification is dependent on bone morphogenetic protein signaling (1, 9, 10). Recently, two groups reported that NOTCH signaling is essential for the specification of TEC common progenitors into mTEC lineage (11, 12). The subsequent differentiation and maturation of immature mTECs into mature mTECs is regulated by NF-κB signaling pathway (13–15). TECs exhibit relative high turnover rates of several weeks (16). The wingless-type MMTV integration site family signaling is critical for the proliferation and expansion of TECs (17). The keratinocyte growth factor promotes thymus regeneration by enhancing the proliferation of TECs (18). Recently, we and another group found that mTOR signaling in TECs controls the proliferation of TECs (19, 20). Although many progresses have been made in understanding the development and proliferation of TECs, the molecular mechanisms that control the survival of TECs are still poorly understood. Several years ago, the STAT3 signaling has been shown to promote the survival and maintenance of mTECs (21, 22). Recently, a group initially reported that the prosurvival protein MCL-1, but not BCL-2 or Bcl-x_L, is essential for TEC survival (23). Later, they further revealed that the death of TEC is not mediated by death receptor or necroptosis pathways but by intrinsic apoptosis pathway (also called the mitochondrial pathway) (24). More studies need to be done to further understand the molecular regulation of TEC survival.

Tuberous sclerosis complex (TSC) is an autosomal dominant disorder that affects multiple organ systems and is caused by loss-of-function mutation in either Tsc1 or Tsc2 gene (25). The TSC is composed of TSC1 and TSC2 protein and serve as the negative regulator of mTOR signaling (25). Many pathological alterations of the disease caused by Tsc1/2 gene mutation could be ascribed to the hyperactivation of mTOR signaling pathway. The activity of mTOR signaling is delicately controlled in organisms (26). Hyperactivation of mTOR signaling due to Tsc1 ablation is implicated in the dysfunction of many immune cells. Tsc1 play critical role in the survival of naive T cells, the differentiation of Th cells and memory CD8⁺ T cells (27–30). Tsc1 ablation is also associated with impaired NK cell development, invariant NKT cell differentiation and abnormal macrophage polarization (31–33). However, the importance of Tsc1 for TEC biology is unknown. Using mouse models with TEC-specific deletion of Tsc1, we found that Tsc1 ablation predominately impaired the development and survival of mTECs and caused severe thymus atrophy. We further showed that Tsc1 ablation led to lysosomal-mediated apoptosis of mTECs because of hyperactivation of mTORC1 activity. Thus, Tsc1 is essential for the homeostasis of mTECs by inhibiting lysosomal-mediated apoptosis via mTORC1-dependent pathway.

TEC-specific Tsc1 knockout (Tsc1 cKO) mice were obtained by crossing Tsc1^flox/flox mice with Foxn1-Cre mice (34). We further crossed Tsc1 cKO mice with Tsc1^flox/floxRaptor^flox/wt mice to obtain Tsc1 cKO Raptor cHE mice. Foxn1-Cre negative littermates served as controls. Foxn1-Cre mice were the generous gifts of Dr. Y. Zhang from Peking University Health Science Center, Beijing, China. Tsc1^flox/flox mice and Raptor^flox/flox mice were kindly provided by Dr. H. Zhang (Institute of Basic Medical Sciences and School of Basic Medicine, Peking Union Medical College and Chinese Academy of Medical Sciences, Beijing, China). The primers used to identify the genetic information of mice were listed in Table I. All mice used in the study were bred under pathogen–free conditions. All animal experiments were performed in accordance with the approval of the animal Ethics Committee of Institute of Zoology, Beijing, China.

For flow cytometry staining, the samples were first incubated with 2.4G2 for 20 min at 4°C. Fluorescein-labeled (FL-1061; Vector Laboratories) Ulex europaeus agglutinin I (UEA I) was purchased from Vector Laboratories. Ki-67-PE (556027), Active caspase-3–PE (550914) and allophycocyanin rat anti-mouse CD107a (LAMP1) (560646) were purchased from BD Biosciences. Alexa Fluor 647–conjugated phospho-S6 ribosomal protein (Ser^235/236) (D57.2.2E, 4851) and Alexa Fluor 647–conjugated phospho-Akt (Ser⁴⁷³) (193H12) Rabbit mAb were purchased from Cell Signaling Technology. The following Abs were purchased from eBioscience or BioLegend: PerCP/Cy5.5, FITC, PE, PE/Cy7, allophycocyanin, allophycocyanin/Cy7, Brilliant Violet 421 or Alexa Fluor 647–conjugated anti-CD45 (clone 30-F11; BioLegend), CD326 (clone G8.8; BioLegend), CD40 (clone 3/23; BioLegend), CD80 (clone 16-10A1; eBioscience), Ly-51 (clone 6C3; BioLegend), autoimmune regulator (Aire; clone 5H12; eBioscience), CD4 (clone GK1.5; BioLegend), CD8 (BioLegend; clone 53-6.7), CD24 (clone M1/69; eBioscience), CD62L (clone MEL-14; eBioscience), CD25 (clone PC61.5; eBioscience), Foxp3 (clone FJK-16s; eBioscience), CD45RB (clone C363.16A; eBioscience), CD44 (clone IM7; BioLegend), CD69 (clone H1.2F3; eBioscience), TCRB (clone H57-597; BioLegend), and cytochrome C (clone 6H2.B4; BioLegend). Surface staining of cell suspensions was performed in PBS containing 0.1% BSA and 0.02% NaN₃ at 4°C for 30 min. Intracellular staining for Ki67, Fxop3, Aire, Bim, Bcl-2, and Mcl-1 was performed using fixation buffer (00-5123-43 and 00-5223-56; eBioscience) and permeabilization buffer (00-8333-56; eBioscience) according to previous experience (35–37). For the staining of Phospho-S6 (4851; Cell Signaling Technology), Phospho-Akt (2337; Cell Signaling Technology), and cytochrome C (612310; BioLegend), after staining with cell surface marker, cells were fixed and permeabilized with the BD Biosciences Cytofix/Cytoperm and Perm/Wash solutions. After cell surface staining, the staining of active caspase-3 (550914; BD Pharmingen), JC-1 (C2006; Beyotime Biotechnology), annexin V and propidium iodide (ZP327; ZOMANBIO) were performed according to the manufacturer’s instructions. The flow cytometry was performed with a Gallios Flow Cytometer (Beckman Coulter, Brea, CA).

Thymic stromal cells were obtained according to our previously publications (35, 38). Briefly, freshly isolated thymi were cut into small pieces, then washed the small pieces with DMEM(containing 2% FBS) thoroughly. After the small pieces settled to the bottom of the pipe, we added 2 ml of 1 mg/ml collagenase/dispase (11097113001; Sigma-Aldrich) with 20 U/ml DNAse I (D5025; Sigma-Aldrich) to resuspend these pieces and incubated at 37°C for 45 min. At the end of the digestion, cell suspensions were gently agitated and then 5 ml of PBS containing 1% FBS and 5 mM EDTA was added to neutralize the digestion. At last, cells were centrifuged and resuspended in DMEM (containing 2% FBS) for flow cytometry analysis.

For immunohistology analysis of the thymus, thymi from Tsc1 cKO mice and littermate control mice were fixed in 4% formalin and embedded in paraffin blocks. Sections (6 μm) were stained with H&E and examined by Vectra Automated Multispectral Imaging System (PerkinElmer). The statistical analysis of the area of cortex and medulla was performed using inForm.

For immunofluorescence analysis, serial sections (6 μm) from OCT-embedded frozen thymi or in vitro–cultured TECs were fixed with 4% polyoxymethylene (P1110; Solarbio Life Science) and blocked with donkey serum, washed in PBS–0.05% Tween (60305ES76; TeaSen) and incubated with Abs. The primary Abs were as follows: rat anti–cytokeratin 8 (K8) (Troma-I; Developmental Studies Hybridoma Bank [DSHB]) diluted by 1:200, rabbit anti-K5 (clone AF 138; Covance) diluted by 1:400, LAMP1 (1D4B; DSHB) diluted by 1:100, and cathepsin B (clone D1C7Y; Cell Signaling Technology) diluted by 1:400. The secondary Abs were as follows: Alexa Fluor 488–conjugated donkey anti-rat IgG (H + L) (712-546-150; Jackson ImmunoResearch Laboratories) diluted by 1:400 and Alexa Fluor 594–conjugated donkey anti-rabbit IgG (H + L) (711-586-152; Jackson ImmunoResearch Laboratories) diluted by 1:400. Nuclei were stained with DAPI (D9542; Sigma-Aldrich). Images were acquired with a laser scanning N-SIM Super-Resolution Confocal Microscope (Nikon, Tokyo, Japan).

The organs of 10-mo-old wild-type (WT) and Tsc1 cKO mice were harvested and fixed in 4% paraformaldehyde, embedded in paraffin (39601095; Leica), sectioned (5 μm), and stained with H&E. For the detection of autoantibodies, sera from 10-mo-old WT and Tsc1 cKO mice were prepared, diluted with PBS (1:30), and incubated with frozen sections of rag2^−/− mice, followed by Alexa Fluor 488–conjugated donkey anti-mouse IgG (H + L) Abs (715-546-150; 1:300; Jackson ImmunoResearch Laboratories).

After isolation of TECs from pooled thymi by method mentioned above, TECs or mTECs were enriched using anti-mouse CD45 microbeads (130052301; Miltenyi Biotec) and then sorted with a MoFlo XDP cell sorter (Beckman Coulter, Brea, CA). Total RNA was extracted using MicroElute Total RNA Kits (R6831; Omega Bio-Tek), and reverse transcription was performed with SuperScript III Reverse Transcriptase (18080-093; Invitrogen) according to the manufacturer’s instructions. Real-time PCR was performed using multiple kits (TaKaRa SYBR Premix Ex Taq, RR420) on a CFX96 apparatus (Bio-Rad Laboratories, Hercules, CA). The primers using in this study are listed in Table II.

Four to six thymi of 4-wk-old WT and Tsc1 cKO mice were pooled and digested to isolate thymic stromal cells. The isolated stromal cells were enriched using anti-mouse CD45 microbeads (130052301; Miltenyi Biotec) then sorted for mTECs using a Fusion cell sorter (BD Biosciences). Total RNA were extracted using TRIzol agent, and first strand cDNA synthesis and cDNA amplification was performed using Smart-Seq2 method. DEGseq was used to identify the differentially expressed genes between WT and Tsc1-deficient mTECs. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis was performed on all the different genes of each cell using the KOBAS online search tool (http://kobas.cbi.pku.edu.cn/). The RNA-sequencing (RNA-seq) datasets for this study can be found in the National Center for Biotechnology Information Sequence Read Archive database (http://www.ncbi.nlm.nih.gov/sra/) with the accession number PRJNA723503.

Because of the rarity of freshly isolated mTECs, Simple Western was used to detect the expression of Bid and some other proteins. mTECs were sorted by cell sorter then were lysed with radioimmunoprecipitation assay (RIPA). Simple Western was performed using 12–230 kDa Separation Module (SM-W002-1; ProteinSimple) according to the manufacturer’s instructions.

The primary and secondary Abs used for Simple Western are as follows: human/mouse BID Ab (AF860; R&D Systems), cleaved caspase-3 (Asp¹⁷⁵) (5A1E) Rabbit mAb (9664; Cell Signaling Technology), human/mouse caspase-8 Ab (AF1650; R&D Systems), and mouse/rat cathepsin L Ab (AF1515; R&D Systems) and anti-goat secondary HRP Ab (043-522; Simple Western), anti-mouse secondary HRP Ab (042-205; Simple Western), and anti-rabbit Secondary HRP Ab (042-206; Simple Western).

mTECs were sorted from pooled thymi and lysed in RIPA. Briefly, cells were washed with cold PBS and lysed in RIPA buffer (50 mM Tris-HCl [pH 7.4], 1% NP-40 [ab142227; Abcam], 0.25% Na-deoxycholate [S817543; Macklin], 150 mM NaCl, and 1 mM EDTA [pH 7.4]) with a protease inhibitor mixture (P8340; Sigma-Aldrich). Protein concentration was determined using a bicinchoninic acid assay. Proteins were analyzed by SDS-PAGE (L5750; Sigma-Aldrich) and transferred onto PVDF membranes (IPFL00010; Merck Millipore). The PVDF membranes were blocked with 5% nonfat dried milk (LP0031; Oxoid) for 1 h and then incubated with primary Abs overnight on a shaker at 4°C. The HPR-coupled anti-mouse (074-1806; KPL International) or anti-rabbit (070-1506; KPL International) secondary Abs were then added and were detected through chemiluminescence (WBKLS0500; Merck Millipore).

The following primary Abs were used: phospho-s6 ribosomal protein (Ser^240/244) (D68F8) Rabbit mAb (5364; Cell Signaling Technology) diluted by 1:10000, β-Tubulin (T5201; Sigma-Aldrich) diluted by 1:1000, LAMP1 (AB 528127, 1D4B; DSHB) diluted by 1:1000, cathepsin B (31718, D1C7Y; Cell Signaling Technology) diluted by 1:1000, and mouse/rat cathepsin L Ab (AF1515; R&D Systems) diluted by 1:1000.

Cellular fractionation was performed using cellular fractionation kit–HT (ab109718; Abcam) according to the protocol described by Huai et al. (39). Briefly, cells were digested, washed once with buffer A, and centrifuged. The pellet was resuspended buffer A at the concentration of 6 × 10⁶ cells/ml, added equal volume of buffer B, and then incubated at room temperature for 7 min on a rotator and centrifuged sequentially at 5,000 × g and 10,000 × g at 4°C for 1 min each. The supernatant was collected as cytosolic fraction. To get the lysosome fraction, the cytosolic fraction was further centrifuged at 100,000 × g for 1 h. The pellet represented the lysosomal fraction.

The detection of reactive oxygen species (ROS) in freshly isolated mTECs was performed using CellROX Deep Red Flow Cytometry Assay Kit (C10491; Thermo Fisher Scientific) according to manufacturers’ instruction. The detection of ROS in cultured TECs was performed using Reactive Oxygen Species Assay Kit (S0033; Beyotime Biotechnology) following manufacturers’ instruction.

For in vitro TEC culture, thymi from WT and Tsc1 cKO neonatal mice were digested as mentioned above. Small thymic fragments from each step were collected and pooled. Fragments were allowed to settle and washed twice with Thymic Epithelial Cell Medium (3911; ScienCell). The remaining thymic explants were plated in 24-well plates with Thymic Epithelial Cell Medium and cultured at 37°C with 5% CO₂. TECs will grow in this medium, and cells gradually die. After ∼8 d, TECs were digested with tyrosine for annexin V and propidium iodide staining.

The lysosome of live cells was stained by LysoTracker Red (C1046; Beyotime Biotechnology) according to the manufacturer’s instructions.

To test the activity of cathepsin B, Magic Red Cathepsin assays was performed. Freshly isolated thymic stromal cells were incubated with the cathepsin B substrate (6133; ImmunoChemistry Technologies) at 37°C for 1 h; after staining with surface Abs, the activity of cathepsin B was examining by flow cytometry.

All data are presented as the means ± SD. A Student unpaired t test for comparison of means was used to compare groups. A p value <0.05 was considered statistically significant.

To investigate the role of Tsc1 in TECs, we generated mice with conditional inactivation of Tsc1 in TECs (Foxn1-Cre; Tsc1^flox/flox; designated as Tsc1 cKO mice in brief; the age-matched Tsc1^fl/fl mice were used as control mice; Table I). The efficient inactivation of Tsc1 locus was confirmed by real-time PCR (Supplemental Fig. 1A, 1B). We first assessed the effect of TEC-specific inactivation of Tsc1 on the development of thymus at embryonic and postnatal stages. After TEC-specific deletion of Tsc1, the thymus size was smaller in comparison with age-matched controls (Supplemental Fig. 1C). Consistently, total thymocyte cellularity reduced significantly from embryonic day 17.5 (E17.5) to 4-wk-old mice compared with WT littermate controls (Fig. 1A). We next investigated the impact of Tsc1 cKO on thymic epithelia. The cell numbers of TECs isolated from Tsc1 cKO mice decreased significantly at E17.5, newborn, and postnatal 1, 2, and 4 wk (Fig. 1B), although the frequencies of TECs only decreased obviously at 2 and 4 wk (Supplemental Fig. 1D, 1E). TECs are composed of cTECs and mTECs, which are responsible for the positive and negative selection of thymocytes, respectively (1). The percentages and cell numbers of mTECs decreased dramatically in Tsc1 cKO mice in comparison with littermate Tsc1^f/f mice (Fig. 1C, 1D, Supplemental Fig. 1F). In contrast, the absolute numbers of cTECs were almost unaffected by the loss of Tsc1, although the relative frequencies of cTECs increased significantly (Fig. 1E, 1F, Supplemental Fig. 1F).

We then examined the anatomical changes of the thymus in Tsc1 cKO mice. We found that the area of the medullary compartment of Tsc1 cKO mice was significantly smaller compared with WT mice as showed by reduced medulla to cortex ratio (Fig. 1G). The medulla and cortex express its specific marker K5 and K8, respectively (1). A histological examination revealed a smaller medulla region in mutant mice as expected (Fig. 1H). What is more, the thymus medullary region in Tsc1 cKO mice appeared tattered and was sparsely dispersed in cortical region (Fig. 1H). The mutant mice also displayed a medulla of damaged architecture compared with controls (Fig. 1I). In contrast, the mutant mice had a cortex with regular architecture that was indistinguishable in comparison with control mice (Supplemental Fig. 2A). In contrast to well-demarcated cortico-medullary junction (CMJ) in WT thymi, the typical circumscribed boundary at CMJ was disrupted in thymi of Tsc1 cKO mice (Fig. 1J). These data collectively indicated that the mTEC compartment was severely impaired after TEC-specific ablation of Tsc1.

We further investigate whether the reduction of mTECs is intrinsic or due to a defect of progenitors. More and more evidence indicated mTECs were derived from progenitors expressing molecular traits specific for cTECs, including β5t, CD205, and IL-7 (40–46). We then evaluated the expression of CD205 on TECs at E17.5; the results showed that the expression of CD205 on TECs is similar between WT mice and Tsc1 cKO mice (Supplemental Fig. 3A). So, the reduction of mTECs may not be due to early stage of mTEC lineage commitment.

We then focused our studies on mTECs. As the maturation of mTECs, the expression of MHC class II (MHCII), CD80, CD40, and Aire is upregulated (47). The percentage of CD40⁺ mTECs was unaffected by inactivation of Tsc1 (Fig. 2A). mTECs are heterogeneous with regard to the expression of MHCII and CD80, enabling their subdivision into MHCII^loCD80^lo mTEC^lo immature subsets and MHCII^hiCD80^hi mTEC^hi mature subsets (1). Similar to the expression of CD40, the frequency of mTEC^hi and mTEC^lo subsets was unaltered after conditional Tsc1 cKO (Fig. 2C, 2D). Aire is responsible for the expression of many tissue-restricted Ags (TRAs) and the induction of immune tolerance (1). Tsc1 ablation in TECs did not change the expression of Aire in mTECs (Fig. 2E). However, because of the overall reduction of mTECs, the absolute cell numbers of CD40⁺, MHCII^hi, mTEC^lo, and Aire⁺ mTECs are all decreased dramatically (Fig. 2B, 2D, 2F). In contrast, both the percentage and the cell numbers of MHCII^hi cTECs were unchanged after Tsc1 ablation (Supplemental Fig. 2B, 2C, 2D). We further examined the impact of Tsc1 deletion on the differentiation of embryonic mTECs, and the results showed that the expression of MHCII on mTECs at E17.5 was comparable between WT and Tsc1 cKO mice (Supplemental Fig. 3B), indicating the primitive mTEC maturation is not obviously affected by Tsc1 deletion. These results collectively demonstrated that the ablation of Tsc1 in TECs did not affect the differentiation and maturation of mTECs but resulted in the overall reduction of mTECs.

The generation, survival, and development of thymocytes depend largely on the thymic epithelial microenvironment (1). We next evaluated the thymopoiesis of 4-wk-old Tsc1^fl/fl and Tsc1 cKO mice. The frequencies of CD4⁺CD8⁻ single-positive (CD4SP) and CD4⁻CD8⁺ single-positive (CD8SP) thymocytes decreased mildly in Tsc1 cKO mice when compared with Tsc1^fl/fl mice and the frequencies of CD4⁻CD8⁻ double-negative (DN) and CD4⁺CD8⁺ double-positive (DP) thymocytes had on obvious change (Fig. 3A). However, the absolute numbers of DN, DP, CD4SP, and CD8SP thymocytes all reduced dramatically in Tsc1 cKO mice compared with Tsc1^fl/fl controls (Fig. 3B). The frequencies of CD69⁺TCRβ^int/hi thymocytes showed a tendency to decrease (Supplemental Fig. 4A), indicating the positive selection of thymocytes is potentially impaired in Tsc1 cKO mice. The positively selected thymocytes subsequently migrate into medullary region to undergo negative selection (1). Thymocytes that survive negative selection proceed further maturation in the thymus before they export to the periphery as naive T cells (48). This postselection maturation is reflected by phenotypic and functional changes, including the increased surface expression of CD62L and downregulated expression of CD24 (49, 50). The frequencies of mature (CD24⁻CD62L⁺) CD4SP and CD8SP thymocytes decreased significantly in Tsc1 cKO mice compared with littermate controls (Fig. 3C). These observations suggested that the postselection maturation of thymocytes was impaired in mice with TEC-specific Tsc1 ablation.

In addition to providing the molecular cues for the development of conventional T cells, mTECs also provide niches for the development of thymic Treg (tTreg) cells (8). The dramatic reduction of mTECs in Tsc1 cKO mice promoted us to examine whether tTreg cell development was impaired. Both the percentage and absolute cell number of CD4⁺CD8⁻Foxp3⁺ tTreg cells reduced considerably in Tsc1 cKO mice compared with WT littermate controls (Fig. 3D, 3E). The mature CD4⁺CD8⁻CD25⁺Foxp3⁺ tTreg cells develop from CD4⁺CD8⁻CD25⁺Foxp3⁻ precursors in the thymus (51). We found that the percentage of CD4⁺CD8⁻CD25⁺Foxp3⁻ tTreg precursors and the ratio of CD4⁺CD8⁻CD25⁺Foxp3⁻ precursors to mature CD4⁺CD8⁻CD25⁺Foxp3⁺ tTreg cells elevated significantly in mice with TEC-specific Tsc1 deletion (Fig. 3F, 3G). These results demonstrated that the development of CD4⁺CD8⁻CD25⁺Foxp3⁺ tTreg cells was blocked at the tTreg precursor stage because of the impaired thymic medulla microenvironment in Tsc1 cKO mice.

The mature naive T cells generated in the thymus populate the peripheral lymphoid organs to perform their functions (52). To assess the thymus output function, we evaluated the recent thymic emigrants that are defined as CD4⁺CD62L^hiCD45RB^int splenocytes (53). The mutant mice displayed a significant, albeit limited, reduction of recent thymic emigrants (Supplemental Fig. 4B). Consistently, both the cell numbers of CD4⁺ and CD8⁺ T cells considerably decreased in the spleen of Tsc1 cKO mice compared with controls (Fig. 3I), although the frequency the CD4⁺ and CD8⁺ T cells only mildly decreased (Fig. 3H). Furthermore, there was noticeable increment of CD44⁺CD62L⁺ or CD44⁺CD62L⁻ effector/memory-like CD4 and CD8 T cells and reduction of CD44⁻CD62L⁺ naive T cells in Tsc1 cKO mice (Supplemental Fig. 4C, 4D). The percentage of Treg cells in the spleen slightly elevated in Tsc1 cKO mice; however, the absolute cell number of Treg cells diminished substantially in Tsc1 cKO mice compared with littermate controls (Supplemental Fig. 4E–G). These data demonstrated the homeostasis of peripheral T cells was disturbed in Tsc1 cKO mice.

mTECs are essential for the establishment of central immune tolerance by expressing the Aire-dependent and Aire-independent TRAs and by providing proper niches for tTreg development (8, 54, 55). The defect of mTEC compartment and the reduction of Treg cells indicated that Tsc1 cKO mice may develop autoimmune disease. We first dissected the impact of Tsc1 ablation on the expression of Aire-dependent and Aire-independent TRAs by comparing our RNA-seq results with published dataset (54–58), and the results showed only a fraction of Aire-dependent and Aire-independent TRAs was disturbed by Tsc1 cKO (Supplemental Fig. 5A, 5B). We then examined the presence of lymphocyte infiltrates and autoantibodies in lung, pancreas, liver, and kidney. The lung of 10-mo-old Tsc1 cKO mice exhibited obvious presence of infiltrated lymphocytes and autoantibodies. The other examined tissues showed no noticeable lymphocyte infiltration and autoantibodies in comparison with age-matched Tsc1^fl/fl mice (Supplemental Fig. 5C, 5D). Thus, we concluded that Tsc1 cKO mice develop mild autoimmune disease, which may be due to the less-changed TRAs expression.

We then explored the causes for the reduction of mTECs in Tsc1 cKO mice. We found the proliferation ability of mTECs was comparable between WT and Tsc1 cKO mice assessed by Ki67 staining (Fig. 4A, 4B), and similar phenomenon was observed as to cTECs (Supplemental Fig. 6A). Tsc1 ablation has been reported to associate with impaired cell survival ability (27, 28). We therefore measured the expression of active caspase-3, a hallmark of apoptotic cell death. Both the newborn and 4-wk-old Tsc1-deficient mTECs exhibited significantly increased active caspase-3 expression in comparison with WT mTECs (Fig. 4C, 4D, Supplemental Fig. 6B). In contrast, the apoptosis of 4-wk-old cTECs was unaltered by Tsc1 deletion (Supplemental Fig. 6C). Although the apoptosis of newborn cTECs increased after Tsc1 ablation, the apoptosis level of newborn cTECs was much lower than 4-wk-old cTECs (Supplemental Fig. 6C). Collectively, these results indicated TEC-specific Tsc1 deletion predominantly promoted the apoptosis of mTECs and slightly affected the apoptosis of cTECs at certain time points of mice after birth. Furthermore, by using an in vitro TEC culture system established in our laboratory (59, 60), we found the apoptosis of TEC was also increased after Tsc1 deletion, as showed by increased annexin V⁺ cells (Fig. 4E). Based on these results, we concluded that the reduction of mTECs in Tsc1 cKO mice was due to increased apoptosis.

To elucidate the molecular events causing the defect survival capacity of Tsc1-deficient mTECs, we performed RNA-seq analysis of mTECs isolated from 4-wk-old Tsc1^fl/fl and Tsc1 cKO mice. The global gene expression profiles comparison between WT and Tsc1-deficient mTECs revealed that 475 probes increased and 380 probes decreased after Tsc1 cKO (Fig. 4F). We then used KEGG analysis method to investigate the affected signaling pathways after Tsc1 ablation by comparing the differentially expressed genes shown in (Fig. 4F. The downregulated top ten pathways included Ag processing and presentation, human T cell lymphotropic virus type 1 infection, and so on (data not shown). The upregulated top 10% pathways included lysosome, metabolic pathway, oxidative phosphorylation, and so on (Fig. 4G). Notably, the apoptosis was among these markedly upregulated pathways (Fig. 4G), which is consistent with the excessive apoptosis of Tsc1-deficient mTECs showed in (Fig. 4C–E. Gene Set Enriched Analysis also revealed the increase of metabolic pathway, oxidative phosphorylation, lysosome, and apoptosis of Tsc1-deficient mTECs compared with Tsc1^fl/fl mTECs (Fig. 4H). These results indicated that Tsc1-deficient mTECs displayed increased cell metabolism, oxidative phosphorylation, lysosomal biogenesis, and apoptosis. Based on the increased apoptosis of Tsc1 cKO mTECs (Fig. 4C–E), we further scrutinized the apoptosis KEGG pathway. We found that the increased apoptosis signaling of Tsc1 cKO mTECs was mainly through the cathepsin–Bid axis (Supplemental Fig. 7), which is termed lysosomal-mediated apoptosis (61). Consistently, it is well known that the abnormal lysosomal biogenesis, increased cell metabolism, and oxidative phosphorylation could trigger lysosomal-mediated apoptosis (29, 39, 61, 62).

Given that Tsc1 ablation increased cell metabolism and oxidative phosphorylation, which would lead to increased ROS production (29), we next examined the ROS production in mTECs. The production of ROS in Tsc1 cKO mTECs increased significantly compared with WT mTECs (Fig. 5A). Consistently, the cultured Tsc1-deficient TECs also displayed elevated ROS production in comparison with WT controls (Supplemental Fig. 8A). According to the increased lysosome pathway from RNA-seq analysis, we then examined the changes of lysosome in Tsc1-deficient mTECs. The expression of lysosome marker (LAMP1) increased markedly in Tsc1-deficient mTECs compared with WT mTECs (Fig. 5B). The expression of many lysosomal biogenesis genes also increased obviously in Tsc1 cKO mTECs, as assessed by real-time PCR (Fig. 5C; Table II). The amount and size of lysosome both increased in cultured Tsc1-deficient TECs than WT controls as assessed by immunofluorescence and LysoTracker staining (Supplemental Fig. 8B, 8C). Lysosomes are the major digestive organelles with an acidic interior within cells (63). The function of lysosomes is largely dependent on the lysosomal proteases; among them, the cathepsins are the best-known proteases (63, 64). Cathepsins are mainly responsible for lysosomal protein degradation, but they also involved in lysosomal-mediated cell death (65). We then detected the expression of cathepsin B in Tsc1-deficient mTECs. The results showed that the expression of cathepsin B increased significantly in Tsc1-deficient mTECs compared with WT mTECs (Fig. 5D). Consistently, the activity of cathepsin B also increased as measured by Magic Red Cathepsin staining assay (Fig. 5E). These data demonstrated that the production of ROS and lysosomal biogenesis increased significantly in Tsc1-deficient mTECs relative to WT mTECs.

Some recent studies illustrated that lysosomal cathepsins are also involved in cell apoptosis, which pivotally depends on the release of cathepsins into the cytosol, a process known as lysosomal membrane permeabilization (LMP) (66). The overproduction of ROS could induce LMP (61), and the enlarged lysosome size also make lysosomes more prone to LMP (61, 63). The increased ROS production and the abnormal lysosomal biogenesis in Tsc1-deficient mTECs reminded us that these cells may undergo LMP. We thus examined the leakage of cathepsins from lysosome into cytosol in WT and Tsc1 cKO cultured TECs. The results showed that the cytosolic cathepsin B and cathepsin L increased obviously in Tsc1 cKO TECs compared with controls (Fig. 5F), indicating Tsc1 ablation resulted in the leakage of cathepsin B and cathepsin L into cytosol. Furthermore, in comparison with WT TECs, Tsc1 cKO TECs not only exhibited an elevated cathepsin B expression but also showed a redistribution of cathepsin B from perinuclear localization to diffused pattern throughout the cytosol (Supplemental Fig. 8D). These results indicated the Tsc1 ablation leads to obvious LMP in TECs.

Apoptosis signaling after LMP often leads to activation of intrinsic pathway and caspase cascade (66). The cytosolic cathepsins cleavage Bid into its active form truncated Bid (tBid), which then translocates to mitochondria so as to induce mitochondria membrane permeabilization and release of cytochrome C and ultimately trigger intrinsic apoptosis pathway (66, 67). We thus detected the cleavage of Bid into tBid in mTECs after Tsc1 deletion, and the results showed that the cleavage of Bid into tBid increased dramatically in Tsc1-deficient mTECs relative to WT controls (Fig. 5G). It has been established that the cleavage of Bid into tBid is mainly mediated by cathepsins in lysosomal-mediated apoptosis or by caspase-8 in Fas-induced apoptosis (61, 68). The expression of Fas and caspase-8 were comparable between Tsc1-deficient mTECs and WT controls (Supplemental Fig. 8E, 8F). These data demonstrated the increased level of tBid was mainly due to the cleavage of Bid by cathepsins. We next examined the mitochondrial membrane potential of Tsc1-deficient mTECs. The results showed that Tsc1 ablation reduced the mitochondrial membrane potential of mTECs significantly as showed by the decreased JC-1 aggregates to monomers ratio (Fig. 5H). Similarly, the in vitro–cultured, Tsc1-deficient TECs also displayed diminished mitochondrial membrane potential in contrast to WT controls (Supplemental Fig. 8G, 8H). In summary, these results illustrated that Tsc1 ablation led to the overproduction of ROS and the malfunction of lysosome. The elevated ROS caused LMP and the release of cathepsins into cytosol, which then cleaved Bid into active tBid that subsequently induced intrinsic apoptosis.

The most important function of TSC1 is to restrain the activity of mTORC1 (26). To elucidate whether the increased apoptosis of mTECs in Tsc1 cKO mice was due to the hyperactivation of mTORC1, we first evaluated the activity of mTORC1 in Tsc1-deficient mTECs. Indeed, Tsc1 ablation resulted in hyperactivation of mTORC1 as assessed by increased phosphorylation of ribosomal protein S6 at Ser^240/244 (Fig. 6A). Next, we generated TEC-specific Tsc1 cKO mice that are heterozygous for Raptor, a scaffold protein of mTORC1 (Foxn1-Cre; Tsc1^flox/floxRaptor^WT/flox, designated as Tsc1 cKO Raptor cHE mice briefly). The genetic reduction of mTORC1 activity in Tsc1 cKO mice markedly increased the size, weight, and cellularity of the thymus compared with those in Tsc1 cKO mice (Fig. 6B–D). We further investigated the impact the heterozygote deletion of Raptor on the development of TECs. The cellularity of TECs was partially rescued after heterozygote deletion of Raptor in Tsc1 cKO mice (Fig. 6E). Furthermore, the genetic knockdown of mTORC1 activity also resulted in partially but significant recovery of the percentage and cellularity of mTECs compared with those in Tsc1 cKO mice (Fig. 6F–H). In contrast, the absolute cell numbers of cTECs were comparable in WT, Tsc1 cKO, and Tsc1 cKO Raptor cHE mice (Fig. 6J), although the frequency of cTECs displayed obvious change in different genotype mice (Fig. 6F, 6I). Importantly, the increased apoptosis of mTECs could also partially reverse in Tsc1 cKO Raptor cHE mice in contrast to Tsc1 cKO mice. These results clearly indicated that the impaired survival of mTECs in Tsc1 cKO mice is due to the hyperactivation of mTORC1.

We found in this study that Tsc1 played a critical role for the survival of mTECs via preventing lysosomal-mediated apoptosis. Deficiency of Tsc1 in TECs reduced the mTEC compartment from embryonic to adult stages. Although Tsc1 deficiency reduced the cell numbers of mTECs dramatically, Tsc1 may be dispensable for the differentiation and functional maturation of mTECs, as evidenced by the intact frequencies of mature mTEC populations (CD40⁺, Aire⁺, and mTEC^hi cells) in Tsc1-deficient mice. It has been reported that mTECs derive from progenitor cells sharing a similarity with the cTEC lineage, including β5t, CD205, and IL-7 (40–46). The similar expression of CD205 on WT and Tsc1 cKO TECs at E17.5 indicated the defect of mTECs may not be due to early stage of mTEC lineage commitment. Also, Tsc1 ablation exhibited little effect on the cell number of cTECs; the transcriptional program of cTECs may be altered as reflected by the reduction of cortex-dependent positive selection and the decrease of cortex-resident DP thymocytes. Recently, some studies discovered the unexpected contribution of several pathways to control TEC development. Similar to the role of Tsc1 in TECs, two studies demonstrated that Stat3 is also essential for the survival immature mTECs (21, 22). Another report revealed that Irf4 is imperative for tTreg homeostasis by regulating the expression of several chemokines and costimulatory molecules on TECs (69, 70). The well-known tumor suppressor p53 is also critical for the homeostasis of TECs, and TEC-specific ablation of p53 disrupts thymopoiesis and central immune tolerance (71). More studies need to be done to systematically uncover the molecular regulation of TEC development and homeostasis.

Tsc1 deficiency resulted in excessive apoptosis of mTECs in vivo, and we further recapitulated this phenomenon using the in vitro TEC culture system. Many publications attributed the excessive apoptosis of Tsc1-deficient cells to downregulation of prosurvival member Bcl-2 and upregulation of proapoptotic member Bim (27, 28, 32, 72). Although, our results indicated that Tsc1 deficiency in mTECs did not alter the expression of Bim (data not shown). Although the expression of Bcl-2 showed a tendency to decrease in Tsc1-deficient mTECs (data not shown), a recent study clearly demonstrated that Bcl-2 is dispensable for TEC survival (23). Thus we ruled out the potential involvement of Bim and Bcl-2 in the defect survival capacity of Tsc1-deficient mTECs. Noticeably, by performing RNA-seq analysis and subsequent experiments, we proved that Tsc1-deificient mTECs undergo lysosomal-mediated apoptosis. Therefore, we revealed a previous unappreciated role of Tsc1 in the regulation of cell apoptosis in mTECs.

We further illustrated that Tsc1 deficiency impaired the survival of mTECs through hyperactivation of mTORC1 because genetic reduction of Raptor could partially ameliorate the increased apoptosis of mTECs. The tight control of mTOR activity is very important for the development of organisms (26). Tsc1-deficient TECs exhibited exorbitant mTORC1 activity, which leads to the increased apoptosis of mTECs. The increased apoptosis of mTECs could partially reverse by heterozygote deletion of Raptor in Tsc1 cKO TECs. These observations are consistent with another report in which the authors showed that the development defect of NK cells caused by Tsc1 ablation could partially rescue by genetic knockdown of Raptor (32). Combining with previous publications from our laboratory (19) and by Wang et al. (20), in which the authors showed that mTOR deficiency caused a poor mTEC development in mice, these results collectively documented the that the tightly control of mTORC1 activity is critical for mTEC development; that is, more or less mTOR activity is harmful to the mTEC development.

The present study also unveiled that Tsc1 ablation promoted the expression of cell metabolism and lysosomal biogenesis–related genes. It is widely acknowledged that mTORC1 promotes cell metabolism (26); however, the role of mTORC1 on lysosomal biogenesis is complicated. Some reports elucidated that mTORC1 inhibits lysosomal biogenesis (73–75), whereas another group reported that mTORC1 activation caused by Tsc2 knockout enhances lysosomal biogenesis (76). Noticeably, two recent studies revealed that epidermal Tsc1 deletion upregulates lysosomal biogenesis (77), and Tsc2 deficiency also promotes lysosomal biogenesis (78), which is coincident with our present study.

It is well known that lysosome is closely related with the degradation of macromolecules, which is associated with autophagy, endocytosis, and phagocytosis (63). However, emerging evidence indicated that lysosome is also involved in cell death, but most of them are under the in vitro conditions (39, 79, 80). We revealed in this study that Tsc1-deficient mTECs underwent lysosomal-mediated apoptosis in vivo. We proved that Tsc1 ablation led to LMP of mTECs. In Tsc1-deficient mTECs, LMP may be induced by two factors: one is the overproduction of ROS (81), the other is the abnormal of lysosomal biogenesis because larger lysosome may breakdown more efficiently with stress (82). The results of LMP is the leakage of cathepsins from lysosome into cytosol and the cytosolic cathepsins cleaved inactive Bid into active tBid that induced mitochondria permeabilization and cell death of mTECs.

In summary, Tsc1 deficiency in TECs caused obvious poor survival of mTECs and disruption of thymus medulla, which resulted in the defect of thymocyte development. Tsc1 ablation in mTECs led to the hyperactivation of mTORC1, which resulted in the increment of oxidative phosphorylation and lysosomal biogenesis. The elevated oxidative phosphorylation produced excessive ROS that caused LMP. The LMP caused leakage of cathepsins into cytosol, and the cytoplasmic cathepsins subsequently cleaved Bid into active tBid. The tBid then translocated to the mitochondria and caused mitochondrial membrane depolarization, which at last initiated the apoptosis of mTECs. The schematic diagram of Tsc1 controlling lysosomal-mediated apoptosis in mTECs was shown in visual abstract.

We thank Dr. Peng Wang for critical reading the manuscript. We thank Dr. Yu Zhang for the gifts of Foxn1-Cre mice. We acknowledge Dr. Hongbing Zhang for the gifts of Tsc1^flox/flox mice and Raptor^flox/flox mice.

The authors have no financial conflicts of interest.