LAMP2 regulates autophagy in the thymic epithelium and thymic stroma-dependent CD4 T cell development

ABSTRACT Within the thymus, thymic epithelial cells (TECs) provide dedicated thymic stroma microenvironments for T cell development. Because TEC functionality is sensitive to aging and cytoablative therapies, unraveling the molecular elements that coordinate their thymopoietic role has fundamental and clinical implications. Particularly, the selection of CD4 T cells depends on interactions between TCRs expressed on T cell precursors and self-peptides:MHC II complexes presented by cortical TECs (cTECs). Although the macroautophagy/autophagy-lysosomal protein degradation pathway is implicated in CD4 T cell selection, the molecular mechanism that controls the generation of selecting MHC II ligands remains elusive. LAMP2 (lysosomal-associated membrane protein 2) is a well-recognized mediator of autolysosome (AL) maturation. We showed that LAMP2 is highly expressed in cTECs. Notably, genetic inactivation of Lamp2 in thymic stromal cells specifically impaired the development of CD4 T cells that completed positive selection, without misdirecting MHC II-restricted cells into the CD8 lineage. Mechanistically, defects in autophagy in lamp2-deficient cTECs were linked to alterations in MHC II processing, which was associated with a marked reduction in CD4 TCR repertoire diversity selected within the lamp2-deficient thymic stroma. Together, our findings suggest that LAMP2 interconnects the autophagy-lysosomal axis and the processing of selecting self-peptides:MHC II complexes in cTECs, underling its implications for the generation of a broad CD4 TCR repertoire. Abbreviations: AIRE: autoimmune regulator (autoimmune polyendocrinopathy candidiasis ectodermal dystrophy); AL: autolysosome; AP: autophagosome; Baf-A1: bafilomycin A1; B2M: beta-2 microglobulin; CTSL: cathepsin L; CD74/Ii: CD74 antigen (invariant polypeptide of major histocompatibility complex, class II antigen-associated); CFSE: carboxyfluorescein succinimidyl ester; CFU: colony-forming unit; CLIP: class II-associated invariant chain peptides; cTECs: cortical TECs dKO: double knockout; DN: double negative; DP: double positive; ENPEP/LY51: glutamyl aminopeptidase; FOXP3: forkhead box; P3 IFNG/IFNγ: interferon gamma; IKZF2/HELIOS: IKAROS family zinc finger 2; IL2RA/CD25: interleukin 2 receptor, alpha chain; KO: knockout; LAMP2: lysosomal-associated membrane protein 2; LIP: lymphopenia-induced proliferation; Lm: Listeria monocytogenes; MAP1LC3/LC3: microtubule-associated protein 1 light chain 3; MHC: major histocompatibility complex; mTECs: medullary TECs; PRSS16/TSSP: protease, serine 16 (thymus); SELL/CD62L: selectin, lymphocyte; SP: single positive; TCR: T cell receptor; TCRB: T cell receptor beta chain; TECs: thymic epithelial cells; UEA-1: Ulex europaeus agglutinin-1; WT: wild-type.


Introduction
The thymus generates T cells expressing α β T cell receptors (TCRs) that recognize foreign antigens presented by self-major histocompatibility complex (MHC) molecules and are tolerant to self-components. This process is developmentally regulated by interactions between TCR expressed on T cell precursors (thymocytes) and self-peptide-MHC complexes presented by thymic epithelial cells (TECs) [1]. TECs express a myriad of soluble and cell-surface ligands that coordinate all stages of T cell differentiation. Importantly, the thymopoietic role of TECs is sensitive to aging and cytoablative therapies, which compromise T-cell responses to pathogens and vaccination in the elderly and immunocompromised patients [2]. Therefore, the identification of novel regulators of TEC functionality is vital to comprehend the foundations of T cell-immunity and to restore thymopoiesis in disorders associated with defective T cell responses. TECs are divided into functionally distinct cortical (cTECs) and medullary (mTECs) subtypes. While cTECs guide T cell lineage commitment and positive selection, mTECs regulate negative selection and T regulatory cell differentiation [3]. The functional segregation between cTECs and mTECs is, in part, determined by their unique antigen processing and presentation properties, which are controlled by specific protein degradation machineries and transcriptional programs [4,5]. However, how these proteolytic machineries coordinate the presentation of selective self-peptide-MHC complexes remains elusive at the molecular level.
In this regard, mTECs employ conventional proteolytic pathways to generate "public" MHC I-and MHC II-bound peptides that negatively select autoreactive thymocytes or drive their deviation into regulatory T cells [4]. These key steps for tolerance induction depend on the distinctive ability of mTECs to express tissue-restricted antigens, a process that is mediated by AIRE and FEZ family zinc finger 2 [1]. Alternatively, cTECs harbor unique antigen processing properties that generate sets of "private" MHC I-and MHC II-bound self-peptides, which are critical for the positive selection of double-positive (DP) thymocytes and their commitment into CD4 or CD8 T cells. The generation of MHC I-bound peptides that select CD8 + T cells depends on the distinct proteolytic activity of PSMB11/β5t (proteasome (prosome, macropain) subunit, beta type 11)containing thymoproteasome [5,6]. For the generation of MHC II-bound peptides, cTECs employ unconventional endogenous MHC II-loading pathways that involve the autophagymediated transport of cytosolic proteins into lysosomes [7,8]. Lysosomal proteases, including CTSL (cathepsin L) and PRSS16/TSSP (protease, serine 16 (thymus)) contribute to the generation of specific self-peptides that select CD4 + T cells [9][10][11][12]. Although autophagy has been coupled to the control of thymic selection [7,8,13], the molecular mechanism that links this catabolic pathway to the generation of selective self-peptides in the lysosomes remains elusive. Our study identifies a role for LAMP2 (lysosomal-associated membrane protein 2) in the thymic stroma as a determinant regulator of CD4 T cell development and TCR repertoire formation.

LAMP2 is dispensable for the establishment of major TEC microenvironments
To identify new candidates involved in the antigen processing capacity of cTECs, we analyzed our transcriptomic data from postnatal cTECs and mTECs [14], together with available data sets from other studies [15,16]. Given that MHC II-bound peptide generation involves the endosomal-lysosomal pathway, we sought genes associated with these vesicular compartments , (WT and lamp2 KO subsets in color and gray, respectively); Right histogram: cTECs (ENPEP/Ly51 + UEA-1 − ), mTEC lo (ENPEP/Ly51 − UEA-1 + CD80 lo ) and mTEC hi (ENPEP/Ly51 − UEA-1 + CD80 hi ). Bars graphs represent mean fluorescence intensity (MFI) of LAMP2 expression. Data representative of 3 independent experiments (n = 12 animals). (C) cTEC and mTEC composition in embryonic day 15.5 (E15.5) and 2-week-old WT and lamp2 KO thymus. Dot plots show a representative ENPEP/Ly51 and UEA staining and graphs represent the average cellularity of c/mTECs in the indicated timepoints from 3 independent experiments (n = 11-14 animals/group). (D) Immunofluorescence analysis of thymic sections from 2-week-old WT and lamp2 KO thymus stained for DAPI (blue), UEA-1 (red) and KRT8 (keratin 8; green). Results in A-C are shown as mean ± SEM. * p < 0.05; ** p < 0.01; *** p < 0.001. that were selectively enriched in cTECs relatively to mTECs. Ctsl (cathepsin L) and Prss16 were expectedly among the top genes upregulated in cTECs. We selected to analyze LAMP2, which was the only member of the LAMP family upregulated (4-fold) in cTECs ( Figure 1(A), Figure S1A). LAMP2 is a lysosomal membrane protein that regulates macroautophagy and chaperone-mediated autophagy (CMA) [17], promoting the fusion of autophagic vacuoles with lysosomes [18,19]. LAMP2 has been also implicated in the regulation of MHC II presentation in human B cells [20,21]. Yet, its involvement in T cell development and selection is unknown.
In man, genetic mutations in LAMP2 are associated with Danon Disease, a fatal X-linked syndrome caused by lysosomal dysfunction [22]. Mice deficient for Lamp2 (lamp2 knockout; lamp2 KO) recapitulate key aspects of the human condition, including the accumulation of autophagic vacuoles in several cell types [18,19,23]. Despite the increased mortality in early adulthood, lamp2 KO mice survive the first weeks of life [19]. Hence, to study the function of LAMP2 in the establishment of major TEC subsets, we analyzed the lamp2 KO thymus during embryonic and postnatal life. We began by analyzing the expression of LAMP2 protein in embryonic wild-type (WT) TECs, and using cells from lamp2 KO mice as a control for the antibody staining ( Figure S1B). Within the postnatal thymus, TECs represented the highest LAMP2-expressing subset, relatively to thymocytes and other non-TEC stromal cells, including endothelial and mesenchymal cells. The corresponding lamp2 KO subsets were included to further validate antibody staining. In line with the transcriptomic data, cTECs significantly expressed higher levels of LAMP2 relatively to mTEC lo/hi subtypes (Figure 1(B)). The deficiency in Lamp2 did not alter the generation of primordial cTECs and mTECs at embryonic day 15 nor affected the cellularity of the main cTEC and mTEC subsets in the 2-weekold thymus (Figure 1(C)). Further analysis of the expression of cTEC-associated markers (LY75/CD205, CD40) or mTEC subpopulations, including mTEC lo (MHCII lo CD40 lo CD80 lo ), mTEC hi (MHCII hi CD40 hi CD80 hi ), CCL21 + and AIRE + cells, revealed no differences between WT and lamp2 KO TECs ( Figure S1C). Additionally, the spatial organization of cortical and medullary areas was normal in the lamp2 KO thymus (Figure 1(D)). Our results indicate that LAMP2 is not essential for the establishment of global cTEC and mTEC compartments.

LAMP2 in thymic stromal cells specifically controls the development of CD4 T cells
The pleiotropic defects of lamp2 KO mice and the broader expression of LAMP2 in several tissues hampered a direct assessment of its role in TEC-mediated T cell selection. To circumvent this aspect, we employed a well-established thymic transplantation model wherein WT or lamp2 KO embryonic thymi were grafted under the kidney capsule of nude adult recipients, referred to hereafter as WT-Nude and lamp2 KO-Nude. In these chimeras, the deficiency in Lamp2 was restricted to TECs, endothelial and mesenchymal cells, allowing the assessment of how recipient-derived T cell precursors (from WT origin) developed and were selected within WT and lamp2 KO thymic stroma, in the absence of confounding effects of lamp2-deficiency in other tissues. Although TECs have a non-redundant role in T cell commitment and selection [4], this approach did not formally exclude a potential contribution for LAMP2 in endothelial and mesenchymal cells in these processes.
Analysis of T cell development 8-10 weeks posttransplantation showed that the percentages of immature double-negative (DN) and DP thymocytes were similar in WT-Nude and lamp2 KO-Nude thymic grafts. While the numbers of total, DN and DP thymocytes appeared slightly reduced, these differences were not statistically significant ( Figure 2(A), Figure S2A), indicating that LAMP2 is dispensable for β-selection and DN-DP transition. Notably, the frequencies of single-positive 4 (SP4) and SP8 thymocytes were altered in lamp2 KO thymi. These changes resulted in a specific and statistically significant, reduction in the number of SP4 but not SP8 cells (Figure 2(A)). Analysis of T cells expressing high levels of TCRB (T cell receptor beta chain) (TCRB hi ) confirmed the decrease of circa 40% in the average number of SP4 cells in the lamp2 KO thymus (1.3x10 6 cells) relatively to the cellularity found in the WT counterparts (2.2x10 6 cells). These changes led to a deviation in SP4:SP8 ratios in the mutant thymi as compared to controls, and endogenous thymi of WT mice (Figure 2(B)). The thymic alterations coincided with changes in the peripheral T-cells, with a statistically significant reduction in the number of lamp2 KO thymus-derived splenic naïve CD4 + T cells. The numbers of effector/memory CD4, and naïve and effector/memory CD8 T cells were moderately reduced, but these changes were not statistically significant ( Figure 2(C)). Together, these results pointed to an impaired CD4 T cell development within the lamp2deficient thymic microenvironment.
To identify the stage at which CD4 + T cell development was altered in lamp2 KO thymus, we examined distinct stages of pre-and post-positive selection, based on the differential expression of TCRB and CD69 expression on thymocytes: Population I (TCRB neg/int CD69 neg ) contains mostly pre-selected DP thymocytes; Population II (TCRB int CD69 int ) harbors TCR-signaled cells initiating positive selection; Population III (TCRB hi CD69 hi ) represents post-positively selected SP thymocytes; Population IV (TCRB hi CD69 neg ) consists of more mature SP cells [24]. Although the frequency of populations II, III and IV within total thymocytes were similar, we found a noticeable alteration in the frequency of SP4 and SP8 in post-positively selected cells (TCRB hi CD69 hi , III) of the lamp2 KO-Nude thymi ( Figure 3(A-B)). This led to a significant reduction in the numbers of post-positively selected SP4, but not SP8 thymocytes, in lamp2 KO-Nude thymi (Figure 3(B)). We further analyzed the expression of CD5 in thymocyte subsets as this correlates with TCR affinity for self-peptide-MHC ligand [25,26]. Stage II DP thymocytes from WT and lamp2 KO thymus expressed comparable CD5 levels, suggesting normal TCR signaling during the early stages of positive selection (Figure 3(C), Figure S2B). Early after positive selection, DP thymocytes transverse through a CD4-+ CD8 int (CD4 int ) transitional stage prior to their commitment into CD4 and CD8 lineage [27]. The frequency and numbers of stage II CD4 int cells in lamp2 KO-Nude mice were comparable to the WT-Nude setting (Figure 3(B)). Yet, the levels of CD5 in stage II CD4 int and III-IV SP4 thymocytes were moderately reduced within the lamp2 KO-Nude thymus, suggesting an attenuation of TCR signaling in cells that completed positive selection (Figure 3(C), Figure S2B). Additionally, stage IV SP4 thymocytes did not accumulate in the lamp2 KO thymus nor presented elevated CD5 levels ( Figure 3(B-C), Figure S2B), arguing against excessive TCR signaling. To study whether lamp2deficiency could affect other steps of CD4 T cell development that depend on mTECs, we analyzed SP4 maturation and the regulatory T cell differentiation [1]. The proportion of immature (CD24 + SELL/CD62L − ) and mature (CD24 − SELL/CD62L + ) SP4 thymocytes were comparable within WT and lamp2 KO thymus. Additionally, the frequency of immature (IL2RA/CD25 + Forkhead box P3 (FOXP3) − and IL2RA/CD25 − FOXP3 + ) and mature CD4 T regulatory (IL2RA/CD25 + FOXP3 + ) cells was similar in WT and lamp2 KO thymi, suggesting that the maturation of T reg was not affected by lamp2-deficiency. Although moderately reduced, the numbers of mature CD4 T regulatory cells were not statistically different ( Figure S2C-D). Contrarily, the cellularity of immature (IL2RA/CD25 + FOXP3 − ) T reg, immature and mature SP4 was statistically diminished in lamp2 KO thymus. The reduction in these subsets may arise from the reduction in SP4 that completed positive selection, which was the subset mostly affected by the deficiency in Lamp2 and included immature T reg and SP. Together, our results suggested a more prominent function for LAMP2 in regulating the development of CD4 + T cells that complete positive selection.
The selection of CD4 + T cells depends on "strong" and persistent TCR-mediated signals [28]. The decreased CD5 levels on SP4 cells could suggest that the reduction in SP4 cells in lamp2 KO thymus was functionally linked to a defective CD4 commitment. In this scenario, MHC IIrestricted thymocytes were misdirected toward the CD8 Numbers on the plots indicate the frequencies of the different subsets. Bar graphs in A and B depict the average absolute cellularity of the thymus and the indicated thymocyte subsets in WT-Nu (gray) and lamp2 KO-Nu (blue). In B, ratio of the frequency of SP4:SP8 is shown in WT-Nu and lamp2 KO-Nu, as well in the endogenous thymus. (C) Analysis of peripheral T cells in spleens from WT-Nu and lamp2 KO-Nu mice. Graphs (left) represent the average % and numbers of TCRB + T cells in the spleen. Dot plots show CD44 and SELL/CD62L expression within CD4 and CD8 T cells. Graphs (right) show average cellularity of naïve (SELL/CD62L + CD44 lo ) and effector/memory (SELL/CD62L ± CD44 hi ) CD4 and CD8 T cells. Data represent an average of 3 independent experiments (n = 10 WT and lamp2 KO ectopic thymi). Results in A-C are shown as mean ± SEM. * p < 0.05; ** p < 0.01; *** p < 0.001. lineage due to an attenuated TCR signaling. To examine this possibility and prevent MHC-I-driven CD8 selection, we generated mice double deficient for Lamp2 and B2m (beta-2 microglobulin) (lamp2 b2m dKO mice). Thymic transplantations were set with b2m KO or lamp2 b2m dKO embryonic thymus grafted into WT mice. Owing to the deficiency in b2m, and consequently MHC-I expression, the selection of MHC-I-restricted SP8 thymocytes was expectedly abolished in b2m KO thymic grafts [29], so that mature TCRB + cells mostly comprised SP4 thymocytes ( Figure 4(A-B)). Notably, the frequency and numbers of SP4 thymocytes were reduced in lamp2 b2m dKO thymus, without a concomitant increase in SP8 thymocytes ( Figure 4(A-B), Figure S3A). Moreover, recently postselected (TCRB hi CD69 hi ) and mature (TCRB hi CD69 neg ) SP4 cells were significantly reduced in frequency and numbers in lamp2 b2m dKO thymic grafts ( Figure 4(C), Figure  S3A). To evaluate the possible contribution of LAMP2 in cortical and medullary negative selection at a polyclonal level, we analyzed the expression of IKZF2/HELIOS (IKAROS family zinc finger 2) on DP (TCRB lo/int IL2RA/ CD25 − FOXP3 − ) and CD4SP (TCRB hi IL2RA/CD25 − FOXP3 − ) thymocytes [26]. The frequency of IKZF2/ HELIOS + cells was similar in both WT-and lamp2 KOderived thymocyte subsets (Figure 4(D), Figure S3B). These observations support that deficiency in Lamp2 does not and lamp2 KO (lamp2 KO-Nu) ectopic thymus grafted into Nude recipients. Dot plots show the analysis of CD69 and TCRB expression on total thymocytes from WT-Nu and lamp2 KO-Nu ectopic thymus. Stage I (TCRB neg/int CD69 neg ), II (TCRB int CD69 int ), III (TCRB hi CD69 hi ) and IV (TCRB hi CD69 neg ) defined pre-, recently-, postselected and mature stages, respectively. Graph represents the average percentages of total populations II, III and IV. In each bar (populations II, III and IV bar) is shown the corresponding frequency of SP4 and SP8 cells, which were calculated based on the frequencies of SP4 and SP8 shown in panel B. (B) CD4/CD8 expression within stages II (TCRB int CD69 int ), III (TCRB hi CD69 hi ) and IV (TCRB hi CD69 neg ), which were defined on total thymocytes (A). Graphs represent the average percentages and absolute numbers of the indicated thymocyte subsets in WT-Nu (gray) and lamp2 KO-Nu (blue). (C) Graphs represent the MFI of CD5 expression in the indicated subsets (defined as shown in Figure S2). Data shown represent an average of 10 ectopic WT and 10 ectopic lamp2 KO thymi from 3 independent experiments. Results in A-C are shown as mean ± SEM. * p < 0.05; ** p < 0.01; *** p < 0.001. redirect MHC II-restricted cells into CD8 lineage, but instead prevents the development of CD4 T cells that completed positive selection.

Alterations in autophagy in lamp2 KO cTECs are associated with changes in MHC II processing
To determine whether the defect in CD4 T cell development could be mechanistically linked to changes in autophagy provoked by the deficiency in Lamp2, we evaluated the autophagic flux in TECs. Ex vivo isolated lamp2 KO cTECs displayed an increased level of Cyto-ID, a tracer that marks AP and AL [30] ( Figure 5(A)), indicative of altered autophagy dynamic. Moreover, lamp2 KO mice were crossed with transgenic mice that ubiquitously expressed a RFP-GFP-MAP1LC3A/LC3 in a pH-dependent manner (RFP-GFP-LC3). In RFP-GFP-LC3 mice, RFP is stable and GFP is quenched in the acidic pH of lysosomal vesicles, allowing the distinction between AP (GFP + RFP + ) and AL (RFP + ) [31]. The RFP-GFP-LC3 expression was predominantly detected in TECs relatively to the hematopoietic and non-TEC stroma of WT and lamp2 KO thymus ( Figure  S4A). Moreover, the proportion of RFP + -GFP + -LC3 TECs was specifically increased in lamp2 KO cTECs, but not mTECs ( Figure 5(B) and Figure S4B). The accumulation of RFP + -GFP + -LC3 cells resulted from an augmented expression of GFP-LC3, but not RFP-LC3, in lamp2 KO cTECs ( Figure 5(B)). We further analyzed the autophagic flux at the single-cell level by employing imaging flow cytometry. lamp2 KO cTECs contained cells with a reduced number of RFP + -LC3 puncta/cell compared to WT counterparts, being the lamp2 KO subset mostly composed of cells with dual RFP + -GFP + -LC3 puncta ( Figure 5(C)). As the status of autophagy flux correlates with the number of autophagic vesicles, these results indicated that AP-lysosome fusion was markedly affected in lamp2 KO cTECs, as reported in other cells deficient in Lamp2 [18,19,23].
We next studied whether changes in LAMP2-mediated autophagy were associated with alterations in self-peptide: MHC II processing in cTECs, which in turn could condition their capacity to select CD4 T cells. MHC II molecules are stabilized in the endoplasmatic reticulum by their binding to the CD74 antigen/Ii (invariant polypeptide of major histocompatibility complex, class II antigen-associated). Subsequently, CD74/Ii is degraded into a fragment called CLIP (class II-associated invariant chain peptides) in the MIIC (MHC II processing compartment) to permit antigenbinding [32,33]. To evaluate MHC II processing and maturation, we employed 15G4 antibody that recognizes I-A b occupied by the CD74/Ii degradation intermediates small leupeptin induced protein or CLIP [34,35]. Strikingly, 15G4 staining was substantially increased in lamp2 KO cTECs, but not in mTECs, relative to WT counterparts ( Figure 5(D)). In line with previous studies [10], mTECs displayed higher 15G4 staining as compared to cTECs ( Figure 5(D)). Control and mutant cTECs, as well as mTECs, expressed comparable levels of MHC II and I molecules (Figure S4C), suggesting a specific role of LAMP2 in the processing of MHC II, but not in its trafficking to the cell membrane. Exploring the fact that Lamp2 resides on X chromosome, we analyzed TECs from  Figure S3B). Data include an average of 2 independent experiments (n = 6 b2m KO and 8 lamp2 b2m dKO-WT). Results in A-D are shown as mean ± SEM. * p < 0.05; ** p < 0.01; *** p < 0.001.
lamp2 WT/KO heterozygous thymus, which due to random X-inactivation in females included WT and lamp2 KO TECs ( Figure S4D). lamp2 KO cTECs presented an augmented 15G4 staining, even within a microenvironment containing WT cells, suggesting that LAMP2 controls MHC II processing in a cell-autonomous manner ( Figure 5(E)). Given the role of CTSL in the degradation of CD74/Ii in cTECs [9,10], we studied if altered MHC II processing resulted from a decrease in CTSL or lysosomal functions. The proteolytic function of CTSL and global lysosomal activity were increased in lamp2 KO cTECs (Figure S4E-F). These results suggest that the augmented levels of MHC II-bound to CD74/Ii degradation intermediates did not result from a decrease in CTSL activity or lysosomal processing capacities. Lastly, we evaluated whether changes in LAMP2-mediated autophagy could be functionally linked to the alterations in MHC II processing. To do so, postnatal day 3-5 WT thymi were treated with two inhibitors, chloroquine and bafilomycin A 1 (Baf-A1), of AP-lysosome fusion. Interestingly, chloroquine and Baf-A1 treatment increased 15G4 staining in WT cTECs ( Figure 5(F)). The elevated 15G4 staining of WT mTECs was not substantially altered in Baf-A1-treated cells (Figure S4G). Worth noting, mTECs are less abundant in the postnatal thymus [14] and were sensitive to apoptosis following culture. Although the results with pharmacological inhibitors should be considered with caution due to additional specific or offtarget effects, our data suggest that inhibition of AP-lysosome fusion in WT cTECs mimicked the phenotype observed on lamp2 KO cTECs, and establish a presumable causality between changes in autophagy flux and MHC II processing.

LAMP2 in thymic stroma controls the generation of a broad CD4 TCR repertoire
Lastly, we determined whether thymic stroma deficient in Lamp2 altered the generation of TCR repertoire diversity in Histograms show representative 15G4 staining on the cell surface of WT and lamp2 KO c/mTECs from lamp2 WT/KO heterozygous mice. Graphs represent the MFI found in indicated subsets in control (gray) and mutant (blue) cells (n = 8 from 4 experiments). (F) Postnatal day 3-5 WT thymus were treated overnight with chloroquine (50 μM) or with Baf-A1 (0.5 μM) for 5 h. Histograms show the representative staining with 15G4 staining in control (light gray) and chloroquine-treated (black) or Baf-A1-treated (black) in WT cTECs. Graphs represent the MFI in control (gray) and treated (dark gray/ black) cTEC (n = 4-6 animals from 3-4 independent experiments). Results in A-F are shown as mean ± SEM. * p < 0.05; ** p < 0.01; *** p < 0.001. developing CD4 T cells, possibly by affecting the MHC IIprocessing and the generation of positive selecting selfpeptides in cTECs. Despite considerable efforts to identify the nature of self-peptides that promote positive selection [5], it remains challenging to conduct this biochemical characterization with sensitivity in rare populations, such as TECs. Due to this limitation, we determined the TCR clonal composition of SP4 thymocytes generated in WT or lamp2 KO thymi. The assessment of TCR diversity could serve as an indirect indicator of the quality of the MHC II-peptide ligandome presented by lamp2 KO cTECs. Given the vast combinations of TCRA/α-TCRBβ chains in SP4 thymocytes in a broad polyclonal setting, we used mice expressing a fixed transgenic TCR Vβ6 (beta6) chain [36] as recipients of thymic grafts. This model referred as TCR:Vβ6Tg, reduced the complexity of the TCR repertoire to the diversity of TCRA chains and favored CD4 + T cell selection, as confirmed by the accumulation of SP4 thymocytes in WT thymic grafts ( Figure 6(A), Figure S5A). The bias toward CD4 lineage may result from the fact that the Vβ6 chain is derived from a MHC II-restricted TCR [36]. Consistent with the results in the fully polyclonal TCR repertoire of Nude and WT recipients, lamp2-deficiency restrained the number of SP4 TCR:Vβ6Tg thymocytes (Figure 6(A)). The cellularity of DP cells was also reduced in TCR:Vβ6Tg model (Figure S5A), possibly due to the ubiquitous expression of MHC-IIrestricted Vβ6 chain that may anticipate the effects provoked by lamp2-deficiency in the DP stage. For repertoire analysis, we isolated RNA from 3 samples of 1.10 6 (FASC sorted) SP4 (CD4 + TCRB + ) thymocytes from WT and lamp2 KO thymus. The Tcra gene was specifically amplified and the generated libraries were subjected to deep sequencing analysis of Tcra transcripts. We recovered 0.7-1,2.10 6 sequences in SP4 cells derived from WT and lamp2 KO grafts, from which 0.6-2.10 5 corresponded to specific TCRs (Figure S5B, first two columns). Further analysis identified 1,3-1,5.10 3 unique TCRs in the WT setting. This number (0,3-0,4.10 3 ) was notably reduced in SP4 cells from lamp2 KO thymi (Figure S5B, third column). Given the variable number of retrieved TCR sequences, the analysis was normalized to 6.10 5 randomized TCR sequences/sample. This threshold was defined by the lower number of sequences identified on one sample (KO2) (Figure S5B). Following unbiased normalization, the number of unique clonotypes remained significantly lower in lamp2 KO-derived SP4 thymocytes relatively to the ones generated from WT thymus ( Figure 6(B), Figure S5B). Additionally, the distribution of TCRA sequences was distinctively different. While clonotypes of WT-derived thymus presented the stereotypical broad distribution expected to be found in a polyclonal repertoire, lamp2 KO counterparts displayed a clear skewing in the TCR catalog ( Figure 6(C)). Further analysis revealed that the clonality index was markedly increased in mutant samples ( Figure S5C).
These results led us to evaluate whether the restricted thymic SP4 TCR repertoire affected the function of peripheral T cells generated within lamp2 KO thymic stroma. To do so, we resorted to the fully polyclonal TCR repertoire model wherein WT and lamp2 KO thymi were grafted in Nude recipients, as described in Figure 2. We probed the proliferative response of splenic CD4 and CD8 T cells of WT-or lamp2 KO-nu mice to polyclonal TCR stimulation in vitro, and found no differences between groups as measured by CFSE analysis (Figure S6A). These results suggest that CD4 and CD8 T cells developing in lamp2 KO thymic stroma respond to a polyclonal TCR-mediated stimulation. Lastly, we determined whether thymic defects in TCR repertoire formation influence the outcome of an immune response to a live intracellular pathogen. We employed a well-defined model of T-cell dependent recall response to Listeria monocytogenes (Lm) [37]. Recipient mice were inoculated with 2.10 3 CFU (colony-forming units) of Lm at 12 weeks post thymic transplantation, to guarantee peripheral T cell reconstitution, and then rechallenged with 1.10 6 CFU 4 weeks later ( Figure S6B). Non-thymic grafted nude recipients did not resist the infection protocol (Figure S6C), highlighting the requirement for a T cell-mediated response in this model. We assessed the T cell reconstitution in the spleen, and the clearance of Lm in the liver of infected WT-nu and lamp2 KO-nu animals 3 days after the secondary challenge. Although reduced, the numbers of naïve and effector/memory splenic CD4 and CD8 T cells were not significantly altered in WT-nu or lamp2 KO-nu ( Figure S6D). All recipient mice were reconstituted. Despite the variability in CFU counts and the absence of complete bacterial clearance in both genotypes, the average CFU counts were slightly elevated in lamp2 KO-nu mice, which presented an increase in the number of animals with a higher bacterial burden (≥ log 10 4) ( Figure S6E). Interestingly, the CFU counts inversely correlated with the amount of IFNG/IFNγ (interferon gamma)-producing CD4 found in lamp2 KO-nude after the recall response. These cells have been shown to provide protection against Lm [37]. A similar association was observed in IFNG/IFNγ-producing CD8, but this correlation was not significant (Figure S6E). Although antigen-specific responses were not monitored, lamp2 KO-nu appeared to present a compromised in vivo T-cell response to secondary infection by Lm. Together, our results suggest that thymic stroma deficient in Lamp2 restricts not only the quantity of selected CD4 + T cells, but also the quality of their TCR repertoire.

Discussion
Following the discoveries on the contributions of CTSL and PRSS16/TSSP to the generation of selecting MHC II:peptides complexes [9][10][11][12], there have been few advances in the mechanism that regulates CD4 + T cell selection. Our findings suggest that LAMP2 may represent a functional liaison that interconnects autophagy and processing of self-peptides-MHC II for selection in cTECs, which in turn regulates the development of CD4 T cells and the diversity of their TCR repertoire. LAMP2 is expressed in TECs, but it was also detected in mesenchymal and endothelial cells of the thymic stroma. As non-TEC subsets do not have a major role in thymocyte selection [4], it is possible to conjecture that the defects in CD4 T cell development in lamp2 KO thymic stroma may arise from LAMP2-dependent failures in TECs. Future studies with TEC-specific conditional knockout mice will provide more precise details on the cell-intrinsic role of LAMP2 in TECs. The strength and duration of the interaction between TCR and MHC II:self-peptide ligandome have implications in thymic selection [27,28]. Thymic stroma deficient in Lamp2 led to a particular reduction in CD4 T cells that completed positive selection. While changes in TCRB int CD69 int and TCRB hi CD69 hi suggest defects in positive selection, alterations in TCRB hi CD69 hi and TCRB hi CD69 neg may reflect survival defects following positive selection, altered lineage commitment or negative selection. Our findings indicate that the development of CD4 T cells was not mechanistically associated with an abnormal redirection of MHC IIrestricted cells into CD8 T cell lineage. Moreover, the reduced CD5 levels in TCRB int CD69 int and TCRB hi CD69 hi SP4s and the lack of alterations in IKZF2/HELIOS in DP and SP4 collectively suggested that TCR signaling and negative selection of SP4 thymocytes were not augmented. Instead, they may support the hypothesis that LAMP2 regulates the development of MHC II-restricted CD4 T cells that completed positive selection in the thymic cortex.
The observation that deficiency in Lamp2 particularly affected autophagy flux in cTECs, but not mTECs nor other non-TEC thymic stromal cells, are in line with the reported high levels of autophagy in cTECs [7,8,38,39] and the role of LAMP2 in lysosome-AP fusion [18,19]. Moreover, autophagic vesicles gain access to the MHC II compartment in TECs [40] and autophagy has been linked with the generation of the peptide repertoire presented by MHC II [7,8,38,39]. We found that CTSL and lysosomal activities were elevated in lamp2 KO cTECs, arguing against the possibility that alterations in CD74/Ii-MHC II processing results from a failure in its proteolytic function. The increased lysosomal activity was also reported in the hippocampus of lamp2 KO mice [41], suggesting that LAMP2-mediated autophagy impacts lysosomal function. Several non-mutually exclusive possibilities might explain the reduction in SP4 selection by lamp2deficient thymic stroma. The inhibition of AP-lysosome fusion in lamp2 KO cTECs can compromise the shuttling of autophagyderived cytoplasmatic selective antigens into the late endosomal MIIC loading pathway. Under these conditions, CTSL [9,10], PRSS16/TSSP [11,12] and other lysosomal proteases may not have access to substrates to produce MHC II-restricted selfpeptides specialized for positive selection of CD4 T cells. This can lead to a compensatory upregulation of lysosomal activities as typically reported in models of lysosomal dysfunction or lysosomal storage disorders [18]. Additionally, the alteration in autophagic flux might also disturb CD74/Ii-MHC II trafficking to MIIC compartment or H2M function, thereby preventing the complete degradation of CD74/Ii or self-peptide loading. Furthermore, CTSL also contributes to the generation of MHC II-bound self-peptides presented by cTECs [9,10,35]. Thus, the increase in CTSL activity in lamp2 KO cTECs could also interfere with the quality of MCH II-peptide ligandome presented by these cells during positive selection. Hence, the incomplete degradation of CD74/Ii-MHC II in lamp2 KO cTECs may result from altered trafficking of CD74/Ii-MHCII to MIIC and/or from a reduction in MHC II-loaded with positive selection-inducing self-peptides, due to an altered antigen processing and peptide repertoire generation. The consequent change in the composition of selective MHC II ligands may lead to a reduction of CD4 T cells that are completing positive selection and a skewing of their TCR repertoire. It will be important to elucidate the biochemical mechanism by which LAMP2 potentially regulates MHC class II-peptide generation in cTECs. Under regular conditions, endogenous antigens may access MHC II molecules through several LAMP2regulated autophagic rotes [17,18]. Future studies should also aim at investigating whether LAMP2-dependent macroautophagy delivers cytosolic antigens into autolysosomes for processing before loading to MHC II, and/or antigens are directly routed into the lysosome via CMA. In this regard, ATG5-dependent autophagy has been linked to positive selection of CD4 T cells [8,39], but it also contributes to mTEC-mediated tolerance induction [38,42]. Worth noting, lamp2-deficiency did not cause major defects in mTEC-dependent function nor disturbed tolerance, as measured by the presence of autoantibodies/lymphocytic infiltrates at 12 weeks post-transplantation (data not shown). Future studies should address whether breaks in tolerance can unfold in aged settings. Alternatively, autophagy may have distinct roles in c/mTEC function, whereby the specificity of the route and its consequence in thymic selection, can also be imposed by distinct downstream elements. While ATG5 participates in early phases of autophagy and may have a broader effect in cTEC-and mTECmediated thymic selection [8,39], LAMP2 mostly regulates later stages of the autophagic process and MHC II processing in cTECs. Future analysis should examine whether changes provoked by lamp2-deficiency in cTECs account for a more dedicated role in the control of CD4 T cell positive selection.
The observations that lamp2-deficiency in thymic stroma induced a skewing, but not a complete block, in CD4 TCR repertoire diversity, suggests that LAMP2 may regulate the generation of particular sets of autophagy-dependent MHC IIbound positive selecting peptides in cTECs. This hypothesis is in line with previous studies [43] supporting that positive selection is limited by the bioavailability of specific selfpeptide/MHC ligands expressed by cTECs. Still, previous studies showed that a single MHC-peptide ligand can positively select multiple TCRs [44,45], indicating that positive selection is relatively permissive regarding the composition of MHCpeptide ligands in cTECs. Assuming that LAMP2 contributes to the generation of particular peptide:MHC II ligands, its deficiency may have variable effects depending on the plasticity of a given TCR. Future analysis employing TCR transgenics will allow defining LAMP2-dependent specificities and the contribution of LAMP2 in positive and negative selection. Additionally, post-selected thymocytes expressing high-affinity TCR intrathymically proliferate in response to their self-peptide /MHC ligands [46]. Thus, the dominance of particular clonotypes within the lamp2-deficient thymus may result from the expansion of SP4 T cells bearing favorable TCR affinities for overrepresented self-peptide/MHC ligands. Importantly, the TCR repertoire analysis was conducted in total SP4 cells, without segregating conventional and regulatory CD4 T cells. Although T reg comprises a small fraction of thymic SP4, one can only presume that their TCR repertoire is also affected. Future analysis is required to clarify the impact of LAMP2 in TCR repertoire formation and the function of T regulatory cells. Lastly, the defect in thymic production imposed by lamp2-deficiency impacted on the peripheral T cell pool, mostly on the numbers of naïve CD4 T cells, and to a less extent on naïve CD8 T cells and effector/memory CD4 and CD8 T cells. In the thymic transplant of Nude mice, the establishment of the peripheral T cell pool depends on thymic output and LIP (lymphopenia-induced proliferation). As thymic CD8 T cell development was not affected, the reduced peripheral CD8 T cell pool may arise indirectly from insufficient CD4 T cell help. In this regard, it was shown that the LIP of CD8 T cells depends on CD4 T cells [47][48][49]. Additionally, our results indicate that CD4 and CD8 T cells, which developed in lamp2-deficient thymic stroma, respond to polyclonal TCRmediated signaling. Yet, lamp2 KO-nude recipients appear to hold a compromised T cell-dependent recall immune response to Lm infection. Future studies should clarify the peripheral consequences of these thymic defects, including a more indepth analysis of T cell homeostasis, antigen-specific responses and tolerance, as well as the role of CD4 T cell in B cell responses and CD8 T cell memory.
In sum, our findings suggest that lamp2-deficiency in thymic stroma conditions the development and TCR repertoire formation of CD4 T cells. These findings provide a novel framework to further investigate how LAMP2 functionally interconnects autophagy and lysosomal generation of selecting self-peptides in cTEC, and its implications for CD4 T cell positive selection. Moreover, our results may also open new avenues for the identification of the "peptidic-self" in the thymus, which remains one of the crucial gaps in our understanding of thymus biology.

Mice
We used lamp2 KO, b2m KO, RFP-GFP-LC3 Tg, TCR:Vβ6 Tg and Nude mice that have been described previously and are all in a C57BL/6 background [19,29,31,36]. Mice were housed under specific pathogen-free conditions, and all animal experiments were performed in accordance with European guidelines (Directive 2010/63/EU). For fetal studies, the day of the vaginal plug detection was designated as embryonic day (E) 0.5.

Thymic kidney transplants
Thymi from WT and lamp2 KO mice were transplanted under the kidney capsule of recipient mice as described [50]. Briefly, thymic lobes from E15.5 WT and lamp2 KO mice were cultured for 2 days with 1.35 mM 2'-deoxyguanosine (SIGMA, D0901) to deplete lymphoid cells. Prior to transplantation recipient mice were pre-treated with the analgesic buprenorphine (0.05 mg/kg of body weight) and anesthetized with a gaseous mixture of 3% isofluorane and oxygen.

TEC and hematopoietic cell isolation
TECs were isolated as described [14,50,51]. Hematopoietic cells from thymus and spleen were prepared by mechanical disruption of the respective tissues. Splenic red blood cells were lysed using erythrocyte lysis solution: 155 mM ammonium chloride (Sigma-Aldrich, A9434), 10 mM potassium bicarbonate (Sigma-Aldrich, 237,205).

CTSL and lysosomal activity assays
CTSL and lysosome activities were respectively monitored using Magic Red CTSL (ImmunoChemistry Technologies, 941) and Lysosome-Specific (Biovision, K448) detection kits, according to the manufacture's protocol. Briefly, cells were either incubated with a magic red staining solution containing the CTSL fluorogenic substrate or Lysosome-Specific Self-Quenched Substrate for 1 h at 37°C in DMEM (10% FBS). Cells were then washed and stained with TEC markers for flow cytometry analysis. In cells with active CTSL, the substrate fluoresces red upon proteolytic cleavage. The fluorescent signal is proportional to the intracellular lysosomal activity in cells with active lysosomes.

TCR sequencing
SP4 thymocytes (1.0 × 10 6 cells) derived from TCR:Vβ6Tg hematopoietic progenitors and developing in WT or lamp2 KO thymus grafts were purified by cell sorting. RNA was extracted using RNeasy Mini kit (Qiagen, 74,106) following the manufacturer's instructions. Full-cDNA library was prepared using Mint-2 kit (Evrogen, SK005), which introduces 5′-adapters to cDNA fragments, according to the manufacturer's instructions. The Tcra gene was then specifically amplified using Platinum TM Taq DNA polymerase high fidelity (Invitrogen, 11,304,011) and a primer pair specific for the 5′-adapter and the C region of the Tcra gene. The sequencing library was prepared using the Nextera kit (Illumina, FC-131-1024), in which each sample was barcoded, and sequenced using 250 bp paired-end illumina MiSeq technology. Highthroughput sequencing were performed at Gene Core facility (EMBL, Germany).

Deep sequencing data analysis
Paired-end 250 bp illumina sequencing data were initially trimmed using Trimmomatic and subsequently merged using PEAR. clonotypeR toolkit was then used to perform TCR sequence annotation and quantitative analysis in R [52][53][54]. Out of the X raw reads obtained from the 6 samples (3 WT and 3 lamp2 KO), we identified X-Y TCR clonotypes from X-Z productive TCR sequences. For the samples to be comparable, the analyses were performed on 60,000 randomly selected TCR sequences for each data set as it was the lowest number of TCR sequences found in a data set. The presented clonality metric is 1− Pielou's evenness index and can vary from 0 to 1 (more diverse to less diverse). The Pielou's evenness corresponds to the Shannon's entropy (using log 2) for each sample divided by the number of unique clonotypes (in log 2) of the same sample [54,55].

In vitro T cell activation
WT and lamp2 KO-derived CD4 and CD8 T cells were sorted, labeled with 1 μM of CellTrace TM CFSE (carboxyfluorescein diacetate succinimidyl ester; ThermoFisher Scientific, C34554) and cultured in the presence of 1 μg/ml anti-CD3 mAb (clone 145-2C11; BD Pharmingen, 553,057) and 5 μg/ml anti-C28 mAb (clone 37.51; BD Pharmingen, 553,294). The precursor frequency of dividing cells (percentage of cells in the initial population that undergone one or more divisions after culture) was calculated as follows: [∑n ≥ 1(Pn/2 n)]/[∑n ≥ 0(Pn/2 n)], where n is the division number that cells have gone through and Pn is the number of cells in division, as described [56].

Listeria infection
Mice were inoculated with Lm reference strain EGDe as described in [57]. Thymic grafted Nude recipients were intravenously infected 12 weeks post-transplantation, through the tail vein with 2.10 3 colony-forming units (CFUs) of Lm. Mice were rechallenged with 1.10 6 CFUs of Lm 28 days after the first priming. Three days after secondary infection, mice were sacrificed, livers were aseptically removed, homogenized in PBS, and homogenates were serially diluted and plated on BHI-agar (Fisher Scientific, 11,708,872). Lm colonies were enumerated after 24 h incubation at 37°C. For the intracellular IFNG/IFN-γ analysis, total spleen cells were stimulated for 5 h at 37°C, using the Cell Activation Cocktail, containing phorbol 12-myristate 13-acetate, ionomycin and brefeldin A (Biolegend,423,304) according to the manufacturer's instructions. After cell surface staining, cells were prepared according to the supplier's protocol (FOXP3 staining kit; eBioscience, 00-5523-00), stained with PE-labeled anti-IFNG /IFNγ Ab (clone XMG1.2; Biolegend, 505,807) and analyzed by flow cytometry.

Statistical analysis
Analysis was performed using Prism 9.1.1 software (GraphPad Software). The two-tailed Mann-Whitney test was used for statistical differences between groups. For multiple comparisons, a two-way ANOVA was used. p < 0.05 was considered significant.