Mice harboring the rag1 S723C mutation (rag1^S/S; Giblin et al., 2009) and the lig4 R278H mutation (lig4^R/R; Rucci et al., 2010) were housed at the Karp Family Research Building under specific pathogen-free conditions. Animal experiments were carried out after approval and in accordance with guidelines from the Subcommittee on Research Animal Care of Children’s Hospital Boston, Harvard Medical School.

Single cell suspensions were prepared from thymuses of lig4^R/R, rag1^S/S, and wild-type (WT) mice. Tissues were homogenized on 70 μm cell strainers (BD Falcon, Bedford, MA, USA) using FACS buffer: Dulbecco’s Phosphate-Buffered Saline (D-PBS, GIBCO from Invitrogen, Grand Island, NY, USA) containing 2% of heat inactivated and filtered fetal calf serum (FCS, from Gemini Bio-Products, West Sacramento, CA, USA). Red blood cell lysis was performed at room temperature by adding 2 ml of Red Blood Cell Lysing Buffer (Sigma Aldrich Inc., St Louis, MO, USA) for 10 min before proceeding with the specific stainings.

Thymic stromal cells were isolated as previously published (Gray et al., 2002) by digesting thymuses from lig4^R/R, rag1^S/S, and WT mice with 0.125% (w/v) collagenase D with 0.1% (w/v) DNAse I (both from Roche, Indianapolis, IN, USA) in RPMI 1640 (GIBCO from Invitrogen, Grand Island, NY, USA).

Thymocytes were incubated with the following antibodies: antigen-presenting cells (APC)-conjugated anti-CD4, biotin-conjugated or PE-conjugated anti-CD8, biotin-conjugated anti-CD4, biotin-conjugated anti-B220, biotin-conjugated anti-CD11b, biotin-conjugated anti-Gr1, FITC-conjugated anti-CD44, PE-conjugated anti-CD25, FITC-conjugated anti-CD69, biotin-conjugated anti-Qa2 (all from BD Biosciences, San Jose, CA, USA). Samples stained with biotin-conjugated antibodies underwent additional incubation with PerCp-conjugated streptavidin. Intranuclear staining for Foxp3 was performed using APC-conjugated anti-mouse/rat Foxp3 staining set (eBioscience, San Diego, CA, USA) following manufacturer’s instructions. At least 20,000 alive cells (defined by physical parameters) were acquired on a FACSCalibur system or FACS Canto (BD Biosciences, San Jose, CA, USA) and analyzed with FLOW-JO software (version 8.3; Treestar Inc.).

Analysis of TECs was performed staining single cell suspensions of stromal cells with PerCp-conjugated anti-CD45, PE-conjugated anti-Ly51, FITC-conjugated anti-MHC-II (BD Biosciences, San Jose, CA, USA). Intranuclear staining for Aire was performed after fixing stromal cells labeled with surface markers, by permeabilization with the BD CytoFix/CytoPerm Fixation/Permeabilization kit (BD Biosciences, San Jose, CA, USA). Anti-Aire antibody (5H12) was a kind gift from Dr. H. Scott (Hubert et al., 2008).

Analysis of thymic DCs was performed staining single cell suspensions of stromal cells with the following antibodies: APC-conjugated anti-CD11c, PE-conjugated anti-CD45RA, biotin-conjugated anti-CD3, anti-Ter119, anti-Gr1, anti-F4/80, anti-CD19, anti-CD11b, anti-CD90 (BD Biosciences, San Jose, CA, USA). Analysis was performed after gating out the biotinylated positive population. At least 100,000 alive cells (defined by physical parameters) were acquired on a FACS Canto (BD Biosciences, San Jose, CA, USA) and analyzed with FLOW-JO software (version 8.3; Treestar Inc.).

Formalin-fixed paraffin embedded tissue sections were stained with hematoxylin and eosin (H&E) and subjected in parallel to immunohistochemistry. Briefly, sections were deparaffinized, rehydrated, and endogenous peroxidase activity was blocked in 0.3% H₂O₂/methanol solution for 20 min prior to heat induced antigen retrieval (when necessary) using a thermostatic bath or a microwave-oven in 1.0 mM EDTA (pH 8.00) or 1.0 mM citrate buffer (pH 6.00) respectively. Sections were then washed in TRIS-base buffer at pH 7.4 and incubated for 1 h with the following reagents diluted in TRIS/1% bovine serum albumin (BSA): rabbit anti-CK5 (1:50; Covance, Berkeley, CA, USA), rat anti-CK8 (1:200; clone TROMA-1; kindly provided by Dr. U.H. von Andrian, Harvard Medical School), rabbit anti-claudin 4 (1:100; Zymed Laboratories, San Francisco, CA, USA), rabbit anti-murine Aire (1:3000; kindly provided by Dr. P. Peterson, University of Tartu, Estonia), and biotin-conjugated Ulex europaeus agglutinin-1 (UEA-1; 1:600; Vector Laboratories, Burlingame, CA, USA). After washes, single immunostains were revealed using the ChemMATE Envision Rabbit/Mouse (DAKO Cytomation, Glostrup, Denmark) or NovoLinkTM Polymer Detection System (NovocastraTM Laboratories Ltd, Newcastle Upon Tyne, United Kingdom) followed by diaminobenzidine (DAB) as chromogen and Hematoxylin as counterstain.

The same procedure was applied to double immunofluorescence stainings prior to the incubation with a secondary swine anti-rabbit FITC-conjugated antibody (1:30; DAKO Cytomation, Glostrup, Denmark) for CK5 and a rabbit anti-rat biotinylated antibody (1:200; Vector Laboratories, Burlingame, CA, USA) followed by Streptavidin–Texas Red (1:100; Southern Biotechnology Associates, Birmingham, AL, USA) for CK8. Sections were then counterstained with DAPI.

Images were acquired with an Olympus DP70 digital camera mounted on an Olympus BX60 microscope using Cell^F imaging software (Soft Imaging System GmbH) and Adobe Photoshop Version 7.0 for the artwork.

RNA isolation was isolated from thymus single cell suspensions using the mirVana miRNA isolation kit, according to the manufacturer’s protocol (Ambion from Applied Biosystems Inc., Foster City, CA, USA). Reverse transcription was then performed with qScript cDNA SuperMix (Quanta BioSciences, Inc., Gaithersburg, MD, USA) following manufacturer’s instructions.

Real-Time PCR for quantitative expression of Aire and TSAs was performed using TaqMan Gene expression assay with the following assays on demand (all by Applied Biosystems Inc., Foster City, CA, USA): Mm00477461_m1 (Aire); Mm00487224_m1 (Cybp1a2); Mm00433188_m1 (Fabp2); Mm00731595_gH (Ins2); Mm00493214_m1 (EpCAM, used as internal control). Reactions were performed using 2 μl of cDNA obtained from 1 μg of thymic RNA in a final volume of 20 μl using TaqMan PCR Master Mix 2x (Applied Biosystems Inc., Foster City, CA, USA) and specific Gene Expression Assay 20x. Amplification was performed in duplicates in the 7500 Real-Time PCR System (Applied Biosystems Inc., Foster City, CA, USA) and results were analyzed with the 7500 Real-Time PCR Software.

Analysis of T cell repertoire in the thymus of WT, rag1^S/S and lig4^R/R mice was performed as previously described (Rucci et al., 2010).

At least five to six mice per group were analyzed. Results are indicated as mean ± SE or SD. p values were determined by unpaired Student’s t-test (p < 0.05 = *;p < 0.01 = **;p < 0.005 = ***;p < 0.001 = ****).

The lig4^R/R and rag1^S/S mice carry mutations that impair different steps in the V(D)J recombination process. In particular, rag1 mutations interfere with the first step of the process, when DSBs are introduced in the DNA (Fugmann et al., 2000. In contrast, lig4 mutations impair the repair of these breaks (Rooney et al., 2004). Despite these differences, both mutations affect V(D)J recombination and cause an early but incomplete block in T cell development.

Both lig4^R/R and rag1^S/S mice showed a significant reduction in thymus size and cellularity (Figure 1A), with a significant decrease in the absolute numbers of thymic CD4⁺ CD8⁺ DP cells as well as of CD4⁺ CD8⁻ and CD4⁻ CD8⁺ SP thymocytes as compared to what observed in WT mice (Figure 1B).

In both mutant models, this block in T cell development was associated with a relative enrichment in the proportion of CD4⁻ CD8⁻ double negative (DN)cells, that was particularly prominent in rag1^S/S mice (% DN cells ± SE: WT 1.78 ± 0.13; lig4^R/R 25.48 ± 3.5; rag1^S/S 90.59 ± 3.27; WT vs. lig4^R/R, p < 0.0001; WT vs. rag1^S/S, p < 0.0001). Developmental progression of DN thymocytes is characterized by an ordered sequence of expression of CD44 and CD25 markers: CD44⁺ CD25⁻ (DN1), CD44⁺ CD25⁺ (DN2), CD44⁻ CD25⁺ (DN3), CD44⁻ CD25⁻ (DN4). Analysis of the distribution of DN thymocytes in lig4^R/R and rag1^S/S mice revealed a severe, but incomplete arrest of thymocyte development at DN3 stage in both models, consistent with a defect in TCRβ rearrangement due to impaired V(D)J recombination (Figure 1C).

Newly generated CD4⁺ and CD8⁺ SP thymocytes undergo sequential stages of maturation in the medulla. The developmental program of maturing SP thymocytes is associated with progressive down-regulation of CD69 and up-regulation of the Qa2 markers on the cell surface (Lucas et al., 1994; Jin et al., 2008).The earliest stages in SP cell development (SP1–SP2) are characterized by expression of CD69 but lack of Qa2 marker (CD69⁺ Qa2⁻ cells). The next step of maturation of SP thymocytes (SP3 stage) is marked by lack of expression of either marker (CD69⁻ Qa2⁻ cells). Finally, in the last stage of differentiation (SP4), mature SP thymocytes acquire expression of Qa2, and hence have a CD69⁻ Qa2⁺ phenotype. Thymocytes that reach this stage are ready to egress from the thymus (Ge and Chen, 1999; Jin et al., 2008). In control mice, the majority of CD4⁺ SP thymocytes are in the SP1–SP2 stages of differentiation; in contrast, both lig4^R/R and rag1^S/S mice showed a significant increase in the proportion of CD4⁺ SP thymocytes with the most mature (SP4) phenotype (Figure 2). A similar pattern was observed for CD8⁺ SP thymocytes (data not shown).This difference in the distribution of SP thymocytes at various stages of differentiation may reflect several, non-mutually exclusive mechanisms, including accelerated intrathymic T cell maturation, homeostatic T cell proliferation in a lymphopenic environment (Datta and Sarvetnick, 2009) and recirculation of mature T lymphocytes that home back to thymus. In order to distinguish between these possibilities, we have analyzed the distribution of SP thymocytes at late stages (18 ± 2 days) of fetal development, when homeostatic proliferation and recirculation of mature lymphocytes should not prevail. As shown in Figure A1 in Appendix, even at this stage of fetal development, SP thymocytes from lig4^R/R and rag1^S/S mice were characterized by a more mature (SP3) phenotype than SP thymocytes from age-matched WT mice. Overall, these data indicate that intra thymic maturation of SP thymocytes is accelerated in lig4^R/R and rag1^S/S mice.

Hypomorphic mutations that affect V(D)J recombination may affect not only the number, but also the TCR repertoire diversity of newly generated thymocytes. We have previously demonstrated that hypomorphic RAG mutations in patients with OS are associated with oligoclonal representation of TCR specificities in the thymus (Signorini et al., 1999). Similarly, a highly restricted T cell repertoire was demonstrated in the thymus of rag1^S/S mice (Figure 3). In contrast, homozygosity for the R278H lig4 mutation allowed generation of a broadly polyclonal repertoire of thymocytes (Figure 3). These differences in size and diversity of the thymocyte pool prompted us to investigate thymus morphology in lig4^R/R and rag1^S/S mice.

Appropriate interaction between elements of the T cell lineage and stromal cells is crucial to maintain thymic architecture and to support the maturation of both TECs and nascent thymocytes. Severe and early blocks in T cell development are associated with lack of thymic cortico-medullary demarcation (Holländer et al., 1995a). Furthermore, inability to maintain an organized thymic architecture may interfere not only with an orderly maturation of thymocytes, but may also impede establishment of self-tolerance (Holländer et al., 1995b; Derbinski and Kyewski, 2005).

Based on this, we investigated in detail the thymic architecture of lig4^R/R and rag1^S/S mice. As shown in Figure 4, staining of thymic tissue with H&E revealed severe depletion of cellularity in both lig4^R/R and rag1^S/S mice. Cortico-medullary demarcation was preserved in lig4^R/R mice, whereas only a rudimentary attempt to form a medulla was noticed in rag1^S/S mice. Differential expression of cytokeratin 8 (CK8) and cytokeratin 5 (CK5) allows distinction between CK8⁺ CK5⁻ cTECs and CK8⁻ CK5⁺ mTECs (Bennett et al., 2002; Takahama, 2006). Analysis of CK5 and CK8 expression by immunohistochemistry and immunofluorescence confirmed significant differences in the degree of thymic architecture abnormalities in lig4^R/R and rag1^S/S mice. In particular, only few CK5⁺ cells were detected in rag1^S/S mice; furthermore, these cells were largely also CK8⁺, a pattern observed in immature TEC progenitors (Bennett et al., 2002). In contrast, tiny but well-defined nests of CK5⁺ CK8⁻ mTECs were appreciated in lig4^R/R mice, consistent with what detected by H&E staining.

The thymic medulla plays an essential role in the tolerance to peripheral antigens. Both mature mTECs and thymic DCs have been implicated in mediating central tolerance by presenting nascent thymocytes with a broad repertoire of TSAs whose expression by mTECs is controlled by the transcription factor Aire (Derbinski et al., 2005; Kyewski and Klein, 2006).

Maturation of mTECs is progressively marked by the expression of claudin-4 (Cld4) and the ligand for UEA-1. Furthermore, terminally differentiated mTECs express high levels of MHC-II molecules and a subset of them also express the transcription factor Aire and TSAs (Hamazaki et al., 2007). Residual presence of Cld4⁺ and of UEA-1⁺ mature mTECs was detected in the thymus of lig4^R/R mice; in contrast, there was no expression of these markers in the thymus from rag1^S/S mice, in keeping with our previous observations that they lack a well-defined thymic medulla (Figure 5). Staining with anti-Aire antibody revealed a relative abundance of Aire⁺ cells in the thymic medulla of WT mice. Aire⁺ cells were detected also in lig4^R/R mice, albeit in low number; in contrast, dramatic depletion of Aire⁺ cells, that were confined to focal areas of cortico-medullary demarcation, was demonstrated in rag1^S/S mice (Figure 5).

Next, we used quantitative real-time polymerase chain reaction (qPCR) to analyze the levels of mRNA specific for Aire and for Aire-dependent TSAs (insulin, cytochrome p450, and fatty acid binding protein) in the thymus of WT, lig4^R/R, and rag1^S/S mice. A significant reduction in the level of these transcripts was observed in the thymus of both mutant mice; this reduction was particularly pronounced in rag1^S/S mice (Figure 6A).

The observed reduction of Aire and TSA mRNA expression in the thymus of lig4^R/R and rag1^S/S mice could reflect either impairment of terminal maturation of mTECs or a general depletion of the mTEC compartment. To distinguish between these two possibilities, we used flow cytometry. It has been shown that cTECs and mTECs can be distinguished based on the expression of MHC-II and Ly-51 markers within the CD45⁻ population of thymic stromal cells (Hubert et al., 2008). Both cTECs and mTECs express MHC-II, but only cTECs express Ly51. Furthermore, based on the levels of expression of MHC-II, it is also possible to discriminate between immature (MHC-II^low Ly-51⁻)and mature MHC-II^hi Ly-51⁻ mTECs. Upon staining of CD45⁻ thymic stromal cells for Ly-51 and MHC-II, we found that the thymuses of both lig4^R/R and rag1^S/S mice were significantly depleted of mature mTECs (% MHC-II^hi mTECs ± SE: WT = 2.87 ± 0.8; lig4^R/R = 1.12 ± 0.4; rag1^S/S = 0.17 ± 0.04; WT vs. lig4^R/R p < 0.005; WT vs. rag1^S/S, p < 0.005). In contrast, both lig4^R/R and rag1^S/S mice showed a relative enrichment in thymic cTECs (% cTECs ± SE: WT = 19.49 ± 3.31; lig4^R/R = 34.36 ± 2.92; rag1^S/S = 33.51 ± 2.23; WT vs. lig4^R/R, p < 0.005; WT vs. rag1^S/S, p < 0.005; Figure 6B). However, this apparent enrichment in cTECs may also reflect the increased number of immature TECs (as demonstrated by co-expression of CK5 and CK8) and/or an increase in fibroblasts or other stromal CD45⁻ Ly51⁺ cells, as reported in other murine models of impaired T cell development (Gray et al., 2002; Rodewald, 2008; Alves et al., 2009).

As mentioned above, fully mature MHC-II^hi mTECs are enriched for Aire expressing cells.In spite of the general reduction in the proportion of mature mTECs, we found that the few CD45⁻ MHC-II^hi Ly51^low mTECs from lig4^R/R and rag1^S/S mice retained the ability to express Aire (% Aire⁺ cells ± SE: WT = 43.05 ± 2.45; lig4^R/R = 48.12 ± 5.11; rag1^S/S = 47.05 ± 5.4; Figure 6B), indicating that the overall impairment of Aire and TSA expression in these mutant models is due to a reduction of the pool of mature mTECs rather than to intrinsic defects in their developmental and gene expression program.

Thymic DCs are the other population of APCs involved in the negative selection of self-reactive thymocytes. Furthermore, a role for thymic DCs in the induction of nTreg cells has been suggested both in mice and in humans (Proietto et al., 2008; Doan et al., 2009). Thymic DCs can be classified into two major subsets: the CD11c⁺ CD45RA⁻ conventional DCs (cDCs) and the CD11c^int CD45RA⁺ plasmacytoid DCs (pDCs). The first subgroup is intrathymically generated from early thymic progenitor cells, whereas pDCs arise extrathymically from partially differentiated precursors (Proietto et al., 2009).

In order to analyze the distribution of thymic cDCs and pDCs in WT, lig4^R/R, and rag1^S/S mice, we stained thymic single cells suspensions with a mixture of monoclonal antibodies against markers specific for T and B lymphocytes, erythroid cells, granulocytes, and macrophages (CD3, CD90, CD19, TER119, Gr-1, CD11b, and F4/80). Upon gating on cells that stained negative for this cocktail of antibodies, we analyzed expression of CD11c and CD45RA to discriminate between cDCs and pDCs. As shown in Figure 7, in WT mice the majority of thymic DCs is composed of cDCs. Although cDCs were more abundant than pDCs also in the thymus of lig4^R/R and rag1^S/S mice, both mutant strains showed a significant enrichment for pDCs as compared to WT mice (Figure 7).

To investigate whether the profound abnormalities of thymic stroma observed in lig4^R/R and rag1^S/S mice could also affect generation of nTreg cells, we analyzed expression of CD25 and Foxp3 within CD4⁺ SP thymocytes. As shown in Figure 8, both lig4^R/R and rag1^S/S mice showed preserved ability to express Foxp3 within CD4⁺ SP thymocytes, and the proportion of Foxp3⁺ cells was actually increased in lig4^R/R mice (% Foxp3⁺ cells ± SD: WT = 4.72 ± 1.3; lig4^R/R = 24.2 ± 3.5; rag1^S/S = 6.25 ± 3.3; WT vs. lig4^R/R p < 0.0001; WT vs. rag1^S/S p < 0.0001; Figure 8). However, it should be noted that both mutant strains display a severe depletion of thymic cellularity, and of CD4⁺ SP T cells in particular (Figure 1). This has obvious implications also on the absolute number of nTreg cells, in particular in lig4^R/R mice. By using immunohistochemistry, we found that a residual number of Foxp3⁺ cells at the cortico-medullary junction were present in lig4^R/R mice, whereas such cells were severely depleted in rag1^S/S mice (data not shown).