Mice

Inbred male and female C57BL/6J (000664) and B6 CD45.1 (002014) mice were obtained from The Jackson Laboratory. Il1r1−/− (003245), Il18−/− (004130), Il18r1−/− (004131), Cas1Δ10 (032662), Cd1d−/− (008881), Ifngr−/− (003288), Prf1−/− (002407), Rag2-eGFP (005688) and GREAT (IFNγ reporter with endogenous poly(A) transcript) mice were obtained from The Jackson Laboratory and bred in-house. Gzmb−/− mice were obtained from G. Hill (Fred Hutchinson Cancer Center) and bred in-house. Il18 flox mice (Il18fl/fl) were obtained from R. Nowarski (Harvard Medical School) and R. Flavell (Yale School of Medicine) and crossed in-house with Zbtb46-Cre+ mice obtained from The Jackson Laboratory (032662) to generate Il18fl/fl:Zbtb-Cre+ (Il18ΔcDC) mice. Il18r1fl/fl mice were obtained from G. Trinchieri (National Cancer Institute) and crossed with Foxn1-Cre+ mice obtained from The Jackson Laboratory (018448) and Ncr1-Cre+ mice obtained from K. Barry (Fred Hutchinson Cancer Center) to generate Il18r1fl/fl:Foxn1-Cre+ (Il18r1ΔTEC) and Il18r1fl/fl:Ncr1-Cre+ (Il18r1ΔNK) mice, respectively. Ifngrfl/fl (025394) and Foxn1-Cre+ (018448) mice were obtained from The Jackson Laboratory and were crossed to generate Ifngrfl/fl:Foxn1-Cre+ (IfngrΔTEC) mice. All experimental mice were between 6 and 10 weeks old. Mice were maintained at the Fred Hutchinson Cancer Research Center and acclimatized for at least 2 days before experimentation, which was performed according to the Institutional Animal Care and Use Committee guidelines.

Cell isolation

Single-cell suspensions of freshly dissected thymuses were obtained and enzymatically digested using 0.15% collagenase D (Sigma 11088882001) and 0.1% DNase I (Sigma 10104159001) in DMEM, as previously described8. Cellularity was calculated using the Z2 Coulter Particle and Size Analyzer (Beckman Coulter). For studies sorting rare populations of cells in the thymus, multiple identically treated thymuses were pooled to isolate sufficient numbers of cells; however, in these instances, separate pools of cells were established to maintain individual samples as biological replicates. The bone marrow was flushed from the femurs and tibias and then passed through a 70-μm filter. Peripheral blood samples were collected into EDTA capillary pipettes (Fisher Scientific). Red blood cell lysis was performed using ACK lysis buffer (A1049201, Fisher Scientific).

Flow cytometry

Cells were stained with antibodies to the following proteins for analysis: CD45 (565967, BD Biosciences), CD31 (102434, BioLegend), CD140a (135907, BioLegend), MHC-II (107620, BioLegend), EpCAM (46-5791-82, BD Biosciences), Ly51 (740882, BD Biosciences), UEA-1 (ZC0426, Vector Laboratories), DLCK-1 (NBP1-77127F, Novus Biologicals), Ly6D (138605, BioLegend), CD104 (123615, BioLegend), CD140a (135921, BioLegend), CD31 (102427, BioLegend), PDPN (127425, BioLegend), CD8a (100714, BioLegend), CD4 (565709, BD Biosciences), TCRβ (109239, BioLegend), CD3ε (100232, BioLegend), CD25 (102030, BioLegend), CD44 (612799, BD Biosciences), NK1.1 (108753, BioLegend), CD49b (561067, BD Biosciences), c-Kit (105811, BioLegend), TCRγδ (118107, BioLegend), CD1d PBS-57 tetramer (National Institutes of Health Tetramer Core), CD11c (35-0114, Tonbo), CD11c (612796, BD Biosciences), CD11b (741722, BD Biosciences), XCR1 (148225, BioLegend), B220 (103232, BioLegend), CD127 (50-1271, Tonbo), Sca-1 (122527, BioLegend), CD135 (135305, BioLegend), CD150 (46-1502-82, eBioscience), CD48 (103427, BioLegend), NKG2D (562800, BD Biosciences), CD49a (741976, BD Biosciences), KLRG1 (138425, BioLegend), CCR-6 (129814, BioLegend), IL-23R (150907, BioLegend), ST2 (566310, BD Biosciences), H-2Kb (116525, BioLegend), IL-18R (25-5183-82, Thermo Fisher), IL-18R (25-5183-82, Thermo Fisher), RAE-1 (130-111-467, Miltenyi Biotec) and streptavidin-APC high concentration (405243, BioLegend). Following fixation and permeabilization (554714, BD Biosciences), cells were stained with antibodies to perforin (154315, BioLegend) and granzyme B (MHGB04, Thermo Fisher). Annexin V and 7-AAD staining (640920, BioLegend) was performed in Annexin V binding buffer (422201, BioLegend). Flow cytometric analysis was performed on a Symphony S6 instrument (BD Biosciences), and cells were sorted on an Aria II cell sorter (BD Biosciences) using FACSDiva (BD Biosciences) or FlowJo (TreeStar) software.

In vivo acute damage models

To induce thymus damage, we subjected mice to SL-TBI at a dose of 550 cGy from a Cs-137 γ-radiation source without hematopoietic rescue. Other models of thymus damage included i.p. injection of 20 mg kg−1 dexamethasone (Sigma-Aldrich D2915), 200 mg kg−1 cyclophosphamide (University of Washington Medical Pharmacy) and 1.5 mg kg−1 LPS (InvivoGen tlrl-eblps). For in vivo studies of rIL-18 administration, C57BL/6J or Ifng-GFP mice were administered 2.5 mg kg−1 rIL-18 (s.c.) either in the absence of other thymus-damaging treatments (day 0) or at 3 days after SL-TBI.

In vivo depletion and transplantation studies

To perform NK1.1+ cell depletion studies, we injected mice with 200 μg (i.p., 10 mg kg−1) of anti-NK1.1 mAb (BioXCell BE0036) on days −1, 1 and 3 following SL-TBI. B6 HCT recipients were subjected to 1,100 cGy TBI (2 × 550 cGy) before transplantation; then, within 24 h, they received an i.v. injection of 5 × 106 to 10 × 106 bone marrow cells. For IL-18 abrogation experiments, mice were dosed with 200 μg (i.p., 10 mg kg−1) of anti-IL-18 mAb (BioXCell BE0237) on days −1, 1, 3, 6, 9, 12, 15 and 18 following transplantation.

Parabiosis

Female C57BL/6 CD45.1+ and CD45.2+ mice were surgically conjoined to establish parabiotic pairs using a modified protocol as previously described12. Briefly, mice were cohoused for 10 days before parabiosis surgery and maintained in a parabiotic state until both members of the pair were subjected to SL-TBI (550 cGy) on experimental day 21, 24 or 27, corresponding to 7, 4 or 1 day(s) before tissue collection. All mice were killed on day 28 after surgery. To distinguish between circulating and tissue-resident cells at the time of collection, we administered 3 μg of anti-CD45 antibody (APC-EF780, BioLegend) into the mice by retro-orbital injection 3 min before killing. Thymuses were collected from both parabionts and analyzed by flow cytometry. Circulating cells labeled by i.v. administration of anti-CD45 antibody were excluded from analysis. Congenic markers (CD45.1 and CD45.2) were used to determine the origin of thymic cells: cells expressing the same CD45 isoform as the assessed mouse-pair thymus donor were classified as intrathymically derived, whereas cells expressing the alternate isoform were considered extrathymically derived.

Protein quantification

For the detection of active IL-1β, active IL-18, granzyme B, perforin, IFNγ and mature Cas-1 (Figs. 1b,c and 5c and Extended Data Fig. 1) in supernatants, thymic tissue was mechanically dissociated in defined volumes of buffer. The resulting supernatant was analyzed using cytokine-specific ELISA kits (IL-1β, Invitrogen #88-7013-22; IL-18, Thermo Fisher #BMS618-3; granzyme B, R&D #DY1865; perforin, Novus Biologicals #NBP3-00452; IFNγ, Thermo Fisher #KMC4021; mature Cas-1, Adipogen #AG-45B-0002-KI01), and absorbance was measured on a Spark 10M plate reader (Tecan).

For the detection of active IL-18 and IL-18BP (Fig. 1b: only IL-18 after cyclophosphamide treatment; Figs. 1d and 2d,e) in whole organs, thymuses were homogenized in RIPA buffer (25 mM Tris (pH 7.6), 150 nM NaCl, 1% NaCl, 1% NP-40, 0.1% SDS, 0.05% sodium deoxycholate, 0.5 mM EDTA) with protease inhibitors (Thermo Fisher A32955) using a Homogenizer 150 (Fisher Scientific) and normalized by mass at a concentration of 20 mg thymus tissue per ml of RIPA buffer. The resulting lysates were analyzed using cytokine-specific ELISA kits (IL-18, Thermo Fisher #BMS618-3; IL-18BP, Abcam ab254509), and absorbance was measured on the Spark 10M plate reader (Tecan).

In vitro cell culture

Coculture experiments were performed by plating 50,000 ex vivo FACS-purified bone marrow Lin− selected or Lin−Sca-1+c-Kit+ FACS-purified cells onto six-well plates confluent with OP9-DLL1 cells in OP9 medium, as previously described17,65. Cocultures were performed in the presence of 5 ng ml−1 Flt-3L (Peprotech, 250-31L) and 1 ng ml−1 IL-7 (Peprotech 217-17), along with either 0, 1 or 10 ng ml−1 rIL-18 (BioLegend 767008). Equal volumes of nonadherent cells were assessed by flow cytometry for differentiation at 10, 14 and 21 days following coculture.

Cytotoxicity assays

Cytotoxicity assays of thymic NK cells (Figs. 5f and 6h) were performed by coculturing thymus-derived FACS-purified NK1.1+IL-18R+CD49b+TCRβ− NK cells with CellTrace Yellow (Thermo Fisher C34573)-labeled RMA-S cells at either a 2:1 or 5:1 effector-to-target ratio in RPMI/10% FBS supplemented with 10 ng ml−1 rIL-15 (BioLegend 566302). Splenic FACS-purified NK1.1+IL-18R+CD49b+TCRβ− NK cells derived from C57BL/6 mice treated with 0.3 mg poly(I:C) (i.p.) (InvivoGen tlrl-picw) 1 day earlier served as positive controls. Cocultures were incubated at 37 °C for 5 h, after which cell death of CellTrace Yellow-labeled RMA-S cells was assessed by flow cytometry according to the expression of Annexin V (BioLegend 640920).

Cytotoxicity assays of thymus-derived TECs as target cells (Fig. 7c) were performed by magnetically enriching thymus-derived CD45− cells and coculturing them with CellTrace Violet (Thermo Fisher C34571)-labeled splenic NK-enriched cells (Miltenyi Biotec 130-115-818) derived from C57BL/6 mice treated with 0.3 mg poly(I:C) (i.p.) (InvivoGen tlrl-picw) 1 day earlier at a 4:1 effector-to-target ratio in RPMI/10% FBS supplemented with 10 ng ml−1 rIL-15 (BioLegend 566302). Cocultures were incubated at 37 °C for 5 h, after which cell death of CD45−EpCAM+MHC-II+Ly51+ cTECs and CD45−EpCAM+MHC-II+UEA-1+ mTECs was assessed by flow cytometry based on the expression of Annexin V and 7-AAD (BioLegend 640920).

Imaging

Formalin-fixed, paraffin-embedded tissues were cut into 4-μm sections, mounted onto positively charged slides and baked for 1 h at 60 °C. The slides were then dewaxed and stained on the BOND RX stainer (Leica) using Leica BOND reagents for dewaxing (Dewax Solution), antigen retrieval and antibody stripping (Epitope Retrieval Solution 2), and rinsing after each step (BOND Wash Solution). The antigen retrieval and antibody stripping steps were performed at 100 °C, while all other steps were conducted at ambient temperature. Endogenous peroxidase was blocked with 3% H2O2, followed by protein blocking with TCT buffer (0.05 M Tris, 0.15 M NaCl, 0.25% casein, 0.1% Tween-20, 0.05% ProClin 300, pH 7.6). Primary antibodies (rabbit polyclonal anti-mouse keratin-14, BioLegend 905301; rat anti-mouse keratin-8, Troma-I Developmental Studies Hybridoma Bank; rabbit anti-mouse NKp46, Abcam 233558) were applied sequentially, followed by the application of the secondary antibodies and the tertiary TSA amplification reagent (Akoya OPAL fluorophore). A high-stringency wash was performed after the secondary and tertiary antibody applications using high-salt TBST solution (0.05 M Tris, 0.3 M NaCl and 0.1% Tween-20, pH 7.2–7.6). Species-specific polymer HRP was used for all secondary antibody applications, including either anti-rabbit HRP (Akoya Opal) or goat anti-rat IgG polymer detection kit (Vector ImmPress). Following the application of the final antibody, the slides were stained with DAPI and coverslipped with Prolong Gold Antifade reagent (Invitrogen/Life Technologies). Slides were cured at room temperature, and whole-slide images were acquired on the Vectra Polaris Quantitative Pathology Imaging System (Akoya Biosciences), spectrally unmixed using Phenoptics Inform software and exported as multi-image TIF files. Tiles were fused, and cellular analysis of the images was performed using the HALO image analysis software (Indica Labs). The cells were first identified based on nuclear recognition of the DAPI stain, and membrane segmentation was assisted by referencing the two cytokeratin stains. Thresholds were set to identify positive cells based on the mean intensity within the cytoplasmic and membrane regions of each cell. Cortical and medullary regions were defined by a random forest classifier, followed by a manual review. Cell populations were quantified within each region, and a nearest-neighbor analysis was performed to determine spatial relationships and provide measurements between cells.

scRNAseq and qPCR

Previously generated and published scRNAseq datasets of thymic CD45− nonhematopoietic cells (GSE240016; 50,890 cells) and Rag2GFP CD45+ hematopoietic cells (GSE244673; 37,879 cells) from 2-month-old mice at steady state and on days 1, 4 and 7 after SL-TBI were used for this study12,28. The CD45− dataset can be viewed at https://thymosight.org/, along with all previously published thymus single-cell sequencing datasets. CellChat (v1.4.0)66 was used with default parameters to predict cell–cell interactions between all subsets using the combined dataset at steady state and on days 1, 4 and 7 after damage, focusing on the IL-18 signaling pathway. Aggregate signal strength was calculated for each IL-18 target cell by combining the CellChat signal quantification for each IL-18 source to an individual target. RNA was extracted from sorted cells using the RNeasy Plus Micro kit (74034, Qiagen). cDNA was synthesized using the iScript gDNA Clear cDNA Synthesis kit (1725035, Bio-Rad) and a Bio-Rad C1000 Touch ThermoCycler (Bio-Rad). RNA expression was assessed using the Bio-Rad CFX96 Real Time System (Bio-Rad), with iTaq Universal SYBR Green Supermix (1725122, Bio-Rad) and the Il18 primer (qMmuCED0061252, Bio-Rad).

Statistics

Statistical analysis between two groups was performed using an unpaired two-tailed t test (Figs. 1c,g–i, 3d,h,j–k, 4f–j, 5b–d,f, 6a,e,f,h–j and 7b–f and Extended Data Figs. 1c, 5e,f, 6b,d,f–i and 7a–d) or a paired two-tailed t test (Fig. 6b,d and Extended Data Fig. 6e). Statistical comparisons among three or more groups in the figures were performed using a one-way ANOVA with Dunnett’s multiple comparison test (Figs. 1a,b,d–f and 6c and Extended Data Figs. 1a and 6c) or Tukey’s multiple comparison test (Figs. 2d,e, 4c–d and 7g and Extended Data Fig. 1b). All statistics were calculated using GraphPad Prism, and display graphs were generated in either GraphPad Prism or R. Information on replicates, error bars and statistical significance can be found in the figures and their corresponding legends.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.