Central tolerance is impaired in the middle‐aged thymic environment

Abstract One of the earliest hallmarks of immune aging is thymus involution, which not only reduces the number of newly generated and exported T cells, but also alters the composition and organization of the thymus microenvironment. Thymic T‐cell export continues into adulthood, yet the impact of thymus involution on the quality of newly generated T‐cell clones is not well established. Notably, the number and proportion of medullary thymic epithelial cells (mTECs) and expression of tissue‐restricted antigens (TRAs) decline with age, suggesting the involuting thymus may not promote efficient central tolerance. Here, we demonstrate that the middle‐aged thymic environment does not support rapid motility of medullary thymocytes, potentially diminishing their ability to scan antigen presenting cells (APCs) that display the diverse self‐antigens that induce central tolerance. Consistent with this possibility, thymic slice assays reveal that the middle‐aged thymic environment does not support efficient negative selection or regulatory T‐cell (Treg) induction of thymocytes responsive to either TRAs or ubiquitous self‐antigens. This decline in central tolerance is not universal, but instead impacts lower‐avidity self‐antigens that are either less abundant or bind to TCRs with moderate affinities. Additionally, the decline in thymic tolerance by middle age is accompanied by both a reduction in mTECs and hematopoietic APC subsets that cooperate to drive central tolerance. Thus, age‐associated changes in the thymic environment result in impaired central tolerance against moderate‐avidity self‐antigens, potentially resulting in export of increasingly autoreactive naive T cells, with a deficit of Treg counterparts by middle age.


Thymic slice preparation
For 2PM imaging, slices were generated from pCX-EGFP thymi. For negative selection assays, slices were generated from C57BL/6, RIP-mOVA, RIP-OVAhi, RIP-mT4, or RIP-mQ4R7 thymi. Dissected thymi were embedded in 4% (w/v) NuSieve GTG lowmelting-temperature agarose (Lonza) in PBS at 37°C. The solidified agarose block was sectioned into 400-µm-thick slices using a VT 1000 S Microtome (Leica) in a bath of icecold PBS, with vibratome frequency set to 70 Hz, speed to 0.20 mm s -1 , and amplitude to 0.6 mm. Slices were collected in DRPMI + 10% bovine calf serum on ice before transfer to 0.4-µm tissue culture inserts (Millipore) in 35-mm Petri dishes containing 1 mL of complete RPMI medium, with or without added peptides.

Two-photon fluorescence microscopy
After incubation for ³ 1 h, pCX-EGFP thymic slices were transferred and secured in an imaging chamber (Harvard Apparatus) on the microscope stage. Perfusion medium, consisting of DRPMI supplemented with 2 g L -1 sodium bicarbonate, 5 mM HEPES, and 1.25 mM calcium chloride, was gravity fed to the stage inlet through a 300-mL IV set, and circulated through the imaging chamber at a flow rate at ~100 mL h -1 , or ~1 drop per second. The perfusion medium was aerated with 95% oxygen and 5% carbon dioxide and maintained at 37°C with a heated microscope stage and inline perfusion heater. Migratory paths for thymocytes were tracked, and mean cell velocity and path straightness calculated using Imaris (v9, Bitplane). The enrichment of thymocytes in the medulla was determined at the first time point for each dataset by measuring the number of thymocytes in manually demarcated cortical and medullary regions.
For each slice, 10 6 isolated cells were stained with either 2 µM CMTPX CellTracker Red or 2 µM Indo1AM (both from Life Technologies) for 30 min at 37°C in 1.5 mL of DRPMI medium (RPMI 1640 without L-glutamine, phenol red, and sodium bicarbonate; Cellgro) supplemented with 0.2 g L -1 sodium bicarbonate and 20 mM HEPES. Cells were washed and incubated in 1.5 mL complete RPMI medium (RPMI 1640 with 2 mM Lglutamine, 50 U mL -1 penicillin, 50 mg mL -1 streptomycin, and 10% (v/v) fetal bovine serum) for 30 min to destain. Cells were washed, combined so that 10 6 CellTracker Redlabeled cells and 10 6 Indo1AM-labeled cells were mixed into each tube, and washed again with complete RPMI medium. Thymocytes were concentrated into 20-µL complete RPMI medium and carefully pipetted onto the surface of each thymic slice before incubation at 37°C 5% CO2 to allow migration of thymocytes into the thymic slice. To avoid color channel bias, 1MO and 12MO thymocyte fluorophores were swapped in different experiments.
The supernatant was transferred into 35 mL of PBS + 2% bovine calf serum and 5 mM EDTA at 4°C, with digestion of the remaining tissue fragments repeated twice to completely dissociate the tissue. The cells were spun down and filtered through a 70-µm nylon mesh to achieve a single-cell suspension.

Statistics
All statistical analyses were performed using Prism (GraphPad) with the corresponding test and multiple-test corrections listed in the Figure Legends.