Thymic Involution Perturbs Negative Selection Leading to Autoreactive T Cells That Induce Chronic Inflammation

Chronic inflammation is a ubiquitous feature of the aging process and implicated in virtually every age-related disease (1, 2). The term “inflammaging” describes the low-level, chronic, and systemic proinflammatory state that accompanies advanced age in the absence of infection (1). Even though clinical manifestations are not obvious, the presence of proinflammatory factors such as IL-6, TNF-α, IL-1, and C-reactive protein are associated with the severity, incidence, and mortality of cardiovascular diseases such as atherosclerosis and myocardial infarction (3), neurodegenerative diseases such as Parkinson’s disease (4) and Alzheimer’s disease (1), and late-life cancers including colitis-associated colon cancer and hepatocellular carcinoma (5). IL-6 and TNF-α are the most predictive inflammatory biomarkers and are highly correlated with “all-cause” morbidity and mortality in the elderly (6). Although the etiology of inflammaging is not fully understood, the source of proinflammatory factors is primarily attributed to the combination of cellular senescence–induced senescence-associated secretory phenotype (SASP) and the persistent activation of immune cells (1, 2, 7).

The persistent activation of immune cells is hypothesized to result from chronic cell death and self-debris serving as damage-associated molecular patterns, leading to innate immune cell activation (1, 8) or repeated life-long exposure to latent viral infections, such as CMV (9, 10), continuously activating both innate and adaptive immune responses. However, it is not known whether thymic involution can lead to the persistent activation of T cells that are capable of inducing inflammaging.

The thymus is a primary lymphoid organ comprised of cortical and medullary thymic epithelial cells (cTECs and mTECs) responsible for the development of thymocytes and the generation of central immune tolerance toward self-tissues. Thymic-driven central immune tolerance is accomplished through two mechanisms: the elimination of autoreactive T cell clones via the process of negative selection (11), and the generation of natural regulatory T cells (nTregs) via diverted differentiation (12, 13). These two processes depend on the TCR–self-peptide–MHC avidity and signal strength, where a weak signal leads to thymocyte survival (14), a strong signal leads to clonal deletion (15), and a moderate signal plus cytokines (IL-2) leads to nTreg differentiation (16, 17). The efficiencies of negative selection and the differentiation of nTregs are dependent on the production and presentation of tissue-specific Ags on MHC, which is, in part, regulated by the autoimmune regulator gene (Aire) in mTECs (18–22).

The thymus undergoes a progressive and age-related involution attributed to the deterioration of the thymic microenvironment (23), which is made of an integrated three-dimensional meshwork of cTECs and mTECs, where TEC differentiation is regulated by the Foxn1 gene (24). It has been reported that defects in mTEC structure and the loss of Aire can affect the maintenance of central immune tolerance (25–27) by leading to the generation of fewer (28) or deficient nTregs (29), and thereby increasing the incidence of autoimmune disease. However, the mechanisms through which thymic involution impacts the two mechanisms of central tolerance (negative selection and nTregs) are not fully understood. Furthermore, whether thymic atrophy alone leads to the release of autoreactive T cells that become persistently activated immune cells and contribute to inflammaging remains unclear.

In this study, we focus on the involvement of thymic involution in inflammaging by utilizing a loxp-Foxn1–ubiquitous promoter-driven Cre-recombinase and estrogen-receptor fusion protein (uCreER^T) conditional knockout (F-cKO) mouse model, which induces accelerated thymic involution of a fully matured thymus while maintaining a young periphery (30), and a naturally aged C57BL/6 mouse model. The F-cKO mouse model allows for the inducible deletion of Foxn1 after the thymus has fully matured, either by administering tamoxifen (TM) or the slow leakage of uCreER^T, resulting in accelerated epithelial-driven thymic atrophy that is comparable with thymic epithelium dysfunction observed in naturally aged C57BL/6 mice (24, 30). Although the slow leakage of uCreER^T results in weak deletion of genomic Foxn1 at ∼1 mo of age (24), observable biological effects including the loss of Foxn1 expression, thymic involution, mTEC disruption, and thymic dysfunction do not become apparent until ∼3–9 mo of age (24) or until induced with the administration of TM (30). We demonstrate that thymic involution disrupts central immune tolerance and results in the release of autoreactive T cells to the periphery. Furthermore, shortly after thymic egress, these autoreactive T cells gain the activated immune cell phenotype and induce systemic low-grade inflammation that is indicative of inflammaging. Finally, we determined that the mechanism responsible for the thymic involution–driven breakdown of immune tolerance results from perturbed negative selection and a reduction in the mTEC expression of Aire rather than defects in the generation of Tregs. Taken together, these results identify thymic involution as a contributing source of inflammaging and a potential therapeutic target for age-related chronic inflammation.

All animal experiments were in compliance with protocols approved by the Institutional Animal Care and Use Committee of the University of North Texas Health Science Center, in accordance with guidelines of the National Institutes of Health. Various gene-manipulated mouse colonies (all on a C57BL/6 background) and their crossbreeding schemes are listed in Supplemental Table I. They are the Foxn1 cKO (fx/fx-uCreER^T mice with induced Foxn1 deletion via TM treatment, termed F-cKO) (30), fx/fx-only (without uCreER^T, same as wild-type [WT] in Foxn1 expression, termed FF-control [Ctr]) (30), Rag2-GFP reporter, RAG^−/− knockout, membrane-bound ovalbumin (mOVA) transgenic (Tg), OT-II TCR Tg, and autoimmunity regulator gene (Aire)^−/− knockout mice. Approximately one- to 2-mo-old F-cKO mice display very faint deletion of Foxn1 exons 5 and 6 as detected by PCR, but they do not differ from fx/fx-only control mice in Foxn1 expression, mTEC maturation, thymic size, and other factors (24). Following induced Foxn1 deletion via TM, ∼1- to 2-mo F-cKO mice display very strong deletion of Foxn1 exons 5 and 6 and undergo accelerated thymic involution (30). Mouse ages are indicated in each figure legend, defined as young (1–2 mo old) and aged (18–22 mo old) groups. Aged WT mice were purchased from the National Institute on Aging.

Erythrocyte-depleted spleen cells from young and aged WT mice or young F-cKO mice were i.v. injected through the retro-orbital route into young RAG^−/− mice (2.5 × 10⁷ cells per recipient mouse). Sixty days after the injection, the representative tissues (such as the lung, liver, salivary gland, as well as sera) of recipient RAG^−/− mice were collected for analysis of inflammatory cell infiltration.

The surgical operation of the kidney capsule transplantation was performed as previously described (31). Intact newborn mouse thymic lobes of fx/fx-uCreER^T and fx/fx-only with and without mOVA-Tg were directly transplanted into young host OT-II⁺ TCR-Tg mice. Three days after the graft, the host mice were i.p. injected with TM (1 mg/10 g body weight/d) for 3 consecutive days to induce deletion of the Foxn1 gene. Two weeks after the last TM injection, the grafted thymi were isolated for FACS analysis of CD4 and CD8, as well as the TCR-Tg (Vα2Vβ5) marker.

The fx/fx-uCreER^T (F-cKO) or fx/fx-only (FF-Ctr) mice (6 wk old) were given 3× TM i.p. injections to induce deletion of the Foxn1 gene. Four weeks after the last TM injection, mice were immunized by s.c. injection of 100 μg IRBP (aa 294–306) P2 peptide emulsified in 100 μl CFA. Ten days following immunization, cells from lymph nodes and spleen of the mice were harvested for IRBP–P2–I-A^b–tetramer (allophycocyanin labeled) enrichment with anti-allophycocyanin microbeads and MACS columns (Miltenyi Biotec), according to published protocols (32). Positively selected cells were counted and then stained with Abs for flow cytometry. P2–I-A^b tetramer was generated by the National Institutes of Health Tetramer Core Facility and provided by Dr. Mark Anderson (University of California, San Francisco).

Single-cell suspensions were prepared from the thymus and spleen of mice using a 70-μm cell strainer. Spleen cells were erythrocyte depleted with RBC lysing buffer (Sigma-Aldrich, catalog no. R7757) and washed with staining buffer. Samples were then treated with Fc receptor blocking Ab 2.4G2. Samples were then stained with specific Ab of cell surface cluster of differentiation markers and or fixed with 2% paraformaldehyde and permeabilized with Triton X-100, as previously reported (33), followed by intracellular staining for Bim (Cell Signaling Technology, no.2819s), Ki67, Foxp3 (eBioscience, kit no. 12-5773-82), and Aire (eBioscience, no. 50-5934-80). TECs were digested following previously published methods (33) and then stained with surface and intracellular Abs. Fluorochrome-conjugated Abs (clone) CD4 (GK1.5), CD8 (53-6.7), CD44 (IM7), Ki67 (16A8), TNF-α (MP6-XT22), CD25 (PC61), TCR Vα2 (B20.1), TCR Vβ5 (MR9-4), CD45 (30-F11), MHC class II (M5/114.15.2), Ly51 (6C3), EpCAM (G8.8), and GITR (YGITR 765) were purchased from BioLegend. Flow cytometry was performed using an LSR II flow cytometer (BD Biosciences) and data were analyzed using FlowJo software.

Lymphocytes from non–lymphoid organs (liver, lung, and salivary gland) were isolated by two-layer density gradient centrifuge on Lympholyte-M (Cedarlane Laboratories, Burlington, ON, Canada, no. CL5031) at 2000 rpm for 15 min. The cells between the two layers were collected for Treg cell staining. Thymocytes were isolated by cell strainer. Single-cell suspensions were stained with fluorochrome-conjugated anti-mouse cluster of differentiation Abs (BioLegend). Intracellular detection of Foxp3 with PE-conjugated anti-Foxp3 (clone FJK-16s, eBioscience) and detection of Bim with anti-Bim followed by secondary allophycocyanin-conjugated anti-rabbit were performed on fixed and permeabilized cells via Cytofix/Cytoperm (eBioscience). Data were acquired and analyzed as above.

fx/fx-uCreER^T and fx/fx-only mice (3–4 mo old) were treated with TM i.p. (F-cKO and FF-Ctr). Two to 3 wk following the final injection, spleens were harvested and passed through a cell strainer to obtain a single-cell suspension. The samples were erythrocyte depleted and washed, and then stained with anti-CD4 and anti-CD25. Cells were sorted on a BD Influx cell sorter (BD Biosciences) and collected into two groups: Tregs (CD4⁺CD25⁺) or T effector cells (Teffs; CD4⁺CD25⁻). Teffs (5 × 10⁴)/Tregs (2.5 × 10⁴) were cocultured in 96-well U-bottom plates with 5 × 10⁴ irradiated APCs (splenocytes from C57BL/6 mice), 1 μg/ml anti-CD3ε, 2 μg/ml anti-CD28, and RPMI 1640 medium totaling 100 μl per well for 72 h. Proliferation of cells was determined using CellTiter 96 AQ_ueous One solution reagent (Promega, catalog no. TB245) following the Promega protocol: adding 20 μl solution per well for the last 2 h of culture, and absorbance was measured at 490 nm using an ELISA 96-well plate reader (BioTek ELx800).

Following adoptive transfer into young RAG^−/− mice, salivary glands were harvested and fixed, cut into 5-μm-thick sections, and stained with H&E.

For evaluation of group differences, the unpaired two-tailed Student t test was used assuming equal variance. A p value <0.05 was considered statistically significant.

Recent evidence suggests that persistently activated immune cells may be a prime cause of age-related chronic inflammation, but the mechanisms leading to the chronic activation of immune cells are not fully understood (6, 9, 10). Utilizing our previously described loxp-Foxn1-uCreER^T conditional knockout mouse model (F-cKO) for accelerated thymic involution (30), we sought to determine whether thymic atrophy could lead to activated T cells without any additional stimuli. Administration of TM into ∼1- to 2-mo-old F-cKO mice allows for accelerated thymic involution that would not emerge from the slow leakage of uCreER^T until ∼3–9 mo of age (24, 30). To differentiate T cells that have egressed from the atrophied thymus versus those that had egressed prior to thymic involution, we crossed our Foxn1-cKO mice with Rag2 promoter–driven GFP-expressing mice, which identifies newly generated T cells or recent thymic emigrants (RTEs) (34–36). The GFP signal is turned on in thymocytes undergoing RAG-dependent TCR recombination and persists for up to 2 wk following thymic egress (35).

Generally, RTEs are not in a highly active state because they have yet to encounter their cognate Ags. As expected, using TM to conditionally knockout the Foxn1 gene in young fx/fx-uCreER^T mice (F-cKO) to induce accelerated thymic atrophy (30) resulted in fewer CD4⁺ and CD8⁺ RTEs owing to decreased thymic output (Fig. 1A, box in top left panel, Supplemental Fig. 1A). However, within this reduced population of F-cKO RTEs, there was a higher proportion of CD44^hi cells (Fig. 1B, left panels, Supplemental Fig. 1B), which represent Ag-experienced T cells (37), and these F-cKO RTEs were more proliferative (Fig. 1B, Supplemental Fig. 1B, Ki67⁺ cells shown in red histograph in right panels). We further ascertained that the RTEs undergoing proliferation (Ki67⁺) were CD44^hi RTEs (Fig. 1A, red boxes, 1C, right panel). These mice are housed together in a clean environment and are specific pathogen-free. Taken together, these results demonstrate that even in the absence of infection, CD4⁺ and CD8⁺ RTEs derived from the atrophied thymus take on an activated immune cell phenotype shortly after encountering the peripheral environment. Therefore, these activated T cells are potentially autoreactive T cells.

To verify whether activated RTEs derived from the atrophied thymus were responding to peripheral self-antigens, we employed an IRBP immunization model to detect and amplify autoreactive T cell clones within a polyclonal T cell repertoire. IRBP is an eye protein expressed in mTECs as a tissue-specific Ag under the control of the autoimmune regulator gene Aire (32, 38–40). Autoreactive T cells can be induced to incite autoimmune uveitis in Aire-deficient mice and is detected in the spleen and lymph nodes following specific tetramer enrichment (41). To expand and detect autoreactive T cells, we first induced thymic involution through Foxn1-cKO. One month following induction of thymic involution, we immunized F-cKO, FF-Ctr, and Aire^−/− (positive control) mice with an IRBP peptide-2 epitope (P2) and 10 d later pooled lymph nodes and spleen to enumerate CD4⁺ T cells reactive to the P2 peptide by P2–I-A^b tetramer enrichment. F-cKO mice showed a significant expansion of activated P2–I-A^b autoreactive T cells, which were mostly undetectable in FF-Ctr littermate controls (Fig. 2A, 2B). This likely explains the source of activated RTEs in our Foxn1-cKO thymic involution model as autoreactive T cell clones responding to peripheral self-antigens.

Next, we wanted to know whether the activated immune cell phenotype resulting from thymic involution was sufficient to induce an elevated proinflammatory state. We focused on the cytokines IL-6 and TNF-α, which are the most indicative of poor prognosis in chronic age-related disease (2, 6). One month following induction of thymic involution, we observed a 2-fold increase in the percentage of CD4⁺TNF-α⁺ T cells in F-cKO mice compared with FF-Ctr littermate controls (Fig. 2C). Additionally, the concentration of IL-6 in the serum was found to be 2-fold higher in F-cKO mice (Fig. 2D) compared with FF-Ctr littermate controls, implicating thymic involution in both local and systemic inflammation. This low-grade (just above baseline), but significant, inflammatory state found in the periphery of Foxn1-cKO mice is consistent with the conditions described in inflammaging.

We also found pathological changes in multiple organs, such as inflammatory cell infiltration in the liver, lung, pancreas, brain, and salivary glands of these Foxn1-cKO mice reported in our previously published study (42), which confirms the presence of an autoreactive activated immune cell phenotype resulting from thymic involution, although neither dominant clinical manifestations nor full-blown autoimmune disease can be observed in these mice. Taken together, these results indicate that thymic involution alone is sufficient to induce a chronic and low-grade inflammatory state.

The detection of autoreactive T cells in the periphery of Foxn1-cKO mice indicates that thymic involution results in the impairment of central immune tolerance, disrupting either negative selection or the function/generation of nTregs. We first wanted to focus on the functional ability of the Foxn1-cKO atrophied thymus to execute negative selection and remove self-reactive TCRs from the developing TCR thymocyte pool. As we previously reported (30), conditional knockout of Foxn1 in TECs of young mice results in a significant decrease of total thymocyte number and causes thymic atrophy. We observed a dramatic reduction in the percentage and total cellularity of the CD4⁺CD8⁺ double-positive (DP) thymocyte subpopulation (Fig. 3A, 3B) in the F-cKO thymus compared with the FF-Ctr. Because of the highly stochastic nature of TCR rearrangement (43), one would presume that a diminished DP subpopulation would be accompanied by equally reduced CD4 and CD8 single-positive (SP) subpopulations. However, we found that the total numbers of CD4 and CD8 SP thymocytes were not reduced in the F-cKO thymus compared with the FF-Ctr, and the frequency of SP thymocytes is increased in the F-cKO atrophied thymus (Fig. 3A, 3B). The increased proportion of SP thymocytes suggests that the F-cKO cannot efficiently perform clonal deletion (negative selection) resulting from the loss of Foxn1 in TECs.

We next attempted to investigate the mechanisms leading to the accumulation of SP thymocytes in the Foxn1-cKO thymus. Aire, which regulates expression of tissue-specific Ags, is one key factor in determining the efficiency of negative selection (21, 44). We previously published that Aire⁺ mTECs are reduced following the loss of Foxn1 (42). In the present study, we further found that the reduction of Aire⁺ mTECs was not simply due to the reduction of total mTECs, but is actually due to a reduction in the frequency of Aire⁺ mTECs among all mTECs (Fig. 3C, 3D), as well as a reduction in the mTEC expression of Aire on a per cell basis, measured by mean fluorescence intensity (Fig. 3E). This disruption in Aire expression following the loss of Foxn1 potentially perturbs negative selection by impairing the expression of Aire-dependent self-antigens (42).

To find direct evidence that the loss of Foxn1 impairs the clonal deletion of autoreactive thymocytes, we used the OT-II–RIP–mOVA model of Aire-dependent negative selection. RIP-mOVA⁺ mTECs express chicken OVA as a mock self-antigen under the control of the RIP by an Aire-dependent mechanism (19, 45). OT-II bone marrow progenitors seed the kidney-grafted thymic lobes and give rise to thymocytes with TCRs that are specific for OVA, MHC class II restricted (CD4), and bind MHC/OVA complexes with a strong affinity. Thymocytes possessing TCRs that bind to MHC/OVA complexes with too strong of an affinity will be signaled to undergo clonal deletion. In our system, host mice do not receive any irradiation as in the conventional bone marrow transplantation method. Therefore, there is minimal risk for damage to the stromal cell niche that may affect Ag presentation. RIP-mOVA mice have been crossbred with Foxn1-cKO mice yielding F-cKO-mOVA, FF-mOVA, and FF-Ctr offspring. Fetal thymic lobes of equal size and cellularity (data not shown) were harvested from F-cKO–mOVA, FF-mOVA, and FF-Ctr newborns and then transplanted under the kidney capsule of young adult OT-II Tg mice. Three days after the transplantation surgery, i.p. injections of TM were administered to the OT-II host mice for 3 successive days to induce the Foxn1 conditional knockout in the kidney-grafted thymus. Two weeks after the final TM injection, the grafted thymic lobes were isolated for FACS analysis (Fig. 4C). As expected, OT-II hosts grafted with newborn thymic lobes from FF-Ctr mice that completely lack the mOVA transgene contain a normal proportion of OT-II–specific CD4 SP thymocytes (CD4⁺CD8⁻Vα2Vβ5⁺) (Fig. 4A, top panel, 4B, filled circles); when the OT-II progenitors were seeded into the grafted FF-mOVA thymus that carries the mOVA transgene and WT levels of Foxn1 (not atrophied), then strong negative selection was observed in OT-II CD4 SP thymocytes (Fig. 4A, middle panel, 4B, filled squares). However, when the OT-II progenitors seeded into the grafted F-cKO–mOVA thymus, which carries the mOVA transgene and the Foxn1-cKO–induced thymic atrophy, the proportion of CD4⁺CD8⁻Vα2Vβ5⁺ cells was increased (Fig. 4A, bottom panel, 4B, open squares) compared with FF-Ctr thymus, indicating that the OT-II CD4 SP thymocytes in F-cKO–mOVA thymus evade clonal deletion. These results provide direct evidence that the loss of Foxn1 and subsequent thymic involution perturbs negative selection, leading to the survival of SP thymocytes that potentially recognize self-antigen.

The manifestation of autoreactive inflammation in the Foxn1-cKO mouse model of accelerated thymic involution suggests a breakdown in the maintenance of central immune tolerance. We revealed measureable dysfunction in the process of negative selection in the Foxn1-cKO thymus. We next addressed whether Foxn1-cKO–driven thymic involution alters the other arm of central immune tolerance, that is, the generation of nTregs. Several studies have shown that the disruption of thymic medullary microstructure and a reduction in medullary size led to the generation of fewer or deficient CD4⁺ nTreg cells (29). Despite the severe depletion of overall thymic cellularity, there was not a significant difference (slightly increased) in the total numbers of thymic CD4⁺CD8⁻CD25⁺Foxp3⁺ nTregs (Fig. 5A, 5C) in the F-cKO thymus compared with FF-Ctr. Instead, the frequency of thymic nTregs was increased in the F-cKO mice (Fig. 5B). These findings suggest thymic involution does not impair the production of nTregs and may actually be a contributing factor toward the age-related accumulation of Tregs (46).

Irrespective to their numbers, Tregs must possess adequate suppressive capabilities to sustain immune tolerance. Whereas many laboratories have shown that aged Tregs retain their suppressive function (46–49), others have reported a functional decline from thymic involution–derived Tregs (29, 50). To determine whether Foxn1-cKO–induced thymic atrophy impairs nTreg function, we tested the ability of peripheral Tregs derived from the Foxn1-cKO atrophied thymus to suppress the proliferation of Teffs. Tregs (CD4⁺CD25⁺) were isolated from the spleens of either F-cKO or FF-Ctr mice, and Teffs (CD4⁺CD25⁻) were isolated from the spleens FF-Ctr mice (Fig. 5D). They were cocultured in the presence of irradiated APCs and CD3ε and CD28 Abs. We did not find any differences in the suppressive ability of Tregs coming from either F-cKO or FF-Ctr mice (Fig. 5E). Additionally, there were no measureable differences in the functional marker GITR (glucocorticoid-induced TNF receptor family-related gene) (12, 51) between F-cKO and FF-Ctr Tregs (Supplemental Fig. 2). These data suggest that the progressive loss of Foxn1 and subsequent thymic involution does not impair the suppressive capacity of Tregs.

Natural aging induces the accumulation of Tregs (the frequency is significantly increased in aged mice and humans) (46), which has been attributed to decreased levels of the proapoptotic gene Bcl2 family member Bim (52–54); however, whether thymic involution influences age-related accumulation of peripheral Tregs remains unclear. As shown in Fig. 5A–C, thymic involution increased the frequency of thymic nTregs, but this result did not appear to carry over into the periphery. We did not observe any peripheral accumulation, including splenic Tregs (Figs. 6A, 6B) and nonlymphoid (lung, liver, salivary gland) Tregs, in the F-cKO mice (Supplemental Fig. 3).

To better answer whether thymic involution imparts any intrinsic changes onto nTregs that could lead to their peripheral accumulation, we performed an adoptive transplantation by infusing a pool of WT naturally aged (>18 mo) T cells into young RAG^−/− mice, which lack T and B cells. Two months after transfer, we found that the age-associated accumulation (frequency and relative ratio) of aged donor Tregs was decreased, matching the levels of the young donor (6 wk) WT control (Fig. 6C, 6D). Furthermore, the diminished expression of Bim observed in the aged Treg population was restored and equivalent to young mice (Fig. 6E). These results suggest that the peripheral microenvironment, rather than the thymus (RAG^−/− mice only have a rudimentary thymus), is responsible for the accumulation of Tregs with age. Moreover, the young peripheral microenvironment is capable of reducing a previously accumulated Treg pool.

Our previous publications have shown that thymic involution leads to inflammatory infiltration in the lung, pancreas, liver, and most frequently in the salivary gland (42). However, we determined that the increased frequency of thymic nTregs does not persist into the periphery of Foxn1-cKO mice, and the age-related accumulation of peripheral Tregs can be reversed by the young microenvironment (Fig. 6). Whether inflammatory infiltration is an intrinsic defect among T cells derived from an involuted thymus or dependent on age-related changes to the peripheral environment is unclear. To test whether the young microenvironment can protect against age-induced autoreactive T cells, we adoptively transferred either WT aged (>18 mo) or WT young (<6 wk) splenocytes into young RAG^−/− hosts. Two months after the infusion, we observed that young hosts infused with young splenocytes generally appeared healthy, but young hosts infused with aged splenocytes showed inflammatory infiltration around salivary blood vessels with severity that progressively worsened as the age of the donor increased. The 18 mo donor was more likely to present with a single inflammatory foci, and hosts given >22 mo donor cells were likely to display multifoci infiltration (Fig. 7A).

To eliminate the effect of peripheral age-related factors (thymus independent) on inflammatory infiltration, we infused a pool of T cells from young F-cKO or FF-Ctr splenocytes into young RAG^−/− mice. Although Foxn1-cKO mice lack any peripheral age-related modifications not directly resulting from thymic involution, we found inflammatory cell infiltration in the salivary glands of the young RAG^−/− host mice infused with F-cKO (atrophied thymus) splenocytes, but infiltration was not observed in hosts infused with FF-Ctr (normal thymus) splenocytes (Fig. 7B).

Taken together, these results indicate that unlike the age-related accumulation of Tregs, inflammatory infiltration from the naturally aged or Foxn1-cKO atrophied thymus cannot be reversed by the young peripheral microenvironment. This indicates that inflammatory infiltration is an intrinsic defect in T cells imprinted during development in the involuted thymus.

Virtually every chronic age-related disease emerges under conditions of chronic inflammation, that is, inflammaging, which provides a highly significant risk factor for both morbidity and mortality in the elderly. Although the characteristics of inflammaging, which describes a chronic systemic low-grade inflammation in the absence of overt/acute infection with advanced age, have been defined (7, 9, 55–57), the precise etiology and mechanisms responsible for inflammaging are largely unknown (1). Current knowledge of the etiology of inflammaging pinpoints latent viral infections giving rise to the persistent activation of immune cells (1), and SASP expressing senescent tissue cells (58), as sources of a chronic inflammation. However, the predisposition toward autoimmunity in the elderly and the contribution of autoreactive T cells in generating a systemic inflammatory environment are widely overlooked in the prevention and treatment of inflammaging. In this study, we identified thymic involution resulting from the loss of Foxn1 as an additional source of activated immune cells (autoreactive T cells) that lead to a state of persistent and low-grade immunopathology with elevated proinflammatory cytokines. Newly generated T cells that have just exited the atrophied thymus of Foxn1-cKO mice adopt an activated and proliferative phenotype shortly after contacting the periphery, despite never encountering foreign Ag. Additionally, we were able to detect autoreactive T cells that specifically respond to the peripheral self-antigen IRBP in the periphery of Foxn1-cKO mice, which suggests that T cell activation and elevated proinflammatory cytokines are the result of self-antigen stimulation. Interestingly, Foxn1-cKO mice do not develop overt autoimmune disease that would account for the elevated inflammatory milieu, but rather present with general and relatively mild infiltration in multiple non–lymphoid organs (42). Furthermore, it is unlikely that extrinsic aging factors cause low-level T cell activation, as these inflammatory T cells cannot be rejuvenated by the young microenvironment. Because these inflammatory T cells do not encounter foreign Ag, they likely result from defects in the thymic-driven generation and maintenance of immune tolerance.

Identification of activated autoreactive T cells in the periphery of Foxn1-cKO mice is evidence of defects in the maintenance of central immune tolerance. In the present study, we further identified impairment in the ability of the Foxn1-cKO involuted thymus to clonally delete OVA TCR Tg thymocytes that recognize the endogenous RIP-mOVA Ag. mOVA expression in mTECs is under the control of the RIP and dependent on Aire (19, 45). Complete knockout of the Aire gene (Aire^−/−) disrupts negative selection and causes systemic autoimmune disease. Notably, even the partial loss of the Aire gene (heterozygous Aire^+/−) leads to a drastic increase in islet-reactive T cells and progression toward diabetes in an insulin TCR Tg mouse model (59). We previously showed mTEC structure disruption and an overall reduction in thymic Aire expression in the Foxn1-cKO mice (42). We report in the present study that the percentage of Aire⁺ mTECs is diminished in Foxn1-cKO mice, and the expression of Aire in mTECs is decreased on a per cell basis. This implies that our observed negative selection impairment associated with thymic atrophy may result from the diminished expression of Aire in mTECs, which supports our previous finding of decreased tissue-specific Ag expression in the Foxn1-cKO thymus (42). Although there is evidence to support both Aire-dependent (21, 44) and Aire-independent (60) nTreg development, we did not observe any defects in the ability of the atrophied thymus to generate nTregs, despite the diminished expression of Aire in Foxn1-cKO thymi. It is possible that the low level of Aire present in the atrophied thymus is sufficient to support nTreg development but is not sufficient to support the strong avidity necessary for the clonal deletion of autoreactive thymocytes. Additionally, it is possible that nTreg generation is maintained in the Foxn1-cKO thymus by Aire-independent thymic dendritic cells (22). We previously published that thymic dendritic cell accumulation shifts from mTEC to cTEC regions in the atrophied thymus (42). However, further work needs to be done to determine whether the loss of Foxn1 and thymic atrophy influence the functional ability of thymic dendritic cells to facilitate negative selection.

It is unclear whether the loss of Foxn1 affects TCR avidity for MHC–self-peptide. The decreased negative selection with enhanced (at least unimpaired) generation of Tregs in the accelerated thymic involution model is probably better explained by an “avidity window” between negative selection (strong TCR signal strength) and deviation into the Treg lineage (moderate TCR signal strength) (61); that is, diminishing the avidity of TCR for MHC–self-peptide should reduce the clonal deletion of negative selection and favor Treg induction in a specific cohort of MHC class II–restricted thymocytes.

In addition to negative selection, Tregs play a critical role in maintaining immune tolerance by inhibiting the activation of autoreactive T cells in the periphery. It has been reported that premature thymic aging, induced by a hypomorphic mutation in the Foxn1 gene (Foxn1^Δ/Δ mice with a germline mutation), weakens the suppressive function of peripheral Tregs (50). However, this is contrary to our observation, which shows that Tregs derived from the Foxn1-cKO thymus can suppress the proliferation of FF-Ctr Teffs just as efficiently as do Tregs derived from the normal FF-Ctr thymus. Although CD4⁺CD25⁺ may not define all Tregs, our results are in agreement with others that show aged Foxp3-GFP Tregs maintain suppressive capabilities (47). Additionally, recent evidence suggests that there is an age-related accumulation of peripheral Tregs with increased suppressive function (46). Age-related Treg peripheral accumulation is suggested to result from decreased expression of the proapoptotic gene Bcl2 family member Bim (52–54). In the present study, we have shown peripheral Treg accumulation and diminished Bim expression in naturally aged WT mice can be normalized by the young peripheral microenvironment and are therefore not due to an intrinsic thymic atrophy-related defect, but potentially due to an extrinsic abnormality in the aged peripheral microenvironment. Although there are many age-related alterations to the peripheral environment, including changes to circulating soluble factors, the stromal compartment, the hematopoietic compartment, and microbiota, the exact mechanisms signaling an age-related decrease in Bim and subsequent accumulation of Tregs remain unclear. Based on these findings, Treg accumulation and function do not appear to significantly contribute to thymic atrophy-driven inflammaging.

Although direct anti-inflammatory interventions, such as the use of low-dose aspirin or statins, to suppress, prevent, and alter the state of chronic inflammation hold great promise for treating multiple age-related diseases (1), targeting the sources of chronic inflammation may be the key to enhancing the prognosis of chronic age-related disease. In this study, we have shown that in addition to developing a strategy for the elimination of senescent cells to suppress SASP (62), reduction or elimination of autoreactive T cells that emigrated from the atrophied thymus is particularly promising to dampen age-related inflammation. We think there are two strategies to do so. One is to rejuvenate the atrophied thymus to restore its function, including functional negative selection and functional generation of naive T cells for TCR diversity. This rejuvenation of thymic involution will also improve immunosenescence, which is a major source of inflammaging (1). Removal of the atrophied thymus is another strategy that can be used to eliminate emigration of the harmful autoreactive T cells. This would likely require particular clinical criteria for the surgery in the elderly, such as ease of thymectomy (i.e., the patient is already having open chest surgery) or if the persistent chronic inflammation would significantly increase the morbidity of other diseases. Although vaccination against persistent infections is another possible strategy to manage the persistent activation of immune cells, it does not address autoreactive immune cell activation.

A caveat to this study is the frequent use of our Foxn1-cKO mouse model of accelerated thymic involution in place of a true natural age-related atrophied thymus from >18 mo WT mice. Progressive thymic involution is an age-related alteration and its causes are almost certainly not due to only the loss of one gene. This complex and multifaceted nature of thymic involution is evidenced by the various processes implicated in age-related thymic atrophy, including the involvement of other genes such as ink4a (63), hormones (64), and adipocyte expansion (65–67). However, because thymic involution occurs alongside total body aging, separating the effects of thymic age from systemic aging (which includes the circulation of soluble factors, T cells, and dendritic cells back into the thymus) is extremely difficult, if not impossible, in an 18 mo WT mouse. The strength of our Foxn1-cKO model lies in the ability to induce the involution of a mature thymus while avoiding the effects of systemic aging. In fact, our model (both TM-induced and slow leakage of uCreER^T) undergoes thymic involution, including a loss of Foxn1⁺ TECs (24, 30), a decline in mature mTEC (24, 30, 42), thymic dendritic cell distribution (42), increased senescent clusters, and increased TAp63⁺ TECs (33) similar to 18 mo C57BL/6 mice. Furthermore, naturally aged C57BL/6 mice and F-cKO splenocytes that were transferred into a young environment induced similar inflammatory infiltrates in the salivary gland (Fig. 7). Additionally, unlike the prenatal Foxn1^Δ/Δ mutation that blocks TEC maturation and reduces Treg function (50), our inducible Foxn1-cKO model of mature thymic atrophy maintains Treg suppressive capacity consistent with naturally aged Tregs (46). However, our Foxn1-cKO model undergoes an extremely accelerated thymic involution that is dissimilar to the 1–3% shrinkage per year observed in the naturally involuting thymus (68). Furthermore, similar to natural age-related thymic involution, the Foxn1-cKO model does not exhibit thymic regrowth that is seen in acute thymic involution following infection and pregnancy (67). However, the Foxn1-cKO model still does not fully recapitulate the physiology of the chronic and extremely gradual nature of natural age-related thymic involution. Therefore, the effects of rapid, as opposed to gradual, involution on defective negative selection and the emergence of chronic inflammation cannot be eliminated. Although the induced conditional knockout of Foxn1 and subsequent accelerated thymic involution in the fully mature thymus may not exactly mimic the naturally aged thymus, there are many molecular, morphological, and functional characteristics shared between the two. Therefore, the Foxn1-cKO model is useful to study thymic involution and is applicable for age-related thymic involution with careful consideration for the limitations of the animal model.

In summary, we have demonstrated that the steady release of autoreactive T cells resulting in tissue-specific inflammation following thymic involution contributes to low-grade chronic inflammation, known as inflammaging. The observed inflammaging is the consequence of a T cell predisposition toward autoimmunity that arises from a decline in Foxn1 expression leading to the disruption of the steady-state thymic medullary compartment (24, 42, 69) and subsequent thymic involution. Thymic involution results in the release of autoreactive T cells that become activated shortly after reaching the periphery and produce low levels of inflammatory cytokines without causing overt autoimmune disease. The autoreactive T cells result from intrinsic defects in negative selection, rather than changes to the extrinsically maintained Treg cell pool. Our studies shed new light on the importance of targeting thymic involution in addition to improving the peripheral immune microenvironment as a potential treatment for eliminating inflammaging and ultimately reducing morbidity and mortality in chronic age-related disease.

We thank Dr. Mark Anderson (University of California, San Francisco) for providing the IRBP peptide and tetramer, Dr. Rance Berg (University of North Texas Health Science Center) for critical reading of the manuscript, Dr. Xiangle Sun (University of North Texas Health Science Center) for flow cytometry technical support, and the National Institutes of Health Tetramer Core Facility for providing the tetramer reagent.

The authors have no financial conflicts of interest.