Thymic atrophy creates holes in Treg‐mediated immuno‐regulation via impairment of an antigen‐specific clone

Abstract

Age‐related thymic atrophy results in reduced output of naïve conventional T (Tcon) cells. However, its impact on regulatory T (Treg) cells is insufficiently understood. Given evidence that thymic Treg (tTreg) cell generation is enhanced in the aged, atrophy thymus and that the aged periphery accumulates peripheral Treg (pTreg) cells, we asked why these Treg cells are unable to effectively attenuate increased autoreactivity‐induced chronic inflammation in the elderly. We designed a mock‐self‐antigen chimera mouse model, in which membrane‐bound ovalbumin (mOVA) transgenic mice, bearing a FoxN1‐floxed gene for induction of conditional thymic atrophy, received OVA‐specific (OT‐II) T‐cell receptor (TCR) transgenic progenitor cells. The chimeric mice with thymic atrophy exhibited a significant decrease in OVA‐specific tTreg and pTreg cells but not polyclonal (pan)‐Treg cells. These OVA‐specific pTreg cells were significantly less able to suppress OVA‐specific stimulation‐induced proliferation in vitro and exhibited lower FoxP3 expression. Additionally, we conducted preliminary TCR repertoire diversity sequencing for Treg cells among recent thymic emigrants (RTEs) from Rag^GFP‐FoxP3^RFP dual‐reporter mice and observed a trend for decreased diversity in mice with thymic atrophy compared to littermates with normal thymus. These data indicate that although the effects of age‐related thymic atrophy do not affect pan‐Treg generation, certain tissue‐specific Treg clones may experience abnormal agonist selection. This, combined with enhanced pan‐pTreg cells, may greatly contribute to age‐related chronic inflammation, even in the absence of acute autoimmune disease in the elderly.

Previously, we have shown that the aged, atrophied thymus perturbs negative selection and relatively enhanced thymic regulatory T‐cell (Treg) generation, but it remains unclear why the increased Treg cells not able to suppress self‐reactive T cell‐induced baseline inflammation (termed inflammaging) in the elderly. Herein, we showed that although polyclonal Treg generation is relatively enhanced, agonist selection of a self‐antigen‐specific Treg clone is reduced in mice with the age‐mimicking atrophied thymus. This may create holes in the aged immuno‐regulatory system, resulting in the contribution of self‐reactive T cells in the manifestations of subclinical inflammaging phenotype in the elderly.

Abbreviations

Ag

antigen

Aire

autoimmune regulator gene

BM

bone marrow

CD4^SP

CD4⁺CD8^{− neg} single positive

CD8^SP

CD4^{− neg}CD8⁺ single positive

cKO

conditional knockout

CreER^T

Cre recombinase and oestrogen receptor fusion protein

cTEC/mTEC

cortical/medullary thymic epithelial cells

FC

floxed‐FoxN1 gene with CreER^T

FF

floxed FoxN1 gene

FCM or FFM

FCmOVA or FFmOVA

FoxN1

forkhead box N1

FoxP3

forkhead box P3

GFP

green fluorescent protein

i.v.

intravenous(ly)

OT‐II

MHC‐II restricted transgenic T‐cell receptor recognizing ovalbumin peptide

RFP

red fluorescent protein

RIP‐mOVA

rat insulin promotor‐driven membrane‐bound ovalbumin

RQ‐MFI

relative quantitative (geometric) mean fluorescence intensity

RTEs

recent thymic emigrants

TCR

T‐cell receptor

Tcon

conventional T cell

Teff

effector T cell

Tg

transgenic

TM

tamoxifen

Treg

regulatory T cell

tTreg/ pTreg

thymic Treg/ peripheral Treg

WT

wild‐type

INTRODUCTION

Age‐related thymic involution or atrophy is a progressive condition observed in most vertebrate animals, resulting in the obvious reduction of naïve conventional T (Tcon) cell output [1]. This contributes to the systemic dysfunction of the aged immune system termed immunosenescence via the broad dampening of T cell‐mediated immune responses [2, 3]. Mounting evidence shows that age‐related thymic atrophy contributes to not only diminished immunity to foreign antigens (Ags), but also enhanced reactivity to self, due to increased output of self‐reactive Tcon cells [4]. Therefore, age‐related thymic atrophy contributes to both aspects of aged immune dysfunction [3, 5]. This dysregulation associated with age‐related thymic involution also affects the generation and function of regulatory T (Treg) cells [6, 7]. However, it is insufficiently understood how Treg cells generated by the atrophied thymus contribute to the age‐related disruptions of immune homeostasis.

Age‐related thymic atrophy is characterized by a primary decline in expression of the thymic epithelial cell (TEC)‐autonomous transcription factor, forkhead box N1 (FoxN1) [8, 9]. Thus, the vast majority of the TECs, specifically in the medullary TEC (mTEC) compartment, retract, resulting in functional decline and influencing the establishment of central tolerance. Central tolerance includes reciprocal mechanisms of negative selection (depletion) of self‐reactive T clones and agonist self‐antigen (Ag) selection (generation) of regulatory T (Treg) clones during late stages of thymocyte development [10, 11]. Our previous studies demonstrated that the atrophied thymus had impaired negative selection, as shown by increased numbers of self‐reactive Tcon cells in the periphery [4], but relatively enhanced Treg cell generation, as shown by an increased proportion of newly generated thymic Treg (tTreg) cells to Tcon cells in the atrophied thymus [6]. Ample evidence also shows that the proportion of peripheral Treg (pTreg) cells accumulate [12, 13] with enhanced suppressive function [7] in the elderly. We therefore wondered why these relatively increased Treg cells are unable to effectively counteract self‐reactive Tcon cell‐driven subclinical autoimmune proneness [4] of inflammaging, referring to a chronic, systemic inflammatory condition in the absence of acute infection observed with advanced age [14].

Thymocyte negative selection of Tcon or agonist selection of Treg TCR repertoires is dependent on presentation of self‐Ags, mainly by mTECs [15, 16], mediated partially by the expression of autoimmune regulator (AIRE) gene [17] or others, such as FEZF2 [18]. Evidence shows that Treg TCR selection in the Aire ^−/− thymus results in defects in certain tissue‐specific [19] but not polyclonal (pan‐) Treg cells [10, 11, 20]. However, these observations may be different for a given self‐antigen clone. For example, one group showed that prostate TCAF3‐specific Treg cells are missing from the repertoire in mice with Aire ^−/− thymus [19, 21], while another group demonstrated that tissue‐specific (such as TRP‐1/TYRP1, melanocyte‐specific) Treg cells are not affected in numbers [22]. Ageing results in mTEC dysfunction, including reduced Aire levels [4, 23, 24]. Thus, by aligning the studies of the aged thymus with Aire ^−/− thymus, we hypothesized that Treg generation in the aged thymus (or age‐mimicking atrophied thymus via an inducible floxed‐FoxN1‐knockout mouse model) [8, 9] follows a similar paradigm as observed in Aire ^−/− mice. Specifically, we postulated that relatively enhanced generation of tTreg clones in the aged thymus [6] is polyclonal Treg TCRs, while certain tissue Ag‐specific Treg clones may not be agonist selected (missing), or, even if selected, may experience an intrinsic functional defect, allowing for low and persistent inflammaging, but not outright autoimmune disease.

In order to address this hypothesis, we generated two mock‐self‐Ag chimera mouse models [4, 6], in which membrane‐bound ovalbumin (mOVA) transgenic mice [25, 26] received mixed wild‐type (WT) and MHC‐II restricted, OVA‐recognizing T‐cell receptor (TCR) transgenic (termed OT‐II TCR‐Tg) progenitor cells [27]. These models circumvent the limitations of attempting to identify a single T‐cell clone within a polyclonal pool. Additionally, each system provided polyclonal precursors along with TCR‐Tg precursors so that our reconstitution models provided a more physiologically relevant intraclonal competition during thymocyte development [28], and the thymic and peripheral stromal cells expressed mOVA‐Tg (driver by the rat insulin promoter, RIP). In addition, these mouse thymuses carried a FoxN1‐floxed gene with (FC) or without (FF) CreER^T [8, 9] for conditional thymic atrophy induction, termed FCmOVA (or FCM) or FFmOVA (or FFM), respectively. These mice experience the hallmarks of thymic ageing (age‐mimicking atrophy due to induced FoxN1 knockout), but possess a young periphery since our experiments are performed on mice 2–3 months of age. This allows us to observe impacts on the Treg cells resulting from thymic involution rather than from aged peripheral conditions.

We observed that with thymic atrophy, this OVA‐specific (OT‐II TCR‐Tg) Treg population was significantly decreased in both the thymus and periphery. Further, we found that the Treg‐enriched cells from mice with thymic atrophy exhibited impaired suppressive capacity against OVA‐specific T effector (Teff) cell proliferation after OVA‐peptide stimulation in vitro. Finally, these OVA‐specific Treg cells displayed lower FoxP3 expression.

Lastly, we postulated that reduction/absence of certain self‐antigen specific Treg clones in the atrophied thymus is potentially due to altered agonist selection of these Treg cells within the Treg repertoire. Indeed, we saw decreased Treg diversity after TCR repertoire sequencing analysis of newly generated Treg cells (based on Rag ^GFP‐FoxP3 ^RFP reporters) in mice with thymic atrophy compared to controls. In sum, these data indicate that although the effects of age‐related thymic involution do not affect pan‐Treg cell generation, agonist selection of certain tissue‐specific Treg clones is impaired, altering the tTreg repertoire of the aged T‐cell regulatory system. This, combined with the enhanced pan‐pTreg cells, this could greatly contribute to age‐related chronic inflammation.

Throughout our study, we refer to pTreg cells based on their physiological location, but based on the current Treg terminology, we need to emphasize that our ‘pTreg’ population does include both tTreg cells (generated in the thymus) and peripherally‐induced pTreg cells described in the current update on Treg nomenclature [29]. The primary reason for our description of pTreg cells based on location is that the current understanding is that at least 80% of peripherally circulating Treg cells are derived from the thymus [30, 31]. Therefore, describing the Treg cells based on their location seemed most appropriate for this study.

METHODS

Mouse models

All mice had a C57BL/6 genetic background, which express CD45.2 or CD45.1/.2 congenic markers. loxp‐floxed FoxN1 knock‐in mice (Jackson Lab #012941) were generated per our previous publications [8, 9], and either do not carry Cre recombinase, termed FF mice, and will maintain a normal thymus after tamoxifen (TM) injection, or carry CMV promoter‐driven CreER^T and have induced thymic atrophy after TM injection, termed FC mice, that is FoxN1 conditional knockout (cKO) mice. RIP‐mOVA (rat insulin promoter‐driven membrane‐bound ovalbumin) mice (Jackson Lab #005431) were crossbred with our FF or FC mice to generate FFmOVA (FFM) and FCmOVA (FCM) mice with inducible FoxN1 knockout. OT‐II TCR‐transgenic (Tg) mice (expressing a TCR that recognizes mOVA in the context of the MHC‐II molecule, I‐A^b) (Jackson Lab #004194) were kept on a Rag sufficient background or crossbred with Rag1 knockout (Rag ^−/−) mice (Jackson Lab #002216). These were crossbred with CD45.1 mice (Charles River #NCI B6‐Ly5.1/Cr) to generate OT‐II/CD45.1/.2 (F1) mice. Rag^GFP reporter mice (Jackson Lab #005688) were crossbred with FoxP3^RFP mice (Jackson Lab #008374) to generate dual‐reporter mice, whose newly generated Treg cells express both GFP and RFP in the CD4 single positive (CD4^SP) population for live fluorescence‐assisted cell sorting (FACS). All animals were housed in our specific pathogen‐free animal facility.

Bone marrow chimera construct

Bone marrow (BM) progenitor cells were harvested from the femur and tibia of young (4–6 week old) OT‐II/CD45.1/.2 (with Rag ^−/−) or WT (CD45.2) mice. Cells were depleted of erythrocytes via ACK lysis buffer and depleted of T cells via anti‐CD4, anti‐CD8 magnetic bead‐assisted cell sorting (MACS) (Miltenyi Biotech LS Columns and PE‐conjugated microbeads). Approximately 10 × 10⁶ cells/mouse were injected intravenously (i.v.) at a ratio of 1:5 WT:OT‐II into lethally (900 Rad) irradiated FFM or FCM recipient mice. One week after immune reconstitution, recipient mice received two doses (1 mg/10 g body weight/day) of tamoxifen (TM) to induce thymic atrophy. Mice received antibiotic water for 1 week prior to irradiation until 1 week post‐irradiation (2 g/l neomycin). A total of 6–8 weeks after immune reconstitution, the mice were killed for further analysis. The workflow was identical for the reporter chimeric mice, except that the host mice were FF and FC mice and each received 10 x 10⁶ BM cells from GFP⁺RFP⁺ donor mice.

Thymus kidney capsule chimera construct

Thymic lobes were harvested from FFM or FCM newborn mice and transplanted under the kidney capsule of young (6–8 week old) OT‐II/CD45.1/.2 (Rag sufficient) host mice [4]. Four to 5 days after surgery, the host mice received three doses of TM as above to induce atrophy of the transplanted thymuses. Approximately 4 weeks after surgery, the mice were killed and the transplanted thymuses were analysed. Surgery was performed with anaesthesia/analgesia under aseptic conditions in accordance with our approved IACUC protocols.

Isolation of T cells from pancreas

Freshly isolated mouse pancreas was digested by cutting into small pieces in 500 µl of DMEM media and 50 µl of digestion buffer (20 µl collagenase, 5 µl DNase, 25 µl DMEM). After 30 min of incubation at 37°C, the tissues were gently mixed via pipetting and then incubated for another 10 min. After incubation, digested tissues were filtered through a 70‐µm filter with flow cytometry staining buffer into a 15‐ml tube and centrifuged at 222 g at 4°C for 5 min. The cell pellets were resuspended and passed through a 40‐µm filter, then centrifuged again. Finally, the cell pellets were washed with staining buffer prior to staining for flow cytometry.

Flow cytometric analysis and fluorescence‐assisted cell sorting (FACS)

Single‐cell suspensions of the homogenized and filtered thymus, spleen and lymph nodes (LNs) were stained with various surface and intracellular markers for flow cytometry. Intracellular staining with FoxP3‐FITC (eBiosciences #53‐5773‐82) utilized the FoxP3/Transcription Factor Staining Buffer Set (eBiosciences #00‐5523‐00) per the kit's protocol. All other antibodies were purchased from BioLegend and include CD4‐APC/Cy7 (Cat#100414), CD25‐PE/Cy7 (#102016), CD45.1‐APC (#110714), Vα2‐APC (#127809), Vβ5‐APC (#139505), CD8‐PE (#100708), FITC isotype control (#400705), PE/Cy7 isotype control (#401908), CD16/32 blocking antibodies (#101302), CD4‐BV711 (#100447), CD8‐APC (#100712), B220‐APC (#103211) and CD19‐APC (#152409). An LSR‐II cytometer (BD Biosciences) was used for sample acquisition, and FlowJo software (v10) was used for data analysis.

For identification of Vα2Vβ5 TCR chains, the same flow cytometry staining channel was used based on results showing that WT mice have <3% of either Vα2, Vβ5, or double‐positive populations and that OT‐II TCR‐Tg mice are enriched for this population (Figure S2).

For FACS, single‐cell suspensions were either MACS bead‐depleted of CD8⁺ T cells and B cells (CD19⁺ and B220⁺) or these markers were labelled for a dump channel. A Sony Cell Sorter (SH800Z) was used for sample acquisition, and purity was verified using pre‐ and post‐sorted fractions analysed on the LSR‐II cytometer and FlowJo software (v10).

In vitro Treg suppression assay

Approximately 6–8 weeks after immune reconstitution, spleen and LNs from the FFM and FCM chimeric mice were collected, stained and sorted via FACS as described above, with a sorting purity of >95% (shown in Figure 3a). Next, the live‐sorted OVA‐Ag‐specific (CD45.1⁺, Vα2⁺Vβ5⁺) Treg‐enriched cells (CD4⁺CD25⁺, 1 × 10⁴/well) were then co‐cultured with irradiated splenic Ag‐presenting cells (APCs, 4 × 10⁴/well) from C57BL/6 mice, and MACS bead negatively sorted (with CD8, CD19, B220, and CD25 antibodies) OVA‐specific Teff cells (CD4⁺ CD25⁻) cells (2 × 10⁴/well) from the spleen of OT‐II TCR‐Tg mice. These co‐cultured cells were treated with or without OVA^323−339 peptide (1 µg/ml) (Bachem #4034255.1) stimulation. After co‐culture for 72 h, cell proliferation was quantified via CellTiter 96 Aqueous One Solution (Promega #PR‐G3582) per the company's protocol, by adding 20 µl of solution to each well for approximately 1·5 h of culture. Then, absorbance at 490 nm was measured using an ELISA 96‐well plate reader (BioTek ELx800).

FIGURE 3.

Suppressive capacity of OVA‐specific pTregs was functionally defective in mice with age‐mimicking thymic atrophy (a) Flow cytometric gating sorting strategy showing purity of specific pTreg cells before (top panels) and after (bottom panels) FACS sorting from BM chimeric mice (Figure 2a). (b) Summarized results of the suppressive capacity of OVA‐specific pTreg cells on CD4^SP effector T (Teff) cells isolated from the spleen of OT‐II TCR‐Tg mice. A one‐way ANOVA with Bonferroni correction was used to determine statistical significance between multiple groups. Each symbol represents cells from an individual animal. Experiment was repeated three times. ***P ≤ 0·001

Treg TCR repertoire sequencing & analysis

Approximately 6–8 weeks after immune reconstitution, lymphocytes from the spleen and LNs of FF and FC Rag^GFP ‐FoxP3^RFP dual‐reporter chimera mice were collected and prepared for FACS. RTE (GFP⁺) Treg (CD4⁺CD25⁺RFP⁺) cells (Gate strategy in Figure 5a) were collected and genomic DNA was isolated from each sorted sample with the Qiagen DNeasy Kit (#69504). TCRβ CDR3 survey‐level sequencing was performed via Adaptive ImmunoSEQ customer service. Information on the primers used by Adaptive can be found in their publication [32]. Data were analysed via the Adaptive ImmunoSEQ data analysis software with the assistance of their technical support team. Per the Adaptive ImmunoSEQ Analyzer, Simpson's clonality index is a way of measuring how polyclonal or monoclonal a population is on a scale of 1 (monoclonal) to 0 (where no two clones are identical) and is calculated as ‘the square root of the sum over all observed rearrangements of the square fractional abundances of each rearrangement’. Observed Richness is simply the number of clones sequenced, termed ‘productive rearrangements’ and relative richness is the number of unique clones divided by the total number of clones in each sample.

FIGURE 5.

Declined TCR repertoire diversity in newly released pTreg cells from mice with age‐mimicking thymic atrophy. RTE‐pTreg cells were FACS sorted from peripheral T cells (lymph nodes and spleen) of dual‐reporter chimera mice, and the CDR3 regions of the TCRβ chain were analysed by TCR repertoire sequencing. (a) Cell sorting strategy for obtaining recent thymic emigrants (RTEs), which express Rag promoter‐driven GFP, and pTreg cells, which express FoxP3‐driven RFP along with CD4⁺CD25⁺ markers. (b) Productive frequency of the top 10 clones (top blue portion of the bars) from three individual FC (with atrophied thymus) and three individual FF (with normal thymus) mice. (c) Results of Simpson's clonality index of RTE‐pTreg cells from FF and FC mice. (d) Observed richness based on total number of productive rearrangements. (e) Relative richness based on unique rearrangements/total productive rearrangements. Simpson's clonality was calculated by ImmunoSEQ, while a Student's t‐test was used to compare two groups. Each symbol represents cells from an individual animal. **P < 0·01

Statistics

RQ‐MFI was used to normalize MFI data to the control (FFM) group by dividing each raw MFI value by the average control group (FFM) MFI for each separate set of experiments. Prism GraphPad software (Prism‐8) was used for data analysis. For evaluating differences between two groups, an unpaired two‐tailed Student's t‐test was used. For comparison between multiple groups, a one‐way ANOVA with Bonferroni correction for multiple comparisons was employed. Differences were considered statistically significant for *P < 0·05, **P < 0·01 and ***P ≤ 0·001.

RESULTS

Thymic atrophy affected agonist selection of a mock‐self‐Ag specific tTreg clone despite relatively normal tTreg polyclones

Previously, we have reported that the aged, atrophied thymus has reduced capacity for negative selection, exhibited by increased self‐reactive T cells [4]; however, it does not impair tTreg cell generation, as displayed by an increased ratio of RTE tTreg vs. tTcon cells either with naturally aged or with age‐mimicking atrophied thymus [6]. This enhancement of tTreg generation may be compensatory, as it seems to result at the expense of negative selection [11, 33]. However, we did not have sufficient evidence to determine whether a given self‐antigen specific Treg clone was also normal or enhanced in the aged, atrophied thymus, although it was shown that a self‐Ag (prostate)‐specific Treg clone (TCAF3) was lost from the Treg repertoire in an Aire ^−/− mouse model [19, 21].

Herein, we examined a mock‐self‐Ag (OVA)‐specific Treg clone in the murine thymus under conditions mimicking natural age‐related thymic atrophy. In order to evaluate this, we utilized a T‐cell system reconstitution model via thymic transplantation, in which a chimera was generated by engrafting FCM (or FFM control) newborn mouse thymic lobes under the kidney capsule of OT‐II TCR‐Tg (Rag sufficient in order to provide an Ag‐specific clone within a polyclonal background) recipient mice (Figure 1a). Three to four weeks after the host mouse progenitors entered the engrafted thymic lobes, we then observed the developmental outcomes of the host progenitors (OT‐II TCR‐Tg) and donor thymus TECs, which express OVA and were induced with/without conditional FoxN1 knockout. Although total cell numbers of tTreg and tTcon cells were reduced in atrophied thymuses, as expected (Figure 1c,d), we found that the mock‐self‐Ag specific Treg clone was quantitatively impaired (in proportion and ratio of Treg/Tcon) in the FCM atrophied thymus, although the generation of pan‐Treg cells remained relative unchanged (Figure 1b,e, right panel). The results imply that although thymic atrophy does not reduce overall pan‐tTreg generation, alterations in thymic function perturb the generation of this mock‐self‐Ag‐specific tTreg clone. In accordance with our previous findings that the atrophied thymus has impaired negative selection [4], we saw that specific tTcon cells were increased in proportion in mice with thymic atrophy (Figure 1f, right panel). The proportion of specific tTreg cells (Figure 1e, right panel) and ratio of specific tTreg/specific tTcon cells (Figure 1g, right panel) in CD4^SP were reduced. In sum, age‐mimicking atrophied thymus impaired negative selection of antigen‐specific tTcon (increased) and agonist selection of tTreg (decreased) cells, but not total, polyclonal tTreg cells.

FIGURE 1.

Selection of OT‐II specific tTreg cells was reduced, in mOVA atrophied thymus immune reconstitution model. (a) Workflow of immune reconstitution via OT‐II TCR‐Tg progenitor cells seeding into FFM or FCM newborn thymuses engrafted under the kidney capsule of OT‐II TCR‐Tg mice. (b) Flow cytometric gating strategy, showing % pan‐tTregs and % OT‐II specific Tregs in the engrafted thymuses. (c) pan‐ and specific tTreg cell numbers in engrafted thymuses, (d) pan‐ and specific tTcon cell numbers in engrafted thymuses, (e) % pan‐ and specific tTreg cells in the engrafted thymuses, (f) % pan‐ and specific tTcon cells in the engrafted thymuses, (g) ratios of pan‐tTregs/pan‐tTcons and specific tTregs/specific tTcons. A Student t‐test was used to determine statistical significance between two groups. All P‐values were calculated by mean ± SD, and each symbol represents an individual animal. ‘NS’, not significant. Experiment was repeated at least four times. *P < 0·05, **P < 0·01 and ***P ≤ 0·001

Thymic atrophy resulted in reduced proportion and number of a mock‐self‐Ag specific pTreg clone in the periphery, despite maintaining normal polyclonal (pan)‐pTreg levels

Given that the generation of mock‐self‐Ag (OVA)‐specific tTreg cells is impaired in the atrophied thymus (Figure 1), we asked whether they can expand to comparable levels in the periphery. Thus, we examined the peripherally circulating (termed for our study, pTreg) cells in our immune reconstitution BM chimera model using OT‐II TCR‐Tg progenitors (on Rag ^−/− background) given with a fixed ratio with WT progenitors to provide a polyclonal set of progenitors to irradiated recipient mice (Figure 2a, flow cytometric gating strategy in Figure 2b). We observed that although total splenic lymphocytes and total pTreg cells were not decreased (Figure 2c), the mock‐self‐Ag‐specific pTreg cells maintained the quantitative reduction found in both proportion (Figure 2d, right panel) and absolute cell numbers (Figure 2e, right panel) in the spleen of FCM mice with atrophied thymus. However, pan‐pTreg cells in the FCM mice with the atrophied thymus still were increased in proportion (Figure 2d, left panel) and showed no significance in absolute number (Figure 2e, left panel) compared to their counterparts (FFM) with normal thymus.

FIGURE 2.

OT‐II specific pTreg cells were reduced, in mOVA recipient mice with age‐mimicking thymic atrophy. (a) Workflow of immune reconstitution via BM chimeras with mixed WT (CD45.2) and OT‐II (CD45.1, Rag ^−/− background) progenitors into FFM or FCM recipient mice. (b) Flow cytometric gating strategy, showing % pan‐pTregs and % OT‐II specific pTregs in the recipient spleen. (c) Absolute splenocyte and pTreg cell numbers from recipient spleens. (d) Proportions of pan‐ and specific pTreg cells in the spleen. (e) pan‐ and specific pTreg cell numbers in spleen. (f) % pan‐ and specific pTcon cells in spleen. (g) pan‐ and specific pTcon cell numbers in spleen (h) Ratios of pan‐pTregs/pan‐pTcons and specific pTregs/specific‐pTcons in spleen. (i) Proportions of specific pTreg or ‐pTeff (OT‐II TCR‐Tg Tcon in the RIP‐mOVA‐Tg pancreas) cells in recipient pancreas. The statistical method is the same as in Figure 1. Each symbol represents cells from an individual animal. Experiment was repeated at least four times. *P < 0·05 and ***P ≤ 0·001

Since the mOVA‐Tg mice have mOVA expression in peripheral tissues, the OT‐II TCR‐Tg T cells encounter a low‐level specific‐Ag stimulation spontaneously. Therefore, CD4^SP pTcon cells could undergo expansion. We found that the Ag‐specific pTcon cells were increased in proportion (Figure 2f, right panel), though not in cell number (Figure 2g, right panel). The pan‐pTcon cells were decreased in both proportion and number (Figure 2f, left panel) in FCM mice, suggesting that the source of the increased Ag‐specific Tcon cells is likely the atrophied thymus rather than peripheral expansion. This resulted in an overall decreased ratio of specific Treg/specific Tcon cells (Figure 2h, right panel), suggesting that the balance could be tipped against self‐tolerance of this mock‐self‐Ag in the mice with thymic atrophy.

In addition, we analysed CD4^SP pTreg and pTcon cells in the pancreas since mOVA‐Tg is driven by the rat insulin promoter, and OVA‐specific Teff cells could induce pancreatitis if the OVA‐specific Treg cells are impaired enough to break central tolerance in combination with specific CD8^SP Teff cells and/or auto‐antibodies [34, 35, 36, 37]. The results showed that self‐Ag‐specific Treg cells were reduced and self‐Ag‐specific Teff cells were increased in the pancreas of mice with thymic atrophy (FCM) (Figure 2i). However, we did not find manifestations of diabetes in these mice based on blood glucose measurement (Figure S4), probably due to lack of intrinsic OVA‐specific CD8^SP Teff and/or antibodies [34, 35, 36, 37]. In sum, these results suggest that this mock‐self‐Ag‐specific Treg clone in the peripheral lymphoid organs and specific tissues indeed has a quantitative defect.

The suppressive capacity of the mock‐self‐Ag specific pTreg clone from mice with thymic atrophy was functionally defective

Given the reduction in proportion and number of these specific Treg cells in the periphery, we asked whether their function of suppressing specific Teff cell proliferation was also defective when assessed at equal cell numbers. We therefore set up a Treg suppression assay using flow cytometrically sorted OVA‐specific Treg‐enriched (CD25⁺) live cells from either FFM or FCM BM chimera mice (model in Figure 2a). In order to exclude any possible alterations to the OVA‐specific Teff cells from mice with thymic atrophy, OT‐II specific Teff cells were taken from unmanipulated OT‐II TCR‐Tg (Rag ^−/− background) mice. These Teff cells were co‐cultured with various groups of pTreg cells (WT, FFM, and FCM). The use of WT pTreg cells served as a control for non‐specific pTreg bystander suppression, since WT mice do not express mOVA as a self‐antigen and therefore should not generate mOVA‐specific Treg cells.

The results showed that sorted OT‐II TCR‐Tg (based on Vα2Vβ5 staining) CD4^SPCD25⁺ pTreg‐enriched cells (Figure 3a), derived from chimera mice with atrophied thymus (FCM), exhibited significantly reduced suppression of OT‐II TCR‐Tg CD4^SPCD25^−neg Teff cells (Figure 3b), compared to their counterparts derived from chimera mice with normal thymus (FFM) (Figure 3b). Surprisingly, the specific Treg cells from FCM chimeras suppressed similarly to the non‐specific (pan) WT Treg control group, that is no Ag‐specific suppressive function (Figure 3b). Therefore, mice with atrophied thymus could contain a compromised OVA‐specific Treg population that possesses not only a quantitative defect in proportion and number, but also exhibit a functional defect, since they cannot efficiently suppress OVA‐specific Teff cells during OVA‐peptide stimulation. In addition, this result was different from our previous report when we observed that pan‐tTreg cells derived from the atrophied thymus possessed normal suppressive function to pan‐Teff cells under non‐specific stimulation via CD3 and CD28 receptors [6].

The defective Ag‐specific Treg clone in mice with thymic atrophy potentially possessed an intrinsic defect

Treg cell function is closely associated with the expression levels of FoxP3 [38]. Solid evidence also shows that once pTreg cells in individuals with autoimmune disease, such as multiple sclerosis patients or mice with experimental autoimmune encephalomyelitis (EAE), encounter self‐Ag, FoxP3 expression is diminished or even undetectable. This phenotype is called Treg instability [39, 40]. Therefore, the expression levels of FoxP3 could be indicative of intrinsic defects associated with autoimmune‐prone Treg cells. To investigate potential intrinsic defects in our mock‐self‐Ag specific Treg population derived from mice with atrophied thymus, we investigated the expression intensity (relative quantitative – (geometric) mean fluorescence intensity, RQ‐MFI) of FoxP3 protein levels in tTreg and pTreg populations using flow cytometry.

Upon investigation, the expression levels of FoxP3 in pTreg cells from the spleen and pancreas (Figure 4) were significantly decreased in OVA‐specific pTreg cells, but not in pan‐pTreg cells in FCM chimeric mice (Figure 4c,d) with low‐level, endogenous OVA‐specific stimulation. This might explain why the OVA‐specific Treg cells from mice with thymic involution demonstrated reduced suppressive function, as shown by the Treg suppression assay (Figure 3).

FIGURE 4.

Reduced FoxP3 expression in OVA‐specific pTreg cells from mice with age‐mimicking thymic atrophy post‐OVA‐specific stimulation. Splenic or pancreatic cells from BM chimeric mice (workflow in Figure 2a) were analysed for FoxP3 expression (RQ‐MFI) in CD4⁺CD25⁺ population after spontaneous (in vivo) mOVA‐Tg stimulation. (a) Flow cytometric gating strategy for MFI of FoxP3 quantification showing % pan‐pTreg cells and % specific pTreg cells (b) Representative histogram of FoxP3 MFI gating for pan‐Treg and specific Treg cells from FCM (blue line) and FFM (yellow line) mice compared to isotype control (red line). (c) Quantification for RQ‐MFI of FoxP3 expression in pan‐ and specific pTreg cells from FCM and FFM spleen. (d). Quantification for RQ‐MFI of FoxP3 expression of pan‐ and specific pTreg cells from FCM and FFM pancreas. The statistic method is the same as in Figure 1. Each symbol represents cells from an individual animal. Experiment was repeated at least two times. *P < 0·05 and ***P ≤ 0·001

Treg development and function normally requires IL‐2 and TGF‐β signalling. We previously found that expression levels of IL‐2 and TGF‐β are normal in the atrophied thymus [6]. Herein, we examined expression levels of IL‐2 receptor‐α (CD25) in the atrophied thymus and peripheral lymphoid organs to determine whether IL‐2 signalling capacity is affected in these Treg cells. CD25 levels were no different for pan‐ and specific tTreg cells in the thymus of transplant chimera mice with and without thymic involution (Figure S1).

These findings suggest that this mock‐self‐Ag‐specific Treg clone potentially possesses an intrinsic defect showing inability to elevate FoxP3 expression upon Ag stimulation, although likely not due to insufficient IL‐2 signalling. This helps explain why these OVA‐specific pTreg cells derived from the atrophied thymus are ill‐equipped to suppress autoimmune Teff reactivity (Figure 3).

Thymic atrophy potentially affects Treg agonist selection resulting in decreased tTreg TCR repertoire diversity

Since the development of the TCR repertoire occurs in the thymus, age‐associated restriction of total TCR repertoire diversity has been at least partly attributed to reduction of naïve T‐cell output from the aged, atrophied thymus [41, 42, 43]. We wanted to know whether the TCR repertoire diversity of Treg cells derived from age‐mimicking atrophied thymus is also declined. We generated a radiation BM chimera via immune reconstitution of BM from Rag ^GFP and FoxP3 ^RFP dual‐reporter mice into host mice carrying TECs with floxed‐FoxN1 and CreER^T for conditional KO (FC) or without Cre‐Tg control (FF) with normal thymus [6]. Six to eight weeks after TM induction of thymic atrophy, the newly generated (recent thymic emigrant, RTE) pan‐pTreg cells, that is Rag ^GFP and FoxP3 ^RFP double‐positive pTreg cells, from host mouse spleen and lymph nodes were flow cytometrically sorted (Figure 5a). DNA sequencing was conducted for repertoire analysis of the TCR β‐chain CDR3 region (Adaptive ImmunoSEQ).

The results showed that the top 10 clones in FC RTE pan‐Treg cells (Figure 5b, left 3 bars) comprised a greater proportion of their total repertoire, compared to FF RTE pan‐Treg cells (Figure 5b, right 3 bars). TCR repertoire diversity via clonality index was used to quantify the clonal dominance, with ‘1’ indicating an entirely monoclonal sample (least possible diversity) and with ‘0’ indicating that no two TCRs are the same (highest possible diversity). The observed increased clonality in the FC group implies higher clonal expansion and thus less diversity (Figure 5c). In addition, there was no statistical significant difference in observed richness (total number of productive rearrangements or clones sequenced), although the FC mice tended to have fewer total productive rearrangements (Figure 5d). When accounting for number of unique rearrangements out of the total (relative richness), there was no difference between groups (Figure 5e). Therefore, although preliminary, we observed a trend for increased RTE pan‐Treg clonality, that is declined TCR repertoire diversity, in mice with age‐mimicking thymic atrophy compared to normal thymus controls. Reduced TCR diversity suggests that certain tissue‐specific Treg clones could be missing from the repertoire.

CONCLUSIONS

Given that increased self‐reactivity in the elderly is at least partially attributed to defects in thymocyte negative selection in the age‐related, involuted thymus [3], we extended our investigation to Treg generation and function associated with the aged, atrophied thymus. Our intent was to decipher the contradiction of enhanced tTreg generation and pTreg accumulation with decreased ability to suppress self‐reactivity in aged individuals. We focused on investigating the differences between pan‐Treg and specific Treg cells, and their TCR repertoires with various chimeric mouse models, carrying mock‐self‐Ag (mOVA) and OVA‐recognizing TCR‐Tg CD4^SP T cells, coupled with age‐mimicking (FoxN1 conditional KO) atrophied thymus.

We identified a quantitative defect in this self‐Ag‐specific Treg clone, exhibited by a significant decline in percentage of tTreg cells, suggesting a shortcoming in tTreg agonist selection, as well as in the percentage and absolute numbers of pTreg cells, suggesting reduced expansion in the periphery. These OVA‐specific Treg cells were significantly less able to suppress OVA‐peptide induced proliferation of OVA‐specific Teff cells in vitro. However, these defects could not be observed in polyclonal (pan)‐Treg cells from the same mice with age‐mimicking thymic involution, which is in agreement with our previous publication [6]. This may partially explain the seemingly contradictory phenomenon of immunosenescence and inflammaging. We do not rule out the role of pan‐pTreg mediated bystander effects, which have been shown to suppress excessive immune responses to both self‐ and non‐self‐antigens to maintain immune homeostasis [44, 45], although this effect is not as powerful as Ag‐specific suppression. This could explain why auto‐reactive Teff cells in the elderly usually induce a low‐level, persistent inflammation (inflammaging) rather than induction of bona fide autoimmune diseases. We further found that these self‐Ag recognizing Treg cells possessed a potential intrinsic defect exhibited as lower FoxP3 expression after mOVA‐Tg stimulation. We postulate that this could be rooted in potential defects during tTreg TCR agonist selection in the age‐mimicking atrophied thymus, resulting in reduced repertoire diversity.

Our model of immune reconstitution with mock‐self‐Ag mOVA‐Tg and OVA‐recognizing TCR‐Tg OT‐II cells follows the current paradigm for thymocyte central tolerance establishment, involving negative selection of clones with high avidity for self‐antigens and the agonist selection of clones with intermediate to high avidity for self‐antigens to develop tTreg cells [11, 46]. Based on the current paradigm and the observations discussed herein, we provide an illustration of our hypothesized mechanism underlying these findings (Figure S3).

tTreg cell development is dependent on multiple signals and cumulative signalling strength or avidity within the thymic microenvironment. In addition to interactions between self‐peptide‐MHC‐II complex and TCR, IL‐2 signalling via IL‐2Rα on Treg cells is also critical. In our previous publication, we demonstrated that expression of IL‐2 was not reduced in the FC nor in the naturally aged atrophied thymus [6]. Similarly, herein, we found that expression of IL‐2Rα, or CD25, was also no different between Treg cells in the normal versus the atrophied thymus. TGF‐β is also an important factor involved with tTreg agonist selection [47], which was not reduced in our previous investigation [6]. Therefore, we believe these changes in certain tissue‐specific Treg cells are mainly due to changes in TCR signalling strength [11, 46, 48, 49, 50]. The reduction in FoxP3 is likely responsible for the observed reduction in suppressive function, as FoxP3 is a master regulator of Treg function. Insufficient IL‐2 signalling has been shown to exacerbate pTreg instability via decreased FoxP3 expression during specific antigen stimulation of certain specific pTreg cells [40, 51], since signalling through CD25 is upstream from peripheral enhancement of FoxP3 expression [52, 53]. However, we did not find reduced expression and IL‐2Rα. Therefore, future study of the epigenetic modifications of FoxP3 expression [54] is needed to clarify this underlying mechanism.

As for other aspects of the thymic microenvironment, other stromal cells, such as thymic dendritic cells, have been shown to decrease in number in the aged thymus [55] and their function appears to be less efficient compared to young thymic‐derived dendritic cells [56]. However, their overall ratio to the number of thymocytes developing in the aged thymus appears to remain intact [56]. It would be interesting to assess how age‐related thymic atrophy impacts promiscuous self‐antigen expression by thymic dendritic cells since they interaction with mTECs to facilitate this process [57]. It is also worth mentioning that other aspects of promiscuous self‐antigen expression for thymocyte selection and central tolerance induction, such as FEZF2, are beginning to be of interest, especially since the mTECs that express AIRE do not appear to overlap with those that express FEZF2 [18]. Moreover, the effects of other age‐related changes to the thymic microenvironment on Treg selection is an under‐investigated field.

Finally, based on our findings in the TCR‐Tg model, we wanted to assess a potential outcome of decreased Treg TCR diversity in a system in which only the thymus was manipulated to mimic thymic ageing, namely our FoxN1 conditional KO mice. Treg TCR repertoire normally encompasses a broad range of self‐ and some non‐self‐recognizing T cells [58, 59]. Reduction of repertoire diversity of the Treg population results in unchecked autoimmunity [60, 61]. The overall T‐cell TCR repertoire diversity declines with age due to thymic atrophy and increased oligoclonal expansion of peripheral memory T cells [41, 42, 43]. Herein, although our initial TCR diversity sequencing data are preliminary, due to limited animals and small sorted cell sample sizes, we observed a trend for declined diversity in the newly generated (RTE) Treg pool from mice with age‐mimicking thymic involution. This finding requires future investigation at the single‐cell level. However, the prospect of a decreased tTreg repertoire diversity as a result of thymic atrophy is in line with current available evidence that certain tissue‐specific Treg TCRs could not be agonist selected from the dysfunctional thymus, such as the missing prostate TCAF3‐specific Treg clone from AIRE gene‐deficient thymus [19, 21], though total (pan‐) Treg clones are not reduced.

Although there are some limitations to using a TCR‐Tg system, such as the irregularities involved during early thymocyte development due to the early TCRα and TCRβ expression, here we studied the impacts of thymic atrophy on the latter stages of Treg agonist selection, which occurs after CD4^SP lineage is determined. We also maintained a polyclonal pool of T‐cell progenitors in these mice to ensure a more physiologically relevant selection process. Given the difficulties of identifying a single Ag‐specific T‐cell clone within a naturally heterogeneous and unmanipulated pool, such a model was necessary and allowed us to avoid other artificial constructs required for such assessments in non‐TCR‐Tg models, such as artificial expansion of a single clone via immunization with cognate antigen [4], or the utilization of a fixed TCR β‐chain [19] in order to ease identification of one clone via diminished overall diversity. These findings from our TCR‐Tg model provided the foundation for our tTreg TCR diversity sequencing in a non‐TCR‐Tg system and are informative for future investigations of antigen‐specific Treg selection in less manipulated models. Further investigation into the identification of various natural Treg clones that may be effected within the polyclonal repertoire of non‐TCR‐Tg mice will shed more light on the observations reported herein.

Taken together, our findings highlight the differential impacts of thymic involution on pan‐ and antigen‐specific tTreg cell generation. In this scenario, the effects of age‐related thymic involution did not affect pan‐Treg cell generation, but resulted in intrinsic impairment of a mock‐self‐Ag specific Treg clone. Taken holistically, some self‐Ag‐specific tTreg clones may even fail to undergo agonist selection under these conditions, possibly creating TCR repertoire holes in the aged T‐cell regulatory system. Although further study is needed, the implications of such Treg repertoire holes would further elucidate the underlying mechanisms of inflammaging and help explain why relatively increased pan‐Treg cells are unable to attenuate inflammaging in the elderly.

CONFLICT OF INTEREST

The authors declare that they have no conflicts of interests.

Supporting information

Fig S1

Fig S2

Fig S3

Fig S4