|Home | About | Journals | Submit | Contact Us | Français|
The thymic medulla plays an essential role in the generation of central tolerance by eliminating self-reactive T-cell clones through thymic negative selection and developing natural regulatory T cells. Age-related FoxN1 decline induces disruption of medullary thymic epithelial cells (mTECs). However, it is unknown whether this perturbs central tolerance to increase autoimmune predisposition in the elderly. Using a loxP-floxed-FoxN1 (FoxN1flox) mouse model, which exhibits a spontaneous ubiquitous deletion of FoxN1 with age to accelerate thymic aging, we investigated whether disruption of steady-state thymic medulla results in an increase of autoimmune-prone associated with age. We demonstrated age-associated ubiquitous loss of FoxN1flox-formed two-dimensional thymic epithelial cysts were primarily located in the medulla. This resulted in disruption of thymic medullary steady state, with evidence of perturbed negative selection, including reduced expression of the autoimmune regulator (Aire) gene and disrupted accumulation of thymic dendritic cells in the medulla, which are required for negative selection. These provoke autoimmune phenotypes, including increased inflammatory cell infiltration in multiple organs in these mice. This finding in an animal model provides a mechanistic explanation of increased susceptibility to autoimmunity in aged humans, although they may not show clinic manifestations without induction.
Thymic microenvironments, composed primarily of an integrated three-dimensional meshwork of thymic epithelial cells (TECs), foster T-cell development by positive (leading to self-MHC restriction) and negative (resulting in elimination of self-reactive T-cell clones in order to tolerate self-antigens in the periphery) selection to shape the T-cell receptor. The TECs can be divided into cortical and medullary TECs (cTECs and mTECs), which are located in cortical and medullary regions, with each forming distinct microenvironments. The thymic medulla provides a unique microenvironment that is essential for the generation of central immune tolerance [1–3] through two major mechanisms: depleting self-reactive T-cell colonies (clonal deletion) by negative selection  and producing CD4+ natural regulatory T cells (nTregs) by “deviated development” [5, 6]. Clonal deletion is mainly dependent on co-operation of both mTECs and thymic dendritic cells (tDCs), in which mTECs serve primarily as a reservoir of peripheral tissue-specific self-antigens, while tDCs take up and present peripheral tissue-specific self-antigens in the unique microenvironment of the thymic medulla [7, 8]. The processes of peripheral tissue-specific self-antigen presentation by mTECs and antigen transfer from mTECs to tDCs are regulated by the autoimmune regulator (Aire) gene [9–12].
Many studies show that defects in mTECs and disruption of thymic medullary microstructure, even a reduction in the size of the medulla, can affect the generation of central tolerance by decreasing the efficiency of Aire and/or tissue-specific self-antigen presentation [13–15], leading to the generation of fewer , or deficient CD4+nTreg cells , and thereby increasing the incidence of autoimmune diseases. The functional thymic epithelium is an organized three-dimensional meshwork. However, in our previous work, we found that loss of epithelial cell-autonomous gene forkhead box N1 (FoxN1) altered the TEC meshwork from a three-dimensional-stereo to a two-dimensional-polarized cystic structure . Cystic thymus is usually associated with autoimmune diseases, seen in several mouse lines including, but not limited to, ALY (severe immunodeficiency) , NOD (insulin-dependent diabetes) , and NZB (lupus-like syndrome)  mice.
Previous findings show that expression of FoxN1 is declined in aged thymus [22, 23]. We confirmed that reduced FoxN1 expression is causally related to age-related thymic involution  in an accelerated thymic aging mouse model  established by crossbreeding a ubiquitous promoter-driven Cre-recombinase and estrogen-receptor fusion protein (uCreERT) transgenic mouse , which has a low-level spontaneous leakage of the CreERT transgene due to incomplete estrogen receptor (ER) blockage in vivo , with a loxP-floxed FoxN1(fx) knock-in mouse . Aged individuals have an increased incidence of autoimmune diseases [27, 28], and many autoimmune diseases such as rheumatoid arthritis  and multiple sclerosis  are associated with age. This, at least in part, is due to aged thymic defects in the context of genetic background and microbial environmental factors.
Recent studies indicate that there is a lifelong need to establish tolerance to self , because T cells are continuously generated from a young to a very old age . Although aged individuals have an atrophied thymus, this atrophied thymus still processes T-cell generation, commencing with receptor-negative T-cell progenitors that need to undergo positive and negative selection in TEC-constituted microenvironment. However, the aged thymic medulla is significantly different from the young one. Therefore, it is largely unknown whether the aged thymic medulla is still able to conduct normal negative selection to generate normal central tolerance, in comparison to a young thymic medulla. If not, whether the generation of central tolerance is perturbed?
In this report, using a FoxN1 decline-induced accelerated thymic aging mouse model , we demonstrate that age-related FoxN1 decline-formed two-dimensional thymic cysts appeared primarily in the thymic medulla, thereby resulting in reduced expression of the Aire gene and Aire-dependent peripheral tissue-specific self-antigen genes in the thymus by a mechanism of decreased gene expression due to a loss of cells. Furthermore, we found that a disorganized medulla disrupted the ability to attract thymic dendritic cells (tDCs) accumulated in the medulla. These phenotypes are associated with perturbation of negative selection, which thereby provokes autoimmune phenotypes, including increased inflammatory cell infiltration and autoantibody deposition in multiple organs, as well as increased antinuclear antibodies in the serum. We, therefore, conclude that an increased susceptibility to autoimmunity in the elderly is potentially associated with aging-induced disruption of thymic medullary microstructure in the postnatal thymus, which results mainly from a decline in FoxN1 expression.
The fx/fx-uCreERT mice (C57BL/6 genetic background) were generated by crossbreeding loxP-floxed FoxN1 (FoxN1flox, fx) mice, described previously  (the Jackson Lab #012941), with CAG-Cre/Esr1 (ubiquitous CreERT expression, uCreERT) transgenic mice (the Jackson Lab, #004682) . The fx/fx-uCreERT mice undergo spontaneous FoxN1 deletion with age  due to incomplete estrogen receptor (ER) blockage in vivo . All young fx/fx-uCreERT mice were genotyped. Primer sequences were reported previously . Mouse ages are indicated in each figure legend, dividing into young (1–2 months old); early middle-aged (6–9 months old); middle-aged (±12 months old); and aged (over 18 months old) groups. Control group was wild type (WT) and fx/fx-only mice, since loxP insertion does not affect FoxN1 gene expression , the fx/fx-only mice without Cre-recombinase are considered same as WT mice. In some experiments, we also used fx/+-heterozygous mice for controls. Aged WT mice were purchased from the National Institute on Aging. All animal experiments were in compliance with the protocols approved by the Institutional Animal Care and Use Committee of the University of North Texas Health Science Center at Fort Worth, in accordance with guidelines of the National Institutes of Health.
Thymic stromal cells were enriched through three-cycle enzymatic digestions using collagenase and DNase-I as previously reported [33, 34]. Total RNA from these thymic stromal cell-enriched thymic cells was reverse transcribed with the SuperScriptIII cDNA kit (Invitrogen). Real-time PCR was performed by standard techniques in a Step-One-Plus real-time PCR system (Applied Biosystems), with TaqMan-probes for Aire, Spt1, and Chrna1 (Applied Biosystems qPCR Cat# Mm00477461_m1, Mm00839568_m1, and Mm00431627_m1, respectively) or SYBR-green reagents [Insulin I, Insulin II, and I-FABP primers previously published ], as well as XCL1 primer . Samples were normalized to 18S RNA or GAPDH internal control. The results were analyzed by the relative quantitative (RQ) gene expression ΔΔCT method, setting the value for average controls as 1.0.
Mouse organs (the thymus, liver, lung, pancreas, brain, and salivary and lacrimal glands) from age-matched fx/fx-uCreERT mice and control mice were fixed, cut into 5 μm-thick sections, and stained with hematoxylin and eosin (H&E).
To detect serum autoantibody, 6 μm cryosections of lacrimal gland, salivary gland, adrenal gland, stomach, pancreas, ovary, prostate, intestine, testicle, and eye were prepared from young RAG−/− and WT (tissues with an enhanced block) mice. After fixation in cold acetone for 5 min, sections were incubated with 0.1% NP-40/TBS for 5 min and blocked with SuperBlock (Thermo scientific #37516) for 15 min at room temperature. The slides were blocked again with Affinipure Fab Fragment Donkey Anti-Mouse IgG (H+L) (1:10) (Jackson ImmunoResearch Laboratories, #715-007-003) for 1hr at 37°C. The sections were divided into two groups: one was incubated with sera from age-matched fx/fx-uCreERT mice, while the other one was incubated with sera from control fx/fx-only mice (1:100 dilutions) at 4°C over night, then stained with Cy3-conjugated donkey anti–mouse IgG (Jackson ImmunoResearch Laboratories, #715-166-151) for 30 min at room temperature. For immunofluorescence staining of thymic tissues, thymic cryosections (6 μm) were fixed in acetone and blocked in 10% donkey serum/TBS. Primary antibodies were rabbit anti-mouse Keratin-5 (K5, Covance), rabbit anti-mouse Claudin-3,4 (Cld3,4, Invitrogen, #34–1700 and #36–4800), rat anti-mouse K8 (Troma-1 supernatant), and rabbit anti-Aire (Santa Cruz, #M-300). Secondary reagents included Cy3-conjugated donkey anti-rabbit IgG, FITC-conjugated donkey anti-rat IgG (Jackson ImmunoResearch Laboratories) or Alexa-Fluor-488-conjugated anti-rabbit IgG (Invitrogen).
Anti-Nuclear Antibody concentration in mouse sera was determined using an ELISA kit (Alpha Diagnostic International, #5210), following the manufacturer’s instructions. In brief, each serum sample was diluted 1:100 and incubated in an antigen coated well. Samples were run in duplicate and the data represent the mean of the two values. The second antibody was HRP-conjugated anti-mouse IgG (H/L), and the substrate was TMB (3,3′,5,5′-tetramethylbenzidine). The absorbance was measured at 450 nm with the BioTek ELx800 ELISA reader. Meanwhile, a concentration standard curve was generated using purified immunoglobulin isotypes supplied in the kit. Concentration (μg/ml) of Anti-Nuclear Antibody in mouse serum samples was calculated against this standard curve.
For evaluation of group differences, the unpaired two-tailed Student’s t-test was used assuming equal variance. Differences were considered statistically significant at values of p < 0.05.
In previous work, we demonstrated that FoxN1flox (fx) deletion mediated by uCreERT leakage in fx/fx-uCreERT mice induced thymic aging characterized by accelerated age-related thymic involution . We also reported that the K14 promoter-driven Cre-recombinase-mediated FoxN1flox deletion, i.e. FoxN1 K14 knockout resulted in increased generation of two-dimensional thymic cysts in the thymic medullary region . We then wanted to determine in which region, the medulla or cortex, uCreERT-induced two-dimensional thymic cysts are primarily located in the fx/fx-uCreERT mouse thymus. Therefore, we observed the thymus of early middle-aged fx/fx-uCreERT mice without tamoxifen (TM) treatment, and detected a large number of two-dimensional thymic cysts (Fig. 1) similar to our previous finding in FoxN1 K14 knockout mice with Cld3,4+ epithelial lining (Figs. 1 C and D). These two-dimensional thymic cysts were mainly located in the medullary region (Figs. 1 B and D), in spite of FoxN1flox deletion was ubiquitous in the both medulla and cortex. The results imply that ubiquitous FoxN1flox deletion-induced thymic deterioration in the postnatal thymus is either limited primarily to or originates in the medulla. This change results in the disruption of the three-dimensional medullary steady state .
The medulla plays a well-characterized role in the generation of central immune tolerance through thymocyte negative selection. One of mechanisms for central tolerance generation involves the presentation of tissue-specific self-antigens, which are expressed in a promiscuous fashion , and is primarily controlled by the Aire (autoimmunity regulator) gene [9–11]. Both Aire and tissue-specific self-antigens including Aire-dependent and Aire-independent antigens are expressed in mature mTECs [9, 37, 38]. Earlier, we found that Aire gene expression was decreased in FoxN1 K14 knockout mice . Here, we confirmed that Aire+ TECs were also reduced in the early middle-aged fx/fx-uCreERT mice without TM treatment (Figs. 2A and B), and expression of the Aire gene itself and most Aire-dependent tissue-specific genes were significantly decreased (Fig. 2C). The results further confirmed that age-related uCreERT-mediated FoxN1flox deletion in the postnatal thymus disrupts Aire gene expression, as well as most tissue-specific self-antigen expression, resulting from the disrupted steady-state medulla. Decreased Aire expression is unlikely due to direct interaction of FoxN1 with Aire, but probably due to gene expression in disrupted medullary epithelial cells. Therefore, we could not use defective mTECs themselves and genes, such as Keratin 5, expressed on these defective mTECs as normal controls in Figure 2. This disruption potentially perturbs negative selection in the medulla. The results coincide with findings from others [14, 15].
Thymic dendritic cells (tDCs) are also important for establishing central immune tolerance because in addition to presenting ubiquitous antigens, tDCs along with mTECs present an array of peripheral tissue-specific self-antigens to direct negative selection [reviewed in [2, 3, 13]] and furthermore, tDCs process and cross-present tissue-specific self-antigens to augment deletion of self-reactive T cells . A recent report shows that tDC accumulation in the medulla is facilitated by XCR1 expression in tDCs and XCL1 expression in mTECs . Disruption of tDC accumulation in the medulla, which is regulated by Aire, elicits autoimmune inflammatory lesions . To determine whether defects in the medulla in early middle-aged fx/fx-uCreERT mice induce dysfunctional tDC accumulation in the thymic medulla, we observed tDC distribution in the thymus of these mice, and found that the distribution of tDCs was reduced in the medullary region and pronounced in the cortical region (arrows in Fig. 3A, right panel), while in the control group (Fig. 3A, left panels) most tDCs were located in the medullary region. Same phenotype was found in the naturally-aged (21-month-old WT) thymus (33) (Fig. 3B right panel), in which FoxN1 expression is decreased (22,23). In addition, we found that XCL1 expression in sorted CD45−MHC-Class-II+ epithelial cells was significantly decreased in naturally-aged mice (Fig. 3C). This may account for the inability of mTECs to attract tDCs. However, a dysfunction of tDCs themselves is unlikely because we did not find a decrease in XCR1, which is expressed in tDCs, by real-time RT-PCR (data not shown).
As FoxN1 decline induces disruption of normal thymic medullary microstructure and homeostasis along with decreased expression of Aire and most tissue-specific self-antigens, and decreased tDC accumulation in the medullary region of the thymus, we set out to determine whether these changes, which perturb negative selection, are enough to elicit autoimmune inflammatory lesions. We observed inflammatory cell infiltration in multiple organs, including the liver, lung, pancreas, brain, and salivary glands in paraffin-embedded sections with H&E staining. We found increased inflammatory cell infiltration in multiple organs in early middle-aged (6~9 months old) fx/fx-uCreERT mice (Figs. 4A–E) in terms of both inflammatory cell infiltrated areas and infiltrated cell cluster numbers (Fig. 4B), particularly in the salivary glands (71.4% positive cases), liver (57.1% positive cases), and lung (38.8% positive cases) (Fig. 4D). These results indicate that an autoimmune-phenotype does exist in early middle-aged fx/fx-uCreERT mice.
We also revealed a spectrum of serum autoantibody deposition by immunofluorescence staining of frozen tissue sections. Sera were collected from early middle-aged (9-month-old) fx/fx-uCreERT and age-matched fx/fx-only control mice. We found that autoantibody deposition occurred mostly in glands and the reproductive system, such as the lacrimal and salivary, the testicles and ovaries (Fig. 5). We also found that antinuclear antibodies in the sera of early middle-aged fx/fx-uCreERT mice were significantly elevated compared to age-matched fx/fx-only control mice (Fig. 6). Higher titers of antinuclear antibodies are indicative of autoimmune predisposition usually caused by disruption of central immune tolerance.
We previously showed that a conditional knockout of FoxN1 in the postnatal thymus causes acute deterioration mainly in the thymic medulla ; progressive decline of FoxN1 expression with age is causally related to age-related thymic involution ; and the naturally-aged atrophied thymus has a disorganized medulla . We also found that deletion of FoxN1 in K14+ TECs induces generation of a large number of two-dimensional thymic cysts in the medulla . However, it is unclear whether progressive ubiquitous decline (mimicking natural aging) of FoxN1 expression with increased age also induces two-dimensional thymic cysts primarily in the medulla. If the cysts are located primarily in the medulla, do these two-dimensional epithelial cysts influence the generation of central tolerance, given the role of the thymic medulla in this process? To determine these issues we used our previously reported FoxN1flox (fx) mice carrying the uCreERT transgene, which undergoes a low level of spontaneous ubiquitous deletion of FoxN1 with increased age. We found that middle-aged fx/fx-uCreERT mice had a disruption of the steady-state thymic medulla characterized by formation of a large number of two-dimensional thymic cysts in the thymic medulla, and exhibited reduced expression of the Aire gene and peripheral tissue-specific self-antigen genes. Furthermore, this disorganized thymic medulla disrupted tDC accumulation in the medulla. Therefore, it is very plausible that negative selection was perturbed in these mice because autoimmune phenotypes were provoked, which included increased inflammatory cell infiltration and autoantibody deposition in multiple organs, as well as increased antinuclear antibodies in the sera. Thus, we concluded that increased autoimmune predisposition in the elderly is possibly associated with aging-induced disruption of steady-state thymic medulla due to the decline of FoxN1 expression.
FoxN1 has distinct roles in the prenatal and postnatal thymus. In the prenatal thymus it controls both cTEC and mTEC patterning and differentiation to generate a functional three-dimensional epithelial meshwork from a two-dimensional epithelial thymic anlage [40, 41], while in the postnatal thymus it mainly controls mTEC differentiation in the medulla or cortico-medullary junction  to maintain epithelial homeostasis in a three-dimensional meshwork  and prevent age-related thymic involution [23, 42]. It is known that the FoxN1-null mutation in germline cells, which results in inborn dysfunction of the thymus with a lack of production of T cells, does not induce autoimmunity but causes primary immunodeficiency. However, a postnatal FoxN1 decline after the thymus is fully developed primarily affects the medulla, resulting in a reduction of naïve T-cell numbers and dysfunctional T-cell output. It is well-known that the thymic medulla is specialized for the generation of central immune tolerance. Negative selection takes place in the medulla to eliminate self-reactive T-cell clones from the T-cell repertoire . The efficiency of negative selection is dependent on the presentation of peripheral tissue-specific self-antigens, so-called “promiscuous gene expression” [37, 43], which is, in part, regulated by the Aire gene in mTECs [38, 44]. Although postnatal mutation in FoxN1 induces deterioration of mTECs  and reduces Aire+mTECs (Fig. 2), it is unlikely that FoxN1 directly controls Aire expression, as Aire mRNA was found in the thymus of FoxN1-null nude mice . However, the thymic medullary steady state is crucial for the efficiency of Aire expression and negative selection.
It is well-known that complete loss of the Aire gene (Airenull) causes autoimmune-polyendocrinopathy-candidiasis ectodermal dystrophy in humans [46, 47] and autoimmune phenotypes in mice [38, 48]. Even in heterozygous Aire+/− T-cell receptor transgenic mice, lack of one copy of Aire led to diminished thymic expression of both the endogenous insulin gene and the exogenous insulin transgene, resulting in a 300% increase in islet-reactive CD4+ T cells that escape thymic deletion, and dramatically increased progression toward diabetes , suggesting importance of dose-dependent Aire expression. However, there is insufficient information on whether thymic architecture abnormality-induced decline in Aire expression or decrease in Aire+ TEC numbers, but not null for Aire gene, can also induce autoimmune phenotypes. The autoimmune-prone NZD mouse strain was reported to have a thymic architecture abnormality-induced decline in Aire expression in the mTEC-high (MHC-class-IIhi,UEA-1hi) subset, and have systemic autoimmunity associated with age . In addition to decline of Aire and tissue-specific self-antigen expression (Fig. 2), conditional deletion of FoxN1flox resulted in the depletion of mTEC-high subsets  and was associated with age , which further indicates that postnatal FoxN1 decline-induced autoimmune predisposition arises due to deterioration of postnatal mTECs, particularly the mTEC-high subset.
mTECs and tDCs are two major antigen-presenting cell types in the medullary region responsible for conducting negative selection [7, 12]. tDCs should be sparse in the cortical region  and accumulated in the medulla in the normal thymus. tDC accumulation in the medulla is regulated by Aire-dependent production of XCL1 expression in the mTECs and XCR1 expression in the tDCs . Transfer of antigens from mTECs to tDCs requires both cell types to be in close proximity and is the regulated by Aire . Decreased XCL1 expression in naturally-aged thymus (Fig. 3C) did indeed result in disrupted tDC medullary accumulation, with large numbers of tDCs present in the cortical region and/or separation of mTECs and tDCs (Fig. 3). These directly perturb negative selection since mTECs cannot efficiently transfer antigens to tDCs, and an excess of cortical tDCs unnecessarily present self-antigens to cortical thymocyte subpopulations, which do not undergo negative selection.
In summary, postnatal FoxN1 decline-induced increase in autoimmune propensity suggests that spontaneous autoimmunity arises from the thymic medullay architectural abnormalities, thereby disrupting thymic medullay steady state. The mechanism is similar to NZB mouse model  and other mouse models . In the postnatal FoxN1 decline model, FoxN1 is not an autoimmunity susceptibility gene, but it plays an important role in postnatal TEC homeostasis to maintain postnatal thymic medullary steady state. Its decline, associated with an increase in age, induces the formation of two-dimensional thymic cysts, even in the naturally-aged thymus. The cysts were primarily located in the medulla, which disrupts the generation of central tolerance by perturbing negative selection, including a reduction of Aire+mTECs, reduction of peripheral tissue-specific self-antigen expression, and deterioration of tDC accumulation in the thymic medullary or transfer of antigens from mTECs to tDCs, thereby provoking autoimmunity. These findings revealed an indirect role of FoxN1 in the postnatal thymus, and the animal model has a potential to be used to study increased autoimmune predisposition in the elderly.
This work was supported by NIAID/NIH grants (R01AI081995 and 3R01AI081995-03S1) to D-M. S.
We have no competing financial interests.