In vitro co-culture systems for studying molecular basis of cellular interaction between Aire-expressing medullary thymic epithelial cells and fresh thymocytes

The mammalian thymus consists of two main compartments, cortex and medulla, that are composed of thymic epithelial cell (TEC) populations. Especially, the cortex is composed of cortical TECs (cTECs) and the medulla is composed of medullary TECs (mTECs). Those TECs share distinct functions. In postnatal thymus, the cortical and medullary compartments are formed under precise cellular process “thymic crosstalk”, which is a bi-directional cellular communication between mTECs and developing thymocytes (van Ewijk et al., 1994; Holländer et al., 1995).

In the cortex, some cTECs expressing distinct major histocompatibility complex (MHC) promote positive selection of immature (CD4⁻CD8⁻) thymocytes that are differentiating to CD4⁺CD8⁺ double-positive (DP) thymocytes. Interaction of T-cell receptor (TCR) on DP thymocytes with MHCs on cTECs triggers differentiation of the thymocytes (Bill and Palmer, 1989; Sha et al., 1988; Zepp and Staerz, 1988). The positively selected DP thymocytes in the cortex are then relocated to medulla.

In the medulla, maturation of DP thymocytes takes place along their thymic export (Takahama, 2006; Anderson et al., 2007). One of the crucial functions of medulla is to establish central self-tolerance by eliminating auto-reactive thymocytes. This negative selection is driven by self-antigen-presenting cells (self-APCs), which are specially selected populations of mTECs and medullary dendritic cells (mDCs).

The self-APCs promiscuously (ectopically) express a wide spectrum of peripheral tissue-specific antigens (TSAs) (Derbinski et al., 2001). The ectopically expressed large repertoires of TSAs are then presented to DP-thymocytes (Gallegos and Bevan, 2004; Ardavín, 1997; Amsen and Kruisbeek, 1998). It should be noted that ectopic expression of TSA genes in self-APCs is regulated, at least in part, by a transcription factor AIRE (autoimmune regulator) that we previously found deficient in the patients with autoimmune disease APECED (also known as APS1; OMIM no. 240300) (Nagamine et al., 1997; Aaltonen et al., 1997).

In the thymus of AIRE-deficient APECED/APS1 patients and of Aire-knockout (KO) mice, ectopic expression of several TSA genes was reduced, leading to impaired negative selection of T cells (Anderson et al., 2002; Ramsey et al., 2002). Moreover, reduced TSA gene expression was found in the mTECs of organ-specific autoimmunity (Gotter and Kyewski, 2004; DeVoss et al., 2006).

Autoimmune disease APECED/APS1 is characterized with a diverse array of clinical features, including functional failures of parathyroid gland and adrenal cortex, a high prevalence of auto-reactive T cells and autoantibodies, and immune destruction of multiple organs (Ahonen et al., 1990; Perniola et al., 2000). However, molecular pathogenesis of APECED/APS1 has not been fully understood. Thus, many questions remain to be clarified; how a single mutation of AIRE gene reflects aberrant expression of TSAs for target organs, how a single transcription factor AIRE regulates ectopic expression of TSAs and how special populations of mTECs/mDCs are selected through thymic crosstalk.

Recently, we established three mouse cell lines (Aire⁺TEC1, Aire⁺TEC2 and Aire⁺DC) that constitutively express Aire gene and exhibit most of the expected characteristics of mTECs/mDCs (Yamaguchi et al., 2011). These Aire-expressing cells (Aire⁺ cells) adhered to fresh thymocytes and induced apoptosis as if they mimic in vivo negative selection of T cells. Moreover, these Aire⁺ cells possessed characteristic features of self-APCs, exhibiting abilities to express a variety of peripheral TSAs, and several essential components of IPSM (immunoproteasome), IS (immunological synapse) and some TNFSFs (tumor necrosis factor super families).

In this study, we made several new findings such as that various critical cellular parameters elevated as Aire⁺cell density increased (semi-confluency vs confluency: sparse cells vs dense cell–cell contacted cells). We postulated that these Aire⁺ cells in culture may mimic in vivo differentiation process of mTECs/mDCs. Moreover, our in vitro co-culture system consisting of fractionated thymocytes and Aire⁺ cell lines implied possible existence of two distinct subtypes of thymocytes that may control the fate (dead or alive) of differentiating Aire⁺ cells. We will present the detailed intercellular interaction data to support these notions and the usefulness of Aire⁺ cell lines for in vitro study on “thymic crosstalk” will be discussed.

All animal experiments were performed in accordance with animal welfare regulations of Laboratory Animal Center, Keio University School of Medicine.

Three lines of Aire⁺ cells (Aire⁺TEC1, TE2 and DC) were established as described previously (Yamaguchi et al., 2011). Those Aire⁺ cells (1×10⁶ cells) were seeded in a 90-mm dish (SUMILON) containing DMEM-high glucose medium supplemented with 10% FBS, 100 units/ml penicillin and 100 units/ml streptomycin.

Those cells were grown at 37°C in 5% CO₂ for 32 hrs to get semi-confluent cultures (0.35×10⁷ cells) and for 72 hrs to get confluent cultures (1×10⁷ cells). Aire⁺TEC1 cells that overexpress FLAG-Aire fusion protein was produced by transfecting plasmid (p3×FLAG/Aire cDNA) as previously described (Yamaguchi et al., 2011). For a negative control of western blotting, mouse A9 skin fibroblast was used. Total RNAs were extracted from Aire⁺ cells using TRIzol reagent (Invitrogen). mRNA was prepared from total RNA using FastTrack MAG Maxi mRNA Isolation Kit (Invitrogen).

Synthesis of 1^st cDNA was carried out by reverse transcription from purified mRNA (0.5 µg) using Superscript III kit (Invitrogen) with oligo (dT₂₀) and random hexamer primer (Roche). qRT-PCR was performed by TaqMan method with Mouse Universal Probe Library Set (Roche), primers for various genes (Tables 1 and 2) and Fast Star Universal Probe Master (ROX) (Roche) on ABI PRISM 7700 Sequence Detection System (Applied Biosystems). Amounts of specific mRNAs were normalized to β-Actin mRNA.

Anti-mouse Aire protein antibody (anti-Aire-pAb): The synthetic peptides corresponding to the amino acid sequence 126–140 (PPRPPTKRKALEEPR) and 541–522 (DDSRPLAETPPFSS) of mouse Aire protein were conjugated with KLH, and used for immunizing mice (A&G Pharmaceutical Inc.). The primary antibodies used include: Mouse anti β-Actin antibody (Millipore) the mouse Aire-pAb. IRDye 800CW-conjugated Goat-anti-mouse IgG (H+L) (LI-COR) was used as second antibody. For western blotting, cells were lysed in 1% SDS-sample buffer and clarified by centrifugation. Protein concentration of cell lysate was determined by DC Protein Assay (BIO-RAD). The protein bands separated on SDS-PAGE were transferred onto PVDF membranes. Aire protein was detected with anti-Aire-pAb and visualized by ODDYSEY imaging system (LI-COR).

Thymus was dissected from BDF1 mouse at age of 3–5 weeks (Oriental Yeast Co., Ltd.), cut into small (1 mm) pieces, mashed by scraping with two sterile slide-glasses, and suspended in DMEM containing 10% FBS and penicillin/streptomycin. They were passed through pre-separation filter (Miltenyi Biotech) at 4°C, pelleted by centrifugation at 1500 rpm and re-suspended in DMEM. Those fresh thymocytes (bulk) were fractionated into four sub-classes regarding expression pattern of surface markers CD4 and CD8: CD4⁺CD8⁻ thymocytes, CD4⁻CD8⁻ thymocytes, CD4⁻CD8⁺ thymocytes and thymocytes without CD4⁻. Separation was performed by MACS Separator (Miltenyi Biotech) using antibody-linked magnet beads such as rat-anti-mouse-CD4MicroBeads and rat-anti-mouse-CD8/MicroBeads (Miltenyi Biotech).

Bulk thymocytes (BDF1; C57BL/6×DBA/2, 6×10⁶ cells) were co-cultured with Aire⁺ cells (BDF1, 3×10⁴ cells). Fractionated thymocytes (1×10⁵ CD4⁻CD8⁻, 1.2×10⁵ CD4⁺CD8⁻, 1.2×10⁵ CD4⁻CD8⁺ and 6.6×10⁶ thymocytes without CD4⁻) were also co-cultured with Aire⁺ cells. They were incubated for 16 hours in 4-well culture slides (BD Falcon) with DMEM containing 10% FBS and penicillin/streptomycin. Fresh PBL was drained from normal mouse (C57BL/6) and Aire-KO mouse (Aire^−/−, C57BL/6), a generous gift from the late Professor Leena Peltonen (Ramsey et al., 2002). They were purified by 40–70% Percoll density gradients. The purified PBLs (0.5×10⁶ cells) were co-cultured with Aire⁺ cells (3×10⁴ cells) for 16 hours in DMEM containing 10%FBS and penicillin/streptomycin.

After 16 hours or 24 hours cultivations, the free thymocytes were removed by gentle decantation and washed out with cold PBS. The co-cultured cells on 4-well culture slides were fixed with 4% paraformaldehyde in PBS (pH 7.4) for 15 min on ice, and treated with a MEBSTAIN Apoptosis Kit Direct (MBL). The specimens treated with TUNEL method were examined under a confocal fluorescence microscope (ECLIPSE E800, Nikon).

Aire⁺ cells were inoculated into 96-well plates and incubated overnight, then bulk thymocytes were added and co-cultured for 24 hours. As control, the same number of Aire⁺ cells and thymocytes were cultured into separate wells. Viable cells were counted by measuring absorbance at 570 nm (A570) after staining with MTT (Roche).

Aire⁺ cells were grown to semi-confluency or confluency in RepCell culture dish (CellSeed). Aire⁺ cells were detached by gentle pipetting at 20°C. The detached cells were washed once with cold PBS and fixed with 4% paraformaldehyde in PBS (pH 7.4) for 15 min on ice. Then, Aire⁺ cells were stained with fluorescent antibodies specific for each of IS components CD40 (Alexa Fluor 647 anti-mouse CD40, BioLegend), MHC-II (Alexa Fluor 488 anti-mouse I-A/I-E, BioLegend), CD80 (Alexa Fluor 488 anti-mouse CD80, BioLegend), CD86 (Alexa Fluor 647 anti-mouse CD86, BioLegend), and control IgG. After washing with cold PBS, surface fluorescence of the stained Aire⁺ cells (5,000 cells each) was measured by a flow cytometry system Guava (Millipore).

In our initial characterization study (Yamaguchi et al., 2011), expression of Aire mRNA was clearly detected in all three Aire⁺ cell lines (Aire⁺TEC1, TEC2 and DC) but endogenous Aire protein could not be detected, perhaps due to poor quality of polyclonal antibodies used. In the present study, we succeeded to produce a new polyclonal anti-Aire antibody (pAb) by immunizing mice with a mixture of synthetic peptides, which correspond to two peptide fragments (126–140 aa and 541–552 aa) of mouse Aire protein. The resulting pAb clearly detected endogenous mouse Aire proteins of Mr = 62 kDa on Western blots in all three Aire⁺ cell lines (Aire⁺TEC1, Aire⁺TEC2 and Aire⁺DC). Also, FLAG-tagged Aire protein was detected (see arrowed three bands and an asterisked band in Fig. 1A, leftside panel). Pre-immune serum or antibodies pre-absorbed with synthetic peptides detected no protein band of Mr = 62 kDa (Fig. 1A, middle and right-side panel). Thus, the newly made anti-Aire pAb had proper immune specificity. Moreover, the protein band of Mr = 62 kDa was not detected in mouse A9 skin fibroblasts. Thus, endogenous Aire protein was detected in all three Aire⁺ cell lines.

Previously, we demonstrated that Aire⁺ cells adhere to fresh thymocytes and kill them by inducing apoptosis (Yamaguchi et al., 2011). Here, we examined how efficiently those Aire⁺ cells kill thymocytes using our in vitro co-culture system. For this, dissected mouse (strain BDF1) thymus was smashed and dissociated thymocytes were filtered through membrane. They were then added into culture plates of Aire⁺ cells or maintained as separate cultures. After incubation for 24 hrs, culture plates were carefully examined under a phase-contrast microscope (Fig. 1B, without the removal of free thymocytes). In the separate culture as control, Aire⁺ cells and thymocytes (bulk) were maintained without significant loss of cell number. However, in the co-culture system (see Aire⁺TEC1 + thymocytes in Fig. 1B), thymocytes (bulk) were substantially reduced while Aire⁺ cells remained unchanged (note cell density difference between separate culture and co-cultures in Fig. 1B).

Direct estimation of “killed thymocytes” is impossible because they are destroyed by apoptosis within 24 hrs. So, we estimated surviving viable cells by measuring absorbance at 570 nm (A570) after MTT-color reaction. Only live cells can change MTT to insoluble formazan and generate blue color at 570 nm. Culture medium alone never affected MTT-color reaction. For all co-cultures, total absorbance at A₅₇₀ (cell viability) was significantly reduced (20∼25%) whereas absorbance of separate cultures remained unchanged (Fig. 1C), indicating significant cell loss in the co-cultures. Under closer microscopic observation, almost equal numbers of live Aire⁺ cells were counted in a small field for both separate culture and co-culture (59 cells vs 61 cells), so that reduction of absorbance at A₅₇₀ (20∼25%) was correlated to killed thymocytes (Fig. 1C). Since absorbance of thymocytes in separate culture was 0.574 (2.58×10⁶ cells), the absorbance difference from Aire⁺TEC1 (A₅₇₀ = 0.300; 2.76×10⁴ cells), TEC2 (A₅₇₀ = 0.297; 2.88×10⁴ cells) and DC (A₅₇₀ = 0.138; 9.60×10⁴ cells) was calculated as 0.274 (1.23×10⁶ thymocytes), 0.277 (1.25×10⁶ thymocytes) and 0.436 (1.96×10⁶ thymocytes), respectively. Then, live cell numbers were calculated as indicated in Fig. 1C legend. Considering both absorbance reduction and live cell number, we calculated that a single Aire⁺ cell killed 20∼45 thymocytes: 44.5 thymocytes for Aire⁺TEC1, 43.2 thymocytes for Aire⁺TEC2 and 20.4 thymocytes for Aire⁺DC. On average, a single Aire⁺ cell appeared able to kill ∼30 thymocytes.

So far, we demonstrated in vitro adherence of fresh thymocytes to Aire⁺ cells and induction of apoptosis. More careful microscopic examination on co-culture revealed that after 16 hrs incubation some thymocytes are still attached on Aire⁺ cell surface (see Aire⁺TEC1 cells in Fig. 1D, with the removal of free thymocytes). TUNEL assay stained these thymocytes green with FITC (green), suggesting they may be in the process of apoptosis (going to die) (see three white arrows in FITC panel of Fig. 1D, 16 h). Most of them clumped on Aire⁺ cells (see three white arrows in PI panel of Fig. 1D, 16 h, also see Merged yellow image). They still retained nuclei (stained with PI: red) and most of them formed clumps on Aire⁺ cell surface (see three white arrows in PI panel of Fig. 1D, 16 h, also see Merged yellow image). On the contrary, Aire⁺TEC1 cells were not stained with TUNEL assay (surrounded by four dashed lines). Thus, FITC-stained (apoptotic) thymocytes attached on Aire⁺ cell surface are still intact at 16 hrs.

After longer cultivation (24 hrs), many Aire⁺TEC1 cells seemed stained with FITC (green) (note Aire⁺TEC1 cells that are seen under thymocytes marked with three open arrowheads in FITC panel of Fig. 1D, 24 h). We assume that fluorescence may have come from the cell debris of destroyed thymocytes (see also Merged yellow–green image). Thus, apoptosis-induced thymocytes were broken after 24 hrs. However, a large clump of thymocytes still adhering to Aire⁺TEC1 cell (marked with a closed white arrowhead in Fig. 1D, 24 h) was not stained with FITC, indicating that these thymocytes are not susceptible to Aire⁺ cell-induced apoptosis. Large clumps of non FITC-stained thymocytes are often observed in co-culture. As control, mouse skin fibroblast A9 and embryonic fibroblast NIH3T3 showed no adherence to thymocytes (Fig. 1D, Control).

In short, apoptotic process of thymocytes is induced within 16 hrs and auto-reactive thymocytes are destroyed during 16–24 hrs. Moreover, some thymocytes can escape from Aire⁺ cell-induced apoptosis (elimination).

To examine if Aire⁺ cells interact with peripheral blood lymphocytes (PBLs), we performed co-culture experiments using Aire⁺ cells and fresh PBLs prepared from either Aire-KO (Aire^−/−) mouse or normal (Aire^+/+) mouse. These freshly drained PBLs were partially purified by Percoll density gradients and added into Aire⁺ cell cultures. After 24 hrs incubation, floating PBLs were carefully removed by gentle decantation (Fig. 2). When PBLs drained from Aire-KO (Aire^−/−) mouse were tested, abundant PBLs adhered to Aire⁺ cells (Fig. 2A, with the removal of free PBLs). Notably, Aire⁺DC line trapped PBLs more efficiently than two other Aire⁺ cell lines. TUNEL assay stained those PBLs green, indicating that they are in the process of apoptosis (see PBLs marked with magenta arrows in Fig. 2A). In the case of PBLs drained from normal (Aire^+/+) mouse, only a few PBLs adhered to Aire⁺ cells (Fig. 2B, with the removal of free PBLs). Thus, those PBLs adhered to Aire⁺ cells are considered to be “auto-reactive PBLs”. Aire-KO (Aire^−/−) mouse had substantially more “auto-reactive PBLs” than normal (Aire^+/+) mouse since thymus of Aire-KO (Aire^−/−) mouse lost ability of negative selection.

We then examined if cell growth states of Aire⁺ cell affects expression levels of Aire gene. To test this, we prepared Aire⁺ cells (Aire⁺TEC1, Aire⁺TEC2 and Aire⁺DC) at two different growth states: rapidly growing state at semi-confluency (SC) and resting state at confluency (C) (Fig. 3). Aire⁺ cells at SC exhibited characteristic shapes of thymic epithelial cells or dendritic cells, whereas all Aire⁺ cells at C formed monolayers of tightly packed cells with cell–cell contact (Fig. 3A).

Unexpectedly, in our preliminary experiments, quantitative Reverse Transcription-PCR (qRT-PCR) analysis showed some fluctuation in expression of housekeeping genes (Gapdh, Hprt1, G6pd2, TATA box-binding protein, γ-Actin and polyubiquitin-C) depending on cell growth states, but expression of β-Actin was consistent (data not shown). Hence, β-Actin was used as a reference gene to compare with Aire gene expression, and each Aire mRNA content was normalized to β-Actin mRNA. Then, the normalized Aire-mRNA expression was compared between two different growth states (SC vs C). It was found that Aire mRNA at C is always higher than SC (Fig. 3B). The increase was significant (60×∼150×) when compared per cell basis. Based on these findings, we postulated that increase of Aire-mRNA expression occurs in accordance with cell density (cell growth states) and it may mimic in vivo developmental process of mTEC (Gray et al., 2007).

Previously we demonstrated that all three Aire⁺ cell lines exhibit ectopic (promiscuous) expression of TSAs (Yamaguchi et al., 2011). It was also reported that Aire-KO (Aire^−/−) mice exhibit reduced expression of TSAs (Anderson et al., 2002).

These findings suggest critical roles of Aire genes in TSA expression. Here, we further examined possible effects of cell growth stage on ectopic expression of five representative TSAs, including Casein alpha (Csna) specific for thyroid, Cytochrome P450 17 (Cyp17) specific for gonad, Fatty acid binding protein 2 (Fabp2) specific for intestine, Insulin II (InsII) specific for pancreas and Thyroglobulin (Tgn) specific for thyroid.

qRT-PCR analysis revealed that these organ-specific TSA genes (Csna, Cyp17, Fabp2, InsII and Tgn) are expressed in all three Aire⁺ cell lines, although expression levels somewhat varied (Fig. 3C). Surprisingly, TSA-mRNA increased (2×∼730×) with increasing cell density (SC vs C) (Fig. 3C). Remarkable increase of TSA-gene expression has been seen during developmental stages of Aire-expressing mTECs in mouse thymus (Gray et al., 2007; Gillard and Farr, 2005; Gäbler et al., 2007), hence we postulated that our in vitro Aire⁺ cell system may mimic in vivo differentiation and maturation process of mTECs and mDCs.

Within antigen-presenting cells (APCs), TSAs are split into small self-antigen peptides through intracellular organelle “immunoproteasome” (IPSM). IPSM is usually re-constructed from constitutive proteasome (PSM) by replacing some subunits with immune types. IPSM plays a crucial role in negative selection of T cells in thymus.

Previously, we observed that all three Aire⁺ cell lines possess IPSMs (Yamaguchi et al., 2011). Here, we measured mRNAs for subunits of IPSM by qRT-PCR analysis and examined if these mRNA contents change at higher cell density (SC vs C) (Fig. 3D). Expression of structural subunits (β1, β2 and β5) of constitutive PSM was detected at significant levels but their contents were not much changed with cell growth states (C/SC: 1.0∼2.6) (Fig. 3D, top panel). However, expression of four subunits of IPSM (β1i and β5i for catalytic subunits of 20S, PA28α and PA28β for regulatory subunits) increased significantly toward C (C/SC: 4.1∼54.0) (Fig. 3D, middle and bottom panels). Thus, more PSMs seem converted to IPSMs toward C. These results again confirmed that Aire⁺ cell lines exhibit characteristic natures of self-APCs, and further suggested that enhanced production of IPSM may be occurring in association with increasing expression of TSA genes.

Promiscuously expressed TSAs are processed to self-antigen peptides within self antigen-presenting cells (self-APCs). Then, these self-antigens are presented to immature thymocytes through MHC molecules with the aid of a bridge “immunological synapse” (IS). IS is composed of CD80, CD86, ICAM1 and so on. And, increased expression of these molecules is considered as indication of the developmental stage of Aire-expressing mTEC. We examined expression of four major components of IS. For this assay, we stained Aire⁺ cells with fluorescently labeled antibodies against CD40, MHC-II, CD80 and CD86. Then, cell surface fluorescence was measured using a flow cytometry system Guava (Millipore). Fluorescence generated by normal control IgGs (Rat IgG for CD40, CD86 and MHC-II and Hamster IgG for CD80) was subtracted. The fluorescence intensity values of Fig. 3E were shown as the subtraction of the measured X-Arithmetic Mean. As shown in Fig. 3E, all three Aire⁺ cell lines expressed IS components. Notably, their expression levels always increased from SC to C. Moreover, Fig. 3F shows surface distribution of IS components, in which each peak shifts to higher intensity (more expression) at C as compared to SC.

It has been reported that single-positive (SP) CD4⁺ thymocytes play critical roles in the development of mTECs in thymus (Irla et al., 2008; Irla et al., 2010). Based on our findings (elevated expression of TSAs and other relevant cellular parameter), we assumed that Aire⁺ cell lines might possess some features of differentiating mTECs. If so, they might still be able to interact with some specialized thymocytes in vitro. Such a “thymic crosstalk” is a reciprocal intercellular communication that controls development of both thymocytes and thymic stromal compartment.

To examine this possibility in vitro, we first prepared four sub-populations of thymocytes: CD4⁺ SP thymocytes (CD4⁺CD8⁻), CD8⁺ SP thymocytes (CD4⁻CD8⁺), CD4⁻CD8⁻ DN thymocytes, and the thymocytes without CD4⁻. Fractionation was done by separating fresh thymocytes using antibody-linked magnetic beads. Then, we performed co-culture experiments (Fig. 4).

To our surprise, when three lines of Aire⁺ cells were co-cultured with CD4⁺ SP (CD4⁺CD8⁻) thymocytes, almost all Aire⁺ cells were killed (Fig. 4A, upper panel). At closer look, shape of floating Aire⁺ cells changed to spherical (Fig. 4A, lower panel, shaded arrowheads) and those were easily removed by gentle PBS-washing. On the contrary, when Aire⁺ cells were co-cultured with CD4⁻ thymocytes (CD4⁻CD8⁻ DN and CD4⁻CD8⁺), all Aire⁺ cells remained alive (Fig. 4B,C, closed arrowheads). Surprisingly, when Aire⁺ cells were co-cultured with thymocytes without CD4⁻ thymocytes, Aire⁺ cells were killed again (Fig. 4D, shaded arrowheads). Thus, in vitro killing of Aire⁺ cells by CD4⁺ thymocytes (CD4⁺CD8⁻ and thymocytes without CD4⁻) was demonstrated. Killing of Aire⁺ cells were not obviously seen when we used bulk thymocytes that contain all CD4⁺/CD4⁻ sub-types (Fig. 4E, closed arrowheads). This may be accounted for by the notion that killing or rescuing Aire⁺ cells are minute events and can be seen only when co-cultured with condensed sub-populations of thymocytes.

Aire-expressing cells in vivo are derived from mTECs and mDCs that are at intermediate stages of differentiation and maturation. Hence, our in vitro observation implied possible existence of two sub-types of thymocytes that may determine the fate (dead or alive) of Aire-expressing mTECs. One type is “CD4⁺ killer” thymocytes, which we assume equivalent to CD40L/RANKL CD4⁺ thymocytes. The other type is a novel “CD4⁻ rescuer” thymocytes that may prevent Aire-expressing mTECs/mDCs from the elimination (Fig. 4F).

It has been reported that tumor necrosis factor-receptor super families (TNFRSFs) with “death domain” are involved in the induction of cell death (Hehlgans and Pfeffer, 2005). Previously, we observed the expression of some TNF family members in all three Aire⁺ cell lines (Yamaguchi et al., 2011). Here, we examined if Aire⁺ cell lines express their receptors (TNFRSFs).

qRT-PCR analysis revealed that a number of TNFRSF members are expressed in all three Aire⁺ cell lines (Fig. 5). Most of 26 members tested were expressed at similar levels. However, expression of five members (Tnfrsf-8, -11a, -11b, -17, -21) was different between two cell types (Aire⁺TECs and Aire⁺DC). On the other hand, no difference was seen for other five members (Tnfrsf-1a, -6, -10b, -12a, and -16).

Among 26 members examined, eight members (Tnfrsf-1a, -6, -10b, -11b, -12a, -16, -21, and -26) are known to possess “death domain” (Hehlgans and Pfeffer, 2005; Tartaglia et al., 1993; Itoh and Nagata, 1993; Pan et al., 1997; Yamaguchi et al., 1998; Chicheportiche et al., 1997; Chao, 1994; Pan et al., 1998; Simonet et al., 1997). Considering the apoptotic ability of death domain-containing TNFRSFs, we assumed that five TNFRSF members (Thfrsf-1a, -6, -10b, -12a and -16) may be involved in killing Aire⁺ cells. Thus, these in vitro results seem to support the TNF/TNFRSF-mediated self-killing of Aire⁺ mTECs.

AIRE gene has been cloned from human chromosome 21 as a pathogenic gene responsible for autoimmune disease APECED/APS-1 that exhibits autosomal recessive inheritance (Nagamine et al., 1997). In human and mouse, it was shown that AIRE gene expression is restricted to only the immune system such as thymus and lymph node, and that Aire protein acts as a unique transcription factor (Nagamine et al., 1997; Pitkänen et al., 2000; Pitkänen et al., 2001). Furthermore, several studies using Aire-deficient mutant mice provided an important evidence that transcription factor Aire plays a critical role in regulating the ectopic (promiscuous) expression of tissue-specific antigens (TSAs) in the mTECs of thymus (Anderson et al., 2002; Ramsey et al., 2002; Liston et al., 2003).

Recently, we established three lines of Aire-expressing cells (Aire⁺TEC1/2 and Aire⁺DC) from abnormally enlarged thymus of the transgenic mice which were produced by injecting a transgene consisting of SV40 T-antigen (SV40-Tag) gene driven by mouse Aire gene promoter into fertilized eggs (Yamaguchi et al., 2011). SV40-Tag protein binds to tumor suppressors p53 and retinoblastoma family (pRb, p107 and p130) that are involved in regulation of cell cycle. Interaction of these proteins with SV40-Tag facilitates cell proliferation and survival, hence by transfecting with SV40-Tag gene a number of transformed cell lines have been established (Gannon and Lane, 1987; Bártek et al., 1992; Ishida et al., 1995; Yanai et al., 1991).

In our transgenic mice, SV40-Tag gene was driven by Aire-gene promoter, hence its expression was allowed only in the cells like mTECs expressing Aire. In all three established Aire⁺ cell lines, endogenous Aire protein was undoubtedly detected (Fig. 1A). To date, a large quantity of Aire-expressing mTECs have never been isolated from mouse thymus, hence newly isolated three Aire⁺ cell lines are only available ones and will be extremely useful for in vitro studying biological aspects of AIRE protein functions at molecular levels.

All three Aire⁺ cells adhered to fresh thymocytes in vitro and induced apoptosis, potentially reproducing in vivo T-cell negative selection occurring in the thymus medulla. Moreover, Aire⁺ cells exhibited characteristic features of “self-APCs”, such as abilities to express TSAs (tissue specific antigens), IPSM (immunoproteasome), IS (immunological synapse) and TNFs/TNFRSFs (tumor necrosis factors/TNF receptors). Thus, Aire⁺ cell lines were considered as “self-APCs” acting in vitro.

Using our in vitro co-culture system, it was calculated that a single Aire⁺ cell appeared able to kill 20–45 (∼30) thymocytes within 24 hrs (Fig. 1B,C). Interestingly, two Aire⁺TEC lines killed thymocytes more efficiently than Aire⁺DC line, probably reflecting expression patterns of TSAs (self-antigens) in those Aire⁺ cells.

A major event occurring in thymus medulla is elimination of auto-reactive thymocytes (T cells). This is critical to secure self-tolerance in immune system. In thymus, auto-reactive thymocytes are apoptosed via “negative selection”. As documented above, Aire⁺ cells induce apoptosis on adhered thymocytes (Yamaguchi et al., 2011). In the co-culture system, TUNEL assay indicated that apoptosis occurs after 16 hrs and DNA fragmentation takes place within 24 hrs (Fig. 1D). Non-thymic cells (mouse fibroblasts NIH3T3 and A9) did not adhere to any thymocytes. Interestingly, some thymocytes were not stained with TUNEL assay (not killed by Aire⁺ cells), suggesting that these minute thymocytes may be “non auto-reactive thymocytes” that will be released into peripheral blood.

In patients with autoimmune disease, it was shown that auto-reactive T cells escape from negative selection in thymus and are drained into peripheral blood (Perniola et al., 2000). Those escaped T cells injure several tissues/organs and disrupt immune tolerance. Such aberrant T cells are found also in peripheral blood lymphocytes (PBLs) of health body, though a minute fraction (1–5%). This peripheral tolerance is mediated through regulatory T cells.

In our in vitro co-culture system, Aire⁺ cells trapped some of fresh PBLs from normal (Aire^+/+) mouse, and many more from Aire-deficient mutant (Aire^−/−) mouse (Fig. 2A,B). These Aire^+/+ and Aire^−/− PBLs trapped by Aire⁺ cells were subject to apoptosis. Thus, all Aire⁺ cell lines are capable of trapping “auto-reactive PBLs” that are rarely present in PBLs of normal (Aire^+/+) mouse and abundant in PBLs of Aire-KO (Aire^−/−) mouse. These findings suggest a possible use of Aire⁺ cells to absorb/trap auto-reactive PBLs in APECED mouse model. Similarly, once human type Aire⁺ cells were developed, they could be used as absorber/trapper to treat patients with autoimmune disease such as APECED/APS1.

Terminal differentiation model of mTECs (Gray et al., 2007; Gillard and Farr, 2005; Irla et al., 2010) postulates that mTECs gradually increase expression of several differentiation-related genes during maturation process in coordination with “thymic crosstalk”. These genes/products include a unique transcription factor “Aire”, a variety of TSAs and essential components of immunological synapse (IS) (MHC-II, CD80, CD86, CD40).

In our in vitro Aire⁺ cell culture system, it was shown that expression of Aire gene and organ-specific TSA genes (Csna, Cyp17, Fabp2, InsII and Tgn) increases as cell density increases (Fig. 3A–C). Furthermore, formation of IPMS- and IS-components increased toward confluency (Fig. 3D–F).

Proteasome (PMS) is a protein complex composed of multiple subunits such as a large subunit (20S-particle) with proteolytic activity and three other subunits (19S-particle, 11S (PA28)-particle and PA200) with regulatory roles (Whitby et al., 2000). In self-APC cells, several of these PMS-subunits are replaced with immune-specific counterparts and reconstructed immune type PMSs (IPSMs) act as a processing machinery by which a variety of TSAs (self-proteins) are digested to small peptides (auto-antigens) (Kloetzel and Ossendorp, 2004; Groettrup et al., 2010). Then, these auto-antigens are bound to MHC molecules expressed on the surface of self-APCs. Then, multiple auto-antigens are presented to a specific type of thymocytes.

In our in vitro system, all three Aire⁺ cell lines remarkably increased expression of IPSM-specific subunits, such as structural subunits (β1i, β2i and β5i) by ∼50× and regulatory subunits (PA28α, PA28β and PA28γ) by ∼6× toward reaching confluent state (Fig. 3D). Thus, increased IPSMs must be contributing to processing of TSAs to auto-antigen peptides within Aire⁺ cells.

Immunological synapse (IS) is required for mediating the binding of self-APCs to T cells (thymocytes). In our system, Aire⁺ cells expressed four IS components (MHC-II, CD80, CD86, CD40) and their expression levels increased at higher cell density. Thus, increased IS expression on Aire⁺ cells must be enforcing the binding of self-APCs to T cells (thymocytes) (Fig. 3E,F).

These results suggest that remolding of PSM to IPSM and formation of IS must be taking place in the proliferating Aire⁺ cell populations, although molecular studies are required.

Previously, we observed that several kinds of TNFSFs are expressed in all three Aire⁺ cell lines (Yamaguchi et al., 2011). In the present study, we further observed expression of their receptors (TNFRSFs). Interestingly, expression of five TNFRSFs (Tnfrsf-8, -11a, -11b, -17 and -21) was low and shows different patterns between two types of Aire⁺ cell lines (Aire⁺TEC and Aire⁺DC) (Fig. 5). However, six other TNFRSFs (Tnfrsf-1a, -6, -10b, -12a, -16 and -26), that are known to possess death domain (Hehlgans and Pfeffer, 2005; Wiens and Glenney, 2011), showed higher expression levels. Therefore, we presumed that the later six TNFRSF members may play more crucial roles as apoptosis-inducing factors to kill Aire-expressing mTECs. Thus, Aire⁺ cell lines that express TNFSFs/TNFRSFs will serve as useful tools for TNF-associated thymic apoptosis mechanism.

Thymocytes (or T cells) are composed of a variety of sub-types with respect to cell surface markers (CD4^+/−, CD8^+/− and so on). SP-thymocytes are shown to be involved in facilitating maturation process of mTECs such as differentiation, proliferation and survival (Irla et al., 2008; Irla et al., 2010; Akiyama et al., 2008; Hikosaka et al., 2008; Villaseñor et al., 2008; Rossi et al., 2007), but little is known about their regulatory role in governing mature-mTEC development. DP-thymocytes are positively selected by cTECs, and through thymic transport mechanism they are sent to cortico-medullary junction, where DP-thymocytes encounter negative selection conducted by self-APCs.

Our in vitro co-culture system demonstrated that CD4⁺ thymocytes kill Aire⁺ cells (Fig. 4A,D), whereas CD4⁻ thymocytes including both CD4⁻CD8⁻ and CD4⁻CD8⁺ are unable to kill Aire⁺ cells (Fig. 4B,C). The sub-population of CD4⁺ thymocytes still contains CD40L⁺/RANKL⁺ thymocytes that are known to apparently determine the fate of Aire-expressing mTECs, and the sub-population of CD4⁻ thymocytes are still heterogeneous as to expression of CD antigens. However, our results implied that some of CD4⁺ and CD4⁻ thymocytes might have distinct ability to determine the fate (dead or alive) of Aire⁺ cells, in a similar way as in vivo mTECs (Irla et al., 2008; Hikosaka et al., 2008). These include two types of thymocytes; “killer” CD4⁺ thymocytes and “rescuer” CD4⁻ thymocytes (Fig. 4F).

“Killer” CD4⁺ thymocytes killed certain Aire⁺ cells, whereas “rescuer” CD4⁻ thymocytes helped certain Aire⁺ cells from the death. Killing of Aire⁺ cells were not obviously seen when we used bulk thymocytes that contain all CD4⁺/CD4⁻ sub-types (Fig. 4E). This may be accounted for by the notion that killing or rescuing Aire⁺ cells are minute events and can be seen only when condensed sub-populations of thymocytes were used for co-culture.

In short, all three Aire⁺ cell lines increased expression of a number of relevant genes at higher cell density, especially at “confluent” state (Fig. 3B,C,E). These implied that Aire⁺ cell lines progress differentiation stages (from intermediate state to mature state) like mTECs in thymus (supplementary material Fig. S1). Thus, our in vitro co-culture system appears to mimic a part of “in vivo thymic crosstalk” and would be useful for further studying at molecular basis.

The authors thank Toagosei Company Limited, Japan, for their generous support. We thank the late Professor Leena Peltonen for the generous gift of an Aire-deficient mouse.

This work was initiated by the funds including a Fund for “Research for the Future” Program from the Japan Society for the Promotion of Science (JSPS) to N.S., Grant-in-Aid for Young Scientists B from the Ministry of Education, Culture, Sports, Science and Technology (MEXT) to Y.Y., the 21st Century COE Program and Global COE Program at Keio University to J.K. and the current work was supported by a “Special Fund” from Toagosei Company Limited, Japan.