Expression of autoimmune regulator gene (AIRE) and T regulatory cells in human thymomas

Abstract

Expression of the autoimmune regulator gene (AIRE) and the presence of CD25⁺/forkhead box p3 (FoxP3)⁺ T regulatory (T_reg) cells were investigated in histologically normal adult thymi and in thymomas using immunohistochemistry and quantitative real-time polymerase chain reaction (PCR). In the normal thymus staining for AIRE was detected in the nucleus of some epithelial-like cells located in the medulla; in thymomas AIRE-positive cells were extremely rare and could be detected only in the areas of medullary differentiation of two B1 type, organoid thymomas. RNA was extracted from 36 cases of thymoma and 21 non-neoplastic thymi obtained from 11 myasthenic (MG⁺) and 10 non-myasthenic (MG^–) patients. It was found that AIRE is 8.5-fold more expressed in non-neoplastic thymi than in thymomas (P = 0.01), and that the amount of AIRE transcripts present in the thymoma tissue are not influenced by the association with MG, nor by the histological type. A possible involvement of AIRE in the development of MG was suggested by the observation that medullary thymic epithelial cells isolated from AIRE-deficient mice contain low levels of RNA transcripts for CHRNA 1, a gene coding for acetylcholine receptor. Expression of human CHRNA 1 RNA was investigated in 34 human thymomas obtained from 20 MG^– patients and 14 MG⁺ patients. No significant difference was found in the two groups (thymoma MG⁺, CHRNA1 = 0.013 ± 0.03; thymoma MG-, CHRNA1 = 0.01 ± 0.03). In normal and hyperplastic thymi CD25⁺/Foxp3⁺ cells were located mainly in the medulla, and their number was not influenced by the presence of MG. Foxp3⁺ and CD25⁺ cells were significantly less numerous in thymomas. A quantitative estimate of T_reg cells revealed that the levels of Foxp3 RNA detected in non-neoplastic thymi were significantly higher (P = 0.02) than those observed in 31 cases of thymomas. Our findings indicate that the tissue microenvironment of thymomas is defective in the expression of relevant functions that exert a crucial role in the negative selection of autoreactive lymphocytes.

Introduction

Thymomas are uncommon tumours derived from the thymic epithelium [1–3]. The World Health Organization (WHO) classification recognizes two major types of thymoma depending on whether the neoplastic epithelial cells and their nuclei have a spindle or oval shape (type A) or whether the cells have a predominantly round or polygonal appearance (type B) [3]. Tumours combining both these features are classified as AB. Type B thymomas are subdivided further on the basis of the extent of the lymphocytic infiltrate and the degree of atypia of the neoplastic cells in B1, B2 and B3. This classification recalls a previous histogenetic classification [4], in which it was proposed that spindle (type A) tumour cells are related to epithelial medullary cells, and that round-type B tumour cells are of cortical origin. Thymomas are well-differentiated tumours. Some of them have an organoid appearance and retain the functions of the normal thymus. In fact, tumour cells and accessory cells are still able to attract T cell precursors into the tumour, and to induce their intratumoral maturation. Organoid thymomas (B1) derive from cortical epithelial cells, have a high content of thymocytes and have a lobular architecture highly reminiscent of that of a normal thymus; in most cases small, rudimentary medullary areas containing a few Hassall's bodies are present.

Thymomas are often associated with autoimmune diseases [1–3]. About 30–45% of patients with thymoma suffer from myasthenia gravis (MG). MG symptoms are caused by autoantibodies against the muscular nicotinic acetylcholine receptor (AChR), which are not produced inside thymomas but in lymph nodes, spleen and bone marrow. It has been proposed that thymoma-derived T cells that circulate to these organs contribute to the pathogenesis of MG. Thus, a defect of thymocyte maturation inside the tumour might facilitate the development of autoreactive T cell clones.

The autoimmune regulatory gene (AIRE) is the gene responsible for APECED (autoimmune polyendocrinopathy–candidiasis–ectodermal dystrophy) or APS1 (autoimmune polyendocrinopathy syndrome type 1) [5,6], a rare autosomal monogenic recessive disorder, characterized by organ-specific autoimmunity, which most commonly affects the parathyroid and the adrenal glands. The AIRE gene is located in chromosome 21q22.3 and encodes a 58-kDa protein involved in nuclear transcription. AIRE is expressed in the thymic medulla, and regulates the negative selection of self-reactive T cells [7]. This result is achieved through the activation of nuclear transcription of ectopic tissue-specific autologous antigens in medullary epithelial cells and in their presentation to medullary thymocytes; peptide recognition by self-reactive thymocytes causes their negative selection. Thus, defective expression of AIRE might play a crucial role in the pathogenesis of autoimmune diseases by favouring the development of self-reactive T cell clones within the thymus.

T regulatory cells (T_reg) are a subset of CD25⁺ CD4⁺ T cells which have inhibitory effects on antigen-specific activation of naive autologous T cells [8,9], and are key controllers of self-reactive cells [10]. The development and function of CD25⁺ T_reg cells requires activation of the forkhead/winged helix transcription factor forkhead box P3 (FoxP3) gene [11], which is currently considered to be a reliable marker for this cell subset. Abrogation of the migration of these cells from the thymus to the periphery from the beginning of their ontogeny results in the development of organ-specific autoimmune diseases [12,13]. CD4⁺ CD25⁺ T_reg cells with suppressive activity have been isolated from the human thymus [14]. It has been suggested that induction of T_reg cells is caused by organ-specific antigens expressed in the thymic medulla, and hence it has been proposed that AIRE expression may influence their development [7]. This view is supported by the recent observation that APECED patients with the AIRE defect have decreased numbers of functionally impaired T_reg cells in the peripheral blood [15].

In the present study, we have investigated the expression of AIRE and the presence of T_reg cells in 36 thymomas and 21 non-neoplastic thymuses obtained from MG⁺ and MG^– patients. Our findings indicate that AIRE is expressed poorly in thymoma epithelial cells, and that thymomas contain reduced numbers of T_reg cells. Taken together, these observations indicate that the tumoral tissue of thymoma represents a poorly organized microenvironment in which the development of self-reactive thymocytes is probably facilitated.

Materials and methods

Patients

The study was conducted according to the informed consensus law of Italy. Patients were operated at the Ospedale Sant'Andrea-Università di Roma ‘La Sapienza’, at the Ospedale ‘Le Molinette’-Università di Torino and at the Istituto Tumori ‘Regina Elena’ of Rome. Thirty-six thymomas (mean age = 58 ± 11, 15M/21F) were removed surgically from 13 myasthenic (mean age = 61 ± 12, 6M/7F) and 23 non-myasthenic patients (mean age = 56 ± 11, 9M/14F), and were classified according to WHO (2004) [3]. Twenty-one non-neoplastic thymi were obtained from 11 myasthenic (MG⁺) (mean age = 33 ± 15, 1M/10F) and 10 non-myasthenic (MG^–) adult patients (mean age = 46 ± 12, 4M/7F). The thymus was removed from the MG⁺ patients as part of the therapeutic strategy; it was the site of follicular hyperplasia in six cases, whereas in the remaining five cases it was histologically normal with evidence of adipose involution. In MG^– patients the fragments of normal thymus were isolated from biopsy material taken for tumour staging (two malignant lymphoma, four lung carcinoma, one papillary carcinoma of the thyroid) or were found adjacent to the thymoma (four patients); in all cases the thymus tissue was histologically normal with evidence of adipose involution. The surgical specimens were formalin-fixed and paraffin-embedded for conventional histology; additional tissue fragments were snap-frozen in liquid nitrogen and stored at −80°C until sectioning.

Immunohistochemistry

Expression of AIRE protein, Foxp-3 protein and CD25 were investigated in frozen sections or in paraffin sections of three normal thymuses, four hyperplastic thymuses and 13 thymomas using immunohistochemistry. The rat monoclonal antibody (mAb) anti-AIRE (clone B1/02-5c11–4) was produced recently at the monoclonal antibody facility at the Walter and Eliza Hall Institute of Medical Research by standard polyethylene glycol fusion. Rat anti-AIRE mAb (B1/02-5C11-4) was generated to a 21 amino acid peptide corresponding to the 20 C-terminal. Most amino acids of mouse AIRE with a C residue at the N-terminus for keyhole limpet hemocyanin (KLH) conjugation [mAire-C- CLQWAIQSMSRPLAETPPFSS (free acid)] were purchased from AusPep (Parkville, Australia). The KLH conjugated mAire-C was used for immunization of two rats, while unconjugated peptide was bound to plastic plates for enzyme-linked immunosorbent assays (ELISAs). Selected unpurified monoclonals with high ELISA titres were screened further by analysing interaction with mouse AIRE and human AIRE on Western blots and immunohistochemistry. The rat anti-AIRE mAb (B1/02-5C11-4) used in this study (working dilution 1:10) is the IgG 2c subtype.

Anti-FoxP3 (mAb 236 A/E7, 1:40 dilution; ABCAM, Cambridge, MA, USA), and anti-CD25 (clone 4C9, 1:50, Novocastra Laboratories, Newcastle upon Tyne, UK) were commercially available. Cryostat sections and deparaffinized sections were preincubated with the blocking solution to prevent non-specific binding, and then with an optimal dilution of the primary antibody. The slides were incubated sequentially with anti-mouse or anti-rat biotinylated immunoglobulins followed by streptavidin–peroxidase complex (Dako LSAB Kit-peroxidase; Dakopatts, Copenhagen, Denmark). Each incubation step lasted 10 min with multiple 5-min TRIS-buffered saline (TBS) washes between each step. The sections were incubated with 0.03% H₂O₂ and 0.06% 3,3′-diaminobenzidine (Dakopatts) for 3–5 min.

Real-time ploymerase chain reaction (PCR)

The presence of AIRE, human CHRNA1, the orthologue of mouse CHRNA1 and FoxP3 RNAs were investigated in RNA extracts obtained from 10 µm frozen sections (50 for sample) of non-neoplastic thymus and thymoma tissue using the RNA fast isolation kit. RNA transcripts were measured by real-time quantitative reverse transcription (RT)–PCR, based on TaqMan methodology, using the ICycler System (Biorad, Milan, Italy). To normalize the amount of total RNA present in each reaction, we amplified the housekeeping gene β-actin. Measurements were performed in triplicate. RNA obtained from phytohaemagglutinin (PHA)-stimulated peripheral blood lymphocytes (PBL) from normal thymus (given from a mixture of eight strongly positive normal thymuses) or from striated muscle were used as positive controls, respectively, for FoxP3, AIRE and CHRNA1. The results are expressed as relative levels of FoxP3, AIRE and CHRNA1 mRNAs referred to the corresponding positive control.

Laser capture microdissection (LCM) was used to isolate cortical and medullary areas in two cases of histologically normal thymus and in two cases of organoid thymomas (B1). The cortex and the medulla were microdissected from 6–8 µm frozen sections using the microdissection laser system SL CUT (Nikon Instruments, Florence, Italy). Samples were collected in test tubes and homogenized in RNeasy lysisbuffer; total RNA was prepared using the Arcturus Pico Pure RNA isolation kit (MWG, Florence, Italy). Gene-specific primers and probes were as follows: AIRE, forward AGGCAACAGTCCAGGAGGTG, reverse TAGGGGTTCCCCAGGTGGAC, probe CTCCTCTCCCGCCGACCTAAGCCC; FoxP3: forward TTCTCGGTATAAAAGCAAAGTTGT, reverse GGCATCGGGTCCTTGTCC, probe TGACAGTTTCCCACAAGCCAGGCT; CHRNA1: forward, TTTCCCTTTGATGAACAGAACTGC, reverse GAATAGGTCACGGAGTGCTCC, probe TGGGCACCTGGACCTACGACGGC.

Macroarray analysis

Frozen 10-µm sections (50 for sample) of thymus or thymoma samples were collected in test tubes and homogenized in RNeasy lysis buffer and total RNA was prepared using the RNA fast kit (Molecular System, San Diego, CA USA). The integrity of the RNA was assessed by denaturing agarose gel electrophoresis and spectrophotometry. Expression of chemokines and cytokines was investigated using a macroarray analysis. Macroarray analysis was performed as described in detail previously (available at http://www.superarray.com). Briefly, total RNA (3 µg) is used as the template for biotin-labelled cDNA probe synthesis and hybridized to human inflammatory cytokines and receptor oligonucleotide arrays (containing 112 human genes) (GEArray Q Series KIT; DBA, Milan, Italy). The reactions were revealed by chemiluminescence and a scanner was used to convert the image into a raw image file. The raw data file was extracted from the image file using an image analysis software (Scanalyse version 2.5 by Michael Eisen, DBA). Each array is spotted with a negative control of pUC18, blanks and housekeeping genes [β-actin, glyceraldehyde-3-phosphate dehydrogenase (GAPDH)]. The relative abundance of a particular transcript was estimated by comparing its signal intensity to the signal derived from GAPDH. To make the data comparable between different arrays we have normalized each signal against the signal of a housekeeping gene.

Differences in RNA expression considered as potentially informative were further investigated using real-time PCR as a confirmatory assay.

Medullary thymic epithelial cells (TEC) from AIRE-deficient mice

Medullary thymic stroma was enriched from the thymi of 6–8-week-old C57Bl/6 and AIRE-deficient mice, as described previously [16]. Briefly, eight to 10 thymi were collected into MT-RPMI-1640, having been trimmed of fat and connective tissue. Following brief agitation using a wide-bore glass pipette, thymic lobules were subjected to enzymatic digestion. Thymi were incubated in 5 ml MT-RPMI-1640 with 0.125% (w/v) collagenase D (Roche, Milan, Italy) and 0.1% (w/v) DNase 1 (Roche) at 37°C for 20 min with agitation by pipetting at 5-min intervals. Cells released into suspension were removed once larger thymic fragments had settled. This step was repeated three to four times with fresh media. Trypsin (Roche) replaced collagenase D in the final digest. Each cell isolation was counted using a Coulter counter and the final two or three enrichments were passed through 100-µm mesh and pooled for further steps. Stromal cells were resuspended in fluorescence activated cell sorter (FACS) buffer with 5 mm ethylenediamine tetraacetic acid (EDTA) and enriched further using CD45-microbeads (Miltenyi Biotec, Bergisch Gladbach, Germany) and the auto magnetic affinity cell sorting (MACS) system (Miltenyi Biotec), as per the manufacturer's instructions. Cells were stained with fluorescein isothiocyanate (FITC)-conjugated anti-CD45.2, phycoerythrin (PE)-conjugated anti-I-A/I-E (MHCII), biotin-conjugated anti-Ly51 and CyChrome-conjugated streptavidin, as described by Gray et al. [16]. Murine thymic epithelial cells (mTECs) were sorted by flow cytometry according to the phenotype: CD45^–, MHCII^hi, Ly51^lo.

RNA extraction and amplification from mTEC

RNA extraction and clean-up was performed using the Qiagen RNeasy Micro Kit (Cat. no. 74007; Milan, Italy) according to the manufacturer's instructions. RNA amplification was performed through the Australian Genome Research Facility (Melbourne, Australia). Briefly, 100 ng of total RNA was amplified using T7-oligo-dT and the Megascript T7 kit (Ambion, Milan, Italy) following the Affymetrix manual (701725 rev5). A second round of cDNA synthesis was performed using some or all of the first-round amplified RNA as indicated.

Reverse transcription was performed using SuperScript III RNase H^– Reverse Transcriptase (Invitrogen, Milan, Italy), with either 500 ng oligo-dT₁₈ (Geneworks, Adelaide, Australia) or 250 ng random primers (Promega, Milan, Italy) per reaction according to the manufacturer's instructions.

Quantitative real-time RT–PCR

Quantitative RT–PCR was performed on the Roche LightCycler 480 using the Roche Universal Probe Library system. Amplification of the housekeeping gene, Actb, was used to normalize the amount of total cDNA in each reaction. Reactions were performed in technical triplicate on two biological replicates and included a no-template control. The gene-specific primers were as follows: Actb, forward AAGGCCAACCGTGAAAAGAT, reverse GTGGTACGACCAGAGGCATAC, UPL probe no. 56; CHRNA1, forward ATCGTCATCAACACACACCAC, reverse CATGATGTTTGGGATAGTGTCG, UPL probe no. 107.

Results

Expression of AIRE in the normal thymus and in thymomas

The presence of AIRE-positive cells was investigated in frozen sections of three histologically normal adult thymi and 13 thymomas using immunohistochemistry. In the normal thymus, staining for AIRE was detected in the nucleus of some epithelial-like cells located in the medulla and around Hassall's bodies (Fig. 1a); in thymomas (AB, B1, B2 and B3), AIRE-positive cells were extremely rare and could be detected only in the areas of medullary differentiation of two organoid thymomas; interestingly, even in tumour tissue AIRE-positive cells were associated with the presence of Hassall's bodies (Fig. 1b).

Fig. 1.

(a) Medullary area of a histologically normal thymus immunostained for the autoimmune regulatory gene (AIRE). Three positive cells are indicated by the arrows. The staining is nuclear and may show a typical punctate patter. (b) B1 organoid thymoma immunostained for AIRE. A single positive cell (arrow) is immunostained in a small area of medullary differentiation. A Hassall's body (HB) is present in the field (× 400, ABC-peroxidase counterstained with haematoxylin).

Expression of AIRE was investigated further at RNA level using real-time PCR. Total RNA was extracted from 36 cases of thymoma and 21 non-neoplastic thymi obtained from 11 myasthenic (MG⁺) and 10 non-myasthenic (MG^–) patients (Fig. 2; Table 1). It was found that AIRE is expressed 8.5-fold more in non-neoplastic thymi than in thymomas (P = 0.01), and that the amount of AIRE transcripts present in the thymic tissue is not influenced by the association with MG; in fact, thymomas from MG⁺ and MG^– patients exhibited similar levels of AIRE RNA. A poor expression of AIRE RNA in thymomas could be confirmed in four cases in which the presence of AIRE RNA was determined in the tumour tissue (AIRE = 0) and in fragments of the paired histologically normal thymus adjacent to the tumour (AIRE = 0.38 ± 0.3).

Fig. 2.

Expression of the autoimmune regulatory gene (AIRE) in 21 cases of non-neoplastic thymus [11 myasthenia gravis (MG⁺) and 10 non-MG^–] and in 36 cases of thymoma (13 MG⁺ and 23 MG^–) as demonstrated by real-time polymerase chain reaction. All samples were tested simultaneously. *P = 0.01, Student's t-test (AIRE mRNAs of normal MG^–, normal MG⁺, hyperplasia MG⁺ versus all thymomas).

Table 1.

The autoimmune regulatory gene (AIRE), forkhead box P3 (FoxP3) and CHRNA1 mRNAs in tissue sections obtained from histologically normal thymus, from thymus with follicular hyperplasia and from thymomas.^*

Histology	Myasthenia gravis	AIRE RNA (mean ± s.d.)	FoxP3 RNA (mean ± s.d.)	CHRNA1 RNA (mean ± s.d.)
Normal thymus	No	0.14 ± 0.2 (n = 10)	21.4 ± 20 (n = 6)	0.19 ± 0.6 (n = 10)
Normal thymus	Yes	0.005 ± 0.004 (n = 7)	36.2 ± 31 (n = 4)	0.01 ± 0.02 (n = 5)
Follicular hyperplasia	Yes	0.31 ± 0.4 (n = 4)	20.4 ± 18 (n = 7)	0.01 ± 0.01(n = 3)
All non-neoplastic thymuses		0.17 ± 0.15	25.4 ± 22	0.12 ± 0.4
Thymoma AB	No	0.01 ± 0.03 (n = 7)	14.77 ± 33 (n = 6)	0.003 ± 0.002 (n = 4)
Thymoma B1	No	0 (n = 6)	25 ± 23 (n = 5)	0.002 ± 0.002 (n = 6)
Thymoma B2	No	0.08 ± 0.1 (n = 5)	14.67 ± 27 (n = 4)	0.0001 ± 0.0001 (n = 3)
Thymoma B3	No	0.02 ± 0.04 (n = 5)	2.89 ± 3 (n = 4)	0.03 ± 0.07 (n = 7)
All thymomas without myasthenia		0.02 ± 0.09	14.9 ± 25	0.01 ± 0.03
Thymoma AB	Yes	0 (n = 3)	0.5 ± 0.7 (n = 2)	0.0001 ± 0.0002 (n = 3)
Thymoma B1	Yes	0.07 ± 0.1 (n = 5)	12.4 ± 15 (n = 5)	0.002 ± 0.003 (n = 5)
Thymoma B2	Yes	0.00015 ± 0.0001(n = 4)	8.3 ± 4 (n = 4)	0.03 ± 0.05 (n = 5)
Thymoma B3	Yes	0 (n = 1)	5.5 (n = 1)	0.0001
All thymomas with myasthenia		0.02 ± 0.09	8.5 ± 10	0.013 ± 0.03
All thymomas with and without myasthenia		0.02 ± 0.08**	12.45 ± 20**	0.01 ± 0.03

^*Total RNA was extracted from frozen sections. RNA transcripts for FoxP3, AIRE and CHRNA1 were measured by real-time quantitative reverse transcription–polymerase chain reaction. To normalize the amount of total RNA present in each reaction, we amplified the housekeeping gene β-actin. Measurements were performed in triplicate. The results are expressed as relative levels of FoxP3, AIRE and CHRNA1 mRNAs present in the samples referred to the expression of these genes in the corresponding positive control.

^**P = 0.01, Student's t-test [AIRE mRNAs of normal non-myasthenia gravis (MG^–), normal MG⁺, hyperplasia MG⁺ versus all thymomas]. P = 0.02, Student's t-test (FoxP3 mRNAs of normal MG^–, normal MG⁺, hyperplasia MG⁺ versus all thymomas). P = 0.05, Student's t-test (CHRNA1 mRNAs of normal MG^–, normal MG⁺, hyperplasia MG⁺ versus all thymomas).

The expression of AIRE RNA in the cortical and medullary areas was investigated using LCM of two histologically normal thymuses and of two organoid (B1) thymomas; in tumour tissue, the presence of Hassall's bodies was taken as a morphological evidence of medullary differentiation. The results of these experiments have confirmed that AIRE RNA is expressed in the medullary areas and provide evidence that expression of AIRE RNA in the small areas of medullary differentiation of organoid thymomas is not significantly different from that of the normal adult thymus (Table 2).

Table 2.

Expression of the autoimmune regulatory gene (AIRE) RNA in the cortex and medulla of two histologically normal thymi, and of two organoid (B1) thymomas.^*

Histology	Age/sex	MG	Cortex	Medulla
Normal thymus	64/F	no	0*	0.11
	29/M	no	0	0.25
Thymoma B1	65/F	yes	0	0.10
	29/M	no	0	0.13

^*Cortical and medullary areas were isolated from frozen sections using laser capture microdissection of two histologically normal thymi, and of two B1 thymomas. AIRE mRNA was measured by real-time quantitative reverse transcription–polymerase chain reaction. The results are expressed as relative levels of AIRE mRNAs. MG: myasthenia gravis.

Thymomas are classified histologically as medullary (A), cortical (B1, B2 and B3) or mixed cortico–medullary (AB) [3,4]. When our cases were grouped according to the histological type it was found that 10 AB thymomas, all characterized by a rich component of spindle medullary-like cells, had levels of AIRE RNA (0.007) even lower than those present in tumours made mainly of cortical-like cells (B1, AIRE = 0.05; B2, AIRE = 0.03; B3, AIRE = 0.01). It should be emphasized that the medullary-like cells of AB thymomas are not organized to form fully developed medullary areas with Hassall's bodies, and that the areas of medullary differentiation with Hassall's bodies present in B1 organoid thymomas represent a small proportion of the tumour.

Expression of AIRE-regulated genes

A possible involvement of AIRE in the development of MG was suggested by the observation that medullary thymic epithelial cells (TEC) isolated from AIRE-deficient mice contain about 10-fold less mRNA for CHRNA 1, a gene coding for the acetylcholine receptor, compared with C57Bl/6 wild-type mice (Fig. 3). Defective expression of AChR peptides by medullary epithelial cells of human thymi or thymomas might be theoretically determinant for the development of autoreactive clones responsible for MG. This possibility was investigated in the 34 cases of human thymoma obtained from 20 MG^– patients and from 14 MG⁺ patients, and in 18 non-neoplastic thymi obtained from 10 MG^– and eight MG⁺ patients. Total RNA extracted from frozen sections was tested for the presence of human CHRNA1, the orthologue of mouse CHRNA 1. No significant difference was found in the two groups (thymoma MG⁺, CHRNA1 = 0.013 ± 0.03; thymoma MG^–, CHRNA1 = 0.01 ± 0.03; non-neoplastic thymi MG⁺, CHRNA1 = 0.02 ± 0.01; non-neoplastic thymi MG^–, CHRNA1 = 0.19 ± 0.6).

Fig. 3.

Relative expression of CHRNA1 in the autoimmune regulatory gene (AIRE)-deficient murine thymic epithelial cells (mTECs) compared to C57Bl/6 wild-type mTECs. Amplification of housekeeping gene, Actb, was used to normalize the amount of total cDNA in each reaction. Reactions were performed in technical triplicate. Error bars = standard deviation.

It has been reported that activation of AIRE regulates the transcription of some chemokine and cytokine genes involved in antigen presentation [17]. In the experiments reported in Table 3 we have evaluated the expression of AIRE-regulated chemokines and cytokines in 26 thymomas and in 15 samples of non-neoplastic thymuses. It was found that interleukin (IL)-4, CCL17, CCL22 and CCL19 were significantly more expressed in the normal thymus compared with thymomas. No association was found between chemokines/cytokines expression and the presence of MG and/or the histological type of the tumour.

Table 3.

Expression of cytokines and chemokines known to be regulated by the autoimmune regulatory gene (AIRE) in sections of thymomas and of histologically normal thymus.

Gene	All thymomas (n = 26)	MG– thymomas (n = 12)	MG+ thymomas (n = 14)	Non-neoplastic thymus (n = 15)	t-test thymus versus thymoma
AIRE	0.02 ± 0.09*	0.02 ± 0.09	0.03 ± 0.09	0.44 ± 0.1	0.03
IL-4	6 243**	3 850	9 232	10 337	0.02
IL-12a	2 703	2 542	2 903	3 008	0.4
IL9	45 275	46 680	43 518	46 159	0.3
CCL17 (TARC)	2 029	1 243	4 296	14 288	6 × 10−7
CCL22 (MDC)	11 545	9 930	13 967	17 138	0.03
CCL19 (ELC)	16 084	14 898	17 564	23 348	0.008

^*AIRE mRNA was evaluated by real-time quantitative reverse transcription–polymerase chain reaction. The results are expressed as relative levels of AIRE mRNAs.

^**Macroarray analysis performed using the human inflammatory cytokines and receptor oligonucleotide arrays. IL: interleukin; MG: myasthenia gravis.

Presence of CD25⁺/FoxP3⁺ T_reg cells

The presence of T_reg cells was investigated using immunohistochemistry in three histologically normal thymi, in four thymi with follicular hyperplasia and in nine thymomas. Tissue sections were immunostained for CD25 and for FoxP3; the mean number of positive cells present in a 10 high power field (HPF) is reported in Table 4. In normal and hyperplastic thymi CD25⁺/FoxP3⁺ cells were much more numerous in the medulla, and were often located at the margins of Hassall's bodies (Fig. 4); apparently, their number was not influenced by the presence of MG (Table 4). FoxP3⁺ cells and CD25⁺ cells exhibited a similar pattern of tissue distribution, but FoxP3⁺ cells were more numerous; this finding may be due to a difference in the intensity of expression of the two antigens, and/or to a difference in the binding affinity of the two antibodies used for immunostaining. FoxP3⁺ and CD25⁺ cells were significantly less numerous in thymomas (Fig. 4), where isolated positive cells were observed in the medullary areas of B1 organoid thymomas.

Table 4.

Tissue distribution of forkhead box P3 (FoxP3)⁺ and CD25⁺ cells in the histologically normal thymus, in the thymus with follicular hyperplasia and in thymomas.^*

		Foxp3		CD25
Histology	No. of cases	Cortex	Medulla	Cortex	Medulla
Histologically normal thymus from MG– patients	3	50.4 ± 37	114 ± 99	6.5 ± 7.2	33.8 ± 18.2
Thymus with follicular hyperplasia from MG+ patients	4	26 ± 15	140 ± 123	4.9 ± 2.7	23.7 ± 11.1
Thymoma	12	7.4 ± 8.5		0.28 ± 0.7**

^*Paraffin sections were immunostained for Foxp3 [monoclonal antibody (mAb) 236 A/E7, 1:40 dilution] and CD25 (mAb clone 4C9, 1:50 dilution). In each case, the number of positive cells was estimated in a 10 high power field (HPF) × 400). The mean ± s.d. is reported in the table.

^**P = 0.001 compared with normal and hyperplastic thymi. MG: myasthenia gravis.

Fig. 4.

Distribution of T_reg cells in the thymus. Tissue sections of a histologically normal thymus (a, c) and of an organoid thymoma (b, d) were immunostained for forkhead box P3 (FoxP3) (a, b) and for CD25 (c, d). In the non-neoplastic thymus the cells stained for FoxP3 in the nucleus and for CD25 at membrane level are numerous and are located at the margins and around a Hassall's body. In the organoid thymoma the stained cells are rare (arrows), and are often detected in the small areas of medullary differentiation.

A quantitative estimate of T_reg cells present in normal and in tumoral thymic tissue was achieved through evaluation of FoxP3 RNA by real-time PCR (Table 1). The levels of FoxP3 RNA detected in six MG^– histologically normal thymi, four MG⁺ histologically normal thymi and seven MG⁺ thymi with follicular hyperplasia were in a similar range, and were significantly higher (P = 0.02) than those observed in 31 cases of thymomas. The levels of FoxP3 RNA detected in 19 cases of MG^– thymoma (14.9 ± 25) were 1.7-fold higher than those observed in 12 cases of MG⁺ thymomas (8.5 ± 10); however, this difference was not statistically significant. Among the histological variants of the tumour, higher levels of FoxP3 RNA were observed in B1 organoid thymomas.

Discussion

In the present study we provide evidence that AIRE is poorly expressed by tumour cells of thymomas. A possible explanation for our findings is that the large majority of human thymomas are made of cortical-like epithelial cells which are not supposed to express AIRE. Cortical and medullary epithelial cells derive from a common precursor cell of endodermal origin [18]. It has been proposed that this dual differentiation capacity is retained by tumour cells of some histological subtypes of thymomas, such as mixed cortico–medullary (AB) and B-type thymomas, particularly in organoid (B1) thymomas. In our series, we have observed that 10 cases of AB thymoma with a prominent spindle cell component contained levels of AIRE RNA which were even lower than those present in cortical thymomas (B1, B2, B3). These unexpected findings may have different explanations. It could be that spindle cells of AB thymomas are not functionally related to medullary epithelial cells; or that spindle cells of AB thymomas are not differentiated enough along the medullary line to retain the capacity to express AIRE; or that the neoplastic transformation per se has interfered with the capacity to express AIRE in spindle thymoma cells.

LCM and immunohistochemistry revealed that expression of AIRE and presence of FoxP3⁺/CD25⁺ cells are more prominent in the small areas of medullary differentiation of B1 organoid thymomas, which contain rudimentary Hassall's bodies. Our observations are consistent with a recent report indicating that Hassall's bodies play a pivotal role in the functional organization of the thymic medulla; in fact, they produce thymic stromal lymphopoietin (TSLP), a cytokine which induces the generation of CD11c⁺ dendritic cells, which in turn allows the development of T_reg cells [19]. Our immunohistological study has confirmed that FoxP3⁺/CD25⁺ T_reg cells and AIRE-positive cells are more numerous at the margins of Hassall's bodies [20,21]; this close spatial relation may represent a necessary prerequisite for the development of a functional interaction. Finally, it has been suggested that AIRE may interfere with the development of T_reg cells [15]; in fact, it was shown that APECED patients have a significant reduction in the number and functional capacity of circulating T_reg cells, perhaps because of a defective formation within the thymus. Taken together, this evidence seems to indicate that AIRE expression and development of T_reg cells are both dependent on the presence of a histologically well-developed thymic medullary area.

Both defective expression of AIRE and poor content of T_reg cells might facilitate the development of thymoma-associated autoimmune diseases. Studies of AIRE knock-out mice have provided direct evidence that Aire has a crucial role in preventing autoimmunity [22,23]. The function of AIRE is that to induce the transcription in medullary epithelial cells of a diverse array of ectopic, organ-specific antigens that are associated with autoimmune diseases; presentation of these antigens to medullary thymocytes is effective in deleting and/or suppressing autoreactive T cells, thus preventing organ-specific autoimmune reactions [7,17]. The potential relevance of this mechanism has been proved further in a recent paper, in which it was shown that spontaneous autoimmunity is prevented by thymic expression of a single self-antigen [24]. In the present study we provide evidence that medullary TEC obtained from AIRE-deficient mice are characterized by reduced expression of CHRNA1, a gene coding for AChR. This observation has provided the rationale to investigate the possibility that poor expression of AIRE in human thymomas is one of the events favouring the development of MG. Unfortunately, we have not been able to find such a correlation, as poor expression of AIRE was present to a similar extent in thymomas obtained from MG⁺ and MG^– patients. In a further characterization of our cases we have investigated the presence of human CHRNA1, coding for the α1 subunit of AChR; our results are similar to those described in a recent paper [25], and indicate that there is no peculiar association between the levels of expression of the α-subunit of AChR and the development of MG. At least three different interpretations can be provided for our findings; either AIRE is not involved in the development of MG; or expression of AIRE is relevant, but needs to be associated with other events which are crucial for the full development of the disease; or our experimental approach was not sensitive enough to highlight the pathogenetic role of AIRE in thymoma-associated MG.

In mice AIRE has the additional function to influence the transcription of genes encoding proteins involved in antigen processing or in antigen presentation [16]; in AIRE-deficient mice the chemokines CCL17 and CCL22 are down-regulated and the cytokines IL4 and IL12a are up-regulated. It has been proposed that the control exerted by AIRE on the expression of these molecules is relevant to increase the capacity of thymic medullary cells to present organ-specific antigens to thymocytes. We have evaluated the RNA expression of some of these chemokines/cytokines in our cases, and we have found that IL-4, CCL17, CCL22 and CCL19 are expressed significantly less in thymomas compared with the non-neoplastic thymi. We are aware that these differences may not be related to the low levels of AIRE transcription in tumour cells; nevertheless, they may indicate that the chemokine/cytokine network is generally poorly developed in the tumoral microenvironment.

T_reg cells are a subset of CD25⁺ CD4⁺ T cells which express the transcription factor FoxP3 [8,10,11]. They have inhibitory effects on antigen-specific activation of naive autologous T cells, are key controllers of self-reactive cells and contribute to the maintenance of immunological tolerance. In the present study we have observed that most cases of thymoma have a poor content of CD25⁺/FoxP3⁺ T_reg cells, but we have also noted that the defect is independent of the association with MG. Our findings are similar to those described in two recent studies addressing to the role of thymic T_reg cells in the development of MG. Ströbel et al. [26] reported that thymomas of MG⁺ and MG^– patients had both reduced numbers of CD25⁺/FoxP3⁺ T_reg cells inside the tumour, but that only MG⁺ thymomas were capable of exporting CD4⁺ naive T cells responsible for the development of the autoimmune disease. The authors proposed that the defect of T_reg cells might contribute to impair peripheral control of autoreactive T cells, even though it is very likely that other major events are involved in the pathogenesis of the disease. Balandina et al. [27] provided evidence that MG⁺ thymi with follicular hyperplasia have normal numbers of T_reg cells which, however, are functionally defective in in vitro assays; they concluded that the impaired function of thymic T_reg cells might be determinant for the development of MG. In the present study we do not provide any information on the functional capacity of the intratumoral T_reg cells; nevertheless, our findings indicate that in most cases of thymoma maturation of intratumoral thymocytes takes place in the absence of histologically recognizable and functionally active medullary areas. It is reasonable to postulate that this profound alteration may contribute to facilitate the development of autoreactive T cells and, hence, of thymoma-associated autoimmune diseases (please note that while this work was under consideration, a related article was published [28]).