Constitutive Activation of p38 and ERK1/2 MAPKs in Epithelial Cells of Myasthenic Thymus Leads to IL-6 and RANTES Overexpression: Effects on Survival and Migration of Peripheral T and B Cells¹

Myasthenia gravis (MG)⁴ is a rare autoimmune disease due to failure of neuromuscular transmission caused by autoantibody targeting to acetylcholine receptors (AChR). The recurrent association with thymic abnormalities, such as thymic hyperplasia and thymoma (60–75% of the cases), is a peculiar feature of MG (1, 2). This, together with the finding that thymectomy reduces the amount of circulating anti-AChR Abs and the rate of clinical relapses, has implicated thymus in the onset and/or the maintenance of the disease (1, 3).

Thymus is a chimeric organ, with each lobule composed of true epithelial spaces (cortex and medulla) where thymocytes differentiate and perivascular spaces (PVS), which are distinct anatomical regions belonging to the peripheral immune system, are devoid of thymocytes and are populated instead by mature lymphocytes and eosinophils, increasing in number with age (4, 5). Thymic epithelial cells (TEC) are essential in building and maintaining this morphofunctional organization by: 1) forming intralobular scaffolds trafficked by thymocytes; 2) modeling tight palisades separating true epithelial from PVS; and 3) regulating thymocyte proliferation and differentiation through cell-cell adhesion and soluble factors secretion (4, 5).

This lobular architecture is severely subverted in thymic hyperplasia, the most frequent disorder associated with MG (50–60% of the cases). In this disease, peripheral lymphocytes home and accumulate in PVS to an extent that interrupts the continuity of the structures interposed between PVS and medullary spaces, including thymic epithelium, thus potentially allowing intercompartment communications of pathogenetic significance (4, 5, 6, 7, 8). Of note, infiltrating lymphocytes include T cells specific for AChR epitopes and B cells that produce anti-AChR Abs and form germinal centers (GC) (4, 5, 6, 7).

Lymphoid infiltration of PVS in MG thymus is likely to be governed by multiple mechanisms, including cytokine and chemokine activity. Among them, IL-6 and RANTES may be of great relevance. In fact, IL-6 is a well-known B cell differentiation/antiapoptotic factor, with additional regulatory functions on proliferation, differentiation and survival of lymphocytes, maintenance of thymic structure in the pre- and postnatal life, and regulation of thymic atrophy during aging (9, 10, 11). RANTES has been observed to regulate the transepithelial migration of T cells (12).

Abnormal production of IL-6 by MG thymic epithelium has been observed “in vivo” and in a limited number of MG-TEC cultured “in vitro” (8, 13, 14). However, the molecular mechanisms leading to this phenomenon still remain to be elucidated. In normal TEC derived from infants, we have reported previously that both the basal and inducible expression of IL-6 are regulated by p38 MAPK (p38) and ERK1/2 MAPK activation, lying downstream of β₁ or β₄ integrin recruitment in TEC/thymocytes and TEC/extracellular matrix (ECM) adhesion and activating NF-κB and NF-IL6 transcription factors (15, 16).

These observations provided us with the rationale for investigating whether MAPKs were involved in the abnormal expression of IL-6 as well as of RANTES genes. Our hypothesis was investigated in TEC cultures derived from early onset MG (EOMG) patients, a subgroup specially characterized by juvenile onset, presence of thymic hyperplasia, benefit from thymectomy, and link to the MYAS 1 locus and to HLA DR3 haplotype (1, 3, 17).

Normal and MG thymic tissues were obtained during corrective cardiovascular surgery or therapeutic thymectomy following informed consent. MG diagnosis was based on clinical, electrophysiological, and pharmacological criteria. MG thymic samples were partly fixed in 10% neutral buffered formalin for 24 h and embedded in paraffin blocks for standard histological examination, partly used fresh to prepare TEC primary cultures. Normal thymic samples were used fresh partly to derive TEC primary cultures, partly to prepare thymocytes by mechanical disruption of tissue specimens followed by Ficoll-Hypaque gradient of cell suspensions. Thymocytes, at least 95% viable, were either used in adhesion assays with MG-TEC (see below) or cultured at 5 × 10⁶ cell/ml for 24 h in RPMI 1640 at 10% FCS in the presence of 5 ng/ml PMA (Sigma-Aldrich) to collect supernatants. Aliquots of frozen thymocytes were also stored.

PBMC were isolated by Ficoll-Hypaque gradient from blood samples obtained following informed consent from MG patients and age-matched normal volunteers. T cells were isolated from macrophage-depleted populations by E-rosetting or MACS separation columns after coating with anti-CD19, anti-CD20, and anti-CD14 mAbs (Miltenyi Biotec) (88–90% CD3⁺ as assessed by cytofluorimetric analysis, performed with a FACScan flow cytometer; BD Biosciences).

The following Abs were used: mAbs anti-CD3, -CD14, -CD20, and FITC-labeled F(ab′)₂ goat anti-mouse Ig (BD Biosciences); anti-ICAM-1, -VCAM-1, anti-integrin-α₃, and anti-β₁ (CD29) (Immunotech); mAb B9-12 (anti-MHC class I), provided by Dr. R. S. Accolla (University of Pavia, Varese, Italy); mAb 3E1 (anti-β₄) (Calbiochem); mAb 34F10 (anti-IFN-γR), gift of Dr. M. Tovey (Viral Oncology, Centre National de la Recherche Scientifique, Villejuif, France); F(ab′)₂ goat anti-mouse Ig (Pierce); and anti-cytokeratin 5 and 8 (Chemicon International). MAPK inhibitors SB202190 and PD098059 were purchased from Calbiochem-Novabiochem.

TEC cultures were derived as previously described (15, 16) from thymic specimens finely minced and trypsinized (0.05% trypsin/0.01% EDTA) at 37°C for 3 h. Cells were collected every 30 min, pooled, plated onto lethally irradiated 3T3-J2 cells (gift of Prof. H. Green, Harvard Medical School, Boston, MA) at 2.5 × 10⁴/cm² and cultured in humidified atmosphere of 5% CO₂ in growth medium composed of DMEM:Ham’s F-12 (3:1 mixture), 10% FCS, insulin (5 μg/ml), transferrin (5 μg/ml), adenine (0.18 nM), hydrocortisone (0.4 μg/ml), cholera toxin (0.1 nM), triiodothyronine (2 nM), epidermal growth factor (EGF) (10 ng/ml), glutamine (4 mM), and antibiotics. Second-passage cells were plated in the absence of feeder-layer cells, grown to confluence, kept at one-third of growth factors and medium complements for 24 h, and then deprived of them for 48 h before use (culture medium). If needed, cultures were additionally deprived of FCS for 12 h (minimal medium). Media were purchased from Seromed and supplements from Sigma-Aldrich. EGF was obtained from Austral Biological. Integrin cross-linking was achieved by treatment with mAb recognizing their extracellular domain and conjugated with goat anti-mouse (gam) F(ab′)₂. Control Ab was the mAb 34F10 recognizing anti-IFN αβ receptors (16). Adhesion of thymocytes, prepared from thymuses of normal adults, was conducted overnight at 1:5 TEC/thymocyte ratio. TEC and TEC supernatants were collected 24 h later after thorough washings of treated and untreated TEC. IL-6 and RANTES were quantified by ELISA (Endogen).

Cytoplasmic and nuclear extracts were prepared as described previously (18). Protein amount was measured by Coomassie protein assay reagent (Pierce). Western blotting was performed with cytosolic extracts (20 μg) fractionated by 12% SDS-PAGE, blotted to nitrocellulose membranes (Bio-Rad) and probed for MAPK-phosphorylated forms by immunoblotting using PhosphoPlus Erk, p38, and SAPK/JNK Ab kit (New England Biolabs) according to the manufacturer’s instructions. Bands were revealed by ECL system (Amersham Biosciences) and signal quantified with an UltroScan Densitometer and the built-in software (LKB Instruments).

Apoptosis was quantified by Cell Death Detection ELISA^plus kit (Roche) following the manufacturer’s instructions. Cell locomotion was studied in 24-Transwell microchemotaxis chambers (Costar) with a 5-μm pore sized polycarbonate filter separating cells (4–5 × 10⁵) from chemoattractant and tested at different concentrations. Cells were cultured overnight in RPMI 1640 medium 10% FCS (Seromed; Biochrom) and subsequently subjected in duplicate to the migration assay in RPMI 1640 medium at 1% BSA in the presence of the chemoattractant at 37°C for 4 h. Migrated cells were sampled at 1-h intervals, counted using a microscope, stained with FITC-labeled anti-CD3, anti-CD14, anti-CD20, and anti-CD56 and phenotyped by cytofluorimetry.

Gene profiling was performed by macroarray (SuperArray), according to manufacturer’s instructions, testing total RNA extracted by standard techniques from N- or MG-TEC. Membranes were exposed to Amersham hyperfilm films (Amersham Biosciences) for 3 and 7 days. Films were scanned by a Quant Image scanner (Molecular Dynamics) and results analyzed by the built-in software. Background ODs were subtracted from samples ODs and results calculated as percentage of housekeeping genes following GE Array analyzer instructions (SuperArray).

Statistical analysis was performed applying the nonparametric Mann-Whitney rank-sum test. Significance was accepted when p < 0.05.

This study was performed with a panel of TEC independently derived from eight hyperplastic thymuses of MG patients (MG-TEC) and from nine control thymuses of age-matched donors undergoing corrective cardiac surgery (N-TEC) (Table I). MG patients matched the clinical, histological, and serological criteria of early onset MG subtype of the disease, with none of them having thymoma, seven showing a generalized MG, and one presenting an ocular MG. This latter patient, as often observed in ocular MG, lacked detectable anti-AChR Abs (1, 2). Mean age of MG patients and normal adult donors was not significantly different (26 ± 10, range 11–36 years and 37 ± 13, range 11–52 years, respectively). Therapy included anticholinesterase drugs for all patients, associated with steroids in two cases.

N- and MG-TEC were grown from frozen primary stocks and, if not otherwise specified, analyzed as confluent monolayers, at the third cultural passage, free of exogenous growth factors and medium complements since 48 h. Their medullary origin was ensured by a cytokeratin 5^pos/cytokeratin 8^neg phenotype (Fig. 1, A and B), which is peculiar of the medullary epithelium, and further supported by the lack of detectable VCAM-1 (data not shown), a marker of the cortical epithelium (19, 20). No differences were scored between normal and pathological cells as regards monolayer morphology (Fig. 1, C and D), the expression of ICAM-1, of β₁ or β₄ integrins (Fig. 1, E and F), or of integrin polarization at the lateral (β₁) or the ventral (β₄) face of the cells (P. C. Marchisio, unpublished observation).

MG-TEC (nos. 1, 3, and 7) and N-TEC (nos. 1, 3, 5, and 8) cultured in the presence (culture medium) or in the absence of FCS for 12 h (minimal medium) were compared considering the expression of MAPK and MAPK signaling pathway-related genes by macroarray. FCS removal, performed to avoid signals from ECM proteins and growth factors, did not affect TEC monolayers adhesion, survival or morphology (V. Antonini, unpublished observation). As shown in Table II, mRNA of MG-TEC kept in culture medium contained abnormally abundant transcripts (range, 0.4- to 6.5-fold greater than N-TEC) of ERK1 and p38 (α, β, and γ isoforms), but not of JNK (data not shown), and of their upstream ERK-activating kinases MKKs (MEK1/2). Transcripts of MKKKs such as B-RAF, MEK kinase (MEKK)1, MEKK4, MEKK3, and MLK3 were also incremented, whereas TAK1 was expressed only in MG-TEC, as well as the serine protein kinase PAK1. The RAS family members N-RAS, H-RAS, and K-RAS were all overexpressed, as well as the ERK phosphatase MKP1/DUSP1 (21). We also performed a parallel analysis on MG-TEC (nos. 1, 3, and 7) and N-TEC (nos. 1, 3, 5, and 8) cultured in the presence of minimal medium, and the results showed that elements of the Ras-ERK1 pathway, MKK3 and ERK1 phosphatase MKP1/DUSP1, were still abnormally expressed in MG-TEC upon serum removal (range, 0.3- to 6.9-fold greater than N-TEC). Overall our data indicate an abnormal expression of many components of the ERK and p38 pathways in MG-TEC.

These observations prompted us to investigate whether MAPKs were also abnormally expressed and activated in MG-TEC. To this purpose, two independent cell lysates obtained from different cultures of all MG- and N-thymuses analyzed were sequentially immunoblotted by using Abs recognizing total or phosphorylated p38 and ERK1/2 MAPK. We also investigated JNK1/2 MAPK (Fig. 2). Densitometry analysis of three independents experiments shown in Table III demonstrated that expression of p38 (p < 0.01) and ERK1/2 (p < 0.001) was significantly greater in MG- than in N-TEC, whereas no differences were detected regarding the expression of JNK.

Moreover, both p38 and ERK1/2 were found to be basally activated in MG-TEC (eight of eight MG-TEC vs one of nine N-TEC in the case of p38 and six of eight MG-TEC vs two of nine N-TEC in the case of ERK1/2) and, in the case of p38, at levels significantly higher than those in N-TEC (p < 0.001). Again, no relevant differences were observed regarding JNK basal phosphorylation, which was detected in four of eight MG-TEC and three of nine N-TEC.

Because p38 and ERK1/2 are known regulators of IL-6 and RANTES gene expression (15, 22), we investigated their production in N-TEC and MG-TEC. Although the basal production of IL-6 and RANTES varied between TEC cultures irrespective of their normal or pathological origin, MG-TEC secreted higher amounts of IL-6 than N-TEC (range, p < 0.05–0.001) (Fig. 3). MG-TEC cultures nos. 1, 2, 3, and 5 appeared to be the greatest IL-6 producers (2- to 3-fold increase). However, even the lowest producers of the MG group produced more IL-6 than those of the N group, indicating that the overall IL-6 production in MG-TEC was up-regulated to higher levels than in normal cells. Low and high producers of RANTES were equally distributed in the N- and MG-TEC groups.

Both IL-6 and RANTES production was instead induced to significantly higher levels in MG-TEC than in N-TEC following stimulation by adhesive stimuli. As shown in Fig. 4, significant increments (range, p <0.05–0.001) were observed following β₄ integrin cross-linking, thymocyte adhesion, or, as observed in a more limited number of cultures (4 MG-TEC for IL-6 and 6 MG-TEC for RANTES), following the treatment with supernatants of PMA-stimulated thymocytes derived from the thymus of normal adults. By contrast, the cross-linking of β₁ integrin failed to function in both N- and MG-TEC (110% of untreated controls ± 46 SD and 119 ± 16 for IL-6, 123 ± 27 and 127 ± 46 for RANTES, respectively, for N- or MG-TEC). The same applied for the cross-linking of the control Ab. When individual cultures were considered, it appeared that this abnormal responsiveness was widely distributed in the MG-TEC group. In fact, all MG-TEC responded abnormally to at least one stimulus and seven of them to more than one stimulus producing abnormal amount of RANTES and IL-6 at the same time. In particular, abnormal amount of both IL-6 and RANTES were secreted in the same supernatants by four of the eight MG-TEC (nos. 3, 4, 5, and 7) after β₄ integrin cross-linking, by five of the eight MG-TEC (nos. 1, 2, 4, 6, and 7) after thymocyte adhesion, and by two of the six MG-TEC tested after treatment with thymocyte supernatants (nos. 4 and 5).

The role of p38 and ERK1/2 in the basal and the inducible production of IL-6 and RANTES of MG-TEC was then investigated by using specific pharmacological inhibitors: SB 202190 (SB), which binds and blocks p38 α and β isoforms, and PD 98059 (PD), which prevents the activation of the ERK-activating kinase MEK-1 (23). The limited number of available cells allowed us to assay only six of the eight MG-TEC and to use only one stimulus, i.e., β₄ cross-linking, chosen because this integrin is recruited in both TEC/thymocyte and TEC/laminin interaction (15, 16). MG-TEC, untreated or treated for 6 h with each inhibitor at the indicated concentrations, were stimulated by β₄ or control mAb cross-linking, washed, and kept in culture for an additional 24 h. Inhibitors, present throughout the experiments, were unable to interfere with TEC surface expression of β₄ integrin (data not shown). As shown in Fig. 5, the basal and the 2-fold increased IL-6 production induced by β₄ cross-linking were both dramatically and dose-dependently reduced by SB treatment (∼80% reduction), whereas PD reduced 40% on the average the basal IL-6 production and brought to basal levels that induced by β₄ integrin cross-linking. When measuring RANTES in the same samples, we found that SB halved the basal production of the chemokine and brought to basal level that induced by β₄ integrin cross-linking whereas PD failed to function.

To understand the functional relevance of RANTES and IL-6 overexpression in our experimental cellular system, we investigated their role in the regulation of lymphocyte migration and apoptosis. To this purpose, we used recombinant RANTES and IL-6 and peripheral lymphocytes obtained from normal controls or from a second group of nine MG patients (mean age, 40.3 ± 5 years; range, 16–60 years), free of immunosuppressive therapy and fulfilling the criteria of EOMG shown in Table I. As shown in Fig. 6,A, macrophage-depleted MG lymphocytes were driven to migrate by RANTES in a dose-dependent manner, showing a greater sensitivity to the chemokine than normal controls. Indeed, 60% of MG lymphocytes migrate in response to 5–10 ng of RANTES as compared with 25% of normal lymphocytes (both with respect to untreated controls). FACS analysis of transmigrated lymphocytes demonstrated that this occurred for both T and B lymphocytes of MG (Fig. 6, B and C). Shown are the results of the 3 h point of the time course. Similar results were obtained at 4 h. Moreover, as shown in Fig. 6 D, IL-6 rescued both N- and MG-T lymphocytes from spontaneous apoptosis. However, rescue from apoptosis was considerably greater for MG-T lymphocytes (25% rescued) than for N-TEC (8% rescue) at both doses of IL-6 used (10² and 10⁴ μg/ml).

In this article, we demonstrate for the first time that thymic epithelial cells of hyperplastic thymus of EOMG patients harbor a complex series of genetic alterations, including disorders of expression and phosphorylation of p38 and ERK1/2 MAPKs. Moreover, we found that as a downstream effect of these alterations both the basal and the inducible expression of IL-6 and RANTES are regulated to higher levels than in normal TEC. The relevance of this phenomenon in the pathogenesis of thymic hyperplasia is underlined by the additional observation that survival and migration of peripheral T and/or B cells of MG patients are regulated by IL-6 and RANTES in a dose-dependent manner.

The most remarkable observations provided by our work is that MG-TEC display abnormalities of p38 and ERK1/2 in that Western blot analysis of all TEC cultures demonstrated that the overall intracellular amount of p38 and ERK1/2 in MG-TEC was significantly greater than that of N-TEC. Moreover, it was evidenced that also the amount of the phosphorylated forms of p38 and ERK1/2 was abnormally increased. This appeared to be solely a consequence of overexpression in the case of ERK1/2 because the fraction of the phosphorylated molecules remained constant, whereas it was due to both overexpression and an increase in the fraction of phosphorylated molecules in the case of p38, thus suggesting that also the mechanism of posttranscriptional control of this kinase could be misruled. It is tempting to speculate that the increased transcription of the ERK1 phosphatase DUSP1 in MG-TEC contributed to lower the phosphorylated fraction of ERK1/2.

Results of gene profiling performed with TEC cultures were in accordance with results of Western blot analysis. In fact, gene profiling puts into evidence that multiple genetic alterations could be implicated in the constitutive overexpression and activation of p38 and ERK1/2 in MG-TEC. We observed an augmented transcription of upstream MEKKs, of members of the Ras family, of B-RAF and of EGFR (data not shown), all previously found to be involved in the overexpression and activation of p38 and ERK1/2 in malignancies (24, 25, 26). Elucidation of EGFR or RAS activity and their possible role in the MG-TEC abnormalities observed by us will require further investigation.

Experiments with selective inhibitors dissected the different regulatory activities of the two kinases, demonstrating that p38 regulated both IL-6 and RANTES expression, played a dominant role over ERK1/2 in the case of IL-6, and exerted a sustained activity in the case of RANTES. ERK1/2 was instead required for IL-6 but not RANTES expression. Although side activities of inhibitors on other kinases could not be excluded (23), our data strongly suggest the crucial role of p38 in the abnormal production of IL-6 and/or RANTES in MG-TEC. In fact, IL-6 and RANTES production was tuned to abnormal levels in all MG-TEC assayed.

The possibility that this observation may be biased in the MG group by the prevalence of females or by the higher percentage of subjects < 20 years than in the control group (37.5 vs 11%, respectively) appears very unlikely because of the following considerations: 1) previous reports showed that estrogens reduce the production of both IL-6 and RANTES and androgens instead increase that of IL-6 (27, 28, 29); and 2) our observations obtained in studies comparing at both polyclonal and clonal level TEC from MG patients, normal adults, and normal infants <5 years of age (V. Antonini, manuscript in preparation) showed that IL-6 and RANTES production of MG-TEC, including those derived from the younger patients examined in this article, were far higher than that of TEC from infants. The possibility that a very different microenvironment of the starting material could account for the differences between MG-TEC and N-TEC reported herein can be therefore ruled out.

The production of IL-6 was found to be already basally increased, unlike that of RANTES; however, both could be raised to abnormally high levels in response to stimuli physiologically present in their microenvironment. The lack of difference between the basal expression of RANTES of N- and MG-TEC may reflect the regulation of the basal and the inducible gene expression of RANTES. According to the literature (30, 31) and to results from our group (data not shown), ERK1/2 activity appears not to be involved in the basal and inducible RANTES gene expression. p38 and JNK are instead the active players, with JNK exerting a dominant role due to its regulation of ATF and Jun transcription factors cooperating with NF-κB and NF-IL6 transactivators (31). As no difference was observed between N- and MG-TEC regarding JNK transcription and expression, it is conceivable that the constitutive activation of p38 did not reach a threshold able to increase RANTES expression basally. This threshold could instead be reached after stimulation where sustained activation of p38 may increase, in a STAT-dependent manner, the activity of the IFN regulatory factor family, whose members govern the inducible expression of the chemokine (31, 32)

One of the major differences between N- and MG-TEC concerned the response to the cross-linking of β₄ integrin, a laminin receptor strongly expressed at the TEC surface throughout their life. We have demonstrated previously that β₄ integrin cross-linking results in an abundant IL-6 secretion in infants (15). In the present study, we found that the strong signaling activity observed in TEC from infants declines in normal adults but not in MG-TEC. By contrast, the signaling activity of β₁ integrin declined in both N- and MG-TEC (16). Age-dependent regulations of β₁ integrin signaling, as in the case of apoptosis and cytokine production (V. Antonini, manuscript in preparation) or further anomalies of β₄ structure or function, may explain these phenomena. If so, β₄ integrin and MAPK alterations may concomitantly sustain the increase of expression of IL-6 and RANTES. Soluble factors secreted by stimulated thymocytes may concur in maintaining high levels of IL-6 and RANTES secretion. In fact, we have observed previously that soluble factors secreted by unstimulated thymocytes failed to induce IL-6 gene expression in normal TEC (15). However, a short exposure of thymocytes to PMA was sufficient to endow the supernatants of thymocytes with a strong inducing activity. Thus, this could mimic an in vivo situation where thymocytes triggered by microenvironment-related stimuli may maximize their secretion of soluble factors, thereby amplifying a genetically determined overresponsiveness.

IL-6 and RANTES exert pivotal functions within the thymus (8, 9, 10, 11, 33). Thus, the concomitant deployment of their activities may critically affect survival, differentiation and proliferation of its lymphoid and nonlymphoid components. As undesired effects of their physiological function, IL-6 might support autoreactive clone proliferation in true epithelial spaces and PVS, RANTES could alter intrathymic migration of double and single positive thymocytes expressing specific receptors such as CCR5, could promote the recruitment of peripheral T and B cell in PVS and then support their progression toward true epithelial spaces.

The complex pathophysiological role of IL-6 in MG has been recently investigated and elucidated in an elegant mouse model. In fact, Deng et al. (34) have analyzed the cellular and humoral immune responses to AChR and the development of clinical experimental autoimmune MG in normally developed IL-6^−/− and wild-type mice, immunized with AChR. Their findings indicate that immunized IL-6^−/− mice develop normally and acquire a normal immune system, but they show reduced size of GC and reduced serum levels of anti-AChR Abs and C3 because the IL-6 physiological effect on GC development, B cell maturation, differentiation, Ab production, and complement C3 mRNA expression is absent. All these observations may explain the suppression of experimental autoimmune MG observed in IL-6^−/− mice and strongly support the pathophysiological role of the aberrant production of IL-6 that we observed in the MG patients analyzed herein.

All the results presented in this article were obtained in a sophisticated experimental system: 1) TEC were originated from medullary epithelium, the subset more directly involved in hyperplastic rearrangements of MG thymus; 2) stimuli provided mimicked the physiological interaction of TEC with other TEC, ECM components, or developing thymocytes; 3) MG-TEC abnormalities were evidenced by comparison with a panel of normal, age-matched controls analyzed at the same stage of differentiation; moreover, the observation that MAPK expression is found in all MG-TEC cultures rules out possible artifacts due to in vitro selection of nonrepresentative phenotypes in primary cultures; and 4) the potential outcome of aberrant production of IL-6 and RANTES was evaluated on peripheral lymphocytes of a panel of EOMG patients. The above setup allowed us to closely mimic an in vivo situation likely to take place within the thymus and to record subtle differences between MG and normal conditions, which may not come to the observation of investigators under less stringent criteria. Thus, our findings may allow us to draw a possible scenario where the genetically altered epithelial cells: 1) facilitate rescue from apoptosis of developing thymocytes through the secretion of abnormally high concentrations of RANTES and IL-6; 2) attract T and B lymphocytes to PVS by increasing RANTES production; and 3) sustain the survival and proliferation of T and B lymphocytes in PVS. Under these circumstances, persistence of autoreactive T cell and B cell clones might be favored, leading to pathogenetic events associated with the development of MG.

In addition, our results raise the interesting possibility that p38 and ERK1/2 inhibitors could be used in the treatment of EOMG to prevent the MAPK-mediated up-regulation of IL-6 and RANTES and consequently the PVS lymphocyte infiltration.

We are indebted to Prof. P. C. Marchisio for immunohistochemical examination of TEC-thymocyte cocultures and helpful criticisms and the Cardiosurgery of Verona, Prof. L. Durelli, and Dr. M. Clerico for providing some of the thymic specimens and blood samples of normal donors and MG patients. We gratefully acknowledge Dr. M. Tovey for gifting 34F10 mAb.

The authors have no financial conflict of interest.