• Interferon;
  • interleukin-12;
  • autoantibodies;
  • myasthenia gravis


  1. Top of page

We have screened for spontaneous anticytokine autoantibodies in patients with infections, neoplasms and autoimmune diseases, because of their increasingly reported co-occurrence. We tested for both binding and neutralizing autoantibodies to a range of human cytokines, including interleukin-1alpha (IL-1α), IL-1β, IL-2, IL-4, IL-6, IL-8, IL-10, IL-12, IL-18, interferon-alpha2 (IFN-α2), IFN-ω, IFN-β, IFN-γ, tumour necrosis factor alpha (TNF-α), transforming growth factor beta-1 (TGF-β1) and granulocyte-macrophage colony stimulating factor (GM-CSF), in plasmas or sera. With two notable exceptions described below, we found only occasional, mostly low-titre, non-neutralizing antibodies, mainly to GM-CSF; also to IL-10 in pemphigoid. Strikingly, however, high-titre, mainly IgG, autoantibodies to IFN-α2, IFN-ω and IL-12 were common at diagnosis in patients with late-onset myasthenia gravis (LOMG+), thymoma (T) but no MG (TMG) and especially with both thymoma and MG together (TMG+). The antibodies recognized other closely related type I IFN-α subtypes, but rarely the distantly related type I IFN-β, and never (detectably) the unrelated type II IFN-γ. Antibodies to IL-12 showed a similar distribution to those against IFN-α2, although prevalences were slightly lower; correlations between individual titres against each were so modest that they appear to be entirely different specificities. Neither showed any obvious correlations with clinical parameters including thymoma histology and HLA type, but they did increase sharply if the tumours recurred. These antibodies neutralized their respective cytokine in bioassays in vitro; although they persisted for years severe infections were surprisingly uncommon, despite the immunosuppressive therapy also used in most cases. These findings must hold valuable clues to autoimmunizing mechanisms in paraneoplastic autoimmunity.


  1. Top of page

‘Spontaneous’ autoantibodies (i.e. occurring without known provocation) can be directed against a spectrum of self-antigens, including serum proteins and DNA. They are rare in healthy subjects and, if present, tend to be low-affinity IgMs [1]. However, especially in autoantibody-mediated disorders, their frequency and titre may be greatly increased [2–4], and they are usually IgGs. Two humorally mediated autoimmune diseases provide striking examples. In myasthenia gravis (MG) and Lambert–Eaton myasthenic syndrome (LEMS), the characteristic muscle weakness is mediated mainly by autoantibodies to the nicotinic acetylcholine receptor (AChR) at the motor endplate in MG [2,5] and the voltage-gated calcium channels in LEMS [2,6,7]. About 10–15% of ‘seronegative’ cases with typical generalized MG instead have autoantibodies to distinct endplate antigens such as the muscle-specific kinase MuSK [8], and not to AChR. Anti-AChR-seropositive MG patients fall into three distinct subsets: about 20% with early-onset MG (EOMG+) (before age 40), 50% with later onset MG (LOMG+) and 10% with thymoma (thymoma-associated MG, TMG+ onset usually after age 35). In addition to anti-AChR, TMG+ patients nearly all have autoantibodies to striated muscle antigens, especially titin [9]. Besides MG, thymomas associate occasionally with bone marrow aplasias, e.g. of red cells or neutrophils [10,11] or with immunodeficiency [12], that apparently have an autoimmune basis [13]. Thymomas are tumours mainly of thymic cortical epithelium, with several histological subtypes [14]. The most common in MG patients (WHO B2) contain typically abundant immature thymocytes. We suspect that AChR-specific helper T cells are selected or autoimmunized there, but induce specific B cell responses only after export to the periphery [15].

There are reports of sporadic antibodies to interferon (IFN) [16] and a number of other cytokines, including IL-1α, IL-2, IL-6 [16], IL-8 [17,18], IL-10 [19], GM-CSF [20] and TNF-α[21] in various autoimmune, malignant or infectious diseases and also in occasional healthy controls [16,22–25]. While many of them can neutralize their respective cytokine in vitro, in vivo effects and clinical significance remain enigmatic, except for the pulmonary alveolar proteinosis syndrome (IPAP) which associates with autoantibodies to GM-CSF or genetic deficiency thereof [26]. We have briefly reported previously [27] a high incidence of spontaneous neutralizing autoantibodies to IFN-α2 and IL-12 in TMG+. Here we summarize a wide-ranging survey of these and other anticytokine autoantibodies in the main MG and LEMS patient subgroups and a variety of other autoimmune, malignant and viral diseases, and suggest novel clues to autoimmunizing cell types that are otherwise very hard to find in humans.


  1. Top of page


The MG and LEMS patients were attending our neurology clinics in London or Oxford. Diagnoses were based on typical clinical and electromyographic features; all MG cases had elevated serum anti-AChR antibodies, apart from the seronegative MG (SNMG+) subgroup. All the TMG+ and TMG cases had histologically confirmed thymomas; the EOMG+ patients had all been thymectomized with no sign of tumours. The LOMG+ cases had mainly been followed for several years without radiological evidence of any mediastinal mass, which effectively excludes thymoma [28]. Many cases were sampled at diagnosis; most of the others were taking alternate day corticosteroids (± azathioprine). The TMG patients had no clinical signs of MG, but a few had low levels of anti-AChR antibodies.

The eight ‘paraneoplastic’ samples (from Dr B. Lang, Oxford) were from patients with breast or ovarian tumours plus cerebellar or sensory syndromes. Other sera were from patients: positive for (a) oligoclonal bands in spinal fluid (multiple sclerosis, ‘MS’) or (b) serum antithyroid microsomal antibodies (all from Prof. H. Chapel, Oxford); (c) with insulin-dependent diabetes mellitus (IDDM) (aged 3–15 years; Dr D. Dunger, Oxford); (d) with pemphigoid (Prof. F. Wojnarowska, Oxford); (e) with rheumatoid arthritis (RA) or systemic lupus erythematosus (SLE) (Prof. D. Isenberg, University College Hospital, London); (f) with histologically proven melanoma (Prof. A. Harris, Oxford) or (g) with small cell lung cancer (SCLC) without LEMS, or with a range of common carcinomas (Dr D. Talbot, Oxford); and (h) with HIV-1, hepatitis B or C infections (Dr J. Kurtz, Oxford). The controls were healthy relatives of our EOMG+ patients and healthy laboratory workers.

Serum or plasma samples (‘sera’) were taken (with informed consent and Ethical Committee approval) and stored at − 20°C/−70°C; they were diluted in the media appropriate for cytokine-specific immunoassays and bioassays (see below).

Interferons and other cytokines

IFNs and other cytokines, donated generously by the stated manufacturers, were used in autoantibody-binding ELISA specific for the following: recombinant human (rhu)IFN-α2a (Hoffmann-La Roche, Basle, Switzerland); rhuIFN-α1/8 (Ciba-Geigy, Basle, Switzerland); natural huIFN-α8 (Glaxo-Wellcome, Beckenham, Kent, UK); rhuIFN-Con1 (Amgen, USA); rhuIFN-β (Renschler, Germany); rhuIFN-ω (Bender & Co., Vienna, Austria); rhuIFN-γ (Roussel-Uclaf, France); rhuGM-CSF (Immunex, Seattle, USA); rhuTGF-β1 and rhuTGF-β2 (Celltrix, USA); rhuIL-1α (Dainippon, Japan); rhuIL-1β (Immunex); rhuIL-2 (Amgen, USA); rhuIL-4 (Schering-Plough, USA); rhuIL-6 (Sandoz, Basle, Switzerland); rhuIL-10 (Schering-Plough, USA); rhuIL-12 (Hoffman-La Roche, Basle, Switzerland) and rhuIL-18 (Hayashibara Biochemical Laboratories, Okayama, Japan).

Binding ELISA for the detection of anti-IFN and anticytokine autoantibodies

Round-bottomed microtitre wells (Dynatech) were coated with IFN or cytokine solutions at 2 µg protein/ml [phosphate-buffered saline (PBS), pH 7·0], 0·1 ml per well, for 2 h at room temperature (22°C). Antigen solution was removed and wells blocked with 1·0% human serum albumin (HSA) in PBS for 0·5 h at 22°C or overnight at 4°C. Patient sera were diluted serially and added to coated wells, incubated for 2 h at 22°C, removed and wells washed four times with 0·1% Synperonic solution. Following washing, antihuman IgG- or antihuman Ig isotype-peroxidase conjugate (Sigma Chemical Co. Ltd), 0·1 ml/well of 1 : 1000 dilution was added to all wells and incubation continued at 22°C for a further 2 h. Wells were washed again four times with 0·1% Synperonic solution before addition of orthophenylene diamine substrate solution. Colour development was terminated after 0·5 h at 22°C by addition of 2 m H2 SO4 (0·05 ml/well).

Antiviral IFN neutralization assays (AVINA)

We pretreated the human glioblastoma cell line 2D9 (provided generously by Dr W. Däubener [29]) with diluted IFN preparations [at 10 laboratory units (LU) per ml] that had been preincubated for 1 h with serial dilutions of test sera [30,31]. The cells were then challenged with encephalomyocarditis virus; after 24 h, the cell monolayers were stained with 0·05% amido blue-black and fixed with 4% formaldehyde solution in acetic acid buffer [30], destained with 0·15 ml of 0·05 m NaOH solution and absorbances were read at 620 nm. The neutralizing antibody titre [32] was the dilution of serum that reduces 10 LU/ml of IFN to 1 LU/ml (the normal end-point of antiviral assays [30]). The cut-off for positivity was a titre of 100, which our healthy controls never exceeded.

Other biological cytokine neutralization assays

We measured antibody neutralization of TNF-α or TNF-β in cytotoxicity assays using the human rhabdomyosarcoma cell line, KYM-1D4 [33]; of TGF-β1 or TGF-β2 in antiproliferation assays using the mink lung epithelial cell line Mv-1-Lu (CCL-64, ATCC) [34]; of GM-CSF in proliferation assays using the human erythroleukaemic cell line TF-1 [35], pulsing for 4 h with [3H]-thymidine after incubation for 48 h, before harvesting and scintillation counting [36]; of interleukin-induced growth stimulatory activity using murine T cell-line D10S (IL-1), CTLL-2 (IL-2), CT-h4S cells (IL-4), murine plasmacytoma cell line B9 (IL-6), murine pro-B cell line transfected with murine IL-10 receptor BaMr (IL-10) and human T-cell line KIT225 (IL-12) [36], pulsing with [3H]-thymidine, harvesting and counting as above [36]; and of IL-18 by inhibition of IL-18-stimulated IFN-γ production by human KG1 myelomonocytic cell line [37]. Assay results for all neutralization assays were analysed in the same way as for IFN (see above).


  1. Top of page

Detection of autoantibodies against IFN-α2 in patients’ sera

Binding antibodies against IFN-α2 were found at strikingly high levels (in ELISAs). They were most prevalent in patients with LOMG+ (32% positive), thymoma alone (TMG; 57%) or particularly with both together (TMG+ 62%; Table 1, Fig. 1), and were rare/low in titre in other autoimmune and malignant diseases and low/modest in titre in viral diseases. Their prevalence was highest of all (∼70%) in the 35 TMG+ patients sampled before any immunosuppressive therapy was started, and was still 56% in those sampled afterwards (Fig. 2). The range of binding levels was very broad, and was only marginally lower in the immunosuppressed cases (Fig. 2). Levels were comparable in some of the LOMG+ patients; also in five of 16 TMG cases who were negative for anti-AChR antibodies, in three of the four who had low anti-AChR (0·5–1·0 nm) and in one with higher levels (>1·0 nm).

Table 1.  Binding and neutralizing autoantibodies to cytokines in the circulation of patients with various autoimmune, malignant and viral diseases
Disease categoryNo. of patients% positive for binding autoantibodies to: (% positive for neutralizing autoantibodies to:)
  • *

    Plasma from 89 patients evaluated; n.d., not done.

TMG+11861·9 (64·4) 8·4 (6·7)44·1 (61·0)51·7 (31·4)12·4* (1·1)*
TMG 2157·1 (52·3) 0 (0)33·3 (28·6)23·8 (19·0)14·3 (0)
LOMG+ 2832·1 (30·8) 3·8 (3·8)46·2 (30·8)17·9 (14·3)29·2 (0)
EOMG+ 34 2·9 (0) 2·9 (0)11·8 (0) 8·8 (0)26·4 (0)
SNMG+ 15 0 (0) 0 (0) 0 (0) 0 (0)13·3 (6·7)
LEMS 31 9·7 (0) 6·5 (0) 3·2 (0) 3·2 (0)19·2 (0)
Paraneoplastic  825·0 (0) 0 (0) 0 (0) 0 (0)n.d. (n.d.)
Polymyositis/mD 11 9·1 (0) 0 (0) 0 (0) 0 (0)13·3 (0)
MS 25 0 (0) 4·0 (0) 4·0 (0) 0 (0) 2·3 (0)
Thyroid 25 0 (0) 4·0 (0) 0 (0) 0 (0)n.d. (n.d.)
IDDM (juvenile) 29 0 (0) 0 (0) 0 (0) 0 (0) 0 (0)
RA/SLE 1729·4 (5·9) 5·9 (0) 5·9 (5·9)17·6 (0)n.d. (n.d.)
Pemphigoid 67 3·0 (0)n.d. (n.d.)n.d. (n.d.) 3·0 (n.d.)n.d. (n.d.)
SCLC 1010·0 (0)10·0 (0) 0 (0) 0 (0)n.d. (n.d.)
Melanoma 36 0 (0) 0 (0) 0 (0) 0 (0)n.d. (n.d.)
Various cancers 24 4·2 (0) 0 (0) 0 (0) 0 (0) 0 (0)
Viral 9422·3 (0)n.d. (n.d.)n.d. (n.d.)29·8 (0)n.d. (n.d.)
Healthy controls 70 1·4 (0) 0 (0) 0 (0) 0 (0) 0 (0)

Figure 1. Binding autoantibodies to IFN-α2 in MG subgroups, and in patients with LEMS, RA/SLE, pemphigoid (PEMPH) or viral diseases (VIRAL) or in healthy controls (NHC). Sera were diluted 10-fold and assayed by ELISA. We consider as positive any values greater than a cut-off 2 s.d. above the mean of the controls’ (this cut-off averaged 0·2 over many assays; dotted line). The percentages of positives for each category are shown in Table 1. In the TMG group, the larger triangles represent those cases positive for anti-AChR antibodies.

Download figure to PowerPoint


Figure 2. Binding autoantibodies in TMG+ patients to type I and II IFNs (assayed by ELISA as for Fig. 1). For IFN-α2, the TMG+ patients have been divided into those already receiving immunosuppressive treatment (closed symbols, left) and those sampled at diagnosis, before any immunosuppressive treatment was given (open symbols, right). We consider as positive any values > 2 s.d. above the mean of the controls’ (average absorbance values ranged from 0·2 to 0·3 depending on the cytokine ELISA). Data for NHC are shown in Fig. 1 for IFN-α2 and non-graphically in Table 1 for IFN-β and IFN-ω; none was positive for IFN-ω, IFN-Con1 and IFN-γ (data not shown).

Download figure to PowerPoint

These antibodies were rare even in the EOMG+ and SNMG+ subgroups and in LEMS (Table 1), and levels were low/modest in the few positives (Fig. 1). In general, these results agree very well with the RIA data from these same MG subgroups (mainly from different individuals) reported recently by Buckley et al. [38]. Modest or low levels of binding autoantibodies to IFN-α2 were seen sporadically in some other autoimmune disorders (Table 1), notably RA/SLE (Table 1, Fig. 1), although not in multiple sclerosis (MS), IDDM or thyroid disease; occasionally also in paraneoplastic syndromes, SCLC, other cancers (of uterus or colon), and particularly in viral infections, mainly with HIV-1. Such borderline binding was detected with only 1/70 (1·43%) healthy controls (Table 1, Fig. 1).

Antibodies neutralizing the antiviral activity of IFN-α2 were much more tightly restricted to the TMG+, TMG and LOMG+ cases (Table 1), where titres were often strikingly high (Fig. 3). Outside these groups, neutralization was found only in the RA/SLE group (one case); it was not detected in the viral disease group despite the modest titres of binding antibodies (Table 1). In the TMG+ group, neutralizing activity coincided closely with positivity in the ELISA (Table 1), including the immunosuppressed cases (Fig. 3). Binding and neutralizing titres also correlated significantly (r2 = 0·375; P < 0·0001), although a few cases appeared to have low levels of either binding or neutralizing antibodies alone. We saw no clear correlation between these titres and anti-AChR levels, sex or onset-age, HLA type (n = 90) or thymoma histology (n = 65) in the TMG+ cases (data not shown).


Figure 3. Neutralization titres against type I IFNs and IL-12 in sera from TMG+ patients. For each cytokine (except IFN-β), the patients already receiving immunosuppressive treatment are shown by closed symbols (left) and those sampled before any immunosuppressive treatment by open symbols (right). Titres of 100 or greater are considered to be positive. No positives were found in the NHC category (Table 1 and data not shown)

Download figure to PowerPoint

Binding and neutralizing activity of sera against a range of different IFNs

Nearly all of the TMG+, TMG and LOMG+ sera containing binding autoantibodies to IFN-α2 were also positive against IFN-Con1, a synthetic IFN-α with a consensus amino acid sequence derived from the 12 IFN-α subtypes [39,40] (Figs 2 and 3). In the TMG+ cases, the binding and neutralizing titres each correlated strongly between IFN-α2 and IFN-Con1 (r2 = 0·8322, P < 0·0001 and r2 = 0·757, P < 0·0001, respectively). These autoantibodies also recognized subtype IFN-α8 and the hybrid subtype IFN-α1/8, produced by rDNA techniques (data not shown). More remarkably, IFN-ω, a natural component of leucocyte IFN [41], has only ∼60% homology to IFN-α, but was still recognized by most of the TMG+, TMG and LOMG+ sera with binding antibodies to IFN-α2 (Table 1, Fig. 2). Its titres correlated less strongly with those for IFN-α2 for binding (r2 = 0·475, P < 0·0001) and especially for neutralization (r2 = 0·242, P < 0·0001), as detailed elsewhere (in preparation).

In stark contrast, very few of these sera were positive against the more distantly related IFN-β, a type I IFN only ∼ 30% homologous to IFN-α subtypes [40] (Table 1, Fig. 2); they appeared to be scattered randomly among the TMG+ group (see Fig. 4). They were even fewer in other MG subgroups or other diseases, where they also failed to neutralize IFN-β, whereas those in the TMG+ cases had low positive neutralizing titres (Table 1, Fig. 3). Moreover, we never detected binding or neutralizing antibodies to the unrelated (type II) IFN-γ in MG (Fig. 2) or in any of the other diseases (n = 409; data not shown).


Figure 4. Lack of correlation between anti-IFN-α2- and anti-IL-12- binding antibodies in the TMG+ group. The arrows signify those few sera that also bind IFN-β, which are widely scattered.

Download figure to PowerPoint

Autoantibodies against IL-12

Binding antibodies to IL-12 showed a similar distribution to those against IFN-α2, with an even stronger bias towards the TMG+, TMG and LOMG+ groups (Table 1, Fig. 5); they were otherwise seen at low titres only in occasional cases of RA/SLE and pemphigoid, and particularly after viral infections (Fig. 5). Although prevalences were slightly lower, antibody levels were also sometimes very high in TMG+, TMG and LOMG+ (Table 1, Fig. 5), again agreeing well with the RIA data [38].


Figure 5. Binding autoantibodies to IL-12 in the same patient groups as in Fig. 1 (and assayed by ELISA in sera diluted 10-fold). Cut-off is as for Fig. 1 (it averaged 0·23 over many assays; dotted line). The percentages of positives for each category are shown in Table 1. In the TMG group, the larger triangles represent those cases which were positive for anti-AChR antibodies.

Download figure to PowerPoint

Many positive sera from TMG+, TMG and LOMG+ groups neutralized IL-12 activity (titres up to 1/10 000), despite immunosuppressive therapy (Fig. 3), although again the proportion was slightly lower than with IFN-α2 (Table 1); so far, we have never detected significant neutralizing activity in other MG subgroups or in any other patient group.

The anti-IL-12 and anti-IFN-α2 binding antibody levels correlated only very weakly (r2 = 0·1197; P < 0·005; Fig. 4). Several sera had very high binding and neutralizing titres against either cytokine alone, very strongly suggesting that these two specificities are independent. Again, the anti-IL-12 antibodies showed no obvious correlation with clinical parameters including thymoma histology or HLA type (data not shown).

Isotypes of autoantibodies to IFN-α2 and IL-12

In selected sera, the binding autoantibodies against both IFN-α2 and IL-12 were mainly IgG, with minor levels of IgM. The IgG1 subclass predominated, with significant smaller proportions of IgG2, IgG3 and IgG 4 (data not shown). Both κ and λ light chains were found, thus excluding monoclonal origins.

Longitudinal analysis of autoantibodies to IFN-α and IL-12 in TMG+ patients

In 20 TMG+ cases studied serially over an average of 8 years, levels of IFN-α2 and IL-12 autoantibodies tended to vary in parallel with the anti-AChR titre, probably reflecting changes in overall immune responsiveness and/or immunosuppressive therapy. Three TMG+ patients who had no detectable IFN-α2 autoantibodies at diagnosis developed them several years later. Similarly, two others, originally with anti-IFN-α2 alone, subsequently developed IL-12 autoantibodies; one with anti-IL-12 alone developed anti-IFN-α2 at a later date. No patients entirely lost their IFN-α2- or IL-12-autoantibodies, although they sometimes decreased during immunosuppressive treatment. In some cases these titres rose strikingly at the time of thymoma recurrence, whereas the anti-AChR autoantibodies remained low, as in the representative case shown in Table 2[38].

Table 2.  Increase in autoantibody titres against IFN-α2, IL-12 and AChR at the time of thymoma recurrence in a representative patient
  • a

    Outside brackets: neutralizing titres; inside brackets: absorbance values in binding ELISAs of 1/10 dilution of sera.

  • b

    Anti-AChR titre expressed as nanomolar. The primary ‘lympho-epithelial’ thymoma was removed from this female MG patient at age 38 .8. A large recurrence (WHO type B2) was removed partially at age 45·8, followed by chemotherapy. She developed pure red cell aplasia at age ∼ 53·6, which resolved on treatment with prednisone plus azathioprine. Over the next ∼ 3 years she has had respiratory and urinary infections, plus oral Candida and genital Herpes zoster.

Two years before recurrence 20 000 (2·26)<<100 (0·00) 4·1
At time of recurrence320 000 (3·36)5000 (2·04)14·1

Binding and neutralizing activity against a range of different cytokines

Over 400 sera were also tested for binding antibodies to IL-1α, IL-1β, IL-2, IL-4, IL-6, IL-10, IL-18, TNF-α, TGF-β1, TGF-β2 and GM-CSF. Very few were detected except: against GM-CSF (Table 1); we found modest levels in 10–30% of the cases with MG or LEMS, but almost none in the MS, IDDM, other cancer or healthy control groups; of all the positives, only two neutralized GM-CSF [20]: against IL-10 in 25% pemphigoid cases (mainly bullous): against IL-18 in about 20% viral disease patients (mainly HIV).


  1. Top of page

We have found: (1) striking prevalences of IFN-α2- and IL-12- binding autoantibodies in patients with late-onset MG, with thymoma but no MG (even without anti-AChR antibodies), and especially with both MG+ and thymoma, whether assayed by ELISA (here) or by RIA [38]; (2) clear cross-reactivity both with the synthetic ‘IFNCon1’ (with a ‘consensus’ IFN-α primary sequence [39]) and with other IFN-α subtypes, e.g. IFN-α8, and IFN-ω, although minimally with the more distantly related IFN-α and not detectably with the unrelated type II IFN-γ; (3) poor correlations between the autoantibodies to IFN-α and IL-12, indicating that they must be separate populations with distinct specificities; (4) clear neutralization by most sera that bound IFN-α and IFN-ω, and many of those binding IL-12, of their respective biological activities in vitro; (5) like the IFN-α autoantibodies that appear following bone marrow transplantation [42], they can clearly persist for many years, despite the immunosuppressive therapies that usually control the patients’ myasthenia; (6) very modest (if any) binding titres − without significant neutralization − against IFN-α alone in RA/SLE, against IFN-α, IL-12 and/or IL-18 after virus infections, against GM-CSF in neuromuscular diseases and against IL-10 in pemphigoid, despite screening many other cytokines in a very broad range of other autoimmune, malignant or viral diseases. As in earlier reports [43,44], we found significant numbers with binding antibodies to IFN-α in our viral disease group (mainly HIV); however, in contrast to Fall et al. [44], we never detected any clear neutralization. The possible clinical consequences of in vitro neutralization of these Th1-inducing cytokines are discussed elsewhere [Zhang et al. submitted].

These findings highlight two completely unexpected coincidences − the strikingly high levels of binding and neutralizing autoantibodies to IFN-α/ω and IL-12, and their occurrence in two separate subgroups of MG patients, with or without thymoma. They thus pose challenging questions about (a) why these particular cytokines are so singularly immunogenic and (b) why only in these particular autoimmune groups.

Immunogenicity of IFNs and IL-12

Clearly, whether they are produced endogenously or administered therapeutically, many cytokines can evoke low-level binding antibodies: the high titre neutralizing responses in MG+± thymoma must be qualitatively different. There are reports of low-level binding or neutralizing autoantibodies to type I and II IFNs and other cytokines in many, apparently healthy, controls [16,22–25]. Moreover, neutralizing autoantibodies to IL-1α, IFN-α, IFN-ω, GM-CSF, and more rarely to IFN-β, but not to IFN-γ, IL-12 or a range of other cytokines, are clearly present in pharmaceutical Ig products derived from plasma pools from apparently healthy donors, although we suspect that they derive from only a few individuals [45,46]. Similarly, sporadic autoantibodies to IL-6, IL-8, IL-10, GM-CSF and TNF-α have been reported in patients with autoimmune diseases or viral infections [16–21]; as in a previous report [19], we also find low levels of binding antibodies to IL-10 in some pemphigoid patients. In addition, we have found binding and occasionally neutralizing autoantibodies to GM-CSF in MG and other autoimmune disorders [20], although not at the high frequency reported for IPAP [26]. Lastly, injection of manufactured cytokines evidently provokes antibodies much more frequently against IFN-α, IFN-β and GM-CSF [47–49] than against IFN-γ and IL-12 [50,51]. Thus we agree that binding antibodies to several cytokines are relatively widespread, although they rarely neutralize significantly in our experience.

Many cytokines − including IFN-α and IL-12 − are induced only transiently during acute or chronic infections, and may never reach levels sufficient to induce tolerance. Possibly, the low levels of autoantibodies against them seen occasionally in the other disease groups, and even in healthy controls [16–25], may result from intercurrent surges in their production, for instance during exacerbations of SLE, or after viral infections. If so, they could be processed distinctively, e.g. by activated dendritic cells [52], as might occur with GM-CSF in the lung in IPAP [26]. There may also be altered processing during apoptosis; motifs for clipping by caspases or granzyme B (activated during apoptosis or cytotoxic reactions) are prevalent in the autoantigens recognized in SLE and such clipping exposes new epitopes [53]. Interestingly, a 30DRHD33 motif for caspase 2 is conserved in most of the IFN-α subtypes and IFN-ω, but not in IFN-β or IFN-γ. Both GM-CSF and IL-12 have two motifs for granzyme B cleavage and two for caspase 8/9; IL-10 has one for granzyme B and two for caspase 2. However, as one such motif is expected per ∼180 residues purely by chance, their significance requires independent support.

In theory, structural similarities among IFN-α, IFN-ω, IL-12 and AChR might be recognized by cross-reactive autoantibodies. Intriguingly, of all known sequences, IFN-α2 bears the highest homology (about 20%) to the AChRα subunit. However: (i) the homology is mainly in its cytoplasmic loop [54], whereas the patients’ AChR antibodies are almost exclusively specific for its extracellular conformation; (ii) the anti-IFN-α2 and -IL-12 antibodies correlate poorly with each other and with anti-AChR levels, being almost undetectable in EOMG+ where anti-AChR titres are highest; (iii) we found no evidence of cross-reactivity either in thymoma cultures (see below) or when we tried to block anti-IFN-α2-containing sera (from LOMG+± thymoma or IFN-α2-treated patients [47]) with a series of recombinant AChR α-subunit polypeptides (A. Meager, unpublished); (iv) because the IFN-α subtypes are heterogeneous and are unrelated to IL-12 in sequence and structure it seems doubly unlikely that both autoantibody populations cross-react with AChR. Nevertheless, cross-reactions by T cells are very hard to exclude.

Novel clues to autoimmunizing mechanisms/cell types

The striking prevalences, titres – and especially the neutralizing activity − of the anti-IFN-α/ω and IL-12 antibodies in MG+ patients ± thymoma suggest some qualitatively distinct autoimmunizing mechanisms/cell types. Conversely, their rarity in the other autoimmune and neoplastic disorders we screened seems highly informative for several reasons: (1) a very early change in the pancreatic islets in IDDM is the increased expression of IFN-α[55]; despite being targets of CD8+ as well as CD4+ T cell attack, these islets clearly do not autoimmunize against IFN-α. (2) Ronnblom et al. [56] have proposed a vicious cycle involving IFN-α in enhancing autoimmunization in SLE, and yet we rarely found neutralizing autoantibodies to IFN-α in such patients (only in 2/40 of SLE and none of 28 RA cases, Meager et al. in preparation). (3) We did not find similar autoantibodies in patients with other tumours, or even in cases with ‘paraneoplastic’ syndromes, implying some distinctive feature in thymomas.

The anticytokine antibodies show even closer associations with thymomas than the anti-AChR antibodies, suggesting selective autoimmunization against IFN-α and/or IL-12 in these tumours. For example, only the anticytokine antibodies increase consistently when the tumours recur, and they are apparently less well controlled by corticosteroids [38]. Moreover, they are produced spontaneously by thymoma cells in culture, having evidently been activated in situ by some cell type(s) expressing these autoantigens in recognizable form (Shiono et al. submitted). By contrast, anti-AChR production in culture requires mitogen stimulation − which fits with the absence of the complete AChR in thymomas and with the notion that helper T cells are primed there only against linear epitotes/isolated subunits [57,58].

The strikingly sharp focus of these neutralizing autoantibodies onto native IFN-α/ω and IL-12, rather than IFN-β and IFN-γ, implicates some strongly immunogenic cell type that naturally produces these cytokines. Whereas IFN-β is derived mainly from fibroblasts and IFN-γ from T cells, large amounts of type I IFN, mainly IFN-α, are derived from virus-stimulated ‘interferon-producing cells’, now known to be synonymous with plasmacytoid’ DC (PDC) [59], a subset found mainly in lymph nodes but also in the human thymus [60], whereas myeloid DC (MDC) particularly produce IL-12, PDC produce type I IFNs, but can also differentiate into strongly IL-12-producing DC [59]. Being highly potent APC for priming naive T and B cells [61], they might also autoimmunize against any endogenous antigens they over-produce or process distinctively, possibly including IFN-α/ω or IL-12 [59,62,63]. Because DC can develop from a common pre-T, pre-NK cell progenitor in the thymus [64] and thymopoeisis is often grossly excessive in thymomas [13–15], DC generation there could also be abnormal. If so, then MDC/PDC also seem likely candidates for maintaining these responses in the periphery; the autoantibodies can clearly persist for years after the thymoma has been removed, even without any sign of recurrence.

Furthermore, PDC/MDC seem a plausible connecting thread between patients with thymoma (TMG+ and TMG) and with LOMG+ in whom the thymus is atrophic and no thymoma can be detected, whether at surgery [28] or after prolonged follow-up [38]. If correct, these ideas may hold also invaluable clues to autoimmunizing mechanisms in LOMG+. Clearly, these must differ in EOMG+ or SNMG+ patients – and even in the LEMS, despite its analogous paraneoplastic and idiopathic subgroups (Table 1) [2,6,7]. In summary, we hypothesize that, because of unknown stimuli or aberrant behaviour, PDC/MDC in thymomas prime helper T cells against linear AChR epitopes and both T and B cells against native IFN-α/ω and IL-12 molecules, possibly a valuable clue to pathogenesis in some LOMG+ cases.


  1. Top of page

We are very grateful to numerous colleagues for kindly providing reagents or samples, also including Drs F. Baggi, S. Berrih-Aknin, A. Batocchi, N. E. Gilhus and M. W. Nicolle, to their laboratory colleagues for locating them, and to Dr M. Bonifati for checking granzyme B motifs. This work was supported by the Sir Jules Thorn Charitable Trust, the Myasthenia Gravis Association/Muscular Dystrophy Campaign and the Medical Rersearch Council.


  1. Top of page
  • 1
    Coutinho A, Kazatchkine MD, Avrameas S. Natural autoantibodies. Curr Opin Immunol 1995; 7: 8128.
  • 2
    Newsom-Davis J. The Hughlings Jackson Lecture: autoimmunity and the nervous system. J Royal Soc Med 1995; 88: 63943.
  • 3
    Kotzin B. Systemic lupus erythematosus. Cell 1996; 85: 3036.
  • 4
    Song YH, Li Y, Maclaren NK. The nature of autoantigens targeted in autoimmune endocrine diseases. Immunol Today 1996; 17: 2328.
  • 5
    De Baets M. Autoimmune diseases against cell surface receptors: myasthenia gravis, a prototype anti-receptor disease. Neth J Med 1994; 45: 294301.
  • 6
    Vincent A, Lang B, Newsom-Davis J. Autoimmunity to the voltage-gated calcium channel underlies the Lambert–Eaton myasthenic syndrome, a paraneoplastic disorder. Trends Neurosci 1989; 12: 496502.
  • 7
    Sher E, Biancardi E, Passafaro M, Clementi F. Physiopathology of neuronal voltage-operated calcium channels. FASEB J 1991; 5: 267783.
  • 8
    Hoch W, McConville J, Helms S et al. Auto-antibodies to the receptor tyrosine kinase MuSK in patients with myasthenia gravis without acetylcholine receptor antibodies. Nature Med 2001; 7: 3658.
  • 9
    Gautel M, Lakey A, Barlow D et al. Titin antibodies in myasthenia gravis: identification of a major immunogenic region of titin. Neurology 1993; 43: 15815.
  • 10
    Ammus SS, Yunis AA. Acquired pure red cell aplasia. Am J Hematol 1987; 24: 31126.
  • 11
    Yip D, Rasko JE, Lee C et al. Thymoma and agranulocytosis: two case reports and literature review. Br J Haematol 1996; 95: 526.
  • 12
    Asherson GL, Webster ADB. Thymoma and immunodeficiency. In: AshersonGL, WebsterADB, eds. Diagnosis and treatment of immunodeficiency. Oxford: Blackwell Publishing, 1980:7898.
  • 13
    Willcox N, Sheppard M, Loehrer P. Thymic tumours and their autoimmune associations. In: SouhamiRL, TannockI, HohenbergerP, HoriotJ-C, eds. Oxford textbook of oncology. Oxford: Oxford University Press, 2002: 2143–52.
  • 14
    Marx A, Müller-Hermelink H-K. From basic immunobiology to the upcoming WHO-classification of tumors of the thymus. Pathol Res Pract 1999; 195: 51533.
  • 15
    Vincent A, Willcox N. The role of T-cells in the initiation of autoantibody responses in thymoma patients. Pathol Res Pract 1999; 195: 53540.
  • 16
    Van Der Meide PH, Schellekens H. Anti-cytokine autoantibodies. epiphenomenon or critical modulators of cytokine action. Biotherapy 1997; 10: 3948.
  • 17
    Amiral J, Marfaing Koka A, Wolf M et al. Presence of autoantibodies to interleukin-8 or neutrophil-activating peptide-2 in patients with heparin-associated thrombocytopenia. Blood 1996; 88: 4106.
  • 18
    Kurdowska A, Miller EJ, Noble JM et al. Anti-IL-8 autoantibodies in alveolar fluid from patients with the adult respiratory distress syndrome. J Immunol 1996; 157: 2699706.
  • 19
    Menetrier-Caux C, Briere F, Jouvenne P et al. Identification of human IgG autoantibodies specific for IL-10. Clin Exp Immunol 1996; 104: 1739.
  • 20
    Meager A, Wadhwa M, Bird C et al. Spontaneously occurring neutralising antibodies against granulocyte-macrophage colony-stimulating factor (GM-CSF) in patients with autoimmune disease. Immunology 1999; 97: 52632.
  • 21
    Sioud M, Dybwad A, Jespersen L et al. Characterization of naturally occurring autoantibodies against tumour necrosis factor-alpha (TNF-alpha): in vitro function and precise epitope mapping by phage epitope library. Clin Exp Immunol 1994; 98: 5205.
  • 22
    Monti E, Pozzi A, Tiberio L et al. Purification of interleukin-2 antibodies from healthy individuals. Immunol Lett 1993; 36: 2616.
  • 23
    Sylvester I, Yoshimura T, Sticherling M et al. Neutrophil attractant protein-1-immunoglobulin G immune complexes and free anti-NAP-1 antibody in normal human serum. J Clin Invest 1992; 90: 47181.
  • 24
    Bendtzen K, Hansen M, Ross C, Svenson M. High avidity autoantibodies to cytokines. Immunol Today 1998; 19: 20911.
  • 25
    Svenson M, Hansen M, Ross C et al. Antibody to granulocyte-macrophage colony-stimulating factor is a dominant anti-cytokine activity in human IgG preparations. Blood 1998; 91: 205461.
  • 26
    Kitamura T, Tanaka N, Watanabe J et al. Idiopathic pulmonary alveolar proteinosis as an autoimmune disease with neutralising antibody against granulocyte/macrophage colony stimulating factor. J Exp Med 1999; 190: 87580.
  • 27
    Meager A, Willcox N, Vincent A, Newsom-Davis J. Spontaneous neutralising antibodies to interferon-alpha and interleukin-12 in thymoma-associated autoimmune disease. Lancet 1997; 350: 15967.
  • 28
    Yamamoto AM, Gajdos P, Eymard B et al. Anti-titin antibodies in myasthenia gravis: tight association with thymoma and heterogeneity of non-thymoma patients. Archs Neurol 2001; 58: 88590.
  • 29
    Däubener W, Wanagat N, Pilz K et al. A new, simple, bioassay for human IFN-γ. J Immunol Meth 1994; 168: 3947.
  • 30
    Meager A. Biological assays for interferons. J Immunol Meth 2002; 261: 2136.
  • 31
    Meager A. Antibodies against interferons: characterization of interferons and immunoassays. In: ClemensMJ, MorrisAG, GearingAJH, eds. Lymphokines and interferons: a practical approach. Oxford: IRL Press, 1987:10527.
  • 32
    Kawade Y. An analysis of neutralization reaction in interferon by antibody: a proposal on the expression of neutralization titre. J Interferon Res 1980; 1: 6170.
  • 33
    Meager A. A cytotoxicity assay for tumour necrosis factor using a human rhabdomyosarcoma cell line. J Immunol Meth 1991; 144: 1413.
  • 34
    Meager A. Assays for transforming growth factor-β. J Immunol Meth 1991; 141: 114.
  • 35
    Kitamura T, Tange T, Terasawa T et al. Establishment and characterization of a unique human cell line that proliferates dependently on GM-CSF, IL-3, or erythropoietin. J Cell Physiol 1989; 140: 3239.
  • 36
    Wadhwa M, Bird C, Dilger P et al. Quantitative biological assays for individual cytokines. In: BalkwillF, ed. Cytokine cell biology: practical approach. Oxford: Oxford University Press, 2000: 20739.
  • 37
    Konishi K, Tanabe F, Taniguchi M et al. A simple and sensitive bioassay for the detection of human interleukin-18/interferon-gamma-inducing factor using human myelomonocytic KG-1 cells. J Immunol Meth 1997; 209: 18791.
  • 38
    Buckley C, Newsom-Davis J, Willcox N, Vincent A. Do titin and cytokine antibodies in MG patients predict thymoma or thymoma recurrence? Neurology 2001; 57: 157982.
  • 39
    Alton K, Stabinsky Y, Richards R et al. Production, characterisation and biological effects of recombinant DNA derived human IFN-α and IFN-γ analogs. In: DeMaeyerE, SchellekensH., eds. The biology of the interferon system. Amsterdam: Elsevier, 1983:11928.
  • 40
    Meager A. Interferons alpha, beta, and omega. In: Mire-SluisAR, ThorpeR, eds. Cytokines. San Diego: Academic Press, 1998:36189.
  • 41
    Adolf GR. Monoclonal antibodies and enzyme immunoassays specific for human interferon (IFN) ω1: evidence that IFN-ω1 is a component of human leukocyte IFN. Virology 1990; 175: 4107.
  • 42
    Prümmer O, Bunjas D, Wiesneth M et al. Antibodies to interferon-α: a novel type of autoantibody occurring after allogeneic bone marrow transplantation. Bone Marrow Transplant 1996; 17: 61723.
  • 43
    Ikeda Y, Toda G, Hashimoto N et al. Naturally occurring anti-interferon-α2a antibodies in patients with acute viral hepatitis. Clin Exp Immunol 1995; 85: 804.
  • 44
    Fall LS, Chams V, Le Coq H et al. Evidence for an antiviral effect and interferon neutralizing capacity in human sera; variability and implications for HIV infection. Cell Mol Biol 1995; 41: 40916.
  • 45
    Wadhwa M, Meager A, Dilger P et al. Neutralising antibodies to granulocyte-macrophage colony-stimulating factor, interleukin-1α and interferon-α but not other cytokines in human immunoglobulin preparations. Immunology 2000; 99: 11323.
  • 46
    Ross C, Svenson M, Hansen M et al. High avidity IFN-neutralising antibodies in pharmaceutically prepared human IgG. J Clin Invest 1995; 95: 19748.
  • 47
    Antonelli G, Currenti M, Turriziani O, Dianzani F. Neutralising antibodies to interferon-alpha: relative frequency in patients treated with different interferon preparations. J Infect Dis 1991; 163: 8825.
  • 48
    Scagnolari C, Bellomi F, Turriziani O et al. Neutralising and binding antibodies to IFN-beta: relative frequency in relapsing–remitting multiple sclerosis patients treated with different IFN-beta preparations. J Interferon Cytokine Res 2002; 22: 20713.
  • 49
    Wadhwa M, Bird C, Fagerberg J et al. Production of neutralising granulocyte-macrophage colony-stimulating factor (GM-CSF) antibodies in carcinoma patients following GM-CSF combination therapy. Clin Exp Immunol 1996; 104: 3518.
  • 50
    Prummer O, Fiehn C, Gallati H. Anti-interferon gamma antibodies in a patient undergoing interferon-gamma treatment for systemic mastocytosis. J Interferon Cytokine Res 1996; 16: 51922.
  • 51
    Atkins MB, Robertson MJ, Gordon M et al. Phase I evaluation of intravenous recombinant interleukin 12 in patients with advanced malignancies. Clin Cancer Res 1997; 3: 40917.
  • 52
    Drakesmith H, Chain B, Beverley P. How dendritic cells cause autoimmune disease. Immunol Today 2000; 21: 2147.
  • 53
    Casciola-Rosen L, Andrade F, Ulanet D et al. Cleavage by granzyme B is strongly predictive of autoantigen status: implications for initiation of autoimmunity. J Exp Med 1999; 190: 81525.
  • 54
    Karlin A, Akabas M. Toward a structural basis for the function of nicotinic acetylcholine receptors and their cousins. Neuron 1995; 15: 123144.
  • 55
    Foulis AK, Farquharson MA, Meager A. Immunoreactive α-interferon in insulin-secreting β cells in type I diabetes mellitus. Lancet 1987; II: 14237.
  • 56
    Ronnblom L, Alm GV. An etiopathogenic role for the type I IFN system in SLE. Trends Immunol 2001; 22: 42731.
  • 57
    Nagvekar N, Moody A-M, Moss P et al. A pathogenetic role for the thymoma in myasthenia gravis: autosensitization of IL-4-producing T cell clones recognizing extracellular acetylcholine receptor epitopes presented by minority class II isotypes. J Clin Invest 1998; 101: 226877.
  • 58
    Willcox N. Myasthenia gravis. Curr Opin Immunol 1993; 5: 9107.
  • 59
    Colonna M, Krug A, Cella M. Interferon-producing cells: on the front line in immune responses against pathogens. Curr Opin Immunol 2002; 14: 3739.
  • 60
    Bendriss-Vermare N, Barthelemy C, Durand I et al. Human thymus contains IFN-α-producing CD11c, myeloid CD11+ and mature interdigitating dendritic cells. J Clin Invest 2001; 107: 83544.
  • 61
    Wykes M, Pombo A, Jenkins C, MacPherson GG. Dendritic cells interact directly with naive B cells to transfer antigen and initiate class switching in a primary T-dependent response. J Immunol 1998; 161: 13139.
  • 62
    Reid SD, Penna G, Adorini L. The control of T cell responses by dendritic cell subsets. Curr Opin Immunol 2000; 12: 11421.
  • 63
    Blanco P, Palucka AK, Gill M et al. Induction of dendritic cell differentiation by IFN-alpha in systemic lupus erythematosus. Science 2001; 294: 15403.
  • 64
    Spits H, Couwenberg F, Bakker AQ et al. Id2 and Id3 inhibit development of CD34+ stem cells into pre-dendritic cells (pre-DC) 2 but not into pre-DC1: evidence for a lymphoid origin of pre-DC2. J Exp Med 2000; 192: 177583.