Article Text
PDF
Abstract
Background: Potential target autoantigens in the demyelinating form of Guillain–Barré syndrome (GBS) include the myelin proteins PMP22, P0 and P2.
Methods: We investigated immunoreactivity to P0, P2 and PMP22 proteins in 37 patients with GBS and 32 healthy controls.
Results: Antibodies to PMP22 or P0 peptides were detected at presentation in only 5 out of 37 patients. In ELISPOT assays, blood mononuclear cells from 15 out of 24 patients with GBS, but none of the control subjects, produced interleukin-10 (IL-10) in response to peptides from proteins P0, P2 or PMP22 (p = 0.0003). The cells from only two patients produced interferon-γ (IFNγ). The cells from 11 patients with GBS had increased IL-10 responses to peptides representing sequences from the extracellular domains of PMP22 before intravenous immunoglobulin (IVIg) treatment (p = 0.006). The cells from 11 patients with GBS, including 7 who responded to the extracellular domains of PMP22, had increased IL-10 responses to the intracellular domain of P0 before (p = 0.005) and those from 9 patients after they had been treated with IVIg (p = 0.01).
Conclusions: Antibodies to P0 and PMP22 protein peptides do occur in GBS but are uncommon. Circulating mononuclear cell IFNγ responses to P0, P2 and PMP22 myelin protein peptides are rare, but IL-10 responses occur significantly more often than in normal subjects. They might be part of a harmful pathogenetic process or represent a regulatory response.
- immune response
- T-cell
- antibody
- myelin protein
- Guillain–Barré syndrome
Statistics from Altmetric.com
Guillain–Barré syndrome (GBS) is an acute monophasic disease with at least three major forms. The acute motor and motor and sensory axonal forms are associated with anti-ganglioside antibodies arising as part of an immune response against bacterial lipo-oligosaccharide-carrying terminal sugar sequences that are identical to the relevant ganglioside.1 The mechanism of presumed autoimmunity in the common acute inflammatory demyelinating polyradiculoneuropathy (AIDP) form is controversial but endoneurial lesions are prominently infiltrated by T cells.2 The target of the T-cell response is unknown but its animal model—experimental autoimmune neuritis3 4—can be induced by immunisation with P0, P2 or PMP22 myelin proteins or by transfer of CD4+ T cells directed against P0 or P2 peptides.5–10 These myelin proteins are therefore the principal candidate protein autoantigens for AIDP. Previous investigations of immunity to these proteins in GBS have given conflicting results.11–16 We have therefore re-examined the question of the role of both humoral and cellular immune responses to human myelin proteins with panels of overlapping peptides covering almost the entire sequence of all three candidate autoantigens in a new cohort of patients with GBS and control subjects.
SUBJECTS AND METHODS
Patients
Patients fulfilling international criteria for GBS17 or its variants were recruited from hospitals throughout the South East of England and assessed on the GBS disability scale from 0 (normal) to 5 (requirement for artificial ventilation).18 The electrophysiology was classified as demyelinating, axonal, equivocal, normal or inexcitable.19 Normal controls were recruited from non-blood relatives of patients and medical school staff. Ethics committee approval was obtained from all the institutions involved and participants gave informed consent.
Blood samples
Blood samples were taken before and after intravenous immunoglobulin (IVIg). Peripheral blood mononuclear cells (PBMCs) were prepared with Lymphoprep and frozen in RPMI-1640 with 0.1% 2-Mercaptoethanol, 10% dimethylsulfoxide and 20% fetal calf serum. For cytokine analysis, 8 × 106 cells/ml were resuspended in RPMI-1640 containing 10% human AB serum.
Antigens
Extracellular domains of PMP2
The first extracellular domain of human PMP22 (ECD1-PMP22) (amino acids 26–72, fig 1) was cloned from vagus nerve obtained during vagotomy, expressed as a glutathione-S-transferase fusion protein (GST–ECD1-PMP22) in Escherichia coli and purified by affinity chromatography. The second extracellular domain of human PMP22 (ECD2-PMP22) was synthesized with an extra N-terminal cysteine residue (amino acids 119–123, fig 1) by Pepceuticals Ltd (Leeds, UK) and conjugated to bovine serum albumin (BSA) using N-maleimido-6-aminocaproyl- (2′-nitro, 4′-sulfonic acid)-phenyl ester (Bachem, St. Helens, UK).
Extracellular domain of P0 (ECD-P0)
The extracellular domain of P0 (ECD-P0, amino acids 29–153, fig 1) was cloned and expressed as a maltose binding protein fusion protein (MBP–ECD-P0) in E. coli.
ELISPOT synthetic peptides
For ELISPOT assays, we used a panel of synthetic overlapping peptides of the peripheral nerve myelin proteins PMP22, P2 and P0 (fig 1) produced by Genosphere Biotechnologies (Paris, France). Most peptides had 25 or 26 amino acids, as shown in figure 1. For ELISPOT analysis, peptides were resuspended in sterile DMSO at 50 mg/ml.
Peptides were pooled in two groups for each myelin protein. PMP22 peptides were divided into panel I (peptides 1, 2 and 3) representing ECD1-PMP22 and part of the transmembrane region of the protein, and panel II (peptides 4, 5 and 6) representing the rest of the transmembrane region and ECD2-PMP22. The P0 peptides were divided into panel I (peptides 1, 2, 3 and 4) corresponding to the majority of ECD-P0, and panel II (peptides 6, 7, 8 and 9) corresponding to the proximal part of the extracellular domain, cytoplasmic tail and transmembrane region. P2 peptides were divided into panel I, consisting of peptides 1, 2, 3 and 4, and panel II, consisting of peptides 5, 6 and 7.
Antibody assays
ELISA against PMP22 and P0 fusion proteins
Each serum sample was tested in triplicate. Commercial antibodies to the fusion proteins (GST and MBP) were positive controls. The antigen solutions were applied at 10 μg/ml in carbonate coating buffer to ELISA plates (Thermolabsystems, Franklin, USA) at 50 μl/well for 2 hours at room temperature. Plates were washed, blocked with 1% fish gelatin and incubated overnight at 4°C. Following this, 50 μl of 1:50 human serum was added to each well and incubated at 37°C for 1 hour. After washing, bound antibody was detected with rabbit anti-human immunoglobulin IgG or IgM polyclonal secondary antibody conjugated to alkaline phosphatase (Dako, Ely, UK) and developed with p-nitrophenylphosphate. Optical density was measured at 405 nm and results expressed as the amount of IgG or IgM bound to the antigen of interest. The amount of bound IgG or IgM was calculated from known amounts of purified IgG or IgM on each plate.
ELISA against Campylobacter jejuni extracts
The GBS sera were tested for anti-Campylobacter jejuni antibodies by ELISA. Responses in two classes of IgG, IgM or IgA were considered to indicate a recent C. jejuni infection.20 21
Thin-layer chromatography with immuno-overlay
The GBS sera were tested for IgG and IgM anti-ganglioside antibodies (GM1, GQ1b, GM2) by thin-layer chromatography with immuno-overlay using a mixture of the gangliosides on aluminium-backed thin-layer chromatography plates.
ELISPOT assays
Thawed PBMCs were dispensed in 48-well plates at 2 × 106 in 0.5 ml RPMI-1640 supplemented with 10% human AB serum. Peptides from each myelin protein were added at a final concentration of 10 μg/ml for each peptide and incubated at 37°C in 5% CO2. Control wells contained RPMI-1640 with 0.1% DMSO. Tetanus toxoid (TT) at a final concentration of 100 ng/ml was the positive control. After 48 hours, non-adherent T cells were harvested, resuspended in RPMI-1640 containing 2% AB serum, washed and brought to a concentration of 106/300 μl. Aliquots of 100 μl of cell suspension were dispensed into 96-well ELISA plates (Nunc Maxisorp; Merck Ltd., Poole, UK) pre-coated with monoclonal anti-IFNγ or anti-IL-10 capture antibody (U-Cytech, Utrecht, The Netherlands) and pre-blocked with 1% BSA in PBS. After incubation at 37°C in 5% CO2 overnight, cells were lysed in ice-cold water; plates were washed in PBST 0.05%, and spots developed according to the manufacturer’s instructions. Spots of 100–120 μm diameter were counted in a BioReader 3000 (BioSys, Karben, Germany) by a blinded observer. The mean number of spots per 3 × 105 cells in triplicate assays was calculated. The mean values of wells with DMSO alone were subtracted from the mean values in test wells. Samples with values >3 SD above the control mean were considered to be positive.
Statistical analyses
We used GraphPad Prism version 4.00 software (GraphPad Software Inc., San Diego, CA, USA) for statistical calculations. The Mann–Whitney or the Wilcoxon matched pairs tests were used to compare distributions and Fisher’s exact test or chi-square test for proportions. Correlations between clinical and laboratory data were sought with the Spearman test. Two-tailed tests of significance have been reported and p<0.05 was considered to be significant.
RESULTS
Subjects
Thirty-seven patients participated in the study (table 1): 11 women and 26 men, mean (SD) age 49.9 (18.0) years. Thirty-two patients fulfilled Asbury criteria for GBS and five had formes frustes, including variants of the Fisher syndrome. Only two patients (6 and 25) had serological evidence of recent C. jejuni infection. There were 32 healthy controls: 6 women and 26 men, mean age 41.8 (18.5) years.
Antibodies
The distribution of IgM and IgG antibodies to ECD1-PMP22 (fig 2a) did not differ significantly between patients and controls. One healthy control serum exhibited high concentrations of IgG (>20 μg/ml) to ECD1-PMP22. At initial presentation, only two of 37 patients with GBS (numbers 9 and 17) and no controls exhibited IgG anti-ECD2-PMP22 concentrations >2 μg/ml (fig 2b). A third patient with GBS (number 31) had an anti-ECD2-PMP22 IgG concentration >2 μg/ml in his second sample. Three of 37 patients with GBS (numbers 10, 17 and 21) exhibited IgM anti-ECD2-PMP22 >2.5 μg/ml levels, which were consistently above the levels in controls (fig 2c). However, the serum of patient 17 contained rheumatoid factor, which may have caused a spuriously high IgM result.
One GBS serum (number 30 in table 1) had IgG anti-ECD-P0 antibodies >5 μg/ml (fig 2D). The GBS serum (number 17) with rheumatoid factor was the only one that had increased anti-ECD-P0 IgM responses.
Cellular responses to PMP22, P2 and P0
ELISPOTs were performed on the mononuclear cells from 24 patients with GBS: 17 men and 7 women, mean (SD) age 51 (20) years. We analysed 22 samples taken before and after administration of IVIg and two samples (patient 30 and 31) taken only before treatment (see table 1). There were 12 healthy controls: 6 men and 6 women, mean age 46 (20) years.
Interferon-γ responses to PMP22, P2 and P0
Most of the samples showed no or very low IFNγ responses to the tested peptides. Only two patients (number 8 positive for P2 II and 36 positive for P0 II) and one control (positive for PMP22 II) produced IFNγ above the cut-off value at initial sampling. Patients 4 and 37 responded to P0 II only after IVIg.
PMP22
We found a significantly increased frequency of IL-10-producing cells against PMP22 I in patients at initial sampling (mean (SD) number of spots 6.0 (7.9) per 3 × 105 cells) compared with healthy controls (0.9 (1.4) per 3 × 105 cells in the controls) (p = 0.02) (fig 3a), but not in patients after IVIg treatment (3.1 (5.7) spots per 3 × 105 cells). Eleven out of 24 (45.8%) patients and no control subjects showed positive IL-10 responses to PMP22 I at initial sampling (p = 0.006) and only 3 out of 22 (13.6%) at second sampling (not significant).
Three patients secreted IL-10 above the levels of healthy controls after stimulation with PMP22 II, in two cases before treatment (patients 11 and 19) and in one only after IVIg (patient 15).
P0
Three patients (4, 11 and 16) showed IL-10 responses to P0 I before treatment, and three patients, two of them different, after IVIg (4, 8 and 27). IL-10 responses to P0 II were more frequent. The total number of spots in patients was increased at initial sampling (mean (SD) 13.9 (14.7) per 3 × 105 cells compared with controls 1.6 (1.7) per 3 × 105 cells, p = 0.01), and persisted after IVIg (17.3 (22.1) spots per 3 × 105 cells, p = 0.06 (fig 3b). Eleven out of 21 (52.4%) patients and no control subjects showed positive IL-10 responses against P0 II at initial sampling (p = 0.005) and 9 of 19 patients (47.4%) at second sampling (p = 0.01), two of which were for the first time. Seven (4, 8, 16, 27, 28, 31 and 37) of the 11 patients who responded to P0 II at initial sampling also responded to PMP22 I.
P2
Patients and controls did not differ in their IL-10 responses to either of two sets of P2 peptides. Only patient 28 had positive IL-10 tests, both before and after treatment (fig 3b).
Correlations between ELISPOT responses to PMP22, P2 and P0
We compared the total number of spots within the whole group of patients for each of the sets of peptides. Increased IL-10 responses to PMP22 I before IVIg showed strong correlations with increased IL-10 responses to PMP22 II (p = 0.02) and P0 I (p = 0.002). After IVIg, there were significant correlations between responses to PMP22 I and PMP22 II (p = 0.01), PMP22 I and P0 I (p = 0.001), P0 II and PMP22 II (p = 0.01), and P0 II and P0 I (p = 0.04). Thus, most positive patients responded to a broad spectrum of myelin proteins rather than one specific peptide.
Correlations between antibody and ELISPOT results
There were no correlations between antibody and the ELISPOT responses against similar regions of myelin proteins (tables 1 and 2). Two patients with GBS with increased IL-10 responses to PMP22 I did not have antibodies to ECD1-PMP22 but did produce either IgM (patient 10) or IgG (patient 31) against ECD2-PMP22. Only one patient (30), whose cells produced significant levels of IL-10 in response to P0 II but none to P0 I, had antibodies in their serum against ECD-P0.
Fifteen of 37 patients with GBS had anti-ganglioside antibodies, 14 had antibodies to GM1, one to GM2 and two to GM1 and GQ1b. There was also no correlation between the presence of antibodies or IL-10 responses to myelin proteins and antibodies to gangliosides (tables 1 and 2).
Correlation with clinical features
There was no difference in the presenting clinical features of patients with GBS with or without anti-ECD2-PMP22 antibodies or IL-10 responses to PMP22 I or P0 II (table 2). Two of the three GBS patients with IgG anti-ECD2-PMP22 antibodies and none of those without experienced a clinical relapse within 1 year of their initial presentation (p = 0.009).
DISCUSSION
Our study analysed the immune responses to a more comprehensive range of myelin protein peptides than has been done before. The ELISA studies showed that IgG or IgM autoantibodies to non-glycosylated peptides from the ECD2-PMP22, but not ECD1-PMP22, were present in only 4 out of 37 (10.8%) patients with GBS at presentation. Previous studies showed different results but used different methodology, not quantifying the amount of IgG and IgM.11 14 Our finding of IgG autoantibodies to non-glycosylated ECD-P0 by ELISA in only 1 of 37 patients with GBS is similar to the results from two western blot studies.13 15 In a recent ELISA study, there was no significant increase in IgG antibodies to peptides corresponding to P0 residues 56–71, 70–85 and 180–199 in 48 patients with GBS compared with 38 healthy controls.16 In the same study, IgG antibodies to P2 residue 14–25 were present in 13/30 patients with GBS at the peak of the disease compared with 2/38 healthy controls (p = 0.002). However, there was no increase in frequency of antibodies to P2 residues 61–70, the minimum component of P2 protein capable of inducing experimental autoimmune neuritis in the rat.16 22 23 As P2 is a cytoplasmic protein that is not known to be exposed at the cell surface, antibodies to it are unlikely to initiate disease and we did not study them.
We did not detect any relationship between the presence of autoantibodies to PMP22 or P0 and particular clinical features, except for an unexpected but significant association between IgG antibodies to ECD2-PMP22 and later relapse. It would be premature to draw conclusions, but a search for a relationship between these antibodies and later relapse and the development of chronic inflammatory demyelinating polyradiculoneuropathy would be worthwhile.
Our study is limited by the absence of controls with other neuropathies, inflammatory neurological conditions or post-infectious conditions, which would enable us to assess the specificity of the responses detected for GBS rather than secondary to nerve damage or the immune response to antecedent infection. All studies conducted so far, including ours, are limited by the difficulty of presenting antigenic epitopes in their native configuration. In our study, the use of recombinant peptides of P0 and PMP22 fused to MBP and GST, respectively, may have altered the conformation and antigenicity of the myelin peptides, particularly when studying antibody responses. Although we successfully expressed whole human PMP22 tagged to a “FLAG” tag in mammalian cells, we were not able to purify this for use in ELISA studies.
Our study suggests that antibodies directed against myelin proteins are likely to be directly involved in demyelination in only a small proportion of GBS cases and it seems difficult to escape the conclusion that the search should now be directed towards other non-peptide autoantigens. The presence of antibodies to gangliosides in 15 of our 37 patients is a possible pointer in this direction, despite the fact that none have so far been linked convincingly to the AIDP form of GBS.1 2
This is the first demonstration of a predominant IL-10 response in the absence of IFNγ responses to myelin proteins in GBS. Thus, 62.5% of patients with GBS showed IL-10 responses to one of two distinct parts of PMP22 or P0 with an almost complete absence of IFNγ responses. Our detection of strong IL-10 responses to the P0 II peptide panel—corresponding to the proximal part of ECD-P0 and the intracellular domain of P0—agrees with previous studies showing increased IL-4 and TGFβ, but not IFNγ responses to P0 peptides, in acute GBS.24 25 In accordance with our results, Csurhes et al. found an absence of IFNγ responses to three PMP22 peptides, which overlapped with our peptides 2 and 5, in GBS.26
We did not unequivocally identify the cell type responsible for the observed increase in IL-10 production. The experimental design, harvesting non-adherent cells and thus reducing the monocyte population, was expected to detect T-cell and predominantly CD4+-cell stimulation. However, type 1 regulatory T cells, Th2-like CD8+ T cells and B cells can also secrete IL-10.27 28 More work will be needed to determine the main source of IL-10 production in GBS.
The observed increase in IL-10 production might indicate an immunoregulatory or inflammatory response. In the early stages of immune response, its secretion is coordinated by increased expression of proinflammatory molecules involved in the innate immune response.29 IL-10 can accelerate the course of some autoimmune diseases,28 such as systemic lupus erythomatosus.30 It enhances immunoglobulin production by naïve and committed B cells and acts as a switch factor for IgG1, IgG3 and IgA production.27 Nevertheless, we could not detect any correlation between increased IL-10 responses to myelin proteins and the presence of autoantibodies against myelin proteins or gangliosides.
The inhibitory role of IL-10 during the adaptive immune response is well documented.31 Increased IL-10 production in patients with GBS may be due to the timing of sampling relative to the generation of the immune response. Patients with GBS have usually had an infection before developing neurological symptoms, so detection of a pro-inflammatory antigen-specific response might be difficult because the anti-bacterial and possible anti-neural immune responses may have already taken place and regulatory mechanisms may have already been triggered.
This study of autoimmune responses to the principal candidate protein antigens indicates that GBS is associated with circulating mononuclear cell IL-10 responses and occasionally antibodies against PMP22 and P0. The almost exclusive generation of IL-10 rather than IFNγ responses suggests that Th2 rather than Th1 cells are involved. There was no correlation between the IL-10 responses to these proteins and antibodies against either these proteins or gangliosides so that the cellular responses arise independently of the humoral responses. Whether these IL-10 responses are pathogenic or regulate and terminate the immune response requires further study.
We thank Dr Gaiping Zhang for conjugating the ECD-PMP22 to BSA and Mr Ian Gray for analysis of anti-C. jejuni antibodies, Action Medical Research, Medical Research Council and Guarantors of Brain for financial support, and our patients with GBS and healthy controls for participating in the study.
REFERENCES
Footnotes
Competing interests: None.