Hypogalactosylation of serum IgG in patients with coeliac disease

Abstract

Coeliac disease (CD) is described as an autoimmune enteropathy associated with the presence of IgG and IgA antigliadin and antitransglutaminase autoantibodies. While of diagnostic significance, the role of these autoantibodies in the immunopathogenesis of CD is elucidated. An inappropriate T cell immune response to gluten is also involved in the pathogenesis of CD, as evidenced by autoantibody switching. The N-glycans released from serum IgG of CD patients and three groups of healthy controls, of differing age ranges, were analysed by NH2-high performance liquid chromatography (HPLC). The fucosylated biantennary N- glycans were the most abundant neutral oligosaccharides; in particular, the agalacto form (G0F) showed a mean value of 42% (s.d. ± 7·4), 30% (s.d. ± 5·9), 26% (s.d. ± 4·2) and 35% (s.d. ± 6·8) for CD patients, healthy children, healthy adults under 40 and healthy adults over 40 years old, respectively. The ratio of asialo agalacto fucosylated biantenna to asialo monogalacto fucosylated biantenna (G0F)/(G1F) for CD patients showed a significant increase compared to healthy children (P < 0·0002), healthy adults under 40 (P < 0·0002) and healthy adults over 40 years old (P < 0·01). Hypogalactosylation was more pronounced for CD patients than for the patients with other autoimmune diseases such as rheumatoid arthritis or psoriatic arthritis.

INTRODUCTION

Coeliac disease (CD) is an autoimmune enteropathy that affects genetically predisposed individuals. The disease is characterized by damage to intestinal mucosa in response to the ingestion of wheat gluten or related proteins from rye and barley [1–3]. The damage results from an immunological reaction to gluten that leads to a flattening of the intestinal epithelium. Typical gastrointestinal symptoms include diarrhoea, abdominal distension, weight loss and failure to thrive. CD is associated with human leucocyte antigen (HLA) alleles [4], and the strong genetic influence to susceptibility is illustrated by a prevalence rate (8–18%) among first-degree relatives of probands and the high concordance rate (approximately 70%) among homozygotic twins [2,4]. The incidence of the disease has been reported to be higher in patients with other autoimmune diseases, such as type 1 diabetes, juvenile chronic arthritis, autoimmune hepatitis and autoimmune thyroid disease [5–8].

It is accepted that both humoral and cellular immune responses are involved in the pathogenesis of coeliac disease. An inappropriate T cell mediated response against ingested gluten [9,10] together with the presence antigliadin antibodies is demonstrable, which is strictly dependent on dietary exposure to gluten [2,10]. The presence of antitissue transglutaminase (tTG) antibodies is also demonstrable and may be a definitive diagnostic indicator for CD [11].

One of the first events in the pathogenesis is inflammation of the intestinal mucosa [2]. Small intestine lesions in CD are characterized by lymphocyte infiltration of the epithelium and increased density of various leucocytes in the lamina propria. A further characteristic of the CD lesion is an accumulation of IgA-, IgM- and IgG-producing plasma cells [12]. Although the specificities of the antibodies produced by these cells have been characterized only partially, in vitro culture of biopsies has demonstrated the presence of antibodies to gliadin and transglutaminase [13,14].

New epitopes are expressed in the subendothelium of intestinal epithelia following the deposition of tTG–gliadin immune complexes to molecules of the extracellular matrix [10]. Thus, antigliadin and antitransglutaminase antibodies are associated directly with the pathology of coeliac disease. The antibodies can be of the IgG or IgA class; however, it is generally accepted that the presence of IgA antibodies is the more specific diagnostic feature of CD; the presence of IgG antibodies is a more sensitive test. The use of combined tests for antigliadin and antitransglutaminase antibodies has been shown to be highly sensitive and specific for diagnosis.

Changes in the N-glycan profile of polyclonal IgG isolated from serum of patients with certain inflammatory and autoimmune diseases, relative to normal individuals, have been reported [15–18], e.g. rheumatoid arthritis (RA), systemic lupus erythematosis (SLE), ankylosing spondylitis (AS), juvenile chronic arthritis (JCA), tuberculosis (TB) Crohn's disease and psoriatic arthritis (PsA), among others. Oligosaccharide analyses revealed a disease related glycosylation patterns with RA (P < 0·0001) and JCA (P < 0·006) patients having predominantly agalactosyl structures, while SLE (P < 0·03–0·0001) and AS (P < 0·025–0·0001) patients exhibited predominantly digalactosyl structures [18].

The human IgG molecule has a conserved N-linked glycosylation site at Asn297 in each of the Cγ2 domains of the Fc region. The attached oligosaccharide is of the complex biantennary type comprised of a core heptasaccharide GlcNAc₂Man₃GlcNAc₂. Variable attachment of outer arm sugars (bisecting N-acetylglucosamine, fucose, galactose and sialic acid) allows for the generation of approximately 30 different oligosaccharide structures [19]. Given random pairing of differentially glycosylated heavy chains there is potential for the generation of > 400 distinct glycoforms of IgG. These glycoforms can differ in their efficacy of effector function activation [20,21].

The IgG molecule is divalent for antigen binding, via the two Fab regions, and is able therefore to form large three-dimensional immune complexes. These complexes present ‘aggregated’ (multivalent) IgG–Fc regions to ligands that activate effector functions. Effector mechanisms mediated through FcγRI, FcγRII and FcγRIII are dependent strictly on IgG–Fc glycosylation. It has been shown that IgG expressing predominantly terminal GlcNAc sugar residues (agalactosyl or G0F–IgG) can activate the complement cascade through mannose binding lectin (MBL) [21]. Age- and pregnancy-related changes in galactosylation levels have also been reported [22–24].

When compared to healthy controls, patients with RA show a significant increase in the level of G0F–IgG; it has been suggested that the G0F value can be an indicator of disease severity and has value for early diagnosis [25]. The fact that G0–IgG (and also IgA: Roos et al. 2001 [26]) is able to activate MBL suggests that immune complexes formed from this glycoform could influence immunoregulatory processes, for self- as well as foreign antigens, following uptake by antigen presenting cells [27]. Recently, Holland et al. [28] reported a profound disregulation of IgG galactosylation for IgG isolated from the sera of patients with acute Wegener's granulomatosis (WG) or microscopic polyanginiitis (MPA). These diseases are associated with the presence of IgG antibodies to neutrophil cytoplasmic antigens (ANCA: antineutrophil cytoplasm antibodies) normally present in neutrophil azurophilic granules. Thus, the presence of G0F–IgG ANCA could both precipitate an inflammatory reaction and potentiate further the generation of auto-antibodies. We have analysed the oligosaccharide profiles of IgG isolated from the sera of patients with another inflammatory, autoimmune condition, namely coeliac disease. We demonstrate a predominance of the G0F–IgG glycoform present in sera having a positive antigliadin and antitransglutaminase antibody titre, compared to three sets of healthy controls, i.e. children under 12 and adults under or over 40 years old.

MATERIALS AND METHODS

Materials

Electrophoresis reagents (electrophoresis grade) were from Bio-Rad Inc. (Richmond, CA, USA). Recombinant PNGase F was from New England Bio Laboratories (Beverly, MA, USA). Acetonitrile (Lichrosolv) and glacial acetic acid were from Riedel-de-Haem (Germany). Nucleosil 5 NH₂-HPLC column was purchased from Supelco (Bellefonta, PA, USA). All other reagents were of analytical grade.

Sample selection

The sera from children, aged 1–12 years, having coeliac disease but no symptoms of other autoimmune disease were collected. All were untreated patients (no drugs but no gluten-free diet), diagnosed following the revised criteria of the European Society of Pediatric Gastroenterology and Nutrition. The clinical features are shown in Table 1. Two sets of samples of autoimmune diseases (RA and PsA) analysed previously in our laboratory were included as positive controls for IgG glycosylation profile changes. Three sets of samples of healthy individuals with age ranges of 1–12 years, adults below 40 and adults above 40 years old were also analysed. Serum samples were obtained from patients and controls after informed consent.

Table 1.

General data from patients of coeliac disease. All were patients diagnosed following the revised criteria of the European Society of Pediatric Gastroenterology and Nutrition [10] without any drug treatment but no gluten-free diet. No other autoimmune diseases were present

Sample	Age (years)	Sex	Antigliadin	Anti-transglutaminase	Biopsy
2	2	M	Positive	Positive	Subtotal villous atrophy
5	2	M	Positive	Positive	Subtotal villous atrophy
9	6	M	Positive	Positive	Total villous atrophy
11	12	F	Positive	Positive	Total villous atrophy
14	3	F	Positive	Positive	Subtotal villous atrophy
21	7	F	Positive	Positive	Subtotal villous atrophy
58	1·5	F	Positive	Positive	Subtotal villous atrophy
70	1·5	M	Positive	Positive	Subtotal villous atrophy
83	1·2	F	Positive	Positive	Subtotal villous atrophy
103	1·8	F	Positive	Positive	Subtotal villous atrophy
104	1·3	F	Positive	Positive	Subtotal villous atrophy
106	2	F	Positive	Positive	Subtotal villous atrophy

Antigliadin and antitransglutaminase determinations

The method used to detect antibodies to gliadin was as described elsewhere [29]. Briefly, white opaque polystyrene microwell strips coated with wheat gliadin (Sigma) were incubated with a dilution 1 : 50 of the serum samples for 20 min. After washing with phosphate buffered saline (PBS)-0·05% Tween 20, the wells were incubated with a protein A-gold conjugated for 10 min. After washing, the reaction was amplified with a silver ion solution, prepared as recommended by the supplier (Amersham Rainham, Essex, UK). The reaction was terminated with distilled water. The resulting colour in each well was compared with that for a control sample used as a threshold in the assay, and each sample classified consequently as negative or positive.

The method used for antibodies to transglutaminase was a one-step immunochromatographic assay described recently [30]. Briefly, a nitrocellulose strip was coated with guinea pig transglutaminase (Sigma) to form a reactive zone. A substance that binds tissue transglutaminase (tTG)–colloidal gold conjugate was absorbed onto the same membrane to form a control zone adjacent to the reactive zone. tTG–colloidal gold conjugate was dried onto an inert fibrous support, which was attached to the plastic backing of the nitrocellulose strip to achieve minimum direct contact with the beginning of the nitrocellulose, close to the reactive zone. In this system, when the conjugated support is dipped in serum or plasma any antibodies to tTG in the sample react with the tTG–colloidal gold conjugate, developing an immune complex that migrates through the membrane strip. The immobilized tTG in the nitrocellulose reacts with the immune complexes, forming a coloured dot in the reactive zone. Excess conjugated and immune complexes react with the control reagent, forming a second coloured dot on the strip. In this assay a positive result, indicating the presence of antibodies to tTG in the samples, is seen as two dots on the strip, and a negative assay shows only the control dot.

IgG purification

IgG purification was performed according to Sumar (1990) [31]. Briefly, 3 ml of Sephadex G25 (Pharmacia, Uppsala, Sweden) were layered onto 0·5 ml of anion exchanger DE52 diethylaminoethyl cellulose (DEAE-cellulose). The column was equilibrated with 10 mm phosphate buffer pH 7·0. Then, 100 µl serum were mixed with an equal volume of 10 mm phosphate buffer, pH 7·0, and applied to the column. The column was washed with 10 mm phosphate buffer pH 7·0 and unbound fractions (containing IgG) were collected. Purity of IgG-containing fractions was determined by sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) with Coomassie staining.

N-glycans release

Proteins were denatured at 70°C for 10 min in 0·1% SDS and 5%β-mercaptoethanol and were cooled to room temperature. Nonidet P-40 detergent (NP-40) was added to a final concentration of 1% before PNGase F addition. PNGase F digests were carried out in the same elution buffer used for IgG purification, at a ratio of 5 U/mg of glycoprotein, and incubated for 16 h at 37°C. After deglycosylation the proteins were precipitated by addition of 3 volumes of cold ethanol and standing for 30 min at − 20°C followed by centrifugation at 3000 g for 10 min. When required, the protein pellet was redissolved in water and then reprecipitated as described above.

Fluorophore labelling of oligosaccharides

Lyophilized oligosaccharides (up to 100 µg) were labelled with 4-ABA as described by Yuen et al. [32]. Briefly, oligosaccharides were added with 4-ABA solution (15 µl; 2 m in 7 : 3 v/v dimethyl sulphoxide/acetic acid) and tetrabutylammonium cyanoborohydride solution (15 µl; 1·5 m, prepared freshly in 7 : 3 v/v dimethyl sulphoxide/acetic acid). After heating at 60°C for 4 h, distilled water (200 µl) was added and excess reagents were removed by extraction five times with ethyl acetate (300 µl).

HPLC running conditions

A Lachrom HPLC system from Merck-Hitachi (Germany) with L7480 flourescence detector was used. HPLC separations were carried on a Nucleosil 5 NH₂ column (4·5 × 250 mm) using a C₁₈ cartridge precolumn: buffer (A): glacial acetic acid (1% v/v) in 70 : 30 (v/v) mixture of acetonitrile-water adjusted to pH 5·5 with triethylamine; buffer (B): glacial acetic acid (1% v/v) adjusted to pH 5·5 with triethylamine; gradient: 5–30% B in 90 min at a flow rate 0·7 ml/min. Fluorescence was measured at λexc 290 nm and λ em 355 nm [32].

Statistical analysis

To perform the statistical analysis a search for a suitable variable to maximize the differences between groups of individuals showing differences in glycosylation profiles was performed. The groups corresponded to children under 12, adults under 40, adults over 40, patients with psoriatic arthritis and patients with rheumatoid arthritis. Additional requirements for variable selection were homocedasticity and normality. Homocedasticity was checked using the Bartlett test and normality using the Komolgorov–Smirnov test. The following variables were studied: G0F, G1F, G2F, G0F/G1F, G0F/G2F, G0F/(G1F + G2F), Ln(G0F), Ln(G1F), Ln(G2F), Ln(G0F/G1F), Ln(G0F/G2F) and Ln[G0F/(G1F + G2F)]. The variable selected was Ln(G0F/G1F) because it maximizes the differences between the groups while showing homocedasticity and normality. The G2F variable has a little variation between groups compared with the within-groups variance; for this reason, its use results in a loss of between-groups resolution power.

To investigate whether a significant difference between groups exist we used an anova test. In order to avoid false significant differences between pairs of groups, post-hoc comparisons were carried out using Tukey's honest significant differences (HDS) test, because it is recognized as conservative.

RESULTS

Patients and negative controls

All CD samples were obtained from children aged 1·2–12 years who were positive for the presence of antigliadin [29] and antitransglutaminase [30] antibodies and confirmed by biopsy, as recommended by the revised criteria of the European Society of Pediatric Gastroenterology and Nutrition [33] (see Table 1). Three groups of negative controls were also studied: healthy children of the same age range and healthy adults under and over 40 years old. As positive control for autoimmune diseases two groups were also included, one of rheumatoid arthritis and the other of psoriatic arthritis.

IgG N-glycan analysis

The predominant oligosaccharide released from native IgG from healthy controls was the monogalactosylated core fucosylated biantenna (G1F), as shown in Table 2. The N-glycans of IgG from sera of patients of coeliac disease showed little, if any, sialylated species (5–10%). The oligosaccharide profiles on NH2-HPLC are shown in Fig. 1a,b. The assignment of N-glycan structure was determined by comparison with the data reported by Yuen et al. [32]. The structures of the most abundant N-glycans, agalacto fucosylated biantenna (G0F), monogalactosylated fucosylated biantenna (G1F) and digalactosylated fucosylated biantenna (G2F), are represented in Fig. 1c.

Table 2.

Analysis of IgG neutral N-glycan. Serum IgG of children healthy controls confirmed as negative for CD and any other autoimmune disease (C); adults under 40 years old (Y); adults over 40 years old (A) and CD patients (P) without any drug treatment but no gluten-free diet; all the patients were positive to antigliadin and antitransglutaminase tests. Mean and standard deviation was calculated for each control group and CD patients

	% G0F	% G1F	% G2F
Healthy children
1C	33	41	26
2C	27	47	26
3C	44	41	15
5C	29	51	20
6C	29	49	21
8C	24	42	34
9C	33	48	19
10C	26	48	26
11C	28	50	22
Mean	30	46	26
s.d.	5·9	3·9	5·5
Healthy adults under 40
1Y	22	52	26
2Y	22	47	31
3Y	30	51	19
4Y	32	50	18
5Y	19	47	34
6Y	29	39	32
7Y	23	41	36
8Y	26	41	33
9Y	24	41	33
10Y	29	47	24
Mean	26	46	29
s.d.	4·2	4·5	6·4
Healthy adults over 40
2 A	36	45	19
3 A	30	44	26
4 A	30	46	24
5 A	30	45	25
6 A	29	46	25
7 A	31	43	26
8 A	32	45	23
9 A	37	43	20
10 A	46	40	14
12 A	49	37	14
13 A	38	41	21
Mean	35	43	22
s.d.	6·8	2·8	4·4
CD patients
2P	54	36	10
5P	30	38	32
9P	42	41	17
11P	37	35	28
14P	37	36	27
21P	38	35	27
58P	44	36	20
70P	34	42	24
83P	54	34	12
103P	41	41	18
104P	48	35	17
106P	42	42	17
Mean	42	38	21
s.d.	7·4	3·1	6·8

Fig. 1.

NH2-HPLC separation of oligosaccharide pools after enzymatic deglycosylation and flourophore labelling of serum IgG from: (a) healthy child as negative control (sample 8C); (b) coeliac disease patient (sample 58P); (c) the most abundant neutral N-glycans structures of human IgG serum. MonoS: monosialylated structures.

All the young healthy controls showed typical N-glycan profiles in which G1F (46%) is the major species, followed by similar proportions of G0F (30%) and G2F (23%) oligosaccharides. By contrast, 10 of 12 children with CD (Tables 2, P samples) showed altered profiles with an increased proportion of the G0F glycan (mean value 42%); G1F and G2F values were 38% and 20%, respectively.

The mathematical expression Ln(G0F/G1F) was used to characterize changes in N-glycosylation profiles of serum IgG of healthy controls and patients (see Materials and Methods). Values in Table 3 show significant differences between the coeliac patients and the three healthy controls, with the most significant differences seen in children and adults under 40 years of age (P < 0·0002); the older negative control group had a higher level of G0F glycans and the P-value was < 0·01.

Table 3.

Tukey's HDS test results for significant differences on Ln (G0F/G1F) between groups including coeliac disease patients and three negative controls, children under 12, adults under 40 and adults over 40 years old. Values in the body of the table show the significance of the differences between the groups of the corresponding row and column. The P-values below 0·05 (considered significant) are shown in bold type

	P values
	0–12	Under 40	Over 40	CD patients
0–12		0·426161	0·142280	0·000182
Under 40	0·426161		0·002312	0·000166
Over 40	0·142280	0·002312		0·009855

To understand better the change in glycosylation profile in coeliac disease a comparison was made with two sets of data for patients with RA and PsA (data not shown) analysed previously in our laboratory, as shown in Fig. 2. Ln(G0F/G1F) increased for each autoimmune disease group, compared to the three healthy control groups. The results shown in Fig. 2 indicate a higher significant difference between all groups of patients and the children control group and young adults than to the group of adults over 40. The marked increase in G0F glycoform in adults over 40 makes it difficult to differentiate this group from those with disease.

Fig. 2.

Plotting of the mathematical expression Ln(G0F/G1F) for three negative controls and coeliac disease, rheumatoid arthritis and psoriatic arthritis groups to characterize G0–IgG glycoform change in autoimmune diseases. Mean + 1·96*s.e. is the confidence interval for each group at a confidence level of 0·95. Standard errors were calculated using pooled variance. Overlapping of confidence intervals of two groups indicates that there is no significant difference between them.

Finally, an opposed tendency of both G1F and G2F to G0F for patients of all the autoimmune diseases under study is shown clearly in Fig. 3. Clearly, G1F and G2F are less sensitive to change than G0F. While for the diseases the mean G0F population is 41%, the negative control G0F population is only 31%, being G1F and G2F, 39% and 21% for the patients and 45% and 24% for the negative controls, respectively.

Fig. 3.

Comparative analysis of the oligosaccharide population as percentage of the total neutral N-glycans from serum IgG in patients of autoimmune diseases (coeliac disease, psoriatic arthritis and rheumatoid arthritis) and negative controls belonging to three different groups: children up to 12 years, adults under 40 years and adults over 40 years old. The points represent mean values, the boxes standard errors and the whiskers confidence intervals at a confidence level of 0·95. Standard errors were calculated using pooled variance.

DISCUSSION

It is accepted that both humoral and cellular immune responses are involved in the pathogenesis of coeliac disease, in particular an inappropriate T cell mediated response against ingested gluten [9,10], and the presence of antigliadin and antitransglutaminase antibodies of IgA or IgG classes is dependent strictly on dietary exposure to gluten [2,10]. While there is a definite association between the presence of antitransglutaminase autoantibodies and CD, their role in the immunopathogenesis of CD is not known [11].

The ease of monitoring the glycosylation status of antibody molecules depends on the number of glycosylation sites present within the molecule. IgA having a relatively complex set of O- and N- linked glycosylation sites is difficult to assay, as changes may be confined to a single site. The IgG molecule has predominantly a single glycosylation site within the IgG–Fc region. For this reason we evaluated only the variation of neutral N-glycan profiles for total polyclonal IgG isolated from the sera of CD patients.

We found, as described previously for other autoimmune diseases such as RA, JCA, WG and MPA, that coeliac disease is characterized by a change in glycosylation profile of total serum IgG (Table 2). Similar to these previous reports, the increase in G0F oligosaccharides is likely to be confined to N-glycans attached to the IgG-Fc region. Our results suggest that, unlike in RA and PsA, the increase of G0F oligosaccharides in CD is shown to be independent of age [22,23] because patients’ ages ranged from 1 to 12 years old, and comparison with a same-age group of healthy children showed significant statistical difference (Table 3, Fig. 2).

As expected, the increase in G0F oligosaccharides was mirrored by a decrease in G1F and G2F oligosaccharides (Fig. 3), as reported previously in RA [34]. Two possible mechanisms may be proposed for the increase in G0F oligosaccharides: (i) a stress-induced deficit in the glycosylation machinery of plasma cells affecting specifically the addition of galactose [35]. An IgG rate-limiting galactosylation reaction was associated initially with the formation of the inter-disulphide bridges between the two heavy chains, restricting the accessibility of the galactosyltransferases to the already-linked N-glycan [36]. Additionally, increased synthesis of IgG could limit the final galactosylation pattern. In B-cells IL6 stimulates the secretion of antibodies to such a degree that serum IgG1 levels can rise 120–400-fold. Moreover, high levels of IL6 are also detected in patients with rheumatoid arthritis, chronic juvenile arthritis, osteoporosis and psoriasis [37]. A close relationship between increased IL6 and altered IgG oligosaccharide structure has also been suggested [38]. (ii) the secretion of normally glycosylated IgG that is acted on by a stress induced galactosidase present in the blood and body fluids: evidence supporting the former but not the latter mechanism has been presented [35,36].

It has been demonstrated that mannose binding lectins (MBL) can bind G0–IgG displaying terminal GlcNAc and activate the lectin pathway of complement. Thus antigen/antibody complexes comprised of a substantial proportion of G0–IgG could generate a proinflammatory response [17]. Also, the removal of terminal galactose was shown to increase the uptake of soluble IgG mediated by the mannose receptor (MR) on the macrophages and dendritic cells [27]. The interaction of MBL and/or MR with G0–IgG self-antigen/antibody immune complexes could potentiate T cell help and the development of class switched, somatically mutated autoantibody responses resulting in pathology.

High levels of G0 glycoform have been observed in advance of clinical pathology and might be perceived as a contributor to, rather than a result of the pathology. In addition, in an animal model of collagen-induced arthritis the passive transfer of G0 autoantibodies was shown to exacerbate the disease [39] and therefore to play a major role in pathogenesis; however, the link between exposure to the G0–IgG glycoform and initiation of inflammation was not revealed. Thus, whether the increase of G0–IgG glycoform is a cause or consequence remains uncertain. Moreover, the wide range of autoimmune diseases, showing increased levels of G0–IgG glycoform with different clinical symptoms, suggests that it might be related to autoimmune processes initiated by diverse factors, all sharing a common inflammatory process.

Contrary to Wormald's [40] initial report, Mimura et al. 2001 [41] did not observe changes in oligosaccharide accessibility, the binding affinity for sFcγRIIb or thermal stability of the IgG1–Fc CH2 domain in G0F–IgG glycoforms, relative to galactosylated structures. It has been reported, however, that the conformation of the oligosaccharide can influence complement activation [42]. A similar mechanism might be considered for the G0F–IgG glycoform in autoimmune diseases, i.e. to diminish complement activation and consequent tissue damage. However, while the G0F glycoform may be less effective at activating the classical pathway via C1q, it may favour activation of the lectin pathway. These issues may reward investigation.

Demonstration is under study of whether these changes in glycosylation profile in coeliac disease is a consequence provoked directly by gluten ingestion in sensitive patients. However, increased levels of G0F–IgG glycoform could have a decisive role in coeliac disease as well as in many other autoimmune disorders.