Subclass Profile of IgG Antibody Response to Gluten Differentiates Nonceliac Gluten Sensitivity From Celiac Disease.

Celiac disease (CD) is an autoimmune enteropathy triggered by exposure to gluten proteins, leading to intestinal inflammation and villous atrophy in genetically predisposed individuals. It is associated with robust B cell and antibody responses to gluten and to the transglutaminase 2 autoantigen.1Caio G. et al.CBMC Med. 2019; 17: 142Crossref PubMed Scopus (315) Google Scholar In contrast, nonceliac gluten or wheat sensitivity (NCGS) is a poorly understood clinical entity defined by onset of symptoms in response to ingestion of gluten-containing food without the prerequisite serologic or histologic features of CD.2Volta U. et al.Gastroenterol Clin North Am. 2019; 48: 165-182Abstract Full Text Full Text PDF PubMed Scopus (27) Google Scholar There are no established biomarkers yet for NCGS, but recent research points to a biological basis, revealing a state of systemic immune activation in conjunction with a compromised intestinal epithelium.2Volta U. et al.Gastroenterol Clin North Am. 2019; 48: 165-182Abstract Full Text Full Text PDF PubMed Scopus (27) Google Scholar,3Uhde M. et al.Gut. 2016; 65: 1930-1937Crossref PubMed Scopus (125) Google Scholar We and others have demonstrated a significant increase in IgG antibody to gluten in NCGS at levels similar to CD.2Volta U. et al.Gastroenterol Clin North Am. 2019; 48: 165-182Abstract Full Text Full Text PDF PubMed Scopus (27) Google Scholar,3Uhde M. et al.Gut. 2016; 65: 1930-1937Crossref PubMed Scopus (125) Google Scholar Accordingly, it has been speculated that an enhanced IgG response to gluten may be a common link between CD and NCGS.2Volta U. et al.Gastroenterol Clin North Am. 2019; 48: 165-182Abstract Full Text Full Text PDF PubMed Scopus (27) Google Scholar However, whether and how B cell reactivity to gluten may differ in these conditions, especially in the context of possible relevance to intestinal pathology, have not been examined. In this study, we extend earlier data to show that the anti-gluten IgG antibody in NCGS is significantly different from CD in subclass distribution and in its relationship to intestinal cell damage. The findings are suggestive of a sustained primary B cell response to gluten in CD, despite the condition’s chronicity, and a more advanced and tolerogenic immune response to gluten in NCGS. Detailed methods are available in the Supplementary Material. Demographic and clinical characteristics of study cohorts are included in Supplementary Table 1. The anti-gliadin IgG response in CD patients was composed primarily of IgG1 and IgG3, which were significantly increased in comparison with the healthy and NCGS cohorts (Figure 1A and C). There was a modest elevation in anti-gliadin IgG2 compared with the healthy group and no comparative increase in the IgG4 subclass (Figure 1B and D). Within the NCGS cohort, however, the lower contributions of anti-gliadin IgG1 and IgG3 in comparison with CD was compensated by significantly elevated IgG4 (compared with CD and healthy cohorts) and IgG2 (compared with healthy cohort) (Figure 1A–D). No significant association was detected in this cohort between any anti-gliadin IgG subclass and the Marsh type, HLA-DQ2/DQ8 status, or eligibility for irritable bowel syndrome or functional dyspepsia diagnostic criteria. The score plot for the principal component analysis of the IgG subclass data demonstrated clustering of the CD and NCGS subjects into discernible groups, further demonstrating the contrasting subclass distributions and suggesting potential biomarker value in these data (Figure 1E). Serum concentrations of intestinal FABP2, a specific marker of intestinal epithelial cell damage,4Pelsers M.M. et al.Clin Chim Acta. 2005; 352: 15-35Crossref PubMed Scopus (358) Google Scholar were similarly elevated in the CD and NCGS groups in comparison with healthy cohort (P < .0001 for each).3Uhde M. et al.Gut. 2016; 65: 1930-1937Crossref PubMed Scopus (125) Google Scholar Within the CD group, only the anti-gliadin IgG3 correlated with FABP2 (Figure 1H). This correlation was similar in strength to that between anti-gliadin IgG3 and anti–transglutaminase 2 IgA (r = 0.505, P = .001). In contrast, FABP2 levels in the NCGS group correlated with anti-gliadin IgG4 and weakly with IgG1 (Figure 1M and J). The observed contrast in the IgG subclass distribution and relationship with FABP2 release in NCGS vs CD are likely reflective of differences in the evolution and disease relevance of B cell immune responses in the 2 conditions. Among IgG subclasses, IgG1 and IgG3 are the most potent activators of complement and efficient at binding a wide range of FcγRs.5Vidarsson G. et al.Front Immunol. 2014; 5: 520Crossref PubMed Scopus (1344) Google Scholar In contrast, IgG2 antibodies generally require higher epitope densities for complement activation and display limited binding to FcγRs.5Vidarsson G. et al.Front Immunol. 2014; 5: 520Crossref PubMed Scopus (1344) Google Scholar IgG4 antibodies contain structural properties that further distinguish them from other immunoglobulin isotypes and IgG subclasses. They bind weakly to Fc receptors and to complement, and are inefficient at crosslinking of antigens or forming immune complexes.5Vidarsson G. et al.Front Immunol. 2014; 5: 520Crossref PubMed Scopus (1344) Google Scholar IgG4 has also been shown to induce an anti-inflammatory M2-like macrophage phenotype through inhibition of interferon-gamma signaling.6Swisher J.F. et al.MAbs. 2014; 6: 1377-1384Crossref PubMed Scopus (23) Google Scholar Considering these properties, the observed increase in the gluten-reactive IgG2 and IgG4 subclasses and the correlation between the IgG4 subclass and FABP2 in NCGS may point to a protective response aimed at dampening the inflammatory effect of other antibodies and immune cells. It is intriguing that these antibody responses are largely absent in CD, where there is instead a correlation between the IgG3 and FABP2. The evolution in subclass switching of the IgG response to an antigen follows a 1-way direction from IgG3 to IgG1, IgG2, and IgG4 over time. Once a B cell has switched to a downstream subclass, it does not return to a preceding one.7Valenzuela N.M. et al.Transplantation. 2018; 102: S7-S13Crossref PubMed Scopus (42) Google Scholar It has been suggested that IgG2 and IgG4 are part of the immunologic memory toward harmless and recurring antigens—an advanced immune response stimulated by a more extensive antigen exposure.8de Jong B.G. et al.Immunol Cell Biol. 2017; 95: 744-752Crossref PubMed Scopus (31) Google Scholar In addition, the variable regions of IgG2 and IgG4 usually display greater levels of somatic hypermutation than IgG1 or IgG3, which can result in higher affinity for target antigens.7Valenzuela N.M. et al.Transplantation. 2018; 102: S7-S13Crossref PubMed Scopus (42) Google Scholar As such, the prominence of the IgG3 subclass and its relationship with the autoimmune response and intestinal cell turnover in CD is suggestive of repeated activation of gluten-specific naïve B cells rather than of memory cells, in response to gluten exposure, despite the chronic nature of the disease. Pathways involved in this phenomenon may represent a source of molecular targets for therapeutic intervention. Possible shortfalls of this study include the lack of other disease controls and the fact that these observational data cannot establish a causal connection between subclass differences and the disease process. These data warrant further examination of the evolution of gluten-reactive B-cell response and subclass switching in CD and NCGS. In addition, information on other aspects of B cell and antibody variability, including affinity, glycosylation profile, and epitope specificity, is expected to contribute to a greater understanding of differences in the immune response to gluten and its relationship with disease pathophysiology in the two conditions. In conjunction with other previously identified markers, these components of the immune response to gluten are expected to provide additional biomarkers that may be informative in the context of stratifying potential disease subsets with varying mechanisms, prognoses, and responses to therapy. The authors thank Dr Mary Ajamian for technical support and critical review of the manuscript. Melanie Uhde, PhD (Data curation: Lead; Formal analysis: Lead; Methodology: Equal; Writing – original draft: Supporting). Giacomo Caio, MD (Data curation: Supporting; Methodology: Equal; Resources: Equal; Writing – review & editing: Supporting). Roberto De Giorgio, MD (Data curation: Supporting; Investigation: Supporting; Resources: Supporting; Writing – review & editing: Supporting). Peter Green, MD (Investigation: Supporting; Methodology: Supporting; Resources: Supporting; Writing – review & editing: Supporting). Umberto Volta, MD (Data curation: Supporting; Funding acquisition: Supporting; Investigation: Supporting; Resources: Supporting; Writing – review & editing: Supporting). Armin Alaedini, PhD (Conceptualization: Lead; Data curation: Lead; Funding acquisition: Lead; Investigation: Supporting; Methodology: Supporting; Project administration: Lead; Resources: Lead; Supervision: Lead; Writing – original draft: Lead). The study included 80 individuals with NCGS who met the diagnostic criteria proposed by an expert group1Catassi C. et al.Nutrients. 2015; 7: 4966-4977Crossref PubMed Scopus (363) Google Scholar and who were identified using a previously described structured symptom questionnaire2Volta U. et al.Gastroenterol Clin North Am. 2019; 48: 165-182Abstract Full Text Full Text PDF PubMed Scopus (27) Google Scholar,3Uhde M. et al.Gut. 2016; 65: 1930-1937Crossref PubMed Scopus (125) Google Scholar (a modified version of the Gastrointestinal Symptom Rating Scale designed to rate symptoms commonly associated with NCGS). All NCGS subjects reported experiencing gastrointestinal or extra-intestinal symptoms after ingestion of gluten-containing foods, including wheat, rye, or barley. The most common gastrointestinal symptoms included bloating, abdominal pain, diarrhea, nausea, and heartburn, and the most prominent extra-intestinal symptoms were fatigue, headache, anxiety, cognitive difficulties, and numbness in arms and legs. The reported symptoms in all subjects improved or disappeared when those foods were withdrawn for a period of 6 months, and recurred when they were re-introduced for a period of up to 1 month. Individuals were excluded if they were already on a restrictive diet or had used nonsteroidal anti-inflammatory drugs within the past 6 months, if they were positive for the CD-specific intestinal histologic findings or the IgA anti-endomysial or -transglutaminase 2 autoantibodies, or if they were positive for wheat allergy-specific IgE serology or skin prick test. A history of autoimmunity or autoantibody reactivity was present in 20 NCGS patients (25%), represented mainly by Hashimoto’s thyroiditis and by antinuclear antibody positivity (demonstrated by indirect immunofluorescence on HEp-2 cells), similar to previously published data.4Pelsers M.M. et al.Clin Chim Acta. 2005; 352: 15-35Crossref PubMed Scopus (358) Google Scholar,5Vidarsson G. et al.Front Immunol. 2014; 5: 520Crossref PubMed Scopus (1344) Google Scholar In addition, 49 patients (61.2%) met criteria for irritable bowel syndrome and 63 (78.8%) for functional dyspepsia according to Rome IV.6Swisher J.F. et al.MAbs. 2014; 6: 1377-1384Crossref PubMed Scopus (23) Google Scholar,7Valenzuela N.M. et al.Transplantation. 2018; 102: S7-S13Crossref PubMed Scopus (42) Google Scholar All patients underwent an esophagogastroduodenoscopy with gastric biopsy to rule out Helicobacter pylori infection. A total of 6 intestinal biopsies, including 2 from the duodenal bulb and 4 from the distal duodenum, were taken from each individual. The study also included 40 patients with biopsy-proven CD and 40 healthy subjects, recruited as part of the same protocol that included the NCGS individuals. All cases of CD were positive for IgA anti-endomysial and IgA anti-transglutaminase 2 autoantibodies, biopsy-proven, and diagnosed according to established criteria.8Rubio-Tapia A. et al.Am J Gastroenterol. 2013; 108: 656-676Crossref PubMed Scopus (1143) Google Scholar Rome IV,6Swisher J.F. et al.MAbs. 2014; 6: 1377-1384Crossref PubMed Scopus (23) Google Scholar,1Catassi C. et al.Nutrients. 2015; 7: 4966-4977Crossref PubMed Scopus (363) Google Scholar Gastrointestinal Symptom Rating Scale,10Svedlund J. et al.Dig Dis Sci. 1988; 33: 129-134Crossref PubMed Scopus (964) Google Scholar and SF-36 Health Survey11Ware Jr., J.E. et al.Med Care. 1992; 30: 473-483Crossref PubMed Scopus (28047) Google Scholar were utilized to evaluate the general health of unaffected controls. Individuals who had a history of liver disease, liver function blood test results (total protein, aspartate transaminase, alanine transaminase, alkaline phosphatase, albumin, globulin, and bilirubin) outside of normal range, or a recent infection were excluded from all cohorts in the study. Cases of IgA deficiency and IgG subclass deficiency were excluded from all cohorts. Inflammatory bowel disease was ruled out in all cases. All samples were collected at time of study entry while participants were on an unrestricted (gluten-containing) diet with written informed consent under Institutional Review Board–approved protocols at St Orsola-Malpighi Hospital, Bologna, Italy. Serum specimens were kept at –80°C to maintain stability. This study was approved by the Institutional Review Board of Columbia University Medical Center. Serum levels of total IgG reactivity to gluten and individual IgG subclass reactivities to gluten were measured separately by an enzyme-linked immunosorbent assay, similar to our earlier studies.4Carroccio A. et al.Gastroenterology. 2015; 149: 596-603.e1Abstract Full Text Full Text PDF PubMed Scopus (64) Google Scholar,5Volta U. et al.Gastroenterology. 2016; 150: 282Abstract Full Text Full Text PDF PubMed Scopus (9) Google Scholar The antigen used for the assays was the Prolamine Working Group reference gliadin, as described and characterized previously.4Carroccio A. et al.Gastroenterology. 2015; 149: 596-603.e1Abstract Full Text Full Text PDF PubMed Scopus (64) Google Scholar,6Mearin F, et al. Gastroenterology doi: 10.1053/j.gastro.2016.02.031.Google Scholar A 2-mg/mL stock solution of the Prolamine Working Group gliadin was prepared in 70% ethanol. Wells of 96-well Maxisorp round-bottom polystyrene plates (Nunc, Roskilde, Denmark) were coated with 50 μL/well of a 0.01-mg/mL solution of protein in 0.1 M carbonate buffer (pH 9.6) or left uncoated to serve as controls. After incubation at 37°C for 1 hour, wells were washed and blocked by incubation with 1% bovine serum albumin in phosphate-buffered saline containing 0.05% Tween-20 for 1.5 hours at room temperature. Serum samples were diluted at 1:300, added at 50 μL/well in duplicates, and incubated for 1 hour. Each plate contained a positive control sample with a high level of relevant IgG subclass reactivity to gluten, as determined in a preliminary screen. After washing, the wells were incubated with horseradish peroxidase–conjugated anti-human IgG1 (Life Technologies, Carlsbad, CA), IgG2 (Life Technologies), IgG3 (Life Technologies), or IgG4 (Southern Biotech, Birmingham, AL) secondary antibodies for 50 minutes. Plates were washed and 50 μL of developing solution, containing 27 mM citric acid, 50 mM Na2HPO4, 5.5 mM o-phenylenediamine, and 0.01% H2O2 (pH 5), was added to each well. Absorbance was measured at 450 nm after 20 minutes. All samples were tested in duplicate. Absorbance values were corrected for nonspecific binding by subtraction of the mean absorbance of the associated bovine serum albumin–coated control wells. The corrected values were first normalized according to the mean value of the positive control duplicate on each plate. The mean antibody level for the healthy control cohort was then set as 1.0 AU and all other results were normalized to this value. The cutoff value for anti-gliadin IgG positivity was assigned as 2 SDs above the mean for the healthy control group. Serum levels of intestinal FABP2 were measured, as we have described previously.13Uhde M. et al.Gut. 2016; 65: 1930-1937Crossref PubMed Scopus (166) Google Scholar FABP2 is a cytosolic protein specific to intestinal epithelial cells that is released into systemic circulation upon cellular damage.15Pelsers M.M. et al.Clin Chim Acta. 2005; 352: 15-35Crossref PubMed Scopus (387) Google Scholar Increased concentrations of circulating FABP2 reflect epithelial cell loss and enhanced enterocyte turnover rate15Pelsers M.M. et al.Clin Chim Acta. 2005; 352: 15-35Crossref PubMed Scopus (387) Google Scholar, 16Pelsers M.M. et al.Clin Biochem. 2003; 36: 529-535Crossref PubMed Scopus (250) Google Scholar, 17Sandler N.G. et al.Gastroenterology. 2011; 141: 1220-1230Abstract Full Text Full Text PDF PubMed Scopus (226) Google Scholar and have been associated with both CD and NCGS.13Uhde M. et al.Gut. 2016; 65: 1930-1937Crossref PubMed Scopus (166) Google Scholar IgA antibody to recombinant human TG2, a sensitive and specific serologic marker for CD,18Rostom A. et al.Gastroenterology. 2006; 131: 1981-2002Abstract Full Text Full Text PDF PubMed Scopus (604) Google Scholar was measured in all serum samples, as outlined previously.13Uhde M. et al.Gut. 2016; 65: 1930-1937Crossref PubMed Scopus (166) Google Scholar HLA genotyping to assess CD genetic predisposition was done as reported previously.12Moeller S. et al.PLoS One. 2014; 9e94677Crossref PubMed Scopus (23) Google Scholar Group differences were analyzed by the Kruskal-Wallis 1-way analysis of variance with post-hoc testing. Correction for multiple comparisons was done using Dunn’s statistical hypothesis testing and the multiplicity-adjusted P values are reported for each comparison. Correlation analysis was performed using Spearman’s r. A multivariate principal component analysis was carried out on the entire IgG subclass dataset to assess clustering. All P values were 2-sided and differences were considered statistically significant at P < .05. Statistical analyses were performed with Prism, version 8 (GraphPad, La Jolla, CA) and Minitab, version 19 (Minitab, Chicago, IL). Supplementary Table 1Demographic and Clinical Characteristics of Study CohortsSubject groupNo. of subjectsAge, y, mean (SD)Female sex, n (%)CD-associated HLA DQ2 and/or DQ8, n (%)Intestinal biopsy histologic grade: Marsh 0; Marsh 1; Marsh 3, n (%)NCGS8034.6 (10.3)aNo statistically significant differences exist among the NCGS, CD, and healthy cohorts.62 (78)aNo statistically significant differences exist among the NCGS, CD, and healthy cohorts.21 (26)bStatistically significant differences exist between the CD and NCGS cohorts (P < .0001 for all comparisons).48 (60); 32 (40); 0bStatistically significant differences exist between the CD and NCGS cohorts (P < .0001 for all comparisons).CD4034.5 (13.7)aNo statistically significant differences exist among the NCGS, CD, and healthy cohorts.30 (75)aNo statistically significant differences exist among the NCGS, CD, and healthy cohorts.40 (100)bStatistically significant differences exist between the CD and NCGS cohorts (P < .0001 for all comparisons).0; 0; 40 (100)bStatistically significant differences exist between the CD and NCGS cohorts (P < .0001 for all comparisons).Healthy4035.0 (12.8)aNo statistically significant differences exist among the NCGS, CD, and healthy cohorts.30 (75)aNo statistically significant differences exist among the NCGS, CD, and healthy cohorts.——a No statistically significant differences exist among the NCGS, CD, and healthy cohorts.b Statistically significant differences exist between the CD and NCGS cohorts (P < .0001 for all comparisons). Open table in a new tab