Antibodies against neo‐epitope of microbial and human transglutaminase complexes as biomarkers of childhood celiac disease

Summary

Tissue transglutaminase (tTG) and microbial transglutaminase (mTG) cross‐link gliadins to form complexes that expose immunogenic neo‐epitopes to produce tTG and mTG‐neo‐epitope antibodies. The aim of this study was to test the diagnostic performance of antibodies against non‐complexed and complexed forms of transglutaminases, to correlate their activities to the intestinal damage and to explore age group dependency in celiac disease (CD). A total of 296 children with untreated CD and 215 non‐celiac disease controls were checked by in‐house enzyme‐linked immunosorbent assays detecting immunoglobulin (Ig)A, IgG or combined detection of IgA and IgG (check) against tTG, AESKULISA ^® tTG New Generation (tTG‐neo) and mTG‐neo (RUO), IgA and IgG antibodies against deamidated gliadin peptide (DGP) and human IgA anti‐endomysium antibodies (EMA) using AESKUSLIDES ^® EMA. Intestinal pathology was graded according the revised Marsh criteria, and age dependencies of the antibody activities were analysed. Using cut‐offs estimated from receiver operating characteristic (ROC) curves, the highest area under curve (AUC) of the TG assays was 0·963 for tTG‐neo check, followed by tTG check (0·962) when the diagnosis was based on enteric mucosal histology. tTG‐neo check was the most effective to reflect the intestinal abnormalities in CD (r = 0·795, P < 0·0001). High levels of anti‐mTG‐neo IgG and anti‐tTG‐neo IgG appeared in the earlier age groups, as compared to anti‐tTG IgG (P < 0·001). Considering antibody diagnostic performance based on AUC, enteric damage reflection and predictability at an early age, the anti‐neo tTG check was the most effective diagnostic biomarker for pediatric CD. The mTG neo check might represent a new marker for CD screening, diagnosis and predictability.

Back to back comparison of 10 various, celiac disease associated antibody were checked in 296 children with untreated CD compared to 215 non‐celiac disease controls.Considering antibodies’ diagnostic performances based on AUC, enteric damage reflection and predictability at an early age, the anti‐neo tTG check was the best diagnostic biomarker for pediatric CD.The mTG neo Check might represent a new marker for CD screening, diagnosis and predictability.

Introduction

Tissue transglutaminase (tTG) is the main autoantigen in celiac disease (CD) 1. Although the presence of circulating immunoglobulin (Ig)A‐tTG is a dominant feature of CD, the mechanism by which autoimmunity against tTG is developed and the pathogenic role played by those autoantibodies is still unresolved 2, 3, 4. Conformational epitopes are important for tTG to be recognized as an autoantigen, and binding of autoantibodies are restricted to a few distinct functional domains on tTG 5, 6, 7, 8. tTG is a post‐translation modifier of proteins resulting in deamidation or transamidation (cross‐linking) of specific glutamine residues, known to be abundant on the gluten/gliadin molecules 9.

In the intestinal lumen, gluten is partially digested, resulting in proteolysis‐resistant peptides which can be deamidated by tTG, leading to the formation of deamidated gliadin peptides (DGPs). DGPs have a stronger binding avidity to major histocompatibility complex (MHC) class II, thus potentiating the immune response and the selection of pathogenic T cell clones. However, it is frequently forgotten that transamidation (75%) occurs at a higher rate than deamidation (25%) in the tTG‐gliadin cross‐talks 10. After transamidation, tTG is covalently linked to gliadin peptides to create neo‐epitope complexes called tTG‐neo. Notably, these neo‐epitopes were described in vivo in small intestine biopsies of CD patients where a pathogenic role was attributed to them 11, 12, 13. The formation of tTG‐neo‐epitopes and their processing by antigen‐presenting cells and presentation to the immune system would support the hypothesis of epitope‐spreading and the ensuing development of autoantibodies against tTG 11, 14.

Microbial transglutaminase (mTG) is a member of the extended family of transglutaminases, having indispensable, multi‐functional, extremely evolutionarily preserved tasks, present in all prokaryotes 9, 15. Functionally, mTG imitates the endogenous human tTG, both capable of cross‐linking numerous molecules with gliadin peptides, and therefore able to present a high variety of neo‐epitopes 9, 14, 15. mTG is a heavily used, unlabeled food additive in processed food industries 14, 15, 16. Despite declarations of its safety for nutritional industrial use, direct evidence for the immunogenicity of the gliadin–mTG complex was recently detected, in contrast to lack of antibodies against the uncomplexed enzyme 14. Recently, it was hypothesized that mTG is a new environmental trigger of CD, based on the following scientific data: (1) it de‐/transamidates gluten such as the endogenous human tTG; (b) being less substrate and pH‐sensitive than its counterpart, it is capable of cross‐linking many more proteins and other macromolecules, changing their three‐dimensional structure/composition/antigenicity/physical and chemical characteristics, resulting in an increased intestinal luminal load presented to the immune system; and (3) its transamidated complexes are covalently linked, making them resistant to luminal proteases, detergents and other enteric luminal hostile factors 14, 17, 18.

The aims of the present study were to test the sensitivity and specificity of antibodies against both non‐complexed and complexed forms of both mTG and tTG compared to DGP antibodies in IgA‐endomysial antibody (EMA)‐positive or pathologically proved CD children compared to non‐CD controls. The second aim was to correlate the various antibody activities to the degree of intestinal atrophy (Marsh score = 0–IIIc) 19; and thirdly, to analyze the longitudinal evolution of antibody titers among the increased age groups, the hypothesis being that antibodies directed against various neo‐epitopes of tTG and mTG gliadin complexes are reliable diagnostic markers, reflect intestinal injury and surge at an early age during CD development, thus potentially playing an important role in CD initiation and progression.

Materials and methods

Study population

Included were 555 children who were referred for an upper endoscopy performed at the Department of Pediatrics, Skåne University Hospital SUS in Malmö, Sweden.

A total of 296 children, median age = 6·7 (range = 0·9–19·0) years, 200 females and 96 males, fulfilled the diagnostic criteria of CD. An additional 215 children, median age = 11·5 (range = 0·9–18·3) years, 117 females and 98 males, had an intestinal biopsy showing Marsh score < 2 and were included as non‐celiac disease controls (DC). The five most common indications for being investigated with endoscopy among the DC were gastroesophageal reflux disease (n = 50), Helicobacter pylori gastritis (n = 49), recurrent abdominal pain (RAP) (n = 22), inflammatory bowel disease (n = 16) and food allergy or intolerance (n = 14). The remaining 44 children, median age = 3·9 (range = 1·9–16·4) years, 30 females and 14 boys, were referred to endoscopy with positive IgA‐tTG, but had normal intestinal features and were diagnosed as potential CD. The study was approved by the ethical committee at Lund University and parents gave their informed consent.

Sample collection

Samples were collected on the day of intestinal biopsy from children who were investigated with an upper endoscopy performed by the pediatric gastroenterology team at the Department of Pediatrics, Skåne University Hospital SUS in Malmö, Sweden. Six small bowel biopsies were collected from the proximal part of the duodenum and two from the bulb, stained with hematoxylin and eosin and anti‐CD3 monoclonal antibodies, according to standard procedures. Diagnosis of CD was established according to the European Society for Paediatric Gastroenterology, Hepatology and Nutrition (ESPGHAN) 1990 criteria 20; that is, an intestinal biopsy showing a Marsh score >2 19.

Immunoassays

The following enzyme‐linked immunosorbent assays (ELISAs) detecting IgA, IgG separated or combined IgA and IgG (check) were used: AESKULISA ^® gliadin (AGA), tTG [tTG; RUO (for in‐house research use only)], AESKULISA^® DGP (DGP), AESKULISA^® tTG ‘New Generation’ (tTG neo‐epitope), AESKULISA^® mTG neo‐epitope and mTG (RUO), according to the manufacturer’s instructions (AESKU.DIAGNOSTICS GmbH, Wendelsheim, Germany). AESKUSLIDES ^® EMA were used for indirect fluorescent IgA and IgG determination of EMA, according to the manufacturer’s instructions (AESKU.DIAGNOSTICS GmbH).

Statistical methods

Statistical analyses were performed using MedCalc for Windows, version 15.0 (MedCalc Software, Ostend, Belgium). All statistical tests were applied at a significance level of 5% (P < 0·05). Two‐tailed P‐values were reported.

Two study outcomes were used for the purpose of this study. The primary outcome was autoimmune CD patients presenting with positive EMA, and the controls were defined as being EMA‐negative. The secondary outcome was biopsy‐proven CD; a case was defined as having a Marsh score > 2 and a control was considered as having a Marsh score < 2. Receiver operating characteristic (ROC) analysis was used to evaluate the diagnostic test performance [area under the curve (AUC), sensitivity and specificity], which means evaluating the usefulness of measurable antibody activity in patients’ sera as markers to discriminate diseased cases from normal controls.

Spearman’s rank correlation was performed to determine the association between the progressive development of intestinal damage (Marsh score) and antibody levels. The Marsh scores have been categorized to subsequent numbers for calculation. The Spearman’s correlation coefficient (rho) confidence interval (CI) was 95%. All calculated P‐values were below 0·0001.

Results

The ROC of the TG antibody assays

The ROC curves of each tTG assay for both study outcomes, EMA‐selected (upper panel) and biopsy‐proven CD (lower panel), are shown in Fig. 1. For better visualization and clarity, the multiple ROC curves in Fig. 1 are summarized in Table 1, where the AUC of the ROC plots are presented in decreasing order. The tTG‐neo check had the highest AUC in the autoimmune (EMA⁺) as well as in the biopsy‐proven CD groups (0·934 and 0·963, respectively).

Figure 1.

Receiver operating characteristic (ROC) graphs of immunoglobulin (Ig)A, IgG and check (IgA and IgG) of tTG, tTG‐neo and microbial transglutaminase (mTG)‐neo antibodies for endomysial autoantibody (EMA‐upper panel) positivity and celiac disease pathology (CD‐lower panel), respectively.

Table 1.

Diagnostic performance of the different antibodies based on presence of endomysial autoantibody (EMA) positivity as the primary outcome and celiac disease (CD) as the secondary outcome, respectively (P < 0·0001)

	EMA			Median	CD			Median
	AUC	Sensitivity (%)	Specificity (%)	(range) U/ml	AUC	Sensitivity (%)	Specificity (%)	(range) U/ml
tTG neo check	0·934	84·1	94·4	300 (1–450)	0·963	90·9	90·4	300 (1–450)
tTG check	0·923	87·0	93·6	300 (1–442)	0·962	93·9	89·4	300 (1–442)
mTG neo check	0·898	81·6	83·1	181 (9–414)	0·903	75·3	88·8	145 (10–414)
DGP check	0·893	86·0	81·0	189 (0–588)	0·909	86·5	81·5	156 (2–588)
tTG neo IgA	0·925	82·7	96·1	300 (0–449)	0·940	86·2	90·7	300 (0–449)
tTG IgA	0·927	86·7	95·4	300 (0–438)	0·959	91·9	91·5	300 (0–438)
mTG neo IgA	0·875	71·2	93·0	26 (0–447)	0·849	78·0	80·7	19 (0–447)
tTG neo IgG	0·894	78·2	88·0	88 (0–444)	0·928	79·7	91·1	79 (1–444)
tTG IgG	0·907	82·3	91·6	23 (0–433)	0·935	86·5	86·5	22 (0–433)
mTG neo IgG	0·877	74·5	84·9	126 (3–449)	0·891	71·6	89·6	112 (5–449)

tTG = tissue transglutaminase; mTG = microbial transglutaminase; DGP = deamidated gliadin peptide; Ig = immunoglobulin.

The diagnostic performance of the TG antibody assays

Comparing all assays, only tTG IgA and tTG‐neo check showed a combined sensitivity and specificity > 90% in the biopsy‐proven celiac patients (Table 1). The tTG check had the highest sensitivity (93·9%), while the tTG‐neo IgA showed the highest specificity (96·1%). In contrast, the mTG‐neo check showed inferior specificity and sensitivity compared to the tTG‐neo check and tTG IgA (Table 1). The best average performance between mTG isotypes was allocated to the mTG‐neo check.

Correlation between antibody activities and mucosal injury

Correlations between antibody levels and Marsh scores are shown in Fig. 2. The tTG‐neo check (r = 0·795, P < 0·0001) was the most effective, followed by tTG‐IgA (r = 0·786, P < 0·0001), as well as the tTG check and tTG‐neo IgA (r = 0·778, 0·757, P < 0·0001, respectively), to reflect the intestinal abnormalities in CD. Comparing the mTG‐neo isotypes, mTG‐neo check was the most useful to reflect the enteric damage.

Figure 2.

Correlation between antibody levels and mucosal Marsh score.

Age dependency of anti‐TG levels

Figure 3 summarizes the IgA (a) and the IgG (b) of the TG antibody activities in correlation to the various age groups.

Figure 3.

Immunoglobulin (Ig)A (a) and IgG (b) antibody titers directed against the different antigens according to the defined age groups.

There was no statistically significant difference between the antibody levels of anti‐tTG IgA and anti‐tTG‐neo IgA among the age groups (P > 0·05). However, a statistically significant difference between anti‐mTG‐neo IgA and anti‐tTG IgA/anti‐tTG‐neo IgA, starting at the < 3 – < 4‐year age group (P < 0·001), was detected. Looking at the values of IgG, there was a substantial difference in the 1–3‐year age group, where significantly higher levels of tTG‐neo and mTG‐neo antibodies than anti‐tTG (P < 0·001) were observed.

Discussion

The present study addresses three topics in CD serological diagnosis: (1) serological comparisons of the well‐established and most frequently used tTG‐IgA to three other antibodies: DGP and the two most recently described tTG and mTG‐neo‐epitopes; (2) back‐to‐back comparison of the antibody performances to reflect the intestinal histological damage in pediatric CD; and (3) the age‐dependent, longitudinal behavior of the antibody activities in the increasing age groups of the participants.

Overall, the antibody’s diagnostic reliability is good, but differential performances exist. While tTG‐neo check had the highest AUC on the ROC graph, the highest sensitivity was allocated to tTG check and the highest specificity was occupied by tTG‐neo IgA. It should be mentioned that anti‐tTG and anti‐tTG‐neo autoantibodies are directed against a single and a complexed endogenous autoantigen of CD, respectively. Conversely, anti‐mTG‐neo antibodies are directed against the gliadin‐docked mTG complex, where the two moieties are exogenous and are heavily consumed as industrial processed food additives, both presenting environmental components 15, 16, 17, 18. In this regard, gluten, which comprises 80% of the wheat protein, is not only the established environmental inducer of CD, but was shown recently to also affect the health of non‐celiac individuals 21, 22. Intriguingly, gluten has a plethora of immunological side effects. Being immunogenic, cytotoxic, proinflammatory, pro‐apoptotic and pro‐oxidative 21, what function does the tTG/mTG docked gliadin moiety of the complex play in activating the mucosal immune systems, boosting the tTG/mTG‐neo antibodies in CD patients’ sera?

Deamidation of gliadin is a major event in CD evolvement 23, and antibodies against the deamidated gliadin are known as CD‐associated biomarkers. In the present study, the anti‐DGP check antibodies were the last to perform in the AUC analysis, thus substantiating their recently published low performances in CD diagnosis 24, 25. Conversely, several groups have shown that tTG cross‐linking of gliadin takes place in CD pathogenesis 26. The cross‐linking ability of the tTG and mTG to dock gliadin is the basis for the changes in the epitopes created and exposed on the new complexes, resulting in tTG‐ and mTG‐neo antibody formation (Fig. 4).

Figure 4.

Schematic hypothesis of the formation of the neo‐epitope complexes and their corresponding antibodies in celiac disease (CD). First part of the figure shows the gluten processing pathway, where cereal products get digested in the intestine by luminal enzymes. (a) During this process gliadin is cleaved into peptides which then reacts with tissue transglutaminase (tTG), in the lamina propria and forms covalently cross‐linked complexes (tTG neo‐epitopes). (b) Before or after digestion of gliadin, the resulting peptides can be further processed by microbial transglutaminase (mTG), which is able to imitate tTG and form cross‐linked peptides (mTG neo‐epitopes). Those tTG deamidated and tTG or mTG cross‐linked peptides and their neo‐epitopes induce antibody formation, resulting in anti‐deamidated gliadin peptide (DGP, tTG, tTG‐neo and mTG neo‐antibodies.

The second current CD‐related topic is the ability of various antibodies to reflect CD intestinal abnormality. Taken as a whole, all the investigated antibodies are good reflectors, but the tTG‐neo check stood out as the foremost one. Taken together, it appears that the check isotypes (IgA + IgG) are more reliable for diagnosis and CD intestinal pathology reflection 27. One of the advantages of the check assay is that its IgG isotype will detect IgA‐deficient CD patients, as an estimated 2% of the CD patients have selective IgA deficiency.

Clinicians and scientists have long debated the gold standard parameters for CD diagnosis. Intestinal biopsy was considered to be the gold standard in the past, but in the last decade incorrect interpretation, non‐uniformity in grading using revised Marsh criteria and the lack of agreement on the cut‐off number of intra‐epithelial lymphocytes, a hallmark of CD histology, raise questions concerning the diagnostic reliability of intestinal pathology 19, 28. Along with the improvement of CD serological reliability, antibodies occupy a higher place in the diagnostic CD flowchart. This is also clearly shown in the new ESPGHAN criteria, published in 2012 29. It is noteworthy that, having been on the market for several decades, tTG–IgA has been explored extensively, and increasing numbers of false positive and negative results are being described 30, 31. These constitute some of the reasons for the need to better explore serological biomarkers, prompting the present study to challenge tTG‐IgA and DGP antibodies against the two neo‐epitope antibodies. The present results reinforce the findings of comparable studies, where tTG–IgA and tTG‐neo were compared back‐to‐back for CD diagnosis 14, 32, 33, and tTG‐neo was found to outperform tTG‐IgA in CD 33.

Being a novel aspect in CD immunogenicity, mTG and its corresponding anti‐mTG‐neo isotypes deserve some discussion. Based on chemical, biochemical, epidemiological and immunological data, mTG was hypothesized recently to be associated with CD induction and development 14, 15, 17, 18. CD children mount a significantly higher incidence and activity level of mTG‐neo antibodies compared to controls, also reflecting their enteric pathology 14. mTG‐neo isotypes were shown not to be the best, but were an acceptable biomarker of CD diagnosis and intestinal injury reflection. After establishing immunogenicity in CD, its actual potential pathogenicity is being currently unraveled. In this sense, mTG was recently shown to exert an anti‐phagocytic effect 34, 35, thus suppressing a crucial component of human immunity. Furthermore, it can be internalized trans‐enterocytically to be deposited beneath the human enteric epithelium, thus facing the immune cells and systems 13. No less importantly, Stricker et al. showed that by heating the mTG–gliadin complexes it becomes more immunogenic, leading to the possibility of neo‐epitopes being exposed during the process 13. In view of the tTG–gliadin subepithelial deposits in CD mucosa 12, could they contain mTG–gliadin complexes? The enzyme can be detected when checked in supermarket processed food products 36, and as its substrates are covalently linked they are resistant to human gut lumen proteases. It seems that the mTG and its gliadin cross‐linking capability and pathogenicities are far from being explored.

Bizzaro et al. have put forward the hypothesis that the autoantibody response against neo‐epitopes could appear during the early phase of CD development. This could lead to the formation of anti‐DGP and anti‐tTG antibodies by epitope spreading in a similar manner to the formation of anti‐nucleosome, anti‐dsDNA and anti‐histone antibodies suggested for systemic lupus erythematosus 10. The hypothesis was substantiated by Tonutti et al., who presented the detection of neo‐epitope tTG autoantibodies in two pediatric patients 6 months or more before other parameters, including tTG–IgA, showed a positive result (Tonutti et al., at the 7th international Congress on Autoimmunity in 2010, Udine, Italy 37). The present age‐dependent antibody activities reinforce the predictability of the tTG and mTG‐neo IgG at an early stage.

Complexed to gliadin, mTG shows an enhanced immunogenicity in children with CD, and anti‐mTG‐neo‐epitope IgG antibodies correlate with intestinal damage to a comparable degree to anti‐tTG‐neo IgA 14, 32. Additionally, mTG and tTG displayed a comparable immunopotent epitope despite the fact that they do not share sequence homology 14, 32. In summary, mTG‐neo IgG presents a new marker for CD, and further studies are needed to explore the pathogenic potential of anti‐mTG antibodies in CD involvement.

The present study has several advantages: the disease control patients had a normal intestinal biopsy in order to rule out CD; the participating children were evaluated in one center. Serological analysis was performed blindly in another country by one technician; the check kits of the CD‐associated antibodies can detect IgA deficiency, due to the incorporated IgG isotype and two acceptable and often‐used CD diagnostic gold standards, meaning that histological criteria and EMA‐positive autoimmunity were compared.

Conversely, several limitations should be mentioned: a multi‐center study would have better represented CD global aspects, an extended cohort would have reinforced the statistical strength and the conclusion reliability and the experimental design allowed only two aspects of CD diagnosis (serology and pathology) to be checked but, epidemiologically, clinical presentation and disease activity were not evaluated, and could have affected the results.

In summary, comparing IgA, IgG and IgA + IgG (check) isotypes of tTG, tTG‐neo, mTG‐neo and DGP check, the tTG‐neo check stood out as the most effective biomarker for CD diagnosis and intestinal damage reflection. Based on the present and published data 14, 32, 33, it is suggested that the serological marker of tTG‐neo check antibodies should be recommended as a first‐line diagnostic screening test for CD in children. The tTG and tTG‐neo IgA assays show similar diagnostic performance, and are both recommended as good screening tests for CD in children. Antibodies against mTG‐neo present a new serological biomarker for CD, but their role in disease development needs to be further evaluated.

Disclosures

T. M. is the CEO of Aesku group and the head of AESKU.KIPP research institute. All the other authors declare no conflicts of interest.

Author contributions

Conceptualization: D. A., A. L. and T. M.; methodology: P. W. and S. N.; formal analysis: A. H. and P. W.; resources: T. M., data curation: D. A. and A. L.; writing – original draft preparation: L. A. and D. A., writing – review and editing: A. L., D. A., P. W., S. N. and M. T. All authors read, approved the final manuscript and consented for publication.