Cocoa Attenuates Gluten‐induced Pathogenesis in a Preclinical Model of Celiac Disease

ABSTRACT

Celiac disease (CeD) is an autoimmune‐mediated disorder triggered by gluten ingestion. Strict adherence to a gluten‐free diet (GFD) is currently the only available treatment, yet it presents nutritional and lifestyle challenges. Polyphenols and other bioactive food compounds have shown potential in modulating inflammation and gut health. This study evaluated the impact of cocoa, a polyphenol‐rich food, on CeD pathogenesis in a preclinical model. DQ8‐D^d‐villin‐IL‐15tg mice with predisposition to CeD were fed either a GFD (REF), a gluten‐containing diet (GLI), or a gluten‐containing diet supplemented with cocoa (GLI+COCOA) for 25 days. Outcomes assessed included intestinal histology, antibody (Ab) levels, cytokine and immunoglobulin profiles and mesenteric lymph node lymphocytes’ phenotypes. Cocoa administration limited gluten‐induced intestinal villous atrophy and reduced anti‐gliadin Abs, while significantly decreasing the secretion of pro‐inflammatory cytokines. Moreover, cocoa normalized the Ig isotype profile dysregulated by gluten intake. Cocoa intake exerts a protective effect against key hallmarks of CeD, attenuating inflammation and limiting the damage to the intestinal structure. These findings support cocoa as a promising complementary dietary strategy to modulate CeD‐related manifestations, although additional translational studies are warranted

Cocoa intake exerts a protective effect against key hallmarks of CeD, attenuating inflammation and limiting the damage to the intestinal structure.

Abbreviations

Ab

Antibody

CeD

Celiac Disease

dSI

Distal Small Intestine

ECS

Extracellular Staining

F

Female

GCD

Gluten‐Containing Diet

GFD

Gluten‐Free Diet

GW

Gut Wash

HOV

Homogeneity of Variances

IC

Intestinal Content

ICS

Intracellular Staining

IFN

Interferon

Ig

Immunoglobulin

IL

Interleukin

M

Male

MID

Monocytes

MLN

Mesenteric Lymph Node

moAb

Monoclonal Antibody

NK

Natural Killer Cell

NKT

Natural Killer T Cell

NMDS

Non‐metric Multidimensional Scaling

TG2

Tissue Transglutaminase 2

Th

T Helper Cell

TNF

Tumor Necrosis Factor

Vh:Cd

Villus Height to Crypt Depth Ratio

1. Introduction

Celiac disease (CeD) is a chronic multi‐organ immune‐mediated enteropathy triggered by exposure to dietary gluten in genetically predisposed individuals [1]. Its average worldwide prevalence ranges from 0.5% to 2% of the general population [2] and it is 2‐2.5 times more frequent in women than men [3]. Even though its incidence has greatly increased in the last 50 years, mainly due to better diagnostic tools and wider screening, it has been calculated that up to 70% of cases remain undiagnosed [4]. Another common gluten‐related disorder is non‐celiac gluten/wheat sensitivity, a condition in which gluten ingestion leads to morphological or symptomatic manifestations despite the absence of CeD [5].

Gluten is a wide term, which includes the main prolamin storage proteins in wheat (gliadin and glutenin), rye (secalin), barley (hordein), and oats (avenin) [6]. These proteins contain high amounts of glutamine and proline residues, which make them water insoluble and increase their resistance to digestion by gastric and pancreatic enzymes.

Currently, the only effective strategy to prevent CeD‐associated symptoms and gastrointestinal (GI) damage is strict adherence to a gluten‐free diet (GFD). This can pose a great challenge from both social and economic perspectives. Moreover, it has been reported that many GFD alternatives have lower nutritional value and that following this diet can lead to nutritional deficiencies [7]. To this end, there is still an active search for therapeutic alternatives to a GFD, which include gluten‐degrading enzymes, probiotics, tight junction modulators, transglutaminase 2 (TG2) inhibitors or immune cell‐targeted therapies, among others [8]. Nevertheless, none of them has proven sufficiently effective to be considered a novel treatment for CeD. A less explored therapeutic approach to CeD is the use of polyphenols, which have been shown to reduce the bioavailability and digestibility of gliadin peptides through its sequestration, to display antioxidant and anti‐inflammatory activity, to improve the intestinal epithelial barrier integrity and modulate its functions, to interact with the gut microbiota and to have immune‐modulatory properties [9].

Cocoa (Theobroma cacao) is a multi‐faceted bioactive food which contains high amounts of the aforementioned health‐promoting polyphenols, especially flavanols, anthocyanidins and proanthocyanidins, but also other bioactive compounds, including methylxanthines (theobromine and caffeine), phytosterols, fibers and several vitamins and minerals [10]. Cocoa powder has been associated with beneficial effects in the context of several diseases, especially cardiovascular and neurodegenerative conditions like diabetes, obesity, cancer, coronary heart disease and stroke [11, 12, 13]. This is due to its widely reported antioxidant [14, 15], anti‐inflammatory, anti‐allergic [16] and microbial‐modulatory activities [17]. Nevertheless, the effect of cocoa intake on prevention or symptom alleviation in CeD has yet to be elucidated, since only very few studies have been focused on this topic [11, 18, 19].

To this end, the goal of this study was to evaluate the potential benefits of cocoa intake alongside a gluten‐containing diet (GCD) on the pathogenesis of CeD using a preclinical model with genetic predisposition to the illness. To achieve this, many relevant biomarkers of this disease were analyzed in a mice model of CeD, including intestinal manifestations, antibody (Ab) response against gliadin and TG2 and extent of villous atrophy.

2. Experimental Section

2.1. Mice and Experimental Design

Animals used in this study were DQ8‐D^d‐villin‐IL‐15tg mice, which were expanded from progenitors kindly provided by V. Abadie from The University of Chicago [20]. Expression of the DQ8 transgenic phenotype and presence of the D^d ‐IL‐15 and villin‐IL‐15 transgenes in all received and newborn animals was assessed through staining of mesenteric lymph node (MLN) lymphocytes and multiplex PCR on ear tissue as previously reported [21]. Mice were maintained in polycarbonate cages in the animal facility of the University of Barcelona (Diagonal Animal Experimentation Unit‐9900030) with ad libitum access to water and to a GFD (CA. 170481 AIN‐76A Purified Diet (Rats/Mice), Envigo, Wisconsin, USA), controlled temperature and humidity and a 12 h:12 h light‐dark cycle (Animal Experimentation Ethics Committee (CEEA) Ref. 186/20). The experimental design consisted of 42 9‐week‐old mice, which were divided in three groups (n = 14 mice per group, 7 females [F] and 7 males [M]) (Figure 1A). The experimental unit was an individual mouse; hence, all replicates were of biological nature. Group size calculation was performed with the Granmo software (https://www.datarus.eu/aplicaciones/granmo/), using intestinal anti‐gliadin IgA Abs as the primary variable to ensure statistically significant differences. No dropout rate and a two‐sided type I error (0.05) were assumed.

FIGURE 1.

(A) Experimental design used to evaluate in vivo the effect of cocoa administration alongside a gluten‐containing diet (GCD) supplemented with gliadin in the context of a preclinical celiac disease (CeD) model (n = 14 mice/group, 7 females and 7 males). (B) Body weight (g), (C) chow (g) and (D) water (mL) intake normalized per 100 g of animal body weight, and fecal variables ((E) weight (%), (F) pH and (G) humidity (%)) measured at the last week of the 25‐day intervention (week 4). ALL, F and M, stand for all animals, only females and only males, respectively. Results are expressed as mean ± SEM. *p ≤ 0.05 versus REF, #p ≤ 0.05 versus GLI. All variables met the assumptions of normality and homogeneity of variance; therefore, a two‐way ANOVA (for ALL animals) or a one‐way ANOVA (for F or M animals) followed by Bonferroni post hoc tests were performed for each independent variable. A significant interaction in ALL animals between treatment group and sex was observed for the following variables: Body weight (p = 0.001), Fecal weight (%) (p = 0.001) (E), Fecal pH (p = 0.009) and Fecal humidity (%) (p = 0.035).

2.2. Dosage Information

Healthy control animals (REF) were maintained on a GFD and administered 100 µL of phosphate buffered saline (PBS) at pH 7.2 using a Hamilton syringe (Hamilton Company, Reno, USA) and a 22G‐gauge gavage tube (Asico, Singapore, Singapore), in order to mimic the handling and administration conditions of the two other groups. Two additional groups received the standard GCD from the animal facility (Teklad 2018 Global 18% Protein Rodent Diet, Inotiv, Lafayette, USA), supplemented with a 20 mg gliadin ball [21] (G3375, Merck, Darmstadt, Germany) by oral gavage thrice a week on alternate days in order to trigger the outcomes derived from gluten exposure in the context of CeD. Out of these two, one of them remained untreated serving as diseases control animals (GLI), while another group followed was additionally supplemented with a 10‐minute prior administration of defatted cocoa (composition in Table 1) at a dose of 5 g/kg animal body weight diluted in PBS 7.2 (GLI + COCOA) by oral gavage using a Hamilton syringe and a 22G‐gauge gavage tube. The human equivalent dose was calculated using the body surface area normalization method, resulting in 0.4 g/kg for a 60 kg adult [22]. The intervention lasted 25 days, during which animal body weight, chow and water consumption, fecal pH and humidity were tracked.

TABLE 1.

Nutritional composition of the defatted cocoa powder used as a supplement (diluted in PBS) to be administered to mice fed with a GCD and gliadin (GLI+COCOA group) at a dose of 5 g/kg body weight.

Component	Values
Moisture	4.0%
Total Fat	11.0%
Saturated	7.0%
Monounsaturated	3.7%
Polyunsaturated	0.3%
Trans	<0.05%
Cholesterol	0.0 mg
Carbohydrates
Total sugars	11.9%
Added sugars	0.9
Starches	11.0%
Polyols	0.0%
Dietary fiber	35.2%
Proteins	23.1%
Organic acids	2.7%
Polyphenols	3.1%
Alkaloids	2.4%
Theobromine	2.2%
Caffeine	0.2%
Ashes	6.5%
Potassium	2.0%
Sodium	0.05%
Salt (sodium × 2.5)	0.1%
Energy Value
Calories (kcal)	318 kcal
Calories (kJ)	1319 kJ

Values are expressed per 100 g of total cocoa weight.

2.3. Biological Sample Collection and Processing

The nutritional intervention lasted 25 days, in order to provide sufficient time for gluten‐derived intestinal and immune alterations to occur. On the 25^th day of intervention (d25) animals were administered and deprived of food and water for 1.5 h to equalize intestinal content. They were subsequentially anesthetized intramuscularly with 90 mg/kg of ketamine (Merial Laboratories S.A. Barcelona, Spain) and 10 mg/kg of xylazine (Bayer A.G., Leverkusen, Germany) and exsanguinated. Blood was placed in Microvette tubes containing the anti‐coagulant EDTA (Sarstedt, Nümbrecht, Germany) and immediately analysed using the automated hematologic analyser Spincell 3 (MonLab Laboratories, Barcelona, Spain). The remaining blood was homogenized and centrifuged (Sigma 112 Mini Centrifuge, Osterode amb Harz, Germany) to obtain plasma. Several relevant organs were weighed and their tissues and contents were stored for further analysis. As CeD is an enteropathy, several intestinal samples were collected from the distal small intestine (dSI), the site showing the greatest damage in this particular mouse model. To assess Ab and autoantibody (auto‐Ab) levels, dSI samples were incubated at 37°C and shaked at 750 rpm in 4 mL of PBS pH 7.2, centrifuged at 538 g (Megafuge 1.0R centrifuge, Heraeus, Hanau, Germany) and the supernatants, i.e., the gut wash (GW), were collected. Moreover, a 0.5 cm dSI tissue samples were immersed in a solution of PBS‐paraformaldehyde 4% (Applichem Panreac ITW Reagents, Castellar del Vallès, Spain) to be used for histological measurements.

2.4. Autoantibody Quantification

Anti‐TG2 and anti‐gliadin Ab levels were evaluated in two different compartments: at systemic level (plasma) and in intestinal samples (GW). Mouse anti‐tissue TG2 IgA Ab ELISA kit and mouse anti‐gliadin Ab (IgA or IgG) ELISA kits (AMSBIO Europe BV, Alkmaar, Netherlands) were used for quantification following the manufacturers’ instructions.

2.5. Cytokine and Immunoglobulin Isotype Quantification

Intestinal samples (tissue and content) and exclusively intestinal contents (IC) from the dSI were homogenized to a final concentration of 90 and 500 mg sample/mL, respectively, using a pellet pestle cordless Motor (Kimble) and PBS pH 7.2. Interleukin (IL)‐4, IL‐5, IL‐6, IL‐12p70, interferon (IFN)‐γ and Tumor Necrosis Factor (TNF)‐α concentrations were quantified using a ProcartaPlex Mouse Essential Th1/Th2 Panel 6‐Plex (Invitrogen, Waltham, United States) in plasma and intestinal samples stored at −80°C. IgA, IgE, IgG1, IgG2a, IgG2b, IgG3, and IgM levels were assessed in plasma and intestinal (GW) samples using a ProcartaPlex Mouse Antibody Isotyping Panel 7‐Plex (Invitrogen). Both tests were performed according to the manufacturer's protocol. A MAGPIX Cytometer (Luminex, Austin, USA) in the Flow Cytometry Unit of the Scientific and Technological Centers of the University of Barcelona (CCiTUB) was then used for data acquisition. To obtain a balance between pro‐inflammatory and anti‐inflammatory Ig isotypes the following formula was used: ∑(IgG2a+IgG3)/∑(IgG1+IgG2b).

2.6. Small Intestine Histology

The assessment of the intestinal morphology was performed by hematoxylin/eosin staining of sections as previously reported [23] using dSI samples, since this region was described to develop the highest atrophy of intestinal villi in the current mouse model of CeD [20, 21]. After staining and mounting, slides were examined using the bright‐field microscope Olympus BX41 (Olympus Corporation, Shinjuku, Tokyo, Japan). The villi height, area and width, as well as the crypt depth were measured using Image J software v.  1.54s6 (Image Processing and Analysis in Java, National Institute of Mental Health, Bethesda, MD, USA) and the villi height/crypt depth (Vh:Cd) ratio was calculated.

2.7. Mesenteric Lymph Node (MLN) Lymphocyte Phenotyping

The MLN were smashed into a sterile 40 µm mesh cell strainer (Fisher Scientific, Waltham, USA) to isolate the lymphocytes [24]. The viability and number of lymphocytes were determined by a Countess Automated Cell Counter (Invitrogen). MLN lymphocytes (5·10⁵ cells) were both intracellularly stained (ICS) and extracellularly stained (ECS). For ECS the following monoclonal Ab (moAb)‐fluorochrome pairs were used: anti‐CD3 APC/Cyanine7, anti‐CD4 PerCP/Cyanine5.5, anti‐CD8a FITC, anti‐CD44 PE, anti‐CD62L APC, anti‐CD127 (IL‐7Rα) PE/Dazzle 594 (BD Life Sciences, San Diego, USA), anti‐CD19 Super Bright 780 and anti‐CD314 (NKG2D) Super Bright 600 (Invitrogen). For ICS, the moAb‐fluorochrome pairs used were anti‐IL‐2 Brilliant Violet 711 and anti‐IFN‐γ Brilliant Violet 421 (BioLegend, San Diego, USA). MLN lymphocyte phenotyping was then performed as previously described [25]. Briefly, Fc receptors were first blocked using purified rat anti‐mouse CD16/CD32 (BD Life Sciences), incubated with the respective mix of EC moAbs in saturating conditions, permeabilized, incubated with the mix of IC antibodies diluted in Brilliant Stain Buffer (Invitrogen) and fixed with 0.5% p‐formaldehyde (Sigma‐Aldrich). A negative control using an isotype‐matched moAb was included in each sample. Analyses were performed by flow cytometry using an Aurora cytometer (Cytek Biosciences, Fremont, USA) in the Flow Cytometry Unit of the CCiTUB and for posterior gating and data analysis the FlowJo v10.8 Software was applied (BD Life Sciences).

2.8. Statistical and Data Analysis

Statistical analysis was performed using IBM SPSS Statistics for Windows, Version 29.0.2.0 (IBM Corp., Armonk, USA). The Shapiro‐Wilk test was used to evaluate if the data followed a normality plot. The Levene test was used to assess data's homogeneity of variance. One or two‐way ANOVA tests followed by post‐hoc analysis with the Bonferroni (homogeneity of variance) or Dunnet T3 (no homogeneity of variance) test were used for sample sets that followed normality. Variables that did not meet the assumption of normality were analyzed using a nonparametric two‐way ANOVA based on the Aligned Rank Transform (ART) procedure, implemented in R (R Core Team (2023). _R: A Language and Environment for Statistical Computing, R Foundation for Statistical Computing, Vienna, Austria) using the ARTool package, which allows testing of main effects and interactions within factorial designs. Results are shown as mean percentages with respect to the REF animals (designed as 100%) ± Standard Error of the Mean (SEM), unless detailed otherwise. R and Rstudio (RStudio Team [2023]. RStudio: Integrated Development for R. RStudio, PBC, Boston, MA) were used for Non‐metric Multi‐Dimensional Scaling (NMDS) ordination of samples based on the type of stimulation using the Bray‐Curtis distance and R package vegan (Oksanen J, et al. (2022). _vegan: Community Ecology Package_. R package version 2.6‐4, https://CRAN.R‐project.org/package=vegan). The non‐parametric ANOSIM test of significance allowed to determine if the differences between two or more groups were significant. The Adonis test was then used as a post‐hoc test for pairwise comparisons. p < 0.05 was used to reject the null hypothesis in all analyses.

3. Results

3.1. Follow‐Up Variables

Animal body weight (Figure 1B) at the final week of the 25‐day intervention (week 4, W4) was significantly higher in gluten‐consuming (GLI), compared to gluten‐free (REF) animals, a phenomenon which was prevented by cocoa intake. A similar pattern was observed for chow consumption, but not water intake, which did not show differences among groups (Figure 1C,D). In terms of fecal variables (Figure 1E–G), which are indicators of intestinal abnormalities, fecal humidity was again higher in those animals from the GLI group, but not in the REF and GLI + COCOA groups. While a decrease in fecal pH was linked to gliadin intake, this was not prevented by cocoa. Hematological profile assessment on d25 showed no differences between the REF and GLI groups, while cocoa intake only led to minor changes, such as slightly higher percentage of lymphocytes (%LYM) and monocytes (%MID), as well as a larger red cell distribution width (%RDW‐CV) (Table S1). The lymphocyte‐to‐monocyte ratio (LMR) remained unchanged regardless of the treatment group.

3.2. Plasma and Intestinal Antibody Levels

As expected in a model of CeD, gliadin‐intake led to significantly higher Ab (anti‐gliadin) and auto‐Ab (anti‐TG2) levels, regardless of Ab type and isotype (IgA or IgG) (Figure 2A–D), indicating successful disease induction. Interestingly, cocoa administration was capable of successfully preventing the increase of anti‐gliadin IgG levels in both plasma and intestinal samples. This preventive effect exerted by cocoa was even stronger for intestinal anti‐gliadin IgA levels, with significantly lower Ab titers present in GLI + COCOA animals, than in those of the GLI group. However, cocoa administration was not able to prevent the increase in plasma anti‐TG2 IgA levels caused by gliadin‐intake in CeD‐predisposed animals.

FIGURE 2.

Anti‐gliadin (A–C) or anti‐transglutaminase (TG2) (D) antibody levels in intestinal samples (A,B) or plasma samples (C,D) from mice at the end of the intervention (day 25) (n = 14/group, 7 females and 7 males). Assessed antibodies belong to the IgG (A,C) or IgA (B,D) isotypes. ALL, F and M, stand for all animals, only females and only males, respectively. Results are expressed as mean ± SEM. For the graphical visualization of results, the value of the reference (REF) group was fixed to a 100 % and used as a baseline for the positive control of CeD (GLI) group and the cocoa supplemented (GLI + COCOA) group. *p ≤ 0.05 versus REF, #p ≤ 0.05 versus GLI. *p ≤ 0.05 versus REF, #p ≤ 0.05 versus GLI. All variables met the assumptions of normality. All of them also displayed homogeneity of variance except for anti‐TG2; therefore, a two‐way ANOVA (for ALL animals) or a one‐way ANOVA (for F or M animals) followed by Bonferroni (or Dunnet T3 for anti‐TG2) post hoc tests were performed for each independent variable. A significant interaction in ALL animals between treatment group and sex was only observed for plasma anti‐gliadin IgG (p = 0.014).

3.3. Cytokine and Immunoglobulin Quantification

As expected in a context of CeD, at the mucosal level, the concentration of the pro‐inflammatory cytokines IL‐6, IFN‐γ, and TNF‐α within the intestine were significantly higher in the gliadin‐consuming group than in animals following a GFD (Figure 3A–C). This increase was effectively prevented by addition of cocoa into the diet in the cases of IFN‐γ and TNF‐α, but not IL‐6. However, in plasma (systemic level) or IC samples, no differences in cytokine concentrations among any of the groups were found (Figures S1 and S2). An NMDS representation of intestinal cytokines (stress = 0.12) showed no overall difference between groups (p = 0.138), while the pairwise comparison showed significant differences between REF‐GLI (p = 0.032), but not GLI‐GLI + COCOA (p = 0.098) or REF‐GLI + COCOA (p = 0.831) (Figure 3P). No differences were found in NMDS analyses of plasma cytokines (Figure S4A).

FIGURE 3.

(A–F) Presence of Th1 and Th2 cytokines in distal small intestine (dSI) tissue samples from celiac disease (CeD) mice at the end of the study (day 25) (n = 14/group, 7 females and 7 males). (G–O) Percentages of immunoglobulin (Ig) isotypes determined in dSI tissue samples from CeD mice. Total IgG (K) was calculated by summing IgG1 (G), IgG2a (H), IgG2b (I) and IgG3 (J). The Th1/Th2 ratio associated to Ig isotypes was obtained through the following ratio: ∑(IgG2a+IgG3)/∑(IgG1+IgG2b). ALL, F and M, stand for all animals, only females and only males, respectively. Results are expressed as mean ± SEM. For the graphical visualization of results, the value of the reference (REF) group was fixed to a 100% and used as a baseline for the positive control of CeD group (GLI) and the cocoa‐supplemented group (GLI + COCOA). *p ≤ 0.05 versus REF, #p ≤ 0.05 versus GLI. *p ≤ 0.05 versus REF, #p ≤ 0.05 versus GLI. The data for the variables IL‐6, IFN‐γ, and TNF‐α did not follow a normal distribution. Hence, statistical significance was evaluated using a nonparametric two‐way ANOVA based on the Aligned Rank Transform (ART) procedure. All Ig data met the assumptions of normality, IgG1, IgG2a, IgG2b, and IgM presented a homogeneity of variance (HOV), while IgG3, IgA, and IgE did not. A two‐way ANOVA (for ALL animals) or a one‐way ANOVA (for F or M animals) followed by Bonferroni (or Dunnet T3 for variables with no HOV) post hoc tests were performed for each independent variable. No interaction was found between treatment group and sex for any of the variables.(P‐Q) Non‐metric multidimensional scaling (NMDS) analysis of intestinal cytokine (P) and Ig isotype (Q) profiles. Statistical significance is set at p ≤ 0.05.

In terms of Igs, intestinal IgG1, IgG2b, total IgGs, IgA, IgE, and IgM were significantly higher in the GLI group, compared to REF animals. Such increases were prevented in all cases by cocoa intake (Figure 3G–O). Conversely, plasma IgG2a, IgG2b, and IgA levels were lower in gliadin‐consuming animals and the Th1/Th2 ratio was higher in both GLI and GLI + COCOA groups compared to the REF animals. Within plasma samples, no differences were observed based on treatment group (Figure S3). An NMDS analysis of intestinal Ig isotypes (stress = 0.13) resulted in a global significance of 0.001, while p‐values for REF‐GLI, GLI‐GLI + COCOA, and REF‐GLI + COCOA comparisons were 0.002, 0.001, and 0.44, respectively (Figure 3Q). No differences were found in NMDS analyses of plasma Ig isotypes (Figure S4B).

3.4. Histology

Morphological assessment of intestinal architecture consistently showed that the CeD‐model was successfully established, as confirmed by an unvarying statistical difference in all histological variables (villi height and width, crypt depth and Vh:Cd) between the REF and GLI groups (Figure 4A–E). In terms of cocoa administration, it was capable of attenuating the morphological damage of the small intestine typically associated with CeD, as observed by decreased values of villi width (Figure 4B) and crypt depth (Figure 4C), and an increase in the Vh:Cd ratio (Figure 4D), compared to the GLI group. No such effects were observed for villi height and area (Figure 4A,E). A microscopic representative image of villi architecture in each group can be visualized in Figure 4F.

FIGURE 4.

For the histomorphological evaluation of distal small intestine (dSI) samples from celiac disease (CeD) mice at the end of the study (day 25) (n = 14/group, 7 females and 7 males). Morphological measurements of villi structure included villi height (A), villi width (B), crypt depth (C), villi area (E), and the villi height to crypt depth (Vh:Cd) ratio (D). ALL, F and M, stand for all animals, only females and only males, respectively. Results are expressed as mean ± SEM. For the graphical visualization of results, the value of the reference (REF) group was fixed to a 100% and used as a baseline for the positive control of CeD group (GLI) and cocoa‐supplemented group (GLI + COCOA). *p ≤ 0.05 versus REF, #p ≤ 0.05 versus GLI. *p ≤ 0.05 versus REF, #p ≤ 0.05 versus GLI. The only variable that met the assumption of normality and homogeneity of variance (HOV) was the Crypt depth. Its statistical analysis was performed through a two‐way ANOVA (for ALL animals) or a one‐way ANOVA (for F or M animals) followed by Bonferroni post hoc test. For the remaining variables (villi height, width and area, and Vh:Cd) a nonparametric two‐way ANOVA based on the Aligned Rank Transform (ART) procedure was used. No interaction was found between treatment group and sex for any of the variables. (F) Representative microscopic images of the H&E‐stained dSI sections of mice showing villi morphology after a dietary intervention consisting of a gluten‐free diet (REF), a gluten‐containing diet (GLI), or the latter supplemented by cocoa (GLI + COCOA). Magnification 100x.

3.5. Mln Lymphocytes Phenotype

Phenotyping of MLN lymphocytes showed no difference on the main populations between treatment groups (Figure 5A–E). Only a smaller proportion of NK CD8α⁻CD4⁺ cells in the GLI + COCOA group, compared to GLI, was noted (Figure 5D). EC and IC phenotyping (Figure 6) revealed a decrease in the percentage of NK CD44⁺CD62L⁻ cells within the GLI group, which was prevented by cocoa intake (Figure 6D–F). A similar pattern was observed for NKT CD44⁺CD62L⁻ and NKT CD44⁻CD62L⁺ cells (Figure 6G–I). Smaller percentages of T IFN‐γ⁻IL‐2^{− −}and T IFN‐γ⁺IL‐2⁻ cells (Figure 6J–L) and of NK IFN‐γ⁻IL‐2⁺ cells (Figure 6M–O) were observed after cocoa administration.

FIGURE 5.

Proportions of the main MLN lymphocyte populations: B (A), B/T cells ratio (B), T (C), NK (D) and NKT (E) cells at the end of the study (day 25) (n = 14/group, 7 females and 7 males). The percentage of CD4⁺ (Th, T helper), CD8α⁺ (Tc, T cytotoxic), double positive (DP), or double negative (DN) cells within the T, NK, and NKT cell populations are also displayed in the corresponding graphs. Results are expressed as mean ± SEM. For the graphical visualization of results, the value of the reference (REF) group was fixed to a 100% and used as a baseline for the positive control of the CeD group (GLI) and the cocoa‐supplemented group (GLI + COCOA). *p ≤ 0.05 versus REF, #p ≤ 0.05 versus GLI. All variables met the assumptions of normality and homogeneity of variance; therefore, a two‐way ANOVA followed by Bonferroni post hoc tests were performed for each independent variable. No interaction was found between treatment group and sex for any of the variables.

FIGURE 6.

Flow cytometry analysis of CD44 and CD62L extracellular (A–I) and of IL‐2 and IFN‐γ intracellular (J–R) expression in mesenteric lymph node (MLN) lymphocytes within the T, NK, and NKT main cell populations at the end of the intervention (day 25) (n = 14/group, 7 females and 7 males). Dot plots show representative expression profiles of the CD44‐CD62L or the IL‐2‐IFN‐γ pairs across experimental groups, with quadrant values indicating the mean percentage ± SEM of cells within each gate. The top left quadrant gates CD44⁻CD62L⁺/IL‐2⁻FN‐γ⁺ cells, the top right CD44⁺CD62L⁺/IL‐2⁺IFN‐γ⁺ cells, the bottom left CD44⁻CD62L⁻/IL‐2⁻IFN‐γ⁻ cells, and the bottom right CD44⁺CD62L⁻/IL‐2⁺IFN‐γ⁻ cells. Color intensity reflects cell density, with red indicating regions of highest cell proportion. *p ≤ 0.05 versus REF, #p ≤ 0.05 versus GLI. All variables followed a normal distribution but did not display homogeneity of variance; therefore, a two‐way ANOVA followed by a Dunnet T3 post hoc test was performed for each independent variable. No interaction was found between treatment group and sex for any of the variables.

4. Discussion

In recent years, polyphenols and other bioactive‐containing compounds have gained popularity as adjuvant approaches for numerous diseases, since they can exert multiple health‐promoting activities simultaneously, due to their recognized antioxidant, anti‐inflammatory and antibacterial effects. In the context of CeD, several functional compounds have been investigated, including polyphenols from a wide range of sources (e.g., green tea, grape seeds, propolis, or turmeric), vitamins, fatty acid, or algae derivatives [8], but only two works have researched the interplay between cocoa and CeD, either in the form of an extract or as the principal component from chocolate. One study reported that procyanindin B2‐enriched cocoa extract led to a significant decrease (77%) of TG2 levels, as well as other pro‐inflammatory cytokines, including IL‐1β, IL‐6, IL‐8, and IL‐15 in an in vitro model of gliadin‐sensitized Caco‐2 cells [18]. Another observational investigation showed that consumption of chocolate in CeD patients led to a decrease on the lymphocyte to monocyte and the platelet to lymphocyte ratio, compared to non‐consumers [19]. Since this is currently the only information regarding the relationship between cocoa intake and CeD, we believed that research on this topic was relevant and a preclinical investigation focused on the preventive effect of cocoa and the mechanisms involved was necessary.

Previous studies have proposed multiple mechanisms through which polyphenols may exert beneficial effects on CeD pathophysiology. Briefly, upon gluten ingestion, the protein is hydrolyzed by digestive enzymes into smaller peptides that can cross the lamina propria, undergo deamidation by the TG2 enzyme, and subsequently trigger an immune response. Polyphenols have been shown to interfere with this cascade at several levels, including the inhibition of digestive and TG2 enzymes, the promotion of epithelial barrier integrity, and their well‐established anti‐inflammatory activity. Among these mechanisms, the most relevant appears to be their ability to form complexes with both native and hydrolyzed gliadins, effectively preventing downstream events by reducing gliadin recognition by enzymes, limiting epithelial translocation, and diminishing immune recognition. Thus, the neutralization of gluten peptides by polyphenols emerges as a promising strategy to mitigate CeD pathophysiology [9, 26].

Overall, the results from the current study demonstrate a beneficial effect of cocoa intake in CeD pathogenesis ‐at a preclinical level‐, as shown by a notable preventive effect on some—but not all– of the standard disease biomarkers, that is, Abs and histological damage of the small intestine, which are recommended in the clinical guidelines to detect, follow and determine remission of CeD [1]. In the current study, administration of cocoa alongside a GCD in a preclinical model of CeD, helped attenuate both biomarkers to some extent. In terms of Abs, cocoa was capable of limiting the generation of both intestinal and plasma anti‐gliadin Abs, but not anti‐TG2 auto‐Abs. Similarly, it limited the progression of villous atrophy, quantitatively evaluated by the Vh:Cd ratio, but did not prevent a decrease in villi height. Hence, the conservation of the Vh:Cd ratio was mainly driven by a preservation of the crypt depth. Notably, the Vh:Cd ratio was ∼3 in cocoa‐consuming animals, consistent with a Marsh‐Oberhuber CeD grade of 0–1, which is considered “normal” to “infiltrative”. Animals solely on gluten were found to have Vh:Cd ratios ∼1‐2, indicative of a Marsh 3a grade entailing “partial villous atrophy”, while those following a GFD had ratios up to 4 [27].

Cocoa intake was also able to limit the increase in intestinal pro‐inflammatory cytokines caused by gliadin intake in the context of CeD. Although the tissue homogenization performed herein, without protease inhibitors nor detergent, cannot confirm that all cytokines present in the tissue were released, the present results align well with a prior study reporting similar findings [18], in which addition of cocoa extract to an in vitro model of CeD significantly reduced the gliadin‐induced secretion of IL‐15, COX‐2, IL‐1β, IL‐6, and IL‐8. The same study reported that this effect was linked to procyanidin B2 and other polyphenols present in the cocoa extracts, but not caffeine or theobromine. In addition, earlier studies have highlighted the ability of cocoa to modulate immune responses, notably by decreasing pro‐inflammatory cytokines under diverse conditions [16, 28, 29, 30.

Evaluation of the intestinal Ig isotype profile showed that gluten intake led to a Th2‐skewed phenotype, commonly associated with allergic reactions [31], as observed by an increase in the concentration of IgG1, IgG2b, IgA, IgE and a decrease in the Th1/Th2 ratio. Although this study employed a model of CeD, not gluten allergy, the coexistence of CeD and gluten IgE‐mediated allergy has been previously reported [32]. Moreover, the significant increase in IgA and total IgG concentration could also be associated with the increase of Ab and auto‐Ab levels from these isotypes. Interestingly, cocoa intake effectively attenuated gluten‐derived increases. Although no previous data exists on the effects of cocoa on the Ig profile in the context of CeD, previous studies in different preclinical models (colitis, allergy, healthy animals) have reported significant reductions in both specific and total Ig levels, including a decrease in serum, fecal and bronchoalveolar lavage fluid (BALF) [16, 29, 33, 34].

While only a few differences were observed as a result of cocoa intake in terms of hematological profile, namely an increase in both lymphocytes and monocytes, the same phenomenon was previously reported after cocoa consumption in a mouse model of intestinal inflammation [29]. Similarly, another relevant finding is the effect of cocoa intake on longitudinal variables assessed throughout the 25‐day long intervention, that is, morphological, food‐behavioral and fecal parameters. In line with current results, weight loss [16, 17] and a decrease in fecal pH [35] and humidity [36] have been previously reported after cocoa intake.

An unexpected outcome of this study was the downregulation of CD44 in both NK and NKT cells from the MLNs, when compared with both control animals and those on a cocoa‐containing diet. While in the context of an autoimmune disease like CeD, CD44 is expected to be overexpressed—playing an important role in lymphocyte activation and migration to inflamed tissues [37] —, it can be hypothesized that the decreased expression of CD44 in cells found in the MLNs might be a reflection of a larger percentage of active lymphocytes having already migrated to the gut in the context of active disease. While not much research exists on this topic, previous studies have reported the ability of polyphenols to modulate the activity of other surface molecules involved in lymphocyte migration, such as the inhibition of LPS‐induced CD62L downregulation by quercetin in vitro [38] or the capacity of resveratrol capacity to limit the expansion of CD62L+ monocytes and prevent its migration to inflamed tissues [39], in line with current results. Conversely, the observed increase in expression of both IL‐2 and IFN‐γ in NK and T cells, respectively, in CeD animals following a GCD, as compared to those in a GFD, is a widely reported characteristic phenomenon from this disease [40, 41]. Inclusion of cocoa to the diet was able to limit the increase of proinflammatory cytokines, as previous research has also shown for other polyphenols [42].

In conclusion, cocoa administration exerted a protective effect in the context of CeD in the current murine model of the disease by limiting intestinal damage, reducing pro‐inflammatory cytokine responses and attenuating gluten‐induced alterations in Ig profiles. While the mechanism underlying these findings was not evaluated in this study, cocoa may exert its beneficial effects through multifactorial actions, potentially including the complexation of cocoa's polyphenols with gliadin peptides, a modulation of the epithelial barrier and broader anti‐inflammatory effects. Altogether, these findings provide a rationale for further investigation of cocoa as a potential complementary dietary strategy to modulate CeD‐related manifestations. However, additional studies—including longer interventions, dose–response analyses, and ultimately clinical evaluation—are necessary before any translational relevance or dietary recommendations for patients with celiac disease can be considered.

Funding

This research received no external funding.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: mnfr70472‐sup‐0001‐SuppMat.pdf.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.