Abstract
Transglutaminase 2 (TG2) plays a pivotal role in the pathogenesis of celiac disease (CeD) by deamidating dietary gluten peptides, which facilitates antigenic presentation and a strong anti-gluten T cell response. Here, we elucidate the molecular mechanisms underlying the efficacy of the TG2 inhibitor ZED1227 by performing transcriptional analysis of duodenal biopsies from individuals with CeD on a long-term gluten-free diet before and after a 6-week gluten challenge combined with 100 mg per day ZED1227 or placebo. At the transcriptome level, orally administered ZED1227 effectively prevented gluten-induced intestinal damage and inflammation, providing molecular-level evidence that TG2 inhibition is an effective strategy for treating CeD. ZED1227 treatment preserved transcriptome signatures associated with mucosal morphology, inflammation, cell differentiation and nutrient absorption to the level of the gluten-free diet group. Nearly half of the gluten-induced gene expression changes in CeD were associated with the epithelial interferon-γ response. Moreover, data suggest that deamidated gluten-induced adaptive immunity is a sufficient step to set the stage for CeD pathogenesis. Our results, with the limited sample size, also suggest that individuals with CeD might benefit from an HLA-DQ2/HLA-DQ8 stratification based on gene doses to maximally eliminate the interferon-γ-induced mucosal damage triggered by gluten.
Similar content being viewed by others
Main
Gluten-containing cereals are essential foods worldwide. However, in up to 2% of individuals1, the ingestion of dietary gluten results in an abnormal immune response in the small intestine and the development of celiac disease (CeD). Predisposing genotypes (human leukocyte antigen (HLA), for example, HLA-DQ2 and HLA-DQ8) are necessary but not sufficient for the manifestation of CeD. Diarrhea, weight loss and malnutrition are classical bowel-related symptoms and signs of CeD, but anemia, osteoporosis and other autoimmune diseases, such as type 1 diabetes, are also frequent manifestations2,3,4.
Currently, a gluten-free diet (GFD) is the only accepted treatment option for individuals with CeD. However, the life-long strict and restrictive GFD is onerous and difficult to follow, and inadvertent gluten ingestion is common5,6, resulting in ongoing symptoms in nearly 50% of treated individuals7,8. Keeping the GFD also has a big impact on quality of life9. Inadvertent gluten ingestion often leads to ongoing duodenal mucosal injury, with inflammation and morphological changes10. Thus, even individuals on a GFD frequently have nutrient imbalances and deficiencies11,12. We have shown that despite having normal duodenal histomorphology, individuals with CeD on a GFD differ from individuals without CeD on the molecular level and display insufficient expression of micronutrient transporter genes13. Thus, adjunctive pharmacological therapy, together with a strict GFD, is needed to efficiently treat CeD.
The CeD autoantigen transglutaminase 2 (TG2) is expressed in the intestine, where it deamidates certain neutral glutamine residues to negatively charged glutamic acid residues in immunogenic gluten peptides14,15,16. These modified gluten peptides are more efficiently presented by HLA-DQ2 or HLA-DQ8 molecules on mucosal antigen-presenting cells, which leads to the activation and expansion of gluten-specific CD4+ type 1 helper T cells and the secretion of proinflammatory cytokines17,18. Eventually, this process leads to villus atrophy, crypt hyperplasia and the production of TG2 IgA.
TG2, being crucial for CeD pathogenesis, is a pertinent target for therapy, and this approach was recently tested in a phase 2, randomized, double-blind, placebo-controlled, dose-finding gluten challenge trial using the oral TG2 inhibitor ZED1227 (ref. 19). In this phase 2 trial, ZED1227 attenuated gluten-induced duodenal mucosal injury, both morphological deterioration and inflammation, and improved symptoms and quality of life scores in individuals with CeD19. Here, we report the results of the molecular histomorphometry assessment of ZED1227 efficacy along with intestinal mucosal transcriptomic analysis. Moreover, as the gene dose of HLA-DQ2 was shown to influence the severity of CeD20,21, we analyzed the efficacy parameters of ZED1227 relative to the HLA-DQ2 gene dose.
Results
ZED1227 prevents gluten-induced transcriptomic changes
Duodenal biopsies were collected from 58 individuals with CeD before (GFD) and after a 6-week gluten challenge combined with treatment with 100 mg of the TG2 inhibitor ZED1227 per day (postgluten challenge drug (PGCd); n = 34) or placebo (PGC placebo (PGCp); n = 24). RNA extracted from the 116 biopsy samples was subjected to transcriptomic next-generation sequencing (NGS) analysis.
Principal component analysis (PCA) performed on all samples using DESeq2-transformed counts of all genes showed a moderate level of separation between groups (GFD drug (GFDd), GFD placebo (GFDp), PGCd and PGCp; Fig. 1a). The PGCp group was clearly discernible, whereas the GFDd, GFDp and PGCd groups tended to cluster closer together. There was a clear cosegregation of transcriptomic profiles and mucosal morphology. Thus, a ratio of villus height to crypt depth (VH:CrD) of <1.2 separated from VH:CrD of ≥1.2 and overlapped with PGCp in the PCA (Fig. 1a). A comparison of the PGCp versus GFDp groups detected 95 differentially expressed genes (DEGs; Fig. 1b,c). Strikingly, only one DEG was detected when the GFDd group was compared to the PGCd group, whereas the comparison of the PGCp and PGCd groups indicated 180 DEGs (Fig. 1b,c and Supplementary Data 1).
a, PCA plot using DESeq2-transformed counts for all samples (n = 115). Green, dark green, violet and orange circles correspond to GFDd (n = 34), GFDp (n = 24), PGCd (n = 34), and PGCp (n = 23) samples, respectively. Yellow, blue and red shaded areas depict samples with a high (H; >2.5), medium (M; 1.2–2.5) and low (L; <1.2) range of VH:CrD, respectively. b, Table showing the number of DEGs (log2 (FC) ≥ | 0.5 | and false discovery rate (FDR) ≤ 0.05) in the indicated comparisons. c, Volcano plot representations comparing DEGs as indicated. The green dots indicate DEGs (FDR ≤ 0.05) above the threshold (log2 (FC) of ≥0.5 and ≤−0.5). The dashed horizontal line represents the FDR threshold of 0.05, and the vertical dashed lines represent the log2 (FC) thresholds (≥| 0.5 |). d, Venn diagram illustrating the number of DEGs that are shared in the PGCp versus PGCd and PGCp versus GFDp comparisons. e, Correlation profile of all detected gene (n = 10,063) log2 (FC) values between PGCp and GFDp and PGCp and PGCd comparisons. f, Pearson’s pairwise correlation heat map analyses of 220 DEGs visualizing the cross-correlations of the transcriptomic profiles of the samples (total n = 115; GFDd n = 34; GFDp n = 24; PGCd n = 34; PGCp n = 23). Samples are organized in the ranking order of increasing VH:CrD ratio (indicated in the scatter charts above the heat map).
Because treating participants with ZED1227 eliminated the gluten-induced gene expression changes entirely, it can be assumed that the majority of the DEGs in the PGCp versus GFDp and PGCp versus PGCd comparisons were shared. Indeed, 56 of 95 (59%) DEGs after the gluten challenge were also differentially expressed, according to the comparison of the PGCp and PGCd groups (Fig. 1d). This analysis suggests that a significant number of genes were ‘uniquely’ differentially expressed after gluten challenge (39 of 95) and between the PGCd and PGCp groups (124 of 180; Fig. 1d). Closer inspection of both ‘uniquely expressed’ DEGs revealed that they were not uniquely differentially expressed in PGCd but, to an extent, were equivalent to those expressed in the GFD group, although this was not sufficiently statistically significant (for example, due to inadequate log (fold change) (FC) or expression level), relative to the PGCp group (Supplementary Fig. 1). When all detected gene log2 (FC) values from the PGCp versus GFDp comparison were compared to those from the PGCp versus PGCd comparison, there was a positive correlation, suggesting a similar pattern of gene expression changes in both groups (Fig. 1e). Accordingly, a Pearson’s pairwise correlation heat map analysis with the 220 selected genes showed that the GFDd, GFDp and PGCd groups had similar features, whereas the PGCp group significantly differed from all groups (Fig. 1f). Similar to the results in Fig. 1a, ranking samples according to VH:CrD ratio made it evident that individuals with the most severe mucosal damage, that is, the lowest VH:CrD ratio, had a very different transcriptomic profile (Fig. 1f).
ZED1227 sustains molecularly assessed intestinal functions
An analysis of the expression data of the 95 DEGs individually after the gluten challenge in the placebo group showed that the expression levels correlated with the VH:CrD ratio (Fig. 2a). Reactome enrichment analysis showed that genes involved in the cellular response to interferon (IFN) signaling, both type 1 (IFNα/IFNβ) and type 2 (IFNγ), were upregulated and overrepresented in the gluten-induced gene expression profile (Fig. 2b, left, and Supplementary Data 2). Transcription motif analyses also indicated that genes harboring motifs for transcription factors transducing IFN signaling (for example, STAT1, RELA and IRF1) were significantly present (Supplementary Fig. 2). Notably, a reactome enrichment comparison of the DEGs in the PGCp versus PGCd groups revealed that the type 2 IFNγ signaling term was no longer statistically significant (Fig. 2b, right, and Supplementary Data 2). Similarly, the Gene Ontology term analyses showed that IFN-mediated inflammatory signaling was enriched in the gluten-induced gene expression profile (Fig. 2c and Supplementary Data 2).
a, Heat map of the 95 DEGs in the PGCp versus GFDp comparison. Samples are ordered by increasing VH:CrD ratio, as depicted in the scatter charts above the heat map (GFDd n = 34; GFDp n = 24; PGCd n = 34; PGCp n = 23). Genes are clustered according to Gene Ontology annotation. The z-score of normalized expression is plotted; OBP, other biological processes. b, Bar plot showing enriched Reactome terms of DEGs in the PGCp group relative to the GFDp and PGCd groups. Enriched terms were determined by overrepresentation analysis. P values were calculated by hypergeometric distribution (one-tailed test) and adjusted for multiple testing using the Benjamini–Hochberg method. Reactome terms with an FDR of <0.05 (–log10 (FDR) > 1.3) were considered enriched. Green and gray dots denote significant and nonsignificant FDRs, respectively. c, Bar plots showing Gene Ontology biological process overrepresentation of DEGs in the PGCp group relative to the GFDp and PGCd groups. A Fisher’s exact overrepresentation test (one tailed) was used to find enriched categories. The obtained P values were adjusted for multiple testing using the Benjamini–Hochberg method. Gene Ontology terms with an FDR of <0.05 (–log10 (FDR) > 1.3) were considered enriched. Green and gray dots denote significant and nonsignificant FDRs, respectively. d, GSZ score analyses were performed for categories including transit-amplifying cells, mature enterocytes, immune cells and duodenal transporters and are presented as box plots, with center lines representing the median, the box boundaries representing the interquartile range and the whiskers representing the minimum and maximum values. Values from individual participants are shown (GFDd + p n = 58; PGCd n = 34; PGCp n = 23). GSZ scores were compared among groups using asymptotic P value estimation, with statistical significance defined as a P value of <0.05 (transit-amplifying cells: GFDd + p–PGCd P = 0.3, PGCp–GFDd + p P = 0.03, PGCd–PGCp P = 0.004; mature enterocytes: GFDd + p–PGCd P = 0.3, PGCp–GFDd + p P = 0.005, PGCd–PGCp P = 5.35 × 10−4; immune cells: GFDd + p–PGCd P = 0.73, PGCp–GFDd + p P = 0.02, PGCd–PGCp P = 0.03; duodenal transporters: GFDd + p–PGCd P = 0.53, PGCp–GFDd + p P = 0.02, PGCd–PGCp P = 0.009).
As gluten challenge impairs enterocyte differentiation and absorptive functions and increases inflammation, we analyzed how ZED1227 protects these cellular processes. Gene sets were formed based on human duodenal single-cell RNA-sequencing data22. Gene set z (GSZ) scores23 were calculated for each sample. Samples in the PGCd group demonstrated the same GSZ score levels in the categories of transit-amplifying cells, mature enterocytes, immune cells and duodenal transporters as samples in the pooled GFDd and GFDp groups (GFDd + p) group (Fig. 2d). Importantly, the PGCp group was consistently significantly different from the PGCd group, indicating that ZED1277 efficiently sustained intestinal functions to a level similar to that observed in individuals in the GFDd + p group. Bulk RNA-sequencing deconvolution that used duodenal single-cell RNA-sequencing data as a reference revealed similar patterns in cell proportion distributions, like a decrease in enterocyte numbers accompanied with a small increase in stem and Paneth cell numbers in the PGCp group (Supplementary Fig. 3a,b). At the same time, markers for cytotoxic intraepithelial lymphocytes (IELs) seemed to not be altered (except HLA-E) by placebo and drug treatment (Supplementary Fig. 3c), probably because of underrepresentation of these cell types in biopsy samples.
ZED1227 can halt the IFNγ response
Reactome and Gene Ontology enrichment analyses (Fig. 2b,c) indicated that IFN signaling was one of the most significantly affected pathways in the gluten challenge. Interestingly, a 100-mg dose of ZED1227 for 6 weeks seemed somewhat insufficient in decreasing the IFNγ response, at least according to the Reactome enrichment analysis (Fig. 2b). We decided to set up an intestinal epithelium-specific IFNγ response gene set to assess how well ZED1227 could inhibit inflammation using an epithelial-specific IFNγ response as a gauge. Human intestinal organoids composed of pure intestinal epithelium were treated with IFNγ, and a DEG set was analyzed against the DEGs induced by gluten challenge. We found that nearly half (43 of 95) of the gluten-induced gene expression changes in CeD were associated with the epithelial response to IFNγ (Fig. 3a and Supplementary Data 3). The GSZ scores calculated based on these 43 genes showed that, on average, ZED1227 inhibited the epithelial IFNγ response, as participants in the PGCd group had significantly lower GSZ scores than participants in the PGCp group (Fig. 3b). However, when the GSZ scores of the PGCd and GFDd + p groups were compared, there was a slight but statistically significant difference. This suggests that either there was a residual IFNγ response in all/many participants in the PGCd group or ZED1227 was not able to inhibit the IFNγ response completely in some individuals. When GSZ scores were calculated for each sample, it was evident that some individuals (4 of 34 participants in the PGCd group) still had an active epithelial IFNγ response even after the high-dose (100-mg) ZED1227 treatment for 6 weeks (Fig. 3c).
a, Venn diagram of all DEGs in human duodenal organoids (n = 3) after a 24-h treatment with 100 U ml–1 IFNγ (violet sphere) and PGCp versus GFD (orange sphere) comparisons. b, GSZ score analyses for the epithelial IFNγ-related gene set (GFDd + p n = 58; PGCd n = 34; PGCp n = 23). The box plot center lines represent the median, the box boundaries represent interquartile range, and the whisker length represents the minimum and maximum range. Values from individual participants are shown. GSZ scores were compared among groups using asymptotic P value estimation, with statistical significance defined as a P value of <0.05 (GFDd + p–PGCd P = 0.05, PGCp–GFDd + p P = 6.07 × 10−6, PGCd–PGCp P = 1.24 × 10−4). c, Bar plot of epithelial IFNγ-related GSZ scores calculated for each sample. The dashed lines represent the threshold, outside of which the gene set was considered to be ‘on’ or ‘off’. The yellow bar below illustrates the samples in which the epithelial IFNγ-related GSZ scores were on and off (GFDd n = 34; GFDp n = 24; PGCd n = 34; PGCp n = 23). d, Expression of TGM2 mRNA in the GFDd, GFDp, PGCd and PGCp groups. The box plot center lines represent the median, the box boundaries represent interquartile range, and the whisker length represents the minimum and maximum range. Values from individual participants are shown. Likelihood ratio test (LRT) P values were calculated using DESeq2, with P values representing adjusted values for multiple testing using the Benjamini–Hochberg method (FDR; GFDd n = 34; GFDp n = 24; PGCd n = 34; PGCp n = 23). e, Expression of TGM2 mRNA in human duodenal organoids (n = 3) treated with 100 U ml–1 IFNγ (I) or mock treated (M) for 24 h. The box plot center lines represent the median, the box boundaries represent interquartile range, and the whisker length represents minimum and maximum range. Values from individual participants are shown. LRT P values were calculated using DESeq2, with P values representing adjusted values for multiple testing using the Benjamini–Hochberg method (FDR; P = 9.48 × 10−17). f, Correlation plot for TGM2 mRNA expression and epithelial IFNγ-related GSZ scores. Each dot represents an individual participant with CeD after gluten challenge. Pearson correlation coefficient values (R) are presented, and the P value (P) was calculated based on the t-distribution under the null hypothesis of no correlation using a two-tailed test; P = 5.57 × 10−8.
IFNγ has been shown to induce TG2 activity in intestinal epithelial cancer cells, and this has been suggested to contribute to CeD pathogenesis24. Similarly, participants in the placebo group after the gluten challenge and concomitant IFNγ response had significantly higher expression of TGM2, whereas in participants treated with ZED1227, TGM2 was expressed at a level similar to that observed in participants in the GFDd group (Fig. 3d). Overproduced interleukin-21 (IL-21) in CeD is known to sustain IFNγ production25, and we also detected an induction in the IL-21 signaling pathway in participants in the PGCp group (Supplementary Fig. 4a,b), but this was not statistically significant. We also found that the expression of TGM2 was positively correlated (R= 0.65) with the epithelial IFNγ response (Fig. 3f). Direct causality was further proven by treating human intestinal duodenal organoids with IFNγ, which resulted in a significant induction of TGM2 mRNA expression (Fig. 3e) that could not be inhibited with ZED1227 treatment (Supplementary Fig. 4c). IFNγ treatment induced TG2 activity in Caco-2 cells, which was inhibited by ZED1227 to the level observed following mock treatment (Supplementary Fig. 4d). These observations could be explained by ZED1227 cell impermeability26 and its binding mainly to enterocyte luminal surfaces27.
ZED1227 prevents activation of gluten-induced immunological pathways
As gluten challenge caused a significant IFNγ response and concomitant upregulation of TGM2 expression and activity, we analyzed gluten challenge-induced immunological pathway alterations and how ZED1227 can inhibit them. Peroxisome proliferator-activated receptor-γ (PPARγ) has been shown to transrepress inflammatory responses28,29. PPARγ is downregulated in celiac mucosa30, and this has been shown to be mediated by TG2 and gliadin31. We also found that PPARG gene expression (Fig. 4a) and the corresponding signaling pathway (Fig. 4b) are significantly less active after gluten challenge in the PGCp group than in the GFD and PGCd groups. We also observed a negative correlation between the expression of TGM2 and PPARG and the expression of PPARG and IEL count (Fig. 4c). This suggests that the mucosal inflammatory response, kept in check by PPARγ, is lifted during the gluten challenge in CeD, and this can be prevented with ZED1227 treatment.
a, Expression of PPARG mRNA in the GFDd, GFDp, PGCd and PGCp groups. LRT P values were calculated using DESeq2, with P values representing adjusted values for multiple testing using the Benjamini–Hochberg method (FDR; GFDd n = 34; GFDp n = 24; PGCd n = 34; PGCp n = 23). b, GSZ score analyses for the PPAR signaling pathway from the KEGG database gene set. GSZ scores were compared among groups using asymptotic P value estimation, with statistical significance defined as a P value of <0.05 (GFDd + p n = 58; PGCd n = 34; PGCp n = 23). c, Correlation plots for TGM2 mRNA expression (top) and IEL density (number of CD3+ cells per 100 enterocytes; bottom) against PPARG mRNA expression. Each dot represents an individual participant with CeD after gluten challenge. The Pearson correlation coefficient (R) is presented, and the P value (P) was calculated based on the t-distribution under the null hypothesis of no correlation using a two-tailed test (TGM2 mRNA expression versus PPARG mRNA expression, P = 1.14 × 10−5; IEL density versus PPARG mRNA expression, P = 2.95 × 10−7). d, Expression of NOS2 mRNA in the GFDd, GFDp, PGCd and PGCp groups. LRT P values were calculated using DESeq2, with P values representing adjusted values for multiple testing using the Benjamini–Hochberg method (FDR; GFDd n = 34; GFDp n = 24; PGCd n = 34; PGCp n = 23). e, GSZ score analyses for selected KEGG, BIOCARTA and Reactome database gene sets. GSZ scores were compared among groups using asymptotic P value estimation, with statistical significance defined as a P value of <0.05 (GFDd + p n = 58; PGCd n = 34; PGCp n = 23). The box plot center lines represent the median, the box boundaries represent interquartile range, and the whisker length represents minimum and maximum range. Values from individual participants are shown. f, Heat map for selected CeD-specific immune cell marker genes detected in Atlasy et al.58. Samples are ordered by increasing IEL density, as depicted in the scatter charts above the heat map (GFDd n = 34; GFDp n = 24; PGCd n = 34; PGCp n = 23). The z scores of normalized expression are plotted; PC, plasma cells; Inf-MF, inflammatory macrophages.
PPARγ inhibits the expression of proinflammatory cytokines, and it also silences inducible nitric oxide (NO) synthase (iNOS/NOS2)32. NOS2 is induced in the mucosa of individuals with active CeD mainly in macrophages and enterocytes33,34,35, leading to a systemic increase of NO in the plasma36.
NO is needed for the responsiveness of natural killer (NK) cells to the NK cell-activating factor IL-12, which stimulates cytotoxicity and IFNγ release37. Our data show that ZED1227 can inhibit gluten challenge-induced NOS2 upregulation (Fig. 4d), resulting in overrepresentation of gene sets involved in the NO–IL-12 and NK cell-mediated cytotoxicity (Fig. 4e) pathways. Also, pathways to antigen presentation and IgA production are normalized following ZED1227 treatment (Fig. 4e). Analysis of the expression of immunological cell gene markers showed that ZED1227 inhibits the infiltration of cell types (especially CD8+ T cells, plasma cells, NK cells and macrophages) involved in the aforementioned inflammatory responses (Fig. 4f).
The effect of HLA-DQ genetic background on treatment outcomes
The fact that some participants treated with ZED1227 in the PGCd group still showed a significant epithelial IFNγ response (Fig. 3c), as a sign of active residual CeD pathophysiology prompted us to study factors behind the incomplete response to treatment. To this end, we performed high-resolution genotyping for HLA class II DQ alleles using the arcasHLA tool38 from aligned sequences obtained from genome-wide 3′ RNA-sequencing data. Five participants had too low coverage either at the HLA-DQB1 or HLA-DQA1 locus, according to RNA sequencing; thus, their allele typing was performed from blood samples collected at the on study inclusion. One participant from the placebo group, however, failed during identification. This participant is marked as ‘not identified’ in Table 1 and was excluded from subsequent analyses.
It is known that HLA-DQ2 gene dose correlates with the strength of the gluten-specific T cell response20; thus, all obtained genotypes were divided into groups by their potential effectiveness in binding and presenting gliadins to T cells39,40. We were able to divide participants into three groups according to their HLA-DQ genotypes, with G1 being the high-gluten-response group and G3 being the low-gluten-response group (Table 1). However, one should note that the group sizes are relatively small.
When examining the changes in mean VH:CrD ratio within genotype groups over time (Fig. 5a), it is evident that the groups exhibit different trajectories of change. Notably, the slope of the G1 group appears to deviate the most from the parallel pattern among the groups for both drug and placebo treatments.
a, The VH:CrD ratio remains higher in the drug group than in the placebo group, regardless of the genotype. Participants (n = 57) were divided into two groups according to the treatment received (drug or placebo). The VH:CrD ratio at PGC is shown as mean ± s.d. b, A two-way ANCOVA was performed with the VH:CrD ratio at PGC as a dependent variable, the VH:CrD ratio at GFD as a covariate and treatment (drug n = 34 and placebo n = 23) and HLA-DQ genotype group (G1, G2 and G3) as independent variables (ANCOVA, F2,50 = 2.2, P = 0.12). Post hoc pairwise multiple comparisons were performed between the drug and placebo groups among HLA-DQ genotype groups. The VH:CrD ratio at PGC is shown as estimated marginal means ± 95% confidence interval (95% CI; drug G1 n = 6; drug G2 n = 14; drug G3 n = 14; placebo G1 n = 2; placebo G2 n = 6; placebo G3 n = 15). c, The G1 genotype group showed weaker recovery after ZED1227 treatment as assessed by VH:CrD ratio. Participants (n = 34) belonging to the drug group were selected for one-way ANCOVA. The VH:CrD ratio at PGCd was used as a dependent variable, and the VH:CrD ratio at GFD was used as a covariate; HLA-DQ genotype group (G1, G2 and G3) served as independent variables (ANCOVA, F2,30 = 5.11, P = 0.012). Post hoc pairwise multiple comparisons were performed between HLA-DQ genotype groups, with P values adjusted by Bonferroni correction. Results are shown as estimate ± 95% CI. d, A two-way ANCOVA plot examining the effects of treatment and HLA-DQ genetic background on PGC epithelial response to IFNγ GSZ score (ANOVA, F2,49 = 0.07, P = 0.93). The epithelial response to IFNγ GSZ score at PGC is shown as estimated marginal means ± 95% CI (drug G3 versus placebo G3 P = 5.50 × 10−4; drug G1 n = 6; drug G2 n = 14; drug G3 n = 14; placebo G1 n = 2; placebo G2 n = 6; placebo G3 n = 15). e, Expression of enterocyte- (APOB, APOA1 and TMSF4), proliferation- (AGR2, MKI67 and CENPF) and inflammation-related (STAT1, GBP1, TGM2, CIITA, PPARG and NOS2) marker genes. Expression is shown as counts grouped by HLA-DQ genotype group (G1, G2 and G3) and are presented as mean (spheres) and s.d. (vertical lines; PGCd G1 n = 6; PGCd G2 n = 14; PGCd G3 n = 14; PGCp G1 n = 2; PGCp G2 n = 6; PGCp G3 n = 15).
The impact of treatment on VH:CrD ratio within different time points (GFD and PGC) across HLA-DQ genetic background groups (G1, G2 and G3) was assessed by fitting repeated-measures analysis of variance (ANOVA). In the placebo group, the interaction term between time point and HLA-DQ genetic groups was statistically significant (P = 0.003; Table 2 and Methods), indicating that HLA-DQ genetic background has an impact on changes in VH:CrD ratio over the course of gluten challenge (Methods). For the drug group, however, the interaction term was not significant (P = 0.06; Table 2 and Methods), suggesting that the drug appears to be effective in reducing the impact of gluten across all genotype groups. However, pairwise comparisons (Table 2 and Methods) performed for the drug group showed that the impact of HLA-DQ genetic background is statistically significant for the G1 group (P = 0.05) and not significant for the G2 (P = 0.07) and G3 groups (P = 0.39).
Given the notable drop in the VH:CrD ratio after ZED1227 treatment in the high-gluten-response genotype group (G1), we analyzed the efficacy of treatments in each genotype group. A two-way analysis of covariance (ANCOVA) was performed to examine the effects of treatment and HLA-DQ genetic background on VH:CrD ratio at PGC. After adjustment for the VH:CrD ratio at GFD, there was no statistically significant interaction between treatment and the HLA-DQ genotype group on the histomorphometry parameters (Methods), and pairwise multiple comparisons show significant differences between the PGC VH:CrD means in all genotype groups between participants receiving drug or placebo (Fig. 5b). This suggests that, despite a substantial decrease in VH:CrD ratio after gluten challenge in the G1 group for participants treated with drug, the VH:CrD ratio was still higher in the drug group than in the placebo group, irrespective of the genotype.
The estimated difference in the VH:CrD ratio for participants treated with drug belonging to the G3 genotype versus the G1 genotype was −0.52 (95% CI of −0.86 to −0.19) with an adjusted P value of 0.01, as assessed by fitting a one-way ANCOVA model. Other estimated differences (G3–G2 and G2–G1) were not significant but showed the tendency of group G2 having the intermediate position between G1 and G3, when judging by VH:CrD ratio (Fig. 5c). Interestingly, the G1 high-risk genotype specifically affected VH and not CrD (Extended Data Fig. 2a,b).
The CeD pathophysiological epithelial IFNγ response was studied with a two-way ANCOVA statistical analysis, and pairwise comparisons showed that participants in the PGCd and G1 genotype groups still had an active IFNγ response and did not statistically differ from the placebo group (Fig. 5d). In fact, in the bar plot presenting four participants in the PGCd group with an IFNγ response in Fig. 3c, three of these participants had the high-gluten-response genotype homozygous HLA-DQ2.5 and one had homozygous HLA-DQ8 associated with an intermediate response to gluten.
The inclination of the G1 group to be highly responsive to gluten and less reactive to ZED1227 was also observed at individual gene expression levels. Reduced expression of enterocyte marker genes (APOB, APOA1 and TM4SF4) and increased expression of proliferation markers (AGR2, MKI67 and CENPF) were observed in participants with G1 genotypes in both the ZED1227- and placebo-treated groups (Fig. 5e). Inflammation-related genes (STAT1, GBP1 and TGM2) showed lower expression in PGCd samples with G2 and G3 genotypes, suggesting that they were more susceptible to ZED1227 treatment. In accordance with the higher residual CeD-associated epithelial IFNγ response in participants in the PGCd and G1 groups (Figs. 3b,c and 5d), these inflammatory genes were more highly expressed in participants treated with either placebo or drug within the genotype group G1. Furthermore, ZED1227 was less able to prevent gluten challenge-induced attenuation of PPARγ-mediated inhibition of NOS2 expression, as the expression of these genes was at the same level in G1 genotypes in the PGCd group as in the G2 and G3 genotypes in the PGCp group (Fig. 5e). Also, the HLA class II transcriptional coactivator CIITA was more highly expressed in individuals with the G1 genotype in the PGCd group (Fig. 5e). Moreover, the G1 group was identified as more pathognomonic when its GSZ scores for ‘transit-amplifying cells’, ‘mature enterocytes, ‘immune cells’ and ‘duodenal transporters’ were assessed (Extended Data Fig. 2c). In addition to IFNγ signaling, molecular pathways for PPAR and lipid signaling seemed to also be affected in the G1 group (Extended Data Fig. 2c).
Molecular histomorphometric analysis of ZED1227 efficacy
We previously created a molecular histomorphometric model to assess gluten-dependent morphological deterioration and healing in the duodenum, that is, VH:CrD, in gene transcriptomic terms13. This model is based on the expression of four genes (ATP8B2, PLA2R1, PDIA3 and TM4SF4), which we showed is significantly correlated with the extent of gluten-induced histological damage13. Scatter plots and partial regression plots for these genes showed that the relationship between gene expression and VH:CrD ratio was linear, and participants in the PGCp group tended to separate from participants in the GFD and PGCd groups (Fig. 6a). Moreover, a comparison of traditional and molecular histomorphometry in the regression scatter plot revealed a high coefficient of determination (R2 = 0.86), indicating that the previously developed molecular histomorphometric tool was able to reliably estimate VH:CrD ratios in this independent study cohort (Fig. 6b). Finally, box plot comparisons of groups with histomorphometric and molecular histomorphometric values indicated that ZED1227 efficiently inhibited gluten-induced mucosal damage in individuals with CeD (Fig. 6c).
a, Scatter plots and regression plots for VH:CrD prediction model genes. A linear regression with 95% CI is shown. Each dot represents an individual participant (GFDd + p n = 58; PGCd n = 34; PGCp n = 23); UMI, unique molecular identifiers. b, Observed versus predicted regression scatter plot for the model predicting VH:CrD. Each dot represents an individual participant (n = 115). A linear regression with 95% CI is shown; R2 = 0.86, F1,114 = 691.6, P < 0.001. The red dashed line represents the ideal regression case, where x = y; r.m.s.d., root mean square deviation. c, Box plot comparisons of groups with histomorphometry (measured VH:CrD) values and molecular histomorphometry (regression model based on RNA expression) values. The box plot center lines represent the median, the box boundaries represent interquartile range, and the whisker length represents minimum and maximum range. Values from individual participants are shown. Two-tailed unpaired Student’s t-tests were used for the PGCd versus PGCp group comparisons (VH:CrD observed: PGCd–PGCp P = 6.41 × 10−4; VH:CrD predicted: PGCd–PGCp P = 4.04 × 10−5; GFDd n = 34; GFDp n = 24; PGCd n = 34; PGCp n = 23).
Discussion
The ability of the TG2 inhibitor ZED1227 (ref. 26) to attenuate gluten-induced mucosal damage was previously reported in a proof-of-concept, randomized, double-blind, placebo-controlled 6-week trial with a daily 3-g gluten challenge19. TG2, the celiac autoantigen14, has a pivotal role in gluten-induced pathogenesis, leading to small intestinal mucosal injury with villus atrophy and crypt hyperplasia, the histological hallmarks of untreated CeD. Here, we sought to assess the efficacy of ZED1227 in preventing gluten-induced mucosal damage at the transcriptomic level. Remarkably, a 100-mg daily dose of ZED1227 inhibited virtually all gluten-induced transcriptomic changes (Fig. 1b,c). Active CeD is accompanied by compromised enterocyte maturation, crypt hyperplasia due to the expansion of transit-amplifying cells41,42,43, immune cell infiltration44,45 and decreased expression of duodenal transporters13,46,47. GSZ23 scores based on published single-cell databases22 clearly indicated that TG2 inhibition efficiently blocked all aforementioned gluten-induced intestinal manifestations in individuals with CeD (Fig. 2d). Our recently published molecular histomorphometry regression model based on genome-wide transcriptomics analysis13 was validated in this independent study sample. We showed a significant accordance between this new molecular tool and the traditional, more subjective biopsy-based microscopic histomorphometry reading. Overall, our transcriptomic findings strongly support the results of the clinical trial with ZED1227, which demonstrated that the inhibition of TG2 activity can efficiently and specifically prevent gluten-induced mucosal damage19. Our data also corroborate the previous findings that gliadin together with active TG2 induces attenuated PPARγ activity, which, together with a concomitant increase in IFNγ, lead to increased mucosal NO production and inflammation30,31,33,34,35,36. We show here that by inhibiting the gliadin deamidation activity of TG2, all these pathogenic immunological changes in CeD can be prevented (Fig. 4). In addition, studies have shown that gluten-derived peptides may have innate immune stimulatory properties, outside the realm of adaptive immunity, which can lead to epithelial stress in CeD48,49. Our data show, however, that halting the adaptive immunity pathway in CeD pathogenesis is sufficient to prevent gluten-induced mucosal damage, as we did not detect any molecular traces of mucosal damage remaining after ZED1227 treatment.
Gene Ontology and Reactome analyses indicated that gluten challenge most significantly affected genes related to the immune response, especially IFN-mediated defense mechanisms (Fig. 2b,c). This is in agreement with previously published transcriptomic analyses of individuals with active CeD compared to individuals on a GFD or healthy individuals47,50,51. Notably, IFNγ secreted by gluten-reactive T cells in the celiac intestine induces TG2 expression and secretion and thus favors the pathogenic autoamplificatory loop of enhanced gluten deamidation by TG2, improved antigenic presentation on HLA-DQ2 or HLA-DQ8 and subsequent gluten-specific T cell activation24. The present study confirms the prominent role of IFN signaling in CeD pathogenesis, in line with findings that nearly half of the gluten-induced gene expression changes in duodenal biopsies can be recapitulated in human intestinal epithelial organoids treated with IFNγ (Fig. 3a). We also detected the suggested autoamplificatory loop in our human data, as TGM2 expression positively correlated with the epithelial IFNγ response (Fig. 3f). Notably, TGM2 expression was induced by IFNγ in human intestinal organoids ex vivo (Fig. 3e), suggesting mutual amplification between these two key players in CeD pathogenesis. The functional relevance of this amplification loop was indeed confirmed in the clinical study in which the inhibition of TG2 activity by ZED1227 in individuals with CeD significantly inhibited both the (epithelial) IFNγ response (Fig. 3b) and TGM2 expression (Fig. 3d), resulting in protection from villous atrophy and intraepithelial lymphocytosis (Fig. 2d).
However, even though TG2 inhibition exhibited significant efficacy, according to a comparison of the transcripts of the placebo/gluten challenge and the gluten challenge/ZED1227-treated group, which showed a transcriptome profile similar to that of the GFD groups, we detected heterogeneity regarding the gluten-induced and IFNγ-dependent cascade of pathogenic events among ZED1227-treated and gluten-challenged individuals with CeD. Four of these individuals still had a modestly active IFNγ response, and the majority (three of four) belonged to the HLA-DQ2.5 homozygous genotype (Fig. 3c). HLA-DQ2.5 homozygous individuals have a fivefold higher risk of developing CeD than HLA-DQ2.5 heterozygous individuals21, which has been linked to the more efficient presentation of deamidated gluten peptides to gluten-specific T cells20. Moreover, homozygosity for HLA-DQ2 predisposes individuals to developing more rapid and severe villous atrophy52 and is associated with malignant complications, such as refractory CeD type 2 and enteropathy-associated T cell lymphoma53,54. Along this line, we also found that individuals belonging to the high-gluten-response HLA-DQ genotype group (G1) were more sensitive to gluten, as their VH:CrD ratios dropped significantly more than individuals belonging to the mid- and low-gluten response groups (G2 and G3) during the gluten challenge, both in the placebo and drug groups (Tables 1 and 2 and Fig. 5a,b). Thus, even after drug treatment, VH:CrD decreased significantly in the G1 versus G2 and G3 genotype groups after the gluten challenge. This was also evident when molecular histomorphometric features were assessed (Extended Data Fig. 2c). We also discovered that PPAR signaling and lipid metabolism, previously reported to be dysregulated in CeD30, were less controlled in the G1 group (Extended Data Fig. 2c). As IFNγ is known to inhibit PPAR and lipid metabolism55, it is conceivable that these are consequences of the overactive IFNγ response in individuals in the G1 group. Nevertheless, duodenal mucosal morphology and, especially, intraepithelial lymphocyte infiltration were significantly healthier in the ZED1227-treated group than in the placebo group, indicating that participants with G1 phenotypes may benefit from a higher dose and/or prolonged treatment with ZED1227. We suggest that the ZED1227 therapy program should include a personalized medicine approach in which HLA-DQ stratification is combined with TG2 dose adjustments, which may lead to an optimal treatment response and a more thorough abrogation of IFNγ-induced mucosal damage. According to our transcriptomic analysis of human intestinal organoids, ZED1227 does not appear to induce significant transcriptomic changes in the organoid model (Supplementary Fig. 5), consistent with the clinical safety observed in the phase 2 challenge study19.
We recognize the limitations of this study. The cohort is relatively modest and characterized by an uneven distribution of HLA-DQ genotypes. This resulted in small G1 subgroups within both the drug and placebo cohorts, which may have implications for statistical power and the generalizability of our results and warrants further corroborative studies. Additionally, we only had one dose of the drug available for this study. The transcriptomic analysis was conducted as an optional component of the study, and RNA isolation was not performed for all drug groups. This decision was made to focus our efforts on the drug group that showed the most significant improvement compared to the placebo group, allowing us to investigate potential transcriptomic changes effectively within the study’s scope.
In conclusion, the strategy to inhibit TG2 activity as a key upstream effector in gluten-induced immune activation in CeD, which has been proven efficient in the clinical study, was mechanistically buttressed by our transcriptomic analysis of the duodenal biopsies of individuals treated or not treated with ZED1227. Importantly, TG2 inhibition prominently prevented the gluten-induced IFNγ response and further downstream pathways that lead to mucosal inflammation, remodeling and villous atrophy. Our analysis also suggests that, based on HLA-DQ2.5 genetics, the dose or dose interval of ZED1227 may have to be adjusted for optimal efficacy, but larger sample sizes are required to confirm this assumption. Moreover, CeD-associated gene expression changes were observable, even on a strict GFD13,56, indicating that complete avoidance of gluten is impossible5,6. In fact, a recent meta-analysis found that 15% of foods labeled as gluten free and 28% labeled as naturally gluten free contained more than 20 mg kg–1 gluten57, the cutoff for qualifying as gluten free. Thus, an adjunctive TG2 inhibition-based therapy combined with a GFD would especially benefit highly gluten-sensitive individuals (possibly carrying a homozygous HLA-DQ genotype) by providing protection against intestinal damage that can occur even in a low-gluten environment.
Methods
Participants and biopsies
PAXgene-fixed and paraffin-embedded biopsies were collected from a multisite, double-blind, randomized, placebo-controlled trial aimed at dose finding and assessing the efficacy and tolerability of a 6-week treatment with ZED1227 capsules versus placebo in individuals with well-controlled CeD undergoing gluten challenge59. Full inclusion and exclusion criteria are published19. Briefly, participants who had a biopsy-proven CeD diagnosis, were on a self-reported strict GFD for at least 1 year and symptom free, showed normalized duodenal histology compared to the initial diagnostic biopsy finding (morphometrically defined as a mean VH:CrD of 1.5 or higher) and tested negative for serum anti-TG2 on study inclusion were included (GFD group; Extended Data Table 1). These participants then underwent a challenge with a cookie containing 3 g of gluten daily for 6 weeks (PGC group). At least 80% compliance was confirmed19.
Biopsy sampling was performed twice on study inclusion (denoted here as GFD) and at the final visit (denoted here as PGC; Extended Data Fig. 1). Duodenal forceps biopsies were immersed in PaxFPE (PAXgene fixative) and processed for paraffin block embedding using a standard formalin-free paraffin-infiltration protocol. For morphology, samples were stained with hematoxylin and eosin and measured using our validated morphometry rules separately for morphology (VH, CrD and VH:CrD)60.
This study used samples from two groups, placebo and the 100-mg ZED1227 group, which represented the highest dose drug group showing the most significant improvement compared to the placebo group. In total, 58 participants (drug group, n = 34; placebo group, n = 24; total number of biopsies = 116) of the 68 participants who had sufficient biopsy samples at both time points in the original trial19 were included, as these exploratory (optional) studies required separate written informed consent. Demographic characteristics and duodenal histomorphometry changes in the form of VH:CrD ratio of the participants in the original cohort and in the present study are presented in Supplementary Tables 1 and 2.
Human organoid cultures
Human duodenal tissues for establishing organoid cultures used in this study were sourced from deidentified surgical specimens (n = 3) of the duodenum obtained from participants who had undergone biopsy procedures unrelated to CeD at Tampere University Hospital. The protocol was approved by the Ethics Committee of Tampere University Hospital (ETL code R18082). Intestinal crypts containing stem cells were isolated following 2 mM EDTA dissociation of tissue samples for 30 min at 4 °C (ref. 61). Crypts were washed in PBS, and fractions enriched in crypts were collected. The supernatant was removed, and the crypt epithelial cells were seeded in 50% Matrigel (diluted with basal culture medium). Crypts were passaged and maintained in WELR500 culture medium, as previously described62. Organoids were treated with 100 U ml–1 IFNγ (Peprotech, 300-02) with or without 50 µM ZED1227 (Zedira) for 24 h and subjected to RNA sequencing to assess any adverse direct side effects to the intestinal epithelium (Supplementary Fig. 5).
Cell culture and treatments
Caco-2 colonic epithelial cells (ATCC, HTB-37; passage 22–35) were grown as standard monolayers in tissue culture flasks in complete MEM 1 g l–1 glucose medium (20% heat-inactivated fetal bovine serum, 1% nonessential amino acids, 1% penicillin–streptomycin, 1% GlutaMAX and 1% sodium pyruvate) at 37 °C in a 5% CO2 atmosphere. Caco-2 cells were treated with 100 U ml–1 IFNγ (Peprotech, 300-02) with or without 50 µM ZED1227 (Zedira) or mock treated with DMSO for 24 h. Cells were collected by trypsinization and lysed in lysis buffer (50 mM Tris (pH 8.0), 150 mM NaCl and 1% IGEPAL) supplemented with 0.2 mM DTT and 1× Complete Protease Inhibitor Cocktail (Roche, 11836170001) and used for the transglutaminase activity assay.
RNA extraction and RNA sequencing
Total RNA was extracted from the PaxFPE-fixed biopsy specimens (n = 116)63 using additional cuttings from the samples on which histomorphometry was previously assessed19. For extraction, an RNeasy kit (Qiagen) was used according to the manufacturer’s instructions. Library preparation and NGS were performed by the Qiagen NGS Service. A total of 10 ng of purified RNA was converted into NGS cDNA libraries. Library preparation was quality controlled using capillary electrophoresis. Based on the quality of the inserts and the concentration measurements, the libraries were pooled in equimolar ratios and sequenced on a NextSeq (Illumina) sequencing instrument according to the manufacturer’s instructions, with 100-bp read length for read 1 and 27-bp read length for read 2. The raw data were demultiplexed, and FASTQ files for each sample were generated using bcl2fastq2 software (Illumina).
RNA from the duodenal organoids was isolated using an RNeasy kit (Qiagen) following the manufacturer’s instructions. RNA purity and concentration were measured using a NanoDrop One spectrophotometer (NanoDrop Technologies). Preparation of the RNA library and transcriptome sequencing was conducted by Novogene. mRNA was purified from total RNA using poly(A) selection and subjected to library construction. Sequencing was performed on an Illumina platform, and 150-bp paired-end reads were generated.
Bioinformatic analyses
Data quality was checked using FastQC. The 3′ adapter sequences were trimmed, reads without adapters were kept, and reads with <15 bp were removed. Reads were aligned to the human genome reference consortium human build 38 (GRCh38) using the splice-aware aligner STAR. For all downstream analyses, genes with low expression (read counts that were equal to the number of samples multiplied by 5) were excluded. One sample with low total reads (1.13 million reads) was excluded, leaving 115 samples for subsequent analyses. The mean total reads for all samples were 3.51 ± 0.07 million reads. A secondary differential expression analysis involving normalization of unique molecular identifier counts and a subsequent pairwise differential regulation analysis was performed using the DESeq2 package64. Pre- and post-treatment samples were compared, and the paired nature of samples was included as a term in the multifactor design formula. The obtained P values were adjusted for multiple testing using the Benjamini–Hochberg method65. Genes with an FDR of <0.05 and | log2 (FC) | of ≥0.5 identified by DESeq2 were assigned as differentially expressed.
Gene Ontology enrichment and Reactome enrichment analyses were performed using topGO66 and ReactomePA67 R packages. GSZ scores, as a particular type of overrepresentation analysis, were calculated as previously described68. For comparison of groups, mean GSZ score asymptotic P value calculation was applied to our datasets69. Gene lists for transit-amplifying cells, mature enterocytes, immune cells and duodenal transporters were retrieved from healthy human duodenal single-cell sequencing analyses published by Busslinger et al.22 or our DEG analysis from human duodenal organoids treated with IFNγ versus mock-treated organoids. Cell-type proportions for CeD biopsy bulk RNA-sequencing data were estimated with the MuSiC analysis toolkit70 using single-cell RNA-sequencing data from duodenal adult biopsies71 as a reference.
Exact HLA genotypes, with a focus on DQ status (HLA-DQA1 and HLA-DQB1 alleles), were determined in silico from RNA-sequencing data using the arcasHLA tool38. FASTQ files were used as input files. The minimum gene read count required for genotyping was set at 5. Due to low expression, low resolution72 (Field1, allele group) was taken into consideration in the subsequent statistical analyses.
Statistical analysis
Statistical tests were conducted as specified in the legends of the respective figures using R version 4.3.0 (R Foundation for Statistical Computing). A repeated-measures ANOVA was used to assess the impact of treatment on VH:CrD ratio within different time points (GFD and PGC) across HLA-DQ genetic background groups (G1, G2 and G3). This analysis comprised 57 participants with identifiable HLA-DQ genotypes. Three null hypotheses were proposed: (1) VH:CrD means are equal across time points, (2) VH:CrD means are equal among HLA-DQ groups, and (3) there is no interaction between these two factors. As a post hoc analysis, multiple pairwise t-tests were used to identify differences between time points for each genotype group. To assess how the impact of the HLA-DQ genotype group on the VH:CrD outcome varies with different time points, a one-way ANOVA model was used. To address multiple testing, a Bonferroni correction was applied to P values (total tests performed = 2). Statistical significance was determined as P < 0.05.
To assess the interaction between treatment groups and HLA-DQ genetic backgrounds on VH:CrD and epithelial response to IFNγ GSZ score at PGC, a two-way ANCOVA was conducted using these values at PGC as the dependent variable, HLA-DQ genetic background (G1, G2 and G3 genotype groups) and treatment (placebo or drug) as independent variables and baseline VH:CrD ratio and epithelial response to IFNγ GSZ score (from the GFD group), respectively, as a covariate. This analysis included 57 participants, with 1 participant from the placebo group excluded due to an unidentified allele type. The study formulated the following two null hypotheses for the two-way ANCOVA analysis: (1) no VH:CrD (epithelial response to IFNγ GSZ) difference at PCG exists between treatment groups (placebo and drug) while accounting for VH:CrD (epithelial response to IFNγ GSZ) at GFD and (2) no VH:CrD (epithelial response to IFNγ GSZ) differences at PCG exist across HLA-DQ genetic backgrounds (G1, G2 and G3 genotype groups) controlling for VH:CrD (epithelial response to IFNγ GSZ) at GFD. For the one-way ANCOVA, only participants in the drug group (n = 34) were selected. The null hypothesis for this analysis was that there is no significant effect of HLA-DQ genetic background (represented by HLA-DQ genotype groups) on VH:CrD within the PGCd group, while adjusting for VH:CrD at GFDd. The one-way ANCOVA regression model included VH:CrD at PGCd as the dependent variable, VH:CrD at GFDd as a covariate and HLA-DQ genotype group (G1, G2 and G3) as independent variables. The same type of approach was used for VH and CrD values. Post hoc pairwise multiple comparisons using estimated marginal means calculation (also known as least-squares means) were conducted between the drug and placebo groups for the two-way ANCOVA as well as between HLA-DQ genotype groups for the one-way ANCOVA. To address multiple testing, the Bonferroni correction was applied to P values (total tests performed = 3). Statistical significance was defined as an adjusted P value of <0.05.
Quantitative real-time PCR
Human duodenal organoids (n = 3) were treated with 50, 100 or 200 U ml–1 IFNγ (Peprotech) and/or 2, 25 and 50 µM ZED1227 (Zedira) for 24 h. Total RNA was isolated using TRIzol Reagent (15596018), following the manufacturer’s instructions, and 500 ng was subjected to cDNA synthesis using an iScript cDNA Synthesis kit (Bio-Rad). Real-time PCR reactions were performed with SsoFast EvaGreen Supermix (1708890, Bio-Rad) and oligonucleotides for human TGM2 (forward: 5′-TGTGGCACCAAGTACCTGCTCA-3′; reverse; 5′-GCACCTTGATGAGGTTGGACTC-3′) and GAPDH (forward: 5′-GTCTCCTCTGACTTCAACAGCG-3′; reverse: 5′-ACCACCCTGTTGCTGTAGCCAA-3′) in triplicate. The results presented were calculated as fold change to the reference sample (nontreated sample), normalized by housekeeping gene expression (GAPDH) as described in Schmittgen and Livak73. Plot whiskers represent the standard error for mean difference between three independent means.
Transglutaminase activity assays in Caco-2 cells
Transglutaminase activity was measured using a hydroxamate-based colorimetric method modified from Folk and Cole74. In short, each reaction contained 75 mM hydroxylammonium chloride, 30 mM Z-Gln-Gly, 10 mM CaCl2 and 10 mM DTT in 200 mM Tris-HCl buffer (pH 8.0) mixed with cell lysate in a final volume of 100 µl. After a 2-h incubation at 37 °C, the reaction was stopped by the addition of 50 µl of stop buffer (1.67% (wt/vol) FeCl3, 4% (wt/vol) trichloroacetic acid and 4% (vol/vol) HCl). The reaction output was measured at 530 nm, and the activity was expressed as nanomoles of hydroxamate produced in 120 min per milligram of total protein, using l-glutamic acid γ-monohydroxamate for the standard curve.
HLA genotyping
Five participants had too low coverage either at the HLA-DQB1 or HLA-DQA1 locus according to RNA sequencing; thus, their allele typing was not performed. For four of those individuals, blood pellet samples stored at −80 °C were available. DNA was extracted from 100 µl of sample using a QIAamp DNA Blood Mini kit (51104, Qiagen) following the manufacturer’s protocol. HLA-DQB1 and HLA-DQA1 typing was performed at the Immunogenetics Laboratory at the University of Turku, and the method was based on an asymmetrical PCR and a subsequent hybridization of allele-specific probes, as previously described75,76.
Molecular histomorphometry regression model
A regression model predicting VH:CrD ratios, developed in our previous study13, was used on the current dataset. Models were evaluated by observed versus predicted regression.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Data availability
Bulk RNA-sequencing data from participant biopsies and patient-derived intestinal organoids described in this study are available in the European Genome–Phenome Archive under accession numbers EGAS50000000337 and EGAS50000000338. Additional data used in this paper include a full single-cell RNA-sequencing dataset of intestinal regions of adult donors (https://www.gutcellatlas.org/), lists of human duodenal cell types and transporter genes expressed along the upper gastrointestinal tract downloaded from supplementary files included within Busslinger et al.22, lists of immune cell marker genes downloaded from supplementary files included within Atlasy et al.58 and pathway gene sets (Reactome, KEGG and BIOCARTA) downloaded from the Human MSigDB Collections at https://www.gsea-msigdb.org/gsea/msigdb/collections.jsp. Source data are provided with this paper. All other data are present in the article and Supplementary Information or are available from the corresponding author upon reasonable request.
Code availability
Code used in this study is freely available on GitHub at https://github.com/IntestinalSignallingAndEpigeneticsLab/Dotsenko-et-al.-2024.
References
King, J. A. et al. Incidence of celiac disease is increasing over time: a systematic review and meta-analysis. Am. J. Gastroenterol. 115, 507–525 (2020).
Lebwohl, B., Sanders, D. S. & Green, P. H. R. Coeliac disease. Lancet 391, 70–81 (2018).
Leffler, D. A., Green, P. H. R. & Fasano, A. Extraintestinal manifestations of coeliac disease. Nat. Rev. Gastroenterol. Hepatol. 12, 561–571 (2015).
Kahaly, G. J., Frommer, L. & Schuppan, D. Celiac disease and endocrine autoimmunity—the genetic link. Autoimmun. Rev. 17, 1169–1175 (2018).
Syage, J. A. et al. Determination of gluten consumption in celiac disease patients on a gluten-free diet. Am. J. Clin. Nutr. 107, 201–207 (2018).
Silvester, J. A. et al. Most patients with celiac disease on gluten-free diets consume measurable amounts of gluten. Gastroenterology 158, 1497–1499 (2020).
Roos, S., Liedberg, G. M., Hellström, I. & Wilhelmsson, S. Persistent symptoms in people with celiac disease despite gluten-free diet: a concern? Gastroenterol. Nurs. 42, 496–503 (2019).
van Megen, F. et al. High disease burden in treated celiac patients—a web-based survey. Scand. J. Gastroenterol. 56, 882–888 (2021).
Itzlinger, A., Branchi, F., Elli, L. & Schumann, M. Gluten-free diet in celiac disease—forever and for all? Nutrients 10, 1796 (2018).
Daveson, A. J. M. et al. Baseline quantitative histology in therapeutics trials reveals villus atrophy in most patients with coeliac disease who appear well controlled on gluten‐free diet. GastroHep 2, 22–30 (2020).
Kreutz, J. M., Adriaanse, M. P. M., van der Ploeg, E. M. C. & Vreugdenhil, A. C. E. Nutrient deficiencies in adults and children with treated and untreated celiac disease. Nutrients 12, 500 (2020).
Vici, G., Belli, L., Biondi, M. & Polzonetti, V. Gluten free diet and nutrient deficiencies: a review. Clin. Nutr. 35, 1236–1241 (2016).
Dotsenko, V. et al. Genome-wide transcriptomic analysis of intestinal mucosa in celiac disease patients on a gluten-free diet and postgluten challenge. Cell. Mol. Gastroenterol. Hepatol. 11, 13–32 (2021).
Dieterich, W. et al. Identification of tissue transglutaminase as the autoantigen of celiac disease. Nat. Med. 3, 797–801 (1997).
Molberg, Ø. et al. Tissue transglutaminase selectively modifies gliadin peptides that are recognized by gut-derived T cells in celiac disease. Nat. Med. 4, 713–717 (1998).
van de Wal, Y. et al. Selective deamidation by tissue transglutaminase strongly enhances gliadin-specific T cell reactivity. J. Immunol. 161, 1585–1588 (1998).
Lundin, K. E. A. & Sollid, L. M. Advances in coeliac disease. Curr. Opin. Gastroenterol. 30, 154–162 (2014).
Schuppan, D., Junker, Y. & Barisani, D. Celiac disease: from pathogenesis to novel therapies. Gastroenterology 137, 1912–1933 (2009).
Schuppan, D. et al. A randomized trial of a transglutaminase 2 inhibitor for celiac disease. N. Engl. J. Med. 385, 35–45 (2021).
Vader, W. et al. The HLA-DQ2 gene dose effect in celiac disease is directly related to the magnitude and breadth of gluten-specific T cell responses. Proc. Natl Acad. Sci. USA 100, 12390–12395 (2003).
Koning, F. Celiac disease: quantity matters. Semin. Immunopathol. 34, 541–549 (2012).
Busslinger, G. A. et al. Human gastrointestinal epithelia of the esophagus, stomach, and duodenum resolved at single-cell resolution. Cell Rep. 34, 108819 (2021).
Törönen, P., Ojala, P. J., Marttinen, P. & Holm, L. Robust extraction of functional signals from gene set analysis using a generalized threshold free scoring function. BMC Bioinformatics 10, 307 (2009).
DiRaimondo, T. R., Klöck, C. & Khosla, C. Interferon-γ activates transglutaminase 2 via a phosphatidylinositol-3-kinase-dependent pathway: implications for celiac sprue therapy. J. Pharmacol. Exp. Ther. 341, 104–114 (2012).
Fina, D. et al. Interleukin 21 contributes to the mucosal T helper cell type 1 response in coeliac disease. Gut 57, 887–892 (2008).
Büchold, C. et al. Features of ZED1227: the first-in-class tissue transglutaminase inhibitor undergoing clinical evaluation for the treatment of celiac disease. Cells 11, 1667 (2022).
Isola, J. et al. The oral transglutaminase 2 inhibitor ZED1227 accumulates in the villous enterocytes in celiac disease patients during gluten challenge and drug treatment. Int. J. Mol. Sci. 24, 10815 (2023).
Pascual, G. et al. A SUMOylation-dependent pathway mediates transrepression of inflammatory response genes by PPAR-γ. Nature 437, 759–763 (2005).
Bailey, S. T. & Ghosh, S. ‘PPAR’ting ways with inflammation. Nat. Immunol. 6, 966–967 (2005).
Simula, M. P. et al. PPAR signaling pathway and cancer-related proteins are involved in celiac disease-associated tissue damage. Mol. Med. 16, 199–209 (2010).
Luciani, A. et al. Lysosomal accumulation of gliadin p31-43 peptide induces oxidative stress and tissue transglutaminase-mediated PPARgamma downregulation in intestinal epithelial cells and coeliac mucosa. Gut 59, 311–319 (2010).
Daynes, R. A. & Jones, D. C. Emerging roles of PPARS in inflammation and immunity. Nat. Rev. Immunol. 2, 748–759 (2002).
ter Steege, J., Buurman, W., Arends, J. W. & Forget, P. Presence of inducible nitric oxide synthase, nitrotyrosine, CD68, and CD14 in the small intestine in celiac disease. Lab. Investig. J. Tech. Methods Pathol. 77, 29–36 (1997).
Maiuri, M. C. et al. Gliadin increases iNOS gene expression in interferon-γ-stimulated RAW 264.7 cells through a mechanism involving NF-κB. Naunyn. Schmiedebergs Arch. Pharmacol. 368, 63–71 (2003).
Daniels, I., Cavill, D., Murray, I. A. & Long, R. G. Elevated expression of iNOS mRNA and protein in coeliac disease. Clin. Chim. Acta 356, 134–142 (2005).
Murray, I. A., Bullimore, D. W. & Long, R. G. Fasting plasma nitric oxide products in coeliac disease. Eur. J. Gastroenterol. Hepatol. 15, 1091–1095 (2003).
Diefenbach, A., Schindler, H., Röllinghoff, M., Yokoyama, W. M. & Bogdan, C. Requirement for type 2 NO synthase for IL-12 signaling in innate immunity. Science 284, 951–955 (1999).
Orenbuch, R. et al. arcasHLA: high-resolution HLA typing from RNAseq. Bioinformatics 36, 33–40 (2020).
Tollefsen, S. et al. HLA-DQ2 and -DQ8 signatures of gluten T cell epitopes in celiac disease. J. Clin. Invest. 116, 2226–2236 (2006).
Tollefsen, S. et al. Structural and functional studies of trans-encoded HLA-DQ2.3 (DQA1*03:01/DQB1*02:01) protein molecule. J. Biol. Chem. 287, 13611–13619 (2012).
Senger, S. et al. Celiac disease histopathology recapitulates Hedgehog downregulation, consistent with wound healing processes activation. PLoS ONE 10, e0144634 (2015).
Juuti-Uusitalo, K., Mäki, M., Kainulainen, H., Isola, J. & Kaukinen, K. Gluten affects epithelial differentiation-associated genes in small intestinal mucosa of coeliac patients. Clin. Exp. Immunol. 150, 294–305 (2007).
Nanayakkara, M., Lania, G., Maglio, M., Kosova, R. & Sarno, M. Enterocyte proliferation and signaling are constitutively altered in celiac disease. PLoS ONE 8, 76006 (2013).
Halstensen, T. S. & Brandtzaeg, P. Activated T lymphocytes in the celiac lesion: non-proliferative activation (CD25) of CD4+ α/β cells in the lamina propria but proliferation (Ki-67) of α/β and γ/δ cells in the epithelium. Eur. J. Immunol. 23, 505–510 (1993).
Kutlu, T. et al. Numbers of T cell receptor (TCR) αβ+ but not of TCR γδ+ intraepithelial lymphocytes correlate with the grade of villous atrophy in coeliac patients on a long term normal diet. Gut 34, 208–214 (1993).
Laforenza, U. et al. Solute transporters and aquaporins are impaired in celiac disease. Biol. Cell 102, 457–467 (2010).
Banaganapalli, B. et al. Exploring celiac disease candidate pathways by global gene expression profiling and gene network cluster analysis. Sci. Rep. 10, 16290 (2020).
Barone, M. V. et al. Growth factor-like activity of gliadin, an alimentary protein: implications for coeliac disease. Gut 56, 480–488 (2007).
Maiuri, L. et al. Association between innate response to gliadin and activation of pathogenic T cells in coeliac disease. Lancet 362, 30–37 (2003).
Bragde, H., Jansson, U., Fredrikson, M., Grodzinsky, E. & Söderman, J. Celiac disease biomarkers identified by transcriptome analysis of small intestinal biopsies. Cell. Mol. Life Sci. 75, 4385–4401 (2018).
van der Graaf, A. et al. Systematic prioritization of candidate genes in disease loci identifies TRAFD1 as a master regulator of IFNγ signaling in celiac disease. Front. Genet. 11, 562434 (2020).
Bajor, J. et al. Classical celiac disease is more frequent with a double dose of HLA-DQB1*02: a systematic review with meta-analysis. PLoS ONE 14, e0212329 (2019).
Al-Toma, A. et al. Human leukocyte antigen-DQ2 homozygosity and the development of refractory celiac disease and enteropathy-associated T-cell lymphoma. Clin. Gastroenterol. Hepatol. Off. Clin. Pract. J. Am. Gastroenterol. Assoc. 4, 315–319 (2006).
Nijeboer, P., van Wanrooij, R. L. J., Tack, G. J., Mulder, C. J. J. & Bouma, G. Update on the diagnosis and management of refractory coeliac disease. Gastroenterol. Res. Pract. 2013, 518483 (2013).
Waite, K. J., Floyd, Z. E., Arbour-Reily, P. & Stephens, J. M. Interferon-gamma-induced regulation of peroxisome proliferator-activated receptor gamma and STATs in adipocytes. J. Biol. Chem. 276, 7062–7068 (2001).
Wapenaar, M. C. et al. The interferon gamma gene in celiac disease: augmented expression correlates with tissue damage but no evidence for genetic susceptibility. J. Autoimmun. 23, 183–190 (2004).
Guennouni, M. et al. Gluten contamination in labelled gluten-free, naturally gluten-free and meals in food services in low-, middle- and high-income countries: a systematic review and meta-analysis. Br. J. Nutr. 127, 1528–1542 (2022).
Atlasy, N. et al. Single cell transcriptomic analysis of the immune cell compartment in the human small intestine and in celiac disease. Nat. Commun. 13, 4920 (2022).
Clinical Trials Register. https://www.clinicaltrialsregister.eu/ctr-search/trial/2017-002241-30/LT
Taavela, J. et al. Validation of morphometric analyses of small-intestinal biopsy readouts in celiac disease. PLoS ONE 8, e76163 (2013).
Sato, T. et al. Long-term expansion of epithelial organoids from human colon, adenoma, adenocarcinoma, and Barrett’s epithelium. Gastroenterology 141, 1762–1772 (2011).
Dotsenko, V., Sioofy-Khojine, A.-B., Hyöty, H. & Viiri, K. Human intestinal organoid models for celiac disease research. in Methods in Cell Biology (Academic Press, 2023). https://doi.org/10.1016/bs.mcb.2023.01.008
Taavela, J. et al. Histological, immunohistochemical and mRNA gene expression responses in coeliac disease patients challenged with gluten using PAXgene fixed paraffin-embedded duodenal biopsies. BMC Gastroenterol. 19, 189 (2019).
Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550 (2014).
Benjamini, Y. & Hochberg, Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J. R. Stat. Soc. Ser. B Methodol. 57, 289–300 (1995).
Alexa, A. & Rahnenfuhrer, J. topGO: enrichment analysis for Gene Ontology. (2022).
Yu, G. & He, Q.-Y. ReactomePA: an R/Bioconductor package for Reactome pathway analysis and visualization. Mol. Biosyst. 12, 477–479 (2016).
Törönen, P., Ojala, P. J., Marttinen, P. & Holm, L. BMC Bioinformatics Robust extraction of functional signals from gene set analysis using a generalized threshold free scoring function. (2009) https://doi.org/10.1186/1471-2105-10-307
Mishra, P. & Holm, L. Gene expression Gene set analysis: limitations in popular existing methods and proposed improvements. 30, 2747–2756 (2014).
Wang, X., Park, J., Susztak, K., Zhang, N. R. & Li, M. Bulk tissue cell type deconvolution with multi-subject single-cell expression reference. Nat. Commun. 10, 1–9 (2019).
Elmentaite, R. et al. Cells of the human intestinal tract mapped across space and time. Nat. 2021 5977875 597, 250–255 (2021).
Naito, T. & Okada, Y. HLA imputation and its application to genetic and molecular fine-mapping of the MHC region in autoimmune diseases. Semin. Immunopathol. 44, 15–28 (2022).
Schmittgen, T. D. & Livak, K. J. Analyzing real-time PCR data by the comparative Ct method. Nat. Protoc. 3, 1101–1108 (2008).
Folk, J. E. & Cole, P. W. Mechanism of action of guinea pig liver transglutaminase I. Purification and properties of the enzyme: identification of a functional cysteine essential for activity. J. Biol. Chem. 241, 5518–5525 (1966).
Kiviniemi, M. et al. A High-Throughput Population Screening System for the Estimation of Genetic Risk for Type 1 Diabetes: An Application for the TEDDY (The Environmental Determinants of Diabetes in the Young) Study. Diabetes Technol. Ther. 9, 460–472 (2007).
Lehmusvuori, A., Kiviniemi, M., Ilonen, J. & Soukka, T. Closed-tube human leukocyte antigen DQA1∗05 genotyping assay based on switchable lanthanide luminescence probes. Anal. Biochem. 465, 6–11 (2014).
Acknowledgements
We thank the individuals who participated for making this study possible. We also thank the expert staff for their participation in sample collection. We thank K.-L. Kolho for providing intestinal biopsies to initiate organoid cultures. This work was Dr. Falk Pharma-sponsored clinical trial supported by the Academy of Finland (310011), the Finnish Cultural Foundation, Mary och Georg C. Ehrnrooths Stiftelse, Päivikki and Sakari Sohlberg Foundation, Laboratoriolääketieteen Edistämissäätiö sr and the Competitive State Research Financing of the Expert Responsibility Area of Tampere University Hospital grant. V.D. was supported by the Finnish Cultural Foundation. D.S. received project related support from the German Research Foundation (DFG) Collaborative Research Center SFB TR355/1 (490846870) project B08 (Treg in celiac disease). The funding sources played no role in the design or execution of this study or in the analysis and interpretation of the data. We acknowledge the Adult Stem Cell Organoid Facility from Tampere University for their service. Ethics approvals TUKIJA dnro 223/06.00.01/2017 and EudraCT 2017-002241-30 were obtained for the Dr. Falk Pharma-funded clinical trial. The study was conducted with deidentified data of the participants who had consented to the use of their anonymized data in research. The protocol to initiate human intestinal organoid cultures from biopsies was approved by the Ethics Committee of Tampere University Hospital (ETL code R18082).
Author information
Authors and Affiliations
Consortia
Contributions
V.D., K.V. and M.M. conceptualized the study. K.V. and V.D. drafted the manuscript. V.D. and K.V. performed data analysis and figure generation. H.H. assisted in statistical analyses. P.H. performed gastroscopies to obtain duodenal biopsies and organoids. B.T., M.H., R.P., J.I., J.T., A.P., J.S., T.Z., R.M., R.G., K.E.A.L. and D.S. assisted in the logistics of data collection and results interpretation. All authors read and approved the final paper.
Corresponding author
Ethics declarations
Competing interests
V.D. and K.V. received funding from Dr. Falk Pharma to Tampere University to conduct the study. B.T., T.Z., R.M. and R.G. are employees of Dr. Falk Pharma. The data presented here are the subject of patent applications EP24173619.8 and EP24173615.6 filed by Dr. Falk Pharma, and B.T., T.Z., R.M., R.G., V.D. and K.V. are inventors on these applications. M.H. and R.P. are employees of Zedira. A.P. is a consultant for JiLab Oy. J.T. is a consultant for Jilab Oy and Dr. Falk Pharma. K.E.A.L. is a consultant for Amyra, Bioniz Pharmaceuticals, Chugai Pharmaceutical, Dr. Falk Pharma, Itrexon Actobios, TOPAS Therapeutics and Takeda California. D.S. is the data and safety monitor for Boehringer Ingelheim (Phil.) and is a consultant for the Dr. Falk Pharma, Takeda, Immunic, Sanofi and TOPAS Therapeutics. J.I. is the owner of Jilab Oy. M.M. is the founder, owner and Chair of the Board of Maki HealthTech (MHT). MHT receives Management/Advisory Affiliation fees from Dr. Falk Pharma and other funding not related to the research from Topas Therapeutics, Calypso Biotech, Vaccitech, ImmunogenX, Equillium and Immunic. MHT holds patents (patent number 7361480 (United States) and European Patent Office Number 1390753) licensed to Labsystems Diagnostics from where MHT receives royalties via Tampere University Hospital. All other authors declare no competing interests.
Peer review
Peer review information
Nature Immunology thanks Bana Jabri and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Primary Handling Editor: S. Houston, in collaboration with the Nature Immunology team. Peer reviewer reports are available.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Extended data
Extended Data Fig. 1 Schematic presentation of the study.
Samples (n = 116; n of patients = 58), in a form of PAXgene fixed and paraffin-embedded biopsies, were collected from the trial, aimed at dose-finding, and assessing the efficacy and tolerability of a 6-week treatment with ZED1227 capsules vs. placebo in subjects with well-controlled celiac disease undergoing gluten challenge. Biopsy sampling was performed twice: on study inclusion (GFDd, n = 34; GFDp, n = 24) and at the final visit (PGCd, n = 34; PGCp, n = 24). Duodenal forceps biopsies were immersed in PAXgene fixative and processed for paraffin block embedding using a standard formalin-free paraffin-infiltration protocol. Created with BioRender.com.
Extended Data Fig. 2 Histomorphometric features and molecular pathways displaying reduced control in G1 genotype.
a, Subjects (n = 34), belonging to drug group were selected for one-way ANCOVA. VH at PGCd used as a dependent variable and VH at GFD as covariate and HLA-DQ genotype group (G1, G2, G3) as independent variables. ANCOVA, F (2, 30) = 6.56, P = .004. Post-hoc pairwise multiple comparisons were performed between HLA-DQ genotype groups, with p values Bonferroni adjusted. Results demonstrated as estimate ± 95% CI. b, Subjects (n = 34), belonging to drug group were selected for one-way ANCOVA. Cr at PGCd used as a dependent variable and CrD at GFD as covariate and HLA-DQ genotype group (G1, G2, G3) as independent variables. ANCOVA, F (2, 30) = 3.6, P = .04. Post-hoc pairwise multiple comparisons were performed between HLA-DQ genotype groups, with p values Bonferroni adjusted. Results demonstrated as estimate ± 95% CI. c, Gene set Z-score was calculated for gene sets enriched in the categories of transit amplifying cells, mature enterocytes, immune cells, duodenal transporters and Reactome database pathways for patients in drug and placebo groups at PCG (PGCd (n = 34), PGCp (n = 23)). GSZ grouped by HLA-DQ genotype group (G1, G2, G3) and presented as mean (spheres) and sd (vertical lines).
Supplementary information
Supplementary Information
Supplementary Tables 1 and 2 and Figs. 1–5.
Supplementary Data 1
List of DEGs in the PGCp versus GFDp comparison, PGCd versus GFDd comparison and PGCp versus PGCd comparison.
Supplementary Data 2
Results of the overrepresentation analysis for Reactome terms for the PGCp versus GFDp and PGCp versus PGCd comparisons and results of the overrepresentation analysis for Gene Ontology biological process terms for the PGCp versus GFDp and PGCp versus PGCd comparisons.
Supplementary Data 3
List of DEGs in the I versus M and Z versus M comparisons.
Supplementary Data 4
Enrichr results for the PGCp versus GFDp and PGCp versus PGCd comparisons (see Supplementary Fig. 2).
Supplementary Data 5
Supplementary Data for Fig. 5
Supplementary Data 6
Source data for Supplementary Tables 1 and 2 and Figs. 1–5.
Source data
Source Data Fig. 1
Statistical source data.
Source Data Fig. 2
Statistical source data.
Source Data Fig. 3
Statistical source data.
Source Data Fig. 4
Statistical source data.
Source Data Fig. 5
Statistical source data.
Source Data Fig. 6
Statistical source data.
Source Data Extended Data Fig. 2
Statistical source data.
Source Data Extended Data Table 1
Statistical source data.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Dotsenko, V., Tewes, B., Hils, M. et al. Transcriptomic analysis of intestine following administration of a transglutaminase 2 inhibitor to prevent gluten-induced intestinal damage in celiac disease. Nat Immunol 25, 1218–1230 (2024). https://doi.org/10.1038/s41590-024-01867-0
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1038/s41590-024-01867-0
