Article Text

This article has a correction. Please see:


Lysosomal accumulation of gliadin p31–43 peptide induces oxidative stress and tissue transglutaminase-mediated PPARγ downregulation in intestinal epithelial cells and coeliac mucosa
  1. Alessandro Luciani1,2,
  2. Valeria Rachela Villella3,
  3. Angela Vasaturo3,
  4. Ida Giardino4,
  5. Massimo Pettoello-Mantovani1,
  6. Stefano Guido2,3,
  7. Olivier N Cexus5,
  8. Nick Peake5,
  9. Marco Londei6,
  10. Sonia Quaratino5,
  11. Luigi Maiuri1
  1. 1Institute of Pediatrics, University of Foggia, Foggia, Italy
  2. 2Dynamic Imaging Microscopy, CEINGE, Naples, Italy
  3. 3Department of Chemical Engineering, University Federico II of Naples, Naples, Italy
  4. 4Department of Biomedical Science, University of Foggia, Foggia, Italy
  5. 5Cancer Research UK Oncology Unit, University of Southampton, Southampton, UK
  6. 6Novartis Pharma AG Translational Medicine, Basel, Switzerland
  1. Correspondence to Professor Luigi Maiuri, Institute of Pediatrics, University of Foggia, viale Pinto 1, Foggia 71100, Italy; maiuri{at}


Background An unresolved question in coeliac disease is to understand how some toxic gliadin peptides, in particular p31–43, can initiate an innate response and lead to tissue transglutaminase (TG2) upregulation in coeliac intestine and gliadin sensitive epithelial cell lines.

Aim We addressed whether the epithelial uptake of p31–43 induces an intracellular pro-oxidative envoronment favouring TG2 activation and leading to the innate immune response.

Methods The time course of intracellular delivery to lysosomes of p31–43, pα-2 or pα-9 gliadin peptides was analysed in T84 and Caco-2 epithelial cells. The effects of peptide challenge on oxidative stress, TG2 and peroxisome proliferator-activated receptor (PPAR)γ ubiquitination and p42/44–mitogen activated protein (MAP) kinase or tyrosine phosphorylation were investigated in cell lines and cultured coeliac disease biopsies with/without anti-oxidant treatment or TG2 gene silencing by immunoprecipitation, western blot, confocal microscopy and Fluorenscence Transfer Resonance Energy (FRET) analysis.

Results After 24 h of challenge p31–43, but not pα-2 or pα-9, is still retained within LAMP1-positive perinuclear vesicles and leads to increased levels of reactive oxygen species (ROS) that inhibit TG2 ubiquitination and lead to increases of TG2 protein levels and activation. TG2 induces cross-linking, ubiquitination and proteasome degradation of PPARγ. Treatment with the antioxidant EUK-134 as well as TG2 gene silencing restored PPARγ levels and reversed all monitored signs of innate activation, as indicated by the dramatic reduction of tyrosine and p42/p44 phosphorylation.

Conclusion p31–43 accumulation in lysosomes leads to epithelial activation via the ROS–TG2 axis. TG2 works as a rheostat of ubiquitination and proteasome degradation and drives inflammation via PPARγ downregulation.

  • Coeliac disease
  • ROS
  • lysosomes
  • tissue transglutaminase
  • PPARγ
  • gastrointestinal immune response
  • immune response
  • inflammatory mechanisms
  • mucosal immunity
  • CD
  • coeliac disease
  • IFN-γ
  • interferon γ
  • TG2
  • tissue transglutaminase
  • PPARγ
  • peroxisome proliferator-activated receptor gamma, ROS, reactive oxygen species, GSH, glutathione
  • MnSOD
  • manganese superoxide dismutase

Statistics from


Coeliac disease is a relatively common pathology that affects 1% of the population, and is precipitated by the ingestion of gluten proteins.1–3 In predisposed individuals, gliadin acts as a stimulator of innate responses, in turn setting the type and intensity of the adaptive immune response4 leading to a powerful activation of lamina propria gliadin-specific T cells.5–7 Several fragments of gliadin have been found to be ‘toxic’ for predisposed individuals and it has also been suggested that different portion(s) of gliadin, ostensibly not those recognised by T cells, modulates this innate activation of coeliac small intestine.7 8 Among these, p31–43 induces the expression of early markers of epithelial activation and leads to enterocyte apoptosis4 both in coeliac intestine and in sensitive intestinal epithelial cell lines.4

The mechanisms by which p31–43 initiates an innate immune response in coeliac duodenum are, however, still elusive. It has been reported that p31–43 can delay epidermal growth factor receptor (EGFR) inactivation through interference with the endocytic pathway,9 this suggesting that gliadin fragments could amplify the effects of trace amounts of EGF. Moreover, p31–43 induces early tissue transglutaminase (TG2) upregulation compared to immunodominant gliadin peptides.10 This indicates that TG2 is not only central to the adaptive response to gliadin as the main deamidating enzyme of the immunodominant epitopes, but is also a key player in the initiation of inflammation in coeliac disease.10 However, it is still unknown how p31–43 induces early TG2 activation in both coeliac intestine and ‘gliadin-sensitive’ intestinal epithelial cells.9 10

TG2 expression is regulated by retinoids, steroid hormones, peptide growth factors and cytokines, which also lead to a time-dependent decrease in TG2 ubiquitination,11 indicating that post-translational control mechanisms also play a role in TG2 regulation. We have demonstrated that in cystic fibrosis (CF) airway epithelia,12 sustained TG2 activation is driven by the increased levels of reactive oxygen species (ROS) caused by the CFTR genetic defect.12 We therefore investigated whether p31–43 was able to drive an intracellular pro-oxidative environment and demonstrate that p31–43 leads to an increase of ROS levels in coeliac duodenum as well as T84 and Caco-2 intestinal epithelial cells.

To understand how p31–43 challenge induces oxidative stress, we analysed the delivery of this peptide to the intracellular degradation systems. It has been reported that the epithelial translocation of the gliadin immunostimulatory peptide 33-mer occurs by transcytosis after partial degradation through a rab5 endocytic compartment.13 Increased transepithelial translocation of the 33-mer was demonstrated in active coeliac disease compared to healthy controls. Moreover, incomplete degradation of the 33-mer and protected transport of the peptide 31–49 occurs in patients with untreated coeliac disease, thus favouring their respective immunostimulatory and toxic effects.14 Here we show that p31–43 is retained within the lysosomes as compared to the more rapid disappearance of the immunodominant peptides pα-2 or pα-9 from the intracellular compartments. Blocking p31–43 endocytosis prevents the increase of ROS.

We have previously reported that in CF airways sustained TG2 activation drives inflammation by inducing cross-linking, ubiquitination and proteasome degradation, of the anti-inflammatory peroxisome proliferator-activated receptor (PPAR)γ,12 a negative regulator of inflammatory gene expression.15 16 Blocking TG2 through specific gene silencing12 or the specific TG2 inhibitors restores a physiological control of inflammation in CF airways.12

We therefore addressed whether the p31–43-induced innate response was mediated by the ROS–TG2 axis via PPARγ downregulation.


Peptide preparation

Biotinylated gliadin peptides p31–43, pα-9 (57–68) or pα-2 (62–75), control human thyroid peroxidase pTPO (535–551) and pTPO (536–547), peptide LGQQQPFAAVQPY (referred to below as pZ), and scrambled peptides with amino acid composition similar to p31–43 (QQGQPFPQPQLQY (referred to below as pX), GLQQFQPPPPQQY (referred to below as pY)) were synthesised by Primm (Milan, Italy) and the purity was determined by reverse-phase HPLC.

Cell lines

Human colon adenocarcinoma T84 or human colorectal carcinoma Caco-2 cell lines were cultured as recommended by American Type Culture Collection (ATCC).

Cell cultures

The cells were challenged with p31–43 for 24 h in the presence or absence of the ROS scavenger EUK-134 (50 μg/ml; Alexis Biochemicals, Florence, Italy), glutathione (GSH) (10 mmol/l; Sigma, Milan, Italy)17 or the TG inhibitor cystamine (400 μmol/l; Sigma). To study internalisation of peptides, cells were challenged for 15, 30, 60 or 90 min, 3 or 24 h at 37°C with the biotinylated peptides (20 μg/ml). To inhibit endocytosis, the cells were pre-treated with the cholesterol-binding agent methyl-β-cyclo-dextrin (M-β-CD, 10 mmol/l; Sigma) or filipin (5 μg/ml; Sigma) and then challenged with biotinylated peptides. Cell lines were also incubated for 6 h with the PPARγ agonist rosiglitazone (10 μmol/l; Alexis Biochemicals) and then challenged with p31–43. The PPARγ antagonist GW9662 (1 μmol/l; Alexis Biochemicals) was added for 24 h to p31–43 in the presence or absence of TG2 siRNA oligos. The proteosome inhibitor MG132 (50 μmol/l; Calbiochem, Milan, Italy) was added 6 h before p31–43 in the presence or absence of TG2 siRNA oligos or EUK134.


Duodenal multiple endoscopic biopsies were performed for diagnostic purposes in seven treated patients with coeliac disease (mean age, 25 years; range, 16–41 years), and five non-coeliac controls affected by oesophagitis (22.4 years; range, 18–29 years). Informed consent was obtained from all individuals, and the study was performed according to the Ethics Committee of Regione Campania Health Authority. One specimen from each patient was used for diagnosis; the other samples were cultured in vitro as described.4

In vitro organ culture of biopsy specimens from patients with coeliac disease and from controls

Duodenal biopsy specimens were cultured as previously reported4 for 3 or 24 h with medium alone, p31–43, pα-9, pα-2, pTPO (20 μg/ml) in the presence or absence of cystamine (400 μmol/l). Duodenal biopsy specimens were also incubated for 6 h with the PPARγ agonist rosiglitazone (10 μmol/l; Alexis Biochemicals) and then re-challenged for 3 h with p31–43.

RNA interference

The cells were transfected with 50 nmol/l human TG2 and scrambled small interfering RNAs (siRNAs) duplex using Hiperfect Transfection Reagent (Qiagen, Milan, Italy) at 37°C for 72 h as previously described.12

Adenoviral vector

Human manganese superoxide dismutase (MnSOD) cDNA was cloned into the shuttle vector pAd5CMVK-NpA.18 MnSOD adenovirus was a gift from Michael Brownlee (Albert Einstein College of Medicine, New York). T84 cell lines were infected with MnSOD or the control adenovirus for 2 h, as previously described.18


Tissue sections were individually incubated with the antibodies anti-phospho-tyrosine (PY99 mAb, 1:80, mouse IgG2b; Santa Cruz Biotechnology, Santa Cruz, California, USA), anti-PPARγ (clone E8, sc-7273, 1:100, mouse IgG1; clone H100, sc-7196, 1:100, rabbit polyclonal IgG; Santa Cruz Biotechnology) and control antibodies.4 12 Cell lines were incubated with the antibodies against phospho-tyrosine, Early Endosome Antigen-1 (EEA-1; 1:100, rabbit polyclonal IgG; Abcam, Cambridge, UK), Rab7 (1:200, rabbit polyclonal IgG; Abcam), ARF-1 (1:100, rabbit polyclonal IgG; Abcam), LAMP-1 (1:100, rabbit polyclonal IgG; Abcam). Two-colour and indirect immunofluorescence were used for the detection of the tested markers by using an LSM510 Zeiss confocal laser scanning unit (Zeiss, Jena, Germany).12 The correlation coefficient analysis was used as value of co-localisation between peptides and EEA-1 or LAMP-1 as described.19

In situ detection of TG2 enzymatic activity

The detection of in situ TG2 activity was performed by detection of biotinylated mono-dansyl-cadaverine (MDC) incorporation by Alexa-546 streptavidin as previously described.4 12

FRET microscopy

For acceptor photobleaching 4-μm frozen tissue sections of biopsy samples were fixed with buffered 2% paraformaldehyde and permeabilised with 0.5% Triton X-100. Upon fixation, tissues were immunostained with Alexa 546/anti-PPARγ (Santa Cruz Biotechnology, Santa Cruz, California, USA) and Cy5/anti-Nε(γ-l-glutamyl)-l-lysine isopeptide (Covolab, Cambridge, UK). Alexa 546 fluorescence was detected before and after Cy5 photobleaching.

Immunoblot and immunoprecipitation

Blots were incubated with anti-phospho-tyrosine (PY99 mAb, 1:500), anti-phospho-p42/p44 MAP kinases (Cell Signaling Technology, Danvers, Massachusetts, USA), PPARγ (clone E8 sc-7273; Santa Cruz Biotechnology, Santa Cruz, California, USA), TG2 (clone CUB7402, 1:100; NeoMarkers, Fremont, California, USA), Nε(γ-l-glutamyl)-l-lysine isopeptide (clone 81DIC2; Covalab), ubiquitin (1:100 clone FL-76, rabbit polyclonal IgG; Santa Cruz Biotechnology), as previously described.12

Immunoprecipitations with anti-TG2 CUB 7402 mAb, anti-phospho-tyrosine and anti-PPARγ were carried out as previously described.12

ROS detection

Cell lines were pulsed with 10 μmol/l 5-(and-6)-chloromethyl-2′,7′-dichlorodihydrofluorescein diacetate acetyl ester (CM-H2DCFDA) (Molecular Probes, Invitrogen, Milan, Italy) and analysed as previously described.12

Statistical analysis

Cells challenged with peptides were compared with those cultured in medium alone, and with those cultured in the presence of the tested inhibitors. All experiments were performed at least in triplicate. Data distribution was analysed, and statistical differences were evaluated using ANOVA Tukey–Kramer test and SPSS 12 software. A p value of <0.05 was considered significant.

Detailed methods are described in the online supplementary information section.


p31–43 is retained within the lysosomes in T84 epithelial cells

T84 cells were incubated with biotinylated p31–43, p-α2, p-α9 (20 μg/ml) for 15, 30, 60, 90 min, 3 h and 24 h at 37°C. The peptides were rapidly detected (upon 15 min) in intracellular vesicles (figure 1A). Prolonged incubation (after 90 min to 24 h) increased the amount of p31–43 (figure 1A), but not of p-α9 (figure 1A) or p-α2 (data not shown), which resided in perinuclear vesicles. Soon after 15 min of challenge, peptides co-localised with EEA-1 (figure 1B), a marker of early endosomes,13 whereas after 60 min they co-localised with Rab7-positive late endosomes (data not shown).20 These data are in agreement with those described for the gliadin peptide 33-mer.13 Co-staining with antibody against Arf1 (data not shown) did not reveal translocation of the peptides to the Golgi complex. After 90 min the internalised peptides co-localised with Lysosomal Associated Membrane Protein-1 (LAMP-1) (figure 1C), a marker of lysosomes.20 The peptide/EEA-1 or peptide/LAMP-1 correlation coefficients of the spatial intensity distributions yielded a fairly high correlation coefficient (figure 1D), indicating that gliadin peptides are internalised and transported through early and late endocytic endosomes to lysosomes. No internalisation of peptides or co-localisation with LAMP-1 were observed when T84 cells were pre-incubated with M-β-CD, which inhibits clathrin-mediated endocytosis,20 or filipin, a specific inhibitor of lipid raft- or caveolae-dependent endocytosis20 (data not shown). This suggests that endocytosis is essential for the delivery of gliadin peptides to the lysosomes for degradation.

Figure 1

p31–43, but not pα-9, accumulates within the lysosomes in T84 cells. (A) Incubation of T84 cells with biotinylated p31–43 and pα-9 from 15 min to 24 h at 37°C and detection by confocal microscopy upon incubation with Alexa 488-conjugated streptavidin. After 15 min both p31–43 and p-α9 were detected in intracellular peripheral vesicles. Prolonged incubation (90 min to 24 h) increased the amount of p31–43, but not of p-α9, in perinuclear vesicles. (B) Co-localisation of p31–43 and p-α9 (green) with EEA-1 (red) in T84 cells upon 15 min of challenge. Both peptides were detected in EEA1-positive vesicles (yellow). (C) Co-localisation of p31–43 and p-α9 (green) with LAMP-1 (red) in T84 cells after 90 min of challenge. Both peptides were detected in LAMP1-positive vesicles (yellow). (D) Quantitative measurement of peptides/EEA-1 or peptides/LAMP-1 co-localisations shown in B and C, respectively. Each bar represents the mean plus SEM of three independent experiments. (E–F) Co-localisation of p31–43 and p-α9 (green) with LAMP-1 (red) in T84 cells after 3 h and 24 of challenge. p31–43, but not pα-9, was detected in LAMP-1 positive vesicles (yellow). (G) Quantitative measurement of peptides/LAMP-1 co-localisations shown in E and F, respectively. Each bar represents the mean plus SEM of three independent experiments. (A–C,E,F), representative results of three independent experiments. Confocal microscopy, DAPI (blue) nuclear counterstaining. Scale bar, 10 μm. DAPI, 4′-diamidine-2′-phenylindol dihydrochloride; EEA-1, Early Endosome Antigen 1; LAMP-1, Lysosomial Associated Membrane Protein 1.

Increasing appearance of the internalised p31–43 in LAMP-1-positive vesicles, mainly limited to perinucelar localisation, was observed after 3–24 h of challenge (figure 1E,F). At these late time points p-α9 (figure 1E,F) or pα-2 (data not shown) were only faintly detected in LAMP-1-positive structures. The coefficients of p-α2 or p-α9 co-localisation with LAMP-1 were significantly lower compared to p31–43 (figure 1G), thus indicating that p31–43, but not p-α9 or p-α2 is retained within the lysosomes. Caco-2 cells showed the same behaviour as T84 after peptide challenge (data not shown). The control pTPO (535–551) (data not shown), pTPO (536–547) and pZ, as well as scrambled pX and pY peptides, behaved like pα-9 and pα-2 in both T84 (supplementary figure 1) and Caco-2 cell lines (data not shown).

p31–43 induces increase of ROS in T84 epithelial cells

T84 cells were pulsed with 10 μmol/l CM-H2DCFDA in the presence of the tested peptides and the levels of ROS were monitored after 90 min, 3 h and 24 h of challenge. The intracellular transport of gliadin peptides through early and late endocytic endosomes did not induce ROS generation (data not shown). Moreover, after 90 min of challenge neither p31–43 nor pα-9, both detected in LAMP-1 positive vesicles (figure 1C), induced significant increase of ROS (figure 2A). At the later time points (3–24 h of challenge), p31–43, but not pα-9 (figure 2A), pα-2, pTPO (535–551) (data not shown) or peptides pTPO (536–547), pX, pY and pZ (supplementary figure 1) induced a time-dependent significant increase of ROS levels, suggesting that the prolonged persistence of p31–43 in LAMP-1 positive vesicles (figure 1E,F) generates a pro-oxidative environment. No increase of ROS was observed after challenge with p31–43 upon M-β-CD or filipin treatment (figure 2B).

Figure 2

p31–43 induces increase of ROS levels in T84 cells. (A) Increase of intracellular ROS in p31–43 stimulated T84 cells. Each bar represents the mean plus SEM of three independent experiments. *p<0.01 versus samples cultured with medium or with p-α9. (B) Inhibition of p31–43 endocytosis by methyl-β-cyclo-dextrin or filipin inhibited p31–43-induced increase of ROS. Each bar represents the mean plus SEM of three independent experiments. *p<0.017 versus samples cultured with p31–43. (C) Immunoblot analysis showed a decrease of p31–43-induced p42/p44 phosphorylation upon treatment with the ROS scavenger EUK-134. Representative result of three different experiments. DCF, XXXXX; ROS, reactive oxygen species.

The generation of a pro-oxidative environment induces activation of different stress sensitive signalling pathways.9 12 21 Since the challenge with p31–43 induces phosphorylation of the extracellular signal-regulated kinases 1/2 (Erk1/2, p42/44 mitogen-activated protein kinases, MAPK),9 we investigated whether p31–43-induced ROS-mediated p42/44 MAPK phosphorylation in T84 cells. We demonstrated that p42–44 MAPK phosphorylation induced by the challenge with p31–43 was prevented by the incubation with the catalase–superoxide dismutase (SOD) mimetic EUK-13412 (figure 2C), as well as by GSH17 or by MnSOD over-expression18 (supplementary figure 2). This further indicates that the generation of a pro-oxidative environment is critical to p31–43 biological activity in T84 cells.

p31–43 induces ROS-dependent inhibition of TG2 ubiquitination

To investigate whether the increased levels of ROS may account for the p31–43-induced increase of TG2 levels,12 we incubated p31–43-challenged T84 cells with EUK-134. We demonstrated that EUK-134 was effective in controlling the increase of TG2 protein levels (figure 3A) and TG2 activity (data not shown) induced by p31–43. The same effects as EUK-134 were observed after incubation with GSH or upon MnSOD over-expression (supplementary figure 2). No increase of TG2 was observed after challenge with p31–43 upon M-β-CD or filipin treatment (data not shown). This indicates that p31–43 internalisation is critical for the induction of ROS-mediated TG2 activation.

Figure 3

ROS-mediated inhibition of TG2 ubiquitination in T84 cells upon p31–43 challenge. (A) Immunoblot analysis of TG2 protein. ROS scavenger EUK-134 inhibited p31–43-induced increase of TG2 protein in T84 cells. (B) Ubiquitin immunoreactivity in TG2 immunoprecipitates. Incubation with EUK-134 increased TG2 ubiquitination upon MG132 treatment in T84 cells. Immunoprecipitation (IP): anti-TG2 antibody; immunoblot (IB): anti-ubiquitin antibody Representative results of three independent experiments. ROS, reactive oxygen species; TG2, tissue transglutaminase.

Since post-translational modifications of TG2, such as ubiquitination, play a role in regulating the levels of TG2 protein11 we investigated whether TG2 ubiquitination was influenced by ROS generation induced by p31–43. We incubated T84 cells with p31–43 in the presence or absence of EUK-134 upon inhibition of proteasome function by MG132,12 then immunoprecipitated TG2 protein and detected ubiquitin co-reactivity. The ubiquitin immunoreactivity on TG2 immunoprecipitates was enhanced upon treatment with EUK-134 (figure 3B). No effects of p31–43 on TG2 ubiquitination were observed when T84 cells were pre-treated with M-β-CD or filipin (data not shown).

p31–43-induced TG2 drives PPARγ cross-linking and proteasome degradation in T84 cells

We investigated whether the p31–43-induced TG2 activation was effective in inducing PPARγ cross-linking and proteasome degradation as we have reported in CF airways.12

We immunoprecipitated PPARγ species from T84 cell lysates after challenge with p31–43 and detected the immunoreactivity with an isopeptide cross-link specifically catalysed by TG2.22 We observed isopeptide immunoreactivity in PPARγ immunoprecipitates (figure 4A). No PPARγ cross-linking was observed in T84 cells challenged with pα-9 or pα-2 (data not shown). TG2 siRNA12 was highly effective in preventing PPARγ cross-linking upon p31–43 exposure (figure 4A). Moreover, pre-treatment with the TG2 inhibitor cystamine,23 as well as with the EUK-134, also prevented TG2-mediated cross-linking of PPARγ (data not shown). These results indicate that p31–43 induces post-translational modifications of PPARγ via the ROS–TG2 axis.

Figure 4

p31–43 challenge induces TG2-mediated PPARγ cross-linking and proteasome degradation in T84 cells. (A) Immunoprecipitates of PPARγ species from whole-cell extracts of T84 cells after challenge with p31–43 were immunoreactive for the anti-isopeptide cross-link antibody. High MW bands ranging from 72 to 130 kDa are evident after p31–43 challenge and significantly reduced upon TG2 siRNA treatment. Immunoprecipitation (IP): anti-PPARγ antibody; immunoblot (IB): anti-isopeptide antibody. (B) Effect of the MG132 treatment on PPARγ ubiquitination upon TG2 siRNA in p31–43-challenged T84cells. Immunoprecipitated PPARγ species from whole-cell extracts of T84 cells immunoreactive for the anti-ubiquitin antibody were evident after challenge with p31–43 upon MG132 treatment. The simultaneous incubation with TG2 siRNA oligos induced a pronounced decrease of ubiquitinated PPARγ protein. Immunoprecipitation (IP): anti-PPARγ antibody; immunoblot (IB): anti-ubiquitin antibody. (C) Immunoblot analysis of PPARγ protein in p31–43-challenged T84 cells. TG2 siRNA as well as cystamine inhibited p31–43-induced PPARγ downregulation. Quantitative analysis (mean, SD) of three different experiments is reported in supplementary figure 3. (D) Confocal microscopy showed a decrease of p31–43-induced tyrosine phosphorylation in T84 cells upon rosiglitazone pre-treatment. (E) Phospho-tyrosine immunoreactivity of PY-99 immunoprecipitates of T84 cells after incubation with p31–43 upon TG2 siRNA in the presence or absence of GW9662. Decrease of PY99 immunoreactivity upon TG2 gene silencing. GW9662 inhibits TG2 siRNA-induced decrease of tyrosine phosphorylation. (F) Confocal microscopy shows an increase of p31–43-induced tyrosine phosphorylation in T84 cells upon TG2 gene silencing followed by GW9662. (D,E) Confocal microscopy, phospho-tyrosine (PY-99 antibody, green), DAPI (blue) nuclear counterstaining. Scale bar, 10 μm (A–F), are the results of three reproducible experiments. DAPI, 4′-diamidine-2′-phenylindol dihydrochloride; PPARγ, peroxisome proliferator-activated receptor γ; TG2, tissue transglutaminase.

We also investigated whether the cross-linking of PPARγ might favour ubiquitination and PPARγ proteasome degradation with reduction of the 55 kDa PPARγ form. Indeed, a striking increase of ubiquinated PPARγ was observed in p31–43-stimulated T84 cells treated with the proteasome inhibitor MG132 (figure 4B). Negligible amounts of ubiquitinated PPARγ were observed in p31–43-stimulated T84 cells upon TG2 gene silencing (figure 4B) as well as after TG2 inhibition by cystamine or after treatment with EUK-134 (data not shown). Moreover, a reduction of the 55-kDa PPARγ form was observed in T84 cells upon p31–43 challenge (figure 4C) and TG2 siRNA was highly effective in preventing p31–43-induced PPARγ downregulation (figure 4C and supplementary figure 3). The incubation with the TG2 inhibitor cystamine (400 μmol/l), showed the same effects as TG2 knock-down (figure 4C). No PPARγ cross-linking or proteasome degradation was observed in T84 cells upon inhibition of p31–43 endocytosis by M-β-CD or filipin (data not shown). These results indicate that the internalisation and lysosomal delivery of p31–43 induces PPARγ downregulation via the ROS–TG2 axis.

The ROS–TG2–PPARγ axis is a master regulator of the epithelial activation to gliadin peptides

To demonstrate that the biological activity of p31–43 was mediated by the downregulation of PPARγ, we pre-incubated T84 cells with the PPARγ agonist rosiglitazone12 for 6 h and then challenged with p31–43. As a matter of fact, PPARγ ligation by agonists favours PPARγ interaction with the nuclear receptor co-repressor (N-CoR) histone deacetylase 3 (HDAC3) complex and thereby blocks its ubiquitination, thus maintaining a repressor condition.23 We monitored the effects of p31–43 on epithelial tyrosine phosphorylation, as revealed by PY-99 antibody, a well-established marker of epithelial activation in T84 cells and in human coeliac disease intestinal mucosa.4 We found that the pre-incubation with rosiglitazone antagonised p31–43-induced tyrosine phosphorylation in T84 cells (figure 4D).

Moreover we incubated p31–43-challenged T84 cells with the PPARγ antagonist GW966212 upon TG2 siRNA. The incubation with GW9662 antagonised the downregulatory effect of the TG2 siRNA on tyrosine phosphorylation induced by p31–43, as detected by immunoprecipitation studies (figure 4E) and confocal microscopy (figure 4F). These data indicate that p31–43 may induce epithelial activation via the ROS–TG2–PPARγ axis.

p31–43 induces TG2-mediated PPARγ cross-linking and reduced protein expression in coeliac duodenum

To investigate whether TG2-mediated PPARγ cross-linking and downregulation occurs in human coeliac intestine, we challenged coeliac biopsies with p31–43, pα-2 or pα-9 in presence or absence of cystamine, a well known TG2 inhibitor already used in vivo in a mouse model of Huntington's disease to control TG2-related manifestations.23 FRET analysis showed that PPARγ interacted with the isopeptide cross-link also in human coeliac disease biopsies (figure 5A). Confocal microscopy showed that PPARγ protein was reduced in coeliac biopsies after challenge with p31–43 (figure 5B) but not with pα-2 or pα-9 (data not shown). Moreover treatment of cultured biopsies with cystamine was highly effective in preventing p31–43-induced PPARγ downregulation (figure 5B). No PPARγ–isopeptide interaction (data not shown) or p31–43 induced PPARγ downregulation was observed in non-coeliac control biopsies (figure 5C), in agreement with our previous data showing that p31–43 fails to induce TG2 upregulation in non-coeliac duodenum.10 Moreover, upon challenge with p31–43 coeliac biopsies showed intense immunoreactivity to the anti-phospho-tyrosine antibody PY99 (figure 5D). Pretreatment with rosiglitazone completely reversed the p31–43-induced phospho-tyrosine immunoreactivity (figure 5D).

Figure 5

The challenge with p31–43 induces PPARγ cross-linking and reduced PPARγ protein in coeliac duodenum. (A) FRET analysis of celiac duodenal mucosa challenged for 3 h with p31–43. The increase of PPARγ–Alexa-546 fluorescence after Nε-(γ-glutamyl)-l-lysine isopeptide–Cy5 photobleaching in enterocytes of p31–43-challenged coeliac biopsies revealed PPARγ–Nε-(γ-glutamyl)-l-lysine isopeptide interaction. (B,C) p31–43 challenge for 3 h induced a marked reduction of PPARγ protein (green) in the enterocytes of duodenal mucosa from celiac patients (B) but not controls (C). The incubation with cystamine prevented p31–43 induced PPARγ downregulation (B). (D) p31–43 challenge for 3 h induced high PY99 immunoreactivity (green) in the enterocytes of coeliac duodenal biopsies which was prevented by treatment with rosiglitazone. (B–D), Confocal microscopy, DAPI (blue) nuclear counterstaining. Scale bar, 10 μm. DAPI, 4′-diamidine-2′-phenylindol dihydrochloride; FRET, Fluorescence Resonance Energy Transfer; HM, high magnification; PPARγ, peroxisome proliferator-activated receptor γ.


An innate response triggered by several ‘toxic’ gliadin fragments4 7 8 is involved in both modulating mucosal damage24–26 as well as in setting the type and intensity of the adaptive immune response in coeliac disease.4 Furthermore, a series of studies has indicated that several gliadin fragments activate, in a non-disease-specific manner, both mouse and human DC cells,27 28 a specific type of DC detected in the intestines of patients with coeliac disease29 and some epithelial cells lines derived from different species.30 31 10 These results have raised a series of questions and the most pressing is to understand why only CD tissues react to this non-T-cell antigenic portion of gliadin.4

Here we demonstrate that ‘gliadin sensitive’ epithelial cells upregulate intracellular ROS upon challenge with p31–43 but not with immunodominant gliadin peptides. These results indicate that p31–43 induces cellular stress. Among the factors able to trigger cellular stress is the accumulation of improperly handled substances within the intracellular compartments.32–35 A string of studies has suggested that some α-gliadin peptides, in particular p31–43(9)13 as well as 33-mer,14 possess the ability to penetrate cells.13 It has been demonstrated that gliadin peptides may be internalised by endocytic uptake and activate some signal transduction pathways.9 Here we demonstrate that p31–43 is delivered to the lysosomes but is retained as late as 24 h after challenge in LAMP1-positive vesicles mainly located in the perinuclear region. The engulfment of lysosomes with p31–43 may induce cellular stress with the generation of a pro-oxidative environment. The relationship between lysosomal garbage and the perturbation of cellular homeostasis has been described in a number of human pathologies such as lysosomal storage diseases,33 36 37 neurodegenerative diseases32 34 and even in ageing.35 However, in T84 or Caco-2 cells as well as in coeliac intestine, the lysosomal machinery of peptide degradation is not generally perturbed since other peptides, such as pTPO and even the gliadin immunodominant pα-2 or pα-9 peptides, are no longer detectable within intracellular organelles after 90 min of challenge. Whatever the mechanism involved in such a puzzling interaction between some gliadin peptides and ‘sensitive’ epithelia, these results further indicate that a complex alteration of the cross-talk between the intestine and its local environment is crucial for the development of coeliac disease.

Our results demonstrate that p31–43 induces TG2 activation via ROS generation. The increased levels of ROS reduce TG2 degradation by the ubiquitin–proteasome system, thus leading to increased TG2 protein levels. TG2 is a multifunctional enzyme with a vast array of biological functions.38 Increased tissue levels of TG2 have been described in a number of human diseases, such as neurodegenerative diseases,39 40 CF12 and even in cancers.41 The upregulation of TG2 has recently been associated with an increased metastatic activity41 or drug resistence.41 One of the TG2 effects is the cross-linking, with consequent functional sequestration and proteasome degradation of several intracellular proteins such as α-synuclein in Parkinson's disease,39 huntingtin in Huntington's disease.42 We have previously reported that in CF TG2 drives the characteristic chronic inflammation via PPARγ downregulation.12

Here we demonstrate that in T84 cells, as well as in coeliac duodenum, p31–43 induces PPARγ downregulation via ROS-mediated TG2. Therefore, an uncontrolled activation of the ROS–TG2 axis, either constitutive, as a consequence of genetic alterations as in CF,12 or induced by triggering factors such as p31–43 in a susceptible target (coeliac mucosa and T84 epithelial cells), leads to PPARγ downregulation with a derangement of the appropriate control of inflammation. PPARγ is a hormone receptor produced by several cell types, including epithelial cells, which negatively regulates inflammatory gene expression by ‘transrepressing’ inflammatory responses15 and even by modulating oxidative stress.43 44 PPARγ has been identified as a major functional receptor mediating the aminosalicylate activity45 46 and PPARγ agonists have been exploited in therapeutic approaches to control inflammation in chronic intestinal inflammatory diseases such as ulcerative colitis47 48 and in experimental models of colitis.49 PPARγ plays a key role in the regulation of the intestinal ‘inflammatory’ homeostasis since is activated by dietary ligands,50 as well as commensal bacteria.51 In this paper we provide the first evidence that TG2-mediated PPARγ downregulation plays a key role in the pathogenesis of coeliac disease. The effects of p31–43 are specific for coeliac intestine and ‘gliadin-sensitive’ cell lines.30 31 10 This might be related to a coeliac disease-specific peptide internalisation or reflect a still unknown coeliac disease-specific perturbation of the machinery of peptide degradation.

The activation of the ROS–TG2 axis is therefore a key pathway of inflammation. TG2 works as a rheostat of ubiquitination and proteasome degradation in inflammation, particularly in the context of coeliac disease. Indeed, in one case, increased PPARγ ubiquitination and degradation leads to increased inflammation, whilst increased TG2 ubiquitination and degradation leads to a reduced inflammatory response. A derangement of TG2 regulation may lead to the amplification of the effects of the stress. The induction of an epithelial pro-inflammatory phenotype may alter the first mucosal defence against ‘toxic’ agents and lead to a wide perturbation of the regulatory mechanisms at the mucosal surface. The increased secretion of inflammatory cytokines may, in turn, derange intestinal permeability52 and enhance the toxic effects of environmental triggers. TG2 may be considered as a main player of the innate response to stress inducers, thus setting the tone of whole mucosal response. Targeting the ROS–TG2 axis might represent a new pathogenic-based approach to antagonise the unwanted effects of gluten in coeliac disease.


The authors wish to thank Fabio Formiggini (Dynamic Imaging Microscopy, CEINGE, Naples, Italy) for the technical support in FRET analysis.


View Abstract


  • Linked articles 169656, 189332.

  • Funding This work was supported by the Coeliac UK and the Rothschild Trust Corporation and Associazione Italiana Celiachia Regione Puglia (# 1400/07, Del. Reg.502 – 08/04/2008).

  • Competing interests None.

  • Ethics approval This study was conducted with the approval of the Regione Campania Health Authority.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles

  • Commentary
    Eduardo Arranz Jose A Garrote
  • Coeliac disease
    Klaus-Peter Zimmer Ina Fischer Thomas Mothes Gabriele Weissen-Plenz Martina Schmitz Herbert Wieser Jürgen Büning Markus M Lerch Paul C Ciclitira Peter Weber Hassan Y Naim
  • Correction
    BMJ Publishing Group Ltd and British Society of Gastroenterology