The recent discovery of an unusually long and potent T cell proliferation agent, the 33-mer peptide from α-2 gliadin, highlighted the importance of understanding the molecular fate of gluten under physiological conditions. In particular, the role of proteolytic resistance in gluten antigenicity became clear. The 33-mer peptide has become a useful probe for DQ2 binding 30, 31
, for peptidase development 32
and for studying the transport of peptide and Celiac pathogenesis at the intestinal level 33
. The goals of the current work were therefore two-fold. On one hand we sought to verify the centrality of the 33-mer in the immunotoxicity of α-2 gliadin via deletion mutagenesis. On the other hand, we wished to identify similar peptides from other gluten proteins, especially the closely related γ-gliadins.
Our studies verified that the 33-mer peptide is the single crucial antigenic component in α-2 gliadin. The digestion mixture of α-2 gliadin was comparably potent to an equivalent amount of the synthetic 33-mer peptide, and 10–20 times more potent than any individual T cell epitope present in the 33-mer sequence. Two deletion mutants of α-2 gliadin lacking the 33-mer were engineered, one excluding only the 33-mer sequence and the other deleting the 33-mer as well as additional C-terminal residues. The resulting proteins were no longer toxic. Taken together with earlier data, the 33-mer was thus confirmed as the major Celiac-toxic segment of α-2 gliadin.
The existence of other long, proteolytically resistant, antigenic peptides in gluten proteins was anticipated from early in silico
analysis using the 33-mer sequence as a probe 14
. To obtain direct experimental evidence for this proposal, here we have identified a 26-mer peptide, FLQPQQPFPQQPQQPYPQQPQQPFPQ, from γ-5 gliadin. This peptide is not as resistant to pancreatic proteolysis as the 33-mer peptide, presumably due to its PQ↓Q repeats, which are weakly susceptible to chymotrypsin cleavage. Like the 33-mer however, the 26-mer peptide is highly resistant to intestinal brush border membrane proteolysis. It is also multivalent; thus far three epitopes, EQPFPEQPE (DQ2-γ-VI), EQPEQPYPE (DQ2-γ-III), EQPEQPFPQ (DQ2-γ-VII) have been identified in the 26-mer 16
. Similar to the 33-mer, the intact 26-mer is more antigenic compared to its smaller monovalent counterparts. Finally, analogous to the 33-mer, the 26-mer is susceptible to PEP breakdown, as illustrated in .
In an attempt to integrate available biochemical and immunological information to derive a global picture of the physiological immunotoxicity of dietary gluten for Celiac Sprue patients, we conducted an in-depth computational analysis of the gluten proteome. Our analysis revealed that long, proteolytically resistant fragments were widespread within the α-gliadin, γ-gliadin, glutenin, hordein and secalin protein families of wheat, rye and barley. In contrast, analogous investigations on sequences of avenins (from oats) and a range of proteins present in bovine myoglobin, chicken ovalbumin, bovine casein and lactoglobulin (from meat, eggs or milk) showed no peptide products longer than 10 amino acids (). Their susceptibility to proteolysis corresponds to their low proline content (an average of 6% in avenins, 2% in myoglobin, 3% in ovalbumin, 12% in casein and 4% in lactoglobulin). Only avenin and casein proteins produced digestion fragments of 8–10 residues. Notably, these avenin fragments contain T cell epitopes recognized by intestinal T cells of oats-intolerant Celiac Sprue patients 14
. Gluten proteins known to be highly antigenic, such as α-2 gliadin, γ-5 gliadin and the M36999 γ-gliadin, appear to release long, proteolytically stable, multivalent peptides as a result of duodenal proteolysis. In contrast, gluten proteins with limited T cell antigenicity seem to be incapable of generating such long, epitope-bearing peptides. Due to the rapid intestinal brush-border membrane proteolysis of peptides shorter than 10 amino acids 12
, the antigenic peptides released from these gluten proteins are likely to be rapidly destroyed by the peptidases of the intestinal brush border membrane.
The putative existence in gluten of long, proteolytically stable peptides that lack any known T cell antigenicity bears some reflection. Perhaps this suggests that the epitope repertoire of gluten responsive T cells in Celiac Sprue patients has not been thoroughly mapped as yet. Alternatively, there may exist non-T cell mediated mechanisms that contribute to the overall gluten enteropathy in Celiac Sprue patients. A few such mechanisms have been proposed based upon the identification of gluten peptides that lack TH
1 reactivity, but are capable of triggering inflammatory reactions in intestinal biopsies derived from Celiac Sprue patients 34–37
. For example, the 13-residue sequence, LGQQQPFPPQQPY, is found in putative physiological peptides (LG)QQQPFPPQQPYPQPQPFPS among α-gliadins, and is known to elicit an innate immune response. However, in no case has a primary molecular target been identified that interacts with the gluten peptide. It is possible that identification of such targets would require use of the physiological forms of these peptides. Finally, gluten sensitivity is occasionally diagnosed in patients who do not have Celiac Sprue 38
. It is conceivable that one or more of these proteolytically resistant peptides plays a crucial role in eliciting a pathological response in those patients. Regardless of the precise mechanism of pathogenicity, peptide length and a high proline content are likely to be common denominators of gluten intolerance. If so, the use of an oral prolyl endopeptidase may be appropriate therapy for such patients, as proposed recently 14, 17, 18, 20, 21
The limitations of the above computational analysis of the physiological fate of gluten must be acknowledged. For example, such analysis does not consider the effect of gastric pepsin on food. Similarly, it does not incorporate complexities resulting from the extensive disulfide crosslinked structure of glutenin, or other secondary structural features of gluten peptides that influence their proteolytic susceptibility. Ultimately, a thorough analysis of gluten proteolysis using experimental proteomic methods will be required in order to obtain a quantitatively accurate picture of the physiological process of gluten digestion. Notwithstanding these limitations, our results have clearly reinforced the notion that protease stability and immunotoxicity of gluten are intimately correlated. This has implications for protein engineering of wheat strains with reduced antigenicity for Celiac Sprue patients 26, 39
, and for the development of high-throughput screening tools for identifying and engineering therapeutically useful enzymes 10, 12, 22, 32, 40