Review Article: Novel Enzyme Therapy Design for Gluten Peptide Digestion Through Exopeptidase Supplementation

ABSTRACT

Background

Dietary peptides are increasingly linked to inflammatory gastrointestinal diseases, exemplified by coeliac disease. Coeliac disease is caused by an acquired immune response to proline‐ and glutamine‐rich gluten peptides, which bottleneck proteolysis and provide substrates for immune recognition. Enzyme therapies aim to eliminate gluten immunogenic peptides as an adjunct to gluten‐free diet.

Aims

To investigate overlooked aspects of enzyme development given difficulties in translating preclinical efficacy into clinical benefit.

Methods

We assessed mode‐of‐action, target organ and drug delivery in the context of digestive physiology and motility for gluten‐digesting enzymes on the market or in development until 1 December 2024.

Results

Most enzymes were gastric endopeptidases specific for proline or glutamine residues. Gastric enzymes may achieve poor enzyme–substrate exposure due to limited mixing and rapid emptying of water‐soluble particles. Moreover, endopeptidases cleave proteins/peptides into shorter peptides but do not systematically cleave protein into absorbable fractions. Natural digestive physiology provides thorough mixing at the intestinal brush border, which produces exopeptidases necessary to fully digest proline‐rich peptides. Despite reduced activity in patients with coeliac disease, exopeptidases remain underexplored as therapeutic agents. Given limited substrate scope and end‐to‐end digestion, exopeptidases are ineffective as single agents, requiring functional combinations. Furthermore, vulnerability to gastric acid requires stabilisation or formulation for rapid enteric release.

Conclusions

Enzymes should be stabilised throughout the gastrointestinal tract including the small intestine. Exopeptidases perform a critical function by systematically generating absorbable fractions, warranting future investigation as therapeutic agents. Sensitive and translational biomarkers are needed to better assess enzyme efficacy in real‐meal conditions.

Exopeptidases are key for completing protein digestion, including proline‐rich peptides, and are mostly produced by enterocytes of the intestinal brush border membrane. In coeliac disease, chronic inflammation leads to reduced brush border exopeptidase activity, exacerbating peptide accumulation. Exopeptidase supplementation could support, enhance and/or replace this critical enzyme function in patients.

1. Introduction

Coeliac disease is a common autoimmune‐like enteropathy caused by an acquired proinflammatory T cell‐mediated immune response to specific peptide fragments derived from certain gluten proteins during digestion. Human gastrointestinal peptidases are slow and inefficient at digesting gluten as they generally lack cleavage activity for proline‐adjacent bonds. Proline residues comprise approximately 14% of amino acids in gluten proteins (glutamine accounts for approximately 32%), but proline is enriched to 30% and 39% in gluten peptides that are immunogenic or immunodominant, respectively, for gluten‐specific T cells in coeliac disease (see Table 1) [1, 2, 3]. The delayed digestive process results in the accumulation of gluten immunogenic peptides that contain proline‐ and glutamine‐rich epitopes which can act as substrates for immune recognition [4, 5]. In patients with coeliac disease, gluten exposure reactivates a specific CD4⁺ T cell‐driven immunological response resulting in acute gastrointestinal symptoms and systemic cytokine release including interleukin‐2 within 1–4 h, expansion of gluten‐reactive CD4⁺ T cells and nonspecific expansion of gut‐homing lymphocytes over a period of days, and intestinal injury developing over a period of weeks to months [6, 7, 8]. Extraintestinal symptoms, including dermatological and neurological symptoms, and certain cancers and neurodegenerative diseases, are associated with untreated coeliac disease [9, 10, 11, 12, 13, 14].

TABLE 1.

Enrichment of proline and glutamine in wheat gliadin and glutenin peptides screened for immunogenicity in HLA‐DQ2.5+ patients with coeliac disease by Tye‐Din et al.

	Library	Immunogenic	Immunodominant
Number of peptides	2152	37	7
Peptide length	20	12 to 16	12
Total amino acids	43,040	476	84
% Glutamine	32	36	32
% Proline	14	30	39

Note: Comprehensive, quantitative mapping of wheat gluten epitopes reported by Tye‐Din et al. provided a hierarchy of gluten peptides stimulating gluten‐specific T cells [1]. Data are derived from Table 1 in Tye‐Din et al. Immunogenic peptides are defined as those recognised by antigen‐specific receptors on B cells or T cells, in this case referring to gluten‐specific CD4⁺ T cells, whereas immunodominant peptides are preferentially recognised over others by T cells. Immunodominant peptides are those stimulating at least 70% the maximal peptide‐stimulated ELISpot response in > 10% of patients, and immunogenic peptides stimulated at least 7% of the maximal response, for details see Tye‐Din et al. [1].

Currently, the only approved treatment for coeliac disease is a lifelong gluten‐free diet, though inadvertent gluten contamination is frequent and a constant risk with ingestion of any food or drink [15]. Carefully performed quantitative duodenal histology or histomorphometry indicates that about half of coeliac disease patients symptomatically well controlled on a gluten‐free diet enrolling in sponsored therapeutics trials have persistent villous atrophy (Marsh 3) and most of the remainder have crypt hyperplasia (Marsh 2) [16, 17]. Consequently, most coeliac disease patients do not experience complete mucosal healing and even the most robust clinical evidence supporting the currently acceptable level of 20 ppm gluten in food labelled ‘gluten‐free’ is controversial [18].

Given the proteolytic resistance of gluten and the inadequacy of the gluten‐free diet alone to restore intestinal health, digestive enzyme therapies have emerged as a promising adjunct treatment for coeliac disease patients, with the aim of supporting or enhancing gluten digestion to destroy gluten immunogenic peptides in the gut lumen before they can reactivate gluten‐specific immunity. To date, these therapies have focused almost exclusively on targeting gluten digestion in the stomach using endopeptidases, whereas supplementing endopeptidase and/or exopeptidase activity in the small intestine has remained underexplored (Table 2). The historic emphasis on targeting gluten digestion in the stomach likely stems from concerns that gluten immunogenic peptide digestion must occur prior to gastric emptying to prevent immune activation in the small intestine. However, research into the timeline of immunological events following gluten ingestion revealed a 1‐ to 4‐h window before peak detection of systemic immune‐activation markers and associated gastrointestinal symptoms [6], suggesting that fast‐acting enteric enzymes could detoxify gluten peptides in time to prevent acute as well as chronic effects of gluten toxicity.

TABLE 2.

Summary of gluten‐digesting enzymes.

Product type	Enzyme name, Company	Enzyme type	Enzyme origin	Potential cleavage sites	Site of action	pH optima (range)	Resistances	Vulnerabilities	Status	Ref
Dietary supplement	AN‐PEP DSM	Endopeptidase	Aspergillus niger	P↓F, P↓Q, P↓Y, P↓L, P↓T, P↓S Proline preferred, followed by alanine (A↓X)	Stomach	Optima: 4–5 Range: 2–8	Pepsin	Trypsin and/or chymotrypsin	On market; also assessed as drug candidate for coeliac disease (Phase IV)	[19, 20, 21]
	Aspergillopepsin Doctor's Best	Endoprotease	Aspergillus spp.	H↓X (X = S,A), Q↓X (X = Q,V,S,E), R↓X (X = V,N), A↓X (X = Y,P), K↓Q, L↓X (X = Q,V,P), F↓X (X = P,E), P↓X (X = V,Q), T↓X (X = I,N), Y↓R, and S↓X (X = F,Q)	Stomach	Optima: 2–3 Range: 2–6	Pepsin a	Denatures at pH ≥ 6.5 Competitive protein substrates	On market	[22, 23, 24]
	DPP‐IV Biocore, National Enzyme Company, Enzymedica, Deerland, others	Exopeptidase	Aspergillus spp.	N‐terminal XP↓ dipeptides preferred (see Table 3)	Unspecified by marketed products	Optima: 7 Range: 4–9	Pepsin	pH < 4	On market	[22, 25]
	Caricain (Gluteguard) Glutagen	Endopeptidase	Carica papaya	P↓X advertised but data not found Cleavage sites for wheat alpha‐amylase/trypsin inhibitors: R↓S, A↓A, S↓S, L↓T, A↓C	Small intestine	Optima: 7 Range: 4.5–7	Trypsin, pepsin (at pH 3)	Pepsin (at pH 2)	On market	[26, 27, 28]
Prescription drug	EP‐B2 + SC‐PEP (Latiglutenase) Entero Therapeutics	Endopeptidases	EP‐B2: Hordeum vulgare	Cleaves after glutamine (Q↓L) preferentially in proline‐rich regions	Stomach	Optima: 4.5 Range: 2.5–6.9	Pepsin	Trypsin	Phase III ready	[29, 30, 31, 32, 33]
			SC‐PEP: Sphingomonas capsulate	P↓Q and P↓Y	Stomach	Optima: 6–7 Range: 1.6–8		Pepsin, trypsin Severe chain‐length restriction for polypeptides (e.g. 30+ amino acids)
	E40 Nemysis Limited	Endopeptidase	Soil actinomycete Actinoallomurus A8	F↓P and Q↓L	Stomach	Optima: 5 Range: 3–6	Pepsin, trypsin	pH > 6	Preclinical	[34]
	Tak‐062 Takeda	Endopeptidase	Computationally designed, derived from bacterial enzyme kumamolisin‐As, from Alicyclobacillus sendaiensis	PQ ↓ dipeptides	Stomach	Optima: 2–4 Range 2.5–6	Pepsin, trypsin	pH > 6	Phase II	[31, 35, 36]
	Neprosin and nepenthesin	Endopeptidases	Carnivorous pitcher plant (Nepenthes spp.)	Neprosin: PQP↓Q, with preference for PQP↓QLP motif	Stomach	Neprosin: Optimum: 3 Range: 2–6	Pepsin		Preclinical	[37]

Note: This table summarises the current landscape of gluten‐digesting enzymes, including on‐market dietary supplements and preclinical and clinical drugs in development for coeliac disease. Potential preferred cleavage sites for gluten proteins are provided for all enzymes where data are publicly available. Assessment of substrate specificity across enzymes has not been standardised, and relative efficacy for a given substrate is not always defined. Likewise, pH activity range does not include loss of efficacy above or below pH optima. This table is not comprehensive of all existing enzymes under development, particularly at the preclinical stage.

Abbreviations: AN‐PEP, Aspergillus niger derived PEP; DPP‐IV, dipeptidyl peptidase IV; (P)EP, (prolyl) endopeptidase.

^a Assumption based on published data, not explicitly defined.

With regard to mode‐of‐action, endopeptidases cleave intrachain peptide bonds generating peptides of varying lengths. In contrast, exopeptidases sequentially cleave peptide bonds at the amino‐ or carboxy‐terminus (‘end‐to‐end’ digestion), generating absorbable fractions (single amino acids, dipeptides or tripeptides). Most exopeptidases of the human gastrointestinal tract are produced by enterocytes at the brush border membrane of the small intestine. Inflammatory damage to the intestinal epithelium in coeliac disease is associated with reduced brush border exopeptidase activity [38, 39], and this enzyme loss is likely to exacerbate gluten peptide accumulation. Therapeutic approaches supporting the digestion capabilities of the intestinal brush border, and in particular exopeptidases, remain overlooked despite the critical role of these enzymes in dietary peptide digestion. In this Review, we discuss the potential role of exopeptidases in achieving thorough gluten peptide degradation and highlight issues of gastrointestinal physiology and motility that are critical to digestive enzyme therapy design for improved therapeutic efficacy.

2. Peptide Digestion Throughout the Gastrointestinal Tract: Motility and Peptidase Activity

Central to digestive enzyme therapy design is ensuring that exogenous enzymes are sufficiently exposed to their target substrates. As such, gastrointestinal motility is a key consideration for gluten‐digesting enzyme therapies, particularly with regard to drug transport and delivery.

3. Gastric Digestion: Motility, Mixing and Pepsin Activity

Protein and peptide digestion involve complex mechanical and enzymatic processes that begin primarily in the stomach where as little as 10%–15% of protein digestion takes place [40]. Pepsin is active only in acidic environments and is the chief proteolytic enzyme of the stomach (Table 3). Pepsin has limited activity on gluten as its preferred substrates, hydrophobic amino acids, are often adjacent to proline for which it has no cleaving activity. In turn, partial digestion of gluten by pepsin generates long gluten peptide fragments in vitro [41, 42] and in the stomach of healthy volunteers [19]. Enzyme performance is dependent on pH, and regional differences in gastric acid secretion create a pH gradient with values higher towards the antrum and lower towards the fundus [43]. Between the fed and fasted states, acidity in the stomach can fluctuate between pH 1 and 8 [43, 44, 45], influenced by meal conditions and other factors [46]. Consequently, gastric pH heterogeneity is an important consideration for enzyme therapies targeting gluten in the stomach. For instance, pepsin has a pH optimum of 2, loses half its activity at pH 3 and is fully inactivated when pH rises above 5 (reviewed elsewhere [47]). Variations in pH across the stomach are also evident from regional differences in pepsin activity [48].

TABLE 3.

Summary of endogenous endopeptidase and exopeptidase activity in the stomach and small intestine.

Enzyme name	Enzyme class	Enzyme commission #	Site of enzyme activity	Production site	Substrate preferences	Examples
Pepsin	Endopeptidases	3.4.23.1	Stomach	Gastric chief cells	Cleaves after hydrophobic or aromatic residues (e.g. phenylalanine, tryptophan and tyrosine)	F, W, or Y↓X
Trypsin		3.4.21.4	Small intestine	Pancreas	Cleaves after basic residues (lysine and arginine)	K, or R↓X
Chymotrypsin		3.4.21.1			Cleaves after hydrophobic residues (phenylalanine, tryptophan, tyrosine, methionine and leucine)	F, W, Y, M, or L↓X
Elastase		3.4.21.36			Cleaves after neutral aliphatic residues (serine, leucine, alanine and valine)	S, L, A, V↓X
Carboxypeptidase A	Exopeptidases	3.4.17.1			Cleaves at C‐terminus of aromatic or aliphatic residues (e.g. leucine, isoleucine, alanine and valine)	X↓L, I, A, or V
Carboxypeptidase B		3.4.17.2			Cleaves at C‐terminus of basic residues (lysine and arginine)	X↓K, or R
Dipeptidyl peptidase IV (DPP‐IV)		3.4.14.5		BBM	Cleaves at N‐terminus of X‐Proline dipeptides (preferred) or X‐Alanine dipeptides	X1P↓X2 or X1A↓X2
Aminopeptidase N (APN)		3.4.11.2			Cleaves at N‐terminal residues, preferentially alanine but including proline	A, or P↓X
Aminopeptidase P (APP)		3.4.11.9			Cleaves at N‐terminal residues before a proline	X↓P
Carboxypeptidase P (CPP)		3.4.17.16			Cleaves at C‐terminal residues with preference for proline	X↓P
Dicarboxypeptidase I (DCP‐I)/Angiotensin‐converting enzyme (ACE)/dipeptidyl peptidase A		3.4.15.1			Cleaves at C‐terminal dipeptides X1↓X2X3 where X2 is not proline and X3 is not aspartic acid or glutamic acid

Note: This table summarises the preferred substrates of gastrointestinal peptidases and includes a subset of brush border membrane (BBM) peptidases with an emphasis on proline‐digesting exopeptidases [49, 50]. Additional information is also available on databases including Brenda‐Enzymes.org and UniprotKB.com.

Limited mixing in the stomach, whereby ingested material is prone to separation based on composition and density [43, 51, 52], provides an additional layer of intragastric heterogeneity. Briefly, different intragastric motor patterns govern stomach motility: fundic contractions lead to an accumulation of food material in the antrum, and antral contraction waves contribute to trituration of food particles, facilitating their mechanical grinding into smaller particles (e.g. < 1 mm) for passage into the duodenum [53]. Poor gastric mixing of ingested material, including drugs, is well documented. Magnetic resonance imaging studies have demonstrated layering of fat fractions in the fundus and solid particles in the antrum [54], and little to no mixing between the chyme and an ingested contrast marker, which largely bypassed the meal when emptied into the duodenum [55]. Moreover, drugs with extended‐release formulation have been shown to float atop gastric content in the fundus [51, 56, 57]. Current clinical trials for enzyme therapies targeting gluten digestion in the stomach have relied on homogenised meals [35] or high drug doses before gluten intake [58], which may have artificially increased intragastric mixing. Studies with ‘typical’ meal conditions will provide an improved understanding of the potential effects of limited mixing on stomach‐targeting ‘glutenases’.

Gastric emptying is a complex process that starts at the beginning of a meal, continues thereafter for several hours and differs fundamentally between solids and liquids. Solid foods undergo trituration followed by constant emptying under enterogastric control, independently of intragastric volume [52]. Notably, larger particles can be retained in the stomach for several hours after a meal through gastric sieving [43, 51, 53, 59]. In contrast, water, low‐calorific liquids and small particles can be rapidly emptied into the duodenum via a dedicated route that can bypass chyme in the stomach [51]. Sometimes referred to as ‘Magenstrasse’, or ‘stomach road’, this phenomenon has been demonstrated in both fasted and fed states, with limited effects of meal fat content [54, 60, 61], and key implications for rapidly dissolving drugs or liquid formulations [54]. For instance, caffeine in water can be emptied within 10 min, with maximum absorption of salivary caffeine reached by 20 min [62]. Additionally, Grossmann et al. demonstrated rapid emptying of manganese‐labelled water in 10–30 min, with emptying most pronounced in fasted or light meal conditions but consistently observed even in high‐fat meals [60].

Overall, limited intragastric mixing and rapid emptying of water and small particles (illustrated in Figure 1) raise challenges for gluten‐digesting enzyme therapies. Features of gastrointestinal physiology such as fluid dynamics and pH gradients that impact enzyme activity are challenging to model in vitro [63], and it is unclear whether exogenous enzymes targeting the stomach can digest clinically relevant gluten dosages in ‘real’ meal conditions before being emptied into the duodenum. Furthermore, the environment of the small intestine, such as rising pH and numerous digestive enzymes, was shown to inactivate or greatly reduce the activity of certain exogenous enzymes [20, 29]. This issue is particularly important for enzyme combinations with different stability criteria, as illustrated by EP‐B2 and SC‐PEP, an enzyme combination under investigation for coeliac disease. Whereas EP‐B2 relies on acidic pH and is vulnerable to inactivation/trypsin digestion in the small intestine [29], SC‐PEP requires neutral pH and is digested by both pepsin [30] and trypsin [31]. This discordance may be further accentuated by reports of EP‐B2 monotherapy enhancing gluten immunogenicity in vitro compared to pepsin alone [64]. Suboptimal performance of SC‐PEP in the stomach has been recognised by other groups working on genetically modifying this enzyme to increase its gastric stability [65]. Ideally, modifications of stability would also account for intestinal conditions to address the possibility of rapid emptying into the duodenum.

FIGURE 1.

Physiological flow of dietary protein and peptide digestion in the stomach and small intestine. (1) Stomach. Mixing and motility: In the stomach, food may separate into poorly mixing layers based on size and density, with larger particles retained for longer periods for digestion in the stomach (gastric sieving) and small particles emptied into the duodenum (gastric emptying). Water, low‐calorific fluids and small particles may be rapidly emptied into the duodenum (light blue arrow), resulting in limited mixing with food in the stomach. Enzymatic digestion: The gastric endopeptidase pepsin cleaves proteins and peptides, which can generate long peptide fragments. (2) Small intestine. Mixing and motility, upper left panel: Inconsistent muscular contractions achieve thorough mixing of chyme in the small intestine. Chyme diffuses around adjacent villi and microvilli that form the brush border membrane, enhancing interactions between brush border membrane enzymes and peptides. Lower left panel: Patients with coeliac disease experience villous atrophy which often extends throughout the duodenum and jejunum [66, 67] and can impair nutrient absorption. Enzymatic digestion. Pancreatic peptidases: Endopeptidases trypsin, chymotrypsin and elastase, and exopeptidases carboxypeptidase A and B, digest peptides into smaller fragments. Brush border membrane peptidases (healthy): Brush border membrane peptidases are important for completing peptide digestion into short, absorbable fractions (single amino acids, dipeptides and tripeptides) that are taken up by enterocytes to reach the circulation. In healthy individuals, brush border enterocytes harbour numerous digestive enzymes including the majority of exopeptidases (e.g. DPP‐IV and APN), most of which are membrane‐bound or secreted in membrane vesicles. In vitro data indicate that the addition of porcine jejunal brush border membrane extracts to gastroduodenal simulations significantly increases the percentage of protein digested [68] and results in higher concentrations of free amino acids [69]. Brush border membrane peptidases (coeliac disease): Patients with coeliac disease can exhibit reduced brush border exopeptidase activity [38, 39, 70] caused by enterocyte death and damage, which may interfere with the critical functionality of the brush border membrane in processing proline‐rich peptides and generating absorbable single amino acids.

4. Small Intestine: Motility and Mixing

Following gastric emptying, the chyme enters the small intestine, the main site for protein and peptide digestion along the length of the duodenum, jejunum and ileum. The interior surface of the small intestine is lined with projections called villi comprised of a monolayer of epithelial cells, the majority of which are enterocytes. Along the apical surface of enterocytes are smaller projections, called microvilli, that collectively form the brush border membrane resulting in a total digestive and absorptive surface area of 4500 square meters [71].

Mixing and motility in the small intestine, as reviewed by Lentle et al. [72], are critical for ensuring thorough peptide digestion. Briefly, mixing is driven by localised folding and kneading of intestinal contents by segmenting and peristaltic contractions, which achieve thorough dispersion of chyme between and around villi. Contractile activity is inconsistent in site, direction and distance of propagation. Postprandial contractile activity achieves a relatively slow rate of luminal mixing, allowing for thorough diffusion of enzymes into the chyme. Higher viscosity of chyme after a meal relative to intermeal periods suggests prolonged residence time of enzymes for thorough digestion. With regard to the spreading of liquids, Grossmann et al. demonstrated rapid and broad dispersion of manganese‐labelled water along the length of the small intestine, with comparable spreading regardless of meal conditions (fasted, light or heavy meals) [60]. Finally, pH remains relatively stable and homogenous in the small intestine, achieving a consistent environment for enzyme activity. For instance, Koziolek et al. showed that while pH fluctuated between 1 and 8 in gastric transit, only slight increases in pH were detected in the proximal (5.9–6.3) and distal (7.4–7.8) small intestine [45].

Small intestinal transit typically lasts around 4–6 h [71, 73, 74] before material is emptied into the large intestine. Epithelial cell turnover occurs every 4–5 days in the mammalian small intestine [75], and mucus secretion and renewal happen more frequently (e.g. every 2–4 h) [76]. The time course of digestion in the small intestine, together with mucosal renewal dynamics, suggests sufficient lengths of enzyme–substrate exposure for thorough digestion.

5. Small Intestine: Pancreatic and Brush Border Membrane Peptidase Activity

The small intestine harbours a diverse array of endopeptidases and exopeptidases, which derive both from pancreatic secretions and brush border enterocytes. These enzymes act in concert to digest peptides into absorbable fractions (Figure 1). Pancreatic endopeptidases (trypsin, chymotrypsin and elastase) and exopeptidases (carboxypeptidases A and B) have distinct substrate specificities (Table 3). Notably, these enzymes are inhibited by proline‐adjacent bonds and have limited cleavage activity for proline‐enriched regions of gluten proteins, as demonstrated by numerous in vitro studies [1, 42, 77, 78, 79, 80, 81]. For instance, digestion of immunodominant α‐ and γ‐gliadin peptides revealed almost all tested peptides to be largely resistant to trypsin, chymotrypsin, elastase and carboxypeptidase A digestion, with ≥ 90% of peptides remaining after 4 h [80]. Chymotrypsin digestion of α‐gliadin in vitro generated peptides that were toxic to enterocytes of coeliac biopsy specimens [77]. Chymotrypsin activity appears limited to ‘immunologically silent’ regions, whereas proteolytically resistant regions harbouring proline‐ and glutamine‐rich immunogenic epitopes remain intact [1]. Indeed, chymotrypsin treatment is routinely used to liberate immunogenic peptides from insoluble wheat gluten for in vitro use in studies of gluten‐specific T cells [82]. Interestingly, bacterial peptidase activity in the gastrointestinal tract also appears to affect gluten digestion, and coeliac duodenal biopsies showed elevated levels of bacterial elastase‐like activity which enhanced the inflammatory response to gluten in murine models [83]. Overall, gluten digestion by endogenous human and bacterial endopeptidases in the gastrointestinal tract can increase the immunogenic properties of gluten peptides.

Brush border exopeptidases are critical for the systematic digestion of short peptides into absorbable fractions. The brush border contains an array of at least 20 known exo‐ and endopeptidases, with exopeptidases being the most abundant and active [50, 84]. Brush border peptidases are anchored to enterocytes or secreted into the lumen via brush border membrane vesicles. Most peptide digestion and absorption occur in the jejunum [85], where approximately 70% of ingested protein remains as peptides [86, 87]. The jejunal architecture, with enlarged villi to further increase surface area [71], maximises enzyme–substrate exposure as chyme diffuses in and around villi. Brush border peptidases have varying substrate preferences (Table 3) and chain‐length restrictions, with preferential activity on substrates up to 20–25 amino acids in length [84]. Notably, several brush border exopeptidases hydrolyse proline‐adjacent bonds, including the aminopeptidases dipeptidyl peptidase IV (DPP‐IV), aminopeptidase N (APN), aminopeptidase P (APP), dipeptidyl carboxypeptidase I (DCP‐I)/angiotensin‐converting enzyme (ACE) and carboxypeptidase P (CPP) [49, 50, 88]. The digestion of proline‐containing peptides by brush border exopeptidases appears to be slow and rate‐limiting. For instance, gliadin digestions in vitro have shown that several peptides stable to gastric and pancreatic peptidases were digested to < 10%–20% of the starting material by rat brush border membrane peptidases, though this process required long (e.g. 4 h) exposures [80]. Furthermore, treatment with brush border membrane extracts from duodenal biopsies resulted in a buildup of DPP‐IV substrates, highlighting the rate‐limiting role of DPP‐IV in the amino‐terminal digestion of gluten immunogenic peptides [79]. Similarly, dipeptidyl carboxypeptidase I (DCP‐I), expressed at low levels in the BBM, appeared rate‐limiting in its carboxy‐terminal digestion [79].

Given the rate‐limiting role of brush border exopeptidases in gluten immunogenic peptide digestion, Hausch et al. have proposed the possibility that subtle deficiencies in their activity may increase the probability of intestinal damage in coeliac patients and that exopeptidase supplementation could compensate for the intrinsically slow processing of proline‐containing peptides by the brush border membrane [79]. Indeed, the critical role of brush border exopeptidases in completing the digestion of proline‐rich peptides underscores a rationale for supplementing this enzyme class to enhance gluten digestion. Furthermore, the small intestine, with its extensive surface area, thorough mixing of food particles with digestive enzymes, relatively stable pH and diverse supporting enzymatic machinery, may be well suited as a target organ for fast‐acting gluten‐digesting enzyme therapies.

6. New Therapeutic Principle: Combinatorial Exopeptidase Supplementation for Gluten Peptide Digestion

Exopeptidases remain underexplored for the treatment of coeliac disease due in part to our incomplete understanding of the human brush border membrane, which is still undergoing characterisation [84, 89, 90, 91], and by commercial unavailability of the necessary enzymes [90, 92]. Standardised models of human digestion, for example INFOGEST (a standardised, physiologically relevant, static in vitro digestion model developed by the INFOGEST international consortium), lack an intestinal brush border membrane step despite its critical importance in peptide digestion [49, 63, 93]. Some studies incorporate brush border membrane extracts from animal models. For instance, porcine [90] and rat jejunum brush border membranes [91] are considered suitable models and have been found to contain a large presence of aminopeptidase N (APN), with levels of APN and DPP‐IV activity comparable between rat and human duodenal biopsies [79]. However, standardised conditions for use of these enzymes, for example incubation times, pH and physiologically relevant enzyme–substrate ratios, are still lacking [49, 63, 69].

Supplementing brush border exopeptidase activity may represent a promising therapeutic avenue when considering the extensive enteropathy that is documented in coeliac disease throughout the duodenum and jejunum [66, 67]. Coeliac intestinal biopsies have revealed incomplete mucosal recovery and decreased activity of certain duodenal and jejunal enzymes (e.g. lactase and sucrase‐isomaltase) even after several years on a gluten‐free diet [94, 95]. Studies assessing changes in brush border peptidase activity in coeliac patients reveal diminished activity of proline‐specific exopeptidase DPP‐IV [38, 39, 70], which was reduced by an average of 70% in coeliac intestinal biopsies compared to controls [39]. Reduced activity of exopeptidases DPP‐IV and alanyl aminopeptidase (aminopeptidase N, APN) has been observed in patients even in disease remission [70]. Interestingly, gene expression profiles, including reduced expression of APN, remained significantly altered in coeliac patients following a strict long‐term gluten‐free diet when compared to healthy controls, even when patients experienced morphological recovery (villus height‐to‐crypt depth ratio) and reduced intestinal inflammation (intraepithelial lymphocytes, IEL) [95]. Some studies have reported differences in gluten peptide digestion between enterocytes from healthy volunteers, patients with coeliac disease managed by a gluten‐free diet, and patients with active coeliac disease, which appeared to stem from differences in brush border peptidase activity [96]. Overall, these studies imply that enterocyte function impaired by chronic inflammation does not necessarily recover even when the inflammation is clinically managed by the gluten‐free diet.

To date, DPP‐IV is the most studied exopeptidase that has been investigated as a gluten‐digesting enzyme supplement, either as a single agent or in combination with endopeptidases. Exopeptidases, however, are ineffective as single agents given their limited substrate scopes and, in turn, rely on peptide trimming at either terminus to facilitate access to target substrates. Indeed, a study on Aspergillus oryzae exopeptidase DPP‐IV and Aspergillus niger endopeptidase PEP (AN‐PEP) found that though neither enzyme efficiently cleaved synthetic gluten immunogenic peptides as a single agent, they efficiently digested these peptides when applied in combination in vitro, including in the presence of competitive substrates (casein) and in whole wheat bread [22]. The authors postulated that an abundance of competitive substrates may slow endopeptidase activity; thus, the addition of an exopeptidase could enhance gluten digestion by clearing short peptides and facilitating endopeptidase activity on longer peptides [22]. In this study, exopeptidase supplementation was considered as a secondary mode‐of‐action to enhance endopeptidase activity. Similarly, Hausch et al. showed that supplementing gluten immunogenic peptide digestion with exogenous fungal DPP‐IV or, separately, with rabbit lung DCP‐I, was sufficient to digest gluten immunogenic peptides completely within 4 h, even when they were resistant to gastric, pancreatic and rat brush border peptidases [79]. However, due to the dependence of DPP‐IV and DCP‐I on access to free amino‐ and carboxy‐termini, respectively, greater emphasis was again placed on gluten peptide cleavage by exogenous prolyl endopeptidase (PEP) supplementation [79] which, by virtue of the endopeptidase mode‐of‐action, is able to access substrates relatively independently.

Evidently, brush border exopeptidases present a critical enzyme mode‐of‐action for the complete digestion of peptides, including proline residues, into absorbable fractions. Numerous studies have demonstrated that higher dosages of exogenous (e.g. fungal) brush border exopeptidases can accelerate and complete gluten immunogenic peptide digestion [79]. Given that gluten peptides less than 9 amino acids long should have low avidity for human leucocyte antigen (HLA)‐DQ molecules involved in coeliac disease (HLA‐DQ2.5, HLA‐DQ2.2 and HLA‐DQ8) [97, 98], the exopeptidase mode‐of‐action presents a systematic approach to neutralising gluten peptide immunogenicity irrespective of what coeliac disease‐predisposing HLA‐DQ haplotype a patient might possess. Therefore, supplementing brush border exopeptidase activity, which is rate‐limiting in gluten immunogenic peptide digestion [79] and exhibits apparent differences between coeliac patients and healthy volunteers [38, 39, 70], represents a promising new avenue for gluten‐digesting enzyme therapeutics. An important consideration will be drug delivery to the small intestine, particularly if these enzymes are sensitive to acidic pH. Specific formulations may be needed to protect enzymes during gastric passage, while ensuring rapid release and sufficient mixing with the chyme upon entry in the duodenum. For instance, pancreatic enzyme replacement therapy has made use of enteric‐coated microspheres, which demonstrated increased mixing over tablets [99].

7. Current Status and Clinical Evaluations of Gluten‐Digesting Enzymes

Several gluten‐digesting enzymes are currently in development as prescription drugs for coeliac disease, while others target noncoeliac gluten sensitivity through the consumer health market. It is important to note that consumer health products claiming to digest gluten, or gluten immunogenic peptides, are not approved as treatments for coeliac disease. In general, gluten‐digesting enzymes share a specificity for proline‐ or glutamine‐containing substrates, an endoproteolytic mode‐of‐action and an acidic pH optimum (Table 2). There are a few exceptions, namely the single exopeptidase DPP‐IV and the endoprotease Caricain acting in the small intestine. Of note, assessment of substrate specificity across gluten‐digesting enzymes has not been standardised, and often cleaving efficacy for a given substrate is not clear despite reported activity. Importantly, low cleaving efficacy can result in delayed or slow activity for a given enzyme (see Table 2). Likewise, enzyme vulnerabilities are not always systematically assessed (e.g. endogenous enzyme digestion and/or irreversible denaturation by pH or bile salts).

To date, only stomach‐acting endopeptidases have advanced to clinical trials. In these trials and in proof‐of‐concept clinical studies with gluten‐digesting enzymes, various approaches have been used to deliver and facilitate enzyme activity in the stomach (Table 4). Aspects of gastric mixing and emptying, however, have been largely overlooked. For instance, a clinical study with AN‐PEP delivered the enzymes as a liquid mixed with a homogenised meal directly into the stomach via catheter [19]. Gastric and duodenal aspirates were then obtained, and the authors reported a significant improvement in gluten immunogenic peptide digestion [19]. It stands to reason that this study design would have artificially increased gastric mixing and enzyme‐to‐substrate exposure. In a follow‐up study, AN‐PEP was taken by patients at home with their typical gluten‐free meals, and no significant differences were observed in serology, symptoms or levels of stool gluten immunogenic peptides [100]. It is unclear to what extent the lack of gluten challenge, and/or the incorporation of realistic meal conditions, affected these outcomes. Real‐meal conditions have been assessed with Latiglutenase. In a phase 2 trial, patients ate their normal gluten‐free diets at home and were instructed to consume a high dose of Latiglutenase, dissolved in approximately 230‐mL water, before and after a gluten challenge (1.2 g Latiglutenase:2 g gluten) [58]. Though drug intake on either side of gluten ingestion may improve mixing, it is worth noting that when the drug is dissolved in water it may be emptied from the stomach faster than anticipated.

TABLE 4.

Summary of clinical data on gluten‐digesting enzymes in healthy volunteers or coeliac disease patients.

Product— Clinicaltrials.gov number—study doi	Approach	Dosing regimen and formulation	Gluten challenge	Diet	Readouts & outcomes
					Vh:Cd	IEL	Serology	Symptoms	GIPs
Latiglutenase NCT01917630 10.1053/j.gastro.2016.11.004	Phase 2, double‐blind, placebo‐controlled, dose‐ranging study n = 494 coeliac patients (412 completed through study week 12) GFD ≥ 1 year	100–900 mg, up to 3× daily with meals for 12 or 24 weeks Powder dissolved in water, consumed in first half of meal	N/A	GFD, at home (no study foods given)	NS	NS	CD3+ NS	Patient‐reported outcome symptom diary NS	N/A
Latiglutenase NCT01917630 10.1002/ygh2.371	Retrospective analysis of the above study Seropositive patients, 900‐mg drug (n = 14) versus placebo (n = 54)	See above	N/A	See above	N/A	N/A	N/A	Improvement in abdominal pain, bloating and constipation NS change in nausea or diarrhoea	N/A
Latiglutenase NCT03585478 10.1053/j.gastro.2022.07.071	Phase 2, double‐blind, placebo‐controlled n = 43 patients completed GFD ≥ 1 yr	1200‐mg drug dissolved in water Drug ingested before and after 2‐g gluten intake	2‐g gluten	Added to regular gluten‐free meal	NS	Significant reduction in mean IEL for study drug vs placebo	NS	NS mean change in symptom severity for abdominal pain, bloating and tiredness 3‐ × 2‐week trend line for these symptoms	Very significant reduction in urine GIPs, indicating 95% gluten degradation by study drug
TAK‐062 NCT03701555 10.1053/j.gastro.2021.03.019	Phase 1, dose‐escalation study Safety, n = 15 healthy participants, 9 coeliac patients GIP degradation, n = 43 healthy participants Gluten‐free for 24 h (healthy participants) Well‐controlled disease on GFD (celiac patients)	100‐, 300‐ or 900‐mg drug Capsules or liquid formulation	1‐ to 6‐g gluten	Complex meals, blended to homogenise, ingested within 10 min of study drug administration	N/A	N/A	N/A	N/A	Gastric samples, median gluten degradation 97%–99%, 20‐ to 65‐min postdose
AN‐PEP NCT01335503 10.1111/apt.13266	Phase N/A Double‐blind, randomised, placebo‐controlled, Crossover study n = 12 healthy volunteers	Dosage (mg) not specified, 1,600,000 Protease Picomol International Dissolved in water	4‐g gluten	Low‐ or high‐calorie meals, blended to homogenise, delivered to the stomach through catheter	N/A	N/A	N/A	NS Mild GI symptoms reported in both AN‐PEP and placebo groups	AN‐PEP significantly reduced gluten levels in gastric and duodenal aspirates
AN‐PEP NCT04788797 10.3748/wjg.v30.i11.1545	Phase 4, exploratory, double‐blind, randomised, placebo‐controlled n = 40 coeliac patients GFD > 2 years	650‐mg drug Capsules Drug taken with each of 3 meals, over 4 consecutive weeks	N/A	GFD, at home (no study foods given)	N/A	N/A	NS	NS	NS difference in stool GIPs

Note: Collected endpoints varied between studies, with the FDA‐recommended endpoints being symptoms and epithelial healing (Vh:Cd).

Abbreviations: AN‐PEP, Aspergillus niger prolyl endopeptidase; GFD, gluten‐free diet; GI, gastrointestinal; GIPs, gluten immunogenic peptides; IEL, intraepithelial lymphocytes; N/A, not applicable; NS, not significant; Vh:Cd, villous height: Crypt depth ratio.

Another important aspect of trial design for gluten‐digesting enzyme therapies is the relevance of clinical endpoints. The Food and Drug Administration Draft Guidance for Industry recommends that trials intending to support marketing approval of coeliac disease therapies assess improvement in symptoms and mucosal inflammation [101]. These and other endpoints have been assessed by authors in enzyme therapy trials (summarised in Table 4), though with apparent discordance between symptomology, histology and enzyme efficacy outcomes. For instance, some studies have sought to directly measure enzyme efficacy with semiquantitative enzyme‐linked immunosorbent assays (ELISA) measuring specific gluten immunogenic epitopes. Interestingly, though several trials have concluded that their endopeptidase enzymes achieve very high levels of gluten degradation, currently these findings have not been corroborated with histology and symptoms [58] (see Table 4). Specifically, though a Latiglutenase study reported 95% + degradation of gluten in urine using ELISA, the authors found no significant improvement in overall symptoms and mucosal healing (villous height‐to‐crypt depth ratio (Vh:Cd)) [58]. The observed discrepancy between these endpoints may be explained, at least in part, by the fact that ELISAs are semiquantitative and are not comprehensive of all coeliac disease‐relevant gluten epitopes [102]. Indeed, urine peptidomic analyses have revealed high variability in gluten peptides between coeliac patients, which the authors postulated may be due to interindividual differences in protein digestion and gastrointestinal damage [102]. In addition, the extent of mucosal damage and associated loss of brush border enzyme activity may further contribute to peptide diversity in coeliac patients when compared to healthy individuals.

Symptoms in patients with untreated coeliac disease are highly variable, ranging from acute nausea and vomiting to chronic gastrointestinal symptoms such as diarrhoea and systemic manifestations such as tiredness and headache [103, 104]. Notably, symptoms in patients with poorly controlled coeliac disease tend to be irritable bowel syndrome (IBS)‐like, which may be due to inflammation in the small intestine and visceral hypersensitivity [105]. Similarly, symptoms may stem from damage to the intestinal brush border membrane, which is associated with loss of brush border digestive enzymes [106], nutritional deficiencies [107, 108] and malabsorption [109]. Nausea and vomiting are relatively sensitive indicators of T cell‐driven systemic cytokine release, in both timing and severity, in patients with coeliac disease after they consume gluten or after systemic (in vivo) administration of peptides that activate gluten‐specific T cells in vitro [6, 7, 110]. In contrast, IBS‐like symptoms without nausea (e.g. bloating and gas) may not be linked to gluten exposure [110]. For example, a large‐scale study with Latiglutenase (no gluten challenge) found no significant differences across endpoints including symptoms [111] but the authors then published a retrospective analysis in a subgroup of seropositive patients reporting improvements in certain symptoms (abdominal pain, bloating and constipation) but not others (nausea and diarrhoea) [112]. However, a follow‐up trial with Latiglutenase incorporating a gluten challenge did not show symptom improvements, including nausea [58] (see Table 4).

Some drugs in development have produced promising early clinical data demonstrating their mode‐of‐action, and these may yet prove successful as adjuncts to the gluten‐free diet. Currently used endpoints, however, may be inadequate given that gluten‐digesting enzymes do not promote active healing of intestinal damage and symptoms resulting from lower gluten exposure are inconsistent. There is a need for new, clinically meaningful endpoints to better assess the efficacy of gluten‐digesting enzymes. Blood interleukin‐2 (IL‐2) has been proposed as a sensitive surrogate biomarker of gluten‐induced immune activation [113, 114]. Improved ultrasensitive detection methods indicate that elevation of interleukin‐2 can be an extremely sensitive measure of gluten exposure in coeliac disease with over 90% of patients on a gluten‐free diet showing elevated plasma interleukin‐2 4 h after 1‐g gluten ingestion (unpublished data presented at conference [115]). Magnitude of interleukin‐2 elevation after 10‐g vital gluten correlates with severity of symptoms such as vomiting and nausea, and is also correlated with gluten peptide‐stimulated whole blood interleukin‐2 release ex vivo, as well as major histocompatibility complex (MHC)‐peptide tetramer‐determined frequency of gluten‐specific T cells in blood (unpublished data presented at conference [116]). Interleukin‐2 release in vivo demonstrates a dose–response relationship with as little as 3‐mg elevating interleukin‐2 in a third of patients. Establishing gluten dose–response curves in patients with and without detoxification by gluten‐digesting enzymes may provide compelling evidence of efficacy and indicate whether very ‘sensitive’ patients might be protected from trace contaminants of gluten.

8. Current and Future Applications

Current gluten‐digesting drugs in development aim to mitigate accidental gluten exposure. This form of therapy could be particularly valuable in situations in which adhering to dietary restrictions is challenging (such as during travel, dining out or attending social gatherings). Enzyme therapies may also apply to patients suffering from villous atrophy with type 1 refractory or nonresponsive coeliac disease, and to those with intermittent symptoms attributable to gluten, particularly when linked to acute gluten reactions such as nausea and vomiting [110]. In this case, rather than acting as a preventative tool, digestive enzyme therapies may take the role of rescue therapies that restore or replace key digestive enzymes lost due to chronic inflammatory damage to the small intestine. This inflammatory damage may have an especially pronounced effect on exopeptidase activity given that most human gastrointestinal exopeptidases originate from enterocytes and that these enzymes provide the gastrointestinal tract with the ability to cleave proline, an amino acid that is abundant in gluten but is also present in multiple dietary peptides associated with gastrointestinal distress (e.g. caseins [117] and collagens [118]).

9. Conclusions

Herein, we have highlighted challenges in the design of orally administered peptidases as therapeutics for coeliac disease. Years of pharmacological development have illustrated the putative effects of gastric emptying on drug transport, particularly for drugs with rapid‐dissolving or liquid formulations, and the risks of poor intragastric mixing between drugs and meal content. In order to ensure that physiologically relevant quantities of gluten peptides are digested in typical meal conditions, drug delivery and gastrointestinal motility should be central to enzyme therapy designs.

In healthy individuals, the proximal small intestine (i.e. jejunum) is the major site of protein and peptide digestion, owing to thorough mixing, a large surface area and a battery of digestive enzymes that together promote optimal enzyme–substrate interactions. As such, fast‐acting enteric enzymes that leverage natural gastrointestinal physiology may sufficiently neutralise gluten immunogenic peptides to prevent an immunological response in the proximal small intestine. While exogenous enzymes will ideally be stabilised for activity throughout the gastrointestinal tract, the processes described herein highlight the importance of enzyme activity in the small intestine. As mentioned, some groups are developing enzymes with broader pH stability, and improvements in computational enzymology are accelerating these efforts.

Gluten‐digesting enzyme therapies aim to neutralise gluten immunogenic peptides by reducing them to harmless fractions. Importantly, exogenous enzymes are likely to be ineffective if they increase the load of gluten peptides that retain or enhance the immunogenicity of gluten compared to endogenous digestion. The exopeptidase mode‐of‐action presents an interesting approach for consistent trimming of gluten immunogenic peptides into harmless, absorbable fractions. Exopeptidases can rapidly digest diverse peptides from one terminus to the other, but given their specific substrate preferences, as single agents they will ultimately encounter residues that they cannot cleave. Consequently, single exopeptidases are ineffective alone but may be highly effective when combined. Indeed, the human brush border membrane contains a diverse range of complementary exopeptidases, many of which, unlike gastric and pancreatic peptidases, have proline specificity. Interestingly, the sensitivity of the intestinal brush border to inflammatory damage and consequent loss of brush border exopeptidase activity in coeliac disease patients [38, 39, 70] may further exacerbate gluten immunogenic peptide buildup and ensuing enteropathy in patients. To date, however, there has yet to be a rational design for exopeptidase supplementation in coeliac disease leveraging complementary exopeptidase activity. Future approaches to enzyme therapy design that incorporate exopeptidase combinations stand to enhance or replace a critical enzyme type in the small intestine, while harmonising with the natural flow of dietary protein and peptide digestion.

10. Literature Search Strategy

For this Review, a broad search strategy was used to identify relevant literature and inform the discussion. Searches were conducted in databases including PubMed, Web of Science and Google Scholar using a range of keywords related to the topic including ‘coeliac disease’/‘celiac disease’ ‘exopeptidase’, ‘endopeptidase’, ‘endoprotease’, ‘glutenase’ and ‘enzyme therapy’. Relevant publications on drug delivery and gastrointestinal physiology were also included. References from key papers and recent reviews were manually screened to identify further relevant studies. While the search was guided by the aim of comprehensively covering the topic, the selection of literature also reflects the authors' perspectives and judgement in highlighting key themes.

Author Contributions

Erin R. Bonner: writing – original draft. Werner Tschollar: writing – review and editing. Robert Anderson: writing – review and editing. Sulayman Mourabit: writing – original draft.

Conflicts of Interest

Erin R. Bonner has a consulting agreement with AMYRA Biotech AG. Werner Tschollar is co‐founder of and shareholder in AMYRA Biotech AG. Robert Anderson has a consulting agreement with AMYRA Biotech. Robert Anderson is a shareholder and director of Novoviah Pharmaceuticals Pty Ltd., a named inventor of, but has no financial interest in patents relating to celiac disease diagnostics and therapies. Robert Anderson is a consultant to Celiac Disease Foundation, Takeda Pharmaceutical Company Ltd./Millennium Pharmaceuticals, Forte Bioscience Inc., Barinthus Biotherapeutics, Immunic AG, Topas Therapeutics GmbH, Universal Cells—ASTELLAS LLC US, DBV (Paris, France, since 2021‐), EVOQ Therapeutics, Allero Therapeutics BV and Kanyos Bio Inc. Sulayman Mourabit is an employee at AMYRA Biotech AG.

Data Availability Statement

The authors have nothing to report.