Molecular and in vivo studies of a glutamate-class prolyl-endopeptidase for coeliac disease therapy

del Amo-Maestro, Laura; Mendes, Soraia R.; Rodríguez-Banqueri, Arturo; Garzon-Flores, Laura; Girbal, Marina; Rodríguez-Lagunas, María José; Guevara, Tibisay; Franch, Àngels; Pérez-Cano, Francisco J.; Eckhard, Ulrich; Gomis-Rüth, F. Xavier

doi:10.1038/s41467-022-32215-1

Download PDF

Article
Open access
Published: 01 August 2022

Molecular and in vivo studies of a glutamate-class prolyl-endopeptidase for coeliac disease therapy

Nature Communications volume 13, Article number: 4446 (2022) Cite this article

8864 Accesses
6 Citations
222 Altmetric
Metrics details

Subjects

Abstract

The digestion of gluten generates toxic peptides, among which a highly immunogenic proline-rich 33-mer from wheat α-gliadin, that trigger coeliac disease. Neprosin from the pitcher plant is a reported prolyl endopeptidase. Here, we produce recombinant neprosin and its mutants, and find that full-length neprosin is a zymogen, which is self-activated at gastric pH by the release of an all-β pro-domain via a pH-switch mechanism featuring a lysine plug. The catalytic domain is an atypical 7+8-stranded β-sandwich with an extended active-site cleft containing an unprecedented pair of catalytic glutamates. Neprosin efficiently degrades both gliadin and the 33-mer in vitro under gastric conditions and is reversibly inactivated at pH > 5. Moreover, co-administration of gliadin and the neprosin zymogen at the ratio 500:1 reduces the abundance of the 33-mer in the small intestine of mice by up to 90%. Neprosin therefore founds a family of eukaryotic glutamate endopeptidases that fulfils requisites for a therapeutic glutenase.

Single-site iron-anchored amyloid hydrogels as catalytic platforms for alcohol detoxification

Article Open access 13 May 2024

Periportal macrophages protect against commensal-driven liver inflammation

Article 24 April 2024

Therapeutic peptides: current applications and future directions

Article Open access 14 February 2022

Introduction

Coeliac disease (CoD) is a chronic autoimmune enteropathy that affects individuals with genetic and environmental sensitization to dietary gluten, a group of cereal prolamin storage proteins rich in proline and glutamine^1,2. Prolamins that trigger CoD include gliadin and glutenin in wheat, hordein in barley, and secalin in rye. Intestinal damage can be inflicted by as little as ~10 mg of dietary gluten per day³, which is <0.1% of the amount found in a typical western diet². CoD is a global health burden across all age ranges, with a worldwide serological prevalence of 1.4%⁴ that increases by 7.5% every year⁵. The disease is caused by partially degraded gluten peptides, including a 33-residue fragment of wheat α-gliadin (33-mer) that is immunogenically the most relevant^2,6. These peptides resist further cleavage by gastric, pancreatic and intestinal brush-border membrane peptidases owing to their high proline content (13 in the 33-mer). In coeliacs, they cross the mucosal epithelium of the small intestine, where the glutamine residues are deamidated by tissue transglutaminase. This enhances the affinity of the peptides for the DQ2.5/DQ2.2 and DQ8 alleles of the human leukocyte antigen (HLA) receptor, which are necessary for the development of CoD². Receptor binding triggers a severe pro-inflammatory autoimmune response mediated by T cells, with intestinal effects including intraepithelial lymphocytosis, crypt hyperplasia, atrophy of small-intestine villi and mucosal inflammation². These lead to the chronic malabsorption of nutrients, diarrhoea, vomiting, bloating, abdominal pain and intestinal lymphomas. Extraintestinal manifestations include delayed puberty, osteoporosis, axonal neuropathy and cerebellar ataxia⁷, which reduce the life expectancy of coeliacs. There is no treatment for CoD, so patients must adhere to a lifelong strict gluten-free diet, which restores the normal architecture of the intestinal villi². However, gluten-free diets do not provide balanced nutrition⁷, and many coeliacs suffer intestinal symptoms even with adherence to such dietary restrictions^8,9. Moreover, gluten is found in most processed foods and medicines, making dietary compliance challenging in western societies². This has created a demand for effective CoD therapies.

One promising approach is the development of endopeptidases that cleave the toxic peptides and would thus act as bona fide glutenases for oral enzyme therapy^10,11,12, reminiscent of lactase tablets for lactose intolerance¹³. Such an approach would also benefit patients suffering from non-coeliac gluten sensitivity, which has a worldwide prevalence of up to 13%, and irritable bowel syndrome, with a prevalence of <0.5%^8,14,15. A candidate glutenase must fulfil certain criteria for clinical application. First, it should work in the stomach during digestion, before the gastric bolus passes into the duodenum and initiates the autoimmune response, and thus must remain stable and active in the acidic gastric environment (pH ~2.5) as well as resisting gastric pepsin. Second, a reasonable dose should efficiently digest gliadin and the 33-mer when combined with pepsin under gastric conditions, which requires the processing of large quantities of dietary protein. Third, it should not harm intestinal structures or inhibit nutrient absorption, and thus ideally should be inactive at the slightly acidic postprandial pH of the duodenum¹⁶.

The therapeutic potential of several glutaminyl and prolyl endopeptidases (PEPs) has been assessed, representing various catalytic classes and diverse sources including bacteria, fungi, insects and germinating cereals^7,10,11,12. These include a serine PEP from Aspergillus niger^10,17; STAN1, a combination of A. niger aspartate aspergillopepsin and Aspergillus oryzae serine dipeptidyl-peptidase IV¹⁸; latiglutenase, a combination of a glutamine-specific cysteine peptidase from barley and a modified serine prolyl-specific oligopeptidase from Sphingomonas capsulata¹⁹; subtilisin-type serine endopeptidases from the natural oral colonizers Rothia aeria and Rothia mucilaginosa¹¹; and the synthetic enzymes KumaMax and Kuma062/TAK-062, developed by the computational redesign of kumamolysin, a serine endopeptidase from the bacterium Alicyclobacillus sendaiensis²⁰. However, none of these candidates fulfils all of the above requirements. The current frontrunners do not show high activity under gastric pH conditions and/or require very high doses or protective modifications, such as PEGylation or microencapsulation. Accordingly, clinical trials have not yet achieved significant clinical remission in coeliacs and have not demonstrated the ability of these enzymes to replace a gluten-free diet¹². Worse, many so-called enzyme preparations currently sold over the counter as CoD dietary supplements do not inactivate toxic gluten peptides and thus represent a hazard for coeliacs²¹.

Neprosin is a 380-residue endopeptidase of unknown class and mechanism, currently assigned to family U74 in the MEROPS database (www.ebi.ac.uk/merops²²). It is a PEP that was discovered in the digestive fluid of the carnivorous plant Nepenthes × ventrata, which traps prey animals in its pitcher^23,24,25,26. The enzyme might have a function in protein metabolism during prey digestion and/or defence²³. In combination with other peptidases from the digestive fluid, it has been identified as part of a potential glutenase preparation²⁴. Purified neprosin is also considered a useful reagent for proteomics^25,26.

Here, we establish a human recombinant production system to produce high yields of neprosin. We determine its mechanism of activation in vitro as well as its thermal stability, pH profile, general proteolytic and peptidolytic activities, and susceptibility to a panel of peptidase inhibitors. We also test cleavage of gliadin and the 33-mer in vitro to evaluate the ability of neprosin to act as a solo glutenase. Moreover, we evaluate the effect of recombinant neprosin on the processing of gliadin in mice. Finally, we report the crystal structure of the neprosin zymogen and its mature form in product-mimicking complexes. These data reveal the mechanism of latency, the overall and active-site architectures, catalytic mechanism and peptidase class, which have been validated by a cohort of mutants.

Results and Discussion

Heterologous expression, autolytic maturation and stability analysis

Previous studies of neprosin mainly used the enzyme purified from pitcher plant fluid because heterologous expression in Escherichia coli produced only a partially impure enzyme with a modest yield^24,25. We were unable to reproduce this approach so we developed a system based on human cells, assuming that eukaryotic post-translational processing is required. This yielded ~10 mg/L of pure well-folded full-length protein with a C-terminal hexahistidine (His₆) tag (41 kDa) or ~8 mg/L with a twin-streptavidin (Strep) tag (43 kDa) (Fig. 1a, b). The protein was properly folded and remained stable for several weeks at 4 °C in a neutral buffer, but lacked proteolytic activity, which we attributed to the full-length protein being the pro-neprosin zymogen. Indeed, it readily underwent autolytic maturation at bond P¹²⁸–S¹²⁹ (residue numbering of neprosin in superscript; UniProt ID C0HLV2) over time when incubated in a highly acidic buffer, yielding the neprosin catalytic domain (CD) and the excised pro-domain (PD) (Fig. 1e). The latter was eventually degraded, and both pro-neprosin and neprosin migrated as monomers when checked by calibrated size-exclusion chromatography (SEC) (Fig. 1d).

**Fig. 1: Protein purification and activation.**

Differential scanning fluorimetry using the thermofluor approach²⁷ revealed a midtransition temperature (T_m) of 68 °C for the mature enzyme (Fig. 2a), which is remarkable for a peptidase that works in an ambient temperature range and is more reminiscent of hyper-thermophilic enzymes²⁸. Furthermore, the T_m of the zymogen was 9 °C higher (Fig. 2a), suggesting the PD promotes stability and, possibly, the correct folding of the full-length protein as reported for other zymogens²⁹. This was supported by our inability to express mature neprosin (without the PD) using the same expression system. Finally, thermofluor studies in the presence of a reducing agent revealed an unfolding process with two transitions, the first occurring at 42–44 °C (Fig. 2b). This indicated the existence of disulfide bonds that stabilize the protein, as discussed in more detail below.

**Fig. 2: Thermal stability, peptidolytic activity and inhibition assays.**

Proteolytic activity

We investigated the effect of pH on the cleavage of fluorescent bovine serum albumin (BSA) by neprosin, using gastric pepsin, an aspartate peptidase, and pancreatic trypsin, a serine peptidase, for comparison (Fig. 2c). The pH optimum of neprosin was 3, close to that of gastric pepsin (pH < 2). By contrast, the pH optimum of trypsin was 8, a value at which both neprosin and pepsin were completely inactive. Pepsin was irreversibly inhibited at neutral pH, as previously reported³⁰, whereas neprosin was reversibly activated and inactivated by switching between pH 2.5 and 9.0. Moreover, neprosin was unaffected by freezing or lyophilization at pH 7.5 for storage, thus recovering its full activity after thawing or resuspension in an acidic buffer, respectively. Finally, neprosin was insensitive to cleavage by pepsin at acidic pH. This profile of activity, efficiency, stability and robustness was therefore consistent with a digestive enzyme that must work over prolonged timescales under varying conditions, precisely the natural environment in the pitchers of carnivorous plants³¹.

To gain further insight into the substrate specificity of neprosin and to guide our cleavage assays, we reanalysed published proteomics data based mainly on purified material, which had identified the enzyme as a bona fide PEP²⁵. We found 3001 unique cleavage sites spanning P₆–P₆′ (substrate and active-site subsite nomenclature based on^32,33), 1863 (62%) of which featured a proline residue in P₁ (Fig. 2g). Proline was also enriched twofold over its natural abundance at P₂ and P₃′, but was strongly disfavoured at P₁′ and P₂′. Glutamate and methionine were enriched threefold at P₂′, alanine was readily accepted throughout P₆–P₆′, and glycine was significantly disfavoured at P₁–P₃′. These data revealed a strong preference for substrates with proline at P₁ and that specific positions within P₆–P₆′ were unsuited for certain amino acids (Fig. 2g).

Cleavage of gliadin and the 33-mer

We investigated the ability of neprosin to digest gliadin in the presence and absence of pepsin by SDS-PAGE and turbidimetry, compared to pepsin alone (Fig. 3a, b). Both enzymes efficiently degraded gliadin separately at concentrations below ~5 μM, the physiological threshold of pepsin³⁴, but optimal results were achieved when both enzymes were combined. Remarkably, the optimal concentration of neprosin was similar to that of gastric pepsin, and orders of magnitude lower than that required for current glutenase candidates. Zymography showed that neprosin degraded gliadin and gelatine, also a dietary protein, with similar efficiency (Fig. 3d, e).

**Fig. 3: Neprosin activity against molecules relevant for coeliac disease.**

Next, we investigated cleavage of the 33-mer, which includes three glutamine residues that are deamidated by transglutaminase and six overlapping immunogenic HLA-DQ2.5 T-cell epitopes^6,11,35, by mass spectrometry (Fig. 3f). We found that peptide at 250 μM was efficiently degraded by 0.5 μM neprosin, a 500:1 ratio, after 20 min at pH 3 (Fig. 3c). No autolytic cleavage products were detected even after overnight incubation, which confirmed the stability of the mature enzyme under acidic conditions. By contrast, pepsin failed to cleave the peptide even after overnight incubation at a 20-fold higher concentration than neprosin, which confirmed the resistance of the 33-mer against digestive peptidases. Analysis of the peptide cleavage fragments generated by neprosin revealed two final products: Q–L–P–Y–P–Q–P (843 Da) and L–Q–L–Q–P–F–P–Q–P (1068 Da). By monitoring the reaction over time (Fig. 3g), we found that cleavage only occurred immediately downstream of five specific proline residues among the 13 present in the 33-mer, preferably at P–Q–P*Q–L–P and always with P–Q–P at the P₁–P₃ subsites, which qualifies the simple specificity for proline at P₁ deduced from indiscriminate proteomics and is in line with disfavouring large hydrophobic residues (leucine, phenylalanine and tyrosine) in P₂ as discussed above²⁵. Overall, our results demonstrate that the 33-mer is degraded at multiple sites featuring the Q–P*Q–L motif. Remarkably, two P–Q dipeptides are also found in BSA, together with five equally favoured P–E sites (see above), which explains why albumin is a suitable substrate for neprosin at low pH.

Finally, we tested the cleavage of a cohort of fluorogenic peptides. We found that peptide FS6 containing a P–L bond (Mca–K–P–L–G–L–Dpa–A–R–NH₂), which is a substrate of matrix metalloproteinases and adamalysins³⁶, was cleaved with modest efficiency according to kinetic analysis (k_cat/K_M = 765 M⁻¹s⁻¹; Fig. 2d). In contrast, peptide variant FS6-QPQL, redesigned to include the neprosin cleavage site of the 33-mer (Mca–Q–P–Q–L–Dpa–A–R–NH₂), was cleaved 30-fold more efficiently, mainly due to k_cat increase (k_cat/K_M = 23,880 M⁻¹s⁻¹; Fig. 2e). Accordingly, neprosin can be defined as a PEP with a more constrained specificity than P–X that efficiently degrades the 33-mer under gastric-like conditions.

Inhibitory profile

Given the unknown catalytic class of the enzyme, we next tested a panel of peptidase inhibitors for their ability to block FS6-QPQL cleavage by neprosin (Fig. 2h). We also followed an approach recently applied to find inhibitors of pyrroline-5-carboxylate reductase³⁷, whose product is proline, and tested a series of proline-containing/mimicking compounds. We found that only pepstatin A and 2-[(4-nitrophenoxy)methyl]oxirane (EPNP) weakly but significantly inhibited neprosin (Fig. 2h), with half-maximal inhibitory concentration (IC₅₀) values of 140 and 480 μM, respectively (Fig. 2i). Given that pepstatin and EPNP-like epoxides are inhibitors of pepsin-type aspartate endopeptidases^38,39, which share no sequence similarity with neprosin, this pointed to an unexpected peptidase type and mechanism of catalysis for neprosin.

Evaluation of neprosin activity in vivo

To investigate the activity of neprosin in vivo, mice were fed a bolus of gliadin 5 min after receiving either the zymogen at a very low mass ratio (1:500 w/w) or vehicle. After 2.5 h, we harvested the contents of three upper gastrointestinal tract segments and measured the concentration of the 33-mer by enzyme-linked immunosorbent assay (Fig. 4). The peptide was substantially less abundant in all segments of the treated animals (61-91%) and by 71% overall. The inactive zymogen is therefore activated upon reaching the stomach and efficiently helps to break down gliadin (and particularly the 33-mer) in vivo while remaining resistant to physiological digestive enzymes. This occurs at much lower concentrations than those of candidate glutenases, and without protective strategies such as PEGylation or microencapsulation. These results are consistent with a previous study reporting that sensitized NOD/DQ8 mice showed a significant decrease in inflammatory markers when fed gliadin that was pre-digested with pepsin and Nepenthes pitcher fluid, which included among other components neprosin and nepenthesin²⁴.

**Fig. 4: Analysis of neprosin activity against gliadin in vivo.**

Structural analysis of latent and mature neprosin

We crystallized pro-neprosin in an orthorhombic space group (Fig. 5a and Supplementary Table 1) and found that the polypeptide was cleaved at the physiological maturation site (P¹²⁸–S¹²⁹). The crystals therefore contained the zymogenic complex of the cleaved PD and the CD (Fig. 5a). We solved the structure by single-wavelength anomalous diffraction, collecting data at the lutetium L_III absorption edge wavelength from a crystal soaked in Lu-Xo4⁴⁰ (Fig. 5b). This soaking led to significant variation in one of the crystal cell axes when compared to native crystals while keeping good diffraction of X-rays (Supplementary Table 1). The final refined model of the derivative complex was used to solve the native pro-neprosin structure by molecular replacement. Moreover, mature neprosin produced two different monoclinic crystal forms, I and II (Fig. 5a and Supplementary Table 1), whose structures were likewise solved by molecular replacement.

**Fig. 5: Structures of pro-neprosin and neprosin.**

Pro-neprosin is a compact oblong molecule of ~55 × ~45 × ~40 Å (Fig. 5c). The N-terminal PD (R²⁵–P¹²⁸) is defined in the final Fourier map from A²⁹ onwards, and features a globular part (A²⁹–G¹¹²) followed by a linker (L¹¹³–P¹²⁸) to the downstream CD (S¹²⁹–Q³⁸⁰). Segment (N¹²²–N¹³¹), which includes the cleaved maturation site, is flexible. The PD features an antiparallel three-stranded β-sheet in which the central strand is bisected by the insertion of the leftmost strand (Fig. 5c, d). The two right strands are connected by a long segment on the top, which includes two short α-helices, a disulfide bond (C⁵²–C⁹⁸), and a disordered 10-residue segment (Y⁷⁷–N⁸⁶) at the back of the molecule. The latter probably results from the protruding glycan chain attached to N¹⁵² within a back strand of the CD (Fig. 5c). A second glycan is attached to N¹⁴⁵ from a cross-over loop on top of the CD. Beyond the last strand of the PD, the chain undergoes a 90° turn and enters the PD/CD linker, which runs in extended conformation along the front surface of the CD.

Atypically for peptidases, which are generally α/β-proteins⁴¹, the CD is an antiparallel β-sandwich, with a seven-stranded strongly-curled front sheet and an eight-stranded back sheet, which provides a scaffold for the former (Fig. 5c, d). Both sheets are interconnected by nine cross-over loops, including long hairpin β12β13, and two further disulfide bonds (C²¹⁹–C²²⁴ and C³⁵⁸–C³⁷⁹) on either side of the sandwich (Fig. 5d). All these elements contribute to a compact and sturdy structure, which explains the remarkable pH stability of neprosin and its ability to resist pepsin digestion. By contrast, the disulfide bonds are not deeply buried in the structure, which explains its sensitivity to reducing agents. The structure of mature neprosin crystal form I (Supplementary Table 1) proved practically identical to the equivalent part of the zymogen, with a core root mean square deviation (RMSD) of 0.62 Å. The only significant difference was encountered at N²³²–Y²³³, which is folded outward in the zymogen to accommodate I¹⁰³ at the beginning of the rightmost strand of the PD. Crystal form II, in turn, was practically indistinguishable from crystal form I (core RMSD = 0.66 Å) except for the tip of loop Lβ21β22, which was spaced apart by 3.8 Å, and the C-terminal tag, which was reoriented owing to crystal packing. Thus, the mature enzyme component is essentially preformed in the zymogen as seen in most peptidases, with the notable exception of chymotrypsin-type serine peptidases^42,43,44.

The active site

We hypothesized that the active-site cleft would be delineated by the PD linker (Fig. 6a) as found in other zymogens^42,44. Moreover, in the structures of mature neprosin crystal forms I and II, the C-terminal segment, which spanned an alanine–isoleucine–alanine tripeptide followed by the His₆-tag, ran along the surface of a symmetry mate, thus mimicking a product complex. Both crystal forms were monoclinic but with different cell constants (Supplementary Table 1), which resulted in variable crystal packing. Even so, the C-terminal tag penetrates the cleft in a similar manner in both crystallographic arrangements but is shifted by three positions, so that H⁴⁰⁴–H⁴⁰⁹ from crystal form I overlaps A⁴⁰¹–H⁴⁰⁶ from crystal form II (Fig. 6b, c). Accordingly, neprosin would possess an extended active-site cleft traversing the concave face of the sheet, which is oblique to the direction of the front-sheet β-strands by ~55° (Fig. 6a–c).

On the search for possible catalytic residues, we were inspired by the functionally analogous pepsin-type aspartic peptidases, which despite their disparate architecture are likewise mainly β-proteins and operate at highly acidic pH⁴⁵. Moreover, the only (weak) neprosin inhibitors we could find are also known to inhibit aspartate peptidases (see above). These enzymes use a pair of aspartic residues bridged by a solvent molecule for catalysis⁴⁶. Indeed, we found a striking pair of glutamate residues (E¹⁸⁸ and E²⁹⁷) bridged by a solvent molecule pinching the bound peptides in the product complexes (Fig. 6b, c). In crystal form II, a clearly resolved second solvent molecule would replace the scissile carbonyl oxygen of a substrate (Fig. 6c). The glutamate pair was similarly arranged in the zymogen structure, albeit slightly farther apart (Fig. 6a). We therefore produced E¹⁸⁸Q, E¹⁸⁸A, E²⁹⁷Q and E²⁹⁷A point mutants of His₆-tagged pro-neprosin for testing (Fig. 1c). These variants did not autoactivate when incubated at acidic pH (Fig. 1f), so activation was triggered in trans using catalytic amounts of mature Strep-tagged wild-type neprosin. Finally, we obtained well-folded and intact mature variants of the E¹⁸⁸Q and E²⁹⁷Q mutants, but not E²⁹⁷A or E¹⁸⁸A (Fig. 1f), and these indeed were catalytically inactive (Fig. 2f). E¹⁸⁸ and E²⁹⁷ may therefore act as a catalytic dyad, revealing that neprosin is a glutamate peptidase, a catalytic class that (in contrast to the aspartate peptidases) has been studied very poorly⁴⁷. This is in agreement with very recent predictions based on bioinformatics studies but not validated experimentally⁴⁸. Our results suggest that the mature neprosin structures mimic upstream product complexes collectively occupying subsites S₆ to S₁′ (Fig. 6b, c), with the two catalytic glutamates plus the bridging solvent molecule poised for reaction. As the PD linker extends for further residues on the right side of the cleft in the zymogen (Fig. 6a), these would correspond to positions up to P₃′. Thus, together with the extra space in the cleft beyond S₃′, neprosin would feature an extended cleft, probably spanning up to 11 subsites (S₆–S₅′), which explains the need for extended peptides beyond the scissile bond (see above).

In the absence of a substrate complex, we constructed a model for the Michaelis complex of neprosin with the P–Q–P*Q–L–P peptide based on the zymogen and product–complex–mimicking structures (Fig. 6d). This model revealed further residues in the proximity of the catalytic glutamates with potential binding or catalytic functions. We therefore mutated residues H¹³⁴, Y¹³⁶, Q¹⁷³, W¹⁷⁵ and Y²¹⁴ (Fig. 6a–d) by replacing them with alanine, and purified the corresponding proteins (Fig. 1c). All the mutants required activation in trans as discussed above (Fig. 1f). The activity of H¹³⁴A, Y¹³⁶A and Y²¹⁴A was ~80% lower than the wild-type enzyme, whereas mutants Q¹⁷³A and W¹⁷⁵A were totally inactive (Fig. 2f). We conclude that H¹³⁴, Y¹³⁶ and Y²¹⁴ are relevant but not critical for catalysis, possibly playing an ancillary role in the catalytic mechanism, whereas Q¹⁷³ and W¹⁷⁵ are essential (see below).

Mechanisms of latency and activation

The PD attaches laterally to the left side of the mature enzyme so that its central β-sheet is rotated ~90° away from the plane of the front sheet. The inter-domain surface has a solvation-free energy gain upon interface formation (ΔⁱG) of –25.8 kcal/mol⁴⁹, indicating a very strong interaction. Furthermore, the complex buries 2176 Å², which exceeds the reported average value of 1910 Å² for protein–protein complexes⁵⁰. The PD (theoretical pI = 9.5; Fig. 5e, bottom) is crescent-shaped and snuggly embraces the CD (pI = 4.3; Fig. 5e, top) under electrostatic complementation, which contributes to activity repression and zymogen stability (pI = 5.9) at neutral or slightly acidic pH values. Moreover, the intimate zymogenic interaction further explains the remarkable stability of pro-neprosin in thermal shift assays (see above). Finally, the importance of the PD was further assessed by testing point mutant A⁶⁰R, designed to destabilize the interface (Fig. 5c, left panel). This mutation prevented the isolation of a folded protein.

Once secreted to the acidic digestive fluid, the protonation of negatively charged residues leads to the repulsion of net positive charges so that the zymogen falls apart under liberation of the preformed mature moiety and the active-site cleft. The S₁ position of the cleft is occupied by K¹¹⁸ from the PD linker in the zymogen structure, which was obtained at pH 7.5. This residue forms a strong salt bridge with catalytic E¹⁸⁸ and a hydrogen bond with Q¹⁷³Nε2, which is essential (see above and Fig. 6a, inset). We therefore produced and tested mutant K¹¹⁸A, which was efficiently overexpressed but underwent partial autolytic maturation in a neutral buffer, conditions under which the wild-type enzyme and other mutants remained intact (Fig. 1c). Subsequent incubation at pH 2.5 completed the activation process (Fig. 1f). As expected, the activity of the mature mutant was similar to that of the wild-type enzyme (Fig. 2f).

Based on the above, we propose that the K¹¹⁸–E¹⁸⁸ pair features a latency plug that may be weakened once the zymogen reaches acidic environment by following a pH-switch mechanism, so the PD linker is pulled out for maturation cleavage. This is reminiscent of the digestive aspartate peptidases pepsin and gastricsin, which feature a lysine residue functionally equivalent to K¹¹⁸^43,51, and of the lysosomal peptidase legumain⁵². The pH-switch mechanism, and the fact that the scissile P₁′–P₁ peptide bond is sandwiched by the Y²¹⁴ side chain so that it is not accessible for cleavage (Fig. 6a), explains why the zymogen linker can bind in the direction of a substrate to the cleft at neutral pH without being cleaved. This contrasts with most zymogens, including digestive aspartate peptidases, in which pro-segments interact in a non-substrate-like manner with the mature enzyme residues as a mechanism to prevent untimely activation^42,43,44. Finally, given that the scissile-bond position in the cleft is occupied by K¹¹⁸–Q¹¹⁹ but maturation occurs at P¹²⁸–S¹²⁹, activation probably occurs in trans by a second enzyme molecule once the PD linker is released from the cleft.

Proposed catalytic mechanism

Based on the preceding results, the catalytic cleavage mechanism of neprosin would proceed as follows. The solvent bridging the E¹⁸⁸ and E²⁹⁷ carboxylates in the product complexes would represent the catalytic water in the ground state (Fig. 6e, step I). The water is closer to E²⁹⁷, which suggests that E¹⁸⁸ may be protonated, as reported for one of the two catalytic aspartates in pepsin-type acidic peptidases⁴⁶. E²⁹⁷ is kept in place by hydrogen bonding with H¹³⁴Nε2 and Y¹³⁶Oη, whereas E¹⁸⁸ is kept in place by hydrogen bonding with Y²¹⁴Oη, T¹⁸⁶Oγ and Q¹⁷³Nε1. During the reaction, the substrate would bind to the active-site cleft in extended conformation (Fig. 6e, step II), with the S₃, S₁ and S₃′ subsites of the cleft being shaped by Y¹⁹⁴, Q¹⁹² and F¹⁶⁹; Y²⁰⁸, Y²⁰⁶, L¹⁷¹, E¹⁸⁸ and Q¹⁷³; and Y¹³⁶, W¹⁷⁵ and E²⁹³, respectively, which are ideal for the accommodation of prolines (Fig. 6d). The substrate main chain would be fixed by hydrogen bonds between its carbonyls and Y²²⁰Oη in P₃, H¹³⁴Nε2 in P₂ (enabled by a 180º rotation around χ₂ upon by substrate binding), and Y¹³⁶Oη in P₁′ (Fig. 6d). Substrate insertion would shift the catalytic solvent further towards E²⁹⁷, which would act as a general base and abstract a proton from it to enhance its nucleophilicity. The protonated E¹⁸⁸ carboxylate, in turn, would bind the scissile carbonyl oxygen (Fig. 6d, e). Thereafter, the polarized solvent would perform a nucleophilic attack on the si-face of the scissile carbonyl carbon, which would result in a tetrahedral gem-diolate reaction intermediate (Fig. 6e, step III). The latter would be stabilized by indispensable W¹⁷⁵Nε1 and Q¹⁷³Nε1, in the critical role of an oxyanion hole⁵³. The intermediate would then resolve by breaking the scissile C–N bond. At this stage, E²⁹⁷ would act as a general acid and protonate the new α-amino nitrogen (Fig. 6e, step IV). Finally, the two cleavage products would leave the cleft and the enzyme would be poised for a new round of catalysis.

Structural similarity with eqolysins

Peptidases were originally assigned to five mechanistic classes: the serine, cysteine, threonine, aspartate, and metal-dependent peptidases⁵⁴. In 2004, the founding glutamate peptidase was structurally characterized, namely scytalidocarboxyl peptidase B (SCP-B) from the dematiaceous fungus Scytalidium lignicolum^55,56,57. Since then, only the closely related aspergilloglutamic peptidase (~50% identical to SCP-B) has been structurally analysed^58,59, and seven others have been functionally assessed, mostly from fungi^{60,61,62,63,64} but one from a bacterium⁶⁵. They are assigned to family G1 in the MEROPS database and are informally known as the pepstatin-insensitive fungal carboxylpeptidase group⁶⁶ or eqolysins⁵⁵. They are thermophilic and pepstatin-insensitive enzymes that function under acidic conditions⁶⁵ and feature a catalytic glutamate acting as a solvent-polarising general base, which is E₁₉₀ in SCP-B (see UniProt ID P15369 for residue numbering in subscript according to the full-length protein, and subtract 54 for the commonly used mature enzyme numbering^55,56). The glutamate is assisted by a glutamine (Q₁₀₇ in SCP-B), hence the family name eqolysins⁵⁵. These residues are invariant within the family and are flanked by very similar residues^66,67.

Archetypal SCP-B is a 7+7 antiparallel β-sandwich that shows overall similarity with the neprosin CD (Fig. 7a). Superposition of neprosin and the bound mature form of SCP-B (Protein Data Bank [PDB] ID 2IFR⁵⁶), whose zymogenic structure is unknown, revealed 140 aligned residues with a rather large core RMSD of 3.0 Å and a sequence identity of only 11%. There are remarkable differences in the connecting loops and the active site, e.g. a large disulfide-linked protruding β-hairpin inserted in the fungal enzyme following the β-strand equivalent to β16 in neprosin (Fig. 7a). Within the active site, the only conserved residue is the catalytic glutamate (E²⁹⁷ in neprosin and E₁₉₀ in SCP-B), as well as the position of the catalytic assistant (E¹⁸⁸ in neprosin and Q₁₀₇ in SCP-B), which lead to variable active-site clefts with disparate substrate trajectories and surface profiles (Fig. 7b, c). Moreover, cross-mutants Q₁₀₇E of SCP-B and E¹⁸⁸Q of neprosin, which mimic each other’s catalytic dyad, are completely inactive, as discussed above and reported in⁶⁸. This explains the different substrate specificities, which in SCP-B leads to the cleavage of F–F, L–Y and F–Y bonds in insulin but not proline-flanking bonds⁶⁸.

**Fig. 7: Structural similarity of neprosin and eqolysins.**

Corollary

Current glutenases have limitations in meeting the stringent criteria for efficient oral enzyme therapy against CoD. Here, our in vitro and in vivo studies showed that recombinant neprosin is a robust pepsin-resistant enzyme that very efficiently degrades gliadin and its 33-mer under laboratory-simulated gastric conditions and in the mouse stomach. Low doses of the enzyme therefore complement gastric pepsin during digestion. Our results demonstrate that the Q–P*Q–L motif of the 33-mer is readily cleaved, which removes all six overlapping immunogenic epitopes by generating peptides too small to stimulate the division of gliadin-specific T cells⁶⁹. The cleavage efficiency of neprosin in vitro under simulated gastric conditions is orders of magnitude higher than that of other glutenases^{18,20,24,70,71,72}. The zymogen is produced at neutral pH, at which it remains stable and is lyophilizable for transport and storage. It only becomes activated after ingestion in the stomach and cleaves toxic components of gluten. Once the gastric bolus exits to the slightly acidic postprandial pH duodenum, it becomes inactive again. Neprosin is therefore a highly promising candidate for further therapeutic development against gluten-sensitive conditions.

Structural and functional studies backed by mutants and activity assays identified neprosin as a pepstatin-sensitive PEP and the only glutamate endopeptidase found in higher eukaryotes. It features a hitherto undescribed pair of catalytic glutamates that are analogous to the aspartates of the otherwise unrelated pepsin-type acidic endopeptidases. Neprosin is produced and secreted as a zymogen, which is activated only in its strongly acidic natural environment, the pitcher plant digestive fluid. Maturation follows a pH-switch mechanism that releases a lysine-mediated latency plug.

Finally, neprosin is pepstatin-sensitive but shares its overall fold with the pepstatin-resistant glutamate peptidases of the eqolysin family, which possess a glutamate–glutamine dyad and are represented by the archetype SCP-B. However, there are differences in the size of the PD and the CD, the active-site environment, the substrate-binding modus and specificity, as well as the chemical mechanism of catalysis. Furthermore, whereas eqolysins are restricted to fungi and bacteria^67,73, potential neprosin orthologues with ~35–40% sequence identity are widely found in (and restricted to) plants, including gluten-containing crops. This suggests the neprosin family may have originated from a SCP-B ancestor by horizontal gene transfer from a bacterium or fungus to a plant, as previously described for other proteins⁷⁴. Transfer would have been followed by divergent evolution within the plant kingdom to modify one of the catalytic residues and the loops decorating the central β-sandwich to adapt to new substrates. By analogy to the eqolysins, neprosin family members could be named eelysins.

Methods

Protein production and purification

A synthetic gene encoding wild-type neprosin from Nepenthes × ventrata, which is 91% identical to the orthologue from Nepenthes alata (UniProt ID A0A1L7NZU4), was inserted into vector pET-28a(+) by GenScript to produce vector pET-28a(+)-proNEP (for plasmids and primers, see Supplementary Table 2). The coding sequence was transferred to vector pCMV to produce vector pS6-proNEP. This conferred ampicillin resistance and added a C-terminal hexahistidine (His₆) tag. The encoded protein is described herein as pro-neprosin. The same plasmid was modified by annealed oligonucleotide cloning to (a) replace the His₆-tag with a twin Strep tag (pS6-proNEP-Strep) for the expression of pro-neprosin-strep, and (b) to remove the PD (pS6-NEP) for the expression of the neprosin CD (S¹²⁹–Q³⁸⁰) plus the C-terminal His₆-tag. The QuikChange Site-Directed Mutagenesis Kit (Stratagene) or inverse PCR-based site-directed mutagenesis were used to generate variants of pS6-proNEP with point mutations A⁶⁰R, K¹¹⁸A, H¹³⁴A, Y¹³⁶A, Q¹⁷³A, W¹⁷⁵A, E¹⁸⁸A, E¹⁸⁸Q, Y²¹⁴A, E²⁹⁷Q and E²⁹⁷A. Plasmids were purified with the GeneJET Plasmid MaxiPrep Kit (Thermo Fisher Scientific), and constructs were verified by DNA sequencing.

Proteins encoded by the pS6-proNEP, pS6-proNEP-Strep and pS6-NEP plasmids, as well as the 11 point mutants, were assessed for overexpression in human Expi293F cells (ThermoFisher Scientific) grown in a Multitron cell shaker incubator (Infors HT) at 37 °C. The cells were transfected with plasmid DNA and harvested after several days for protein purification. Cell-conditioned medium was cleared by centrifugation and supplemented with imidazole, incubated with nickel-nitrilotriacetic acid (Ni-NTA) resin (Invitrogen), subjected to batch affinity chromatography purification (AC), and washed extensively with buffer containing 20 mM imidazole. Proteins were eluted with the same buffer containing 300 mM imidazole. For pro-neprosin-strep, the Ni-NTA resin was replaced with Strep-Tactin XT Superflow suspension resin (IBA Life Sciences), and proteins were eluted in buffer containing 50 mM d-biotin (VWR Life Science). Fractions containing the protein were pooled and concentrated before size-exclusion chromatography (SEC) in a Superdex 75 10/300 GL column (GE Healthcare), which was attached to an ÄKTA Purifier liquid chromatography system (GE Healthcare).

Proteins were concentrated by ultracentrifugation in Vivaspin filter devices (Sartorius Stedim Biotech). Approximate protein concentrations were determined by measuring the absorbance at 280 nm (A₂₈₀) using a BioDrop-DUO Micro Volume (Biochrom), and applying the appropriate theoretical extinction coefficients. Moreover, protein purity was assessed by sodium dodecylsulfate polyacrylamide gel electrophoresis (SDS-PAGE) followed by staining with Coomassie (Thermo Fisher Scientific). Protein identity was determined by peptide mass fingerprinting and N-terminal Edman sequencing at the Protein Chemistry Service and the Proteomics Facility of the Centro de Investigaciones Biológicas (Madrid, Spain), respectively. Finally, mature wild-type neprosin was lyophilized, stored at –20 °C, and reconstituted by dissolving in Milli-Q water.

For activity assays, the filtered conditioned medium of wild-type neprosin and the point mutants was supplemented with 3 mM reduced glutathione and 0.3 mM oxidized glutathione, the pH was adjusted with 20 mM Tris·HCl pH 8.0, and the mixture was incubated with cOmplete His-Tag Purification Resin (Roche). The resin was collected in an open column and the bound protein was washed with 10 mM Tris·HCl pH 7.0, 300 mM sodium chloride, and was then eluted with 100 mM glycine pH 2.5, 300 mM sodium chloride.

Autolytic activation of pro-neprosin

Wild-type and mutant mature forms of neprosin or neprosin-strep were obtained by autolysis. Protein samples eluted from Ni-NTA or Strep-Tactin columns were dialysed against buffer, diluted twofold with 100 mM glycine pH 2.5, and incubated at 37 °C for up to 16 h. Reactions were stopped at specific time points (0 min, 10 min, 20 min, 30 min, 1 h, 2 h and overnight) by boiling aliquots in reducing/denaturing SDS sample buffer, followed by SDS-PAGE. Mature neprosin was buffer-exchanged to 20 mM Tris·HCl pH 7.5, 250 mM sodium chloride in a PD10 column followed by SEC in a Superdex 75 10/300 GL column with the same buffer. Protein purity and identity were assessed as stated above.

Trans-activation of pro-neprosin mutants

To obtain mature neprosin point mutants from zymogens that do not autoactivate, the purified pro-proteins (H¹³⁴A, Y¹³⁶A, Q¹⁷³A, W¹⁷⁵A, E¹⁸⁸A, E¹⁸⁸Q, Y²¹⁴A, E²⁹⁷Q and E²⁹⁷A) were incubated with activated neprosin-strep at a 20:1 weight ratio overnight at 37 °C. Pro-neprosin-strep was previously buffer-exchanged to 100 mM glycine pH 3.0, 150 mM sodium chloride for activation. Cleaved samples were buffer-exchanged and purified by reverse affinity chromatography, concentrated and purified by SEC.

Protein stability assays

Pro-neprosin and mature neprosin were analysed by differential scanning fluorimetry using an iCycler iQ real-time PCR detection system (Bio-Rad). Samples were prepared at 0.5 mg/mL, in the presence or absence of 5 or 10 mM tris(2-carboxyethyl)phosphine (TCEP) as a reducing agent, and supplemented with 5× SYPRO Orange Protein Stain (Thermo Fisher Scientific). The temperature of midtransition (T_m) was determined as the average of duplicate measurements of the midpoint value of the stability curve.

Proteolytic activity and pH profile

We incubated 10 μM of the fluorescent protein substrate DQ Red BSA (Thermo Fisher Scientific) with 0.15 μM neprosin in 100 μL buffer at pH 2–8. Fluorescence was monitored using an Infinite M2000 microplate fluorimeter (Tecan) at 37 °C. We tested 0.5 μM bovine trypsin (Sigma-Aldrich) and porcine pepsin (Fluka) for comparison. Each assay was carried out in triplicate.

Cleavage studies with fluorogenic peptides and determination of kinetic parameters

The kinetic parameters of FS6-QPQL peptide (Mca–Q–P–Q–L–Dpa–A–R–NH₂; GenScript) cleavage by wild-type neprosin (25 nM final enzyme concentration), as well as those of the FS6 peptide (Mca–K–P–L–G–L–Dpa–A–R–NH₂; Sigma-Aldrich) by neprosin at 100 nM final enzyme concentration, were determined in reactions containing 100 mM glycine pH 3.0 and substrate concentrations of 1–75 μM (FS6-QPQL) or 2.5–75 μM (FS6) at 37 °C. The fluorescence signal, representing cleavage product formation, was recorded over time for each substrate concentration and the initial rate (v₀) was derived from the slope of the linear part of the curve. Using a range of substrate concentrations and a surplus of peptidase, we measured the fluorescence signal generated after full substrate turnover and calculated the corresponding fluorescence units per picomole of cleaved substrate. These values were plotted against substrate concentration and fitted to the hyperbolic Michaelis-Menten equation (v = V_max·[S]/{K_M+[S]}) by nonlinear regression using GraphPad⁷⁵ and SigmaPlot⁷⁶ to determine the maximum velocity (V_max), the Michaelis substrate affinity constant (K_M), the turnover rate (k_cat = V_max/[E_total]), and the catalytic efficiency (k_cat/K_M) of the cleavage reaction. All experiments were carried out in triplicate.

The peptidolytic activity of wild-type neprosin was compared to that of the mutants K¹¹⁸A, H¹³⁴A, Y¹³⁶A, Q¹⁷³A, W¹⁷⁵A, E¹⁸⁸Q, Y²¹⁴A and E²⁹⁷Q (140 ng) using 10 μM of the fluorogenic FS6-QPQL peptide in 100 mM glycine pH 3.0, 150 mM sodium chloride at 37 °C, shaking in a Synergy H1 microplate reader (BioTek). To ensure identical sample treatment, all protein variants were activated with neprosin-strep, which was then removed by reverse affinity chromatography as stated above. The protein concentration was estimated from the surface of the A₂₈₀ SEC curves and corrected based on the ε₂₈₀ values. Fluorescence values after 30 min were used as activity endpoints. Experiments were carried out in triplicate and differences were analysed for statistical significance using GraphPad.

Cleavage of gliadin in vitro

Wheat gliadin (Sigma-Aldrich) was prepared in 100 mM glycine pH 2.5 and variable concentrations of pepsin (0.05–10 μM) from porcine gastric mucosa (Fluka), neprosin (0.05–2 μM), or mixtures of 0.5 μM pepsin and 0.05–2 μM neprosin were used to digest 10 mg/mL gliadin slurries. Reactions were monitored by turbidimetry in 96-well plates (Corning) at 37 °C in a microplate spectrophotometer (BioTek). Reactions were quenched by boiling in SDS sample buffer before analysis by SDS-PAGE. Gliadin degradation by neprosin was also analysed by zymography using SDS-PAGE gels containing either wheat gliadin or teleostean gelatin (Sigma-Aldrich), which was used as a control, at 0.1 mg/mL. Pro-neprosin was also tested, which became activated to the mature form during the assay. Proteins were renatured by washing the zymograms with 2.5% Triton X-100 in 100 mM glycine pH 2.5, 200 mM sodium chloride. After further washes with the same buffer plus 0.02% Brij-35, the zymograms were incubated overnight in the same buffer, rinsed briefly with water, and stained with Coomassie.

Cleavage of the 33-mer peptide in vitro

Cleavage of the 33-mer peptide of wheat α-gliadin (LQLQPFPQPQLPYPQPQLPYPQPQLPYPQPQPF, 3911 Da) was monitored using an AutoFLEX III MALDI-TOF mass spectrometer. The peptide (from GenScript) was dissolved in water to a concentration of ~20 mg/mL and stored at –20 °C. The cleavage reaction was carried out with ~1 mg/mL (~250 μM) substrate in 100 mM glycine pH 3.0 at 37 °C by adding 0.5 μM neprosin or 10 μM pepsin. Reactions were stopped at different time points (0 min, 10 min, 20 min, 45 min, 1 h and overnight) and samples were then diluted 1:10 with water, mixed with an equal volume of the 2,5-dihydroxybenzoic acid matrix at 10 mg/mL in a solution containing 30% acetonitrile and 70% 0.1% trifluoroacetic acid, and spotted on a ground steel plate (Bruker). Mass spectra were acquired in positive reflectron mode at 21 kV total acceleration voltage.

Liquid chromatography-mass spectrometry (LC-MS/MS) data analysis

We reanalysed the cleavage specificity data of endogenous neprosin or recombinant material obtained from Escherichia coli deposited at Chorus (Project ID 1262²⁵). LC-MS/MS raw files were converted to MGF format, and data were processed using TANDEM, Comet and MS-GF+, as implemented in SearchGUI⁷⁷. Results were evaluated using PeptideShaker⁷⁸ with a false discovery rate of 1%. Data were non-specifically searched for hits against the human proteome in UniProt (March 2020) using a mass tolerance of 20 ppm for both MS1 and MS2, fixed cysteine carbamidomethylation, and variable methionine oxidation. Up to 50 missed cleavages or a maximum of 5500 Da were tolerated for the parental peptide mass.

Inhibition assays

On the search for neprosin inhibitors, we assayed the broad-spectrum cOmplete Inhibitor Cocktail (Roche); the metallopeptidase inhibitors 1,10-phenathroline, phosphoramidon, marimastat, and captopril (all from Sigma-Aldrich); the serine peptidase inhibitor 4-(2-aminoethyl)-benzenesulfonyl fluoride (AEBSF; Sigma-Aldrich); the aspartate peptidase inhibitors pepstatin A (Sigma-Aldrich), methyl-2-[(2-diazoacetyl)amino]hexanoate (DAN; Chemical Abstracts Service (CAS) 7013-09-4; Bachem 4010441), and ENPN (CAS 5255-75-4; Apollo Scientific OR26560); as well as the proline-containing/mimicking compounds 2-acetyl-1-methylpirrole (AMP; CAS 932-16-1; Sigma-Aldrich 160865); (S)-tert-butyl-2-(3-ethoxy-3-oxopropanoyl)pyrrolidine-1-carboxylate (BEOPC; CAS 109180-95-2; Fluorochem 387901); and N-boc-glycylproline (BGP; CAS 14296-92-5; Bachem 4003703). Inhibition of the cleavage of the FS6-QPQL peptide was investigated by pre-incubating 100 nM neprosin in 100 mM glycine pH 3.0 with 100 μM of each tester compound for >1 h at 37 °C. We then added 10 μM of the substrate and the residual activity was monitored for 4 h as an increase in fluorescence. Differences were analysed for statistical significance using GraphPad. The positive control in the absence of inhibitors (100% activity) contained the same final concentration of dimethyl sulfoxide that was used to solubilize the inhibitors. In addition, half-maximal inhibitory concentration (IC₅₀) values were determined for pepstatin A and ENPN by measuring the activity of 50 nM neprosin in the presence of 10 μM of substrate and inhibitor concentrations of 5–500 μM and 5–5000 μM, respectively, to obtain the inhibition curves. These curves were analysed by nonlinear regression using GraphPad.

Evaluation of gliadin processing by neprosin in vivo

Experimental procedures involving mice followed the institutional guidelines for the care and use of laboratory animals and the ARRIVE guidelines. Protocols were approved by the Ethical Committee for Animal Experimentation of the University of Barcelona (CEEA-UB/Ref. 186/20-P2) and the Government of Catalonia (PAMN/Ref. 11485), which followed Directive 2010/63/EU for the protection of animals used for scientific purposes. The sample size was estimated by the Appraising Project Office’s program from the Universitat Miguel Hernández of Elx (Alacant, Spain). We used 5-week-old male and female C57BL/6 mice (n = 16) purchased from Janvier and housed at the animal facility of the Faculty of Pharmacy and Food Science of the University of Barcelona in a controlled environment (20–24 °C, 40–60% relative humidity) and a 12-h photoperiod, with lights on at 8 a.m. and lights off at 8 p.m. Animals were housed in cages with large Souralit 1035 fibrous particles as bedding (Bobadeb), and tissue paper (Gomà-Camps) and cardboard climbing structures for cage enrichment. Animals had free access to water and RM3 (P) SQC diet (Special Diet Services).

After 1 week for acclimation, two groups of mice were randomly selected, each comprising four males and four females (n = 8 per group) and were marked neprosin (N) or vehicle (V). Animals were not fasted to account for the physiological transit time, and food and water were removed only 1 h before oral gavage. Group N mice were fed 50 μL pro-neprosin in vehicle (0.2 mg/mL in 20 μM Tris-buffered saline pH 7.5, 150 μM sodium chloride), whereas group V mice were fed 50 μL vehicle alone. After 5 min, all mice were fed 50 μL gliadin slurry containing 5 mg wheat gliadin (Sigma-Aldrich) at 100 mg/mL in 10% ethanol solution using small-volume Hamilton syringes and adapted oral probes. The enzyme:gliadin ratio (1:500) was calculated based on our in vitro results, which had shown that neprosin digests gliadin at a 1:500–1000 ratio at 37 °C over a period of 90 min. Given that gastrointestinal transit in mice causes a bolus to reach the small intestine after 1–3 h, with some content already entering the large intestine⁷⁹, we selected 2.5 h as the optimal endpoint to assess the degradation of gliadin in the upper gastrointestinal tract. Animals were then euthanized by cervical dislocation and the contents of the stomach, proximal small intestine and distal small intestine were removed, weighed, and frozen at –20 °C.

Samples were suspended in phosphate-buffered saline (pH 7.2) at a concentration of 200 mg/mL, homogenized with a Kimble Pellet Pester Cordless Motor (DWK Life Sciences), and extracted first with buffer at 50 °C for 40 min and then with 80% ethanol at 20–25 °C for 1 h. The mixtures were centrifuged (2000 × g, 10 min, 4 °C), and the aqueous layer between the particulate and fat layers was removed. The 33-mer content in each diluted extract was analysed using the AgraQuant Gluten G12 ELISA test kit (Romer Labs), which has a detection limit of 2 ppm, according to the manufacturer’s instructions. The G12 antibody detects the 33-mer but no other gliadin degradation fragments⁸⁰. Final amounts were normalized taking into account the sample weight and results were expressed as mean ± SEM. The Statistical Package for Social Sciences (SPSS v22.0; IBM) was used for statistical analysis. The data showed homogeneity of variance (Levene’s test) and followed a normal distribution (Shapiro-Wilk test), so we applied conventional one-way analysis of variance (ANOVA).

Crystallization and diffraction data collection

We screened for crystallization conditions at the joint IBMB/IRB Automated Crystallography Platform using the sitting-drop vapour diffusion method. Optimal pro-neprosin crystals (~20 mg/mL in 20 mM Tris·HCl pH 7.5, 150 mM sodium chloride) were obtained at 20 °C with 0.1 M sodium acetate pH 4.0, 22% polyethylene glycol (PEG) 6000, 10% isopropanol as the reservoir solution. Crystals were harvested using cryo-loops (Molecular Dimensions), rapidly passed through a cryo-buffer consisting of reservoir solution plus 15% (v/v) glycerol, and flash-vitrified in liquid nitrogen for data collection. A lutetium derivative of pro-neprosin was obtained by soaking native crystals for 5 min in cryo-buffer supplemented with 100 mM of the Lu-Xo4 crystallophore (Polyvalan)⁴⁰ and flash-vitrifying them without back soaking. X-ray diffraction data were collected from native crystals at 100 K on a Pilatus 6M-F pixel detector at beamline I04-1 of the Diamond Light Source (Harwell, UK). Lutetium derivative data were recorded on a Pilatus 6M detector at beamline XALOC of the ALBA synchrotron (Cerdanyola, Catalonia, Spain) operated with the Generic Data Acquisition (GDA) software.

The mature neprosin–product complex (crystal form I) was obtained at a protein concentration of ~16 mg/mL in 20 mM Tris·HCl pH 7.5, 250 mM sodium chloride at 4 °C using 10% PEG 1000, 10% PEG 8000 as the reservoir solution. Crystals were cryo-protected with the same reservoir solution plus 15% (v/v) glycerol prior to flash-vitrification in liquid nitrogen. X-ray diffraction data at 100 K were collected at beamline ID30B of the ESRF synchrotron (Grenoble, France) using a Pilatus 6M detector. The mature neprosin–product complex in crystal form II was obtained at the same protein concentration but in 0.1 M glycine pH 3.0, 150 mM sodium chloride at 20 °C using 0.1 M sodium citrate tribasic pH 5.6, 0.5 M ammonium sulfate, 1 M lithium sulfate as the reservoir solution. Crystals were cryo-protected with a solution containing 20% (v/v) glycerol. Diffraction data were collected at beamline XALOC on a Pilatus 6M detector.

Diffraction data were processed using Xds⁸¹ and Xscale, and were transformed to MTZ format using Xdsconv for the Phenix⁸² and Ccp4⁸³ program suites. All crystals contained a monomer in the crystal asymmetric unit and Supplementary Table 1 provides essential statistics on data collection and processing.

Structural solution and refinement

The structure of pro-neprosin was solved by single-wavelength anomalous diffraction using data collected from a lutetium derivative crystal at the L_III-absorption peak wavelength (1.34 Å) by applying the Autosol protocol of the Phenix package. The resulting Fourier map was then subjected to further density modification with wARP/ARP⁸⁴. A starting model for Lu-Xo4 was obtained by energy minimization applied to the coordinates of the metal-chelating moiety of the compound as found in its complex with Tb³⁺ (Protein Data Bank [PDB] ID 6FRO, residue name 7MT) using Chimera⁸⁵. The resulting coordinates in PDB format were combined with a Lu³⁺ ion for model building. Thereafter, several rounds of manual model building in Coot⁸⁶ alternated with crystallographic refinement using the Refine protocol of Phenix and the BUSTER⁸⁷ program. The final model comprised pro-neprosin residues A²⁹–Q³⁸⁰ except S⁷⁶–Y⁸⁵ and N¹²²–N¹³¹ plus three extra C-terminal residues from the purification tag (A⁴⁰¹–I⁴⁰²–A⁴⁰³); two Lu-Xo4 moieties at roughly half occupancy; two N-linked glycan chains totalling five sugar residues attached to N¹⁴⁵ and N¹⁵², respectively; two acetate molecules; and 180 solvent molecules.

The structure of native pro-neprosin was solved by molecular replacement using the Phaser crystallographic software within Ccp4 and the protein coordinates of the lutetium derivative crystal structure. Subsequent model building and refinement proceeded as described above. The final model included residues A²⁹–Q³⁸⁰ except Y⁷⁷–Y⁸⁵ and N¹²²–N¹³¹ plus two extra C-terminal residues from the purification tag (A⁴⁰¹–I⁴⁰²), two N-linked glycan chains totalling four sugar residues, as well as eight acetate, one isopropanol, four glycerol and 257 solvent molecules.

The structure of a product complex of native mature neprosin in crystal form I was also solved by molecular replacement, using the coordinates of fragment T¹³²–I⁴⁰² from native pro-neprosin. Subsequent model building and refinement proceeded as described above. The final model spanned residues T¹³²–Q³⁸⁰ plus the entire C-terminal tag (A⁴⁰¹–I⁴⁰²–A⁴⁰³+H⁴⁰⁴–H⁴⁰⁹), two N-linked glycan chains totalling seven sugar residues, plus one triethylene glycol and 171 solvent molecules. The structure of a product complex of native mature neprosin in crystal form II was also solved by molecular replacement, using fragment T¹³²–Q³⁸⁰ of the crystal form I complex. The final model included residues T¹³²–Q³⁸⁰ plus the C-terminal tag except H⁴⁰⁹ (A⁴⁰¹–I⁴⁰²–A⁴⁰³+H⁴⁰⁴–H⁴⁰⁸), as well as two N-linked glycan chains totalling four sugar residues plus one nickel cation, three sulfate anions, one tetraglycine and one glycine, as well as 250 solvent molecules. The nickel ion, presumably from the Ni-NTA resin used for purification, was tentatively assigned based on short liganding distances to two histidine residues (~1.8 Å), which were closer to those reported for tetrahedrally-coordinated nickel ions (1.88 Å on average) than to those of the more abundant lithium (2.03 Å) from the reservoir solution⁸⁸. A tetraglycine was tentatively placed in an adequate density region based on the capacity of this amino acid to oligomerize under certain conditions⁸⁹. Supplementary Table 1 provides essential statistics on the final refined models, which were validated using the wwPDB Validation Service at https://validate-rcsb-1.wwpdb.org/validservice and deposited with the PDB at www.pdb.org (access codes 7ZU8, 7ZVA, 7ZVB and 7ZVC).

Miscellaneous

Structural superpositions and structure-based sequence alignments were calculated using the SSM program within Coot. Figures were prepared using Chimera. Structure-based similarity searches were performed with Dali⁹⁰. Protein interfaces were calculated with PDBePISA at www.ebi.ac.uk/pdbe/pisa. The interacting surface of a complex was defined as half the sum of the buried surface areas of either molecule.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability

All data and reagents are freely available from the authors upon reasonable request and signature of non-disclosure and material transfer agreements for non-profit usage by academic groups. Source data are provided with this paper. Atomic coordinates are available from the Protein Data Bank under codes 7ZU8, 7ZVA, 7ZVB and 7ZVC. Source data are provided with this paper.

References

Gee, S. On the coeliac affection. St Barth Hosp. Rep. 24, 17–20 (1888).
Google Scholar
Lindfors, K. et al. Coeliac disease. Nat. Rev. Dis. Prim. 5, 3 (2019).
Article PubMed Google Scholar
Monachesi, C. et al. Quantification of accidental gluten contamination in the diet of children with treated celiac disease. Nutrients 13, 190 (2021).
Article CAS PubMed Central Google Scholar
Singh, P. et al. Global prevalence of celiac disease: systematic review and meta-analysis. Clin. Gastroenterol. Hepatol. 16, 823–836(e822) (2018).
Article PubMed Google Scholar
King, J. A. et al. Incidence of celiac disease Is increasing over time: a systematic review and meta-analysis. Am. J. Gastroenterol. 115, 507–525 (2020).
Article PubMed Google Scholar
Hausch, F., Shan, L., Santiago, N. A., Gray, G. M. & Khosla, C. Intestinal digestive resistance of immunodominant gliadin peptides. Am. J. Physiol. Gastrointest. Liver Physiol. 283, G996–G1003 (2002).
Article CAS PubMed Google Scholar
Balakireva A. V., Zamyatnin A. A. Properties of gluten intolerance: gluten structure, evolution, pathogenicity and detoxification capabilities. Nutrients 8, 644 (2016).
El-Salhy, M., Hatlebakk, J. G., Gilja, O. H. & Hausken, T. The relation between celiac disease, nonceliac gluten sensitivity and irritable bowel syndrome. Nutr. J. 14, 92 (2015).
Article PubMed PubMed Central CAS Google Scholar
Daveson, A. J. M. et al. Baseline quantitative histology in therapeutics trials reveals villus atrophy in most patients with coeliac disease who appear well controlled on gluten-free diet. GastroHep 2, 22–30 (2020).
Article Google Scholar
Shan, L. et al. Structural basis for gluten intolerance in celiac sprue. Science 297, 2275–2279 (2002).
Article ADS CAS PubMed Google Scholar
Wei, G., Helmerhorst, E. J., Darwish, G., Blumenkranz, G. & Schuppan, D. Gluten degrading enzymes for treatment of celiac disease. Nutrients 12, 2095 (2020).
Article CAS PubMed Central Google Scholar
Kivelä L., et al. Current and emerging therapies for coeliac disease. Nat Rev Gastroenterol Hepatol 18, in press (2021).
Suchy, F. J. et al. National Institutes of Health Consensus Development Conference: lactose intolerance and health. Ann. Intern Med 152, 792–796 (2010).
Article PubMed Google Scholar
Molina-Infante, J., Santolaria, S., Sanders, D. S. & Fernández-Banares, F. Systematic review: noncoeliac gluten sensitivity. Aliment Pharm. Ther. 41, 807–820 (2015).
Article CAS Google Scholar
Creed, F. Review article: the incidence and risk factors for irritable bowel syndrome in population-based studies. Aliment Pharm. Ther. 50, 507–516 (2019).
Article Google Scholar
Clarysse, S. et al. Postprandial evolution in composition and characteristics of human duodenal fluids in different nutritional states. J. Pharm. Sci. 98, 1177–1192 (2009).
Article CAS PubMed Google Scholar
König, J., Holster, S., Bruins, M. J. & Brummer, R. J. Randomized clinical trial: effective gluten degradation by Aspergillus niger-derived enzyme in a complex meal setting. Sci. Rep. 7, 13100 (2017).
Article ADS PubMed PubMed Central CAS Google Scholar
Ehren, J. et al. A food-grade enzyme preparation with modest gluten detoxification properties. PloS one 4, e6313 (2009).
Article ADS PubMed PubMed Central CAS Google Scholar
Kulkarni, A., Patel, S., Khanna, D. & Parmar, M. S. Current pharmacological approaches and potential future therapies for Celiac disease. Eur. J. Pharm. 909, 174434 (2021).
Article CAS Google Scholar
Pultz, I. S. et al. AGA Abstracts 1125. Kuma062 effectively digests gluten in the human stomach: results of a phase 1 study. Gastroenterology 158, S–218 (2020).
Article Google Scholar
Krishnareddy, S., Stier, K., Recanati, M., Lebwohl, B. & Green, P. H. Commercially available glutenases: a potential hazard in coeliac disease. Ther. Adv. Gastroenterol. 10, 473–481 (2017).
Article CAS Google Scholar
Rawlings, N. D. & Bateman, A. How to use the MEROPS database and website to help understand peptidase specificity. Protein Sci. 30, 83–92 (2021).
Article CAS PubMed Google Scholar
Lee, L., Zhang, Y., Ozar, B., Sensen, C. W. & Schriemer, D. C. Carnivorous nutrition in pitcher plants (Nepenthes spp.) via an unusual complement of endogenous enzymes. J. Proteome Res 15, 3108–3117 (2016).
Article CAS PubMed Google Scholar
Rey, M. et al. Addressing proteolytic efficiency in enzymatic degradation therapy for celiac disease. Sci. Rep. 6, 30980 (2016).
Article ADS CAS PubMed PubMed Central Google Scholar
Schräder, C. U. et al. Neprosin, a selective prolyl endoprotease for bottom-up proteomics and histone mapping. Mol. Cell Proteom. 16, 1162–1171 (2017).
Article Google Scholar
Schräder, C. U., Ziemianowicz, D. S., Merx, K. & Schriemer, D. C. Simultaneous proteoform analysis of histones H3 and H4 with a simplified middle-down proteomics method. Anal. Chem. 90, 3083–3090 (2018).
Article PubMed CAS Google Scholar
Ericsson, U. B., Hallberg, B. M., DeTitta, G. T., Dekker, N. & Nordlund, P. Thermofluor-based high-throughput stability optimization of proteins for structural studies. Anal. Biochem 357, 289–298 (2006).
Article CAS PubMed Google Scholar
Vieille, C. & Zeikus, G. J. Hyperthermophilic enzymes: sources, uses, and molecular mechanisms for thermostability. Microbiol Mol. Biol. Rev. 65, 1–43 (2001).
Article CAS PubMed PubMed Central Google Scholar
Eder, J. & Fersht, A. R. Pro-sequence-assisted protein folding. Mol. Microbiol 16, 609–614 (1995).
Article CAS PubMed Google Scholar
Raufman J. P. Pepsin. In: Encyclopedia of Gastroenterology. (ed Johnson L. R.). 1st edn. Academic Press—Elsevier (2004).
Butts, C. T., Bierma, J. C. & Martin, R. W. Novel proteases from the genome of the carnivorous plant Drosera capensis: structural prediction and comparative analysis. Proteins 84, 1517–1533 (2016).
Article CAS PubMed PubMed Central Google Scholar
Schechter, I. & Berger, A. On the size of active site in proteases. I. Papain. Biochem Biophys. Res Commun. 27, 157–162 (1967).
Article CAS PubMed Google Scholar
Gomis-Rüth, F. X., Botelho, T. O. & Bode, W. A standard orientation for metallopeptidases. Biochim Biophys. Acta 1824, 157–163 (2012).
Article PubMed CAS Google Scholar
Roberts, N. B., Sheers, R. & Taylor, W. H. Secretion of total pepsin and pepsin 1 in healthy volunteers in response to pentagastrin and to insulin-induced hypoglycaemia. Scand. J. Gastroenterol. 42, 555–561 (2007).
Article CAS PubMed Google Scholar
Qiao, S. W. et al. Antigen presentation to celiac lesion-derived T cells of a 33-mer gliadin peptide naturally formed by gastrointestinal digestion. J. Immunol. 173, 1757–1762 (2004).
Article CAS PubMed Google Scholar
Neumann, U., Kubota, H., Frei, K., Ganu, V. & Leppert, D. Characterization of Mca-Lys-Pro-Leu-Gly-Leu-Dpa-Ala-Arg-NH₂, a fluorogenic substrate with increased specificity constants for collagenases and tumor necrosis factor converting enzyme. Anal. Biochem 328, 166–173 (2004).
Article CAS PubMed Google Scholar
Christensen, E. M. et al. In crystallo screening for proline analog inhibitors of the proline cycle enzyme PYCR1. J. Biol. Chem. 295, 18316–18327 (2020).
Article CAS PubMed Google Scholar
Umezawa, H., Aoyagi, T., Morishima, H. & Matsuzaki, M. Hamada M. Pepstatin, a new pepsin inhibitor produced by Actinomycetes. J. Antibiot. (Tokyo) 23, 259–262 (1970).
Article CAS Google Scholar
Abell, A. D., Hoult, D. A., Bergman, D. A. & Fairlie, D. P. Simple cis-epoxide-based inhibitors of HIV-1 protease. Bioorg. Med Chem. Lett. 7, 2853–2856 (1997).
Article CAS Google Scholar
Engilberge, S. et al. Crystallophore: a versatile lanthanide complex for protein crystallography combining nucleating effects, phasing properties, and luminescence. Chem. Sci. 8, 5909–5917 (2017).
Article CAS PubMed PubMed Central Google Scholar
Stawiski, E. W., Baucom, A. E., Lohr, S. C. & Gregoret, L. M. Predicting protein function from structure: Unique structural features of proteases. Proc. Natl Acad. Sci. USA 97, 3954–3958 (2000).
Article ADS CAS PubMed PubMed Central Google Scholar
Khan, A. R. & James, M. N. Molecular mechanisms for the conversion of zymogens to active proteolytic enzymes. Prot. Sci. 7, 815–836 (1998).
Article CAS Google Scholar
Khan, A. R., Khazanovich-Bernstein, N., Bergmann, E. M. & James, M. N. G. Structural aspects of activation pathways of aspartic protease zymogens and viral 3C protease precursors. Proc. Natl Acad. Sci. USA 96, 10968–10975 (1999).
Article ADS CAS PubMed PubMed Central Google Scholar
Arolas, J. L., Goulas, T., Cuppari, A. & Gomis-Rüth, F. X. Multiple architectures and mechanisms of latency in metallopeptidase zymogens. Chem. Rev. 118, 5581–5597 (2018).
Article CAS PubMed Google Scholar
Fujinaga, M., Chernaia, M. M., Tarasova, N. I., Mosimann, S. C. & James, M. N. G. Crystal structure of human pepsin and its complex with pepstatin. Protein Sci. 4, 960–972 (1995).
Article CAS PubMed PubMed Central Google Scholar
Wlodawer A., Gutschina A., James M. N. G. Chapter 2—Catalytic pathways of aspartic peptidases. In: Handbook of Proteolytic Enzymes (eds Rawlings N. D., Salvesen G. S.). 3rd edn. Academic Press (2013).
Rawlings N. D., Barrett A. J. Chapter 1—Introduction: aspartic and glutamic peptidases and their clans. In: Handbook of Proteolytic Enzymes (eds Rawlings N. D., Salvesen G. S.). 3rd edn. Academic Press (2013).
Ting, T.-Y., Baharin, A., Ramzi, A. B., Ng, C.-L. & Goh, H.-H. Neprosin belongs to a new family of glutamic peptidase based on in silico evidence. Plant Physiol. Biochem 183, 23–35 (2022).
Article CAS PubMed Google Scholar
Krissinel, E. & Henrick, K. Inference of macromolecular assemblies from crystalline state. J. Mol. Biol. 372, 774–797 (2007).
Article CAS PubMed Google Scholar
Janin, J., Bahadur, R. P. & Chakrabarti, P. Protein-protein interaction and quaternary structure. Q Rev. Biophys. 41, 133–180 (2008).
Article CAS PubMed Google Scholar
Herriott, R. M., Bartz, Q. R. & Northrop, J. H. Transformation of swine pepsinogen into swine pepsin by chicken pepsin. J. Gen. Physiol. 21, 575–582 (1938).
Article CAS PubMed PubMed Central Google Scholar
Dall, E. & Brandstetter, H. Mechanistic and structural studies on legumain explain its zymogenicity, distinct activation pathways, and regulation. Proc. Natl Acad. Sci. USA 110, 10940–10945 (2013).
Article ADS CAS PubMed PubMed Central Google Scholar
Robertus, J. D., Kraut, J., Alden, R. A. & Birktoft, J. J. Subtilisin; a stereochemical mechanism involving transition-state stabilization. Biochemistry 11, 4293–4303 (1972).
Article CAS PubMed Google Scholar
Boon, L., Ugarte-Berzal, E., Vandooren, J. & Opdenakker, G. Protease propeptide structures, mechanisms of activation, and functions. Crit. Rev. Biochem Mol. Biol. 55, 111–165 (2020).
Article CAS PubMed Google Scholar
Fujinaga, M., Cherney, M. M., Oyama, H., Oda, K. & James, M. N. G. The molecular structure and catalytic mechanism of a novel carboxyl peptidase from Scytalidium lignicolum. Proc. Natl Acad. Sci. USA 101, 3364–3369 (2004).
Article ADS CAS PubMed PubMed Central Google Scholar
Pillai, B. et al. Crystal structure of scytalidoglutamic peptidase with its first potent inhibitor provides insights into substrate specificity and catalysis. J. Mol. Biol. 365, 343–361 (2007).
Article CAS PubMed Google Scholar
Kondo, M. Y. et al. Studies on the catalytic mechanism of a glutamic peptidase. J. Biol. Chem. 285, 21437–21445 (2010).
Article CAS PubMed PubMed Central Google Scholar
Sasaki, H. et al. The three-dimensional structure of aspergilloglutamic peptidase from Aspergillus niger. Proc. Jpn Acad. Ser. B—Phys. Biol. Sci. 80, 435–438 (2004).
Article ADS CAS PubMed Central Google Scholar
Sasaki, H. et al. The crystal structure of an intermediate dimer of aspergilloglutamic peptidase that mimics the enzyme-activation product complex produced upon autoproteolysis. J. Biochem 152, 45–52 (2012).
Article CAS PubMed Google Scholar
Jara, P. et al. Cloning and characterization of the eapB and eapC genes of Cryphonectria parasitica encoding two new acid proteinases, and disruption of eapC. Mol. Gen. Genet 250, 97–105 (1996).
CAS PubMed Google Scholar
Poussereau, N., Creton, S., Billon-Grand, G., Rascle, C. & Fevre, M. Regulation of acp1, encoding a non-aspartyl acid protease expressed during pathogenesis of Sclerotinia sclerotiorum. Microbiology 147, 717–726 (2001).
Article CAS PubMed Google Scholar
Moon, J. L., Shaw, L. N., Mayo, J. A., Potempa, J. & Travis, J. Isolation and properties of extracellular proteinases of Penicillium marneffei. Biol. Chem. 387, 985–993 (2006).
Article CAS PubMed Google Scholar
O’Donoghue, A. J. et al. Inhibition of a secreted glutamic peptidase prevents growth of the fungus Talaromyces emersonii. J. Biol. Chem. 283, 29186–29195 (2008).
Article PubMed PubMed Central CAS Google Scholar
Rolland, S. et al. pH controls both transcription and post-translational processing of the protease BcACP1 in the phytopathogenic fungus Botrytis cinerea. Microbiology 155, 2097–2105 (2009).
Article CAS PubMed Google Scholar
Jensen, K., Ostergaard, P. R., Wilting, R. & Lassen, S. F. Identification and characterization of a bacterial glutamic peptidase. BMC Biochem 11, 47 (2010).
Article CAS PubMed PubMed Central Google Scholar
Oda, N. et al. Nucleotide sequence of the gene encoding the precursor protein of pepstatin insensitive acid protease B, scytalidopepsin B, from Scytalidium lignicolum. Biosci. Biotechnol. Biochem 62, 1637–1639 (1998).
Article CAS PubMed Google Scholar
Stocchi, N., Revuelta, M. V., Castronuovo, P. A. L., Vera, D. M. A. & ten Have, A. Molecular dynamics and structure function analysis show that substrate binding and specificity are major forces in the functional diversification of Eqolisins. BMC Bioinforma. 19, 338 (2018).
Article CAS Google Scholar
Kataoka, Y. et al. Catalytic residues and substrate specificity of scytalidoglutamic peptidase, the first member of the eqolisin in family (G1) of peptidases. FEBS Lett. 579, 2991–2994 (2005).
Article CAS PubMed Google Scholar
van de Wal, Y., Kooy, Y. M., Drijfhout, J. W., Amons, R. & Koning, F. Peptide binding characteristics of the coeliac disease-associated DQ(α1*0501, β1*0201) molecule. Immunogenetics 44, 246–253 (1996).
Article PubMed Google Scholar
Siegel, M. et al. Rational design of combination enzyme therapy for celiac sprue. Chem. Biol. 13, 649–658 (2006).
Article CAS PubMed Google Scholar
Gass, J., Bethune, M. T., Siegel, M., Spencer, A. & Khosla, C. Combination enzyme therapy for gastric digestion of dietary gluten in patients with celiac sprue. Gastroenterology 133, 472–480 (2007).
Article CAS PubMed Google Scholar
Mitea, C. et al. Efficient degradation of gluten by a prolyl endoprotease in a gastrointestinal model: implications for coeliac disease. Gut 57, 25–32 (2008).
Article CAS PubMed Google Scholar
Sims, A. H., Dunn-Coleman, N. S., Robson, G. D. & Oliver, S. G. Glutamic protease distribution is limited to filamentous fungi. FEMS Microbiol Lett. 239, 95–101 (2004).
Article CAS PubMed Google Scholar
Keeling, P. J. Functional and ecological impacts of horizontal gene transfer in eukaryotes. Curr. Opin. Genet Dev. 19, 613–619 (2009).
Article CAS PubMed Google Scholar
Swift, M. L. GraphPad Prism, data analysis, and scientific graphing. J. Chem. Inf. Comput Sci. 37, 411–412 (1997).
Article CAS Google Scholar
Kornbrot, D. Statistical software for microcomputers: SigmaPlot 2000 and SigmaStat2. Br. J. Math. Stat. Psychol. 53, 335–337 (2000).
Article PubMed Google Scholar
Barsnes, H. & Vaudel, M. SearchGUI: a highly adaptable common interface for proteomics search and de novo engines. J. Proteome Res 17, 2552–2555 (2018).
Article CAS PubMed Google Scholar
Vaudel, M. et al. PeptideShaker enables reanalysis of MS-derived proteomics data sets. Nat. Biotechnol. 33, 22–24 (2015).
Article CAS PubMed Google Scholar
Padmanabhan, P., Grosse, J., Asad, A. B., Radda, G. K. & Golay, X. Gastrointestinal transit measurements in mice with 99mTc-DTPA-labeled activated charcoal using NanoSPECT-CT. EJNMMI Res 3, 60 (2013).
Article PubMed PubMed Central CAS Google Scholar
Morón, B. et al. Toward the assessment of food toxicity for celiac patients: characterization of monoclonal antibodies to a main immunogenic gluten peptide. PloS one 3, e2294 (2008).
Article ADS PubMed PubMed Central CAS Google Scholar
Kabsch, W. Xds. Acta Crystallogr sect. D. 66, 125–132 (2010).
Article CAS Google Scholar
Adams, P. D. et al. PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr sect. D. 66, 213–221 (2010).
Article CAS Google Scholar
Winn, M. D. et al. Overview of the CCP4 suite and current developments. Acta Crystallogr sect. D. 67, 235–242 (2011).
Article CAS Google Scholar
Langer, G., Cohen, S. X., Lamzin, V. S. & Perrakis, A. Automated macromolecular model building for X-ray crystallography using ARP/wARP version 7. Nat. Protoc. 3, 1171–1179 (2008).
Article CAS PubMed PubMed Central Google Scholar
Huang, C. C., Meng, E. C., Morris, J. H., Pettersen, E. F. & Ferrin, T. E. Enhancing UCSF Chimera through web services. Nucl. Acids Res 42, W478–W484 (2014).
Article CAS PubMed PubMed Central Google Scholar
Casañal, A., Lohkamp, B. & Emsley, P. Current developments in Coot for macromolecular model building of electron cryo-microscopy and crystallographic data. Protein Sci. 29, 1069–1078 (2020).
Article PubMed PubMed Central CAS Google Scholar
Smart, O. S. et al. Exploiting structure similarity in refinement: automated NCS and target-structure restraints in BUSTER. Acta Crystallogr sect. D. 68, 368–380 (2012).
Article CAS Google Scholar
Kuppuraj, G., Dudev, M. & Lim, C. Factors governing metal-ligand distances and coordination gemoetries of metal complexes. J. Phys. Chem. B 113, 2952–2960 (2009).
Article CAS PubMed Google Scholar
Campbell, T. D., Febrian, R., Kleinschmidt, H. E., Smith, K. A. & Bracher, P. J. Quantitative analysis of glycine oligomerization by ion-pair chromatography. ACS Omega 4, 12745–12752 (2019).
Article CAS PubMed PubMed Central Google Scholar
Holm, L. & Laakso, L. M. Dali server update. Nucleic Acids Res 44, W351–W355 (2016).
Article CAS PubMed PubMed Central Google Scholar

Download references

Acknowledgements

We are grateful to Laura Company, Roman Bonet, Xandra Kreplin and Joan Pous from the joint IBMB/IRB Automated Crystallography Platform and the Protein Purification Service for assistance during purification and crystallization. Plasmid pCMV was kindly provided by Jan J. Enghild, Århus University, Denmark. The authors also would like to thank the ESRF, DIAMOND, and ALBA synchrotrons for beamtime and the respective beamline staff for assistance during diffraction data collection. This study was supported in part by grants from Spanish and Catalan public and private bodies (grant/fellowship references PID2019-107725RG-I00 to F.X.G.-R., A.R.B., U.E. and T.G.; BES-2016-076877 to S.R.M., BES-2015-074583 to L.A.M., Beatriu de Pinós 2018BP00163 to U.E., 2017SGR3 and Fundació La Marató de TV3 201815 to F.X.G.-R., U.E., A.R.B. and T.G.). The authors thank Richard M. Twyman for editing the manuscript.

Author information

These authors contributed equally: Laura del Amo-Maestro, Soraia R. Mendes.

Authors and Affiliations

Proteolysis Laboratory; Department of Structural Biology, Molecular Biology Institute of Barcelona (CSIC), Barcelona Science Park; c/Baldiri Reixac, 15-21, 08028, Barcelona, Catalonia, Spain
Laura del Amo-Maestro, Soraia R. Mendes, Arturo Rodríguez-Banqueri, Laura Garzon-Flores, Tibisay Guevara, Ulrich Eckhard & F. Xavier Gomis-Rüth
Section of Physiology; Department of Biochemistry and Physiology; Faculty of Pharmacy and Food Science, University of Barcelona, Av. Joan XXIII, 27-31, 08028, Barcelona, Catalonia, Spain
Marina Girbal, María José Rodríguez-Lagunas, Àngels Franch & Francisco J. Pérez-Cano
Research Institute of Nutrition and Food Safety (INSA-UB), University of Barcelona, Av. Prat de la Riba, 171, 08921, Santa Coloma de Gramenet, Catalonia, Spain
Marina Girbal, María José Rodríguez-Lagunas, Àngels Franch & Francisco J. Pérez-Cano

Authors

Laura del Amo-Maestro
View author publications
You can also search for this author in PubMed Google Scholar
Soraia R. Mendes
View author publications
You can also search for this author in PubMed Google Scholar
Arturo Rodríguez-Banqueri
View author publications
You can also search for this author in PubMed Google Scholar
Laura Garzon-Flores
View author publications
You can also search for this author in PubMed Google Scholar
Marina Girbal
View author publications
You can also search for this author in PubMed Google Scholar
María José Rodríguez-Lagunas
View author publications
You can also search for this author in PubMed Google Scholar
Tibisay Guevara
View author publications
You can also search for this author in PubMed Google Scholar
Àngels Franch
View author publications
You can also search for this author in PubMed Google Scholar
Francisco J. Pérez-Cano
View author publications
You can also search for this author in PubMed Google Scholar
Ulrich Eckhard
View author publications
You can also search for this author in PubMed Google Scholar
F. Xavier Gomis-Rüth
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

F.X.G.R. conceived and supervised the project; L.d.A.M., T.G., L.G.-F. and S.R.M. produced and purified proteins, generated mutants, and performed in vitro studies; L.d.A.M., A.R.B. and T.G. crystallized proteins; A.R.B. and U.E. collected diffraction data; U.E. performed experiments, analysed data and supervised workers; F.X.G.R. solved and refined crystal structures; F.J.P.C., M.G., M.J.R.L., and À.F. performed animal experiments, and F.X.G.R. wrote the manuscript with contributions from all authors.

Corresponding author

Correspondence to F. Xavier Gomis-Rüth.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks Hans Brandstetter and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Reporting Summary

Source data

Source Data

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

del Amo-Maestro, L., Mendes, S.R., Rodríguez-Banqueri, A. et al. Molecular and in vivo studies of a glutamate-class prolyl-endopeptidase for coeliac disease therapy. Nat Commun 13, 4446 (2022). https://doi.org/10.1038/s41467-022-32215-1

Download citation

Received: 16 May 2022
Accepted: 21 July 2022
Published: 01 August 2022
DOI: https://doi.org/10.1038/s41467-022-32215-1

This article is cited by

A family of di-glutamate mucin-degrading enzymes that bridges glycan hydrolases and peptidases
- Yoshiki Narimatsu
- Christian Büll
- Ramon Hurtado-Guerrero
Nature Catalysis (2024)

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.