Hydrolysis of Gluten-Derived Celiac Disease-Triggering Peptides across a Broad pH Range by RmuAP1: A Novel Aspartic Peptidase Isolated from Rhodotorula mucilaginosa

Abstract

An aspartate peptidase with proteolytic activity toward gluten was identified from an isolated red yeast Rhodotorula mucilaginosa strain. This peptidase consists of 425 amino acids, comprising an N-terminal signal peptide, a propeptide, and a C-terminal catalytic domain. The catalytic domain, termed RmuAP1_CD, could be secreted by the recombinant oleaginous yeast Yarrowia lipolytica, whose genome contains the expression cassette for RmuAP1_CD. RmuAP1_CD exhibited optimum activity at pH 2.5 when acting on bovine serum albumin. Moreover, it facilitated the hydrolysis of gluten-derived immunogenic peptides (GIPs), which are responsible for triggering celiac disease symptoms, across a pH range of 3.0–6.0. The preferred cleavage sites are P–Q–Q–↓–P–Q in the 26-mer and P–Q–L–↓–P–Y in the 33-mer GIPs. Conversely, porcine pepsin cannot hydrolyze these two GIPs. The ability of RmuAP1_CD to degrade GIPs under acidic conditions of the stomach indicates its potential as a viable oral enzyme therapy for celiac disease.

Introduction

Celiac disease (CeD) is an autoimmune disease triggered by the daily ingestion of gluten in genetically predisposed individuals who carry the DQ2 or DQ8 HLA class II alleles.^1,2 The primary pathological characteristics of CeD are intestinal villous atrophy and crypt hyperplasia, associated with typical gastrointestinal symptoms including chronic diarrhea, stomach bloating, vomiting, and abdominal cramping.³ Additionally, CeD can manifest through a range of extraintestinal symptoms, such as migraine, cerebellar ataxia, chronic fatigue, dermatitis herpetiformis, arthritis, osteoporosis, iron-deficiency anemia, and dental enamel hypoplasia. Notably, this disease is also linked to the development of enteropathy-associated intestinal lymphoma.^4,5 Hence, patients with CeD may experience a variety of discomforts, leading to a reduced quality of life. A systematic review and meta-analysis in 2018 revealed that the prevalence of CeD based on serologic tests for antitissue transglutaminase and/or antiendomysial antibodies was 1.4%; however, the result was 0.7% based on biopsy, with prevalence values at 0.4% in South America, 0.5% in Africa and North America, 0.6% in Asia, and 0.8% in Europe and Oceania.⁶ A more recent immunohistopathological screening involving 12,981 Norwegians revealed that 1.47% of the adult population had CeD; notably, 75% of these cases were previously undiagnosed.⁷ This survey indicated that the conventional diagnostic methods for CeD, such as serological tests and biopsies, may occasionally lack the necessary sensitivity, resulting in the possibility of overlooking diagnoses.

Gluten is a group of water-insoluble proteins in wheat, barley, and rye grains. A subgroup of proteins in gluten is called prolamin because of the high proline and glutamine content. To be specific, prolamin in wheat, barley, and rye is named gliadin, hordein, and secalin, respectively. The rich content of proline makes prolamin resistant to being fully digested by pepsin in the stomach. Some of the pepsin-resistant peptides might trigger immune responses in susceptible individuals after being modified by tissue transglutaminase 2 in the lamina propria of the duodenum.⁸ The notable gluten-derived immunogenic peptides (GIPs) include the 26- and 33-mer peptides generated from γ5- and α2-gliadins, respectively.

To date, the only effective prescription for CeD patients is a lifelong gluten-free diet (GFD). Nonetheless, patients may not see improvement in their health condition despite adhering to the GFD due to inadvertent consumption of hidden gluten. In patients undergoing a GFD regimen, consumption of as little as 20 mg of gluten per day can harmfully sustain the inflammation status of the small intestine. Therefore, new therapies are urgently needed. Among the developing approaches, oral gluten hydrolase therapy, which aims to destroy GIPs before they enter the duodenum, has been considered promising. When combined with a GFD regimen, the oral peptidase can help to minimize the effects of small accidental gluten consumption. Besides CeD, nonceliac gluten sensitivity (NCGS), a condition characterized by similar intestinal and extraintestinal symptoms, is also related to the ingestion of gluten-containing foods.⁹ Individuals with NCGS may also benefit from taking oral gluten hydrolases.

Gluten hydrolases from various origins with the potential to eliminate gluten contamination have been reported. Notable examples are as follows: EP-B2 from barley,¹⁰ nepenthesin and neprosin from Nepenthes,^11,12 AN-PEP and aspergillopepsin from A. niger,¹³ DPP-IV from A. oryzae,¹⁴ SE-PEP from Sphingomonas capsulata,¹⁵ subtilisin Carlsberg from Bacillus licheniformis,¹⁶ Rmep from Rothia mucilaginosa,¹⁷ and Kuma030, an engineered mutant of kumamolisin, from Alicyclobacillus sendaiensis.¹⁸ The proteolytic properties of these and many other peptidases have been summarized in several recent reviews.¹⁹⁻²² While some of the gluten hydrolases are available on the dietary supplement market, few have undergone clinical trials. IMGX-003 (Latiglutenase, a formula containing EP-B2 and SE-PEP) and TAK-062 (an engineered variant of Kuma030) have completed phase II and phase I trials, respectively.^23,24 The clinical trials showed that IMGX-003 could reduce gluten-induced intestinal mucosal damage and symptom severity, and TAK-062 effectively degraded gluten intake in the stomach. Despite the positive results, the peptidase-to-gliadin ratios employed in these trials were notably elevated. To be specific, participants were orally administered up to 1.2 g of IMGX-003 and 2 g of gluten daily, while 0.9 g of TKA-062 and 3 g of gluten were given each day during the treatment phases. The utilization of high peptidase-to-gluten ratios in these trials suggests that there is still room for improvement in the digestive efficiency of peptidases in the stomach. In brief, an ideal oral peptidase for CeD patients should possess the following characteristics: (1) being able to take action in the stomach, particularly at pH < 4.0, and (2) strong catalytic efficiency toward GIPs so that a smaller dose is sufficient in practice.

To obtain peptidases that could be beneficial for treating CeD, a serine peptidase, called Bga1903, from Burkholderia gladioli was recently characterized in our laboratory.²⁵ Bga1903 belongs to the S8 peptidase family according to MEROPS classification²⁶ and exhibits notable activity in breaking down the 33- and 26-mer peptides, with a particular preference for glutamine located at the P1 position. Active-site modifications based on the crystal structure of Bga1903 further enhanced the proteolytic activity of Bga1903 toward GIPs.²⁷ It is known that the gastric pH after a meal could decrease from pH 6.0 down to 2.0 in about 2 h.²⁸ Therefore, the insignificant activity of Bga1903 at pH < 3.5 may compromise its potential value in helping CeD patients. Screening for microorganisms capable of secreting acidic peptidases was thus continued using agar plates at pH 3.0. An aspartic peptidase secreted by an isolated R. mucilaginosa strain caught our attention. The heterogeneous expression, purification, and activity characterization of this aspartic peptidase, named RmuAP1 hereafter, are described in this report.

Materials and Methods

Proteins and Peptides

Gliadin (product no. G3375), gelatin from fish skin (product no. G7765), and bovine serum albumin (BSA, product no. A7906) were purchased from Sigma-Aldrich (St. Louis, MO). Porcine pepsin was purchased from Roche (Basel, Switzerland). The 33-mer (LQLQPFPQPQLPYPQPQLPYPQPQLPYPQPQPF) and 26-mer (FLQPQQPFQQPQQPYPQQPQQPFPQ) peptides were synthesized by Mission Biotech (Taipei, Taiwan).

Microorganisms

To isolate microorganisms capable of producing acid proteases, soils from the ambient environment were collected, and the diluted suspension was spread on the screening agar plate, which contained 0.3% (w/v) gliadin, 50 mM McIlvaine buffer [pH 3.0], and 1.5% (w/v) agar. The colony with a halo was purified using the streaking method, and its taxonomic classification was determined based on the basic local alignment search tool (BLAST) analysis of the internal transcribed spacer (ITS) regions, including ITS1 and ITS2. To estimate the extent of protease activity, the radial ratio of the halo to the colony was calculated using the ImageJ software program.²⁹ The R. mucilaginosa strain isolated in this study was subsequently maintained in yeastextractpeptonedextrose (YPD) medium (10 g/L yeast extract, 20 g/L peptone, and 20 g/L dextrose, pH 6.5).

Escherichia coli DH5α, used in molecular cloning and plasmid construction, was cultivated in Luria–Bertani medium (10 g/L tryptone, 5 g/L yeast extract, and 10 g/L NaCl, pH 7.0) with antibiotics when necessary.

Yarrowia lipolytica Po1g [MatA, leu2-270, ura3-302::URA3, xpr2-332, and axp-2], the host for recombinant protein expression, was purchased from Yeastern Biotech (Taipei, Taiwan). Y. lipolytica Po1g was cultivated regularly in the YPD medium unless otherwise indicated.

Identification of RmuAP1

The isolated R. mucilaginosa strain was streaked on an agar plate containing 0.3% gliadin and 50 mM McIlvaine buffer [pH 3.0]. Approximately 300 mg of the agar located in the transparent zones surrounding the R. mucilaginosa colonies was blended with 1 mL of sodium acetate buffer (30 mM, pH 5.0) in an Eppendorf tube. The agar was thoroughly crushed using a plastic pestle. After a 10 min centrifugation at 18,000g, the clear supernatant from the mixture was concentrated to 100 μL using a 10 kDa NanoSep centrifugal tube (Port Washington, NY). The concentrated liquid was then analyzed for protein identification by liquid chromatography (LC)–tandem mass spectrometry (LC–MS/MS).

Molecular Cloning

A complementary deoxyribonucleic acid (cDNA) library of the isolated R. mucilaginosa strain was synthesized from messenger ribonucleic acid extracted from the strain grown in YPD broth by using Moloney murine leukemia virus reverse transcriptase (Invitrogen, Carlsbad, CA). RmuAP1 was amplified from the cDNA library by polymerase chain reaction (PCR) using primers 5′-CGGGATCCACACGCGATAATCACCAACCAAC-3′ and 5′-CCCAAGCTTAAAGTTCCGGAGAGTCGCTGC-3′ that match the untranslated regions of the ortholog (GenBank: TKA53319.1) in Rhodotorula sp. CCFEE 5036. The engineered restriction sites EcoRI and HindIII were used to insert amplified RmuAP1 into plasmid pETDuet-1 (Novagen, Madison, WI, USA). The authentic nucleotide sequence of RmuAP1 was determined by the Sanger sequencing method. To express RmuAP1 in Y. lipolytica Po1g, the open-reading frame of RmuAP1 was amplified by PCR using primers 5′-CACGTGAATGCCTTCATTCGCC-GCCTCTGCCGC-3′ and 5′-GGAGGTACCTTAAGCGAGCTTCGAGAAGCCG-3′. The PCR-amplified fragment, with PmlI and KpnI restriction sites engineered at the 5′ and 3′ ends, respectively, was then inserted into the protein expression plasmid pYLEX1 (Yeastern Biotech, Taipei, Taiwan). The recombinant plasmid was further modified to include a (His)₆-coding tag at the 3′ end of RmuAP1 by PCR-based insertion mutagenesis using the Q5 site-directed mutagenesis kit (New England Biolabs, Ipswich, MA) to create pYLEX1/RmuAP1. Finally, pYLEX1/RmuAP1 was linearized with NotI and introduced into the competent cells of Y. lipolytica Po1g by the chemical method according to the protocol recommended by Yeastern Biotech. The recombinant yeast carrying the expression cassette of RmuAP1 in the genome was selected on leucine-free yeast nitrogen base (YNB) agar plates (20 g/L glucose, 1.7 g/L YNB [without amino acids], 5 g/L ammonium sulfate, and 15 g/L agar, pH 5.4). The integration of RmuAP1 into the host’s genome was confirmed by PCR.

Heterologous Expression and Purification of RmuAP1

The recombinant Y. lipolytica Po1g, which contains the (His)₆-tagged RmuAP1 in the genome, was cultivated in modified YPG media (50 mM McIlvaine buffer [pH 5.0], 10 g/L yeast extract, 20 g/L peptone, and 2% (v/v) glycerol) at 28 °C and 150 rpm for 4 days. After a 15 min centrifugation at 10,000g, 4 °C, the cell-free supernatant was mixed with 10 volumes of equilibrium buffer (20 mM sodium phosphate [pH 6.0], 0.5 M NaCl, and 40 mM imidazole). Then, the mixture was loaded onto a HisPrep FF 20 mL column (GE Healthcare Life Science, Marlborough, MA). After an excessive wash with equilibrium buffer, the mature RmuAP1 (RmuAP1_CD) was eluted using a buffer, pH 3.6, consisting of 60% equilibrium buffer [pH 6.0] and 40% 0.1 M glycine–HCl [pH 2.5]. The fractions with peptidase activity, as indicated by the development of clear zones on the gliadin-containing agar plate, were pooled and concentrated, and the buffer was replaced with 50 mM sodium acetate [pH 5.0]. The purified RmuAP1_CD with 50% glycerol was stored at −20 °C for later use. The protein concentration was calculated on the basis of the absorbance at 280 nm according to the Bougure–Beer law (OD = ε × M × 1 cm), where the molar extinction coefficient (ε) of RmuAP1_CD was estimated to be 60,850 M^–1 × cm^–1 by adding ε of tryptophan residues (5500 M^–1 cm^–1) and tyrosine residues (1490 M^–1 cm^–1) under 280 nm.³⁰

Activity Characterization of RmuAP1_CD

The peptidase activity was regularly assayed by using BSA as the substrate. The 1 mL reaction mixture, consisting of 10 μg of the purified RmuAP1_CD and 5 mg of BSA, was incubated under specified conditions for 1–1.5 h. The reaction was stopped by the addition of cold trichloroacetic acid to a final concentration of 10% (w/v). The absorbance at 280 nm of the clarified supernatant after centrifugation at 20,000g, 4 °C, 10 min was measured with the JASCO V550 UV/VIS spectrophotometer (Tokyo, Japan) to estimate the aromatic residues released from BSA. Each assay was performed in triplicate.

To determine the pH dependence of the peptidase activity, the reactions were performed at a pH within the range of 1.0–8.0 (50 mM K/HCl, pH 1.0–2.2; 50 mM glycine–HCl, pH 2.2–3.0; and 50 mM McIlvaine, pH 2.5–8.0), 37 °C. The pH stability of RmuAP1_CD was determined by preincubating the peptidase at different pH buffers, 37 °C, for 3 h. Subsequently, the residual peptidase activity was assayed at pH 2.5. To determine the temperature dependence of the peptidase activity, the reactions were performed at temperatures in the range of 4–60 °C, pH 2.5, for 1 h. The thermostability of RmuAP1_CD was determined by preincubating the peptidase at different temperatures (4–60 °C), pH 2.5, for 3 h. Then, the residual peptidase activity was assayed at 37 °C and pH 2.5.

To determine the effects of potential inhibitors on the peptidase activity, purified RmuAP1_CD was preincubated with various potential inhibitors at 37 °C for 3 h. The inhibitors and their final concentrations at the preincubation step were 5 mM ethylenediaminetetraacetic acid (EDTA), 1 mM phenylmethylsulfonyl fluoride (PMSF), 1 mM pepstatin A, 5 mM β-mercaptoethanol (β-ME), 1% (v/v) Triton X-100, and 0.1% (w/v) sodium dodecyl sulfate (SDS). The residual peptidase activity toward BSA at pH 2.5 was then determined according to the regular assay described previously.

Gliadin Hydrolysis Assay by Peptidases

The reaction that contained gliadin (7.5 mg/mL), the purified RmuAP1_CD (2–256 μg/mL) or porcine pepsin (8–1024 μg/mL), and 50 mM McIlvaine buffer [pH 2.5] was performed at 37 °C for 1.5 h. The reaction was stopped by the addition of protein loading buffer, followed by heating at 95 °C for 5 min. The gliadin hydrolysate was then analyzed by SDS-15% polyacrylamide gel electrophoresis (PAGE). The protein fragments on the gel were visualized by staining with a Coomassie Blue R-250.

GIP Hydrolysis Assay by Peptidases

The reaction that contained 26- or 33-mer peptide (1 mg/mL), the purified RmuAP1_CD (25 μg/mL) or porcine pepsin (500 μg/mL), and 50 mM McIlvaine buffer at the indicated pH (2.5–6.0) was performed at 37 °C for 3 h. The reaction was stopped by heating the solution to 95 °C for 10 min. The degradation degree of the peptide was analyzed with a Gilson HPLC system (Middleton, WI) using a C18 column (Ascentis Express 25 cm × 4.6 mm, 5 μm, Supelco, Bellefonte, PA). The elution program consisted of (1) a 5 min wash with water and (2) a linear gradient from 20 to 60% acetonitrile over a period of 12 min. Both water and acetonitrile used in the program contained 0.1% (v/v) trifluoroacetic acid. The flow rate was set at 1 mL/min, and the elution was monitored at 220 nm.

The identity of cleaved products of the 26- and 33-mer peptides, resulting from the action of RmuAP1_CD at pH 3.0 after a 3 h incubation, was analyzed by LC–MS/MS, in which the sequences of the GIPs were used as the searching database. Each of the degraded peptides was aligned to the peptide substrate (26- or 33-mer), from which the scissile bond could be assigned. In case of ambiguous assignments due to the repetitive sequence along the peptide substrate, the scissile bond was tentatively assigned to the one proximate to the N-terminus. The frequency (%) of the scissile bond was then calculated by the count of that specific cut in the total cut number, as determined according to the formula

LC–Tandem Mass Spectrometry

LC–MS/MS was conducted by the Biotechnology Center at National Chung Hsing University using an UltiMate 3000 RSLCnano LC system (Thermo Fisher Scientific, Waltham, MA) coupled with a Sciex TripleTOF 6600 mass spectrometer (Applied Biosystems, Waltham, MA). The trypsin-digested products of proteins secreted by the isolated R. mucilaginosa strain were separated on an Acclaim PepMap C18 analytical column (Thermo Fisher Scientific, Waltham, MA) through a gradient elution program with increasing acetonitrile concentration. Mass spectrometry was performed in positive mode with an electrospray ionization source, employing the following settings: ion spray voltage +2500 V, interface heater temperature 150 °C, declustering potential 80 V, and column oven temperature 35 °C. The information-dependent acquisition mode was utilized with a scan range set to 65–1800 m/z and an acquisition time of 0.1 s. For database searching, Mascot version 2.3.02 (Matrix Science, London, UK) was employed with the following parameter settings: monoisotopic mass values, peptide mass tolerance of ±0.05 Da, fragment mass tolerance of ±0.03 Da, and a maximum of 0 missed cleavages. The analysis of proteolytic fragments of GIPs by RmuAP1_CD followed the same method, except that the initial trypsin pretreatment step was omitted.

In Silico Protein Structure Prediction and Analysis

The search of homologous proteins to RmuAP1_CD was performed using the protein BLAST, NCBI. The functional domain prediction of RmuAP1_CD was performed on the Web site InterPro.³¹ The secondary structure of RmuAP1_CD was predicted using the online free software ESPript 3.0.³² The 3D protein structure of RmuAP1_CD was predicted using ColabFold,³³ a freely available tool based on AlphaFold2.^34,35 The 3D structure model with the highest predictedlocaldistancedifferencetest (pLDDT) value was taken as the optimum structure of RmuAP1_CD. The superimposition of the predicted structure of RmuAP1_CD and pepsin (PDB ID: 1PSA) was visualized using PyMOL.³⁶

To create credible docking models for the peptidases, the peptidyl ligand PQQPQ was initially constructed using simplified molecular input line entry system strings and the UCSF Chimera interface.^37,38 The peptidase’s binding pocket was then identified with the assistance of the PeptiMap Server.³⁹ Subsequently, site-targeted docking between the peptidase and the ligand was conducted using AutoDock Vina.⁴⁰ AutoDock Tools managed polar hydrogens, and Gasteiger charges were applied to each molecule. Following the placement of the ligand within the pocket, simulations were executed ten times, each with varying random seeds, until a suitable conformation was achieved. The docking poses were then screened based on structural correlations between the peptidase and the ligand. Finally, the molecular conformations were exported, and the docking results were visualized using PyMOL. The hydrogen bonding interactions between the ligand and the peptidase were visually assessed through LigPlot⁺.⁴¹

Results

Identification of RmuAP1

Efforts for isolating microorganisms capable of secreting gluten hydrolases under acidic conditions in our laboratory resulted in the acquisition of a handful of colonies with varying morphologies on the screening agar plates. These colonies could grow at pH 3.0, relying on gliadin as the sole carbon and nitrogen source. The appearance of clear zones surrounding the colonies indicated that gliadin in the agar was degraded by secreted peptidases (Figure 1). The taxonomic classification through a BLAST analysis of the ITS sequences identified these strains as Acremonium sp., Cladosporium sp., Penicillium citrinum, Pseudocercosporella fraxini, R. mucilaginosa, Saccharomycopsis fibuligera, and Simplicillium sp. Among these, the R. mucilaginosa strain was selected for further scrutiny due to its relatively high ratio of the radius of the clear zone to the colony (Table S1). Proteins secreted by this isolated R. mucilaginosa strain were identified by LC–MS/MS using the available databases of Rhodotorula sp. CCFEE 5036 (taxid:1965284), which was isolated from McMurdo Dry Valleys, Antarctica.⁴² The best hit was a peptidase (GenBank: TKA53319.1) of the A1 family. Therefore, the R. mucilaginosa peptidase identified in this screen was called RmuAP1.

Figure 1.

Screening for microorganisms secreting gliadin hydrolase. The indicated microbial strains were cultivated on the gliadin-containing agar plate. The radial ratio (mean ± SD) of the surrounding clear zones to colonies of each strain is presented.

The cDNA encoding RmuAP1 was obtained as described in Materials and Methods. The nucleotides and the deduced amino acid sequence (Figure S1) revealed that RmuAP1 is completely identical to the hypothetical protein (GenBank entry TKA53319.1) of Rhodotorula sp. CCFEE 5036. Protein BLAST using the nonredundant protein sequence database found another completely identical protein, which is annotated as type I transmembrane sorting receptor (GenBank: KAG0663087.1) encoded by the presumable gene “PEP1_2” of the R. mucilaginosa strain KR isolated from kefir. The second-best hit was an aspartate peptidase (GenBank: KWU42276.1) deduced from the genome of Rhodotorula sp. JG1b, isolated also from McMurdo Dry Valleys, Antarctica. RmuAP1 and the C-terminal half of KWU42276.1 share 96% identity in their amino acid sequences. The next hit (82% identity) was an aspartate peptidase (GenBank: QCC30103.1) hypothetically produced by Rhodotorula taiwanensis strain RS1. RmuAP1 comprises 425 amino acid residues and belongs to the aspartic peptidase family A1. It was predicted to contain an N-terminal signal peptide (residues 1–20), a propeptide domain (residues 21–110), and a C-terminal catalytic domain (residues 111–425) (Figure S1) by analysis using the Web site InterPro.³¹

The catalytic domain of RmuAP1 (aliased as RmuAP1_CD) shares 35% identity with porcine pepsin in their amino acid sequences (Figure 2a). The pairwise alignment indicates the presence of the catalytic aspartate in the conserved motifs DTGS and DTGT. The tertiary structure of RmuAP1_CD, predicted by the ColabFold algorithm, has a bilobal fold similar to that of other pepsin-like peptidases (Figure 2b). According to this predicted model, an extended active-site cleft is formed between the two lobes, with each of the two catalytic aspartates situated separately in each lobe, and a β-hairpin structure, the flap, projects over the catalytic cleft. Two disulfide bonds, each in one lobe, are formed between C42 and C47 and between C253 and C284. Overall, the Cα traces of RmuAP1_CD and porcine pepsin can be perfectly superimposed (Figure 2c).

Figure 2.

Primary, secondary, and tertiary structures of RmuAP1_CD. (a) Pairwise comparison of RmuAP1_CD and porcine pepsin with regard to their primary amino acid sequences. Secondary structural elements are shown on top of the sequence alignment. The consensus residues in the alignment are presented on a blue background, except for the catalytic aspartate-containing motif D-T-G-S/T, which is presented on a peach-colored background. The β-hairpin sequence, the flap, is boxed by a red line rectangle. The distinct residues presumably involved in subsite formation, consequently affecting the substrate selection, are indicated by the symbol (∧). (b) The tertiary structure of RmuAP1_CD, predicted by ColabFold, is presented in the ribbon diagram. The catalytic aspartates are shown in magenta sticks and the disulfide bonds in green sticks (S: golden). The flap is shown in yellow. (c) Superimposition of the tertiary structures of RmuAP1_CD and porcine pepsin (PDB: 1PSA) in Cα traces was performed by PyMOL. RmuAP1_CD, green; pepsin, magenta.

Preparation of Recombinant RmuAP1_CD

To facilitate future applications as an oral therapeutic peptidase, it is preferable to produce RmuAP1_CD in large quantities using organisms that are generally recognized as safe (GRAS). Considering this requirement, Y. lipolytica was selected as the host in this study. Y. lipolytica is an ascomycetous yeast known for its ability to degrade lipids and proteins. By utilizing this host, we aim to ensure safety and scalability in the production of RmuAP1_CD. The pYLEX1/RmuAP1 plasmid (Figure 3a) was linearized and introduced into Y. lipolytica Po1g. The recombinant yeast that harbors the integrated RmuAP1 expression cassette in the genome was selected by using a solid medium plate without leucine. The expression of RmuAP1 in the recombinant yeast would be driven by the hybrid Hp4d promoter constitutively,⁴³ and the protein would have an engineered (His)₆-tag at the C-terminus. The accumulation of an additional protein of ∼35 kDa in the culture medium suggested that Y. lipolytica Po1g can recognize the signal peptide of RmuAP1 and facilitate the secretion of RmuAP1_CD into the medium (Figure 3b). Western blotting analysis confirmed the authenticity of the secreted protein to be part of RmuAP1 (Figure 3c). Moreover, the peptidase activity exhibited by the culture medium of Y. lipolytica Po1g depended largely on the ∼35 kDa protein (data not shown). Accordingly, this ∼35 kDa protein was postulated to represent RmuAP1_CD.

Figure 3.

Heterologous expression of RmuAP1. (a) Map of plasmid pYLEX1/RmuAP1. The intact open reading frame of RmuAP1, with an engineered hexahistidine-coding sequence, was placed under the hybrid promoter Hp4d. A sequence encoding hexahistidine was engineered at the 3′ end of the open reading frame of RmuAP1. (b) Proteins secreted by the recombinant Y. lipolytica Po1g. Proteins in the culture medium were analyzed by SDS-PAGE. pYLEX1/RmuAP1 denotes the sample from the yeast carrying the RmuAp1 expression cassette, while pYLEX1 serves as a control. The open arrow points to the protein band presumably representing RmuAP1_CD. (c) Western blotting analysis of the samples same as those in panel (b) by using rabbit anti-(His)₆ antibody (Recenttec, Tokyo, Japan) as the primary antibody and alkaline phosphatase-conjugated goat antirabbit IgG antibody (GeneTex, Irvine, CA) as the secondary antibody.

Purification of RmuAP1_CD was achieved by immobilized metal affinity chromatography (IMAC) as described in Materials and Methods. Because RmuAP1_CD is stable in an acidic buffer at pH values as low as 2.5, elution of the recombinant protein from the column was performed by using a decreasing pH gradient. The ∼35 kDa protein, which appears to be homogeneous, as indicated on the gel of SDS-PAGE, was obtained after the purification process (Figure 4). The production yield of purified RmuAP1_CD was estimated to be up to 15 mg/L of culture broth.

Figure 4.

Purification of RmuAP1_CD. RmuAP1_CD in the culture medium was purified by IMAC. The samples before and after the purification were analyzed by SDS-PAGE. The open arrow points to the band representing RmuAP1_CD.

Catalytic Properties of RmuAP1_CD

Besides gluten, a broad range of substrates such as BSA and casein could be degraded by RmuAP1_CD. Because the solubility of gluten was limited at neutral pH, BSA was utilized as the substrate when the general catalytic properties of RmuAP1_CD were characterized. The pH profile of the purified RmuAP1_CD had maximal activity at pH 2.5 and declined to an inactive level when pH ≥ 5.0 (Figure 5a). RmuAP1_CD was relatively stable, retaining ≥85% activity, after a 3 h preincubation in buffers of pH values ranging from 3.0 to 6.0 (Figure 5b). RmuAP1_CD exhibited its greatest stability at pH 4.0 but was labile when the pH value increased to 7.0. The temperature profile indicated that the purified RmuAP1_CD had optimal performance at 37 °C (Figure 5c), and the peptidase remained fully active after a 3 h preincubation at temperatures no higher than 37 °C (Figure 5d).

Figure 5.

Proteolytic activity of RmuAP1_CD on bovine serum albumin. (a) pH-dependence of the activity. The relativity activity at pH 2.5 was taken as 100%. K/HCl was used for pH 1.0–2.2, glycine–HCl for pH 2.2–3.0, and McIlvaine buffer for pH 2.5–8.0. (b) Residual activity after preincubation at the indicated pH, 37 °C, for 3 h. The relative residual activity at pH 4.0 was taken as 100%. (c) Temperature dependence of the activity. The relativity activity at 37 °C was taken as 100%. (d) Residual activity after preincubation at the indicated temperature, pH 2.5, for 3 h. The relative residual activity at 20 °C was taken as 100%. The data are the mean of three independent assays. Error bars indicate standard deviations.

The impact of several potential inhibitors of peptidases was examined (Table 1). The proteolytic activity of RmuAP1_CD was unaffected by EDTA and PMSF, indicating that RmuAP1_CD is neither a metallopeptidase nor a serine peptidase. By contrast, the peptidase activity was drastically inhibited by pepstatin A, confirming RmuAP1_CD as an aspartic peptidase, classified in peptidase family A1. The reducing agent β-ME did not influence the peptidase activity, implying that the two disulfide bonds are not critical for the protein stability. Detergents such as 1% (v/v) Triton X-100 and 0.1% (w/v) SDS moderately inhibited the proteolytic activity of RmuAP1_CD. The inhibitory effects of the tested agents on pepsin were analogous to those on RmuAP1_CD.

Table 1. Effects of Chemicals on the Proteolytic Activity of RmuAP1_CD and Pepsin.

	RmuAP1CD (%)	pepsin (%)
control (no additive)	100	100
EDTA (5 mM)	101 ± 2.3	95 ± 2.5
PMSF (1 mM)	96 ± 4.5	81 ± 8.9
pepstatin A (1 mM)	6 ± 1.8	7 ± 1.6
β-ME (5 mM)	109 ± 0.8	90 ± 1.0
Triton X-100 (1%, v/v)	44 ± 9.2	25 ± 4.9
SDS (0.1%, w/v)	19 ± 4.5	30 ± 4.0

RmuAP1_CD-Catalyzed Hydrolysis of Gliadin

The hydrolysis of gliadins catalyzed by purified RmuAP1_CD or pepsin at pH 2.5 was assessed. Gliadin (7.5 mg/mL) was treated with a series of increasing amounts of RmuAP1_CD (2–256 μg/mL) or pepsin (8–1024 μg/mL) at 37 °C for 1 h. The degree of gliadin hydrolysis was then analyzed by SDS-PAGE (Figure 6). Gliadins in the assays could be effectively degraded by RmuAP1_CD at a dosage of ≥64 μg/mL, whereas pepsin at a dosage of 512 μg/mL was minimally required to achieve the degradation of gliadins. Notably, RmuAP1_CD removed the small degraded gliadins, centered around 11 kDa, more efficiently than pepsin.

Figure 6.

Hydrolysis of gliadins catalyzed by RmuAP1_CD and porcine pepsin. Gliadins (7.5 mg/mL) were digested by (a) RmuAP1_CD at the indicated concentrations (2–256 μg/mL) or by (b) pepsin at the indicated concentrations (8–1024 μg/mL). All the reactions were performed at pH 2.5 for 90 min. Then, the degradation degree of gliadins was analyzed by SDS-PAGE.

Removal of GIPs is a critical matter in terms of CeD treatment. Thus, the effectiveness of the purified RmuAP1_CD in removing the 33- and 26-mer peptides was examined under the reaction conditions described in the Materials and Methods. The high-performance liquid chromatography (HPLC) chromatograms of the reaction products indicate that both of the immunogenic peptides could be effectively degraded by RmuAP1_CD at pH ranging from 4.0 to 6.0 (Figure 7a,b). The 26-mer peptide could be degraded even when the pH was dropped to 3.0. By contrast, the 33-mer and 26-mer peptides were not degraded by pepsin, even with a significantly higher dosage (approximately 20-fold) of pepsin.

Figure 7.

Hydrolysis of GIPs by RmuAP1_CD and porcine pepsin. The 33-mer peptide (a) and 26-mer peptide (b) at the concentration of 1 mg/mL were separately digested by RmuAP1_CD (25 μg/mL) or pepsin (500 μg/mL) at the indicated pH, 37 °C, for 3 h. The degradation of the peptides was analyzed by reverse-phase HPLC under the conditions described in Materials and Methods.

To determine the identity of the degraded fragments of GIPs, the products of GIPs catalyzed by RmuAP1_CD at pH 3.0 were subjected to LC–MS/MS. The determined fragments were aligned along the sequence of the GIP substrate to show the potential scissile bonds (Figure 8). Besides, the peptide bonds preferentially cut by RmuAP1_CD could be determined according to the observed frequency of every scissile bond as described in Materials and Methods. To the 26-mer peptide, the most prominent scissile bond was P–Q–Q–↓–P–Q, accounting for 62% of the total cuts, followed by F–L–Q–↓–P–Q, which accounted for another 19% of cuts. To the 33-mer peptide, the primary scissile bond was P–Q–L–↓–P–Y, accounting for 37% of the total cuts, followed by L–Q–L–↓–Q–P, which represented another 14% of cuts. Apparently, glutamine was the most favorable residue at the P1 position in the 26-mer peptide, whereas leucine was the favorable residue at the P1 position in the 33-mer peptide. This analysis also suggests that proline was an acceptable residue at the P1′ position. Overall, the cleavage patterns suggest the removal of epitopes presented on the two GIPs by RmuAP1_CD.

Figure 8.

Cleaved products of GIPs by RmuAP1_CD. The reactions were performed at pH 3.0 for 3 h, and the identity of the digested products was analyzed by LC–MS/MS as described in Materials and Methods. The amino acid sequences of the 26-mer (a) and 33-mer (b) are shown, and the epitopes critical for the development of celiac disease are underlined in colors. The detected fragments, totaling 933 and 992 from the 26- and 33-mer peptides, respectively, are aligned with the sequences of the substrate peptides from the N to C termini. In case of ambiguous assignments due to the repetitive sequence along the peptide substrate, the degraded peptide was tentatively aligned to the position proximate to the N-terminus. The frequency (%) of each specific cut is indicated in the table on the right side.

The ability of RmuAP1_CD to digest GIPs at sites with proline at the P3 and P1′ positions prompted us to ask whether gelatin, which is rich in proline, could also be digested. To explore this, a zymographic test, incorporating gelatin into neutral SDS-PAGE, was conducted. The results confirmed that gelatin could be digested by RmuAP1_CD rather than by pepsin (Figure S2).

Molecular Docking

A molecular docking approach was used to predict binding patterns for the peptidyl ligand PQQPQ in the catalytic clefts of both RmuAP1_CD and pepsin. The interaction between PQQPQ and RmuAP1_CD may involve as many as six hydrogen bonds, associated with the flap residues G74 and D75, as well as body residues Y188, G215, and A293 (Figure 9). In contrast, pepsin forms only two hydrogen bonds, contributed by the flap residue G76. The residues T77, Y189, G217, and V291 in pepsin, spatially corresponding to residues D75, Y188, G215, and A293 in RmuAP1_CD, are situated outside effective ranges of distance and angle for the formation of hydrogen bonds.

Figure 9.

Molecular docking of the peptidyl ligand PQQPQ to RmuAP1_CD and pepsin. The residues that form hydrogen bonds to the ligand are shown in green sticks (C: green; O: red; and N: blue). The aspartic dyads are shown in magenta sticks (C: magenta; O: red; and N: blue). Residues located in the flap are labeled with an asterisk (*) and those involved in S1′ with a hash (#). Hydrogen bonds, assessed through LigPlot⁺, are presented as cyan dashed lines.

Discussion

In this study, RmuAP1 was identified as a gluten hydrolase. The BLAST analysis unveiled the presence of many homologous aspartate peptidases across various species within the Rhodotorula genus. Presumably, these aspartate peptidases are secreted and play a crucial role in extracting nutrients from protein substrates in the environment to support the growth of Rhodotorula spp. Despite the abundant predictions of Rhodotorula aspartate peptidases through in silico analysis, RmuAP1 is the first characterized Rhodotorula peptidase with the ability to break down 26- and 33-mer peptides under acidic conditions. When compared to pepsin, which lacks proteolytic activity toward GIPs, RmuAP1_CD was demonstrated as a promising oral peptidase to assist the complete digestion of gliadins in the stomach.

The ultimate aim of this research is to provide an effective oral peptidase for people who cannot tolerate gluten consumption. Therefore, in addition to the catalytic quality of RmuAP1_CD, it is equally crucial to develop an efficient and scalable production system for RmuAP1_CD. In this study, Y. lipolytica was chosen as the host for expressing RmuAP1_CD on the basis of the organism’s GRAS status and its highly effective secretory pathways.⁴⁴ The successful production of RmuAP1_CD by Y. lipolytica (Figure 3) and the ease of purification of RmuAP1_CD from the culture medium through IMAC (Figure 4) set the foundation for the future development of RmuAP1_CD as an oral therapeutic peptidase.

The activity-pH profiles of RmuAP1_CD differ between those of BSA and GIPs. RmuAP1_CD exhibited an optimum pH of 2.5 for the hydrolysis of BSA and remained at less than 5% activity when pH was ≥5.0 (Figure 5). In contrast, the optimal pH for GIPs hydrolysis shifted upward to around 4.0, with activity still evident at pH 6.0 (Figure 7). Consequently, it is intriguing to explore the underlying reasons for these distinct catalytic behaviors.

The proteolytic activity of pepsin on BSA within a pH range spanning from 1.0 to 5.5 was investigated decades ago.⁴⁵ The optimum pH for BSA was found to be about 1.7, and the activity dropped to 10% of its peak value as the pH increased to 4.0. Nonetheless, the action of pepsin on HCl-denatured BSA, obtained after a 10 min preincubation at pH 0.8, displayed a broad range of activity, peaking at pH 3.0 and maintaining 65% activity at pH 4.5. This finding, in conjunction with the observation that BSA underwent a conformational shift from its natural state to an expandable form as the pH decreased from neutral to around 4.5–3.5, led the authors to propose that pepsin’s inefficiency at pH values above 4.0 stemmed from the limited presence of BSA in its expandable form. In simpler terms, the rate-limiting step in pepsin’s overall action on BSA was the transformation of the substrate from its native state to unfolded forms under pH 5.0–7.0. Hence, the difference in the pH profile of RmuAP1_CD for BSA compared to that for GIPs might be due to the necessity of a critical conformational change in the catalytic process of BSA but not GIPs. After all, GIPs already existed in expandable conformations within the studied pH range.

The specificity of pepsin had been investigated for a period about half a century ago, suggesting a preference for hydrophobic residues at both the P1 and P1′ positions.⁴⁶ A more recent statistical analysis of the pepsin-catalyzed reactions toward a wide variety of protein substrates concluded that the preferential cleavage occurs after leucine and phenylalanine and before isoleucine.⁴⁷ Additionally, the presence of proline at positions P1 and P2 prohibits cleavage, and its occupation at P2′ also results in minimal cleavages.⁴⁷ Therefore, the poor ability of pepsin to digest GIPs may arise from its unfavorable interaction with proline residues. In this study, RmuAP1_CD appears to tolerate proline residues near the scissile bonds (Figure 8). The Cα backbone traces of the predicted 3D structure of RmuAP1_CD and porcine pepsin can be superimposed (Figure 2C). Therefore, dispositions in the subtle differences of the residues constituting the catalytic cleft may account for the discrepancy in recognizing and digesting GIPs. Although the authentic complex conformation of RmuAP1_CD and the peptidyl substrate segments in GIPs remains to be determined, the molecular docking models (Figure 9) support the notion that RmuAP1_CD has a binding ability stronger than that of pepsin to accommodate the pentapeptide PQQPQ.

Studies on pepsin’s structure and mutagenesis have indicated that the substrate specificity of pepsin is affected by nearest-neighbor hydrophobic residues surrounding the catalytic dyad and by residues in the flap.⁴⁸⁻⁵⁰ The sequence of RmuAP1_CD and porcine pepsin was aligned on the structural base (Figure 2a). The comparison reveals several differing residues constituting the S1, S2, S3, and S1′ subsites. These residues in RmuAP1_CD include but are not limited to D27 (S1), D75 (S1), S80 (S3), S113 (S3), L118 (S1), A212 (S1′), V220 (S2), and V285 (S2) (residue numbering is based on the sequence listed in Figure 2a). Further investigation is warranted to explore the role of these residues in substrate selection.

When compared to previously reported gluten hydrolases, RmuAP1_CD demonstrates noteworthy advantages. It exhibits the capacity to enzymatically break down both 26- and 33-mer peptides within the pH range of 3.0 to 6.0 (Figure 7), aligning closely with the postprandial pH fluctuations in the human stomach, which typically span from 6.0 to 2.0.²⁸ This underscores its promising role as an oral peptidase for individuals with CeD. However, it is essential to acknowledge the limitations associated with RmuAP1_CD. Notably, its enzymatic activity toward the degradation of GIPs, particularly the 33-mer peptide, is diminished when the pH falls below 3.0. Addressing this limitation is imperative for realizing the full potential of RmuAP1_CD in the context of CeD treatment. In conclusion, this study sets a crucial foundation for future research aiming to enhance the catalytic efficiency of RmuAP1_CD on GIPs at extremely acidic pH through directed evolution and rational design approaches.

Glossary

Abbreviations Used

CeD

celiac disease

GFD

gluten-free diet

GIPs

gluten-derived immunogenic peptides

NCGS

nonceliac gluten sensitivity

BSA

bovine serum albumin

LC–MS/MS

LC–tandem mass spectrometry

RmuAP1_CD

catalytic domain of Rhodotorula mucilaginosa aspartic protease 1

EDTA

ethylenediaminetetraacetic acid

PMSF

phenylmethylsulfonyl fluoride

β-ME

β-mercaptoethanol

SDS

sodium dodecyl sulfate

ITS

internal transcribed spacer

GRAS

generally recognized as safe

IMAC

immobilized metal affinity chromatography

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jafc.3c04750.

• Cleavaged peptides by RmuAP1 (XLSX)

• Methods and the detailed results of radius measurement of the selected colonies, nucleotide and amino acids sequences of RmuAP1, and zymographic analysis for the proteolytic activity of RmuAP1_CD on gelatin (PDF)

Author Contributions

Yuhan Zhang: data acquisition; investigation; validation; writing—original draft preparation; and software. Wei-Ming Leu: conceptualizations; writing—review and editing. Menghsiao Meng: funding acquisition; project administration; supervision; and writing—review and editing.

This work was supported by the Ministry of Science and Technology, Taiwan, ROC [grant number 109-2313-B-005-018-MY3].

The authors declare no competing financial interest.