Sequence and Structural Analysis of human gut microbial Prolyl Oligopeptidases (POPs): towards design of therapeutics for Celiac Disease

doi:10.21203/rs.3.rs-3972746/v1

Download PDF

Article

Sequence and Structural Analysis of human gut microbial Prolyl Oligopeptidases (POPs): towards design of therapeutics for Celiac Disease

https://doi.org/10.21203/rs.3.rs-3972746/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 23 Aug, 2024

Read the published version in Scientific Reports →

You are reading this latest preprint version

Background

Celiac Disease (CD) is a common autoimmune disorder where the patients are unable to digest gluten, which is present in foods made up of wheat, barley and rye. Whilst diagnosis happens late in 80% of the cases, avoidance of such foods appears to be the common solution. Alternative management strategies are required for the patients and their families since CD is also genetically carried over. Probiotic solutions and the consumption of appropriate enzymes, such as prolyloligopeptidases (POPs), from gut-friendly bacteria could reduce the disease burden and provide a better lifestyle for CD patients.

Results

We have examined around 5,000 gut bacterial genomes and identified nearly 4000 non-redundant putative POPs. A select set of 10 gut bacterial POP sequences were subject to three-dimensional modelling, ligand docking and molecular dynamics simulations where stable interactions were observed between the POPs and gluten peptides.

Conclusions

Our study provides sequence and structural analysis of potential POP enzymes in gut bacterial genomes, which form a strong basis to offer probiotic solutions to CD patients. In particular, these enzymes could be lead future therapeutics for this disease.

Biological sciences/Biophysics

Biological sciences/Computational biology and bioinformatics

Health sciences/Gastroenterology

S9 protein family

Domain Architecture

Prolyl Oligopeptidase

Oligopeptidase B

Celiac Disease

Gluten Intolerance

Serine proteases are enzymes that cleave peptide bonds in proteins, where serine serves as the nucleophilic amino acid at the active site(1). Prolyl Oligopeptidases (POPs) are a type of serine proteases which specifically cleave Pro-X peptide bonds and are widely distributed in all the kingdoms of life (2, 3). This enzyme family differs from trypsin-subtilisin-like serine proteases in their catalytic triad arrangement and specificities for small peptide substrates. POPs hydrolyse peptides consisting of less than 30 residues from the carboxy-terminal of Proline residues(2, 3).

Recent studies showed that POPs have an essential role in the treatment of diseases such as Alzheimer's, amnesia, depression, cancer and celiac disease(4, 5).

Bacterial POPs are pharmaceutically necessary enzymes, and their functional and evolutionary details are not fully studied. POP family proteins are typically around 700 residues in length in most species(4). Studies on crystal structure bacterial POPs from Myxococcus xanthus, Sphingomonas capsulata, and Aeromonas punctata have revealed a two-domain architecture(6, 7). The crystal structures suggest that bacterial POPs consist of a catalytic α/β hydrolase domain and a β-propeller domain. The α/β hydrolase domain in POPs is characterized by a short helical N-terminal region (~ 70 residues) and a larger C-terminal region that encompasses the catalytic triad. This catalytic region is less conserved within other α/β hydrolases in terms of its sequence but is structurally highly superimposable. The propeller domain is based on a radially arranged seven-fold repeat of four-stranded antiparallel β sheets. In the case of POPs, this domain is considered to be of the “open-velcro” topology, where the first and seventh blades are connected only through hydrophobic interactions. Notably, the catalytic triad composed of Ser, Asp, and His residues is concealed at the interface between the two structural domains(3) (Additional File 1). POPs are classified into the S9 family according to the MEROPS(8) database. The S9 family is divided into four subcategories. Different members have different substrate specificity. The active site residues are in the order Ser, Asp, and His in sequence. The motifs around Ser at the active site are characteristic for each subfamily: GGSXGGLL, where X is typically Asn or Ala (S9A), GWSYGGY (S9B), GGSYGG (S9C) and GGHSYGAFMT (S9D)(9). POPs cleave all naturally occurring peptides, suggesting their regulatory role towards neuronal peptides and hormones containing Proline(6). Due to their substrate specificity for proteolysis and their implications in human diseases, bacterial pathogenesis, and detoxification, these are important targets in pharmaceutical industries.

POPs and Celiac Disease (CD)

CD is a chronic health condition and an autoimmune disorder which afflicts 1% of humans in the world and six to eight million Indian populations. This condition results from an uncontrolled immune reaction to wheat gluten, along with similar proteins found in rye and barley. This disease occurs in people with human leukocyte antigen (HLA)-DQ2 or DQ8 alleles, which trigger immunogenic response to gluten. When consumed, gluten proteins are broken down into peptides by proteases within the gastrointestinal tract. The enzyme tissue transglutaminase(tTG) then modifies these peptides through deamidation, converting glutamine residues to glutamic acid(10, 11). Consequently, these altered peptides attach to HLA-DQ2 or -DQ8 molecules, triggering T-cell responses that induce inflammation in the small intestine, damaging the intestinal lining. This inflammatory process ultimately gives rise to the symptoms of CD, including diarrhoea and malnutrition, and requires lifelong treatment(12). CD could also co-occur with other autoimmune conditions and can lead to neurological disorders and cancer(4).

The immune response in celiac disease is triggered by the association of HLA with proline-rich peptides(prolamins) in gluten. The most prominent serotype is HLA-DQ2.5 (DQA1*05:01). Research has shown that DQ2.5 has a broader capacity to bind gluten peptides. Additionally, stable complexes are formed between gluten T-cell epitopes and DQ2.5, contributing to the heightened risk associated with DQ2.5(13, 14). Gluten consists of two main components, Gliadins and glutenins. Gliadin exists in three main types of gliadin (α, γ, and ω), which triggers the immune response in celiac disease(15). Epitopes derived from these subtypes are similar in terms of proline content.

Gluten peptides, being rich in prolines, cannot be readily cleaved by gastric enzymes such as trypsin and chymotrypsin. An approach to address the effects of CD is the effective degradation of immunoreactive peptides (released from glutens) using microbial enzymes such as prolyl oligopeptidases, which have been shown to cleave proline-rich peptides. Several in vitro and in vivo studies have shown that supplementation of Proline or Glutamine-specific endoproteases in the diet significantly enhances the ability of these endogenous proteases to detoxify dietary gluten(6, 7).

It has been shown that prolyl oligopeptidase from Flavobacterium meningosepticum (FM-POP) is capable of breaking down gluten, which would otherwise trigger the immune responses in vitro(16). Prolyl oligopeptidases from Sphingomonas capsulate and Myxococcus xanthus were also studied and have comparable properties(6, 7). The POP from Sphingomonas capsulata is a potential oral therapeutic candidate due to its stability under acidic conditions.

Although the crystal structures of some bacterial POPs are known, the structure and distribution of POPs in gut microbiome are largely unknown. Here, we emphasise that gut bacterial species, which are/can be used as probiotics, could be used as therapeutics for celiac disease. Hence, we have examined the distribution of putative POPs within genomes of bacteria, which are abundant in gut microbiome, using computational techniques and mathematical models. In addition, we have examined the structural compatibility of select gut bacterial POPs to serve as therapeutic agents for CD through three-dimensional modelling and docking of gluten peptides.

Data collection

The human gut microbiome data were collected and assembled from six online repositories. We have extensively collected, filtered, and assembled data from these six databases (please see Methods for details) to identify POP homologues.

Pipeline

The methodology is illustrated in Fig. 1 (Individual sequences taken at each stage are given in Additional File 2). All genomes accessible from the aforementioned six studies were incorporated into the analysis. The resulting non-redundant filtered dataset encompasses a comprehensive compilation exceeding 5,000 genomes. This includes diverse strains co-existing within the human gut for specific species. Subsequent to rigorous data mining and meticulous refinement processes (please see Methods for our sequence searches), a subset of 4,400 sequences was selected. By subsequently applying a filter to target sequences featuring the two-domain architecture of propeller and catalytic domains (POP), 734 sequences were identified for subsequent in-depth investigation. Ultimately, ten sequences bearing potential probiotic significance (as evidenced in the literature) were chosen for structure modelling. The protein which served as the template (PDB code 3IVM)(17) is from the bacterium Aeromonas punctata in the ligand-bound “closed” form of the protein. Further, a refined set of six sequences was selected based on the binding affinity with the ligands and the probiotic nature of the bacterial species of origin. These six proteins were further subjected to a thorough evaluation concerning their therapeutic relevance in the context of celiac disease using Molecular Dynamics simulations.

The selected sequences are either probiotics or show probiotic potential

Probiotic products typically take the form of oral supplements or food-based products containing microorganisms, typically bacteria. We hypothesize that a gut bacterium could be regarded as a potential therapeutic in this context if it is generally used as a probiotic and also retain genes which have sequence signature of POP domains. These criteria could suggest that such enzymes are capable of degrading the harmful celiac disease-containing peptides. We evaluate the effectiveness of six potential bacterial sequences (Table 1) to degrade the known celiac disease epitopes(18)(Fig.6C). The most commonly used strain for gram-negative probiotics is Escherichia coli Nissle 1917(19). Escherichia coli is also one of the common bacteria of the human gut Microbiome. We have evaluated the sequence WP_021564606 from E.coli.

Bacteria within Prevotella species are one of the most important members of the gut microbiome, especially in Indian populations(20). Studies show that the genomes of Prevotella species recovered from human gut metagenomes show that the most prevalent and relatively abundant species are primarily related to P. copri and P. stercorea. These species possess distinct gene repertoires likely reflecting adaptations to metabolic niches(21).

We could not find any genes with potential POP domain in any of the strains of Prevotella copri through our sequence searches, but genes coding for POP domains with specific two-domain architecture were found in P. stercorea and other Prevotella species.

Recent studies also suggest that Prevotella species could be used as next-generation probiotics(22).

We have evaluated the sequence WP_040596915 from Prevotella Stercorea and WP_081457517 from Prevotella salivae for our analysis.

Growing attention is being directed towards Faecalibacterium prausnitzii, an abundantly prevalent bacterial species within the gut. This heightened interest arises from its potential significance in fostering gastrointestinal well-being. This species is shown to have anti-inflammatory properties as it can induce a tolerogenic cytokine profile(23).

Research shows that Faecalibacterium prausnitzii could be used as a next generation probiotic in gut disease improvement(24). Hence, we have modelled the sequence VDR34335 of Faecalibacterium prausnitzii, corresponding to the POP domain, for our analysis. Clostridium species are a predominant cluster of commensal bacteria in our gut, and they exert considerable effects on our intestinal homeostasis. Clostridium cadaveris is usually considered non-pathogenic and does not produce toxins, unlike other species of Clostridium(25). Likewise, we have modelled the POP domain in one sequence (WP_035771429) from Clostridium cadaveris for our analysis.

The commercial Bacillus probiotic strains in use are B. cereus, B. clausii, B. coagulans, B. licheniformis, B. polyfermenticus, B. pumilus, and B. subtilis. These probiotic bacteria have demonstrated their ability to significantly improve digestive health, and they even have a role in preventive healthcare with their ability to lower lipid levels and improve immune response(4).

Almost all the POP sequences from Bacillus strains found in our analysis from gut microbiome consists of only the catalytic domain. However, we could discover POP sequence from B. subtilis, which contain the two-domain architecture and hence can be considered for protein modelling and subsequent study. This sequence belongs to S9A, and it is an oligopeptidase B, identified by the conserved sequence motif (GGSXGGLL) around the active site serine. Oligopeptidase B also has a similar domain architecture as POP but differs in the motif around the active site (as previously mentioned in the introduction). Studies show that Oligopeptidase B can also cleave proline-rich peptides(26). Considering the widespread use of Bacillus strains as probiotics, we have included the following oligopeptidase B sequence from B. subtilis for docking and molecular dynamics study. Although the strain from which the sequence is selected is not found in the above six studies, other studies show that B. subtilis is a part of the gut microbiome(27). We have included the POP domain in the sequence MCF7751424 from B. subtilis for our analysis.

Table 1: The selected species and the POP sequences. Functionally important residues have been marked.

Species	Sequence_ID	Tyr	Ser	Trp	Asp	Arg	His
Prevotella stercorea	WP_040596915	479	559	600	643	645	678
Clostridium cadavaris	WP_035771429	450	530	571	614	616	649
E.coli	WP_021564606	463	544	585	627	629	660
Alistepis Sahii	WP_149885823	479	559	599	642	644	677
Faecalibacterium prausnitzii	VDR34335	456	536	577	619	621	654
Bacilus Subtilus	MCF7751424	475	555	599(E)	640	642(Q)	675

Distribution of POP family (S9 family proteins) in gut microbiota across bacterial phyla

Fig. 2 shows the distribution of POPs that could be identified in the gut microbiome as present in our dataset. Studies have shown that gut microbiome consists of bacteria from four predominant phyla - namely Bacteroidetes, Actinobacteria, Firmicutes and Proteobacteria(28). Our study also reports similar results (Figure 2 and Additional file 3). Along with the above mentioned four phyla, bacteria from two other phyla (namely Fusobacteria and Verrucomicrobia) were also found to contain POP domain. However, no genes of the POP family (featuring the distinctive dual-domain architecture) were detected within these two phyla. Consequently, only genes coding for POP originating from the dominant four phyla were included in the analysis presented in Fig 2. We observe that nearly 50% of the identified POP homologous sequences are from phylum Bacteroidetes. Proteobacteria and Firmicutes constitute almost the same number of POP family genes (~20%). Finally, the phylum Actinobacteria has around (13%) of POP family genes (Fig.2A). Fig. 2B represents the distribution of bacterial phyla when data is segregated per species. We notice most genomes from Firmicutes have a single gene coding for a protein belonging to the POP family. In contrast, for the phyla Actinobacteria and Proteobacteria, the number varies from 1 to 3. Finally, the phyla Bacteroidetes have many genomes that encode for multiple POPs, where the number of genes varies from 1 to 10; however, most of the genomes were observed to have more than four POP genes (Fig. 2B).

Gut Microbiome genes containing POP domains show Diverse Domain Architecture

Diverse domain architectures were found in the genes of the gut microbiome that contain the sequence signature of POP domains (Fig. 3). The most common one constitutes sequences having only a catalytic domain. Sequences having domain architecture DPP_IV and catalytic domain (DPPIV_N-Peptidase_S9), as well as sequences having domain architecture as propeller domain and catalytic domain (Peptidase_S9_N-Peptidase_S9) are also common. Other frequently occurring domains are alpha-beta hydrolase 2, PD40, and esterase.

Alpha-beta hydrolase and esterase are protease sequences which share remote homology with S9 family proteins or act as intermediates. Hence, it is not uncommon that some of these domains are found in S9 family peptidases. In addition, many other domains were also present in very low frequency (present in one or two proteins) (details in Additional File 4). However, we chose candidate POPs which contain the classic two-domain architecture (N-terminal propeller domain and catalytic domain) for structural studies.

Modelled Structure of POP domain shows Stable Conformation

The structure of POP from the bacteria Aeromonas caviae is known in both open and closed conformation. This domain (closed form: PDB ID 3IVM) showed nearly 30 to 40% identity to the selected sequences and was hence considered as a template for structure modelling of the sequences. The alignment of the query and template shows the preservation of conserved residues, along with the catalytic triad, which is present towards the C-terminal of the sequence in the catalytic domain. There was good agreement between the secondary structural elements. The structures were modelled using Modeller(29,30). For each sequence, 100 different models were generated, and the model which showed the lowest DOPE score was selected for further analysis. The modelled structures were evaluated by Ramachandran plot analysis using PROCHECK(31). More than 98% of residues were in the core allowed region of the Ramachandran plot, indicating the better quality of models, while few residues were in the disallowed region (Fig. 4). A representative model from Prevotella stercorea is shown in Fig. 4B. The modelled catalytic domain is shown in blue while the modelled propeller domain is shown in red. The superimposed structure of the model with the template structure (Fig 4A) shows an RMSD value of 0.18, which indicates that the modelled structure is quite similar to the template structure. The per residue energy of the modelled structure is negative (Fig 4C), and the Ramachandran plot (Fig 4D) shows only six residues in the disallowed region. The statistics of all the other modelled structures are available in Additional File 5.

Inhibitor and substrate docking

Four of the potent celiac disease-causing epitopes (Proline containing sequence derived from Gluten subtype α and γ) which are recognised by HLA DQ2.5 were considered for peptide docking analysis.

Z-Pro-Prolinal (ZPR) (detailed structure in additional file 6) is known to be a natural inhibitor for POPs. We first analysed the crystal structure of the 3IVM (ligand (ZPR) bound form of POP). Apart from the active site residue, the closed form is shown to bind to three other stabilizing residues (Tyr 458, Trp 579 and Arg 624). It is known from the literature that Trp 579 is the critical residue that packs with the Proline ring of the substrate, and Arg 624 is known to be involved in hydrogen bonding to the substrate(3). In the docking analysis, only those docking poses which show interaction with the catalytic site and the stabilizing residues were considered. ZPR and the four most potent celiac disease-causing peptides were considered for docking analysis. Due to the presence of slight insertions and deletions within the analysed POP sequences, there is no exact correspondence between the residue numbers of sequences and the crystal structure residues. To address this disparity, equivalent residues were determined through a multiple sequence alignment encompassing all selected sequences and the template (Fig. 5 and Table 1). Notably, all active site residues and stabilizing residues exhibit 100% conservation across the POP sequences. The only exception is in an Oligopeptidase B in Bacillus subtilis, where Arginine and Tryptophan are not conserved, although the catalytic triad remains preserved.

The docking of all the selected sequences with above mentioned celiac disease-causing epitopes were performed using Schrodinger software (for details, please refer to Methods section). A representative 2D-docking diagram is shown in Fig. 6A, along with the hydrogen bonding patterns with the critical residues Arg, His, Tyr and Ser. The 3D-model of protein-peptide docking is shown in Fig. 6B. The heatmaps for docking score for all peptides and sequences are shown in Fig. 6C. Bacteria in Prevotella and Faecuilibacterium show the lowest average binding energy, considering all the peptides. All the peptides show similar average binding energy for all the sequences.

Molecular Dynamics Simulation shows Stable Binding of Peptides

Next, MD simulation was run for 100 ns on all six protein-peptide complexes in order to assess the stability of docking complexes. The peptide PQPELPYPQ, which is a part of the gluten, was selected as a representative peptide for MD analysis. A representative MD run is shown for a Prevotella POP in Fig 7. The RMSD plot shows the stability of the complex over the 100 ns simulation period (Fig 7A). It was also observed that the simulation converges, and the RMSD values stabilize around 20ns and do not fluctuate much for a duration of 100ns. Ligfit prot (Fig 7A) indicated that the RMSD of the protein-ligand complex remains stable and is comparable to the protein RMSD, and hence, it indicates a stable protein-ligand complex for the whole duration of 100 ns.

The major interacting residues which form H-bonds with the ligand and remain for more than 50% of the simulation time are Lys 150, Lys 197, Asp 273, Phe 482, Asn 699 and Trp 600 (Fig 7B, 7C). Trp 600 is one of the essential residues known to bind the ligand. Another essential residue mentioned earlier was Arg 645. A hydrogen bond with Arg 645 was observed, but not for more than 50% of the simulation time. However, Arg 645 was found to be generally interacting for almost 90% of the simulation time, including hydrophobic interactions. Few other residues interact through water bridges or hydrophobic interaction as well (Fig. 7C). At each point during simulation time, the protein remains in contact with the ligand (Fig 7D), which suggests the stability of the complex.

MD simulation for 100 ns was run for other protein-peptide sequences, and the complexes were found to be stable for the entire simulation time (Additional File 7). The equivalent residues of Tyr 479, Trp 600 and Arg 645 are found to be interacting in other selected POPs as well (details in Additional file 7). As mentioned earlier, these peptides are the actual epitopes which act as T-cell epitopes, which eventually cause CD.

In our analysis, we observed that POPs from the above-mentioned sequences are able to bind to these epitopes. Additionally, our molecular dynamics simulation study indicates the stability of these interactions. Hence, these POP sequences could be potentially used to degrade CD-causing peptides. Additionally, since all of the selected sequences are from probiotics/gut bacteria, it would be easier to design experiments and should be safe for human consumption. Consequently, we propose that these POPs from these select probiotic bacteria could potentially act as therapeutics and offer relief from the symptoms of CD.

MD Simulation of Prevotella in Acidic Environment

One requirement for a candidate for effective therapeutics for CD is that the protein should be able to show activity in both the acidic environment of the stomach and at neutral pH in the intestine. It is absolutely important that a candidate therapeutic for CD should be stable in both acidic (in the stomach) and in neutral pH (in the intestine). Therefore, we next evaluated the stability in an acidic environment and chose the POP identified from Prevotella. The MD simulation was redone at 2.5 pH, simulating the acidic environment of the stomach for 200ns. The RMSD diagram reveals that the protein-peptide complex is stable for the whole duration of 200ns (Fig 8A). In addition, the percentage conservation of secondary structure shows that most of them remain stable during the simulation time(Fig 8B). The protein-ligand interacting graph (Fig 8C and 8D) shows that several important residues, such as Arg 645, Tyr 479, and Ser 559(Active site residue), retain hydrogen bond interaction and are in contact with the ligand for a significant amount of time (40-90% of the time).

In this study, we have performed an exhaustive sequence search for POP family genes within the proteomes of the human gut microbiota. Our study revealed that POP family genes are present in all bacterial phyla of the human gut and show diverse domain architectures. Most of the genomes have one or two POP genes, while there is a significant representation of genomes from the phyla Bacteroidetes, and they often contain four or five POP genes. We have evaluated the therapeutical properties associated with the POPs for the treatment of CD. We have selected six sequences from the gut bacterial genomes, which are also probiotics and retained good interactions with the peptide (as shown by docking scores). We demonstrated their potential by employing rigorous docking CD-causing peptides followed by molecular dynamics analyses. These selected sequences fulfil many criteria for an effective oral enzyme therapy for CD. First, they have strong signatures to retain POP domains even if they are barely characterised. They are stable under gastro-intestinal conditions. They are also able to bind to the epitopes derived from the gluten peptides, which otherwise activate the T cells and cause celiac disease. Finally, as these sequences are derived from human gut bacteria which are also already established probiotics or next generation probiotics, these could be regarded as safer for human consumption.

This research contributes valuable insights as well as lays the foundation for further investigation and experimental analyses. This could prove instrumental in the development of novel therapeutics for celiac disease.

Identification of POP homologues

POP homologues were identified directly by searching for annotation in the cumulative set of genomes from six databases (as mentioned earlier). CD-hit (100%)(32)identity was used to make the unique set of sequences. The six databases used to make the non-redundant sequences are:

The BIO-ML(33) is a repository of human gut bacteria. This repository comprises 7758 bacterial isolates, each accompanied by 3632 paired genome sequences. This comprises the majority of phylogenetic diversities observed within the human gut.

2. The HMP(Human Microbiome Project)(34) is a collection of projects collecting and annotating the genomes of microbiota found in humans. The subset of these genomes, which correspond to the microbiota which were present in the gut, was taken into consideration for our analysis.

3. The Unified Human Gastrointestinal Genome (UHGG)(35) is a collection of genomes which considered various studies and comprises 204,938 genomes from 4,644 gut prokaryotes. The non-redundant genomes which passed the quality control and are available in ENA accession ERP116715 were considered.

4. HumGut(36) is a comprehensive collection of human gut prokaryotic genomes filtered by metagenome data. It comprises data from Refseq(37) and UHGG(35) collection. This serves as an augmentation to UHGG data.

5. The Culturable Genome Reference (CGR)(38) is a collection of 1,520 non-redundant, high-quality draft genomes generated from more than 6,000 bacteria cultivated from faecal samples of healthy humans.

6. The database of ‘unculturable’ human microbiota built by Browne et al. under the NCBI accession ID PRJEB10915 was also taken for the database generation. It consists of 137 distinct bacterial species, 45 of which are potentially novel species.

Due to the possibility of repetition of genome data across these studies, we have considered the superset removing the redundancies. For the above studies, the final set of genomes which were available for download were considered. (Additional file 8)

Validation of identified POP homologues

HMMER(39) is a software suite that uses Hidden Markov Models(HMMs) to perform homology searches, helping identify evolutionarily related sequences to annotate and identify domains within a given protein sequence. It does so by building profile HMMs for consensus protein sequences which can then be used to infer and annotate the domains in a novel protein sequence. All the annotated POP homologues identified were further validated using HMMSCAN, which is a part of the HMMER suite. HMMSCAN identifies domains in the search sequence against a database of HMMs of known domains, and we have searched against the PFAM(40) database. Only those sequences which have a catalytic domain (S9_catalytic) are selected for further analysis.

3D Modelling of peptides

The CD-causing epitopes are small and around nine amino acids in length. These sequences are known, but for molecular dynamics and docking, a 3D model is required. PEP- FOLD(41) online server was used to predict the 3D model of peptides. PEP-FOLD is a de novo approach aimed at predicting peptide structures from amino acid sequences.

Structure modelling

The structures of selected POPs were modelled using MODELLER(29,30) which is a homology-based structure modelling tool. The template for homology modelling was selected using BLAST against PDB, and PSIPRED(42) (Protein Structure Prediction Server) was used for predicting secondary structural elements. A set of 10 models were generated, from which the lowest energy structure according to DOPE (Discrete optimized protein energy)(43) score was used for further processes. DOPE (Discrete Optimized Protein Energy) method is a statistical potential optimized for model assessment. Ramachandran plot analysis was done using PROCHECK(31). PROSA(44) software was used to further validate the structure.

Molecular Docking

For molecular docking simulations, we employed two distinct software tools, AutoDock and Glide. In AutoDock(45), we utilized AutoDockTools (ADT) for the preparation of both the protein and ligand. The docking grid was strategically positioned to encompass the entire protein structure, while the docking parameters were configured with 100 Genetic Algorithm (GA) runs employing the Lamarckian genetic algorithm.

In the case of Glide(46), which is a part of the Schrodinger suite, we prepared the modelled structures for docking using the Protein Wizard module, which handled essential tasks such as protonation, hydrogen bond optimization, and restrained minimization. A docking grid was defined, spanning a 30 Angstrom box centred around the six conserved residues of interest. To estimate the binding affinities, we employed the Prime MM-GBSA (Molecular Mechanics/Generalized Born Surface Area) tool within the Schrodinger suite, which calculated ΔG (Gibbs free energy) values for each docked pose, allowing us to rank and prioritize potential binding interactions.

Molecular Dynamics Simulation

To validate the stability of our protein-ligand complexes, we conducted Molecular Dynamics (MD) simulations using the Desmond(47) package within the Schrodinger suite. These simulations were executed over a duration of 100-200 nanoseconds. Additionally, we utilized Simulation Interaction Diagrams (SIDs) to analyse and interpret the results obtained from the0 MD simulations, shedding light on the critical interactions and conformational changes througho0ut the simulation trajectories. Simulations were performed at two different pH (7.6 and 2.5) using a tool called PROPKA(48), in the Schrödinger suite.

POP – Prolyl Oligopeptidase ,OPB – Oligopeptidase B, DPP IV – Dipeptidyl peptidase-4

HMM – Hidden Markov Model, DA – Domain Architecture, CD – Celiac Disease

Acknowledgements

The authors would like to acknowledge NCBS (TIFR), SERB, IBAB and DBT for infrastructural and other support.

Funding

This work was supported by PhD fellowship from Department of Biotechnology (DBT). Authors thank NCBS for infrastructural facilities. RS is a J.C. Bose National Fellow (JBR/2021/000006) from the Science and Engineering Research Board, India. RS would also like to thank Bioinformatics Centre Grant funded by the Department of Biotechnology, India (BT/PR40187/BTIS/137/9/2021) and the Institute of Bioinformatics and Applied Biotechnology for the funding through her Mazumdar-Shaw Chair in Computational Biology (IBAB/MSCB/182/2022).

Author's contributions

RS designed the experiments and conceived the idea. DR, SN performed the experiments, analysed the data and wrote first draft of the manuscript and RS, DR and SN improved it.

Availability of data and materials

All the data are available in supplementary material.

Competing interests

The authors declare that they have no competing interests.

Consent for publication

Not applicable

Ethics approval and consent to participate

Not applicable

Hedstrom L. Serine Protease Mechanism and Specificity. Chem Rev. 2002 Dec 1;102(12):4501–24.
Polgár L. The prolyl oligopeptidase family. Cell Mol Life Sci CMLS. 2002 Feb;59(2):349–62.
Fülöp V, Böcskei Z, Polgár L. Prolyl Oligopeptidase. Cell. 1998 Jul;94(2):161–70.
Penttinen A, Tenorio-Laranga J, Siikanen A, Morawski M, Roßner S, Arturo Garcia-Horsman J. Prolyl Oligopeptidase: A Rising Star on the Stage of Neuroinflammation Research. CNS Neurol Disord - Drug Targets. 2011 May 1;10(3):340–8.
Shan L, Marti T, Sollid LM, Gray GM, Khosla C. Comparative biochemical analysis of three bacterial prolyl endopeptidases: implications for coeliac sprue. Biochem J. 2004 Oct 15;383(2):311–8.
Gass J, Ehren J, Strohmeier G, Isaacs I, Khosla C. Fermentation, purification, formulation, and pharmacological evaluation of a prolyl endopeptidase from Myxococcus xanthus : Implications for Celiac Sprue therapy. Biotechnol Bioeng. 2005 Dec 20;92(6):674–84.
Xiao B, Zhang C, Song X, Wu M, Mao J, Yu R, et al. Rationally engineered prolyl endopeptidases from Sphingomonas capsulata with improved hydrolytic activity towards pathogenic peptides of celiac diseases. Eur J Med Chem. 2020 Sep;202:112499.
Rawlings ND, Barrett AJ, Thomas PD, Huang X, Bateman A, Finn RD. The MEROPS database of proteolytic enzymes, their substrates and inhibitors in 2017 and a comparison with peptidases in the PANTHER database. Nucleic Acids Res. 2018 Jan 4;46(D1):D624–32.
MEROPS - the Peptidase Database [Internet]. [cited 2024 Jan 3]. Available from: https://www.ebi.ac.uk/merops/cgi-bin/famsum?family=S9
van de Wal Y, Kooy Y, van Veelen P, Peña S, Mearin L, Papadopoulos G, et al. Selective deamidation by tissue transglutaminase strongly enhances gliadin-specific T cell reactivity. J Immunol Baltim Md 1950. 1998 Aug 15;161(4):1585–8.
Molberg Ø, Mcadam SN, Körner R, Quarsten H, Kristiansen C, Madsen L, et al. Tissue transglutaminase selectively modifies gliadin peptides that are recognized by gut-derived T cells in celiac disease. Nat Med. 1998 Jun;4(6):713–7.
Dunne MR, Byrne G, Chirdo FG, Feighery C. Coeliac Disease Pathogenesis: The Uncertainties of a Well-Known Immune Mediated Disorder. Front Immunol. 2020 Jul 8;11:1374.
Sollid LM, Qiao SW, Anderson RP, Gianfrani C, Koning F. Nomenclature and listing of celiac disease relevant gluten T-cell epitopes restricted by HLA-DQ molecules. Immunogenetics. 2012 Jun;64(6):455–60.
Vader W, Stepniak D, Kooy Y, Mearin L, Thompson A, van Rood JJ, et al. The HLA-DQ2 gene dose effect in celiac disease is directly related to the magnitude and breadth of gluten-specific T cell responses. Proc Natl Acad Sci U S A. 2003 Oct 14;100(21):12390–5.
Ribeiro M, Nunes-Miranda JD, Branlard G, Carrillo JM, Rodriguez-Quijano M, Igrejas G. One Hundred Years of Grain Omics: Identifying the Glutens That Feed the World. J Proteome Res. 2013 Nov 1;12(11):4702–16.
Osorio CE, Wen N, Mejías JH, Mitchell S, Von Wettstein D, Rustgi S. Directed-Mutagenesis of Flavobacterium meningosepticum Prolyl-Oligopeptidase and a Glutamine-Specific Endopeptidase From Barley. Front Nutr. 2020 Feb 18;7:11.
Li M, Chen C, Davies DR, Chiu TK. Induced-fit Mechanism for Prolyl Endopeptidase. J Biol Chem. 2010 Jul;285(28):21487–95.
Sollid LM, Tye-Din JA, Qiao SW, Anderson RP, Gianfrani C, Koning F. Update 2020: nomenclature and listing of celiac disease–relevant gluten epitopes recognized by CD4+ T cells. Immunogenetics. 2020 Feb;72(1–2):85–8.
Behnsen J, Deriu E, Sassone-Corsi M, Raffatellu M. Probiotics: Properties, Examples, and Specific Applications. Cold Spring Harb Perspect Med. 2013 Mar 1;3(3):a010074–a010074.
Prasoodanan P. K. V, Sharma AK, Mahajan S, Dhakan DB, Maji A, Scaria J, et al. Western and non-western gut microbiomes reveal new roles of Prevotella in carbohydrate metabolism and mouth–gut axis. Npj Biofilms Microbiomes. 2021 Oct 7;7(1):77.
Yeoh YK, Sun Y, Ip LYT, Wang L, Chan FKL, Miao Y, et al. Prevotella species in the human gut is primarily comprised of Prevotella copri, Prevotella stercorea and related lineages. Sci Rep. 2022 May 31;12(1):9055.
Huang F, Sardari RRR, Jasilionis A, Böök O, Öste R, Rascón A, et al. Cultivation of the gut bacterium Prevotella copri DSM 18205 ^T using glucose and xylose as carbon sources. MicrobiologyOpen. 2021 Jun;10(3):e1213.
Lopez-Siles M, Duncan SH, Garcia-Gil LJ, Martinez-Medina M. Faecalibacterium prausnitzii: from microbiology to diagnostics and prognostics. ISME J. 2017 Apr;11(4):841–52.
He X, Zhao S, Li Y. Faecalibacterium prausnitzii: A Next-Generation Probiotic in Gut Disease Improvement. Chen T, editor. Can J Infect Dis Med Microbiol. 2021 Mar 5;2021:1–10.
Schade RP, Van Rijn M, Timmers HJLM, Dofferhoff ASM, Klaassen CHW, Meis JFGM. Clostridium cadaveris bacteraemia: Two cases and review. Scand J Infect Dis. 2006 Jan;38(1):59–62.
Mattiuzzo M, Gobba CD, Runti G, Mardirossian M, Bandiera A, Gennaro R, et al. Proteolytic Activity of Escherichia coli Oligopeptidase B Against Proline-Rich Antimicrobial Peptides. J Microbiol Biotechnol. 2014 Feb 28;24(2):160–7.
Hong HA, Khaneja R, Tam NMK, Cazzato A, Tan S, Urdaci M, et al. Bacillus subtilis isolated from the human gastrointestinal tract. Res Microbiol. 2009 Mar;160(2):134–43.
MetaHIT Consortium (additional members), Arumugam M, Raes J, Pelletier E, Le Paslier D, Yamada T, et al. Enterotypes of the human gut microbiome. Nature. 2011 May 12;473(7346):174–80.
Šali A, Blundell TL. Comparative Protein Modelling by Satisfaction of Spatial Restraints. J Mol Biol. 1993 Dec;234(3):779–815.
Webb B, Sali A. Protein Structure Modeling with MODELLER. In: Kaufmann M, Klinger C, Savelsbergh A, editors. Functional Genomics [Internet]. New York, NY: Springer New York; 2017 [cited 2024 Jan 2]. p. 39–54. (Methods in Molecular Biology; vol. 1654). Available from: http://link.springer.com/10.1007/978-1-4939-7231-9_4
Laskowski RA, MacArthur MW, Moss DS, Thornton JM. PROCHECK: a program to check the stereochemical quality of protein structures. J Appl Crystallogr. 1993 Apr 1;26(2):283–91.
Li W, Godzik A. Cd-hit: a fast program for clustering and comparing large sets of protein or nucleotide sequences. Bioinformatics. 2006 Jul 1;22(13):1658–9.
Poyet M, Groussin M, Gibbons SM, Avila-Pacheco J, Jiang X, Kearney SM, et al. A library of human gut bacterial isolates paired with longitudinal multiomics data enables mechanistic microbiome research. Nat Med. 2019 Sep;25(9):1442–52.
The Integrative HMP (iHMP) Research Network Consortium. The Integrative Human Microbiome Project. Nature. 2019 May;569(7758):641–8.
Almeida A, Nayfach S, Boland M, Strozzi F, Beracochea M, Shi ZJ, et al. A unified catalog of 204,938 reference genomes from the human gut microbiome. Nat Biotechnol. 2021 Jan;39(1):105–14.
Hiseni P, Rudi K, Wilson RC, Hegge FT, Snipen L. HumGut: a comprehensive human gut prokaryotic genomes collection filtered by metagenome data. Microbiome. 2021 Dec;9(1):165.
O’Leary NA, Wright MW, Brister JR, Ciufo S, Haddad D, McVeigh R, et al. Reference sequence (RefSeq) database at NCBI: current status, taxonomic expansion, and functional annotation. Nucleic Acids Res. 2016 Jan 4;44(D1):D733–45.
Zou Y, Xue W, Luo G, Deng Z, Qin P, Guo R, et al. 1,520 reference genomes from cultivated human gut bacteria enable functional microbiome analyses. Nat Biotechnol. 2019 Feb;37(2):179–85.
Eddy SR. A Probabilistic Model of Local Sequence Alignment That Simplifies Statistical Significance Estimation. Rost B, editor. PLoS Comput Biol. 2008 May 30;4(5):e1000069.
Mistry J, Chuguransky S, Williams L, Qureshi M, Salazar GA, Sonnhammer ELL, et al. Pfam: The protein families database in 2021. Nucleic Acids Res. 2021 Jan 8;49(D1):D412–9.
Maupetit J, Derreumaux P, Tuffery P. PEP-FOLD: an online resource for de novo peptide structure prediction. Nucleic Acids Res. 2009 Jul 1;37(Web Server):W498–503.
McGuffin LJ, Bryson K, Jones DT. The PSIPRED protein structure prediction server. Bioinformatics. 2000 Apr 1;16(4):404–5.
Eswar N, Eramian D, Webb B, Shen MY, Sali A. Protein Structure Modeling with MODELLER. In: Kobe B, Guss M, Huber T, editors. Structural Proteomics [Internet]. Totowa, NJ: Humana Press; 2008 [cited 2023 Dec 29]. p. 145–59. (Walker JM, editor. Methods in Molecular Biology; vol. 426). Available from: http://link.springer.com/10.1007/978-1-60327-058-8_8
Wiederstein M, Sippl MJ. ProSA-web: interactive web service for the recognition of errors in three-dimensional structures of proteins. Nucleic Acids Res. 2007 May 8;35(Web Server):W407–10.
Morris GM, Huey R, Lindstrom W, Sanner MF, Belew RK, Goodsell DS, et al. AutoDock4 and AutoDockTools4: Automated docking with selective receptor flexibility. J Comput Chem. 2009 Dec;30(16):2785–91.
Friesner RA, Banks JL, Murphy RB, Halgren TA, Klicic JJ, Mainz DT, et al. Glide: A New Approach for Rapid, Accurate Docking and Scoring. 1. Method and Assessment of Docking Accuracy. J Med Chem. 2004 Mar 1;47(7):1739–49.
Bowers KJ, Sacerdoti FD, Salmon JK, Shan Y, Shaw DE, Chow E, et al. Molecular dynamics---Scalable algorithms for molecular dynamics simulations on commodity clusters. In: Proceedings of the 2006 ACM/IEEE conference on Supercomputing - SC ’06 [Internet]. Tampa, Florida: ACM Press; 2006 [cited 2024 Jan 3]. p. 84. Available from: http://portal.acm.org/citation.cfm?doid=1188455.1188544
Bas DC, Rogers DM, Jensen JH. Very fast prediction and rationalization of p K _a values for protein–ligand complexes. Proteins Struct Funct Bioinforma. 2008 Nov 15;73(3):765–83.

No competing interests reported.

AdditionalFiles.zip

Download PDF

Journal Publication

published 23 Aug, 2024

Read the published version in Scientific Reports →

Editorial decision: Revision requested
20 Apr, 2024
Reviews received at journal
28 Mar, 2024
Reviews received at journal
24 Mar, 2024
Reviewers agreed at journal
17 Mar, 2024
Reviewers agreed at journal
17 Mar, 2024
Reviewers agreed at journal
15 Mar, 2024
Reviewers agreed at journal
15 Mar, 2024
Reviewers invited by journal
15 Mar, 2024
Editor assigned by journal
15 Mar, 2024
Editor invited by journal
02 Mar, 2024
Submission checks completed at journal
02 Mar, 2024
First submitted to journal
20 Feb, 2024

You are reading this latest preprint version

Sequence and Structural Analysis of human gut microbial Prolyl Oligopeptidases (POPs): towards design of therapeutics for Celiac Disease

Status:

Journal Publication

Version 1

Abstract

Figures

Introduction

Results and Discussion

Data collection

Pipeline

The selected sequences are either probiotics or show probiotic potential

Conclusion

Methods

Abbreviations

Declarations

References

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1