In-Silico Subcellular Localization and Functional Analysis of Computationally Predicted Virulent Proteins in X. Oryzae pv. Oryzae Strain PXO99A Causal Organism of Bacterial Leaf Blight (BLB) of Rice (O. Sativa L.)

Dharmendra Kashyap^* and Aafreen Khan
^*Correspondence: Dharmendra Kashyap, Department of Microbiology & Bioinformatics, Atal Bihari Vajpayee Vishwavidyalaya, Bilaspur University, Bilaspur, Chhattisgarh, India, Tel: 07752-220007, Email:

Keywords

O. Sativa L. • X. Oryzae pv. Oryzae • Bacterial leaf blight • Virulent proteins • Subcellular localization prediction • Pfam/Interpro

Introduction

Rice (O. Sativa L.) is a major source of food for the world’s population with an estimated global production of 508 million tons in the year 2020 [1]. The production of rice in India for the year 2020 is estimated to be 102.36 million tonnes [2]. For Indians rice is an important part of the diet with more than 70% of people directly dependent on it as a major source of calories. In India rice is cultivated in an area of 32.179 million hectares with a productivity of 3.18 metric tons per hectare [3]. Rice is a semi-aquatic plant and is infected by a large no of pathogens like fungi, bacteria, nematodes and viruses. These pathogens affect the productivity and quality of rice from moderate to severe levels. The bacterial leaf blight disease is one of the most severe diseases caused by bacteria X. oryzae and is observed throughout the Asian continent [4]. The losses due to disease may rise to 40% in susceptible varieties [5,6]. The Xanthomonas genus is plant-specific yellow-pigmented microbes, some of which are economically important phytopathogens that devastate crops such as citrus plants, rice, beans, grape and cotton. These organisms are almost exclusively found associated with their plant hosts [7,8]. The X. oryzae, a Gram’s negative bacterial plant pathogen, enters through the natural opening like stomata or from wounds on the plant surface. After entry, the pathogens start the process of host recognition, colonization and reproduction [9]. The infection is driven by virulent proteins that have a role in the initiation of infection and colonization of the host plant. The virulent proteins are crucial for the mechanism of infection [10]. The virulent proteins show a high level of expression in bacterial cells which undergoes a process of colonization and infection. Identification of such proteins is vital for the understanding of the pathogenicity of the pathogen. Due to advances inv computational biological sciences, it has become easy to identify such proteins through computational methods. Various tools like MP3, VirulentPred server and VICM server are available free of cost for users to identify such pathogenic proteins. Virulence Factor Database in Bateria (VFDB) is another resource that can be used for the prediction of virulent protein which is also freely available for users. The subcellular location of proteins is quite significantly related to their biological functions [11]. The Gram’s negative bacteria have five main subcellular localization sites it includes the inner membrane, the outer membrane, the periplasm, the cytoplasm and the extracellular space. Identification of such proteins, associated with the disease cycle and their subcellular prediction is crucial for the understanding of pathogenesis. Various online servers like BUSCA, CELLO, Gneg-mPLoc, PSORTb, PSL-Pred, SLP-Local, ngLOC, Gram-LocEN, CELLO2GO are available free to users for this purpose. These servers apply different techniques for the prediction of subcellular localization of proteins. Using multiple methods eliminates the chance of unilateral prediction of subcellular localization. The identified proteins were again subjected to Pfam/Interpro analysis for further functional analysis. The present research work is an in-silico approach for the prediction of virulent proteins, their subcellular localization and functional prediction of involved proteins in the virulence/pathogenicity process for X. oryzae, a potent pathogen of O. Sativa L.

Materials and Methods

Sequences

X. oryzae pv. oryzae PXO99A contains 5.24 Mbp of genome size (largest in X. oryzae) and contains 3907 proteins estimated through bio-project PRJNA131967. The Refseq no is NC_010717.2. The necessary files containing protein sequences, nucleotide sequences were downloaded from the NCBI server. The files were used for all sorts of computational analyses.

Prediction of virulence factor

MP3 standalone software: It is a UNIX-based software tool used for the prediction of pathogenic proteins in Genomic and Metagenomic data. MP3 Software works on two approaches Support Vector Machines (SVM) and Hidden Markov Model (HMM) for analysis of protein sequences. The standalone version of the software works on Linux operating systems and has been developed at Meta-informatics Laboratory, Metagenomics and Systems Biology Group, Department of Biological Sciences Indian Institute of Science Education and Research, Bhopal, M.P, India.

Virulentpred server: VirulentPred server is a bacterial virulent protein prediction server based on a double-layer cascade support vector machine. The first layer of SVM classifiers is trained with various sequence characteristics such as amino acid composition, dipeptide composition, composition of higher order dipeptides and remote and optimizes evolutionary relationships with Position Specific Iterated BLAST (PSIBLAST). The second layer received the binary score output generated by the top 5 modules of the first layer in order to train the SVM model of the second layer.

VICMpred server: VICMpred is a web-server for the broad functional classification of proteins of Gram’s negative bacteria into virulence factors, information molecules, cellular processes and metabolism molecules. The VICMpred server applies SVM-based methods for patterns, amino acid and dipeptide composition of protein sequences of bacteria. VICMpred server allows the users to predict the function of a protein as virulence factors, information molecules, cellular process and metabolism from its amino acid sequences.

Local blast with VFDB: The selected sequences passing the above criteria will be locally BLASTed against the Virulence Factor Database in Bateria (VFDB). The VFDB has been developed by Jin Qi and co-workers at the State Key Laboratory for Molecular Virology and Genetic Engineering, Beijing, China, The Wellcome Trust Sanger Institute, Wellcome Trust Genome Campus, Hinxton, Cambridge, UK and National Center of Human Genome Research, Beijing, China.

Servers for prediction of subcellular localization of identified pathogenic proteins

BUSCA server: The Bologna Unified Subcellular Component Annotator (BUSCA) web server, available at http://busca.biocomp.unibo.it/, integrates several established methods such as DeepSig, TPpred3, PredGPI, BetAware and ENSEMBLE3.0 and has been developed by the Bologna Biocomputing Group for the Prediction of the specific subcellular localization based on the protein sequence.

CELLO v.2.5 server: CELLO is a multiclass SVM-based subcellular localization prediction tool designed for Gram-negative, Gram-positive and Eukaryotic cells are available at http://www.csbio.sjtu.edu.cn/bioinf/Gneg-multi/. The server predicts subcellular localization by considering the amino acid composition, the di-peptide composition, the partitioned amino acid composition and the sequence composition of given protein sequences.

Gneg-mPLoc: The Gneg-mPLoc tool, available at http://www.csbio.sjtu. edu.cn/bioinf/Gneg-multi/, is part of Cell-PLoc 2.0 webserver and is specialized for prediction of subcellular localization of Gram’s negative bacteria. The GnegmPLoc uses information of gene ontology, functional domain and sequential evolution for the purpose.

PSORTb v.3.0 server: PSORTb server available at PSORTb Subcellular Localization Prediction Tool - version 3.0 is developed and made available by The Brinkman Laboratory. PSORTb v.3.0 consists of several modules for analysis, each of which analyzes one biological feature known to influence or be characteristic of subcellular localization.

PSL-Pred server: The PSLpred server, available at http://www.imtech. res.in/raghava/pslpred/, developed for prediction of subcellular localization of bacterial proteins in Gram’s negative bacteria is a hybrid approach-based method integrating PSI-BLAST and three SVM modules based on compositions of residues, dipeptides and the physicochemical properties.

SLP-Local server: The SLP-Local server, available at https://sunflower. kuicr.kyoto-u.ac.jp/~smatsuda/slplocal.html, predicts the subcellular location of proteins from the amino acid sequence. The method employed by the server is the local compositions of amino acids and twin amino acids and local frequencies of the distance between successive amino acids like basic, hydrophobic and others for prediction of subcellular localization.

ngLOC server: The ngLOC server is primarily an n-gram based Bayesian classifier on-line server for the prediction of subcellular localization of proteins each in prokaryotes and eukaryotes from the protein sequences the usage of bayesian type approach that fashions the density distribution of amino acids. The distributions are decided from datasets of experimentally annotated subcellular localization of proteins from diverse organism.

Gram-LocEN server: Gram-LocEN is an interpretable multi-label online predictor which makes use of unified capabilities to yield sparse and interpretable answers for large-scale prediction of proteins of various species, along with Gram’s negative and Gram-positive microorganisms. Given a question protein series in a selected species, a hard and fast of GO phrases are retrieved from newly created compact databases, specifically ProSeq-GO. The frequencies of GO occurrences are used to formulate frequency vectors with a dimensionality of 8,000+. Based on the chosen essential GO terms, the user of Gram-LocEN determines where a protein resides is located.

CELLO2GO server: CELLO2GO server, available at http://cello.life.nctu. edu.tw/cello2go/, is a publicly accessible web system for the detection of various properties of a target protein and its subcellular location. The server can be used for detailed Gene Ontology (GO) annotations, including subcellular locations, for queried proteins by combining the CELLO location prediction and BLAST homology search approaches.

Methods

The total number of protein sequences encoded by X. Oryzae pv. Oryzae PXO99A in FASTA format were subjected to MP3 software analysis. The software predicts the virulence of sequence for a range of 0 to 1. The number 0 and 1 denotes the to which extent the predicted protein has a probability of being virulent protein. The value 1 was chosen to predict virulence as we intended for those proteins only that have a very high chance of involvement in the pathogenesis process. An only a positive value of Support Vector Machine (SVM) prediction was taken as a criterion for the determination of virulent protein as negative values reflect non- virulent proteins only. The second criterion used is the detection of protein as pathogenic by Pfam database. The selected protein from this analysis was subjected to the next analysis of the VirulentPred server and VICM server. The proteins which were predicted by a vote of a majority were taken as virulent proteins and were selected for further analysis and prediction for subcellular localization. The selected sequences were BLASTed locally against the Virulence Factor DataBase (VFDB) with an E-value less than 0.05 and query coverage of more than 70% and more for selection as virulent sequences. The selected sequences passing all of the above selection criteria were then subjected for analysis on various web servers for prediction of the subcellular localization. The selected sequences were subjected to Pfam analysis using webserver of European Bioinformatics Institute and is available at https://www.ebi.ac.uk/Tools/ hmmer/. The HMMER web server is a fast and responsive homology search engine. This site is designed to provide a near interactive search for most searches along with intuitive and interactive display of results.

Results and Discussion

Prediction of virulent proteins

Using MP3 software, we were able to predict 499 protein sequences to be virulent with the assignment of the Pfam domain as a pathogenic protein. The highest SVM predicted score was 7.26 and the lowest was 1.38. The predicted pathogenic proteins were subjected to VirulentPred server analysis and 406 of the 499 proteins were predicted to be virulent. The VICM server predicted that only 62 of the 406 proteins were virulent. The MP3 software and both servers predicted 116 proteins as pathogens by majority vote. Table 1 contains the information of Predicted 116 proteins from MP3, VirulentPred and VICM server.

The proteins identified here were annotated as membrane proteins like transporter protein, membrane proteins, extracellular protein. These proteins have a role in the transportation of vital elements, signalling and passing of signals and secretory proteins, having a role in modulation of the infectious process and countering the defense mechanism of the host.

The VFDB local BLAST analysis

The VFDB local BLAST analysis with E-values of 0.05 and query coverage of 70% also confirmed that 116 proteins are a good candidate to be treated as virulent proteins. We can interpret that most of the proteins from the Table 2 we can interpret that a wide variety of proteins have been identified as virulent.

Prediction of subcellular localization of protein

The subcellular localization of selected 116 proteins through various web server analysis yielded diverse results. A summarized result is given here. We can clearly analyze that most of the identified proteins belong to extracellular, plasma membrane and cytoplasmic proteins, extracellular proteins are mostly secreted proteins, plasma membrane proteins (internal or external) play a role in channeling of metabolites across the membrane They play a role in the transmission of signals across the cell membrane as well. Cytoplasmic proteins are important metabolites that play a role in modulating the information required for pathogenesis. Therefore, we can confirm that through in silico analysis could find some of the proteins which may play a defined role in the pathogenesis of X. oryzae in causing bacterial leaf blight in rice (O. Sativa L.) (Table 3).

Conclusion

Results from Pfam/Interpro analysis using webserver of European Bioinformatics Institute is given here in Table 3. The function of these proteins was referenced by Pfam database and Interpro server analysis. From the above Table 3 it can be seen that most of the identified proteins play a role in host recognition as a part of the outer membrane protein, in the movement of bacteria as a part of the pathways responsible for formation and assembly of flagella or cilia, in the energy regulation as part of cytochrome proteins, part of the secretion system of type II, III, IV, transporters of various secreted proteins, as exotoxins and proteolytic enzymes play a role in the breakdown of the host cell wall, as heat shock protein plays a role in combating the fluctuations in ambient temperature. The proteins identified here are good candidates for further computational and wet laboratory analysis for analysis of various pathways responsible for infection and survival processes of the pathogen.

Acknowledgement

Above work was carried out in the Bioinformatics lab of Dept. of Microbiology and Bioinformatics of Atal Bihari Vajpayee Vishwavidyalaya, Bilaspur, Chhattisgarh, India.

Conflict of Interest

None.

References

1. Ben Hassen, Tarek, Hamid El Bilali and Mohammad S. Allahyari. "Impact of COVID-19 on food behavior and consumption in Qatar."Sustainability12 (2020): 6973.

Google Scholar, Crossref, Indexed at

2. Syed-Ab-Rahman, Sharifah Farhana, Lilia Costa Carvalhais and Dzolkhifli Omar. "Development of plant-based emulsion formulations to control bacterial leaf blight and sheath brown rot of rice."Heliyon6 (2020).

Google Scholar, Crossref, Indexed at

3. Shaheen, Rozina, Muhammad Zahid Sharif, Luqman Amrao and Aiping Zheng, et al. "Investigation of bacterial leaf blight of rice through various detection tools and its impact on crop yield in Punjab, Pakistan."Pak J Bot51 (2019): 307-312.

Google Scholar, Crossref, Indexed at

4. Joshi, Johnson Beslin, Loganathan Arul, Jegadeesan Ramalingam and Sivakumar Uthandi. "Advances in the Xoo-rice pathosystem interaction and its exploitation in disease management."J Biosci45 (2020): 1-17.

Google Scholar, Crossref, Indexed at

5. Shen, Yuwei and Pamela Ronald. "Molecular determinants of disease and resistance in interactions of X. oryzae pv. oryzae and rice."Microbes Infect 4 (2002): 1361-1367.

Google Scholar, Crossref, Indexed at

6. Furutani, Ayako, Minako Takaoka, Harumi Sanada and Yukari Noguchi, et al. "Identification of novel type III secretion effectors in X. oryzae pv. oryzae."Mol Plant-Microbe22 (2009): 96-106.

Google Scholar, Crossref, Indexed at

7. Jensen, L. Juhl, Ramneek Gupta, Nikolaj Blom and D. Devos, et al. "Prediction of human protein function from post-translational modifications and localization features."J Mol Biol319 (2002): 1257-1265.

Google Scholar, Crossref, Indexed at

8. Salzberg, Steven L., Daniel D. Sommer, Michael C. Schatz and Adam M. Phillippy, et al. "Genome sequence and rapid evolution of the rice pathogen X. oryzae pv. oryzae PXO99 A."BMC Genom 9 (2008): 1-16.

Google Scholar, Crossref, Indexed at

9. Gupta, Ankit, Rohan Kapil, Darshan B. Dhakan and Vineet K. Sharma. "MP3: A software tool for the prediction of pathogenic proteins in genomic and metagenomic data."PloS one9 (2014): e93907.

Google Scholar, Crossref, Indexed at

10. Garg, Aarti and Dinesh Gupta. "VirulentPred: A SVM based prediction method for virulent proteins in bacterial pathogens."BMC Bioinform 9 (2008): 1-12.

Google Scholar, Crossref, Indexed at

11. Saha, Sudipto and G. P. S. Raghava. "VICMpred: an SVM-based method for the prediction of functional proteins of Gram-negative bacteria using amino acid patterns and composition."Genom Proteom Bioinform 4 (2006): 42-47.

Google Scholar, Crossref, Indexed at