A Prediction Of Protein-protein Interaction Network Of Rice Sheath Blight Pathogen And Functional Validation Of Transcription Factor STE12 Gene

Rhizoctonia solani is a very common soilborne pat hogen with a great diversity of host plants, such as rice, maize, and wheat. Rice sheath blight, one of three fungal diseases in rice caused by Rhizoctonia solani Kuhn, is becoming more and severe in world wide. Rhizoctonia solani belongs to Deuteromycontina in a sexual form while it is called Thanatephorus cucumeris at sexual form affiliated to Basidiomycolina. R. solani is the imperfect state of the basidiomycete fungus that does not produce any asexual spores （called conidia） and only occasionally produce sexual spores （basidiospores）. So, the studies of the genetics and pathology of this pathogen have been experimentally hindered by the lack of an effective transformation system. Unfortunately, only a few pathogenic genes have been identified in this pathogen. The mechanisms for overcoming the plant immune system remain poorly understood. We also have not identified the resistance genes in rice varieties.Traditional research is time consuming, labor intensive, tedious, and ineffective at producing a holistic understanding of a complete biological process because it relies on a single gene or protein4. Within cells, the proteins interact with each other and function as a complex network. Therefore, every protein can be regarded as part of a complex PPI network. Moreover, the interacting protein pairs within one cell can be enormous, making it impossible to verify every interaction experimentally. Therefore, an effective way to complement experimental approaches to advance our knowledge about this pathogenic filamentous fungus in biological processes is needed. Identifying the protein-protein interaction （PPI） maps of R. solani can provide insights into the potential pathogenic mechanisms and assign putative functions to unknown genes. The high quality of the network was revealed by comprehensive methods, including yeast two-hybrid experiments. In addition, we identified and characterized the transcripition factor gene STE12 in R. solani AG1 IA by using the reverse genetic approaches according the network. Our results are as followings:1. Two widely computational methods, the interolog and domain interaction-based methods, are used to generate the PPI of R. solani AG1 IA. Interolog approach can be described as the transfer of known interactions from model organisms to other species based on comparative genomics. Through this approach, we used the reported data of the PPIs from five model organisms, D. melanogaster, C. elegans, H. sapiens, S. cerevisiae, and M. oryzae, and identified the orthologs between R. solani AG1 IA and these species using the Inparanoid algorithm. Totally,8,312 interactions were inferred among the 1,991 R. solani AG1 IA proteins. Additionally, we utilize the DDI information available in the DOMINE database and used the interacting Pfam domain pairs from DOMINE to predict PPIs. In our work,1,336,786 interactions were annotated between the 4,780 proteins associated with the DDIs in the DOMINE database. Although various successful computation methods were used to predict the DDIs, a significant number of false positives and negatives were also obtained. To reduce the number of false positives, we only use the DDIs complementation with the interolog method. Furthermore, we selected interactions with a confidence score of at least 0.5, as provided by Inparanoid, and interactions predicted using both the interolog and DDI methods to obtain our core interactions network. This network contained a total of 6,705 interactions among 1,773 proteins, and 3,166 （47.12%） of these interactions were supported by DDIs. Our results indicated that approximately two-thirds of the interactions were from the PPI data of M. grisea. The reason may be the M. grisea are evolutionarily closer to R. solani.2. Because there was a high false positive rate in the current large-scale experimental PPI data, the PPIs based on the interolog method unavoidably contained a large number of false positives. Moreover, due to the absence of comprehensive experimental PPI data in R. solani to verify the protein interactions, we employed an experimental approach to evaluate the reliability of the predicted interaction network. The entire interaction network was assessed by gene ontology （GO） annotations. In the network, a pair of physiologically interacting proteins would be expected to have related, but not identical functions. Therefore, they should share some, but not all of their GO annotations. We count the interactions for which the connected proteins share common GO terms to evaluate the interaction dataset and compare the proportion of the interactions that shared at least one GO term in the predicted network and a randomly generated network of the same size. It showed that the proportion of interactions sharing the GO annotations at any level of the GO hierarchy is higher than the largest percentage in the randomized networks. This result suggests that the predicted PPI network indeed preferentially connects the functionally related proteins sharing GO terms at any level of the GO hierarchy. To further validate the predicted interactions, we investigated whether the interacting proteins tend to have correlated gene expression profiles. Moreover, we analyzed the transcriptome of R. solani AG1 IA at six time points during the infection process and found that 10,103 genes were expressed. Among those genes, many have correlated gene expression profiles, which were up-regulated or down-regulated together. The level of co-expression of an interacting protein pair can be assessed using Pearson correlation coefficients （PCCs）. We calculated the PCCs for the predicted interactomes and compared them to those of a same size random network. The result showed that the number of PPIs with higher PCCs in the predicted network was significantly larger than that of the random protein pairs. This result demonstrates that the protein interaction pairs in the predicted PPI network prefer to be co-expressed, implying that the predicted PPI network was more reliable than random protein pairs.3. To validate the predicted PPI network accuracy in laboratory, we selected a hub, AG1IA_O5961, and randomly selected its eleven partners using a yeast two-hybrid assay, which is a well-established genetic approach for testing PPIs in vivo. Among the selected predictions, two candidate proteins encoded by the AG1IA₀3263 and AG1IA_O6O33 genes could active the yeast reporter showed self-activated. Four pairs clearly displayed interaction:AG1IA_O5961 and AG1IA₀0962, AG1IA₀5961 and AG1IA₀0847, and AG1IA₀5961 and AG1IA₀0059, AG1IA₀5961 and AG1IA₀5186. To reduce the false positive results from the yeast two-hybrid experiment, we used a stringent selection condition in our experiment, making it inevitable that we missed some weak or transient interactions. In fact, we find another three pairs of interaction showed weakly interactions, which are AG1IA₀5961 and AG1IA₀4857, AG1IA₀5961 and AG1IA₀1360, AG1IA₀5961 and AG1IA₀2195. Taken together, the results of the yeast two-hybrid experiments on the selected prediction pairs demonstrated that our predictions were reliable, and our predicted PPI data can provide useful guideline for future research.4. Based on an analysis of the PPI networks, we can better understand the web of interactions that takes place inside a cell, as well as the basic internal components and organization at the network level. One method to better understand the entire network is to partition it into a more manageable series of sub-networks. We selected some multiple-core networks with core genes related to pathogenic and secreted proteins and their interacting partners for further analysis. The expression of pathogenic genes during infection is crucial for overcoming the innate immunity of rice and for the maintenance of its parasitic lifestyle. Specially, the secretome provides clues for understanding the complex relationship between plants and their pathogens. We also slected some core gene and their partner to further analysis, and the results indicated that many of their interacting partners shared similar or related functions with the core genes. Thus, we can predict the functions of previously uncharacterized members of the network.5. To detect the pathogenic genes in a precise way, the gene expression profiles of R. solani AG1 IA were determined 10,18,24,32,48, and 72 h after its invasion into rice. The expression of some pathogenic genes was generally altered following the invasion of the pathogen into the rice. By integrating the protein interaction map with the gene expression data, we can predict the pathogenic genes of R. solani AG1 IA in a precise manner. The genes that tightly interact with known pathogenic genes and the genes that are differentially expressed after the invasion of rice are candidate pathogenic genes. Using this approach, we choose genes with significant expression changes and their interacting partners to generate a sub-network. The network demonstrates that many proteins that interact with the core proteins associated with pathogenicity either increased or decreased their expression during the infection process. In particular, the secreted proteins acted as ribosomal proteins, and many of their interacting partners were ribosomal proteins.6. In our work, we predicted 158 interactions among 20 genes from the MAPK pathways of R. solani AG1 IA and compared it with another rice destructive fungus, M. oryzae, in which 437 interactions were observed between 40 genes. The pathways of the two fungi share some components of the network, such as Fksl, Rhol, and Pbs2. However, the pathways also have many different elements, such as Gpe1, Ste4, Ste18, and Ssk2, which are only found in R. solani AG1 IA, and Cdc42, Bckl, Slt2, and Hogl, which are only found in M. oryzae. Furthermore, a detailed comparison analysis of the CWI pathway was performed. The CWI pathway is highly conserved and plays a crucial role in the maintenance of CWI to retain fungal growth, survival, and pathogenesis in response to stress. R. solani AG1 IA and M. oryzae have two common core components, Rho1 and Pkc1. A variety of proteins interact with Rhol in both R. solani AG1 IA and M. oryzae. Most of the proteins that interact with Rho1 have similar functions and are involved in similar biological processes between the two rice destructive funguses. However, the two fungi were also shown to have some different proteins. Together, these findings highlight the diversity and complexity of signal regulation in fungi, which can contain both the highly conserved MAPK phosphorylation cascades as well as a variety of regulatory genes. These differences may be due to the differences in lifestyle and infection process between the two fungi.7. In order to identify the function of Fus3/Kssl-homolog in development and pathogenic process, a STE12 gene of R. solani AG1 IA were identified, which located downstream in MAPK pathway. STE12 included 3,218 bp DNA sequence with 2,643 bp coding region, encoded 880 amino acids, the molecular weight was 96.0494KDa and consisted of 6 exons and 5 introns. Ste12 was detected in R. solani AG1 IA and contains two C-terminal C2H2 zinc finger motifs, a Ste domain, and a novel protein structure found when compar ing it with Ste 12 in other fungi. Sequence analysis revealed that STE 12 shared highly homologous with other STE12-like genes of other plant pathogens.8. In order to identify the function of STE 12 in R. solani AG1 I A, we constructed the fungal integrative transformation vector pSH75-STE12 and the null knockout mutants were generated through fungal protoplast transformation. The gene knockout mutants were confirmed by PCR, Semiquantitative RT-PCR, qRT-PCR and Southern blot. Semiquantitative RT-PCR and qRT-PCR analysis of expression levels of STE12 showed that the expression levels of STE 12 in transformants were greatly decreased. To further characterize homologous recombination of transforming DNA in the filamentous fungus R. solani AG1 IA, the Southern blot was performed. The results showed that STE12 gene was only single copy in R. solani AG1 IA genome. Compared with the wild type AG1 IA, â–³ STE 12 led to a slightly decreased vegetative growth, the number of selertium also decreased, the aging of mycelium delayed and somatic incompatibility reactions in transformants and R. solani AG IA on PDA.9. Pathogenicity of the wild-type and the disruptant was analysed by the leaf sheath intact method. The results showed that the â–³STE12 disruptant showed a significant decrease in pathogenicity or even loss of pathogenicity. So the STE12 gene was important to the plant infection in R. solani AG IA. In addition, the STE12 gene is one of the downstream transcription factors of MAPK and involved in the regulation of the plant infection in R. solani AG IA. So we can predicted there might be some other genes in STE12 downstream which regulated by STE12 involving in the process of vegetative growth and pathogenicity in R. solani AG IA.