Consensus Co-expression Network Analysis And Machine Learning Model Investigate Ethylene Regulated Cell Wall Degradation Mechanisms In Kiwifruit

Cell wall degradation and remodeling is the key factor causing fruit softening during ripening.Ethylene significantly promotes most fruit softening through cell wall degradation.To explore the mechanism underlying postharvest cell wall metabolism,taking kiwifruits after ethylene treatment and during ethylene-24 hour-treatment as materials to conduct physiological and transcriptomic researches.The improved consensus co-expression network analysis(CCNA)and machine learning model was employed together with molecular biology approaches.Finally,the regulation mechanism of cell wall degradation after ethylene treatment was identified,and the time series response model during ethylene treatment was constructed as well as the hierarchical cascade regulation mechanism of ethylene response was predicted.The dominating results were listed as following:1.The improvement of data analysis.CCNA was combined with weighted gene co-expression network analysis(WGCNA)by introducing external physiological data(including firmness,cell wall materials,cellulose,hemicellulose,water soluble pectin,covalent binding pectin and ionic binding pectin),85 consensus clusters were generated,and 10 candidate genes were predicted,including 4 transcription factors(Ad ZAT5,Ad DOF3,Ad NAC083 and Ad MYBR4)and 6 cell wall metabolism-related structural genes(Ad GAL1,Ad MAN1,Ad PL1,Ad PL5,Adβ-Gal5 and Ad PME1).The regulatory relationships among candidate genes were predicted by correlation analysis.2.We clarified that Ad ZAT5 regulated Ad PL5 and Adβ-Gal5 to affect kiwifruit postharvest softening after ethylene treatment during storage.Dual luciferase system demonstrated that zinc finger protein(Ad ZAT5)positively regulated pectin lyase gene(Ad PL5)and β-galactoside gene(Adβ-Gal5)by binding their promoters.Electrophoretic mobility shift assay(EMSA)further confirmed that Ad ZAT5 directly bound promoters of Ad PL5 and Adβ-Gal5 with GAGGGAGAAGAGAAGACAA and CTTAGTCAGCCTCCTCCGC.Subcellular localization experiments demonstrated that Ad ZAT5 localized in nuclear like most transcription factors,and transient overexpression experiments in kiwifruit fruits verified the regulatory effect of Ad ZAT5 on Ad PL5 and Adβ-Gal5 in vivo.3.The quick response motifs and genes on ethylene during ethylene treatment were predicted.The contents of total soluble solids,firmness,cell wall materials,water soluble pectin,covalent binding pectin,ionic binding pectin and RNA sequencing in time series sampling during ethylene treatment were measured.The important pupative cis-regulatory elements(p CREs)were characterized by random forest model(RF).Then,CCNA results were combined with characteristic p CREs to predict three ethylene initial(0-2 h)response elements,which were TACGCG,AACGCG and ACGCGG.What is more,transcription factors and softening-related genes underlying early response(0-8 h),middle response(10-12 h)and late response(16-24 h)stages were predicted.Thus,a hierarchical cascade regulation model of different genes based on time response model was proposed.