Omics And Fluorescence Excitation-emission Matrices Techniques For The Diagnosis Of Iron Nutrition In Hydroponic Gynura Bicolor DC.

Iron is an essential micronutrient for plants,and its deficiency can seriously affect plant pigment synthesis,photosynthesis and respiration.In the nutrient-controlled hydroponic mode,if the culture solution is not changed for a long time,it will lead to an imbalance in nutrient distribution ratio and cause iron deficiency in plants.As an iron-rich food,iron deficiency in Gynura bicolor DC.(G.bicolor),known as hong-feng-cai,will result in lower iron and anthocyanin content in the edible parts,seriously affecting its nutritional value.In order to establish a diagnostic method for iron nutrition in G.bicolor,this paper explores the differences in gene expression and metabolites of iron-deficiency G.bicolor at the molecular level based on the analysis of physiological differences in iron-deficiency G.bicolor,respectively,based on transcriptomics and metabolomics technologies,reveals the diagnostic mechanism of fluorescence spectroscopy of iron nutrition in G.bicolor,clarifies the best diagnostic pathway for iron nutrition in hydroponic G.bicolor,establishes a EEM-based diagnostic method for early iron nutrition in hydroponic G.bicolor,and provides a new method for the development of nondestructive nutrition diagnosis in hydroponic plants.The main contents and conclusions are as follows:(1)Physiological differences between normal and iron-deficient G.bicolor were compared.The morphological indexes,chlorophyll content and major mineral elements were detected in control and test groups of G.bicolor.The results showed that iron deficiency led to retarded growth,shorter plants,shorter leaves and a significant decrease in fresh and dry weight of G.bicolor;chlorophyll content in new leaves was significantly reduced by iron deficiency;iron content in new leaves,stems and roots was significantly reduced by iron deficiency,and iron deficiency seriously interfered with the accumulation of calcium,magnesium and zinc in stems and roots of G.bicolor.(2)The differentially expressed genes of iron-deficiency G.bicolor were analyzed.Differentially expressed genes were analyzed in new leaves,old leaves and roots of irondeficiency G.bicolor using transcriptomics techniques.The results showed that iron deficiency caused significant down-regulation of genes related to the metabolic pathways of porphyrin and chlorophyll metabolism,anthocyanin biosynthesis and flavonoid biosynthesis in new leaves of G.bicolor;iron deficiency caused significant up-regulation of genes related to amino acid biosynthesis,starch and sucrose metabolic pathways in roots of G.bicolor;iron deficiency significantly affected metabolic pathways involved in the synthesis of secondary metabolites in both roots and new leaves.(3)Differential metabolites of iron-deficient G.bicolor were analyzed.The differential metabolites in new leaves,old leaves,roots and root exudates of G.bicolor with iron deficiency were analyzed using metabolomics techniques.The results showed that iron deficiency caused a decrease in the content of 4-aminobutyric acid,vitamin B6,nicotinate and citrate in new leaves of G.bicolor and reduced the nutritional quality of new leaves;iron deficiency caused a significant increase in the content of sucrose and citrate in roots of G.bicolor;iron deficiency led to significant differences in the accumulation of three metabolites with fluorescence effects,oleic acid,pyridoxine and vitamin B6,in the root exudates of G.bicolor,providing a theoretical basis for diagnosing iron nutrition in G.bicolor based on the information of fluorescence differences in root exudates.(4)The EEM characteristics of control and iron-deficiency(test group)G.bicolor root exudates were analyzed,and a qualitative diagnostic model of iron nutrition in G.bicolor based on the EEM of root exudates was established.The results showed that 1)the fluorescence components of root exudates of G.bicolor in the control and test groups were significantly different,with the control group containing EEM components,namely tryptophan-like(Ex/Em =260/350 nm),tyrosine-like(Ex/Em = 275/305 nm)and humic-like fluorophores(Ex/Em =325/440 nm),while the test group contained only the latter two fluorescence components;2)parallel factor analysis(PARAFAC),alternating trilinear decomposition algorithm(ATLD),principal component analysis(PCA)and K-nearest neighbor(KNN)were used to downscale the EEM of G.bicolor root exudates in the control and test groups,respectively,and combined with support vector machine(SVM)and auto-encoder(AE)to establish a qualitative diagnostic model of iron nutrition in G.bicolor,in which the diagnostic model established by PCA combined with SVM method had the best classification effect,with an overall accuracy of 87.0%;3)convolutional neural network(CNN),long and short-term memory network(LSTM)and bidirectional long and short-term memory network(Bi LSTM)were used to extract the depth features of the EEM of root exudates of G.bicolor in the control and test groups,respectively,and a deep learning model for iron nutrition diagnosis of G.bicolor was established,in which the LSTM model had the best diagnostic effect with an overall accuracy of 95.7% and good recognition ability for samples in both the control and test groups,proving that EEM of root exudates combined with deep learning algorithms for iron nutrition diagnosis of G.bicolor were proved to be feasible.(5)A EEM-based method for early iron nutrition diagnosis of circulating hydroponically grown G.bicolor was proposed.Using EEM,the root exudates accumulation solution(circulating solution)of hydroponically grown G.bicolor was examined to explore the feasibility of early iron nutrition diagnosis.The results showed that 1)the fluorescence components of circulating solution from control and test samples were significantly different,and the test samples contained four fluorescence components,namely microbial-like humic fluorophores(Ex/Em = 325/365 nm),humic fluorophores(Ex/Em = 340/430 nm),fulvic acid,humic-like fluorophores(Ex/Em = 265,370/455 nm)and protein fluorophores(Ex/Em = 275/315 nm),while the control samples contained only the latter three fluorescence components;2)the emission spectrum at excitation wavelength of 325 nm and the excitation spectrum at emission wavelength of 365 nm were extracted based on the results of fluorescence component analysis,and the extracted excitation spectrum and emission spectrum were used in combination with SVM and KNN to establish a qualitative diagnostic model for early iron nutrition in hydroponically grown G.bicolor,in which The best performance of the early diagnosis model was established by combining the emission spectrum at 365 nm excitation wavelength with the SVM method,and the overall accuracy was94.0%;3)deep learning models for early diagnosis of iron nutrition in hydroponically grown G.bicolor was established using CNN,LSTM and Bi LSTM to extract the depth features of the EEM of G.bicolor circulating solution,in which the Bi LSTM model gave the best diagnosis results with an overall accuracy of 96.0%;4)In order to improve the training speed and accuracy of the deep learning models,the network parameters of the deep learning models in(4)were learned using the transfer learning method,and this network parameter was used to reconstruct the deep learning models for early diagnosis of iron nutrition in hydroponically grown G.bicolor for qualitative analysis,in which the transfer learning combined with CNN and LSTM models gave the best results for qualitative analysis with an overall accuracy of 97.0% for both.The recall rate of the models iron deficiency samples both reached 100%,and the early iron nutrition diagnosis of circulating hydroponic G.bicolor was initially achieved.