Optimization of salicylic acid and chitosan treatment for bitter secoiridoid and xanthone glycosides production in shoot cultures of Swertia paniculata using response surface methodology and artificial neural network
Optimization of salicylic acid and chitosan treatment for bitter secoiridoid and xanthone glycosides production in shoot cultures of Swertia paniculata using response surface methodology and artificial neural network
复制标题
DOI：
10.1186/s12870-020-02410-7
复制
发表时间：
2020-05-19
影响因子：
5.3
通讯作者：
Pandey, Devendra Kumar
中科院分区：
文献类型：
--
作者：
Kaur, Prabhjot;Gupta, R. C.;Pandey, Devendra Kumar
本研究采用响应面法（RSM）和人工神经网络（ANN）建立了水杨酸（SA）和壳聚糖（CS）两个自变量对獐牙菜芽培养物生产獐牙菜苷(I)、獐牙菜苷（II）和芒果苷（III）的线性、二次效应和交互效应的预测模型。这些化合物是獐牙菜属植物的主要治疗代谢物，在制药工业中具有重要的作用和需求。结果在1/2 MS改良培养基（BA、KN各2.22 mM， NAA各2.54 mM）上，不同浓度的SA和CS激发子显著影响了(I)、（II）和（III）化合物的产率。在RSM中，利用五因子-三水平全因子CCD计算了线性、二次和双向交互模型的不同响应变量。在人工神经网络建模中，将13组CCD矩阵分成3个子集，以大约8:1:1的比例进行训练、验证和测试。在9 mM和12 mg L- 1 （SA）和（CS）诱导下，经14 d处理后，金针藤茎部(I)（0.435%）、（II）（4.987%）和（III）（4.357%）的产量均达到最佳水平。在优化研究中，(I)为0.170 ~ 0.435%；（II）在SA （1 ~ 20 mM）和CS （1 ~ 20 mg L- 1）的变化范围内显示1.020 ~ 4.987%和（III）高达2.550 ~ 4.357%的差异。总体而言，与RSM （r(2) = 99.8%）相比，利用人工神经网络模型优化激发子促进赛环烯醚酮和山酮苷生成（r(2) = 100%）的结果更为显著。
Background In this study, response surface methodology (RSM) and artificial neural network (ANN) was used to construct the predicted models of linear, quadratic and interactive effects of two independent variables viz. salicylic acid (SA) and chitosan (CS) for the production of amarogentin (I), swertiamarin (II) and mangiferin (III) from shoot cultures of Swertia paniculata Wall. These compounds are the major therapeutic metabolites in the Swertia plant, which have significant role and demand in the pharmaceutical industries. Results Present study highlighted that different concentrations of SA and CS elicitors substantially influenced the % yield of (I), (II) and (III) compounds in the shoot culture established on modified 1/2 MS medium (supplemented with 2.22 mM each of BA and KN and 2.54 mM NAA). In RSM, different response variables with linear, quadratic and 2 way interaction model were computed with five-factor-three level full factorial CCD. In ANN modelling, 13 runs of CCD matrix was divided into 3 subsets, with approximate 8:1:1 ratios to train, validate and test. The optimal enhancement of (I) (0.435%), (II) (4.987%) and (III) (4.357%) production was achieved in 14 days treatment in shoot cultures of S. paniculata elicited by 9 mM and 12 mg L- 1 concentrations (SA) and (CS). Conclusions In optimization study, (I) show 0.170-0.435%; (II) display 1.020-4.987% and (III) upto 2.550-4.357% disparity with varied range of SA (1-20 mM) and CS (1-20 mg L- 1). Overall, optimization of elicitors to promote secoiridoid and xanthone glycoside production with ANN modeling (r(2) = 100%) offered more significant results as compared to RSM (r(2) = 99.8%).