采用DoE设计方法优化可溶性肿瘤坏死因子I型受体蛋白的聚乙二醇修饰工艺

Optimization of procedure for PEGylation of soluble tumor necrosis factor α receptor Ⅰ by Design of Experiment

目的 采用DoE设计方法对人可溶性肿瘤坏死因子I型受体（soluble tumour necrosis factorαreceptors I,sTNFα-R玉）的聚乙二醇（PEG）修饰工艺参数进行优化,以期获得稳定、可控的工艺参数。方法 采用UNICORN（DoE）软件中Central Composite Face Centered（CCF）模型对sTNFα-R玉的3个PEG修饰工艺参数,即溶液pH值（pH）、反应时间（Time）和PEG与蛋白质量比（PEGrate）进行优化,设置4个响应值（多PEG修饰蛋白峰面积、单PEG修饰蛋白峰面积、未修饰蛋白峰面积及单PEG修饰蛋白峰面积与多PEG修饰蛋白峰面积比值）揭示其化学反应过程,确定参数操作空间;应用蒙特-卡罗模拟法（Monte-Carlo Simulation）对优化参数进行风险评估。结果3因子17组试验结果的重复性较好,中心点重复组响应值介于最低值和最高值之间,且随机分布,符合DoE试验设计的要求;4个响应值所对应的回归系数（R2）均〉0.75,预测准确性（Q2）〉0.5,模型有效性（Model Validity）〉0.25,重复性（Reproducibility）〉0.5,模型预测与实验值之间偏差较小,具有可靠的预测准确性;17组试验结果标准残差在允许范围（-4～4）之间,线性较好,预测值与观测值之间相关性较好,建立的模型拟合度高,可较为准确地预测试验结果。溶液pH值对多PEG修饰产量影响较小,随着Time和PEGrate的升高,多PEG修饰产物增加,单PEG修饰产物经二次或多次修饰形成多PEG修饰产物,减少Time和PEG用量可减少多PEG修饰产物;单PEG修饰产物随着pH上升,产量有所增加,与Time呈二次项关系,在Time.PEGrate交互作用中,减少PEG用量可增加单PEG修饰产物量;减少Time和降低PEG用量可提高单PEG修饰产物与多PEG修饰产物的比值,有利于下游纯化。确定了工艺参数的操作空间,当pH为5.5时,反应时间控制在30 h以上,PEG与蛋白质量比控制在3.7以下,符合预设值;pH在6.5时,反应时间控制在22～35 h之间,PEG与蛋白质量比控制在2.5～4.0之间,符合预设值;蒙特-卡罗模拟法对在最佳操作参数（反应时间34.1 h,溶液pH值6.5,PEG与蛋白质量比2.5）时模型预测的响应值（单PEG修饰产物峰面积值5 635.82和单PEG修饰产物与多PEG修饰产物比值3.226 5）的风险评估,10万次模拟试验可信度分别为99.825%和99.982%,试验风险较小,工艺可控。结论 确定了sTNFα-R玉蛋白PEG修饰稳健的工艺参数。

Objective To optimize the technological parameters for PEGylation of soluble tumor necrosis factor α receptor Ⅰ （sTNFα-R Ⅰ ） by Design of Experiment （DOE） so as to obtain stable and controllable parameters. Methods Three technological parameters of PEGylation of sTNFα-R Ⅰ , pH, time and PEGrate, were optimized by using the Central Composite Face Centered （CCF） model in UNICORN （DOE） software, based on which the chemical reaction process was indicated by four response values, i.e. peak area of multiple PEGylation product, peak area of single PEGylation product, peak area of unmodified protein as well as ratio of peak area of single PEGylation product to that of muhiple PEGylation product, and the control space of the parameters were determined. The risks of optimized parameters were evaluated by Monte-Carlo Simulation method. Results The results of 17 groups of tests on three technological parameters showed high reproducibility. The response value of central point repeat group was between the minimum and maximum and distributedrandomly, which met the requirement for DoE. All the regression coefficients （R^2） corresponding to the four response values were more than 0. 75, while the predicted accuracy （Q^2） was more than 0. 5, the model validity was more than 0. 25, and the reproducibility was more than O. 5. The deviations between the predicted and measured values were small, indicating a reliable accuracy. The residual standard deviations of 17 groups of test results were within the permitted range （-4 - 4 ）, and good linearity was observed. The predicted value showed high relationship to the measured value, indicating high degree of fitting of the established model which was suitable for accurate prediction of test result. The pH value of solution showed no significant effect on multiple PEGylation rate. However, along with the increasing time and PEGrate, multiple PEGylation rate increased, and single PEGylation product formed multiple PEGylation product by PEGylation for 2 times or more. Decreasing time and PEGrate decreased multiple PEGylation product. Single PEGylation product increased with the increasing pH value, which showed a quadratic term relationship to the time. In Time＇ PEGrate interaction, decreasing PEG used increased the single PEGylation product, while decreasing time and PEG used increased the ratio of single PEGylation product to multiple PEGylatioin product, which was beneficial to downstream purification. The control spaces of the param- eters were determined. At pH 5. 5, the reaction time was controlled at more than 30 h, while PEGrate at less than 3. 7, which were consistent with the designed values. However, at pH 6. 5, the reaction time was controlled at 22 - 35 h, while PEGrate at 2. 5 - 4. 0, which were consistent with the designed values. The optimal reaction time, pH value and PEGrate were 34. 1 h, 6. 5 and 2. 5 respectively. Under the optimal condition, the peak area of single PEGylation product was 5 635. 82, while its ratio to that of multiple PEGyl_ation product was 3. 226 5. The rehabilities of the two response values evaluated by Monte-Carlo Simulation method in I00 000 simulation tests were 99. 825% and 99. 982% respectively, indicating low risk of test and controllability of procedure. Conclusion Stable technological parameters for PEGylation of s TNFα-R I were determined.