Model-assisted Process Development Of Continuous Chromatography And Its Applications For Antibody Separation

Monoclonal antibody(mAb)is regarded as the most important biotechnological drug.The traditional batch-based manufacturing methods often possess some limitations like huge investment,high risk,as well as high consumption of energy,substrate,and buffer.Changing from the traditional process into continuous manufacturing is the trend,which can improve product quality and process efficiency,promote equipment miniaturization and process flexibility,and reduce waste discharge and production cost.Main difficulties of continuous bio-manufacturing lie in the separation units,especially in continuous chromatography,which is complicated with many parameters and hard to design and optimize.Aiming at the complexity and particularity of continuous chromatography,this thesis developed mechanism model and artificial intelligence approach,constructed new methods of model-assisted design of continuous chromatography,evaluated the influences of key factors,and established new process optimization methods,to provide some guidance for the process development and application of continuous chromatography.Firstly,a mechanism model of chromatography was constructed to characterize the different modes of continuous capture process(including twin-column CaptureSMB,three columns 3C-PCC and multi-column continuous chromatography)for antibody affinity capturing.A general rate model considering parallel diffusion was used to predict the breakthrough curves under different operating conditions.Combined with the process model,the performances of continuous chromatography were calculated,and the interactive influence of different operating parameters on productivity and capacity utilization was analyzed.The results showed that the productivity changed with the key operation parameters by three phases,and different continuous capture modes showed different rules.The corresponding optimization methods were proposed.With the model-assisted optimization,the maximum productivity of CaptureSMB process with MabSelect SuRe resin was 33.3 g/L/h,and the capacity utilization was 94.3%.The maximum productivity of 3C-PCC process with MabSelect SuRe LX resin was 34.5 g/L/h,and the capacity utilization was 97.6%.The results were validated by experiments.Based on above results,the strategy of model-assisted continuous chromatography process optimization was proposed,and the software package was developed,which was then applied on the process development for the continuous capture of mAb.Secondly,due to the calculation speed of mechanism model,an artificial neural network approach was introduced to establish the hybrid model of mechanism model and artificial neural network,which was then utilized on the analysis of continuous capture process.The structure and data coupling method of the artificial neural network were optimized.Compared with the mechanism model,the calculation speed of the hybrid model was increased by 2000 times,reaching the millisecond level which is suitable for the multi-parameter process analysis and comparison under complex conditions.The hybrid model was then used to evaluate the influences of resin's main properties on the maximum productivity and working window under varying modes of continuous capture.It was found that different feeding concentrations have different rules,and the resin selection and optimization strategy of continuous chromatography under different feeding concentrations were obtained.In order to facilitate the application of hybrid model,an integrated software package was developed,which has been applied in the process development of continuous chromatography.Thirdly,for the frontal chromatography of antibody polishing,a new continuous mode of twin-column continuous frontal chromatography,Flow2,was studied,in which focused on the separation of mAb monomer and aggregate.Combining Flow2 process model with lumped kinetic model and DLVO theory,a new Design Procedure method based on the process understanding was established to characterize the performance of Flow2 process and achieve multi-parameter and multi-objective optimization.The results indicated that Flow2 process has significant advantages over single-column frontal chromatography.The process yield of Flow2 was three times higher than that of the single-column frontal chromatography at low loading flow rate and can attain a higher productivity and yield at higher salt concentration.The productivity and yield of Flow2 can reach 130 g/L/h and 99%,respectively,which is 36.5%higher than single-column frontal chromatography.In addition,the parameter sensitivity analysis also proved that Flow2 has a better robustness.Finally,SuperPro Deisgner software was used to simulate the entire manufacturing process of mAb in both batch and continuous mode.The investment and operation cost were calculated and compared.The results indicated that with the increasing of batch scale,the cost decreased significantly,and the economic hot spots shifted from equipment-dependent costs to raw materials and consumables.For the whole process optimization,three main bioreactors operated in staggered mode corresponding to one downstream line can effectively improve the overall process efficiency.Furthermore,several continuous mAb manufacturing processes(upstream continuous,downstream continuous,and integrated upstream and downstream continuous)with annual production of 1000 kg were constructed and compared with the batch process.The results showed that the upstream perfusion culture can significantly reduce the equipment-dependent costs of both upstream and downstream as well as simplify the strain amplification process,while the downstream continuous capture chromatography can reduce the column volume by 10 times,reduce the consumption of Protein A resin by 33%,hence saving the cost of mAb by 11%.The integration of upstream and downstream continuous process demonstrated the merits of both processes which can reduce 30%mAb cost than that of the batch process.To conclude,this thesis used the mechanism model,artificial intelligence method and process simulation tools to assist the process analysis and optimization of continuous chromatography.The influence of key parameters was evaluated,the best working windows were determined,and the rational design strategies were developed.The model-assisted continuous chromatography process design methods established in this thesis are universality and can be further extended to other separation processes,which would be useful for the rational development and application of continuous bioprocess.