The paper proposes a modular-based approach to constraint handling in process optimization and control. This is partly motivated by the recent interest in learning-based methods, e.g., within bioproduction, for which constraint handling under uncertainty is a challenge. The proposed constraint handler, called predictive filter, is combined with an adaptive constraint margin and a constraint violation cost monitor to minimize the cost of violating soft constraints due to model uncertainty and disturbances. The module can be combined with any controller and is based on minimally modifying the controller output, in a least squares sense, such that constraints are satisfied within the considered horizon. The proposed method is computationally efficient and suitable for real-time applications. The effectiveness of the method is illustrated through a realistic case study of glycosylation constraint satisfaction in continuous monoclonal antibody biosimilar production using Chinese hamster ovary cells, employing a metabolic network model consisting of 23 extracellular metabolites and 126 reactions. In the case study, the average constraint-violation cost is reduced by more than 60% compared to the case without the proposed constraint-handling method.

A novel approach of design of experiment (DoE) is developed for the optimization of key substrates of the culture medium, amino acids, and sugars, by utilizing perfusion microbioreactors with 2 mL working volume, operated in high cell density continuous mode, to explore the design space. A mixture DoE based on a simplex-centroid is proposed to test multiple medium blends in parallel perfusion runs, where the amino acids concentrations are selected based on the culture behavior in presence of different amino acid mixtures, and using targeted specific consumption rates. An optimized medium is identified with models predicting the culture parameters and product quality attributes (G0 and G1 level N-glycans) as a function of the medium composition. It is then validated in runs performed in perfusion microbioreactor in comparison with stirred-tank bioreactors equipped with alternating tangential flow filtration (ATF) or with tangential flow filtration (TFF) for cell separation, showing overall a similar process performance and N-glycosylation profile of the produced antibody. These results demonstrate that the present development strategy generates a perfusion medium with optimized performance for stable Chinese hamster ovary (CHO) cell cultures operated with very high cell densities of 60 × 106 and 120 × 106 cells/mL and a low cell-specific perfusion rate of 17 pL/cell/day, which is among the lowest reported and is in line with the framework recently published by the industry.

BACKGROUND: Tangential flow filtration (TFF) systems are widely used cell retention devices in perfusion cultures, but significant challenges occur with their prolonged operation. A well-known and common issue includes membrane fouling, which leads to reduced permeate flow and increased product retention. The fouling behavior in hollow fibers have been studied widely and the application of alternating flow profiles, cell lysis, shear stress, and membrane pore size have been found to be major contributors to this phenomenon. RESULTS: In this work, different process set ups and conditions were tested using magnetic levitation pumps for low shear TFF systems in small-scale perfusion bioreactors (200 mL). A novel concept based on the application of reverse flow across the hollow fiber using two magnetically levitating pumps was validated with a Chinese Hamster Ovary cells (CHO) cell line producing a recombinant monoclonal antibody. Product sieving could be improved by 30% when a dynamic recirculation flow was applied. Furthermore, minimal product retention was achieved by reversing the flow of two alternating pumps with short cycle times. Besides this, a correlation was found between the passage of high molecular weight species to the harvest stream of the perfusion process and the flow direction, as well as the degree of product sieving. CONCLUSION: TFF with a reverse flow is a valuable alternative to an alternating tangential flow (ATF) system for overcoming antibody retention and it can be used at various scales and at a constant bioreactor volume. The comparison of ATF and TFF showed differences in product yield and purity and is, therefore, an important point for process design.

Monoclonal antibodies (mAb) are used as therapeutics and for diagnostics of a variety of diseases, and novel antibodies are continuously being developed to find treatments for new diseases. Therefore, the manufacturing process must accommodate a range of mAb characteristics. Acid-sensitive mAbs can severely compromise product purity and yield in the purification process due to the potential formation of aggregates. To address this problem, we have developed an integrated downstream process for the purification of pH-sensitive mAbs at mild conditions. A calcium-dependent Protein A-based ligand, called Z_Ca, was used in the capture step in a 3-column periodic counter-current chromatography operation. The binding of Z_Ca to antibodies is regulated by calcium, meaning that acidic conditions are not needed to break the interaction and elute the antibodies. Further, the virus inactivation was achieved by a solvent/detergent method, where the pH could remain unchanged. The polishing steps included a cation and an anion exchange chromatography step, and screening of the capture and polishing steps was performed to allow for a seamless integration of the process steps. The process was implemented at laboratory scale for 9 days obtaining a high yield, and a consistently pure drug substance, including high reduction values of the host cell protein and DNA concentrations, as well as aggregate levels below the detection limit, which is attributed to the mild conditions used in the process.

In this study, we demonstrated the first, to our knowledge, integrated continuous bioprocess (ICB) designed for the production of acid-sensitive monoclonal antibodies, prone to aggregate at low pH, on pilot scale. A high cell density perfusion culture, stably maintained at 100 × 10⁶ cells/ml, was integrated with the downstream process, consisting of a capture step with the recently developed Protein A ligand, Z_Ca; a solvent/detergent-based virus inactivation; and two ion-exchange chromatography steps. The use of a mild pH in the downstream process makes this ICB suitable for the purification of acid-sensitive monoclonal antibodies. Integration and automation of the downstream process were achieved using the Orbit software, and the same equipment and control system were used in initial small-scale trials and the pilot-scale downstream process. High recovery yields of around 90% and a productivity close to 1 g purified antibody/L/day were achieved, with a stable glycosylation pattern and efficient removal of impurities, such as host cell proteins and DNA. Finally, negligible levels of antibody aggregates were detected owing to the mild conditions used throughout the process. The present work paves the way for future industrial-scale integrated continuous biomanufacturing of all types of antibodies, regardless of acid stability.

Raman spectrum based predictive models provide a process analytical technology (PAT) tool for monitoring and control of culture parameters in bioprocesses. Steady-state perfusion cultures generate a relatively stable metabolite profile, which is not conducive to modeling due to the absence of variations of culture parameters. Here we present an approach where different steady-states obtained by variation of the cell specific perfusion rate (CSPR) between 10 and 40 pL/(cell * day) with cell densities up to 100 × 106 cells/mL during the process development provided a dynamic culture environment, favorable for the model calibration. The cell density had no effect on the culture performance at similar CSPR, however a variation in the CSPR had a strong influence on the metabolism, mAb productivity and N-glycosylation. Predictive models were developed for multiple culture parameters, including cell density, lactate, ammonium and amino acids; and then validated with new runs performed at multiple or single steady-states, showing high prediction accuracy. The relationship of amino acids and antibody N-glycosylation was modeled to predict the glycosylation pattern of the product in real time. The present efficient process development approach with integration of Raman spectroscopy provides a valuable PAT tool for later implementation in steady-state perfusion production processes.

Recombinant protein therapeutics have become an indispensable part of modern medicine to treat a wide variety of diseases. Their manufacture is mostly based on mammalian cells with fed-batch production bioreactors and downstream batch purification unit operations. Although significant improvements have been made over the last decades to yield high production titers, concerns about their high capital investment costs and limitations in production flexibility have been raised. As a result, the biopharmaceutical industry is slowly transitioning to continuous biomanufacturing with perfusion bioreactors and integrated continuous downstream processes. The potential to maintain very high cell densities over extended process times in a perfusion culture offers higher volumetric productivities in comparison to fed-batch. This intensified production of recombinant proteins can be performed in smaller volumes, thus reducing equipment size and capital costs. However, new challenges are emerging from this shift to continuous bioprocessing. These include the lack of available scale-down models to streamline process development programs, limited knowledge on the optimization of culture medium and perfusion feed strategy, and shortage of tools for the in-line monitoring of relevant culture parameters to ensure stable perfusion operation with consistent product quality.

This thesis presents tools for the development of intensified perfusion processes with mammalian cells. Monoclonal antibody producing Chinese Hamster Ovary (CHO) cells or erythropoietin producing Human Embryonic Kidney (HEK293) cells were used as model organisms in the studies. Efforts were made to develop a small-scale perfusion system of 200 mL optimized for long-term steady-state cultivations with achievable cell densities of at least 10⁸ cells/mL. It was shown that volumetric productivities increased linearly with the cell density, thus demonstrating successful process intensification by high cell density perfusion. The small scale of this perfusion bioreactor enabled the conduction of a larger number of experiments with reduced workload and material consumption in a shorter timeframe in the following studies. To control the formation of byproducts, such as lactate, and the N-glycosylation of antibodies, which is an important quality attribute, a novel feed strategy for sugars including glucose, mannose and galactose was developed and implemented. With this feeding strategy, the sugars are delivered to match a target cell specific consumption rate, independent of the cell specific perfusion rate (CSPR). Furthermore, a high cell density perfusion process at a CSPR of 15 pL cell^-1 day^-1 was developed with high specific antibody productivity after sequential scanning of multiple steady states with various cell densities and perfusion rates. The dynamics in culture parameters from these development runs allowed the calibration of predictive models based on Raman spectroscopy, to enable real-time monitoring of multiple parameters in a perfusion culture. It was shown for example that the N-glycosylation profile was predicted with sufficient accuracy in validation experiments. Moreover, the perfusion process developed on small scale was transferred to a 30 L pilot-scale process, where it was successfully operated for 20 days. The steady state operation ensured a consistent product quality and the integrated continuous downstream process removed most impurities, while ensuring a high recovery yield. Finally, a strategy for perfusion medium optimization in CHO cell cultures operated with very low CSPR was presented by utilizing microbioreactors in combination with design of experiments methodology.

To conclude, the strategies presented in this thesis provide a new toolbox for the development and control of intensified perfusion processes, and supports the biopharmaceutical industry in their efforts to swiftly adapt continuous bioprocessing in commercial manufacturing of recombinant proteins.

Rekombinant proteinbaserade bioläkemedel har blivit en oumbärlig del av modern medicin för behandling av en mängd olika sjukdomar. Deras tillverkning är huvudsakligen baserad på däggdjursceller producerade i fed-batch odlingar i bioreaktorer, följt av nedströms reningsprocess utfärd batchvis. Trots att betydande förbättringar har skett under de senaste decennierna för att öka deras produktionstitrar, är deras höga kapitalinvesteringskostnader och begränsningar i produktionsflexibilitet problematiska. Till följd av detta övergår den biofarmaceutiska industrin sakta mot kontinuerlig tillverkning med perfusionsbioreaktorer och integrerade kontinuerliga nedströmsprocesser. Potentialen att upprätthålla mycket höga celltätheter under långa processtider i en perfusionsodling ger högre volymetrisk produktivitet jämfört med fed-batch. Denna intensifierade produktion för rekombinanta proteiner kan utföras i mindre volymer vilket minskar storlek på utrustning och kapitalkostnader. Nya utmaningar uppstår dock vid övergång till kontinuerlig process. Dessa inkluderar brist på nedskalningsmodeller för att effektivisera processutveckling, begränsad kunskap om optimering av odlingsmedium och strategier for perfusion, samt brist på verktyg för in-line monitoring av relevanta odlingsparametrar för att säkerställa stabil perfusionsproduktion med konsekvent produktkvalitet.

Denna avhandling presenterar verktyg för utveckling av intensifierade perfusionsprocesser baserade på däggdjursceller. Monoklonal antikropps-producerande Chinese Hamster Ovary celler (CHO) eller erytropoietin-producerande humana embryonala njurceller (HEK293) användes som modellorganismer i studierna. Ett småskaligt perfusionssystem på 200 mL utvecklades, optimerat för långtidsodlingar i stabilt tillstånd med celltätheter på minst 10⁸ celler/ml. Den volymetriska produktiviteten ökade linjärt med celltätheten, vilket visade framgångsrik processintensifiering med perfusion vid hög celltäthet. Tack vare den lilla storleken på perfusionsbioreaktorn, kunde ett större antal experiment med minskad arbetsbelastning och materialförbrukning genomföras inom en kortare tidsram i följande studier. För att kontrollera skapandet av biprodukter, såsom laktat, och N-glykosyleringen av antikroppar, vilket är en viktig kvalitetsegenskap, implementerades en ny feedingsstrategi för sockerarter inklusive glukos, mannos och galaktos. Med denna feedingsstrategi levereras sockerarterna för att matcha en målcellspecifik konsumtionshastighet, oberoende av den cellspecifika perfusionshastigheten (CSPR). Vidare utvecklades en perfusionsprocess med hög celltäthet vid en CSPR på 15 pL cell^-1 dag^-1 med hög specifik antikroppsproduktivitet efter sekventiell skanning av flera steady-states med olika celltätheter och perfusionshastigheter. Dynamiken i odlingsparametrar från dessa utvecklingskörningar möjliggjorde kalibrering av prediktiva modeller baserade på Raman-spektroskopi, för att möjliggöra realtidsmonitorering av flera parametrar i en perfusionsodling. Till exempel förutspåddes N-glykosyleringsprofilen med tillfredsställande noggrannhet i valideringsexperiment. Dessutom överfördes den perfusionsprocessen utvecklad i liten skala till en 30 L pilotskalaprocess, där den drevs framgångsrikt i 20 dagar. Driften i steady-state säkerställde en konsekvent produktkvalitet och den integrerade kontinuerliga nedströmsprocessen tog bort de flesta föroreningar, samtidigt som ett högt utbyte uppnåddes. Slutligen presenteras en strategi för optimering av perfusionsmedium för CHO-cellodlingar drivna vid mycket låg CSPR, genom att använda mikrobioreaktorer i kombination med Design of Experiment-metodik.

Sammanfattningsvis presenterar denna avhandling en verktygslåda med strategier för utveckling och kontroll av intensifierade perfusionsprocesser, samt stöder den biofarmaceutiska industrin i deras anpassning till kontinuerlig bioprocessteknik för kommersiell tillverkning av rekombinanta proteiner.

Cell clarification represents a major challenge for the intensification through very high cell density in the production of biopharmaceuticals such as monoclonal antibodies (mAbs). The present report proposes a solution to this challenge in a streamlined process where cell clarification and mAb capture are performed in a single step using magnetic beads coupled with protein A. Capture of mAb from non-clarified CHO cell suspension showed promising results; however, it has not been demonstrated that it can handle the challenge of very high cell density as observed in intensified fed-batch cultures. The performances of magnetic bead-based mAb capture on non-clarified cell suspension from intensified fed-batch culture were studied. Capture from a culture at density larger than 100 × 106 cells/ml provided an adsorption efficiency of 99% and an overall yield of 93% with a logarithmic host cell protein (HCP) clearance of ≈2–3 and a resulting HCP concentration ≤≈5 ppm. These results show that direct capture from very high cell density cell suspension is possible without prior processing. This technology, which brings significant benefits in terms of operational cost reduction and performance improvements such as low HCP, can be a powerful tool alleviating the challenge of process intensification. 

The N-linked glycosylation pattern is an important quality attribute of therapeutic glycoproteins. It has been reported by our group and by others that different carbon sources, such as glucose, mannose and galactose, can differently impact the glycosylation profile of glycoproteins in mammalian cell culture. Acting on the sugar feeding is thus an attractive strategy to tune the glycan pattern. However, in case of feeding of more than one carbon source simultaneously, the cells give priority to the one with the highest uptake rate, which limits the usage of this tuning, e.g. the cells favor consuming glucose in comparison to galactose. We present here a new feeding strategy (named ‘TAFE’ for targeted feeding) for perfusion culture to adjust the concentrations of fed sugars influencing the glycosylation. The strategy consists in setting the sugar feeding such that the cells are forced to consume these substrates at a target cell specific consumption rate decided by the operator and taking into account the cell specific perfusion rate (CSPR). This strategy is applied in perfusion cultures of Chinese hamster ovary (CHO) cells, illustrated by ten different regimes of sugar feeding, including glucose, galactose and mannose. Applying the TAFE strategy, different glycan profiles were obtained using the different feeding regimes. Furthermore, we successfully forced the cells to consume higher proportions of non-glucose sugars, which have lower transport rates than glucose in presence of this latter, in a controlled way. In previous work, a mathematical model named Glycan Residues Balance Analysis (GReBA) was developed to model the glycosylation profile based on the fed carbon sources. The present data were applied to the GReBA to design a feeding regime targeting a given glycosylation profile. The ability of the model to achieve this objective was confirmed by a multi-round of leave-one-out cross-validation (LOOCV), leading to the conclusion that the GReBA model can be used to design the feeding regime of a perfusion cell culture to obtain a desired glycosylation profile.