Improving Upstream Bioprocessing Efficiency with Digital Tools
Digitalization holds great promise for increasing the efficiency, quality, and reproducibility of upstream bioprocessing. From automating a bioprocess run, process optimization through modeling and statistical methods, digital solutions for equipment maintenance, and cloud-based data handling, digitalization can benefit many facets of bioproduction.
An eBook produced in partnership between Eppendorf and GEN (Genetic Engineering & Biotechnology News) explores the value and adoption of bioprocess digitalization with articles on the people behind the digital tools, examples of process automation and how AI can transform bioprocessing workflows.
People are the Key to Digital Evolution of Biopharma
There is a clear trend towards the adoption of digital manufacturing in the biopharmaceutical industry, with experts citing globalization, supply chain complexity, price, and cost pressure as key drivers for embracing this innovation. Véronique Chotteau, PhD, of Sweden’s Royal Institute of Technology (KTH) cell technology group, says the newfound willingness to invest in innovative manufacturing technologies reflects the emergence of a new generation of executives who understand their benefits for complex cell-based bioproduction. As well, increasing interest in continuous bioprocesses require advanced tools for on/in-line monitoring (process analytical technology) and feedback controls compared to legacy batch operation.
However, Chotteau recognizes there is a shortage of staff with digital manufacturing expertise, which could slow momentum particularly with more challenging modalities like cell and gene therapy. Overcoming siloed knowledge across disciplines and increasing academic support in bioprocessing will go a long way to addressing the personnel shortfall in years to come and increase competency across the digital space in biomanufacturing.
Bioprocess Feed Automation Enabled through Software Scripts
In cell-based bioproduction, optimizing the culture feeding is critical to maintaining a high volumetric productivity and is a central part of upstream bioprocess development. In this article, author Ulrike Rasche of Eppendorf’s SE Bioprocess Center in Germany discusses how to automate culture feeding with digital technologies to improve bioprocess efficiency.
Fed-batch fermentation is the most common mode of operation in the bioprocess industry where cells or microorganisms are grown in bioreactor systems and nutrients are incrementally added throughout the duration of the culture period to keep cells in exponential growth phase. Substrate limitation typically dictates the start of culture feeding but it is important that the substrate is not added in excess to limit unwanted by-product accumulation while fully supporting cell growth.
Classically, investigators must sample the bioreactor at designated time intervals to assess the metabolism of the cells to determine the start of feeding but this analog method is laborious and prone to failure if time points are missed, which can negatively impact the culture’s productivity. Implementing an automated feeding process offers numerous advantages:
- Prevent nutrient depletion
- Generate a more stable macro-environment
- Reach higher product yields
- Reduce manual workload
- Improve standardization
To effectively implement automated feeding, a device that drives the feed solution in the bioreactor automatically and programmable bioprocess software (for example DASware® control) are needed. Typically, a feed pump that is driven by the bioprocess control software with scripting functionality. Software scripts are valuable tools to implement precise bioprocess control routines to enable process automation under user-defined conditions. The author presents two software script strategies designed to optimize the culture feeding process: time-based feeding and sensor-based feeding.
Strategy 1: Time-based feeding
Here, a defined volume of feed solution is added to the culture at a specific time increment in the culture period. For instance, exponential feeding aims to extend the exponential growth phase for an actively growing culture to achieve higher biomass density, while effectively avoiding nutrient depletion and toxic byproduct buildup. The feed rate is described by the script text entered into the bioprocess control software containing a formula defining the exponential feed profile.
Strategy 2: Sensor-based feeding
In a sensor-based feeding strategy, feed media is added to the culture based on an inline or online sensor reading that is fed into the bioprocess control software to create an automated feedback loop. Based on sensor value, the software script activates an actuator, which controls the process parameter of interest at setpoint. Different parameters can be used to trigger automated feeding regimens including:
- Feeding based on a dissolved oxygen (DO) spike
- Feeding based on substrate concentration (i.e., glucose)
- Feeding based on the respiratory quotient (the quotient of CO2 produced and O2 consumed by a culture)
Details on how to implement these strategies is described in the article and how software scripts can be implemented to create customized process control.
PID controlled Constant RQ Fermentation of Pichia pastoris in the DASbox® Mini Bioreactor System
Pichia pastoris is a commonly used protein expression system in biomanufacturing with advantages over Escherichia coli because of its ability to carry out human-like post-translation modifications like glycosylation. Additionally, compared to mammalian cell lines like CHO, P. pastoris fermentation offers a faster growth rate and uses less complex (and expensive) culture media, which makes it ideal for large-scale bioreactor production. One of the key strategies to optimize production yield is to fine-tune the feeding strategy. A convenient and well-established sensor-based approach triggered by a dissolved oxygen (DO) spike in the culture media is frequently used but new research indicates that feeding based on respiratory quotient (RQ) can provide better productivity while avoiding unwanted ethanol formation.
This article presents a study demonstrating the feasibility of the constant RQ-based feeding using the Eppendorf DASbox Mini Bioreactor System compared against a DO spike-based strategy. A GA4 exhaust gas analyzer unit including humidity sensors was connected to the bioreactor to measure oxygen (O2) and carbon dioxide (CO2) concentration. An automatic RQ based PID feeding script on the DASware control 5 was used to trigger feeding once the RQ value drops below 1, indicating that the glucose has been consumed by the culture.
Compared to DO-based feeding, feeding control based on constant RQ of P. pastoris fermentation proved to be more time and cost effective allowing for rapid feed optimization with minimal byproduct formation.
Stars in Alignment for Artificial Intelligence in Bioprocessing
The final article of the eBook examines the use of artificial intelligence (AI) for process development and biomanufacturing control. In the manufacturing plant, AI can be used for laboratory automation, efficient document processing, and process control driven by the increased interest (and adoption) of continuous manufacturing initiatives. According to Vikas Revankar, head of Software and Automation at MilliporeSigma, data-driven continuous operations will help to significantly improve product quality, reduce production costs and shorten the time to market.
Because AI and machine learning (ML) technology can detect patterns in complex datasets that are difficult to observe for an operator, it provides new avenue to gain insight into production processes that can be used to improve and/or automate them. While biotechnology often lags behind other sectors in embracing digital tools, the use of process analytical technology (PAT) is well-established, paving the way for AI and ML adoption. Many companies already have the information technology infrastructure in place to support PAT that can be leveraged to implement AI.
In order to realize the full potential of these technologies, access to data to train AI mathematical models and clearer regulatory expectations are needed. According to Jens Smiatek, PhD, an AI and data management expert at the University of Stuttgart, the main challenges for ML- and AI-based methods are missing guidelines to extend their application into GMP environments. There will also be a requirement for personnel to cultivate different skillsets to adapt to these new tools and companies must be cognizant of making them user-friendly to capitalize on their benefits. All of the components are in place to bring about an AI revolution in bioprocessing.
To download a copy of the eBook, please visit: Upstream Bioprocessing – Improving Efficiency through Digital Tools