Robotic search for optimal cell culture in regenerative medicine
Curation statements for this article:-
Curated by eLife
Evaluation Summary:
The manuscript by Kanda GN, Natsume T et al. describes a robotic artificial intelligence system with a batch Bayesian optimization algorithm that allows to optimise and reliably repeat cell culture protocols. The authors utilise induced pluripotent stem cell-derived retinal pigment epithelial cells as a model culture system of broad interest in regenerative medicine. They demonstrate that the robotic system with Bayesian algorithm accelerates the optimisation of cell culture protocols and increases the quality and quantity of cell products, compared with manual operations - these results will likely inform and strongly impact modern cell culture strategies in regenerative medicine.
(This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1, Reviewer #2 and Reviewer #3 agreed to share their name with the authors.)
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (eLife)
- Computational and Systems Biology (eLife)
Abstract
Induced differentiation is one of the most experience- and skill-dependent experimental processes in regenerative medicine, and establishing optimal conditions often takes years. We developed a robotic AI system with a batch Bayesian optimization algorithm that autonomously induces the differentiation of induced pluripotent stem cell-derived retinal pigment epithelial (iPSC-RPE) cells. From 200 million possible parameter combinations, the system performed cell culture in 143 different conditions in 111 days, resulting in 88% better iPSC-RPE production than that obtained by the pre-optimized culture in terms of the pigmentation scores. Our work demonstrates that the use of autonomous robotic AI systems drastically accelerates systematic and unbiased exploration of experimental search space, suggesting immense use in medicine and research.
Article activity feed
-
-
Evaluation Summary:
The manuscript by Kanda GN, Natsume T et al. describes a robotic artificial intelligence system with a batch Bayesian optimization algorithm that allows to optimise and reliably repeat cell culture protocols. The authors utilise induced pluripotent stem cell-derived retinal pigment epithelial cells as a model culture system of broad interest in regenerative medicine. They demonstrate that the robotic system with Bayesian algorithm accelerates the optimisation of cell culture protocols and increases the quality and quantity of cell products, compared with manual operations - these results will likely inform and strongly impact modern cell culture strategies in regenerative medicine.
(This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private …
Evaluation Summary:
The manuscript by Kanda GN, Natsume T et al. describes a robotic artificial intelligence system with a batch Bayesian optimization algorithm that allows to optimise and reliably repeat cell culture protocols. The authors utilise induced pluripotent stem cell-derived retinal pigment epithelial cells as a model culture system of broad interest in regenerative medicine. They demonstrate that the robotic system with Bayesian algorithm accelerates the optimisation of cell culture protocols and increases the quality and quantity of cell products, compared with manual operations - these results will likely inform and strongly impact modern cell culture strategies in regenerative medicine.
(This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1, Reviewer #2 and Reviewer #3 agreed to share their name with the authors.)
-
Reviewer #1 (Public Review):
The manuscript by Kanda GN, Natsume T et al. describes the development of a robotic artificial intelligence system with a batch Bayesian optimisation algorithm with the aim to optimise and reliably repeat cell culture protocols in an automated fashion.
The authors successfully achieve the overall goal of demonstrating the capacity and potential of the robotic system for protocol optimisation, reliable repetition and efficient cell product generation.
The major strength of the study is the advanced laboratory automation system developed, with seemingly superior flexibility, precision and efficiency to generate advanced cell products, compared with other systems described previously, or compared with manual handling.
The authors utilise induced pluripotent stem cell-derived retinal pigment epithelial cells as …
Reviewer #1 (Public Review):
The manuscript by Kanda GN, Natsume T et al. describes the development of a robotic artificial intelligence system with a batch Bayesian optimisation algorithm with the aim to optimise and reliably repeat cell culture protocols in an automated fashion.
The authors successfully achieve the overall goal of demonstrating the capacity and potential of the robotic system for protocol optimisation, reliable repetition and efficient cell product generation.
The major strength of the study is the advanced laboratory automation system developed, with seemingly superior flexibility, precision and efficiency to generate advanced cell products, compared with other systems described previously, or compared with manual handling.
The authors utilise induced pluripotent stem cell-derived retinal pigment epithelial cells as a proof-of-principle, model culture system. Using this system, they convincingly demonstrate the potential of this automated platform for multiple cell culture applications of broad interest in regenerative medicine.
The work presented here demonstrates the successful implementation of laboratory automation for a cell culture product used for regenerative medicine purposes. LabDroid Maholo robotic platform integrated advanced robotics capabilities to perform precise handling of equipment to replace skilled human operations in a repeatable and error-free manner. Integrated AI-based image analysis with batch Bayesian optimisation algorithm helped to identify required process changes to the cell culture protocol to improve the differentiation of induced pluripotent stem cells (iPS) cells to retinal pigment epithelial cells (RPE).
-
Reviewer #2 (Public Review):
Laboratory automation in life science laboratories to replace complex operations performed by skilled scientists with mechanisation or automation remains a major technical challenge. Since most of the cell culture process steps remain as an art and the parameters are undecoded to be amenable to a robotic system to operate, complex cell culture protocols remain as manual processes. Recent developments in laboratory automation advanced to perform simple process steps like cell culture media addition or removal, cell seeding and passaging using robotic platforms. Determination of cell confluence and assessment of the level of differentiation typically needs to be conducted offline and the process parameters to be fed into the robotic system manually to command further steps.
Natsume et.al. demonstrated by this …
Reviewer #2 (Public Review):
Laboratory automation in life science laboratories to replace complex operations performed by skilled scientists with mechanisation or automation remains a major technical challenge. Since most of the cell culture process steps remain as an art and the parameters are undecoded to be amenable to a robotic system to operate, complex cell culture protocols remain as manual processes. Recent developments in laboratory automation advanced to perform simple process steps like cell culture media addition or removal, cell seeding and passaging using robotic platforms. Determination of cell confluence and assessment of the level of differentiation typically needs to be conducted offline and the process parameters to be fed into the robotic system manually to command further steps.
Natsume et.al. demonstrated by this work that the cell culture techniques conducted by skilled scientists can be replaced with laboratory automation. Work revealed that an effective combination of robotics and artificial intelligence can replace complex human operations of cell culture techniques with robots even in a complex cell differentiation process. While other similar studies demonstrated the use of robotics in cell culture protocol to mechanise repetitive process steps or liquid handing this work utilised an advanced robotic system with precision movements and the capacity to perform complex tasks required in a cell culture process. The robotic arms undertake several activities required in a cell culture lab with comparable precision to human counterparts yet with high repeatability of a robot. The capability of the robot to handle a micropipette, small tubes or using an incubator demonstrates its refined mechanical abilities. An effective combination of artificial intelligence with robotics enabled this platform to conduct analytics in real-time and the algorithms predicted the best process parameters to be used to improve the cell culture. Cell culture images generated by the integrated imaging suite within the Maholo system were analysed to calculate the rate of differentiation of iPS cells into RPE cells. In general, the integration of all necessary systems in a single platform with robotic arms that can perform refined activities like a human made the system highly suitable for the application.
This work showed that the LabDroid Maholo system is superior compared to the existing systems used in bioscience and cell culture applications. It is evident from the presented data that the system operates with high precision and repeatability with less chance of errors to generate differentiated RPE cells. Since the rate of differentiation consistently increased with the optimisation predicted by the Bayesian algorithm cell culture system which ensures the reliability of the system for this purpose. Iterative process improvements were undertaken based on the Bayesian algorithm resulting in an increased rate of differentiation. Results obtained from the screening experiments were validated multiple times with statistical significance, which proves the reliability of the Bayesian algorithm and robotics.
Although the LabDroid Maholo system presents a huge capability to replace several activities of a skilled scientist, the work presented here was a rather well-suitable process to exploit the capabilities of this highly precise robotic system. Despite the robot's ability to undertake complex laboratory tasks, an initial set of work including the cell seeding was conducted manually before it was suitable for a robot to operate. Although it is not far away from assigning such activities to a robot to conduct, it needs to be done to make the full process to be automated.
It is noteworthy that process parameters used in this work included duration of trypsinization, pipetting strength and speed along with the concentration of differentiation agents and their exposure time. With such parameters alone, it is not realistic to claim that the system can be utilised to optimise the cell culture or differentiation programme. Since a differentiation programme is developed by comparing against or modelling an in-vivo system, a lot more elements of that real biological environment and complex signalling mechanism need to be taken into consideration. Incorporation of such biological information into the presented robotic platform is impractical. In that sense, searching for optimal cell culture conditions must be considered within the limits of the existing system. Thus it may be argued that the capacity of the Maholo LabDriod system is limited to a pre-defined set of process parameters in terms of its process optimisation capacity.
Maholo LabDriod system utilises a precise robotic system with analytical power but here some components and operations remained traditional e.g. usage of manual micropipettes and especially to use the same pipette multiple times to dispense adequate volume. This seemed to be a combination of the most advanced system with an old tool. Dedicated electronic pipettes with defined volumes would have been used instead. While the process utilised here presented with a visible differentiation marker (melanin expressing cells upon differentiation) made it possible to measure differentiation and adjust the process parameters accordingly. In other cell culture products, where no such visible markers are available to determine differentiation, immunostaining or flow cytometry or other ways of determination will be required. In such cases, the LabDroid system might not work as efficiently as in the work presented here.
Although it was not directly within the scope of this work, another major aspect that is not considered in this study is the mode of cell culture. Due to the requirement of the large scale of cells required for regenerative medicine purposes, suspension-based cell culture replaces adherent cell cultures in several products. It would be interesting to see how the LabDroid Maholo system performs with suspension-based cultures. Although the Batch Bayesian optimization predicts millions of potential combinations of parameters, it is not clear how the selected 48 conditions provided the best differentiation conditions. It is not clear if the study did address selecting the parameters based on the Design of Experiment principle and if the selected parameters cover all necessary process parameter combinations.
Overall, the Maholo LabDroid system brings in more capable robotic control to mechanise and automate highly skilled human operations in the cell culture process in comparison to previously used robotic systems elsewhere. The cell culture imager and AI-based analytical capability provide options to identify process parameters for optimization and improved differentiation process.
It is fully appreciated that the robotic system is potentially the most advanced one used for cell culture protocols. The way it handles micropipettes and uses both arms to dispense liquid or to open the incubator etc. provides a view about how this type of robot is going to emerge. The study demonstrates the effective use of a combination of robotics and artificial intelligence to replace complex human operations of cell culture techniques with robots. Integrated AI could successfully analyse the state of cell culture to identify cell differentiation markers and enabled live modifications to the cell culture system to improve differentiation. Certainly, the Maholo LabDroid system brings in more capable robotic control to mechanise and automate highly skilled human operations in comparison to previously used robotic systems elsewhere.
-
Reviewer #3 (Public Review):
In this article authors have developed a robotic system coupled with Artificial Intelligence (AI) to autonomously differentiate hiPSC into RPE. The goal is to reduce human errors and optimise culture conditions for clinical applications.
The paper is well written and experiments are rigorously conducted. The development of such approaches and protocols would clearly have a great impact on the quality of future cell therapy clinical applications and are developed by many groups demonstrating the need.
-
-