From chimeric antigen receptor (CAR) T cells to CD34+ stem cells, cell therapies save thousands of people’s lives. Made of immune cells carefully isolated from patients with cancer and engineered into lifesaving treatments, cell therapies are becoming more and more common in the clinic.

While cell therapies hold enormous potential for treating diverse conditions, from different kinds of cancer to autoimmune diseases, manufacturing enough of them is difficult. Often, when new cell therapy companies spin out of academic research labs, they need to scale up their cell therapy manufacturing capabilities quickly to make enough of their drug to run a clinical trial, for example. However, that fast manufacturing scale-up is not easy.

I would challenge anybody to tell me that we should have conversations with a husband, a wife, children about their various family members: ‘I'm sorry. I can't treat you. I can't treat your family member because it's just too expensive.’ That's not going to be a conversation that any of us should be having. 
- Matthew Hewitt, Charles River Laboratories

“We have challenges in terms of recruiting people to do the manufacturing, retaining them, especially in the Bay Area, where there's a lot of competition,” said Jonathan Esensten, an immunologist at Sheba Medical Center and formerly of the University of California, San Francisco (UCSF). 

The need for highly trained people to create these therapies also drives up the manufacturing cost, which in turn increases the price of the therapies themselves, limiting patient access. “I would challenge anybody to tell me that we should have conversations with a husband, a wife, children about their various family members: ‘I'm sorry. I can't treat you. I can't treat your family member because it's just too expensive.’ That's not going to be a conversation that any of us should be having,” said Matthew Hewitt, who is the vice president and technical officer of cell and gene therapy and biologics at Charles River Laboratories.

The other problem with producing large amounts of cell therapies is that many of the steps needed to create them require researchers to manually transfer cells from one vial to another, which can introduce batch variability as well as an increased risk of microbial contamination. “Anytime humans touch things in good manufacturing practices (GMP), it comes with a cost,” Hewitt added. “This is not related to anybody. It's just a general risk with humans.”

Now, a team of engineers at the biotechnology company Multiply Labs may have found a way to both decrease risks of contamination and batch effects while also removing the need for an expensive labor force. Their solution? Robots.

“People always say that robots are very useful for the ‘three Ds,’ which are dull, dirty, and dangerous jobs,” said Fred Parietti, a mechanical engineer as well as the cofounder and chief executive officer of Multiply Labs. But, he added, the main difference for pharmaceutical manufacturing is that it’s “not dirty at all. The opposite: it must be sterile.”

With their robotic system, the team at Multiply Labs hopes that by helping researchers scale up their production of high-quality cell therapies, more patients will benefit from these lifesaving treatments.

Lending a robotic hand (or arm)

When he was a graduate student at the Massachusetts Institute of Technology (MIT), Parietti wanted to find a way to apply robotic technology to help as many people as possible. “I never particularly liked military applications,” he said. “I also don't like consumer robots because those are too simple. Carpet cleaning robots — that's too boring.” 

While he was at MIT, he met Alice Melocchi, a chemical engineer and now cofounder and chief scientific officer at Multiply Labs, who told him how difficult it was to manufacture drugs. “She really opened my eyes that pharmaceutical companies typically focus so much on discovering new therapies because the biology is so complicated,” he said. “That has left pretty much a fantastic, I think, opportunity for robots to be useful, for engineers to be useful, in the pharmaceutical space,” he said.

There are multiple examples of automation in drug manufacturing, but most of those systems perform one specific set of procedures. Adding new features depends on whether the company that makes that machine decides to do that.

“With robotics, you have the ability to put together not one all-in-one machine, but several different machines that each do one thing really well and control them with a robot,” said Esensten, who led the UCSF team that collaborates with Multiply Labs for cell therapy manufacturing.

This modular set up is exactly how Multiply Labs’ robots work. Large three-foot long cubes sit stacked on top of one another as well as side by side. Between two rows of these stacked cubes sits a robotic arm that can move to each of the cubes along a metal rail. Inside each of the different cubes sits an instrument or machine that’s involved in one particular step in the therapy production process. Because all of the cubes and the robot arm reside inside a single closed system, the samples remain in a sterile condition throughout the entire time the robot is working. 

The robots are precise; they use syringes to collect tiny amounts of liquid and record timestamps automatically. But most importantly, once they’re programmed, they complete the same task the same way every single time.

Robots do, however, have trouble when asked to pick up something that has a flexible or soft structure, so handling bags or soft tubing is not ideal. For those cases, Parietti and his team developed rigid cartridges and cassettes for the robots to grab onto and move around from one module to another. The robots can already make small molecule drugs, and now the Multiply Labs team is extending their capabilities to cell therapies.

Renske ten Ham, a health economist who specializes in gene and cell therapy development at University Medical Centre Utrecht and who is not associated with Multiply Labs, thought that using robots to automate cell therapy manufacturing was a compelling step forward.

“It's an interesting development, and it's a next step in automation —  not just having one part, but stringing them all together,” she said. “I always wonder what the people who assess the quality and the safety and the reproducibility and the inspectorate, for example, think of these kinds of innovations. Because you can imagine that if you have a manufacturing line going, and you have your products licensed and everything, then introducing this would have to require you to either go a few steps back or get regulators involved to make sure that you're not out of specs on a lot of things.”

For the Multiply Labs team, the modular nature of the robots addresses these concerns. “These instruments are already used today in manufacturing cell therapies, so if you're already using four or five instruments, and that's your process, you don't need to change it. You just need to select essentially the right number of modules to replicate the same set of instruments, but now they're operated by the robots,” said Parietti. “Our approach to automation is always very respectful of the biology team, so essentially, we adapt the robots to the biology. We don't ask them to adapt biology to the robots.”

In theory, there should be no difference between cell therapies produced by a robot or a human worker, but of course, the Multiply Labs team needed to prove that.

Robots versus humans

While the engineers at Multiply Labs had robotics expertise, they needed collaborators with expertise in cell therapy manufacturing. In addition to Esensten’s team at UCSF, researchers at Cytiva, Thermo Fisher Scientific, and Hewitt and his colleagues at Charles River Laboratories joined Multiply Labs’ Robotic Manufacturing Consortium. The team from Cytiva focused on integrating automation of bioreactors with the robotic system; the Thermo Fisher Scientific team helped with incubator automation, and the experts from Charles River Laboratories worked on quality control testing automation.

Our approach to automation is always very respectful of the biology team, so essentially, we adapt the robots to the biology. We don't ask them to adapt biology to the robots. 
– Fred Parietti, Multiply Labs

“The goal was to do a relatively simple task, which was to expand primary human T cells. So, to take cells and to stimulate them and then to grow them in culture for 10 or 12 days,” Esensten said. “We considered many different possible things that we could demonstrate, and in the end, we chose something very simple because when you have something that's so new, there are so many variables at play. We wanted to keep the biology as simple as possible.”

While the robots easily interacted with the instruments in each of the modules and transferred liquids between them, they had trouble with one specific task that researchers in the lab perform without even thinking about it: resuspending cells in a flask. “You just know how to move your hand in a way that will put all those cells into a homogeneous suspension. For a robot, you can't just say, ‘Oh, just swirl.’ There are many parameters that the robot has to be taught to do,” Esensten explained.

The engineers started by programing the robots to move flasks back and forth with a simple shaking motion, but that didn’t work. They then had it shake harder, but the cells stayed firmly put at the bottom of the flask. “Then we realized this is not what people do. People don't just shake in one direction as hard as possible. It's a much more gentle motion, but a much more complicated motion with multiple movements at the same time,” said Parietti.

In the end, the Multiply Labs team took a cue from the movies. They had the scientists wear motion capture gloves as they moved the flasks so that they could map the movement and program the robots to move in the exact same way.

With the resuspension problem solved, the team found that there was no difference in whether a human or a robot expanded the human CD8+ T cells, which they reported in a new paper (1). Cell yield, viability, and phenotype of the expanded cells were all comparable between both the manual and robot manufacturing approaches.

“Was that surprising to me as a scientist? Not really, but we needed to show that formally,” said Esensten. 

Hewitt echoed the sentiment. “If you start showing wildly different measures coming out of it, then we start to have questions about why we can't recreate the data that we were able to do manually,” he said. “I was happy to see that it was as boring as we would have hoped.”

One clear advantage of the robotic system, though, was evident in its ability to keep the cells free of microbial contamination. The one contamination case that occurred during the course of the study happened during the manual process, not the robotic one.

While this paper showed that the robots can expand T cells just as well as humans can, Parietti said that’s just the first step. During the rest of 2024, the Multiply Labs team plans to bring the robots to different pharmaceutical companies so that researchers can compare the robotic versus manual manufacturing process in their own facilities. 

“This year basically is the year of all these pilot demonstrations, showing equivalence not just in [cell] expansion like in the [paper], but in the complete end-to-end process on a variety of processes,” said Parietti. “If this year goes well, then next year, that paves the way for actual commercial deployment. Because after the full end-to-end process works on the robots, then people have enough data to say, ‘Okay, it works. Now we can use it to actually make drugs for patients.’”

The rest of the consortium members, including Esensten and Hewitt, are excited to continue working with the Multiply Labs team to move these robotic systems forward. Parietti is already thinking about how robotics can improve the manufacturing of other important treatments.

“There are so many new therapeutic modalities. There are gene therapies. There is mRNA, and I think all of them need automation,” said Parietti. “As biology progresses, there are so many new therapies, and we find all of them would really benefit from automation to reach more patients more affordably. To me, this is very very exciting.”

Reference

1. Melocchi, A. et al. Development of a robotic cluster for automated and scalable cell therapy manufacturing. Cytotherapy (2024).