messenger RNA

The art and science of taking the RNA scale-up recipe from the lab to a manufacturing plant

“There’s a difference between baking a normal loaf of bread and a gigantic loaf 10,000 times bigger,” says Adam Shanebrook, a biotechnologist and pilot plant director at GreenLight Biosciences. “You’ll burn the outside and leave the inside uncooked.” His point: You can’t simply scale up chemical and biological processes. You have to be clever about it.

That is very much on the mind of researchers at GreenLight as they scale up their manufacturing from laboratory to commercial production for their RNA-based solutions targeted at Colorado potato beetles, a crop pest, and varroa mites, which decimate bee colonies. The company’s innovative cell-free RNA synthesis was shown to work at laboratory scale some years ago in Medford, Massachusetts, but GreenLight has opened a pilot plant at an old Kodak factory in Rochester, New York to make it work at a far greater scale.

Producing RNA in a lab or a factory typically involves gene-editing a cell, often a harmless bacterium. But the process doesn’t just manufacture RNA, it also makes the rest of the bacterium. That means that it is wasting a lot of energy and raw materials in making unwanted byproducts. What’s more, those byproducts are mixed in with the RNA you actually want, so you have to either separate them—an expensive and difficult process—or deal with an impure final product, full of debris left over from the cells.

The cell-free process

The cell-free process works rather differently. It uses harmless strains of E. coli—long used in the pharmaceutical industry to make things like human insulin—to make the starting materials for the cell-free reaction. To prepare for that, short rings of DNA called “plasmids” are made to provide assembly instructions for the cell-free RNA production step. Additionally, E. coli are used to make the enzymes that catalyze the production and assembly of the RNA molecule.

Then those products are put into a large vat containing a “soup” of nucleosides, the building blocks of RNA and DNA, and the enzymes pluck the floating nucleosides out of the soup, energize them so that they snap together like magnets, and assemble them in a line according to the DNA plasmids’ instructions. This has the advantages that it is faster than bacterial-cell production—perhaps 10 times faster, requiring about three hours per batch as opposed to 30 or so; it is cheaper; and it brings forth a far purer product, approximately 50-65% pure.

By 2017, that was all shown to work in the lab. But although it might be easy to make an 8-inch baguette in a home oven, it doesn’t mean you could successfully make a 65-foot-long loaf in a gigantic mega-oven.

Building up to a 1,250-litre reactor

Karthik Ramachandriya is a biochemical engineer and process development director at GreenLight. He holds up a small object, the size of a pen lid. “The first experiments were done in a 50-microlitre reactor,” he says—that is 0.05 millilitres, a barely visible quantity. “Then we scaled that up to 250ml.” He holds up a coffee cup. “It wasn’t that challenging, going from a microreactor to a coffee cup.” Then the team tried it in a bench-scale reactor that contained 10 litres. Finally, in 2019, the company started to build the Rochester plant, trying a 150-litre reactor and eventually a 1,250-litre one.

“We figured out how to do it all at the bench scale in Medford,” says Eric Otto, a biotechnologist and GreenLight’s vice president of manufacturing. But at both stages—making the plasmids and enzymes, and making the RNA—there were challenges associated with scaling up.

With enzyme and plasmid production, “our target audience was the good E. coli that we’re trying to make happy,” Otto says. “It’s like brewing beer, but a bit more complicated. You need to maintain what the organism needs and provide sufficient oxygen and other nutrients while removing the carbon dioxide so the fermentation doesn’t get sick.”

But maintaining those conditions is a different job in a 150-litre tank than it is in a one-litre one. “In the lab, you shake it in a tube,” says Shanebrook. “But at a manufacturing plant you can’t just shake it up when it’s a whole tank.”

“The mixing is different,” agrees Erin Cobb, a process engineer. “It’s not instantaneous any more. It’s not perfectly uniform. How you’re delivering oxygen to the cells is different; it’s easy at a small scale using bottled oxygen, but that is not economical when you scale up, and it’s a combustion risk, so you have to use pressurized air.” You can simply shake a one-litre flask to get all the nutrients to where they need to be, and the whole thing will automatically be at a uniform temperature. 

Pumps, fork trucks, and conveyor belts

Cobb adds that there are also logistical issues. If you’re adding a kilo of salt to a reactor, she says, “I can lift that with my hand. Can I do that with something 10 times bigger? I can pick up the 10-litre tank and dump it out; I can’t with the 1,000-litre. I need pumps, fork trucks, conveyor belts.”

Even prosaic tasks like cleaning become more complicated. “The spatula that I scooped the raw material into the small tank with,” she says: “I can take it over to the sink, wash it in the autoclave,” a machine that sterilizes equipment with steam. “You can’t put a 1,000-litre tank in an autoclave. You need a whole sterile system, you have to heat it to 120°C, you need sterilized air going into the system.”

There’s also a human resources issue. As the process becomes larger and reliant on many more skillsets, you need different kinds of labor. Setting up the Rochester plant was a case in point. “You need to know about utilities, fire alarms, sprinklers, heating, ventilation, floor drains,” says Shanebrook. “You’re dealing with hundreds of people in skilled trades. People who fix the roof, coat the floor.” 

Within GreenLight itself, there are many brands of expertise—engineers, biotechnologists, chemists, entomologists, agronomists—all of whom need to understand each other’s specialty to some degree. “The most important thing in scaling up and scaling down is understanding the people who have worked on the process, being really comfortable asking them questions,” says Cobb. “Can I call anytime and ask, does this look weird? Is it supposed to look like this?” 

A plant that can scale up

There are other complications. Although the Rochester plant is a large targeted RNA manufacturing plant, it is itself a pilot for a much bigger facility—hundreds of times larger, 20,000 litres or more. So there’s no point solving a problem at a 150-litre scale if it wouldn’t then work at a much greater one. For instance, it might be easy to find ways to remove waste products from a one-litre tank, so that the bacteria can thrive. But unless the solution would also work at a much larger scale, then it’s not a solution at all. The pilot plant is designed as far as possible to mimic the realities of a larger, commercial plant. “You need to understand the limitations of a 20,000-litre reactor when we’re developing a process at a one-litre reactor,” says Ramachandriya.

“This is our playground,” says Shanebrook. “In the commercial plant, it’ll be designed for purpose: your pump goes from A to B, it’s dedicated to this function.

“But in the pilot plant, you don’t know what you need, and you need to accept that. You need to be flexible, use hoses that go from A to B, or A to C, or D to A. You take the changes in stride.”

When you do meet a problem in the pilot plant, says Ramachandriya, what’s important is going back to the smaller scales and understanding it properly. “If it’s a biological function, we go back to the tiny scale,” he says. “If it’s a process function, we do it at the coffee-cup or the one-litre size.” If your E. coli are unhappy and producing too much acid, you can probably explain that in the microreactor. If your pipes are clogging because the RNA solution gets too viscous, you’ll need the larger reactors, with their own pipes and pumps, to see where it’s going wrong.

GreenLight’s reactor energizes nucleotiodes

What’s also important as you scale up, says Ramachandriya, is keeping costs down. For instance, when you’re ordering thousands of litres of nucleotide solution or nutrients for your E. coli, you don’t want to be spending too much per litre. The nucleotides are made from yeast byproducts and energized by enzymes in GreenLight’s own reactor; they can be bought in bulk. “The yeast industry is very large and produces a variety of products from its specialized process, and we gain from the economies of scale,” says Otto.

Shanebrook adds that the scaling project isn’t done yet. You have to foresee future demand: if, as GreenLight hopes, its products become widely used, then even the planned 20,000-litre reactor will not be enough. “It’s reading the tea leaves and planning for expansion,” he says. “As the manufacturing person, you don’t want to be the limiting step. If the business side says, ‘We want a bigger plant,’ you want to be able to say, ‘We have a plan for this;’ it’s a common problem that business says we want this next month, and the technologists say it’ll take two years.”

At some point, scaling up further in a single plant becomes impossible. “In baking,” says Shanebrook, returning to his earlier metaphor, “sometimes you reach a scale, X number of tons, where you can’t scale up any more. Instead you number up.” That is, you build more plants, doing the same thing. It has advantages of redundancy in case of disaster.

GreenLight is not at that stage yet. But Shanebrook, in particular, is already thinking about ways to continue the scale-up process. At the moment, at each scale—whether 50-microlitre or 1,250-litre—the RNA is produced in batches: a reactor is filled with raw materials, left to do its thing, and the final product is harvested. But perhaps the future is continuous production: “You can put all that stuff in the pot of soup for three hours,” he says, “or you can have a long pipe that takes three hours to flow from the input to the output.” 

The ability to run continuously is a scenario for the future, he hopes. But the scale-up has already been spectacular, dealing in hundreds of thousands of times the volume the GreenLight team did just a few years ago. 

“It’s sort of like I sat with the home chefs,” says Cobb, “and then learned to cater the wedding.”