Making biologics on demand

As a U.S. Army doctor stationed in Afghanistan in 2003, Geoffrey Ling had a hard time accessing medicines his patients needed. “The combat support hospital was nothing more than a tent city,” he says. “When you’re in a situation like that, you have limited supplies.”

In particular, biopharmaceuticals, like insulin, were hard to obtain and store.

“I wanted to have a machine that could make any drug at any time in any quantity for providers in situations like I was in, austere situations where they have to take care of patients,” Ling says. “Instead of carrying around boxes and boxes of drugs that they may never use, they would just carry some basic reagents and make what they needed.”

When he joined the U.S. Defense Advanced Research Projects Agency (DARPA) as a program manager, he found himself in a position to do something about it. In 2010, DARPA leadership let him start a battlefield medicine program. He challenged researchers to come up with ways of manufacturing drugs, both small molecules and biopharmaceuticals, in under 24 hours.

Earlier this year, two DARPA-funded research teams—one from Massachusetts Institute of Technology and one from the University of Maryland, Baltimore County—answered the second half of Ling’s call, reporting modular systems capable of manufacturing protein therapeutics on demand. The systems and others like them are expected to have far-reaching effects: In addition to allowing biopharmaceuticals, also known as biologics, to be produced in remote settings for military physicians, the same type of technology might make it easier and cheaper to supply drugs and vaccines around the globe. And in developed countries that already have access to biologics, the underlying technologies could pave the way for truly personalized medicines.

Answering the call

One of DARPA’s goals in synthesizing biologics on demand is to eliminate the need for refrigerating drugs. “Insulin was at the top of my list because insulin requires refrigeration,” Ling says. Cold storage needs energy, and that’s a big problem when you are dealing with extreme environments that lack infrastructure, like the military faces, he adds. But making insulin and other biologics on the spot and then giving them to patients immediately would eliminate the need for a refrigerator.

DARPA also wants to reduce the need to stockpile biologics as countermeasures in the event of chemical, biological, radiological, or nuclear attacks. Such drugs rarely need to be used, and they need to be replaced as they expire.

“There’s a time clock on those molecules, and they may become less and less active” during storage, says Brad Ringeisen, deputy director of DARPA’s biological technologies office and program manager for battlefield medicine. “This concept of making something where you need it, when you need it was really attractive to us because you might be able to greatly reduce the need to stockpile.” Smaller stockpiles could be kept for immediate use, and more drugs could be made on the spot in response to epidemics or attacks.

Biologics are normally synthesized in batches in large-scale reactors that are thousands of liters in volume by cells genetically engineered to produce desired proteins. In the case of Escherichia coli bacteria, the cells must be lysed to recover the proteins. In the case of yeast or Chinese hamster ovary (CHO) cells, the cells secrete the proteins, which simplifies purification. Switching one of these systems from making one molecule to another can take months. Such large-scale reactors are most efficient when used to produce proteins for a large patient population. Companies aim to optimize their supply chain over an approximately two-year time frame, but getting such forecasts right can be challenging.

“Traditionally, the manufacture of broader classes of biopharmaceuticals—from enzymes to hormones to cytokines—requires one-off, bespoke processes and unique facilities designed for each molecule,” says J. Christopher Love, a chemical engineer in the Koch Institute for Integrative Cancer Research at MIT. He leads one of the DARPA-funded teams that developed a system to meet Ling’s challenge. Love and his group wanted to design an on-demand biologics manufacturing system that could be easily switched from producing one molecule to another. Doing that has “really required thinking deeply about the biology of the host cell that produces the products,” Love says.

The system he and his team came up with, called Integrated Scalable Cyto-Technology, or InSCyT, uses Pichia pastoris yeast cells to produce various drugs in a benchtop manufacturing system (Nat. Biotechnol. 2018, DOI: 10.1038/nbt.4262). Pichia’s genome is small enough, Love says, that for about $30, you can sequence the whole thing. “The accessibility of that biology makes it possible to think about tuning the host to make the molecules of interest with the precision that’s necessary for biopharmaceuticals.” With new sequencing and genome-editing tools, researchers can quickly adapt the yeast to make new proteins.

Knowing the yeast’s biology well also enables Love and his team to predict what proteins from the host cells might contaminate their biologic product. Typically, the biopharma industry programs CHO cells to pump out selected biologics. The industry uses CHO cells because they are mammalian cells that can perform the posttranslational modifications often needed for therapeutic proteins. But these cells, more complex than yeast, generate about 2,000 contaminant proteins in addition to the desired therapeutic. With yeast, only about 200 host proteins end up in the cell culture medium with the product, making purification easier.

Love points out that his team’s InSCyT system has all the components of conventional biomanufacturing, just at a smaller scale. “We think a lot about right-scale manufacturing,” he says. “What we’ve demonstrated has all the same elements for manufacturing that one might generally expect in a typical facility.”

InSCyT consists of three modules. Fermentation takes place in a production module. The cell culture medium from that module, with the protein product, flows through tubing to a chromatography system in a purification module. Finally, the purified protein is filtered in a formulation module.

In contrast with Love’s system, the other DARPA-funded system to meet Ling’s challenge works without cells. Called Bio-MOD, for Biologically-derived Medicines on Demand, the system, designed by chemical engineer Govind Rao at UMBC and his team, fits in an 89-cm suitcase (Nat. Biomed. Eng. 2018, DOI: 10.1038/s41551-018-0259-1). The researchers are now working on an even smaller version that fits in a briefcase.

Instead of using cells, Bio-MOD works with freeze-dried extracts from CHO cells. The extracts contain gene transcription and translation machinery from the cells that Bio-MOD then harnesses to synthesize protein therapeutics. Compared with Love’s system, Rao’s is similarly modular, including single-use protein production and purification modules that can be inserted into the suitcase.

A key advantage of cell-free systems is speed. “If you need something in a hurry and it’s truly patient specific, you need it made on demand in a few hours. This is the only way you can realistically do it,” Rao says. In contrast, Love’s cell-based system can do end-to-end production of hundreds to thousands of doses of protein biologics in about three days.

The other advantage of cell-free systems is that refrigeration is not needed during transport. The freeze-dried cell extracts that Bio-MOD uses are like powdered milk, Rao says. Fresh milk is a perishable product that has to be refrigerated. “You need a cold chain,” he says. “Once people figured out you could make powdered milk, it revolutionized nutrition throughout the world. You could have something that was shelf stable. You just added water when you needed it, and it was ready.”

The cell extracts work similarly. “You just add buffer and DNA; in four to six hours, the expression of the protein product is done,” Rao says.

Rao’s group has shown that even cells from human blood can serve as a source of cell-free extracts (Sci. Rep. 2018, DOI: 10.1038/s41598-018-27846-8). He sees such extracts being particularly useful for making vaccines. In such instances, the researchers could draw blood from individuals, make the extract, use it in the Bio-MOD system to produce a protein such as a vaccine antigen, and inject it back into the same person.

This process would be especially helpful for dealing with outbreaks and epidemics. If you could make the vaccine at the point of care, you could administer it to all the people in the vicinity right away and “nip potential outbreaks in the bud,” he says. Injecting the vaccine back into the same person would reduce or even eliminate the need for screening for viruses or immunogenicity.

Rao’s system has some drawbacks. Cell-free production systems are generally suited to making small amounts of proteins. And more purification is required to remove the cell debris.

DARPA’s Ringeisen sees the benefits of both the MIT and UMBC systems, as well as their distinct applications.

Rao’s freeze-dried product “can be stored pretty robustly on the shelf at relatively elevated temperatures and humidity,” Ringeisen says. “You can imagine that being used in a very field-forward, austere kind of environment.” Ringeisen sees Bio-MOD as appropriate for special operations missions involving small numbers of individuals who need protection from specific threats.

Love’s benchtop system, on the other hand, could be used to replace biopharmaceutical stockpiles. The infrequency of chemical and other types of attacks means those stockpiles are rarely used, Ringeisen says. “You could almost think of this as additive manufacturing for pharma, where you could produce what you need when you need it.” He envisions scenarios in which an outbreak or exposure event leads to switching on production in nearby facilities. “Chris Love has shown that after about 24 hours he is able to grow up these Pichia yeast strains and start producing hundreds of doses a day just with his laboratory-scale operation. You can imagine scaling that up to make thousands of doses a day.”

Both projects demonstrated that they could produce a variety of molecules, including protein therapeutics and antigens for vaccines. Among them was granulocyte colony-stimulating factor (G-CSF), a protein that is administered in response to radiation exposure. A generic version of G-CSF has already been synthesized and been U.S. Food & Drug Administration approved, so it’s a well-understood protein therapeutic that the MIT and UMBC teams used to test their systems.

Cell-free innovations

Organizations besides DARPA are interested in on-demand biomanufacturing to make it easier and cheaper to provide biologics around the globe. Most of those other organizations and researchers are focused on designing cell-free systems because of their ability to work in a multitude of environments. The first time that freeze-dried cell extracts were used to produce proteins was by Bradley C. Bundy’s group at Brigham Young University in 2014 (BioTechniques, DOI: 10.2144/000114158). Around the same time, James Collins and coworkers at MIT also created freeze-dried cell extracts, which they eventually applied to on-demand, portable biomanufacturing (Cell 2016, DOI: 10.1016/j.cell.2016.09.013).

Bundy’s team has focused on using extracts from E. coli, an abundant and common type of bacterium. “We’re using an E. coli-based system because we want this to be a very low-cost system so more people can have access to lifesaving protein therapeutics,” Bundy says. Bundy has recently started collaborating with Rao’s team.

Collins says his cell-free system is aiming to improve global health. “It would open up portability for health care workers in low-resource areas,” he says. “It would open up portability for military personnel or space travel or hikers or athletes.” But for larger companies, he says, “outside of offering easy-to-use, inexpensive research tools for making molecules on the spot, I don’t think our platform would figure into production.” That’s because large companies typically need to produce sizable quantities of biopharmaceuticals and need sufficient resources to run the corresponding live-cell bioreactors. Cell-free platforms are better suited for producing modest quantities in areas of limited resources, he says.

Perhaps validating Collins’s skepticism, only one mainstream biopharma company—Sutro Biopharma—is focusing on cell-free systems. On the discovery side, the company uses cell-free production to quickly make molecules for testing. For example, it uses such systems to make antibody-drug conjugates.

“Because it’s fast, we can evaluate many different constructs,” says Shabbir Anik, chief technical operations officer at Sutro.

Still, Sutro is also using cell-free manufacturing to take products through to commercialization. The firm says one advantage of a cell-free system is that it can use the same expression system for multiple new products. The company makes the cell extract itself and places it in inventory. Within 24 hours of loading extract into a bioreactor, Sutro can produce proteins. With cell-based systems, it can take days or weeks to have product. Sutro’s first antibody-drug conjugate manufactured by the cell-free process is now in a Phase I clinical study.

Sutro also wants to use cell-free systems to make biopharmaceuticals that would otherwise be difficult to make in cells. For example, such systems can be engineered to add nonnatural amino acids to proteins. Or they can make proteins that are toxic to cells. “We’ve made things like cytokines and growth factors and peptides that degrade cellular systems,” Anik says.

But cell-free systems face challenges too. Many protein therapeutics are decorated with sugars, also called glycans or glycosyl groups, that are hard for cell-free systems to add. Earlier this year, a team led by Michael C. Jewett of Northwestern University and Matthew P. DeLisa of Cornell University reported a method for getting E. coli cell-free extracts to add sugars to proteins in a single reaction mixture (Nat. Commun. 2018, DOI: 10.1038/s41467-018-05110-x). Jewett is now collaborating with Rao’s Bio-MOD team.

And there are costs associated with acquiring cells to make the extracts and with purifying them after they’re made. “You have to start with cells to get to extracts,” MIT’s Love says. “And all that cell debris has to be removed.” Breaking open, or lysing, the cells to get the gene transcription and translation machinery leaves behind unwanted membranes.

Making it personal

On-demand biomanufacturing has the potential to lower the cost of producing biologics for smaller patient cohorts, even to the point of individual patients.

In the Netherlands, that goal is well on its way to becoming reality. Huub Schellekens and coworkers at Utrecht University have been running a pilot program in which they make biological therapeutics on demand for individual patients.

Economics is one of the drivers. Prices are increasing at the same time that the average effectiveness of new drugs is decreasing, Schellekens says. The current biopharmaceutical industry “is not really designed for the next step in pharmaceutical care, which is personalized medicine,” he says.

Schellekens’s team is working on a method for making therapeutics for individual patients. “We are mainly producing biologics in the hospital pharmacy,” Schellekens says. “We can produce most expensive drugs at 5% of their current prices and still make a profit.” Such a model can be thought of as the biologic equivalent of a compounding pharmacy.

Schellekens and his group are starting with treatments for lymphoma, a cancer of the white blood cells that can become resistant to antibody therapies. Because of this resistance, doctors need another way to target a patient’s cancer cells. Once a new biomarker for the cells is found, Schellekens’s team could produce a bespoke monoclonal antibody to match it and be ready to treat the person.

The idea is to make the amount needed for a year’s worth of treatment. The dosage forms made by the pharmaceutical industry often result in the need to discard unused products. “If you make it for an individual patient, we can give him or her the exact amount they need. We don’t need to throw away anything,” Schellekens says.

The approach is a “no-brainer” for expensive orphan drugs, such as some enzyme replacement therapies, Schellekens says. But there will still be a place for a pharmaceutical industry. “There will always be a need for a pharmaceutical industry for large patient populations and for mass production of drugs,” Schellekens says.

His team is focusing on off-patent drugs for now, but he thinks he will be able to sidestep patent issues regardless. Such treatment is known as “magistral” treatment, meaning it was made or prescribed to fit the needs of a particular case. “Magistral treatment is driven by the obligation every doctor has to treat patients,” Schellekens says. “As a medical doctor, I’m obliged to treat patients in need. That is a duty that cannot be overwritten by somebody on economic grounds.”

Schellekens sees such drug production being implemented on a regional basis for general applications and in specialized treatment centers such as cancer centers. “It would be very inefficient if every hospital were to make all possible biologics.”

Back in the U.S., efforts are ongoing to advance the DARPA-funded technologies.

Ling, the former army doctor and DARPA program manager, is spearheading one of them. He’s now the CEO of On Demand Pharmaceuticals, a start-up that’s licensing technologies for both small-molecule drugs and biologics. “I want to see this work go to fruition,” Ling says. “The only way you’re going to get to patients is if a commercial enterprise decides to go ahead and make the thing and actually get it distributed.”

Love’s team is in the process of forming its own spin-off to move its platform toward commercial development. But he’s unwilling to provide details at this time.

Ringeisen anticipates that the systems DARPA is funding could be operational within 18 months. That’s an ambitious timeline, Ringeisen admits. But, he adds, “if it weren’t ambitious, it wouldn’t be DARPA.”