IL-2 treatment can be dangerous. Here’s how drug firms are trying to fix it

In his first years as an oncologist, Jonathan Drachman treated people with kidney cancer and melanoma. There were few treatment options for them in the 1990s, he says, and cancer immunotherapy—the manipulation of a person’s immune system to kill off cancer cells—was still new.

But there was a drug called Proleukin that, for better or worse, gripped his attention. Proleukin, also called aldesleukin, was a therapeutic version of interleukin-2 (IL-2), a small human protein called a cytokine that could attach to specific kinds of immune cells and stimulate them to grow and divide. Some of these immune cells could kill foreign invaders, including cancer cells.

Proleukin was an early cancer immunotherapy, and when it worked, Drachman says, it worked well. About 15% of people treated went into remission, he says, and of those, up to 8% survived for years, even decades. Drachman, now the CEO of Neoleukin Therapeutics, a company developing an IL-2-like therapy, says he couldn’t have been more excited.

“It was one of the first immunotherapies that actually worked in anyone,” he says.

But there was a catch. Proleukin could be dangerous. At a time when many cancer treatments could be taken without checking into the hospital, Drachman says, a person on Proleukin had to be admitted to the cancer ward and, almost always, the intensive care unit. With a half-life so short that it started disappearing almost as soon as the infusion was over, the drug had to be given at high doses every few hours, for days to be effective.

For many people, that high dose induced shakes and chills as their bodies responded to the cytokine, Drachman says. Many developed vascular leak syndrome: their blood vessels leaked fluid, their bodies swelled, their blood pressures dropped quickly, and they became confused or lost consciousness altogether. Nearly 75% of patients treated with Proleukin eventually developed antibodies to the drug, raising fears that the antibodies would decrease efficacy. Many people couldn’t tolerate the treatment.

For every person seemingly cured, Drachman says, there were as many who died of their cancers.

But Drachman kept thinking about those 15% of people who went into remission. He and others wondered how many more people IL-2 could successfully treat if researchers could figure out how to engineer out the toxicity. It would be years before technology would catch up—before biology, computing, and protein chemistry were advanced enough to build a better IL-2.

These new IL-2s have been attracting the drug industry’s attention, and companies have inked a flurry of six- and seven-figure deals over the past few years to develop the molecules for the clinic. Many of these molecules aim to rev up the immune system’s cancer response. Others aim to calm the overactive response at the root of autoimmune diseases. But none of these tweaks is a sure shot. A few clinical trials of IL-2 drug candidates have failed or been canceled.

The IL-2 drug candidates under investigation are new proteins that don’t exist in nature. Some are molecules that resemble IL-2 but have amino acid changes that strengthen some actions and weaken others. Others are IL-2s with antibodies and other chemical compounds attached that thwart some immune interactions and facilitate others.

All are trying to solve therapeutic IL-2’s fundamental problems—a short half-life that requires toxic dosing levels, nonspecific binding that stalls efficacy and causes side effects, and a tendency to spur the production of antibodies that could affect the efficacy of the drug or native IL-2.

Some scientists say the clinical trials are as much a test of the technologies that have gone into these engineered IL-2s as they are of the candidates’ abilities to successfully treat cancer or autoimmunity.

Traditional IL-2 works, says Giovanni Abbadessa, an oncologist who leads early clinical development of cancer immunotherapies at Sanofi, which paid $2.5 billion for Synthorx, a biotechnology firm developing an engineered IL-2. But the cytokine could work better, he says. The question is: Which of the many biochemical tweaks or chemical additions will overcome IL-2’s limitations—without creating new ones?

“I think it’s really fascinating to see how different companies have taken different technological strategies and approaches to find different ways to do the same thing,” Abbadessa says. “And of course, I’m afraid only clinical trials—larger clinical trials—will tell us which one is the better way.”

One molecule, two cell fates

About 45 years ago, while studying immune cells called T cells, researchers working with Robert Gallo at the US National Cancer Institute discovered a molecule that made T cells grow and divide. It was originally called T-cell growth factor and was eventually renamed IL-2.

A few years later, while trying to understand how IL-2 stimulates T cells to multiply, a team at the National Cancer Institute that included Warren Leonard, as well as a team led by eventual Nobel Prize winner Tasuku Honjo, discovered a molecule on the surface of T cells to which IL-2 binds.

But when researchers tried to re-create the effect of that binding, they couldn’t. IL-2, bound to that molecule (eventually called α), wasn’t enough to get T cells to multiply. It would take years and dozens of scientists to figure out that IL-2 stimulates immune cells in a complicated way, using a receptor that can have two prongs, β and γ, or three, α, β, and γ. It was eye opening that IL-2 uses a multiprong receptor, Leonard says, but it was even more revelatory when researchers later discovered that the way IL-2 attaches to the receptor is the source of its therapeutic potential.

“It was always in the back of our minds that this could be something of down-the-road therapeutic utility,” Leonard says of understanding how IL-2 works. “But, at least at the time, we were driven by more basic-science considerations.”

IL-2 stimulates many kinds of immune cells, including T cells and natural killer (NK) cells, says Jamie Spangler, a bioengineer at Johns Hopkins University who is developing antibody-coupled IL-2s as possible therapies. T cells can take on special tasks, including becoming effectors, which drive the immune response, or regulatory cells, restraining the immune response. Natural IL-2 stimulates both effector and regulatory T cells to multiply, providing an army that allows the immune system to do its job, but with checks and balances.

Indeed, providing checks and balances seems to be IL-2’s most fundamental task, Spangler says. During the years of unraveling how the cytokine works came a surprising finding: mice missing the IL-2 gene can still mount an immune response; IL-2, it appears, has backups that can take the cytokine’s place. But Spangler says, these mice have serious autoimmune problems.

It turns out that IL-2 binds very tightly to the α piece of the receptor, which is primarily found on the regulatory T cells, which dampen the immune response. And IL-2 binds less tightly to the β and γ pieces, which are found on both effector and regulatory T cells. So binding to all three parts of the receptor allows IL-2 to stimulate cells that reduce the immune response. Binding to only β and γ allows IL-2 to ramp up the response.

In those mice that can’t make IL-2, other cytokines spur the multiplying of effector cells and the other immune cells which fight infection or cancer. But those same experiments suggest that IL-2 is critical in keeping the immune response in check.

What this means for IL-2 as a therapy is that building an IL-2-like compound that stimulates mainly effector cells to grow could spur a potent anticancer immune response, Spangler says. And building an artificial IL-2 to preferentially stimulate regulatory cells could create something that mainly fights autoimmunity.

“It’s a fascinating molecule,” she says of IL-2, “because it’s sort of got this two-faced function.”

But only in the past few years has the technology developed to allow scientists to take advantage of IL-2’s two faces, Spangler says. “Up until maybe 10–15 years ago, we didn’t have the tools and the wherewithal to actually hone the activities of IL-2 in a targeted way,” she says. “I think that’s really what’s been a part of this whole renaissance.”

Bulking up a skinny cytokine

Most of the industry effort to modify IL-2 focuses on developing it for cancer immunotherapy, says Nikolaos Sgourakis, a researcher at the University of Pennsylvania who studies the structure of cytokines. The scientists building IL-2 as a cancer agent are trying to make something that won’t bind to the α piece of the receptor or are trying to boost IL-2’s interaction with the β and γ pieces.

But in addition to building something that is specific to effector T cells and other cancer-killing immune cells, these scientists have to build something that is long lasting, has few side effects, and doesn’t inadvertently prompt the antibody-making arm of the immune system to attack it.

Among the scientists working on ways to prevent IL-2 from binding to the α part of the receptor is Spangler. Her lab is developing therapeutic IL-2s that are attached to antibodies. One of these constructs blocks the ability of IL-2 to connect with the α part of the receptor, keeping IL-2 from preferentially stimulating regulatory T cells to multiply and potentially spurring cancer-killing effector cells to multiply instead.

For several years, Roche has likewise sought to turn IL-2 into a therapeutic that can’t bind to the α part of the IL-2 receptor. The company is doing this through muteins—versions of the IL-2 protein with altered, or mutated, amino acid sequences that change how the molecule interacts with its receptor. Roche calls its IL-2 mutein IL-2v, for IL-2 variant.

Pablo Umaña, who leads cancer immu­notherapy projects at Roche, says the firm’s scientists also tried coupling IL-2v to antibodies that attach to cancer cells that have small immune beacons called tumor antigens on their surfaces. The idea, he says, was to keep the company’s IL-2 near the tumor so IL-2 could stimulate effector T cells.

The researchers ran into problems. One of their IL-2-antibody constructs became immunogenic—in trials, people given the drug developed antibodies against it. Another was well tolerated but didn’t have the cancer-killing effect the company was looking for. The trials were stopped.

Now, Umaña says, Roche is working on an IL-2-antibody construct that attaches to two points on the same T cell: the β and γ parts of the IL-2 receptor as one point, and PD-1, an immune-regulating protein that many cancer immunotherapies attempt to override to make T cells more effective at killing, as the other. Umaña hopes the dual-action molecule, now in Phase 1 trials, will strengthen IL-2’s ability to stimulate effector cells. The idea is that the antibody’s binding to PD-1 will create a strong interaction that helps solidify the weaker IL-2 interaction with its receptor on effector cells, urging those cells to divide.

“We call this our third-generation IL-2 variant,” he says.

Companies are also trying ways to get IL-2 to stick around longer in the bloodstream so they can reduce the high doses that lead to toxicity and side effects. One strategy is to pegylate the molecule. Attaching polyethylene glycol (PEG) to specific points in the amino acid chain of IL-2 makes the construct bigger, lengthening its half-life, says Willem Overwijk, vice president of oncology research at Nektar Therapeutics.

Nektar’s lead IL-2 candidate, bempeg­aldesleukin, or bempeg, features six strands of PEG attached to lysines on aldesleukin, the active ingredient in Proleukin.

Overwijk says the therapy is injectable and doesn’t become active until some of the PEG chains fall off, freeing up the part of the protein that binds preferentially to the β and γ pieces on cancer-killing effector T cells and NK cells. Overwijk says people with tumors who have been treated with bempeg have many effector T cells around their tumors and few regulatory cells.

And unlike with Proleukin, people treated with bempeg are not likely to require hospitalization.

“That’s a huge difference for the patient. They can go home basically right after; they don’t have to stay in the hospital,” Overwijk says. “It’s a totally different drug now.”

Bempeg is in testing as a stand-alone therapy and as a combination with other cancer immunotherapies. In general, the trials have been small, and Nektar has had to grapple with investors concerned about trial results as well as manufacturing issues that may have affected some early results. Nektar reports that in Phase 2 trials of bempeg combined with Opdivo, a checkpoint inhibitor from Bristol Myers Squibb, about one-third of the people in its 38-person cohort saw all their cancerous lesions disappear after treatment. Many of the trial participants lived 30 months after treatment started before their cancers started progressing again.

Unearthing unearthly IL-2s

Many researchers trying to bring modified IL-2s to the cancer clinic use aldesleukin, which only slightly varies from natural IL-2, as a starting point for mutations, or they couple natural IL-2 to another molecule.

But several companies are exploring molecules that act like IL-2 but are completely artificial. Neoleukin’s lead compound does what IL-2 is supposed to do to stimulate effector T cells, but it’s smaller and more compact, Drachman says.

NL-201, Neoleukin’s IL-2-mimicking compound, came out of the Institute for Protein Design at the University of Washington. Carl Walkey, an executive at Neoleukin who was previously at the institute, says the team used computational biology to envision what an ideal IL-2 would look like as a cancer treatment. Researchers sorted through an astronomical combination of amino acid sequences to understand how altering amino acids in the IL-2 molecule might affect its structure and function, Walkey says.

They then made versions of the best candidates and chose NL-201 for its activity. The molecule barely looks like natural IL-2, yet it binds to the β and γ parts of the IL-2 receptor with enough affinity to potentially stimulate effector T cells to divide and kill.

“IL-2 has these funny geometries. It bends and twists in funny ways” because it has so many different jobs across many cell types, Walkey says. NL-201 “is much more ideal. It’s straighter. It’s more compact.”

NL-201 is also stabler than the native cytokine, he says; it has a rich hydrophobic center and a hydrophilic exterior that makes it soluble. It’s an “idealized protein structure,” Walkey says.

The US Food and Drug Administration seems to be moving carefully with what Walkey says is the first completely engineered, made-from-scratch protein to approach the clinic. In January, FDA officials asked Neoleukin to make some changes to its clinical trial protocols to better measure dosages. Drachman says he hopes to be in trials sometime this year.

For Sanofi, IL-2 immunotherapy dreams rest on the unnatural amino acid N₆-[(2-azidoethoxy)carbonyl]-l-lysine, or azidolysine. The novel amino acid was developed by Synthorx cofounder Floyd Romesberg and engineered into an IL-2 candidate called THOR-707. Sanofi now owns Synthorx and THOR-707. This unnatural lysine has a PEG attached to it that blocks the molecule from interacting with the α part of the IL-2 receptor. THOR-707 is in early trials in people with a wide range of tumors.

“I see it as a cleaner approach to try and have a selective drug,” Abbadessa says, comparing it with other modes of engineering IL-2. “We hope that it will work.”

Then there’s Bright Peak Therapeutics, which has taken protein building completely out of cells. Rather than coaxing bacterial or animal cells to spit out modified IL-2s, which is the norm, Bright Peak’s IL-2-like molecule is made by chemically stitching together small peptides.

Taking the protein-assembly process out of cells allows the company’s scientists to better understand the growing amino acid chain as it’s assembling and folding, says CEO Sef Kurstjens. Company scientists can more easily incorporate things like noncanonical amino acids.

Vijaya Pattabiraman, Bright Peak’s cofounder and an organic chemist, says the company’s chemists can stitch together a peptide and modify an amino acid on that peptide to improve its half-life. Researchers can build the next section of the protein and, if needed, modify it.

The scientists can then join these pieces with all the different elements they want to build their perfect molecule, and once it’s folded, the scientists can add molecules like PEG, if that’s what their design calls for. This process doesn’t require microbes or cells that have to determine how to add an artificial amino acid.

“It can be anything that chemists and biologists can think about and dream of,” Pattabiraman says of a molecule built using Bright Peak’s technology. “There’s no other way you can really introduce multiple noncanonical amino acids within the protein and get the biology that we want.”

Steering IL-2 toward autoimmunity

For all of IL-2’s potential as a cancer therapeutic, Hopkins’s Spangler says we shouldn’t forget its potential as a strong suppressor of the immune response. Because IL-2 binds so strongly to the α part of the receptor, low doses can trigger the multiplication of regulatory T cells, which dampen the immune response, making the cytokine an enticing drug candidate for autoimmune disorders. She is developing an antibody-coupled IL-2 to skew toward activating regulatory T cells.

But modulating IL-2 to activating regulatory T cells is a more nuanced task than treating cancer, Penn’s Sgourakis says. T cells aren’t the only cells that express the α part of the IL-2 receptor. Any effort to boost the activity of regulatory T cells could also stimulate nontarget cells. And, he says, effector cells can start producing that α segment, scooping up a therapeutic IL-2 meant to tamp down the immune system.

As IL-2 and its receptor move in space, he says, there are fleeting conformations—shapes that last just an instant but could mediate some of the most important interactions IL-2 might have with a molecule that affects the immune response. Using nuclear magnetic resonance, Sgourakis’s team is trying to capture the fleeting moments to design mutants to mimic those interactions, including ones that could lead to IL-2-based autoimmune therapies.

“You really have a much more narrow set of conformations, or of IL-2 forms that could be potentially useful for an autoimmune disease treatment,” he says. “Trying to make an IL-2 that targets cancer immunotherapies is like hitting a free throw. Trying to make a molecule that targets autoimmunity or dampens a certain set of immunity is like trying to hit a half-court shot.”

Still, a few companies are trying. This includes Amgen, which is developing its own IL-2 products, and Merck & Co., which in February announced it is buying Pandion Therapeutics for $1.85 billion. Pandion’s pipeline includes two IL-2-based molecules, plus the technology to deliver those molecules and others to specific tissues.

Pandion cofounder Jo Viney says turning therapeutic IL-2 toward autoimmunity should require lower doses than used in cancer treatment and should thus lower the risk of side effects. But using existing IL-2 is still daunting, she says. While a person with cancer would take IL-2 therapy for about a week, autoimmune diseases are chronic, and IL-2 might have to be given over long periods. IL-2 could activate other immune cells, which could counteract the autoimmune effort. And with a short half-life, IL-2 would be an almost daily treatment, she says.

But on the plus side, the use of IL-2 therapies for autoimmunity could unleash a more comprehensive shutdown signal on an overreacting immune system, she says; in contrast, current autoimmune therapies tend to be molecules and antibodies that target only one component of an autoimmune response.

Pandion and Amgen are using similar technology to build autoimmune-specific IL-2: coupling IL-2 muteins to antibody fragments that extend IL-2’s circulation time.

Making Pandion’s candidate, PT101, involved testing many amino acid mutations in the native IL-2 protein, Viney says. The firm’s scientists found that acidifying a critical asparagine destabilized IL-2’s binding to the β piece of the receptor. Adding four other amino acids to the IL-2 mutein created a molecule that would preferentially bind to the α piece of the IL-2 receptor, favoring the growth of regulatory T cells.

Viney says that because Pandion’s IL-2 is coupled to the antibody fragment, the therapy could be given to people every few weeks instead of every day. In a small Phase 1 trial, people who received PT101 produced many more regulatory T cells than people who were untreated. The experimental biologic seemed to produce few side effects and no antidrug antibodies. By the end of 2021, Viney says, PT101 should be in trials for ulcerative colitis and lupus.

Amgen’s lead IL-2 construct, efavaleukin alfa, is in testing for lupus and graft-versus-host disease. The company ended a trial of the drug candidate—a fusion of a mutein IL-2 and an antibody—as a “business decision,” according to ClinicalTrials.gov. Nektar is also exploring pegylated IL-2 as an autoimmune therapy.

While people wait to see which, if any, of these novel IL-2s will become the next generation of IL-2 immunotherapy, Neoleukin’s Drachman says Proleukin is still in occasional use as a late-stage therapy for people who have not responded to other treatments. He says he recently spoke to a man with melanoma whose doctors pulled him off Proleukin after his heart stopped during treatment. He went on to receive checkpoint inhibitors, immunotherapies that block some of the regulatory functions of the immune system, and has survived for several years.

He is one of the lucky ones. Even with checkpoint inhibitors, Drachman says, fewer than half of people with melanoma go into stable remission. Melanoma is a tough cancer to treat. It’s resistant to many drugs, Drachman says, but not IL-2. This is why developing a safer version is so important, he says.

Sgourakis and others believe that the technologies in play in the effort to create a better IL-2 may also lead to other, more tolerable and potent cytokines and immune modulators to treat diseases beyond cancer and autoimmunity. What structural and computational biology reveal could guide scientists to molecules that affect a host of immune functions.

It’s an enticing possibility, Sgourakis says. “This field is coming of age.”