mRNA pioneers refocus on therapeutics

When the COVID-19 pandemic erupted in early 2020, messenger RNAs (mRNAs) designed to produce therapeutic proteins were moving towards clinical trials. Companies in the field — including Moderna, BioNTech, CureVac, Arcturus and Translate Bio—all pivoted to vaccines. Now that the success of the Moderna and BioNTech/Pfizer COVID-19 vaccines has validated the mRNA technology platform, they’re doubling down on the goal of converting human cells into protein factories to treat disease. “The wind is at our backs now, and we must use that newfound momentum to change the outcome for other diseases, too,” says John Androsavich, global head, RNA medicine lead at Pfizer.

Indeed, a growing pipeline of mRNA therapeutics is entering clinical trials, in particular systemic delivery of protein replacement therapy for rare monogenic disorders, and cytokines or antibodies for cancer immunotherapy (Table 1). mRNA offers potential safety advantages over gene therapies because there is no genome integration or permanent modification. And compared with recombinant proteins, there are manufacturing process advantages, as well as the opportunity to improve drug characteristics. In an interview with Nature Reviews Drug Discovery last year, BioNTech CEO Uğur Şahin highlighted the potential to dose the company’s modified IL-2 mRNA (BNT151) three times weekly compared with three times daily for commercial recombinant interleukin (IL)-2 (Proleukin). For short-lived bispecific antibody constructs, mRNA-encoded versions can have better pharmacokinetics than recombinant versions.

Unlike recombinant proteins, mRNA can reach any subcellular compartment. “And we can embed the protein in the cellular membrane,” says Susanne Rauch, who oversees CureVac’s molecular therapy programmes. “You can’t do that with just delivering the protein.”

But several technical obstacles remain. The most advanced molecule, Moderna’s vascular endothelial growth factor A (VEGFA)-encoding mRNA for myocardial ischaemia, was recently dropped by clinical development partner AstraZeneca, despite meeting its primary endpoint of safety and tolerability in a small phase IIa trial, with a trend towards improvement in three exploratory efficacy endpoints. Other clinical trials have been terminated for lack of efficacy. As vaccine euphoria gives way to therapeutic reality, drug developers in the field are crafting solutions to the problems — particularly related to delivery, immunogenicity and duration of protein expression, which present much greater hurdles for mRNA therapeutics than for vaccines.

Delivering the message

The success of the two mRNA vaccines for COVID-19 culminated decades of research into stabilizing, delivering and efficiently translating synthetic mRNA. In vitro transcription (IVT) of mRNA basically involves mixing a linearized plasmid DNA template with ribonucleotides, then adding RNA polymerase. It doesn’t require living cells, which is a major process advantage over recombinant proteins and many vaccines. However, synthetic mRNA is intrinsically unstable in vivo, and although it doesn’t need to enter the cell nucleus like gene therapies, it still has to cross the cell membrane and emerge intact into the cytoplasm. mRNA vaccines successfully employ nucleoside modifications to the mRNA sequence to improve stability, and encapsulation in lipid nanoparticles to facilitate intracellular delivery.

mRNA therapeutics are composed of the same basic package as mRNA vaccines: a modified single-stranded mRNA encoding the protein, usually encapsulated in a lipid nanoparticle. But it’s much easier to make vaccines work than therapeutics. Treating disease usually requires a lot more protein, expressed over a longer period of time, implying heavier and more frequent dosing, raising the risk of toxicity.

That toxicity is mainly immunological. Mammalian mRNAs have modified nucleosides, but the innate immune system sees synthetic unmodified mRNA as viral RNA, triggering pattern recognition receptors and leading to type 1 interferon and inflammatory cytokine production. A key advance was the incorporation of pseudouridine nucleosides into synthetic mRNA to dampen immunogenicity and boost translation, reported by Katalin Karikó, Drew Weissman and colleagues at the University of Pennsylvania in 2005. Some innate immune system activation remains, which actually helps mRNA vaccines, because it triggers cellular immunity. But immunogenicity is an absolute liability for an mRNA therapeutic, especially for the kind of chronic therapy needed in many diseases.

Companies have historically tried to solve the immunogenicity problem in two ways. One is to incorporate non-natural nucleosides — for example, replacing all uridines in the mRNA sequence with N1-methylpseudouridine, a key element of both the Moderna and the Pfizer/BioNTech COVID-19 vaccines. This masks the mRNA from the immune system, by preventing binding to pattern recognition receptors, among other mechanisms. Other companies employ different modifications; Ethris, for example, uses 5-methylcytidine and 2-thiouridine nucleosides, in a portion of its mRNAs.

The second approach is sequence optimization, which typically involves switching out uridine nucleosides for guanosines and cytidines in specific codons, without changing the amino acid sequence, among other strategies. Until recently, some companies, including CureVac and Translate Bio (now Sanofi), deliberately avoided modified nucleosides in favour of sequence optimization and other strategies, but they're now exploring modifications.

Another potential toxicity issue is the accumulation of lipid nanoparticle carriers in the liver, given the frequent, chronic dosing required for many mRNA therapeutics. In September 2019, Translate Bio discontinued its mRNA candidate for ornithine transcarbamylase deficiency, because of the “pharmacokinetic and safety profile… related to the first-generation liver lipid nanoparticle.” (The FDA had earlier placed a hold on the clinical trial.)

Newer lipid nanoparticles appear safer, because they are engineered for rapid degradation — for example, by incorporating esters, which have biodegradable bonds, in the tails of the ionizable lipid. Arcturus’s lipid nanoparticle has a nitrogen in its head group, instead of carbon. “Carbon is like a rock, it’s very hard to biodegrade that carbon core,” says Arcturus CSO Pad Chivukula.

Proteins that last

Duration of protein expression — because mRNA is inherently short-lived — is a key challenge for mRNA therapeutics. Nucleoside modifications and sequence optimization to replace rare codons with those that have abundant complementary transfer RNAs in target tissues can help, but companies are applying a whole suite of changes to their mRNAs.

For example, they’re testing different 5' and 3' untranslated region (UTR) sequences, which influence mRNA translation and stability by interacting with RNA binding proteins. Ethris’s custom UTRs, says company CEO Carsten Rudolph, “add two things. High yields of protein, of the translation, and high yields during the in vitro transcription” of the mRNA. These two goals — high levels of synthetic mRNA production and high levels of mRNA translation — can conflict, requiring careful balancing, he notes. Companies also use their own analogues of the mRNA 5' cap, the 7-methylguanosine linked to the first nucleotide of the RNA transcript that blocks nuclease degradation of the mRNA and helps with translation. “I believe they will be game-changers, because one of the main issues in mRNA therapeutics is how long the mRNA stays around,” says National Cancer Institute molecular biologist Shalini Oberdoerffer. “Cap analogues are very attractive in that sense.”

Circular RNA (circRNA) is earlier in development, but may have even more impact. CircRNAs are very long-lived, because they lack the ends where exonucleases normally initiate mRNA degradation. Companies can now circularize full-length mRNA. A co-founder of circRNA startup Orna Therapeutics in 2019 described circRNAs with comparable protein expression in vitro as modified linear mRNA. Whether circRNA evades innate immunity is controversial, but companies are enthusiastic. In August, Orna inked a circRNA collaboration with Merck worth up to $3.5 billion. “People are showing that these things can last weeks to months,” says Chivukula, whose company is also actively exploring circRNA. “We are looking at everything that can help the duration.”

Some companies also alter the amino acid sequence of the encoded protein. For example, Arcturus, says Chivukula, is removing ubiquitination sites on the protein, to improve protein longevity by delaying degradation. Changing the amino acid sequence brings a theoretical risk of immune activation, but this is something Moderna hasn’t seen to date, said Lisa Rice, a senior scientist in the rare diseases division of Moderna, at the May 2022 annual meeting of the American Society of Gene & Cell Therapy (ASGCT). “But yet to be proven,” she added.

At the ASGCT meeting, Rice described a series of iterative improvements in an mRNA encoding the enzyme that is deficient in the rare disease phenylketonuria (PKU). Her team performed codon optimization; changed the tail geometry of the mRNA to prevent deadenylation of the poly(A) tail and slow mRNA decay; identified protein sequences that extended protein half-life; and altered the lipid nanoparticle composition to better enable mRNA escape from endosomes once inside the cell. Combining these manipulations, Moderna went from a one-day reduction in toxic levels of the amino acid phenylalanine to a five-day reduction, following a single dose in a mouse model of PKU. But that outcome, if it holds true in humans, may still not be good enough for even weekly dosing.

Moderna remains optimistic. “There is some evidence to suggest that, because of metabolic rates etcetera, we would have greater durability in humans than what we see in mouse models,” says Ruchira Glaser, senior vice president and therapy area head at Moderna. It’s also possible, she adds, that intermittent expression could still be therapeutic.

To get around the protein durability issue, some companies are selecting diseases where protein expression is already exceptionally long-lived, or where durable expression isn’t needed. For example, Sanofi and Ethris are both targeting primary ciliary dyskinesia, a rare genetic lung disease that impairs the ability of cilia to clear mucus, leading to recurrent infections and lung damage. “Biology helps us here,” says Ethris’s Rudolph. Lung epithelial cells make cilia when they undergo cell division. “If you deliver the messenger RNA at the right time to the cells when they go through this differentiation step,” says Rudolph, “and you then incorporate the [encoded normal] protein in the cilia, the protein will last the lifetime of the cell.”

Monogenic metabolic disorder treatments often involve enzymes that are small in size, with expression marginally elevated in a fraction of cells. This “has a higher probability of success today than other indications that require large structural proteins to be delivered to the vast majority of diseased cells,” says Androsavich.

But safety, with multiple doses, is the first hurdle to overcome for systemic delivery. Early signs are positive. Multiple intravenous infusions of Moderna’s mRNA for propionic acidemia, one of several life-threatening inherited metabolic diseases known as organic acid disorders, were well tolerated in an ongoing phase I study, with five patients taking the complete set of ten doses.

Moderna’s ongoing trial in a similar disease, methylmalonic acidemia, is close behind. Arcturus has opened a phase II trial in Europe of its mRNA for ornithine transcarbamylase deficiency, with interim data expected by early 2023. Its goal, says Chivukula, is “to prove that chronic therapy is potentially safe.”

Spreading the message

Looking further out, can mRNA therapeutics get beyond the liver? This has also been challenging for antisense oligonucleotide and small interfering RNA drugs. Local delivery has a place; Moderna, for example, is exploring intracardial and intratumoral injection, and aerosolized lipid nanoparticles for inhalation. But most of its drugs use intravenous delivery, with lipid nanoparticles binding to circulatory proteins that go to the liver. To avoid this, one approach is to engineer lipid nanoparticles with affinity for other tissues. But rodent models are unreliable predictors of lipid nanoparticle delivery efficiency in humans, warns Androsavich. “It’s one of the biggest issues the field is facing,” he says. “Some may not even be aware of it, especially as the field swells with inexperienced new entrants.”

Another method takes advantage of the fact that the more than 2,600 known microRNAs vary widely in activity across cell types. By incorporating microRNA target sites into mRNA UTRs, some selective expression can be achieved. “These target sites allow microRNA to bind to our UTRs, when present, and degrade the mRNA if we don’t want mRNA expression in that tissue cell type,” said Rice at the ASGCT meeting.

How much further can mRNA engineering take the field? “The focus here needs to be on progress, not perfection,” says Androsavich. “I don’t think there is one discovery that will make mRNA rock solid, for instance. It’s going to have to be a combination of innovations: mods, circularization, UTR design, codon optimization, cap analogues, protein engineering — all need to come together in an optimized mRNA drug product.”