Researchers synthesize drug molecules from the genome's 'dark transcriptome' for the first time

Omar F. Khan had been thinking about a number: 45%. That is roughly how much of the human genome produces long strings of RNA that do not encode proteins. These molecules, called long noncoding RNA (lncRNA), are transcribed from DNA but never translated into the proteins that do most of the cell's visible work. About 40,000 lncRNA transcripts have been catalogued so far. Their collective function is so poorly understood that researchers sometimes call them the "dark transcriptome."

Khan, a professor of engineering at the University of Toronto, saw an opportunity hiding in that darkness. Messenger RNA (mRNA) therapies had already proven that you could synthesize RNA in the lab and deliver it to the body as medicine. The COVID-19 vaccines were the most visible example. But mRNA works by instructing cells to build specific proteins. What if you could take a similar approach with lncRNA, molecules that do not build proteins but instead regulate gene expression in highly specific ways?

Three sequences, three different anti-inflammatory effects

Khan's team, led by PhD student Janice Pang, started by searching the literature for lncRNA sequences linked to inflammation. Chronic or excessive inflammation underlies conditions from sepsis to arthritis to cardiovascular disease. If lncRNA sequences naturally regulate inflammatory pathways, synthesizing and delivering them could provide a way to shut down inflammation when it spirals out of control.

They selected three candidates: GAPLINC, MIST, and DRAIR. Each had been identified by other researchers as potentially involved in inflammatory regulation, but none had been made outside a living cell. Using in vitro transcription synthesis, chemical modifications, and high-performance liquid chromatography purification, the Toronto team produced the first synthetic copies of all three sequences.

They then packaged the lncRNA into nanoparticles and delivered them to human cell cultures and to mice with inflammatory disease. Each sequence reduced inflammation, but through different mechanisms. All three worked by decreasing production of specific cytokines, the signaling proteins that trigger inflammatory cascades. But each targeted different cytokines, suggesting narrow, specific mechanisms of action.

Evolved specificity as a drug design advantage

Traditional drug development is slow and expensive partly because candidate molecules frequently fail due to off-target effects or poor compatibility with human biology. Khan argues that lncRNA offers a structural advantage: these molecules have been refined by millions of years of evolution inside the human body. They are inherently biocompatible. Their mechanisms of action tend to be narrow and specific, which reduces potential side effects and enables lower doses.

The team went further by exploring structural and chemical modifications to each lncRNA to increase potency. This is delicate work, since the three-dimensional shape of these molecules matters to their function, and changing too much can destroy activity. Through careful optimization, Pang and the team found modifications that actually increased potency, meaning even lower doses could achieve the same anti-inflammatory effect.

A library of 40,000 candidates

The three inflammation-targeting molecules are a proof of concept. The larger significance, as Khan frames it, is that the dark transcriptome represents a vast library of naturally evolved molecules that could be mined for therapeutic purposes. Where traditional drug discovery screens synthetic chemical libraries numbering in the millions, lncRNA offers a different starting point: molecules that already exist in the body and already interact with specific biological pathways.

The approach draws on the same delivery technology that made mRNA vaccines possible: lipid nanoparticles that package RNA and carry it into cells. The Toronto team's extensive expertise in RNA delivery systems was critical to getting lncRNA into target cells in sufficient quantities.

What remains uncertain

The work is early-stage. The anti-inflammatory effects were demonstrated in cell cultures and mouse models, not in human patients. The jump from mouse to human is uncertain for any drug candidate, and lncRNA therapeutics face additional challenges. These are large, complex molecules whose three-dimensional structures may behave differently in different cellular contexts. Whether they can be manufactured at scale, stored stably, and delivered effectively in clinical settings are open questions.

The study also does not address long-term safety. Introducing exogenous lncRNA into the body could, in theory, interfere with endogenous gene regulation in unexpected ways. The specificity that Khan highlights as an advantage could also be a limitation: if the mechanism of action is too narrow, the therapeutic benefit in complex diseases with multiple inflammatory drivers may be modest.

And while 40,000 lncRNA transcripts have been catalogued, functional annotation for most of them remains sparse. Finding the ones with therapeutic potential will require extensive screening, much of it in model systems that may not predict human responses.

But as a proof of principle, the study published in Science Signaling demonstrates something that had not been done before: taking a naturally occurring lncRNA sequence, making it in the lab, and showing that it works as a drug. If the approach scales, it could open an entirely new class of therapeutics derived from a part of the genome that, until recently, most researchers had dismissed.