RNA: The Forgotten Molecule That Controls Everything and Is Now Rewriting Medicine

For most of the twentieth century RNA was taught as the messenger — the courier that carried instructions from DNA to the machinery that builds proteins. Useful, necessary, and entirely secondary.

Two numbers show how completely that view collapsed. The reference annotation of the human genome, GENCODE release 50, lists 19,442 protein-coding genes and 35,885 long non-coding RNA genes. There are nearly twice as many genes whose product is RNA that never becomes a protein as there are genes for proteins at all.

And the clinic has followed. Registry searches on 2 October 2026 for mRNA vaccine, antisense oligonucleotide and siRNA return 771, 217 and 208 studies. The molecule once dismissed as a courier now underpins approved vaccines and drugs.

The courier picture is wrong, not merely incomplete. RNA regulates which genes are expressed and when. It catalyses the reaction that builds every protein in your body. It may even have been the original molecule of life, carrying genetic information and doing chemistry before DNA and proteins existed.

This is the story of what RNA is, what it does, where it may have come from, and where RNA medicine is going.

What RNA Is — and How It Differs from DNA

RNA — ribonucleic acid — is a close chemical relative of DNA. Like DNA, it is a polymer built from nucleotides, each carrying one of four nitrogen-containing bases. Like DNA, it encodes information in the sequence of those bases. Like DNA, it folds into specific three-dimensional structures that determine its function. The foundational molecule it partners with is covered in full in our guide to what DNA is and how it works.

But three chemical differences make RNA fundamentally distinct from DNA — and those differences explain almost everything about what RNA can do that DNA cannot.

The first difference is the sugar. DNA uses deoxyribose — a sugar that lacks a hydroxyl group at the 2′ position of the ring. RNA uses ribose, which has that hydroxyl group. This small chemical difference makes RNA less stable than DNA — it is more vulnerable to hydrolysis, to enzymatic degradation, to spontaneous breakdown. DNA’s stability makes it an ideal long-term archive. RNA’s relative instability makes it ideal for dynamic, responsive, short-lived molecular signals — messages that can be sent, read, and quickly cleared.

The second difference is a single base. Where DNA uses thymine (T), RNA uses uracil (U). Uracil pairs with adenine exactly as thymine does, so the information content is equivalent. The difference is metabolic — thymine requires an extra chemical synthesis step to produce, which may reflect evolutionary constraints on where each molecule appeared first.

The third and most consequential difference is structure. DNA is almost always double-stranded — its stability and information-storage function depend on the complementary base pairing that holds the two strands together. RNA is mostly single-stranded. But a single-stranded molecule can fold back on itself, forming hairpins, loops, bulges, and complex three-dimensional structures through internal base pairing. This folding capacity is what gives RNA its extraordinary functional versatility. A single RNA molecule can carry genetic information, fold into a specific three-dimensional structure, and use that structure to catalyse a chemical reaction — all three at once. No other known biological molecule can do all of this.

The Many Lives of RNA: A Family Portrait

Messenger RNA (mRNA) is the most familiar member: a single-stranded copy of a gene, made by RNA polymerase in a process called transcription and carried from the nucleus to the ribosomes, which translate it into protein. mRNAs are deliberately short-lived, made when a protein is needed and degraded when it is not, so gene expression can be adjusted quickly; most last hours, though some survive only minutes and others days.

Ribosomal RNA (rRNA) is the most abundant RNA in any cell — it makes up roughly 80 percent of total cellular RNA. Ribosomal RNA forms the core of the ribosome, the molecular machine that translates mRNA into protein. The critical discovery that changed biology’s understanding of ribosomes — and of RNA more broadly — came from structural studies of the ribosome in the late 1990s and early 2000s.

Researchers including Thomas Steitz, Ada Yonath, and Venkatraman Ramakrishnan, who shared the 2009 Nobel Prize in Chemistry, showed through X-ray crystallography that the catalytic centre of the ribosome — the site where peptide bonds are actually formed — is made entirely of RNA, not protein. The ribosome is a ribozyme. Proteins in the ribosome are structural scaffolding. The chemistry of life — the assembly of every protein in every organism on Earth — is catalysed by RNA.

Transfer RNA (tRNA) is the adapter between the language of nucleic acids and the language of proteins. Each tRNA carries one amino acid and a three-base anticodon that pairs with a matching codon on the mRNA, so the ribosome adds the right amino acid each time. There are fewer kinds of tRNA than codons, because some tRNAs can read several codons through a looser pairing called wobble.

Small nuclear RNA (snRNA) is a component of the spliceosome — the molecular machine that removes introns (non-coding sequences) from pre-mRNA and joins the remaining exons together to produce the final mature mRNA. This process, called RNA splicing, is universal in eukaryotic cells and can occur in different ways for different transcripts — alternative splicing allows a single gene to produce multiple different proteins, enormously expanding the protein-coding capacity of the genome without requiring additional genes. It is estimated that over 95 percent of human multi-exon genes undergo alternative splicing.

The Non-Coding Revolution: Rethinking “Junk”

For decades after the central dogma was established, the working assumption was that the important RNAs were the ones that carried protein-coding information — mRNAs — and the structural components of ribosomes and the translation machinery. The vast majority of the genome, which did not code for proteins, was assumed to produce little of consequence — junk DNA making junk RNA.

That assumption has been seriously challenged. In 2012 the ENCODE project reported that about three-quarters of the human genome is capable of being transcribed into RNA, and most of those transcripts do not code for protein. How many of them do something useful is still argued over, but some clearly matter.

MicroRNAs (miRNAs) are short RNAs, typically 21 to 23 nucleotides long, that bind matching sequences in target mRNAs and either trigger their destruction or block their translation. Victor Ambros and Gary Ruvkun discovered the first one in the roundworm C. elegans in 1993 and shared the 2024 Nobel Prize in Physiology or Medicine for it. miRNAs have since been found in virtually every multicellular organism studied; each can regulate many target genes, and together they are thought to influence most human protein-coding genes. Their disruption is implicated in cancer and in heart, brain and metabolic disease.

Long non-coding RNAs (lncRNAs) are RNA molecules longer than 200 nucleotides that do not encode proteins. GENCODE counts 35,885 lncRNA genes, and some other catalogues list well over 50,000, though the functions of most remain unknown. Those that have been studied reveal diverse biology. XIST, perhaps the best characterised, coats one of the two X chromosomes in female cells and shuts it down into a compact structure called the Barr body, equalising gene dosage with males.

HOTAIR recruits chromatin-modifying complexes to specific places in the genome, silencing genes in a way that goes wrong in several cancers. NEAT1 helps build nuclear structures called paraspeckles that hold certain RNAs in the nucleus during stress. Some researchers now see lncRNAs as important regulators of how the genome is folded and read, a layer explored in depth in our article on epigenetics: how your environment shapes the way your genes work.

Circular RNAs (circRNAs), long dismissed as rare oddities and only recently recognised as widespread, form closed loops with no free ends. That shape protects them from the enzymes that destroy linear RNA, so they are unusually stable and abundant in long-lived cells such as neurons. Some soak up specific miRNAs; others bind proteins or are translated. Their stability in blood and spinal fluid makes them promising biomarkers.

The RNA World: Where Life Began

RNA’s versatility — its ability to both carry genetic information and catalyse chemical reactions — has a profound implication for the origin of life. Modern cells use three distinct types of molecule: DNA to store genetic information, RNA to carry it, and proteins to catalyse reactions. The interdependence of these three systems raises an obvious question: which came first? You need proteins to replicate DNA, but you need DNA to make proteins. The system cannot have bootstrapped itself from scratch.

The RNA world hypothesis, named by the Nobel laureate Walter Gilbert in 1986, proposes a resolution: RNA came first. Before DNA, before proteins, there was RNA — molecules that could store genetic information and catalyse chemical reactions simultaneously. In the prebiotic environment of early Earth, RNA molecules that could replicate themselves and catalyse useful chemistry would have been subject to natural selection, gradually evolving greater complexity and efficiency. DNA eventually superseded RNA as the genetic archive — more stable, less susceptible to hydrolysis — and proteins took over most catalytic functions from ribozymes, leaving RNA in its current intermediary role.

The evidence for the RNA world is substantial and growing. The ribosome — the universal protein-synthesis machine — is a ribozyme, catalysing the most fundamental reaction in all of cell biology. Many essential metabolic cofactors, including ATP, NADH, and Coenzyme A, are nucleotide-based or contain nucleoside moieties, consistent with the hypothesis that they are molecular fossils of an RNA-based metabolism. And ribozymes — catalytic RNA molecules — can in principle carry out a wide variety of chemical reactions, as demonstrated through laboratory evolution experiments.

The most direct support comes from Gerald Joyce’s group at the Salk Institute, which has spent many years engineering RNA enzymes that copy other RNA strands. In a landmark 2024 study, published in PNAS, the team showed an RNA polymerase ribozyme copying other functional RNAs accurately enough for Darwinian evolution to occur: mutations arose, beneficial variants were selected, and the population adapted.

“We’ve long wondered how simple life was at its beginning and when it gained the ability to start improving itself. This study suggests the dawn of evolution could have been very early and very simple.”
— Nikolaos Papastavrou, research associate, Joyce Lab, Salk Institute. Salk Institute press release, 4 March 2024 (study in PNAS, DOI: 10.1073/pnas.2321592121).

The experiment did not create life, but it showed that the core feature of life, self-copying molecules subject to natural selection, is possible with RNA alone, without DNA or protein. Prebiotic chemists are now testing how RNA’s building blocks could have formed together from simple molecules such as hydrogen cyanide.

The mRNA Revolution: From COVID Vaccines to Cancer Cures

The most widely publicised RNA story of the past decade is the development of mRNA vaccines. But the full scope of what mRNA technology is becoming extends far beyond COVID-19, and the science behind it is more remarkable than the headlines convey.

The central idea of mRNA medicine is straightforward: instead of delivering a drug directly, deliver the genetic instructions for the cell to make the drug itself. An mRNA molecule encoding a therapeutic protein — an antigen for vaccination, an enzyme for replacing a missing one, a cytokine for cancer immunotherapy — is packaged into lipid nanoparticles that protect it from degradation and carry it into cells, where it is translated into the therapeutic protein. The mRNA is then degraded by normal cellular mechanisms, leaving no permanent genetic changes.

The intellectual foundation was laid over decades by scientists who were initially ignored or disbelieved. Katalin Karikó spent years at the University of Pennsylvania arguing that mRNA could be a therapeutic tool, facing scepticism, funding cuts, and a demotion before her work was vindicated. The crucial breakthrough came in 2005, when she and Drew Weissman discovered that incorporating modified nucleosides — specifically pseudouridine — into synthetic mRNA sharply reduced the immune response that synthetic mRNA normally triggered, allowing it to be taken up and translated without being immediately destroyed. This discovery, which earned Karikó and Weissman the 2023 Nobel Prize in Physiology or Medicine, was the key that unlocked practical mRNA medicine.

The COVID-19 pandemic provided the first proof of concept at scale. The Pfizer-BioNTech and Moderna vaccines, developed with unprecedented speed using mRNA technology that had been in development for years, demonstrated that mRNA could be delivered safely and effectively in billions of doses. Their success transformed the field’s ambitions entirely.

A June 2026 review in Nature Reviews Drug Discovery, “Towards mRNA therapeutics 2.0”, drew lessons from a decade of clinical trials in enzyme replacement for rare diseases, cancer immunotherapy, genome editing and immune-cell reprogramming. Its authors argue that the next stage depends on delivery systems that work inside the body, formulations quiet enough for the immune system to allow repeated doses, and ways to reach organs other than the liver.

In cancer, the evidence is early but real. In a 2024 first-in-human study published in Cell, an mRNA formulation from Elias Sayour’s group at the University of Florida triggered rapid immune activation and glioma-specific immune responses in patients with glioblastoma, one of the most aggressive brain tumours. In October 2025 a Nature study from researchers at MD Anderson and the same Florida group reported that, across large retrospective cohorts, patients who received a COVID-19 mRNA vaccine within 100 days of starting checkpoint immunotherapy lived significantly longer; in mice, the vaccines made tumours more responsive to that treatment. The connection between RNA immunotherapy and tumour genetics is developed in our article on the genetics of cancer.

RNA Interference: Silencing Genes with Molecular Precision

In 1998, Andrew Fire and Craig Mello discovered that double-stranded RNA — RNA with two complementary strands — could silence specific genes in C. elegans with extraordinary potency. The silencing was sequence-specific: double-stranded RNA matching any given gene would knock down the expression of that gene dramatically. They called the phenomenon RNA interference, or RNAi. The 2006 Nobel Prize in Physiology or Medicine went to Fire and Mello for this discovery — one of the fastest Nobel Prize recognitions in the history of the award.

RNAi turned out to be a widespread biological phenomenon, conserved across plants, animals, and fungi, that operates through a defined molecular pathway. Double-stranded RNA is processed by an enzyme called Dicer into small interfering RNA (siRNA) duplexes of approximately 21 nucleotides. These siRNA molecules are loaded into a multi-protein complex called RISC — the RNA-induced silencing complex. RISC uses one strand of the siRNA as a guide to find complementary sequences in target mRNAs and cleaves them, preventing their translation into protein. MicroRNAs, described above, work through a closely related pathway, using RISC to silence target mRNAs through imperfect complementarity rather than the perfect match required for siRNA cleavage.

The therapeutic potential of RNAi was immediately apparent: if you could deliver a synthetic siRNA matching any disease gene to the relevant cells, you could silence that gene with molecular precision. The challenge was delivery — naked siRNA is rapidly degraded in the bloodstream and does not efficiently enter cells. After years of intensive research into delivery strategies, the field reached a milestone in 2018 when the FDA approved patisiran (Onpattro) — the first siRNA drug — for the treatment of transthyretin amyloidosis, a progressive and fatal disease caused by the accumulation of misfolded transthyretin protein. Patisiran uses lipid nanoparticles to deliver siRNA targeting TTR mRNA to liver cells, silencing the gene and dramatically reducing transthyretin protein levels.

More siRNA drugs have followed, and the pipeline has expanded to include the liver, eye, and increasingly the central nervous system. Inclisiran, which uses a GalNAc-conjugated siRNA to silence PCSK9 in the liver — reducing LDL cholesterol levels by approximately 50 percent with just two injections per year — is approved in multiple countries and is one of the most promising cardiovascular therapeutics in recent years. The connection between this approach and the gene editing pipeline is direct: where CRISPR permanently edits the PCSK9 gene, inclisiran suppresses it reversibly through RNA interference. For the full story of how CRISPR and gene editing are changing medicine, see our article on what is CRISPR? the gene editing revolution rewriting human medicine.

The Numbers Behind the Reversal

GENCODE, the annotation used by Ensembl and most human genomics work, currently records 19,442 protein-coding genes against 35,885 long non-coding RNA genes, plus a further 7,608 small non-coding RNA genes. Of the 78,733 genes annotated in total, only about a quarter code for protein. On the older definition — a gene is a stretch of DNA specifying a protein — three quarters of what we now call genes would not qualify.

A caution is warranted here, and it is one the field takes seriously. Annotating a stretch of DNA as a long non-coding RNA gene means a transcript has been detected from it. It does not automatically mean that transcript does something. Some almost certainly represent transcriptional noise rather than function, and distinguishing the two is unfinished work. The figure of 35,885 is a count of catalogued transcripts, not a count of confirmed jobs.

The clinical side needs less hedging. Registry searches on 2 October 2026 return 771 studies for mRNA vaccine, 217 for antisense oligonucleotide and 208 for siRNA, 1,196 in all, against 1,181 when this article was first written on 22 August: fifteen registrations in six weeks, the pace of an established programme rather than a boom. A registered study is a declaration of intent, not a result, but antisense and siRNA drugs are approved and in use, and mRNA vaccines have been given in billions of doses.

What makes RNA attractive as a drug is precisely what made it seem unimportant as biology. It is transient. A protein drug or a gene edit is a lasting intervention; an RNA therapeutic delivers an instruction that the cell executes and then degrades. For a vaccine that is ideal. For a disease requiring permanent correction it is a limitation, which is why the same molecule underpins both the fastest-moving and the most repeat-dose-dependent therapies in medicine.

RNA in Epigenetics: The Regulator of Regulators

RNA and epigenetics are more deeply intertwined than is often appreciated. The non-coding RNAs that regulate genome organisation — lncRNAs, small interfering RNAs derived from repetitive elements, and PIWI-interacting RNAs (piRNAs) — are among the primary mechanisms through which the epigenome is established and maintained.

PIWI-interacting RNAs are a class of small non-coding RNAs, 24 to 31 nucleotides long, expressed primarily in the germline — the egg and sperm cells — where they silence transposable elements: the “jumping genes” that make up approximately half the human genome and would, if left unsilenced, insert themselves into random locations in the genome and potentially disrupt gene function. piRNAs silence these transposable elements through a combination of RNA interference and DNA methylation, protecting genome integrity in the cells that pass genetic information to the next generation.

Polycomb repressive complex 2, one of the main complexes that silence developmental genes, binds RNA, and some lncRNAs are thought to help guide it to the genes it silences, though how much this matters is still debated. The link between RNA and chromatin is one of the most active areas of epigenetics research, with direct relevance to how some biological information appears to be inherited across generations.

RNA as Biomarker and Diagnostic Tool

RNA’s dynamic nature — the fact that it reflects the current activity of the genome rather than its static sequence — makes it an ideal diagnostic tool. The transcriptome, the complete set of RNA molecules expressed in a cell or tissue at a given moment, is a readout of cellular state with extraordinary information content.

Single-cell RNA sequencing (scRNA-seq), developed and rapidly refined in the 2010s, can now measure the transcriptome of thousands or millions of individual cells simultaneously, generating maps of cell type and cell state at single-cell resolution. The Human Cell Atlas project — an international effort to create a reference map of every cell type in the human body using scRNA-seq and related technologies — has already characterised millions of cells from dozens of organs and tissues, discovering new cell types and revealing previously unknown cellular diversity in every organ examined.

Cell-free RNA in blood and other body fluids — including fragments of mRNA, miRNA, and circRNA — can be detected and quantified by liquid biopsy, providing non-invasive windows into tissue-specific gene expression. Specific miRNA profiles in blood are being studied as early signs of certain cancers. The stability of circular RNAs makes them particularly attractive as liquid biopsy biomarkers for diseases including Alzheimer’s disease, where circRNA profiles in cerebrospinal fluid may reflect brain-specific gene expression changes.

The connection between RNA biomarkers and computational genomics is direct: the same transformer architecture now being trained to read the regulatory genome predicts RNA expression levels across cell types as one of its primary outputs, connecting the sequence of the genome to the transcriptome it produces. That convergence of AI and molecular biology is the subject of our article on how large language models actually work — the systems that learned to read text are now learning to read RNA.

What Comes Next: The RNA Therapies of the 2030s

Circular RNA therapies are moving toward the clinic. Their closed loops resist degradation, which could mean longer-lasting protein production from a single dose than conventional linear mRNA, and several academic and commercial groups are developing them for protein replacement.

RNA base editing — using adenosine deaminase acting on RNA (ADAR) enzymes, either endogenous or delivered therapeutically, to make specific A-to-I edits in RNA rather than in DNA — offers a form of genetic correction that is inherently transient and reversible. Unlike DNA base editing, RNA base editing does not permanently alter the genome: the edited RNA is eventually degraded and replaced by transcripts from the unchanged DNA. This transience is both a limitation and a safety advantage, depending on the application. Several companies are developing ADAR-based RNA editing approaches for liver and other organ diseases. For how this fits into the wider editing landscape, see our article on gene editing in 2026.

RNA medicine now spans mRNA, siRNA, antisense oligonucleotides, circular RNA and RNA editing, each suited to different diseases and different delivery problems.

The connection between this therapeutic revolution and the foundational genetics described in our guide to what is DNA? is direct and profound. DNA carries the instructions. RNA reads them, executes them, regulates them, and — it now turns out — can be targeted to fix them when they go wrong. The molecule that was once dismissed as a molecular middleman has turned out to be the most therapeutically accessible layer of the genome. RNA did not just turn out to be more important than we thought. It turned out to be central to almost everything.

Frequently Asked Questions

Sources

1. Papastavrou, N., Horning, D.P. & Joyce, G.F. (2024). RNA-catalyzed evolution of catalytic RNA. PNAS, 121(11). doi.org/10.1073/pnas.2321592121
2. Salk Institute (4 March 2024) — Modeling the Origins of Life: New Evidence for an RNA World
3. Nature Reviews Drug Discovery — Towards mRNA Therapeutics 2.0 (June 2026)
4. Chemistry World — How RNA Reveals Clues to Life’s Origins (February 2026)
5. Nature Communications — Organ-Selective mRNA Delivery via Lipid Nanoparticles (December 2025)
6. Nature Reviews Molecular Cell Biology — The RNA World Hypothesis (2022)
7. Nobel Prize — Fire and Mello: RNA Interference (2006)
8. Nobel Prize — Karikó and Weissman: mRNA Modifications (2023)
9. Nobel Prize — Ambros and Ruvkun: MicroRNA Discovery (2024)
10. NCBI Bookshelf — The RNA World and Origins of Life (Molecular Biology of the Cell)
11. GENCODE release 50 — human gene annotation statistics, EMBL-EBI (checked 2 October 2026)
12. ClinicalTrials.gov — registered RNA therapeutic studies (queried 22 August and 2 October 2026)
13. Djebali, S. et al. (2012). Landscape of transcription in human cells. Nature, 489, 101–108. doi.org/10.1038/nature11233
14. Mendez-Gomez, H. R. … Sayour, E. J. (2024). RNA aggregates harness the danger response for potent cancer immunotherapy. Cell, 187(10), 2521–2535. doi.org/10.1016/j.cell.2024.04.003
15. Grippin, A. J. … Sayour, E. J. & Lin, S. H. (2025). SARS-CoV-2 mRNA vaccines sensitize tumours to immune checkpoint blockade. Nature, 647, 488–497. doi.org/10.1038/s41586-025-09655-y