m6A: Impact on RNA Stability, Splicing, and Cellular Health

N6-methyladenosine, commonly abbreviated as m6A, is the most abundant internal chemical modification found on messenger RNA in mammals, and it acts as a master switch that influences whether a given RNA molecule gets translated into protein, chopped up for recycling, spliced into alternative forms, or shuttled to a specific location in the cell. Far from being a static mark, m6A is added and removed dynamically by dedicated enzymes, and specialized “reader” proteins interpret each mark to trigger distinct downstream effects. The modification touches nearly every aspect of cellular life, from how stem cells decide their fate to how neurons strengthen or weaken their connections, and its dysregulation is increasingly linked to cancer, metabolic disease, and neurodegeneration.

How the Mark Gets Written and Erased

m6A is installed on RNA by a molecular machine built around two partner proteins, METTL3 and METTL14, which lock together to form an obligate pair. METTL3 is the catalytically active half, using a cofactor called SAM to transfer a methyl group onto adenosine, while METTL14 stabilizes the RNA substrate and helps position it correctly.¹ A handful of additional proteins join this complex to help guide it to its targets, and the modification tends to land in a loose consensus sequence known as DRACH, where the methylated adenosine sits in the middle. Not every adenosine in every transcript gets methylated; the process is selective, shaped by both sequence context and the local landscape of chromatin and other regulatory signals.

Removal of m6A is handled by two “eraser” enzymes, FTO and ALKBH5, both of which are iron-dependent oxidative demethylases. Despite acting on the same mark, they work through fundamentally different chemistry. ALKBH5 strips the methyl group off in a single step, converting m6A directly back to unmodified adenosine and releasing formaldehyde rapidly. FTO, by contrast, follows a two-step path: it first oxidizes m6A into an intermediate form called hydroxymethyladenosine, and only slowly converts that to plain adenosine with delayed release of formaldehyde.² Structural studies have shown that ALKBH5 prefers substrates containing the same consensus motif where m6A typically appears, and a specific trio of amino acids within the enzyme (Arg130, Lys132, and Tyr139) enables its rapid, one-step demethylation by forming a transient covalent intermediate with the methyl group.³ The existence of these two kinetically distinct erasers means the cell can fine-tune not just whether m6A is removed, but how quickly, producing different intermediate products along the way.

Readers Decide What the Mark Means

An m6A mark on its own does nothing. Its biological consequences depend entirely on which “reader” protein recognizes it. The best-studied family of readers are the YTH-domain proteins, and each member channels the signal in a different direction. YTHDF2, for instance, is a major driver of RNA destruction. When it binds an m6A-marked transcript, it recruits degradation machinery through at least three independent routes: it can trigger deadenylation (shortening the protective tail at the end of the RNA), promote decapping (stripping the protective cap at the start), or stimulate direct internal cleavage of the RNA by an endoribonuclease complex.⁴ Different segments of the YTHDF2 protein handle each pathway: one stretch interacts with a bridging protein called HRSP12 to activate the endoribonuclease route, an adjacent segment binds the helicase UPF1 to promote decapping, and a third region recruits the deadenylase complex CNOT1.⁵

Not all readers spell doom for a transcript, though. Some, like the IGF2BP family, do the opposite: they bind m6A-marked RNAs and stabilize them, extending their half-life. IGF2BP3, for example, was found to stabilize the mRNA for a kinase called MYLK in an m6A-dependent manner, and that stabilization had tangible metabolic consequences in mice, suppressing fat cell formation and improving insulin sensitivity on a high-fat diet.⁶ The fact that one modification can either accelerate or prevent RNA decay, depending on which reader protein arrives first, is central to understanding why m6A’s effects are so context-dependent.

How m6A Steers Splicing Decisions

Before a messenger RNA is finalized, it undergoes splicing, a process where certain segments (exons) are kept and others (introns) are removed. Which exons make the final cut can drastically change the protein that gets made. m6A influences these decisions through the nuclear reader protein YTHDC1, which promotes the inclusion of specific exons by recruiting one splicing factor (SRSF3) while blocking a competitor (SRSF10) from accessing the same stretch of RNA.⁷ When YTHDC1 is depleted, the exon-inclusion patterns shift to resemble what you’d see if SRSF10 were running the show unopposed, and restoring a normal (but not a m6A-binding-defective) YTHDC1 fixes the problem. So the modification doesn’t just passively sit near a splice site; it actively tips the balance between competing splicing programs.

Recent work has added another layer: YTHDC1 also recognizes m6A marks deposited by METTL16, a methyltransferase distinct from the METTL3/14 complex, on RNAs that are still physically associated with the chromosome being transcribed. This coupling between methylation and cotranscriptional splicing turns out to be important for genes involved in cell cycle progression and DNA repair during organ development.⁸ And beyond YTHDC1, m6A can reshape splicing indirectly by altering the physical structure of the RNA itself. When a methyl group is added, it can destabilize local hairpin structures, exposing stretches of RNA that were previously buried. This structural rearrangement, termed the “m6A switch,” allows proteins like HNRNPC, which help process precursor RNA, to gain access to binding sites they could not reach before.⁹

Getting RNA to the Right Place and Keeping Translation Running

YTHDC1 does double duty. Beyond splicing, it helps m6A-marked transcripts exit the nucleus and reach the cytoplasm where they can be translated. Knocking it down causes methylated mRNAs to pile up inside the nucleus while the cytoplasm becomes depleted of those same transcripts.¹⁰ In other words, m6A doesn’t just affect what a transcript says (via splicing) or how long it lasts (via decay); it also controls where the transcript ends up.

Once in the cytoplasm, m6A can influence how efficiently a transcript gets translated into protein. Normally, translation kicks off when a cap-binding complex latches onto the protective cap at the front of the mRNA and recruits ribosomal machinery. But when m6A sits in the 5′ untranslated region of a transcript, ribosomes can assemble and begin translating without the cap-binding complex at all.¹¹ This cap-independent route becomes especially valuable during cellular stress, when the normal cap-dependent pathway gets shut down. Research has shown that the level of m6A on an mRNA quantitatively determines how resistant its translation is to inhibition of the mTOR pathway, a central growth-signaling hub that cells dial down during nutrient deprivation or other stress conditions. Transcripts with enough m6A essentially maintain their translation output even when the rest of the cell’s protein production is faltering.¹²

Stress Granules and the Cellular Emergency Response

When a cell faces acute stress — heat shock, oxidative damage, viral infection — it rapidly sequesters nonessential mRNAs into dense cytoplasmic clusters called stress granules. These granules act as temporary holding pens, protecting RNA while freeing up the cell’s translational machinery for survival tasks. m6A-modified mRNAs are enriched in stress granules, and the reader proteins YTHDF1 and YTHDF3 play a structural role in forming these condensates. Both proteins contain intrinsically disordered regions that help them undergo phase separation, the biophysical process by which stress granule components coalesce from the surrounding cytoplasm. Super-resolution imaging has shown YTHDF proteins clustering at the junctions between stress granule cores, reducing the energy barrier needed for the granule to form in the first place.¹³ When YTHDF1 and YTHDF3 are depleted, stress granule assembly is impaired and m6A-marked mRNAs fail to be recruited properly. This places m6A at the heart of the cell’s triage system during emergencies.

Stem Cell Fate and Embryonic Development

Embryonic stem cells face a fundamental choice: keep dividing as undifferentiated copies of themselves or commit to becoming a specialized cell type. m6A is critical for nudging them toward differentiation. When METTL3 is genetically inactivated in mouse or human embryonic stem cells, the resulting loss of m6A on key transcripts causes prolonged expression of Nanog, a master regulator of the stem cell state. The cells become trapped in self-renewal, unable to properly exit pluripotency and commit to specific developmental lineages.¹⁴ In normal development, m6A marks the transcripts of pluripotency genes and accelerates their degradation, clearing the decks so differentiation programs can take over. When that degradation signal is missing, the pluripotency transcripts linger, and the cell stays stuck.¹⁵ This makes m6A a kind of molecular timer: by controlling how fast certain mRNAs are cleared, it determines the pace at which a stem cell transitions to a new identity.

Cancer and Leukemia Stem Cells

The same machinery that governs normal stem cell transitions can be hijacked in cancer. In acute myeloid leukemia (AML), METTL14 is required for the self-renewal of leukemia stem cells, and knocking it down promotes terminal differentiation and death of leukemic blasts. Limiting dilution assays in mice showed that METTL14 knockout reduced leukemia stem cell frequency roughly sevenfold. Key targets that METTL14 keeps active in AML cells include the oncogenic transcription factors MYB and MYC, which drive proliferation and block differentiation.¹⁶

On the reader side, YTHDF2 is overexpressed across a broad range of human AML subtypes and is essential for both disease initiation and propagation. It works by shortening the half-life of transcripts that would otherwise prime leukemia stem cells for programmed cell death. One such target encodes the tumor necrosis factor receptor TNFRSF2; when YTHDF2 is removed, that transcript accumulates and sensitizes the cells to apoptosis.¹⁷ Meanwhile, on the eraser side, small-molecule inhibitors of FTO have shown the ability to impair self-renewal in glioblastoma stem cells and slow tumor progression in both AML and glioblastoma mouse models.¹⁸ The picture that emerges is that cancer cells co-opt different nodes of the m6A network depending on context, sometimes relying on writers to maintain oncogenic transcripts, sometimes on readers to silence pro-death ones.

Metabolism and Fat Tissue

m6A regulation extends well beyond the nucleus and into whole-body metabolism. In a striking demonstration of tissue-specific effects, researchers deleted METTL14 separately in brown fat and white fat of mice and observed opposite metabolic outcomes. Losing METTL14 in brown adipose tissue improved insulin sensitivity and protected against diet-induced glucose intolerance in both male and female mice. Deleting it in white adipose tissue, on the other hand, diminished insulin sensitivity and worsened metabolic dysfunction on a high-fat diet.¹⁹ The improvement in the brown fat knockout appeared independent of body weight changes or the classic heat-generating function of brown fat, suggesting that m6A controls a distinct metabolic signaling axis within those cells.²⁰ These findings highlight why blanket manipulation of m6A across the whole body could be counterproductive: the same modification can be metabolically protective in one tissue and harmful in another.

Synaptic Plasticity and Neurodegeneration

Neurons are among the most m6A-rich cell types, and the modification plays a direct role in synaptic plasticity, the process by which connections between neurons strengthen or weaken in response to experience. In rats, pharmacological inhibition of METTL3 with the compound STM2457 blocked long-term potentiation (LTP), a cellular correlate of memory formation, without affecting baseline synaptic transmission. Conversely, experimentally raising m6A levels, whether by activating METTL3/14 or inhibiting FTO, also attenuated LTP as well as long-term depression. The results fit a model in which m6A acts as a stabilizing signal: some is needed to consolidate plasticity, but too much clamps down on further change.²¹

This has implications for aging and disease. Studies in aged mice found a broad decline in m6A levels in the brain, and many of the transcripts that lost methylation were linked to synaptic function, including those encoding the signaling kinase CaMKII and the glutamate receptor GluA1. Reduced m6A on those transcripts corresponded with decreased local protein synthesis at synapses, and similar m6A changes were observed in brain tissue from Alzheimer’s disease patients.²² While this does not prove that m6A loss causes cognitive decline, it suggests that the erosion of epitranscriptomic marks with age could weaken the molecular infrastructure needed to maintain synaptic strength.

Immunity and Viral Camouflage

Viruses face an immediate problem upon entering a cell: the innate immune system is primed to recognize foreign RNA and sound the alarm. Some viruses, it turns out, use m6A as camouflage. Research on nonsegmented negative-sense RNA viruses found that m6A methylation on viral RNA acts as a molecular marker that helps the host immune system distinguish self from nonself, and that viruses exploit this by methylating their own transcripts to mimic host RNA and dampen the immune response.²³ This strategy appears conserved across multiple viral families, suggesting it evolved independently more than once. From a therapeutic standpoint, stripping m6A from viral RNA or blocking the enzymes that install it on viral transcripts could potentially unmask viruses and boost the immune response, though that idea remains early-stage.

Non-Coding RNA Gets the Mark Too

m6A is not limited to messenger RNA. Long non-coding RNAs and other non-coding transcripts also carry the modification, and this methylation can regulate their stability, localization, and function. In cancer contexts, METTL3-mediated m6A has been found to stabilize specific oncogenic long non-coding RNAs, boosting their expression and promoting tumor progression in liver and nasopharyngeal cancers among other types.²⁴ Because non-coding RNAs themselves regulate gene expression through diverse mechanisms, m6A on these molecules creates additional layers of control, a modification on a regulator that regulates other regulators.

Crosstalk with Other Epigenetic Systems

m6A does not operate in isolation. It communicates extensively with histone modifications and DNA methylation, the classic epigenetic marks that sit on chromosomes rather than RNA. m6A can influence histone methylation and acetylation states, and histone marks can reciprocally affect the expression or targeting of m6A-related enzymes.²⁵ This bidirectional crosstalk means that changes in one system can ripple through the other, amplifying or buffering gene expression changes in ways that neither system would produce alone.²⁶ For researchers trying to understand gene regulation in disease, this interplay complicates the picture considerably: a drug that targets m6A writers or erasers may have unintended consequences on chromatin state, and vice versa.

Mapping the Marks at Single-Nucleotide Resolution

For years, the field’s main tool for locating m6A sites was antibody-based immunoprecipitation, which could identify the general region of an mRNA that carried the mark but couldn’t pinpoint the exact nucleotide. This limited researchers’ ability to study the functional consequences of individual m6A sites. Newer methods have largely solved this problem. One approach, called m6A-REF-seq, leverages an RNA-cutting enzyme (MazF) that is blocked by m6A, providing single-nucleotide resolution without relying on antibodies.²⁷ Another, m6A-SAC-seq, uses selective chemical labeling to achieve quantitative, transcriptome-wide mapping at single-base resolution.²⁸ These technical advances have been transformative: they reveal not just where m6A is, but how abundant it is at each site, which matters because the functional outcome often depends on the fraction of transcripts carrying the mark at a given position.

Toward Therapeutic Targeting

Given m6A’s involvement in cancer, metabolic disease, and neurological dysfunction, pharmaceutical companies have moved to develop small molecules that target the writer and eraser enzymes. The most advanced program focuses on METTL3 inhibition for cancer. In mouse models of AML, a METTL3 inhibitor blocked leukemic expansion, extended survival, and reduced AML cells in peripheral blood after transplantation, all without significant toxicity or weight loss.²⁹ Based on those preclinical results, a compound called STC-15 became the first METTL3 inhibitor approved for a phase 1b/2 clinical study in cancer patients, marking a milestone for the field.³⁰

The challenge ahead is specificity. METTL3 methylates thousands of transcripts, and broadly inhibiting it will affect many cellular processes simultaneously, as the metabolic studies showing opposite effects in brown versus white fat vividly illustrate. Future drugs will likely need to be combined with biomarkers that predict which patients’ tumors are genuinely dependent on m6A dysregulation. Early work on pharmacological tools targeting synaptic m6A in the brain suggests another therapeutic frontier, but the fact that both too little and too much m6A can impair plasticity warns that dosing precision will be critical.

Conservation Across Life

m6A is not a recent evolutionary invention. The core regulatory machinery, including the writers, erasers, and readers, is conserved across the plant kingdom, from mosses to flowering crops.³¹ In plants, m6A governs growth, development, and stress responses, performing analogous roles to its functions in animals. This deep conservation suggests that RNA methylation was already an important regulatory layer in the common ancestor of plants and animals, hundreds of millions of years ago. It also means that insights from plant biology can sometimes inform animal research and vice versa, an unusual bridge in molecular biology that has accelerated progress in both domains.