How long does mRNA last in a cell?
Half-lives from bacteria to human cells, and what happens to vaccine mRNA
Matic Broz, PhD11 min
A messenger RNA in a human cell usually lasts a few hours. Published medians for mammalian cells run from about 2 to 10 hours, depending on the cell line and how the decay was measured. The shortest-lived messages disappear within minutes, while globin mRNA in developing red blood cells lasts for days. Bacterial mRNA is far more fleeting: the median transcript in E. coli has a half-life of about 5 minutes.
Vaccine mRNA is a different case. It is chemically modified and packaged in lipid nanoparticles, and small human studies have found traces of it in blood for up to 28 days and in lymph nodes for up to 60 days after injection. Those are the latest detections in a few people, not a typical lifetime, and the sections below explain both sets of numbers.
How long does mRNA last in a cell?
Most cellular mRNAs have half-lives between a few minutes and several hours, with the shortest typical values in bacteria and the longest in mammalian cells. In E. coli, Bernstein and colleagues measured a median half-life of 4.7 minutes in rich medium and 5.4 minutes in minimal glucose medium, with 99% of transcripts between 1 and 18 minutes.[1]
Budding yeast sits in between, although its figure depends strongly on the assay. A pulse-chase study with the label 4-thiouracil found a median of 18 minutes.[2] A later method from the same laboratory, which labels RNA without first loading and then washing out the label, reported an average of only 4.8 minutes.[3]
Mammalian cells keep their mRNAs much longer, but the published medians spread across a fivefold range. The same mouse fibroblast line, 3T3, gives three answers: 110 minutes in a 2024 subcellular labeling study, 274 minutes in a 2009 study with a one-hour 4-thiouridine pulse, and 9 hours in a 2011 study that also measured protein turnover.[4][5][6] Human cell lines fall in the same window, from 200 minutes in K562 leukemia cells to about 10 hours in HepG2 liver cells.[4][7]
| Organism and cells | Half-life | Method | Study |
|---|---|---|---|
| E. coli, rich medium | 4.7 min | Rifampicin transcription arrest | Bernstein et al. (2002) |
| E. coli, minimal glucose medium | 5.4 min | Rifampicin transcription arrest | Bernstein et al. (2002) |
| Budding yeast | 4.8 min (mean) | 4-thiouracil labeling to steady state | Chan et al. (2018) |
| Budding yeast | 18 min | 4-thiouracil pulse-chase | Munchel et al. (2011) |
| Mouse 3T3 fibroblasts | 110 min | 4sU labeling by compartment (TimeLapse-seq) | Ietswaart et al. (2024) |
| Human K562 cells | 200 min (3.3 h) | 4sU labeling by compartment (TimeLapse-seq) | Ietswaart et al. (2024) |
| Mouse NIH-3T3 fibroblasts | 274 min (4.6 h) | One-hour 4sU pulse | Friedel et al. (2009) |
| Human B cells | 315 min (5.3 h) | One-hour 4sU pulse | Friedel et al. (2009) |
| Human HeLa cells | 6.9 h (mean) | BrU pulse-chase (BRIC-seq) | Tani et al. (2012) |
| Mouse embryonic stem cells | 7.1 h | Actinomycin D transcription arrest | Sharova et al. (2009) |
| Mouse NIH-3T3 fibroblasts | 9 h | 4sU pulse labeling | Schwanhäusser et al. (2011) |
| Human HepG2 cells and fibroblasts | About 10 h | Actinomycin D transcription arrest | Yang et al. (2003) |
Much of the spread comes from the methods. Transcription-arrest experiments add a drug such as actinomycin D and follow the remaining mRNA, but shutting down transcription disturbs the cell. Labeling methods leave transcription running and tag new RNA, but the label can stress cells, and the result depends on how long the pulse lasts and which kinetic model is fitted.[10] A half-life worth quoting therefore names its cell type and assay.
What does mRNA half-life mean?
An mRNA half-life is the time it takes for half of a population of one transcript to be degraded. It is not an expiry date for each molecule. If decay is exponential, 50% remains after one half-life, 25% after two and about 3% after five, so a transcript with a 7-hour half-life still leaves a measurable trace a day and a half later.
The average lifetime of a single molecule is longer than the half-life by a factor of about 1.44, the reciprocal of the natural logarithm of 2. A median half-life of 9 hours therefore corresponds to an average mRNA lifetime of about 13 hours.
The proteins an mRNA encodes usually outlast it. Schwanhäusser and colleagues measured both in the same mouse fibroblasts and found a median protein half-life of 46 hours, about five times the 9-hour mRNA median, with no correlation between the two half-lives gene by gene.[6] Protein half-lives vary even more widely, which is why protein made from a short-lived message can persist long after the message is gone.
Which mRNAs are shortest- and longest-lived?
Short-lived mRNAs tend to encode regulators. In mouse embryonic stem cells, fewer than 100 of 19,977 measured genes had half-lives under 1 hour, among them the growth regulator Myc and the developmental genes Hes5 and Foxa2. Transcription factors were enriched among short-lived mRNAs, and metabolic and structural genes among long-lived ones.[9] A HeLa study found the same split around 4 hours, with regulatory mRNAs below it and housekeeping mRNAs above.[8]
Some messages last only minutes. FOS mRNA, made when growth factors stimulate a fibroblast, is degraded within minutes of reaching the cytoplasm, driven by at least two independent instability elements, one of them in its coding region.[11] In E. coli, the quickest transcripts have half-lives of 1 to 2 minutes.[1] A short life lets the cell switch such genes off soon after the signal stops.
At the other end are the globin mRNAs that red blood cell precursors use to make hemoglobin. Jiang and colleagues summarize in vivo estimates of 24 to 60 hours for human α-globin mRNA and more than 16 to 20 hours for β-globin mRNA. These messages keep being translated for up to a week after the cell has condensed and expelled its nucleus and can no longer make new RNA.[12]
| mRNA | Half-life | Cells |
|---|---|---|
| FOS, an immediate-early gene | Minutes | Growth factor-stimulated fibroblasts |
| Fastest bacterial mRNAs | 1 to 2 min | E. coli |
| Myc, Hes5, Foxa2 and other regulators | Under 1 h | Mouse embryonic stem cells |
| Regulatory mRNAs | Under 4 h | Human HeLa cells |
| Housekeeping mRNAs | 4 h or more | Human HeLa cells |
| β-globin | More than 16 to 20 h | Erythroid cells |
| Human α-globin | 24 to 60 h | Erythroid cells, in vivo estimates |
What makes some mRNAs last longer than others?
An mRNA's lifespan is set mostly by how fast the cell removes its protective ends. In eukaryotes, decay usually begins when enzymes shorten the poly(A) tail. The RNA then loses its 5′ cap and is digested from the 5′ end, or it is degraded from the 3′ end by the exosome. Sequences and proteins bound to the message speed up or slow down each step.[13]
AU-rich elements in the 3′ untranslated region are the best-known accelerators. Across mouse stem cell genes, these elements and binding sites for PUF proteins were associated with shorter half-lives, while more exon junctions per length of coding sequence were associated with longer ones.[9]
Codon choice matters too, though how much is debated. In yeast, Presnyak and colleagues found stable mRNAs enriched in optimal codons, and swapping optimal codons for synonymous non-optimal ones destabilized a message dramatically.[14] Chan and colleagues later concluded that competition between translation initiation and decay, not codon optimality, is the main determinant in yeast.[3]
Sequence tools can change or inspect these features but cannot predict a half-life inside a cell. Codon optimization rewrites a coding sequence with synonymous codons for a target organism, and the RNAfold web server, part of ViennaRNA, predicts the secondary structure that also influences stability. Both help compare designs; neither replaces a decay measurement.
How long does mRNA from a vaccine last in the body?
Vaccine mRNA is built from the same parts that stabilize natural messages. The 4,284-nucleotide mRNA in the Pfizer-BioNTech vaccine BNT162b2 has a 5′ cap, a 5′ untranslated region taken from human α-globin mRNA, and a segmented poly(A) tail, and every uridine is replaced with N1-methylpseudouridine.[15] In mice, mRNA containing the related pseudouridine was translated more efficiently, persisted at higher levels in the spleen 24 hours after injection, and avoided the interferon response that unmodified mRNA triggered.[16]
The spike protein the vaccine encodes appears in blood quickly and clears within weeks. In 13 recipients of Moderna's mRNA-1273, the S1 part of spike was detectable from day 1, peaked on average 5 days after the first injection, and was undetectable in everyone by day 14. Whole spike was found in 3 people, on average 15 days after the first dose. After the second dose, neither was detected through day 56, apart from one sample taken the day after that injection.[17]
The mRNA itself lingers longest in the lymph nodes that drain the injected arm. In lymph node biopsies from 7 people taken 7 to 60 days after the second dose, Röltgen and colleagues found vaccine mRNA in germinal centers on days 7, 16 and 37, with a weaker signal on day 60, and spike protein also present at day 60.[18] An autopsy study found vaccine RNA in the armpit lymph nodes of 8 of 11 tested patients who died within 30 days of vaccination and in none of the 8 who died later. It was not found in the spleen or liver.[19]
Traces also turn up in blood. Sequencing plasma from 108 patients being monitored for hepatitis C found full-length or partial vaccine sequences in 10 samples, up to 28 days after vaccination.[20]
| What was measured | Sample | People | Latest detection |
|---|---|---|---|
| S1 spike fragment | Plasma | 13 | Undetectable in all by day 14 after the first dose |
| Whole spike protein | Plasma | 13 | Average day 15 after the first dose, in 3 people |
| Vaccine mRNA | Plasma | 108 | Day 28, in 10 people |
| Vaccine RNA | Armpit lymph nodes at autopsy | 20 | Within 30 days; none after 30 days |
| Vaccine mRNA and spike protein | Armpit lymph node biopsies | 7 | Day 60 after the second dose, weaker signal |
These windows describe how long sensitive assays can find something, not how long a working message survives. The autopsy assay amplified short stretches of about 160 to 190 nucleotides and could not rule out degraded RNA, and the plasma study counted partial sequences as well as full-length ones.[19][20] None of these studies measured a half-life, and each followed a small group.
How much RNA is in a human cell?
A typical mammalian cell contains about 10 to 30 picograms of total RNA. Roughly 80% to 90% of that RNA mass is ribosomal RNA, while mRNA contributes only about 1% to 7%, depending on the reference and cell type.[21][22]
Reference estimates put the mRNA pool at about 300,000 to 1 million molecules in a mammalian cell. A direct calibration of RNA sequencing with single-molecule imaging estimated about 300,000 mRNA molecules in a human U2-OS cell.[21][23]
An average mammalian mRNA is about 1,700 to 2,000 nucleotides long in commonly used reference estimates.[21][22] This is an average across many transcripts, not a standard size. Cells contain tens of millions of RNA molecules in total because short, abundant tRNAs and the rRNAs in millions of ribosomes far outnumber most individual mRNAs.
That small pool still controls protein production, because each transcript can be read repeatedly and often by several ribosomes at once. The broader RNA types differ greatly in abundance, size, function, and stability.
