Not the One Who Copies, but the One Who Permits — 1,639 Transcription Factors and a Paused Polymerase
The question in transcription is not what does the copying. It is who gets the door opened. And once the door is open, the machine stops once more.
An earlier piece settled the difference between transcription and translation in a line: whether the language changes.
That is an exam answer. The question that remains thirty years later is elsewhere.
Not what does the copying, but who gets the door opened.
Inside the dome it is almost all water
Take a cell as a dome 50 metres across. Five million times. Human cells run 10–20 micrometres; 10 is the reference here.
Inside is not empty. Nor is it packed with protein.
Close to 99 percent of the molecules in a cell are water.
By mass water is about 70 percent and protein 15–20, but counted as molecules water dominates overwhelmingly. Proteins the size of quail eggs drift and collide through a fog of it.
At the centre is a room 30 metres across: the nucleus. A human somatic nucleus is 5–6 micrometres, which is 30 metres here.
Inside it, forty-six threads a centimetre thick, each averaging 215 kilometres in length.
Not held together by negative charge
One line in the draft needs correcting.
“Both are negatively charged. So they are always unstable. So neither can exist alone.”
The first half is right and the second is backwards.
Both strands are negative because of their phosphates. Like charges repel. Negative charge is not what joins the strands; it is what pushes them apart.
Two forces hold them.
| Force | What it is |
|---|---|
| Hydrogen bonds | two for A-T, three for G-C |
| Base stacking | hydrophobic interaction of bases piled on one another |
Quantitative work finds stacking contributes more to thermal stability. The helix does not stand on hydrogen bonds alone.
So what of the repulsion? It is neutralised. Cations such as magnesium bind, and histones wrap it in positive charge.
This is where the need for histones becomes exact. Not protection — without neutralising the repulsion, that length does not fit in that room.
And eight histones form an octamer, not an octagon. The shape is closer to a spool.
A blueprint guarded too well
The consequence is this.
A single cell actually expresses roughly 11,000 to 13,000 protein-coding genes. The human genome has about 20,000, so a little over half.
The rest stays wound. That is how a liver cell and a neuron carry the same blueprint and become different bodies.
Guarding the blueprint succeeded. The cost is that the blueprint is not easily read.
Visitors with a pass — 1,639 of them
A transcription factor is a protein that recognises a specific DNA sequence and binds it.
The human set was catalogued in 2018. 1,639.
And they do not copy. They do not touch the DNA. They do one thing.
They bind and make a signal.
Co-factors arrive at the bound site — histone acetyltransferases such as p300/CBP. Acetyl groups attach to lysines, positive charge is neutralised, and the winding loosens.
The mechanism from the previous instalment: held by charge, released by erasing charge.
The ones that bind a closed door
Which creates a problem. For a transcription factor to reach its site, the site must be open. To open it, a transcription factor must bind.
Chicken and egg.
The resolution is the pioneer factor. A small number of transcription factors — FoxA1, Oct4, Sox2 — can bind DNA while it is still wrapped on a nucleosome.
| Type | Where it binds |
|---|---|
| Ordinary factor | regions already open |
| Pioneer factor | closed chromatin, directly |
Most visitors enter through an open door. Pioneers open it.
They sit at the points where cell fate changes — which is why Oct4 and Sox2 appear in the experiments that reprogram somatic cells back to pluripotency.
Only then does the machine arrive
With the door open, RNA polymerase comes.
The machine asks no questions of meaning. It judges nothing good or bad. It copies exactly as far as it is opened and permitted.
It runs at 1–4 kilobases per minute, averaging around 2.
And then it stops again.
Permission is needed twice
RNA polymerase II begins transcribing, advances 30 to 60 bases, and halts. It waits there.
Promoter-proximal pausing.
A gene in this state is ready and not started. When the signal comes, transcription resumes within seconds.
| Stage | Point of control |
|---|---|
| ① Chromatin opens | transcription factors · acetylation · remodellers |
| ② Polymerase binds | promoter |
| ③ Pause released | signals such as P-TEFb |
Genes that must switch on fast during development are the ones held this way. Opening and launching are separate decisions.
Permission is not given once. It is given twice.
Accuracy is budgeted
How accurate is transcription?
| Error rate | |
|---|---|
| DNA replication | 1 in 10^9 to 10^10 |
| Transcription | 1 in 10^5 to 10^6 |
More than a thousandfold apart. Transcription is far sloppier.
The reason is in an earlier piece. RNA degrades within minutes to hours. There is no case for spending master-copy accuracy on something that will be gone.
What is kept and what is discarded get different budgets. Life does not make everything equally accurate.
Still unknown
Where a transcription factor will bind is not yet predictable.
The same binding motif occurs hundreds of thousands of times across the genome, and only a small fraction is actually bound. What makes the difference is only partly explained.
A model in which transcription machinery gathers by liquid–liquid phase separation into condensates was proposed in 2017, and observations both for and against keep accumulating. It is unresolved.
The writer and the one who permits
Fixed here as this article’s definition.
Transcription is not the making of information but the permitting of it.
The copying machine does not judge. Judgement has already finished at the door. And it is passed twice: whether to open, and whether to launch.
Life moves less by who writes than by who permits.
In short
- Close to 99% of the molecules in a cell are water. The nucleus is 30 metres at this scale
- The two strands are held by hydrogen bonds and base stacking, not by negative charge — stacking contributes more
- Negative charge pushes apart. Histones sit on the side that neutralises the repulsion
- One cell expresses 11,000–13,000 protein-coding genes out of about 20,000
- There are 1,639 human transcription factors. They do not copy; they signal
- Pioneer factors bind closed chromatin directly; ordinary factors need an open door
- RNA polymerase II pauses 30–60 bases in and waits. Permission is given twice
- Transcription errs a thousand times more than replication, because the copy is disposable
This piece is rebuilt for search readers from “Copying” of the serial essay The Great Corona Swindle. The original’s point of view is kept; the mechanisms and figures have been reset against primary literature.
Frequently asked
- Why do the two strands of DNA stay together?
- Not because of negative charge. Both strands are negative from their phosphates, so they repel each other. What holds them is hydrogen bonding between base pairs (two for A-T, three for G-C) and base stacking — the hydrophobic interaction of bases piled on one another, which contributes more. The negative repulsion destabilises the helix, and positive charges such as magnesium ions and histones neutralise it.
- Do transcription factors do the copying themselves?
- No. A transcription factor binds a specific DNA sequence and generates a signal. It recruits co-factors that acetylate histones, and the chromatin loosens as a result. The actual reading of DNA into RNA is done by RNA polymerase. The human transcription factors number 1,639 in the current catalogue.
- Can closed regions ever be read?
- Special transcription factors called pioneer factors — FoxA1, Oct4, Sox2 among them — can bind DNA wrapped on a nucleosome. Most transcription factors enter only through doors already open; pioneer factors open the door.
- Why is transcription less accurate than replication?
- Transcription errs at about one base in 100,000 to 1,000,000; DNA replication at one in a billion to ten billion. The difference is over a thousandfold. RNA is a disposable copy degraded within minutes to hours, so the accuracy budget spent on it is smaller. Accuracy itself is budgeted.
References
- Lambert, S. A. et al. (2018). The Human Transcription Factors. Cell, 172(4), 650–665. — The catalogue of 1,639 human transcription factors and their DNA-binding domains.
- Zaret, K. S. & Carroll, J. S. (2011). Pioneer transcription factors: establishing competence for gene expression. Genes & Development, 25(21), 2227–2241. — Factors that bind closed chromatin directly.
- Adelman, K. & Lis, J. T. (2012). Promoter-proximal pausing of RNA polymerase II: emerging roles in metazoans. Nature Reviews Genetics, 13(10), 720–731. — The polymerase stopping 30–60 bases after initiation and waiting.
- Jonkers, I. & Lis, J. T. (2015). Getting up to speed with transcription elongation by RNA polymerase II. Nature Reviews Molecular Cell Biology, 16(3), 167–177. — Elongation rates of 1–4 kilobases per minute.
- Gout, J. F. et al. (2013). Large-scale detection of in vivo transcription errors. PNAS, 110(46), 18584–18589. — Measured transcription error rates against replication.
- Yakovchuk, P., Protozanova, E., Frank-Kamenetskii, M. D. (2006). Base-stacking and base-pairing contributions into thermal stability of the DNA double helix. Nucleic Acids Research, 34(2), 564–574. — Base stacking contributes more to helix stability than base pairing.
- Uhlén, M. et al. (2015). Tissue-based map of the human proteome. Science, 347(6220), 1260419. — How many genes are actually expressed per tissue — generally 11,000–13,000.
Published September 30, 2026