11.1 Chromatin remodelling #
11.1.1 The SWI/SNF idea #
A promoter on a nucleosome is a poor landing site for the general factors. Remodellers use ATP to slide, eject, or unwrap nucleosomes. The motor is a Snf2-family ATPase. It does not acetylate or methylate tails. Those are the writers of Chapter 2.
Bromodomains on the complex read acetyl-lysine, so the mark recruits the remodeller, and opening the DNA helps further writers and transcription factors. ISWI, CHD, and INO80 complexes share the ATPase logic and differ in the product and in how they are addressed. Repression can be a nucleosome parked over the start site, or compaction by readers of H3K27me3 and H3K9me3. Naked DNA is not the eukaryotic default.
11.2 How transcription factors read DNA #
The study of How transcription factors read DNA provides fundamental principles governing cellular architecture, molecular interactions, and physiological regulation.
Dna Transcription
11.2.1 Four recognition strategies #
| Domain | Fold | What touches the bases | Anchor |
|---|---|---|---|
| Helixβturnβhelix | two helices and a turn | the second helix, in the major groove | LacI, CAP, Ξ» Cro; homeodomains are the eukaryotic relative |
| C2H2 zinc finger | Ξ²Ξ²Ξ± module, Zn2+ on two Cys and two His | one helix per finger; fingers are arrayed | TFIIIA, Sp1 |
| Leucine zipper | coiled coil, Leu every seventh residue | the basic regions, and only as a dimer | FosβJun. The leucines do not bind DNA |
| Nuclear receptor | two zinc modules of another class, plus a ligand-binding domain | the zinc modules, as a dimer on a hormone-response element | glucocorticoid receptor and other steroid receptors |
| Basic helixβloopβhelix | a dimer fold often confused with the zipper | basic regions | MyoD; its developmental job is not this course |
Contacts are hydrogen bonds and shape fit to base edges, mostly in the major groove, plus non-specific phosphate contacts. A steroid is not a transcription factor.
11.2.2 Combinatorial control #
A eukaryotic control region is a cluster of sites, not one operator. Output depends on which factors are present, which partners they dimerise with, and whether the complex brought a coactivator or a corepressor. Coactivators include Mediator contacts and acetyltransferases such as p300/CBP. Corepressors include deacetylases. The same protein can activate one gene and repress another because the company changes.
The interferon-Ξ² enhanceosome is the molecular case: several activators assemble cooperatively on a short region, and the set, not any one protein, is the switch. Stripe enhancers in the Drosophila embryo make the same point in space. About 20,000 protein-coding genes do not include a private gene for each cell type, because combinations scale faster than a list of types.
11.3 Methylation, and what epigenetic means #
11.3.1 CpG, DNMT, and a limit on the word #
Vertebrate regulatory methylation is 5-methylcytosine, almost only at CpG. CpG islands, about a kilobase of unusually high CpG density, often at promoters, are usually unmethylated where the gene can be expressed. Methylation in a gene body is not the same signal and can coexist with transcription.
After replication the site is hemimethylated. UHRF1 recognises that state and recruits DNMT1, which methylates the new strand. DNMT3A and DNMT3B write new methylation. TET enzymes oxidise 5-methylcytosine and open a route back, so the mark is reversible. MeCP2 reads methyl-CpG; mutations in its gene cause Rett syndrome.
Epigenetic, as marked in this course, means a difference in activity or phenotype transmitted through cell division that is not a change in DNA sequence. Maintenance methylation is the clear case. A hormone receptor occupying a promoter only while hormone is present is regulation, not epigenetic inheritance, unless the state continues after the signal has gone. Transient acetylation is not, by itself, epigenetic. The word is not a synonym for chromatin, nor for any control that is not a mutation. In the mammalian germ line most methylation is erased and reset, so an ordinary exposure does not become a mark in grandchildren merely because DNMT1 exists. Imprinting is a real parent-of-origin exception: at IGF2/H19, a control region methylated on the paternal chromosome causes the two genes to be expressed from opposite parental alleles.
11.4 Small RNAs #
11.4.1 miRNA, siRNA, Dicer, and RISC #
MicroRNAs are genomic, usually pol II transcripts with a hairpin. Drosha and DGCR8 crop it in the nucleus to a pre-miRNA of about 70 nucleotides. Exportin-5 carries it out. Dicer cuts the loop off and leaves a duplex of about 22 bp.
Small interfering RNAs come from long double-stranded RNA, viral or experimental or, in some organisms, copied by an RNA-dependent RNA polymerase. Dicer cuts that duplex directly into pieces of about 21β23 bp.
Complementarity decides the outcome. Extensive pairing lets the slicer Argonaute, AGO2 in humans, cut the target. That is typical of an siRNA. Partial pairing, the animal microRNA pattern, is concentrated in the seed, nucleotides 2β8, often at several sites in a 3β² untranslated region. The mRNA is not cut. Translation is repressed, the tail shortens, and the mRNA decays. A knockdown by this route destroys RNA. It does not change the gene, and it is not the knockout in Chapter 12.
11.5 A long non-coding RNA with one job #
11.5.1 Xist #
Long non-coding RNAs are defined by length and by lacking a long open reading frame, not by a shared mechanism. Xist is the example that has a demonstrated job.
It is transcribed from the X-inactivation centre of the X that will be silenced. The RNA is many kilobases long and does not encode the silencing proteins. It coats that chromosome in cis and does not coat the other X. The coat recruits Polycomb, so H3K27me3 rises, acetylation falls, and DNA methylation follows. The Barr body is the compact chromosome a microscope sees. In the mouse, the antisense RNA Tsix opposes Xist on the future active X. Other long non-coding RNAs exist, a few with documented roles, many without one. Xist does not licence a claim that the class controls the genome.
11.6 Key terms, summary, and self-check #
Key terms. SWI/SNF; helixβturnβhelix; zinc finger; leucine zipper; nuclear receptor; DNMT1; CpG; Dicer; seed sequence; Xist.
Summary. SWI/SNF spends ATP to move nucleosomes and reads acetyl marks; it does not write them. Helixβturnβhelix proteins, zinc fingers, leucine zippers, and nuclear receptors read DNA by different folds, and the zipper dimerises rather than touching the bases. Eukaryotic genes respond to combinations of factors and cofactors. DNMT1 copies CpG methylation onto the new strand; that copying is what “epigenetic” should mean, not regulation in general. Dicer loads a guide into RISC, and complementarity chooses slicing or repression. Xist RNA coats one X in cis and brings Polycomb. It is one long non-coding RNA with one job.
Self-check.
1. What does SWI/SNF change about a nucleosome, and which covalent mark does it not itself write?
2. Which part of a leucine-zipper protein binds DNA, and which part only dimerises?
3. Why is DNMT1 maintenance methylation epigenetic, while a hormone-bound receptor at a promoter is not, by itself?
4. What feature of the pairing decides whether RISC cuts an mRNA or only represses it?
5. What does Xist coat, and why does that not define a mechanism for every long non-coding RNA?
—
—
—