RNA that controls RNA
Across this rung you have watched a raw transcript get capped, tailed, and spliced, and you have met the startling idea that RNA can fold into a catalyst — a ribozyme — which is the seed of the RNA-world hypothesis that RNA came first. This last guide closes the loop with the most modern twist of all: RNA can also *govern* RNA. Short RNA molecules, far too small to code for any protein, act as guides that recognise a target message by its sequence and shut it down. The slogan for the whole RNA rung — RNA is far more than a passive messenger — finds its sharpest proof here, where RNA becomes the regulator pulling the strings.
Here is the core idea in one breath, before any names. A double-stranded RNA is a red flag in a cell — your own genes are copied into single-stranded messages, so a long double strand usually means a virus or a gene gone haywire. Cells evolved a system that treats that double strand as a wanted poster: cut it into tiny pieces, keep one piece as a mug shot, and then patrol the cell destroying every single-stranded RNA whose sequence matches. Because matching is done by plain Watson-Crick base-pairing (A-U / G-C), the system is exquisitely *sequence-specific* — it silences exactly the gene the double strand came from, and leaves the thousands of other messages untouched. That whole sequence-targeted silencing system is [[rna-interference-pathway|RNA interference]], or RNAi.
The shredder and the hunter: Dicer, Argonaute, RISC
A search-and-destroy operation needs just two pieces of kit: a shredder to chop the wanted poster into pocket-sized guides, and a hunter that takes one guide and goes looking for matches. The cell has exactly these, and naming them is half the battle. The shredder is an enzyme called Dicer — picture a precise pair of molecular scissors that grabs a long double-stranded RNA and cuts it into uniform little blocks about 21 to 23 base pairs long, each with a tidy two-nucleotide overhang at its ends. The hunter is a protein called Argonaute. Argonaute on its own is empty-handed; loaded with one guide strand it becomes the active core of the RISC complex (RNA-Induced Silencing Complex). The whole apparatus — Dicer feeding guides into Argonaute-RISC — is the engine introduced in Dicer, Argonaute and RISC.
Here is the elegant part. Each little block Dicer makes is itself double-stranded, but only one of its two strands is useful as a guide — the strand whose sequence is *complementary* to the target message. Argonaute keeps that one, called the guide strand, and discards the other (the passenger strand) — much as you keep a key's working face and throw away the blank. Now Argonaute holds a single short strand and dangles it like bait. As a message RNA drifts past, Argonaute lets the guide try to base-pair with it. A mismatch and the message floats on, ignored. A match and the message is caught, pinned against the guide as a short RNA-RNA double helix — and its fate is sealed.
long double-stranded RNA (virus / experimental)
===========================
===========================
| Dicer chops
v
~21-23 bp pieces, 2-nt overhangs (these are siRNAs)
=====================--
--=====================
| load into Argonaute, toss passenger strand
v
RISC = Argonaute + single guide strand
guide 3'-...UACGGAU...-5'
||||||| <- base-pairs to a matching mRNA
target mRNA 5'-...AUGCCUA...-3' --> CUT (siRNA) or BLOCK (miRNA)- Trigger: a long double-stranded RNA appears — from a virus, a transposon, a fold-back transcript, or one an experimenter adds on purpose.
- Dice: Dicer chops the double strand into ~21-23 bp fragments — the small interfering RNAs (siRNAs).
- Load: one strand of each fragment is handed to Argonaute as the guide; the passenger strand is thrown away. Argonaute-plus-guide is RISC.
- Search and silence: RISC scans passing mRNAs; where the guide base-pairs a match, the message is cut and destroyed (siRNA) or blocked from translation (miRNA).
microRNAs: the genome's own dimmer switches
So far the trigger has been a foreign or accidental double strand. But cells also turn the very same machinery on themselves — deliberately, and as a routine part of running their genes. A [[molbio-microrna|microRNA]] is a tiny RNA, only about 22 nucleotides, that the genome *encodes on purpose* to throttle its own messages. Picture a busy kitchen full of recipe slips (the mRNAs), and a quiet inspector who stamps certain slips with 'cook less of this'. MicroRNAs are those inspectors. They never become protein themselves; their whole job is to ride in Argonaute and dial down how much protein chosen genes produce.
A microRNA is born as a longer transcript that folds back on itself into a hairpin — a stem of paired bases with a loop at the top — and that hairpin is trimmed (in two steps, the second by Dicer) down to the mature ~22-nucleotide guide. Now comes the key difference from siRNA, and it changes the outcome. A microRNA usually matches its target only *partially* — often just a short core of about six bases, the 'seed', pairs perfectly, while the rest is loose. That imperfect pairing is too weak to trigger a clean cut, so instead of slicing the message RISC mostly *jams* it: it blocks the ribosome from translating, and recruits factors that strip off the protective poly-A tail so the message decays faster. The effect is a dimmer, not an off-switch — a gene turned down rather than abolished. This is the microRNA layer of gene regulation that the later regulation rung will build on.
Do not underestimate this layer because each microRNA only nudges. Because pairing needs only a short seed, one microRNA can match — and gently restrain — hundreds of different messages at once, and the human genome encodes well over a thousand microRNAs. The result is a vast, overlapping web of fine adjustments that helps set how much of each protein a cell makes, sharpening developmental decisions and damping noise. It is a humbling correction to the old 'one gene makes one protein, full stop' picture you met early in this ladder: even after a gene is transcribed and spliced, a crowd of tiny RNAs still gets a vote on how loudly it is finally expressed.
When is a message kept, and when is it scrapped?
Silencing by small RNAs is only one of several ways a cell decides how long a message survives. It is worth a brief look at the wider picture of mRNA stability, because the amount of any protein depends not just on how fast its message is *made* but on how fast it is *destroyed*. Every mRNA has a half-life — some last minutes, some many hours — set largely by signals in its untranslated tails. The 5' cap and the poly-A tail you met earlier in this rung act as protective bookends; as long as both are intact, the message looks 'fresh' and is read again and again. Strip the tail (as a microRNA can encourage) or remove the cap, and degrading enzymes move in. A message is kept exactly as long as its protective marks hold up.
Cells also run a quality-control inspection that throws out messages with a particular defect — a job done by [[nonsense-mediated-decay|nonsense-mediated decay]], or NMD. The flaw it hunts for is a stop codon in the wrong place: a *premature* stop, far earlier than the message's true end (a 'nonsense' codon, hence the name). Such a message would be translated into a short, broken, possibly toxic protein, so the cell shreds it before it can be read in earnest. Remarkably, the cell knows where the *real* end should be thanks to a clue left by splicing: when introns are removed, the spliceosome stamps each junction with a marker protein. On a normal message the ribosome sweeps all those markers off as it translates; if it hits a stop codon while markers still lie downstream, the stop is in the wrong place, NMD is triggered, and the faulty message is destroyed.
From cell defence to lab tool and medicine
The moment biologists understood that a short double-stranded RNA can silence any matching gene, they realised the cell had handed them a programmable off-knob. If you want to know what a gene does, feed the cell an siRNA matching that gene's message, let the cell's own Dicer-and-RISC machinery destroy it, watch the protein level fall, and see what breaks. This use of the pathway as a research method is [[molbio-rna-interference|RNA interference as a tool]] — a knockdown, since it lowers a gene rather than deleting it. It became the standard way to switch genes off across thousands of labs, and you can now run it genome-wide, testing every gene one at a time in a large-scale screen. It is worth being honest about its place today: gene editing with CRISPR often gives a cleaner, permanent knockout, but RNAi remains valuable precisely because it is reversible and tunable — a dimmer rather than a deletion.
If you can design an siRNA to silence any chosen gene, you can in principle silence a *disease* gene — and that hope has become real medicine. A handful of [[antisense-sirna-therapeutics|siRNA drugs]] are now approved, each a short synthetic RNA designed to shut down the message of a gene that, when overactive or making a toxic protein, causes illness. The practical hurdles have been real and instructive: a bare RNA is fragile and is chewed up in the bloodstream, struggles to cross into cells, and can trip the body's innate alarms for foreign double-stranded RNA. The breakthroughs came from chemically armouring the RNA against degradation and packaging or tagging it so it reaches the right tissue — most successfully the liver. These drugs do not touch your DNA; they act on the message, so the effect is potent but must be re-dosed as the silencing fades. That last point is the same honest theme: RNAi turns a gene down, it does not erase it.
Step back and see how far the RNA rung has carried you. You began with RNA as a humble courier of the gene-expression message; you watched it get capped, tailed, and spliced; you saw it fold into a catalyst; and now you have seen it become a regulator that searches out and silences other RNAs by sequence alone — a system the cell uses for defence and fine control, and one we have repurposed into a research tool and a class of drugs. That is the throughline of this whole rung: RNA is no passive messenger. With that in hand you are ready to climb on — to the ribosome that reads these messages into protein, and to the broader machinery of gene regulation where microRNAs take their place among many switches.