Seeing the invisible: the size-sorting gel
By now you can do astonishing things to DNA — cut it at a chosen sequence, splice pieces from different sources into one recombinant molecule, drop that into a vector, and let bacteria copy it. But a tube of DNA is a clear, colourless drop. You cannot see whether the cut worked, whether the right insert went in, or how big anything is. The first thing every molecular biologist needs is a way to *see* the invisible — and the humble workhorse that does it is gel electrophoresis.
The trick rests on one lucky fact you met back on the nucleic-acids rung: the sugar-phosphate backbone carries a steady negative charge all along its length, no matter the sequence. So in an electric field, *every* piece of DNA drifts the same way — toward the positive electrode. Pour a slab of jelly-like gel (agarose, a seaweed extract), punch little wells in one end, load your samples, and switch on the current. The gel is a dense microscopic mesh, like a thicket. Short fragments are nimble and thread through quickly, travelling far; long fragments snag and lag near the start. Run for a set time and the molecules fan out by size.
Stain the gel with a dye that slips between the stacked bases and glows under UV light, and the result is a ladder of sharp bright bands — each band is a pile of fragments that are all the same length. Run a lane of known size-standards alongside, and you can read each fragment's length straight off the picture. This single tool lets you confirm a restriction digest cut where you expected, check that a PCR made the right product, or cut a wanted band out of the gel to purify it. RNA runs the same way. Proteins have no uniform charge, so you first coat them with the detergent SDS to give them a matching negative charge — then they too separate purely by size (this version is called SDS-PAGE).
The blotting trio: naming one molecule in the crowd
A blot turns the anonymous smear into a single labelled answer, and it leans on a principle you already know intimately: complementary strands stick. Recall hybridization — cool two strands whose sequences match and they zip back together, A reaching across to T and G to C, finding their partner in a sea of other molecules. The same lock works between a DNA strand and a complementary RNA. So if you make a short, labelled single strand whose sequence is the complement of your target — a probe — it will base-pair only where your sequence lives, lighting up that band and no other.
- Separate: run your molecules out on a gel so they spread by size into bands.
- Blot: transfer the bands out of the soft gel onto a sturdy membrane, keeping their positions exactly — buffer wicks up through the gel by capillary action and carries the molecules onto the membrane, like ink soaking into blotting paper.
- Probe: bathe the membrane in your labelled probe; it hybridizes only to the matching band.
- Detect: read out the label — by radioactivity, colour, or light — and your one sequence appears as a single visible band at a known size.
Edwin Southern invented this in 1975 for DNA — the Southern blot, the first way ever to spot one specific sequence inside a whole complex genome, used for decades to confirm a gene's presence, count its copies, or catch a rearrangement. Apply the very same recipe to RNA and you get the Northern blot: it measures *whether a cell is using a gene right now and how much*, because the band's brightness reflects how much messenger RNA is present, and its position reveals the transcript's size — handy for spotting alternative splicing, where one gene yields RNAs of different lengths — a reminder that 'one gene, one protein' is outdated. The third sibling, the Western blot, detects *protein*, and here the probe must change because proteins do not base-pair. Instead of a complementary strand you use an antibody — an immune protein that recognizes its target by three-dimensional shape, the molecular recognition you met on the protein rung. A primary antibody sticks to your protein and a tagged secondary antibody amplifies the signal, so a band appears wherever it bound. Be honest about its weak point: a Western is only as trustworthy as its antibody, which can bind the wrong protein and paint a misleading band. (The 'Northern' and 'Western' names are puns on Southern, not compass directions.)
Southern -> DNA probe = labelled complementary DNA Northern -> RNA probe = labelled complementary nucleic acid Western -> protein probe = antibody (recognizes shape) all three: separate on gel -> blot to membrane -> probe -> detect
Watching genes glow: reporters, fusions, and tags
Blots tell you about a snapshot — a sample you ground up and ran on a gel. But often the real question is *when and where* a gene switches on inside a *living* cell or embryo, watched in real time. The problem is that most genes are invisible: their products do nothing you can see. The clever fix is to attach an easy-to-see 'tattletale' gene whose activity you *can* detect — a reporter gene.
The idea: hook a reporter up to the control sequence — the promoter — of the gene you are studying. Then whenever that promoter fires, the reporter is made too, and its glow or colour 'reports' the activity for you. Three reporters are famous. GFP (green fluorescent protein), borrowed from a jellyfish, glows green under blue light, so you can literally watch a gene switch on in a living cell, no grinding required. lacZ makes an enzyme that turns a colourless substrate blue — the very colour change behind the blue-white screening you used to pick successful clones earlier in this rung. And luciferase, the firefly enzyme, emits light, giving an extremely sensitive, quantitative readout. GFP was such a transformative reagent that its discovery and development won the 2008 Nobel Prize in Chemistry.
There is a second, subtler way to use these markers. Instead of just reporting a promoter's activity, you can *fuse* the marker directly onto a protein of interest — splice the two coding sequences in-frame so the cell builds a single fusion protein carrying both parts. Fuse GFP onto a protein and you can follow that exact protein moving around inside a living cell in real time, glowing as it goes. The trade-off is honest and worth remembering: the marker is foreign extra mass, and once in a while it perturbs the host protein's folding, activity, or location — which is why fusions are usually placed at one end of the protein rather than buried in the middle.
The same fusion idea, shrunk down, gives you the tag — a short standard handle sewn onto a protein so you can always grab it or recognize it in a crowd. An *epitope tag* (FLAG, HA, Myc) is a tiny peptide that a ready-made antibody binds tightly, so you can detect or locate any tagged protein even when no good antibody exists for the protein itself — perfect for spotting it on a Western blot with one standard antibody. An *affinity tag* binds a specific matrix: the classic His-tag (six histidines, 5'...catcatcat...3' added in frame) sticks to a nickel column, so you flow a messy cell extract over the column, everything else washes away, and you elute one pure protein in a single step. Tags fish a single protein out of thousands — the everyday currency of protein purification.
Editing one word: site-directed mutagenesis
Now turn from watching to engineering. If you want to learn what one word in a sentence does, the cleanest experiment is to change *just that one word* and read what happens, leaving everything else untouched. Site-directed mutagenesis is that experiment for genes — making one chosen, deliberate change at a precise spot in a cloned sequence, then asking how it reshapes the protein.
The classic trick reuses base-pairing once more. You synthesize a short primer that matches your target region *almost* perfectly — except it carries the exact base change you want. You anneal it to your cloned gene: the single mismatched base bulges out, but the rest of the primer grips firmly enough to stay bound. A DNA polymerase then extends from the primer and copies the whole plasmid, faithfully building your designed change into the new strand. Remove the original unmutated template, and you are left with plasmid carrying *precisely* the mutation you specified — one substituted base, a deletion, or an insertion, exactly where you placed it. Modern kits make this routine.
What does this buy you? Site-directed mutagenesis turned proteins into editable text. Suspect a particular amino acid in an enzyme's active site does the catalysis (an idea from the enzymes guide)? Change just that one residue and watch the activity die — that single clean swap confirms the residue's role, where bulkier experiments could not. The same tool lets you repair a disease mutation in a cloned gene, or deliberately engineer a protein with improved properties. Charles Hutchison and Michael Smith pioneered it, and Smith shared the 1993 Nobel Prize for handing biology a way to rewrite a gene one letter at a time.
The payoff: making real protein, like insulin
Everything so far — the gel that lets you see, the blot that names, the reporter and tag that let you watch and grab, the mutagenesis that lets you edit — serves the great payoff of this whole rung: making a useful protein on purpose, in quantity. That step is recombinant protein expression. Cloning a gene only puts an instruction on the shelf; expression is getting a host cell to actually *read* it and build the product, often by the bucketful.
You build it on an *expression vector* — a plasmid that, unlike a plain cloning vector, carries the signals a host needs to actually make protein: a strong promoter to drive heavy transcription, a ribosome-binding signal, and usually a switch so you can turn expression on at the chosen moment. You insert your gene — typically as intron-free cDNA, made by reverse transcriptase reading the mRNA, so a bacterium that cannot splice still gets a clean, ready-to-read coding sequence — transform the host, grow a big culture, then flip the switch (for example, add the inducer IPTG) and command the cells to pour out your protein following the same DNA -> RNA -> protein flow they use for their own genes.
This is the step that delivered the practical promise of recombinant DNA. Before it, human proteins for medicine had to be scraped in tiny amounts from animal tissue. Recombinant insulin (Humulin, 1982) was the first recombinant drug: the human insulin gene, expressed in bacteria, giving a limitless, consistent, *human* supply that replaced pig and cow insulin. One honest caveat decides the host you pick: E. coli is the cheap, fast workhorse, but it cannot perform many eukaryotic finishing touches — adding sugar chains (glycosylation) or forming certain folds — so human therapeutic proteins that need those touches are made in yeast or mammalian cells instead, chosen so the protein folds and is modified correctly.
Step back and the toolkit forms a single arc. You cut and pasted a gene into a vector; you saw the pieces on a gel; you named a specific molecule with a blot; you watched a gene fire with a reporter and grabbed its protein with a tag; you fine-tuned the protein with mutagenesis; and finally you commanded a cell to manufacture it. This is the moment molecular biology stops merely describing life and starts *building* with it — the engineering science the next rungs, with PCR and sequencing, will sharpen into a precise and astonishing craft.