AP Biology · Unit 6 · Topic 6.2
Three models of replication predicted three different results, and one experiment separated them. Run all three and see which one matches.
Learning objective 6.2.A Describe the mechanisms by which genetic information is copied.
In 1958 nobody knew which of these three was right. Each makes a different prediction about where DNA would settle in a density gradient after each round of copying. Run all three at once and watch when they separate.
Set the slider to 1. The conservative tube shows two separate bands and the other two tubes each show one band in the middle. The real result was a single intermediate band, so the conservative model was finished — but two models were still standing, and no amount of staring at that first result would have separated them.
Set the slider to 2. Now the two surviving models come apart: the semiconservative tube has split into two bands while the dispersive tube still has one, moved a little higher. The real result was two bands. That is the whole experiment.
It is worth noticing what made this work. The experiment was not designed to prove the semiconservative model. It was designed so that the three models would give visibly different answers — and the second round was necessary because the first could not.
Run the semiconservative model out to 4 rounds and watch the intermediate band. It never disappears, and it never grows: it holds exactly two molecules for ever, because there were exactly two original strands and each one stays whole inside one daughter molecule.
DNA polymerase can only add nucleotides to a 3′ end, so a new strand can only grow 5′ to 3′. The two templates at a fork point in opposite directions. One of them can be read the way the fork is opening; the other cannot, and has to be copied backwards in pieces.
The two templates are chemically identical — they are both just DNA. Nothing about one of them is harder to copy than the other. The entire difference comes from the fact that they point in opposite directions while the enzyme can only work in one, so the fork opens toward one template's readable direction and away from the other's.
Drag the length slider and watch which readouts move. The leading strand stays at one piece however long the region is; every other number scales with the length. That contrast is the point of the tab.
This is why the count of primers is not the same as the count of fragments: the leading strand needs one to get started, and then never needs another, while the lagging strand needs a fresh one for every fragment.
Base pairing on its own is not accurate enough to copy a genome. What makes replication reliable is that three separate mechanisms each catch most of what the previous one missed — and their error rates multiply.
| Stage | Errors per base | Errors per human genome | What it does |
|---|
This is the deepest reason replication can be accurate, and it is the same fact as Topic 6.1's pairing rule seen from a different angle. If you know one strand you know the other exactly, so a damaged or wrongly inserted base can always be corrected by reading its partner. A single-stranded genome would have nothing to check itself against.
An error survives only if it gets past the pairing, and past proofreading, and past mismatch repair. Independent hurdles multiply, exactly as the gamete counts did in Topic 5.2 — so three checks that each let one in a hundred through give one in a million overall.
Replication is not perfect, and it is worth being careful about what to conclude from that. The errors that survive are mutations, and mutation is the only source of genuinely new alleles — the raw material every other process in Units 5 and 7 depends on.
So the accuracy is not a target the cell is failing to reach. A perfect copier would produce a population with no genetic variation at all, and Topic 6.7 looks at what a surviving error actually does.
Answer these in your own words before you open the answers.
1.
2.
3.
4.
5.
6.
7.
8.
| Model | After one round | After two rounds | Ruled out by |
|---|
A question that gives you gradient results and asks which model they support is asking you to read this table backwards. Count the bands, and count the rounds.