Yeast two-hybrid
Screen for protein–protein interactions inside yeast.
Our ruleSimulations prepare you for the bench but don't replace hands-on practice.
1A transcription factor cut in two
Eukaryotic activators like GAL4 have two separable parts: a domain that binds a specific DNA sequence and a domain that recruits the transcription machinery. Neither does anything alone: a DNA-binding domain sits on the promoter and nothing happens; an activation domain floats in the nucleus and never reaches a promoter.
23-AT is the stringency dial
HIS3 reporters leak. On plates lacking histidine, a reporter strain forms small colonies whether or not anything is interacting, so the assay needs a way to raise the bar. 3-amino-1,2,4-triazole is a competitive inhibitor of the HIS3 gene product (imidazoleglycerol-phosphate dehydratase): adding it means the cell must make more of the enzyme before it can grow, so more 3-AT demands more reporter output.
DNA-binding fusion
GAL4-BD or LexA fused to your protein, on a plasmid with one auxotrophic marker (commonly TRP1). Must be tested alone for autoactivation.
Activation-domain fusion
GAL4-AD or VP16 fused to a candidate, or to a whole cDNA library, on a plasmid with a second marker (commonly LEU2).
Growth as readout
GAL-promoter HIS3 plus 3-AT; often ADE2 as well, so a true positive is His⁺ Ade⁺ and the two can be scored together.
Specificity
Rescue the prey plasmid and retransform it with the bait and with an empty bait vector. Only a prey that needs the bait is a candidate.
3What makes it say yes when the answer is no
- An autoactivating bait: the fusion itself recruits transcription machinery, so the reporter is on with an empty prey vector. The commonest artefact, and the reason the bait-alone plate is the first thing to set up.
- An autoactivating or sticky prey: certain library clones (ribosomal proteins, heat-shock proteins, zinc-finger fragments) come out of almost every screen.
- Reporter reversion and plasmid loss: a his3 revertant grows regardless, and a cell that has lost the bait plasmid can still carry a self-activating prey.
- Uneven loading: a heavy patch of a negative control out-grows a thin spot of a real interaction, which is why strains are normalised to the same OD600 and spotted as a dilution series next to a non-selective loading plate.
4And what makes it say no when the answer is yes
A two-hybrid negative is weak evidence. The interaction has to happen in a yeast nucleus, between fusion proteins, at whatever level the plasmid's promoter gives: a protein that needs a post-translational modification yeast does not make, a membrane protein that never reaches the nucleus, a fusion that blocks the interaction surface, or a pair that only binds as part of a larger complex will all score negative.
- Autoactivation
- Reporter activation by the bait (or prey) fusion on its own, without any interaction partner. Measured with the opposite empty vector.
- 3-AT
- 3-amino-1,2,4-triazole, a competitive inhibitor of the HIS3 product, used at 1–50 mM to set how much reporter output a colony needs.
- Prey rescue
- Recovering the prey plasmid from a positive yeast colony (usually by transformation of E. coli with a yeast DNA prep) so it can be sequenced and retested.
- Library coverage
- The fraction of a library's independent clones present at least once among the transformants: 1 − e^(−n/N), so three-fold over the library gives about 95%.
Key terms
Practise these as flashcards →The protein fused to the DNA-binding domain, held at the reporter promoter.
The protein (or library of proteins) fused to the activation domain, which must be recruited to the promoter to be seen.
A yeast strain with its own GAL4 and GAL80 deleted, auxotrophic for histidine (and often adenine), carrying reporters behind GAL4 binding sites.
Strains normalised to the same density and spotted as a ten-fold dilution series on selective and non-selective plates, so growth can be compared row to row.
β-galactosidase activity normalised to cell density and reaction time, so numbers from different cultures can be compared.
The network of interactions in an organism, large parts of which were first mapped by two-hybrid screens at genome scale.
When to use it
- You have a protein and no idea what it binds, and want an unbiased, cheap screen over a whole library of candidates.
- You want to map which fragment or domain of two known interactors is responsible, by testing a panel of truncations against each other.
- You need a genetic selection rather than a biochemical measurement, for example to select interaction-disrupting point mutations, or to screen a mutant library for separation-of-function alleles.
Compared with alternatives
Prefer co-IP when you want evidence that the two proteins associate at native levels in the cell type of interest; it is the natural follow-up to any two-hybrid hit.
A pull-down with recombinant protein tests whether the interaction is direct, which a two-hybrid positive only suggests.
Use PLA or proximity labelling when the question is whether the proteins are close together in situ, including in a compartment yeast cannot model.
For affinity and kinetics you need purified protein and a biosensor: two-hybrid growth is not a Kd, and colony size is at best a rank order.
Display formats screen far larger libraries for binders, but select on the surface rather than in a nucleus, so they suit antibody and peptide discovery rather than mapping cellular complexes.
Sub-techniques
Choosing one here switches the whole page, including the protocol and the lab, to it.
Instead of one named prey, a whole cDNA or genomic library of activation-domain fusions is transformed into the bait strain, so a single selection asks which of a million clones can bring the activation domain to the promoter. Scale, coverage and a specificity retest are what separate a screen from a list of artefacts.
- Scale first: the transformation has to be plated as a dilution on doubly selective medium and counted, because coverage (1 − e^(−n/N)) is the only thing that makes a negative screen meaningful. Three-fold over the library is the usual target.
- Stringency has to be titrated on the bait alone before the screen, and then held: the lowest 3-AT that keeps the bait-alone plate blank.
- Most primary His⁺ colonies are not interactors. Picking the largest, best-separated colonies enriches for real ones, but the retest is what decides.
- The retest is a rescue-and-retransform: the prey plasmid is recovered and paired with the bait and with an empty bait vector. A second reporter does not test specificity, because an autoactivating prey turns both reporters on.
- Prey clones that come out of unrelated screens in the same lab (ribosomal, chaperone and zinc-finger fragments) should be treated as suspects whatever the retest says.