Biology Atlas

Map / Assays

assay · intermediate · partly written

Yeast two-hybrid (Y2H)

Yeast two-hybrid asks whether two proteins can touch: each is fused to half of a yeast transcription factor, and if they bind, the halves come together and switch on a reporter gene that lets the yeast grow.

Also called: Y2H, two-hybrid, yeast two-hybrid screen

What it is

Yeast two-hybrid tests whether two proteins can touch. Each is fused to half of a yeast transcription factor. If they bind, the halves reunite and switch on a reporter gene, so the yeast cell grows. It answers a yes or no about contact, not about strength, timing or the human cell. 12

Reported, not yet replicated

Why a negative is weak evidence more

The fusion is a peptide bond at one end of each protein. In HuRI, sensitivity fell where a known interface lay within 20 angstrom of the terminus fused to the activating half; the authors say this "might stem from" steric constraints, so it is a statistical association with known structures, not a per-pair demonstration. In a Varicella-Zoster screen the four N/C orientations found 182, 89, 149 and 144 interactions with only 15 common to all four. Vector choice alone changed bacterial results from about 140 to 47 interactions. So repeating a screen with the same construct re-asks the same blocked question. 234

Reported, not yet replicated

Scale and coverage deep

HuRI reports 52,569 interactions among 8,275 proteins from 17,408 genes screened in nine screens (three Y2H versions) and estimates it covers 2-11% of the binary interactome; the space of possible human pairs is over 150 million. Among HuRI pairs that fall in a complex with a known structure, 90% are direct contacts, against 81% for literature binary pairs and under 50% for complex maps. The earlier HI-II-14 map reported about 14,000 interactions among 4,303 proteins and described itself as depleted of interactions among proteins with predicted transmembrane helices. 25

Reported, not yet replicated

Question answered

Can protein A and protein B physically contact each other, in a yeast nucleus, when each is fused to one half of a transcription factor? A positive says contact is possible in yeast; it does not say the two proteins meet in a human cell. 12

Reported, not yet replicated

Readout physical

Growth (or colour) of a yeast strain on selective medium. Protein A is fused to the DNA-binding half of the yeast transcription factor GAL4 and protein B to its activating half. If A and B bind, the two halves are brought together and a reporter gene regulated by a GAL4 binding site is switched on. Reporters such as HIS3 let the cell grow without histidine; a drug (3-AT) that blocks the His3 enzyme is added to raise the bar. 13

Reported, not yet replicated

Not measured

Affinity, kinetics and stoichiometry: it gives a yes or no for one construct. It misses interfaces within about 20 angstrom of the terminus fused to the activating half, protein pairs that need a human modification or a third protein, proteins of which only one isoform per gene was tested, and (by the authors' account) membrane proteins. For HuRI the authors found no interactions at all for half of the tested proteins. A "no" is a statement about one construct, not about the pair. 25

Reported, not yet replicated

Setting

Where in cascade

Sample prep

Each open reading frame is cloned in frame with a GAL4 domain (as bait or as prey) and the two plasmids are brought into one yeast cell, typically by mating. Because a fusion can block an interface, careful screens repeat the test with the domain on the N- or C-terminus of each protein. 13

Reported, not yet replicated

Conditions

Everything happens in the yeast nucleus with overexpressed fusion proteins. Raising the 3-AT concentration removes noise and signal together: in one viral screen most interactions disappeared at 25 mM and a bait that activated the reporter by itself gave clear results only at 10 or 25 mM. Vector choice changes the answer: two vector sets applied to the same bacterial proteins gave about 140 and 47 interactions, with only 6 found by both (the abstract says 140, the text 138). 34

Reported, not yet replicated

Throughput

Genome scale is possible. HuRI screened 17,408 genes in nine screens with three Y2H versions and reported 52,569 interactions among 8,275 proteins. A permuted four-orientation screen of Varicella-Zoster virus proteins ran about 20,000 tests. The class "high" is our reading of these scales, not a stated rate. 23

Our inference

Time per run

A provider of a contract library screen states about 2 months for the service (bait cloning and sequencing, self-activation and toxicity test, library screen, validation, up to 5 positive clones). This is the provider's statement on its own page, fetched 2026-11-17. 6

Reported, not yet replicated

Engineerable variants

Permuted fusion orientation

Run all four N/C-terminus combinations of bait and prey. In the Varicella-Zoster screen the four combinations found 182, 89, 149 and 144 interactions; pairwise overlap between screens ran from 17% to 43%; 15 interactions (13 in one figure legend) appeared in all four; four screens found about twice as many interactions as one. The authors put typical Y2H false negatives at 70-90%. 3

Reported, not yet replicated

Several vector versions

HuRI used three Y2H versions because they detect complementary sets of interactions; three screens of each version doubled the number of detected interactions and proteins compared with nine screens of version 1 only. 2

Reported, not yet replicated

Titrating the reporter stringency

Raise 3-AT in steps (0, 1, 3, 10, 25, 50 mM in one study) to see which interactions survive; autoactivating baits are tested alone and dropped if they cannot be titrated out. 37

Reported, not yet replicated

Split-ubiquitin (MYTH) for membrane proteins

A variant that reassembles ubiquitin at a membrane instead of GAL4 in the nucleus is described as the workaround for membrane proteins. The page that describes it could not be opened. 5

Unverified: from our own report, source not opened

Use cases

HuRI, the Human Reference Interactome

An all-against-all binary map of human proteins. It reports 52,569 interactions among 8,275 proteins from a space of over 150 million possible pairs (17,408 x 17,409 / 2, our arithmetic) and puts its coverage at 2-11% of the binary interactome. The authors retested Y2H hits in two further assays (MAPPIT and GPCA) and report that they retest at a rate on par with high-confidence literature binary pairs, while single-evidence literature pairs retest significantly worse. The same programme made the map and the retest, so this is not independent confirmation. 2

Reported, not yet replicated

Yeast: two competing genome-wide screens

The comprehensive Ito screen found 4,549 two-hybrid interactions among 3,278 proteins and states that its data do not largely overlap with an earlier genome-wide screen (Uetz 2000). 8

Reported, not yet replicated

Before after

Status quo before

Before 1989 interactions were found by methods such as crosslinking, co-immunoprecipitation and chromatography, which start from a protein you already hold. This framing is recorded in our research notes as the context of the 1989 paper; the abstract was not re-read for that exact wording. 1

Unverified: from our own report, source not opened

What forced it

By 2001 the authors of the yeast screen wrote that a comprehensive description of interactions would contribute to interpreting "fully sequenced genomes, which are flooded with novel genes of unpredictable functions". 8

Reported, not yet replicated

The trick

GAL4 has two separable and functionally essential domains, a DNA-binding one and an activating one. Fuse one to protein X and the other to protein Y; if X and Y bind, they bring the domains back together and a reporter gene switches on. 1

Reported, not yet replicated

Why it worked

Proof of principle was the yeast proteins SNF1 and SNF4, which gave high reporter activity only when both hybrids were present. Because the readout is the growth of a yeast cell, it can be pooled and automated, which is what let groups run genome-wide screens such as the 2001 yeast map. 18

Reported, not yet replicated

Data availability

HuRI (Human Reference Interactome) download public

Interaction tables in TSV and PSI-MI formats, released under CC BY 4.0, including the HuRI map and related literature and union sets. 9

Caveats. The download page notes a moratorium on publishing binary screening data that its authors have not yet released beyond 10 interactions or one protein's interactions. No code is linked on the page. Licence of derived products was not checked against the exact files.

Reported, not yet replicated

Contract Y2H library screen (one provider) cro

One provider describes a screen as "request a quotation", about 2 months, including a self-activation and toxicity test and up to 5 positive clones. 6

Caveats. No price is published. Only one provider page was opened, so this says nothing about other providers, quality, reproducibility, IP terms or lock-in. A hit from one construct says nothing about orientation effects unless the provider tests several.

Reported, not yet replicated

Teach this

In one sentence

Yeast two-hybrid turns "do these two proteins touch?" into "does this yeast cell grow?" by splitting a transcription factor in two and gluing each half to one of your proteins. 1

Reported, not yet replicated

The five-minute version

Take GAL4, a yeast protein with a DNA-grabbing half and a gene-switching half. Neither half works alone. Fuse one half to your bait and the other to your prey and put both in yeast. If bait and prey touch, the halves meet on the DNA and a reporter gene turns on, so the cell grows on selective medium. The reporter can be made harder to trigger to remove noise, but that removes real signal too. Because each protein carries a fused tail, an interface near that end can be hidden, so careful screens repeat the test in several orientations and vector versions. 132

Reported, not yet replicated

What a listener will ask

If it is negative, does that mean they do not interact?

No. It means they did not trigger the reporter in that construct, in yeast. In one viral screen four orientations together found about twice as many interactions as one. 3

Reported, not yet replicated

Does a positive mean they bind in my cells?

It means contact is possible in a yeast nucleus with overexpressed fusions. HuRI authors retested their hits in other assays for that reason. 2

Reported, not yet replicated

Why is it used for whole-proteome maps?

The readout is yeast growth, so pairs can be tested in pools and in arrays; HuRI covered 17,408 genes in nine screens. 2

Reported, not yet replicated

The common wrong idea

A Y2H "no" means the proteins do not interact.

A Y2H "no" is a statement about one fusion construct in yeast; the interface can be hidden by the fusion.

The one citation to hand over: 2

Still open here

Not yet written: Cost, Capital, ML role.

Linked from

Sources

  1. Fields S, Song O. A novel genetic system to detect protein-protein interactions. Nature (1989). paper; opened 2026-11-17 all records citing this
  2. Luck, K., Kim, D.-K., Lambourne, L., et al.. A reference map of the human binary protein interactome. Nature (2020). paper; opened 2026-11-17 all records citing this
  3. Stellberger T, et al.. Improving the yeast two-hybrid system with permutated fusions proteins: the Varicella Zoster Virus interactome. Proteome Science (2010). paper; opened 2026-11-17 all records citing this
  4. Rajagopala SV, Hughes KT, Uetz P. Benchmarking yeast two-hybrid systems using the interactions of bacterial motility proteins. Proteomics (2009). paper; opened 2026-11-17 all records citing this
  5. Rolland, T., Tasan, M., Charloteaux, B., et al.. A proteome-scale map of the human interactome network. Cell (2014). paper; opened 2026-11-17 all records citing this
  6. Biost: yeast two-hybrid screening service. Biost. vendor-page, not peer reviewed; opened 2026-11-17 all records citing this
  7. Y2H screening guidelines. Yeast Resource Center. organisation-page, not peer reviewed; opened 2026-11-17 all records citing this
  8. Ito T, et al.. A comprehensive two-hybrid analysis to explore the yeast protein interactome. PNAS (2001). paper; opened 2026-11-17 all records citing this
  9. Human Reference Interactome download page. Interactome Atlas. database, not peer reviewed; opened 2026-11-17 all records citing this
Where it sits on the map
History of this record (3)

This record has been changed 3 times, on the dates below. The note kept against each change is written for the people who maintain the atlas, in their own shorthand, so it is not shown here; it is in the record itself.

Throughput class
high
Emerging
no
Status
partly written
Last updated
2026-11-17
Record
content/assays/yeast-two-hybrid.json