Proximity labelling
The original method: a promiscuous mutant of the E. coli biotin ligase (BirA*) fused to the bait, with biotin added for 15–24 hours. Slow, so the result is a time-average over a large part of a cell cycle, but the enzyme is benign, the substrate is non-toxic, and it works in almost any cell that can be grown for a day.
Our ruleSimulations prepare you for the bench but don't replace hands-on practice.
BioID
The original method: a promiscuous mutant of the E. coli biotin ligase (BirA*) fused to the bait, with biotin added for 15–24 hours. Slow, so the result is a time-average over a large part of a cell cycle, but the enzyme is benign, the substrate is non-toxic, and it works in almost any cell that can be grown for a day.
How it differs from the standard method
- Labelling windows of many hours, so it cannot resolve anything that changes in minutes.
- A long window integrates transient encounters, which widens the effective neighbourhood and saturates the nearest proteins.
- The culture grows through the window, so density and cell-cycle distribution change while you measure.
- BioID2, a smaller ligase from another bacterium, is less disruptive on a small bait and needs less biotin.
1Label first, lyse afterwards
The standing complaint about immunoprecipitation is that it only sees what survives lysis. A buffer gentle enough to keep a transient or membrane-embedded interaction is also gentle enough to keep the background, and a buffer stringent enough to clean the lane takes the interaction apart.
2The output is a neighbourhood
What the enzyme tags is everything that was close, for long enough, while the substrate was present. Labelling falls off with distance, roughly exponentially over a radius of ten or twenty nanometres, and has nothing to do with affinity.
Build and validate the line
Fusion at a known expression level, imaged against the untagged protein, shown to label on a streptavidin blot.
Label
Add the substrate; the window is set by the enzyme and decides what is averaged over.
Stop
Cold washes, or a radical quencher for APEX2, so the labelling ends when you meant it to.
Capture
SDS lysis, free biotin removed, streptavidin beads, washes with urea and salt, on-bead digestion.
Identify
Mass spectrometry of fusion and control samples in replicate; enrichment is the fold change over the control.
3Radius and time
The radius is a property of the reactive species. The biotin ligases make biotinyl-5′-AMP, which leaks from the active site and reacts with lysines nearby; APEX2 makes phenoxyl radicals from biotin-phenol and hydrogen peroxide, which react with tyrosines and die within a millisecond.
4The controls define enrichment
No protein is enriched on its own; it is enriched over something. Two controls do two different jobs.
- Labelling radius
- The distance over which labelling falls to about a third of its value at the enzyme: roughly 10 nm for the biotin ligases and 20 nm for APEX2. It is a fall-off, not a hard edge.
- Enzyme-alone control
- The same enzyme, expressed at a similar level and targeted to the same place as the fusion, with no bait attached. The denominator for enrichment.
- Endogenous biotinylation
- The cell's own biotin-dependent carboxylases (pyruvate carboxylase, propionyl- and methylcrotonyl-CoA carboxylase, acetyl-CoA carboxylase), which carry biotin permanently and are captured from any lysate.
- Spectral count
- The number of mass spectra assigned to a protein. It rises with abundance times the labelled fraction, so it measures how much was captured, not how tightly it was bound.
- Neighbourhood
- The set of proteins within a few labelling radii of the bait, averaged over the window. Contains binders, complex members that never touch the bait, and passers-by.
5Where it goes wrong
The enzyme is 27 to 35 kDa, which is enough to change where a small protein goes, and an overexpressed bait spills out of its normal location. Proximity labelling then faithfully reports the neighbourhood of the wrong place, and no control can subtract that, because the control has no bait to mislocalise.
Key terms
Practise these as flashcards →The R118G mutant of the E. coli biotin ligase, which releases its reactive intermediate instead of handing it to its one natural substrate. The enzyme behind BioID.
The activated biotin that the ligases make. It diffuses a short distance from the enzyme and reacts with nearby lysine amines.
The APEX2 substrate. With hydrogen peroxide, the peroxidase turns it into a phenoxyl radical that reacts with electron-rich side chains, mostly tyrosine.
Pulling the biotinylated proteins out of an SDS lysate. The bond survives denaturants, so the washes can remove essentially everything that is not covalently labelled.
The ratio of a protein's signal in the fusion sample to its signal in the enzyme-alone control. The only number that defines a neighbour.
Stopping the reaction at the end of the window: cold washes for the ligases, ascorbate, Trolox and azide for APEX2.
An enzyme divided into two inactive halves on two baits, which only labels where the two baits meet: a way of mapping the neighbourhood of a specific complex rather than of one protein.
When to use it
- Mapping the neighbours of a protein whose interactions do not survive lysis: membrane proteins, components of insoluble structures, transient enzyme–substrate encounters.
- Mapping the protein content of a compartment that cannot be purified cleanly, by targeting the enzyme to it and comparing with a matched control elsewhere.
- Following how a neighbourhood changes after a stimulus, with TurboID or APEX2, where the labelling window can be shorter than the response.
- Generating candidate interactors to test by more direct methods, when a co-IP keeps coming out empty or dominated by background.
Compared with alternatives
Captures the complexes that survive lysis, with no fusion protein and no enzyme; prefer it for endogenous, stable interactions and when a validated antibody exists.
Reports proximity of two named proteins in situ, cell by cell and with location; prefer it when the candidates are known and the question is where and when they meet.
Tests whether two purified proteins bind directly, which proximity labelling cannot show; the natural follow-up for a candidate neighbour.
The identification step downstream of the capture; the quality of the neighbourhood depends on its quantification and replicate design.
Used before the experiment to show the fusion is where the untagged protein is, and after it to confirm a candidate neighbour co-localises.
Sub-techniques
Choosing one here switches the whole page, including the protocol and the lab, to it.
The original method: a promiscuous mutant of the E. coli biotin ligase (BirA*) fused to the bait, with biotin added for 15–24 hours. Slow, so the result is a time-average over a large part of a cell cycle, but the enzyme is benign, the substrate is non-toxic, and it works in almost any cell that can be grown for a day.
- Labelling windows of many hours, so it cannot resolve anything that changes in minutes.
- A long window integrates transient encounters, which widens the effective neighbourhood and saturates the nearest proteins.
- The culture grows through the window, so density and cell-cycle distribution change while you measure.
- BioID2, a smaller ligase from another bacterium, is less disruptive on a small bait and needs less biotin.