BioID, BioID2, miniTurbo, TurboID, and Split-TurboID: A Practical Selection Guide

The Five Enzymes at a Glance

All five share the same conceptual starting point: E. coli biotin ligase BirA, mutated to release reactive bioAMP nonspecifically rather than transferring it to a single defined substrate. From that shared ancestor, each enzyme took a different evolutionary or engineering path.

References: Roux et al. (2012); Kim et al. (2016); Branon et al. (2018); Cho et al. (2020). For background on how the bioAMP labeling mechanism works at the molecular level, see our TurboID proximity labeling principles and workflow guide.

Labeling Speed: Where the Family Splits in Two

Labeling time comparison across the five biotin ligase proximity labeling enzymes (Branon et al., 2018; Cho et al., 2020).

Speed is the sharpest dividing line in this family. BioID and BioID2 require 16–24 hours of continuous biotin supplementation for robust labeling in mammalian culture. That window is simply too long for experiments where the biology moves faster than the assay.

TurboID changed the equation. In the original benchmarking study, TurboID produced in one hour more biotinylated material than BioID generated in 18 hours (Branon et al., 2018). A 10-minute pulse is sufficient for proteome-scale labeling in HEK293 cells—short enough to capture receptor-proximal protein recruitment within minutes of ligand addition, or to profile which proteins associate with a chromatin locus during a specific cell cycle window.

miniTurbo sits between the two groups: roughly 1.5–2 fold less active than TurboID, typically requiring 30–60 minutes for robust labeling. Its lower constitutive activity means the “off” state before biotin addition is cleaner, which makes it better suited for comparative before/after experimental designs where a dirty baseline would obscure the stimulus effect.

When speed is not the constraint—stable, long-lived protein complexes, overnight incubations, static proteome maps—BioID’s 18 hours of labeling is not a liability. Hundreds of published bait proteins validated in BioID is a real asset when reviewers start asking hard questions.

Tag Size: Small Enzymes for Sensitive Baits

For most bait proteins, a 35 kDa tag is acceptable. For small proteins, structured termini, or subunits within tightly packed complexes, it can displace interaction partners or mislocalize the fusion entirely.

BioID2 addressed this by starting from Aquifex aeolicus BirA—at 233 amino acids, substantially smaller than E. coli BirA’s 321 residues—and introducing a single R40G promiscuity mutation. The result is a ~26 kDa enzyme that also requires lower concentrations of exogenous biotin and shows improved subcellular targeting fidelity when fused to nuclear envelope proteins (Kim et al., 2016). Labeling kinetics, however, are unchanged: BioID2 is still an 18-hour experiment. Creative Proteomics offers a dedicated BioID2 interaction analysis service for projects where tag size is the primary constraint.

miniTurbo takes the same approach from within the TurboID lineage. Deleting TurboID’s N-terminal domain yields ~27 kDa while preserving most of the core catalytic mutations. The speed advantage survives—miniTurbo is still far faster than BioID2—though labeling efficiency drops approximately 50% relative to full-length TurboID.

Need a smaller tag and sub-hour labeling? miniTurbo is the only option in this family that delivers both.

Background: The Cost of High Activity

Higher catalytic output has a price. TurboID’s efficiency extends to endogenous biotin—the nanomolar concentrations present in standard DMEM are enough to drive low-level biotinylation before any exogenous supplement is added. This shows up as a “baseline” signal in the no-biotin control that is higher than what BioID produces, and several groups have noted concerns about persistent labeling and potential fusion protein instability in some contexts (May et al., 2020).

miniTurbo is more restrained. Without exogenous biotin, its constitutive activity is comparable to BioID—and that cleaner off-state is often the reason it wins out in comparative designs (treated vs. untreated, stimulus vs. no stimulus). TurboID’s background can be managed with proper controls; it cannot be eliminated.

BioID and BioID2 sit at the low-background end of the spectrum—the same slow kinetics that limit their temporal resolution also limit their constitutive activity. If stringent inducibility is the first requirement, either is easier to control for, at the cost of temporal precision.

For control strategies that work across all five enzymes, see our guide on distinguishing true interactors from background in TurboID experiments.

Temperature and Organism Compatibility

This criterion is often non-negotiable, and it eliminates two enzymes from consideration immediately for many researchers.

BioID was engineered in and for E. coli at 37°C. Below that, activity drops sharply: in yeast cytosol at 30°C, BioID labeling is undetectable (Branon et al., 2018). For Drosophila (25°C), C. elegans (20–25°C), or plant systems, neither BioID nor BioID2 should be used without independent activity validation—the streptavidin blot will show signal from endogenous carboxylases regardless, creating a false impression that the experiment is working.

TurboID and miniTurbo were evolved in the yeast secretory pathway at 30°C. Both have been validated in Drosophila melanogaster, C. elegans, Saccharomyces cerevisiae, multiple plant species (Arabidopsis, rice, N. benthamiana), and zebrafish. TurboID also functions in the mammalian ER lumen, where the oxidizing environment compromises BioID activity (Branon et al., 2018).

If your model organism grows below 37°C, TurboID or miniTurbo is the only viable choice from this enzyme family. Full stop.

Split-TurboID: A Specialized Tool for Contact Sites

Split-TurboID reconstitutes enzymatic activity only at organelle contact zones, restricting labeling to the ~10-30 nm contact interface.

Full-length proximity labeling enzymes have one spatial limitation: if the bait protein resides in multiple locations, labeling occurs in all of them. A protein at ER–mitochondria contacts is also present elsewhere on the ER membrane and the mitochondrial outer membrane—targeting full-length TurboID to it would produce a composite proteome, not a contact-site-specific one.

Split-TurboID solves this by fragmenting TurboID into two inactive pieces. N-TurboID on the ER membrane and C-TurboID on the outer mitochondrial membrane cannot function independently. Only at the narrow zone where both membranes are within contact distance do the fragments reconstitute an active enzyme. Using this design, Cho et al. (2020) identified over 100 proteins at ER–mitochondria contacts, validated eight by independent biochemical methods, and found many with no prior annotation to that compartment.

Split-TurboID is not a drop-in replacement for full-length TurboID. Two fusions to design and validate instead of one, system-dependent reconstitution efficiency, and additional troubleshooting complexity at every stage. Reserve it for questions where full-length approaches have already been shown to be insufficiently specific.

Full Comparison and Decision Framework

Enzyme selection decision tree: match your primary experimental constraint to the right proximity labeling enzyme.

• Use TurboID when you need maximum signal, temporal resolution (pulse <60 min), or broad organism compatibility including invertebrates and plants.
• Use miniTurbo when you need a smaller tag, cleaner inducibility, or when 30–60 min pulses are scientifically acceptable.
• Use BioID2 when tag size is the primary constraint and slow labeling (18 h) is acceptable for your experimental question.
• Use BioID when you want the deepest published validation record and background control is more important than speed.
• Use Split-TurboID only for organelle contact sites, interaction-dependent conditional labeling, or cell–cell interface proteomics where full-length approaches lack sufficient spatial specificity.

Once you have selected an enzyme, the next decisions are experimental design, controls, and replication strategy—covered in detail in our guides on TurboID experimental design for PPI studies and TurboID control design. Comparing proximity labeling against AP-MS or APEX2? See our guide on TurboID vs APEX2 vs AP-MS.

Four Mistakes That Are Easy to Make

Running TurboID without a no-biotin control. Standard culture media contains biotin. TurboID will label at low levels before you add any exogenous supplement, and streptavidin beads will pull down endogenous biotinylated carboxylases regardless. Without a no-biotin control processed identically to experimental samples, separating genuine proximity signal from these two sources is not possible.

Using BioID in Drosophila or C. elegans. The streptavidin blot will show signal—from endogenous carboxylases—and it looks convincing. Proximity labeling activity is essentially zero at 25°C with BioID. This organism compatibility limitation is well-documented and should not require an independent experiment to re-discover.

Treating miniTurbo and TurboID as equivalent at 10-minute pulses. They are not. At 10 minutes, TurboID produces substantially more biotinylated material. The two enzymes become more comparable at 30–60 minute pulses, and miniTurbo’s cleaner background often makes it the better choice at that timescale.

Starting a project with split-TurboID. Two fusions to validate instead of one, plus system-dependent reconstitution efficiency. The standard approach is to first confirm that full-length TurboID fused to each component localizes and labels correctly, then switch to split format. Starting with split-TurboID and seeing no signal leaves two independent failure modes to diagnose simultaneously.

How Creative Proteomics Can Help

Selecting the right enzyme is step one. Validating construct design, establishing controls, and generating a publication-quality dataset requires expertise across molecular biology, cell biology, and mass spectrometry—often in the same project.

Creative Proteomics offers dedicated services for the full proximity labeling toolkit, including our TurboID service and BioID proximity labeling service. Our scientists can advise on enzyme selection, construct design, and control strategy based on your specific bait protein, organism, and biological question.

Not sure where to start? Reach out through our online inquiry page and we will help map the right approach to your project goals.

Frequently Asked Questions

What is the main difference between TurboID and BioID?

TurboID carries 15 directed-evolution mutations that increase catalytic output 7–26 fold relative to BioID, reducing the required labeling time from 18–24 hours to approximately 10 minutes. TurboID is also active at 25°C, making it compatible with invertebrate and plant systems where BioID produces no detectable labeling. The trade-off is higher background from constitutive activity with endogenous biotin.

Why would I choose BioID2 over BioID?

BioID2 offers a smaller enzymatic tag (~26 kDa vs. ~35 kDa for BioID), derived from Aquifex aeolicus BirA rather than E. coli BirA. It also requires lower concentrations of exogenous biotin and shows improved subcellular targeting precision. Labeling kinetics remain slow (16–18 hours), so BioID2 does not offer the temporal resolution of TurboID or miniTurbo.

When should I use miniTurbo instead of TurboID?

miniTurbo is the better choice when you need a smaller tag (~27 kDa), when your experimental design requires clean inducibility—miniTurbo produces significantly less background without exogenous biotin compared to TurboID—or when 30–60 minute pulse times are compatible with your biological question. For maximum signal at 10-minute pulses, TurboID is preferred.

Can TurboID and BioID be used in the same organism?

Both work in mammalian cell culture at 37°C. Only TurboID and miniTurbo are validated in Drosophila, C. elegans, yeast, zebrafish, and plants at 20–30°C; BioID activity is undetectable in yeast cytosol at 30°C.

What is split-TurboID and when is it needed?

Split-TurboID divides TurboID into two catalytically inactive fragments that reconstitute an active enzyme only when brought into close proximity by a protein–protein interaction or organelle–organelle contact. It is specifically needed for mapping proteomes at organelle contact sites or cell–cell interaction interfaces where full-length enzyme fusions would label entire compartments non-specifically.

Does TurboID have higher background than BioID?

Yes. TurboID's high catalytic activity enables partial utilization of nanomolar endogenous biotin in standard culture media, producing low-level constitutive biotinylation without exogenous supplementation. miniTurbo has lower constitutive activity and a background profile closer to BioID, making it preferable for comparative before/after experimental designs.

Which enzyme identifies the most proteins by LC-MS/MS?

TurboID typically produces the largest candidate list per experiment. However, enrichment specificity matters more than candidate count: a well-controlled miniTurbo dataset at a 45-minute pulse will outperform a poorly controlled TurboID experiment in the proportion of candidates representing genuine proximity neighbors.

Can I switch from BioID to TurboID mid-project?

Technically yes, but the two enzymes differ enough in background profile, labeling radius, and kinetics that cross-enzyme comparisons are not straightforward. If switching is necessary, re-establish all experimental conditions—bait construct, expression level, controls, biotin pulse—from scratch in the new enzyme system rather than trying to normalize data from both.

Related Services

For Research Use Only. Not for use in diagnostic procedures.

Author: CAIMEI LI | Senior Scientist at Creative Proteomics