Split-TurboID enables contact-dependent proximity labeling in cells

Background:

In addition to classical methods such as cell fractionation and immunoprecipitation, proximity labelling (PL) has become an important tool to study the compartmentalisation of protein networks and macromolecular complexes in living cells. Currently, PL involves targeting promiscuous enzymes such as APEX^1,2, BioID³ or TurboID⁴, to a protein complex or an organelle of interest and initiating biotinylation of endogenous proteins localised within a few nanometers of the promiscuous enzyme. Biotinylated proteins may be harvested using streptavidin beads and further identified using mass spectrometric methods. Although current PL techniques have been valuable to map the proteomes of various organelles, they do not achieve the spatial specificity required to identify proteins at organelle contact sites. Variants of PL enzymes split and fused to two different organelles could offer the specificity required to only biotinylate proteins in the vicinity of organelle contact sites.

Key Findings:

Cho et al screened 14 different split sites on the previously described TurboID enzyme⁴, with each pair fused to FKBP and FRB proteins, which associate upon induction with rapamycin (Fig 1A). The screen found an optimal fragment pair (split at the 73/74 split) whose reconstitution, and further biotinylation activity, depended on the presence of rapamycin. Impressively, split-TurboID performed with significantly higher activity than previously described split PL enzymes (split-BioID and Contact-ID), as well as the original full-length BioID^5,6,7.

With this tool, Cho and colleagues targeted the two TurboID fragments (still fused to FKBP and FRB in order to enable rapamycin-induced dimerization) to a transmembrane domain ER membrane resident protein, Cb5, and Tom20, an outer mitochondrial membrane resident protein (Fig 1B); expression of these constructs in HEK 293T cells showed compartment-specific localisation. Incubation of these cells with biotin for 4 h led to biotinylation activity, which was further enhanced upon addition of rapamycin.

 

Finally, biotinylated proteins (from streptavidin-enriched samples) were labelled with isotopically distinct tandem mass tags (TMT) and the relative abundance of the proteins was quantified across the different samples using mass spectrometry. The samples comprised of split-TurboID with or without rapamycin addition, full-length TurboID control as well as single replicates of appropriate controls. The mass spectrometric data were filtered, and 67 proteins were identified in the dataset with rapamycin, while 63 proteins were identified without rapamycin. Interestingly, there were over 70 “orphan” proteins, which had not been assigned to ER-mitochondria contacts previously. Cho and colleagues further validated eight split-TurboID-identified ER-mitochondria contact orphan proteins, using overexpression and imaging, as well as biochemical fractionation. These proteins have been annotated to span a range of diverse functions, and their function at ER-mitochondria contact sites could have exciting biological implications.

Overall, Cho and colleagues have built upon previous split proximity labelling tools to attain better biotinylation activity, while maintaining spatial specificity.

What I like about this preprint:

As someone with an interest in metabolic compartmentalisation, I can foresee this tool being very useful in determining changes in the assembly of macromolecular complexes or organelle contacts depending on functional changes within the cell (for example, under nutrient stress conditions). This can be especially relevant when assessed under disease settings, such as metabolic disorders or cancer.

Aside from being a great tool, I appreciated the methodical nature of this preprint. Each step clearly built upon the previous and I also liked the comparisons of split-TurboID with previously described split PL enzymes, and clear evidence of why split-TurboID is better.

Questions for the authors:

1. The time scales for split-TurboID are significantly longer than full-length TurboID (4 h vs 1 min to attain similar biotinylation activity). Does this have to do with the time it takes to reconstitute split-TurboID or is this a reflection of decreased biotinylation activity due to the fact that only when the two membranes are in contact, can there be biotinylation?
2. As there was more biotinylation activity in split-TurboID upon induction with rapamycin, can there be false positives found by forcing the two membranes together, as opposed to letting them reconstitute freely?
3. Do you foresee any functional manipulations (such as addition of chemical inhibitors, etc) affecting reconstitution affinity of the split-TurboID fragments? Is there any way of predicting this a priori?

References:

1. S. S. Lam, et al., Directed evolution of APEX2 for electron microscopy and proximity labelling. Nat. Methods. 12, 51-54 (2014).
2. H. W. Rhee, et al., Proteomic mapping of mitochondria in living cells via spatially restricted enzymatic tagging. Science. 339, 1328-1331 (2013).
3. K. J. Roux, et al., A promiscuous biotic ligase fusion protein identifies proximal and interacting proteins in mammalian cells. J. Cell Biol. 196, 801-810 (2012).
4. T. C. Branon, et al., Efficient proximity labelling in living cells and organisms with TurboID. Nat. Biotechnol. 36, 880-898 (2018).
5. S. De Munter, et al., Split-BioID: a proximity biotinylation assay for dimerization-dependent protein interactions. FEBS Lett. 591, 415-424 (2017).
6. I. M. Schopp, et al., Split-BioID a conditional proteomics approach to monitor the composition of spatiotemporally defined protein complexes. Nat. Commun. 8 (2017).
7. C. Kwak, et al., Contact-ID, a new tool for profiling organelle contact site, reveals proteins of mitochondrial-associated membrane formation. PNAS. (2020) (in press).
8. K. F. Cho, et al., Split-TurboID enables contact-dependent proximity labelling in cells. BioRxiv. (2020).

Tags: cell biology, organelle contacts, proximity labelling, turboid

doi: https://doi.org/10.1242/prelights.18799

Read preprint (No Ratings Yet)