A Toolbox for Efficient Proximity-Dependent Biotinylation in Zebrafish Embryos

Abstract

Understanding how proteins are organized in compartments is essential to elucidating their function. While proximity-dependent approaches such as BioID have enabled a massive increase in information about organelles, protein complexes, and other structures in cell culture, to date there have been only a few studies on living vertebrates. Here, we adapted proximity labeling for protein discovery in vivo in the vertebrate model organism, zebrafish. Using lamin A (LMNA) as bait and green fluorescent protein (GFP) as a negative control, we developed, optimized, and benchmarked in vivo TurboID and miniTurbo labeling in early zebrafish embryos. We developed both an mRNA injection protocol and a transgenic system in which transgene expression is controlled by a heat shock promoter. In both cases, biotin is provided directly in the egg water, and we demonstrate that 12 h of labeling are sufficient for biotinylation of prey proteins, which should permit time-resolved analysis of development. After statistical scoring, we found that the proximal partners of LMNA detected in each system were enriched for nuclear envelope and nuclear membrane proteins and included many orthologs of human proteins identified as proximity partners of lamin A in mammalian cell culture. The tools and protocols developed here will allow zebrafish researchers to complement genetic tools with powerful proteomics approaches.

Keywords: BioID; Proximity-dependent biotinylation; TurboID; miniTurbo; zebrafish embryo.

Graphical abstract

Fig. 1

mRNA injection-based expression identifies LMNA proximal proteins in the early zebrafish embryo.A, in vitro transcribed mRNA coding for TurboID fused to GFP was injected into the single cell stage zebrafish embryo and GFP fluorescence was visualized at 6, 12, and 48 hpf (magnification 16×). B, TurboID-LMNA or miniTurbo-LMNA was expressed by mRNA injections (with the corresponding GFP fusions used as negative controls). Injected embryos were collected at 12 and 48 hpf. The number of proteins detected at BFDR ≤1% under each of the conditions is shown. C, list of the 18 high-confidence proximal interactors shared by all four conditions in (B). D, the 25 most abundant proteins (by spectral counts) with a BFDR ≤1% from the 12 h TurboID-LMNA condition are listed, alongside their abundance across each of the conditions.

Fig. 2

Inducible transgenic expression of TurboID-LMNA or miniTurbo-LMNA identifies LMNA proximal proteins.A, cassettes compatible with the Tol2 trangenesis system were generated that express the selected fusion proteins (TurboID-LMNA and GFP, miniTurbo-LMNA and GFP) downstream of the heat shock inducible Hsp70 promoter. The cassettes also contain GFP under the control of a cardiomyocyte specific Cmlc2 promoter to allow selection of integration-positive embryos. B, TurboID-GFP and miniTurbo-GFP transgenic embryos were heat shocked at 60 hpf for 1 h and imaged at 72 hpf. Integration-positive embryos were selected prior to heat shock based on expression of GFP in the heart (yellow arrow). GFP expression 12 h following heat shock is indicated by white arrowheads, and GFP expression in the eye by red arrows (magnification 16×, Scale Bar = 1 mm). C, immunofluorescence microscopy on embryos collected at 72 hpf following 1 h heat shock at 60 hpf time point was performed as detailed in Experimental Procedures. FLAG staining to detect the bait expression and streptavidin-fluorophore (Strep) to detect biotinylation. DAPI was used to stain the nucleus. (Upper panels Scale bar 50 μm, magnification 20×; Lower panels scale bar 5 μm, magnification 40×). D, the scatterplot shows all proteins detected with a SAINTexpress Bayesian FDR ≤1% with all least one of the baits (miniTurbo on the y-axis, TurboID on the x-axis); the dashed lines represent the different fold changes listed at the top. Preys are color-coded based on uniquely scoring as high-confidence interactors with either miniTurbo (green) or TurboID (blue) while preys scoring as high-confidence with both baits are coded in red. E, list of the 27 common partners. F, the 25 most abundant proteins (by spectral counts) with a BFDR ≤1% from the TurboID-LMNA experiment are listed, alongside their abundance across both conditions.

Fig. 3

Functional enrichment of LMNA proximal partners.A, schematic showing all the high-confidence interactors annotated with the term nuclear envelope or nuclear membrane. The color coding indicates in how many of the experiments this interactor was identified. B, overlap between all high-confidence proteins identified using both TurboID and miniTurbo by mRNA injection or transgenic expression (left), their mapped human orthologs (middle), and the human orthologs annotated as “nuclear envelope” by GO CC (right).

Fig. 4

Parameters influencing the recovery of proximal partners in transgenic experiments.A, the total spectral counts from both replicates for the 54 proteins identified with a BFDR ≤1% in the 800 μM preload condition were summed for each condition and the total compared. B, the 25 most abundant proteins (by spectral counts) with a BFDR ≤1% from the earlier TurboID-LMNA experiment (Fig. 2F) are listed, alongside their abundance across conditions. C, schematic of the timeline for biotin supplementation, induction of expression, and collection of embryos.