The engineered ascorbate peroxidase (APEX2) has been effectively employed in mammalian cells to identify protein–protein interactions. APEX2 fused to a protein of interest covalently tags nearby proteins with biotin-phenol (BP) when H2O2 is added to the cell culture medium. Subsequent affinity purification of biotinylated proteins allows for identification by MS. BP labelling occurs in 1 min, providing temporal control of labelling. The APEX2 tool enables proteomic mapping of subcellular compartments as well as identification of dynamic protein complexes, and has emerged as a new methodology for proteomic analysis. Despite these advantages, a related APEX2 approach has not been developed for yeast. Here we report methods to enable APEX2-mediated biotin labelling in yeast. Our work demonstrated that high osmolarity and disruption of cell wall integrity permits live-cell biotin labelling in Schizosaccharomyces pombe and Saccharomyces cerevisiae respectively. Under these conditions, APEX2 permitted targeted and proximity-dependent labelling of proteins. The methods described herein set the stage for large-scale proteomic studies in yeast. With modifications, the method is also expected to be effective in other organisms with cell walls, such as bacteria and plants.
Proximity-dependent labelling techniques that detect protein–protein interactions have been developed in recent years. For example, the proximity-dependent biotin identification (BioID) method has been successfully used in biochemical screens for interaction partners in a native environment [1–3]. This approach is based on fusion of bacterial biotin ligase mutant BirA to a targeting protein. After an overnight incubation with biotin, proteins biotinylated by BirA can be recovered by affinity purification using Streptavidin beads and identified by MS. This simple technique has been used to identify proteins that interact weakly or transiently with the protein of interest. However, a long half-life (>minutes) of the BirA product, a biotin-adenylate ester, could be a drawback since it leads to a large labelling radius, making it hard to distinguish true and false-positives. Moreover, the slow reaction of BirA (requiring 18–24 h) makes it difficult to investigate highly dynamic cellular events or acute cellular responses upon environmental changes.
To improve the spatial and temporal specificity, a method that uses an engineered variant of ascorbate peroxidase (APEX2) was developed [4–6]. In the presence of H2O2, APEX2 catalyses the conversion of biotin-phenol (BP) to biotin-phenoxyl radicals that covalently link biotin to electron-rich amino acid side chains within a short radius. As in BioID-based analysis, the biotinylated proteins are purified using Streptavidin beads and then analysed by MS. Biotin labelling in this case is achieved after only 1 min as opposed to a longer incubation in the BioID approach. The biotin-phenoxyl radical is short-lived (<1 ms), generating a small labelling radius (<20 nm). Given its small labelling radius and short labelling time, APEX2-based proximity labelling has the potential to provide higher spatial and temporal specificity. Indeed, the APEX2 tool has emerged as a powerful system to provide proteomic maps of proteins present in challenging compartments such as mitochondrial intermembrane space, ER–PM (plasma membrane) junctions, primary cilia and the synaptic cleft [5,7–10]. Besides spatial information, the APEX2 technique is also expected to offer temporal proteomic maps as well, uncovering signalling dynamics in living biological systems. A combined approach of APEX2 with stable isotope labelling by amino acids in cell culture-based quantitative proteomics allowed for higher spatial and temporal resolution .
Yeast is a powerful model organism for studying protein function and cellular processes . With short generation times and well-characterized genomes, a major advantage of yeast systems is the ability to perform high-throughput genetic screens to identify genes and pathways. Given the ease of the genetic manipulation of yeast, application of APEX2 to yeast proteomics studies would provide another powerful tool for addressing complex biological questions. Although well established in the mammalian system, APEX2 has not been adapted for yeast to date.
In this paper, we describe methods to permit APEX2-based live-cell biotin labelling in yeast. We observed that applying a high osmotic solution such as 1.2 M sorbitol to fission yeast Schizosaccharomyces pombe dramatically increased APEX2-dependent live-cell biotinylation. We used the APEX2 system to confirm known protein–protein interactions. In cells expressing a Dsc5-APEX2 fusion protein, we observed specific biotinylation of its known direct binding partner Cdc48 . A similar labelling protocol using Zymolyase to remove the cell wall allowed APEX2-mediated biotinylation in the budding yeast Saccharomyces cerevisiae, demonstrating that the APEX2 approach can also be extended to this organism. Our results open the door to application of the APEX2 system to proteomic studies in the two most common yeast model organisms.
We obtained Edinburgh minimal medium (EMM) and complete supplement mixture (CSM) from MP Biomedicals; yeast extract, peptone, agar from Becton Dickson; oligonucleotides from Integrated DNA Technologies; complete EDTA-free protease inhibitor from Roche; amino acids, D-sorbitol, D-galactose, yeast nitrogen base without amino acids, sodium ascorbate, (S)-6-hydroxy-2,5,7,8-tetramethylchromane-2-carboxylic acid (Trolox) from Sigma–Aldrich; Prestained Protein Standards from Bio-Rad Laboratories; BP from Iris Biotech; Streptavidin magnetic beads from Promega; Gibson Assembly Master Mix from NEB BioLabs; Zymolyase from Seikagaku distributed by MP Biomedicals.
Strains and media
Wild-type haploid S. pombe (h-leu1-32 ura4-D18 ade6-M210 his3-D1) and derived strains were grown to exponential phase at 30°C in EMM (32.2 g/l EMM plus 20 g/l glucose supplemented with 225 mg/l each of adenine, uracil, leucine, histidine and lysine) . S. pombe dsc5∆ strain (h-leu1-32 ura4-D18 ade6-M210 his3-D1 Δdsc5-D1::kanMX6) was previously described . Wild-type S. cerevisiae cells (MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0) were grown in synthetic complete (SC) medium (20 g/l glucose, 6.7 g/l yeast nitrogen base without amino acids, 0.89 g/l CSM) with 2% sucrose at 30°C . Galactose (2% w/v) was used to induce APEX2 protein expression. Standard yeast transformations were performed as previously described [14,15].
We obtained anti-Flag M2 from Sigma (F1804). Antiserum to Cdc48 was a kind gift of R. Hartmann-Petersen (University of Copenhagen). IRDye800CW goat anti-mouse and rabbit IgG, IRDye680RD goat anti-mouse and rabbit IgG, and IRDye800CW Streptavidin were from LI-COR.
All oligonucleotides used in PCR analyses are listed in Supplementary Table S1. A plasmid containing Flag-APEX2 under control of S. pombe adh1+ promoter  was generated via Gibson Assembly protocol by assembling two PCR-amplified fragments of Flag-APEX2 from pcDNA3 APEX2-NES (Plasmid #49386, Addgene) (by primers oJH39/40) and adh1+ promoter-driven vector pJB114 (by primers oJH41/42) following manufacturer's protocol (NEB). The dsc5-Flag-APEX2 and dsc5-ΔUBX-Flag-APEX2 were generated via Gibson Assembly by assembling two PCR-amplified fragments of Flag-APEX2 from pcDNA3 APEX2-NES (using oJH116/117 and oJH113/117 respectively) and adh1+ promoter-driven dsc5 vector pREB30 (oJH118/119 and oJH118/114 respectively). A plasmid containing Flag-APEX2 under control of S. cerevisiae GAL1 promoter was generated via Gibson Assembly by assembling two PCR-amplified fragments of GAL1-driven vector pRS425 (a gift from Susan Michaelis Lab, Johns Hopkins University) (by oJH124/125) and Flag-APEX2 (oJH126/127) from pcDNA3 APEX2-NES.
Equal amounts of protein (30–50 μg) from each sample were loaded on an SDS/PAGE gel. Gels were transferred to nitrocellulose using Trans-Blot Turbo Transfer system (Bio-Rad Laboratories), blocked using a solution of 5% milk-PBST (PBS with 0.05% Tween 20), and probed with primary antibody followed by mouse or rabbit IgG (IRDye800CW or IRDye680RD) or IRDye800CW Streptavidin. Blots were scanned through channel 700 or 800 using an Odyssey CLx imager from LI-COR and quantified using LI-COR Image Studio 4.0 Software.
The original BP labelling protocol for mammalian cells was modified for yeast [4,9]. Exponentially growing cells (5–10×108) were harvested and washed with water. Cell pellets were resuspended in 1 ml of 1.2 M sorbitol dissolved in H2O. BP (2.5 mM) was added and cells were incubated for 1 h at room temperature. H2O2 (1 mM) was added for 1 min to initiate BP labelling and cells were quickly spun down. To stop the reaction, the solution was aspirated off and cell pellets were then washed four–five times with a quenching solution consisting of 5 mM Trolox, 10 mM sodium azide and 10 mM sodium ascorbate in 1.2 M sorbitol dissolved in H2O. Cell pellets were washed with 1 ml of 1.2 M sorbitol once more, and lysed by addition of 27 mM NaOH, 1% (v/v) 2-mercaptoethanol for 10 min on ice. Total protein was precipitated with trichloroacetic acid (20% w/v) by incubating on ice for 10 min and centrifuging at 20000 g for 10 min at 4°C. The protein pellet was washed with 0.5 ml ice-cold acetone once. The protein pellet was then solubilized by sonication at 30% of max power for 10 s (Fisher Scientific Model 500 Sonic Dismembrator) in 200 μl lysis buffer (25 mM Tris pH 7.4, 150 mM NaCl, 2% SDS) containing Complete Protease Inhibitor EDTA free. After denaturation by heating at 75°C for 15 min, protein samples were either directly analysed by Western blotting or subjected to enrichment of biotinylated proteins. For enrichment, protein samples (1–2 mg in 0.5–1 ml) were first dialysed in dialysis buffer (25 mM Tris pH 7.4, 150 mM NaCl, 0.2% SDS) using 3.5 K MWCO dialysis tubing (Thermoscientific) for 2–3 h at room temperature, and then incubated with 50–100 μl Streptavidin-coated magnetic beads for 1–2 h at room temperature. Dialysis was performed to remove intracellular unreacted free BP, which may compete with biotinylated proteins for binding to Streptavidin beads. Streptavidin beads were then washed 6× with Wash buffer (25 mM Tris pH 7.4, 150 mM NaCl, 0.3% SDS) followed by 1× wash with 2 M urea/50 mM Tris pH 7.4, 1× wash with 1 M KCl and 1× wash with Wash buffer. Biotinylated proteins were eluted by incubating the beads with 30 μl 2× SDS sample loading buffer [4% (w/v) SDS, 0.2% (w/v) Bromophenol Blue, 20% (v/v) glycerol, 100 mM Tris pH 6.8, 10% (v/v) 2-mercaptoethanol] and heating to 75°C for 15 min. For Zymolyase incubation, cell pellets from exponentially growing cells (5–10×108) were resuspended in 500 μl of Zymolyase-100T solution (1000 units/ml of Zymolyase in 1.2 M sorbitol dissolved in H2O). After 10 min or at the same time as Zymolyase treatment, BP (2.5 mM) was added and the cells were further incubated for 20–60 min at room temperature rotating before addition of H2O2.
RESULTS AND DISCUSSION
Optimization of APEX2-mediated biotin labelling in fission yeast
To apply the APEX2 system to yeast, we first asked whether APEX2 is expressed and active in fission yeast. To express APEX2 in S. pombe, a plasmid carrying Flag-APEX2 gene under adh1+ promoter was transformed into leucine-auxotrophic wild-type strain. Four transformants were tested for Flag-APEX2 expression by Western blotting. All isolates tested expressed Flag-APEX2 with a predicted molecular weight of ∼28 kDa (Figure 1A). A single isolate was used for subsequent analysis. To examine whether APEX2 is active and mediates biotinylation in yeast, we initiated labelling by adding 2.5 mM BP to the culture medium of Flag-APEX2 expressing cells. After 1 h incubation at room temperature, H2O2 was added for 1 min and then the labelling was terminated. Streptavidin blot of whole cell lysates revealed endogenous biotinylated proteins in untreated cells, and H2O2 treatment for 1 min had no effect despite using 5-fold more BP than used in mammalian cells (Figure 1B, lanes 1–2). Because the BP probe is thought not to have high membrane permeability, we suspected that BP might not be entering the cell. To improve uptake, we disrupted cell wall structure using Zymolyase treatment. Flag-APEX2 expressing cells were pre-incubated with Zymolyase-100T for 10 min at room temperature, and then 2.5 mM BP was added for 1 h. Streptavidin blot of cell lysates showed specific H2O2-dependent biotinylation of proteins (Figure 1B, lanes 3–4). Thus, the BP probe is delivered to cells when the yeast cell wall architecture is compromised. In this experiment, APEX2 biotinylation is presumably non-specific, occurring on cytosolic proteins in proximity to soluble APEX2.
APEX2-based biotin labelling of yeast proteome in S. pombe
Because Zymolyase treatment was performed in 1.2 M sorbitol, we next tested whether BP delivery required Zymolyase, sorbitol or both. The APEX2 labelling reaction was performed using control or Flag-APEX2 cells. Streptavidin blot analysis of cell lysates showed that numerous endogenous proteins were biotinylated when cells were pre-treated with Zymolyase in 1.2 M sorbitol in a reaction that required APEX2, BP and H2O2 (Figure 1C). Surprisingly, we observed identical results when cells were pre-incubated with 1.2 M sorbitol alone (Figure 1C, lanes 7–9), indicating that the high osmotic sorbitol solution may permit BP labelling. Indeed, significant rearrangement in cell wall architecture occurs upon osmotic shock . Also, it has been previously reported that transient pre-incubation of intact fission yeast cells with a high osmotic solution (e.g. 2 M sorbitol) before electrical application dramatically improved DNA transformation efficiency of S. pombe . Thus, remodelled cell wall structure in response to sudden change in osmolarity might improve BP uptake in a similar way as for nucleic acid uptake. In support of this hypothesis, we observed little to no biotinylation above the background from endogenous biotinylated proteins when cells were incubated with the lower concentrations of sorbitol solution (0–600 mM) (Figure 1D, lanes 1–5). In contrast, when cells were incubated in a high osmolar solution of 1 M KCl we observed labelling equal to 1.2 M sorbitol (Figure 1D, lanes 5–6), suggesting that an increase in osmolarity promotes BP uptake. Alternatively but not exclusively, high osmolarity may prevent BP export out of the cell. Sorbitol incubation and Zymolyase treatment are routinely used for yeast experiments and have been demonstrated not to affect yeast cell viability [18–20]. Thus, this condition for BP delivery is expected to be non-toxic.
To our surprise, incubation of cells in EMM containing 1.2 M sorbitol did not result in APEX2-dependent protein biotinylation (Figure 1E), indicating that the EMM culture medium may contain an inhibitory factor(s). To ask whether glucose in the EMM interferes with uptake of BP, we examined the effect of 1.2 M sorbitol solution on biotin labelling in the absence or presence of glucose. Streptavidin blot analysis of cell lysates showed that the presence of 2% glucose (w/v) in 1.2 M sorbitol solution completely inhibited biotin labelling compared with 1.2 M sorbitol alone (Figure 1F, lanes 2 and 4). Also, we observed no APEX2-specific biotinylated proteins when cells were incubated in 2% glucose solution. Thus, glucose blocked BP accumulation in the cell. Development of structural variants of the BP probe that bypass glucose inhibition may improve BP delivery into fission yeast.
We performed additional experiments to optimize BP labelling, varying BP concentration (Figure 1G) and H2O2 labelling time (Figure 1H). Streptavidin blot analysis of lysates from cells incubated with different concentrations of BP showed that protein biotinylation was maximal at 2.5 mM BP (Figure 1G). A requirement for a higher concentration of BP in fission yeast compared with mammalian cells (2.5 mM compared with 0.5 mM) may be due to the different physical barriers . We observed that 1 min of H2O2 labelling was optimal, and that longer H2O2 treatment did not enhance the protein biotinylation level (Figure 1H). From these experiments, we established a labelling protocol for APEX2 in fission yeast.
APEX2 proximity-dependent biotin labelling detects specific protein–protein interactions in fission yeast
APEX2-catalysed biotin-phenoxyl radicals are short-lived, thus acting within a small radius from APEX2 . This property permits the detection of specific and direct protein–protein interactions . To examine whether APEX2 allows detection of specific protein interactions in fission yeast, we fused APEX2 to C-terminus of the Golgi protein Dsc5 (Dsc5-Flag-APEX2). Dsc5 is a subunit of the Golgi Dsc E3 ligase complex and contains a UBX domain, which interacts with the AAA-ATPase Cdc48 . Our previous in vitro and in vivo binding assays demonstrated that the Dsc5 UBX domain directly binds Cdc48, and that Dsc5 lacking the UBX domain (Dsc5ΔUBX) failed to recruit Cdc48 to the Dsc E3 ligase . Based on these data, we predicted that Dsc5-APEX2 would label Cdc48 with biotin, but that Dsc5ΔUBX-APEX2 would not label Cdc48 (Figure 2A).
Specificity of APEX2-mediated proximity-dependent biotin labelling in S. pombe
To test the utility of the APEX2 system, yeast cells expressing Dsc5-Flag-APEX2, Dsc5ΔUBX-Flag-APEX2 or Dsc5 were generated, and expression of APEX2 fusion proteins was verified by Western blotting (Figure 2B). We further developed a protocol to purify biotinylated proteins from cells (Figure 2C). Briefly, following a 1 h pre-incubation of cells with BP in 1.2 M sorbitol labelling was initiated by the addition of H2O2 for 1 min. After quenching with anti-oxidant reagents, cells were lysed in a denaturing buffer. To remove unreacted free BP, we dialysed the cell lysate, and biotinylated proteins were then enriched using Streptavidin beads. After stringent washing with different buffers, biotinylated proteins were eluted from beads by heating samples in SDS at 75°C. Whole cell lysates and bead-bound fractions were then analysed by Western blot to detect proteins of interest. Streptavidin and anti-Flag blot analysis showed that both Dsc5-APEX2 and Dsc5ΔUBX-APEX2 induced biotinylation of multiple endogenous proteins, but that Dsc5 alone without APEX2 fusion did not (Figure 2D). Importantly, Cdc48 was specifically recovered from cells expressing Dsc5-APEX2, but not Dsc5ΔUBX-APEX2. Quantification of Western blot band intensities showed that ∼2% of total Cdc48 (average from two biological replicates) was recovered from cells expressing Dsc5-APEX2. In contrast, only ∼0.1% and ∼0.01% of Cdc48 was recovered from Dsc5ΔUBX-APEX2 and control cells respectively, demonstrating that APEX2 detects specific protein–protein interactions in fission yeast.
Cell wall removal enables APEX2-mediated biotin labelling in budding yeast
Next, we sought to develop a similar approach for the budding yeast S. cerevisiae. To express APEX2 in S. cerevisiae, we generated cells carrying a plasmid expressing Flag-APEX2 gene under the control of the galactose-inducible GAL1 promoter. Yeast cells carrying an empty vector (EV) under the same promoter served as a negative control. Western blotting of whole cell lysates revealed that expression of Flag-APEX2 protein was induced in a galactose-dependent manner (Figure 3A). We then subjected cells to the same sorbitol labelling procedure optimized for S. pombe. To our surprise, this same method did not result in protein biotinylation above the background from endogenous biotinylated proteins (Figure 3B). The general composition of the cell wall in S. cerevisiae and S. pombe yeast cells is similar, but substantial differences exist [21,22]. Thus, osmotic shift may have different effects on cell permeability in the two organisms.
APEX2-based biotin labelling of yeast proteome in S. cerevisiae
Next, we tested whether cell wall removal using Zymolyase would enable labelling. After 10 min pre-incubation of cells with Zymolyase in 1.2 M sorbitol solution, we added the BP probe, incubated cells further for an additional 1 h and treated cells with H2O2 for 1 min. Under these conditions, we observed APEX2-dependent protein biotinylation (Figure 3C), demonstrating that cell wall removal permitted BP labelling in S. cerevisiae. BP uptake after Zymolyase treatment occurs rapidly, because 20 min BP incubation following 10 min pre-treatment with Zymolyase allowed protein biotinylation to a similar level that was obtained from 60 min incubation (Figure 3D, lanes 3 and 5). In this experiment, we pretreated with Zymolyase for 10 min prior to the addition of BP to allow spheroplast formation, but BP may be also added at the same time as Zymolyase.
In this report, we describe a method for the application of APEX2-dependent proximity biotinylation in yeast. To date, the APEX2 system has only been used in mammalian cells, but our results demonstrate that the APEX2 technique can also identify protein–protein interactions in yeast. This new technique expands the toolkit for detection of protein–protein interactions in these model organisms. Importantly, this technique opens the door for detailed studies examining whether APEX2 methodology can be extended to large-scale protein–protein interaction screens. In addition to specific binding partners, APEX2 fusion proteins will also label non-binding, protein neighbours due to effects of molecular crowding. However, specific binding partners should be labelled with higher efficiency, and these will differ among APEX2 fusion proteins. Indeed, this methodology may be most powerful when applied on a large scale since non-specifically labelled proteins will be common among samples, but specific binding partners will not. Application of software like that developed to analyse data from affinity purification-based approaches should identify high confidence interacting proteins for subsequent validation . Should the APEX2 technique extend to large-scale proteomic studies, this methodology will allow construction of spatial and temporal molecular maps of biological pathways and networks that can be paired with existing datasets from these well-studied model organisms.
Jiwon Hwang and Peter Espenshade conceived the study. Jiwon Hwang performed all experiments. Jiwon Hwang and Peter Espenshade analysed data and wrote the manuscript.
We thank the Ting Lab for sharing their mammalian protocol and advice. We thank Eric Spear and Susan Michaelis for providing the GAL1-driven plasmid and S. cerevisiae strain and assisting with S. cerevisiae transformation. We are grateful to all members of the Espenshade Lab for insightful discussions and general assistance.
This work was supported by the National Institute of Health [grant number R01HL077588].