Marked by association: techniques for proximity-dependent labeling of proteins in eukaryotic cells

Abstract

Various methods have been established for the purpose of identifying and characterizing protein–protein interactions (PPIs). This diverse toolbox provides researchers with options to overcome challenges specific to the nature of the proteins under investigation. Among these techniques is a category based on proximity-dependent labeling of proteins in living cells. These can be further partitioned into either hypothesis-based or unbiased screening methods, each with its own advantages and limitations. Approaches in which proteins of interest are fused to either modifying enzymes or receptor sequences allow for hypothesis-based testing of protein proximity. Protein crosslinking and BioID (proximity-dependent biotin identification) permit unbiased screening of protein proximity for a protein of interest. Here, we evaluate these approaches and their applications in living eukaryotic cells.

Introduction

In the wake of the genome revolution, we are faced with the daunting task of revealing not just the identity but also the function of all those genes, or more specifically the function of the proteins they encode. Far more often than not these functions remain partially or completely unknown. One common approach to provide clues as to the function of a protein is to identify which other proteins it associates with. With this knowledge of PPIs in hand, one can often make testable hypotheses as to the nature of the protein under study. To take this relatively simple paradigm further, it is clear that proteins do not exist solely in simple complexes but in vast interconnected networks, which we are only beginning to uncover. It is the field of systems biology that is most in need of a confident understanding of PPIs. Unfortunately, we are far from reaching this goal given the vast complexity of PPIs that vary dramatically with variables including cell type, proliferation state, environmental influences or age. Thankfully, there have been considerable advances in the variety and sophistication of methods to study protein associations. With more sensitive mass spectrometry combined with computational advances, it has become commonplace to identify tens or even hundreds of protein candidates from single experiments. And approaches such as protein-fragment complementation (e.g., split-GFP or split-luciferase) and forster (or fluorescence) resonance energy transfer (FRET) have enabled real-time observation of protein proximity in live cells. No one single method reigns supreme, but each contributes unique advantages to the growing toolbox available to the study of PPIs and, ultimately, protein function.

There are a variety of fundamental approaches to investigate PPIs. Some rely on the in vitro maintenance of direct physical protein interactions that have occurred in vivo. Such methods range from simple co-immunoprecipitation, where antibodies are utilized to isolate protein complexes, to sophisticated tandem affinity complex purification in which fusion proteins containing dual affinity tags are utilized to isolate intact protein complexes. These approaches often take advantage of mass spectrometry to identify and even quantify the composition of protein complexes [1]. Other methods rely on the observation of events that are triggered in vivo by protein interactions or proximity. Yeast-2-hybrid (Y2H) is the most widely utilized of these methods and utilizes the expression of selectable reporter genes in response to protein proximity [2]. Other examples include split-luciferase or GFP assays that rely on the generation of functional reporter proteins from non-functional fragments fused to proteins that reside in close proximity. Proximity ligation assay combines fluorescent antibody-based protein labeling, PCR amplification and fluorescent in situ hybridization to assess the proximity of two proteins or epitopes in fixed cells [3].

Why proximity-dependent protein labeling?

Another approach to investigate PPIs whose methods have expanded in recent years is based on proximity-dependent labeling of proteins in living cells. Proximity-dependent protein labeling refers to any general method to investigate protein proximity by specifically labeling proteins based on their proximity to each other. For the purpose of this discussion, labeling refers to a covalent modification of the protein. Intrinsic to these approaches is a fundamental limitation, namely that most by their very nature do not directly test for physical interactions between proteins. Instead, the presence of this labeling implies a spatial proximity that can be used to provide candidate protein interactors. These candidate interactions should be tested with other more direct methods if they are to be designated as genuine physical PPIs. Detection of PPIs is not the only useful application of these methods. The ability to observe protein proximity can provide valuable information about protein dynamics, allow the monitoring of post-translation modifications within a subpopulation of proteins, and can be used to reveal the constituency of discrete cellular structures. Some of these methods can also be applied to investigate intra-molecular proximity. To derive physiologically relevant information, techniques in proximity-dependent labeling are most useful when applied in living cells. Therefore, this review focuses on applications in living eukaryotic cells.

Variations on a theme: hypothesis-driven pursuit of protein interactions

Hypothesis-based analysis of PPIs occurs when specific pre-defined PPIs are investigated. The identities of these proteins must already be known a priori, and thus necessitates a pre-existing rationale to suspect potential interaction. Examples of such methods include common techniques such as FRET and protein-fragment complementation assays for which genetically engineered fusion proteins must be generated to test the interaction. Of course, these same technical principles can also be applied to non-hypothesis-based screens that use the power of genetics, for instance the protein-fragment complementation-based Y2H, to test large numbers of interactions. Advantages of the hypothesis-based approach, as compared to screening (often called ‘fishing’), include a wide selection of specific methods to choose from and the flexibility to test them in a variety of conditions, since the experiments are traditionally far simpler to perform than a large screen. Disadvantages include the obvious limitation of testing often binary interactions where the identity of the proteins must be known and the bias that this generates for the investigator to generate a positive interaction.

Among hypothesis-based methods of proximity-dependent protein labeling, there are a handful of examples. Most of these are variants of the BirA/BioTag system that harnesses BirA, the biotin ligase from E. coli. BirA is a dual function protein that in E. coli serves both to transcriptionally regulate the biotin synthetic operon as well as to biotinylate a subunit of the acetyl-coA carboxylase [4]. It is the latter function that has proven methodologically useful for protein isolation and analysis. Small biotin acceptor tags (BATs) have been designed such that they contain a lysine specifically biotinylated by BirA [5, 6]. Fusion proteins containing these BATs are co-expressed with BirA, often in eukaryotic cells, leading to the specific biotinylation of the fusion protein. As biotinylation is a relatively rare protein modification and is amenable to high-affinity capture with avidin/streptavidin, this system permits robust enrichment of the fusion protein for protein complex purification or chromatin immunoprecipitation. To accomplish hypothesis-based proximity-dependent protein labeling, the Ting group pioneered the use of a simple modification of the BioTag system in which BirA was fused to one protein of interest and co-expressed with a BAT-fusion protein [7]. Thus, biotinylation can only occur if both proteins are in close physical proximity (Fig. 1). Modifications to the BAT tag were made to reduce its affinity to BirA to avoid stabilizing the interaction or generation of an artificial interaction. This approach is conducive to monitoring interactions by both fluorescence microscopy and western blot analysis of protein biotinylation. In proof-of-principle experiments, which utilized the rapamycin-mediated interaction of FRB and FKBP as well as the interaction between Cdc25C and 14.3.3ε as positive controls, labeling was induced by cellular incubation with excess biotin for periods as short as 1 min. A variant of this system, named PUB-MS (proximity utilizing biotinylation and mass spectrometry) and developed by the Ogryzko group, used modified BAT sequences to facilitate mass spectrometry-based analysis [8]. Several PPI models were used to test PUB-MS including self-oligomerizing proteins such as TAP54α and HP1γ as well as the characterized binary interaction of KAP1 and HP1γ. Advantages of PUB-MS include an ability to utilize variable BAT sequences that theoretically permit evaluation of multiplex interactions. The Ogryzko group also utilized a pulse-chase approach, capitalizing on the permanence of the biotinylation, to monitor the fate of labeled proteins over a period of time [8]. This advantageous application likely extends to any of the proximity-dependent methods that permanently modify proteins in living cells.

Fig. 1.

Hypothesis-based methods for proximity dependent protein labeling. These methods rely on the co-expression of two fusion proteins, one fused to a ligase (Protein A) and the other to an acceptor peptide (Protein B). To test for an interaction between Protein A and Protein B the fusion proteins are co-expressed and labeling of protein B is evaluated. Ligases include BirA and LplA that specifically label an acceptor peptide sequence. The label is either biotin for BirA or a coumarin fluorophor for LplA. When the BirA, BAT combination is applied to cell surface proteins expressed in distinct cell populations for intercellular labeling this method is termed BLINC. The BirA/BAT, PUB-MS and BLINC methods utilize the BirA/BAT combination, whereas ID-PRIME uses the LplA/LAP combination. Both ID-PRIME and BLINC permit imaging of protein proximity in populations of live cells

A fundamentally similar approach also generated by the Ting group is termed BLINC (biotin labeling of intercellular contacts) [9]. BLINC is designed to monitor protein interactions at the cell surface and can be applied intercellularly (Fig. 1). Specifically, these experiments used the BirA and BAT system fused to cell surface receptors neurexin and neuroligin found at synaptic junctions of rat hippocampal neurons. One cell population expressed the BirA-fusion protein and the other the BAT-fusion protein. Upon cell contact and in the presence of biotinoyl-AMP, or biotin and ATP, the BAT-fusion protein can be covalently labeled with biotin permitting visual identification of the interaction with the BirA-fusion protein on the other cell. BLINC does not trap interactions and permits spatiotemporal assessment of protein interactions on the surface of live cells. Drawbacks to the method include the irreversibility of the biotin visualization due to the high affinity of biotin to streptavidin and the time required to visualize the biotinylation with streptavidin may be slower than the internalization rate of the biotinylated proteins. This means that not all of the biotinylated proteins may be observed and not all of the proteins that are biotinylated are actively interacting with the partner protein. However, BLINC provides a powerful tool to monitor dynamic protein interactions at the cell surface in live cells.

The Ogryzko group recently reported a modified version of PUB-MS called PUB-NChIP (proximity utilizing biotinylation with native chromatin immuno-precipitation; ChIP) in which a nuclear protein of interest is fused to BirA and co-expressed with a BAT-tagged histone [10]. As initially applied, the BAT was fused to specific histones and the ligase to Rad18, an E3 ubiquitin protein ligase associated with DNA repair. The net effect of PUB-NChIP is the biotinylation of specific histones associated with the protein of interest. This labeling permits the isolation of DNA associated with the subpopulation of histones proximate to the protein of interest. This method, similar to native ChIP [11], avoids the crosslinking necessary for conventional ChIP since core histones are relatively tightly associated with DNA. This preserves the ability to analyze posttranslational modifications on the histones since lysines are the most frequent targets of crosslinking with ChIP and also common sites of posttranslational modification. Other advantages of PUB-NChIP include the ability to utilize histone variants associated with specific functional states (e.g., active or repressed chromatin), the potential to observe post-translational modification of the labeled population and the capacity to perform pulse-chase experiments to monitor protein turnover. It is worth noting that these last two advantages extend beyond the simple purpose of detecting PPIs and may apply to other methods of proximity-dependent labeling such as BioID. One potential caveat with PUB-NChIP is the potential presence of naturally biotinylated histones [12]; however, these appear to exist in such extremely low abundance that they are unlikely to significantly impact the results and can easily be controlled for.

A technique called ID-PRIME (interaction-dependent probe incorporation mediated by enzymes), also developed by the Ting group, was developed to permit imaging of protein–protein interactions inside living cells [13]. ID-PRIME capitalizes on a mutant form of the E. coli lipoic acid ligase, LplA^W37V, to specifically attach a coumarin fluorophore to an LplA acceptor peptide (LAP). In principle, this approach is similar to the BirA-BAT system in that the ligase and acceptor peptide are independently fused to proteins chosen in a hypothesis-based manner and these are co-expressed in cells. Where ID-PRIME differs is by providing a method to visualize evidence of past or current interactions in living cells. Since the cells are loaded for a period of ~10 min and then unloaded to clear the unattached fluorophore for 30–50 min, what is observed is evidence of past interactions that may or may not remain. This system was tested with a multiple positive controls, including rapamycin-mediated interaction of FRB and FKBP and the leucine zipper domains of Jun and Fos. Although not a real-time observation of protein proximity, as is possible with FRET for example, ID-PRIME does permit a modest temporospatial analysis of protein proximity that suggests the evidence of a PPI.

These methods all share a need to express at least two fusion proteins, one containing a modifying enzyme and the other with the receptor for that modification. As compared to the size of GFP, a commonly accepted fusion protein that permits fluorescence-based imaging, both BirA and LplA are larger. At their largest dimension, GFP is ~4.6 nm (27 kDa) [14], BirA is ~7.0 nm (35 kDa) [15], and LplA is ~7.5 nm (38 kDa) [16]. This added bulk may impart undesired targeting or stability issues and, depending on placement in the fusion protein, may interfere with the normal protein function and interactions due to steric hindrance. An additional complication is that all of these methods rely on the proximity of the enzyme with the acceptor protein. The spatial configuration needed for successful labeling remains unclear, but is likely to vary on a case-by-case basis and will depend on factors such as the relative orientation of proteins to each other, accessibility of the acceptor sequences, and the presence, length and flexibility of any linker sequences between the enzyme or receptor and the bait. It is not difficult to imagine situations where sizeable proteins may interact, but, depending on the placement of the enzyme and acceptor tag, no labeling would occur. Such false negatives must always be considered when using these approaches. The main advantage of these approaches is that, by the covalent modification of one of the candidates, they can generate a permanent mark of protein proximity in live cells. This mark can be used to monitor distribution, natural post-translational modifications, and fate of these proteins.

Fishing for friends with proximity-dependent labeling

Often, researchers cannot rely on the hypothesis-based method to test PPIs. Sometimes, there is a need for methods to detect interactions that are not already known or surmised. These fall into the broad category known as fishing or screening, in which the bait is defined but the identities of the protein interactors (prey) are unknown. In some instances, such as Y2H, there are large libraries of prey that are screened for an interaction with the bait. Other methods attempt to probe PPIs that have occurred with endogenous prey under physiological conditions. The most common of these approaches is affinity complex purification. In this situation, prey are identified by bottom-up mass spectrometry. It is advances in bottom-up mass spectrometry that have most dramatically facilitated these PPI fishing expeditions. There are two fundamentally different methods to fish for candidate interactions in vivo that both rely on proximity-dependent labeling, protein crosslinking, and BioID.

Protein crosslinking: proximity-dependent tethering

Crosslinking is by far the most utilized method of proximity-dependent protein labeling and a detailed evaluation is beyond the scope of this review, but has been well covered elsewhere [17–24]. For the purposes of detecting PPIs, crosslinking is essentially a variant of affinity complex purification that seeks to overcome the loss of interactions during solubilization of the protein complexes. This is accomplished by the covalent crosslinking of proximate proteins (or other molecules) while the cells are intact and essentially alive (Fig. 2). A variety of reagents can be used to obtain this physical crosslinking, but all are based on the use of molecules with at least two reactive groups that serve to tether proteins to their neighbors. Following this crosslinking procedure, the approach follows that of affinity complex purification in which a protein is isolated via some specific property (e.g., an affinity purification domain or epitope) and the proteins to which it is physically tethered can be identified by mass spectrometry. Goals of chemical crosslinking in proteomics are twofold: the most common one is identification of constituents in protein complexes while the other is to map interface sites between or within proteins. It is this latter goal that makes chemical crosslinking unique in its potential to step beyond the simple identification of protein complex constituents. This relies on the detection of specific sites where inter- or intra-molecular links have occurred. These linked peptides signify a site where those sequences are in close proximity and can be used to identify actual interfaces of PPIs or potentially intra-molecular interfaces within a protein. This application of crosslinking is more comprehensively covered elsewhere [25–30].

Fig. 2.

Overview of protein crosslinking to identify PPIs and/or map protein interfaces. Live cells are treated with a crosslinking reagent to covalently connect proximate proteins. These fixed cells are lysed and crosslinked protein complexes are released into solution for affinity purification of the desired complex(es). The purified complex(es) is digested into peptide fragments which are analyzed by mass spectrometry to identify the constituents of the complex and/or by using inter- or intra-crosslinked peptides to determine the structure and/or interfaces of the proteins in the complex

There are numerous reagents spanning distances ranging from <3–40 Å that can be utilized for chemical crosslinking. These include formaldehyde, NHS esters, carbodiimides and benzophenones. Reactions can occur through a variety of chemistries to sites within proteins such as side chains or the N-termini. Some of these reagents can be photoactivated to permit the user to control the timing of the reaction by exposing samples to UV light. More sophisticated approaches include the incorporation of unnatural amino acids with photo-reactive side chains to covalently trap proximate proteins [31]. Another such method uses protein interaction reporters (PIR) that contain an affinity motif and can be cleaved in a two-step process by low-energy MS [32]. The first step is to identify the ions with crosslinks and their nature (intra- or inter-molecular) and the second is to release the individual peptides for identification.

TRAP-crosslinking (targeted releasable affinity probe) is a notable method that limits protein crosslinking to only to one specific protein of interest [33]. This approach generated by the Mayer laboratory is based on the fusion of a cell permeable photo-activatable benzophenone crosslinker to a fluorescent biarsenical probe (the TRAP probe) that binds to engineered tetracysteine motifs within the protein of interest. A fusion protein is expressed in vivo that contains one or more of these short tetracysteine motifs. The TRAP probe can be added to the cells to fluorescently label the proteins and, when photo-activated, to crosslink proximate proteins. Finally, with the use of dithiols to break the association with the tetracysteine tag, the fluorescent probe can be released from the protein of interest where it remains on the interacting protein to facilitate identification of the proteins, and potentially the specific binding site, by mass spectrometry. TRAP was validated in vitro and in vivo in a prokaryotic system and applied to cultured myocytes to identify an interaction between phospholamban and fibronectin. Clearly, the main advantage of TRAP is its ability to focus only on the protein of interest among the complex mixture of proteins found in living cells. Advantages include ability to release the biarsenical fluorophore from the protein of interest with high concentrations of dithiols. Potential limitations include palmitoylation and oxidation of the tetracysteine motif [34], and uncertainty of consequences to normal protein behavior during the period of photoactivation (~2 h). In principle, this approach has the potential to provide a powerful tool to fish for proximate proteins in a physiological setting while avoiding many of the drawbacks associated with conventional crosslinking.

For the purposes of proximity-dependent labeling, protein crosslinking typically relies on affinity capture of the crosslinked proteins. Highly mobile proteins that engage in transient interactions would require different crosslinking conditions than immobile proteins in a large complex or protein matrix. If the protein of interest has subpopulations with variable properties, it is difficult to imagine optimizing the crosslinking conditions to capture relevant interactions that span all subpopulations. Successful crosslinking is a delicate balance, with concentrations of crosslinking reagents and duration of crosslinking affecting the outcome. Too much crosslinking can lead to protein ‘loss’ due to fixation in large insoluble complexes, whereas too little reduces the capture of weak and transient interactions. There also exists the possibility of masking epitopes or otherwise interfering with affinity complex purification. Ideally, the net effect of such crosslinking is akin to taking a snapshot of interactions at the time of fixation. As such, crosslinking excels at the identification and potentially the structural characterization of relatively discrete and stable protein complexes. Where it faces limitations is with transient interactions, especially those that occur in low abundance. For example, if only 1 % of a kinase (the bait) interacts with a specific substrate (the prey) at any one point in time it may be difficult to identify that kinase–substrate interaction with conventional crosslinking, especially if the bait is a protein of relatively low abundance.

BioID: following the protein footprints

Another method of screening or ‘fishing’ for protein interactions has recently been reported [35]. Called BioID (proximity dependent biotin identification), this approach capitalizes on a mutant prokaryotic biotin ligase capable of promiscuous biotinylation [36, 37]. When expressed in cells, the BioID fusion protein biotinylates proximate proteins, permitting their selective isolation with avidin/streptavidin-based biotin affinity capture (Fig. 3). These proteins can be identified by immunoblot analysis if hypothesis-based interactions are being investigated or by mass spectrometry if ‘fishing’ for candidates. BioID generates a history of protein footprints that occurred over a period of time in a relatively natural cellular setting. In this way, BioID is very different from crosslinking approaches that capture the snapshot of protein proximity. In theory, transient interactions will accumulate biotin labeling over time leading to an increased chance of identification with BioID as compared to crosslinking.

Fig. 3.

Outline of the BioID method. A BioID-fusion protein, consisting of a bait protein fused to a promiscuous biotin ligase, is expressed in cells. Biotin is added to the cells for a period of labeling, during which time proteins proximate to the BioID-fusion protein are biotinylated. The cells are subsequently lysed, all of the proteins are denatured and the biotinylated proteins are selectively isolated for identification by mass spectrometry. The identified proteins are those proteins that were proximate to the bait during the labeling period and thus represent candidates PPIs

Since biotinylation is a covalent modification, biotin–avidin interactions are nearly covalent [38] and there is no need to maintain protein–protein interactions, the proteins can be solubilized and affinity captured under extremely stringent conditions. An obvious benefit of this nearly covalent affinity capture is the ability to apply extremely stringent wash conditions (including buffers containing 2 % SDS or 500 mM NaCl) that remove many contaminants typically detected with affinity complex purification. Another advantage of BioID is its applicability to insoluble proteins (e.g., intermediate filaments or integral membrane proteins) that are traditionally difficult to study by conventional affinity complex purification or Y2H.

Supplemental biotin in the cell media is necessary for robust biotinylation by the BioID fusion protein, perhaps due to a reduced affinity of the mutant ligase to biotin and/or the limited levels founds in tissue culture media. This requirement for excess biotin provides a mechanism to temporally regulate the biotinylation process to meet the experimental needs. Based on the initial reported use of BioID, it appears that biotinylation is saturated prior to 24 h, whereas biotinylation becomes noticeably detected above background within 1 h [35]. These times may be sufficient for temporally restricted experiments such as selective observation of cell-cycle-specific protein behavior.

So far, there have only been two published applications of BioID. In the first, a constituent of the nuclear lamina was fused to BioID (BioID-lamin A) and expressed in human cells [35]. Over 100 protein candidates were identified and ranked according to abundance. The majority of those candidates fit with what is known about the protein constituency of the nuclear lamina. Also identified were several uncharacterized proteins, the most abundant of which was shown to predominantly reside at the nuclear envelope. This protein, named SLAP75, represents one of a very few non-lamin proteins known to specifically localize to the nuclear envelope despite the lack of a transmembrane domain. This may reflect the predicted strength of BioID to identify weak and/or transient PPIs often missed by other methods. The second application of BioID was in the unicellular eukaryote, Typanosoma brucei [39]. To identify novel constituents of the discrete insoluble cytoskeletal structure called the bilobe, Morriswood et al. fused BioID to TbMORN1, one of the few known bilobe proteins. By analyzing insoluble proteins biotinylated by BioID–TbMORN1, seven novel bilobe proteins and two novel flagellar attachment proteins were identified. These results demonstrate the utility of BioID in identifying protein constituents of discrete intractable cellular structures and its versatility in application within divergent model systems.

BioID is not without its drawbacks and limitations [40]. There is the obvious need to express at least low levels of an exogenous fusion protein. As mentioned above, the BirA enzyme does add to the size of the protein and could compromise its targeting and/or function. During the biotinylation process, the irreversible covalent modification of primary amines may impact the function of labeled proteins, at least by blocking biotinylated sites (which include lysines) from alternative modifications. Another limitation of BioID relates to how the results can be interpreted. In theory, false negatives could arise from proteins without proximate reactive primary amines. Positive candidates do not prove direct interaction with the BioID fusion protein. Labeled candidates may instead reside in close proximity to the fusion protein, but not physically interact. This is the result of BioID’s mechanism of promiscuous biotinylation. The wild-type BirA uses biotin and ATP to generate biotinoyl-AMP [41]. BirA holds on to that reactive biotin molecule until it is covalently attached to a very specific substrate. The mutant BirA (R118G) used in BioID has reduced affinity to biotinoyl-AMP [42] and is thought to prematurely release the reactive biotin molecule [36, 37]. It is this reactive biotin that labels proximate proteins. What remains unclear is the extent to which the reactive biotin can diffuse away from the ligase before reacting with a protein or being otherwise hydrolyzed. In vitro, a similar adenylate is reported to be relatively stable (nonenzymatic hydrolysis rate ~0.7 × 10⁻³ per second) implying a potential for distal labeling [43]. The hydrolysis rate of biotinoyl-AMP in vivo remains unknown; however, it is likely to be rapid due in part to the high concentration of reactive sites (e.g., reactive primary amines on proteins) in a biological setting. By interpreting the proteomic data from BioID-LaA, a labeling distance of <20 nm has been proposed [35]; however, this is based on only one report and may vary with each application and cellular environment. Future studies will be needed to resolve the practical range of biotinylation by BioID. To facilitate successful application of BioID, a detailed protocol that includes extensive analysis of its strengths and limitations has been published [40].

Conclusions

Techniques involving proximity-dependent labeling do not currently represent commonly utilized ‘front-line’ approaches to detect and monitor PPIs. With the exception of protein crosslinking, none have been used more than once or twice in the literature, and are often presented in a proof-of-principle format. It remains to be seen whether any of the hypothesis-based techniques reach utilization levels similar to the more widespread approaches such as FRET or protein-fragment complementation. And although certainly more established, techniques in protein crosslinking face challenges in widespread acceptance, simplicity of use, and data interpretation, especially for the more sophisticated variants. As the newest member of this group, the fate of BioID remains unclear. Further studies are needed to resolve issues surrounding the spatial dimensions of labeling that impact data interpretation and provide a clearer picture as to the strengths and limitation of BioID. Hopefully, the same vision and creativity that led to the inception of these methods will ultimately lead to their appropriate adoption and creative use by the broader scientific community.

Non-Standard Abbreviations

BAT

Biotin acceptor tag

BioID

Proximity-dependent biotin identification

BLINC

Biotin labeling of intercellular contacts

ChIP

Chromatin immuno-precipitation

FRET

Forster (fluorescence) resonance energy transfer

ID-PRIME

Interaction-dependent probe incorporation mediated by enzymes

LAP

LplA acceptor peptide

PIR

Protein interaction reporter

PPI

Protein–protein interaction

PUB-MS

Proximity utilizing biotinylation and mass spectrometry

PUB-NChIP

Proximity utilizing biotinylation with native ChIP

TRAP

Targeted releasable affinity probe

Y2H

Yeast-2-hybrid