|Home | About | Journals | Submit | Contact Us | Français|
Two hybrid systems are one of the most popular, preferred, cost effective and scalable in vivo genetic approaches for screening protein-protein interactions. A number of variants of yeast and bacterial two hybrid systems exist, rendering them ideal for modern, flexible proteomics-driven studies. For mapping protein interactions at genome scales (that is, constructing an interactome), the yeast two hybrid system has been extensively tested and is preferred over bacterial two hybrid systems, given that users have created more resources such as a variety of vectors and other modifications. Each system has its own advantages and limitations and thus needs to be compared directly. For instance, the bacterial two hybrid method seems a better fit than the yeast two hybrid system to screen membrane associated proteins. In this chapter, we provide detailed protocols for yeast and bacterial two hybrid systems as well as a comparison of outcomes for each approach using our own and published data.
Most of the key biological processes of the cellular machinery, including transcription, translation, protein quality control, transport, and signaling are dependent on protein–protein interactions (PPIs) that occur either individually or in multiprotein complexes. For example, the 26 S proteasome requires interactions among more than thirty different proteins organized in three sub-complexes. Together, these proteins serve a common set of functions.
The mapping of protein-proteins interactions on a genome or proteome scale remains challenging and has been comprehensively explored only in a few model organisms such as Escherichia coli, budding yeast, and humans. Numerous biochemical and genetic approaches have been developed to study protein-protein interactions. This review focuses on The Yeast (Y2H) and Bacterial (B2H) Two Hybrid methods which are among the most explored in vivo genetic tools to map PPIs. The full extent of biochemical and genetic approaches in use is beyond the scope of this chapter but several detailed surveys have been published (e.g. PMID: 24393018, PMID: 22688816, PMID: 24693427).
Both Y2H and B2H are methods that have been used to map and identify PPIs networks in many organisms (Mehla et al., 2015 JB, Yi Wang, 2010). They can be used on a small scale or as high-throughput screening approaches. Two-hybrid methods are based on the principle of restoring protein activity upon non-covalent reconstitution of split protein fragments. One fragment of a modular protein is fused to a protein X, and the other fragment is fused to a protein Y. The resulting fusion proteins may be referred to as a bait and prey. If the proteins X and Y interact upon co-expression, the modular protein is reconstituted, regains its activity, and its activity is detected through a reporter gene. Although both the B2H and Y2H work on similar principles, there are fundamental differences and both have distinct advantages and limitations. Different users may choose one over another depending on their needs, available resources, properties of the proteins to be screened, and ease of screening. This chapter describes a detailed protocol for high-throughput yeast and bacterial two hybrid methods and a comparison of their use and some typical results.
The B2H system may be used as an alternative to Y2H screening. The B2H system has been applied to screening of proteins from different sources (bacterial, mammalian, viral), of different size and cellular location (cytoplasmic, peripheral, inner membrane) and of different functions (transporters, enzymes, regulatory proteins, etc.). Although we do not know of any systematic comparison of the two systems, the B2H may be preferable to Y2H when screening bacterial proteins as the screening conditions are more biologically similar to the proteins' natural environment.
The most commonly used B2H system is the Bacterial Adenylate Cyclase-based Two-Hybrid System (BACTH) which is based on the reconstitution of adenylate cyclase activity in Escherichia coli (Fig 1). The catalytic domain of bacterial adenylate cyclase has two sub-domains: a 25 kDa fragment (T25, or cyaAT25) with a catalytic site and a 18 kDa fragment (T18, or cyaAT18) with a binding site for calmodulin. These domains are not forming a functional protein when they are physically separated. The test proteins (X and Y) are fused to the T25 and T18 domains, respectively, allowing interactions between them to bring the domains in close proximity and – in the presence of calmodulin – restore adenylate cyclase activity. The restored activity leads to cyclic adenosine monophosphate (cAMP) synthesis. The cAMP produced by the reconstituted adenylate cyclase forms a complex with catabolite activator protein (CAP), inducing the transcriptional activation of catabolic operons such as lactose or maltose and yielding a characteristic phenotype in a cya- E. coli strain (usually BTH101/DHM1) on selective or indicator media plates.
Bacterial two hybrid screens as described by Karimova et al. (1998) frequently employ E. coli BTH101 and/or DHM1 as reporter strains. Both strains are adenylate cyclase deficient but only the latter of these two strains contains a recA1 mutation, potentially improving plasmid stability. Store these stocks at −80°C.
BTH101: F−, cya-99, araD139, galE15, galK16, rpsL1, hsdR2, mcrA1, mcrB1
DHM1: F−, cya-854, recA1, endA1, gyrA96, thi1, hsdR17, spoT1, rfbD1, glnV44 (AS)
As described above, bacterial two hybrid screens use vectors allowing for expression of hybrid proteins containing either T25 or T18 adenylate cyclase fragments. The vectors pKT25 (Kanr, cyaAT25 fragment) and pUT18 (Ampr, cyaAT18) are available in several variants (see table 1).
Genes encoding the proteins of interest (X and Y) are amplified by PCR and cloned by traditional or recombination cloning (e.g. using the Gateway system, Hartley et al. 2000). The latter method is preferred for preparation of clone sets as it can be performed with higher cloning efficiency than traditional cloning. With Gateway cloning, the attL-flanked ORFs can be recombined into the attR-flanked BACTH-DEST plasmids (pST25 and pUT18C) using the LR reaction to generate attB-flanked ORFs within the BACTH vectors. Expression clones may be prepared with the vectors pST25, pSTM25, pUT18, and pUTM18, among others. Outer membrane (OM) bound, peripheral or extracytoplasmic proteins should be transferred to pSTM25 or pUTM18 B2H expression vectors. These modified vectors have an extra transmembrane segment fused downstream to T25 and T18 cyclase domains to detect interactions for OM or peripheral proteins (Ouellette et al., 2014). The Gateway vectors should be propagated in ccdB resistant E. coli strains at 30°C.
E. coli cultures may be grown with LB media. For LB/X-Gal indicator plates, add X-Gal (5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside) to 40 µg/ml. IPTG may also be included in these plates (at 0.5 mM) to improve protein expression and hence observation of interactions.
MacConkey agar with 1% maltose may also be used for indicator plates.
Where specified, ampicillin (Amp) and spectinomycin (Spc) should each be used at 100 µg/ml.
One liter of Z buffer is prepared as follows:
16.1 g Na2HPO4 • 7H2O (or 8.5 g anhydrous)
5.5 g NaH2PO4 • H2O
0.75 g KCl
0.246 g MgSO4 • 7H2O
Add H2O to the final volume and adjust to a pH of 7 if needed. Store at 4°C.
Add 2.7 µl β-mercaptoethanol (β-ME) per ml of Z buffer immediately before use in an assay. Retain a portion of Z buffer without β-ME added (see review by Battesti and Bouveret 2012 for further details.)
ONPG (2-Nitrophenyl β-D-galactopyranoside) is used to assay β-galactosidase activity and hence the strength of a protein interaction. Solutions may be prepared as 4 mg/ml in Z-buffer and should be prepared fresh on the day of each assay.
This procedure also requires the following solutions: 0.1% SDS, chloroform for cell lysis and 1M Na2CO3 to stop the enzymatic reaction.
See Fig. 2 for a summary of this procedure.
Expression constructs of test proteins encoding the T25-X (or X-T25) and T18-Y (or Y-T18) fusions are co-transformed into competent BTH101 E. coli. Competent cells can be prepared on the same day as transformation. Routine heat-shock based transformation protocols are ideal, especially for high-throughput screens. See Chung et al. (1989) for a basic protocol.
Select the co-transformants for further screening on indicator plates at 30°C.
Indicator or selective media plates can be used to detect positive interactions between test proteins. The positive interactions should produce specific phenotypes on respective indicator plates, i.e. blue colonies on LB-X-Gal or red on MacConkey-maltose. This may take up to 4 days. If no interaction (negative result) occurs, all colonies should be colorless on indicator plates. This method can be scaled up for screening libraries.
The same batch of cells can be used for both plating on indicator plates and for quantification using the β-galactosidase assay.
The interactions between proteins (X and Y) can be quantified by measuring β-galactosidase activity in liquid cultures using the classical Miller assay (Miller 1972). The β-galactosidase activity is usually expressed in Miller units (one unit corresponds to 1 nmol of ONPG hydrolyzed per min at 28°C) per mg of bacterial dry weight.
This protocol is used to determine the level of β-galactosidase activity in a bacterial two hybrid interaction screen and may be scaled up for high-throughput use in 96 well plate format. The level of galactosidase activity will vary depending on the level of interaction for strongly and weakly interacting proteins. Thus, we can quantify the strength of an interaction between two proteins. The time required for the assay to develop reliable readings/color can vary depending on the strength of the interactions, especially if the interaction is very weak.
Prepare overnight cultures of each co-transformed culture to be assayed. Transfer cultures to a 96 well plate after their overnight growth. Set an incubator or water bath to 30°C.
The Y2H assay is an in vivo genetic method developed by Fields and Song (1989) to screen for binary protein-protein interactions. In this method, the yeast transcription factor Gal4 is split into a DNA-binding domain (DBD) and an activation domain (AD). Protein pairs are fused to the DBD and AD and the resulting DBD-X and AD-Y fusions are generally referred to as bait and prey. Only the simultaneous co-expression of both the bait and an interacting prey will reconstitute the full transcription factor and permit transcription (Fig. 3).
The subjects of a two hybrid screen may be random libraries, such as fragments of genomic or cDNA, or defined clone sets, usually in array format. Arrays are preferred as they are easily scalable (i.e., a full 96 well plate of expression constructs can be screened against one protein of interest on a single plate) and do not require sequencing to verify the identity of each interactor. Using arrays may also help to decrease the incidence of false positive results.
The following protocol uses the yeast strains AH109 (for baits) and Y187 (for preys) for high-throughput yeast two hybrid screens. These strains are frequently used and grow consistently, especially when provided with supplemental adenine in media (up to 100 mg/L).
AH109: MATa, trp1-901, leu2-3, 112, ura3-52, his3-200, gal4Δ, gal80Δ, LYS2::GAL1UAS-GAL1TATA-HIS3, GAL2UAS-GAL2TATA-ADE2, URA3::MEL1UAS-MEL1TATA-lacZ (James et al. 1996).
Y187: MATα, ura3-52, his3-200, ade2-101, trp1-901, leu2-3, 112, gal4Δ, met−, gal80Δ, URA3::GAL1UAS-GAL1TATA-lacZ (Harper et al. 1993).
The selection mechanism used for diploid cultures in yeast two hybrid screens will depend upon the strains and vectors in use. Many methods employ nutritional selection, i.e. if the two yeast strains are auxotrophs for leucine and tryptophan, the yeast two hybrid vectors must contain complementing genes, e.g. LEU2 and TRP1, and diploid cultures must be isolated on media lacking both amino acids. For yeast two hybrid screens using HIS3 as a reporter gene, selective yeast media may be supplemented with 3-AT (3-amino-1,2,4-triazole) at concentrations between 1 mM and 50 mM. 3-AT acts as a competitive inhibitor of the HIS3 enzyme (imidazoleglycerol-phosphate dehydratase) which will therefore reduce background growth, allowing for easier distinction of positive colonies from background.
Selective solid yeast medium, in single well plates (e.g. Nunc Omnitrays) (including –L, –T, –LT, –LTH media)
10X TE buffer (0.1 M Tris-HCl, 10 mM EDTA), pH 7.5
YEPDA medium is prepared as:
|Adenine hemisulfate salt||0.01%|
Sterilize by autoclaving. For solid medium, include 1.6% agar and pour 40–50 mL of sterile medium into each single-well Omnitray plate. Allow to solidify.
See Fig. 4 for a summary of this procedure.
These following methods have been designed to be performed using 96-well plates (or 96-colony plates). The protocol may be used with manual colony transfer or in high-throughput with the aid of a laboratory automation system, such as a Beckman Coulter Biomek 2000, FX or FXP, along with a high density plate replicator attachment (e.g. 96/384 HDR Tool Body, V&P Scientific) and compatible pins (“thin” pins of 0.381 to 0.457 mm or “thick” pins of 1.143 to 1.58 mm in diameter). Plates may also be replicated manually using a 96-pin or 384-pin plate replicator. Solid agar media should be prepared in single-well 86 × 128 mm Omnitray microtiter plates (Thermo Fisher/Nunc).
Plate replicator pins must be sterilized between each pinning or they may introduce cross-contamination. Rinse the tips of each pin in 20% bleach (60 s), sterile dH2O (20 s), ethanol (≥95% v/v; 60 s), and sterile dH2O (20 s).
Two-hybrid methods traditionally define screened fusion proteins as baits and preys. This distinction is ensuring that the two vectors contain different genes for selection. Otherwise, nearly any protein may be used as a bait or prey.
Some ORF collections and libraries provide sequences (including, in some cases, nearly complete genomes) in two hybrid vectors. These vectors may require subcloning or a recombination-based cloning procedure (e.g. Gateway cloning) to produce a full set of expression vectors for two hybrid screens. ORFeomes are available for viral genomes (e.g., herpesviruses; Uetz et al. 2006), several bacterial genomes (e.g., E. coli; Rajagopala et al. 2010), and some eukaryote species (e.g. C. elegans (Lamesch et al. 2004), human (Rual et al. 2004), and Arabidopsis thaliana (Gong et al. 2004).
If a bait protein is screened against a random genomic (or cDNA) library of prey proteins in a library screen identification of the interacting prey proteins requires isolation and sequencing of positive clones. For further technical details, see Mehla et al. (2015).
Arrays have certain advantages for Y2H screens, especially to validate the positive interactions in a library screen (Mehla et al., 2015 A Tool for Mapping Protein–Protein Interactions). Because certain baits may self-activate (i.e. produce positives without an interactions), usually the preys are arrayed in a defined order before the actual Y2H screen or retest. Prey proteins usually do not generally produce self-activation without a bait protein.
For genome-wide screens, automation is strongly recommended, given that manual work is not only straining, and error prone, but also time consuming and tedious. However, automation can be an expensive option. Collaborating with a lab already in possession of automation equipment may be an alternative.
The ORFs from entry vectors are assembled into Gateway-compatible prey or bait vectors by Gateway cloning. The Gateway-compatible bait and prey vectors are shown in table 1. The prey and bait vectors are transformed into mating-competent, haploid yeast strains Y187 (mating type ‘α’) and AH109 (mating type ‘a’), respectively.
This method is optimized for high-throughput transformation of bait or prey plasmid clones into respective yeast strains and can be scaled up and down as required.
Strains, such as Y187, may require 3 to 4 days to yield noticeable transformant growth. The specific transformants (each carrying one specific prey vector), are arrayed on selective media plates in a 96- or 384-format in duplicates or quadruplicates. For freeze stocks, the arrays may be cultured in rich (YEPDA) or selective liquid media, resuspended in 20% glycerol, and stored at −80°C for long-term storage. For routine use, the arrays can be stored on solid selective medium for up to 3 months at 4°C and fresh arrays can be propagated using storage plates prior to Y2H screen.
For long term storage, the master prey array should be kept on selective plates (e.g -Leu plates). The master copy of the array should only be used to make fresh copies on YEPDA agar plates for mating. Copying the array onto fresh selective plates every 2 weeks should protect against plasmid loss and contamination.
The bait auto activation test is a prerequisite prior to Y2H screen and is performed to measure the background potential of baits to activate transcription of the reporter gene (here: HIS3) in the absence of prey protein (or presence of empty prey vector). This will help to optimize actual Y2H screen conditions, saving time and resources.
After the bait self-activation test, the actual Y2H screens are performed as described below.
The screening plates should be monitored daily for positive colonies. Typical positive interactions should appear within 3 to 5 days.
The main concern associated with the Y2H screen is the possibility of a large number of false positives. The use of different vectors and retesting of all the positive interactions through repeated mating can help to reduce the number of false positives. Only the reproducible interactions in retest Y2H screens will be counted as positive interactions. Moreover, screening the same baits and preys using multiple vectors provides additional evidence for positive interactions. Alternative methods are also available for confirmation of Y2H interactions (for protocols see Serebriiskii et al. 2005; Rajagopala et al 2007).
The Y2H screens can capture only direct pairwise interactions, including weak interactions. Y2H screens are a heterologous system for bacterial proteins which comes with some limitations. For example, non-physiological level of protein expression, absence of necessary cofactors, ligands or chaperones for functional activity and issues with the translocation of bacterial proteins to the yeast nucleus. This leads to low sensitivity and coverage and up to 70% of PPIs are not detected even in comprehensive screens. The specificity of the Y2H screens is enhanced with the advancement of multi vector and next generation Y2H screens for mapping membrane interactomes (Stellberger et al. 2010, PMID: 22610515).
The B2H methods are developed to be performed in their natural host and thus can detect both direct and indirect protein interactions. The adenylyl cyclase based B2H is the most commonly used B2H technique and is suitable for detecting interactions between cytoplasmic, membrane proteins and periplasmic proteins. Most of the B2H studies are performed at small scale. Traditionally, the yeast two hybrid (Y2H) system has been the preferred choice over B2H, for detecting protein–protein interactions on genome wide scale. However, there is no objective reason why the B2H should not be applicable on a large-scale.
Two hybrid methods are popular fixtures in the biomedical literature, as a chronological search of PubMed reveals (Fig. 5). Here, we show the total number of publications per year containing the phrase “two hybrid” anywhere in their title or abstract. At its peak in the early 2000’s, two hybrid methods were highlighted in more than 900 publications each year, with yeast two hybrid approaches providing the majority of those methods. Bacterial two hybrid methods have been comparatively less popular but have been modified to be used in concert with other methods (Serebriiskii et al. 2005) or to more easily confirm expression of the desired fusion proteins (Battesti and Bouveret 2008). Some of the difference in popularity between yeast and bacterial two hybrid methods may simply be due to the time since their introduction: Karimova et al. did not publish details of their bacterial two hybrid method until 1998, nearly a decade after Fields and Song described Y2H (see Fig. 6 for counts of citations per year to this paper alone). It should also be noted that two hybrid methods of any type may not always serve as a project’s primary data source: 370 publications from 2015 in PubMed included bacterial two hybrid methods though just 40 in the same year mentioned the method in their title or abstract. Some publications may use two hybrid approaches to generate further experimental evidence and may therefore avoid discussing the method in detail. Though perhaps increasingly seen as a basis for new approaches rather than a core methodology, two hybrid methods remain prominent among protein-protein interaction screens.
Likely owing to their popularity and ease of implementation, two hybrid methods have not only remained popular approaches but have contributed more than a third of all interactions in the IntAct molecular interaction database (Fig. 7). These counts include all interactions in the database – even if the interaction was studied by more than one group or with more than one method – and therefore serve as a summary of publicly available interaction data and the methods responsible. For some species, two hybrid methods are especially popular: more than half of the protein-protein interactions in IntAct among bacterial proteins were identified using two hybrid approaches.
The main concerns of two hybrid methods are the rate of false positives and the bait self-activation especially for Y2H screens. Despite its extensive use, two hybrid methods have rarely been explored simultaneously or in parallel. To better understand the experimental outcomes of B2H and Y2H methods, an extensive study should be done to compare different types of vector, types of proteins and parameters affected by the host. The types of proteins may be further classified in terms of their functions and location. More extensive studies are required to comprehend the outcomes and suitability of two hybrid methods. We tried to get an idea of the outcomes of B2H and Y2H screens by screening the same set of interacting proteins by both B2H and Y2H (Fig. 8). A protein “X“ was tested against a set of 12 prey partners by both methods. Few positive interactions are captured in our B2H screens, however not even a single interaction was detected by Y2H method. There can be several reasons for this result. The most plausible explanation is that the fusion proteins in both systems are sufficiently different to allow an interaction in one system but not the other. Rajagopala et al. (2009) as well as Chen et al. (2010) have shown that small differences even within the Y2H can cause substantial outcomes in the set of detected PPIs. Given that only a fraction of all interactions are detected in each screen, this explains even non-overlapping resulting PPIs in different screens. Most of the proteins selected in our example are membrane proteins which is another confounding reason that not a single interaction was captured in Y2H screen. Also, the function of the interacting proteins might have some implications. For instance, directly or indirectly, the selected proteins are predicted to be involved in the LPS assembly or outer membrane asymmetry in bacteria and thus could lead to lower mating efficiency and even may cause toxicity in the yeast cells. The interactions detected by Y2H screens took place in the nucleus and thus the membrane or outer membrane proteins which are having hydrophobic transmembrane region/domains could not be able to reach the nucleus and ultimately no interaction (Auerbach and Stagljar 2005). In B2H screens, the interactions are not restricted to the nucleus and only the protein complementation is required to detect the positive interactions. That could give the B2H screens a certain advantage for screening inner membrane (or even peripheral membrane) proteins unless a specialized membrane yeast two hybrid assay is used. That is why the interactions shown here are captured only by B2H screens.
Also, the proper folding of the membrane proteins might need the phospholipids bilayers and thus can lead to false negatives or to no interaction in Y2H screens. These limitations could be overcome by screening only the cytoplasmic domain of the membrane proteins or using a low copy number plasmid to avoid any toxicity (if any). Not only the localization of the proteins causes false negatives, also the specific condition (e.g oxidative, cofactors etc.) or posttranslational modification required for the proper folding/activity of the proteins can lead to higher rate of false negatives. Another reason for higher false negatives in Y2H screens, could be the toxicity of interacting proteins upon overexpression (Brückner et al. 2009).
Affinity purification coupled with mass spectrometry (AP–MS) is the most popular biochemical method for detecting PPIs. In AP/MS, the tagged bait proteins are used to pull out interacting prey proteins. The two hybrid methods are widely used and have produced large number of interactions available in different databases (Fig. 7). Two hybrid methods are preferred because of following:
However, several features of two hybrid assay impose limitations on the type and nature of interactions that can be analyzed.
No single two hybrid method is best suitable for every individual or lab. The selection of methods for mapping protein-protein interactions is dependent on the needs, and resources availability. A comprehensive analysis of these methods in terms of nature of proteins, host and methodology, might help in planning and choosing a better approach for mapping protein-protein interactions.