Bimolecular Fluorescence Complementation (BiFC): A Colorful Future in Drug Discovery
By Y. John Shyu, Kazuhito Akasaka and Chang-Deng Hu, Department of Medicinal Chemistry and Molecular Pharmacology, School of Pharmacy and Pharmaceutical Sciences & the Purdue Cancer Center, Purdue University
Protein-protein interactions play pivotal roles in carrying out many cellular functions and are the basis of all signal transduction pathways in cells (1). Given that small molecule inhibitors of protein-protein interactions are valuable tools for dissecting complex biological signaling pathways and for therapeutic purposes, protein-protein interactions, in theory, are ideal molecular targets for drug development. Unfortunately, the search for small molecule inhibitors of protein-protein interactions has been avoided because of an unsuccessful history(2). However, recent progress has shown that protein-protein interactions can be targeted through the binding of small molecules to few ‘hot spot’ contacts or by causing allosteric inhibition(3-11). Given that tremendous progress in genome-wide interaction mapping has been made(12-18), these successes will undoubtedly inspire the search for small molecule inhibitors of protein-protein interactions in the near future.
Over the past decade, high-throughput screening (HTS) has evolved as an important tool for identifying small molecule inhibitors or activators(2). The identified small molecules that perturb specific protein functions are valuable tools for dissecting complex biological processes in cells(19-23). Consequently, these inhibitors or activators also have the potential to be developed into drugs for treatment of human diseases. While progress has been considerably made in fields such as synthesizing high-quality small molecule libraries, increasing library size, developing screening instrumentation and miniaturizing screening formats(24), development of robust, specific and cost-effective screening methods has become one of the crucial factors to the success of HTS(25, 26).
Traditionally, the screen can be designed as either a phenotype-based screening or a target-based screening(20). Given that target identification in phenotype-based screening is a time-consuming step, target-based screening may offer a shortcut to the final success, particularly in drug discovery. This is especially true in the post-genome era since the mechanisms underlying many human diseases are better understood at the molecular level. Unfortunately, the available methods used for screening of small molecule inhibitors that target protein-protein interactions are largely limited to in vitro assays such as fluorescence polarization-based assay, fluorescence resonance energy transfer, or artificial systems such as yeast two-hybrid based screening. Several limitations likely contribute to the low success rate of these screenings. First, these assays use purified proteins to perform screening in vitro or in yeast or bacteria. Thus the screening environment is not physiologically relevant. Given that protein-protein interactions are often tightly regulated by complex mechanisms such as posttranslational modifications, identified small molecule inhibitors may be effective in vitro, but not effective in vivo. Second, most of these assays are designed to screen for a pair of interaction and many candidates from primary screening target reporters or probes rather than the interaction of two proteins. Hence, a large number of false positive candidates will have to be eliminated by secondary or tertiary screenings. Third, some compounds may exhibit general cellular toxicity or general inhibition of the system when an in vivo screening system such as yeast two-hybrid is used. Therefore, a better method for screening of small molecule inhibitors of protein-protein interactions under normal cellular context is in urgent need.
Principle of BiFC
Taking advantage of the intrinsic fluorescence property of green fluorescent protein (GFP) and its variants yellow fluorescent protein (YFP) and cyan fluorescent protein (CFP), a novel bimolecular fluorescence complementation (BiFC) assay and a multicolor BiFC assay have been developed (27, 28). These assays have been successfully used for the study of protein interactions in vitro, and direct visualization of protein interactions in living cells (27-34). The principle of the BiFC assay is to split a fluorescent protein into an N-terminal and C-terminal fragments and fuse them to a pair of potential interacting proteins (Fig. 1A). Once the two proteins interact with each other, their interaction brings two fragments in proximity and reconstitutes an intact fluorescent protein, allowing direct visualization of fluorescence (BiFC signal). Hence, the observed BiFC signal reflects potential interactions between two proteins. The positions that can be used for splitting fluorescent proteins depend on individual fluorescent proteins. While the positions between residues 154 and 155 and between residues 172 and 173 can be used to split YFP, CFP, BFP, Cerulean, and Venus for fluorescence complementation(28, 35, 36), the position between residues 157 and 158 works for GFP (37). Interestingly, the positions between residues 154 and 155 and between residues 168 and 169 works for a monomeric red fluorescent protein (mRFP1), which has 225 residues, but the corresponding positions do not work for CFP, BFP, GFP and YFP(38).
Shyu et al. Figure 1
Fig 1. A. Proteins A and B are fused to N-terminal (VN) and C-terminal (VC) fragments of Venus. The interaction between proteins A and B brings VN and VC together and reconstitutes an intact fluorescent protein Venus.
Fig 1. B. Non-fluorescent N-terminal fragments of Venus (VN) and Cerulean (CrN) and C-terminal fragment of CFP (CC) are fused to proteins A, B and C, respectively. The interaction between proteins A and C brings VN and CC together and reconstitutes a fluorescent protein VN-CC (Venus-like), and the interaction between proteins B and C reconstitutes a fluorescent protein CrN-CC (Cerulean-like). These two different bimolecular fluorescent complexes can be spectrally resolved using filters optimized for both YFP and CFP. The molecular models were created using PyMol (Delano, W.L. The PyMol Molecular Graphics System, 2002, www.pymol.sourceforge.net/ ) with the structures of GFP from the Protein Data Bank (accession code 2G5Z).
Because protein folding and chromophore formation of YFP and CFP are sensitive to high temperatures(39), visualization of YFP- and CFP-based BiFC signals require a pre-incubation of transfected cells at 30oC for a few hours to overnight prior to detection of fluorescent signals(40). This extra stress to cells apparently limits their use for visualization of protein interactions under physiological conditions. To overcome this problem, we have recently demonstrated that Venus, Citrine and Cerulean, three mutant proteins of YFP and CFP with improved properties(41-43), can be used for BiFC and multicolor BiFC analyses without a preincubation at low temperatures(36). Furthermore, the use of these improved fluorescent proteins has significantly shortened the incubation time and increased the specificity compared to YFP-based BiFC analysis.
Applications of BiFC
Over the past few years, the BiFC assay has become a widely used assay for visualization and identification of protein-protein interactions in living cells. This has been extensively reviewed recently by Kerppola(34), and will not be discussed here in details. The following are summary of the current trend for the use of BiFC and BiFC-based technologies.
1. Multiple Protein Interactions: Although identification of individual interacting proteins are important, each protein interacts with multiple partners and sometimes multiple partners compete for the binding to the same protein. To meet this need, a multicolor BiFC assay was developed by using different combinations of fluorescent proteins (Fig. 1B)(28). Using this assay, it was demonstrated that Fos-Jun heterodimer formation may be favored to Jun-ATF2 heterodimer formation. The multicolor BiFC assay was also used to study the dynamics of Myc/Max/Mad family of proteins and competitions between G protein βγ dimer formation(44, 45). Given the fact that more fluorescent proteins are being identified and more BiFC-competent fluorescent proteins such as mRFP1 for BiFC analysis become available(38, 46), we anticipate that multicolor BiFC assay will be widely used for visualization of multiple protein complexes.
2. Multiple Families of Proteins: The initial development of BiFC used antiparallel leucine zipper or basic region leucine zipper proteins(35, 37), now the BiFC assay has been successfully used for many different families of proteins(34, 47). They range from nuclear-localized transcription factors to cytosolic proteins and membrane-bound receptors. The size also ranges from small peptides such as ubiquitin to large proteins. Detailed introduction of these applications can be found from two recent reviews(34, 47).
3. Development of BiFC-Based Technologies: While BiFC is mainly used for visualization or identification of protein interactions, the BiFC principle has recently been explored to study ubiquitination(48), RNA-protein interaction(49), DNA-DNA hybridization(50), cDNA library screening(51, 52), conformational change(53), topology of proteins(54), and visualization of RNA in single living cells(55, 56). We have also recently developed a BiFC-based fluorescence resonance energy transfer (BiFC-FRET) system for visualization and identification of ternary complexes in living cells(57). There is no doubt that more applications will be explored in the near future.
4. Multiple Model Systems: In addition to the wide use of BiFC assay for visualization and identification of protein interactions in living mammalian cells, it has now been successfully applied to the study of protein interactions in in vitro, E. coli, plant, yeast, fungi and C. elegans(34, 47). To facilitate BiFC analysis in different model systems, BiFC cloning vectors have been constructed for BiFC analysis in mammalian system(35, 36), plant(58), yeast(59), and C. elegans(60)