The Analysis of Protein-Protein Interactions in Plants by Bimolecular Fluorescence Complementation

Following the complete genome sequencing of different plant species such as Arabidopsis (Arabidopsis thaliana), rice (Oryza sativa), and Physcomitrella (Physcomitrella patens), as well as advances toward deciphering entire proteomes, the need for a reliable way to identify protein-protein interactions is becoming a major task for the future. Bimolecular fluorescent complementation (BiFC) is a noninvasive fluorescent-based technique that allows detection of protein-protein interactions in living cells, and furthermore can be used to determine subcellular localization of the interacting proteins, and if it changes over time, without requiring addition of external agents. BiFC is based upon reconstitution of split nonfluorescent GFP variants, primarily yellow fluorescent protein (YFP), to form a fluorescent fluorophore (Ghosh et al., 2000; Hu et al., 2002). The technique has become increasingly popular due to its simplicity, ease of use, and the capability to carry out experiments with regular epifluorescence or confocal laser scanning microscopes (CLSMs). In this Update, we first discuss the principles of BiFC and its major advantages and disadvantages. We then describe the adaptation of BiFC to plant systems, provide practical suggestions for its use, and review protein-protein interactions that have been identified and confirmed in plants using this technique. Finally, additional potential exploitations of BiFC are discussed.

Due to lack of space we did not discuss other fluorescent-based techniques for detection of protein-protein interactions, such as fluorescent resonance energy transfer, and refer the readers to a recent review on fluorescent resonance energy transfer and BiFC (Bhat et al., 2006). Discussion of additional protein fragment complementation assays techniques can be found in a recent review (Remy and Michnick, 2007). We also apologize to those colleagues whose work we have not cited due to lack of space.

THE PRINCIPLES AND DEVELOPMENT OF BiFC

The BiFC Principle

BiFC is based upon tethering split YFP or other GFP variants to form a functional fluorophore. The association of the split YFP/GFP/cyan fluorescent protein (CFP) molecule does not occur spontaneously and requires interaction between proteins or peptides that are fused to each of the fluorophore fragments (Fig. 1). Upon interaction of these fused proteins/peptides, the split fluorophore fragments can interact to form a fluorescent protein that has the same spectral properties as the unsplit YFP (or other GFP variants; Figs. 1 and 2). If the proteins that are fused to the split fluorophore fragments do not interact, reconstitution of the YFP/GFP/CFP usually does not take place and no fluorescence is detected.

Figure 1.

BiFC can be used to determine subcellular localization of protein complexes. A, Under physiological conditions reconstitution of a fluorescent YFP molecule can only take place following interaction between proteins or peptides that are fused to YN and YC fragments. B, A BiFC assay showing that complexes of AtROP6 and the ROP-interacting coiled-coil scaffold protein ICR1 are localized in the plasma membrane of N. benthamiana leaf epidermal cells. C, A BiFC assay showing that the complexes between the Atrop6mS mutant, in which the prenyl-acceptor Cys was changed to Ser, and the ROP-interacting coiled-coil scaffold protein ICR1 is localized in the cytoplasm and nuclei. D, Possible configurations for testing BiFC. YN, The N-terminal fragment of YFP; YC, the C-terminal fragment of YFP. Yellow-shaded squares symbolize potential fluorescence resulting from protein-protein interaction leading to YFP reconstitution.

Figure 2.

Fluorescence emission spectra of intact, BiFC reconstituted YFP and autofluorescence of plant tissues. Samples of YFP (yellow line) and ROP-ICR1 (green line) were excited at 488 nm and the fluorescent spectra were determined using the spectral scanning module of a Leica TCS-SL CLSM. Fluorescence intensities are shown in arbitrary units relative to the maximal fluorescence, which received a value of 1. Autofluo, Autofluorescence (red line).

THE DEVELOPMENT OF THE BiFC ASSAY

Mutational studies have uncovered sites within the GFP molecule that allow insertions without perturbing fluorescence characteristics (Abedi et al., 1998; Baird et al., 1999). These findings led to the discovery that split GFP fragments could be reconstituted in vitro and in Escherichia coli when fused to interacting antiparallel coiled-coil peptides (Ghosh et al., 2000). The same study showed that each GFP fragment by itself is insoluble when expressed in E. coli; however, the reconstituted GFP protein complex is soluble and stable, with a low dissociation coefficient (Ghosh et al., 2000). Insertion of a peptide spacer within GFP to which a calmodulin (CaM) and a CaM target peptide M13 were fused at its N- and C-terminal ends, respectively, resulted in a chimeric protein, designated Pericam. Reconstitution of GFP fluorescence in Pericam takes place when Ca²⁺ binds the CaM moiety that in turn interacts with the M13 peptide. Pericam was used to monitor dynamic changes in Ca²⁺ levels in different cellular compartments of HeLa and cardiomyocytes (Nagai et al., 2001; Robert et al., 2001). It was further shown that when the two fragments of GFP were completely separated by deleting the spacer peptide between them, they could interact to reconstitute a fluorescing GFP in a Ca²⁺-dependent fashion (Nagai et al., 2001). These studies of split GFP in E. coli and Pericam in HeLa and cardiomyocytes prompted the development a split YFP system, designated BiFC, which was used for determining subcellular localization of protein complexes in mammalian cells (Hu et al., 2002). The work of Hu et al. (2002) demonstrated that BiFC is a simple, reliable, and efficient tool to examine protein-protein interactions in living cells, allowing determination of their subcellular localization. Following this work, BiFC became a widely used method. In the BiFC systems that were subsequently developed, including versions adapted for plants (Bracha-Drori et al., 2004; Walter et al., 2004), the YFP molecule was split between amino acids 154 and 155, based on the design of Hu et al. (2002) or between residues 174 and 175 (Citovsky et al., 2006).

The dynamics of the split YFP reconstitution has been investigated to elucidate the pathway for fluorescent molecule formation (Hu et al., 2002). Formation of a complex between split YFP fragments takes place with a t_1/2 of about 60 s. The complex formation between proteins fused to the split YFP fragments, however, occurs with t_1/2 shorter than 1 s. Subsequent maturation of the complex to form a florescent YFP molecule takes additional 3,000 s (50 min; Hu et al., 2002; Kerppola, 2006b). The relatively slow maturation time of the reconstituted YFP and other GFP variants is a major disadvantage of the original BiFC system, as will be discussed below. To date, to our knowledge, no studies on the dynamics of BiFC in plants have been published and it is not known whether the time for YFP complex formation in plant and mammalian cells are similar.

BASIC DESIGN OF BiFC VECTORS

Proteins under study can be expressed as either N-terminal or C-terminal fusions with the split YFP fragments, often referred to as YN and YC, respectively (Hu et al., 2002; Bracha-Drori et al., 2004; Citovsky et al., 2006; Kerppola, 2006a). The resulting combinations of interacting pairs of proteins are depicted in Figure 1D. Using different combinations of YN and YC fusion pairs (Fig. 1D) is advisable since the orientation of the fusion can greatly affect YFP complex formation (Bracha-Drori et al., 2004). It is also recommended to place a flexible spacer between the split YFP fragment and the proteins under investigation, to alleviate structural constraints that might compromise YFP complex formation (Hu et al., 2002; Kerppola, 2006a). Currently, however, there are no accurate predictions for designing ideal spacers. Furthermore, the structure of the proteins under study and the spacers used can greatly affect YFP complex formation and fluorescence emitted. Mature YFP complex formation could be detected even when the interaction surface between two given proteins is small and the interactions are weak or transient (Hu et al., 2002; Kerppola, 2006a, 2006b; Lavy et al., 2007; Morell et al., 2007). Thus, use of strong promoters should be avoided.

MULTICOLOR BiFC

Protein complementation assays (PCAs) can take place between split GFP (Ghosh et al., 2000), YFP (Hu et al., 2002), and CFP (Hu and Kerppola, 2003) protein pairs, which have different excitation-emission spectra. In an expansion of their original BiFC system, Hu et al. (2002) have introduced multicolor BiFC, demonstrating that PCAs can be performed between the complementing fragments of YFP and CFP as well as between YFP and GFP fragments. The resulting GFP-YFP and CFP-YFP complexes have unique, resolvable emission spectra (Hu and Kerppola, 2003). Multicolor BiFC assays are useful when complex formation of a given protein with different interacting partners is examined. Careful analysis of fluorescence intensities has enabled prediction of preferable interactions (Hu and Kerppola, 2003). Multicolor BiFC assays in plants have not yet been described.

EQUIPMENT REQUIRED FOR BiFC EXPERIMENTS

A foremost advantage of the BiFC assay is its simplicity and the ability to carry out experiments with either a regular epifluorescence microscope equipped with the relevant filter sets and a CCD camera, or with a CLSM. Background fluorescence does not usually constitute a problem because the signals are strong enough, especially when using split YFP. In cases of weak fluorescence, however, appropriate filter sets, such as the Ziess Pinkel Set 40, can be used to resolve true YFP fluorescence from autofluorescence (for more detail see Bracha-Drori et al., 2004). Certain CLSMs have spectrum scan capabilities that enable determination of fluorescence spectra (Fig. 2; Bracha-Drori et al., 2004). Furthermore, a fluorescent spectrophotometer can be used to determine whether the relevant spectra are emitted by the sample. Finally, examination of fluorescence resulting from nonlegitimate interaction between a candidate protein fused to either YN or YC with the reciprocal YN or YC only, or preferably with a mutated noninteracting protein, should be determined.

ADVANTAGES AND PITFALLS OF BiFC

BiFC has several major advantages. (1) The assay is simple and does not require sophisticated dedicated equipment. (2) There is either no or low background signal because a fluorescing YFP would only form after interaction between proteins fused to split fragments. (3) BiFC enables determination of the subcellular localization of interacting protein complexes as well as the mutual affect of interacting partners on the subcellular localization of the complex. (4) BiFC is a sensitive assay, enabling detection of weak and transient interactions, primarily due to the stability of the reconstituted YFP complexes (Hu et al., 2002).

However, the assay suffers from several pitfalls that must be taken into account. (1) The slow maturation time of the reconstituted GFP/YFP/CFP compromises detection of dynamic changes in protein-protein interactions in real time (Ghosh et al., 2000; Hu et al., 2002; Kerppola, 2006a, 2006b). However, this problem can be alleviated by using the Venus variant of YFP, which matures within few seconds (Miyawaki et al., 2003, 2005). BiFC and multicolor BiFC experiments with fragments derived from Venus and Cerulean, a modified variant of CFP, have been successful (Shyu et al., 2006). (2) The stability of the reconstituted YFP complexes (Hu et al., 2002) hampers the ability to analyze the dynamics of protein-protein dissociation. This can lead to detection of nonspecific interactions when expression levels of the split YFP fragments are high. (3) The molecular properties of chimeric fusion proteins could be different from that of the native proteins.

These disadvantages of the BiFC system require careful consideration of the following issues. Although BiFC-based systems have been used successfully for monitoring dynamic changes in Ca²⁺ concentration (Nagai et al., 2001; Robert et al., 2001), the stability of the reconstituted YFP/GFP/CFP complexes and the slow maturation times of fluorescent dyes other than Venus may compromise the results. Most BiFC assays in plants are, however, carried out using transient expression systems in which transformed tissues are typically analyzed after several or even 24 to 48 h. This allows enough time for protein maturation. To alleviate the problem of nonspecific interactions, expression levels should be kept low. Using negative controls in the form of noninteracting point mutants of tested proteins is essential. To ascertain that lack of fluorescence is not due to low expression, it is crucial to monitor expression levels of the relevant proteins with antibodies (Fig. 3; Bracha-Drori et al., 2004; Walter et al., 2004). Thus, quantification of the fluorescent signal could become very useful, provided it is normalized to the protein expression levels. The validity of the BiFC results should be verified by determining expression patterns, colocalization assays, and if applicable, genetic analysis and determination of protein-protein interactions in plants by an independent method. The bottom line is that as with any other experimental systems, using the right controls and proper calibration are essential.

Figure 3.

Protein immunoblots reveal expression of split YFP fragments in the BiFC assay. Protein immunoblots decorated with anti-hemagglutinin (HA) tag (left section) and anti-GFP (right section) monoclonal Abs were used for detection of YC and YN AP1 fusion proteins, respectively. The weaker bands observed with the anti-GFP Abs were due to the lower specificity of these Abs toward the YN moiety. Comparable expression levels of the YN and YC AP1 fusion proteins were detected in blots decorated with anti-AP1 Abs (K. Shichrur and S. Yalovsky, unpublished data).

WORKING WITH BiFC IN PLANTS

The Application of BiFC in Plants

Successful application of BiFC in plants was first described in three publications (Bracha-Drori et al., 2004; Tzfira et al., 2004; Walter et al., 2004). The data from all three articles demonstrated that BiFC can be used to detect expression of different proteins in different subcellular compartments following transient expression in onion (Allium cepa) epidermis or tobacco (Nicotiana tabacum) leaves (Tzfira et al., 2004), infiltration of Agrobacterium cells into leaves of Nicotiana benthamiana and Arabidopsis (Bracha-Drori et al., 2004; Walter et al., 2004), or protoplast transformation (Walter et al., 2004). It was demonstrated that fluorescence resulting from a specific interaction was much stronger than the fluorescence from nonspecific interactions at comparable protein expression levels (Bracha-Drori et al., 2004; Walter et al., 2004). The reconstituted YFP fluorescence was distinguished from autofluorescence by using either an epifluorescence microscope equipped with the appropriate filter sets or by using the spectral scanning capability of the CLSM (Fig. 2; for details see Bracha-Drori et al., 2004). Finally, complex formation was also verified by immunoprecipitation using monoclonal antibodies directed against peptides that were engineered into the expression cassettes (Bracha-Drori et al., 2004). Interestingly, the reconstituted YFP fluorescence resulting from the interaction between the α- and β-subunits of the enzyme farnesyltransferase had different intensities depending on the orientation of the fusion of each of the subunits to the YFP fragments (Bracha-Drori et al., 2004). This finding highlighted the importance of examining different fusion orientations when testing protein-protein interactions by BiFC. Subsequently, several additional BiFC systems for plants were described (Uhrig et al., 2007) and the usefulness of the BiFC assay to determine subcellular localization of protein complexes was expanded and highlighted (Citovsky et al., 2006). Recently, a split monomeric red fluorescent protein system for plants was described (Jach et al., 2006).

UTILIZATION OF BiFC TO DETERMINE PROTEIN-PROTEIN INTERACTIONS IN PLANTS AND THEIR SUBCELLULAR LOCALIZATIONS

Below we describe several studies in which results obtained with BiFC have led to new insights and understandings of biological processes. Table I presents a list of protein-protein interactions that have been verified or identified by BiFC in plants.

Table I.

Case studies of protein-protein interaction in plants using BiFC

a, b, c, d, and e denote the interacting protein pairs and their subcellular localization in cases that two or more interactions were presented in the same study. w, x, y, and z denote transformation methods. w, Coexpression in cell culture; x, agroinfiltration into leaf epidermal cells; y, particle cobombardment into leaf epidermal cells; z, plasmid cotransformation into leaf protoplasts.

Interactor 1	Interactor 2	Source Plant	Subcellular Localization	Experimental System	Reference
Signaling
Farnesyltransferase α-subunit (PFTA)	Farnesyltransferase β-subunit	Arabidopsis	Cytoplasm	N. benthamianax; Arabidopsisx	Bracha-Drori et al. (2004)
14-3-3 proteins phospho-Ser/phospho- Thr-binding modules	14-3-3 proteins phospho-Ser/phospho- Thr-binding modules	Arabidopsis	Cytoplasm and nucleus	N. benthamianax	Walter et al. (2004)
EID1 (empfindlicher im dunkelroten Licht, hypersensitivein far-red light)	Arabidopsis S-phase kinase-related protein1, F-box protein	Arabidopsis	Nucleus	Etiolated mustard (Sinapis alba) seedlingsy	Stolpe et al. (2005)
Tandem-pore Kþ channels (AtTPK1, AtTPK5, and AtKCO3)	Tandem-pore Kþ channels (AtTPK1, AtTPK5, and AtKCO3)	Arabidopsis	Vacuolar membrane	Arabidopsisz and tobaccoz BY2z	Voelker et al. (2006)
Cell death suppressor BAX INHIBITOR1	CaM7	Arabidopsis	Cytoplasm	N. benthamianax	Ihara-Ohori et al. (2007)
GID1	SLR1	Arabidopsis	Nucleus	N. benthamianax	Ueguchi-Tanaka et al. (2007)
Transcription factor WRKY51	Transcription factor WRKY71	Rice	Nuclei of aleurone cells	Barleyy	Xie et al. (2006)
Prolamin-box binding factor (WPBF), a DOF transcription factor	TaQM	Wheat (Triticum aestivum)	Nucleus	Oniony	Dong et al. (2007)
Cell polarity
Members of the actin- nucleating ARP2/3 complex and subunits of the SCAR/WAVE complex	Members of the actin-nucleating ARP2-ARP3 complex and Rac-like GTPases	Arabidopsis	Plasma	Oniony	Uhrig et al. (2007)
AtSCAR2a			Membranea
SPIKE1b	AtROP7a
HSPC300c	AtSCAR2b		Cytoplasmb,c
	BRICK1c
AtROP6	ICR1	Arabidopsis	Plasma membrane	N. benthamianax	Lavy et al. (2007)
AtROP9
AtROP10
ICR1	AtSEC3 exocyst subunit	Arabidopsis	Cytoplasm and nuclei	N. benthamianax	Lavy et al. (2007)
Chloroplast
Component of the import apparatus to the chloroplasts inner envelope (AtTic110)	Component of the import apparatus to the chloroplasts inner envelope (AtTic110)	Arabidopsis	Chloroplast envelope	Arabidopsisz	Bedard et al. (2007)
Transcriptional regulation
Basic Leu zipper (bZIP) transcription factor (bZip63)	bZIP transcription factor (bZip63)	Arabidopsis	Nucleus	N. benthamianax; Arabidopsisz	Walter et al. (2004)
FLOWERING LOCUS T	bZIP transcription factor (FD)	Arabidopsis	Nucleus	N. benthamianax	Abe et al. (2005)
DNA WITH ONE ZINC FINGER (DOF)	GAMYB	Barley (Hordeum vulgare)	Nucleus	Oniony	Diaz et al. (2005)
bZIP transcription factor AtbZIP53	bZIP transcription factor AtbZIP10	Arabidopsis	Nucleus	N. benthamianax	Weltmeier et al. (2006)
MYB transcription factor CPC	BLH transcription factor GL3	Arabidopsis	Nucleus	Oniony	Jach et al. (2006)
STM	BLH	Arabidopsis	Nucleus and cytoplasm	Leek (Allium porrum)y	Cole et al. (2006)
The Dornrosch (also known as ENHANCER OF SHOOT REGENERATION1)	Arabidopsis class III HD-ZIP gene family PHAVOLUTA	Arabidopsis	Nucleus	Leeky	Chandler et al. (2007)
BLH1 TALE homeodomain protein	AtOFP1 ovate family protein	Arabidopsis	Nucleus and cytoplasm	N. benthamianax	Hackbusch et al. (2005)
Disease resistance and viral plant interaction
Bacterial virulence protein VirF	Virulence protein VIP1	Agrobacterium tumefaciens	Nucleus	Tobaccoz; oniony	Tzfira et al. (2004)
Virulence protein VIRE2
Nuclear import-mediator importin αa	VirD2a	Arabidopsis and cucumber (Cucumis sativus)	Nucleusa,b,d	BY-2w, Arabidopsisz, tomato (Solanum lycopersicum)z, tobaccoz	Citovsky et al. (2006)
VIP1b	VIP1b		Chloroplastsc,d	N. benthamianax; oniony
CUCUMBER CHROMOPLAST D (CRD) proteinc	CRDc		Plasmodesmae
Yellow leaf curl virus (TYLCV) capsid protein (CP)d	TYLCV CPd
Tobacco mosaic virus (TMV) movement proteine	Arabidopsis calreticulin CRTe
Arabidopsis homolog of kinase interacting sequence domain specific to the β-subunits (AKINbg)	AKINb subunits	Arabidopsis	Nucleus and cytoplasm	N. benthamianax	Gissot et al. (2006)
NPR3 paralog of nonexpressor of PR GENE1	Transcription factor TGA	Arabidopsis	Nucleus	Oniony	Zhang et al. (2006)
Tomato bushy stunt virus P19 protein	RNA-binding proteins (ALY)	Arabidopsis	Nucleus	N. benthamianax	Canto et al. (2006)
The coat protein of Prunus necrotic ringspot virus	The coat protein of Prunus necrotic ringspot virus	Prunus necrotic ringspot virus	Unclear	N. benthamianax	Aparicio et al. (2006)
TMV coat protein	TMV movement protein	TMV	Cytoplasm	N. benthamianax	Bazzini et al. (2007)
Viral protein RNA-binding protein	Translation eukaryotic initiation factor iso 4E [eIF(iso)4E]	Turnip mosaic virus and Arabidopsis	Nucleus	N. benthamianax	Beauchemin et al. (2007)
Chromatin structure
FIE	MEA	Arabidopsis	Nucleus and cytoplasm	N. benthamianax	Bracha-Drori et al. (2004)
Methyl CpG-binding domain AtMBD5	Ran GTPase protein AtRAN3	Arabidopsis	Nucleus	BY2 cellsw	Yano et al. (2006)
miR
DCL1 dsRNA-binding domain protein, HYL1	Zinc-finger domain protein SE	Arabidopsis	Nuclear D bodies	N. benthamianax	Fang and Spector (2007)
Protein topology
Viral membrane-associated movement proteins		PMTV genus Pomovirusa	Endoplasmic reticulum membrane	N. benthamianax	Zamyatnin et al. (2006)
TGBp2 of Potato mop top virusa	TGBp2 of Potato mop top virusa	BYV genus Closterovirusb
(p6) of Beet yellows virusb	(p6) of Beet yellow virusb

BiFC proved to be useful for determining the mutual effect of interacting proteins on their subcellular localization (Fig. 4A). The ROP GTPase-interacting coiled-coil scaffold protein INTERACTOR OF CONSTITUTIVE ACTIVE ROP1 (ICR1) makes cytoplasmic-localized homooligomers but it is recruited by ROPs to the plasma membrane. In turn, ICR1 interacts and recruits specific sets of proteins, including SEC3 exocyst subunit to the plasma membrane, regulating cell polarity. The BiFC experiments demonstrated how posttranslational lipid modifications of the ROPs determine the subcellular localization of the complex (Lavy et al., 2007; Fig. 1). In the same study, the colocalization of CFP-ROP9 together with a reconstituted YFP complex formed by the interaction between ICR1 and SEC3 showed that the BiFC assay could be used to demonstrate colocalization between three proteins using only two fluorescent dyes: CFP and the reconstituted YFP (Lavy et al., 2007). BiFC was used to demonstrate that ALY proteins, which interact and inhibit translocation of the tomato bushy stunt virus P19 protein from the nucleus to cytoplasm, compromise its antisilencing activity (Canto et al., 2006). SHOOTMERISTEMLESS (STM) is a transcriptional regulator that functions in maintenance of the shoot apical meristem and is localized by default to the cytoplasm. BiFC assays revealed that nuclear localization of STM depends on its heterodimerization with BELL1-LIKE HOMEODOMAIN (BLH) proteins (Cole et al., 2006).

Figure 4.

Different uses of BiFC. A, Interactions can occur only if two proteins are localized in the same compartment. Thus, how one protein affects the subcellular localization of the complex can be tested, as exemplified in Figure 1. The stability of the bimolecular complexes formed in BiFC should enable tracking of subcellular targeting pathways of proteins that may otherwise be difficult to observe. B, Protein interaction may depend on the presence of a cofactor or another protein(s) (Ueguchi-Tanaka et al., 2007). Screening cDNA vector libraries or chemical libraries can be used to identify protein and nonprotein mediators. Expression in relevant mutant backgrounds and sequestration or removal of cofactors could be used as controls. C, Some protein-protein interactions depend on posttranslational modifications. For example, a ubiquitin-based BiFC system has been described and proteins showed different localizations upon interaction with ubiquitin or the related Sumo protein (Fang and Kerppola, 2004). The use of mutated interacting proteins or mutated genetic backgrounds can aid in analyzing requirements for posttranslational modifications. In addition, site-directed mutagenesis could be used to remove the modification site and to map the domain mediating the interaction. D, Protein interactions depend on expression times and patterns. Using endogenous promoters to express a particular BiFC enables analysis of the temporal and spatial interaction pattern, as has been demonstrated in C. elegans (Zhang et al., 2004). Coexpression patterns of genes could be studied by expressing known interacting protein partners under respective promoters. In this case, the YFP fluorescence should only be detected in the cells where the two promoters are active.

Determination of subcellular localization of protein complexes by BiFC has opened new avenues toward understanding basic cellular processes in plants. The function of microRNAs (miR) in gene expression regulation has been at the forefront of research in plant and nonplant systems ever since they were first discovered. Much is still not known about the mechanisms of miR formation and processing. In a recent study, BiFC was used to demonstrate that DICER-LIKE1 (DCL1), HYPONASTIC LEAVES1 (HYL1), and SERRATE (SE), three proteins involved in miR processing or storage, are assembled and localized in nuclear dicing bodies (Fang and Spector, 2007). Plant scientists have used Agrobacterium tumefaciens for three decades to transform plants. Yet, many steps of T-DNA-mediated transformation are not well understood. BiFC was used to show that the bacterial F-box-containing protein VirF interacts with the plant protein VIP1 in the nucleus. This interaction likely leads to degradation of VIP1 and another bacterial proteinVirE2 by the proteasome (Tzfira et al., 2004). Extensive BiFC and yeast (Saccharomyces cerevisiae) two-hybrid assays were used to map the interactions between the Arp2/3 complex and its regulating SCAR/WAVE complex subunits, between the SCAR2 subunit and several ROPs, and between SCAR/WAVE subunits and the SPIKE1 ROPGEF (Uhrig et al., 2007). This study revealed internal interactions between subunits of each Arp2/3 and SCAR/WAVE complex as well as interactions taking place between the complexes. That the SCAR2 subunit of SCAR/WAVE interacted with several ROPs suggested that it is a ROP effector, and that SPIKE1 interacted with several SCAR/WAVE subunits indicated that it could be part of this complex. A new study exemplifies the usefulness of BiFC for determination of protein-protein interactions that depend on a third component (Fig. 4B). BiFC and yeast two-hybrid assays showed that the interaction between the soluble GA receptor GA-INSENSITIVE DWARF1 (GID1) and the DELLA repressor protein SLENDER RICE1 (SLR1) requires GA (Ueguchi-Tanaka et al., 2007).

Localization of protein complexes with BiFC has yielded surprises and new insights. Complexes of the polycomb (PcG) chromatin regulating complex proteins FERTILIZATION INDEPENDENT ENDOSPERM (FIE) and the SET domain MEDEA (MEA) were detected in both the nuclei and cytoplasm (Bracha-Drori et al., 2004), suggesting that some PcG complexes may function also outside the nucleus.

The topology of two membrane proteins were tested with BiFC. The amino (YN) and carboxy (YC) fragments of YFP were cloned at different positions along the P6 movement protein of Beet yellow virus and the TGBp2 protein of Potato mop top virus (Zamyatnin et al., 2006). Unfortunately, in this study the YN and YC were not fused to interacting protein partners and therefore the measured fluorescence may have resulted from nonspecific interactions.

POSSIBLE PITFALLS OF BiFC ASSAYS IN PLANTS

BiFC assays have failed to confirm data obtained using in vitro or yeast two-hybrid assays. For example, homodimerization of APETALA1 (AP1), and heterodimerization of AP3 and PISTILLATA (PI) MADS-box floral regulators and their Antirrhinum homologs had previously been demonstrated by in vitro and yeast two-hybrid assays (Egea-Cortines et al., 1999; Honma and Goto, 2001). We used BiFC to examine AP1 homodimerization and heterodimerization of AP3 and PI. Surprisingly, no BiFC signals were detected with different combinations of split YFP pairs, even though protein expression was verified by immunoblotting (Fig. 3). Furthermore, efforts to immunoprecipitate YN-AP1 YC-AP1 homodimers were unsuccessful (K. Shichrur and S. Yalovsky, unpublished data). These negative results do not prove that homodimerization of AP1 and heterodimerization of AP3 and PI did not take place. Rather, the structure of the AP1 and AP3-PI homo- and heterodimers may have prevented successful reconstitution of YFP. Alternatively, the interaction between these MADS-box proteins in plant cells may require components that were not present in leaf epidermal cells, the tissue in which the BiFC assays were carried out, or certain components present in these cells may have interfered with the complex formation. However, the data do show that ectopic expression of split YFP fragments that accumulated in the same subcellular compartment (the nucleus in the case of AP1, AP3, and PI) does not necessarily lead to reconstitution of an active fluorophore, as was previously claimed (Zamyatnin et al., 2006).

FUTURE PROSPECTS

The use of the BiFC assays could be expanded in several directions. These include determination of protein-protein interactions in time and space by using endogenous promoters of genes of interest. In addition, high-throughput screens for interacting proteins in plant cells and determination of gene expression patterns could be facilitated using BiFC.

Using endogenous promoters for gene expression in BiFC could solve several problems, as specified below. Avoiding overexpression should alleviate nonspecific interactions and will enable testing interaction at physiologically relevant conditions. Furthermore, in some cases interaction between two proteins is indirect, requiring additional factor(s) that may only exist in certain tissues or cells (Fig. 4B). Some interactions may only occur following certain stimuli, whose effects could be obscured by overexpression. Under physiologically more relevant conditions, quantification of fluorescence signals should be more readily achievable and relevant. The use of specific promoters would be invaluable for developmental studies (Fig. 4D). It is of concern that the low expression levels often observed with native promoters could potentially compromise detection of YFP signal above background fluorescence. However, the proven sensitivity of the BiFC assay together with the newer and more sensitive YFP variant, Venus, previously adapted for BiFC (Shyu et al., 2006), should make the use of specific and weaker promoters feasible. Furthermore, as discussed above, YFP and the background fluorescence emitted from plant tissues have different wavelength spectra that can be resolved (Fig. 2; see Bracha-Drori et al., 2004 for more details). In Caenorhabditis elegans labeling of specific cells was achieved by using specific promoters to drive expression of split YFP fusion proteins (Zhang et al., 2004). Similarly, in plants BiFC could be used to detect at the cellular level the expression pattern of two genes (Fig. 4D).

The simplicity and sensitivity of BiFC makes it an attractive system for high-throughput protein-protein interaction screens in plants. A split GFP system has been used for screening interacting partners of the protein kinase PKB/Akt in COS cells (Remy and Michnick, 2004). The following problems should be evaluated before initiating such screens in plant cells.

Background fluorescence should be quantified to select between specific and nonspecific interacting partners. What is the preferred experimental system? Should it be, for example, tissue culture cells, pollen, seedlings, or mature plants? Unlike mammalian cells, plant cells do not adhere to the bottom of culture plates, making it much more difficult to screen efficiently for individual cells/colonies. This problem may be alleviated by using FACS if the fluorescence is strong enough. It would be necessary to devise an efficient and reliable screening method including high transformation efficiency, reduce transgene silencing, and develop reliable methods for plasmid recovery. Finally, BiFC should be compared to other PCA systems.

A successful application of a split luciferase in plants has been described (Fujikawa and Kato, 2007). Large-scale screens for rare recombination events in Arabidopsis have been carried out using luciferase as a reporter (Jelesko et al., 1999) in conjunction with a sensitive single photon counting CCD camera. Thus, a luciferase-based PCA system might be more adaptable for high-throughput screens.

The popularity and usage of BiFC in plants systems is rapidly expanding and undoubtedly, new protein-protein interactions will be revealed. The quality and reliability of these data will depend on carrying out all the necessary controls, reducing the expression levels, and primarily using BiFC where its greatest capability lies—for the detection of subcellular localization.