Visualization of protein interactions in living Drosophila embryos by the bimolecular fluorescence complementation assay
© Hudry et al; licensee BioMed Central Ltd. 2011
Received: 23 November 2010
Accepted: 28 January 2011
Published: 28 January 2011
Protein interactions control the regulatory networks underlying developmental processes. The understanding of developmental complexity will, therefore, require the characterization of protein interactions within their proper environment. The bimolecular fluorescence complementation (BiFC) technology offers this possibility as it enables the direct visualization of protein interactions in living cells. However, its potential has rarely been applied in embryos of animal model organisms and was only performed under transient protein expression levels.
Using a Hox protein partnership as a test case, we investigated the suitability of BiFC for the study of protein interactions in the living Drosophila embryo. Importantly, all BiFC parameters were established with constructs that were stably expressed under the control of endogenous promoters. Under these physiological conditions, we showed that BiFC is specific and sensitive enough to analyse dynamic protein interactions. We next used BiFC in a candidate interaction screen, which led to the identification of several Hox protein partners.
Our results establish the general suitability of BiFC for revealing and studying protein interactions in their physiological context during the rapid course of Drosophila embryonic development.
Since its cloning in 1992 , the green fluorescent protein (GFP) from the jellyfish Aequorea victoria has become a potent tool in biology. GFP has been manipulated and modified and a set of derived fluorescent proteins showing different spectral properties and stabilities are now available. GFP proteins can tolerate peptide insertions without perturbing its fluorescence characteristics [2, 3]. This led to the finding that split GFP fragments can reconstitute a functional fluorescent protein when fused to interacting peptides . First applied in Escherichia coli  and Hela cells , the visualization of protein interactions in vivo was definitively validated by the work of Hu et al., who investigated the interactions between transcription factors in mammalian cells .
This protein interaction assay, termed BiFC for bimolecular fluorescence complementation, uses a number of protein variants derived either from the GFP [yellow fluorescent protein (YFP), Venus and Cerulean]  or from other fluorescent proteins (mRFP1 , mCherry  and Dronpa ). Protein interactions are visualized using a standard epifluorescence microscope and the analysis does not require complex data processing. Combining peptides of different fluorescent proteins for complementation (termed multicolour BiFC) expanded the number of protein interactions that could be simultaneously visualized [7, 11, 12].
BiFC is by now a widely used assay for testing the protein interaction status in cultured cells  and plants . In contrast, only a handful studies have used the technology for the visualization of protein interactions in animal model organisms. In vertebrates, BiFC has been performed in Xenopus laevis and Danio rerio to visualize the nuclear translocation of a Smad2/Smad4 complex in embryonic explants [15, 16] or embryos , respectively. In invertebrates, BiFC was used to reveal interactions between proteins of gap junctions  or between leucine zipper polypeptides  in Caenorabditis elegans and between odorant receptors , actin nucleation proteins  or transcription factors  in Drosophila melanogaster. Importantly, these latter studies were only undertaken in adults.
The absence of BiFC applications in the embryos of these model organisms might appear somewhat surprising, given that they have been extensively used to address questions relating to mechanisms underlying developmental control, and that protein interactions are central to the developmental processes. Among the reasons that may explain this are technical constraints, such as the short time window characterizing worm and fly embryogenesis, and the sensitivity of the technique when fusion proteins are expressed at the same level as the endogenous proteins of interest, a prerequisite for drawing physiologically relevant conclusions.
In this work, we question the suitability of the BiFC technology in conditions where proteins are expressed at physiological levels in the Drosophila embryo, using the partnership between the Hox protein AbdominalA (AbdA) and the PBC class Extradenticle (Exd) protein as a paradigm. Hox proteins are present in all bilaterians and play major roles during embryonic development by controlling diversified morphogenesis along the anteroposterior axis . Hox and PBC proteins are both homeodomain (HD) containing transcription factors and have been shown to work in many instances by assembling regulatory complexes on cis-regulatory sequences of target genes .
Here, we establish experimental parameters that make BiFC compatible with both the physiological levels of protein expression and the short time window characterizing Drosophila embryogenesis. Our work demonstrates that BiFC is a specific and sensitive method which allows the visualization of spatial interaction dynamics between AbdA and Exd. The technology was further used for assessing the in vivo interaction status between AbdA and several candidate partners, including non-HD containing transcription factors. Results indicate that the methodology is generally suited for the study of interactions between different types of transcription factors.
Fusion protein constructs and transgenic lines
The DNA fragments coding for the N-terminal (VN: 1-173) and C-terminal (VC: 155-238) moieties of Venus were generated by polymerase chain reaction (PCR) and cloned into EcoRI-XhoI or XhoI-XbaI restriction sites of the pUAST  and pUASTattB  vectors for 5' and 3' fusions, respectively. The DNA inserts to obtain fusions downstream the Venus fragment were cloned into XhoI-XbaI sites, while inserts to generate fusions upstream the Venus fragment were cloned into EcoRI-XhoI sites. Inserts coding for Hox, Exd and candidate partner proteins were generated by PCR from full length complimentary DNAs (cDNA). Exd fusion proteins were also HA-tagged. In all constructs, a linker of five amino acids was added to separate the Venus fragment from the protein of interest. Primers used are available upon request.
Cerulean N-terminal (CN: 1-173) and C-terminal (CC:155-238) coding fragments  were cloned into EcoRI-XhoI sites of pUAST for generating 5' fusions with AbdA and Exd proteins. mCherry  N-terminal (mCN: 1-159) and C-terminal (mCC:159-237) coding fragments were cloned into EcoRI and XhoI sites of pUASTattB, respectively, for generating 5' fusions with AbdA and Exd.
All constructs were sequence-verified before fly transformation and transgenic lines were established either by the ΦC-31 integrase  Ultrabithorax (Ubx), AbdA and Exd variants, as well as fusions with split mCherry fragments] or by classical P-element (nlsVN, VC, HthVN, VCHth, VNTFIIbeta, BIP2VN, TshVN, VNBin, as well as AbdA and Exd constructs with split fragments of Cerulean) mediated germ line transformation.
The abdAGal4 line was generated by replacing a P-lacZ insertion (line HC7JA1 ) by a P-Gal4 element . Other Gal4 drivers used are: armadillo(arm)-Gal4, paired(prd)-Gal4, engrailed (en)-Gal4, 24B-Gal4, breathless (btl)-Gal4 and Ubx-Gal4 M1  drivers. UAS-abdA, UAS-Bip2, UAS-β-Galactosidase and exd XP11 lines were obtained from the Bloomington Stock Centre (IN, USA); the UAS-Tsh line was kindly provided by S Kerridge.
Cuticle preparations and immunostaining
Embryo collections, cuticle preparations and immunodetections were performed according to standard procedures [29, 30]. The antibodies used were: chicken anti-GFP (Promega, WI, USA; 1/500), mouse anti-ß-galactosidase (Molecular Probe, Invitrogen, CA, USA; 1/500), rat anti-HA (Molecular Probe, 1/500), rabbit anti-AbdA (Dm.Abd-A.1, 1/1000), mouse anti-Ubx (FP3.38, 1/100).
In vivoquantification of protein expression levels
Experimental conditions allowing physiological levels of protein expression were established by two steps. First, the VC-AbdA construct was expressed with the arm-Gal4 driver at different temperatures. The anti-AbdA fluorescent immunostaining was compared between the A2 segment of wild-type embryos and the T2 segment of embryos ectopically expressing VC-AbdA. Pictures of at least 10 different embryos were taken with a LSM510 Zeiss confocal microscope under fixed parameters of acquisition and average levels of pixel intensities were determined with the ImageJ software. The ratio between the experimentally induced AbdA protein levels in T2 and the wild-type AbdA protein levels in A2 was determined, allowing the selection of the temperature for which the induced AbdA expression with arm-Gal4 was close to the endogenous AbdA expression levels found normally in A2.
Secondly, we established experimental conditions for using the abdAGal4 driver. This was performed with the VC-AbdA construct: intensities of the GFP fluorescent immunostaining were measured and compared to embryos expressing VC-AbdA with the arm-Gal4 driver in order to adjust the physiological conditions of expression. In our conditions, abdA-Gal4 is 20% more expressed than arm-Gal4. Since arm-Gal4 led to 80% of the AbdA endogenous expression levels in the A2 segment, we concluded that our experimental conditions with abdA-Gal4 are close to physiological levels of expression.
BiFC visualization in living embryos
Fly crosses for BiFC analyses were set up at the defined temperature over night. After the removal of the flies, the embryos were kept at 4°C for 28 h before live imaging. In order to visualize the complementation between split Venus fragments, living embryos were dechorionated and mounted in the halocarbon oil 10S (commercialized by VWR, Pennsylvania, USA). In order to quantify the BiFC signals, unsaturated images of ectodermal fluorescence were taken in embryos of the desired stage (with a minimum of 10 embryos by condition) using a LSM510 confocal microscope (Zeiss, Jena, Germany). For Venus fluorescence, filters were adjusted at 500 nm for excitation and 535 nm for emission. Identical parameters of acquisition were applied between the different genotypes. The number and intensity of the all pixels (for each embryo) were measured using the histogram function of the ImageJ Software. The quantification of fluorescence complementation was shown for each condition by boxplot representation using R-Software. Boxplot depicts: the smallest value, lower quartile, median (green line), upper quartile and largest value for each condition. Black points correspond to individual measures. Cerulean and mCherry BiFC signals were taken after identical maturation times of 28 h with the AxioImagerZ1 microscope (Zeiss), using specific filters for excitation and emission wavelengths (440/475 nm for Cerulean, 580/610 nm for mCherry).
Test of the lethality induced by incubation times at 4°C
A fixed number of 100 embryos were collected over a 1h period at room temperature and left to develop for 4h, which allows having a high proportion of stage 9/10 embryos. Embryos were next placed at 4°C for different times before being returned to resume and complete development at room temperature. The embryonic lethality rate was deduced by counting the number of hatching embryos that gave rise to first instar larvae. For each incubation time the experiments were repeated twice and the results were stably reproduced.
Protein expression and electrophoretic mobility shift assays (EMSAs)
The fusion constructs were subcloned in the PcDNA3 vector and sequence-verified. Proteins were produced with the TNT T7-coupled in vitro transcription/translation system (Promega). Production yields of AbdA and Exd fusion proteins were estimated by 35S-methionine labelling. The amount of Hox proteins used in the band shift assays is indicated in the figure legends. EMSAs were performed as described previously . We used double strands radiolabelled DllRcon 5'-TATTTGGGCCATAAATCATTCCCGCGGACAGTT-3'  and PRS 5'-TTAGCGCGGGCGCATCAATCAATTTTCG-3' probes . The rabbit anti-AbdA and mouse anti-Exd antibodies were used at a 1/100 dilution for the 'supershift' experiments.
Influence of the fusion topology on AbdA-Exd complex formation in vitro
Indeed, EMSAs allow one to measure both DNA binding and protein interactions. Each of the split fragments of the Venus fluorescent protein was fused to the N-terminus or the C-terminus of AbdA and Exd (Figure 1a). Importantly, all fusion proteins were constructed with an identical five amino acids linker region of about 20 angstroms long (see Methods). Although this short linker length could limit the efficiency of fluorescence complementation , it was preferred over longer linkers in order to avoid artificial complementation in the absence of protein interactions. We considered this to be of particular importance for transcription factors, since the proximity of binding sites on the DNA target, or chromatin looping, could bring proteins in close proximity without implying direct interactions between them. EMSAs were performed with two different probes, both of which corresponded to consensus DNA target sites. Consensus binding sites were preferred over physiological sites since they probably better reflect the general DNA binding properties of a given transcription factor. The first probe, called Dllcon (for Distalless consensus), corresponds to a consensus Hox-Exd DNA binding site for the AbdA monomer and the AbdA-Exd heterodimer . The second one, called PRS (for Pbx1 recognition sequence), corresponds to a monomer binding site for PBC proteins in vitro . Together, these probes allow us to measure monomer DNA-binding activities and heterodimer formation separately, enabling conclusions to be drawn on the AbdA/Exd interaction potential.
EMSAs with single proteins show that the fusion topology is not neutral for DNA binding. In AbdA, the VN (Venus) fragment strongly inhibits monomer DNA binding while the VC fragment has minor effects (Figure 1b, compare lanes 2-3 to 4-5 and the quantitative diagram). In Exd, it is the position of the fusion that is important, with fusions at the C-terminus affecting the DNA binding activity more strongly than fusions at the N-terminus (Figure 1c, compare lanes 3 and 5 to lanes 2 and 4, and the quantitative diagram). For heterodimer formation, EMSAs with all eight possible combinations between AbdA and Exd fusion proteins showed that the protein topologies providing the highest levels of monomer binding did not necessarily yield to highest levels of heterodimeric complexes (Figure 1d). For example, the best combination for heterodimeric formation corresponds to AbdA-VC/Exd-VN, although these two fusion proteins did not show the strongest monomer DNA binding activities (compare the red and green quantitative diagrams in Figure 1d). This observation suggests that fusion topologies also affect the AbdA-Exd interaction. In order to confirm this point, we compared the efficiency of complex formation between proteins with equivalent monomer binding affinities such as the AbdA/VN-Exd and VC-AbdA/VN-Exd combinations (green quantitative diagrams in Figure 1e). Results show that the VC-AbdA fusion protein formed fewer complexes with VN-Exd than did the unfused AbdA (Figure 1e, compare lanes 3-4 to 1-2). As both AbdA proteins have equivalent DNA binding properties on Dllcon (Figure 1b, lanes 1 and 4), we conclude that the fusion of the VC fragment at the N-terminus of AbdA impaired protein interactions with Exd.
Our extensive in vitro analysis highlights the fact that the topology of the fusions can drastically affect complex formation between AbdA and Exd. Although this was expected in general, it was not possible to predict which topology would actually be the most neutral to protein activity. Therefore, EMSA may constitute a valuable rapid assay for choosing the best match of fusion proteins for BiFC analysis in the case of DNA binding transcription factors.
Expressing fusion proteins at endogenous levels
Influence of the fusion topology on AbdA and Exd functions
In conclusion, fusions with Venus fragments can affect activity in vivo, potentially leading to hypoactive proteins, as observed for VN-AbdA and Exd-VC fusion proteins. Interestingly, however, the in vivo activity of the proteins correlates well with their in vitro activity assessed by EMSA.
Establishing experimental parameters in order to visualize BiFC from the AbdA/Exd complex assembly in the Drosophilaembryo
Additional File 2: Additional File 2. Live imaging of a developing embryo expressing the VC-abdominalA (AbdA) and VN-extradenticle (Exd) fusion proteins with the abdA-Gal4 driver. Live imaging was acquired from stage 10 to stage 14 of embryogenesis. (MPG 12 MB)
From these experiments, we concluded that the best incubation time at 4°C to observe BiFC from the VC-AbdA/VN-Exd complex assembly was 28 h. Under these conditions, BiFC signals were strong enough and embryonic viability was not significantly affected. This incubation time was systematically applied for further BiFC assays.
Influence of the fusion topology on BiFC signals
As several protein motifs lying on both sides of the HD of AbdA could potentially interact with Exd , it was difficult to predict the best appropriate fusion insertion site in AbdA for BiFC with Exd. In order to test whether the choice of insertion sites could be predicted from the knowledge of the emplacement of known protein interaction domains, we performed BiFC between Exd and its well known nuclear translocation partner Hth. The domains required for Exd-Hth interactions have been well characterized and are located N-terminally in both proteins ( and Additional File 3). Thus, we tested the influence of fusion insertions at the N-terminus or C-terminus of Hth (Additional File 3), expecting that the fusion closest to the Exd-interaction domain would more strongly affect BiFC. Surprisingly, we observed the reverse, as it was the C-terminus fusion that produced no BiFC, even under high levels of protein expression and confocal parameters rose at their maximum of sensitivity (Additional File 3).
Altogether, these results highlight the fact that the critical influence of fusion topology on BiFC efficiency is not predictable, even when interacting domains are known, and that the absence of BiFC signals for a given protein complex is difficult to interpret unless different fusion topologies have been tested. However, as shown for AbdA and Exd, previous in vitro protein interaction assays like EMSAs can be used for selecting the appropriate fusion topology for further BiFC assays.
Specificity of BiFC in the living Drosophilaembryo
In order to be exploitable, all complementation assays are based (with enzymes or fluorescent proteins) on the property of the spontaneous association of split fragments when brought together upon protein interactions . This property explains that the complex formation between fusion proteins is generally stabilized by the association of split fragments, allowing the visualization of weak and transient interactions. Importantly, this property also highlights the need to express fusion proteins at physiological levels and the necessity of using negative controls of BiFC for each protein complex.
In order to analyse the properties of split Venus fragments in the Drosophila embryo, we first expressed the VN and VC moieties as such. As expected, when expressed together, these two fragments led to strong BiFC signals which were visible after short incubation times at 4°C (Additional File 4). The level of fluorescence did not increase with longer incubation times (green-dotted curve in Figure 4b), suggesting that the VN/VC association was not regulated, leading to the rapid self-assembly of all VN and VC fragments. The level of fluorescence was stronger at later developmental stages (Additional File 4), providing a background level of unspecific BiFC, and which was probably due to the accumulation of peptide synthesis during embryogenesis.
Self-assembly of Venus moieties did not take place when one of the Venus fragments was fused to a protein, as previously described . Indeed, even after 28 h of incubation, VN/VC-AbdA or VC/VN-AbdA did not produce any BiFC signal or at extremely low levels in late stage embryos (Additional File 4). These results highlight that the affinity between Venus fragments is not strong enough to induce artificial self-assembly in the context of a protein fusion.
Nevertheless, the inherent property of Venus fragments for self-assembly stresses the necessity of performing controls to validate BiFC results. We validated the specificity of BiFC between AbdA and Exd using competition experiments as well as by using mutant fusion proteins impaired in their ability to interact.
As a second validation control, we searched for a condition where AbdA-Exd interactions could be drastically affected or abolished. As EMSAs previously revealed that AbdA/Exd complex assembly critically depends on monomer DNA binding, we mutated the amino acid 51 of the HD in the VC-AbdA and VN-Exd fusion proteins. These mutations are expected to abolish DNA binding. Accordingly, EMSAs showed that complex formation is strongly affected when one of the two partners contains the HD mutation and that no complex can be formed when both fusion proteins contain the HD mutation (Figure 6b). This result establishes that complex formation between AbdA and Exd cannot occur in the absence of DNA binding in vitro, which is not necessarily the case for interactions between AbdA and other cofactors (see below). The effect of HD mutations was next analysed on BiFC in the Drosophila embryo. We verified that HD-mutated fusion proteins were properly expressed, at levels equivalent to the wild type fusion proteins (Additional File 5). BiFC with the abdA-Gal4 driver showed that combinations involving one partner protein mutated in the HD produced weak fluorescent signals (Figure 6c). No BiFC was obtained when both fusion proteins were mutated (Figure 6c), even when expressed at higher levels with the engrailed (en)-Gal4 driver (Additional File 5). Thus, BiFC with the HD-mutated forms of AbdA and Exd recapitulates the previous in vitro observations.
Altogether, these data establish that BiFC cannot occur in the absence of AbdA-Exd interactions and that the complementation between split Venus fragments is not sufficient to promote artificial protein complex assembly.
BiFC can reveal a spatial control of protein interactions
Since the association between split fragments of Venus is likely to stabilize protein interactions, as noticed in vitro , we analysed whether this could extend the life time of the AbdA/Exd complex. This issue was addressed by taking advantage of the abdA-Gal4 driver which is progressively turned off during late embryonic stages in dorsal-most epidermis, while endogenous AbdA remains unchanged (Figures 7b-b'). We hypothesized that stabilisation of heterodimer complex formation by fluorescence complementation will lead to BiFC signals that will not follow the temporal dynamic of abdA-Gal4 driver expression, and will rather mimic the sustained endogenous AbdA expression pattern. This was clearly not the case since we observed loss of BiFC in dorsal most parts of late stage embryos, even for weak expression levels of fusion proteins (as assessed with the anti-HA staining against the VN-Exd-HA-tagged construct: Figures 7b-b'). This result shows that BiFC detection accurately follows dynamics of fusion proteins expression and/or degradation.
Multicolour BiFC in the Drosophilaembryo
Nevertheless, BiFC with Cerulean appears to be exploitable in the Drosophila embryo. This offers the possibility to test for multicolour BiFC, allowing the simultaneous visualization of two different protein complexes. This property is based on the capacity of the VN fragment to complement with the C-terminal fragment of Cerulean, producing a Venus-like fluorescent signal (Figure 8c and ). Multicolour BiFC was tested by co-expressing two Hox proteins, Ultrabithorax (Ubx) and AbdA, fused respectively to the N-terminal fragment of Venus (VN) or Cerulean (CN), with the Exd protein fused to the C-terminal fragment of Cerulean (CC). Expression of these three fusion proteins led to simultaneous Venus and Cerulean fluorescent signals in the same nuclei, corresponding to the VN-Ubx/CC-Exd and CN-AbdA/CC-Exd complexes, respectively (Figure 8d). Of note, these fusion proteins were expressed at twice their physiological level with an engrailed-Gal4 driver to thwart the possible competition of endogenous proteins. This result establishes that multicolour BiFC can be performed in the Drosophila embryo for simultaneous visualization of multiple protein interactions.
Suitability of BiFC for in vivoidentification of candidate interacting partners
The role of the PBC class proteins as Hox cofactors is well established. However, several Hox functions are PBC-independent, and other factors have been reported to interact with Hox proteins  and/or to assist them in the regulation of common downstream target genes . One class of such protein partners that has long been postulated comprises Hox proteins themselves. For example, Hox-Hox interactions have been proposed to occur in Drosophila to explain a phenomenon of quantitative competition and mutually inhibitory functions . Interactions among different vertebrate Hox proteins have also been described in vitro .
Here, we took advantage of BiFC to directly address the interaction status between AbdA and different candidate interacting proteins in vivo. Of note, all candidate interacting proteins were fused to the VN fragment, allowing analysis with VC-AbdA, which we previously showed as best suited for BiFC analysis. The position of the VN fragment in fusion proteins was dictated by cloning constraints since the best appropriate fusion topology cannot be predicted, as demonstrated above.
We next investigated whether proteins previously described as putative Hox cofactors could indeed directly interact with AbdA. Candidate proteins were chosen on the basis of genetic and/or biochemical analysis but direct evidence for in vivo interactions with AbdA had not been established. They were also chosen to cover distinct functions, such as region- or tissue-specific transcription factors or proteins of the basal transcription machinery.
We first tested Teashirt (Tsh), which is a zinc finger transcription factor. Tsh has been genetically described to act with Antennapedia (Antp) and Hox proteins of the Bithorax complex for specifying the embryonic trunk region . More recently, Tsh was also shown to directly interact with Sex combs reduced for specifying the prothorax in Drosophila . We observed that the VC-AbdA/Tsh-VN complex produced strong BiFC signals (Figure 9b), indicating that Tsh acts through direct interactions with AbdA. Moreover, AbdA/Tsh complexes accumulate in a few nuclear foci, suggesting that their assembly is tightly regulated within the nucleus (Figure 9b). Finally, BiFC with the HD-mutated form of AbdA showed that AbdA-Tsh interactions do not depend on AbdA DNA-binding (Figure 9b).
The second putative partner analysed is a tissue-specific transcription factor, Biniou (Bin), which contains a forkhead DNA-binding domain. Bin has been proposed to act as a Hox cofactor for the regulation of several target genes in the visceral mesoderm . BiFC, indeed, showed that AbdA interacts with Bin in few nuclear loci (Figure 9c) and in a DNA-binding-dependent manner (Figure 9c).
Finally, we tested the interaction potential of AbdA with two candidate proteins belonging to the basal transcriptional machinery, TFIIbeta and BIP2. TFIIbeta has been shown to interact with Ubx in a yeast two-hybrid assay , while BIP2 has been shown to interact with Antp . We found that AbdA/TFIIbeta (Figure 9d), but not AbdA/BIP2 complexes (Figure 9e), produced BiFC signals. Fluorescent signals were not affected upon the HD mutation in AbdA, suggesting that AbdA-TFIIbeta interaction did not depend on AbdA DNA-binding (Figure 9d).
Our results raised two limits of the BiFC for a candidate interaction screening. First, even if we observed that BIP2-VN was able to produce strong BiFC signals with VC-Ubx (Figure 9e), the absence of BiFC between VC-AbdA and BIP2-VN is not conclusive, since it could simply result from unfavourable fusion topologies, as previously described. Secondly, we noticed that the HD mutation in AbdA, which abolished interactions with Exd, did not affect BiFC with some candidate partners. This stresses the difficulty of finding appropriate negative controls when molecular cues required for protein complex assembly are not known. As already mentioned, competition experiments can provide suitable controls. We reasoned that specific interactions should be competitive against BiFC resulting from the VC-AbdA/VN-AbdA complex assembly, which can thus be used as an experimental test to validate interactions between AbdA and a given protein partner.
In order to validate the observed BiFC between AbdA and Tsh, we co-expressed VC-AbdA and VN-AbdA with wild-type Tsh, which we designed as a 'cold' competitive partner (not fused to a Venus fragment). We observed that this co-expression led to a drastic loss of BiFC (Figure 9f), which can be explained by the titration of VC-AbdA/VN-AbdA dimers outcompeted by cold (non fluorescent) VC-AbdA/Tsh and VN-AbdA/Tsh complexes.
The same type of experiment was performed with BIP2 to validate absence of BiFC between AbdA and BIP2. We observed that co-expressing the cold wild-type BIP2 protein led to a strong decrease of BiFC resulting from the VC-AbdA/VN-AbdA complex assembly (Figure 9f). We concluded that BIP2 can interact with AbdA despite unproductive BiFC assay, which in turn suggests that, in contrast to Tsh, BiFC with BIP2 was not visualized because of unfavourable fusion topologies.
Establishing BiFC as a specific protein interaction assay in the Drosophilaembryo
By using Hox partnership as a paradigm, we have demonstrated the suitability of BiFC for the study of transcription factors interactions in the Drosophila embryo. Our extensive analysis highlights the need for careful selection of fusion topologies, which is of particular importance when considering that absence of BiFC signal can simply result of an unfavourable fusion topology. There is no simple rule for predicting the arrangement of fusion topologies that will produce maximal signals, even when the protein interaction domains are known. For example, in the case of Hth, the N-terminus Exd interaction domain was paradoxically more affected by fusions at the C-terminus of the protein. Similarly, the best fusion topologies in AbdA and Exd cannot be predicted, although Hox/PBC interfaces have been well characterized by structural studies. Since these structures were obtained with portions of proteins, it can simply not be informative in the context of the three dimensional arrangements of the full length protein. Appropriate fusion constructs can, however, be rapidly identified with the help of in vitro interaction assays, limiting the number of Drosophila transgenic lines that need to be generated. In the case of AbdA and Exd, we showed that EMSA provides a mean to select the best suited fusion proteins for subsequent BiFC analysis in the Drosophila embryo. The impact of fusion topologies on BiFC also underlines the need of using the same fusion topology when comparing protein interaction efficiencies.
We showed that the VN and VC fragments were able to self assemble with a typical kinetic when expressed as single peptides. Although this property was expected for a visualisation method based on complementation, it also emphasized the need of control experiments for validating BiFC results. Here, we provided several evidences demonstrating the specificity of BiFC for the study of protein interactions in the Drosophila embryo. This includes the extremely weak tendency of VN and VC fragments to self-associate when one fragment is fused to AbdA, the destabilization of complex formation through protein competition, the absence of BiFC signals in absence of interactions, the spatial restriction of BiFC in the tracheal system or absence of BiFC between AbdA and BIP2.
We found that the time required for visualizing BiFC signals can be provided by keeping embryos at 4°C, ideally for a period no longer than 28 h. This procedure allows the visualization of protein interactions within the short time window of Drosophila embryonic development and suggests that the methodology could be generally suited for the study of protein interactions in the context of other rapid developmental processes. Finally, we showed that multicolour BiFC is applicable to the Drosophila embryo, extending further the potential of the methodology for in vivo simultaneous study of multiple protein interactions.
BiFC, a simple and sensitive method allowing in vivoanalyses of protein interactions under physiological conditions
BiFC is a commonly used method in cultured cells and in plants for studying protein interactions. Surprisingly, this method remains rarely used in living animal embryos. In contrast, the visualization method based on fluorescence resonance energy transfer (FRET) has been more thoroughly used in animal models, although it requires complex instrumentation and data processing. FRET analysis is based on energy transfer between two fluorophores placed in close proximity, allowing real time detection of complex formation and dissociation. The detection of this energy transfer usually requires high levels of protein expression  and is crucially dependent of the distance between the donor and acceptor molecules . This constrain may explain that FRET is mainly used to monitor conformational changes, processes of great importance in signalling . Accordingly, FRET has only been used in the context of biosensors [50, 51] and as a caspase activity indicator  in Drosophila.
In this work, we establish BiFC as a complementary approach to FRET. We show that the sensitivity of BiFC enables it to work at physiological levels of expression. This is a crucial property of the methodology, since quantitative differences in protein interactions play important roles in controlling in vivo protein functions. In order to validate BiFC as a unique visualization method for measuring physiologically relevant protein interactions, we have generated a P-Gal4 driver by replacing a P-lacZ insertion in abdA cis-regulatory sequences. This driver reproduces the abdA expression profile and we selected experimental conditions allowing expression levels close to endogenous AbdA. Importantly, this insertion affects endogenous abdA expression, which offers the advantage of performing BiFC in cells containing normal doses of AbdA. Loss of the endogenous competitive protein can also be crucial for visualising weak protein interactions by BiFC. The huge number of genomic P insertions in Drosophila are thus of great utility for performing BiFC under physiological conditions.
Identification of interacting protein partners using BiFC
Given the sensitivity, specificity and simplicity of BiFC, the method should in principle be suited for the identification of interacting protein partners. We investigated the suitability of the method for revealing protein interactions of different natures, using a candidate gene approach. Although the expression level of all candidate cofactors was not systematically verified, several transgenic lines were tested in each case. This includes: Hox-Hox interactions that do not involve a TALE (three amino acids loop extension) homeodomain as in the case of AbdA/Exd interactions; interactions with proteins of the basal transcription machinery; and interactions with zinc-finger or forkhead transcription factors. The nature of these interacting partners, involved in different transcriptional processes, illustrates the potential of Hox proteins to accommodate interactions with various classes of transcription factors.
We tried to validate the specificity of these interactions by using a DNA-binding deficient form of AbdA. We observed that this mutation abolished interactions with Exd or Bin but had no effect on interactions with Ubx, Tsh or TFIIbeta. This result stresses that the DNA binding of Hox proteins is not a mandatory step for recruiting protein partners in vivo. It also illustrates the difficulty of validating BiFC observations by using mutated proteins when the molecular cues required for the complex assembly are not known. We showed that an alternative way to validate BiFC observations is through competition experiments. This was achieved by co-expressing the candidate partner as a 'cold' protein with the VC-AbdA and VN-AbdA fusion proteins. In the context of specific interactions, the titration of AbdA fusion proteins by the cold competitive partner will affect BiFC resulting from the VC-AbdA/VN-AbdA complex assembly, as observed with Tsh. The same approach can also be instructive in absence of BiFC, as shown with BIP2, revealing an unfavourable arrangement of fusion topologies for visualising protein interactions.
Interestingly, interactions between AbdA and candidate partners have distinct and typical nuclear localisations, suggesting that the nuclear architecture has important roles in controlling Hox protein interactions and functions. It also illustrates the power of BiFC for deciphering unanticipated mechanisms regulating protein interactions within a proper physiological context.
Finally, the variety of protein interactions visualized through BiFC indicates that the methodology could be used more globally  for screening for novel interacting partners in a developing organism. This could readily be achieved combining BiFC with the strength of Drosophila genetics.
We have explained how to perform multicolour BiFC with physiological levels of protein expression in the live Drosophila embryo. We provided experimental procedures to circumvent the limitations of BiFC and we designed control experiments to validate BiFC results. We also revealed the importance of choosing the best match of fusion proteins, which can be rapidly achieved with transcription factors through EMSA experiments. Finally, we described how to use BiFC for finding novel cofactors through a candidate gene approach. Altogether, our work demonstrates that BiFC is not only a physiologically relevant protein interaction test but also a powerful method for studying protein interaction dynamics during Drosophila embryogenesis.
bimolecular fluorescence complementation
C-terminal fragment of Cerulean
cyan fluorescent protein
N-terminal fragment of Cerulean
electrophoretic mobility shift assay
fluorescence resonance energy transfer
three amino acids insertion
polymerase chain reaction
N-terminal fragment of venus
C-terminal fragment of Venus
N-terminal fragment of mCherry
C-terminal fragment of mCherry
We thank Tom K Kerppola, Xavier Morin, Cédric Raoult, Stephane Zaffran, Manfred Frasch, Alain Vincent, Steve Kerridge, Michèle Crozatier, Ernesto Sanchez-Herrero, Dimitris Papadopoulos and the Bloomington stock centre for providing cDNAs, fly stocks and antibodies. We also thank Jonathan Ewbank and René Rezsohazy for their critical reading of the manuscript. This work was supported by the CNRS, grants from 'l'Association pour la Recherche contre le Cancer' (ARC), 'la Ligue Nationale Contre Le Cancer' (LNCC), 'l'Agence Nationale pour la Recherche' (ANR).
- Prasher DC, Eckenrode VK, Ward WW, Prendergast FG, Cormier MJ: Primary structure of the Aequorea victoria green-fluorescent protein. Gene. 1992, 111 (2): 229-233.View ArticlePubMedGoogle Scholar
- Abedi MR, Caponigro G, Kamb A: Green fluorescent protein as a scaffold for intracellular presentation of peptides. Nucleic Acids Res. 1998, 26 (2): 623-630.PubMed CentralView ArticlePubMedGoogle Scholar
- Baird GS, Zacharias DA, Tsien RY: Circular permutation and receptor insertion within green fluorescent proteins. Proc Natl Acad Sci USA. 1999, 96 (20): 11241-11246.PubMed CentralView ArticlePubMedGoogle Scholar
- Ghosh Indraneel, Hamilton AD, Lynn Regan: Antiparallel Leucine Zipper-Directed Protein Reassembly: Application to the Green Fluorescent Protein. J Am Chem Soc. 2000, 122: 5658-5659.View ArticleGoogle Scholar
- Nagai T, Sawano A, Park ES, Miyawaki A: Circularly permuted green fluorescent proteins engineered to sense Ca2+. Proc Natl Acad Sci USA. 2001, 98 (6): 3197-3202.PubMed CentralView ArticlePubMedGoogle Scholar
- Hu CD, Chinenov Y, Kerppola TK: Visualization of interactions among bZIP and Rel family proteins in living cells using bimolecular fluorescence complementation. Mol Cell. 2002, 9 (4): 789-798.View ArticlePubMedGoogle Scholar
- Shyu YJ, Liu H, Deng X, Hu CD: Identification of new fluorescent protein fragments for bimolecular fluorescence complementation analysis under physiological conditions. Biotechniques. 2006, 40 (1): 61-66.View ArticlePubMedGoogle Scholar
- Jach G, Pesch M, Richter K, Frings S, Uhrig JF: An improved mRFP1 adds red to bimolecular fluorescence complementation. Nat Methods. 2006, 3 (8): 597-600.View ArticlePubMedGoogle Scholar
- Fan JY, Cui ZQ, Wei HP, Zhang ZP, Zhou YF, Wang YP, Zhang XE: Split mCherry as a new red bimolecular fluorescence complementation system for visualizing protein-protein interactions in living cells. Biochem Biophys Res Commun. 2008, 367 (1): 47-53.View ArticlePubMedGoogle Scholar
- Lee YR, Park JH, Hahm SH, Kang LW, Chung JH, Nam KH, Hwang KY, Kwon IC, Han YS: Development of bimolecular fluorescence complementation using Dronpa for visualization of protein-protein interactions in cells. Mol Imaging Biol. 2010, 12 (5): 468-478.View ArticlePubMedGoogle Scholar
- Hu CD, Kerppola TK: Simultaneous visualization of multiple protein interactions in living cells using multicolor fluorescence complementation analysis. Nat Biotechnol. 2003, 21 (5): 539-545.PubMed CentralView ArticlePubMedGoogle Scholar
- Grinberg AV, Hu CD, Kerppola TK: Visualization of Myc/Max/Mad family dimers and the competition for dimerization in living cells. Mol Cell Biol. 2004, 24 (10): 4294-4308.PubMed CentralView ArticlePubMedGoogle Scholar
- Kerppola TK: Bimolecular fluorescence complementation (BiFC) analysis as a probe of protein interactions in living cells. Annu Rev Biophys. 2008, 37: 465-487.PubMed CentralView ArticlePubMedGoogle Scholar
- Ohad N, Shichrur K, Yalovsky S: The analysis of protein-protein interactions in plants by bimolecular fluorescence complementation. Plant Physiol. 2007, 145 (4): 1090-1099.PubMed CentralView ArticlePubMedGoogle Scholar
- Hagemann AI, Xu X, Nentwich O, Hyvonen M, Smith JC: Rab5-mediated endocytosis of activin is not required for gene activation or long-range signalling in Xenopus. Development. 2009, 136 (16): 2803-2813.PubMed CentralView ArticlePubMedGoogle Scholar
- Saka Y, Hagemann AI, Piepenburg O, Smith JC: Nuclear accumulation of Smad complexes occurs only after the midblastula transition in Xenopus. Development. 2007, 134 (23): 4209-4218.PubMed CentralView ArticlePubMedGoogle Scholar
- Harvey SA, Smith JC: Visualisation and Quantification of Morphogen Gradient Formation in the Zebrafish. PLoS Biol. 2009, 7 (5): e101-View ArticleGoogle Scholar
- Chen B, Liu Q, Ge Q, Xie J, Wang ZW: UNC-1 regulates gap junctions important to locomotion in C. elegans. Curr Biol. 2007, 17 (15): 1334-1339.PubMed CentralView ArticlePubMedGoogle Scholar
- Zhang S, Ma C, Chalfie M: Combinatorial marking of cells and organelles with reconstituted fluorescent proteins. Cell. 2004, 119 (1): 137-144.View ArticlePubMedGoogle Scholar
- Benton R, Sachse S, Michnick SW, Vosshall LB: Atypical membrane topology and heteromeric function of Drosophila odorant receptors in vivo. PLoS Biol. 2006, 4 (2): e20-PubMed CentralView ArticlePubMedGoogle Scholar
- Gohl C, Banovic D, Grevelhörster A, Bogdan S: WAVE forms hetero- and homo-oligomeric complexes at integrin junctions in Drosophila visualized by bimolecular fluorescence complementation. J Biol Chem. 2010, 285 (51): 40171-40179.PubMed CentralView ArticlePubMedGoogle Scholar
- Plaza S, Prince F, Adachi Y, Punzo C, Cribbs DL, Gehring WJ: Cross-regulatory protein-protein interactions between Hox and Pax transcription factors. Proc Natl Acad Sci USA. 2008, 105 (36): 13439-13444.PubMed CentralView ArticlePubMedGoogle Scholar
- McGinnis W, Krumlauf R: Homeobox genes and axial patterning. Cell. 1992, 68 (2): 283-302.View ArticlePubMedGoogle Scholar
- Mann RS, Lelli KM, Joshi R: Hox specificity unique roles for cofactors and collaborators. Curr Top Dev Biol. 2009, 88: 63-101.PubMed CentralView ArticlePubMedGoogle Scholar
- Brand AH, Perrimon N: Targeted gene expression as a means of altering cell fates and generating dominant phenotypes. Development. 1993, 118 (2): 401-415.PubMedGoogle Scholar
- Bischof J, Maeda RK, Hediger M, Karch F, Basler K: An optimized transgenesis system for Drosophila using germ-line-specific phiC31 integrases. Proc Natl Acad Sci USA. 2007, 104 (9): 3312-3317.PubMed CentralView ArticlePubMedGoogle Scholar
- Bender W, Hudson A: P element homing to the Drosophila bithorax complex. Development. 2000, 127 (18): 3981-3992.PubMedGoogle Scholar
- de Navas L, Foronda D, Suzanne M, Sánchez-Herrero E: A simple and efficient method to identify replacements of P-lacZ by P-Gal4 lines allows obtaining Gal4 insertions in the bithorax complex of Drosophila. Mech Dev. 2006, 123 (11): 860-867.View ArticlePubMedGoogle Scholar
- Alexandre E, Graba Y, Fasano L, Gallet A, Perrin L, De Zulueta P, Pradel J, Kerridge S, Jacq B: The Drosophila teashirt homeotic protein is a DNA-binding protein and modulo, a HOM-C regulated modifier of variegation, is a likely candidate for being a direct target gene. Mech Dev. 1996, 59 (2): 191-204.View ArticlePubMedGoogle Scholar
- Hu N, Castelli-Gair J: Study of the posterior spiracles of Drosophila as a model to understand the genetic and cellular mechanisms controlling morphogenesis. Dev Biol. 1999, 214 (1): 197-210.View ArticlePubMedGoogle Scholar
- Popperl H, Bienz M, Studer M, Chan SK, Aparicio S, Brenner S, Mann RS, Krumlauf R: Segmental expression of Hoxb-1 is controlled by a highly conserved autoregulatory loop dependent upon exd/pbx. Cell. 1995, 81 (7): 1031-1042.View ArticlePubMedGoogle Scholar
- Gebelein B, Culi J, Ryoo HD, Zhang W, Mann RS: Specificity of Distalless repression and limb primordia development by abdominal Hox proteins. Dev Cell. 2002, 3 (4): 487-498.View ArticlePubMedGoogle Scholar
- Lu Q, Wright DD, Kamps MP: Fusion with E2A converts the Pbx1 homeodomain protein into a constitutive transcriptional activator in human leukemias carrying the t(1;19) translocation. Mol Cell Biol. 1994, 14 (6): 3938-3948.PubMed CentralView ArticlePubMedGoogle Scholar
- Merabet S, Kambris Z, Capovilla M, Bérenger H, Pradel J, Graba Y: The hexapeptide and linker regions of the AbdA Hox protein regulate its activating and repressive functions. Dev Cell. 2003, 4 (5): 761-768.View ArticlePubMedGoogle Scholar
- Azpiazu N, Morata G: Functional and regulatory interactions between Hox and extradenticle genes. Genes Dev. 1998, 12 (2): 261-273.PubMed CentralView ArticlePubMedGoogle Scholar
- Abu-Shaar M, Ryoo HD, Mann RS: Control of the nuclear localization of Extradenticle by competing nuclear import and export signals. Genes Dev. 1999, 13 (8): 935-945.PubMed CentralView ArticlePubMedGoogle Scholar
- Peifer M, Wieschaus E: Mutations in the Drosophila gene extradenticle affect the way specific homeo domain proteins regulate segmental identity. Genes Dev. 1990, 4 (7): 1209-1223.View ArticlePubMedGoogle Scholar
- Kerppola TK: Visualization of molecular interactions by fluorescence complementation. Nat Rev Mol Cell Biol. 2006, 7 (6): 449-456.PubMed CentralView ArticlePubMedGoogle Scholar
- Merabet S, Hudry B, Saadaoui M, Graba Y: Classification of sequence signatures: a guide to Hox protein function. Bioessays. 2009, 31 (5): 500-511.View ArticlePubMedGoogle Scholar
- Mann RS, Affolter M: Hox proteins meet more partners. Curr Opin Genet Dev. 1998, 8 (4): 423-429.View ArticlePubMedGoogle Scholar
- Bondos SE, Tan XX, Matthews KS: Physical and genetic interactions link hox function with diverse transcription factors and cell signaling proteins. Mol Cell Proteomics. 2006, 5 (5): 824-834.View ArticlePubMedGoogle Scholar
- Lamka ML, Boulet AM, Sakonju S: Ectopic expression of UBX and ABD-B proteins during Drosophila embryogenesis: competition, not a functional hierarchy, explains phenotypic suppression. Development. 1992, 116 (4): 841-854.PubMedGoogle Scholar
- Zappavigna V, Sartori D, Mavilio F: Specificity of HOX protein function depends on DNA-protein and protein-protein interactions, both mediated by the homeo domain. Genes Dev. 1994, 8 (6): 732-744.View ArticlePubMedGoogle Scholar
- Fasano L, Röder L, Coré N, Alexandre E, Vola C, Jacq B, Kerridge S: The gene teashirt is required for the development of Drosophila embryonic trunk segments and encodes a protein with widely spaced zinc finger motifs. Cell. 1991, 64 (1): 63-79.View ArticlePubMedGoogle Scholar
- Taghli-Lamallem O, Gallet A, Leroy F, Malapert P, Vola C, Kerridge S, Fasano L: Direct interaction between Teashirt and Sex combs reduced proteins, via Tsh's acidic domain, is essential for specifying the identity of the prothorax in Drosophila. Dev Biol. 2007, 307 (1): 142-151.View ArticlePubMedGoogle Scholar
- Zaffran S, Küchler A, Lee HH, Frasch M: biniou (FoxF), a central component in a regulatory network controlling visceral mesoderm development and midgut morphogenesis in Drosophila. Genes Dev. 2001, 15 (21): 2900-2915.PubMed CentralPubMedGoogle Scholar
- Prince F, Katsuyama T, Oshima Y, Plaza S, Resendez-Perez D, Berry M, Kurata S, Gehring WJ: The YPWM motif links Antennapedia to the basal transcriptional machinery. Development. 2008, 135 (9): 1669-1679.View ArticlePubMedGoogle Scholar
- Sapsford KE, Berti L, Medintz IL: Materials for fluorescence resonance energy transfer analysis: beyond traditional donor-acceptor combinations. Angew Chem Int Ed Engl. 2006, 45 (28): 4562-4589.View ArticlePubMedGoogle Scholar
- Schmid JA, Birbach A: Fluorescent proteins and fluorescence resonance energy transfer (FRET) as tools in signaling research. Thromb Haemost. 2007, 97 (3): 378-384.PubMedGoogle Scholar
- Gordon S, Dickinson MH: Role of calcium in the regulation of mechanical power in insect flight. Proc Natl Acad Sci USA. 2006, 103 (11): 4311-4315.PubMed CentralView ArticlePubMedGoogle Scholar
- Lissandron V, Rossetto MG, Erbguth K, Fiala A, Daga A, Zaccolo M: Transgenic fruit-flies expressing a FRET-based sensor for in vivo imaging of cAMP dynamics. Cell Signal. 2007, 19 (11): 2296-2303.View ArticlePubMedGoogle Scholar
- Takemoto K, Kuranaga E, Tonoki A, Nagai T, Miyawaki A, Miura M: Local initiation of caspase activation in Drosophila salivary gland programmed cell death in vivo. Proc Natl Acad Sci USA. 2007, 104 (33): 13367-13372.PubMed CentralView ArticlePubMedGoogle Scholar
- Ding Z, Liang J, Lu Y, Yu Q, Songyang Z, Lin SY, Mills GB: A retrovirus-based protein complementation assay screen reveals functional AKT1-binding partners. Proc Natl Acad Sci USA. 2006, 103 (41): 15014-15019.PubMed CentralView ArticlePubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.