to BioTechniques free email alert service to receive content updates.
An improved bimolecular fluorescence complementation assay with a high signal-to-noise ratio
Yutaka Kodama*1 and Chang-Deng Hu2
1Department of Medicinal Chemistry and Molecular Pharmacology, Purdue University, West Lafayette, IN, USA
2Purdue University Center for Cancer Research, Purdue University, West Lafayette, IN, USA

*Y.K.'s current address is Plant Functional Genomics Research Group, RIKEN Plant Science Center, Tsurumi-ku, Yokohama 230-0045, Japan
BioTechniques, Vol. 49, No. 5, November 2010, pp. 793–805
Full Text (PDF)
Supplementary Material

Protein-protein interactions (PPIs) play crucial roles in various biological processes. Among biochemical, genetic, and imaging approaches that have been used for the study of PPIs, visualization of PPIs in living cells is the key to understanding their cellular functions. The bimolecular fluorescence complementation (BiFC) assay represents one of these imaging tools for direct visualization of PPIs in living cells. The BiFC assay is based on the structural complementation of two nonfluorescent N- and C-terminal fragments of a fluorescent protein when they are fused to a pair of interacting proteins. Although over 10 different fluorescent proteins have been used for BiFC assays, the two nonfluorescent fragments from all of these fluorescent proteins can spontaneously self-assemble, which contributes to background fluorescence and decreases the signal-to-noise (S/N) ratio in the BiFC assay. Here we report the identification of a mutation, I152L, that can specifically reduce self-assembly and decrease background fluorescence in a Venus-based BiFC system. This mutation allows a 4-fold increase in the S/N ratio of the BiFC assay in living cells. This improved Venus-based BiFC system will facilitate PPI studies in various biological research fields.

Protein-protein interactions (PPIs) are important to many cellular processes. Direct visualization of PPIs in living cells and organisms has become key in fully understanding their cellular roles. Several assays such as fluorescence resonance energy transfer (FRET) (1), bioluminescence resonance energy transfer (2), and bimolecular fluorescence complementation (BiFC) (3,4) have been developed to visualize PPIs in living cells. These assays can provide information regarding when and where PPIs occur in the cell.

The BiFC assay, as originally reported for visualization of PPIs in living mammalian cells, is based on the complementation of two nonfluorescent N- and C-terminal fragments from an enhanced yellow fluorescent protein (EYFP) (4). When the two fragments are brought into proximity by an interaction between two proteins fused to the fragments (Figure 1), the reconstituted fluorescence can be easily observed by any fluorescence microscopy (4). Since the EYFP-based BiFC assay was sensitive to higher temperatures, the Venus protein, a YFP variant, was recently used in the BiFC assay (5). Venus, which has fast and efficient maturation, was generated from EYFP with several mutations: F46L, F64L, M153T, V163A, and S175G (6). Unlike EYFP, Venus is less sensitive to the environment (6), and fluorescence intensity of the Venus-based BiFC was ~10× higher than that of EYFP-based BiFC (5). Other fluorescent proteins—such as cyan (5,7), green (3), red (8-11), and photoswitchable dronpa (12)—have been also used for BiFC assays (13,14). The basic principle of the BiFC assay has also been utilized to study post-translational modifications, protein folding, protein aggregation, protein conformational change, and protein topology (14). In addition, multicolor BiFC and BiFC-based FRET assays have been developed to visualize multiple protein interactions (7,10,11,15,16). Despite these successes, one common limitation of the BiFC assay is the self-assembly between the two nonfluorescent fragments derived from most, if not all, fluorescent proteins (14). This self-assembly contributes to false-positive fluorescence and decreases the signal-tonoise (S/N) ratio in the BiFC assay, making data interpretation difficult. To date, only one mutation, T153M, a mutation known to affect the protein folding of Venus (6), in the Venus N-terminal fragment was reported to reduce the spontaneous self-assembly in Xenopus embryos (17). However, it remains to be determined whether the T153M mutation can be used in other BiFC applications.

Figure 1. Schematic diagram of Venus-based BiFC. (Click to enlarge)

We searched for mutations that can specifically reduce the spontaneous self-assembly of the two nonfluorescent fragments in the Venus-based BiFC assay. Here we report the identification of a mutation, I152L, that can dramatically reduce the self-assembly between the Venus N- and C-terminal fragments, and thus enable BiFC assay visualization of PPIs with a high S/N ratio in living cells.

Materials and methods

Site-directed mutagenesis and plasmid construction

pMyc and pHA vectors (Clontech, Palo Alto, CA, USA) were used as backbone plasmids, and PCR-ligation-PCR mutagenesis (18) was used for all mutations. For Venus-V150L, Venus-I152L, and VenusT153M, pHA-Venus(A206K) vector was used as a first PCR template. Two distinct mutated cDNA fragments for Venus-V150L were amplified by PCR with primer sets: (5′-AATTGTACCCGCGGGCCCACC-3′/5′-GTCGGCGGTGATATAGAGGTTGTGGCTGTTG-3′) and (5′-CAACAGCCACAACCTCTATATCACCGCCGAC-3′/5′-CACTGCATTCTAGTTGTGGTTTG-3′). For Venus-I152L, the following primer sets were used for the first PCR: (5′-AATTGTACCCGCGGGCCCACC-3′/5′-CTGCTTGTCGGCGGTGAGATAGACGTTGTGG-3′) and (5′-CCACAACGTCTATCTCACCGCCGACAAGCAG-3′/5′-CACTGCATTCTAGTTGTGGTTTG-3′). For Venus-T153M, the following primer sets were used for the first PCR: (5′-AATTGTACCCGCGGGCCCACC-3′/5′-GCCGTTCTTCTGCTTGTCGGCCATGATATAGACGTTG-3′) and (5′-CAACGTCTATATCATGGCCGACAAGCAGAAGAACGGC-3′/5′-CACTGCATTCTAGTTGTGGTTTG-3′). The result-ing two cDNA fragments were mixed and used as a second PCR template. To obtain a full-length cDNA encoding Venus, a second PCR was performed with a primer set: (5′-AATTGTACCCGCGGGCCCACC-3′/5′-CACTGCATTCTAGTTGTGGTTTG-3′). To make VN155-V150L, VN155-I152L, and VN155-T153M, pMyc-VN155 vector was used as a PCR template. cDNA fragments for VN155-V150L, VN155-I152L, and VN155-T153M wereamplified by PCR with respective primer sets: (5′-AATTGTACCCGCGGGCCCACC-3′/5′-ATATGCGGCCGCTCAGGCGGTGATATAGAGGTTGTGGCTG-3′), (5′-AATTGTACCCGCGGGCCCACC-3′/5′-ATATGCGGCCGCTCAGGCGGTGAGATAGACGTTGTGGCTG-3′) and (5′-AATTGTACCCGCGGGCCCACC-3′/5′-ATATGCG GCCGCTCAGGCCATGATATAGACGTTGTGGCTG-3′). The amplified cDNA fragments were cloned into the pMyc vector digested with KpnI/NotI and confirmed by DNA sequencing. For constructions of bJun fusion proteins, the cDNA fragment encoding bJun was subcloned into the pMyc vector digested with EcoRI/KpnI. For constructions of bFos and ΔbFos fusion proteins, cDNA fragments encoding bFos and ΔbFos were subcloned into the pHA-VC155(A206K) vector digested with EcoRI/NotI (5).

Fluorescence intensity and S/N ratio

COS-1 cells cultured in 12-well plates were transfected with appropriate plasmids (0.25 µg each) as described previously (5). Fluorescent images were similarly acquired 24 h post-transfection with C (EX430/25, EM470/30), Y (EX500/20, EM535/30) and R (EX572/35, EM632/60) filters (5). To calculate the S/N ratio of BiFC, the bZIP domain of c-Jun (bJun, amino acid residues 257–313)/the bZIP domain of c-Fos (bFos, amino acid residues 118–211) and bJun/ΔbFos (a mutant version of bFos with a deletion of residues 179–193) (19) were used as positive and negative interaction pairs, respectively. Cerulean (20) and RFP-Q66T (8) were used as controls for normalization of the protein expression level of Venus and Cerulean-BiFC, respectively. bJun was fused to the N-terminal fragment of Venus (e.g., VN155), and bFos or ΔbFos was fused to the C-terminal fragment of Venus (e.g., VC155). To quantify BiFC efficiency, fluorescence intensity of BiFC complex was divided by the intensity of the control protein (Cerulean or RFP-Q66T) to normalize the transfection and protein expression levels. More than 50 cells from each experiment were quantified, and the median was presented. For calculation of the S/N ratio, a median of the BiFC efficiency with the positive interaction was further divided by a median of the BiFC efficiency with the negative interaction. All experiments were performed at least three times, and error bars represent standard error of the fluorescence intensity (BiFC efficiency) or the S/N ratio for three independent experiments. Student's t-test was performed by using GraphPad Prism 5 (La Jolla, CA, USA). ImageJ ( and Microsoft Excel (Redmond, WA, USA) were used for measurement of fluorescence intensity and calculation of the S/N ratio, respectively.

Emission and excitation spectra

Measurements of emission and excitation spectra peaks were previously reported (5). COS-1 cells expressing Venus, bJunVN155/bFosVC155, or bJunVN155-I152L/bFosVC155 were subjected to spectra measurement by using the FluoroMax-3 spectrofluorometer (HORIBA, Tokyo, Japan).

Temperature resistance

Cells cultured at 37°C for 24 h after transfection were treated with 20 ng/mL cyclo-heximide for 30 min to inhibit protein synthesis and then were incubated at 37°or 30°C for 2 h, followed by fixation with 3.7% formaldehyde for 30 min. The fluorescence intensity was then measured.

Immunoblotting analysis

Transfected cells were harvested after image acquisition and lysed in 40 µL 1× Laemmli sample-loading buffer. One-third of the total lysate from one well was subjected to SDS-PAGE analysis (10% gel), followed by transfer to a nitrocellulose membrane. Myc-, hemagglutinin (HA)-, and Flag-tagged proteins were detected by immuno-blotting analysis with anti-Myc (Clontech), anti-HA, and anti-Flag antibodies (Sigma-Aldrich, St. Louis, MO, USA), respectively. A horseradish peroxidase–conjugated anti mouse antibody (GE Healthcare, Uppsala, Sweden) was used as a secondary antibody.

Results and discussion

Three pairs of N- and C-terminal fragments—VN155/VC155, VN173/VC173, and VN173/VC155—have been used for the Venus-based BiFC assay (5). Because the VN155/VC155 pair showed higher S/N ratio and fluorescence intensity when compared with the VN173/VC173 and VN173/VC155 pairs (Supplementary Figure S1), we chose the VN155/VC155 for improvement in this study. To know false-positive fluorescence in the VN155/VC155-based BiFC assay, we performed a VN155/VC155-based BiFC assay in COS-1 cells. When the VN155 and VC155 fragments were only coexpressed in COS-1 cells, bright fluorescence could be detected (Figure S2). Likewise, coexpression of a negative PPI pair, basic region leucine zipper (bZIP) domain of c-Jun (bJun) and an interaction-defective bZIP domain of c-Fos (ΔbFos) fused to VN155 and VC155, respectively, also produced readily detectable fluorescence (Figure S2). Such false-positive fluorescence was estimated to be 20–50% of the fluorescence derived from the positive PPI pair (bJunVN155 and bFosVC155) (Figure S2), and the S/N ratios of Venus-based BiFC assays were in a range of 3–7 (Figure S1). Unless specified, VN and VC refer to VN155 and VC155 fragments, respectively, throughout the paper.

Because the structural information of BiFC complexes is not available, we utilized the crystal structure of the Venus protein to predict mutations that may reduce the self-assembly of the N- and C-terminal fragments of Venus. Based on the crystal structure and topology of Venus (21), the 7th β-strand in VN is associated with two β-strands, 8th and 10th, in VC in the BiFC assay (Figure 1). These associations are mediated by hydrogen bonds of the backbones and possibly hydrophobic effect of the side chains of the amino acid residues (22) (Figure S3), thereby contributing to the self-assembly between VN and VC. If this were the case, amino acid substitutions within the 7th β-strand to disrupt some of these interactions would weaken the strength of the association. To be useful for BiFC assays, the same mutations should not affect the complementation efficiency driven by positive PPIs. To attain this aim, we sought to reduce hydrophobicity of the 7th β-strand without significant alteration of the overall structure. Based on the prediction of solvent accessibility of these residues (Figure 2A), valine 150 and isoleucine 152 were selected as candidate residues for the substitutions (Figure 2A). Valine and isoleucine are hydrophobic amino acids, and their structures are closely similar to that of leucine, although both valine and isoleucine prefer to form a β-strand whereas leucine prefers to form an α-helix (23). In addition, the hydropathy index of the leucine side chain is lower than that of valine and isoleucine (24). Thus, we predicted that a V150L or I152L mutation could reduce the self-assembly between VN and VC without significantly altering the overall structure (Figure 2B). We also tested the T153M mutation, which was reported to reduce the spontaneous self-assembly between VN and VC in Xenopus (17).

  1    2    3