Association of Fluorescent Protein Pairs and Its Significant Impact on Fluorescence and Energy Transfer

Abstract Fluorescent proteins (FPs) are commonly used in pairs to monitor dynamic biomolecular events through changes in proximity via distance dependent processes such as Förster resonance energy transfer (FRET). The impact of FP association is assessed by predicting dimerization sites in silico and stabilizing the dimers by bio‐orthogonal covalent linkages. In each tested case dimerization changes inherent fluorescence, including FRET. GFP homodimers demonstrate synergistic behavior with the dimer being brighter than the sum of the monomers. The homodimer structure reveals the chromophores are close with favorable transition dipole alignments and a highly solvated interface. Heterodimerization (GFP with Venus) results in a complex with ≈87% FRET efficiency, significantly below the 99.7% efficiency predicted. A similar efficiency is observed when the wild‐type FPs are fused to a naturally occurring protein–protein interface system. GFP complexation with mCherry results in loss of mCherry fluorescence. Thus, simple assumptions used when monitoring interactions between proteins via FP FRET may not always hold true, especially under conditions whereby the protein–protein interactions promote FP interaction.


Construction and production of Bcl3-sfGFP/Venus-P50 complex
The sfGFP-Bcl3 and p50-venus (see above for constructs) were cloned into pET11a and the pCA24N vectors, respectively. The two plasmids where cotransformed by electroporation into E.coli BL21 (DE3) (NEB), which where used to inoculate 1L flasks of 2xYT (Melford) supplemented with 100 µg/mL carbenicillin and 35 µg/mL chloramphenicol. Cultures were grown at 37 °C until an OD600 of 0.6 was achieved, and protein expression induced by addition of 0.4 mM IPTG. The cells where grown overnight at 27 °C. Cells were harvested and were lysed using a French press. The resulting lysate was clarified by centrifugation and loaded onto a 5 mL HisTrapHP TM (GE Healthcare) column equilibrated in equilibration-wash buffer (20 mM HEPES, 300 mM NaCl, 10 mM imidazole, pH8.0) then allowed to drain by gravity. Bound uncomplexed protein (sfGFP-Bcl3 and p50-venus) were eluted by washing the column with 250 mM imidazole. Complex (Bcl3-sfGFP/p50-Venus) was eluted by washing the column with 500 mM imidazole. Samples containing the complex were then loaded onto a HiLoad TM 26/600Superdex TM S200 pg column equilibrated in 20 mM HEPES, 300 mM NaCl, pH8.0. Purity was checked via SDS-PAGE analysis. Fluorescence analysis was performed as outlined in the main manuscript.
Construction and production of mCherry variants. The gene encoding wt mCherry resident within the pBAD plasmid was used to generate the K198TAG mutation by whole plasmid site-directed mutagenesis (Forward primer 5'-CGGCGCCTACAACGTCAACATCTAGT-3' and reserve primer 5'-GGCAGCTGCACGGGCTTCTT-3') using Q5 polymerase (NEB, USA). E. coli Top10 cells were transformed and used to inoculate 1L flasks of LB media supplemented with 100 µg/mL ampicillin, 25 µg/mL tetracycline. To incorporate azF, cells were also co-transformed with pDULEcyanoRS and cultured in the presence of 0.1 mM azF. Cultures were grown for 1 hour at 37°C in a shaking incubator before expression was induced by addition of 0.1% of arabinose and incubated for 24 hours at 25°C. Cultures were kept in the dark until after dimerisation with SCO-K containing variants.
Cells were harvested via centrifugation and resuspended in 20 mL of 50 mM Tris-HCl pH8.0, 1 mM EDTA. The cells were lysed using a French press and the resulting lysate was clarified by centrifugation. Cell lysates were then loaded onto a 5 mL HisTrapHP™ (GE Healthcare) equilibrated in lysis buffer. Bound protein was eluted by washing the column in 250 mM Imidazole. Samples were then loaded onto a Superdex 75 column equilibrated in 50 mM Tris-HCl pH8.0 and purity was checked via SDS-PAGE analysis. Concentrations of monomer variants were determined using the Bio-RAD DC Protein Assay using wild type wt mCherry as a standard and correlated to the 280 nm absorbance. The quantum yield of mCherry 198azF was calculated as described previously (1) using WT mCherry as the reference sample.
Conjugation of mCherry with non-proteinaceous molecules. Conjugation of mCherry 198azF with the ncAA SCO-K was performed in 50 mM Tris-HCl pH 8.0) with 5 µM protein and 200 µM of ncAA. Samples were left for 4 hr at room temperature and analysed by absorbance and fluorescence spectroscopy. Conjugation of mCherry 198azF with Cy3 DBCO (Click Chemistry Tools, USA) was performed with equimolar concentrations of protein and dye (5 µM). The absorbance and fluorescence emission were recorded immediately after sample mixing. The sample was then left at room temperature overnight. Following overnight incubation, the absorbance and fluorescence were performed again, and the reaction mix was run on SDS page gel.  Figure S1. Representative sample of dimeric sfGFP 204x2 single molecule time course traces (raw and Cheung-Kennedy filtered) coupled with paired intensity frequency histograms (generated from Cheung Kennedy filtered data) to the right of each trace.

Supporting
Supporting Figure S2. Structural analysis of sfGFP 204x2 . (a) Overlay of the two observed dimers in 5NI3 with the WT sfGFP (grey). The Ca RMSD of WT sfGFP with the azF monomer is 0.193 Å and 0.165 Å with the SCO-K monomer. (b) Overlay of chromophores from sfGFP WT (grey), the sfGFP 204azF monomer unit (green) and the sfGFP 204azF monomer unit (cyan). (c) Overlay of best fitting model calculated previously (6) (coloured grey) and the determined structure of sfGFP 204x2 (orange). Overlap of residues involved in the triazole crosslink are shown to the right as spheres. The Ca RMSD between the model and determined structure is 5.6 Å. Figure S3. Long range tunnels and bond networks in sfGFP 204x2 . (a) CAVER (9) analysis of sfGFP 204x2 . The tunnel linking the CROs are coloured magenta. (b) Long range H-bond network linking the two CROs. The subscript to the residues denotes the monomer the residues are from. The sfGFP 204azF monomer is coloured green and the sfGFP 204SCO is coloured cyan. Additional information. Analysis of sfGFP 204x2 using CAVER 21 suggests that a tunnel linking the two CROs is present (Figure 5a). The pathway extends beyond CRO SCO to exit around residues 23, 54 and 136. An extended putative interaction polar network involving the SCO and triazole moieties, water and amino acids spans the cavity and links the two CROs. Compared to the GFP 148x2 , the network is not as coherent or direct. Figure S4. Water molecules (red spheres) closely associated with the CRO (grey sticks) in sfGFP WT monomer.

Supporting
Supporting Figure S5. B-factor changes on dimerisation. B-factors are represented in the "putty" format with increased thickness corresponding to increase in B-factor and thus apparent flexibility. Shown is sfGFP WT together with the monomer units (sfGFP 204azF and sfGFP 204SCO ) of the sfGFP 204x2 dimer. The regions shown comprise the dimer interface. Selected residues important for function are also shown.   Figure S10. (a) Calculating k 2 through chromophore dipole vector positioning using the equation k 2 = (Cos T -3Cos DCos A ) 2 (12) . The dipole orientations for the chromophores have been reported previously (13,14) and are shown in Figures 4 and S6.