Engineering a new tripartite split-ccGFP system from Corynactis californica for detecting protein-protein interactions

Abstract

Protein-protein interactions (PPIs) are critical to a range of biological processes and, consequently, aberrant interactions are implicated in many disorders. The study of the complex networks of PPIs promises to elucidate undiscovered roles in cellular processes and the mechanisms of disease. To accomplish this, tools to effectively sense PPIs are necessary. Effective PPI sensors must rapidly detect interactions in real-time with high sensitivity without perturbing the proteins of interest (POIs) under study. Split fluorescent proteins have previously been used to successfully monitor PPIs, in part due to the small size of the tags. Here, we developed an optimized tripartite split GFP system based on Corynactis californica GFP (ccGFP) to detect PPIs in vitro. In this sensor system, ccGFP fragments ccGFP10 and ccGFP11 are tagged to two POIs. PPIs can then be detected via fluorescence by complementation to the third fragment, ccGFP1-9, which reconstitutes functional ccGFP. The optimized ccGFP system shows improved detection kinetics and pH and temperature stability compared to a previous system. We then validated the sensor by monitoring PPIs in two model systems: attractive/repulsive coiled-coils and rapamycin-inducible FRB/FKBP heterodimerization. Finally, we developed an anti-tripartite ccGFP single-chain variable fragment (scFv), which could enable versatile detection of identified protein-protein complexes.

Keywords: Antibody; Directed evolution; Green fluorescent protein (GFP); Protein detection; Protein engineering; Protein fragment complementation; Protein tagging; Protein-protein interactions; Single-chain variable fragment (scFv); Split fluorescent protein.

Fig. 1

A schematic diagram showing the principle of the tripartite split ccGFP system in detecting protein-protein interactions similar to the previously reported tripartite split GFP. Proteins A and B are fused to ccGFP10 and ccGFP11 fragments, respectively. If proteins A and B interact, upon complementation with the detector fragment, ccGFP1-9, they will reconstitute functional ccGFP, causing a fluorescent signal (top). If proteins A and B do not interact, there will be no fluorescent signal, as functional ccGFP will not associate together (bottom). Individual β fragments (represented throughout as cylinders) do not individually fluoresce. Only when the 11 β-strands of ccGFP associate into a β-barrel does the complex generate fluorescence (right).

Fig. 2

Complementation of the tripartite split ccGFP in vitro and in vivo. (a) Complementation curves of ccGFP1-9 OPT with the engineered ccGFP10-SR-ccGFP11 fusion protein (light green) and the starting point ccGFP1-9 with the ccGFP10-SR-ccGFP11 starting point fusion protein (dark green). (b) Comparing complementation kinetic rates of engineered tripartite split ccGFP with the engineered tripartite split GFP system. Fluorescence was measured every 3 min for 15 h. Points show average of n = 2 measurements, with error bars ± 1 S.D. (c) In vivo solubility screen of 8 Pylobacterium test proteins expressed in a ‘‘sandwich’’ configuration with N-terminal ccGFP10 and C-terminal ccGFP11 (i.e., ccGFP10-POI-ccGFP11, POI: protein of interest, from 1 to 8, see Table S3) from the pTET-GFP10/11 plasmid assayed with either ccGFP1-10 (top) or ccGFP1–9 OPT (bottom) expressed from a pET vector in BL21(DE3) E. coli. Fluorescence pictures are of E. coli colonies on plates after 1.5 h Antet induction (expressing the “sandwich”) followed by IPTG induction (expressing the ccGFP1-9 OPT or ccGFP1-10) (sequential induction) (SEQ) or 3 h co-induction (CO). Legend (bottom) indicates solubility of the tagged proteins determined by SDS-PAGE [as a gradient from soluble (green; soluble fraction 1.0) to insoluble (red; soluble fraction 0.0)]. See Cabantous, et al., Nguyen, et al., and Table S3 for additional information regarding protein solubility determination.

Fig. 3

Characterization of protein-protein interactions using coiled-coil heterodimerization. (a) Kinetics of heterodimerization of E1-ccGFP11 with detector ccGFP1-9 and either ccGFP10-E1 (orange) or ccGFP10-K1 (green). 100 nM of E1-ccGFP11 and 6 µM of either ccGFP10-E1 or ccGFP10-K1 was incubated together with 8 µM ccGFP1-9. Fluorescence was measured every 3 min for 15 h. (b) Response after 1 h of coiled-coil heterodimerization as E1-ccGFP11 concentration was varied with constant concentrations of ccGFP10-K1 (green, labeled K/E) or ccGFP10-E1 (orange, labeled E/E) and ccGFP1-9. E1-ccGFP11 was varied from 12.5 to 1600 nM, while ccGFP1-9 was at 8 µM and ccGFP10-K1 or ccGFP10-E1 was kept at 6 µM. Black lines show the line of best-fit from linear regression, with equations and R² values shown. Points display average of n = 2 measurements, with error bars showing ± 1 S.D.

Fig. 4

Application of the tripartite split ccGFP system to study the rapamycin-inducible FRB/FKBP12 interaction. (a) The ccGFP10 and ccGFP11 tags were fused to FRB and FKBP12 proteins, respectively. Rapamycin (RAP) ligand binding brings both protein fusions into proximity, permitting ccGFP fluorescence reconstitution upon addition of ccGFP1-9 OPT (top). Without rapamycin ligand binding, ccGFP does not reconstitute, resulting in limited fluorescence signal (bottom). (b) Fluorescence progress curves for ccGFP1-9 OPT complementation with soluble extracts of ccGFP10-FRB and FKBP12-ccGFP11 fusions in the presence or absence of rapamycin. (c) Effect of altering the concentrations of ccGFP10-FRB and FKBP12-ccGFP11 (320, 160, and 40 nM each) initiated by the addition of rapamycin to 150 nM final concentration [(+) RAP, red] and no rapamycin [(-) RAP, black]. (d) Rapamycin dose response curve (0.1 to 2500 nM) for FRB/FKPB12 binding in vitro, measured as final fluorescence after 1 h. The black line shows a model fit to the Hill equation. As appropriate, points/bars show average of n = 2 measurements, with error bars showing ± 1 S.D.