Split Green Fluorescent Proteins: Scope, Limitations, and Outlook

Abstract

Many proteins can be split into fragments that spontaneously reassemble, without covalent linkage, into a functional protein. For split green fluorescent proteins (GFPs), fragment reassembly leads to a fluorescent readout, which has been widely used to investigate protein-protein interactions. We review the scope and limitations of this approach as well as other diverse applications of split GFPs as versatile sensors, molecular glues, optogenetic tools, and platforms for photophysical studies.

Keywords: bimolecular fluorescence complementation; biosensor; green fluorescent protein; photochemistry; split protein.

Figure 1

General schematic of protein-fragment complementation assays. The engineered split protein fragments X and Y (shown in shades of red) are genetically fused to two proteins whose interaction is of interest (proteins A and B, shown in shades of green). Upon interaction of proteins A and B, the effective concentration of the split protein fragments increases such that fragments X and Y form a noncovalently bound complex and regain native activity, creating the assay’s protein–protein interaction–dependent readout.

Figure 2

Green fluorescent protein (GFP) structure and topology. (a) Ribbon structure of GFP (PDB ID: 2B3P) (103) highlighting the chromophore environment and the proximity of the N- and C-termini. The internal α-helix that contains the chromophore is shown in green, while β-strands 4, 7, 10, and 11 are shown in black, orange, blue, and red, respectively. (b) Topology of GFP’s 11 β-strands and internal α-helix (ih). Figure adapted with permission from Reference .

Figure 3

Photophysical properties of the green fluorescent protein (GFP) chromophore. (a) Absorbance (black) and fluorescence emission (green) spectra of superfolder GFP (S65) (59). The protonated A state and deprotonated B state absorb at 393 and 467 nm, respectively. (b) Several factors influence the structure and photophysical properties of the GFP chromophore. Mutations to Y66 and nearby residues can tune chromophore absorption and fluorescence across the visible spectrum. Modulating the chromophore’s pK_a (acid dissociation constant) is beneficial for various microscopy applications and biosensor development. In most reversibly switchable fluorescent proteins, the chromophore isomerizes from the fluorescent cis to the nonfluorescent trans conformation when irradiated with blue light. The chromophore then undergoes either thermal relaxation or violet light–driven isomerization back to its original state. Finally, the chromophore can convert to a red fluorescent species from a green fluorescent precursor (termed photoconversion) or convert to a fluorescent species from a nonfluorescent precursor (termed photoactivation; not shown).

Figure 4

Schematic diagram illustrating split protein reassembly between recombinant GFP1–10 and a synthetic GFP11 peptide with subsequent chromophore maturation (PDB ID: 2B3P) (103). Mutations at E222 tune the photophysical properties of the chromophore. Note that the 3D structure of the truncated protein shown in gray is not currently known. Figure adapted with permission from Reference .

Figure 5

Schematic illustrating the required experimental steps for the synthetic control of green fluorescent protein, applied to either (a) the strand 7, 10, or 11 system or (b) the internal α-helix system. The loop between the terminal secondary structural element and the rest of the protein is proteolytically cleaved. The noncovalently bound complex is denatured, and the fragments are separated with size exclusion chromatography. Refolding of the larger fragment, referred to as the denatured truncated protein (shown in gray as if folded), with synthetic peptide (shown in red) corresponding to the missing structural element yields a fluorescent species resembling the native protein. Point mutations on the synthetic peptides that cause color shifts and/or protonation state changes can be introduced in this manner. Figure adapted with permission from Reference .

Figure 6

Strand binding and photodissociation of the strand-10 circularly permuted split green fluorescent protein (GFP). (a) When refolded in vitro, the truncated protein with strand 10 removed can bind synthetic peptides similar to strand 10. If the peptide contains the T203Y mutation responsible for the yellow color of yellow fluorescent protein (YFP), addition of aliquots of this peptide leads to a green-to-yellow color shift that signifies binding. Note that the structure of the truncated protein shown as a gray barrel is not currently known, although it does contain a mature chromophore. (b) In the presence of excess synthetic peptide containing the T203Y mutation, the proteolytically cleaved but still noncovalently attached native (T203) strand 10 does not dissociate spontaneously but does dissociate upon irradiation with blue light. The strand exchange is evident by a shift in the absorption spectrum similar to that observed by direct addition in panel a. Figure adapted with permission from Reference .

Figure 7

(a) Structure (PDB ID: 6B7R) (25) and (b) light activation of the strand-11 truncated protein. The N-terminal His tag (shown in gray) binds as a new β-strand to the vacant spot left by removal of the native 11th strand. Light irradiation displaces the bound His tag and allows for binding of added synthetic peptides (shown in green). Figure adapted with permission from Reference .

Figure 8

Two-tailed version of green fluorescent protein (GFP). (a) Cartoon illustrating the design of the two-tailed GFP. The rest of the barrel (strands 11 through 9) are flanked by two strand-10 peptides connected by linkers. A point mutation at position 203 on strand 10 leads to either a green (T203) or yellow (T203Y) fluorescent protein when bound. (b) Application of the two-tailed GFP engineered as a light-activated ratiometric protease sensor by combining light-driven photodissociation of the cut strand and strand replacement. Figure adapted with permission from Reference .

Figure 9

Potential energy curves for photodissociation, highlighting relevant parameters. Branching points are shown as circled numbers and are color coded to match their associated processes: ❶ represents the excited-state barrier partitioning fluorescence and isomerization; ❷ represents the photochemical funnel, which divides aborted from successful isomerization; and ❸ represents branching of strand dissociation and thermal relaxation. Abbreviations: C, noncovalently bound fluorescent protein complex with a cis chromophore in the ground electronic state; C’, noncovalently bound fluorescent protein complex with a trans chromophore in the ground electronic state; D, truncated protein containing a trans chromophore in the ground electronic state with the dissociated strand removed; P, dissociated strand (peptide); γ, population branching ratio at the photochemical funnel; E_a,diss, energy barrier for thermal strand dissociation; E_a,fwd, energy barrier in the excited-state to reach photochemical funnel; FC, Franck-Condon excitation from the ground- to the excited-state; k_BC, rate constant for excited-state barrier crossing; k_diss, rate constant for strand dissociation; k_FI, rate constant for fluorescence; k_IC, rate constant for nonradiative internal conversion without passing through the photochemical funnel; k_th, rate constant for thermal relaxation from C’ to C. Figure adapted with permission from Reference .