Development of an miRFP680-Based Fluorescent Calcium Ion Biosensor Using End-Optimized Transposons

Abstract

The development of new or improved single fluorescent protein (FP)-based biosensors (SFPBs), particularly those with excitation and emission at near-infrared wavelengths, is important for the continued advancement of biological imaging applications. In an effort to accelerate the development of new SFPBs, we report modified transposons for the transposase-based creation of libraries of FPs randomly inserted into analyte binding domains, or vice versa. These modified transposons feature ends that are optimized to minimize the length of the linkers that connect the FP to the analyte binding domain. We rationalized that shorter linkers between the domains should result in more effective allosteric coupling between the analyte binding-dependent conformational change in the binding domain and the fluorescence modulation of the chromophore of the FP domain. As a proof of concept, we employed end-modified Mu transposons for the discovery of SFPB prototypes based on the insertion of two circularly permuted red FPs (mApple and FusionRed) into binding proteins for l-lactate and spermidine. Using an analogous approach, we discovered calcium ion (Ca²⁺)-specific SFPBs by random insertion of calmodulin (CaM)-RS20 into miRFP680, a particularly bright near-infrared (NIR) FP based on a biliverdin (BV)-binding fluorescent protein. Starting from an miRFP680-based Ca²⁺ biosensor prototype, we performed extensive directed evolution, including under BV-deficient conditions, to create highly optimized biosensors designated the NIR-GECO3 series. We have extensively characterized the NIR-GECO3 series and explored their utility for biological Ca²⁺ imaging. The methods described in this work will serve to accelerate SFPB development and open avenues for further exploration and optimization of SFPBs across a spectrum of biological applications.

Keywords: biliverdin-binding fluorescent protein; cell signaling; directed evolution; fluorescence microscopy; heme oxygenase; protein engineering; transposons.

Figure 1

Generation of linker-shortened libraries using end-modified Mu transposons. (a) Sequences of the end-modified Mu transposon ends, presented with the top strand in the 5′ to 3′ orientation. The chloramphenicol resistance gene (CmR) is shown in black. The efficiency of each end-modified transposon was calculated by comparing colony number, relative to Mu-BsaI, in the primary library. (b) Schematic representation of linker lengths in both the plasmid (above) and the protein structure (below). The target domain is depicted in blue, and the inserted domain is in red. “AA” stands for amino acids. (c) Schematic representation of key steps in the assembly of cpmApple-inserted LldR-LBD libraries. Colors are consistent with panels (a) and (b). Integration of Mu-cpmApple results in a 5 base pair duplication at the target site (underscored). In the representative example shown, P112 is identified as the insertion site, and V113 and Q353 are considered as “linkers” between cpmApple and LldR-LBD.

Figure 2

Characterization of LldR-LBD and PotD Libraries with cpRFP Insertions. (a) Fluorescence change (ΔF/F_min) of variants from cpmApple-inserted LldR-LBD libraries in response to 10 mM l-lactate in screening buffers. Negative ΔF/F_min values indicate an inverse response. Error bars represent ± standard deviation (n = 3). (b) ΔF/F_min values of variants from the cpFusionRed-inserted LldR-LBD library, represented as in (a). (c) Distribution of ΔF/F_min values for functional insertions along the PotD sequence in response to 10 mM spermidine in screening buffers. (d) Positions of insertions (spheres) for functional cpmApple (left) and cpFusionRed (right) insertions into PotD (PDB 1POY). Spermidine is shown in a sphere representation (carbon in green and nitrogen in blue).

Figure 3

Development of NIR-GECO3 via directed evolution. (a) Schematic representation of the structure and response of NIR-GECO3 variants. (b) Relative fluorescent brightnesses of miRFP680 coexpressed with various HO-1 mutants, normalized to wild-type HO-1. (c) Genealogy of NIR-GECO3 variants.

Figure 4

Characterization of NIR-GECO3 variants. (a) Excitation (emission at 710 nm) and emission spectra (excitation at 630 nm) of purified NIR-GECO3.5 protein in the presence (39 μM) and absence of Ca²⁺. (b) Absorbance spectra of purified NIR-GECO3.5 protein in the presence (39 μM) and absence of Ca²⁺. (c) ΔF/F_min as a function of Ca²⁺ concentration for NIR-GECO3 variants. (d) Two-photon excitation spectra of purified NIR-GECO3.5 protein in the presence (39 μM) and absence of Ca²⁺. ΔF/F_min is indicated on the right axis. GM, Goeppert-Mayer units. (e) BV binding kinetics of purified NIR-GECO1, NIR-GECO2, and NIR-GECO3.5 proteins. The peak absorbance (at 680 nm for NIR-GECO1/2 and 660 nm for NIR-GECO3.5) is plotted against time and normalized to the maximum absorbance (obtained by fitting the data to an exponential curve). (f) pH titration of purified NIR-GECO3.5 protein in the presence (39 μM) and absence of Ca²⁺. ΔF/F_min is represented on the right axis. Values are normalized to their maximum. (g) ΔF/F_min of NIR-GECO3 variants in HeLa cells where F is the fluorescence with ionomycin/EGTA (2 μM + 1 mM) and F_min is the fluorescence with ionomycin/Ca²⁺ (2 μM + 4 mM) (n = 8 cells per group). (h) Normalized effective brightness of NIR-GECO3 variants, relative to NIR-GECO2G, measured from approximately 2000 cells transfected with NIR-GECO-P2A-EGFP-3NLS (3 technical replicates) using flow cytometry. Data represents the median value. (i) F_BV+/F_BV- value of NIR-GECO3 variants. The F_BV+ group had 12.5 μM BV added 2 h before measurement. Flow cytometry of approximately 2000 cells transfected as in (h), with brightnesses relative to EGFP.

Figure 5

Two-photon imaging of NIR-GECO3.5. (a) Representative trace of NIR-GECO3.5 response in a HEK293T cell in response to 100 μM ATP, using 1220 nm excitation. (b) Photobleaching curves for NIR-GECO3.5 and iBB-GECO1 in HEK293T cells at both 940 and 1220 nm excitation. Laser power was adjusted so that the obtained fluorescence intensity was similar level (Figure S14c). Areas between thinner dotted lines represent s.e.m (n = 57 for NIR-GECO3.5 at 940 nm, n = 75 for NIR-GECO3.5 at 1220 nm, n = 55 for iBB-GECO1 at 940 nm, n = 55 for iBB-GECO1 at 1220 nm). (c) Representative two-photon excitation (940 nm) images of NIR-GECO3.5 (left) and XCaMP-G (right) expressed in live mouse brain and imaged in slices. Scale bar, 10 μm. (d) Representative time courses of ΔF/F for in vivo imaging of spontaneous neuronal activity with both NIR-GECO3.5 and XCaMP-G. A Savitzky-Golay filter was applied to NIR-GECO3.5, and a median filter was applied to XCaMP-G.