Engineering of a fluorescent chemogenetic reporter with tunable color for advanced live-cell imaging

Abstract

Biocompatible fluorescent reporters with spectral properties spanning the entire visible spectrum are indispensable tools for imaging the biochemistry of living cells and organisms in real time. Here, we report the engineering of a fluorescent chemogenetic reporter with tunable optical and spectral properties. A collection of fluorogenic chromophores with various electronic properties enables to generate bimolecular fluorescent assemblies that cover the visible spectrum from blue to red using a single protein tag engineered and optimized by directed evolution and rational design. The ability to tune the fluorescence color and properties through simple molecular modulation provides a broad experimental versatility for imaging proteins in live cells, including neurons, and in multicellular organisms, and opens avenues for optimizing Förster resonance energy transfer (FRET) biosensors in live cells. The ability to tune the spectral properties and fluorescence performance enables furthermore to match the specifications and requirements of advanced super-resolution imaging techniques.

Fig. 1. The engineered protein tag pFAST is a fluorescent chemogenetic reporter with tunable color.

a Structures of the chromophores of the HBO, HBT, HBP, HBR, HBIR, and HBRAA series positioned according to the wavelengths of maximal emission and maximal absorption of their assembly with pFAST. The fluorescence quantum yield (ϕ), molar absorptivity at maximal absorbance (ε), and thermodynamic dissociation constant (K_D) values of their assembly with pFAST are indicated into brackets below the molecules. b Normalized absorption (left) and fluorescence (right) spectra of the different fluorogenic chromophores bound to pFAST. Spectra were recorded in pH 7.4 PBS at 25 °C. Fluorescence spectra were measured by exciting at the maximal absorption wavelength. c Solutions containing the different pFAST:chromophore assemblies illuminated with 365 nm light. Source data for spectra are provided as a Source Data file.

Fig. 2. Engineering of promiscuous FAST variants.

a Directed evolution allowed the generation of oFAST, tFAST, and pFAST. Yeast-displayed libraries of variants of FAST were screened in presence of the indicated fluorogenic chromophore by fluorescence-activating cell sorting (solid arrows). Additional mutations were introduced by rational design (dotted arrows). b–d K_Ds (thermodynamic dissociation constants) and fluorescent quantum yields (ϕ) of the clones isolated from the selections with b HBO-3,5DM, c HBT-3,5DM, and d HBP-3,5DM. Filled triangles are variants from library A; black squares are variants from library B; gray triangles are the variants used for the generation of the DNA shuffling library B and unfulfilled triangles and squares are variants generated by rational design. Values are also given for FAST for comparison. e–g Confocal micrographs of HeLa cells expressing cytoplasmic e oFAST labeled with 10 μM HBO-3,5DM, f tFAST labeled with 5 μM HBT-3,5DM and g pFAST labeled with 5 μM HBP-3,5DM. Experiments were repeated 1 time (e, f) and three times (g) with similar results. Scale bars 10 μm. See Supplementary Table 12 for imaging settings. h K_Ds (thermodynamic dissociation constants) and emission wavelength of the different fluorogens in presence of FAST variants. Fluorescent quantum yields are given in %. The circle diameter reflects the value of the fluorescence quantum yield, while the different coloring systems are used to tell apart the different variants. Source data for graphs are provided as a Source Data file.

Fig. 3. Structural model of pFAST.

a Structural model of pFAST and FAST generated by homology modeling using the crystal structure of the Halorhodospira halophila Photoactive Yellow Protein PYP (PDB: 6P4I) (See also Supplementary Fig. 6a). Mutations of pFAST relative to FAST are indicated. b Root Mean Square Fluctuation (RMSF) of the residues within pFAST and FAST during molecular dynamic simulations. c Polar and apolar interactions network involved in HBP-3,5DM binding and recognition within pFAST. Source data for graphs are provided as a Source Data file.

Fig. 4. Selective imaging of pFAST fusions in various cellular localizations in mammalian cells and cultured neurons.

a Confocal micrographs of HeLa cells expressing pFAST fusions (either with H2B or to a transmembrane domain) in presence of the entire set of fluorogenic chromophores. Scale bars 10 μm. Experiments were repeated > 3 times with similar results. b Confocal micrographs of life and fixed HeLa cells expressing pFAST fused to: histone H2B, lyn11 (inner membrane-targeting motif), LifeAct (actin-binding peptide domain), mito (mitochondrial targeting motif), and to microtubule-associated protein (MAP) 4 and labeled with 5 μM HBP-3,5DM. Note that control experiments showed that the mislocalization of MAP4-pFAST observed in fixed cells (*) was due to MAP4 fixation rather than pFAST (see also Supplementary Fig. 12). Experiments were repeated >3 times with similar results. Scale bars, 10 μm. c Depth color coded three-dimensional reconstruction (from 81 optical sections over 12.8 μm) of live HeLa cells expressing lyn11-pFAST fusion and labeled with 5 μM HBR-3,5DOM. d Confocal micrographs of live dissociated hippocampal neurons expressing pFAST fused to lyn11, LifeAct, and MAP4 and labeled with 10 μM HBR-3,5DOM. Experiments were repeated >6 times with similar results. Scale bars, 50 μm. The Lyn11-pFAST image results from the maximum intensity projection of five optical sections. a–d See Supplementary Table 13 for imaging settings.

Fig. 5. Selective imaging of pFAST fusions in chicken embryo.

a–g Plasmids encoding H2B-pFAST and b mKO, c, d EYFP, e, f EGFP, and g mCerulean fused to H2B were electroporated in each side of the neural tube in ovo at embryonic day 2 (E2, HH stage 13–14). mCherry or EGFP reporters were co-injected with each construct to monitor electroporation efficiency. 24 h later, embryos with homogeneous bilateral reporter expression in the neural tube were dissected and imaged before and after addition of the indicated fluorogenic chromophore. Experiments were repeated two times with similar results. h–j mito-pFAST (mitochondrial targeting motif) and memb-iRFP670 (membrane-targeting motif); i H2B-pFAST and pact-mKO (PACT domain of pericentrin) and j H2B-pFAST, pact-mKO, and mb-iRFP670 were electroporated in the neural tube in ovo, at embryonic day 2 (E2, HH stage 13–14). 24 h later, embryos were dissected, and imaged in presence of h 1 μM HBR-3,5DOM, i 5 μM HBP-3,5DOM, and j 1 μM HMBR using a spinning disk confocal microscope. Time-lapse showing cell division presented correct localization of the proteins (see also Supplementary Movies 2–4). Experiments in h–j were repeated 3, 2, and 2 times with similar results, respectively. Scale bars, 10 μm. a–j See Supplementary Table 13 for imaging settings.

Fig. 6. Fluorescence lifetime imaging of mTurquoise2 and EGFP donors using pFAST as acceptor in FRET experiments.

a Model illustrating the FRET experiments of mTurquoise2-pFAST and EGFP-pFAST tandems before and after addition of HBR-3,5DOM or HBIR-3M chromophores. b–e Representative fluorescence (left panels) and relative lifetime images (right panels) of live U2OS cells expressing b, c mTurquoise2-pFAST and d, e EGFP-pFAST before and after addition of imaging buffer, 10 μM of HBR-3,5DOM or 10 μM of dark chromophore HBIR-3M (mTurquoise2 and EGFP channels). Experiments were repeated eight times with similar results. The box plots in c and e show the variation of the lifetime of the donor (ΔLifetime) calculated per cell after addition of chromophores (n = 8 cells). Whiskers represent the highest and lowest values. f Model illustrating the mode of action of the autophosphorylated AURKA biosensor. The complete sequence of AURKA is located between the donor (mTurquoise2) and the acceptor (pFAST). When AURKA is autophosphorylated on Thr288, the kinase brings mTurquoise2 and pFAST in close proximity allowing FRET detection after addition of chromophores. Of note, the real three-dimensional orientations of the two reporters are not known. g, h Representative fluorescence (left panels) and relative lifetime images (right panels) of live U2OS cells synchronized in mitosis expressing mTurquoise2-AURKA-pFAST after g sequential addition of 1 μM of HBR-3,5DOM and 10 μM of dark-competitor HBIR-3M and h simple addition of 10 μM HBIR-3M (mTurquoise2 channel). Experiments were repeated six times with similar results. i–k The graphs show mTurquoise2 donor i relative lifetime per cell (median values are reported) and j, k ΔLifetime calculated per cell after addition of chromophores (n = 6 cells, box plots with whiskers representing the highest and lowest values). b, d, g, h All scale bars are 10 μm. The artificial ‘fire’ color represents pixel-by-pixel lifetimes. See Supplementary Table 13 for imaging settings. Source data for graphs are provided as a Source Data file.

Fig. 7. STED nanoscopy of pFAST-tagged proteins in live mammalian cells.

a–h Confocal and STED micrographs of live HeLa cells expressing a–d Lyn11-pFAST and e–h MAP4-pFAST. The graphs d, h show gain in resolution. Experiments with Lyn11-pFAST and MAP4-pFAST were repeated ten and 15 times, respectively, with similar results. i–k Two-color confocal and STED micrographs of live HeLa cells expressing MAP4-pFAST (green) and labeled with 1 μM Silicon-Rhodamine (SiR)-actin (magenta) using a single 775 nm depletion laser. Experiment was repeated ten times with similar results. Cells were treated with 10 μM HBR-3,5DOM before imaging. All scale bars are 5 μm (see Supplementary Table 13 for imaging settings). Source data for graphs are provided as a Source Data file.

Fig. 8. STED nanoscopy of pFAST-tagged proteins in live neurons and astrocytes.

a, b Confocal and STED micrographs of live dissociated hippocampal neuron expressing LifeAct-pFAST. c, d Confocal and STED micrographs of a neurite growth cone expressing LifeAct-pFAST. e–g Confocal and STED micrographs of a live astrocyte expressing MAP4-pFAST. Cells were treated with 10 μM of HBR-3,5DOM before imaging. Experiments with LifeAct-pFAST and MAP4-pFAST were repeated 10 and 20 times, respectively, with similar results. The graphs show gain in resolution. All scale bars are 5 μm, except the one in e that is 20 μm. a–g See Supplementary Table 13 for imaging settings. Source data for graphs are provided as a Source Data file.