Comparison of spectral FRET microscopy approaches for single-cell analysis

Abstract

Förster resonance energy transfer (FRET) is a valuable tool for measuring molecular distances and the effects of biological processes such as cyclic nucleotide messenger signaling and protein localization. Most FRET techniques require two fluorescent proteins with overlapping excitation/emission spectral pairing to maximize detection sensitivity and FRET efficiency. FRET microscopy often utilizes differing peak intensities of the selected fluorophores measured through different optical filter sets to estimate the FRET index or efficiency. Microscopy platforms used to make these measurements include wide-field, laser scanning confocal, and fluorescence lifetime imaging. Each platform has associated advantages and disadvantages, such as speed, sensitivity, specificity, out-of-focus fluorescence, and Z-resolution. In this study, we report comparisons among multiple microscopy and spectral filtering platforms such as standard 2-filter FRET, emission-scanning hyperspectral imaging, and excitation-scanning hyperspectral imaging. Samples of human embryonic kidney (HEK293) cells were grown on laminin-coated 28 mm round gridded glass coverslips (10816, Ibidi, Fitchburg, Wisconsin) and transfected with adenovirus encoding a cAMP-sensing FRET probe composed of a FRET donor (Turquoise) and acceptor (Venus). Additionally, 3 FRET "controls" with fixed linker lengths between Turquoise and Venus proteins were used for inter-platform validation. Grid locations were logged, recorded with light micrographs, and used to ensure that whole-cell FRET was compared on a cell-by-cell basis among the different microscopy platforms. FRET efficiencies were also calculated and compared for each method. Preliminary results indicate that hyperspectral methods increase the signal-to-noise ratio compared to a standard 2-filter approach.

Keywords: FRET; Fluorescence; Hyperspectral; Microscopy; Signature; Spectral; Spectroscopy.

Figure 1.

Stitched HIFEX and confocal fluorescence images. (Left) The HIFEX image shows all fluorescing molecules excited by the excitation-scanning system. (Right) The confocal image shows all fluorescing molecules excited by the lasers on the confocal system. Each image is a sum of the intensity through the spectral data cube.

Figure 2.

Regions of interest selected for FRET estimation comparison among microscopy platform. (Left) The stitched confocal fluorescence image from Figure 1, provided as a reference for the selected cells. (Right) The same confocal fluorescence image with five cells highlighted for FRET estimation. The selected regions are color-coded for ease of identification. 1. Red. 2. Blue. 3. Green. 4. Magenta. 5. Yellow.

Figure 3.

Zoomed images of the fifth selected region from Figure 2. (Left) A zoomed image of “cell 5” from the HIFEX image. (Center) A zoomed image of “cell 5” from the confocal image. (Right) An image of the same “cell 5” captured using the HIFEX approach with a region of interest drawn to coincide with the cell shown in the confocal image. As the HIFEX method utilizes widefield microscopy, its images potentially include light from out-of-focus cells. By selecting only the regions which also appear in the confocal image, a more appropriate comparison may be made between FRET estimations.

Figure 4.

Linearly unmixed excitation-scanning (HIFEX) spectral images corresponding to the donor and acceptor molecules with respect to the dichroic filter cube used to obtain the images. All pixel intensities were scaled such that the maximum displayed value was 30,000 (of the total available 65,535) to provide a visual representation of the differences in fluorescence intensities among the unmixed images. (Top Left) The unmixed donor image acquired using the 458 nm dichroic filter cube. (Bottom Left) The unmixed acceptor image acquired using the 458 nm dichroic filter cube. (Top Right) The unmixed donor image acquired using the 495 nm dichroic filter cube. (Bottom Right) The unmixed acceptor image acquired using the 495 nm dichroic filter cube. The unmixed acceptor image acquired using the 458 dichroic filter cube is the dimmest image, as the acceptor is difficult to excite with wavelengths shorter than 475 nm. However, the unmixed acceptor acquired using the 495 dichroic filter cube is the brightest image, as the acceptor can potentially be excited directly or through FRET.

Figure 5.

The excitation and emission spectra of the donor and acceptor FRET molecules. Excitation spectra are shown as dashed lines. Emission spectra are represented by solid lines. The donor spectra are shown as blue and the acceptor spectra are shown as orange. Of particular interest for the HIFEX method are the excitation spectra at the 450 nm and 495 nm regions, as the intensities plotted demonstrate the difficulty of exciting the acceptor when using the 458 nm dichroic filter cube and the change in preferential molecule excitation around 475 nm.