Quantitative mapping of fluorescently tagged cellular proteins using FCS-calibrated four-dimensional imaging

Abstract

The ability to tag a protein at its endogenous locus with a fluorescent protein (FP) enables quantitative understanding of protein dynamics at the physiological level. Genome-editing technology has now made this powerful approach routinely applicable to mammalian cells and many other model systems, thereby opening up the possibility to systematically and quantitatively map the cellular proteome in four dimensions. 3D time-lapse confocal microscopy (4D imaging) is an essential tool for investigating spatial and temporal protein dynamics; however, it lacks the required quantitative power to make the kind of absolute and comparable measurements required for systems analysis. In contrast, fluorescence correlation spectroscopy (FCS) provides quantitative proteomic and biophysical parameters such as protein concentration, hydrodynamic radius, and oligomerization but lacks the capability for high-throughput application in 4D spatial and temporal imaging. Here we present an automated experimental and computational workflow that integrates both methods and delivers quantitative 4D imaging data in high throughput. These data are processed to yield a calibration curve relating the fluorescence intensities (FIs) of image voxels to the absolute protein abundance. The calibration curve allows the conversion of the arbitrary FIs to protein amounts for all voxels of 4D imaging stacks. Using our workflow, users can acquire and analyze hundreds of FCS-calibrated image series to map their proteins of interest in four dimensions. Compared with other protocols, the current protocol does not require additional calibration standards and provides an automated acquisition pipeline for FCS and imaging data. The protocol can be completed in 1 d.

Figure 1:. Principle of fluorescence correlation spectroscopy (FCS) for point confocal microscopes.

(a) The excitation laser beam is positioned to a specific location of a cell. Fluorophores entering the confocal volume are excited and the number of photons emitted is recorded. Shown are two cells expressing different levels of the fluorescent protein mEGFP. Scale bar 10 μm. (b) The fluctuations of the fluorophores diffusing in and out of the confocal volume cause fluctuations in the number of photons. (c) Computing the self-similarity of the fluctuations in (b) yields auto-correlation functions (ACFs). The amplitude of the ACF is directly proportional to the inverse of the number of molecules observed on average in the confocal volume.

Figure 2:. Workflow for FCS-calibrated imaging

(a) Setup of the multi-well chamber with four different samples and optimization of the water objective correction for glass-thickness and sample mounting using a fluorescent dye and the reflection of the cover glass. dye: fluorescent dye to estimate the effective confocal volume, WT: WT cells, mFP: WT cells expressing the monomeric form of the FP, mFP-POI: cell expressing the fluorescently labeled POI. (b) Computation of parameters of the effective confocal volume using a fluorescent dye with known diffusion coefficient D_dye and similar spectral properties as the fluorescent protein. Fitting of the ACF to a physical model of diffusion yields the diffusion time τ_D and structural parameter κ. The two parameters are used to compute the focal radius w₀ and the effective volume V_eff (Box 2). (c) Images and photon counts fluctuations are acquired for cells that do not express a fluorescent protein (WT), cells expressing the mFP alone (mEGFP), and cells expressing the tagged version of the POI, mFP-POI (mEGFP-NUP107). The image fluorescence intensity (FI) at the point of the FCS measurement is recorded. Scale bar 10 μm. (d) The ACF is fitted to a physical model to yield the number of molecules in the effective volume. WT concentrations are set to 0. After background and bleach correction concentrations are computed using the V_eff estimated in (a-b) and the Avogadro constant N_a.Data is quality controlled with respect to the quality of the fit and the amount of photobleaching using objective parameters. (e) Concentrations estimated from FCS and image fluorescence intensities are plotted against each other to obtain a calibration curve (black line).

Figure 3:. Example of image quantification using FCS calibration.

(a) The fluorescence intensity at each pixel, I_p, is converted to concentrations and protein numbers (for 3D and 4D images). The images are acquired with the same imaging parameters as the images used to compute the FCS calibration curve. Shown is data for a HeLa Kyoto cell line with endogenous NUP107 tagged N-terminally with mEGFP. Scale bar 10 μm. (b) The quantitative distribution of a POI can be derived by using markers for cellular structures. DNA stained with SiR-DNA is used to compute a 3D chromatin mask of the nucleus. In the equatorial plane, a three pixel wide rim defines the nuclear envelope. Fluorescence intensities and protein numbers on the nuclear envelope can then be calculated. Scale bar 10 μm. (c) mEGFP-NUP107 average fluorescence intensity on the nuclear envelope. Data shows results obtained on two different microscopes. The boxplots show median, interquartile range (IQR), and 1.5*IQR (whiskers), for n = 16–22 cells and 2 independent experiments on each microscope. System 1: LSM880, FCS and imaging using the 32 channel GaAsP detector. System2: LSM780, FCS using the APD detector (Confocor-3), imaging using the 32 channel GaAsP detector. (d) Conversion of fluorescence intensity to protein numbers using the corresponding calibration curve obtained for each experiment and microscope system (not shown). The protein density on the nuclear envelope has been computed according to Eq. S24 (Supplementary Note 6). The gray shadowed boxes show the expected protein numbers for NUP107 ,.