Exploring Molecular Dynamics with Raster Image Correlation Spectroscopy (RICS)

For these experiments, U2OS cells were transfected with either 1xGFP, 2xGFP, 3xGFP, or 4xGFP oligomers. The cells were imaged the day after transfection with the ZEISS LSM 980 laser scanning confocal using the solid state 488 nm laser to excite GFP at 1 % laser power. For the RICS acquisition, settings with a pixel size of 50 nm and frame speed of 8.19 μs were used, with a time series of 100 frames (unless explicitly mentioned otherwise). The ZEISS QUASAR detector was used in photon counting mode for the measurements comparing diffusion across the GFP oligomers. Ten cells were imaged for each GFP oligomer. The same cells were imaged with the detector in photon counting and integration mode (5 cells per condition were imaged).

Figure 4 depicts the average spatial correlation function, with each spatial correlation function calculated separately from each image in the experimental image series. The purpose of the RICS spatial correlation function is to inform the user what the likelihood is to find molecules at a distance (ξ, ψ) from their current location, given their diffusion constant and given the fact that the laser is imaging the molecules by raster scanning over them with a given scan speed along a line and between lines. The center correlation point (at ξ = 0 and ψ = 0; Figure 4A) depicts the maximal amplitude of the autocorrelation function, N (Figure 4B). The ξ axis is the fast-scanning axis of the confocal microscope along a line. The ψ axis is the slow scanning axis of the microscope as it scans line after line.

The spatial correlation function surface plot (Figure 4A) provides direct information on the molecular mobility. Fast diffusing molecules will exhibit little spatial correlation along the ψ axis, while slower diffusing molecules exhibit more correlation along the ψ axis. An ideal RICS correlation has correlation both in ξ and ψ, yet more pronounced correlation in ξ than in ψ. This visually translates to faster movements being displayed as an asymmetric, line-like correlation, whereas slower movements as a symmetric, circular correlation function (equal correlation in ξ and ψ). If the correlation is a perfect circle, then molecules do not diffuse significantly within one image scan.

The maximal amplitude of the correlation function (indicated in red, peak of the surface plot, Figure 4B) defines the number (N) of fluorescent molecules in the confocal volume, from which the concentration of the diffusing molecule can be assessed by N = 1/G. The larger the amplitude, the lower the concentration. The standard deviation on the correlation function points that surround the actual correlation (the blue carpet, Figure 4A&B) informs on the quality of the RICS data. The lower the standard deviation (more homogeneously blue and smooth carpet), the better the data.

RICS analysis fits the experimental data to a predefined model to extract the diffusion coefficient (D), and the number of molecules (N). During RICS analysis, for each data point of this average correlation function, the standard deviation is also calculated to allow for error-weighted analysis. In Figure 4D (lower graph), two sections of the experimental RICS data (error bars are the standard deviation) and fit model (solid lines) are shown: the blue section along the fast-scanning axis, G(ξ ,0), and the red section along the orthogonal slow scanning axis, G(0,ψ). In the top part of Figure 4C, the weighted residuals are shown, which are the difference between the data and fit model, divided by the error.

Visual inspection of the fit results allows qualitative assessment of the goodness-of-fit: In Figure 4C, color scaling depicts the weighted residuals. A more intense blue or red color indicates that the fit model value at a given location of the spatial correlation is lower or higher, respectively, than the experimental correlation data value at that location. The absence of any color for a given fit model in the left graph illustrates a perfect fit, i.e., weighted residuals equal to zero. The lower the error on the experimental data, the better the fit model needs to fit the experimental correlation data to lead to a small, weighted residual.

When imaging with a laser scanning microscope, the signal is collected by the detector. The detector can be operated in either photon counting or integration mode. The choice between the two modes depends on the specific experimental requirements and the characteristics of the sample being studied.

Photon counting mode is typically used when the sample is relatively dim, and the signal-to-noise ratio needs to be optimized. In photon counting mode, each detected photon is precisely counted, which can be used to calculate the diffusion coefficient of the fluorescent particles in the sample. Photon counting mode is also useful for detecting low levels of fluorescence, as the sensitivity of the detector is typically higher than in integration mode.

Conversely, integration mode has a higher dynamic range, making it the preferred choice for brighter samples. In integration mode, the detector integrates the signal from each pixel over a certain period of time, resulting in a continuous output signal that represents the average intensity of the fluorescent particles in the sample. Integration mode is useful for measuring the distribution and concentration of fluorescent particles in the sample, as it provides a more accurate measure of the average intensity of the fluorescence signal.

To investigate if the mode of detection affects the calculated diffusion coefficient, the same cells were imaged with ZEISS Spectral RICS in both photon counting and integration mode. The RICS analysis revealed that there is no significant difference in the diffusion coefficient calculated regardless of the mode of acquisition. This provides evidence that either mode of the detector can be used to accurately characterize protein behavior.