Fluorescence microscopy is a powerful tool that has brought much insight into the structure and function of biological cells and molecules. As researchers press further to gain greater understanding at this level, experiments are moving beyond the single fluorophore to incorporation of multiple fluorophores in a single imaging experiment.
While the use of multiple fluorophores- is conceptually straightforward, analysis of the resulting images can be extremely complicated because of the significant overlap of fluorescence emission signals (Figure 1). Going beyond more than four fluorophores makes quantitative signal extraction using standard filter based techniques virtually impossible. This emission overlap can even be problematic with a single dye where the peak emission is masked by an intrinsic signal such as auto-fluorescence
To address the problems posed by overlap of fluorescence emission signals, Optical Insights has developed the Spectral-DV™ imaging system, capable of acquiring full spectral information at each pixel in the image. As described in the next section, the measured full spectrum of emission can be used to accurately determine the quantity of each fluorophore at each pixel in the image.
Spectral unmixing is a mathematical technique inherited from the optical remote sensing community to determine the quantity of component spectra from a measured spectrum. For example, if you know that your sample was stained with Alexa Fluor 555 and Texas Red [Molecular Probes, Eugene, OR], then you also know (assuming no auto-fluorescence) that at each pixel in your fluorescence image, there will be some combination of the Alexa Fluor 555 and Texas Red emissions (Figure 2a). What you really want to know is the quantity of each fluorophore at each pixel. However, because of the considerable overlap of the two fluorescence emissions, an accurate calculation will only be possible in this instance if the full spectrum of emission has been measured.
Given that the fluorescence emission curves for Alexa Fluor 555 and Texas Red are known, the amount of each in the measured spectrum can be determined at each pixel using different spectral unmixing algorithms (Figure 2b).
Applications for Spectral Imaging & Unmixing:
• Separation of specific fluorescence and auto-fluorescence
• Quantitative separation of true FRET signal from bleed through signal of donor
• Colocalization in the presence of multiple fluorophores
• Pixel classification based on spectral signature
• Spectral karyotyping (SKY)
• Multi-Color FISH (mFISH)
EXAMPLE APPLICATION - UNMIXING GREEN FLUORESCENCE AND AUTOFLUORESCENCE
When investigating cells from organs such as the liver or the brain, there is often the difficulty of separating the fluorescence signal of interest from the auto-fluorescence of the tissue. The auto-fluorescence tends to have a broad emission curve making it difficult to find a wavelength region where only the emission of the fluorophore being used is present. While this scenario presents difficulties for bandpass filter based systems, it is easily dealt with using spectral imaging and subsequent spectral unmixing.
To demonstrate this capability, a liver section was stained with Alexa Fluor 488 phalloidin. It was desired to visualize the actin filaments separately from the rest of the tissue. Figure 3 shows how the image appears when a simple 500 nm longpass filter is used to image the sample.
From the image, it is difficult to determine which structures have been stained with the Alexa Fluor 488. In other words, it is difficult to see the actin filaments separately from the general auto-fluorescence of the tissue.
To solve this problem, the Spectral-DV™ imaging system was used to acquire a spectral stack of images, a few of which are shown in figure 4. The spectral stack of images provides a full spectrum measurement at each pixel in the image. This data was then processed using Optical Insights' Melange™ spectral imaging software package to unmix the Alexa Fluor 488 and the auto-fluorescence emission signals.
Figure 5 shows the intensity image due to auto-fluorescence. Figure 6 shows the Alexa Fluor 488 fluorescence separated from the auto-fluorescence. The actin filaments are clearly visible in the image. The unmixed signals can be overlaid to show the Alexa Fluor 488 signal on top of the auto-fluorescence signal (figure 7).
Micron Optics, 240 Cedar Knolls Road , Cedar Knolls, NJ 07927
*GIF animation at top is a conoscopic image of quartz captured through an E600pol microscope using the Nikon DXM-1200 digital camera. Courtesy of Daniel Sparling, former employee and now clergyman in training