Friday, March 18, 2011

Fluorochrome signal calibration in QFISH analysis

It is desirable that independently from the imaging system or from the fluorochrome used (FITC, Cyanine, ...), the intensity of the fluorescent probe depends only on the number of targets in the biological sample.
Spectral calibration:
In order to compare the FISH signal obtained with different fluorochromes as FITC, Cy3, Cy5, the following assay can be performed with conventional material: 
  • Prepare a glass slide, and lay down a  2micro-L drop of 100 micro-M of fluorescent solution (stock solution of the molecular probe).
  • Heat the slide and mount with an anti-fading solution.
  • For each fluorochrome, take a serie of images of the microscopic field of varying time exposure and compute the mean gray level of each image.
 The following result was obtained with a leica DMR microscope equiped with a KAF1600 CCD camera:
      The data were linearly fitted and the sensitivity of a given fluorochrome was set as the slope of the linear response (gray level per time unit for a give probe concentration). Cy3 yields 713 gray level per 1/10 sec, Cy5:403 gl/time and FITC:297. Thus taking Cy3 as reference:
      Relative sensitivity of fluorescent probe.
      Color compensation
      In fluorescent microscopy, several fluorochromes are present on a biological sample. The light emitted by each fluorochrome is observed in a spectral window "opened" by a filter in the microscope.

      DAPI stains the DNA, with the DAPI block filters (the "blue window""or the blue channel), when you watch in the microscope oculars to see a DAPI stained metaphase, it looks blue. Thus,if you try to watch DAPI stained chromosomes in a green channel (with a FITC band pass filter), you should see nothing: this is true with naked eyes but wrong with a CCD camera with long exposure time. In the following example, images of a DAPI stained metaphase were taken in different spectral channels with different exposure time. Excepted for DAPI, the filters set were from omega filters(alpha vivid serie):
      DAPI:blue channel filter A
      FITC:green channel alpha vivid XF100-2
      Cy3:orange channel alpha vivid XF108-2
      Cy5: near infrared channel alpha vivid XF110-2
        DAPI spectral overlap displayed with false color.Top left:DAPI viewed through filter A (0.1s). Top right:DAPI viewed through green channel (8s). Bottom left: DAPI viewed through orange channel (8s).Bottom right:DAPI viewed through deep red channel for 30s.
        As the exposure time increases, DAPI is visible in the green channel, in the orange channel and even the near infra red channel after 30 sec of exposure. 
        The same is true for other fluorochromes, as FITC. The following example is the result of  a FITC labelled telomeric PNA probe hybridizing V79 cell line chromosomes.
        FITC spectral overlap in blue, orange and red channels
        To address this issue, Castleman (Digital Image processing, Prentice Hall, p556-561), define a color compensation technique relying on linear algebra:


        where E a diagonal matrix specifying the relative exposure time of each channel and C a matrix accounting for the spectral spread through the channels with cij the proportion of signal  from fluorophore j visible in the channel i.
        x  is a 1 by m (m fluorochromes) vector representing an ideal image with no spectral spread.
        y is a 1 by m (m fluorochromes) vector representing the observed image with no spectral spread.
        b  is a 1 by m vector accounting for the background.
        To compute an ideal image with no spectral spread, the following equation has to be solved:
        with C' and E' the inverse matrix of C and E.
        For example, with four fluorochromes (DAPI, FITC, Cy3, Cy5), the EC matrix was:

        graphical representation of the EC matrix