Wayne,

A few comments:

>I have been told I can safely say that if I have an image where the
>intensity of a signal in one subcellular compartment relative to another
>(in the same image) is 4 fold (e.g.. pixel values averaging 50 and 200),
>then the corresponding protein (to which our antibody is directed)
>concentration should be 4 times higher in one compartment relative to
>the other (assuming accessibility etc. are equal).

My reaction to this is that it depends. First, you are correct in that you
must account for any offsets in the quantitizations that may be due to PMT
dark current counts or other electronic offsets. These should be subtracted
from your raw pixel counts. Second, you must be sure that you are in the
linear range of operation of the PMT. Both of these factors can be
determined by calibration. On our BioRad 1024, I have used reference
solutions consisting of fluorescein from 5 to 500 uM in 50% glyerol:water
plus 2.5 mg/ml phenylenediamine at pH 8.5. These give nice linear behavior.
(Accumulates are necessary for the lower concentrations). When using this
approach, I was careful to keep the objective (63x 1.4 NA) focused to
within 10 um of the inner surface of the coverslip to avoid problems with
spherical aberration, which can change the size of the sampled excitation
volume. So that is part of the maybe. When you image a real sample,
different compartments may be at different depths and have slightly
different index of refractions so that volume being sample is not
identical. The signal you acquire will depend on the number of fluors you
excite, not the concentration, so there is the possibility of some error.

>
>Biorad has told us that the relationship between two intensity values at
>a given setting in an image should be linear.
>

For our 1024, using the fluorescein reference solutions, I agree with the
BioRad claim with the caveats stated above. One more thing, we have also
done this kind of calibration with FITC labeled goat anti-mouse, and you do
have to use a very concentrated solution and lots of integration. The
reason is that the concentration of fluor is going to be very low. A 1
mg/ml solution of antibody with an average of four fluors per molecule will
have a fluor concentration of less than 30 uM.






Mario M. Moronne, Ph.D.
Dept. of Subcellular Structure M/S 6-2100
Lawrence Berkeley National Laboratory
1 Cyclotron Rd.
Berkeley, CA
94720

ph (510) 486-4236
FAX (510) 486-5664
[log in to unmask]  or [log in to unmask]