 |
 |
 |
 |
1. Standardization of the instrument. We standardize by setting a
fluorescent bead set each day to exactly the same channel value.
2. Proper compensation. We use an automated compensation setting
program so that there is no "judgment" involved.
But the most important item is
3. Saturating antibody reagents.
We set up a single set of reagents to cover the entire study (i.e., premixed).
Also, they were carefully titred under the staining conditions used.
This is extremely important, since nonsaturating reagents can lead to
variations in fluorescence due to time of incubation as well as small
errors in the pipetting of the reagent. (Both of these become
irrelevant under saturating conditions). Note that for a vast majority
of reagents, cell number is almost irrelevant since antibody is in vast
excess to antigen.
Attaining saturating reagents is difficult for many PE reagents and
most Cy5PE reagents, since virtually all manufacturers bottle these
reagents much too dilute (and yes, I have tested many manufacturer's
reagents, please salesmen don't EMail me to tell me that YOU sell
saturating reagents. or else I'll publish the titration curves for
your reagents!). Biotin and FITC reagents are usually saturating.
We overcame this by getting bulk conjugated reagents and preparing the
stains ourselves. For some PE and most of the Cy5PE reagents, we had
to include competing "cold" (unlabeled) antibody to bring the signal
back on-scale on the flow cytometer as well as maintain saturating
conditions.
Finally, some antigens simply do not saturate until extremely high
concentrations (i.e., >50 ug/ml), for instance virtually every
anti-human CD20 I have tested. When using these reagents, you have to
be as careful as possible to maintain a constant incubation time and
temperature, and pipette reagent and cell volumes very carefully.
In our study, we achieved typically 10% variation in most antigen
fluorescences (as measured on the same person over a 9 month period).
Some antigens were more variable, some were less.
We found very interesting changes in densities associated with HIV
progression, showing that the process is well worthwhile.
After all, in FACS data there is so much information, why should we
limit ourselves to only population frequencies?
mr
|
|
|
|