Problems with colour compensation
These detailed instructions are for the FACScans with FACSMates, but the principles also apply to setting up 3 (or even 4) colour analysis on the FACSort.
1) It is important to set the colour compensation accurately because the fluorescent dyes have significant overlaps in their spectrum. In particular, strong FITC signals (FL1) will overlap into FL2, PhycoErythrin (PE on FL2) will overlap mostly into FL3, and tandem conjugates (that are chemical conjugates of PE with Texas Red eg. "Duochrome" or PE with Cy5 eg. "Quantum Red" normally measured in FL3) will vary greatly in their efficiency of coupling (possibly even between batches) and therefore their background emission in FL2. Therefore FL2-FL1 compensation is most important for two colour analyses while FL2-FL1, FL2-FL3 and FL3-FL2 are all critical for three colour setups.
2)If you are running three colour analyses you must use the 5 paramter 1024 resolution setting on the FACSMate Main Menu page, together with Instrument Control Settings at this resolution (settings saved with 4 paramter 256 resolution can not be used or even restored in this mode). You can use such three colour setups also to run two colour stained samples as the FL2-FL3 and FL3-FL2 compensations are then irrelevant.
3) The colour compensation settings are acutely dependent on the sensitivity of each detector - you cannot change either the detector or amplifier gain of any of the fluorescence parameters (FL1, FL2 or FL3) without also recalibrating the compensation values. (You can, however, still adjust FSC or SSC).
4) To check or adjust the compensation run you should use control samples labelled with each colour singly. Ideally these should be fixed standards that you have run before rather than single colour stains on the day (although it is a good idea to run each single colour controls in your experiment too), as you should not attempt to use compensation settings to adjust for cross-reactions between colour reagents or other artifacts in your experiment. The ideal standard for each colour has a mixture of positive and negative cells, or a broad smear of reactivity from negative to positive, over the range of sensitivity for the reagents you intend to run. I also have available some beads labelled with FITC or PE for setting up two colour compensation, and some blue beads for checking 4 colour settings - please ask.
5) Run the standards from the FACSmate with the Data Aquisition mode set to "No data", the number of events set to a maximum of 50,000, and at a slowish rate (eg 200 cells per second) so that you have plently of time to make adjustments while the sample is running. It is also best to use a tight gate round your cells to avoid confusion with dead cells or other cells with a high background which will confuse your compensation settings (a cell with a non-specific background gives an artifactual higher degree of cross-talk between colours and is usually very difficult to compensate for).
6) While running the single FITC standard, select FL1 on the vertical axis of the dot plot (move there with the cursor keys and hit PREVIOUS or NEXT until you get FL1) and FL2 on the horizontal axis.
7) Make sure that the right hand edge of the positive population is directly above the right hand edge of the negative population. If it is leaning to the right then you need to increase the compensation of FL2-FL1 slightly (use the touch buttons on the front panel of the FACScan - you may need to switch the panel on first also by pressing the appropriate touch panel). The setting of FL2-FL1 is usually only 0 to 17% - suspect a problem if you need to set it much higher (one likely reason is that you have set the detector for FL2 too high compared to FL1 - try reducing it). If in doubt, it is better to overcompensate slightly as you will not be able to compensate out completely any real double-positive cells (but overcompensation using FL2-FL1 will reduce the apparent level of fluorescence of weakly FL2 positive cells).
8) When you are happy with the FITC staining, move onto the next sample (if your samples are in order you can go to the next by pressing F6 [Abort Sample]) which should be the single PhycoErythrin (PE) standard. The single PE positive population should now be horizontal with the top edge in line with the negative population (assuming you still have FL1 vertical and FL2 horizontal). With too little compensation on FL1-FL2 the population will be bent upwards, while with too much compensation all the dots will appear squashed onto the bottom axis. Again, use the touch panel of the FACScan to select FL1-FL2 and make the necessary adjusments. Normal values for FL1-FL2 are in the range 0-2%, and the correct compensation can be quite sensitive to even minor adjustments of less than 1%. If in doubt, it is better to overcompensate slightly as you will not be able to compensate out completely any real double-positive cells (but overcompensation using FL1-FL2 will reduce the apparent level of fluorescence of weakly FL1 positive cells).
9) If you are running two colour analysis then you have finished compensating and are ready to run samples (make a note of the compensation in the FACScan LOG BOOK for future reference, or Save the instrument settings as detailed above if they are completely different to any others in common use).
10) If compensating for three colours you should also adjust the FL3-FL2 compensation while running the single PE stained sample. Change the on screen dot-plot so that you have FL3 vertical (leaving FL2 still horizontal) and check again that your cells fall exactly horizontal - if not adjust FL3-FL2, increasing the value if the positive cells bend upwards. Settings for FL3-FL2 are extremely dependent on the FL3 sensitivity (amplifier setting), and a small change in the value can have a very large effect, but are often in the range 5-50%. If you cannot achieve enough compensation you will have to turn down the sensitivity of FL3 and try again. Again, overcompensation is better than under.
11) Finally, run through the singly stained third colour sample (eg. Quantum Red [QR]) and assuming you still have FL3 vertical and FL2 horizontal, adjust the setting of FL2-FL3 to ensure the positive population is directly over the negative population, increasing the value if there is any tendency for the positive population to lean to the right. This setting appears to be most dependent on the quality of the tandem conjugate used as well as the sensitivity of FL2, but will often have a value between 5-25%.
12) You are now ready to run - record you settings in the FACScan LOG BOOK and pray for Good Luck! A double Glen Farclas 105 does wonders at this point.
Generally speaking, if you think your samples are not correctly compensated for, DO NOT try and compensate for any inadequacies in them directly, but run through some well characterised samples or single colour standards to check the machine settings as above. Also, make sure that there is not a flow problem by checking there is enough sheath, perform a few drain and fills, and double check that the scatter looks normal.
The usual problem with multicolour analysis is that you can't seem to get rid of cells that fall on the diagonal ie. apparently erroneous double staining cells. There are three principle reasons why this can occurr:
1) Non-specific staining due to inadequate blocking (especially of Fc receptors - use heat inactivated normal rabbit serum or Pharmingens "Fc block" for direct conjugates) or poor quality reagents (especially if aggregated or precipitated - try microfuging and/or 0.2micron filtering).
2) Cross reactions between reagents - this is a particular problem if you are using indirect staining for any of your colours, and is often due to capture and/or cross-reactivity by the antiglobulin conjugate. You can use methods to compete out this cross-reactivity if necessary.
3) Autofluorescence - this can either be genuine autofluorescence of the cells, which is difficult to solve other than using brighter labelling reagents to swamp it out, or aquired autofluorescence during culture especially with phagocytic or endocytic cell types (use phenol red free medium and 0.2micron filtered sera). Other sources of aquired autofluorescence are over or wrong fixation (especially if cells are clumped - but this will show on the scatter), or contaminating dyes (PI etc) because the FACS was not properly washed between samples or after use (try running some 70% ethanol followed by water first).
scobbold@dunn1.path.ox.ac.uk updated 18.12.97.
and is a subset of the Sir William Dunn School Web Pages