I was ask to post a summary of my responses to this question. So here it is : There were various opinions to this case, but nobody had the answer for the "Perfect" control. A short sammery : 1. You have to use isotypcontrol for publishing 2. Use both negative control, unstained for setting gates and then look for isotype staining 3. Isotype and antibody have to be from the same company And the autheneic answers : Not a good idea. If your test antibody and your control antibody are isotype (and species) matched, then that is the only control you can use. I'm not sure you could even publish the results if you didn't use a control antibody. The isotype and test antibodies must be from the same host. There are about as many opinions on the use, abuse and validity of isotype controls as there are controls!! To be absolutely correct, each antibody panel should have a corresponding isotype control which is protein, fluorochrome matched, and source to the corresponding antibody. Controls will vary by F-P ratio, but it is not a perfect world in the land of Flow QA/QC. Since the majority of the antibodies are G1, I set up a single control choosing the maximum protein concentration (500ng for CD34) for each of the fluorochromes. Additional controls should be set up if (1)antibody sources differ within the panel, ie if using a Coulter mAb conjugated to FITC, do not use a BD isotype as the F-P ratios may differ or (2) antibody is other than G1 isotype, ie CD5 IgG2a or CD15 IgM. I also set up a non-stained 'autoflouorescent' population of cells as a second negative control to set up the PMTs and negative quadrants. Using isotypes to set gates, ie negative populations, can be misleading, as antigens do not show bimodal expression patterns that are wither negative or positive. Many antigens are expressed on negative cells and the brighter the conjugated antibody the 'less negative' these cells will be. Thus the flourochrome choice becomes important, although many times limited to availability. I'm personally not a true believer in isotypes, however I run them to pacify the various investigators. My experience with isotypes and quad stats is that sometimes a population is positive for the isotype, however not in the same location as the positive population and subtraction of events would not be relevant to evaluation of the positive population. In addition, the majority of my analysis is using sequential logical gating and this solves the problem, as if there are non-specific events they will be in the same location as the positive population. Use both. A control antibody will then be compared to the unstained cells, and THIS is what you can use estimate the nonspecific binding. The assumption is that your test antibody has the same background. I would use cells not stained with any antibody as a negative control. However, like the antibody-stained cells, the control should also be incubated with the fluorochrome-labeled second-step antibody I agree with the person who mentioned this to you. Every ab is different and leads to different efficiency in binding to the same Fc receptor. Furthermore, the conjugation of both abs will be different. There is no good control. We set our flow with an unstained sample and control it (that) with an irrelevant ab, because journals really want to see that control. So for the sake of publishing your data take the isotype matched antibody anyway. For intracellular staining unstimulated cells might do the trick. And the summary of the discussion in 1998 on : http://www.cyto.purdue.edu/hmarchiv/98/index.html Kerstin ***************************************** Kerstin Beck Institut für Medizinische Mikrobiologie und Hygiene Abteilung Virologie Hermann-Herder-Strasse 11 D-79104 Freiburg Telefon : +49-(0)761-203-6646 Fax : +49-(0)761-203-6603 beckk@ukl.uni-freiburg.de *****************************************
This archive was generated by hypermail 2b29 : Wed Apr 03 2002 - 11:57:31 EST