Calman Prussin wrote- >While we are on the subject of counting angels on the heads of pins: > >In addition to limiting frequencies of positive cells, LIGAND DENSITY can >also be a problem. My 2 part question is: > >1. What is the lower limit of detection of surface molecules using a bench >top flow cytometer and typical staining, compensation techniques? Units can >be in molecules of FITC or PE, # of receptors, angels, etc. > >2. What techniques have you personally used to enhance this lower limit of >detection and what S/N enhancement did you get? A critical factor here is the autofluorescence of the unstained cells. Lymphocytes excited at 488 nm have autofluorescence equivalent to a few hundred molecules of fluorescein in the green (520-530 nm) channel and probably equivalent to at least 200 molecules of phycoerythrin in the yellow (575-585 nm) channel. The autofluorescence distribution is also relatively broad, so you won't get a clean separation between unstained cells and cells bearing hundreds of molecules of label. It is generally accepted that most of the autofluorescence observed with 488 nm excitation in unfixed cells comes from flavins; reaction products of formaldehyde with amines add to fluorescence in fixed cells. Autofluorescence is substantially lower when longer wavelength excitation (above the absorption region of flavins) is used. B-D has modified FACScans, Caliburs, etc. for internal use (and for selected customers) by substituting a 532 nm (green) frequency-doubled YAG laser for the argon laser, improving sensitivity substantially. I have built instruments (Cytomutts) with green YAG lasers for my lab and for Frank Mandy in Ottawa; the Luminex 100 incorporates essentially the same excitation and collection optics and detectors. Using the Mutts, we found that the fluorescence of unstained lymphocytes topped out at around 75 molecule-equivalents of phycoerythrin. Frank's lab has done repeated analyses of cells stained with mixtures of phycoerythrin-labeled fluorescent antibody and "cold" antibody prepared by bleaching the phycoerythrin-labeled antibody (this preserves stoichiometry, since reaction kinetics differ significantly between the MW 400,000 PE-labeled IgG and unlabeled MW 160,000 IgG, but not between the fluorescent and bleached PE-antibody, which have the same 400,000 MW). The Mutts can reliably discriminate cells bearing 120 or more molecules of antibody-bound PE (the antibodies are B-D's 1:1 PE conjugates) from unstained cells - and the Luminex system (which has a less powerful YAG laser) is not much less sensitive; it can generally discriminate cells (or beads) with 200-300 bound PE molecules. No magic here; I'm sure you could get equal or better performance from a FACSCalibur or 'Scan with a green laser, and I'd expect better because B-D's flow cell design collects more light than mine. The signal-to-noise enhancement associated with going from 488 nm to 532 nm excitation is a factor of 4 or 5. If you really want the ultimate in sensitivity, you need to select gating antibodies which don't bleed into your primary measurement channel, because the crosstalk from spectrally overlapping dyes contributes to background noise, even when compensation is done (Bob Hoffman of B-D has produced some informative presentations on this topic). The Mutts have red lasers and can use Cy5, Cy5.5, APC and its tandems, etc. for gating when the primary quantitative measurement is done with a PE-antibody. PE-Cy5 should be avoided because of apectral overlap; PerCP and the PerCP-Cy5.5 tandem from B-D should be OK for gating but I personally haven't tried them. Slow flow systems such as those at Los Alamos have managed detection of single molecules of phycoerythrin, and one can generally expect better photoelectron statistics and improved sensitivity if flow is slowed down. However, the single molecule measurements are made in solution, in which case the autofluorescence background associated with a cell is not present; if you're measuring cells, you still have to have a signal well above autofluorescence. -Howard
This archive was generated by hypermail 2b29 : Wed Apr 03 2002 - 11:57:21 EST