Limitations of flow cytometry for the specific detection of bacteria in mixed populations.

Flow immunofluorescence (FIF) techniques were established for the specific detection of the bacteria Escherichia coli, Legionella pneumophila and Bacillus anthracis spores after staining with fluorescein-conjugated antibacterial antibody. For each bacterial type, a comparison was made of gating on narrow forward angle (NFA) light scatter and on the red fluorescence (Red Flu) signal available from staining with the nucleic acid dye propidium iodide. No universal gating method was found, since Bacillus spores did not take up propidium iodide and only a part of the Legionella population gave detectable NFA scatter signals. The efficiency of detecting bacteria stained with antibody remained constant with differing concentrations of the specific bacterium, and the estimate of the count for specific bacteria expressed as a fraction of the total cytometer count fell sharply with bacterial concentration. This effect was apparently due to cytometer noise inherent in the high sensitivity of detection needed for particles as small as these bacteria. The noise did not originate in the photomultipliers and was evidently the result either of light scatter from sub-micron particles in the sheath fluid or scatter from optical components. Part of the noise could be removed by selective gating, but there remained a noise component overlapping with the NFA scatter and Red Flu signals from the heterologous bacteria, i.e., those not stained with specific antibody. In consequence, at the low bacterial concentrations used no meaningful cytometer count could be obtained for the excess of the unstained bacteria and the proportion of specific bacteria in the mixed population could not, therefore, be calculated.