Accurate counting of neurons in frozen sections: some necessary precautions.

In 30 microns frozen sections of rat midbrain the retrograde axonal transport of diamidino yellow, a fluorescent tracer, was used to demonstrate a population of neurons in the substantia nigra. However, when visualisation was carried out using the routine Nissl method a significant proportion of neurons failed to stain. As the presence of the retrograde tracer did not affect Nissl staining of such cells, such incomplete staining, with consequent underestimation of neuronal populations, is probably a common error in similar material. Further investigation revealed that the proportion of such unstained neurons was greater when the staining time was short, when stain concentration was low, or when section thickness was increased. Some stains were worse in this respect than others. Cresyl fast violet resulted in the highest proportion of unstained neurons, thionin resulted in the lowest proportion. It was concluded that the rate of diffusion of the stain into the section was the main factor limiting the staining of neurons present. Staining with pure thionin at 0.1% concentration for at least 3 minutes and with sections no thicker than 30 microns is one regime which would avoid this problem.