Requirements for obtaining unbiased estimates of neuronal numbers in frozen sections.

The use of frozen sectioning is a convenient and rapid means of observing the results obtained using fluorescent retrograde tracers. Quantitation of these results using the biased stereological methods currently available can be associated with large errors. A recently developed stereological tool, the optical dissector, provides unbiased and efficient results, however, the requirements for its use in frozen sections has not previously been established. In this study, a comparison was made of neuron numbers, estimated using the optical dissector method, in the motoneuron pool retrogradely labelled from the rat sciatic nerve with either Fast Blue or Tetramethylrhodamine dextran (fluoro-ruby) in methacrylate embedded and frozen spinal cord specimens. Despite over 50% shrinkage in the frozen sections, compared with virtually no shrinkage in the methacrylate sections, no significant difference in labelled motoneuron numbers was observed, provided this shrinkage was taken into account. Correction for section shrinkage is therefore essential in order to use the optical dissector with confidence to count fluorescent labelled neurons in frozen tissue.