Cautionary observations on preparing and interpreting brain images using molecular biology-based staining techniques

Though molecular biology-based visualization techniques such as antibody staining, in situ hybridization, and induction of reporter gene expression have become routine procedures for analyzing the structures of the brain, precautions to prevent misinterpretation have not always been taken when preparing and interpreting images. For example, sigmoidal development of the chemical processes in staining might exaggerate the specificity of a label. Or, adjustment of exposure for bright fluorescent signals might result in overlooking weak signals. Furthermore, documentation of a staining pattern is affected easily by recognized organized features in the image while other parts interpreted as "disorganized" may be ignored or discounted. Also, a higher intensity of a label per cell can often be confused with a higher percentage of labeled cells among a population. The quality, and hence interpretability, of the three-dimensional reconstruction with confocal microscopy can be affected by the attenuation of fluorescence during the scan, the refraction between the immersion and mounting media, and the choice of the reconstruction algorithm. Additionally, visualization of neurons with the induced expression of reporter genes can suffer because of the low specificity and low ubiquity of the expression drivers. The morphology and even the number of labeled cells can differ considerably depending on the reporters and antibodies used for detection. These aspects might affect the reliability of the experiments that involves induced expression of effector genes to perturb cellular functions. Examples of these potential pitfalls are discussed here using staining of Drosophila brain. (C) 2003 Wiley-Liss, Inc.