In search for a gold-standard procedure to count motor neurons in the spinal cord.

Counting motor neurons within the spinal cord and brainstem represents a seminal step to comprehend the anatomy and physiology of the final common pathway sourcing from the CNS. Motor neuron loss allows to assess the severity of motor neuron disorders while providing a tool to assess disease modifying effects. Counting motor neurons at first implies gold standard identification methods. In fact, motor neurons may occur within mixed nuclei housing a considerable amount of neurons other than motor neurons. In the present review, we analyse various approaches to count motor neurons emphasizing both the benefits and bias of each protocol. A special emphasis is placed on discussing automated stereology. When automated stereology does not take into account site-specificity and does not distinguish between heterogeneous neuronal populations, it may confound data making such a procedure a sort of "guide for the perplex". Thus, if on the one hand automated stereology improves our ability to quantify neuronal populations, it may also hide false positives/negatives in neuronal counts. For instance, classic staining for antigens such as SMI-32, SMN and ChAT, which are routinely considered to be specific for motor neurons, may also occur in other neuronal types of the spinal cord. Even site specificity within Lamina IX may be misleading due to neuronal populations having a size and shape typical of motor neurons. This is the case of spinal border cells, which often surpass the border of Lamina VII and intermingle with motor neurons of Lamina IX. The present article discusses the need to join automated stereology with a dedicated knowledge of each specific neuroanatomical setting.