The scarring process occurring after adult central nervous system injury and the subsequent increase in the expression of certain extracellular matrix molecules are known to contribute to the failure of axon regeneration. This study provides an immunohistochemical analysis of temporal changes (8 days to 1 year) in the cellular and molecular response of the Swiss mouse spinal cord to a dorsal hemisection and its correlation with the axonal growth properties of a descending pathway, the serotoninergic axons. In this lesion model, no cavity forms at the centre of the lesion. Instead, a dense fibronectin-positive tissue matrix occupies the centre of the lesion, surrounded by a glial scar mainly constituted by reactive astrocytes. NG2 proteoglycan and tenascin-C, potential axon growth inhibitors, are constantly associated with the central region. In the glial scar, tenascin-C is never observed and the expression of chondroitin sulphate proteoglycans (revealed with CS-56 and anti-NG2 antibodies) highly increases in the week following injury to progressively return to their control level. In parallel, there is an increasing expression of the polysialilated neural cell adhesion molecule by reactive astrocytes. These molecular changes are correlated with a sprouting process of serotoninergic axons in the glial scar, except in a small area in contact with the central region. All these observations suggest that while a part of the glial scar progressively becomes permissive to axon regeneration after mouse spinal cord injury, the border of the glial scar, in contact with the fibronectin-positive tissue matrix, is the real barrier to prevent axon regeneration.

Purkinje cells can survive axotomy for as long as 18 months without retracting their severed axons. During this period of time, the fate of the terminal bulbs of axotomized Purkinje cell axons and their relationship with the glial scar were determined. Terminal axonal sprouting begins three months after the lesion and continuously increases up to 18 months (the longest survival time studied), when the sprouts establish synaptic contacts, mainly on granule cell dendrites at the glomeruli. Cellular changes in the glial scar were analyzed to determine whether the late onset and continuous increase of axonal sprouting could be correlated with an increase of permissive factors and/or a decrease of inhibitory factors for axonal growth. Activated macrophages disappeared much earlier than did the initiation of sprouting. Myelin and its associated neurite growth inhibitory molecules began to decrease from three months after the lesion. This decrease was uneven and not correlated spatially with the sprouting. Reactive astrogliosis was heterogeneous: only some of the reactive astrocytes expressed PSA‐NCAM, the embryonic form of the neural cell adhesion molecule, a permissive substratum for neurite outgrowth. The expression of PSA‐NCAM occurred concurrently with sprouting in the area of gliosis containing Purkinje cell sprouts. Moreover, the ultrastructural study showed that the majority of sprouts (75%) were totally ensheathed by astrocytic processes. Thus, long‐term glial scars are permissive to axonal sprouting, suggesting that reactive astrocytes, either through the expression of permissive molecules or by preventing direct contact between axonal elements and myelin inhibitory molecules, regulate the sprouting. J. Comp. Neurol. 408:399–418, 1999. © 1999 Wiley‐Liss, Inc.