The main results were described in three publications (Livet et al., Neuron 35:877, 2002; Haase et al., Neuron 35:893, 2002; Helmbacher et al., Neuron 39:767, 2003). This series of studies used as a model system a particular group of motor neurons in the spinal cord, which innervate two flank muscles called cutaneus maximus (CM) and latissimus dorsi (LD). Based on project result 12225 (above), we knew that HGF signalling is required for the proper innervation of these muscles and analysed the signalling pathways required downstream of the Met receptor. These three papers concerned the coordinated signalling within motor neurons required to achieve muscle innervation. One key aspect of motor neuron positioning is physical segregation into motor pools within the cord, each innervating a single muscle. Neither the mechanism by which these pools are formed, nor the link between cell body grouping and target innervation, were known. The CM and LD pools specifically express the PEA3 transcription factor. In collaboration with S. Arber (Basel) and T. Jessell (NY), we showed that in the absence of PEA3 these motoneurons are mispositioned within the spinal cord and do not correctly innervate their target muscles. Downstream of PEA3, we identified potential effectors (cadherin 8, semaphorin 3E).
Upstream, we showed that the neurotrophic factor GDNF, expressed specifically in the brachial plexus and then in the CM and LD muscles, is necessary to trigger expression of PEA3. Consequently, GDNF and PEA3 knockout mice are phenocopies at this level, and this new signalling pathway provides an unexpected link between the peripheral and central factors that contribute to the formation of a motor unit. Lastly, we discovered an additional level of regulation of the formation of the same pools. After initial induction by GDNF, the recruitment of the complete set of PEA3 motoneurons is brought about by a non-cell-autonomous action of the HGF/Met system, thus explaining the deficits first observed in Met mutant mice.
These results show that interactions between different neurotrophic factors and signalling pathways are remarkably complex, and can be specific to very limited neuronal ensembles. This undoubtedly has consequences for strategies involving neurotrophic treatments of neurodegenerative disease.