We were also motivated to study early synaptic rearrangement because of uncertainty about its role in circuit development. In particular, we were interested to know whether early synaptic rearrangements are ostensibly

minor refinements that “functionally validate” or “error correct” connectivity patterns (Cowan et al., 1984 and Jacobson, 1969) or perhaps have a more central role of specifying the connectivity. In this work, we use techniques that give direct measures both NSC 683864 chemical structure of the size of motor units (divergence) and the number of axons that innervate each muscle fiber (convergence). Our results show that at birth, axons transiently project to nearly an order of magnitude more muscle fibers

than later and that each neuromuscular junction is innervated by roughly 10-fold more axons. The many extra axonal branches originate from the same neurons that provide the few branches that ultimately survive development and are spatially intermingled with the surviving branches. Thus, it is likely that local interactions at each postsynaptic target cell, such as those mediated ALK inhibitor by activity-dependent synaptic competition, not only underlie the final stages of minor refinement in the second postnatal week in mice but also the massive early loss of synaptic connections beginning just before birth. In order to reconstruct motor axon arbors in fetal and very young animals, we used “YFP-H” mice that we had previously found expressed cytoplasmic yellow fluorescent protein (YFP) in very small numbers of motor axons (Feng et al., 2000). Because of the developmental regulation of the promoter used in these transgenic because animals (from the thy1 gene), our previous studies detected very faint or no fluorescence in these and other subset-expressing lines prior to postnatal day (P) 7 ( Keller-Peck et al., 2001).

However, when we amplified the signal by fluorescent immunohistochemistry, we could clearly detect YFP-expressing axons in very young animals ( Figure 1), albeit rarely. We surveyed ∼4,000 neck muscles (the sternomastoid, cleidomastoid, and clavotrapezius) between embryonic day (E) 16 and P4 and found 23 in which a motor axon arbor was labeled sufficiently well that all of the branches were visible to each terminal. We discarded approximately ten other motor axons in which the labeling was deficient or in which inadvertent damage to the muscle precluded quantifying the full complement of branches. The 23 well-labeled motor units were reconstructed by stitching together confocal image stacks obtained at the diffraction limit using high numerical aperture (NA) oil objectives.