The chief aim of this study was to test the hypothesis that there is a developmental period of the Drosophila larval neuromuscular junction during which synaptic morphological and physiological phenotypes are consolidated. We tested this idea using conditional expression of a transgene that was previously shown to disturb both the development and the physiology of this synaptic connection. We observed that there are indeed sensitive periods of synaptic development during the first one to two days of larval life, but that synaptic physiology seems to be more sensitive to disturbance than synaptic morphology. These are important data that extend our knowledge of critical periods of neural development to the level of synapses of identifiable neurons in a genetically tractable model system.
The idea of a critical period in neural development is a central tenet in the field of neurobiology. The classic work of Lorenz in the behavioral realm and Hubel and Wiesel on the development of ocular dominance columns are the first instances in the larger field of neural plasticity, which includes synaptic competition and synaptic refinement . Indeed, plasticity in the visual system has been one of the most extensively studied models of critical periods in development, although critical periods have been observed in a wide variety of mammalian sensory systems . Analysis of the mouse NMJ has shown activity-dependent synaptic competition to establish the mature NMJ [24, 25] but there are relatively fewer studies of critical periods on the motor side of the nervous system.
Rarer are studies on critical periods in neural development of invertebrates. Axonal pruning during development within the CNS of Drosophila, particularly the olfactory system and the mushroom bodies, provide some examples [26–29] although this development has not strictly been shown to have activity-dependent critical periods. Additionally, studies on neural rearrangements during insect metamorphosis are informative in this regard, having a strongly-timed developmental component [30, 31]. An interesting description of critical periods within the mechanosensory system of C. elegans using a sensory deprivation design  illustrates the likelihood of critical periods in invertebrate sensory systems.
A useful comparator study is that of Jarecki and Keshishian . They examined the role of neural activity in synaptic development, using synaptic target selection as their measurement. The Drosophila NMJ is characterized by precise muscle target selection by the innervating motor neurons. Jarecki and Keshishian observed that the occurrence of ectopic targeting can be influenced by manipulating neural activity; reducing activity leads to an increase in the appearance of inappropriate synaptic connections. Importantly, they showed that manipulations performed during embryogenesis and the first larval instar was important to observe the ectopic synapses, thus indicating a sensitive period in synaptic target selection during this early time window. Their observation is largely in congruence with our results, albeit using a different molecular perturbation; we have additionally added the profile of native NMJ target neurons and the physiological changes to this early critical period in Drosophila NMJ development.
Although our results are quite clear that there is a sensitive time window during early development, our interpretation of these data are somewhat limited. Firstly, the primary cause of the synaptic phenotypes induced by dNSF2E/Q is not well understood. While NSF is well known as an ATPase that disassembles the SNARE complex following vesicle fusion, NSF has been increasingly shown to have non-SNARE roles in the cell [34, 35]. Indeed, we have shown that expression of dNSF2E/Q is associated with a variety of phenotypes at this NMJ for which one would not normally invoke SNARE disassembly as a cause. Such phenotypes include alteration of the synaptic actin cytoskeleton and impaired synaptic vesicle dynamics . Our earlier genetic studies [37, 38] also identified a variety of genes that interact with dNSF2E/Q NMJ overgrowth, including genes that encode cytoskeletal elements, transcription factors, and signaling proteins. While the present study does not shed light on the phenotypic action of dNSF2E/Q, the transgene has proven to be a useful tool for dissecting sensitive developmental time windows.
Additionally, although we have demonstrated control of the synaptic phenotypes using the Gal4 / Gal80 competition approach, the use of temperature to regulate a temperature sensitive mutation of Gal80 limits our ability to more precisely define when the critical period starts and ends. Use of other tools such as the Geneswitch system  or the newer Q system  may help in this regard.
With these limitations in mind, it is useful to project the types of synaptic changes that may occur during the consolidation process. It is particularly interesting that it appears that once the NMJ starts to develop under normal conditions, it continues to develop normally even in the face of a perturbing factor. This implies that the cytoskeletal and cell adhesion mechanisms necessary for proper synaptic architecture are present at the required levels and resistant to the changes induced by dNSF2E/Q. Alternatively, repair mechanisms that counteract the action of dNSF2E/Q are fully established in the first two days of synaptic development. Delineating between these possibilities will be the subject of future studies.