UC Berkeley

Synapses are the cellular structures that allow neurons throughout the nervous system to communicate, perform calculations, and store information. Due in part to their small size, complex spatial organization, heterogeneity, stochastic behavior, and plastic nature, monitoring the behavior of networks of synapses in intact neural circuits remains an enormous challenge. Yet, this level of understanding will be critical for identifying the central principles and the specific molecular mechanisms that govern neural function, orchestrate activity in networks of neurons and ultimately direct organismal behavior.

To begin to answer some of these fundamental questions regarding the activity of synapses in intact networks, I turned to the Drosophila larval neuromuscular junction (NMJ). The NMJ is a model glutamatergic synapse that directly controls locomotor behavior in the developing larva. A single larval muscle cell receives primary excitatory input from two glutamatergic axons with different morphologies and functions. Each input forms hundreds of individual heterogeneous synaptic contacts with the target muscle cell. Furthermore, each larva only uses a few hundred motor neurons and muscle cells to accomplish all of its behaviors. Thus, this single system can provide diverse information about synaptic function at multiple levels important to the function of the motor system, from the release of a single vesicle of neurotransmitter (quantum) at a single synapse, all the way up to large-scale coordination of muscles during complex locomotive behavior. Through this work, I implemented several generations of optical tools that provide quantal, single-synapse resolution measurements of multiple modes of synaptic transmission and synaptic plasticity. I generated the experimental and analytic methods that allow these measurements, for the first time, to be applied in vivo in the live, behaving animal to monitor quantal synaptic transmission during native activity. I also combined these optical measurements with the precise optical manipulation of intracellular signaling, genetic mutations, and cell-specific alterations in gene expression or neuronal function. Using combinations of these techniques, I was able to uncover several novel homeostatic mechanisms that help to stabilize synaptic and neuronal function.

First, I utilized a unique combination of optopharmacology (light-controlled ligands) and optogenetic pharmacology (light-controlled proteins) to manipulate the signaling of native and orthogonal presynaptic metabotropic glutamate receptors (mGluRs), while simultaneously monitoring single-synapse quantal transmission at hundreds of synapses. I was able to demonstrate that it is possible to track changes in synaptic strength at multiple synapses, during several forms of plasticity. This included monitoring the number of active synapses, quantal amplitudes and presynaptic quantal release probability (Pr, the likelihood of quantal release), all important factors that work together to regulate synaptic strength. Global activation of mGluRs throughout the NMJ, using an orthogonal, light-agonized mGluR, LimGluR2, produces a powerful and uniform suppression of Pr. However, the NMJ contains a highly heterogeneous collection of synapses that may introduce spatial variability in presynaptic mGluR activation. To address this, I used either local LimGluR2 photoactivation or local glutamate uncaging to activate the native mGluR, DmGluRA. With these methods, I was able to show that local excesses of glutamate and presynaptic G protein signaling, produce a combination of local and global, autoreceptor-dependent, negative feedback suppression of Pr. This mechanism likely stabilizes and balances synaptic strength throughout the NMJ, while maintaining synapse-to-synapse heterogeneity.

Next, I extended these optical measurements to both of the convergent glutamatergic inputs to demonstrate that there are enormous, input-specific differences in spontaneous glutamate release, basal evoked Pr, and short-term plasticity. These differences were found to shape the even more divergent behavior of these two inputs in vivo, during locomotion, in the intact larva. By manipulating the postsynaptic muscle sensitivity to glutamate, I was able to show that there is a long-term, retrograde, homeostatic increase in synaptic strength that only occurs at the physiologically dominant input, while not affecting the neighboring, modulatory input. Mechanistically, this input-specificity was found to be due to differential postsynaptic Ca2+ activity and different activity of postsynaptic Ca2+/calmodulin-dependent kinase II (CaMKII), which likely functions to spatially and temporally integrate postsynaptic activity, shaping the retrograde signals to only one of the presynaptic motor neurons. These signals then appropriately adjust glutamate release in order to maintain effective contraction of the muscle.

I then combined this input-specific, synapse-specific quantal framework with multiple mutations that affect presynaptic active zone (AZ) structure and function. I found that these presynaptic mutations are phenotypically heterogeneous, when observed with such high spatial precision and under multiple, interrelated synaptic transmission modes. These mutations produced differential effects on spontaneous transmission, locally and globally. They differentially affected presynaptic Pr, with one mutation increasing the Pr so much, that single-synapse basal transmission switched from univesicular to multivesicular release. I found that short-term plasticity was less affected by basal Pr, as has been suggested by classic models. Rather, plasticity mechanics are more strongly governed by some, as-yet-to-be-identified, input-specific, frequency-dependent factor. I then demonstrated that these alterations in synaptic function are not entirely due to cell autonomous changes at the NMJ specifically, but reflect the systemic, pan-neuronal function of these proteins. Furthermore, neurotransmitter release-defective mutant NMJs have very different behaviors when considered during traditional, semi-intact recording conditions versus when they are observed truly in vivo. In fact, I demonstrated that input-specific synaptic phenotypes are completely inverted when considered under these two recording conditions, suggesting a phenotypic complexity that we are only beginning to uncover.

Finally, in response to the conclusion that mutations may not produce cell autonomous, synapse-autonomous phenotypes, as predicted, I have begun to develop a platform for understanding molecular mechanisms, and homeostatic compensation more broadly. This system combines quantal imaging in vivo and under controlled levels of activity, with cell-specific gene knockdown and network-level activity analysis. This system will provide a full phenotypic characterization of the function of any gene within a specific cell type in the nervous system, while simultaneously identifying the synapse-specific, input-specific, and cell-specific compensatory mechanisms induced by neuronal dysfunction.

Together, this work demonstrates the exquisite precision and diversity in the regulation of synaptic strength employed by even a relatively simple network of synapses, governing a straightforward process of muscle contraction. In particular, this work uncovers several homeostatic mechanisms that the larva can use to stabilize NMJ function in response to diverse perturbations. These changes alter the activity of individual synapses, individual neurons, and the behavior of the entire motor system. Because many of the molecular mechanisms at work within the Drosophila NMJ are conserved throughout many animals, including the most common mammalian central synapses, the framework provided by this work will help to shape our understanding of the broader mechanisms that regulate synaptic function, and as a consequence regulate neuronal circuit function and behavior.