The intricate web of synaptic connections between neurons is thought to be the physical substrate that stores information in the brain, with memories encoded by the selective increase in the strength of some synapses in the context of others that remain unchanged or decreased. Changes in synaptic strength are associated with changes in synaptic structure . Over the last 5 decades researchers have sought to find these changes in the memory trace or neural engram , but as yet no methods are available to examine synaptic strength in an intact animal across the whole brain. We have succeeded in generating genetically encoded high affinity probes for three synaptic proteins that allow synaptic connections to be visualized and manipulated in real time in vivo for the first time [3, 4]. Such probes now enable us to answer fundamental questions about diverse topics, ranging from synaptic plasticity, development and memory, to addiction and neurodegenerative disease, opening many possible avenues of inquiry with significant therapeutic applications.
We have created a family of vital molecular imaging probes known as FingRs (Fibronectin intrabodies generated with mRNA display), that bind to endogenous PSD95, Gephyrin or Rim, all proteins that form the supporting scaffold for receptors and ion channels at pre- or postsynaptic sites. FingRs fused with fluorescent proteins can be expressed in neurons either in culture or in transgenic animals. Because the amount of PSD95, Gephyrin and Rim at synapses is proportional to the physiological strength of that synapse, the fluorescence produced by labeled FingRs can be used to map the precise location and strength of synaptic connections. FingRs that are fused to an E3 ligase cause their target protein to be ubiquitinated and then degraded rapidly, specifically and inducibly. Elimination of synaptic scaffolding proteins results in receptors being released from the synapse and causes the synapse to be functionally erased. Thus, using FingRs we can not only visualize synaptic connections, but modify them with extraordinary temporal and spatial precision.
Current Projects in the lab:
Sleep: We are currently imaging distributions of endogenous PSD95 and Gephyrin, labeled with FingRs, in the brains of living zebrafish using two photon laser scanning microscopy (Fig. 4) and two photon light sheet (SPIM) microscopy (in collaboration with Scott Fraser at USC). We convert these images into a dataset consisting of the spatial coordinates, synapse strengths and polarities for each FingR-labeled synapse (Fig. 5), which we refer to as the “synaptome”. The software for generating synaptomes from stacks of images was developed in our collaboration with Carl Kesselman, a computer scientist at USC. This software can also translate and rotate a synaptome, allowing it to be aligned with a synaptome collected from a fish in a different orientation. This allows us to compare synaptomes taken of the same brain at different times and calculate the changes between them. By comparing the synaptomes measured before and after sleep vs. before and after a period of wakefulness we will be able to directly test one of the most controversial hypotheses in sleep research – that during sleep synapses are “renormalized” by having their strength reduced in a proportional manner so that the overall pattern of synaptic inputs is preserved . A number studies using indirect methodology have obtained conflicting results, with some finding that sleep decreases synaptic strength , while at least one finds a correlation between sleep and strengthened synaptic connections . By directly measuring changes in synapses between cells in different brain areas, we hope to settle this critical question and further explore the role of sleep in shaping synaptic structure.
Synaptic encoding of information:In 1894 Ramon y Cajal first suggested that information is stored in the changing patterns of synaptic connections within the brain . Although many experiments over the last 120 years have supported this statement, it has not been possible technically to examine how information is encoded in synaptic structure. We propose to use FingRs to measure changes in the synaptome of a juvenile zebrafish that occur when the fish learns a specific type of information through a behavioral manipulation. We will start with the simplest paradigm that has been shown to elicit learning in larval zebrafish, sound habituation. By repeatedly exposing zebrafish to loud, sudden noises, the startle response can be habituated so that the fish no longer responds to the stimulus. It has been documented that this form of learning is protein synthesis- and NMDA receptor-dependent . By changing the characteristics of the tones to which the fish are habituated and then correlating those characteristics with changes in the synaptome, we will probe how different sound qualities are encoded in the zebrafish brain. A second, more sophisticated behavioral paradigm that we will explore is fear conditioning. In particular, we can pair a conditioned stimulus, such as being placed in a particular part of the tank, with an electric shock. By comparing the changes in synaptic structure that occur when locations are paired with the shock, we can identify changes in the visual cortex and elsewhere that mediate the association of a particular place with an aversive stimulus. We can also pair visual stimuli with a positive unconditioned stimulus, such as allowing a zebrafish to observe additional zebrafish . By comparing synaptic changes with the same conditioned stimulus and different unconditioned stimuli, we will be able determine in a precise and quantitative manner exactly how the brain assigns positive or negative associations with specific memories of specific places.
Mapping neuronal projections: In order to truly understand brain function it is necessary to understand the circuitry that underlies the generation of patterns of neuronal activity. A first step in this process is to map the connections of multiple neurons simultaneously onto their targets. Although several different techniques for tracing neuronal connections have been developed in the last decade, none is capable of tracing the anterograde or retrograde connections of two or more closely spaced neurons simultaneuously in vivo [13, 14]. By combining PSD95, Gephyrin and Rim FingRs with a variant of the Brainbow technique, we are developing a procedure for tracing the anterograde or retrograde connections of groups of neurons. In particular, we will label a group of neurons so that each has all of its presynaptic and postsynaptic terminals as well as its nucleus labeled with one of up to one hundred different colors. Thus, individual synapses will be labeled with two different colors – one corresponding to the presynaptic site and a second corresponding to the postsynaptic site. By matching the corresponding pre- and postsynaptic colors with nuclei of the same color, it will be possible to trace projections from nucleus to presynaptic site to postsynaptic site to the nucleus of the postsynaptic cell. Using software that computes the pre- and postsynaptic synaptomes, matches them at synapses and then traces projections in this manner we will be able to simultaneously trace neuronal projections to and from groups of closely positioned cells in parallel in vivo.
Sculpting synaptic connectivity: In addition to using FingRs to visualize endogenous proteins, we have also used them to target functional domains, such as E3 ligases, which catalyze the transfer of ubiquitin onto lysine residues of proteins that are to be degraded. For instance, the Gephyrin FingR can be fused to an E3 Ligase domain (Gephyrin.FingR-E3 Ligase) to create a protein that binds to Gephyrin and then mediates its ubiquitination and degradation. This construct works very efficiently, with virtually 100% of Gephyrin degraded within 24 hours of transfection (Fig. 6). Furthermore, the degradation of target is very specific. There is virtually no reduction in the amount of GABA receptor expressed in cells after degradation of Gephyrin, a result that we have shown in dissociated culture using western blots, immunocytochemistry and electrophysiology. Remarkably, expression of Gephyrin FingR-E3 Ligase resulted in an 80% decrease in frequency of IPSCs and a 30% decrease in their amplitude (Fig. 7). Thus, using the Gephyrin.FingR-E3 Ligase we can drastically reduce phasic inhibition onto individual neurons, while leaving tonic inhibition intact. Experiments with PSD95 FingR-E3 Ligase show similar results for excitatory synapses.
Because the ablating FingR is a bifunctional molecule (comprised of the FingR and the E3 ligase domain) it is possible to make it inducible using the Rapamycin, FKBP, FRB system . For instance, we fused FKBP to the FingR (labeled with GFP) and FRB to the E3 ligase. Expression of these two fusion proteins does not affect expression of endogenous Gephyrin. However, addition of Rapamycin leads to reconstitution of a working ablating FingR, which in turn leads to the efficient degradation of the endogenous target protein within less than 3 hours. We have also used this system with a caged version of rapamycin developed by Alex Deiters  to efficiently degrade Gephyrin in neurons exposed to low levels of UV light for 10 minutes (Fig. 8). Therefore, we can specifically and efficiently change the synaptic connectivity between neurons with extremely precise temporal and spatial control.
FingRs have significant advantages over traditional forms of knocking out proteins, which rely on deleting nucleic acids such as DNA (targeted gene deletion) and RNA (siRNA). Although these methods have proven to be very specific and effective, they are indirect; essentially the source of the protein is eliminated so that when the protein is degraded it is not replaced. However, elimination of the protein can take a week or more, which limits the type of experiments that can be performed, and allows for compensatory up-regulation of gene expression, which can mask the effects of eliminating the protein. In dramatic contrast, ablating FingRs directly mediate degradation of the target protein, which results in elimination of that protein in minutes to hours. Using light activation we can also delete proteins within very circumscribed and targeted subcellular locations, for instance, eliminating individual, or arbitrary combinations of, synapses. We can, as well, target proteins that are in specific conformations or that have specific post-translational modifications. Finally, FingRs allow endogenous proteins to be visualized before and during ablation. Thus, we can now eliminate a specific protein in a particular cell or subcellular region within a living animal, watch as that protein is degraded, and then measure the precise effect of manipulating that protein before compensatory mechanisms have obscured the effect.
1. Transformative R01 "Dynamic mapping of the complete synaptome using recombinant probes"
2. Eureka R01 "Recombinant antibodies for cytoplasmic, nuclear and transmembrane proteins"
3. R01 "Molecular probes to visualize endogenous synaptic proteins in vivo"
4. Human Frontiers Science Program "Probing and controlling single neuron synaptic function in the brain with light, intrabodies and sensors".
5. McKnight Technolgical Innovations in Neuroscience award "Ablating intrabodies- tools for direct ablation of endogenous proteins".