Final Report Summary - SYNAPSEMAP (Dynamic super-resolution mapping of neurotransmitter release events and synaptic glutamate receptors)
Information transfer between nerve cells in the brain occurs via specialized contacts called synapses. The chemical synapse is classically depicted as having a post-synaptic scaffold, where receptor ion channels are anchored opposite vesicular neurotransmitter release sites of a presynaptic terminal. In this fellowship, I focused my attention on the common excitatory chemical synapse, where glutamate released from the presynaptic terminal binds to alpha-amino-3-hydroxyl-4-isoxazole-propionic acid (AMPA)-type glutamate receptors to depolarize the membrane of the post-synaptic nerve cell.
Previous studies from the host laboratory have revealed several important details regarding the fundamental organization and operation of excitatory synapses, which illustrate the motivation for this fellowship. Firstly, AMPA receptors (AMPARs) are not randomly distributed in the post-synaptic density (PSD) but are in fact organized in discrete submicron-sized clusters, called nanodomains. Secondly, AMPARs are not chronically anchored to the PSD, as is commonly depicted, but are in dynamic equilibrium with a large pool of freely diffusing non-synaptic surface AMPARs. Furthermore, the rates of diffusion and exchange of AMPARs between these two pools can effectively out-compete recovery of receptors from desensitization, thus bypassing what is effectively a low-pass filter to incoming glutamate and modifying short-term changes in synaptic responses.
In light of these observations, the nanoscale positioning of vesicular glutamate release relative to AMPAR nanodomains has become the next crucial question. This is especially evident when considering evidence that the glutamate concentration drops steeply with distance from the release location, and that the occupancy of AMPAR gating states (e.g. active, desensitized) varies considerably with glutamate concentration. During this fellowship I have studied the relationship between the location of putative markers of presynaptic vesicular glutamate release with the nanodomains of postsynaptic AMPARs, and the impact that this could have on short-term synaptic plasticity. During my fellowship I used a variety of different experimental approaches including molecular biology, two-colour super-resolution methods and slice electrophysiology. I have followed the working plan proposed in Annex I. In the summary below I present some of the results for each one of the aims of my project.
SUPER RESOLUTION STRATEGIES TO IMAGE RELEASE LOCATIONS AT PRESYNAPTIC TERMINALS
After trying many approaches to image release locations, including the Q-dot and fluorophore quenching methodologies, I found that the most straightforward and effective approach to image potential sites of vesicular glutamate release was to use cutting-edge PhotoActivated Localization Microscopy (PALM) in living cultured hippocampal neurons expressing EOS-tagged presynaptic protein candidates, which typically reside close to vesicle fusion events. I tested five different proteins: Syntaxin, Cask, Munc13, CaV2.1 and CaV2.2 (Figure 1A).
I made fusions of candidate proteins with the photoconvertable protein mEOS2 and performed super-resolution imaging using PALM on transfected cultured neurons. Despite being able to see some enrichment in putative active zones, Cask and Syntaxin were surprisingly mobile with unimodal distributions of diffusion coefficients in the presynaptic terminal that were indifferent from their diffusive properties globally in the cell. Furthermore, for these proteins I could not identify hotspot regions of confinement of these proteins within the active zone. In contrast, Munc13 gave a distribution of diffusion coefficients that was shifted to lower, less mobile values in putative presynaptic elements compared to globally, and these diffusion properties were associated with areas of confinement within the active zone (Figure 1B). PALM results for the two calcium channels were even better, with highly distinct synaptic diffusion (Figure 1C) and confinement (Figure 1D). Therefore, PALM imaging of calcium channels provided a suitable means to identify putative release locations beyond the diffraction limit in living hippocampal neurons.
TRAJECTORY MAPS OF ENDOGENOUS AMPA RECEPTORS RELATIVE TO PUTATIVE RELEASE LOCATIONS
The lab previously established a continuous antibody labeling single particle tracking (SPT) method called U-PAINT that can obtain high numbers of synaptic trajectories of endogenous plasma membrane proteins. I combined U-PAINT using an ATTO647N-coupled antibody against the extracellular domain of GluA2, with PALM of CaV2.2 as described above to provide simultaneous two-colour super resolution live-cell imaging. GluA2 detections showed sub-synaptic nanodomain organization, and their trajectory tracks clearly indicated substantial diffusion in the postsynaptic structures (Figure 1E). PALM detections of CaV2.2 partially matched with GluA2 confinement areas and overlapped considerably with the GluA2 trajectories (Figure 1E). These images provide the first insight into the dynamic behavior of AMPA receptors with relation to putative vesicular glutamate release locations.
PROBING THE OCCUPANCY OF AMPA RECEPTOR GATING STATES FOLLOWING SYNAPTIC GLUTAMATE RELEASE
My trajectory maps of hippocampal synapses indicate that AMPA receptor diffusion is important for the receptors to fully cover the area of active zone where release locations might be found. To understand how this is important for synaptic transmission I have combined cross-linking (X-linking) methods to immobilize the AMPA receptors, with pharmacological and genetic methods to reveal the receptor gating states during synaptic transmission. I expressed recombinant pHGFP-tagged AMPA receptors that are biotinylated on a specific N-terminal peptide sequence by a coexpressed ER-localized biotin ligase. When these AMPARs traffic to the neuronal plasma membrane their biotinylated region is exposed to the extracellular milieu and is available to X-link with tetravalent biotin-binding proteins, such as avidin (Figure 2A). Indeed, fluorophore-coupled avidin surface labeled live cultured neurons expressing biotinylated pHGFP-tagged AMPA receptors and prevented fluorescence recovery after photobleaching (FRAP) at spines (Figure 2B). The intrinsic function of these recombinant receptors, X-linked or not, were functionally identical to untagged receptors (Figure 2C).
I then expressed these biotinylated receptors in organotypic slices to confirm that X-linking could indeed be achieved in this more intact hippocampal preparation (Figure 3A,B) and that the receptors were functionally incorporated at CA3-CA1 synapses as monitored by the rectification of evoked EPSCs (Figures 3C,D). I also confirmed that X-linking of biotinylated AMPA receptors led to a specific decrease in the paired-pulse response of CA3-CA1 synapses at 20 Hz (Figure 3E), consistent with previous reports from the host laboratory using microiontophoretic glutamate application to spines after antibody X-link in cultured neurons. I was then in a good position to conduct the experiments for the third objective. I examined 20 Hz EPSP trains onto CA1 pyramidal cells expressing biotinylated AMPARs +/- X-link, either in control condition, in the presence of the low affinity antagonist gamma-DGG alone (to prevent receptor gating in response to low glutamate concentrations) or in combination with cyclothiazide (to prevent desensitization of all receptors). I found that gamma-DGG reveals a greater difference +/- X-link, which suggests that the size of synaptic glutamate transient modifies the effect of receptor diffusion on short-term plasticity (Figure 4A). To corroborate that this relates to its interaction with desensitization, when I apply cycothiazide the difference +/- X-link is gone, with the responses for both facilitating more than in the control condition (Figure 4A). To take this finding further I created biotinylated AMPA receptor desensitization mutants. The ET/YR mutant enters into a long-lived desensitized state and X-linking this mutant causes a substantial long-lasting depression of EPSCs (Figure 4B). In contrast, the LY mutant cannot desensitize and X-linking it has no effect (Figure 4C).
In summary, diffusion effectively resets synapses to bypass inevitable desensitization to synaptically released glutamate at hippocampal synapses. The pharmacology data is indicates that some of this desensitization could occur to low glutamate concentrations sensed by receptors not directly under vesicle release locations, which is consistent with only partially overlapping localization of AMPA receptor nanodomains with putative vesicular glutamate release locations.
(Please find 4 figures in the attached PDF document)
Previous studies from the host laboratory have revealed several important details regarding the fundamental organization and operation of excitatory synapses, which illustrate the motivation for this fellowship. Firstly, AMPA receptors (AMPARs) are not randomly distributed in the post-synaptic density (PSD) but are in fact organized in discrete submicron-sized clusters, called nanodomains. Secondly, AMPARs are not chronically anchored to the PSD, as is commonly depicted, but are in dynamic equilibrium with a large pool of freely diffusing non-synaptic surface AMPARs. Furthermore, the rates of diffusion and exchange of AMPARs between these two pools can effectively out-compete recovery of receptors from desensitization, thus bypassing what is effectively a low-pass filter to incoming glutamate and modifying short-term changes in synaptic responses.
In light of these observations, the nanoscale positioning of vesicular glutamate release relative to AMPAR nanodomains has become the next crucial question. This is especially evident when considering evidence that the glutamate concentration drops steeply with distance from the release location, and that the occupancy of AMPAR gating states (e.g. active, desensitized) varies considerably with glutamate concentration. During this fellowship I have studied the relationship between the location of putative markers of presynaptic vesicular glutamate release with the nanodomains of postsynaptic AMPARs, and the impact that this could have on short-term synaptic plasticity. During my fellowship I used a variety of different experimental approaches including molecular biology, two-colour super-resolution methods and slice electrophysiology. I have followed the working plan proposed in Annex I. In the summary below I present some of the results for each one of the aims of my project.
SUPER RESOLUTION STRATEGIES TO IMAGE RELEASE LOCATIONS AT PRESYNAPTIC TERMINALS
After trying many approaches to image release locations, including the Q-dot and fluorophore quenching methodologies, I found that the most straightforward and effective approach to image potential sites of vesicular glutamate release was to use cutting-edge PhotoActivated Localization Microscopy (PALM) in living cultured hippocampal neurons expressing EOS-tagged presynaptic protein candidates, which typically reside close to vesicle fusion events. I tested five different proteins: Syntaxin, Cask, Munc13, CaV2.1 and CaV2.2 (Figure 1A).
I made fusions of candidate proteins with the photoconvertable protein mEOS2 and performed super-resolution imaging using PALM on transfected cultured neurons. Despite being able to see some enrichment in putative active zones, Cask and Syntaxin were surprisingly mobile with unimodal distributions of diffusion coefficients in the presynaptic terminal that were indifferent from their diffusive properties globally in the cell. Furthermore, for these proteins I could not identify hotspot regions of confinement of these proteins within the active zone. In contrast, Munc13 gave a distribution of diffusion coefficients that was shifted to lower, less mobile values in putative presynaptic elements compared to globally, and these diffusion properties were associated with areas of confinement within the active zone (Figure 1B). PALM results for the two calcium channels were even better, with highly distinct synaptic diffusion (Figure 1C) and confinement (Figure 1D). Therefore, PALM imaging of calcium channels provided a suitable means to identify putative release locations beyond the diffraction limit in living hippocampal neurons.
TRAJECTORY MAPS OF ENDOGENOUS AMPA RECEPTORS RELATIVE TO PUTATIVE RELEASE LOCATIONS
The lab previously established a continuous antibody labeling single particle tracking (SPT) method called U-PAINT that can obtain high numbers of synaptic trajectories of endogenous plasma membrane proteins. I combined U-PAINT using an ATTO647N-coupled antibody against the extracellular domain of GluA2, with PALM of CaV2.2 as described above to provide simultaneous two-colour super resolution live-cell imaging. GluA2 detections showed sub-synaptic nanodomain organization, and their trajectory tracks clearly indicated substantial diffusion in the postsynaptic structures (Figure 1E). PALM detections of CaV2.2 partially matched with GluA2 confinement areas and overlapped considerably with the GluA2 trajectories (Figure 1E). These images provide the first insight into the dynamic behavior of AMPA receptors with relation to putative vesicular glutamate release locations.
PROBING THE OCCUPANCY OF AMPA RECEPTOR GATING STATES FOLLOWING SYNAPTIC GLUTAMATE RELEASE
My trajectory maps of hippocampal synapses indicate that AMPA receptor diffusion is important for the receptors to fully cover the area of active zone where release locations might be found. To understand how this is important for synaptic transmission I have combined cross-linking (X-linking) methods to immobilize the AMPA receptors, with pharmacological and genetic methods to reveal the receptor gating states during synaptic transmission. I expressed recombinant pHGFP-tagged AMPA receptors that are biotinylated on a specific N-terminal peptide sequence by a coexpressed ER-localized biotin ligase. When these AMPARs traffic to the neuronal plasma membrane their biotinylated region is exposed to the extracellular milieu and is available to X-link with tetravalent biotin-binding proteins, such as avidin (Figure 2A). Indeed, fluorophore-coupled avidin surface labeled live cultured neurons expressing biotinylated pHGFP-tagged AMPA receptors and prevented fluorescence recovery after photobleaching (FRAP) at spines (Figure 2B). The intrinsic function of these recombinant receptors, X-linked or not, were functionally identical to untagged receptors (Figure 2C).
I then expressed these biotinylated receptors in organotypic slices to confirm that X-linking could indeed be achieved in this more intact hippocampal preparation (Figure 3A,B) and that the receptors were functionally incorporated at CA3-CA1 synapses as monitored by the rectification of evoked EPSCs (Figures 3C,D). I also confirmed that X-linking of biotinylated AMPA receptors led to a specific decrease in the paired-pulse response of CA3-CA1 synapses at 20 Hz (Figure 3E), consistent with previous reports from the host laboratory using microiontophoretic glutamate application to spines after antibody X-link in cultured neurons. I was then in a good position to conduct the experiments for the third objective. I examined 20 Hz EPSP trains onto CA1 pyramidal cells expressing biotinylated AMPARs +/- X-link, either in control condition, in the presence of the low affinity antagonist gamma-DGG alone (to prevent receptor gating in response to low glutamate concentrations) or in combination with cyclothiazide (to prevent desensitization of all receptors). I found that gamma-DGG reveals a greater difference +/- X-link, which suggests that the size of synaptic glutamate transient modifies the effect of receptor diffusion on short-term plasticity (Figure 4A). To corroborate that this relates to its interaction with desensitization, when I apply cycothiazide the difference +/- X-link is gone, with the responses for both facilitating more than in the control condition (Figure 4A). To take this finding further I created biotinylated AMPA receptor desensitization mutants. The ET/YR mutant enters into a long-lived desensitized state and X-linking this mutant causes a substantial long-lasting depression of EPSCs (Figure 4B). In contrast, the LY mutant cannot desensitize and X-linking it has no effect (Figure 4C).
In summary, diffusion effectively resets synapses to bypass inevitable desensitization to synaptically released glutamate at hippocampal synapses. The pharmacology data is indicates that some of this desensitization could occur to low glutamate concentrations sensed by receptors not directly under vesicle release locations, which is consistent with only partially overlapping localization of AMPA receptor nanodomains with putative vesicular glutamate release locations.
(Please find 4 figures in the attached PDF document)
