Study of dendritic spine compartmentalization : a correlative fluorescence light microscopy-electron microscopy approach

Neurons communicate with each other through synapses. Most excitatory synapses contact small protrusions called dendritic spines. Spines are connected to dendrites by a very thin stalk called the “spine neck” which restricts diffusion between the spine head and its parent dendrite. In consequence, dendritic spines form biochemical micro-compartments. Compartmentalization inside spines is thought to be important for synaptic function, since strong compartmentalization could influence concentration of activated molecules close to synapses during repetitive synaptic stimulations, and also increase depolarization in spine heads. But it is not fully understood how and to what extend spines compartmentalize biochemical signalings.
With two-photon microscopy we measured diffusion coupling between spine heads and parent dendrites of CA1 pyramidal neurons using fluorescence recovery after photobleaching of Alexa dye. Since dendritic spines are below the diffraction limit of light microscopy, it is not possible to measure their detailed morphology with two-photon microscopy. To investigate how spines ultrastructure regulates diffusional coupling to the dendrite, we needed informations about diffusion time constant and spine morphology from the same spine.
We developed a correlative (two-photon microscopy / electron microscopy) approach to reconstruct the precise morphology of dendritic spines where diffusional coupling measurements took place. We found that the outer shape of dendritic spines predicts the diffusional coupling of small molecules. However their diffusional speed in the cytoplasm of spines is 5 times slower than in dendrites. The impact of dendritic spines on electrical compartmentalization depends on spine neck resistance. There is a controversy between studies focusing on dendritic spines morphology (low neck resistance estimates) and studies focusing on synaptic physiology (high neck resistance estimates). All estimates from morphology rested on the assumption that the cytoplasm inside spines and dendrites has homogenous diffusional properties and thus the same resistivity. Here we show that this assumption is not correct. In consequence, we estimate that spine necks resistance approaches 1 G? in some spines, sufficiently high to compartmentalize electrical signals.
For the correlative experiments we used Alexa, a small molecule (1 kDa) roughly the size of ATP or GTP. We were also interested to see if larger molecules like calmodulin (16 kDa) or PKA (38 kDa) behave in the same way. In contrast to Alexa, we found that the diffusional coupling of PA-GFP (27 kDa) and Dextran (70 kDa) could not be predicted from spine shapes. Thus, in addition to the high viscosity of the cytoplasm in all spines, some spines seem to contain an additional size filter that selectively blocks the diffusion of larger molecules. This filter might be important in regulating metaplasticity.
Theoretically, dye particles and other molecules should concentrate in high viscosity compartments. We tested this prediction by creating synthetic images based on 3D reconstructions from our EM data. Indeed, we found that spines appear too bright in the two photon images. Thus, the differences in diffusion speed between spines and dendrites result in different particles densities, making dendritic spines ‘protein enrichment devices’. Finally, we found that the coefficient of diffusion in the cytoplasm is not a static value, but that the viscosity of the entire neuron increases in response to strong depolarization. In summary, dendritic spines appear to be even more complex than previously thought, as we found a new function and a new level of regulation in their functionality. In the light of our findings, the disagreement of previous estimates of spine neck resistance can be readily explained by local differences in cytoplasmic viscosity.