Now single spines: monitoring neuronal membrane potential with submicron and submillisecond resolution

Abstract

Since the spectacular and mysterious morphology of neurons was visualized by Cajal, scientists have wondered what happens when an action potential winds through the trajectory of the dendritic tree and reaches fine branches and synaptic spines. However, until recently we have not had a way to measure the time course, size and shape of a back-propagating action potential through a neuron's fine branches. Such capability is important to our understanding of how information is processed in the dendritic tree and stored at the synapses. Pioneered by Antic & Zecevic (1995), the ‘inside dye’ technique allows us to monitor membrane potential from hundred of locations on a single neuron. The method is to infuse a voltage-sensitive dye (VSD) into a neuron with a patch pipette, allowing the dye to bind to the cellular membrane from inside, and to measure the dye signals that reflect the transmembrane potentials at different locations on the neuron. Voltage-sensitive dyes work differently from calcium dyes; they have to bind to the cellular membrane in order to measure the voltage across the membrane. Any unbound intracellular dye only adds noise. Thus the signal-to-noise ratio increases in finer branches, because the ratio of membrane area to intracellular volume increases. This method works better in fine branches than in the soma, which is the opposite of the patch-electrode techniques, which work better on larger structures. In a recent issue of The Journal of Physiology, Holthoff et al. (2010) have demonstrated improved methods with greatly increased signal-to-noise ratio. Terminal dendritic branches and individual synaptic spines can be monitored with a sensitivity that was not imagined by the original founders of voltage imaging. As described in Holthoff et al. (2010), the magic is achieved with the help of lasers. At fine neuronal branches the low density of photons limits the signal-to-noise ratio (S/N). The illumination intensity from an arc lamp is limited mainly by factors related to focusability of the emitted light. Solid state lasers, on the other hand, are, in biology, a practically limitless source of intensity. A second advantage of a laser source is the ability to use the wavelength where the dye has the largest fractional fluorescence change. As the report shows, the overall effect of using a near optimal wavelength at high intensity resulted in a S/N improvement by a factor of 10–40. This improvement resulted in a method with sub-micrometer and sub-millisecond resolution (Fig. 1). Figure 1 Locations now accessible for membrane potential measurements What's next? Clearly, this new tool will enable neuroscientists to monitor electrical signals from axons, axon collaterals and axon terminals, as well as from terminal dendrites and individual dendritic spines, all important but tiny parts of neurons that are difficult to probe for electrical signals with any other measurement technique. There are a number of unresolved questions related to the physiology and pathology of these structures that can now be explored by direct measurement. One prominent example is the unresolved electrical role of dendritic spines in neuronal signal processing. Another area of research that will be greatly facilitated by high-sensitivity voltage imaging is the characterization of the currently unknown natural input to a dendritic tree (the spatio-temporal pattern of synaptic inputs). This information is critical for understanding the input–output function of any neuron. More ‘natural’ experiments should also be possible. Thus far, the action potentials in imaging experiments were evoked by some form of artificial stimulation of a neuron. It would be far more interesting to monitor action potentials generated naturally by the network. In a brain slice there are a variety of network events such as up–down states, theta oscillations and gamma oscillations. In these events it would be important to see if the shape of action potentials changes as well as where each action potential is initiated. In addition, most of the inside dye studies label and measure one neuron at a time. Can one inject a number of electrophysiologically identified pyramidal and fast-spiking neurons and watch their interactions during gamma oscillations? It would be possible to directly test the hypothesis that when oscillation frequency increases, the activations of pyramidal and GABAergic interneurons become more in phase (Mann & Mody, 2010). Finally, when the method is used in vivo, with the help of genetically encoded sensors and two-photon microscopy, one should be able to watch spike timing-dependent plasticity (Markram et al. 1997) at fine branches in which the temporal relationships can be largely different from location to location in the same neuronal pair. Several strategies for improving the fractional fluorescence change are under investigation. For example, the signal would be increased if unbound dye is quenched. Or a FRET mechanism or protein sensors may generate larger signals (Bradley et al. 2009). Thus the remarkable achievement of Holthoff et al. may not be the end of the line.