Analysis of calcium channels in single spines using optical ̄uctuation analysis

Most synapses form on small, specialized postsynaptic structures known as dendritic spines. The in ̄ux of Ca ions into such spinesÐthrough synaptic receptors and voltage-sensitive Ca channels (VSCCs)Ðtriggers diverse processes that underlie synaptic plasticity. Using two-photon laser scanning microscopy, we imaged action-potential-induced transient changes in Ca concentration in spines and dendrites of CA1 pyramidal neurons in rat hippocampal slices. Through analysis of the large trial-to-trial ̄uctuations in these transients, we have determined the number and properties of VSCCs in single spines. Here we report that each spine contains 1±20 VSCCs, and that this number increases with spine volume. We are able to detect the opening of a single VSCC on a spine. In spines located on the proximal dendritic tree, VSCCs normally open with high probability (,0.5) following dendritic action potentials. Activation of GABAB receptors reduced this probability in apical spines to ,0.3 but had no effect on VSCCs in dendrites or basal spines. Our studies show that the spatial distribution of VSCC subtypes and their modulatory potential is regulated with submicrometre precision. We imaged spines and dendrites of CA1 pyramidal neurons ®lled with a Ca-sensitive ̄uorophore (Fig. 1). Action potentials propagate into these dendrites and trigger Ca in ̄ux (D[Ca]MP) by opening VSCCs. We measured action-potential-induced ̄uorescence transients simultaneously in spine heads and in small dendrites, under conditions where changes in ̄uorescence are proportional to changes in the concentration of intracellular free Ca, D[Ca]. The tortuous spine neck serves to diffusionally isolate spine heads from their parent dendrites over timescales of 20±100 ms (ref. 6). Therefore, immediately after an action potential, D[Ca] in spines will be due exclusively to opening of VSCCs on the spine head, with no contribution from dendritic VSCCs. Following an action potential, D[Ca] rose quickly (, 2 ms), showing that VSCCs exist in spine heads and dendrites (Fig. 1b). We determined the types of VSCCs present by the application of pharmacological blockers (Fig. 1c). v-Conotoxin-MVIIC (CTx), a blocker of N/P/Q-type VSCCs, had no effect on D[Ca]AP (dendrites: 95 6 5% of control, spines: 96 6 8%, n = 4). CTx diffused into the slice and blocked presynaptic Ca in ̄ux, as it reduced synaptic potentials by more than 90% in these same cells (EPSP, Fig. 1c). Blockade of L-type channels by nimodipine reduced D[Ca]AP in dendrites but not spines (dendrites: 86 6 8% of control, spines: 99 6 4%, n = 5). Blockade of NMDA-Rs had no effect on D[Ca]AP (dendrites: 95 6 3% of control, spines: 97 6 3% in spines, n = 4). Because blocking L-, N-, Pand Q-type VSCCs did not affect D[Ca]AP in the spine, and low-voltageactivated channels are inactivated at resting membrane potentials of our experiments (-60 to -68 mV), action-potential-evoked in ̄ux of Ca in spines is probably primarily through R-type channels. Because the areas of spine membranes are small (, 0.5 mm), spines are expected to contain fewer VSCCs and to show larger ̄uctuations in D[Ca]AP than dendrites. Larger trial-to-trial ̄uctuations in ̄uorescence were observed in spines than on their parent dendrites (Fig. 1d). Fluctuations were stationary and subsequent trials were uncorrelated (Fig. 2a), indicating that the number of VSCCs does not vary during the experiment. As the low mobility of membrane proteins prevents diffusion of VSCCs into and out of spines during an interstimulus interval (5 s), our measurements have suf®cient time resolution to detect changes in the number of channels if they occur. We analysed trial-to-trial ̄uctuations further to extract the number and opening probability of VSCCs in individual spines. Dark noise (before shutter opening), baseline ̄uorescence (between shutter opening and action potential initiation), and response ̄uorescence (after action potential initiation) were measured during each trial (Fig. 2b, c). We selected for analysis pairs of spine heads and parent dendrites whose ̄uctuations in baseline ̄uorescence were no greater than expected from photon shot noise and dark noise. In these structures, signals were not contaminated by noise sources such as movement of the spine, ̄uorophore bleaching, photodamage, and run-down of calcium currents. Furthermore, ̄uctuations in resting [Ca] due to spontaneous VSCC openings were negligible. Following a back-propagating action potential, the variance of the ̄uorescence increased and was larger than expected from shot noise (Fig. 2b, c). The additional variance was caused by stochastic opening of VSCCs in the spine and dendrite. In a spine containing N channels that open independently with probability p per action potential, the number of channels opened by an action potential is governed by the binomial distribution.