Glutamate-triggered translocation of PKCγ into spines of hippocampal neurons. (A) Schematic view of the reversible plasma membrane–translocation process of signaling proteins. C2, C1, and PH domains are the main classes of known lipid interaction domains that can reversibly translocate signaling proteins to the plasma membrane in response to increases in Ca²⁺, diacylglycerol, and PIP_n lipids, respectively. PIP_n refers to PIP₂ or PIP₃. (B) Confocal microscope images showing the translocation of endogenous PKC (green in the merged image) into spiny structures that also correspond to sites stained with the postsynaptic marker PSD95 (red). A control cell is shown in the top panel, whereas the bottom panels show a different cell that was fixed and stained for 1 min after the application of glutamate. Arrows show examples of sites containing endogenous PSD95, to which PKCγ translocates after stimulation. (C) YFP-PKCγ also translocates to local sites that partially overlap with the postsynaptic marker CFP-PSD95. After stimulation, the fluorescence intensity of YFP-PKCγ was markedly enriched in spines as well as in other dendritic processes compared with nearby plasma membrane sites. Images are shown for two different cells. The left panel shows the magnified image of a dendrite of a neuron transfected with YFP-PKCγ before (top) and 102 s after stimulation with 30 μM glutamate (bottom). Images of CFP-conjugated PSD95 from corresponding time points are shown in the middle panel. In the third panel, the images of YFP-PKCγ (green) and CFP-PSD95 (red) before and after stimulation were overlaid to show the translocation of PKC into spiny structures that protrude from the main dendrite that also contain the PSD95 protein (arrows). Another cell is shown in the series of panels on the right.

Kinetics of PKCγ translocation into spines are delayed and depend on spine morphology. (A) Time courses showing kinetics of translocation of GFP-conjugated PKCγ in four dendritic spines of the same neuron (Fig. S2, available at http://www.jcb.org/cgi/content/full/jcb.200503118/DC1). (B) Histograms depicting the observed delay for PKCγ translocation into dendritic spines (bottom) compared with the translocation of PKCγ to the plasma membrane of the cell soma (top). The time required for individual spines (n = 100 spines from 11 cells) and soma plasma membranes (n = 27 cells) to attain half-maximal PKCγ accumulation was measured from the point of stimulus addition. (C) Histograms showing the distribution of translocation half-times for spines with or without necks. The translocation of GFP-PKCγ to spines that have a neck (scored only if the spine showed a well-defined narrow neck joining a mushroom-shaped spine (see Fig. S2) is significantly delayed compared with spines with no neck. The half-time of translocation was derived from an exponential fit of these traces (see Materials and methods). 21 spines without a neck and 15 spines with a neck from five different neurons were examined. Individual translocation traces in spines without a neck (green) and with a neck (blue) are also shown from three different neurons.

Measurement of the diffusion coefficient of dendritic PKCγ in the presence and absence of stimulation is sufficiently fast to explain the observed translocation behavior. (A, left) Four images taken from a time series of 20 used for the measurement of the PKCγ diffusion coefficients. The Gaussian bleach pulse was applied between the first and second image. (right) Linescan analysis of the recovery of the bleach profile from the series of images shown on the left. (B) The same experiment as in A performed 8 s after stimulation with glutamate and with a longer time course to account for the slower recovery of PKCγ (a total of 70 images were acquired during this experiment). Arrows point to the site of photobleaching within the branch. (C) Bar representation of the dendritic PKCγ diffusion coefficients measured in the absence (n = 15) and presence (n = 9) of glutamate stimulation. In the presence of glutamate, the diffusion of PKC was ∼16 times slower than the diffusion of PKC in the cytosol in the absence of stimulation (0.33 ± 0.07 SEM and 5.45 ± 0.85 SEM, respectively), and it was also slower than the diffusion of membrane-associated YFP-CAAX (diffusion coefficient 0.52 ± 0.04 SEM; n = 13). Error bars represent the SEM for these experiments. (D) The diffusion coefficients from C were used to produce a schematic view illustrating the range of action of PKCγ in the absence (top) and in the presence (bottom) of a stimulus.

Spine necks delay the translocation of PKCγ and reduce the accessibility of a membrane-bound marker protein. This suggests that spine necks isolate spines against short stimuli while still enabling enrichment of PKCγ as a result of prolonged responses or integrated receptor and electrical activity. (A) Photobleaching of YFP-PKCγ in individual spines is used to measure the rate of PKCγ exchange across the spine neck before (top) and after (bottom) stimulation with 30 μM of the glutamate agonist NMDA. Arrows indicate the spine that was photobleached. (B) Comparison of the average photobleaching recovery rates in unstimulated spines (n = 17) and in spines after glutamate addition (n = 6). (C) Test for potential membrane-diffusion barriers in the neck of dendritic spines. Photobleaching measurements of a membrane-inserted YFP-CAAX protein showed recovery rates comparable to that of PKCγ in the presence of glutamate (n = 12 spines), suggesting that the neck of spines delays diffusion access within the membrane but that there is no actual physical barrier to membrane diffusion.Error bars show SEM. (D) Example of a cell used in the measurements for C. Four confocal images showing the slow recovery of YFP-CAAX fluorescence after photobleaching of two spines (arrows) situated near each other on a thick dendrite.

Translocation of GFP-PKCγ to the plasma membrane in response to glutamate application and electrical field stimulation. (A) Series of confocal images showing the reversible glutamate-triggered (30 μM) plasma membrane translocation of GFP-PKCγ. Glutamate was added after the first image. (B) A train of electrical pulses (50 Hz, lasting 1 s, applied every 2 s, from t = 24 s in the time series) induced translocation of GFP-PKCγ to the plasma membrane (see also Video 1 and Fig. S1, available at http://www.jcb.org/cgi/content/full/jcb.200503118/DC1). (C) The glutamate-induced change in relative fluorescence intensity, corresponding to the relative GFP-PKCγ concentration at the soma plasma membrane, is shown as a function of time for three cells from separate experiments. (D) The same glutamate stimulation (30 μM) that was used to evoke GFP-PKCγ translocation in the experiment shown in C was applied to different, nontransfected cells loaded with the Ca²⁺ indicator Fluo-3 AM. This stimulation generates Ca²⁺ signals that have similar kinetics to GFP-PKCγ translocation (three representative traces from one experiment are shown; n = 10 experiments).

Quantitative analysis of the stimulus-induced relative increase in PKCγ concentration in thin branches. Elevated PKCγ concentration in these branches persists even after PKCγ dissociates from the membrane of main branches and soma. (A) Images from two time series show that glutamate triggers a delayed translocation of PKCγ from thicker into thinner dendritic processes. The thick and thin arrows in the top panel mark thick and thin dendritic processes, respectively. The arrows on the bottom panel indicate thinner branches into which PKCγ translocates after the stimulus. The final image shows retention of PKCγ in many of the thin branches emanating from the thick branches and soma. (B) Comparison of the relative time course of GFP-PKC translocation measured at three locations along a thin branch versus the time course measured in the cytosol and at the plasma membrane of the cell soma. Each region of interest was 2 μm wide, and regions were positioned at 2-μm intervals along the thin branch. (right) The diagram demonstrates the position of each region placed in the soma and along the thin branch. The color of the region and the symbols drawn inside the regions of interest mirror those in the traces in the graph on the left. (C) Ratio imaging comparing GFP-PKCγ in branches to an RFP cytosolic marker before (top) and after (bottom) stimulation. The original GFP and RFP images (left), as well as the pseudocolored ratio images, are shown (right). (D) Series of topographical images showing the overlaid color-coded ratio of PKC/RFP generated using a perimeter mask to highlight changes occurring in thin branches and processes at the edge of the cell. Areas that appear red indicate an enrichment of PKCγ relative to the cytosolic marker, whereas green depicts regions of low PKCγ concentration.

Simulation and modeling of the translocation process into thin branches reveals a geometric attraction process that can suppress subthreshold stimuli, integrate activity, and create molecular memory. (A) Simulation of the PKCγ translocation in a dendritic arbor during transient stimulation. The schematic images were drawn by an analysis program based on data generated from simulating the diffusion of PKCγ in the dendritic arbor after stimulation. In these schematic images, regions of elevated PKCγ concentration are shown in red. After the stimulus has been applied, a rapid translocation of PKCγ to the plasma membrane of the thick branch is observed. These simulations suggest that delayed translocation of PKCγ into thin branches and the prolonged localization in branches after the response to the stimulus has ceased can be explained by surface-to-volume arguments and the measured diffusion coefficient of PKCγ. (B) Simulated concentration profiles of PKCγ at different points along a thin dendritic branch as a function of time (for a 120-s period after the application of the stimulus). The points were chosen according to the experimental results in Fig. 4 B. (C) Schematic illustration of the parameters used for the construction of the thin branch attraction simulation. (D) Measuring the relative SVR (surface-to-volume parameter) in the dendritic arbor. The membrane marker GFP-Lyn, consisting of the myristoylation/palmitoylation motif from Lyn kinase, was coexpressed with a cytosolic RFP. The original images, in addition to the pseudocolored ratio image, are shown.

A model of the spine translocation process shows that spine necks provide a means to suppress subthreshold stimuli, integrate activity, and create local memory of past activity. (A) Diagram illustrating the parameters used to simulate translocation of PKCγ into dendritic spines. (B) Simulation of PKCγ translocation curves for two spines with different morphologies. Translocation was compared between a spine that only has a small neck or no neck at all and one that has a restricted neck. The spine with the small neck (*) is a fast-loading synapse and filled rapidly with PKC, unlike the spine with the restriction (#), which loaded much more slowly. (C) Schematic images were drawn by a MatLab program based on simulations of PKCγ translocation from a main branch into three different synaptic spines. One spine had no neck constriction, and two spines had neck constrictions because of long or narrow necks (see Materials and methods for model descriptions). Translocation into the spine without a neck was rapid and reversible, whereas the spines with restricted necks underwent a slow PKCγ recruitment behavior, and the protein remained in the spine for a much longer time than the spine with no neck. Red indicates a high concentration of PKCγ at the plasma membrane or in spines, whereas green shows a low concentration of PKCγ.