Image acquisition and analysis

Fixed-cell images were acquired using a Zeiss LSM510 confocal microscope. Z-stacks of 6–12 images were taken at 2048 × 2048 resolution, optical slice depth of 1 μm per image, and z-step of 0.37 μm. At the time of acquisition, laser power was adjusted so that all spines were below the threshold of saturation. Analysis was by ImageJ software (NIH). Maximum intensity projections were processed to smooth contours and a binary mask was obtained after edge detection. The cross-sectional area, and number of spines was calculated. For each condition, 90–100 μm sections of secondary dendrite from each neuron were analysed, 4–6 neurons per experiment, from three separate experiments, resulting in 700–1400 spines per condition. Experiments were both imaged and analysed with the experimenter blind to the experimental conditions. Statistical analyses were performed in Excel (Microsoft). The Student's t-test was performed on spine density data. Cumulative plots were analysed using Kolmogorov–Smirnov test (K–S test).

Time lapse live confocal images of dendritic spines were acquired using a Nikon Eclipse Ti-E microscope. Z-stacks of 15–20 images were taken at various time points at 512 × 512 resolution with a z-step size of 0.4 μm. Neurons were continually perfused at 35°C with HBS at a flow rate of 4 ml/min. For chem-LTD, buffer was switched to HBS containing 20 μM NMDA and 20 μM glycine for 3 min, followed by return to normal HBS. For PMA treatment, perfusion buffer was HBS containing 50 μM D-AP5, 5 μM NBQX and 0.5 μM PMA. Baseline perfusion buffer was the same minus PMA. Glutamate receptor antagonists were present to prevent any PKC-stimulated release of endogenous glutamate from activating postsynaptic glutamate receptors (Calabrese and Halpain, 2005). Analysis was by ImageJ. Maximum intensity projections were produced for each time point and the images were processed and analysed as above.