In some experiments, the pipette solution included ∼0.1%–0.3% biocytin (Sigma). Alexa Fluor 488 or 594 fluorescence or biocytin labeling with immunoperoxidase reaction was used for post hoc verification of the localization of neurons along the proximodistal axis of CA3. Series resistance was <30 MΩ. CA3 neurons had resting membrane potentials between −60 and −76 mV (average: −68.0 ± 0.2 mV, n = 381). CA3PCs were usually kept at −68

to −72 mV, unless otherwise indicated. GSK1349572 molecular weight CA1 PCs were held at ∼−65 mV unless otherwise indicated. A dual galvanometer-based two-photon scanning system (Prairie Technologies) was used to image Alexa Fluor 488-loaded neurons and to uncage glutamate at individual dendritic spines (Losonczy and Magee, 2006, Losonczy et al., 2008 and Makara et al., 2009). Two ultrafast pulsed laser beams (Chameleon Ultra II; Coherent) were used, one at 920 nm for imaging Alexa Fluor 488 and the Fasudil price other at 720 nm to photolyze MNI-caged-L-glutamate (Tocris; 10 mM) that was applied through a puffer pipette with an ∼20- to 30-μm-diameter, downward-tilted aperture above the slice using a pneumatic ejection system (Picospritzer III [Parker Hannifin] or PDES-02TX [NPI]). Laser beam intensity was independently controlled with electro-optical modulators (Model

350-50, Conoptics). Local dendritic spikes were evoked by uncaging of MNI-glutamate at a spatially clustered set of visually identified spines (see also Supplemental Experimental

Procedures) with the highest synchrony our system allows, using 0.2 ms uncaging duration with 0.1 ms intervals between synapses (termed synchronous stimulation), unless otherwise indicated. For clarity, throughout the paper the term “Na+ spike” refers to local Na+ channel-mediated dendritic spikes, whereas the axosomatically generated spike is termed action potential or AP. Dendritic Na+ spikes were usually evoked using 5–20 synapses. NMDAR-mediated nonlinearity was measured in dendritic segments 61–193 μm (basal dendrites) or 152–315 μm (apical dendrites) from the soma. NMDA spikes were generated using 20–40 synapses activated synchronously (0.2 ms out uncaging duration and 0.1 ms intervals between synapses), except in some experiments (Figures 4G and 4H) in which longer intervals (1–5 ms) were used. To avoid variability in kinetic parameters of NMDA spikes due to distance-dependent distortion of voltage responses, we performed pharmacological experiments, spatiotemporal distribution experiments, and paired-pulse experiments exclusively on basal dendrites. For determining input-output function, the expected amplitude was calculated as the arithmetic sum of the physiologically sized unitary gluEPSPs at the given temporal input pattern. Unitary gluEPSPs were measured repeatedly (usually two to five times) interleaved with synchronous stimulations, using 205–420 ms intervals between the individual synapses (see also Supplemental Experimental Procedures).