(a) Representative voltage traces recorded simultaneously at the soma and a second-order dendrite (see schematic), first during an STDP protocol pairing rCA3 EPSPs with a single AP evoked by brief somatic current injection (2 ms; 4 nA; left), followed by pairing with three evoked APs delivered at a frequency of 200 Hz (right). A 1 AP STDP protocol failed to potentiate EPSPs, whereas 3 AP STDP resulted in LTP (top traces). Note that the 1 AP STDP protocol was generally insufficient to generate a supralinear dendritic response, whereas the 3 AP STDP protocol caused dendritic spikes visible in the dendritic recording and to a lesser extent, because of cable filtering, in the somatic recording. Inset: Area under the evoked responses was plotted to distinguish between linear and supralinear events. (b) Supralinear responses with the 3 AP STDP protocol were prevented following NMDAR blockade (D-AP5, 1.4±3.2%, n=5, P<0.0001). Inset: supralinear responses are prevented when NMDARs are blocked. (c) Faster EPSP rise times in dendrite versus soma (n=5, P=0.008, paired t-test) indicates that for these experiments a majority of the synapses activated by stimulation of rCA3 collaterals were located at or near the dendritic branch recorded from. The green line denotes the pooled average of all cells. (d) The NMDAR contribution generated by the different STDP protocols was estimated by calculating the ratio between the areas under the voltage traces in the absence and presence of D-AP5 (n=5, P=0.0001, paired t-test). (e) When the STDP protocol did not adequately activate NMDARs to generate a supralinear response, LTP was not induced (n=5, P=0.7; paired t-test). (f) A hyperpolarizing pulse applied during the 3 AP STDP protocol decreased the probability of evoking an NMDA spike and prevented LTP induction. As a result, a supralinear response was not generated, either in the soma (blue trace) or the dendrite (brown trace). Furthermore, the rCA3 EPSP was not potentiated (black traces: rCA3 EPSP recorded in the soma). When the same protocol was repeated, but without the hyperpolarizing pulse, supralinear responses were generated and LTP was induced. (g) Pooled data showing faster rise times for the rCA3 EPSP in the dendritic as compared with the somatic recording (n=5, P=0.0008; paired t-test), indicating that primarily dendritic inputs were stimulated. (h) Quantification of the contribution mediated by NMDARs to the recorded responses shows that the STDP protocol that included a hyperpolarizing pulse (3AP STDPhyper) inhibited the generation of a supralinear signal versus without the hyperpolarization (3AP STDP, n=5, P=0.0001, paired t-test). (i) In addition, the STDP protocol that included a hyperpolarizing pulse greatly reduced the magnitude of LTP at the rCA3 synapse (n=5, P=0.0002, paired t-test).