Somatic Depolarization Enhances Hippocampal CA1 Dendritic Spike Propagation and Distal Input-Driven Synaptic Plasticity

Dendritic spike propagation efficacy is modulated by the somatic membrane potential

To explore the regulation of dendritic spike propagation in hippocampal CA1 PNs, we obtained dual whole-cell current clamp recordings from the soma and distal apical dendrite of these cells, ∼250–300 µm from the soma (at the border of SR and SLM), in acute hippocampal slices from adult mice (Fig. 1A). We triggered a distal dendritic spike by injecting a 350-pA, 100-ms depolarizing current step through the dendritic patch pipette and recorded the resultant voltage change at the soma (Fig. 1B). In the dendrite, the resultant spike had a fast rising component, typically associated with voltage-gated Na⁺ channels (Kim and Connors, 1993; Golding and Spruston, 1998), followed by a slower plateau potential, typically generated by voltage-gated Ca²⁺ channels (Amitai et al., 1993; Kim and Connors, 1993; Schiller et al., 1997; Golding et al., 1999). When the somatic membrane potential was held initially at −70 mV, near the typical CA1 resting potential, the dendritic spike (32.4 ± 5.3 mV in amplitude) produced only a very small somatic depolarization (0.65 ± 0.08 mV, n = 5; Fig. 1B–D), for a dendritic attenuation factor (the amplitude of the peak somatic depolarization during the dendritic current step, divided by the peak amplitude of the dendritic spike) of 49.8 ± 5.7, consistent with previous findings (Jarsky et al., 2005; Sun et al., 2014) .

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Somatic resting membrane potential determines the efficacy of dendritic spike propagation to the soma. A, Schematic of the experimental setup. B, Dendritic (top) and somatic (bottom) recording of a dendritic spike, elicited by current injection into the dendrite at varying somatic membrane potentials before (black) and after adding 500 nm TTX (red) followed by 200 μm Cd (blue) to the bath solution; insets show an expanded view of the traces within the dashed box. C, Peak depolarization of the somatic membrane potential in response to a dendritic spike at varying somatic baseline potentials before (black) and after adding TTX (red) and Cd (blue) to the bath. D, Comparison of peak somatic membrane depolarization at −70- and −55-mV baseline somatic membrane potential before (black) and after adding TTX (red) and Cd (blue) to the bath, all data are paired as all conditions were recorded consecutively in each of the patch-clamped cells. E, Attenuation factor of the dendritic calcium spike, as it propagates to the soma, plotted against the somatic baseline potential before (black) and after adding TTX (red) and cadmium (blue) to the bath. *p < 0.05, **p < 0.01, ***p < 0.001; multiple t-test after 2way ANOVA.

However, depolarization of the somatic resting potential by constant current injection dramatically enhanced the efficacy of dendritic spike propagation by nearly an order of magnitude. Thus, with the soma held initially at −55 mV, a distal dendritic spike produced a 6.46 ± 0.90 mV somatic depolarization, decreasing the attenuation factor to 7.79 ± 1.19 (p < 0.005, n = 5; Fig. 1B–D). Upon further depolarization of the soma, a dendritic spike typically elicited a burst of somatic APs (Fig. 1B). Changes in somatic membrane potential had no impact on dendritic spike threshold, amplitude or duration at the distal dendritic recording site (Table 1). Somatic depolarization also caused no significant change in the passive properties of the cell in either compartment (Table 1). The signal attenuation factor between the distal dendritic recording site and the soma showed an S-shaped dependence on membrane potential, decreasing continuously to an asymptotic value as the somatic membrane potential was depolarized (p < 0.0001 with Friedman test, n = 5; Fig. 1E). We term this phenomenon depolarization enhancement of dendritic spike propagation (DEDSP).

To determine the ionic nature of the dendritic spikes and the mechanism by which somatic depolarization enhanced their propagation, we examined the effects of antagonists of voltage-gated Na⁺ and Ca²⁺ channels. Inhibition of voltage-gated Na⁺ channels with TTX (500 nm) blocked the fast phase of the dendritic spike but left the slower plateau potential unchanged (indicative of Ca²⁺ spikes; Amitai et al., 1993; Kim and Connors, 1993; Schiller et al., 1997; Golding and Spruston, 1998; Golding et al., 1999). Somatic depolarization enhanced the propagation to the soma of these Ca²⁺ spikes in a manner similar to that seen under control conditions. In TTX, the somatic amplitude of the dendritic spike was 0.66 ± 0.13 mV at a somatic resting potential of −70 mV compared with 4.36 ± 0.69 mV at a resting potential of −55 mV (p < 0.05, n = 5; Fig. 1B–D), indicating that DEDSP does not require voltage-gated Na⁺ channels. Combined application of the voltage-gated Ca²⁺ channel blocker Cd²⁺ (200 μm) and TTX fully eliminated the dendritic spike, even in the distal dendrite (0.69 ± 0.11 mV at −70 mV vs 1.51 ± 0.31 mV at −55 mV, p = 0.07; Fig. 1B–D), confirming that the plateau potentials were Ca²⁺ spikes. Importantly, with dendritic spikes blocked, somatic depolarization had only a minimal effect on propagation of the passive dendritic depolarization.

Modulation of dendritic spike propagation efficacy during somatic membrane depolarization is dependent on A-type K⁺ channels

Previous studies have implicated voltage-gated A-type Kv4.2/4.3 K⁺ channels in regulating dendritic excitability. These channels are present in the apical dendrites of CA1 PNs where they regulate the back-propagation of somatic APs (Hoffman et al., 1997; Gasparini et al., 2007), curtail the duration of dendritic plateau potentials, and confine their spread throughout different dendritic branches (Cai et al., 2004). Moreover, these channels rapidly inactivate over a voltage range that matches the effect of somatic depolarization we observed (Hoffman et al., 1997; Ramakers and Storm, 2002; Kim et al., 2007), providing a potential mechanism for how depolarization may enhance dendritic spike propagation.

To examine directly the role of A-type K⁺ channels in DEDSP, we compared the effects of membrane depolarization on dendritic spike propagation before and after bath application of the A-type K⁺ channel blocker, 4-AP (5 mm). Application of 4-AP greatly enhanced spike propagation at a somatic holding potential of −70 mV and largely occluded the enhancement in spike propagation with membrane depolarization (Fig. 2A–D). Although some residual effect of depolarization on propagation efficacy remained in the presence of 4-AP (p < 0.005 with Friedman test; Fig. 2B,D), nonlinear fits to the data showed that 4-AP significantly diminished the effect of voltage (p < 0.01 with extra-sum-of-squares F test; Fig. 2B,D). In the presence of 4-AP, the slope of a linear regression of the dendritic attenuation curve (Fig. 2D) was reduced by 93.8%, from −1.94 to −0.12 (p < 0.0001). Furthermore, while somatic depolarization from −70 to −55 mV caused a large and highly significant increase in the amplitude of the dendritic spike at the soma (0.97 ± 0.2 mV compared with 4.94 ± 0.78 mV; p < 0.005, n = 5; Fig. 2C), depolarization caused only small, nonsignificant increase in the somatic response in the presence of 4-AP (4.42 ± 0.43 vs 5.79 ± 0.55 mV; p = 0.063, n = 5; Fig. 2C). Finally, 4-AP produced a similar enhancement of dendritic Ca²⁺ spike propagation in the presence of TTX, confirming that voltage-gated Na⁺ channels are not required (Fig. 2B–D).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

The voltage dependence of the propagation efficacy of dendritic spikes is mediated by 4-AP-sensitive and phrixotoxin-2-sensitive K⁺ channels. A, Dendritic (top) and somatic (bottom) recording of a dendritic spike, elicited by current injection into the dendrite at varying somatic membrane potentials before (black) and after adding 5 mm 4-AP (green) and 500 nm TTX (orange) to the bath solution, insets show a zoom of the traces of somatic membrane potential within the squares. B, Peak depolarization of the somatic membrane potential during the dendritic spike at varying somatic baseline potentials before (black) and after adding 4-AP (green) and TTX (orange) to the bath. C, Comparison of peak somatic membrane depolarization at −70- and −55-mV baseline somatic membrane potential before (black) and after adding 4-AP (green) and TTX (orange) to the bath; all data are paired as all conditions were recorded consecutively in each of the patch-clamped cells. D, Attenuation factor of the dendritic calcium spike, as it propagates to the soma, plotted against the somatic baseline potential before (black) and after adding 4-AP (green) and TTX (orange) to the bath. E–G, Same as B–D but comparing data before (black) and after (red) application of 500 nm phrixotoxin-2. H, Dendritic (top) and somatic (bottom) recording of a dendritic spike, elicited by current injection into the dendrite at varying somatic membrane potentials before (black) and after adding 500 nm phrixotoxin-2 (red). I, Voltage traces in 3 example cells, patch-clamped at the soma and recorded from at different dendritic locations (black, 60-µm distance from soma; red, 110-µm distance; blue, 200-µm distance) in response to a 300-ms somatic current step (100–400 ms), adjusted in each cell to provide ∼15-mV somatic depolarization. J, Depolarization (black) with estimated A-type K⁺ channel availability according to Hoffman et al. (1997); (steady-state inactivation shown in red) at various dendritic locations when the soma is depolarized by 15 mV (closed circles) or when the soma is not depolarized (open circles). **p < 0.01, ***p < 0.001; multiple t-test after 2way ANOVA.

As 4-AP is also known to block Kv1 delayed-rectifier channels (Castle et al., 1994), we examined the effect of the specific Kv4.2/3 blocker phrixotoxin-2 (Diochot et al., 1999; Chagot et al., 2004). Under control conditions, the somatic depolarization in response to a dendritic spike increased significantly from 1.26 ± 0.36 mV at a somatic membrane potential of −70 to 4.56 ± 0.26 mV when the somatic membrane was depolarized to −55 mV (p < 0.005, n = 5; Fig. 2E–H). Similar to the action of 4-AP, 500 nm phrixotoxin-2 caused a significant increase in the amplitude of the somatic response to a dendritic spike at a somatic resting potential of −70 mV (4.28 ± 0.37 mV; p < 0.005 vs control in absence of phrixotoxin-2, n = 5). Moreover, the enhancement in dendritic spike propagation normally seen on depolarization of the membrane to −55 mV was largely abolished by the toxin (5.12 ± 0.41 mV at – 55 mV, p = 0.094, n = 5; Fig. 2F), with the slope of the best linear fit to the somatic response–voltage relation reduced by 91.6%, from 1.67 to 0.14 (p < 0.0001 with linear regression test; Fig. 2G).

Since neither 4-AP nor phrixotoxin-2 altered dendritic passive membrane properties, dendritic spike threshold, amplitude or duration (Table 2), we surmised that somatic depolarization might enhance dendritic spike propagation by enhancing the resting inactivation of A-type K⁺ channels. As a first test of this idea, we measured the effects of somatic depolarization on the proximal dendritic membrane voltage of CA1 PNs to determine whether the extent of depolarization was likely sufficient to alter channel inactivation. We performed multiple dual patch clamp recordings from the soma and from the apical dendrite at various distances from the soma. In each cell, we injected 300-ms current steps into the soma to depolarize the somatic membrane voltage by ∼15 mV while recording from the apical dendrites. Consistent with previous observations on spatial attenuation of voltage changes (Bar-Yehuda and Korngreen, 2008), we saw a strong depolarizing effect of a 15-mV somatic depolarization on the proximal dendrites, but little effect in the distal dendrites (Fig. 2I,J). Using the inactivation vs voltage plot for A-type K⁺ channels from Hoffman et al. (1997) and correcting for liquid junction potential (−6.2 mV), we found that the measured effect of somatic depolarization should markedly increase steady-state inactivation of the A-type K⁺ channels in the proximal dendrites within 100–150 µm of the soma (Fig. 2J). Thus, in the proximal dendrite (60 µm from the soma), resting inactivation of A-type K⁺ channels was predicted to increase from 38% when the soma was at −70 mV to 78% when the soma was depolarized to −55 mV. In contrast, in the distal dendrite (>200 µm from the soma), we expect only a small increase in resting inactivation, from 45% to 50% (Fig. 2J). From these data, we conclude that somatic depolarization should result in a marked increase in resting inactivation of A-type K⁺ channels in the proximal apical dendrite that allows for more efficient propagation of dendritic spikes during somatic depolarization, while distal dendritic locations (the site of dendritic spike initiation) should be little affected by alterations in somatic voltage.

A computational model shows the impact of A-type K⁺ channels on propagating dendritic spikes in the apical trunk

To further explore the dynamics of the interaction between A-type K⁺ channels and the propagation of the dendritic calcium spike, we created a three compartment NEURON model of a simplified pyramidal cell, including a soma, apical dendritic trunk, and apical dendritic tuft compartments (Fig. 3A; see Materials and Methods). Passive parameters were adjusted to reflect our experimental data on input resistance and time constant in all three compartments. Active conductances were distributed according to previous physiological and histochemical observations (see Materials and Methods) and consisted of R-type and T-type voltage-gated Ca²⁺ channels, A-type K⁺ channels, HCN channels, delayed rectifier K⁺ channels, medium afterhyperpolarization Ca-dependent K⁺ channels, and an exponentially decaying Ca²⁺ extrusion mechanism. Voltage gated Na²⁺ channels were not introduced into the model to minimize the number of independently adjustable parameters as DEDSP was observed in the presence of TTX (Figs. 1, 2). This reduced model was able to reproduce the DEDSP phenomenon. Depolarization of the somatic resting membrane enhanced the amplitude of the dendritic spike at the soma, from 1.9 mV when the soma was held at −72.3 to 8.4 mV when the soma was depolarized to −52.5 mV, a 4.5-fold increase (Fig. 3B,E). Consistent with the experimental data, this change in the somatic depolarization was dependent on A-type K⁺ channels. Reducing the A-type K⁺ channel conductance to 0 greatly enhanced the somatic amplitude of the dendritic spike at the negative resting potential (7.4 mV when the soma was held at −72.3 mV) and markedly reduced the effect of somatic depolarization to enhance spike amplitude to a 1.3-fold increase (Fig. 3C,E). The model also reproduced our finding that DEDSP is dependent on actively propagating signals as a subthreshold dendritic depolarization is only minimally responsive to prior manipulation of the somatic membrane potential (Fig. 3D,E).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Computational model of a three compartment CA1 PN, reproducing key features of DEDSP. A, Schematic of the three-compartment model. B, Voltage traces recorded from the tuft (top), trunk (at position 0.33 where the soma is 0 and the tuft is 1, middle), and soma (bottom) in response to a 500-pA current step injected into the tuft compartment (top gray current traces) while the somatic membrane potential was modified by long-lasting current steps into the soma, ranging from −100 to +200 pA (bottom gray current traces). C, Same as B except that the A-type K⁺ conductance was set to 0 (simulating block of A-type K⁺ channels). D, Same as B, except that the dendritic current injection was reduced to 250 pA to generate a subthreshold event in the tuft compartment. E, Peak depolarization of the somatic membrane in response to the dendritic spike at varying somatic membrane potentials before the dendritic spike onset (measured at 700 ms during the simulations shown in B, C) with a realistic level of A-type K⁺ channel conductance (black), with the A-type K⁺ conductance set to 0 (green), and in response to a subthreshold dendritic current injection with A-type K⁺ channels present (blue). F, The percentage of inactivated A-type K⁺ channels in the soma (black), trunk (at position 0.33, dark gray) and tuft (light gray) at varying somatic membrane potentials (measured at 700 ms during the simulations shown in B, C). Lines depict linear regression fits.

The modeling results also support the view that somatic depolarization enhanced dendritic spike propagation by increasing A-type K⁺ channel inactivation. Increasing amounts of somatic depolarization that simulate our experimental values produced a progressive increase in channel inactivation that was much more pronounced in the soma and proximal dendrites than in the distal dendrites (Fig. 3F). In the soma, the fraction of A-type K⁺ channels that were inactivated at a resting potential of −72 mV increased markedly when the soma was depolarized to −53 mV, from 0.14 to 0.60. The effect of somatic depolarization also enhanced A-type channel inactivation in the proximal portion of the apical dendrite (1/3 of the distance from the soma to tuft), from 0.24 to 0.33. These changes in inactivation in the soma and proximal dendrite were sufficient to have a large impact on spike regeneration along the trunk and thus the resulting depolarization seen in the soma.

The importance of A-type K⁺ channel inactivation in enabling dendritic spike propagation was further supported by examining the effect of reducing inactivation by shifting the mid-point voltage of channel inactivation by 10 mV, from its normal value of −56 to −46 mV. This resulted in a nearly complete blockade of dendritic spike propagation, with little effect on spike generation at its point of origin in the tuft. Thus, the voltage shift in inactivation reduced the peak amplitude of the dendritic spike in the proximal apical dendrite from 14.5 to 5.0 mV with the soma held at −72.3 mV, and from 44.9 to 5.9 mV with the soma held at −52.5 mV. In contrast, the shift in inactivation only reduced spike amplitude in the tuft by 1.1 mV, from 89.9 to 88.8 mV (while changes of the somatic holding potential had a negligible effect of <0.05 mV in either case). It should be noted, that this model is meant to provide a proof-of-principle and to determine the likelihood that A-type K⁺ channel inactivation contributes to the DEDSP mechanism. It is likely that other ion channels that were not included in the model may also contribute to this phenomenon.

Somatic depolarization enhances propagation of synaptically evoked dendritic spikes but not subthreshold PSPs

Next, we investigated whether somatic depolarization can also enhance the propagation of dendritic spikes generated in a more physiological manner through synaptic stimulation of the perforant path (PP) inputs to the distal CA1 PN dendrites. We placed an extracellular stimulating electrode in the SLM of CA1 (at least 200 μm from the dendritic patch location), the distal dendritic site of the PP inputs, and applied either a single stimulating pulse to elicit a subthreshold postsynaptic potential (PSP) or a strong, brief high-frequency burst of stimuli to elicit a dendritic spike (four or five stimuli at 200 Hz; Fig. 4A).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

A-type K⁺ channels affect active and passive signals differently. A, Example traces of dendritic and somatic responses to single PP stimulation, which evokes a passively propagated PSP (left), and a train of five pulses of PP stimulation at 200 Hz, which evokes an actively propagating dendritic spike (right), when the soma is held at either −70 or −57.5 mV before (black) and after (green) application of 5 mm 4-AP. B, Peak depolarization of the somatic membrane potential during a passively propagating PSP, evoked by extracellular stimulation (top) and a Δ-EPSP waveform (bottom), at varying somatic baseline potentials before (black) and after adding 4-AP (green). C, The attenuation factor of passively propagating dendritic PSPs evoked by extracellular stimulation if the soma is held at −70 mV (left) or at −57.5 mV (right) before (black) or after (green) application of 4-AP. Data are paired as all conditions recorded consecutively in each of the patch-clamped cells. D, Amplitude of passively propagating PSPs, evoked by single PP stimulation at the dendrite (top) and the soma (bottom) before (black) and after (green) bath application of 5 mm 4-AP. E, Half width of passively propagating PSPs, evoked by single PP stimulation at the dendrite (top) and the soma (bottom) before (black) and after (green) bath application of 4-AP. F, Peak depolarization of the somatic membrane potential during an actively propagating dendritic spike at varying somatic baseline potentials before (black) and after adding 4-AP (green). G, Same as C but for actively propagating signals. H, Amplitude of actively propagating dendritic spikes, evoked by high-frequency PP stimulation at the dendrite (top) and the soma (bottom) before (black) and after (green) bath application of 4-AP. I, Half Width of actively propagating dendritic spikes, evoked by high-frequency PP stimulation at the dendrite (top) and the soma (bottom) before (black) and after (green) bath application of 4-AP. *p < 0.05; Wilcoxon test.

Propagation of the subthreshold PSP evoked by a single PP stimulus or a subthreshold PSP-like signal evoked by dendritic current injection (Δ-EPSP) showed little dependence on somatic depolarization. Thus, the PSP amplitude at the soma was attenuated by a factor of 9.49 ± 1.67 at −70 mV and 8.96 ± 1.45 at −57.5 mV (p = 0.94, n = 7; Fig. 4A–C). In contrast, the propagation of a dendritic spike evoked by a brief train of synaptic stimuli (five stimuli at 200 Hz) was greatly enhanced by somatic depolarization in a manner similar to that observed when we evoked a dendritic spike through direct current injection (Fig. 1). Thus, the synaptically evoked spike amplitude attenuation at the soma was 33.4 ± 4.0 at −70 mV versus 9.8 ± 1.2 at −57.5 mV, p = 0.015, n = 7 (Fig. 4A,F,G). Moreover, with sufficient somatic depolarization, the synaptically evoked dendritic spike also led to burst firing of somatic APs.

The application of 4-AP (5 mm) caused a small but significant increase in the subthreshold PSP amplitude in both the dendrite and the soma (in both cases p = 0.03, n = 7; Fig. 4A,D). This change in amplitude may be because of presynaptic or postsynaptic effects or a combination of both (Migliore et al., 1999; Migliore, 2003; Palani et al., 2010). However 4-AP did not alter dendritic propagation of the subthreshold PSP, at either a somatic voltage of −70 or −57.5 mV (Fig. 4B,D; dendritic attenuation was 8.71 ± 1.34 at −70 mV and 8.44 ± 1.34 at −57.5 mV, p = 0.57, n = 7; Fig. 4A–C). In contrast, the propagation of the synaptically evoked dendritic spike was greatly enhanced by bath application of 4-AP at more negative but not depolarized somatic membrane potentials. In 4-AP, spike attenuation at −70 mV (10.53 ± 1.05, n = 7) was significantly less than that in control conditions (33.4 ± 4.0; p = 0.0156; Fig. 4F,G). In contrast, at −57.5 mV attenuation in 4-AP (9.02 ± 1.10) was similar to that in the absence of the drug (9.78 ± 1.19, p = 0.109, n = 7; Fig. 4F,G). 4-AP had no significant impact on the spike amplitude in the distal dendrite (p = 0.08, n = 7; Fig. 4H) nor did it alter PSP or dendritic spike duration (Fig. 4E,I), indicating that A-type K⁺ channels are more important in regulating spike propagation along the apical dendrites in SR than spike generation in distal dendrites in SLM. This is consistent with previous pharmacological results (Cai et al., 2004) and the pattern of expression of Kv4.2 (Maletic-Savatic et al., 1995).

Depolarization enhances dendritic spike firing and somatic bursting during paired activation of distal and proximal synaptic inputs

Paired activation of distal PP and proximal SC synaptic inputs can synergistically interact to generate firing of AP bursts and to induce Hebbian (Takahashi and Magee, 2009) and non-Hebbian forms of synaptic plasticity, including ITDP (Dudman et al., 2007; Basu et al., 2013, 2016). ITDP is most strongly induced when paired stimuli are delivered at 1 Hz for 90 s and when the distal input precedes the proximal input by 20 ms (−20 ms pairing interval,). During the standard 90-s induction protocol from a somatic potential of −70 mV, paired stimulation initially fails to elicit a dendritic spike. However, as the train of paired stimuli continues, the probability of spike firing increases, reflecting some unknown form of short-term intrinsic or synaptic plasticity. At a resting potential of −70 mV, the dendritic spikes fail to elicit somatic APs (Basu et al., 2016). This contrasts to the in vivo situation where dendritic plateau potentials trigger a burst of somatic AP output (Grienberger et al., 2014). As the CA1 PN somatic resting potential in vivo is depolarized by 10–15 mV relative to that in ex vivo slices, we examined the influence of somatic depolarization on dendritic spike firing and propagation in response to paired distal and proximal stimuli at the −20-ms ITDP pairing interval.

We obtained dual apical dendritic and somatic whole-cell patch clamp recordings and either blocked A-type K⁺ channels or manipulated the somatic membrane potential during the 90-s period of 1-Hz paired stimulation (Fig. 5A,B). To control for potential effects of the channel blockers or somatic depolarization on subthreshold PSP amplitude, we adjusted the strength of both the PP and SC stimulating currents to elicit approximately the same sized synaptic responses to the initial test stimuli and during the ITDP induction protocol (comparison of dendritic PP and SC PSPs from a holding potential of −70 mV in the presence and absence of 4-AP and from holding potential of −55 mV in absence of 4-AP using a Kruskal–Wallis test; PP PSPs, p = 0.62; SC PSPs, p = 0.09; n = 5).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

The impact of dendritic spikes at the soma during paired PP-SC stimulation. A, Schematic of the experimental setup, showing the somatic and dendritic patch pipettes (black) as well as the PP (orange) and SC (blue) stimulating electrodes. B, Example traces of the somatic (bottom) and dendritic (top) responses to ITDP induction stimuli for all three tested conditions [soma was held at −70 mV (left); 4-AP was present in the bath during induction (middle); soma was held at −55 mV during induction stimulation (right)]; black traces show examples of subthreshold PSPs, red traces show examples in which the SC stimulus elicited a dendritic spike. C, Average number of dendritic spikes in 10-s bins during the induction period. The 90 paired stimuli were separated into nine bins of 10 stimuli each and the number of dendritic spikes in each bin was counted [black, soma was held at −70 mV; green, 4-AP (5 mm) was present in the bath during induction; red, soma was held at −55 mV during induction stimulation]. Straight lines show linear regression fits. D, Dendritic spike count during the last 10 stimuli of the induction (black, soma was held at −70 mV; red, soma was held at −55 mV during induction stimulation; green, 4-AP was present in the bath during induction). E, Attenuation of the somatic depolarization when a dendritic spike was elicited by the paired PP-SC stimulation (black, soma was held at −70 mV; green, 4-AP was present in the bath during induction; red, soma was held at −55 mV during induction stimulation). F, Distribution of the number of somatic APs caused by dendritic spikes during paired PP-SC stimulation when 4-AP was present in the bath (green) or when the soma was depolarized to −55 mV during the induction period (red). *p < 0.05, **p < 0.01; Mann-Whitney test after Kruskal-Wallis test.

As described above, with the membrane at −70 mV a 1-Hz train of paired stimuli of distal and proximal inputs elicited dendritic spikes that propagated poorly to the soma (Fig. 4B; Basu et al., 2016). Application of 4-AP at a resting potential of −70 mV or depolarization of the membrane to −55 mV enhanced the propagation of dendritic spikes elicited by paired synaptic stimuli (Fig. 5E). In addition, either depolarization or 4-AP increased the probability that a pair of stimuli would elicit a dendritic spike at various times during the train (Fig. 5C). More specifically, the rate of increase of dendritic spike firing throughout the time course of ITDP induction, measured as the slope of the linear regression (0.36 when the soma was held at −70 mV, n = 5), was enhanced by both 4-AP (to 0.52, p = 0.038 compared with −70 mV with multiple comparison test, n = 5) and somatic depolarization to −55 mV (to 0.67; p = 0.021, compared with −70 mV with multiple comparison test, n = 5; Fig. 5C). With the membrane at −70 mV the number of dendritic spikes elicited by the last 10 paired stimuli of the train increased from 3.8 ± 0.4 in the absence of 4-AP to 6.4 ± 0.5 in the presence of 4-AP (p = 0.079). Depolarization to −55 mV in the absence of 4-AP caused an even larger increase in the number of dendritic spikes, to 8.2 ± 0.8 (p = 0.016 relative to −70 mV without 4-AP; Fig. 5D).

Of interest, when somatic APs were blocked by local application of TTX to the soma while the soma was depolarized to −55 mV, the number of dendritic spikes still increased significantly during the 90 s of paired stimulation (see below) and increased relative to dendritic spike firing at a somatic potential of −70 mV. Thus, the short-term plastic process that enhances dendritic spike firing during the train of paired stimuli does not require somatic AP firing but rather may be triggered by dendritic spiking alone. Moreover, as neither somatic depolarization nor 4-AP alters the local threshold in the dendrite for firing a dendritic spike in response to a single current injection (Tables 1, 2), the effect of 4-AP or somatic depolarization to enhance the probability of dendritic spike firing during the train of paired stimuli (relative to spike firing with the normal resting potential of −70 mV in the absence of 4-AP) likely reflects their action to enhance dendritic spike propagation.

Although the probability of dendritic spike firing increased progressively during the 90 s of paired stimulation with the resting membrane at −70 mV (Fig. 5C), the dendritic spikes were ineffective in eliciting a somatic AP. Of the 104 dendritic spikes elicited by synaptic pairing, only two triggered a somatic AP (1.9%, n = 5 cells). Moreover, in these two instances, the soma fired only one or two APs. Application of 4-AP or somatic depolarization greatly increased the probability that a dendritic spike would elicit one or more somatic APs. Thus, in the presence of 4-AP, 24 out of 158 dendritic spikes elicited a somatic AP (15.2%; p = 0.0004 with χ² test relative to control, n = 5 cells). Moreover, when a dendritic spike was successful in activating the soma, it now triggered a burst of 2 or more spikes, with an average of 2.67 ± 0.14 somatic APs per dendritic spike. Depolarization to −55 mV was even more effective in triggering a burst of somatic actions potentials, with 231 out of 268 dendritic spikes elicited by paired stimulation evoking one or more somatic APs (86.2%, p < 0.0001 with χ² test relative to control and 4-AP, n = 5 cells), with an average of 2.87 ± 0.08 APs per burst (p < 0.0001 with Mann–Whitney test relative to 4-AP; Fig. 5F).

At least two factors may contribute to the effect of somatic depolarization to enhance the firing of somatic APs in response to a dendritic spike elicited by paired stimulation. The first is the effect of somatic depolarization to enhance dendritic spike propagation. The second is the effect of somatic depolarization to bring the membrane closer to threshold. We believe that both factors are important as somatic spikes were never triggered by paired stimuli that failed to trigger a dendritic spike, even with the soma held at −55 mV. Additionally, 4-AP caused a significant increase in the amplitude of the somatic response to a dendritic spike (5.93 ± 0.53 to 9.39 ± 1.24 mV, p < 0.05 with Mann–Whitney test, n = 5; Fig. 5E), while not significantly changing somatic AP threshold (Table 2), indicating that the enhanced propagation of dendritic spikes itself leads to increased firing of somatic APs.

Manipulations that enhance dendritic spiking enhance the induction of ITDP

Our finding that somatic depolarization enhanced dendritic spike firing during paired distal and proximal synaptic stimulation suggests that somatic resting potential may also regulate the induction or expression of ITDP. When the membrane was depolarized to −55 mV, the ITDP protocol induced a final level of potentiation to 312.4 ± 9.5% (n = 5) of its initial level, similar to that seen with the membrane held at −70 mV (297.2 ± 19.6%, n = 7; p = 0.76 relative to –55 mV). However, membrane depolarization greatly accelerated the time course of onset of potentiation. With the soma held at −70 mV, the ITDP induction protocol resulted in a slowly developing enhancement in the PSP amplitude over a period of 20 min, as previously described (Dudman et al., 2007; Basu et al., 2013, 2016), with the PSP increasing to only 138.5 ± 8.1% (n = 7) of its initial value 2 min after the end of the ITDP induction protocol. However, with the membrane held at −55 mV, the PSP increased to 225 ± 11.2% (n = 5) of its initial value at this time (p = 0.0025 relative to −70 mV; Fig. 6A–E). Application of 4-AP (with the membrane at −70 mV) also sped the onset of ITDP, with the PSP increased to 184.6 ± 14.5% (p = 0.014 relative to −70 mV with no 4-AP, n = 5) after 2 min (Fig. 6E), with no enhancement in ITDP magnitude after 20 min (PSP increased to 308.5 ± 32.7%, p = 0.88, n = 5). Importantly, membrane depolarization did not alter the tuning of ITDP induction to the −20-ms pairing interval (p = 0.0019 with Kruskal–Wallis test, n = 5/time interval; Fig. 6F). This characteristic input timing dependence indicates that the plasticity mechanism enhanced by 4-AP or somatic depolarization is indeed related to ITDP, rather than some other form of long-term plasticity, such as LTP or spike-timing-dependent plasticity (STDP).

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Manipulating the somatic membrane potential affects onset of ITDP expression. A, Somatic PSP response to SC stimulation before and after ITDP induction, normalized to the response before induction (orange and blue symbols represent the time points that were used for averages in later panels). B, Example traces of PSPs in response to PP (left) and SC stimulation before (black) and 2 min (orange) and 20 min (blue) after ITDP induction. C, D, Same as A, B, respectively, except that in this case, the soma was depolarized to −55 mV during the induction stimuli. E, Comparison between the normalized response to SC stimulation 2 min (orange filled bars) and 20 min (blue filled bars) after ITDP induction at a membrane potential of −70 mV (black) or −55 mV (red) or if 4-AP was bath applied during ITDP induction (green). F, Normalized response to SC stimulation 20 min after ITDP induction at a membrane potential of −55 mV at varying PP-SC intervals during the induction. *p < 0.05, **p < 0.01; Mann-Whitney test after Kruskal-Wallis test.

If dendritic spikes are critical for the induction of ITDP, we predicted that it might be possible to induce ITDP with fewer than 90 paired stimuli when dendritic spiking is enhanced. We therefore examined the effect of varying the number of paired stimuli on the magnitude of ITDP with the soma held at −70 mV in the absence and presence of 4-AP or with the soma held at −55 mV (no 4-AP). At a holding potential of −70 mV (in the absence of 4-AP), 80 stimulus pairs were required to produce a significant potentiation of the PSP measured 20 min after the induction protocol (p = 0.011 with Mann–Whitney test after Kruskal–Wallis test), with 90 stimulus pairs required to double the size of the PSP (an increase of 197.2 ± 19.6%, n = 7; Fig. 7A,B). However, in the presence of 4-AP only 50 stimulus pairs were required to produce a significant potentiation (p = 0.039 with Mann–Whitney test after Kruskal–Wallis test), with 60 stimuli sufficient to cause a doubling of the PSP (PSP increase to 207.5 ± 27.3% of its initial level, n = 5; Fig. 7A,C). Fifty stimulus pairs were also sufficient to cause a significant increase in the PSP in the presence of the Kv4.2/3 channel blocker phrixotoxin-2 (the SC response increased to 200.4 ± 17.8% after 20 min, p = 0.048, n = 4; Fig. 7A). The effect of somatic membrane depolarization to −55 mV was even more pronounced, with a significant potentiation being produced after only 20 stimuli (p = 0.031 with Mann–Whitney test after Kruskal–Wallis test), and 40 stimuli sufficient to more than double the PSP (223.8 ± 22.2%, n = 5; Fig. 7A,D).

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

Somatic membrane depolarization or block of A-type K⁺ channels enhances the efficiency of ITDP induction. A, Change of the somatic PSP response to SC stimulation after ITDP induction, if the number of induction stimulus pairs is reduced to 10 (left), 50 (middle), or is kept at the usual 90 stimuli (right), colors indicate the condition of the cell during induction (black, soma was held at −70 mV; green, 4-AP was present in the bath during induction; red, soma was held at −55 mV during induction stimulation; blue, 500 nm phrixotoxin-2 was present in the bath during induction with 50 stimuli). B, Average change in the somatic response to SC stimulation 20 min after ITDP induction, if the number of induction stimulus pairs is incrementally decreased from 90 pairs to 10 in steps of 10 and soma was held at −70 mV during induction. C, Same as B but in the case that 4-AP was bath applied during ITDP induction. D, Same as B but in the case that the soma was held at −55 mV during ITDP induction. E, Correlation between number of dendritic spikes during a given ITDP induction protocol and the increase in the SC response after 15 min when the soma is depolarized to −55 mV. F, Same as E but when the soma is held at −70 mV and 4-AP is present in the bath. For E and F, each point is an individual cell. G, Correlation between the total number of dendritic spikes during ITDP induction and the SC response 2 min after ITDP induction (normalized to PSP before induction) within a stimulus range that yielded a total of 13–27 dendritic spikes when the cell soma was held at −70 mV (black), or at −55 mV under control conditions (red) or during somatic application of TTX (blue). *p < 0.05; Mann-Whitney test after Kruskal-Wallis test.

We observed a strong correlation between the number of dendritic spikes during the induction period and ITDP expression, especially over the range of 15–30 dendritic spikes. This was observed either when the soma was depolarized to −55 mV during ITDP induction (nonparametric Spearman correlation factor r = 0.66, p = 0.009 in this range, n = 15 cells; Fig. 7E) or when 4-AP was present in the bath (r = 0.78, p = 0.002, n = 14 cells; Fig. 7F). This correlation supports the view that dendritic spike firing during the paired synaptic stimulation plays an important role in the induction of ITDP. For technical reasons, most of the ITDP experiments were performed with only somatic recordings. As a result, we could not evaluate the correlation under control conditions, as dendritic spikes were only detectable in somatic recordings under conditions that favored spike propagation (Fig. 5B).

Given the correlation between the number of dendritic spikes and ITDP, the effect of somatic depolarization to enhance the induction of ITDP is likely to result, at least in part, from the increased firing of dendritic spikes. Might the enhancement in dendritic spike propagation to the soma also contribute to the enhancement in induction of ITDP? To address this question, we plotted the onset of ITDP with the soma at either −70 or −50 mV as a function of the number of dendritic spikes observed during the induction protocol (Fig. 7G). We found that the onset of ITDP elicited by the same number of dendritic spikes was significantly greater when the membrane was depolarized. A linear fit to the ITDP versus spike number relation yielded y intercept values of −0.20 ± 0.43 at −70 mV and 0.35 ± 0.50 at −55 mV [p = 0.0005 with linear regression analysis, n = 7 (−70 mV) and 5 (−55 mV); Fig. 7G]. This result suggests that depolarization can enhance ITDP onset independently of enhancing dendritic spike firing, consistent with a role for DEDSP enhanced dendritic spike propagation. This effect of depolarization was not visible in the SC response 20 min after ITDP induction (data not shown), which is consistent with our earlier finding that after 20 min, the increase in the SC response reaches a plateau level (Fig. 6).

As the dendritic spikes can trigger AP output with the membrane held at −55 mV but not with the membrane at −70 mV, we examined whether the enhanced induction of ITDP from the more positive somatic potential was a result of the enhanced firing of somatic spikes. We locally applied TTX to the soma during paired stimulation to block the firing of somatic APs without altering dendritic spikes or SC PSPs (Fig. 8A–C). Under these conditions, somatic depolarization was still able to enhance the induction of ITDP, which displayed a characteristic rapid onset (Fig. 8D) and ability to be induced with only 50 paired stimuli (Fig. 8E). Similarly, even in the presence of TTX, somatic depolarization enhanced the relation between ITDP at 2 min and the number of dendritic spikes during induction (y intercept 0.34 ± 0.15, p = 0.0058 relative to −70 mV with linear regression analysis, n = 4; Fig. 7G). These results clearly indicate that an increase in the efficacy of dendritic spike propagation and/or generation is sufficient to increase the efficacy of ITDP induction. However, as the rate of development of ITDP with the membrane at −55 mV is somewhat greater in the absence of TTX than in the presence of TTX, increased somatic spike firing is also likely to be important.

• Download figure
• Open in new tab
• Download powerpoint

Figure 8.

Depolarization of the somatic membrane during ITDP induction in the presence of TTX. A, Left, Example traces of PSPs in response to a single SC stimulus before (black), after 5 min (green), and after 10 min (red) of TTX, applied to the soma through a local puffing pipette. Middle, The average response to SC stimulation during local somatic TTX application remains unchanged (n = 6). Right, Same as the middle graph but showing the responses of the individual cells. B, Example traces of the somatic response to ITDP induction stimulation in those cases where a dendritic spike occurred (red) or did not (blue). The cell soma was depolarized to −55 mV, and TTX was applied to the soma (compare to Fig. 4B). C, Average occurrence of dendritic spikes during the induction period. The 90 paired stimuli were separated into nine bins of 10 stimuli each and the number of dendritic spikes in each bin was counted (black, soma was held at −70 mV; red, soma was held at −55 mV during induction stimulation; green, 4-AP was present in the bath during induction; blue, TTX was applied to the soma and the soma was held at −55 mV during induction stimulation). D, Change of the somatic PSP response to SC stimulation after 50 ITDP induction stimuli when TTX was applied to the soma and the soma was depolarized to −55 mV during induction (n = 4). E, Same as D but after 90 induction stimuli (n = 3).

Dendritic spikes are necessary and sufficient to induce ITDP in conjunction with SC stimulation

To probe more directly the importance of dendritic spikes, we determined whether they were sufficient and/or necessary to induce ITDP. We conducted dual patch clamp recordings at the soma and distal apical dendrite of CA1 PNs as before. However, instead of the normal ITDP induction protocol, we omitted the PP stimulus and administered 90 SC stimuli at a frequency of 1 Hz, either alone or paired with a dendritic spike elicited by direct dendritic current injection. The 90 SC stimuli alone caused no significant increase in the PSP in the following 20 min (113.5 ± 5.7%, p = 0.65, n = 8; Fig. 9A–D). Next, we examined the effect, in the same cell, of pairing each of the 90 SC stimuli with the injection of a depolarizing 2-ms, 400-pA current step into the dendrite 3–5 ms after the SC stimulus. We chose this delay as it was within the time window in which we observed the onset of the dendritic spike during paired PP-SC stimulation at the −20-ms interval (Fig. 5B, upper traces) and roughly coincided with the peak of the SC PSP measured in the dendrite. The pairing of SC synaptic stimulation and distal dendritic current injection evoked a dendritic spike with a high probability (89.6%, n = 450 stimuli in 5 cells). Importantly, the pairing also induced a large ITDP-like enhancement in the SC PSP measured in the distal dendrite, which progressively increased in amplitude within the following 20 min to 194.8 ± 18.0% of its value before pairing (from 6.23 ± 0.76 mV before pairing to 11.21 ± 0.84 mV after pairing, p = 0.008, n = 8; Fig. 9A–D).

• Download figure
• Open in new tab
• Download powerpoint

Figure 9.

Dendritic spikes are sufficient and likely necessary for ITDP induction. A–D, Pairing SC stimulation with a distal dendritic spike enhances the SC PSP, similar to ITDP. A, Example traces of a modified ITDP induction protocol, where a SC stimulus is either paired with dendritic current injection sufficient to elicit a dendritic spike (blue) or is not paired with a dendritic spike (red). Top, Dendritic patch-clamp recording. Bottom, Somatic recording. B, The SC PSP of the same cell before any stimulation (black), 20 min after 90 single SC stimuli were administered at 1 Hz (red), 20 min after 90 SC stimuli were combined with dendritic current injection to elicit dendritic spikes at 1 Hz (blue) and after subsequent bath application of 2 μm SR95531 and 1 μm CGP55845 (green). C, Time course of the change in the SC PSP over the course of the experiment; the bars indicate the time of the 1-Hz SC stimulation first without (red) and then with (blue) dendritic current injection. D, Summary of the SC PSP amplitude before and after the two stimulation modes described above; dots indicate single data points (n = 8). E–H, Pairing SC stimulation and dendritic spike suppresses IPSP through CB1 receptor activation, similar to ITDP. E, Average SC PSP amplitude 20 min after SC stimulation only, before (black) and after (green) application of SR95531 and CGP55845 (n = 6). GABA receptor antagonists increase the PSP because of blockade of IPSP. F, Average SC PSP amplitude 20 min after combined SC stimulation and dendritic current injection before (blue) and after (green) bath application of SR95531 and CGP55845 in a subset of cells (n = 4). GABA receptor antagonists fail to alter PSP, consistent with view that paired SC+dendritic spikes induces ITDP, resulting in suppression of IPSP. G, Change in SC PSP response after inducing ITDP with 90 SC stimuli at 1 Hz, each one combined with dendritic current injection to elicit dendritic spikes in the presence of 2 μm SR95531 and 1 μm CGP55845; green bar indicates the time at which SR95531 and CGP55845 are present in the bath (n = 4). GABA receptor antagonists reduced effect of pairing to enhance the PSP, consistent with view that PSP enhancement, like ITDP, is because of suppression of IPSP. H, Change in SC PSP response after inducing ITDP with 90 SC stimuli at 1 Hz, each one combined with dendritic current injection to elicit dendritic spikes with 2 μm AM251 present in bath (orange bar) during the stimulation (n = 5). Reduction in effect of SC+dendritic spikes to enhance PSP indicates dependence on CB1 receptors, similar to ITDP. I–L, Suppressing dendritic spikes during ITDP synaptic pairing protocol inhibits induction of ITDP and the reduction in IPSP. I, Example traces of the dendritic voltage signal during ITDP induction showing effect of hyperpolarizing current during suprathreshold events. Black, A PSP that does not reach dendritic spike threshold did not trigger current injection. Orange, A PSP that reached dendritic spike threshold without delivery of a hyperpolarizing current step. Light blue, A PSP that reached dendritic spike threshold with delivery of a hyperpolarizing current step, preventing firing of a dendritic spike. J, Time course of SC PSP amplitude when dendritic spikes are electrically blocked during ITDP induction. The green data points indicate measurements after bath application of SR95531 and CGP55845, indicating presence of a large IPSP. K, SC PSP amplitude before (black) and after (green) bath application of SR95531 and CGP55845 when dendritic spikes were electrically prevented during the ITDP protocol, indicating large IPSP. L, Same as I when dendritic spikes were allowed to fire during ITDP induction, indicating loss of IPSP. *p < 0.05; Mann-Whitney test.

Does the potentiation induced by paired SC stimulation and dendritic spiking reflect the same plasticity mechanism as ITDP induced by paired synaptic stimulation? ITDP largely results from the long-term-depression of the component of the feedforward IPSP evoked by SC stimulation (iLTD) that is mediated by cholecystokinin-positive (CCK⁺) interneurons (accounting for ∼75% of the potentiation of the PSP), with a smaller component caused by LTP of the EPSP (eLTP; accounting for ∼25% of the potentiation; Basu et al., 2013). To determine whether pairing of SC stimulation with dendritic spiking also suppressed the IPSP, we assessed the extent of inhibition by applying blockers of GABA_A (2 μm SR95531) and GABA_B (1 μm CGP55845) receptors on the PSP amplitude before and after induction of plasticity. Normally, the application of the GABA receptor antagonists produced a nearly 3-fold increase in the peak depolarization during the SC PSP because of blockade of the rapid feedforward IPSP (peak SC PSP increased to 282.6 ± 40.9% of initial value, p = 0.031, n = 6; Fig. 9E). However, 20 min following the induction of ITDP with paired synaptic stimulation, the blockers increased the PSP to only 133.0 ± 14.2% of its value relative to control (p = 0.125, n = 4; Fig. 9L), indicating a pronounced decrease in feedforward inhibition. We observed a similar loss of sensitivity to GABA_A and GABA_B receptor blockade (SR95531 and CGP55845) 20 min after pairing SC stimulation with dendritic spikes, with the blockers increasing the PSP to only 114.3 ± 2.5% of control (PSP increased from 10.78 ± 1.29 to 12.35 ± 1.60 mV, p = 0.13, n = 4; Fig. 9F).

To further confirm that the plasticity induced by pairing SC stimulation with dendritic spike firing was related to ITDP, we applied the induction protocol with inhibition continuously blocked (using SR95531 and CGP55845). This largely eliminated the enhancement of the PSP, with only a small residual potentiation (127.5 ± 6.8%, n = 4), which likely reflects the small component of eLTP recruited by ITDP induction (Fig. 9G). These results thus support the view that the plasticity induced by pairing SC stimulation with dendritic spiking largely resulted from ITDP. In contrast, if the pairing protocol induced conventional Hebbian LTP or STDP, it should have been greatly enhanced with inhibition blocked (because of increased postsynaptic depolarization).

As a third approach to determine the nature of the plasticity mechanism, we examined the effects of bath application of the endocannabinoid CB1 receptor blocker AM251 (2 μm) during the ITDP induction protocol. Because the iLTD component of ITDP results from the activation of CB1 receptors on presynaptic terminals of CCK⁺ interneurons, AM251 blocks the majority of ITDP induced by paired synaptic stimulation (Basu et al., 2013). Consistent with these results, application of AM251 also greatly reduced the synaptic potentiation produced by pairing dendritic spikes with SC stimulation (144.3 ± 11.9%, n = 5, which is significantly less than the potentiation in the absence of the blocker seen above; p = 0.0062), with the residual potentiation likely resulting from eLTP (Fig. 9H).

We thus conclude that dendritic spikes in conjunction with SC stimulation are sufficient to induce ITDP. But are dendritic spikes necessary? Can ITDP be induced during paired distal and proximal synaptic stimulation if dendritic spikes are prevented? Addressing this question is complicated by the fact that pharmacological blockade of active dendritic events would also compromise synaptic transmission, which is essential for ITDP induction. We therefore suppressed the generation of dendritic spikes electrically by using an online computer-controlled program to inject negative current into the dendrite to prevent dendritic spiking whenever the membrane depolarized to the dendritic spike threshold. This prevented dendritic spikes from occurring while keeping interference with summation of distal and proximal PSPs at a minimum, as long as they remain below the threshold for negative current injection (Fig. 9I, dotted line). Prevention of dendritic spikes in this manner almost fully suppressed ITDP evoked by the standard paired synaptic stimulation protocol. Twenty min after delivering the induction protocol with dendritic spikes suppressed, the SC-evoked PSP was increased to only 126.5 ± 7% of its initial value, compared with the 274.3 ± 34.2% increase under normal condition (p < 0.05 with Mann–Whitney test, n = 4; Fig. 9I–K).

To assess the extent of inhibition when the ITDP induction protocol was delivered with dendritic spiking suppressed, we applied GABA_A and GABA_B receptor antagonists (SR95531 and CGP55845) 20 min after paired stimulation. After pairing, the blockers still caused a large increase in the net SC PSP amplitude (to 310.8 ± 40.0%, p < 0.0001, n = 4; Fig. 9I–K), similar to their effect under baseline conditions. Thus, the prevention of dendritic spiking suppressed the induction of iLTD. In contrast, when we allowed dendritic spikes to occur normally during the ITDP induction protocol, we observed an increase of the SC response to 252.9 ± 34.1% of baseline (p = 0.045, n = 4), and subsequent application of SR95531 and CGP55845 caused only a slight, statistically insignificant increase in PSP amplitude (325.1 ± 32.6% from baseline but only 133.0 ± 14.2% from the post-ITDP SC response, p = 0.125, n = 4; Fig. 9L), consistent with previous results showing that ITDP largely suppresses feedforward inhibition (Basu et al., 2013). Together, these results indicate that dendritic spikes are both necessary and, in conjunction with SC stimulation, sufficient to induce ITDP.