Postsynaptic spiking determines anti-Hebbian LTD in visual cortex basket cells

Abstract

Long-term plasticity at pyramidal cell to basket cell (PC → BC) synapses is important for the functioning of cortical microcircuits. It is well known that at neocortical PC → PC synapses, dendritic calcium (Ca²⁺) dynamics signal coincident pre-and postsynaptic spiking which in turn triggers long-term potentiation (LTP). However, the link between dendritic Ca²⁺ dynamics and long-term plasticity at PC → BC synapses of primary visual cortex (V1) is not as well known. Here, we explored if PC → BC synaptic plasticity in developing V1 is sensitive to postsynaptic spiking. Two-photon (2P) Ca²⁺ imaging revealed that action potentials (APs) in dendrites of V1 layer-5 (L5) BCs back-propagated decrementally but actively to the location of PC → BC putative synaptic contacts. Pairing excitatory inputs with postsynaptic APs elicited dendritic Ca²⁺ supralinearities for pre-before-postsynaptic but not post-before-presynaptic temporal ordering, suggesting that APs could impact synaptic plasticity. In agreement, extracellular stimulation as well as high-throughput 2P optogenetic mapping of plasticity both revealed that pre-before-postsynaptic but not post-before-presynaptic pairing resulted in anti-Hebbian long-term depression (LTD). Our results demonstrate that V1 BC dendritic Ca²⁺ nonlinearities and synaptic plasticity at PC → BC connections are both sensitive to somatic spiking.

Keywords: action potential backpropagation; calcium imaging; inhibitory interneurons; plasticity; spike-timing dependent plasticity; synapse; visual cortex.

Figure 1

APs backpropagate decrementally but actively in BC dendrites. (A) Sample patched BC filled with Alexa 594 illustrate how dendritic processes were readily identified. Region in the red box is magnified in (B). (B) We imaged Ca²⁺ transients using line scans at randomized locations along BC dendrites. In this sample experiment, line scans were acquired in the order denoted by the numbers. Ca²⁺ transients at line scan 3 (circled) is shown in (C,D). (C) dG/R was recorded in response to a train of APs delivered while the BC was at rest (light blue) or depolarized (depol; dark blue). Sample Ca²⁺ transients were averaged across 20 sweeps at one dendritic location. The region in the dotted box is magnified in (D). Shading denotes the SEM. (D) The dG/R integral was taken over a 275-ms-long window (grey bar) to measure dendritic Ca²⁺ transients due to bAPs (dark red traces). The 100-ms-long baseline period (black bar) was set to zero. (E) Ca²⁺ transients due to bAP diminished with distance from the soma (LMM, p < 0.0001). Depolarization boosted Ca²⁺ transients (rest, 72 ± 5; depolarized, 100 ± 6; LMM, p < 0.001), implying the involvement of active conductances in AP backpropagation. Different marker styles denote individual PCs. Red markers represent sample locus shown in (B–F). Somatic depolarization boosted bAPs more with distance (Pearson’s r = 0.48, p < 0.001), suggesting active backpropagation. If boosting was passively mediated, the correlation ought to be non-existent or negative. Different marker styles denote individual PCs. Red marker indicates sample locus shown in (B–E).

Figure 2

More than 75% of PC → BC synapses were located within 100 μm. (A) Sample reconstructed connected PC → BC pair. Yellow circles indicate contact points between the PC axon and the BC dendrite. Inset: EPSPs recorded from the BC (blue trace) in response to PC spiking (pink trace) are shown in the inset. (B) Sholl plot (blue trace) across N = 12 PC → BC pairs revealed that 75% of BC dendrites were restricted to within a radius of ~110 μm from the BC soma (dotted line). The majority of putative synaptic contacts (purple) were also found within this region. Dark blue: Sholl analysis mean, pale blue: Sholl analysis SEM, grey: cumulative normalized Sholl plot. Box plot: median, first, and third quartiles, with whiskers representing minimum and maximum. (C) Out of 29 putative synaptic contacts (purple markers), 25 were located below the BC soma (blue circle; putative contact y-coordinate −77 ± 20 μm vs. zero, p < 0.001, Wilcoxon rank).

Figure 3

Synchronous EPSP-AP pairing elicited supralinear Ca²⁺ transients. (A) As indicated by this representative sample patched BC, the extracellular stimulation electrode was placed near the image BC dendrite (red box). (B) The dendritic region highlighted by the red box in (A) was targeted for frame scans. The dotted red box indicates the boundaries of the frame scan. (C) Ca²⁺ transients were detected by frame scans in response to extracellular stimulation and somatic current injection (onset denoted by the black arrow). The black bar represents the baseline period. The grey bar represents the period over which dG/R was integrated. The three images are sample movie frames positioned at their time points. (D) We paired pre-and postsynaptic activation in two different ways, synchronously or asynchronously. For each of the two ways, we used three conditions: presynaptic activity alone (“EPSPs”), postsynaptic activity alone (“APs”), or both (“Both”). Black arrows indicate the time of stimulation onset. (E) Ca²⁺ nonlinearity was assessed by comparing both to the arithmetic sum of EPSPs + APs. For synchronous pairing, postsynaptic activation followed presynaptic activation by Δt = 10 ms. For asynchronous pairing, postsynaptic activation preceded presynaptic activation by Δt = −90 ms. Solid line: mean. Shaded region: SEM. Black bar: baseline. Grey bar: integration period. Black arrows indicate stimulation onset, corresponding to black arrows in (D). (F) Ca²⁺ supralinearity resulted from synchronous but not asynchronous pairing (p < 0.001, LMM; synchronous Both, 7.8% ± 1% vs. Sum, 5.6% ± 0.7%; p < 0.001; asynchronous Both, 6.4% ± 1% vs. Sum, 6.8% ± 1%; p = 0.14). Open circles: individual dendritic segments. Closed circles: mean ± SEM. (G) Synchronous and asynchronous pairing elicited different Ca²⁺ supralinearities in BC dendrites (Synch, 143% ± 5.4% vs. Asynch, 89.8% ± 4.6%; p < 0.001). Ca²⁺ supralinearity denotes the ratio dG/R_Both over dG/R_Sum. Filled circles: individual dendritic segments. Box plot: median, first, and third quartile, with whiskers showing minimum and maximum.

Figure 4

Synchronous EPSP-AP pairing elicited PC → BC LTD. (A) PC → BC synapses were identified using 2P optogenetics. A BC (green) was patched and monitored for EPSPs while surrounding ChroME-expressing PCs (red) were activated with femtosecond laser spiral scans. (B) Multiple ChroME expressing PCs (red fluorescence) were sequentially activated by 2P excitation to find connections onto the patched BC (green circle), an approach we call optomapping (Chou et al., 2024). Connected PCs were used in long-term plasticity experiments with synchronous Δt = 10 ms (blue circles) and asynchronous Δt = −90 ms pairings (pink circles). Sample PCs in (C) and (D) are indicated by triangles. (C) Synchronous pairing at this sample PC → BC synapse elicited LTD (after/before, 76%, p < 0.05). During the induction (grey), the presynaptic PC was activated by laser light (cyan rectangles) and the postsynaptic BC was activated by somatic current injections Δt = 10 ms later. Periods indicated by pink and red bars were used to quantify EPSPs (pink and red traces). (D) Asynchronous pairing at this sample PC → BC connection did not evoke detectable plasticity (after/before, 98%, p = 0.84). During the induction (grey), the presynaptic PC was activated (cyan rectangles) Δt = −90 ms after the postsynaptic BC. Periods indicated by light and dark blue bars were used to quantify EPSPs (light and dark blue traces). (E) Pooled across postsynaptic cells, synchronous pairing consistently evoked LTD (after/before, 83% ± 2%), while asynchronous pairing yielded no discernible plasticity (after/before, 99% ± 3%). Synchronous and asynchronous pairing yielded different long-term plasticity outcomes at PC → BC synapses (p < 0.001). (F) The change in PPR (ΔPPR) was indistinguishable for synchronous (0.056 ± 0.03) and asynchronous pairing (0.022 ± 0.04; p = 0.16). Open circles: individual connections. Box plot: median, first, and third quartile, with whiskers denoting minimum and maximum. (G) LTD caused 1/CV² to decrease, suggesting that LTD was expressed presynaptically via reduced release (Brock et al., 2020). Open circles: individual connections. Filled circle: mean ± SEM.