Developmental switch in the contribution of presynaptic and postsynaptic NMDA receptors to long-term depression

Abstract

NMDA receptor (NMDAR) activation is required for many forms of learning and memory as well as sensory system receptive field plasticity, yet the relative contribution of presynaptic and postsynaptic NMDARs over cortical development remains unknown. Here we demonstrate a rapid developmental loss of functional presynaptic NMDARs in the neocortex. Presynaptic NMDARs enhance neurotransmitter release at synapses onto visual cortex pyramidal cells in young mice [before postnatal day 20 (P20)], but they have no apparent effect after the onset of the critical period for receptive field plasticity (>P23). Immunoelectron microscopy revealed that the loss of presynaptic NMDAR function is likely attributable in part to a 50% reduction in the prevalence of presynaptic NMDARs. Coincident with the observed loss of presynaptic NMDAR function, there is an abrupt change in the mechanisms of timing-dependent long-term depression (tLTD). Induction of tLTD before the onset of the critical period requires activation of presynaptic but not postsynaptic NMDARs, whereas the induction of tLTD in older mice requires activation of postsynaptic NMDARs. By demonstrating that both presynaptic and postsynaptic NMDARs contribute to the induction of synaptic plasticity and that their relative roles shift over development, our findings define a novel, and perhaps general, property of synaptic plasticity in emerging cortical circuits.

Figure 1.

Postsynaptic NMDAR function can be blocked with either hyperpolarization or the inclusion of MK-801 in the postsynaptic recording pipette (iMK-801). A, Model depicting the commonly used experimental protocol for detecting functional presynaptic NMDARs. After blocking postsynaptic NMDARs with strong hyperpolarization (or iMK-801), the role of presynaptic NMDARs in synaptic transmission can be tested by blocking the remaining presynaptic NMDARs with bath application of APV. B, Normalized I–V relationship at the L4→L2/3 synapse of pharmacologically isolated NMDAR-mediated EPSCs. Dotted line is a fit to the linear portion of the I–V relationship. Note the strong block of NMDAR currents by hyperpolarization (n = 11; average age of animals, ∼P24). C, Top, APV blocks the NMDAR component of mEPSCs recorded at −60 mV in 0.1 mm Mg²⁺. Bottom, APV has no postsynaptic effect on the amplitude or kinetics of mEPSCs recorded at −80 mV in 1 mm Mg²⁺, suggesting that the mEPSC currents are mediated by AMPA receptors and that the NMDAR component is nonexistent or negligible. D, Synaptic I–O relationship for L2/3 cells recorded at +24 mV with iMK-801 (open circles) or without iMK-801 (filled circles). Inset, Representative traces for control and iMK-801 I–O curves. Even at this depolarized holding voltage, iMK-801 blocks >96% of the NMDAR current evoked at 20 μA, which is our average stimulation intensity. E, Synaptic I–V relationship for the same cells shown in D showing that, with iMK-801 and a stimulation of 30 μA, the NMDAR current is completely blocked (n = 4) compared with control (n = 5) at −65 mV. Inset, Representative traces for the I–V recordings.

Figure 2.

Presynaptic NMDARs tonically increase the probability of neurotransmitter release onto L2/3 pyramidal cells in the mouse visual cortex. A, Example recording from an L2/3 pyramidal cell from a P16 mouse demonstrating that 100 μm APV reversibly reduces mEPSC frequency. Events are indicated by asterisks. B, Example experiment demonstrating that the reduction in mEPSC frequency (freq.) by APV is reversible. C, Cumulative probability histograms from an L2/3 pyramidal cell at P16 demonstrating that APV application reversibly increases mEPSC interevent interval (C₁) without affecting amplitude (C₂). D, Representative data from a single cell demonstrating that neither interevent interval (D₁) nor amplitude (D₂) changed in the absence of APV treatment. The control cell was recorded for the same duration as experiments in which APV was applied.

Figure 3.

Developmental loss of functional presynaptic NMDARs in visual cortex. A, Averaged data demonstrating that APV strongly reduced mEPSC frequency in L2/3, L4, and L5 pyramidal cells in mice aged P7–P11 and P13–P20, suggesting the presence of functional presynaptic NMDARs. d,l-APV had no effect in L2/3, L4, or L5 cells in mice aged P23–P30 or P72–P90. d-APV was used in a subset of experiments (where indicated) at P13–P20 in L2/3 and showed the same effect on mEPSC frequency. B, The reduction in mEPSC frequency was reversed with washout of APV. No significant rundown in mEPSC frequency was observed in control cells not exposed to APV but recorded for a similar duration. C, D, No significant changes in mEPSC amplitude were observed in any of the groups. Asterisks in A and B indicate significance of p < 0.0045 corrected for multiple tests using the Bonferroni method, and sample sizes are given within the bars. The mEPSC frequency and amplitude in A–D were normalized to the averaged baseline values before APV application.

Figure 4.

The presence of presynaptic NR1 is downregulated with development. A, Electron micrograph in visual cortex L2/3 of a P16 mouse demonstrating an NR1-positive terminal (t+) making a synapse with a spine that is also NR1 positive (s+). An NR1-negative presynaptic terminal (t−) making an asymmetric synapse onto an NR1-negative spine (s−) is present in the same field. Scale bar: A–D, 250 nm. Arrowheads highlight aggregations of DAB in presynaptic terminals. B, In a section from L2/3 of another P16 mouse, a diffusely labeled terminal (t+) is seen forming a synapse onto a spine (s+) that contains NR1 label at the postsynaptic density. C, At P27, most synapses exhibit postsynaptic, but not presynaptic, NR1. An unlabeled terminal (t−) forms a synapse onto a labeled dendrite (d+). D, In a section from another P27 mouse, an NR1-positive terminal (t+) makes a synapse onto an NR1-positive spine (s+). E, Scatter plot from four mice (2 at each age) quantifying the selective loss of presynaptic, but not postsynaptic, NR1 over development. Note that 30% of terminals still contain NR1 at P27.

Figure 5.

Presynaptic NMDARs are required for tLTD at L4→L2/3 synapses in young (A, Example of tLTD induced at L4→L2/3 synapses by AP–EPSP pairings in a P13 mouse. R_input and V_m do not change significantly over the time course of the experiment. B, Averaged data from control cells demonstrating the depression in EPSP slope (p < 0.007) induced with AP–EPSP pairings. C, Strong tLTD could be induced with AP–EPSP pairings even when postsynaptic NMDARs were blocked by inclusion of MK-801 in the internal recording solution (iMK-801) (p < 0.018). This suggests that postsynaptic NMDARs are not required for tLTD induction in these young mice. D, Induction of tLTD was prevented by bath application of the NMDAR antagonist APV (p > 0.798), arguing that presynaptic NMDARs are required for the induction of tLTD.

Figure 6.

Analysis of synaptic depression indicates that tLTD induction in P13–P17 mice is expressed as a decrease in release probability. A, Sample waveform of six pulses evoked by 30 Hz stimulation before and after AP–EPSP pairings. This example recording was made with iMK-801. B, In the control conditions, the rate of synaptic depression at L4→L2/3 connections was reduced after tLTD produced by AP–EPSP pairings. This suggests that the pairing protocol caused a lasting reduction in neurotransmitter release. The STD index was used here as a measure of synaptic depression (see Materials and Methods). C, AP–EPSP pairing induction of tLTD made with iMK-801 also produced a reduction in the rate of synaptic depression, indicating that the pairing-induced reduction in neurotransmitter release did not require activation of postsynaptic NMDARs. D, There was no significant change in synaptic depression when tLTD is blocked by bath application of APV, suggesting a role for presynaptic NMDARs in the tLTD. Data are from the same cells as in Figure 5.

Figure 7.

Postsynaptic NMDARs are required for tLTD induction in critical period mice (P23–P30). A, Example of tLTD induced at L4→L2/3 synapses by AP–EPSP pairings in a P27 mouse. B, In P23–P30 mice, AP–EPSP pairings induced a small degree of tLTD (p = 0.02). C, No tLTD was induced when postsynaptic NMDARs were blocked by iMK-801 (p > 0.79), suggesting that the activation of postsynaptic NMDARs is required for the full expression of the long-term depression. D, No tLTD was induced when inhibition was left intact (no picrotoxin in the bath ACSF) (p = 0.30).

Figure 8.

Analysis of synaptic depression indicates that tLTD induction in P26–P28 mice can be expressed as a decrease in release probability. A, Sample waveform of six pulses evoked by 30 Hz stimulation before and after AP–EPSP pairings. B, The rate of synaptic depression at L4→L2/3 connections was reduced after tLTD produced by AP–EPSP pairings.