Extracellular proteolysis by matrix metalloproteinase-9 drives dendritic spine enlargement and long-term potentiation coordinately

Abstract

Persistent dendritic spine enlargement is associated with stable long-term potentiation (LTP), and the latter is thought to underlie long-lasting memories. Extracellular proteolytic remodeling of the synaptic microenvironment could be important for such plasticity, but whether or how proteolytic remodeling contributes to persistent modifications in synapse structure and function is unknown. Matrix metalloproteinase-9 (MMP-9) is an extracellular protease that is activated perisynaptically after LTP induction and required for LTP maintenance. Here, by monitoring spine size and excitatory postsynaptic potentials (EPSPs) simultaneously with combined 2-photon time-lapse imaging and whole-cell recordings from hippocampal neurons, we find that MMP-9 is both necessary and sufficient to drive spine enlargement and synaptic potentiation concomitantly. Both structural and functional MMP-driven forms of plasticity are mediated through beta1-containing integrin receptors, are associated with integrin-dependent cofilin inactivation within spines, and require actin polymerization. In contrast, postsynaptic exocytosis and protein synthesis are both required for MMP-9-induced potentiation, but not for initial MMP-9-induced spine expansion. However, spine expansion becomes unstable when postsynaptic exocytosis or protein synthesis is blocked, indicating that the 2 forms of plasticity are expressed independently but require interactions between them for persistence. When MMP activity is eliminated during theta-stimulation-induced LTP, both spine enlargement and synaptic potentiation are transient. Thus, MMP-mediated extracellular remodeling during LTP has an instructive role in establishing persistent modifications in both synapse structure and function of the kind critical for learning and memory.

Fig. 1.

Persistence of TBP-triggered spine expansion and LTP requires MMP proteolysis. (Ai). A representative CA1 neuron showing the position of the local stimulating (Stim) and whole-cell recording (Rec) electrodes. (Scale bar: 50 μm.) (Aii and Aiii) Representative images (Aii) and EPSP traces (Aiii) showing effects of the MMP inhibitor (Inhibitor II) on TBP-induced spine expansion and potentiation. The spines and EPSP traces are shown before (1), 5 min after (2), and 45 min after (3) TBP, and correspond to times (in minutes) indicated in the population data shown in B and C. In this and subsequent figures, expanded and stable spines are indicated by arrowheads and arrow, respectively. (Scale bars: 1 μm, 5 mV, and 50 ms.) (B) Population data showing that immediate TBP-triggered spine expansion is unaffected by MMP inhibitors (143.5 ± 7.4% vs. 145.7 ± 6.0% at 5 min post-TBP, P > 0.1), but persistence of expansion is blocked (filled triangles) in comparison with control neurons (open triangles). (C) Population data showing initial TBP-triggered potentiation is also unaffected by MMP inhibitors (filled circles, 144.9 ± 17.3% inhibitor-treated vs. 153.3 ± 19.7% untreated at 5 min post-TBP, P > 0.4), but persistence of potentiation is blocked in comparison with long-lasting potentiation after TBP in control neurons (open circles). All statistical tests are Mann–Whitney U tests. Gray lines in B and C indicate duration of bath-applied Inhibitor II.

Fig. 2.

MMP-9 drives persistent spine expansion and LTP at naive spines. (A) A representative experiment from a single CA1 neuron showing persistent spine expansion and synaptic potentiation after brief puffing of MMP-9 onto naive spines. Numbers indicate corresponding times on the recording. (Scale bars: 1 μm, 5 mV, and 50 ms.) (B) Population data showing that MMP-9 induces a persistent potentiation (open circles, 213.1 ± 38.9% at 75 min post-MMP-9 exposure; P < 0.001, n = 10). Neurons exposed to inactive pro-MMP-9 show no changes in EPSPs (filled circles). (C) Population data showing that MMP-9 induces persistent spine expansion (open triangles, 143.9 ± 6.1% of the baseline level at 75 min post-MMP-9 exposure; n = 95 spines/10 cells; P < 0.001). Such effects on spine size are not seen in response to inactive pro-MMP-9 (filled triangles). All statistical tests are paired t tests. In this and subsequent figures, black bar indicates duration of local MMP-9 puffing.

Fig. 3.

MMP-9-induced structural and functional plasticity is mediated by integrins. (A) Preincubation with a neutralizing β1-integrin antibody prevents MMP-9-induced spine expansion. (B) Preincubation with β1-integrin antibody also prevents MMP-9-induced synaptic potentiation.

Fig. 4.

MMP-9 drives integrin-dependent cofilin phosphorylation in dendritic spines. (A) Representative Western blotting showing levels of phosphocofilin (p-cof) and total cofilin (cof) in untreated control sections, ones treated with MMP-9, ones treated with the β1-integrin neutralizing antibody alone, or ones treated with MMP-9 in the presence of the β1-integrin-neutralizing antibody. Quantitative analysis shows significant elevation in levels of phosphocofilin after MMP-9 exposure (*, P < 0.001 in comparison with other conditions, n = 5 rats, ANOVA) that is blocked by the β1-integrin-neutralizing antibody. (B) Confocal microscope images taken from area CA1 stratum radiatum showing phosphocofilin immunofluorescent puncta in untreated control slices or ones exposed to MMP-9. (Scale bar: 5 μm.) The graphs indicate significantly greater numbers and sizes of puncta in the MMP-9 stimulated slices in comparison with control slices (*, P < 0.01, unpaired t tests, n = 5 rats). (C) High-power confocal images from an MMP-9-stimulated slice showing that labeling for the postsynaptic density marker PSD-95 (green), which is concentrated in spines, codistributes with that for phosphocofilin (red), as indicated by the yellow pixels shown in the merge and Insets (arrows). (Scale bar: 5 μm.)

Fig. 5.

Blocking postsynaptic exocytosis or protein synthesis prevents MMP-9-induced potentiation and destabilizes MMP-9-induced spine expansion. (A) Sample images and EPSP traces showing the transient spine expansion (arrowheads) and small increase in EPSP size in a representative neuron internally loaded with light chain of BoTox (postsynaptic). (Scale bars: 1 μm, 5 mV, and 50 ms.) Numbers indicate corresponding times in A–C. (B) Population data showing that synaptic potentiation is largely absent in neurons loaded with BoTox after exposure to MMP-9 (119.1 ± 11.5% of baseline at 75 min after MMP-9 puffing vs. 99.2 ± 5.1% at 0 min before MMP-9 puffing, P > 0.1, paired t test). (C) Initial spine expansion in response to MMP-9 occurred normally in neurons loaded with BoTox (142.8 ± 7.0% at 15 min post-MMP-9 exposure, P > 0.3, Mann–Whitney U test), but was transient. At 75 min postpuffing, when the experiments were terminated for technical reasons, spine volume had decayed significantly from initial peak values (115.2 ± 1.8% at 75 min post-MMP-9 puffing, P < 0.01, paired t test, n = 108 spines/13 cells). At this time, the average spine volume remained significantly larger than baseline (P < 0.05, paired t test). We speculate that spine volume would continue to decay to baseline over time. (D) MMP-9-induced potentiation is significantly reduced by protein synthesis inhibition (black circles). The ghostline (gray circles) indicates comparative control values (MMP-9 alone) shown in Fig. 2B. (E) MMP-9-induced spine expansion in the presence of cycloheximide (black triangles) is transient, reaching control values initially (129.2 ± 4.3% MMP-9 alone vs. 138.7 ± 8.6% MMP-9 plus cycloheximide immediately after MMP-9 exposure, P > 0.2, Mann–Whitney U test), but then decays to baseline values (at 60 min, 95.4 ± 7.6% of baseline; n = 72 spines/8 cells). The ghostline (gray triangles) indicates comparative control values (MMP-9 alone) shown in Fig. 2C. Gray bars indicate duration of bath-applied cycloheximide.