Transient ECM protease activity promotes synaptic plasticity

Abstract

Activity-dependent proteolysis at a synapse has been recognized as a pivotal factor in controlling dynamic changes in dendritic spine shape and function; however, excessive proteolytic activity is detrimental to the cells. The exact mechanism of control of these seemingly contradictory outcomes of protease activity remains unknown. Here, we reveal that dendritic spine maturation is strictly controlled by the proteolytic activity, and its inhibition by the endogenous inhibitor (Tissue inhibitor of matrix metalloproteinases-1 - TIMP-1). Excessive proteolytic activity impairs long-term potentiation of the synaptic efficacy (LTP), and this impairment could be rescued by inhibition of protease activity. Moreover LTP is altered persistently when the ability of TIMP-1 to inhibit protease activity is abrogated, further demonstrating the role of such inhibition in the promotion of synaptic plasticity under well-defined conditions. We also show that dendritic spine maturation involves an intermediate formation of elongated spines, followed by their conversion into mushroom shape. The formation of mushroom-shaped spines is accompanied by increase in AMPA/NMDA ratio of glutamate receptors. Altogether, our results identify inhibition of protease activity as a critical regulatory mechanism for dendritic spines maturation.

Figure 1. Enzymatic activity of recombinant auto-activating MMP-9 initiates morphological changes in dendritic spines that are concluded by the subsequent inhibition of proteolytic activity.

(A) Representative images of live cell imaging of 21 DIV hippocampal neurons that expressed EGFP. Dendrites were imaged before (CTRL) and after treatment (40 minutes of incubation with iaMMP-9/aaMMP-9). Examples of spines that became elongated after MMP-9 activity are indicated by red arrows. Relative changes (mean ± SEM) in the length/width ratio of dendritic spines 40 min after iaMMP-9 or aaMMP-9 treatment. Incubation with iaMMP-9 did not induce prominent changes in dendritic spine shape (n_spine = 267; length/width, -0.014 ± 0.024), whereas 40 min incubation with aaMMP-9 caused elongation of the spines (n_spine = 838; length/width, 0.050 ± 0.015; p = 0.02). (B) Representative images of live cell imaging of 21 DIV hippocampal neurons that expressed EGFP. Dendrites were imaged before (CTRL) and after treatment (40 minutes of incubation with aaMMP-9 and then GM6001/DMSO for 30 minutes, 70 minutes with GM6001/DMSO).The inhibition of MMP activity by GM6001 blocked elongation of the spines (n_spine = 488; length/width, 0.017 ± 0.020; p = 0.006). The GM6001 solvent (DMSO) did not stop the elongation process induced by aaMMP-9 application (n_spine = 370; length/width, 0.103 ± 0.023). Comparing to control (incubation with DMSO itself for 70 min) the incubation with aaMMP-9+DMSO induced significant changes in dendritic spine length/width (n_spine = 266; length/width, 0.036 ± 0.021; p = 0.0425). The application of GM6001 for 70 minutes did not induce a significant change in dendritic spines morphology (n_spine = 339; length/width, 0.028 ± 0.019).

Figure 2. MMP-9 inhibition in transgenic rats that overexpress MMP-9 induces the maturation of dendritic spines.

(A) Representative images of fragments of DiI-stained dendrites in the CA1 area of the hippocampus from transgenic rats that overexpressed aaMMP-9 (MMP-9 TG) and wild type (WT) animals. The dendritic spines of MMP-9 TG rats were longer and thinner compared with WT. Application of the MMP inhibitor GM6001 caused the maturation of dendritic spines in TG animals, but it did not affect the shape of dendritic spines in WT animals. (B) The bar plot shows a significant decrease in the length/width ratio of dendritic spines in MMP-9 TG rats after GM6001 administration (n_rats = 4; length/width, 1.72 ± 0.04; p < 0.001) compared with dendritic spines in DMSO-treated MMP-9 TG rats (n_rats = 3; length/width, 2.35 ± 0.056). The dendritic spines of MMP-9 TG rats were significantly longer and thinner (n_rats = 3; length/width, 2.35 ± 0.056; p < 0.001) compared with WT in the control state (DMSO; n_rats = 4; length/width, 1.95 ± 0.054).

Figure 3. Inhibition of MMP-9 activity reduces the number of AMPAR-silent synapses presumably through the recruitment of AMPARs to the synapse.

(A–C) Schematic representation of the minimal stimulation protocol that was used to calculate the number of silent synapses. The graphs show 40–50 consecutive stimulation trials, resulting in either successes (gray, blue or red circles) or failures (black circles) in triggering AMPAR- and NMDAR-mediated EPSCs, recorded at −60 mV and +45 mV, respectively. (D) Summary bar graph that shows an increase in the number of silent synapses in rats that overexpressed MMP-9. Application of the MMP inhibitor GM6001 (25 μM) decreased the number of AMPAR-silent synapses. Numbers in brackets represent the number of recorded cells. (E) Example average traces of AMPA EPSCs that were recorded at −60 mV and composite AMPA and NMDA EPSCs that were recorded at +45 mV, peak-scaled to the size of the NMDA response in WT rats (shown in gray). The relative amplitude of AMPA responses decreased in MMP-9 TG animals (red) and increased upon MMP-9 inhibition with GM6001 (blue). (F) Summary bar graph that shows a decrease in the ratio of AMPA/NMDA EPSCs amplitudes in rats that overexpressed MMP-9. Application of the MMP inhibitor GM6001 increased the relative size of AMPAR-mediated currents. Numbers in brackets represent the number of recorded cells.

Figure 4. Long-term potentiation in TG rats that overexpress MMP-9 is altered and can be rescued by applying an MMP inhibitor.

(A) The figure shows the time course of maximal EPSP slopes normalized to baseline in the CA1 region of the hippocampus. Long-term potentiation that was induced by high-frequency stimulation (HFS; 3× 100 Hz; black arrow) of the Schaffer collaterals in the presence of DMSO in slices from MMP-9 TG rats had a lower magnitude (filled red squares; n_rats = 6; 130.8% ± 6.2% of baseline) compared with control slices from WT rats (filled gray circles; n_rats = 6; 175.7% ± 8.4% of baseline; p = 0.0004). The inhibitor of metalloproteinases, GM6001, was applied immediately after HFS and improved LTP that was evoked in slices from MMP-9 TG rats (filled dark blue squares; n_rats = 6; 203.1% ± 16.8% of baseline) compared with slices from the same animals that were treated with the solvent of inhibitor only (filled red squares; p = 0.001). GM6001 application during the recordings from slices that were obtained from WT rats caused LTP destabilization 30 min after LTP induction (filled light blue circles; p = 0.002, compared with DMSO-treated slices from WT rats; filled gray circles). The light blue line marks the time when GM6001 was present in the experimental solution. Error bars represent the SEM. (B) Representative traces of fEPSP 10 min before (black) and 15 and 90 min after (gray) the induction of LTP are shown. Scale bars = 2 mV and 5 ms.

Figure 5

cLTP elevates TIMP-1 protein expression (A) Representative Western blot on conditioned media obtained from 21 DIV dissociated hippocampal cultures that were stimulated with cLTP. An increase in the level of TIMP-1 protein occurred 10 min after cLTP (1.78 ± 0.12; p = 0.008) stimulation compared with control media that were treated with DMSO (0.65 ± 0.14). (B) Representative Western blot of cell lysates from 21 DIV dissociated hippocampal cultures. Enhanced proteolysis of β-dystroglycan occurred 10 min after cLTP stimulation compared with the control (DMSO).

Figure 6. Inhibition of endogenous MMP-9 is required for the maturation of dendritic spines in dissociated hippocampal cultures.

(A) Results of the DQ-gelatin assay. MMP-9 activity (red) was inhibited by TIMP-1 (blue). The effect of MMP-9 inhibition by TIMP-1 was blocked by incubation with iaMMP-9 (gray). (B) Representative images of live cell imaging of 21 DIV hippocampal neurons that expressed EGFP, revealing changes in dendritic spine morphology (indicated by red arrows) after incubation with cLTP+iaMMP-9. The same set of dendritic spines was analyzed before and after stimulation. (C) Relative changes (mean ± SEM) in the length/width ratio of dendritic spines incubated with cLTP+iaMMP-9. The spines became longer and thinner after 10 min cLTP+iaMMP-9 treatment (n_spine = 370; length/width, 0.069 ± 0.026; p < 0.001 and p = 0.004) compared with stimulation without sequestration of the endogenous inhibitor (cLTP only; n_spine = 641; length/width, −0.101 ± 0.021) and control (DMSO, vehicle of cLTP; n_spine = 379; length/width, −0.025 ± 0.021). After 40 min, the changes in dendritic spine length were persistent. cLTP+iaMMP-9 stimulation increased the length/width ratio (n_spine = 370; length/width, 0.120 ± 0.027; p < 0.001 and p = 0.01) compared with cLTP (n_spine = 641; length/width, −0.253 ± 0.022) and the control (n_spine = 379; length/width, 0.032 ± 0.024). (D) Relative changes (mean ± SEM) in the head width of dendritic spines. Dendritic spine heads after 10 min cLTP were wider (n_spine = 641; head width, 0.133 ± 0.018; p < 0.001 and p < 0.001) compared with dendritic spines in cLTP+iaMMP-9 treatment (n_spine = 370; head width, −0.019 ± 0.021) and control (n_spine = 364; head width, 0.044 ± 0.018). Stimulation for 40 min with cLTP induced enlargement of head width (n_spine = 641; head width, 0.091 ± 0.017; p < 0.001 and p < 0.001) compared with cLTP+iaMMP-9 (n_spine = 370; head width, −0.027 ± 0.023) and the control (n_spine = 364; head width, −0.016 ± 0.020).

Figure 7. Inhibition of endogenous MMP-9 is required for LTP maintenance.

(A) The figure shows the time course of maximal EPSP slopes normalized to baseline measured in the CA1 area of the hippocampus. High-frequency stimulation in slices from WT rats that were treated with iaMMP-9 (open red circles; n_rats = 5) evoked LTP at a similar level (180.2% ± 11.5% of baseline) as in untreated slices from wild type rats (filled gray circles; n_rats = 6; p = 0.560 for the first 45 min after LTP induction). After approximately 45 min, LTP on slices from WT rats treated with iaMMP-9 declined to the level that was previously observed in slices from MMP-9 TG rats that were treated with DMSO (filled red squares; n_rats = 6; p = 0.026, compared with untreated slices from wild type animals for the last 45 min of recording; p = 0.358, compared with untreated slices from MMP-9 TG rats). Black arrows mark the time of application of HFS. The open gray line marks the time when iaMMP-9 was present in the experimental solution. Error bars represent the SEM. (B) Representative traces of fEPSP 10 min before (black) and 15 and 90 min after (gray) LTP induction are shown. Scale bars = 2 mV and 5 ms.

Figure 8. Transient ECM protease activity promotes synaptic plasticity.

Proteolytic activity per se initiates the promotion of structural and functional plasticity, which requires subsequent endogenous enzymatic inhibition to be concluded. Thus, spines first become elongated based on MMP activity. Subsequently, extracellular proteolysis is terminated (due to TIMP-1), resulting in dendritic spine growth that is expressed as expansion of its head and the incorporation of AMPARs to previously silent synapses, a process that is required for LTP maintenance.