Input-Specific Metaplasticity in the Visual Cortex Requires Homer1a-Mediated mGluR5 Signaling

Abstract

Effective sensory processing depends on sensory experience-dependent metaplasticity, which allows homeostatic maintenance of neural network activity and preserves feature selectivity. Following a strong increase in sensory drive, plasticity mechanisms that decrease the strength of excitatory synapses are preferentially engaged to maintain stability in neural networks. Such adaptation has been demonstrated in various model systems, including mouse primary visual cortex (V1), where excitatory synapses on layer 2/3 (L2/3) neurons undergo rapid reduction in strength when visually deprived mice are reexposed to light. Here, we report that this form of plasticity is specific to intracortical inputs to V1 L2/3 neurons and depends on the activity of NMDA receptors (NMDARs) and group I metabotropic glutamate receptor 5 (mGluR5). Furthermore, we found that expression of the immediate early gene (IEG) Homer1a (H1a) and its subsequent interaction with mGluR5s are necessary for this input-specific metaplasticity.

Keywords: H1a; NMDA receptor; mGluR5; metabotropic glutamate receptor; metaplasticity; visual cortex.

Figure 1.. mGluR5 activity is required for experience-dependent downregulation of mEPSCs

(A) Experimental design. Mice were normally reared until opening of the critical period (~P21), then either given PBS or MPEP. For NR and DE conditions, PBS or MPEP (once daily i.p.) were given for 2 days, while for LE condition PBS/MPEP injection was given once after 2 days of DE and 30 min prior to light exposure. Arrows: time of each injection. (B, C) MPEP prevents experience-dependent reduction in mEPSC amplitudes. Top: Average mEPSC traces. Bottom left: Comparison of average mEPSC amplitude (B, PBS: NR=11.6±0.2 pA, n=15; DE=12.65±0.3 pA, n=17; LE=11.4±0.3 pA, n=13; ANOVA, F(2,42)=6.57, **p<0.01; C, MPEP: NR=12.9±0.4 pA, n=15; DE=12.3±0.3 pA, n=14; LE=12.9±0.2 pA, n=14; ANOVA, F(2,40)=1.074, p=0.35). Bottom right: Estimated population density plot of mEPSC amplitudes. X-axis: Estimated probability density, which is the probability per mEPSC amplitude estimated from each measured data fitted with the Gaussian kernels. Y-axis: shared with the left panel. For MPEP group, there was no significant change in the average mEPSC amplitudes, but a significant difference in the variance of the estimated population density probability (see Table S1) reflecting a change in the variance of mEPSC amplitudes in the population. (D) MPEP treatment (once daily i.p. for 2 days) in NR mice increases mEPSC amplitude in a multiplicative manner. (Mann-Whitney test: NR-PBS vs. NR-MPEP, p<0.0001; NR-PBS-scaled vs. NR-MPEP, p>0.01) (E, F) mEPSC frequency does not change across groups. Left: Comparison of average mEPSC frequency (E, PBS: NR=4.0±0.4 Hz, n=15; DE=3.5±0.2 Hz, n=17; LE=3.0±0.2 Hz, n=13; ANOVA, F(2,42)=2.07, p=0.14; F, MPEP: NR=4.1±0.4 Hz, n=15; DE=3.2±0.3 Hz, n=14; LE=3.9±0.3 Hz, n=14; ANOVA, F(2,40)=1.57, p=0.09). ns: Not statistically significant. Right: Estimated population density plot of mEPSC frequencies. X-axis: Estimated probability density, which is the probability per mEPSC frequency estimated from each measured data fitted with the Gaussian kernels. Also see Figure S1 and Table S1.

Figure 2.. Visual experience produces input-specific weakening of IC inputs to L2/3 neurons requiring NMDAR and mGluR5 activity

L4Cre mice (≥P21) were injected with AAV to express ChR2-tagged with YFP or mCherry in V1 L4 neurons. Virus was incubated for ≥6 weeks and recordings were done at P70-120. (A) Confocal image of a V1 slice used for recording showing ChR2-YFP in L4 (green) and a recorded neuron (red, biocytin filled) in L2/3. Left: lower magnification image. Right: higher magnification image of the L2/3 neuron showing a pyramidal shaped soma (red) and ChR2-YFP axons around (green). (B) Sr²⁺-mEPSCs recording example traces. L4 inputs were activated by blue LED stimulation (5 ms duration, blue arrow head). Spontaneous events were recorded before the evoked response during a 450 ms window (red-dashed line). Sr²⁺-desynchronized LED-evoked events were recorded 50 ms after the evoked response in a 400 ms window (red-solid line). The strength of evoked inputs was calculated by subtracting out the spontaneous events (see STAR methods for details). IC inputs were activated by a stimulating electrode placed laterally in L2/3, and quantified using the same method. (C,D) Changes in visual experience regulate the strength of lateral IC inputs to L2/3 neurons, but not FF inputs from L4. Left: Schematics of the experiment. Lateral IC inputs were activated via electrical stimulation (C), while L4 inputs were activated by ChR2 expression in L4Cre mice (D). Middle: Comparison of calculated average evoked Sr²⁺-mEPSC amplitude (C, IC inputs: NR=13.1±0.6 pA, n=12; DE=16.2±0.6 pA, n=12; LE=13.8±0.8 pA, n=13; ANOVA, F(2,34)=5.096, p<0.05; Newman-Keuls multiple comparison *p<0.05; D, FF inputs: NR=17.6±1.6 pA, n=13; DE=16.8±1.4 pA, n=12; LE=15.3±0.9 pA, n=13; ANOVA, F(2,35)=0.5751, p>0.72). Right: Estimated population density plot of Sr²⁺-mEPSC amplitudes. X-axis, Estimated probability density. Bottom: Average evoked Sr²⁺-mEPSC traces. (E,F) LE-induced reduction in lateral IC inputs to L2/3 neurons is dependent on NMDAR and mGluR5. Left: Comparison of calculated evoked Sr²⁺-mEPSC amplitude from IC inputs (E) and FF inputs (F) to L2/3 neurons of LE mice which received saline, NMDAR antagonist (CPP) or mGluR5 inverse agonist (MPEP) injection 30 min before reexposure to light following 2 days of DE (E, IC inputs: LE+Saline=12.8±0.9 pA, n=12; LE+CPP=18.2±1.4 pA, n=12; LE+MPEP=16.1±0.8 pA, n=15; ANOVA, F(2,36)=6.411, p<0.01; Newman-Keuls multiple comparison **p<0.01, *p<0.05; F, FF inputs: LE+Saline=15.2±0.8 pA, n=12; LE+CPP=15.11±1.0 pA, n=13; LE+MPEP=13.1±0.7 pA, n=17; ANOVA, F(2,39)=2.304, p>0.11). Right: Estimated population density plot of Sr²⁺-EPSC amplitudes. X-axis, Estimated probability density. Bottom: Average evoked Sr²⁺-mEPSC traces. Also see Table S1.

Figure 3.. H1aKOs lack visual experience-dependent synaptic weakening in V1 L2/3 neurons

(A) Schematics of the experiments. H1aWT and H1aKO mice were reared normally until P21~P30. A group of mice were put in the darkroom for 2 days (DE), and another group of mice underwent the same 2 day DE followed by 2 hours LE. (B,C) H1aKO lack experience-dependent regulation of mEPSCs. Top: Average mEPSC traces. Bottom left: Comparison of average mEPSC amplitude (B, H1aWT: NR=11.2±0.2 pA, n=20; DE=12.7±0.3 pA, n=14; LE=11.2±0.2 pA, n=14; ANOVA, F(2,45)=11.07, p=0.0001; Newman-Keuls multiple comparison ***p<0.001, ****p<0.0001; C, H1aKO: NR=12.5±0.4 pA, n=15; DE=12.8±0.4 pA, n=14; LE=12.3±0.4 pA, n=12; ANOVA, F(2,38)=0.4, p=0.67). Bottom right: Estimated population density plot of mEPSC amplitudes. X-axis, Estimated probability density. (D) Cumulative probability graph plotting mEPSC amplitudes from NR H1aKO (blue solid line) and NR H1aWT (black solid line) are statistically significantly different while H1aWT mEPSC amplitudes that are scaled up by a scaling factor (1.12) is not different from those of H1aKOs (Mann-Whitney test: NR-H1aWT vs. NR-H1aKO, p<0.0001; NR-H1aWT-scaled vs. NR-H1aKO, p>0.4). (E) No change in average mEPSC frequency. First panel: Comparison of average mEPSC frequency of H1aWTs (NR=4.2±0.4 Hz, n=20; DE=3.5±0.2 Hz, n=14; LE=4.6±0.3 Hz, n=14; Kruskal-Wallis test, p>0.09). Second panel: Estimated population density plots of mEPSC frequency (X-axis, Estimated probability density). Third panel: Comparison of average mEPSC frequency of H1aKOs (NR=5.0±0.4 Hz, n=15; DE=5.2±0.6 Hz, n=14; LE=4.4±0.3 Hz, n=12; ANOVA, F(2,38)=0.65, p=0.53). Fourth panel: Estimated population density plots (X-axis, Estimated probability density). Also see Figure S2 and S3; Table S1.

Figure 4.. H1a knockout does not grossly alter developmental changes in mEPSCs of V1 L2/3 neurons

(A) Schematics of the experiment. Top: mEPSCs were recorded from H1aWT and H1aKO mice at P11 (before eye opening) and during the critical period (P23-32). Bottom: H1aWTs either received daily injections of PBS or MPEP from P14 until P23-32. (B) Average mEPSC traces. (C,D) Developmental changes in mEPSC amplitude (C, H1aWT: P11=18.8±0.8 pA, n=21; P23-32=11.2±0.2 pA, n=20; unpaired t-test, ****p<0.0001; H1aKO: P11=16.9±0.9 pA, n=16; P23-32=12.5±0.4 pA, n=15; unpaired t-test, ***p<0.001) and frequency (D, H1aWT: P11=1.5±0.1 Hz, n=21; P23-32=4.2±0.4 Hz, n=20; unpaired t-test, ****p<0.0001; H1aKO: P11=1.0±0.2 Hz, n=16; P23-32=5.0±0.4 Hz, n=15; unpaired t-test, ****p<0.0001). Average mEPSC amplitude and frequency of H1aWT treated with PBS (orange circle) and MPEP (red triangle) are overlaid on each graph. Note: For all panels, the datasets for P23-P30 group are a replot of the NR data shown in Figure 3.

Figure 5.. Postnatal H1a expression is required for visual experience-dependent regulation of mEPSCs.

(A) Genotype scheme for producing conditional H1aKO. Survival surgeries were performed to inject AAV containing Cre in mouse V1 L2/3 neurons (age at injection P21-30). To localize H1a primarily in excitatory neurons, Cre expression was driven by CaMKII promoter and GFP was tagged as a reporter for expression. Control mice were injected with only CaMKII-driven GFP. MEPSC recordings were performed in GFP-expressing neurons at P28-45. (B-D) Lack of visual experience-dependent plasticity of mEPSCs in acute postnatal H1aKO neurons (H1b/c^+/−;H1a^−/−). First panel: Confocal image of a recorded neuron (red, filled with biocytin) also expressing Cre-GFP (green). Second panel: Comparison of average amplitude of mEPSCs (B, H1b/c^+/+;H1a^+/−: NR=11.8±0.3 pA, n=17; DE=13.1±0.4 pA, n=14; LE=11.1±0.4 pA, n=16; ANOVA, F(2,44)=8.031, p<0.005; C, H1b/c^+/−;H1a^−/−: NR=13.7±0.6 pA, n=17; DE=14.3±0.4 pA, n=14; LE=13.1±0.6 pA, n=12; ANOVA, F(2,40)=1.143, p>0.3; D, H1b/c^+/−;H1a^+/−: NR=11.7±0.3 pA, n=12; DE=14.7±0.6 pA, n=16; LE=12.8±0.5 pA, n=14; ANOVA, F(2,39)=8.229, p<0.005; Newman-Keuls multiple comparison *p<0.05, **p<0.01, ***p<0.001). Third panel: Estimated population density plot of mEPSC amplitudes (X-axis: Estimated probability density). Fourth panel: Average mEPSC traces. Fifth panel: Comparison of average mEPSC frequency (B, H1b/c^+/+;H1a^+/−: NR=6.1±0.4 Hz, n=17; DE=5.5±0.5 Hz, n=14; LE=5.7±0.6 Hz, n=16; ANOVA, F(2,44)=0.3479, p>0.7; C, H1b/c^+/−;H1a^−/−: NR=5.8±0.5 Hz, n=17; DE=5.1±0.5 Hz, n=14; LE=5.6±0.5 Hz, n=12; ANOVA, F(2,40)=0.5313, p>0.5; D, H1b/c^+/−;H1a^+/−: NR=4.6±0.6 Hz, n=12; DE=5.2±0.4 Hz, n=16; LE=5.8±0.6 Hz, n=14; ANOVA, F(2,39)=1.202, p>0.3). Also see Figure S4 and Table S1.

Figure 6.. mGluR5-H1a interaction is indispensable for homeostatic decrease in mEPSC amplitude in V1 L2/3 neurons

mEPSC recordings were done in V1 slices from juvenile mice (P23-32). (A,B) TSKI lack visual experience-dependent homeostatic plasticity. Left: Average mEPSC amplitude comparison (A, TSWT: NR=10.9±0.3 pA, n=18; DE=12.7±0.3 pA, n=12; LE=11.4±0.2 pA, n=16; ANOVA, F(2,43)=8.295, p<0.001; Newman-Keuls multiple comparison, ***p<0.001, **p<0.01; B, TSKI: NR=12.6±0.4 pA, n=18; DE=12.2±0.3 pA, n=16; LE=13.1±0.2 pA, n=15; ANOVA, F(2,46)=1.716, p>0.19). Middle: Estimated population density plot of mEPSC amplitudes (X-axis, Estimated probability density). Right: Average mEPSC traces. TSKIs showed a significant change in the variance of the estimated population density probability of mEPSC amplitudes (see Table S1). (C) Cumulative probability of mEPSC amplitudes of NR TSWT (black solid line) overlaid with that of NR TSKI (blue solid line) showing a multiplicative increase. Cumulative probability curve of TSWT mEPSC amplitudes multiplied by a scaling factor (1.16; black dashed line) overlaps with the TSKI curve. Mann-Whitney test: NR-TSWT vs. NR-TSKI, p<0.0001; NR-TSWT-scaled vs. NR-TSKI, p>0.9. Also see Figure S3. (D) No change in mEPSC frequency across groups. First panel: Average mEPSC frequency in TSWT (NR=3.8±0.3 Hz, n=18; DE=4.7±0.4 Hz, n=12; LE=4.0±0.5 Hz, n=16; ANOVA, F(2,43)=0.9937, p>0.37). Second panel: Estimated population density plot of mEPSC frequencies (X-axis: Estimated probability density). Third pane: Average mEPSC frequency in TSKI (NR=3.4±0.3 Hz, n=18; DE=3.4±0.3 Hz, n=16; LE =4.2±0.4 Hz, n=15; ANOVA, F(2,46)=1.798, p>0.17). Fourth panel: Estimated population density plot (X-axis, Estimated probability density). (E,F) FRKIs fail to weaken synaptic strength with LE. Left: Comparison of average mEPSC amplitude (E, FRWT: NR=11.5±0.4 pA, n=18; DE=13.0±0.4 pA, n=26; LE=11.8±0.4 pA, n=17; ANOVA, F(2,58)=4.600, p<0.05; Newman-Keuls multiple comparison *p < 0.05; F, FRKI: NR=13.6±0.8 pA, n=14; DE=13.2±0.5 pA, n=12; LE=13.8±0.6 pA, n=17; ANOVA, F(2,40)=0.1870, p>0.8). Middle: Estimated population density plot (X-axis: Estimated probability density). Right: Average mEPSC traces. (G) FRKIs display a non-multiplicative increase in basal mEPSC amplitudes compared to FRWT. Mann-Whitney: NR-FRWT (black solid line) vs. NR-FRKI (blue solid line), p<0.0001; NR-FRKI vs. NR-FRWT-scaled (black dashed line, scaling factor=1.19), p<0.0001. (H) No change in mEPSC frequency across groups. First panel: Comparison of average mEPSC frequency of FRWTs (NR=6.2±0.6 Hz, n=18; DE=5.6±0.3 Hz, n=26; LE=5.4±0.4 Hz, n=17; ANOVA, F(2,58)=0.9287, p>0.4). Second panel: Estimated population density plot (X-axis: Estimated probability density). Third pane: Average mEPSC frequency in FRKI (NR=6.1±0.6 Hz, n=14; DE=5.1±0.4 Hz, n=12; LE=6.0±0.4 Hz, n=17; ANOVA, F(2,40)=1.252, p>0.29). Fourth panel: Estimated population density plot (X-axis: Estimated probability density). Also see Table S1.

Figure 7.. Input-specific depression of IC inputs to L2/3 is dependent on mGluR5-H1a interaction.

(A,B) FRKI;L4Cre mice lack depression of IC inputs, but display aberrant increase in FF inputs, with LE. First panel: Schematics of the experiment. Second panel: Comparison of calculated average evoked Sr²⁺-mEPSC amplitude measured in L2/3 neurons of FRKI;L4Cre mice (P70-120) (A, IC inputs: NR=13.7±0.8 pA, n=9; DE=15.7±1.2 pA, n=12; LE=16.3±1.2 pA, n=10; ANOVA, F(2,28)=1.361, p>0.27; B, FF inputs: NR=15.5±0.6 pA, n=12; DE=14.3±1.3 pA, n=10; LE=19.6±1.7 pA, n=13; ANOVA, F(2,32)=4.647, p<0.05; Newman-Keuls multiple comparison *p<0.05). Estimated population density plot (X-axis, Estimated probability density) is shown next to the bar graphs. Bottom: Average Sr²⁺-mEPSC traces. (C,D) NMDAR activity is involved in the aberrant increase of FF input in FRKI;L4Cre mice. Systemic injection of NMDAR antagonist CPP (10 mg/kg, i.p.) 30 min prior to LE does not alter the strength of IC inputs (C, LE+Saline=17.2±1.5 pA, n=12; LE+CPP=18.3±1.2 pA, n=10; unpaired t-test, p=0.560), but prevents the aberrant increase in Sr²⁺-mEPSC amplitudes for FF inputs (D, LE+Saline =20.9±1.1 pA, n=10; LE+CPP=17.4±0.9 pA, n=10; unpaired t-test, *p=0.026). Estimated population density plots (X-axis: Estimated probability density) are shown next to each bar graph. Bottom: average Sr²⁺-mEPSC traces. Also see Table S1.