LTPE→I in stratum radiatum is dependent on the activation of NMDA receptor and astrocytic metabolism.

(A) Left: superimposed representative averaged EPSPs recorded 10 min before (dark traces) and 35-45 min after (red traces) LTP induction; Right: superimposed representative averaged EPSPs recorded 10 min before (dark traces) and 35-45 min after (red traces) LTP induction while in the presence of NMDA receptor antagonist D-AP5.

(B) Normalized slope before and after the TBS stimulation protocol in control conditions and in the presence of NMDA receptor antagonist D-AP5.

(C) The summary data measured 35-45 min after LTP induction (Control: 220.2±33.94, n=10 slices from 5 mice; D-AP5: 105.2±4.404, n=6 slices from 3 mice; t=2.580 with 14 degrees of freedom, p=0.0218, two-tailed unpaired t-test).

(D) Left: superimposed representative averaged EPSPs recorded 10 min before (dark traces) and 35-45 min after (red traces) LTP induction; Right: superimposed representative averaged EPSPs recorded 10 min before (dark traces) and 35-45 min after (red traces) LTP induction while astrocytic metabolism was disrupted.

(E) Normalized slope before and after the TBS stimulation protocol in control conditions and in the presence of astrocytic metabolism inhibitor FAC.

(F) The summary data measured 35-45 min after LTP induction (Control: 193.5±21.84, n=11 slices from 5 mice; FAC: 103.8±7.278, n=6 slices from 6 mice; t=2.943 with 15 degrees of freedom, p=0.0101, two-tailed unpaired t-test).

Astrocyte Ca2+ is involved in the induction of LTPE→I.

(A) Representative image of the astrocytic syncytium illustrating that Ca2+ clamping reagents and Alexa Fluor 594 diffuse through gap junctions surround the patched interneuron.

(B) Left: superimposed representative averaged EPSPs recorded 10 min before (dark traces) and 35-45 min after (red traces) LTP induction; Right: superimposed representative averaged EPSPs recorded 10 min before (dark traces) and 35-45 min after (red traces) LTP induction when Ca2+ concentration was clamped.

(C) Normalized slope before and after the TBS stimulation protocol in control conditions and when Ca2+ concentration was clamped.

(D) The summary data measured 35-45 min after LTP induction (ACSF: 162.7±22.21, n=6 slices from 3 mice; Ca2+ clamp: 111.1±7.323, n=7 slices from 4 mice; Mann-Whitney U Statistic= 3.000, p=0.008, Mann-Whitney Rank Sum Test; ACSF: 197.5±33.83, n=6 slices from 4 mice; Ca2+ clamp+D-serine: 195.1±31.75, n=7 slices from 4 mice; Mann-Whitney U Statistic= 19.000, p=0.836, Mann-Whitney Rank Sum Test; ACSF: 172.2±19.49, n=8 slices from 4 mice; patch: 189.9±38.11, n=6 slices from 3 mice; Mann-Whitney U Statistic= 22.000, p=0.852, Mann-Whitney Rank Sum Test).

Activation of astrocytic CB1 receptors causes an increase in astrocytic Ca2+ signals and is involved in the induction of LTPE→I.

(A) Representative images of a GCaMP6f+ astrocyte before (left) and after (right) theta burst stimulation (TBS).

(B) Kymographs and △F/F traces of cells with Ca2+ signals evoked by the activation of Schaffer collateral with a TBS in GCaMP6f+ astrocyte.

(C) Representative images of a GCaMP6f+ astrocyte before (left) and after (right) theta burst stimulation (TBS) when in the presence of CB1 receptor blocker AM251. .

(D) Kymographs and △F/F traces of cells with Ca2+ signals evoked by the activation of Schaffer collateral with a TBS in GCaMP6f+ astrocyte when in the presence of CB1 receptor blocker AM251.

(E) Kymographs and △F/F traces of cells with Ca2+ signals evoked by the activation of Schaffer collateral with a TBS in GCaMP6f+ astrocyte when in the presence of α1-adrenoceptor blocker terazosin.

(F) Summary plots for experiments in (B)-(E) illustrating that Ca2+ signals elicited by TBS is mediated by the activation of CB1 receptors (Control pre: 0.3275±0.1083 △ F/F, Control post: 1.313±0.2483 △F/F, n=45 from 20 cells of 5 mice; z= 3.369, p<0.001, Wilcoxon Signed Rank Test; AM251 pre: 0.3541±0.07219 △F/F, AM251 post: 0.3418±0.07240 △F/F, n=28 from 13 cells from 5 mice; z= −0,797, p=0.432, Wilcoxon Signed Rank Test; Terazosin pre: 0.5725±0.1749 △F/F, Terazosin post: 1.73±0.4661 △F/F, n=25 from 11 cells from 4 mice; z=3.215, p=0.001, Wilcoxon Signed Rank Test).

(G) Upon: superimposed representative averaged EPSPs recorded 10 min before (dark traces) and 35-45 min after (red traces) LTP induction; below: superimposed representative averaged EPSPs recorded 10 min before (dark traces) and 35-45 min after (red traces) LTP induction while in the presence of CB1 receptor antagonist AM251.

(H) Normalized slope before and after the TBS stimulation protocol in control conditions and in the presence of CB1 receptor antagonist AM251.

(I) The summary data measured 35-45 min after LTP induction (Control: 180.2±13.76, n=12 slices from 5 mice; AM251: 118.9±7.406, n=10 slices from 4 mice; t=3.697 with 20 degrees of freedom, p=0.0014, two-tailed unpaired t-test; Control: 169.3±16.89, n=12 slices from 6 mice; AM251+D-serine: 177.4±14.27, n=10 slices from 5 mice; t=0.3561 with 20 degrees of freedom, p=0.7255, two-tailed unpaired t-test).

D-serine release from astrocyte potentiates NMDAR-mediated responses

(A) Example traces of NMDAR-mediated EPSPs before and after bath application of D-serine in stratum radiatum interneurons.

(B) Normalized amplitude before and after add D-serine in stratum radiatum interneurons

(C) The percentage of potentiation in NMDAR-mediated responses (percentage of potention: 155.523 ± 11.234, n=6 slices from 3 mice; t=-4.942 with 6 degrees of freedom, p=0.00260, two-tailed paired t-test).

(D) Left: example traces of NMDAR-mediated EPSPs before and after the TBS stimulation protocol in stratum radiatum interneurons .Right: example traces of NMDAR-mediated EPSPs before and after the TBS stimulation protocol in stratum radiatum interneurons when in the presence of D-serine.

(E) Normalized amplitude before and after the TBS stimulation protocol in stratum radiatum interneurons during ASCF and after application D-serine.

(F) The summary data measured the percentage of potentiation in NMDAR-mediated responses (Control: 155.5±11.23, n=7 slices from 3 mice; D-serine: 121.1±6.917, n=8 slices from 4 mice; FAC: 112.8±4.931, n=7 slices from 3 mice; p=0.0036, F(2, 19)=7.653, ANOVA with Dunnett’s comparison) and AMPAR-mediated responses after TBS(Control: 132.357±12.305, n=5 slices from 3 mice; D-serine: 136.012±10.611, n=8 slices from 4 mice; t=-0.224 with 10 degrees of freedom, p=0.827, two-tailed unpaired t-test).

Astrocytic activation induces De Novo LTPE→I.

(A) Representative images of a GCaMP6f+ astrocyte before (left) and after (right) CNO application.

(B) Example confocal images which showing co-expression of GCaMP6f and mCherry constructs confirm that representative cell in (A) is co-expression hM3Dq.

(C) Kymographs and ΔF/F traces of cells with Ca2+ signals evoked by bath application of CNO in GCaMP6f+ and mCherry+ astrocyte.

(D) Summary plot illustrating that CNO is effective to activate astrocytes (Baseline: 0.6167±0.07762 ΔF/F, CNO: 1.456±0.1504 ΔF/F, n=36 from 19 cells of 3 mice; t=6.411 with 35 degrees of freedom, p<0.0001, two-tailed paired t-test).

(E) Upon: superimposed representative averaged EPSPs recorded 10 min before (dark traces) and 35-45 min after (red traces) CNO application when astrocyte only express mCherry; below: superimposed representative averaged EPSPs recorded 10 min before (dark traces) and 35-45 min after (red traces) CNO applicaion when hM3Dq expressed astrocytes were activated by CNO

(F) Relative EPSP slope before and after CNO application when astrocytes only express mCherry and when astrocytes express hM3Dq.

(G) The summary data measured 35-45 min after LTP induction (mCherry: 100.6±5.651, n=5 slices from 3 mice; hM3Dq: 172.8±10.28, n=7 slices from 4 mice; t=5.474 with 10 degrees of freedom, p=0.0003, two-tailed unpaired t-test).

(H) Upon: superimposed representative averaged EPSPs recorded 10 min before (dark traces) and 35-40 min after (red traces) CNO application; middle: superimposed representative averaged EPSPs recorded 10 min before (dark traces) and 35-40 min after (red traces) CNO application when in the presence NMDA receptor blocker D-AP5; below: superimposed representative averaged EPSPs recorded 10 min before (dark traces) and 35-40 min after (red traces) CNO application when the glycine sites of NMDA receptors were saturated by D-serine.

(I) Relative EPSP slope before and after CNO application when astrocytes were activated by CNO, and when in the presence of NMDA receptor antagonist D-AP5 and in the presence of D-serine.

(J) The summary data measured 35-45 min after LTP induction (hM3Dq+CNO: 197.1±27.57, n=6 slices from 3 mice; hM3Dq+CNO+D-AP5: 101.3±6.886, n=6 slices from 3 mice; hM3Dq+CNO+D-serine: 104.2+8.166, n=7 slices from 4 mice; p=0.0011, F(2, 16)=10.80, ANOVA with Dunnett’s comparison).

Impaired hippocampus-dependent long-term memory in knockdown γCaMKII mice

(A) Upon: superimposed representative averaged EPSPs recorded 10 min before (dark traces) and 35-45 min after (red traces) LTP induction in WT group; middle: superimposed representative averaged EPSPs recorded 10 min before (dark traces) and 35-45 min after (red traces) LTP induction in AAV-mDLx-scramble shRNA group; below: superimposed representative averaged EPSPs recorded 10 min before (dark traces) and 35-45 min after (red traces) LTP induction in AAV-mDLx-γCaMKII shRNA group.

(B) Normalized slope before and after the TBS stimulation protocol in control, scramble shRNA and γCaMKII shRNA group.

(C) The summary data measured 35-45 min after LTP induction (Control: 196±21.56, n=12 slices from 5 mice; γCaMKII shRNA: 121.3±10.64, n=14 slices from 6 mice; scramble shRNA: 199.6±21.42, n=11 slices from 5 mice; p=0.0040, F(2, 34)=6.527, ANOVA with Dunnett’s comparison).

(D) Schematic illustration of the experimental design for contextual and cued fear conditioning test, indicating the time line of the experimental manipulations.

(E) The freezing responses was measured in context A before training in WT, bilaterally injected γCaMKII shRNA and scramble shRNA mice (Control: 17.5±2.975, n=10 mice; γCaMKII shRNA: 20.9±2.275, n=15 mice; scramble shRNA: 20.29±2.341, n=15 mice; p=0.6378, F(2, 37)=0.4553, ANOVA with Dunnett’s comparison).

(F) The freezing responses was measured 24 h after training in WT, bilaterally injected γCaMKII shRNA and scramble shRNA mice (Control: 52.4±6.695, n=10 mice; γCaMKII shRNA: 30.64±4.574, n=15 mice; scramble shRNA: 49.27±4.899, n=15 mice; p=0.0105, F(2, 37)=5.170, ANOVA with Dunnett’s comparison).

(G) The freezing response was measured 28 h after training in WT, bilaterally injected γCaMKII shRNA and scramble shRNA mice (Pre Control: 19.5±2.171, n=10 mice; Pre γCaMKII shRNA: 25.01±3.016, n=15 mice; Pre scramble shRNA: 21.94±3.519, n=15 mice; p=0.4986, F(2, 37)=0.7093, ANOVA with Dunnett’s comparison; Tone 1 Control: 57.9±5.153, n=10 mice; Tone 1 γCaMKII shRNA: 56.09±3.319, n=15 mice; Tone 1 scramble shRNA: 55.32±3.368, n=15 mice; p=0.9002, F(2, 37)=0.1055, ANOVA with Dunnett’s comparison; Tone 2 Control: 59.75±3.825, n=10 mice; Tone 2 γCaMKII shRNA: 60.95±3.482, n=15 mice; Tone 2 scramble shRNA: 60.34±4.485, n=15 mice; p=0.9802, F(2, 37)=0.01996, ANOVA with Dunnett’s comparison).

EGFP is expressed in GABAergic interneurons in the stratum radiatum of hippocampus.

(A) Schematic illustrating the procedure of AAV2/9 microinjections into the stratum radiatum of the hippocampus.

(B) Confocal images showing EGFP (green) and GAD67 (red, white arrow) expressing cells in the stratum radiatum of the hippocampus.

(C and D) mDLx::EGFP was expressed in >98% of interneurons in stratum radiatum of the hippocampus, with >98% specificity.

EGFP+ interneurons have normal sEPSC, excitability and PPR.

(A) Relationship of injected current to number of APs in postulated interneurons and EGFP+ interneurons. Inset: traces of membrane responses to current injection.

(B) The number of APs at up state membrane potentials (Vm) values (Control: 14.62±1.474, n=13 slices from 6 mice; EGFP: 15.61±1.458, n=18 slices from 6 mice;, t=0.4684 with 29 degrees of freedom, p=0.643, two-tailed unpaired t-test).

(C) Interneuron resting membrane potentials (Control: −63.85±1.779 mV, n=13 slices from 6 mice; EGFP: −61.44±1.539 mV, n=18 slices from 6 mice; t=1.018 with 29 degrees of freedom, p=0.317, two-tailed unpaired t-test).

(D) Left: example traces of sEPSCs measured in stratum radiatum of hippocampal postulated interneurons of WT mice; Right: example traces of sEPSCs measured in stratum radiatum of hippocampal EGFP+ interneurons of virus injected mice.

(E, F) Cumulative distribution plots and summary of sEPSC amplitude and frequency in postulated interneurons and EGFP+ interneurons (Amplitude: 16.89±1.846 pA of control, 18.28±2.003 of EGFP, n=10 slices from 6 mice, t=0.5125 with 18 degrees of freedom, p=0.6146, two-tailed unpaired t-test; Frequency: 12.41±3.058 Hz of control, 13.78±2.601 Hz of EGFP, n=10 slices from 6 mice, t=0.3405 with 18 degrees of freedom, p=0.7374, two-tailed unpaired t-test).

(G) Example paired-pulse traces measured in postulated interneurons and EGFP+ interneurons.

(H) Summary of paired-pulse ratio (Control: 1.602±0.1443, n=8 slices from 3 mice; EGFP: 1.677±0.1794, n=8 slices from 3 mice; t=0.3267 with 14 degrees of freedom, p=0.7488, two-tailed unpaired t-test).

Stratum radiatum interneurons show linear rectifying AMPARs and a big NMDAR-mediated component.

(A) Current-voltage (I-V) relation of AMPAR-mediated EPSCs in stratum radiatum interneurons. Inset: averaged EPSCs traces at −90, −60, −30, 0 and 60 mV, showing the times at which the two components were measured.

(B) AMPA rectification index (EPSC amplitude at 60 mV / EPSC amplitude at −60 mV) in eight interneurons from stratum radiatum indicates linear rectifying AMPARs (AMPA retification index: 0.8959±0.09362).

(C) I-V relation for the NMDAR-mediated EPSCs in stratum radiatum interneurons.

(D) NMDA / AMPA ratio (NMDAR-mediated EPSC amplitude at 60 mV / AMPAR-mediated EPSC amplitude at −60 mV) in eight cells from stratum radiatum (NMDA/AMPA ratio: 0.8664±0.2075).

LTPE→I in stratum oriens is not dependent on astrocytic metabolism and the activation of NMDA receptor.

(A) Left: superimposed representative averaged EPSPs recorded 10 min before (dark traces) and 35-45 min after (red traces) LTP induction; Right: superimposed representative averaged EPSPs recorded 10 min before (dark traces) and 35-45 min after (red traces) LTP induction when astrocytic metabolism was disrupted.

(B) Normalized slope before and after the TBS stimulation protocol in control conditions and in the presence of astrocytic metabolism inhibitor FAC.

(C) The summary data measured 35-45 min after LTP induction (Control: 233±52.74, n=5 slices from 3 mice, FAC: 233.5±33.06, n=6 slices from 6 mice; D-AP5:220.9±38.94, n=8 slices from 4 mice; p=0.96, F(2, 16)=0.03277, ANOVA with Dunnett’s comparison).

Stratum oriens interneurons show inwardly rectifying AMPARs and a negligible NMDAR-mediated component.

(A) Current-voltage (I-V) relation of AMPAR-mediated EPSCs in stratum oriens interneurons. Inset: averaged EPSCs traces at −90, −60, −30, 0 and 60 mV, showing the times at which the two components were measured.

(B) AMPA rectification index (EPSC amplitude at 60 mV / EPSC amplitude at −60 mV) in six interneurons from stratum oriens indicate inwardly rectifying AMPARs (AMPA retification index: 0.2406±0.05594).

(C) I-V relation for the NMDAR-mediated EPSCs in stratum oriens interneurons

(D) NMDA /AMPA ratio (NMDAR-mediated EPSC amplitude at 60 mV / AMPAR-mediated EPSC amplitude at −60 mV) in six cells from stratum oriens (NMDA/AMPA ratio: 0.1132±0.05370).

GCaMP6f was expressed in astrocytes.

(A and B) Representative immunohistochemistry images showing colocalization of GCaMP6f and GFAP (A), but not NeuN (B).

(C and D) gfaABC1D::GCaMP6f was expressed in >70% of astrocyte in stratum radiatum of the hippocampus, with >99% specificity.

(E) Quantification of the percentage of cells expressing NeuN that also expressed GCaMP6f.

hM3Dq was expressed in astrocytes.

(A) hM3Dq(mCherry) colocalization with an GCaMP6f in the stratum radiatum of the hippocampus.

EGFP-γCaMKII shRNA is expressed in GABAergic interneurons in the stratum radiatum of hippocampus.

(A) Confocal images showing EGFP γCaMKII shRNA (green) and GAD67 (red, white arrows) expressing cells in the stratum radiatum of the hippocampus. These images indicate that the γCaMKII shRNA, which is driven by an interneuron specific promoter (mDLx), is expressed specifically in interneurons.

knockdown γCaMKII from interneurons have no effect on their sEPSC, excitability and PPR

(A) Relationship of injected current to number of APs in postulated interneurons, expressing scramble shRNA and γCaMKII shRNA interneurons. Inset: traces of membrane responses to current injection.

(B) The number of APs at up state membrane potentials (Vm) values (Control: 15.95±1.904, n=10 slices from 4 mice; scramble shRNA: 15.00±1.430, n=10 slices from 4 mice; γCaMKII shRNA: 14.90±1.410, n=10 slice from 5 mice; p=0.4986, F(2, 27)=0.4756, ANOVA with Dunnett’s comparison).

(C) Interneuron resting membrane potentials (Control: −66.03±0.8127, n=10 slices from 4 mice; scramble shRNA: −64.44±1.398, n=10 slices from 4 mice; γCaMKII shRNA: −67.23±0.9803, n=10 slice from 5 mice; p=0.2128, F(2, 27)=1.639, ANOVA with Dunnett’s comparison).|

(D) Left: example traces of sEPSCs measured in stratum radiatum of hippocampal postulated interneurons of WT mice; Right: example traces of sEPSCs measured in stratum radiatum of hippocampal EGFP+ interneurons of AAV-mDLx-shRNA γCaMKII expressed mice.

(E, F) Cumulative distribution plots and summary of sEPSC amplitude and frequency in postulated interneurons, expressing scramble shRNA and γCaMKII shRNA interneurons (Frequency: 11.62±2.328 Hz of control, n=6 slices from 3 mice, 11.69±2.462 Hz of scramble shRNA, n=8 slices from 4 mice, 11.12±2.015 Hz of γCaMKII shRNA, n=8 slices from 4 mice, p=0.9805, F(2, 19)=0.01969, ANOVA with Dunnett’s comparison; Amplitude: 15.23±1.846 pA of control, n=6 slices from 3 mice, 14.57±1.382 of scramble shRNA, n=8 slices from 4 mice,16.23±1.821 of γCaMKII shRNA, n=8 slices from 4 mice, p=0.8459, F(2, 19)=0.1688, ANOVA with Dunnett’s comparison).

(G) Example paired-pulse traces measured in postulated interneurons of WT mice and EGFP+ interneurons of AAV-mDLx-shRNA γCaMKII expressed mice.

(H) Summary of paired-pulse ratio (Control: 1.679±0.1190 n=9 slices from 3 mice; scramble shRNA: 1.688±0.1405, n=7 slices from 3 mice; γCaMKII shRNA: 1.622±0.1477, n=7 slice from 3 mice; p=0.9361, F(2, 20)=0.06628, ANOVA with Dunnett’s comparison).