All experiments followed the European Convention for the Protection of Vertebrate Animals used for Experimental and other Scientific Purposes 1986 86/609/EEC and were approved by the Ethical Committee of the Institute of Higher Nervous Activity and Neurophysiology, Russian Academy of Sciences (IHNA RAS). The mice were purchased from the Nursery for laboratory animals of the Branch of the Institute of Bioorganic Chemistry of the Russian Academy of Sciences in Pushchino. The mice were maintained in a temperature-controlled vivarium (22 ± 2°C) under a 12-h light/dark cycle (lights on at 08.00 h) with food and water ad libitum. All efforts were made to minimize animal suffering and to reduce the number of animals used.

Horizontal brain slices (300 μm thick) containing the ventral hippocampus and entorhinal cortex were prepared from the brains of 25–35-day-old C57Bl/6 female and male mice killed by decapitation. The slicing chamber contained an oxygenated ice-cold solution (modified from Dugue et al., 2005) composed of the following (in mM) K-gluconate, 140; N-(2-hydroxyethyl) piperazine-N′-ethanesulfonic acid (HEPES), 10; Na-gluconate, 15; ethylene glycol-bis(2-aminoethyl)-N,N,N′,N′-tetraacetic acid (EGTA), 0.2; and NaCl, 4 (pH 7.2 with KOH). Brain slices were cut using a Vibratome (Leica VT1000S, Germany). Slices were incubated for at least 40 min at 35°C before being stored at room temperature in artificial CSF (ACSF) containing the following (in mM): NaCl, 125; NaHCO₃, 25; KCl, 2.5; NaH₂PO₄, 1.25; MgCl₂, 3.9; CaCl₂, 1; and glucose, 25; bubbled with 95% O₂, and 5% CO₂.

Electrophysiological recording was performed in an acrylic glass perfusion chamber (Luigs and Neumann, Germany) with the bath temperature kept at 30 ± 2°C and perfused at a constant rate of 3 ml/min. Patch electrodes (resistance 4–5 MΩ) were pulled from borosilicate capillary glass (Narishige PC-100 Puller, Japan) and were filled with either a polyamine-free or a polyamine-containing solution. The polyamine-free solution consisted of the following (in mM): Cs-gluconate, 110; CsCl, 30; HEPES, 10; NaCl, 8; EGTA, 0.2; MgATP, 4; Na₃GTP, 0.3; and phosphocreatine, 10 (pH 7.3 with CsOH) osmolarity ~290 mOsm. The polyamine-containing solution was identical except for the addition of 10 μM spermine.

CA1 pyramidal cells were identified visually using an Olympus microscope fitted with infrared differential interference contrast optics (Olympus BX51WI). Whole-cell recordings from these neurons were made in a voltage-clamp mode using the ELC-03XS amplifier (NPI Electronic, Tamm, Germany) and Clampex software (Axoclamp, Molecular Devices). Cells were held at −70 mV. Cells with unhealthy morphology and resting membrane potential above −50 mV (before correction for the liquid junction potential) were excluded from the experiments. To evoke synaptic current, glass electrodes filled with ACSF were placed in the dendritic region of stratum radiatum and stratum oriens, ~50–100 μm from the cell body, to stimulate the inputs at interstimulus intervals of 6 s. Inhibitory synaptic transmission was blocked during recordings by adding 50 μM picrotoxin to the perfusion ACSF. In all the experiments except AMPA/NMDA ratio measurements, the NMDA-mediated component was blocked by adding 50 μM APV to the ACSF. For the experiments with NOS inhibition, slices were incubated for 40–120 min in L-NAME, 15 min in 3-bromo-7-nitroindazole or carboxy-PTIO before being placed in the perfusion chamber. Whole-cell recordings typically started 5–10 min after break-in, when the balance between intracellular milieu and patch solution was established and a steady-state current was reached, except the experiments with GluR2-lacking AMPA receptor inhibition. The stimulation intensity was adjusted to produce an EPSC with an amplitude of ~50 pA at the beginning of each recording. The experiments were not started if there was an unstable baseline. Series resistance was monitored, and data from cells in which series resistance varied by >15% during recording were discarded from the analysis. In all the experiments, the command voltage was corrected for the liquid junction potential (−10 mV).

ODQ from Sigma-Aldrich was prepared as a 25-mM stock solution in DMSO and diluted down to achieve a final bath concentration of 30 μM. 3-Bromo-7-nitroindazole (3-Br-7-ni; Enzo Life Sciences) was dissolved as a 100-mM stock in DMSO and diluted down to achieve a final bath concentration of 50 μM. Nω-Nitro-L-arginine methyl ester hydrochloride (L-NAME; Sigma-Aldrich) was prepared as a 200-mM stock in milli-Q water diluted down to achieve a final bath concentration of 200 μM. Carboxy-PTIO (2-(4-Carboxyphenyl)-4,4,5,5-tetramethylimidazoline-1-oxyl-3-oxide potassium salt) from Enzo Life Sciences was prepared as a 100-mM stock in DMSO and diluted down to achieve a final bath concentration of 50 μM. Naspm (1-naphthylacetyl spermine) was prepared as a 100-mM stock in milli-Q water and diluted down to achieve a final bath concentration of 200 μM (Tocris Bioscience). Picrotoxin from Sigma-Aldrich was prepared as a 100-mM stock in DMSO and diluted down to achieve a final bath concentration of 50 μM. DL-2-Amino-5-phosphonopentanoic acid sodium salt (APV) from Tocris was prepared as a 100-mM stock in milli-Q water and diluted down to achieve a final bath concentration of 50 μM. CNQX disodium salt from Tocris Bioscience was prepared as a 100-mM stock in milli-Q water and diluted down to achieve a final bath concentration of 50 μM.

Results are presented as mean ± standard error (S.E.M.) of n cells. All statistical tests were performed using SigmaPlot 11.0 (Systat Software Inc., USA). For pairwise comparisons (Figures 1B,C), one-way ANOVA was used. For multiple comparisons (Figures 1F, 3A,B, 4C,D, 5C,D, 6, Supplementary Figure 1), we used two-way ANOVA. A significant main effect or interaction was followed by post-hoc comparison using Multiple Comparisons vs. Control Group (Holm–Sidak test); for between-subject analysis, the untreated cells were scored as “Control Group.” For Figures 2, 5A,B, and Supplementary Figure 2, we used two-way repeated-measures ANOVA. A significant main effect or interaction was followed by post-hoc comparisons using Multiple Comparisons vs. Control Group (Holm–Sidak Test). For between-subject analysis, the control cells were scored as “Control Group,” and for within-subject analysis all measurements were compared to the first EPSC in the train. A probability level of 0.05 or less was considered statistically significant (*p ≤ 0.05, **p ≤ 0.005).

First, to evaluate the contribution of GluA2-lacking AMPARs to currents in apical and basal compartments of CA1 pyramidal neurons, we measured the RI by stimulating apical and basal dendrites with glass electrodes located as shown in Figure 1A. The rectification index reflects inward rectification of GluA2-lacking AMPARs and is scored as the ratio of the current amplitude measured at −70 mV and +35 mV (EPSC_{–70 mV}/EPSC_{+35 mV}). We found that the rectification index in responses to stimulation of apical dendrites (RI = 3.4 ± 0.2) is significantly higher than in responses to stimulation of basal dendrites (RI = 2.5 ± 0.2; p = 0.005, one-way ANOVA, n = 14, Figure 1C). The current–voltage (IV) characteristics for each of the inputs supported the difference revealed by RI measurement; the apical IV demonstrated significantly higher inward rectification (p = 0.035, one-way ANOVA, n = 7, Figure 1B). The most common approach to detecting the contribution of GluA2-lacking AMPARs to synaptic transmission is to test sensitivity to CP-AMPAR antagonists, for example, 1-naphthylacetyl spermine trihydrochloride (Naspm). Therefore, we tested whether the responses to stimulation of apical and basal inputs are sensitive to extracellular Naspm (200 μM) application (Figures 1D–G). To evaluate the degree of amplitude reduction, we averaged the values from each cell in the 5-min segment at the end of the recordings (Figure 1F). We observed a decrease in AMPAR currents in both of the inputs (73% ± 4 of the baseline in str. radiatum and 83% ± 6.5 in str. oriens, n = 10), which may indicate the presence of CP-AMPARs in the basal and apical dendrites of CA1 pyramidal neurons. However, comparison to the untreated cells (n = 6) showed significant difference only in apical (73% ± 4 vs. 109% ± 7, p = 0.001, two-way ANOVA) but not in basal inputs (83% ± 6.5 vs. 94% ± 4, p = 0.354, two-way ANOVA). Even at physiological resting membrane potentials (~−70 mV), a substantial portion of GluA2-lacking channels could still be blocked by polyamines during single unitary EPSCs. Thus, we used a spermine-free intracellular solution in these experiments. As a result of the endogenous polyamine washout from the dendrites, the amplitude of responses increased up to 20 min. To avoid subsequent spike generation, we added 30 μM QX-314 to the intracellular solution. Due to the different morphology of apical and basal dendrites (Benavides-Piccione et al., 2020), the response amplitude at the basal inputs tends to stabilize earlier than at the apical inputs.

In addition, Rozov and colleagues have shown previously that application of several high-frequency stimuli to the input with CP-AMPARs causes relief of the polyamine block which, in turn, results in polyamine-dependent facilitation (PdF; Rozov and Burnashev, 1999; Rozov et al., 2018). We applied a similar protocol to the apical and basal inputs of CA1 pyramidal cells and found that this type of short-term plasticity was characteristic for synapses of these cells. We applied four stimuli with an interstimulus interval of 33 ms to each of the inputs at the beginning of the recording and 20 min later. We used spermine-free and spermine-containing intracellular solutions in order to test whether polyamine washout would affect the EPSCs. For experiments with polyamine washout, we adjusted the stimuli strength after 20 min to avoid spikes due to potential space clamp problems. Figures 2A,B demonstrate the results of these experiments. When we used a spermine-free intracellular solution, the 3rd and 4th EPSCs in the train were significantly lower 20 min after polyamine washout than at the beginning of the recording at apical dendrites (p = 0.037 for the 3rd EPSCs and p = 0.014 for the 4th EPSCs, two-way RM ANOVA, n = 8), and we observed the same tendency at basal dendrites (p = 0.117 for the 3^d EPSCs and p = 0.066 for the 4th EPSCs, two-way RM ANOVA, n = 8), whereas with the presence of polyamines in the patch pipette we did not observe such differences (apical: p = 0.46 for the 3rd EPSCs and p = 0.101 for the 4th EPSCs; basal: p = 0.47 for the 3rd EPSCs and p = 0.26 for the 4th EPSCs, two-way RM ANOVA, n = 6). Thus, we showed polyamine-dependent facilitation at the synapses of CA1 pyramidal cells, which indicates the presence of CP-AMPARs at both apical and basal dendrites of these cells. However, experiments measuring the current–voltage characteristics of these inputs revealed that the CP-AMPA receptors contribute more to synaptic transmission at apical dendrites of CA1 pyramidal neurons than basal dendrites.

Since nitric oxide can affect the incorporation of CP-AMPARs into the postsynaptic membrane of cells, we tested whether the blockade of its synthesis by various NO synthase inhibitors affects the characteristics of the apical and basal synapses of CA1 pyramidal cells. Figure 3B shows the rectification indices of the apical and basal inputs under the NO-synthase inhibitor L-NAME (200 μM), the selective inhibitor of neuronal NO-synthase 3-bromo-7-nitroindazole (50 μM), and the NO scavenger carboxy-PTIO (50 μM). The latter two inhibitors were dissolved in DMSO, and we also tested the effect on RI of DMSO alone (25 μl). A decrease in the intracellular concentration of nitric oxide significantly reduced RIs at the apical dendrites compared to control cells (DMSO: 3.1 ± 0.1, p = 0.321, n = 8; L-NAME: 2.6 ± 0.2, p = 0.025, n = 8; 3-br-7-ni: 2.4 ± 0.3, p = 0.007, n = 10; PTIO: 2.4 ± 0.3, p = 0.002, n = 7, two-way ANOVA), but not basal dendrites (DMSO: 2.1 ± 0.2, p = 0.28, n = 8; L-NAME: 2.6 ± 0.4, p = 0.78, n = 8; 3-br-7-ni: 2.1 ± 0.2, p = 0.24, n = 10; PTIO: 1.8 ± 0.1, p = 0.07, n = 7, two-way ANOVA), in various ways, thus leveling the significant difference between these two inputs. The current–voltage characteristics of the studied inputs under NOS blockade by L-NAME demonstrated a loss of inward rectification at synaptic inputs in apical but not basal dendrites (p = 0.031 for apical, p = 0.4 for basal, two-way ANOVA, n = 7, Figure 3A). We also compared the AMPA/NMDA ratio at apical and basal dendrites with or without NOS inhibition (Figure 5D). In control cells, this ratio was significantly higher at apical dendrites than at basal dendrites (6.7 ± 0.6 vs. 4.9 ± 0.6, n = 5, p = 0.049, two-way ANOVA), and this difference disappeared after NOS inhibition (4.3 ± 0.8 vs. 3.9 ± 0.8, n = 5, p = 0.788, two-way ANOVA), supporting the results of the experiments with RI measurement.

Additionally, NOS inhibition prevented the drop in the current amplitude during CP-AMPAR blockade by Naspm in str. radiatum (109% ± 7.3 vs. 3% ± 4, n = 10, p = 0.001, two-way ANOVA, Figure 4C); however, in str. oriens we did not find a significant difference in the last 5 min of recording (89.4% ± 3.2 vs. 83% ± 6.5, n = 10, p = 0.264, two-way ANOVA, Figure 4D). Interestingly, the increase in the current amplitude at the beginning of the recording, which is associated with polyamine washout in both str. radiatum and str. Oriens, persisted after incubation in L-NAME (Figures 4A,B). This might be due to either specific mechanisms associated with polyamine washout or the possibility of NO-dependent regulation of CP-AMPAR sensitivity to polyamines. In the latter case, polyamines could still block the receptor pore but the blockade is more easily relieved which causes smaller EPSC increase and faster baseline stabilization.

Next, we tested how the decrease of nitric oxide in the cell would affect polyamine-dependent facilitation at the studied inputs in the presence of spermine in the recording pipette (Figures 5A,B). NOS inhibition significantly affects the 4th EPSCs at apical dendrites (p = 0.006, two-way RM ANOVA, n = 7), but not at basal dendrites (p = 0.499, two-way RM ANOVA, n = 7). PdF recording was performed at the fifth minute of recording after a stable baseline was reached. Considering the wide range of nitric oxide action in the presynapse of cells (Hardingham et al., 2013), we tested whether the discovered effect was due to the presynaptic effect of nitric oxide. One of the possible approaches for evaluation of the presynaptic contribution to synaptic transmission is paired-pulse ratio measurement (Schulz et al., 1994; Christie and Jahr, 2006). Recording of paired-pulse ratio at the synapses of CA1 pyramidal cells during NOS inhibition did not reveal significant differences in these cells compared to control cells (control: 1.6 ± 0.1 in str. radiatum, and 1.4 ± 0.06 in str. oriens, n = 8; L-NAME: 1.5 ± 0.1 in str. radiatum, and 1.4 ± 0.07 in str. oriens, n = 7; 3-bromo-7-ni: 1.4 ± 0.06 in str. radiatum, and 1.4 ± 0.1 in str. oriens, n = 7). Thus, the results suggest that the disappearance of PdF in str. radiatum after inhibition of nitric oxide synthesis was associated with the interaction of nitric oxide with CP-AMPARs.

We examined whether the sGC-dependent pathway is involved in this interaction by treating slices with the sGC inhibitor ODQ (30 μM). We measured RIs under this condition and found that sGC inhibition did not affect the RIs of neither apical (RI = 3.5 ± 0.2, n = 5) nor basal (RI = 2.5 ± 0.3, n = 5) inputs (Figure 6, p = 0.027, two-way ANOVA). Next, we blocked NOS and sGC simultaneously and found that treating with L-NAME also leveled the RIs of apical (RI = 3.3 ± 0.3, n = 5) and basal (RI = 3 ± 0.4) inputs, as in the cells treated with the NOS inhibitors alone (p = 0.680, two-way ANOVA).