Abstract
Inhibitory modulation of glutamatergic information processing is a prerequisite for proper network function. Among the many groups of interneurons (INs), somatostatin-expressing interneurons (SOM-INs) play an important role in the maintenance of physiological brain activity. We have previously shown that somatostatin (SOM) causes a reduction in pyramidal cell (PC) excitability. However, the mechanisms of action of the peptide on cortical synaptic circuits are still unclear. To understand the effects of the neuropeptide SOM on cortical synaptic circuits, we performed a detailed side-by-side comparison of its postsynaptic effects on PCs, SOM-INs, and layer 1 interneurons (L1-INs) in the anterior cingulate cortex of male and female mice and found that SOM produced pronounced postsynaptic effects in PCs while having little to no effect on either IN type. This comparison allowed us to link the observed postsynaptic effects to SOM-induced modulations of glutamatergic and GABAergic synaptic transmission and to trace the impact of the neuropeptide on the neuronal circuitry between these three cell types. We show here that SOM depresses glutamatergic synaptic transmission via a presynaptic mechanism while exerting a differential impact on GABAA receptor- and GABAB receptor-mediated transmission at the pre- and postsynaptic level resulting in a shift of inhibition in L2/3 PCs from L1-INs to SOM-INs. In summary, this study unravels a novel aspect by which SOM modulates synaptic signaling between PCs, L1-INs, and SOM-INs.
Significance Statement
Somatostatin-expressing interneurons (INs) represent a distinct class of GABAergic INs whose main function is a modulation of top-down inputs into the cortex. Despite a large body of evidence investigating the behavior-specific activity of SOM-INs within the cortex, our understanding of the nature of SOM-IN–mediated modulation of synaptic transmission remains limited. Specifically, the function of the secreted neuropeptide within cortical circuits is not understood. By analyzing the effect of SOM not only on pyramidal cells but also on GABAergic INs, the study presented here provides detailed insight into the pre- and postsynaptic mechanisms of SOM in modulating local and distant information flow.
Introduction
Within the cerebral cortex, the neuropeptide somatostatin (SOM) is predominantly expressed by a distinct class of GABAergic interneurons (INs) termed somatostatin-positive INs (SOM-INs). SOM is thought to be released upon burst-firing or high-frequency action potential discharge of the respective presynaptic neuron (Tapia-Arancibia and Astier, 1989; van den Pol, 2012; Liguz-Lecznar et al., 2016). The biological actions of the peptide are conveyed via the activation of five distinct somatostatin receptor (SSTR) subtypes, SSTR1 to SSTR5, all of which are G-protein–coupled receptors (Moller et al., 2003; Tulipano and Schulz, 2007). In the postsynaptic target cell, the binding of the ligand to SSTRs activates inwardly rectifying potassium conductances (GIRK) through a direct interaction of the Gβγ subunit of the receptor with the corresponding channels (Whorton and MacKinnon, 2013). As a consequence, the postsynaptic cell becomes hyperpolarized, and its excitability is reduced. In the brain, the actions of SOM are mostly mediated via SSTR2 (Stumm et al., 2004; Meis et al., 2005; Riedemann and Sutor, 2019). In agreement with that, autoradiography assays, in situ hybridization, and single-cell sequencing studies identified SSTR2 as being prominently expressed in the cerebral cortex or as being expressed by the majority of cortical PCs (Fehlmann et al., 2000; Videau et al., 2003; Tasic et al., 2016). In addition, SSTRs are expressed by different types of GABAergic INs (Lukomska et al., 2020).
Cortical GABAergic INs are typically classified into molecularly, morphologically, and electrophysiologically distinct groups: The group of parvalbumin-positive INs (PV-INs) is the largest. PV-INs innervate the proximal dendrites and somata of pyramidal cells (PCs) and mostly provide feedforward inhibition. In contrast, other groups of GABAergic INs preferentially innervate the distal dendrites of PCs and provide feedback and feedforward inhibition (Gonzalez-Burgos et al., 2005; Tan et al., 2008; Kubota et al., 2015). These IN groups are the abovementioned SOM-INs and INs either expressing vasoactive intestinal peptide (VIP-INs) or not. The non–VIP-expressing INs can be subclassified into INs expressing Id2 (Id2-INs) or Reelin-INs (non–VIP-INs resp. Reelin-INs; Lee et al., 2015; Tremblay et al., 2016; Zhou et al., 2017; Nigro et al., 2018; Williams and Riedemann, 2021; Machold et al., 2023). PV-INs, SOM-INs, and Id2-INs are present in all cortical layers but layer 1 (L1), whereas VIP-INs and Reelin-INs are preferentially located in supragranular cortical layers, including L1. All IN types not only inhibit PCs but also inhibit each other to a variable degree leading to the disinhibition of PCs or, more accurately, to a shift of inhibition either from the soma to the dendrites or possibly, from one specific dendritic site to another (Tan et al., 2008; Gentet et al., 2012; Letzkus et al., 2015; Karnani et al., 2016). To date, we lack an understanding of how different neuropeptides such as SOM modulate these cortical circuits. It is known that at least 75% of all GABAergic fibers within L1 originate from SOM-INs (Shlosberg et al., 2003), rendering L1 a likely site for SOM release by SOM-INs. L1 contains INs [the so-called layer 1 interneurons (L1-INs)] that are mostly of the non-VIP type (Schuman et al., 2019), and their somata lie in close vicinity to axons of SOM-INs, possibly allowing for an impactful modulation of L1-INs by SOM. In addition, L1 contains a high density of horizontal fibers not only from other cortex areas but also from subcortical regions such as the brainstem or thalamus. These fibers target L2/3 PCs and GABAergic INs. We tested here whether SOM preferentially modulates these inputs at the distal dendrites of L2/3 PCs, the postsynaptic innervation site of presynaptic SOM-INs. Furthermore, we investigated how SOM modulates the synaptic circuitry between L2/3 PCs, SOM-INs, and L1-INs in the anterior cingulate cortex.
Materials and Methods
Animals
Experiments were performed on mice from three different lines: (1) a transgenic mouse line [FVB-Tg(GadGFP)45704Swn/J] in which a subset of SOM-INs expresses the enhanced green fluorescent protein (eGFP; Oliva et al., 2000); (2) a transgenic mouse line, where all GABAergic INs express channelrhodopsin-2 and YFP (VGAT-ChR2-YFP; Zhao et al., 2011); and (3) a crossbreed of lines 1 and 2. Animals were purchased from The Jackson Laboratory and were bred in the institute’s animal facility. All experiments were approved by the authors’ institutional committees on animal care and were performed according to the German Animal Protection Law, conforming to international guidelines on the ethical use of animals.
Slice preparation
The mean age of animals used in this study was postnatal day 36. Animals of either sex were used in the study. The preparation of coronal slices (thickness, 300 µm) was performed as previously described (Riedemann et al., 2018; Riedemann and Sutor, 2019). After the cutting procedure, the slices were collected and submerged in artificial cerebrospinal fluid (ACSF) containing the following (in mM): NaCl (125), KCl (3), NaH2PO4 (1.25), NaHCO3 (25), CaCl2 (2), MgCl2 (2), and D-glucose (25 mM). The ACSF was continuously perfused with 95% O2/5% CO2 to maintain a pH of 7.4. The osmolarity of the ACSF ranged between 310 and 323 mOsmol. Before recording, the slices were left to recover from the cutting procedure for 30 min at 28°C.
Whole-cell patch-clamp recordings
A single slice was transferred to a recording chamber mounted on the stage of an upright microscope (BX-RFA-1-5, Olympus). The ACSF temperature was maintained at 27–28°C with the help of a temperature controller (TC-324B, Warner Instruments). All cells were visualized with a Dodt contrast tube (DGC, Scientifica) that was attached to the microscope. SOM-INs were additionally visualized by eGFP expression with the help of a fluorescence lamp (pE-300, CooLED) and epifluorescence optics for green fluorescence (Chroma Technology). Images were taken and displayed using a software-operated microscope camera (Evolve 512 Delta, Teledyne Photometrics). Electrodes were fabricated from borosilicate glass capillaries (OD, 1.5 mm; ID, 0.86 mm; Hugo Sachs Elektronik-Harvard Apparatus) and were filled with a solution containing (in mM) K-gluconate (135), KCl (4), NaCl (2), EGTA (0.2), HEPES (10), Mg-ATP (4), Na-GTP (0.5), and phosphocreatine (10) or with a solution containing (in mM) KCl (139), NaCl (2), EGTA (0.2), HEPES (10), Mg-ATP (4), Na-GTP (0.5), and phosphocreatine (10). All solutions had an osmolarity of 290–300 mOsm and a pH of 7.3. A silver/silver chloride pellet served as a reference electrode. Somatic whole-cell recordings were made in current- or voltage-clamp mode using an ELC-03XS amplifier (npi electronic). After the rupture of the membrane, the electrode capacitance and series resistance were compensated as previously described (Riedemann et al., 2016a). The resting membrane potential (RMP) of all cells ranged between −58 and −85 mV.
Electrical or optical stimulation
Stimulation was performed by either placing a monopolar glass electrode into cortical layer 1 or by applying a short focal laser pulse (wavelength, 473 nm; light pulse diameter, 5 µm) onto cortical layer 1 perpendicular to each recorded cell. Stimulation current intensities ranged between 10 and 500 µA, and the pulse duration was 100 µs. Laser intensities ranged between 1 and 15 mW and lasted for 1–3 ms. Evoked responses were recorded after a single stimulus, after a paired stimulus [two pulses at 100, 50, 30, and 20 ms interpulse interval (IPI)], or after 10 stimuli (10 pulses at 20 ms IPI). All stimulation protocols were applied at intervals of 15–30 s.
Drug application
All drugs were either obtained from Tocris Bioscience or Biotrend. Blockers of glutamatergic neurotransmission were NBQX (10 µM) and D-AP5 (20 µM). The GABAAR-mediated transmission was blocked by the addition of 10 µM bicuculline. GABABR-mediated signaling was inhibited by SCH50911 (10 µM). Action potentials (APs) were blocked by tetrodotoxin (TTX) at a concentration of 0.5 µM. Somatostatin-14 (SOM), the pan-SSTR antagonist cyclosomatostatin, and the selective SSTR2 antagonist CYN15048 were added at a concentration of 1 µM. All drugs were either applied by addition to the bath or by local pressure application.
Bath perfusion
Antagonists of GABAergic (bicuculline, SCH50911), glutamatergic transmission (D-AP5, NBQX), or SSTR antagonists (cyclosomatostatin, CYN154806) were added to the bath at least 10 min prior to recording and were left inside the bath until the end of the recording session. In some cases, antagonists were washed out to test the reversibility of the effect. SOM and/or baclofen were applied for 5–10 min at a concentration of 1 µM (SOM) or 30 µM (baclofen). The perfusion rate was set to 3 ml/min.
Pressure application
GABA (10 mM), glutamate (10 mM), and baclofen (300 µM) were applied via focal pressure application. To this end, a glass capillary was positioned in L1 perpendicular to the recorded cell, and the pressure was adjusted to 300–500 mbar. The pressure pulse lasted for 1 s and the application intervals ranged between 180 and 240 s.
Data acquisition and analysis
Voltage and current signals were amplified, filtered at 20 kHz (current-clamp recordings) or at 3–5 kHz (voltage-clamp recordings), and digitized at sampling rates of 10–50 kHz. Data acquisition and generation of command pulses were accomplished by an analog–digital converter (CED 1401 Power 3, Cambridge Electronic Design) in conjunction with the signal data acquisition software (Version 6, Cambridge Electronic Design). Data analysis was performed using IGOR Pro 9 (WaveMetrics) together with the NeuroMatic IGOR plugin (www.neuromatic.thinkrandom.com). Fingerprint electrophysiological properties of recorded neurons were assessed as previously described (Riedemann et al., 2018). SOM- and/or baclofen-induced differences in the holding current were analyzed at a membrane potential of −60 mV with either a K-gluconate-based or a K-chloride–based intracellular solution. Voltage dependence of SOM- and/or baclofen-induced current was assessed by subtracting the voltage-ramp–induced current under the control condition (average of 5) from the voltage-ramp–induced current after drug application (average of 5). Spontaneous postsynaptic currents (sEPSCs) were analyzed at a holding potential of −65 mV and were automatically detected using the algorithm provided by the NeuroMatic plugin (Version 3.0; Rothman and Silver, 2018). In voltage-clamp recordings, the detection threshold was set to −10 pA, and in current-clamp recordings, the detection threshold was set to 0.35 mV. We analyzed the frequency, amplitude, decay time, and rise time of all synaptic events under control condition (5 min) and after drug application (5 min). Evoked synaptic responses and agonist-induced responses (GABA, glutamate, SOM, and baclofen) were assessed by determining the peak voltage or current and by subtracting the baseline voltage or current from this peak value. Synaptic coupling between pairs of neurons was tested by injecting at least 20 suprathreshold depolarizing current steps into the presynaptic neuron and analyzing the corresponding voltage changes in the postsynaptic neuron. Synaptic coupling was defined as successful if the average voltage trace of the passive neuron showed a postsynaptic response within a (peak-to-peak) latency of <7 ms. The cross-correlation of neuronal activity between pairs of neurons was quantified in IGOR Pro. To this end, 80- to 100-s-long voltage traces obtained from pairs of neurons under control condition and 5 or 15 min after bath application of 1 µM SOM were analyzed for correlation assessing the Pearson’s linear correlation coefficient r according to Nazemi and Jamali (2018). Statistical analyses were performed by initially testing for a normal distribution of data points using the D'Agostino and Pearson’s omnibus normality test. In case of a normal distribution of data points, paired or unpaired two-tailed students’ t tests were performed for comparisons between two different conditions. In case of a nonnormal distribution of data points, Mann–Whitney or Wilcoxon signed-rank tests were performed. Comparisons between multiple conditions were performed by one-way ANOVA or repeated-measures ANOVA with Tukey’s multiple-comparisons tests when data points were normally distributed. In case of a nonnormal distribution of data points, Kruskal–Wallis or Friedman test with Dunn's multiple-comparisons tests were applied. Significance levels were p < 0.05 (*), p < 0.01 (**), and p < 0.001 (***). Principal component analysis of electrophysiological features of PCs, SOM-INs, and L1-INs was performed the “prcomp” function implemented in R (https://www.r-project.org/). All variables were scaled to unit variance prior to analysis. Unless stated otherwise, data are presented as line or dot plots showing individual cell responses and mean ± SEM. The sample size (n) is indicated on the diagram and/or in the figure legend. Graphs were prepared in IGOR Pro, in GraphPad Prism, or in RStudio (Version 1.3.1093, https://www.r-project.org/).
Results
Somatostatin depresses glutamatergic synaptic transmission
We initially characterized L2/3 PCs, SOM-INs, and L1-INs according to their passive and active membrane properties. The fingerprint electrophysiological properties of each neuron type are provided in Table 1 and are depicted in Figure 1.
Summary of fingerprint electrophysiological properties in L2/3 PCs, SOM-INs, and L1-INs. A, Representative voltage traces (bottom recording trace) upon injection of a series of hyper- and depolarizing current steps (top recording trace) into five individual L2/3 PCs (left panel). Pie chart showing the distribution of AP discharge patterns (right panel). B, Depiction of representative voltage traces of five individual L2/3 SOM-INs upon injection of a series of hyper- and depolarizing current steps. Note the large sag potential (asterisk) in SOM-INs (left panel). Pie chart showing the variability of firing patterns (right panel). C, Representative current–voltage traces of five individual L1-INs. The relative distribution of different firing patterns in all L1-INs is shown as a pie chart in the right panel. Continuous, continuously spiking; discontinuous, discontinuously spiking; transient, transiently spiking; late, late spiking; nonlate, nonlate spiking. D, Principal component analysis of 11 electrophysiological features derived from 237 L2/3 PCs (86 animals, 43♂, 43♀; mean age, P39), 88 SOM-INs (39 animals, 22♂, 17♀; mean age, P44), and 70 L1-INs (37 animals, 19♂, 18♀; mean age P50).
Summary of the fingerprint electrophysiological properties of PCs, SOM-INs, and L1-INs (mean ± SD)
Next, we characterized synaptic inputs onto L2/3 PCs, SOM-INs, and L1-INs after electrical stimulation of L1 (Fig. 2A). Electrical stimulation in L1 excites all intact horizontal fibers within a given coronal brain slice; however, the composition of excited fibers is not known. In our slice preparation, these axons are mostly excitatory fibers from other cortex areas and GABAergic fibers from local INs. We initially performed an input–output analysis and compared synaptic recruitment of L2/3 PCs, SOM-INs, and L1-INs by excitatory projection fibers in L1 and found that L1-INs are more easily recruited compared with either PCs or SOM-INs. No difference in synaptic recruitment was observed between SOM-INs and L2/3 PCs (Fig. 2B,C). In agreement with Marek et al. (2018), we found that this stimulation evoked a postsynaptic response in L2/3 PCs that was composed of three distinct parts: An initial fast depolarization followed by a slower depolarization that was followed by a late-onset hyperpolarization (Fig. 2D, orange trace). Furthermore, electrical stimulation of L1 in the presence of blockers of glutamatergic transmission disclosed a fast-onset and late-onset hyperpolarizing response in L2/3 PCs that were kept at a membrane potential of −60 mV (Fig. 2E). Given that electrical stimulation of L1 induced an antidromic spike in 20 out of 22 SOM-INs and in 8 out of 10 L1-INs (Fig. 2A,F,G), it is suggested that either IN type contributes to the evoked triphasic postsynaptic response observed in L2/3 PCs. In addition, we tested which of these three postsynaptic components were affected by SOM. Therefore, we electrically stimulated L1 and recorded evoked postsynaptic potentials (ePSPs) in L2/3 PCs that were kept at defined membrane potentials (ranging from −100 to −50 mV). By plotting the amplitudes of each postsynaptic response component as a function of the respective membrane potential, we analyzed the nature of these ePSP components and found that the two slower-onset responses had reversal potentials between −65 and −55 mV and between −100 and −90 mV, respectively, indicative of a presumed GABAAR- and GABABR-mediated response (Fig. 2H–K), suggesting that inhibition in L1 is equally conveyed by GABAA and GABAB receptors. In addition, we showed that SOM treatment resulted in a depression of the initial excitatory postsynaptic response (Fig. 2I) at high membrane potentials and in a depression of both inhibitory postsynaptic responses at depolarized membrane potentials (Fig. 2J,K).
Synaptic recruitment of L2/3 PCs, SOM-INs, and L1-INs by excitatory fibers in L1. A, Schematized synaptic circuitry between L2/3 PCs, SOM-INs, and L1-INs. All three cell types receive glutamatergic inputs from other brain areas via L1. In addition, L1 contains numerous GABAergic fibers of SOM-INs and L1-INs. We placed a monopolar glass electrode into L1 which stimulates these glutamatergic and GABAergic fibers. B, Single ePSPs of an L2/3 PC (blue), a SOM-IN (green), and an L1-IN (magenta) after electrical stimulation in L1. C, Input–output curve of ePSPs in PCs (n = 45, 27 animals, 14♂, 13♀; mean age, P29), SOM-INs (n = 14, 8 animals, 5♂, 3♀; mean age, P30), and L1-INs (n = 10, 3 animals, 3♂; mean age, P30). Mean ePSP amplitude ± SEM, 0 µA: PC 0.30 ± 0.05, SOM-IN 0.28 ± 0.11, L1-IN 0.22 ± 0.07; 50 µA: PC 0.83 ± 0.17, SOM-IN 0.02 ± 0.10, L1-IN 6.03 ± 1.97; 100 µA: PC 2.71 ± 0.44, SOM-IN 2.50 ± 0.90, L1-IN 9.07 ± 2.29; 150 µA: PC 5.05 ± 0.64, SOM-IN 3.31 ± 0.91, L1-IN 11.73 ± 2.19; 200 µA: PC 7.56 ± 1.34, SOM-IN 5.93 ± 1.55, L1-IN 12.14 ± 1.71; 250 µA: PC 8.91 ± 1.52, SOM-IN 10.29 ± 1.94, L1-IN 12.56 ± 1.76; 300 µA: PC 8.68 ± 1.28, SOM-IN 9.90 ± 1.94, L1-IN 13.43 ± 1.35. At 100 µA, the ePSP amplitude in L1-IN is significantly higher compared with PCs (p < 0.05, two-way ANOVA with Bonferroni’s post test). At 150 µA, the ePSP amplitude in L1-INs is significantly higher compared with either PCs or SOM-INs (L1-IN vs PC, p < 0.05; L1-IN vs SOM-IN, p < 0.01, two-way ANOVA with Bonferroni’s post test). D, Single ePSPs of an L2/3 PC after L1 stimulation with different stimulation intensities (average of 3). The early-onset (open arrow) and the later-onset depolarizing response (solid arrow) and the late-onset hyperpolarizing response (open asterisk) are exemplified in the orange trace. E, Single traces (average of 3) of ePSPs recorded in the same PC (MP, −60 mV) as shown (in D) in the absence (gray) or presence (black) of 10 µM NBQX and 20 µM D-AP5. F, G, Electrical stimulation in L1 evokes antidromic spikes in 20 out of 22 SOM-INs and in 8 out of 10 L1-INs. H, Single evoked ePSPs (average of 3) in an L2/3 PC that was kept at different membrane potentials before (black) and after addition of SOM (red). I, Individual amplitudes of the initial fast depolarization under control condition (black) and after addition of SOM (red) plotted as a function of the membrane potential (mean ePSP amplitude −100 mV: control, 11.72 ± 2.25 mV; SOM, 9.71 ± 2.29 mV, p < 0.05, two-way ANOVA with Bonferroni’s posttest. −90 mV: control, 10.60 ± 2.03 mV; SOM, 9.22 ± 2.07 mV. −80 mV: control, 9.08 ± 1.66 mV; SOM, 8.09 ± 1.64 mV. −70 mV: control, 7.54 ± 1.44 mV; SOM, 6.88 ± 1.29 mV. −60 mV: control, 5.37 ± 1.08 mV; SOM, 5.29 ± 1.00 mV. −50 mV: control, 3.84 ± 0.90 mV; SOM, 3.57 ± 0.75 mV). J, Amplitudes of the presumed GABAAR-mediated response under control condition (black) and after addition of SOM (red) plotted as a function of the membrane potential (mean ± SEM; mean ePSP amplitude −100 mV: control, 2.53 ± 0.74 mV; SOM, 1.67 ± 0.50 mV. −90 mV: control, 2.66 ± 0.8 mV; SOM, 1.92 ± 0.58 mV. −80 mV: control, 2.49 ± 0.65 mV; SOM, 2.02 ± 0.60 mV. −70 mV: control, 1.65 ± 0.60 mV; SOM, 1.74 ± 0.58 mV. −60 mV: control, −0.03 ± 0.59 mV; SOM, 0.73 ± 0.53 mV. −50 mV: control, −3.22 ± 0.77 mV; SOM, −1.49 ± 0.79 mV, p < 0.001, two-way ANOVA with Bonferroni’s posttest). K, Amplitudes of the presumed GABABR-mediated response under control condition (black) and after addition of SOM (red) plotted as a function of the membrane potential (mean ± SEM; mean ePSP amplitude −100 mV: control, 0.23 ± 0.10 mV; SOM, −0.05 ± 0.14 mV. −90 mV: control, −0.10 ± 0.12 mV; SOM, −0.3 ± 0.10 mV. −80 mV: control, −0.57 ± 0.15 mV; SOM, −0.59 ± 0.12 mV. −70 mV: control, −1.18 ± 0.24 mV; SOM, −0.83 ± 0.20 mV. −60 mV: control, −2.17 ± 0.45 mV; SOM, −1.30 ± 0.33 mV; p < 0.01, two-way ANOVA with Bonferroni’s posttest. −50 mV: control, −3.68 ± 0.84 mV; SOM, −2.26 ± 0.52 mV, p < 0.001, two-way ANOVA with Bonferroni’s posttest). Data in H–K derived from 8 animals, 6♂ and 2♀, with a mean age of P29.
Next, we recorded pharmacologically isolated evoked excitatory postsynaptic currents (eEPSCs) in L2/3 PCs after electrical stimulation of L1 (Fig. 3A). Cells were clamped at a holding potential of −65 mV, and the GABAAR blocker bicuculline (10 µM) was added to the bath at least 15 min prior to a baseline recording. Given that L1 exerts a strong inhibitory control over L2/3 PCs (Fig. 2E), we observed prolonged eEPSC responses in the presence of 10 µM bicuculline, similar to what has previously been reported (Shlosberg et al., 2003) in L5 PCs after removal of L1 in acute brain slices. For comparison, the addition of 1 µM bicuculline resulted in an increase in ePSP amplitude but did not cause prolonged responses (Fig. 3B,C, arrow, Extended Data Fig. 3-1). After a 10 min baseline recording, we applied SOM to the bath for 10 min and continued the recording. Bath application of SOM significantly reduced the eEPSC amplitude. In contrast, perfusion of acute slices with the pan-SSTR blocker cyclosomatostatin (Cyclosom) tended to increase it (Fig. 3D). Comparing the eEPSC amplitudes of SOM- and Cyclosom-treated cells, we found a significantly larger eEPSC amplitude in Cylcosom-treated neurons, suggesting that endogenously released SOM modulates glutamatergic synaptic transmission in L1 (Fig. 3E).
SOM depresses glutamatergic transmission by depressing presynaptic glutamate release. A, Schematized experimental setup. B, Representative single traces (average of 3) of evoked eEPSCs during baseline recording (dark gray, CON) and after addition of Cyclosom (yellow). C, Representative single traces of eEPSCs during baseline recording (dark gray, CON) and after addition of SOM (red). An exemplary current-clamp recording of eEPSPs in the presence of 1 µM bicuculline is shown in Extended Data Figure 3-1. D, The mean eEPSC amplitude plotted as a function of time. E, Dot plot showing the normalized current amplitude just before drug application in comparison with that 10 min after addition of drug (mean normalized eEPSC amplitude control, 98.8 ± 4.4; SOM, 77.6 ± 5.2; p = 0.0265, paired t test; n = 8, 2 animals, 2♂; mean age, P36; mean normalized eEPSC amplitude control, 101.0 ± 3.1; Cyclosom, 119.5 ± 14.5; p = 0.2061, Wilcoxon signed-rank test; n = 11, 5 animals, 2♂, 3♀; mean age, P36; mean normalized eEPSC amplitude SOM vs mean eEPSC amplitude Cyclosom: p = 0.0294, unpaired t test). F, Schematized experimental setup. G, Average uEPSPs (bottom trace) recorded in the postsynaptic cell before (black) and after bath application of SOM (red). Individual uEPSP is depicted in light gray (CON) and light red (SOM). The presynaptic spike is depicted in the top trace. H, The uEPSP amplitude (average of 2) of the cell shown in G plotted as a function of trial number. I, Line plot showing mean uEPSPs under control condition and after SOM exposure (mean uEPSP control, 0.37 ± 0.08; mean uEPSP SOM, 0.25 ± 0.06; p = 0.0117, paired t test; n = 5, 4 animals, 2♂, 2♀; mean age, P44). J, Schematized experimental setup (left panel). Single representative current traces (average of 2) upon focal application of glutamate in the presence of TTX before (black) or after bath application of SOM (red) and block of AMPA and NMDA receptors (light gray, right panel). K, Line plot showing individual (and mean) glutamate-induced current responses under control condition and after addition of SOM (mean current amplitude control, −469.2 ± 133.1 pA; SOM, −430.5 ± 156.3 pA, p = 0.4185, paired t test, n = 6, 2 animals, 2♀; mean age, P26). L, Single ePSP traces (average of 3) after paired-pulse stimulation at different IPIs before (black) and after (red) SOM bath application. M, Line plot depicting the paired-pulse ratios (PPR) under control condition and after SOM bath application. Mean PPR ratio at IPI 100 ms: control, 125.4 ± 7.8; SOM, 144.9 ± 12.9 (n = 13, 10 animals, 5♂, 5♀; mean age, P24). IPI 50 ms: control, 144.5 ± 6.9; SOM, 160.7 ± 11.3 (n = 15, 10 animals, 5♂, 5♀; mean age, P24). IPI 30 ms: control, 151.2 ± 6.7; SOM, 169.5 ± 11.5 (n = 15, 10 animals, 5♂, 5♀; mean age, P24). IPI 20 ms: control, 159.9 ± 8.0; SOM, 186.1 ± 11.84 (n = 15, 10 animals, 5♂, 5♀; mean age, P24). SOM increases the PPR at an IPI of 100 and 20 ms (IPI 100 ms control vs SOM, p = 0.0107, Wilcoxon signed-rank test; IPI 50 ms control vs SOM, p = 0.1525, paired t test; IPI 30 ms control vs SOM, p = 0.0334, paired t test; IPI 20 ms control vs SOM, p = 0.0197, paired t test).
Extended Data 3-1
SOM depresses the amplitude of eEPSPs in L2/3 PCs. Left panel: Depiction of single eEPSPs in the same L2/3 PC before the addition of 1 µM bicuculline (CON, black), after addition of bicuculline (bic, orange) and after application of 1 µM SOM (SOM, red). Right panel: Depiction of a histogram showing the eEPSP amplitude over the recording duration. Download Extended Data 3-1, TIF file.
In order to test whether SOM also modulates local information flow within L2/3 of the aCC, we performed dual recordings between pairs of synaptically coupled PCs. We averaged unitary excitatory postsynaptic potentials (uEPSPs) in the postsynaptic PC after evoking at least 15 APs in the presynaptic PC and compared the mean uEPSP amplitude under control conditions with that after bath application of SOM. We found that SOM significantly reduced the mean uEPSP amplitude (Fig. 3F–I), indicating that SOM depresses local and distant information flow between PCs.
To test whether this SOM-induced decrease in eEPSC amplitude was due to a postsynaptic depression of glutamatergic currents or a presynaptic inhibition of glutamate release, we synaptically isolated PCs by bath application of the sodium channel blocker TTX and focally applied glutamate to the apical tuft of L2/3 PC dendrites in L1 (Fig. 3J,K). Recorded cells were clamped at a holding potential of −65 mV, and glutamate was puffed at an interval of 180 s to ensure that the glutamate-driven response returned to baseline levels prior to consecutive drug applications. Glutamate was applied three times under control conditions (i.e., in the presence of TTX) before the addition of SOM to the bath. Bath application of SOM failed to reduce the amplitude of the glutamate-driven response rendering a postsynaptic SOM effect on glutamate receptors highly unlikely.
In addition, SOM increased the paired-pulse ratio (PPR) at an IPI of 20, 30, and 100 ms (Fig. 3L,M) and reduced the frequency but not the amplitude of pharmacologically isolated sEPSCs recorded in L2/3 PCs (Fig. 4).
SOM depresses spontaneous excitatory transmission in L2/3 PCs. A, Top panel, Representative current trace recorded from a PC under control condition showing spontaneous sEPSCs in the presence of bicuculline. Bottom panel, Depiction of an expanded current trace of the same cell shown in the top panel. B, Top panel, Individual current trace of the same PC as in A after bath application of SOM. Bottom panel, Expanded current trace of the corresponding trace in the top panel. C, Cumulative probability plot of cell shown in A and B depicting the sEPSC frequency under control condition (black) and after bath application of SOM (red). D, The line plot depicts all sEPSC frequencies before and after addition of SOM. Mean sEPSC frequency control, 0.59 ± 0.17 Hz; mean sEPSC frequency SOM, 0.24 ± 0.06 Hz, p = 0.028, paired t test. E, Average sEPSC trace of all events detected under control condition (black) and after SOM treatment (red) in the cell depicted in A and B. F, The line plot depicts the amplitude of all sEPSCs recorded before and after SOM exposure. Mean sEPSC amplitude control, 17.13 ± 1.64 pA; mean sEPSC amplitude SOM, 17.31 ± 1.51 pA, p = 0.542, Wilcoxon signed-rank test. Data in Figure 4 are derived from 8 animals, 3♂ and 5♀, with a mean age of P27.
Next, the possibility of endogenously released SOM was probed by analyzing spontaneous excitatory transmission under control condition and in the presence of Cylcosom or the specific SSTR2 blocker CYN154806. Bath application of either antagonist had no effect on the sEPSC frequency or amplitude, perhaps due to the fact that PC activity in acute brain slices is too little to disclose the effect of endogenous SOM release on spontaneous excitatory transmission (Fig. 5).
SSTR antagonism has no effect on spontaneous excitatory transmission in L2/3 PCs of the aCC. A, Top panel, Representative 30 s current trace of sEPSC recording under control condition in the presence of 10 µM bicuculline. Bottom panel, Expanded current trace of the corresponding trace depicted in the top panel. B, Top panel, Representative 30 s current trace of sEPSC recordings of the same neuron as in A after addition of 1 µM cyclosomatostatin (Cylcosom). Bottom panel, Expanded current trace of the corresponding trace depicted in the top panel. C, Cumulative probability plot of the sEPSC frequency of cell depicted in A and B under control condition (dark gray) and after Cyclosom treatment (yellow). D, The line plot depicts the sEPSC frequencies before and after addition of the drug. Mean sEPSC frequency control, 0.31 ± 0.14 pA; mean sEPSC frequency Cyclosom, 0.34 ± 0.19 pA, p = 0.895, paired t test. E, Average sEPSC trace of all events detected under control condition (dark gray) and after addition of Cyclosom (yellow) in the same neuron depicted in A and B. F, Line plot showing sEPSC amplitudes before and after addition of Cyclosom. Mean sEPSC amplitude control, 13.79 ± 0.55 pA; SOM, 14.17 ± 0.91 pA, p = 0.538, paired t test (n = 7, 3 animals, 3♂; mean age, P31). G, Representative 30 s current trace of sEPSC recordings under control condition in the presence of 10 µM bicuculline. Bottom panel, Expanded current trace of the corresponding trace depicted in the top panel. H, Top panel, Representative 30 s current trace of sEPSC recordings of the same neuron as in G after addition of 1 µM CYN154806 (CYN). Bottom panel, Expanded current trace of the corresponding trace depicted in the top panel. I, Cumulative probability plot of the sEPSC frequency of cell shown in G and H under control condition (dark gray) and after CYN154806 exposure (yellow). J, The line plot depicts the sEPSC frequencies before and after addition of the drug. Mean sEPSC frequency control, 0.43 ± 0.10 pA; SOM, 0.42 ± 0.07 pA, p = 0.907, paired t test. K, Average sEPSC trace of all events detected under control condition (dark gray) and after addition of CYN (yellow) in the same cell shown in G and H. L, The line plot depicts the mean sEPSC amplitudes before and after addition of the drug. sEPSC amplitude control, 12.80 ± 0.61 pA; SOM, 12.77 ± 0.38 pA, p = 0.950, paired t test (n = 7, 3 animals, 2♂, 1♀; mean age, P27).
In line with the above hypothesis that SOM relies on neuronal activity to inhibit glutamatergic synaptic transmission, SOM neither affected the frequency nor the amplitude of miniature EPSCs (mEPSCs) recorded in the presence of TTX and bicuculline (Fig. 6). In summary, the data presented so far indicate that SOM inhibits presynaptic glutamate release.
SOM depresses miniature excitatory currents in L2/3 PCs. A, Top panel, Representative current trace of miniature EPSC (mEPSC) recordings under control condition and in the presence of bicuculline and TTX. Bottom panel, Expanded current trace of the corresponding trace shown in the top panel. B, Top panel, Representative current trace of mEPSC recordings of the same neuron as in A in the presence of SOM. Bottom panel, Expanded current trace of the corresponding trace in the top panel. C, Cumulative probability plot of the mEPSC frequency of cell shown in A and B under control condition (black) and after bath application of SOM (red). D, The line plot depicts the mEPSC frequencies before and after addition of SOM. Mean mEPSC frequency control, 0.25 ± 0.12 Hz; mean mEPSC frequency SOM, 0.31 ± 0.17 Hz, p = 0.859, Wilcoxon signed-rank test. E, Average mEPSC trace of all events detected under control condition (black) and after addition of SOM (red) in the same neuron depicted in A and B. F, The line plot shows all mEPSC amplitudes recorded before and after SOM treatment. Mean mEPSC amplitude control, 10.82 ± 1.10 pA; mean mEPSC amplitude SOM, 11.44 ± 1.39 pA, p = 0.41, paired t test. Data in Figure 6 are derived from 6 animals, 3♂ and 3♀, with a mean age of P29.
SOM depresses evoked GABAergic synaptic transmission in L2/3 PCs
Our next aim was to determine whether SOM inhibited GABAergic synaptic transmission in a similar fashion. Therefore, we initially performed an input–output analysis of evoked inhibitory postsynaptic potentials (eIPSPs) recorded from L2/3 PCs following optogenetic activation of GABAergic fibers in L1 using the VGAT-ChR2-YFP mouse line. PCs were kept at a membrane potential of −60 mV. As mentioned previously, this allowed us to observe a biphasic inhibitory postsynaptic response with an early and a late peak (Fig. 7A,B), indicative of a GABAAR- and a GABABR-mediated synaptic response (Fig. 7C) that is presumably mediated by activation of SOM-INs and L1-INs as optogenetic stimulation in L1 evoked a spike in 100% of all SOM-INs (14 out of 14) and in 80% of all L1-INs (8 out of 10) tested (Fig. 7D). Further, channelrhodopsin-2 (ChR2) responsiveness in SOM-INs was tested in a crossbreed between VGAT-ChR2-YFP and FVB-Tg(GadGFP)45704Swn/J mice by triggering light-induced APs by trains of light pulses at different IPIs. Light pulse frequencies ranged from 20 to 50 Hz, and we could show that these stimulation frequencies elicited light-induced APs in SOM-INs (Fig. 7E), when the laser was focused in L1. Optogenetic stimulation either by a single pulse or by 10 pulses at 50 Hz revealed that SOM reduced the early as well as the late inhibitory response (Extended Data Fig. 8-1).
Laser stimulation in L1 elicits inhibitory postsynaptic potentials in L2/3 PCs and light-evoked APs in SOM-INs and L1-INs. A, Schematized experimental setup. B, Representative single traces (average of 3) of optically evoked IPSPs in L2/3 PCs at different stimulation intensities. Note the initial fast (asterisk) and the later-onset (circle) hyperpolarization (left panel). Input–output curve of the normalized average early and late eIPSP amplitudes at different stimulation intensities (right panel). Data derived from 5 to 20 L2/3 PCs (2–7 animals; age range, P28–P34). eIPSP amplitude early: 1 mW, 1.00 ± 0.00 mV; 2 mW, 0.95 ± 0.33 mV; 3 mW, 3.29 ± 0.72 mV; 5 mW, 4.77 ± 1.19 mV; 7 mW, 5.8 ± 1.45 mV; 10 mW, 6.15 ± 1.75 mV. eIPSP amplitude late: 1 mW, 1.00 ± 0.00 mV; 2 mW, 1.00 ± 0.21 mV; 3 mW, 2.40 ± 0.39 mV; 5 mW, 3.15 ± 0.49 mV; 7 mW, 4.28 ± 0.60 mV; 10 mW, 4.72 ± 0.81 mV. C, Representative single traces (average of 3) of optically evoked IPSPs (five pulses at 50 Hz) under control conditions (black), after bath application of bicuculline (dark gray) and after bath application of SCH50911 (light gray) in the same PC. The cell was kept at a membrane potential of −60 mV. D, Top panel, Schematic of recording situation. Bottom panel, Optically evoked APs in SOM-INs and L1-INs. E, Top panel, Schematized recording situation. Bottom panel, Light stimulation at different frequencies (as indicated) evokes APs in SOM-INs. For comparison, an evoked AP train following 10 pulses at 50 Hz with a monopolar glass electrode placed in L1 is depicted on the far right.
We next isolated the GABAAR-driven response by bath application of the GABABR blocker SCH50911 and found that SOM significantly reduced GABAAR-mediated postsynaptic responses evoked by a single light or electrical pulse (Fig. 8A,B). In contrast, SOM had no effect on the maximal GABAAR-mediated response following 10 pulses at 50 Hz (Fig. 8C). Next, using optogenetic stimulation, we recorded evoked GABABR-mediated responses in the presence of bicuculline or using electrical stimulation in the presence of NBQX, D-AP5, and bicuculline. SOM significantly reduced the amplitude of evoked GABABR-driven responses following a single pulse or 10 pulses at 50 Hz (Fig. 8D–F).
SOM depresses GABAergic synaptic transmission at the pre- and postsynaptic levels. A, Schematized recording setup. B, Left panel, Representative single traces (average of 3) of evoked GABAAR-mediated postsynaptic potentials before (black) and after bath application of SOM (red) in the presence of SCH50911 following a single laser pulse. Middle panel: amplitudes of single GABAAR-mediated responses as a function of time. The red bar indicates the SOM exposure. Right panel, The line plot depicts the amplitudes of GABAAR-mediated eIPSPs before and after addition of SOM (SCH50911, 5.43 ± 0.78 mV; SOM, 4.54 ± 0.79 mV; p = 0.025, paired t test; n = 11, 6 animals, 6♂; mean age, P57). C, Left panel, Representative single traces (average of 3) of evoked GABAAR-mediated postsynaptic potentials before (black) and after SOM treatment (red) in the presence of SCH50911 following 10 laser pulses at 50 Hz. Right panel, The line plot depicts the amplitudes of GABAAR-mediated eIPSPs before (black) and after (red) addition of SOM (SCH50911, 7.69 ± 1.21 mV; SOM, 7.15 ± 1.29 mV, p = 0.764, paired t test; n = 10, 7 animals, 5♂, 2♀; mean age, P61). D, Representative ePSP trace (average of 3) after electrical stimulation in L1 before (gray) and after addition of NBQX and D-AP5 (purple). The recorded PC had a RMP of −78 mV. E, Left panel, Representative single traces (average of 3) of the same neuron shown in D. RMP, −60 mV. Depiction of evoked GABABR-mediated postsynaptic potentials before (black) and after SOM exposure (red) in the presence of NBQX, D-AP5, and bicuculline after a single pulse. Middle panel, Amplitudes of GABABR-mediated responses after a single stimulus as a function of time. SOM application indicated by a solid red rectangle. Right panel, The line plot depicts the eIPSPGABAB amplitudes before (black) and after (red) addition of SOM (bicuculline, 5.33 ± 1.20 mV; SOM, 3.96 ± 0.90 mV; p = 0.0206, paired t test; n = 12, 7 animals, 3♂, 4♀; mean age, P51). F, Left panel, Representative single traces (average of 3) of the same neuron depicted in D and E, showing evoked GABABR-mediated postsynaptic potentials before (black) and after bath application of SOM (red) in the presence of NBQX, D-AP5, and bicuculline following 10 pulses at 50 Hz. Right panel, The line plot depicts the mean eIPSPGABAB amplitudes before (black) and after (red) addition of SOM to the bath (bicuculline, 8.07 ± 1.17 mV; SOM, 5.82 ± 1.13 mV; p = 0.0012, paired t test; n = 11, 7 animals, 3♂, 4♀; mean age, P51). More information on SOM-induced inhibition of combined GABAAR- and GABABR-mediated transmission is provided in Extended Data Figure 8-1. G, Schematized experimental setup. H, Left panel, Representative single current traces (average of 2) in response to a GABA puff before addition of TTX (CON, black), after addition of TTX (TTX, gray), and after exposure to SOM (SOM, red). Right panel, Line plot showing the mean GABA puff-induced current responses under TTX condition (CON) and after SOM treatment (mean GABA response TTX, −25.97 ± 8.49 pA; mean GABA response SOM, −26.01 ± −10.39 pA; p = 0.557, Wilcoxon signed-rank test, n = 10, 6 animals, 2♂, 4♀; mean age, P26). I, Left panel, Representative single traces (average of 2) of baclofen puff-induced currents under control condition in the presence of TTX (CON, black), after addition of SOM to the bath (SOM, red), and after SOM washout (wash, gray). Right panel, Mean baclofen puff-induced current responses under control condition and after SOM bath application as line plot (mean baclofen current control, 65.87 ± 15.76 pA; mean baclofen current SOM, 42.74 ± 8.77 pA, p = 0.0375, paired t test; n = 10, 5 animals, 4♂, 1♀; mean age, P30).
Extended Data 8-1
SOM depresses evoked GABAergic transmission in L2/3 PCs. Representative optogenetically evoked IPSPs (eIPSPs, average of 3) induced by a single light pulse (left panel) and after 10 light pulses (stimulation frequency 50 Hz, right panel) under control (black) condition and after bath application of SOM (red). Download Extended Data 8-1, TIF file.
In addition, PCs were synaptically isolated by bath perfusion with TTX, and GABA was focally applied to the apical tuft of L2/3 PC dendrites in L1. GABA was applied under control conditions (i.e., in the presence of TTX) and after bath application of SOM. SOM bath application had no effect on the postsynaptic GABAAR-mediated response pointing to a presynaptic SOM-induced inhibition of GABA release (Fig. 8G,H). However, repeated pressure application of the GABABR agonist baclofen at an interval of 240 s triggered an outward current in PCs that became significantly reduced after bath application of SOM (Fig. 8I). In summary, these data suggest that SOM not only acts presynaptically to block GABA release from GABAergic INs but also induces a postsynaptic depression of GABABR-mediated responses in L2/3 PCs. Therefore, we next focused on (1) postsynaptic SOM effects in L2/3 PCs and (2) presynaptic SOM effects on GABAergic synaptic transmission.
SSTR and GABABR signaling pathways converge on GIRK channel activation in cortical PCs
We and others have previously shown that SOM leads to a reduction in the excitability of projection neurons (Tallent and Siggins, 1997; Hu et al., 2017; Riedemann and Sutor, 2019; Brockway et al., 2023). We investigated here whether SOM acted directly on L2/3 PCs and which SSTR conveyed the previously observed loss in excitability. Therefore, we monitored SOM-induced changes in holding current as an indirect read-out of excitability. We clamped L2/3 PCs at a holding potential of −60 mV and added SOM to the bath after a baseline recording lasting for 5–10 min. In agreement with previous data by us and others (Meis et al., 2005; Hu et al., 2017; Riedemann and Sutor, 2019), SOM induced an outward current that could not be blocked by preincubation with TTX, indicating a direct postsynaptic effect. Preincubation of cells either with Cyclosom or with CYN154806 significantly reduced the SOM-induced outward current (Fig. 9A,B). Previous studies have reported that the GABABR agonist baclofen also induces an outward current in neurons (Howe et al., 1987; Sodickson and Bean, 1996). Therefore, we compared the magnitude of the SOM- with that of the baclofen-mediated outward current. To this end, we applied either SOM or baclofen or a sequential application of either agonist (Fig. 9C). We observed the following effects: (1) Baclofen induced a significantly larger outward current compared with SOM (Fig. 9C,D), indicating a higher intrinsic efficacy of the drug. (2) SOM failed to induce an additional outward current after the pretreatment of slices with baclofen. (3) In contrast, baclofen was still able to induce an additional outward current in PCs that had been pre-exposed to SOM (Fig. 9C,D).
SOM induces GIRK channel activation in L2/3 PCs. A, Representative current traces of L2/3 PCs either treated with SOM alone or in combination with TTX, CYN154806, or Cyclosom (as indicated). B, The dot plot depicts the amplitudes of the SOM-induced outward currents under the different recording conditions shown in A. The mean outward current is depicted as a line: SOM, 33.87 ± 3.36 pA (n = 25, 12 animals, 6♂, 6♀; mean age, P27); SOM + TTX, 28.47 ± 2.74 pA (n = 17, 9 animals, 5♂, 4♀; mean age, P29); SOM + CYN, 6.73 ± 1.88 pA (n = 7, 5 animals, 2♂, 3♀; mean age, P29); SOM + Cylcosom, 13.69 ± 2.85 pA (n = 9, 4 animals, 3♂, 1♀; mean age, P27); SOM versus SOM + CYN, p < 0.001; SOM versus SOM + Cyclosom, p < 0.01; SOM + TTX versus SOM + CYN, p < 0.01; SOM + TTX versus SOM + Cyclosom, p < 0.05; one-way ANOVA with Tukey's multiple-comparisons test. C, Concatenated current trace of a PC receiving voltage ramps at a frequency of 0.1 Hz after sequential addition of SOM and baclofen to the bath. D, The dot plot shows the drug-induced outward currents. The mean outward current is shown as a line: SOM, 33.87 ± 3.36 pA (n = 25, 12 animals, 6♂, 6♀; mean age, P27); SOM + Bac, 86.96 ± 13.56 pA (n = 7, 5 animals, 2♂, 3♀; mean age, P24); Bac, 74.4 ± 8.78 pA (n = 7, 5 animals, 2♂, 3♀; mean age, P31); Bac + SOM, 78.2 ± 10.8 pA (n = 6, 4 animals, 2♂, 2♀; mean age, P28). SOM versus Bac, p < 0.001; SOM versus SOM + Bac, p < 0.001; SOM versus Bac + SOM, p < 0.001; one-way ANOVA with Tukey's multiple-comparisons test. E, Left panel, Voltage command ranging from −120 to −20 mV. Right panel, Representative single voltage-ramp–induced current traces (average of 5) of a recording from a PC before exposure to SOM (CON, black), after bath application of SOM (SOM, red), and after addition of baclofen (SOM + baclofen, gray). F, Pooled I–V relationship of drug-induced currents shows that the drug-induced current reverses between −90 and −80 mV. G, I–V relationship of PCs recorded in the presence (BaCl2, black) or absence of BaCl2 (SOM, red). Pretreatment with BaCl2 blocked the SOM-induced outward current.
By applying a ramp protocol (ranging from a membrane holding potential of −120 to −20 mV), we then confirmed that the SOM- and baclofen-mediated outward currents were driven by an increased potassium conductance (Howe et al., 1987; Sodickson and Bean, 1996; Luscher et al., 1997; Sun et al., 2002; Wang et al., 2010; Bou Farah et al., 2016; Gerrard et al., 2018; Riedemann and Sutor, 2019), as the drug-induced current displayed a reversal potential ranging between −80 and −90 mV and was blocked by bath perfusion with barium chloride (30 µM; Fig. 9E–G).
In summary, these results show that GABABR and SSTR signaling pathways both converge on GIRK channel activation and that SOM-induced GIRK channel activation might lead to its desensitization and hence a postsynaptic depression of GABABR-mediated currents in the apical dendrites of L2/3 PCs.
SOM increases spontaneous GABAergic transmission onto L2/3 PCs by a differential effect on SOM-INs and L1-INs
We tested next whether SOM had any effect on spontaneous GABAergic synaptic transmission onto L2/3 PCs. Recordings were performed using a high-chloride internal solution resulting in a chloride equilibrium potential close to 0 mV, and spontaneous inhibitory postsynaptic currents (sIPSCs) were isolated by the addition of NBQX and D-AP5 to the bath. We analyzed the frequency and amplitude of sIPSCs in the absence or presence of SOM and found that bath application of SOM had no effect on the sIPSC amplitude but, surprisingly and in contrast to our hypothesis that SOM blocks presynaptic GABA release, significantly increased the sIPSC frequency (Fig. 10A–F). However, SOM had no impact on the mIPSC amplitude or frequency suggesting an activity-dependent mechanism (Fig. 10G–L). Therefore, to better understand the underlying mechanisms of SOM-induced increases in spontaneous GABAergic transmission, we tested whether SOM activated SOM-INs and L1-INs, both of which are located presynaptically to L2/3 PCs.
SOM increases the sIPSC frequency in L2/3 PCs. A, Top panel, Representative current trace of a PC under control condition showing sIPSCs. Bottom panel, Expanded current trace of the corresponding trace in the top panel. B, Top panel, Individual current trace of the same PC shown in A after bath application of SOM. Bottom panel, Expanded current trace of the corresponding trace in the top panel. C, Cumulative probability plot of the sIPSC frequency of cell shown in A and B under control condition (black) and after bath application of SOM (red). D, The line plot depicts the mean sIPSC frequencies before and after addition of SOM. Mean sIPSC frequency control, 3.50 ± 0.74 Hz; mean sIPSC frequency SOM, 3.72 ± 0.75 Hz; p = 0.0394, Wilcoxon signed-rank test (n = 18, 8 animals, 5♂, 3♀; mean age, P26). E, Average sIPSC trace of all events detected under control condition (black) and after SOM (red) in the same PC shown in A and B. F, The line plot depicts the amplitudes of all sIPSCs recorded before and after SOM bath application. Mean sIPSC amplitude control, 33.94 ± 2.37 pA; mean sIPSC amplitude SOM, 33.96 ± 2.61 pA; p = 0.992, paired t test test (n = 18, 8 animals, 5♂, 3♀; mean age, P26). G, Top panel, Representative mIPSC recordings under control condition. Bottom panel, Expanded current trace of the corresponding trace shown in the top panel. H, Top panel, Representative mIPSC recordings of the same neuron as in G in the presence of SOM. Bottom panel, Expanded current trace of the corresponding trace in the top panel. I, Cumulative probability plot of the mIPSC frequency of cell shown in G and H under control condition (black) and after SOM exposure (red). J, The line plot depicts the mean mIPSC frequencies of all cells before and after addition of SOM. Mean mIPSC frequency control, 2.12 ± 0.40 Hz; mean mIPSC frequency SOM, 2.04 ± 0.37 Hz; p = 0.46, paired t test (n = 14, 5 animals, 2♂, 3♀; mean age, P28). K, Average mIPSC trace of all events detected under control condition (black) and after addition of SOM to the bath (red) in the same neuron shown in G and H. L, The line plot shows the amplitudes of all mIPSCs recorded before and after SOM bath application. Mean mIPSC amplitude control, 29.28 ± 2.79 pA; mean mIPSC amplitude SOM, 30.02 ± 2.31 pA; p = 0.52, paired t test (n = 14, 5 animals, 2♂, 3♀; mean age, P28). M–O, Comparison of SOM-induced outward currents recorded in an L2/3 PC (M), a SOM-IN (N), and an L1-IN (O). P, The dot plot shows the amplitudes of SOM-induced outward currents. Amplitude outward current PC, 33.87 ± 3.36 pA (n = 25, 12 animals, 6♂, 6♀; mean age, P27); SOM-IN, 2.35 ± 0.66 pA (n = 29, 15 animals, 10♂, 5♀; mean age, P30); L1-IN, 14.93 ± 3.65 pA (n = 19, 12 animals, 6♂, 6♀; mean age, P31). PC versus SOM-IN, p < 0.001; PC versus L1-IN, p < 0.001; L1-IN versus SOM-IN, p < 0.01; one-way ANOVA with Tukey's multiple-comparisons test.
We compared SOM-induced holding currents in L2/3 PCs, SOM-INs, and L1-INs. Unexpectedly, SOM failed to induce a detectable outward current in SOM-INs, whereas it generated an outward current in L1-INs with an amplitude significantly smaller compared with that of L2/3 PCs (Fig. 10M–P).
In addition, we analyzed whether SOM had any effect on evoked AP discharge in the different cell types by injecting a suprathreshold depolarizing current step into the cells before and after bath application of SOM. We found that the SOM-induced effects on evoked AP discharge were significantly different between L2/3 PCs and SOM-INs and between L1-INs and SOM-INs, supporting the above finding that SOM does not negatively affect the excitability of SOM-INs (Fig. 11A–D). Moreover, we observed that SOM increased spontaneous AP discharge in ∼50% of SOM-INs, but only in a minor fraction of L1-INs or PCs (Fig. 11E), suggesting that a subpopulation of SOM-INs is responsible for the SOM-induced increase in sIPSC frequency observed in L2/3 PCs.
Cell-type–specific effect of SOM on the excitability of L2/3 PCs, SOM-INs, and L1-INs. A–C, Single traces of evoked APs upon injection of a suprathreshold depolarizing current step into a PC (A), a SOM-IN (B), and an L1-IN (C) before (black) and after (red) addition of SOM. D, Summary of the SOM-induced effect on evoked AP firing as dot plot (mean normalized AP discharge (% of control) PCs, 69.1 ± 9.1 (n = 20, 9 animals, 5♂, 4♀; mean age, P32); SOM-IN, 111.3 ± 8.9 (n = 29, 15 animals, 10♂, 5♀; mean age, P30); L1-IN, 73.9 ± 9.9 (n = 33, 20 animals, 12♂, 8♀; mean age, P47). SOM-IN versus PC, p < 0.05; SOM-IN versus L1-IN, p < 0.01; one-way ANOVA with Dunn's multiple-comparisons test. E, Raster plot showing spontaneous AP discharges in six individual SOM-INs before and after SOM exposure (indicated by red boxes).
The possibility of endogenously released SOM and its effect on spontaneous inhibitory inputs onto PCs was tested by recording sIPCS before and after the addition of Cyclosom or CYN154806 to the bath. In agreement with the hypothesis that SOM depresses presynaptic GABA release, we found that either blocker caused a significant increase in the sIPSC frequency in PCs (Fig. 12), suggesting that blockage of endogenous SOM release enhances GABAergic transmission in a subset of GABAergic INs that are located presynaptic to PCs and whose activity is most likely controlled by SOM-INs.
SSTR inhibition increases inhibitory synaptic transmission in L2/3 PCs. A, Top panel, Depiction of representative sIPSC recording from an L2/3 PC under control condition (CON). Bottom panel, Expanded current trace of the corresponding trace shown in the top panel. B, Top panel, Depiction of sIPSC recording from the same cell shown in A after addition of cyclosomatostatin (Cyclosom) to the bath. Bottom panel, Expanded current trace of the corresponding trace shown in the top panel. C, Cumulative probability plot of sIPSC frequency of cell shown in A and B under control condition (black) and after Cyclosom exposure (yellow). D, The line plot depicts the sIPSC frequencies of all recorded PCs before and after addition of Cyclosom. Mean sIPSC frequency control, 4.52 ± 1.13 Hz; mean sIPSC frequency Cyclosom, 6.75 ± 1.68 Hz; p = 0.0355, paired t test (n = 9, 3 animals, 2♂, 1♀, mean age = P24). E, Top panel, Representative 30 s current trace of sIPSC recording under control condition (CON). Bottom panel, Expanded current trace of the corresponding trace shown in the top panel. F, Bottom panel, Representative 30 s current trace of the same cell depicted in E showing sIPSC recording after addition of CYN154806 (CYN). Bottom panel, Expanded current trace of the corresponding trace shown in the top panel. G, Cumulative probability plot of the sIPSC frequency of cell shown in E and F under control condition (black) and after CYN treatment (yellow). H, Summarized data as a line plot. Mean sIPSC frequency control, 3.32 ± 0.88 Hz; mean sIPSC frequency CYN, 4.45 ± 1.30 Hz, p = 0.0051, Wilcoxon matched-pair signed-rank test (n = 11, 5 animals, 2♂, 3♀; mean age, P30).
Next, we recorded pharmacologically isolated sIPSCs in SOM-INs and L1-INs before and after addition of SOM to the bath. In agreement with the above hypothesis that blockage of endogenous SOM release disinhibits GABergic INs that are located presynaptic to L2/3 PCs, we found that SOM exposure significantly increased the sIPSC frequency in L1-INs (Fig. 13A–D) but decreased the sIPSC frequency in SOM-INs (Fig. 13G–J). In addition, SOM had no effect on the sIPSC amplitude in L1-INs but increased the latter in SOM-INs (Fig. 13E,F,K,L).
SOM increases the sIPSC frequency in L1-INs but decreases the sIPSC frequency in SOM-INs. A, Top panel, Representative 30 s current trace of an L1-IN under control condition showing sIPSCs. Bottom panel, Expanded current trace of the corresponding current trace shown in the top panel. B, Top panel, Individual 30 s current trace of the same L1-IN as in A after SOM exposure. Bottom panel, Expanded current trace of the corresponding trace in the top panel. C, Cumulative probability plot of the sIPSC frequency of cell shown in A and B under control condition (black) and after SOM (red). D, The line plot depicts the sIPSC frequencies before and after addition of SOM. Mean sIPSC frequency control, 1.38 ± 0.47 Hz; mean sIPSC frequency SOM, 1.62 ± 0.56 Hz; p = 0.0049, Wilcoxon signed-rank test (n = 12, 7 animals, 3♂, 4♀; mean age, P85). E, Average sIPSC trace of all events detected under control condition (black) or after SOM treatment (red) in the same neuron shown in A and B. F, The line plot depicts the amplitudes of all sIPSCs recorded in L1-INs before and after SOM treatment. Mean sIPSC amplitude control, 34.81 ± 3.58 pA; mean sIPSC amplitude SOM, 34.93 ± 2.80 pA; p = 0.9487, paired t test test (n = 12, 7 animals, 3♂, 4♀; mean age, P85). G, Top panel, Representative current recording of a SOM-IN under control condition. Bottom panel, The current trace of the corresponding current trace shown in the top panel. H, Top panel, Individual current trace of the same SOM-IN as in G after bath application of SOM. Bottom panel, Expanded current trace of the corresponding trace in the top panel. I, Cumulative probability plot of the sIPSC frequency of cell depicted in G and H under control condition (black) and after bath application of SOM (red). J, The line plot depicts the sIPSC frequencies before and after addition of SOM. Mean sIPSC frequency control, 1.52 ± 0.31 Hz; mean sIPSC frequency SOM, 1.24 ± 0.29 Hz; p = 0.0084, Wilcoxon signed-rank test (n = 16, 7 animals, 5♂, 2♀; mean age, P70). K, Average sIPSC trace of all events detected under control condition (black) or after SOM treatment (red) in the same neuron depicted in G and H. L, The line plot depicts the amplitudes of sIPSCs recorded before and after SOM bath application. Mean sIPSC amplitude control, 25.37 ± 2.02 pA; mean sIPSC amplitude SOM, 29.31 ± 2.11 pA; p = 0.0157, paired t test test (n = 16, 7 animals, 5♂, 2♀; mean age, P70).
Taken together, these results suggest that the observed SOM-induced increase in sIPSC frequency in PCs is—at least partially—mediated by an increased activity of presynaptic SOM-INs.
In order to further corroborate the hypothesis that SSTR antagonism disinhibits GABAergic INs located presynaptic to L2/3 PCs, we determined the synaptic coupling probability within the circuitry between L1-INs and L2/3 PCs and SOM-INs.
SOM increases correlated activity between PCs and decreases correlation between SOM-INs and PCs
Dual recordings between pairs of L2/3 PCs showed a synaptic coupling probability of 13% (15 out of 113 pairs recorded, Fig. 14A,B,K). In addition, we observed disynaptic inhibition of 8 out of 15 synaptically coupled L2/3 PCs (Fig. 14B). Importantly, this type of lateral inhibition is preferentially mediated by SOM-INs (Silberberg and Markram, 2007). In agreement with that, we found a coupling probability of 19% from L2/3 PC onto SOM-IN (6 out of 32 pairs recorded, Fig. 14C,D,K), and synaptic coupling from SOM-IN to PC was detected in 13% of recorded pairs (4 out of 31, Fig. 14E,F,K). Given the fact that SOM-INs innervate the distal dendrites of PCs (Lee et al., 2015; Nigro et al., 2018), the observed unitary postsynaptic response in L2/3 PCs after SOM-IN activation was rather small, and the true synaptic coupling rate is likely to be higher. The coupling probability of L1-INs to PCs was 36% (4 out of 11 recorded pairs) and PCs exhibited a coupling ratio of 18% (2 out of 11 pairs) with L1-INs (Fig. 14G,H,K). These results confirm that either inhibitory IN type is located presynaptic to L2/3 PCs.
Synaptic connectivity between L2/3 PCs, L2/3 SOM-INs, and L1-INs of the aCC. A, Schematized recording setup. B, Presynaptic AP discharge in presynaptic PC evokes uEPSPs in postsynaptic PC. Disynaptic inhibition is observed when the postsynaptic PC is held at different membrane potentials to disclose different components of the postsynaptic response. Individual responses are depicted in light gray; the mean postsynaptic response is shown in black. C, Schematized recording situation between PC and SOM-IN. D, APs in presynaptic PC evoke uEPSPs in postsynaptic SOM-IN. Individual responses are depicted in light gray; the mean postsynaptic response is shown in black. E, Paired recording between a presynaptic SOM-IN and a postsynaptic PC. F, AP discharge in presynaptic SOM-IN evokes an inhibitory response in postsynaptic PC with a membrane potential of −55 mV. Individual responses are depicted in light gray; the mean postsynaptic response is shown in black. G, Recording situation of reciprocally connected L1-IN with PC. H, APs in presynaptic L1-IN evoke postsynaptic potentials in PC and vice versa. Individual responses are depicted in light gray; the mean postsynaptic response is shown in black. I, Paired recordings between L1-IN and SOM-IN. J, Presynaptic APs in SOM-IN evoke postsynaptic responses in L1-IN. Individual responses are depicted in light gray; the mean postsynaptic response is shown in black. K, Heatmap showing the overall connectivity between L2/3 PCs, L2/3 SOM-INs, and L1-INs of the aCC (PC-PC, n = 113, 68 animals, 35♂, 33♀; mean age, P38; SOM-IN-PC, n = 32, 27 animals, 16♂, 11♀; mean age, P45; SOM-IN-L1-IN, n = 20, 16 animals, 9♂, 7♀; mean age, P43; PC-L1-IN, n = 11, 6 animals, 3♂, 3♀; mean age, P33).
In order to test the possibility of disinhibition, we next recorded from pairs of SOM-INs and L1-INs. SOM-INs and L1-INs exhibited a coupling probability of SOM-IN to L1-IN of 20% (4 out of 20, Fig. 14I–K) while the coupling probability of L1-INs to SOM-INs was 5% (1 out of 20), suggesting the existence of disinhibitory circuit motifs in the supragranular layers of the aCC.
Taken together, these results strongly indicate that SOM increases spontaneous GABAergic transmission onto L2/3 PCs by differentially acting on two types of dendrite-targeting inhibitory INs that are mutually connected. In agreement with this, we found that either IN type received spontaneous inhibitory input at a comparable frequency and amplitude (Fig. 15).
Comparison of the magnitude of inhibitory inputs onto L2/3 PCs, SOM-INs, and L1-INs. A, A scatter plot showing the sIPSC frequencies recorded in PCs (n = 38, 16 animals, 9♂, 7♀; mean age, P27), SOM-INs (n = 16, 7 animals, 5♂, 2♀; mean age, P70), and L1-INs (n = 12, 7 animals, 3♂, 4♀; mean age, P85). L2/3 PCs received significantly more inhibitory inputs compared with either IN type. Mean sIPSC frequency PC, 3.69 ± 0.5 Hz; mean sIPSC frequency SOM-IN, 1.52 ± 0.31 Hz; mean sIPSC frequency L1-IN, 1.38 ± 0.47 Hz. PC versus SOM-IN, p < 0.05; PC versus L1-IN, p < 0.01; one-way ANOVA with Dunn's multiple-comparisons test. B, A scatter plot showing the sIPSC amplitudes in PCs, SOM-INs, and L1-INs. Mean sIPSC amplitude PC, 34.1 ± 2.1 pA; mean sIPSC amplitude SOM-IN, 25.4 ± 2.0 pA; mean sIPSC amplitude L1-IN, 34.8 ± 3.6 pA. PC versus SOM-IN, p < 0.05, one-way ANOVA with Tukey's multiple-comparisons test.
We next asked ourselves whether the SOM-dependent synaptic modulation described above had any influence on network activity. To this end, we initially characterized synaptic inputs onto these three different cell types in the absence of pharmacological blockers and found that the level of detected synaptic activity was highest in SOM-INs. Likewise, the mean amplitude of spontaneous postsynaptic potentials (sPSPs) was highest in SOM-INs and lowest in L2/3 PCs (Fig. 16). Next, we tested whether SOM had any influence on sPSP frequency and/or kinetics in the absence of any pharmacological blockers. SOM had no impact on the sPSP properties (frequency, amplitude, duration, decay time) in PCs (Fig. 16C,D). In SOM-INs, SOM did not alter sPSP frequency, sPSP amplitude, or sPSP duration but significantly prolonged the sPSP decay time. In L1-INs, SOM did not modulate sPSP frequency but significantly reduced sPSP amplitude while increasing sPSP duration and sPSP decay time (Fig. 16G,H).
Summary of the frequency and kinetics of postsynaptic potentials in L2/3 PCs, L2/3 SOM-INs, and L1-INs. A, A dot plot showing the frequencies of spontaneous postsynaptic potentials (sPSPs) recorded in PCs (n = 50, 19 animals, 9♂, 10♀; mean age, P43), SOM-INs (n = 34, 16 animals, 10♂, 6♀; mean age, P64), L1-INs (n = 31, 18 animals, 11♂, 7♀; mean age, P53). L2/3 PCs received synaptic inputs at the lowest frequency (mean sPSP frequency L2/3 PCs, 0.45 ± 0.08 Hz; SOM-INS, 1.75 ± 0.24 Hz; L1-INs, 1.60 ± 0.29 Hz). Comparison frequency PC versus L1-IN: p < 0.001; Kruskal–Wallis test with Dunn's multiple-comparisons test. Comparison frequency PC versus SOM-IN: p < 0.001, Kruskal–Wallis test with Dunn's multiple-comparisons test). B, A dot plot showing the amplitudes of the spontaneous postsynaptic potentials (sPSPs) recorded in PCs, SOM-INs, and L1-INs. Mean sPSP amplitude L2/3 PCs: 0.64 ± 0.02 Hz; SOM-INs, 1.08 ± 0.05 Hz; L1-INs, 0.78 ± 0.04 Hz. Comparison frequency PC versus L1-IN: p < 0.05, Kruskal–Wallis test with Dunn's multiple-comparisons test. Comparison frequency PC versus SOM-IN: p < 0.001, Kruskal–Wallis test with Dunn's multiple-comparisons test. Comparison frequency SOM-IN versus L1-IN: p < 0.001, Kruskal–Wallis test with Dunn's multiple-comparisons test. C, Averaged sPSP under control condition (dark gray, CON) and 5 (red, SOM) and 15 (light red, wash) min after the onset of SOM bath application taken from the same L2/3 PC. D, Line plots of sPSP amplitude (CON, 0.64 mV ± 0.02; 5 min SOM, 0.66 mV ± 0.02; 15 min SOM, 0.65 mV ± 0.02), sPSP frequency (CON, 0.45 ± 0.08 Hz; 5 min SOM, 0.48 ± 0.09 Hz; 15 min SOM, 0.58 ± 0.11 Hz), sPSP duration (CON, 29.2 ± 2.1 ms; 5 min SOM, 28.6 ± 2.4 ms; 15 min SOM, 29.6 ± 2.5 ms), and sPSP decay time (CON, 36.9 ± 2.4 ms; 5 min SOM, 36.5 ± 2.8 ms; 15 min SOM, 37.9 ± 3.1 ms) under control condition and 5 and 15 min after the onset of SOM bath application. E, Averaged sPSP under control condition (dark gray, CON) and 5 (red, SOM) and 15 (light red, wash) min after the onset of SOM exposure taken from the same L2/3 SOM-IN. F, Line plots of sPSP amplitude (CON, 1.08 mV ± 0.05; 5 min SOM, 1.05 mV ± 0.04; 15 min SOM, 1.06 mV ± 0.05), sPSP frequency (CON, 1.75 ± 0.24 Hz; 5 min SOM, 1.46 ± 0.21 Hz; 15 min SOM, 1.46 ± 0.19 Hz), sPSP duration (CON, 34.5 ± 2.6 ms; 5 min SOM, 32.7 ± 2.6 ms; 15 min SOM, 36.6 ± 2.9 ms), and sPSP decay (CON, 38.3 ± 2.0 ms; 5 min SOM, 37.1 ± 2.1 ms; 15 min SOM, 42.5 ± 2.8 ms; 5 min vs 15 min SOM, p < 0.05; repeated-measures ANOVA with Tukey's multiple-comparisons test) time under control condition and 5 and 15 min after the onset of SOM bath application. G, Averaged sPSP under control condition (dark gray, CON) and 5 (red, SOM) and 15 (light red, wash) min after the onset of SOM application taken from the same L1-IN. Right panel, Line plots of sPSP amplitude (CON, 0.78 mV ± 0.04; 5 min SOM, 0.73 mV ± 0.04; 15 min SOM, 0.74 mV ± 0.04; CON vs 5 min SOM, p < 0.05; CON vs 15 min SOM, p < 0.05, repeated-measures ANOVA with Tukey's multiple-comparisons test), sPSP frequency (CON, 1.60 ± 0.29 Hz; 5 min SOM, 1.66 ± 0.27 Hz; 15 min SOM, 2.00 ± 0.32 Hz), sPSP duration (CON, 22.4 ± 1.9 ms; 5 min SOM, 22.3 ± 2.0 ms; 15 min SOM, 26.3 ± 2.5 ms; 5 min vs 15 min SOM, p < 0.05; Friedman test with Dunn's multiple-comparisons test), and sPSP decay time (CON, 27.6 ± 2.5 ms; 5 min SOM, 27.0 ± 3.0 ms; 15 min SOM, 32.6 ± 3.5 ms; 5 min vs 15 min SOM, p < 0.01; Friedman test with Dunn's multiple-comparisons test) under control condition and 5 and 15 min after the onset of SOM application.
The impact of SOM on correlated neuronal activity between the above cell types was tested by dual recordings. After a baseline recording of at least 5 min, SOM was added to the bath for 5–7 min, and the recording continued for at least 15 min. We found that SOM exposure increased correlated activity between pairs of PCs (Fig. 17A–C). In pairs of L1-INs and PCs, we observed that correlated activity remained stable after initial SOM exposure but decreased 15 min after the onset of SOM bath application (Fig. 17D–F). In agreement with the fact that different IN subtypes are driven by discrete ensembles of PCs (Karnani et al., 2016), we could not detect correlated activity between pairs of SOM-INs and L1-INs under control conditions, and SOM exposure had no effect on correlated activity between SOM-INs and L1-INs (Fig. 17G–I).
SOM increases correlated activity between PCs and decreases the correlation between INs and PCs. A, Scheme for recording pairs of L2/3 PCs. B, Top panel, Voltage traces of dual recordings of two PCs (gray and black) before (CON) and after bath application of SOM (SOM). Bottom panel, Voltage traces during baseline recording and ∼5 min after addition of SOM to the bath. C, The line plot shows Pearson's linear correlation coefficients under control condition and 5 and 15 min after the onset of SOM application. Mean r control, −0.04 ± 0.05; mean r 5 min SOM, 0.13 ± 0.05; mean r 15 min SOM −0.05 ± 0.05. Control versus 5 min SOM, p = 0.015, paired t test; control versus 15 min SOM, p = 0.339, paired t test; 5 min SOM versus 15 min SOM, p = 0.281, paired t test (n = 18, 11 animals, 5♂, 6♀; mean age, P43). D, Recording pairs of L2/3 PC and L1-IN. E, Top panel, Voltage trace of an L1-IN (gray) and a PC (black) before and after application of SOM. Bottom panel, Voltage traces during baseline recording and ∼5 min after addition of SOM to the bath. F, The line plot shows the mean Pearson's linear correlation coefficients under control condition and 5 and 15 min after the onset of SOM bath application. Mean r control, 0.15 ± 0.05; mean r 5 min SOM, 0.20 ± 0.05; mean r 15 min SOM, −0.03 ± 0.05. Control versus 5 min SOM, p = 0.368, paired t test; control versus 15 min SOM, p = 0.0081, paired t test; 5 min SOM versus 15 min SOM, p = 0.005, paired t test (n = 10, 5 animals, 2♂, 3♀; mean age, P31). G, Recording pairs of SOM-IN and L1-IN. H, Top panel, Voltage trace of an L1-IN (gray) and a SOM-IN (black) before and after application of SOM. Bottom panel, Voltage traces during baseline recording and ∼5 min after addition of SOM to the bath. I, The line plot shows the mean Pearson's linear correlation coefficients under control condition and 5 and 15 min after the onset of SOM bath application. Mean r control, −0.01 ± 0.04; mean r 5 min SOM, 0.09 ± 0.05; mean r 15 min SOM, −0.03 ± 0.05. Control versus 5 min SOM, p = 0.258, paired t test; control versus 15 min SOM, p = 0.771, paired t test; 5 min SOM versus 15 min SOM, p = 0.184, paired t test (n = 8, 5 animals, 4♂, 1♀; mean age, P31). J, Recording pairs of SOM-IN and PC. K, Top panel, Voltage trace of a SOM-IN (gray) and a PC (black) before and after application of SOM. Right panel, Voltage traces during baseline recording and ∼5 min after addition of SOM to the bath. L, The line plot shows the Pearson's linear correlation coefficients under control condition and 5 and 15 min after the onset of SOM bath application. Mean r control, 0.35 ± 0.06; mean r 5 min SOM, 0.13 ± 0.08; mean r 15 min SOM, 0.13 ± 0.07. Control versus 5 min SOM, p = 0.037, paired t test; control versus 15 min SOM, p = 0.027, paired t test; 5 min SOM versus 15 min SOM, p = 0.998, paired t test (n = 11, 9 animals, 6♂, 3♀; mean age, P35).
The correlated activity between pairs of SOM-INs and PCs decreased during SOM exposure and remained decreased thereafter (Fig 17J–L).
Discussion
We have shown here that SOM, by acting pre- and postsynaptically, modulates the neuronal circuitry between L1-INs, L2/3 PCs, and L2/3 SOM-INs in the aCC. This study revealed a differential effect of SOM on these three different cell types, and we could show that SOM failed to reduce the intrinsic excitability of SOM-INs despite reducing that of PCs and L1-INs. In agreement with this finding, we detected increased sIPSC frequencies in PCs and L1-INs that are most likely mediated by presynaptic SOM-INs and paired recordings between SOM-INs and either PCs or L1-INs confirmed synaptic connections from SOM-INs onto either cell type. At the same time, we observed that SOM, by depressing presynaptic GABA release, decreased the frequency of inhibitory inputs onto SOM-INs. SOM-mediated inhibition of presynaptic GABA release was further confirmed by the finding that SOM exposure caused a depression of evoked GABAAR-mediated postsynaptic responses in L2/3 PCs whereas focal GABA application in the presence of SOM had no such effect. In addition, we report here that SOM leads to a depression of GABABR-mediated synaptic transmission in L2/3 PCs, and the data presented here suggest that this decrease is the consequence of a combined pre- and postsynaptic effect on resp. in PCs themselves, due to (1) SOM-induced modulation of presynaptic GABA release and (2) GIRK channel inactivation or desensitization in PCs in response to continuous SOM exposure. Desensitization of SSTRs in response to continuous agonist exposure cannot be entirely ruled out; however, previous work by us points to desensitization of GIRK channels rather than SSTRs upon ligand binding (Riedemann and Sutor, 2019). Importantly, a similar loss of postsynaptic GABABR-mediated responses was reported in a GIRK knock-out mouse (Luscher et al., 1997), driving the hypothesis that the combined activation of SSTR and GABABR reduces postsynaptic GIRK channel activity (Turecek et al., 2014).
Methodological considerations
There are some limitations to the study that should be kept in mind when interpreting the results of the present work. First, the age range of animals used in this study is rather large, and endogenous SOM release and SSTR expression might vary across this age range. Second, PV-INs represent a major source of inhibitory inputs to PCs. We cannot entirely exclude that SOM-dependent modulations of sIPSCs are the result of a SOM-induced change in (somatic and synaptic) PV-IN properties. It is however suggested here that the excitability of PV-INs is reduced by SOM as (1) they receive reduced excitatory drive upon SSTR activation (Song et al., 2020) and (2) possibly become hyperpolarized after SOM treatment (Brockway et al., 2023). The latter study did not differentiate between IN types; however, there is a higher likelihood of sampling PV-INs in comparison with SOM-INs or VIP-INs given the higher absolute numbers of the former. Third, we recorded from SOM-INs using the FVB-Tg(GadGFP)45704Swn/J mouse line (“GIN”) that labels a subpopulation of SOM-INs. We have previously characterized these SOM-INs in the aCC in great detail (Riedemann et al., 2016b, 2018) and used them as a proxy of all SOM-INs in the present work. We are aware that SOM-IRES-Cre mouse lines label subtypes of SOM-INs that cannot be found in the mouse line used by us (Hu et al., 2013; Nassar et al., 2015; Large et al., 2016; Nigro et al., 2018; Hostetler et al., 2023). This SOM-IN subtype (“quasi fast-spiking” SOM-INs) that is absent in the GIN mouse line is mostly, but not entirely, confined to the deeper cortical layers and only shows sparse expression in L2/3. We can therefore not entirely rule out that the effects described in the present work can only be attributed to a specific subtype of SOM-INs. Fourth, we did not use a genetic mouse line to label specific subpopulations of L1-INs. Rather, we recorded unbiasedly from all neurons present in L1. Analysis of SOM-induced changes in L1-INs properties should be refined by specifically targeting defined subsets of L1-INs.
Continuous SOM exposure reduces GABABR-mediated signaling and modulates the synaptic strength between INs and PCs
SOM and baclofen both activate GIRK channels in the postsynaptic cell. We found here that the amplitude of the SOM-induced outward current was larger in PCs compared with L1-INs and not detectable in SOM-INs. Cell-type–specific baclofen effects were also reported on different hippocampal IN types, and in agreement with our findings, baclofen did not induce an appreciable outward current in hippocampal SOM-INs (Booker et al., 2017, 2018), leading to an uncoupling of SOM-INs from the active network.
The SOM-induced actions on L2/3 PCs seem to be primarily mediated by the activation of SSTR2, consistent with the previous work by us and others (Stumm et al., 2004; Meis et al., 2005; Riedemann and Sutor, 2019). However, a detailed analysis of SSTR expression on GABAergic INs revealed prominent expression of SSTR1, SSTR3, SSTR4, and SSTR5 on PV-, SOM-, and VIP-INs. Therefore, the activation of other SSTR subtypes cannot be ruled out (Lukomska et al., 2020). Given that L1-INs are reported to be the primary mediators of GABABR-mediated inhibition in PCs (Olah et al., 2009; Abs et al., 2018; Schulz et al., 2021), SOM-induced GIRK channel inactivation in PCs results in a smaller L1-IN–mediated inhibition of PCs, that is, SOM weakens the synaptic transmission between L1-INs and PCs. SOM-INs, on the other hand, inhibit PCs to a large extent, but not exclusively, via activation of GABAARs (Urban-Ciecko et al., 2015; Abs et al., 2018; Cao et al., 2020; Donato et al., 2023), and we have shown here that SOM preferentially weakens the GABABR-mediated synaptic transmission, that is, it is likely that SOM does not affect the synaptic strength between SOM-INs and PCs. It is therefore suggested that SOM-INs, by activity-dependent release of the peptide, gain control over L1-INs and become the predominant modulator of excitatory information flow between PCs. This hypothesis is supported by the finding that optogenetic silencing of cortical SOM-INs enhances the synaptic transmission between L2 PCs (Urban-Ciecko et al., 2015) and increases their action potential firing rate (Gentet et al., 2012).
On a network level, this could potentially mean that the lateral inhibition and/or feedback inhibitory motifs provided by SOM-INs gain importance over feedforward inhibitory motifs provided by L1-INs upon endogenous release of SOM. Previous studies have suggested that SOM-INs play an important role in modulating top-down inputs into the cortex (Kvitsiani et al., 2013; Zhang et al., 2014; Kim et al., 2016; Naka et al., 2019; Wilmes and Clopath, 2019), and behavioral experiments in mice suggest that either bilateral injection of SOM into the visual cortex and/or optogenetic activation of GABAergic long-range efferents from the (posterior) cingulate to the visual cortex enhance visual discrimination and promote center-surround suppression (Zhang et al., 2014; Song et al., 2020). Likewise, SOM-INs promote sustained spectral surround suppression in excitatory neurons of the auditory cortex (Lakunina et al., 2020). It is therefore suggested that SOM-INs, by endogenous release of the peptide, promote lateral inhibition of PCs. This hypothesis is supported by the fact that SOM-INs contribute to the L1-IN inhibition and recordings from NDNF1 cells suggest that SOM-INs are important inhibitors of the NDNF1-IN activity (Abs et al., 2018). We propose here that there exists a disinhibitory SOM-IN-to-L1-IN circuit that possibly gains importance upon the endogenous release of SOM.
Such a scenario would encompass that top-down inputs into the aCC would preferentially activate L1-INs and that SOM-INs only become recruited upon higher activation. Indeed, input–output analyses in PCs, SOM-INs, and L1-INs after electrical L1 stimulation suggest that L1-INs become recruited well before SOM-INs and PCs. In agreement with such a hypothesis, previous studies using optogenetic stimulation (Abs et al., 2018) and virus tracing techniques (Wall et al., 2016) showed that L1-INs received a strong corticocortical input.
SOM increases correlated activity between L2/3 PCs by modulating dendro-somatic integration
We showed here that SOM exposure resulted in an inhibition of glutamatergic as well as GABAergic synaptic transmission in the aCC. As a result, we found that in an open network, that is, with intact GABAergic and glutamatergic transmission, the overall impact of SOM on synaptic excitability of PCs is balanced. It is rather suggested that SOM, by the mechanisms described above, modulates the dendritic integration in apical dendrites of L2/3 PCs. In support of this hypothesis, it could be shown that the activation of GABABRs in apical dendrites of L5 PCs influences the dendro-somatic synergy between feedback and feedforward inputs (Schulz et al., 2021). This SOM-mediated modulation of dendro-somatic synergy between different presynaptic inputs onto PCs could well lead to the observed enhanced correlated neuronal activity between pairs of PCs. In line with the finding that SOM has differential effects on PCs and L1-INs compared with SOM-INs, we find that SOM decreases the correlated activity between PCs and GABAergic INs, suggesting that SOM-INs and L1-INs become uncoupled from the network. In vivo recordings of L2/3 PCs and SOM-INs of the barrel cortex during quiet wakefulness showed that membrane potential fluctuations of the neighboring PCs are correlated whereas those of PCs and SOM-INs are anticorrelated (Gentet et al., 2012). In contrast to that, we observed a correlated neuronal activity between PCs and SOM-INs under in vitro conditions. Correlated membrane potential fluctuations between L2/3 PCs and SOM-INs were also reported by Neske et al. (2015) in in vitro recordings of barrel cortex neurons, suggesting that synaptic inputs that are no longer present in ex vivo recordings are necessary for the anticorrelated activity between SOM-INs and PCs.
Ultimately, pre- and postsynaptic effects of SOM on other GABAergic IN types such as PV-INs must be investigated to fully understand how SOM influences the cortical microcircuit and corticocortical information processing. Given that SOM-INs are reported to inhibit all other IN types (Jiang et al., 2015; Tremblay et al., 2016; Williams and Riedemann, 2021), differentially modulated pre- and postsynaptic GABAA/BR activation represents an interesting possibility, whereby SOM-INs could regulate GABA release and the strength of synaptic inhibition and inhibitory circuit motifs.
Footnotes
We thank Gabi Horn for her excellent technical assistance. This work was funded by two grants from the Friedrich-Baur Stiftung (04/16, 03/21).
The authors declare no competing financial interests.
- Correspondence should be addressed to Therese Riedemann at therese.riedemann{at}med.uni-muenchen.de.
