Phylogenetic divergence of GABAB receptor signaling in neocortical networks over adult life

Abstract

Cortical circuit activity is controlled by GABA-mediated inhibition in a spatiotemporally restricted manner. GABA_B receptor (GABA_BR) signalling exerts powerful slow inhibition that controls synaptic, dendritic and neuronal activity. But, how GABA_BRs contribute to circuit-level inhibition over the lifespan of rodents and humans is poorly understood. In this study, we quantitatively determined the functional contribution of GABA_BR signalling to pre- and postsynaptic domains in rat and human cortical principal cells. We find that postsynaptic GABA_BR differentially control pyramidal cell activity within the cortical column as a function of age in rodents, but minimally change over adult life in humans. Presynaptic GABA_BRs exert stronger inhibition in humans than rodents. Pre- and postsynaptic GABA_BRs contribute to co-ordination of local information processing in a layer- and species-dependent manner. Finally, we show that GABA_BR signalling is elevated in patients that have received the anti-seizure medication Levetiracetam. These data directly increase our knowledge of translationally relevant local circuit dynamics, with direct impact on understanding the role of GABA_BRs in the treatment of seizure disorders.

Fig. 1. Age- and species-dependent changes in neuronal excitability.

A Example reconstructions of L2/3 and L5 neurons from rat and human neocortex. Inset, voltage response of example cells in response to −500 to +500 pA current injections (100 pA steps, 500 ms duration). B Current-voltage responses of L2/3 (black) and L5 (blue) neurons from 1-month-old (left; L2/3: 25 cells from 14 rats, L5: 25 cells from 13 rats), 6–8-month-old (middle left; L2/3: 35 cells from 20 rats, L5: 31 cells from 20 rats), and 12–14-month-old rats (middle right; L2/3: 13 cells from 5 rats, L5: 11 cells from 5 rats), and human neocortex (right; L2/3: 37 cells from 20 cases, L5: 16 cells from 9 cases). C The slope of the current-frequency (FI) plot between 0 and 500 pA from recorded neurons between layers and age of the same rats (L2/3 vs. L5: p = 2.6 × 10⁻⁹ (1-month), 2.55 × 10⁻⁹ (6–8 months), 0.599 (12–14 months), L2/3: P = 0.0004 (1 vs. 12–14 months), 0.0335 (1 vs. 6–8 months), 0.0191 (6–8 vs. 12–14 months); Tukey post hoc tests). C’ Comparison of FI slopes of L2/3 and L5 PCs in adult rats (6–14 months) and humans. D Membrane capacitance of recorded neurons in rats for the same age groups (L2/3 vs. L5: p = 0.0025 (1-month), 0.0006 (6–8-month), 0.0364 (12–14 months), L2/3: P = 0.0729 (1 vs. 12–14 months), 0.0029 (1 vs. 6–8 months), 0.834 (6–8 vs. 12–14 months); Tukey post hoc tests). D’ Comparison of membrane capacitance L2/3 and L5 PCs in adult rats (6–14 months) and humans. All data is shown as either mean ± SEM (B) or boxplots (25–75% range) with the median indicated and whiskers indicating maximum/minimum ranges (C and D). All data are overlain by responses of individual cells (circles). Statistical test results are shown above graphs from 2-way ANOVA (2-sided, B) and LMM models (C and D), with results from post-hoc 2-sided Tukey tests: *P < 0.05, **P < 0.01; ***P < 0.001.

Fig. 2. Pharmacologically activated postsynaptic GABA_BR currents in rat and human neocortex.

A Whole-cell currents from L2/3 (black) and L5 (blue) PCs from 1-, 6–8-, 12–14-month-old rats, and human neocortex following bath-application of baclofen and CGP. Number of cells and baseline current (grey line) indicated. B Peak baclofen current from L2/3 and L5 in 1-month (L2/3: 10 cells/8 rats; L5: 11 cells/5 rats), 6–8-month (L2/3: 22 cells/13 rats; L5: 19 cells/13 rats), and 12–14-month-old (L2/3: 12 cells/5 rats; L5: 11 cells/5 rats) rats (L2/3 vs. L5: p = 7.0 × 10⁻⁶ (1-month), 0.317 (6–8-month), 0.105 (12–14-month). B’ Peak baclofen current of adult rats and humans (L2/3: 34 cells/19 cases; L5: 12 cells/8 cases). C Baclofen current-density for rat neurons (L2/3 vs. L5: p = 0.0002 (1-month), 0.282 (6–8-month), 0.762 (12–14-month). C’ Baclofen current density for adult rat and human neurons. D Scatter-plot of baclofen current density for rat L2/3 (black) and L5 (blue) and age (months) showing linear regression (solid lines) with 95% confidence intervals (dashed lines) and R² and P-values (F-test). E Scatter-plot of baclofen current density from human L2/3 (black) and L5 (blue) plotted against age (years) in the same format. F Time-course plots of L2/3 (black) and L5 (blue) PCs from 1-month (L2/3: 8 cells/5 rats; L5: 6 cells/3 rats) and 6–8-month (L2/3: 8 cells/5 rats; L5: 7 cells/5 rats) following bath-application of CGP. G GABA_BR-mediated tonic-currents plotted for rat PCs. Data is shown as mean ± SEM (A, F), or boxplots (25–75% range) with median indicated and whiskers indicating maximum/minimum ranges (B, C, G) or individual data (filled circle) as a scatter-plot (D, E). Box-plots are overlain by responses of individual cells (open circles). Statistics shown from LMM (B, C, G) or Pearson correlation (D, E), with comparisons from 2-sided Tukey post-hoc tests: *P < 0.05, **P < 0.01; ***P < 0.001. Source data are provided as a Source Data file.

Fig. 3. GABA release activates GABA_BRs with age-dependent differences.

A Immunofluorescence for GABA_B1 (green) and NeuN (magenta) in rat and human neocortex, with layers (dashed lines) and depth (arrow). Scale = 100 μm. B GABA_B1 intensity normalised to cortical depth for 1 (black, n = 5), 6–8 (grey, n = 3) and 12–14-month-old (grey, n = 3) rats, with SEM (dotted lines). C GABA_B1 intensity for human (magenta, 6 patients) and adult rat (black, 6–14-months). D Schematic of IPSC recordings, with control and CGP from L2/3 (black) and L5 (blue). E IPSC amplitudes in 1 (L2/3: 19 cells/11 rats; L5: 17 cells/8 rats), 6–8 (L2/3: 22 cells/13 rats; L5: 17 cells/13 rats), and 12–14-month-old (L2/3: 12 cells/5 rats; L5: 11 cells/5 rats) rats. E’ IPSC amplitude for adult rats and humans (L2/3: 19 cells/13 cases; L5: 8 cells/5 cases). F IPSC 20–80% rise-time in 1-month (L2/3: 18 cells/10 rats; L5: 5 cells/5 rats), 6–8-month (L2/3: 15 cells/13 rats; L5: 7 cells/6 rats), and 12–14-month-old (L2/3: 11 cells/5 rats; L5: 3 cells/2 rats) rats. Rise-time could not be measured for all IPSCs. F’ 20–80% rise-time in adult rat and humans (L2/3: 14 cells/10 cases; L5: 6 cells/4 cases). G IPSC decay time-constant from 1 (L2/3: 18 cells/10 rats; L5: 6 cells/5 rats), 6–8 (L2/3: 15 cells/13 rats; L5: 7 cells/6 rats), and 12–14 (L2/3: 11 cells/5 rats; L5: 3 cells/2 rats) rats (L5: 1 vs. 12–14 months, p = 0.035, L5: 6–8 vs.12–14 months, p = 0.0013; 12–14-month: L2/3 vs. L5: P = 0.0048). G’ Decay time-constants of adult rat and human (L2/3: 17 cells/10 cases; L5: 5 cells/5 cases). H IPSC charge-transfer 1 (L2/3: 19 cells/11 rats; L5: 17 cells/8 rats), 6–8 (L2/3: 20 cells/13 rats; L5: 15 cells/13 rats), and 12–14-month-old (L2/3: 12 cells/5 rats; L5: 11 cells/5 rats) rats. H’ IPSC charge-transfer for adult rat and humans (L2/3: 22 cells/13 cases; L5: 8 cells/5 cases). Data shown as boxplots (25–75% range) with median indicated and whiskers indicating maxima/minima, showing individual points. Statistics from LMM with 2-sided Tukey post-hoc tests: *P < 0.05, **P < 0.01; ***P < 0.001. Source data are provided as a Source Data file.

Fig. 4. Perisomatic GABA_BRs differentially inhibit AP discharge in L2/3 and L5 PCs over development.

A Schematic of experimental set-up depicting puff-application of 50 μM baclofen (red) to the perisomatic region during whole-cell recordings. IR-DIC image confirming somatic recording (record) and puff-pipette location in L5. B Baclofen puff-mediated currents recorded at −65 mV from L2/3 (black) and L5 PCs (blue), from 1- and 6–8-month-old rats, and humans. Baclofen puff (red arrow) and baseline (grey line) are shown for reference. C Action potential (AP) output of L2/3 (black, upper) and L5 (blue, lower) PCs from 1-month old rats held at +25 pA above rheobase following baclofen puff under control conditions (left) or in CGP (right). D AP firing during baclofen puff recordings for L2/3 (7 cells/4 rats) and L5 PCs (7 cells/4 rats), compared in the presence of CGP (grey or light blue). Right, AP discharge at 2–3 s after baclofen puff (U = 2, p = 0.0023). Number of cells is shown in parentheses. E, F The same analysis performed in 6–8-month-old rats (L2/3: 5 cells/4 rats; L5: 5 cells/4 rats; U = 10, p = 0.651). G, H The same analysis performed in human cortex (L2/3: 6 cells/5 cases; L5: 5 cells/5 cases; U = 9, p = 0.873). Data is shown as either mean ± SEM (left and middle panels) or boxplots (25–75% range) with median indicated and whiskers indicating maximum/minimum ranges with individual cell data overlain (right panels). Statistics are shown as ns P > 0.05, **P < 0.01, all from 2-sided Mann–Whitney tests. Source data are provided as a Source Data file.

Fig. 5. Presynaptic GABA_BRs strongly inhibit synaptic inputs to human cortical PCs.

A Recording configuration for L2/3 PCs, showing stimulus (lightning) applied to L1 or L4. B EPSCs evoked by L1 stimulation in L2/3 PCs were recorded at −70 mV for control (Ctrl) and following application baclofen (Bacl.) and CGP, the latter underlain by control (grey). Time-course of EPSC amplitude following baclofen and CGP wash-in is shown below, with baseline indicated (grey dashed line). C Data in the same form, but for L4 inputs (light grey). D Baclofen-mediated inhibition of L1 (black) and L4 (grey) inputs to L2/3 PCs during the last minute of wash-in from 6–8-month-old rats (8 cells/4 rats) and humans (L1: 15 cells/7 cases; L4: 14 cells/7 cases). E Recording configuration for L5 PCs with stimulus applied to L1 or L5. F EPSCs evoked by L1 stimulation in L5 PCs under control (Ctrl), and following baclofen, and CGP, the latter underlain by control (blue traces). Time-course of EPSC amplitude following baclofen and CGP wash-in is shown below. G Data in the same form, but for L5 inputs (light blue). H Baclofen-mediated inhibition of L1 (blue) and L5 inputs (light blue) inputs to L5 PCs during the last minute of wash-in from 6–8-month-old rat (L1: 9 cells/4 rats; L5: 9 cells/4 rats) and human (L1: 10 cells/3 cases; L5: 10 cells/3 cases). I Human EPSC recordings from L1 stimulation in L2/3 PC following L1 stimulation, before (black) and after (red) CGP wash-in for 10 EPSCs have been amplitude scaled for brevity to clarify CGP effects. J Short-term plasticity (STP) of EPSCs, normalised to the first, at 10 Hz (6 cells/3 cases). K STP of EPSCs at 20 Hz (6 cells/3 cases). Data is shown as mean ± SEM (B, C, F, G, J, K) or boxplots (25–75% range) with the median indicated and whiskers indicating maximum/minimum ranges (D, H). All statistics are shown from either LMM (D, H) or 2-way repeated measures ANOVA (2-sided, J, K). Source data are provided as a Source Data file.

Fig. 6. GABA_BR activation attenuates oscillatory activity in L2/3 and L5 of the human cortex.

A Local field potential (LFP) signals filtered into theta (θ), beta (β), low-gamma (Lo-γ), or high-gamma (Hi-γ) frequency ranges for L2/3 (black) and L5 (blue) during control (kainic acid and carbachol—KA+CCh) and during the last 2 min of 2 µM (light blue and grey) or 20 µM (lightest blue and grey) baclofen (Bacl.) application. B Power-spectra of L2/3 control (black; 8 slices/8 cases), 2 µM (grey; 5 slices/5 cases) or 20 µM (light grey; 8 slices/8 cases) baclofen recording, or following CGP (red; 3 slices/3 cases), or in L5 for control (blue; 8 slices/8 cases) 2 µM (light blue; 5 slices/5 cases) and 20 µM (lightest blue; 8 slices/8 cases) baclofen. C Area-under-curve (AUC) for KA+CCh recordings from L2/3 (black) and L5 (blue). D Effect on AUC (% change from KA+CCh) for frequency bands following 2 μM (dark grey, 5 slices/5 cases) and 20 μM (light grey; 8 slices/8 cases) baclofen. E The same data but for L5 (2 μM Bacl.; light blue, 5 slices/5 cases) and (20 μM Bacl.; lightest blue; 8 slices/8 cases). F CGP wash-in on AUC change for L2/3 (3 slices/3 cases) and L5 (2 slices/2 cases), not tested due to insufficient replicates. G Filtered LFP signals from L2/3 of 6–8-month-old rats in KA+CCh (dark green), 2 µM (mid green) or 20 µM (lightest green) baclofen. Right, power spectra of adult rat LFP recording for control (black; 6 slices/6 rats) and 2 μM (mid green; 5 slices/5 rats), or 20 μM baclofen (mid green; 6 slices/6 rats). H Comparison of AUC change following 2 μM (left) and 20 μM (right) baclofen for L2/3 of human (black) and rat (green) recordings. Data are shown as mean ± SEM (B, G), or boxplots (25–75% range) with the median indicated and whiskers indicating maximum/minimum ranges with individual data points overlain. Statistics are shown from 2-way ANOVA (2-sided, C–E) or 3-way ANOVA (H). Source data are provided as a Source Data file.

Fig. 7. GABA_BR activation de-synchronises beta/gamma coupling in human cortical circuits.

A Schematic of auto-correlation and cross-correlation analysis strategy. B Example plots of auto-correlation strength as a function of lag-time (ms) in LFP recordings from L2/3 (black) and L5 (blue) of the human cortex under control conditions (Kainic Acid, KA [0.6–2 µM] + Carbachol, CCh [60–200 µM]). C Example plots of cross-correlation strength between prominent oscillations in LFP recording from human L2/3 under control conditions (black) and following 2 µM (dark grey) or 20 µM (light grey) baclofen bath application. D Comparison of log-transformed correlation strength at zero lag (no time delay) between key oscillatory bands in L2/3 revealed significant frequency and baclofen effects. E Similar observations on cross-correlation strength were observed in human cortex L5 with respect to frequency and baclofen application. Data are shown as boxplots (25–75% range) with the median indicated and whiskers indicating maximum/minimum ranges; with individual values overlain. Statistical tests were performed with two-way ANOVA (two-sided). Source data are provided as a Source Data file.

Fig. 8. Pre-surgical LEV leads to elevated GABA_BR signalling in L2/3 of the human neocortex.

A Responses of human L2/3 PCs to depolarising current (500 pA, 500 ms), from seizure-free patients (black), patients with seizures and received LEV (red, SZ&LEV), or patients not experiencing seizures, but receiving LEV (purple). B Current–frequency responses (0–500 pA, 500 ms), from control (24 cells/15 cases), SZ&LEV (24 cells/11 cases), and LEV (13 cells/4 cases). C GABA_BR IPSCs evoked in L2/3 PCs from control (22 cells/13 cases), SZ&LEV (19 cells/10 cases), and LEV (8 cells/4 cases), superimposed response in CGP (light traces) and IPSC amplitudes for each patient group. D Whole-cell current following baclofen and CGP wash-in for control (27 cells/17 cases), SZ&LEV (18 cells/11 cases), and LEV (6 cells/4 cases) patients. E Baclofen current-density for control (34 cells/18 cases), SZ&LEV (23 cells/13 cases), and LEV (7 cells/4 cases) patients (Ctrl vs. SZ&LEV: p = 0.014; Ctrl vs. LEV-alone: p = 0.048; SZ&LEV vs. LEV-alone: p = 0.705; Tukey post hoc tests). F Baclofen current-density in slices pre-treated for 3 h with 100 μM LEV or vehicle, from control (Veh: 6 cells/4 cases; LEV: 8 cells/5 cases) and SZ&LEV patients (Veh: 7 cells/5 cases; LEV: 7 cells/4 cases). G Power spectra from L2/3 LFP recordings in KA+CCh from control (8 slices/8 cases), SZ&LEV (6 slices/6 cases), and LEV (4 slices/4 cases) patients. H Relative change in AUC from LEV-only patients compared to control following 2 μM (left) or 20 μM (right) baclofen wash-in. All data is shown as either mean ± SEM (B, D, G) or boxplots (25−75% range) with the median indicated and whiskers indicating maximum/minimum ranges (C, E, F, H) with individual values overlain. Statistics from LMM (C), 1-way ANOVA (E) with Holm–Sidak tests, 2-way (2-sided, B, F), or 3-way ANOVA (H).