Control of impulsivity by Gi-protein signalling in layer-5 pyramidal neurons of the anterior cingulate cortex

Abstract

Pathological impulsivity is a debilitating symptom of multiple psychiatric diseases with few effective treatment options. To identify druggable receptors with anti-impulsive action we developed a systematic target discovery approach combining behavioural chemogenetics and gene expression analysis. Spatially restricted inhibition of three subdivisions of the prefrontal cortex of mice revealed that the anterior cingulate cortex (ACC) regulates premature responding, a form of motor impulsivity. Probing three G-protein cascades with designer receptors, we found that the activation of G_i-signalling in layer-5 pyramidal cells (L5-PCs) of the ACC strongly, reproducibly, and selectively decreased challenge-induced impulsivity. Differential gene expression analysis across murine ACC cell-types and 402 GPCRs revealed that - among G_i-coupled receptor-encoding genes - Grm2 is the most selectively expressed in L5-PCs while alternative targets were scarce. Validating our approach, we confirmed that mGluR2 activation reduced premature responding. These results suggest G_i-coupled receptors in ACC L5-PCs as therapeutic targets for impulse control disorders.

Fig. 1. Activation of the G_i-cascade in pyramidal cells of the ACC reduces premature responding.

a Approach of narrowing down a pharmacological target, which starts with genetically targeted manipulation of neural activity of a brain area (left), a specific cell-type within that area (mid left), intracellular signalling within that cell-type (middle), and is followed by the identification of activity-regulating genes (especially receptors) within that cell-type whose therapeutic action is finally validated pharmacologically. G_i, G_q and G_s refer to G-protein cascades. b Selective transfection of the infralimbic (IL, purple), prelimbic (PrL, lilac) and anterior cingulate cortex (ACC, blue) subregions of the PFC with the fusion protein hM4Di-mCherry in CamKIIα-Cre mice, as indicated by native mCherry-fluorescence in a brain slice that contains all three subregions (approximate anterior-posterior, AP, distance from Bregma: 1.8 mm; scale bar, 1 mm). c Operant cycle of the 5-CSRTT (see Methods; ITI, inter-trial interval; SD, stimulus duration). d Temporal order of surgery and chemogenetic/behavioural testing that each mouse of the cohort underwent, including the duration for each test phase; all tests were conducted within-subject with vehicle and 10 mg/kg CNO applications separated by 3–7 d. e, g, i Absolute values of premature responses (as % of the number of trials; indicator of impulsivity) after injection of vehicle (Veh) or CNO for the four subgroups of the cohort [ACC (blue), PrL (lilac), IL (purple), controls (Ctrl; black)] and 5-CSRTT protocols named above the panels. Black stars indicate significant drug-group interaction (RM-ANOVA); coloured stars indicate Sidak paired post-hoc test between vehicle and CNO conditions in the colour-coded group. CNO-induced changes in the 5-CSRTT baseline protocol (f), attention challenge (h) and impulsivity challenge (j), respectively, measured as log₁₀-transform of the within-subject ratio (value after CNO/value after vehicle) for relevant behavioural performance parameters on the 5-CSRTT are shown for the four groups as indicated in the legend; asterisks indicate one-sample t test against 0 (identity of parameter under both conditions); accuracy is multiplied by 10 as this parameter shows comparatively small variations. N-numbers stated in each panel. See Supplementary Tables 3–6 for reasons for varying N-numbers across experiments, further statistical analysis, and absolute response numbers. *P < 0.05; **P < 0.01; ***P ≤ 0.001; error bars, s.e.m. in e, g, i, otherwise C.I.; individual dots or dot-lines represent subjects.

Fig. 2. Activation of the G_i-cascade in ACC L5-PCs reduces premature responding in parametric challenges of impulsivity.

a Transfection (left) and Cre-dependent expression (right) of hM4Di-mCherry (G_i), hM3Dq (G_q) and rM3Ds (G_s) in ACC L5-PCs in coronal slices collected at ~0.6 mm anterior of bregma; scale bar, 1 mm. b Laminar expression of hM4Di-mCherry (red); scale bar, 1 mm. c Temporal order of surgery and chemogenetic/behavioural testing, including the duration for each test phase; all tests except for the EPM were conducted within-subject with vehicle and CNO applications separated by 3-7 d. d–f Premature responses after injection of vehicle (Veh) or CNO for the four subgroups of the cohort [G_i (blue), G_s (orange), G_q (red), mCherry-controls (Ctrl; black), N-numbers in brackets] and 5-CSRTT challenge condition indicated above. Black stars indicate significant CNO-group interaction (RM-ANOVA); coloured stars indicate Sidak paired post-hoc test between vehicle and CNO conditions in the colour-coded group. g CNO-induced changes in the 9s fITI-challenge measured as log₁₀-transform of the within-subject ratio (value after CNO/value after vehicle) for relevant behavioural parameters on the 5-CSRTT are shown for the 4 groups indicated in the legend with N-numbers stated in brackets. h Same data and N-numbers as in g depicting the parametric response profile of individual groups; grey line indicates 0-lattitude (no CNO-induced change). i CNO-induced changes of premature responses (as in g) shown for the different parametric challenges in order of their execution, including dose-response (DR) experiments in the fITI- and vITI-challenges as indicated. N-numbers for the first three protocols are stated in d–f, for the other three experiments N-numbers are as follows in chronological order; fITI-DR: G_i (13), G_q (7), G_s (9), Ctrl (20); repeated 9s-fITI: G_i (12), G_q (7), G_s (10), Ctrl (18); vITI: G_i (8 for 1 mg/kg, 6 for 10 mg/kg), G_q (8), G_s (4), Ctrl (15). Asterisks in g and i indicate one-sample t-tests against 0. See Supplementary Fig. 2 for related accuracy data and Supplementary Tables 3 and 7–9 for reasons for varying N-numbers across experiments, further statistical analysis, and absolute response numbers. 1 mg/kg CNO was used for the G_q-group and a subset of Ctrl mice, 10 mg/kg was used for the remainder, except for DR experiments for which doses are indicated in i. *P < 0.05; **P < 0.01; ***P ≤ 0.001; error bars, s.e.m. in d–f, otherwise C.I.; individual dots or dot-lines represent subjects.

Fig. 3. Activation of the G_i-cascade in ACC L5-PCs reduces premature responding in a pharmacological challenge of impulsivity.

a Premature responding and accuracy after injection of vehicle (0) or indicated doses of Ro 63-1903 (Ro) in a separate cohort of 18 wildtype mice. Error bars, C.I. b Experimental schedule applying Ro as impulsivity challenge and CNO  before testing in the 5-CSRTT in a subset of the DREADD-transduced Rbp4-Cre cohort evaluated in Fig. 2. %premature responses (c) and attentional accuracy (d) observed after application of 3 mg/kg Ro 63-1908 (Ro) and additional application of CNO (Ro+CNO). 1 mg/kg CNO was used for the G_q-group and a subset of Ctrl mice, 10 mg/kg CNO was used for the remainder. In a, c, d, grey asterisks indicate significant main effect of drug condition across all groups (RM-ANOVA), coloured asterisks below horizontal lines indicate significant paired post-hoc comparisons between the indicated drug condition (vehicle and 3 mg/kg Ro in a, Ro+Veh and Ro+CNO in c, d) in the groups indicated by the colour (Sidak); coloured asterisks on the right indicate significant pairwise post-hoc comparisons between the group indicated by the colour (G_i, blue; G_q, red; Ctrl, black) at the Ro+CNO condition (Sidak). e Response profiles of Ro- (dashed lines of reduced opacity) or Ro+CNO-induced (solid lines) changes relative to the vehicle/vehicle condition measured as log₁₀-transform of the within-subject ratio (value after drug/value after vehicle) for relevant behavioural performance parameters on the 5-CSRTT are shown for the three groups as indicated in the legend; grey solid line indicates 0-lattitude (no CNO-induced change). N-numbers for c–e stated in d. See Supplementary Tables 3 and 7–10 for reasons for varying N-numbers across experiments, further statistical analysis, and absolute response numbers, respectively. *P < 0.05; **P < 0.01; ***P ≤ 0.001; individual dot-lines represent subjects.

Fig. 4. The G_q- and the G_i-cascades in ACC L5-PCs have antagonistic effects on locomotor activity.

a Experimental set-up. Locomotor activity in a novel open-field assessed by total distance moved (b) and total amount of time spent immobile (c) during the 45 min test after injection of vehicle or CNO. Group identity and N-numbers stated in b. Black stars on top line indicate significant drug-group interaction (RM-ANOVA); coloured stars on horizontal lines below indicate a significant difference between vehicle and CNO conditions in the group coded by the colour (Sidak paired post-hoc test); coloured stars below data points indicate significant difference to the corresponding value in the Ctrl group (Sidak post-hoc test). d Exploration/anxiety ratio assessed by the preference to visit the open arm of an elevated plus-maze after injection of CNO. Group identity and N-numbers stated in colour legend. See Supplementary Tables 3 and 7 for statistics and reasons for varying N-numbers across experiments, respectively. 1 mg/kg CNO was used for the G_q-group and a subset of Ctrl mice, 10 mg/kg was used for the remainder. *P < 0.05; **P < 0.01; ***P ≤ 0.001; error bars, C.I.; individual dots or dot-lines represent subjects.

Fig. 5. Differential gene expression analysis across 402 GPCR-encoding genes within the mouse ACC.

a, b Differential gene expression values (log₂-transformed ratios of gene expression in target set of ACC L5-cells relative to a non-target set of GABAergic, L2/3/6 excitatory and non-neuronal ACC cells, N = 3514; left axis), difference of share of cells with expression in target set and non-target set (Beta, blue dots, right axis), and average expression level in non-target set (log₂(CPM + 1); grey scale) are displayed for all GPCR-encoding genes that are among differentially expressed transcripts in target set (Bonferroni-corrected P < 0.05) and are expressed at least threefold higher in target set compared to the non-target set. Horizontal dotted line indicates a 10fold higher expression in the target set (referring to left axis). The calculation has been performed by either selecting tdTomato-positive excitatory cells extracted from an Rbp4-Cre::Ai14 line (i.e. using the same selection as for the chemogenetic experiments in Figs. 2–4; N = 676; a), or by selecting clusters of L4/5 intertelencephalic-projecting (IT) cells according to the metadata (N = 1238; b). c t-SNE maps of cell-type clusters of the complete dataset involving 20 cortical regions with the cells from the ACC target (green, arrow) sets indicated; overlaid black and grey names indicate clusters of excitatory and non-excitatory cells, respectively. d, e Expression levels—colour-coded according to the continuous scale below—of Grm2 (d) and Grm3 (e) projected onto the ACC cells of the t-SNE map shown in c indicate high expression of those genes majorly in excitatory cells, with Grm2 (compared to Grm3) showing a higher specificity for L4/5 IT cells. f Same plot as (c) but for displaying L4/5-IT cells across all analysed cortical areas (shown in different colours). Expression of Grm2 (g) and Grm3 (h) across L4/5–IT cells of all analysed regions. Gene expression analysis and creation of t-SNE maps have been performed in CytosploreViewer. See Supplementary Data 2 for details of sets, results and analysed GPCR genes, and Supplementary Fig. 3 for the same analyses as in a, b) using a reduced number of cell types for the non-target set.

Fig. 6. Reduction of impulsivity by the mGluR2/3 agonist LY379268 and the mGluR2-PAM TASP0433864.

Absolute values of %premature responses (left) and attentional accuracy (right) after injection of vehicle (0) or various doses of LY379268 (LY; a, c) or TASP0433864 (TASP; e, g), as indicated, in the parametric 9s fITI (a, e) or pharmacological Ro 63-1908 (Ro, c, g) challenge conditions. The same group of 18 male C57BL/6 mice was used for every LY-experiment and the same group of 13 male C57BL/6 wildtype mice was used for every TASP-experiment with contributing N-numbers indicated in the accuracy panels. Black stars indicate significant effects of dose; grey stars indicate significant paired post-hoc comparisons between dose-levels (Sidak). (b, d, f, h) Drug-induced changes in the corresponding experiment identified in a, c, e, g measured as log₁₀-transform of the within-subject ratio (value after drug(s)/value after vehicle) for relevant behavioural performance parameters on the 5-CSRTT are shown for the tested doses and drug combinations as indicated in the colour legend. Asterisks indicate one-sample t-tests against 0, colour-coded according to the dose. For the Ro-challenges (d, h), the results of paired t-tests between the two drug conditions are also indicated by black asterisks below bars. See Supplementary Tables 10, 11 (for LY) and 13-14 (TASP) for statistics, further analysis, and reasons for varying N-numbers across experiments. N-numbers are stated in each panel. *P < 0.05; **P < 0.01; ***P ≤ 0.001; error bars, C.I.; individual dots or dot-lines represent subjects.

Fig. 7. Activation of the G_i-cascade in pyramidal cells of the ACC reduces premature responding in a pharmacological model of impulsivity.

a Transfection of the ACC with a CamKIIα-hM4Di-mCherry vector (red) shown in different slices along the AP-axis (as indicated); scale bar, 1 mm. b Premature responding and c attentional accuracy in Tacr1-knockouts (KO, grey) and littermate wildtype controls (WT, black) in impulsivity-promoting challenges, as indicated: variable ITI (vITI), combined reduction of stimulus duration (1 s) and fixed-duration increase of ITI (7 s), and fixed ITI increase (9 s fITI). Note that only 15 WT participated in the vITI and combined challenge, only 25 KO participated in vITI and 9s-fITI challenge. d, e Premature responding and accuracy after injection of vehicle (V) or CNO (C) (d), and double-injections of Ro/Veh (V) or Ro/CNO (C) as indicated (f) during the vITI protocol in KO (brown), WT (black), or KO with hM4Di-transfection in the ACC (Rescue, green). Coloured asterisks indicate significant paired post-hoc comparisons (Sidak) within the indicated group; black asterisks on horizontal lines indicate significant within-drug difference between rescue and control groups (Sidak). Note that in e, premature responses, there is also a significant drug-group interaction and effect of group in overall RM-ANOVA, and a significantly lower impulsivity in both KO groups compared to WT under Ro/vehicle condition (Sidak), not indicated. f CNO-induced changes in the 5-CSRTT Ro-challenge measured as log₁₀-transform of the within-subject ratio (value after CNO/value after vehicle) for relevant behavioural performance parameters on the 5-CSRTT are shown for the three groups as indicated in the legend; asterisks indicate one-sample t-test against 0. See Supplementary Tables 3, and 15, 16 for reasons for varying N-numbers across experiments, statistics, and absolute numbers of responses. N-numbers stated in each panel. Doses of 10 mg/kg CNO were used throughout. *P < 0.05; **P < 0.01; ***P ≤ 0.001; error bars, C.I.; individual dots or dot-lines represent subjects.