Transcriptional and physiological adaptations in nucleus accumbens somatostatin interneurons that regulate behavioral responses to cocaine

Abstract

The role of somatostatin interneurons in nucleus accumbens (NAc), a key brain reward region, remains poorly understood due to the fact that these cells account for < 1% of NAc neurons. Here, we use optogenetics, electrophysiology, and RNA-sequencing to characterize the transcriptome and functioning of NAc somatostatin interneurons after repeated exposure to cocaine. We find that the activity of somatostatin interneurons regulates behavioral responses to cocaine, with repeated cocaine reducing the excitability of these neurons. Repeated cocaine also induces transcriptome-wide changes in gene expression within NAc somatostatin interneurons. We identify the JUND transcription factor as a key regulator of cocaine action and confirmed, by use of viral-mediated gene transfer, that JUND activity in somatostatin interneurons influences behavioral responses to cocaine. Our results identify alterations in NAc induced by cocaine in a sparse population of somatostatin interneurons, and illustrate the value of studying brain diseases using cell type-specific whole transcriptome RNA-sequencing.

Fig. 1

NAc somatostatin interneuron activity regulates cocaine-induced locomotor activity and reward. a AAV-DIO-EYFP or AAV-DIO-ChR2-EYFP were injected bilaterally into NAc of Sst-Cre mice 3 weeks prior to optogenetic stimulation or suppression, respectively, with the following parameters: Stimulation (473 nm, 20 Hz frequency, 1 burst of 10 pulses, 4 ms pulse width, every 10 s) or Suppression (473 nm, 20 Hz frequency, 1 burst of 20 pulses, 49 ms pulse width, every sec). b There is no difference between the locomotor activity of the three groups during habituation while the lasers remain off (repeated measures two-way ANOVA: no effect of activity F _(2,104) = 1.953. p = 0.147; n = 8,8,8) or day F_(4,104) = 0.3617. p = 0.9383; n = 8,8,8). c The activity of NAc somatostatin interneurons (EYFP control, Stimulation, or Suppression) controls the locomotor activity of mice after repeated cocaine injections. There is a significant effect of activity (repeated measures two-way ANOVA: significant effect of activity F_(2,21) = 5.647. p < 0.05; n = 8,8,8) and Day (Significant effect of Day F _(4,84) = 8.856. p < 0.001; n = 8,8,8). d Experimental design for unbiased CPP with optogenetics. Groups were counterbalanced to ensure that cocaine was paired evenly in both the gray and striped sides of the CPP chambers. e Stimulation of NAc somatostatin interneurons enhances cocaine CPP (Student's t-test: EYFP Pre-test vs EYFP Post-test, **p < 0.01; EYFP Pre-test mean = −16.50 ± 91.33 SEM vs EYFP Post-test mean = 376.4 ± 91.33 SEM (n = 8); CHR2-Stimulation Pre-test vs CHR2-Stimulation Post-Test, ****p < 0.0001 CHR2-Stimulation Pre-test mean = 51.00 ± 77.37 SEM vs CHR2-Stimulation Post-Test mean = 655.1 ± 53.93 SEM (n = 7); EYFP Post-test vs. CHR2-Stimulation Post-test, *p < 0.05 n = 8,7). f Cocaine CPP is blocked when NAc somatostatin interneurons are suppressed during training with cocaine. (Student’s t-test; AAV-DIO-EYFP Pre vs Post: **p < 0.01, EYFP Pre-test mean = −7.11 ± 66.2 SEM vs EYFP Post-test mean = 348.3 ± 90.41 SEM (n = 9)); AAV-DIO-CHR2 Pre vs Post: P > 0.05, n = 14). Data are represented as ± SEM

Fig. 2

Cocaine induces 1100 DETs in NAc somatostatin interneurons. a Coronal section of anterior forebrain showing EGFP-F/SST overlap in NAc. Scale in a = 200 µm; Scale in inset = 25 µm. b Representative FACS gating used to isolate EGFP(+) nuclei from NAc of SST-TLG498 mice. We isolated ~5000 nuclei per mouse. c Differential expression analysis identifies 1100 cocaine-regulated transcripts in NAc somatostatin interneurons (p < 0.05). 10 mice were injected with saline and 10 with cocaine every day for 7 days. Mice were analyzed 1 h after the final injection and tissue was flash frozen and stored before nuclear FACS. On average, ~1–2 ng of RNA were collected per mouse as determined by analysis on RNA Pico Bioanaylzer chips (Agilent). 0.5–2 ng were then used to generate indexed libraries for sequencing using Clontech SMARTer Stranded Total RNA-seq library preparation kits. All libraries were analyzed for mean insert size and molar concentration prior to sequencing with V4 chemistry (Illumina) at Beckman Coulter Genomics (Now: GeneWiz). Samples were pooled in groups of 8 and all pools were sequenced across multiple lanes. We performed multiplexed Illumina HiSeq to obtain > 2 × 10⁷ paired reads per sample with read length of 125 × 2 bp. Individual transcripts are depicted as individual squares. Transcripts above the x axis are upregulated, while transcripts below the x axis are downregulated. Transcripts are color coded by transcript type and transcript types are quantified for relative abundance in top right inset pie chart. d DETs are enriched for gene ontology terms related to multiple neuronal compartments demonstrating different ways in which genes can regulate cellular function (Bonferoni p < 0.05). Several of the genes implicated in regulating these cellular compartments (Ankr, Map2, NrCam) are implicated in regulating neuronal structural and functional plasticity. e Cell type-specific expression of GABAergic neuron and other interneuron subtype transcriptional markers. Npy neuropeptide Y, Nos1 nitrous oxide synthase 1, Lhx6 Lim-containing homeobox 6, Cck cholecystekinin, Pv parvalbumin, Chat choline acetyltransferase, Vip vasoactive intestinal peptide

Fig. 3

JunD expression in NAc somatostatin interneurons regulates locomotor behavior. a There is a positive correlation in our RNA-sequencing data between JunD expression levels and locomotor activity across saline- and cocaine-treated mice on the last day of testing (Pearson Correlation: r = 0.46, n = 20, p = 0.043). Green circles represent JunD expression levels from individual mice. b Representative images used for quantification of JUND corrected total cell fluorescence (CTCF) in EGFP(+) cells. Each image is a single plane in a z-stack to ensure true fluorescence signal (not the sum of multiple planes). Scale bar = 25 µm. c JUND protein expression is increased in EGFP(+) cells of cocaine-treated mice (1 h after last dose) compared to saline controls (Student's t-test with Welch’s Correction: ***p < 0.0001; Saline mean CTCF = 1807 ± 138.9 SEM (n = 20 cells/3 mice) vs Cocaine mean CTCF = 2920 ± 207.7 SEM (n = 23 cells/3 mice). Welch’s correction: t = 4.54, df = 37.77). d Stereotaxic targeting of HSV-LSL-JUND + HSV-LSL-mCHERRY or HSV-LSL-ΔJUND + HSV-LSL-mCHERRY in NAc of Sst-Cre mice. e qPCR quantification of transgene expression in NAc of Sst-Cre mice infected with HSV-LSL-JUND or HSV-LSL-mCHERRY. ∆JunD is a truncated form of JunD that lacks the N-terminus, making it a dominant-negative protein. (Student’s t-test with Welch’s correction: *p < 0.05; HSV-LSL-JUND mean = 0.01859 ± 0.008903 SEM (n = 8) vs HSV-LSL-ΔJUND mean = 0.0007039 ± 0.0004589 SEM (n = 10) Welch’s Correction: t = 2.259, df = 16). f JunD overexpression in NAc somatostatin interneurons increases locomotor activity in an open field (Students t-test: *p = 0.051; mCherry mean = 303.4 ± 26.25 SEM vs + JunD mean = 395.5 ± 35.62 SEM (n = 11,11)). g Dominant-negativeΔJunD expression in NAc somatostatin interneurons decreases locomotor responses to cocaine (Repeated Measures Two-way ANOVA: Significant effect of Virus F _(1,20) = 4.937. p < 0.04; n = 10,10). Data are represented as ± SEM

Fig. 4

The excitability and rheobase of NAc somatostatin interneurons are altered by repeated cocaine exposure. a Expression profiling of neuronal ion channels in NAc somatostatin interneurons. Average expression of each transcript is shown as a vertical line for each treatment group with blue intensity representing level of expression. White colored transcripts are lowly or not expressed whereas dark blue transcripts are more highly expressed. Significant changes in the transcription of specific channels after cocaine are shown under the heat map with the change in expression depicted by the font color of the gene name, such that yellow represents increased expression, while blue indicates decreased expression. b Representative traces of spike number obtained by 25 pA current injection in EYFP(+) NAc somatostatin neurons, as a measure of intrinsic neuronal excitability. c Whole-cell quantification of current-induced spike numbers in EYFP(+) NAc somatostatin neurons showed that repeated cocaine administration resulted in decreased cellular excitability with 25 and 50 pA current injections as compared to saline-treated controls (Two-way ANOVA, Bonferroni posthoc-test: cocaine treatment F_{(1, 65)} = 7.02, p = 0.0101; current injection F_{(1, 65)} = 269.89, p < 0.0001; interaction F_{(1, 65)} = 0.10, p = 0.7561; n = 31–36 cells per group, 5 mice per group). d Whole-cell quantification of current injection-induced first spike numbers in EYFP(+) NAc somatostatin neurons showed that increased current was needed to induce the first spike (rheobase) in repeated cocaine-treated mice compared to saline controls (Gaussian fit (Shapiro–Wilk) normality test p < 0.05, two-tailed Mann–Whitney test p = 0.0415; n = 31–36 cells per group, 5 mice per group). Data are represented as ± SEM