Anatomical and molecular characterization of parvalbumin-cholecystokinin co-expressing inhibitory interneurons: implications for neuropsychiatric conditions

Inhibitory interneurons are crucial to brain function and their dysfunction is implicated in neuropsychiatric conditions. Emerging evidence indicates that cholecystokinin (CCK)-expressing interneurons (CCK+) are highly heterogenous. We find that a large subset of parvalbumin-expressing (PV+) interneurons express CCK strongly; between 40 and 56% of PV+ interneurons in mouse hippocampal CA1 express CCK. Primate interneurons also exhibit substantial PV/CCK co-expression. Mouse PV+/CCK+ and PV+/CCK- cells show distinguishable electrophysiological and molecular characteristics. Analysis of single nuclei RNA-seq and ATAC-seq data shows that PV+/CCK+ cells are a subset of PV+ cells, not of synuclein gamma positive (SNCG+) cells, and that they strongly express oxidative phosphorylation (OXPHOS) genes. We find that mitochondrial complex I and IV-associated OXPHOS gene expression is strongly correlated with CCK expression in PV+ interneurons at both the transcriptomic and protein levels. Both PV+ interneurons and dysregulation of OXPHOS processes are implicated in neuropsychiatric conditions, including autism spectrum (ASD) disorder and schizophrenia (SCZ). Analysis of human brain samples from patients with these conditions shows alterations in OXPHOS gene expression. Together these data reveal important molecular characteristics of PV-CCK co-expressing interneurons and support their implication in neuropsychiatric conditions.


Allen Clusters
Columns sum to 1
expressed only in PV+ & CCK-Cre+ cells Supplemental Figure 8. Genetically targeted inactivation of PV+/CCK+ GABAergic inhibitory interneurons in mouse CA1 impairs cognitive memory behaviors.(a) Schematic illustration of genetic inactivation of PV+/CCK+ interneurons using a cross between CCK-IRES-Cre mice and PV-Flp mice (PV-Flp; CCK-Cre).(b) The resulting progeny are injected with AAV-Syn-Con/-Fon-hM4Di-mCherry-WPRE into CA1.Histological analysis in coronal brain sections verifies spatially restricted hM4Di-mCherry expression in CA1.Scale bars is 50µm.The viral injection experiment was independently repeated in 18 mice, with similar results obtained each time.(c) Scheme for the experimental design of the objectlocation memory (OLM) task with hM4Di/C-NO-mediated inhibition of PV+/CCK+ interneurons in CA1 during the learning phase.The box represents the open-field arena, and the black filled circles indicate the training (left) and test (right) object locations.Before the experiment, mice were handled and habituated to the con-text in the absence of objects.Mice received a single intraperitoneal injection of control saline or experimental CNO treatment (1.4 mg per kg) 1 hour before the learning session.(d) The discrimination index (DI) for the testing session 24 h after training.CNO-treated mice did not show a preference for the moved object in contrast to saline-treated controls.Data are present-ed as the mean ± S.E.M. *p < 0.05 (Mann-Whitney test).(e) The interaction index for the testing session 24 h after training.CNO-treated mice did not show a preference for the moved object in contrast to saline-treated controls.Data are presented as the mean ± S.E.M. *p < 0.05 (Mann-Whitney test).(f) Scheme for the experimental design of the fear renewal paradigm with hM4Di/CNO-mediated inhibition of PV+/CCK+ interneurons in CA1 during the fear renewal session.The boxes represent the shock box contexts, and the vertical lines indicate the foot shock bars.Mice received a single intraperitoneal injection of control saline (n=10) or experimental CNO treatment (1.4 mg per kg)(n=10) 1 hour before the fear renewal session.(g) Freezing behavior across all sessions.Mice first acquired fear conditioning by receiving paired sound and foot shock in the first context.The following 2 days, fear extinction sessions were performed in another context where sound was present, but without foot shock.One week later spontaneous recovery and fear renewal were measured.Fear renewal was significantly reduced in PV-Flp; CCK-Cre treated with CNO (n=9) 1 hour after the spontaneous recovery session as compared to transgenic mice treated with saline (n=9).Data are presented as the mean ± S.E.M. *p < 0.05 (two-tailed t-test).PV-Flp; CCK-Cre mice performed normally in allsessions.In G, data for CNO-treated mice is in red and saline is shown in black as separate groups, although treatments were not performed until after the spontaneous recovery session.
of transcriptomes of PV+/CCK+ and PV+/CCK-GAB-Aergic inhibitory interneurons in mouse.(a) (Samples) Individual sn-RNA-Seq datasets were processed as follows.Cells exhibiting an extremely high or low number of features were eliminated.Count matrices were log-normalized and scaled.Highly variable genes were identified and used as features for PCA dimensionality reduction.Next, anchors were identified and datasets were integrated via the Seurat framework.The data was projected to two dimensions via UMAP for visualization.(Cell Types) The integrated datasets were clustered, and cell types were identified via expression of interneuron gene markers (Sst, Pvalb, Vip, Lamp5, and Sncg).These are characteristically divided into the MGE (SST, PV) and CGE (VIP, LAMP5, and SNCG) lineages.(Pvalb) Pvalb gene expression, imputed using MAGIC, shows strong localization of Pvalb expression to the PV subtype.(Cck) Cck gene expression, imputed using MAGIC, is exhibited in both SNCG and PV interneurons.(b) Individual datasets are further analyzed through the following pipeline.Cck expression is imputed using MAGIC, and PV interneurons were identified as PV+/CCK+ if Cck expression is above the 75th percentile, and PV+/CCK-if Cck expression is below the 25th percentile.Differential expression testing between the groups identified >50 genes differentially expressed in at least four datasets.Many of these genes were associated to differences in interneuron communication (Cck, Nlgn1, Ptprd, Cox6a2, Grid2, Pcdh9).(c) Cck and Cox6a2 expression (post imputation) are highly correlated in the PV+ cell type (0.73), but not in any other interneuron subtype.(d) Gene co-expression analysis via WGCNA identified several highly coexpressed modules.Genes exhibiting coexpression less than 0.15 were removed from the visualization.The rest clustered into 6 groupings, including a COX grouping, that was also highly co-expressed with CCK.(e) Differentially expressed genes and genes in the COX/CCK module were analyzed for their associated ontologies.Overexpressed ontologies included Axon/Dendrite, Regulation of Synaptic Transmission, Mitochondrial ATP synthesis, Electron Transfer, and Mitochondrial respiration.These ontologies were associated to mostly non-overlapping gene groupings, possibly indicating largely independent functional subclusters within the COX/CCK module.(f) Gene expression of typical CCK+ or PV+ interneuron function exhibited significant differences between PV+ and PV-(SNCG+) groupings, as would be expected from their disparate lineages.Additionally, Cck, Tac1 and Cox6a2 expression exhibit large differences between PV+/CCK+ and PV+/CCK-groupings.Data are presented as violin plots. of chromatin accessibility peaks in hippocampal PV+ interneurons.(a-b) Nucleosome signal and TSS enrichment scores for Multiome samples.TSS enrichment was thresholded at 3.5 (in our experience Signac produces lower average TSS scores compared with alternative methods such as ArchR and SnapATAC).(c) Percentage of differentially expressed peaks overlapping known enhancer and promoter regions, divided by cluster.(d) Peak to gene linkages for each cluster marker (excluding Brinp3 as none were identified), along with coverage tracks for each cluster.Only nearby (within 1,000,000 bp of annotated gene center) candidate linkages are shown.

Supplemental Figure 7. Comparisons of firing patterns and spike shapes of PV+/C- CK+, GABAergic inhibitory interneurons in mouse CA1.
Summary graphs of electrophysiological properties including, action potential (AP) frequency, AP amplitude, AP threshold, AP adaptation index, AP risetime, AP half width, AP falltime, after hyperpolariza-The boxes are the 25th and 75th percentiles of the data with the median line in the middle.

Table 4 . Percentages of inhibitory neurons with neurochemical co-expression in different hippocampal regions.
: The table shows percentages of the number of neurochemical co-expressing neurons (e.g.(CCK&Dlx5/6)+ and PV+) relative to the number of neurons labeled by single marker in each corresponding anatomical structures in mouse hippocampus.The data presented as mean ± SD in percentage. Note