Single-Nucleus Transcriptomics of the Nucleus Accumbens Reveals Cell-Type-Specific Dysregulation in Adolescent Macaques with Depressive-Like Behaviors

Abstract

Adolescent depression is increasingly recognized as a serious mental health disorder with distinct clinical and molecular features. Using single-nucleus RNA sequencing, we identified cell-specific transcriptomic changes in the nucleus accumbens (NAc), particularly in astrocytes, of adolescent macaques exhibiting depressive-like behaviors. The level of diacylglycerol kinase beta was significantly reduced in neurons and glial cells of depressed macaques, while FKBP5 levels increased in glial cells. Disruption of GABAergic synapses and disruption of D-glutamine and D-glutamate metabolism were linked to depressive phenotypes in medium spiny neurons (MSNs) and subtypes of astrocytes. Communication pathways between astrocytes and D1/D2-MSNs were also disrupted, involving factors like bone morphogenetic protein-6 and Erb-B2 receptor tyrosine kinase-4. Bulk transcriptomic and proteomic analyses corroborated these findings, and FKBP5 upregulation was confirmed by qRT-PCR, western blotting, and immunofluorescence in the NAc of rats and macaques with chronic unpredictable mild stress. Our results highlight the specific roles of different cell types in adolescent depression in the NAc, offering potential targets for new antidepressant therapies.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12264-025-01412-5.

Introduction

The adolescent period is often overlooked as a vulnerable time for the onset of major depressive disorder (MDD), affecting 4%–5% of this demographic [1]. Moreover, young patients with MDD experience greater functional impairments than adults [2], along with diminished treatment responsiveness [3] and elevated suicide risk, which is the leading cause of mortality in this demographic [4]. Research indicated that neuron-glia interactions are involved in the underlying pathophysiology of MDD including neurotransmitter function [5], synapse function [6], and cortical structure [7]. However, the symptom profiles and brain molecular dysfunctions are notably different between adolescent and adult depression [8, 9]. Therefore, exploring cell-type-specific molecular pathology in adolescent depression may lead to improved diagnosis and treatment for this population.

Single-nucleus RNA sequencing (snRNA-seq) effectively reveals gene expression patterns in major cell types and has identified novel subtypes linked to neuropsychiatric disorders, including Alzheimer's disease [10], autism spectrum disorder [11], and depression [12]. Various cell types are disturbed in MDD, especially during critical adolescent development [13], but no snRNA-seq studies have examined brain tissues from this population. The applicability of rodent models in preclinical research into complex neuropsychiatric disorders has been a topic of debate in recent years [14]. Non-human primate (NHP) models of depression are suitable for translational scrutiny of snRNA-seq studies, given that the brain structure, cell types, and transcription signature are more similar to those of humans [15, 16]. In recent years, an increasing number of researchers have conducted single-cell transcriptome analyses in the brain using NHP samples [15, 17]. Our group has recently identified a subpopulation of pro-inflammatory microglia in the dorsolateral prefrontal cortex of adult macaques exhibiting depressive-like behaviors through the integration of snRNA-seq and spatial transcriptomic analysis [18]. An NHP depression model has previously been developed in adolescent macaques by using early life stress [9, 19], which is highly suitable for studying adolescent depression due to their similarities to humans during this developmental stage [20, 21]. Thus, the adolescent NHP depression model may become a useful model for studying the pathogenesis of MDD during adolescence by snRNA-seq analysis.

The nucleus accumbens (NAc), a key part of the basal ganglia, regulates reward and pleasure processing, and its dysfunction is linked to various neuropsychiatric disorders, including depression [22, 23]. Clinically, the NAc is a common treatment target in the comorbidity of depression and addiction [24], as well as the treatment target region of the rapid-acting antidepressant ketamine [25]. Moreover, the presence of heightened chronic stressors during adolescence, such as academic pressure and interpersonal challenges, has been linked to disruptions in the NAc reward circuitry [26]. The influence of early-life stress on NAc functionality is thought to originate from alterations in genome-wide transcription, leading to alterations in synaptic function and increased susceptibility to depression [27, 28], which may also be a contributing factor to the increased prevalence of depression in adolescents during this development period [29]. Recently, snRNA-seq has been successfully applied to generate gene expression atlases of the NAc under the physiological conditions in the human and mouse brain [22, 30]

In this study, we applied snRNA-seq analysis to elucidate the cell-type-specific differential expressed genes (DEGs) and functions in NAc tissue from adolescent macaques with depressive-like behaviors, using our developed adolescent NHP depression model [19]. Then, the cell subtypes and cell-cell communications of neurons and astrocytes were also identified in the NAc of depressed adolescent macaques. Subsequently, a multi-omic data integration approach was used, encompassing snRNA-seq, bulk sample RNA-seq, and proteomics. Ultimately, the expression of FKBP5 in the NAc of adolescent CUMS (chronic unpredictable mild stress) macaques and rats was validated by molecular biology techniques. The experimental design and analysis workflow is illustrated in Fig. 1.

Fig. 1.

Workflow of the experiment and analysis. Nuclei are extracted from the NAc of five adolescent cynomolgus macaques presenting with depressive-like behaviors and five age- and gender-matched controls. Then, single nuclei are captured for RNA sequencing (10X Genomics). All cells are subjected to unsupervised clustering and cell-type annotation. Bioinformatic analyses are applied to detect cell-specific dysregulation in adolescent depression, including quality control, cell annotation, differential expression, pathway analysis, database analysis, and STRING DB network analysis. Moreover, cell subtype analysis for neurons and astrocytes is conducted, including QuSAGE and cell-cell communications. Finally, multi-omics data integration of snRNA-seq, whole NAc bulk RNA-seq, and proteomics are applied.

Materials and Methods

Adolescent Cynomolgus Macaques with Depressive-Like Behaviors

The NAc samples from all of the ten adolescent cynomolgus macaques included in this study were collected in our previous studies [9, 19]. The details of the rearing conditions, stress procedures, and ethical statements were also reported in our previous studies [9, 19]. In summary, all the adolescent male cynomolgus macaques (n = 10, aged 21–52 months) were purchased from and fed in Zhongke Experimental Animal Co., Ltd (“Zhongke”) located in Suzhou, China. The macaques were maintained under constant rearing conditions (temperature, ≥18°C; light/dark cycle, 12 h; humidity, 40%–70%), provided with water ad libitum, and fed a diet of fresh fruit, vegetables, and macaque chow twice daily. Veterinary care, daily cleaning, and yearly health inspections were provided by Zhongke. The procedures were conducted in accordance with the “Guide for the Care and Use of Laboratory Animals” [31] and “The Use of Non-Human Primates in Research” [32]. The study protocol was approved by the Ethics Committee of Chongqing Medical University (approval number: 20180705). Ten healthy adolescent male cynomolgus macaques were paired according to age and body weight, and each member of a pair was randomly assigned to either the CUMS group (n = 5) or the unstressed control group (CON, n = 5). The macaques in the CUMS and CON groups were housed in single cages in two separate rooms. The macaques in the CUMS group experienced 7 days of chronic unpredictable mild stress (e.g., noise, space restriction, and cold stress) for five cycles, while the macaques in the CON group were reared as usual. The detailed results of the body weight, cortisol levels, behavioral observations, and tests (attempt for apple test and human intruder test) at the end of the study in both CUMS and CON groups have been published [19].

Isolation of Nuclei from Frozen Brain Tissue

All ten animals were deeply anesthetized and euthanized by pentobarbital sodium injection (30 mg/kg). The entire brain was excised and rapidly frozen in liquid nitrogen [33]. Subsequently, each frozen brain was sectioned coronally, and the NAc was dissected in accordance with the macaque brain atlas [34, 35]. The NAc tissues were double-checked by two investigators. The nuclei of the NAc were isolated and purified as previously described [36], with modifications. The frozen tissue was homogenized in Nuclei Lysis Buffer which consisted of (in mmol/L) 250 sucrose, 10 Tris-HCl, 3 MgAc2, 0.1 EDTA, 0.1% Triton X-100 (Sigma-Aldrich, USA), and 0.2 U/μL RNase Inhibitor (Takara, Japan). The sucrose concentration was adjusted to facilitate the purification of the nuclei. The concentration of snRNA-seq was adjusted to ~1000 nuclei/μL.

10× snRNA-Seq Pipeline

The snRNA-Seq libraries were generated using the 10X Genomics Chromium Controller Instrument and Chromium Single Cell 3’ V3 Reagent Kits (10X Genomics, Pleasanton, USA). The cell nuclei were concentrated to 1000 nuclei/μL, and subsequently loaded into each channel in order to generate single-cell Gel Bead-In-Emulsions (GEMs). Following the reverse transcription phase, the GEMs were disrupted and the barcoded cDNA was purified and amplified. The amplified barcoded cDNA was fragmented, A-tailed, ligated with adaptors, and subjected to index PCR amplification. The final libraries were quantified utilizing the Qubit High Sensitivity DNA assay (Thermo Fisher Scientific, USA), and the size distribution of the libraries was determined using a High Sensitivity DNA chip on a Bioanalyzer 2200 (Agilent, USA). All libraries were sequenced using a Novaseq6000 (Illumina, USA) instrument with a 150-bp paired-end run.

snRNA-Seq Statistical Analysis

The fast algorithm was applied with the default parameters, which entailed filtering the adaptor sequence and removing low-quality reads, in order to achieve a clean data set [37]. Subsequently, the feature-barcode matrices were obtained by aligning the reads to the mouse genome (GRCm38 Ensembl: version 100) using CellRanger v6.1.1. A down-sampling analysis was conducted on the mapped barcoded reads per cell of each sample, and the aggregated matrix was then generated. The number of expressed genes in the cells ranged from 200 to 10,000. Mitochondrial UMI (unique molecular identifier) rates <10% were deemed to indicate satisfactory cell quality, and thus mitochondrial genes were excluded from the expression table. The Seurat package (version: 4.0.3, https://satijalab.org/seurat/) was used for cell normalization and regression based on the expression table in accordance with the UMI counts of each sample and the percentage mitochondrial rate, in order to obtain the scaled data. As the samples were processed and sequenced in batches, we applied the Mutual Nearest-Neighbor approach to eliminate the potential for batch effects. Subsequently, the top 10 principal components were used for the construction of Uniform Manifold Approximation and Projection (UMAP) plots. A graph-based cluster method (resolution = 0.8) was applied to obtain an unsupervised cell cluster result based on the top 10 principal components. The marker genes were calculated using the FindAllMarkers function with a Wilcoxon rank-sum test algorithm, according to the following criteria: (i) lnFC >0.25; (ii) P-value <0.05; (iii) min.pct >0.1. In order to gain greater insight into the specific cell types present, the clusters belonging to the same cell type were subjected to further analysis, including re-UMAP, graph-based clustering, and marker analysis.

Differential Analysis in snRNA-Seq Between CUMS and CON Macaques

In order to identify DEGs in different cell types including neurons, astrocytes, microglia, oligodendrocytes, and oligodendrocyte precursor cells (OPCs), as well as D1- and D2-MSNs among the CUMS and CON groups, the function FindMarkers with the Wilcoxon rank-sum test algorithm was applied under following criteria: (i) logFC >0.25; (ii) FDR <0.05; (iii) min.pct >0.1. A Gene Ontology (GO) analysis was conducted to elucidate the biological implications of DEGs in neurons, astrocytes, microglia, oligodendrocytes, and OPCs [38]. The GO annotations were downloaded from the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov/), the UniProt database (http://www.uniprot.org/), and the Gene Ontology website (http://www.geneontology.org/). Fisher’s exact test was applied to identify the significant GO categories, and the FDR was used to correct the P-values. The DisGeNET and PsyGeNET databases were used to establish links between the DEGs and mental illness in the literature by using the disgenet2r and psygenet2r R packages, respectively [39, 40]. Subsequently, Fisher’s exact and hypergeometric tests were used to identify significant enrichments of the DEGs. STRING DB network analysis was used to plot the interactions between protein-coding DEGs in at least three cell types [41]. The results of the network were visualized using Cytoscape (3.8.2). To ascertain the relative activation of a given gene set comprising D1-MSNs, D2-MSNs, and subtypes of astrocytes, we applied QuSAGE (2.16.1) analysis [42]. In order to facilitate a comprehensive examination of the molecules involved in the communication between astrocytes and D1/D2-MSNs communication molecules, a cell communication analysis was applied using the public repository CellPhoneDB [43], which provides information on ligands, receptors, and their interactions. The mean and Cell Communication significance (P-value <0.05) was calculated based on the interaction and the normalized cell matrix achieved by Seurat Normalization.

Multi-omics Data Integration Analysis of snRNA-seq, Bulk RNA-seq, and Proteomics in the NAc

Bulk RNA-seq

Total RNA was extracted from NAc tissue using TRIzol® Reagent according to the manufacturer’s instructions (Invitrogen), and genomic DNA was removed using DNase I (TaKaRa). RNA degradation and contamination were monitored on 1% agarose gels, and the RNA quality was determined using the 2100 Bioanalyser (Agilent Technologies) and quantified using the ND-2000 (NanoDrop Technologies). A sequencing library was constructed using high-quality RNA samples (OD260/280 = 1.8-2.2, OD260/230 ≥2.0, RIN ≥8.0, 28S:18S ≥1.0, >1 μg). RNA purification, reverse transcription, library construction, and sequencing were performed according to the manufacturer’s instructions (Illumina, San Diego, CA). After quantification by the Turner Biosystems TBS380, the paired-end RNA-seq sequencing library was sequenced with the Illumina NovaSeq 6000 sequencer (2 × 150-bp read length). Raw paired-end reads were trimmed and quality-controlled using fastp (https://github.com/OpenGene/fastp) with default parameters [37]. Clean reads were then aligned separately to the reference genome with orientation mode using HISAT2 software (http://ccb.jhu.edu/software/hisat2/index.shtml) [44]. The mapped reads of each sample were assembled using StringTie (https://ccb.jhu.edu/software/stringtie/) in a reference-based approach [45]. Bulk RNA-seq DEGs of paired samples from the CUMS and CON groups were calculated by DESeq2 (P-value <0.05).

Bulk Proteomics

Liquid chromatography: We used a nanoElute liquid chromatography system (Bruker Daltonics). Peptides (200 ng of digest) were separated within 90 min at a flow rate of 300 nL/min on a 25 cm × 75 μm column with a laser-pulled electrospray emitter packed with 1.5 μm ReproSil-Pur 120 C18-AQ particles (Dr. Maisch, Germany). Mobile phases A and B were water and acetonitrile with 0.1 vol% FA, respectively. The %B was linearly increased from 2 to 22%, followed by an increase to 37% within 8 min and a further increase to 95% within 5 min before the last 7-min 95% process.

Mass spectrometry and generation of a spectral library: All fraction samples were analyzed on a hybrid trapped ion mobility spectrometry (TIMS) quadrupole time-of-flight mass spectrometer (MS) (TIMS-TOF Pro, Bruker Daltonics) via a CaptiveSpray nano-electrospray ion source. The MS was operated in data-dependent mode for the ion mobility-enhanced spectral library generation. We set the accumulation and ramp time at 100 ms each and recorded mass spectra in the range of 100 to 1700 m/z in positive electrospray mode. The ion mobility was scanned from 0.6 to 1.6 Vs/cm². When performing data independent acquisition (DIA), we defined quadrupole isolation windows as a function of the TIMS scan time to achieve seamless and synchronous ramps for all applied voltages. We defined up to eight windows for single 100-ms TIMS scans according to the m/z-ion mobility plane. During Parallel Accumulation Serial Fragmentation (PASEF) MSMS scanning, the collision energy was ramped linearly as a function of the mobility from 59 eV at 1/K0 = 1.6 Vs/cm² to 20 eV at 1/K0 = 0.6 Vs/cm². Raw files were processed using a developmental version of Spectronaut (v14.2, Biognosys). The ion mobility-enhanced library was generated from data-dependent acquisition (DDA)-PASEF raw data using Spectronaut’s database against a UNIPROT human database.

Validation of FKBP5 Expression Levels in the NAc

Chronic Unpredictable Mild Stress in Rats

The CUMS protocol has been previously described in detail in our laboratory [46]. A total of 16 male adolescent Sprague-Dawley rats (21 days old) were obtained from Chongqing Medical University and housed under controlled conditions. After one week of acclimatization, the rats were randomly divided into a CUMS group (n = 8) and a CON group (n = 8). Rats in the CUMS groups were subjected to two mild stressors daily for four weeks, with no repeated stressors on consecutive days, while rats in the control group were reared as usual. The sucrose preference test (SPT), forced swimming test (FST), open field test, and elevated plus-maze were applied at the end of experiments and were reported in our previous study [46].

Immunofluorescence

The rats were perfused transcardially with 300 mL of cold phosphate-buffered saline (PBS) followed by 300 mL of cold 4% paraformaldehyde (PFA). Then, the brains were post-fixed in PFA for 24 h at 4°C followed by an additional 48 h of dehydration in 15% and 30% sucrose at the same temperature. The brain samples were sectioned at 20 μm on a cryostat microtome (RWD Minus FS800) at −20°C. The sections were washed with PBST (0.3% Triton X-100 in PBS) for 15 min (5 min × 3 times) and then blocked in QuickBlock™ Blocking Buffer (Beyotime, P0220) for 10 min at room temperature. Subsequently, the sections were incubated with primary antibody overnight at 4°C. The primary antibody was diluted as FKBP5 (mouse, 1:500, Proteintech, 67874-1-lg). The sections were then washed three times in PBS and incubated with secondary antibodies for 2 h at room temperature. The following secondary antibodies were applied: goat anti-mouse (1:500, AF488; Abcam, ab150113). Finally, the sections were washed three times with PBS (5 min each time). Fluorescence images were captured on a fully automatic digital slide scanning system (Zeiss Axioscan 7).

Quantitative Real-Time PCR (qRT-PCR)

The total RNA of the NAc samples was extracted using the Total RNA Extraction Kit (Tiangen, DP419) according to the manufacturer’s instructions. cDNA was then synthesized using the PrimeScript™ RT Reagent Kit (Takara, RR047A). For qRT-PCR, the SYBR Green qPCR Master Mix (MCE, HY-K0523) was used, and the cDNA of the targeted genes was amplified using SYBR Green in a real-time fluorescence quantitative PCR analyzer (QuantGene 9600). Samples were assayed in triplicate, with beta-actin serving as an internal control. The primer sequences utilized in this study were as follows: beta-actin (forward: 5’-CCGCGAGTACAACCTTCTTG-3’; reverse: 5’-TGACCCATACCCACCATCAC-3’) and FKBP5 (forward: 5’-GTGGTTGAGCAGGGAGAAGATAT-3’; reverse: 5’-TTCCATTCGACAACTTGCCTTTG-3’).

Western Blotting

The NAc tissue was lysed in RIPA buffer (Beyotime), and phenylmethylsulfonyl fluoride was added prior to use. The lysate was then supplemented with protease and phosphatase inhibitors. The resulting supernatant was obtained by centrifugation at 12,000 rpm at 4 °C for 15 min. Approximately 20 mg of protein was subjected to SDS-PAGE gel electrophoresis and subsequently transferred to polyvinylidene fluoride membranes. The membranes were blocked for 1 h at room temperature with 5% non-fat milk. Following the blocking step, the membranes were incubated with the primary antibodies overnight at 4 °C. The primary antibodies against FKBP5 for macaques (rabbit, CST, 12210), FKBP5 for rats (mouse, Proteintech, 67874-1-1g), and GAPDH (rabbit, Abcam, ab181602) were diluted at 1:1000, 1:5000, and 1:8000, respectively. On the next day, the membranes were washed three times and incubated with the corresponding HRP-conjugated secondary antibody for 1 h. The bands were then detected by enhanced chemiluminescence.

Statistical Analysis

As we matched pairs with age and weight between the CUMS group (n = 5) and the CON group (n = 5), the results of body weight, behavior observations, attempt for apple test, human intruder test, and hair cortisol level of all the ten cynomolgus macaques in both groups were assessed by the Wilcoxon Signed Rank Test (SPSS Statistics for Windows, Version 25.0, IBM Corp., Armonk, NY, USA). Also, the characteristics of the summary of the quality control and number of cells between the CUMS and CON groups in snRNA-seq were assessed by the Wilcoxon Signed Rank Test (SPSS Statistics for Windows, Version 25.0). Spearman's correlation analyses were also applied using SPSS Statistics for Windows (Version 25.0). The results of behavioral tests and the expression of FKBP5 at the mRNA or protein level were assessed by the Mann-Whitney test (SPSS Statistics for Windows, Version 25.0). The bar graph was generated by GraphPad Prisma (Version 8.0.2). The statistical significance level was set at P <0.05.

Results

Identification of Adolescent Macaques with Depressive-Like Behaviors in the CUMS Group

The adolescent macaque stress-induced model of depressive-like behavior was reported in our previous study [19]. Briefly, there was no significant difference in body weight and behavioral results between the CUMS and CON groups at baseline and the midpoint (three stress cycles). At the end of the experiment (five stress cycles), CUMS macaques were characterized by a significantly higher frequency and longer duration of huddle posture, as well as a lower frequency and shorter duration of locomotion than the CON group. There was no significant difference in body weight between the CUMS and CON groups. In the attempt for the apple test, CUMS macaques showed a significantly shorter duration of attempts compared to controls. In the human intruder test, CUMS macaques showed a significantly longer duration of anxiety-like behaviors in the staring phase compared to controls. Notably, the hair cortisol levels were significantly higher in the CUMS macaques than in the controls at the endpoint and also increased significantly from baseline to the endpoint (details in Supplementary Table S1). All these findings suggested that the CUMS macaques exhibited depressive-like behaviors compared to the CON group.

snRNA-seq Profiling of the NAc in Adolescent Macaques

To identify the NAc cell types in adolescent macaques in the CUMS and CON groups, tissues from the CUMS and CON groups (five per group) were isolated for nuclear extraction and snRNA-seq. After quality control, we sequenced to an average depth of >770 million reads per sample, capturing a total of 66,554 UMIs (Supplementary Table S2). There were no significant differences between the CUMS and CON groups in the estimated number of nuclei (Wilcoxon Signed Rank Test, P = 0.893), mean reads per nucleus (Wilcoxon Signed Rank Test, P = 0.686), and median genes per nucleus (Wilcoxon Signed Rank Test, P = 0.345). After normalization and elimination of mitochondrial DNA and doublets, we obtained a total of 59,686 nuclei for analysis (Supplementary Fig. S1A-C). For dimension reduction and graph-based clustering in the initial ‘low-resolution’ clustering analysis, we applied UMAP to identify seven major cell types using a combination of known markers reported in previous studies (Fig. 2A) [12, 47]: neurons (Neu; n = 36,557; 61.25%; SYT1, SNAP25), astrocytes (Ast; n = 4,709; 7.89%; AQP4, CLDN10, MFGE8, GFAP), microglia (Mic; n = 2,720; 4.56%; CSF1R, MRC1, PTPRC, DOCK8), oligodendrocytes (Olig; n = 10,975; 18.39%; MBP, MOG, PLP1), oligodendrocyte precursor cells (OPCs; n = 3,538; 5.93%; PDGFRA, PCDH15, SOX6), and other cells including stromal cells (n = 823; 1.38%; DCN, FLT1) and monocytic cells (n = 364; 0.61%; C1QA, LYZ). The majority (98.01%) of the cells in the NAc were neurons and glial cells (Ast, Mic, Olig, and OPCs), with a ratio of 1.67:1. The non-neuronal populations of the NAc in adolescent macaques are similar to other brain regions in macaques [48], as well as the NAc in humans and mice [22, 30]. The cell type-specific marker genes were visualized using a dot plot (Fig. 2B) and a UMAP plot (Supplementary Fig. S2A–G), and the top 5 marker genes were also differentially expressed between the cell populations (Fig. 2C). In addition, there were no significant differences between the CUMS and CON groups in seven major cell types (Fig. 2B; Supplementary Table S3).

Fig. 2.

Identification of the cell types in the NAc of adolescent cynomolgus macaques. A UMAP analysis depicting 59,686 nuclei in 7 major cell types after normalization and quality control. Neu, Neuron; Ast, Astrocytes; OPCs, Oligodendrocyte precursor cells; Olig, Oligodendrocytes; Mic, Microglia; Mono, Monocytic cells; Str, Stromal cells. B Cell-types are annotated in accordance with the expression of the well-established marker genes for each cell type. The size of the dots represents the percentage of cells expressing the marker genes, and their color intensity represents the average expression of the marker genes. The marker genes displayed at the bottom are colored according to the cell types. The bar plot on the right lists the total number of each cell type in the CUMS (red) and CON (blue) groups. C Heatmap depicting the top 5 marker genes for each cell type. Yellow, genes with high expression: purple, genes with low expression in each cell type. All the marker genes below are colored according to cell types.

Cell-Type-Specific Transcriptomic Alterations in Adolescent Macaques with Depressive-Like Behaviors

To identify the DEGs in five major cell types (Neu, Ast, Mic, Olig, and OPCs), the gene expression profiles of each type were compared between CUMS and CON macaques. A total of 2,158 DEGs were identified in all five major cell types (|log2FoldChange| >0.25, FDR <0.05, min.pct >0.1), including 1,135 in Ast, 337 in Neu, 247 in Mic, 220 in OPCs, and 219 in Olig (Fig. 3A; Supplementary Table S4, Supplementary Fig. S3A-E). Among all the DEGs, 1,361 were upregulated and 797 were downregulated (Fig. 3A; Supplementary Table S4). The Ast cells (1135 DEGs) exhibited a greater number of robust DEGs than other glial cells (DEGs ranging from 219 to 247) or neurons (337 DEGs) (Fig. 3A). In addition, the majority of DEGs were specific to a single cell type, with only 314 DEGs (14.6%) shared by at least two cell types (Supplementary Fig. S4A). Nevertheless, DGKB and ENSMFAG00000040922 were downregulated in all five major cell types, whereas FKBP5 was upregulated in all four glial cell types - Ast, Mic, OPCs, and Olig (Supplementary Fig. S4B). Diacylglycerol kinase beta (DGKB) is the subtype of DGK that converts diacylglycerol to phosphatidic acid and has recently been considered a potential therapeutic target for neuronal diseases [49]. The FKBP5 gene encodes FK506-binding protein 51, which has been identified as a key regulator of endocrine stress responses, and also as a potential therapeutic target for depression [50].

Fig. 3.

DEGs of the major cell types in adolescent macaques with depressive-like behaviors. A The numbers of up-regulated (red) and down-regulated (blue) DEGs (|log2FoldChange| >0.25, FDR <0.05, min.pct >0.1) in adolescent macaques with depressive-like behaviors. B Biological pathway of GO analysis in at least 2 cell types (color intensity represents -log₁₀ FDR). ***FDR <0.001, **FDR <0.01, *FDR <0.05. C The DEGs in at least two cell types are associated with six or more depression-related terms based on DisGeNET. The colors in the gene circle indicate the degree of altered gene expression in specific cell types, while the colors of the lines denote the categories of association with diseases. D A total of 259 DEGs in at least two cell types are identified in PsyGeNET as being associated with mental illness. The term "total" represents the sum of all genes that are identified as overlapping the database for a specific disorder, and are therefore colored green. The term "100% association" denotes genes that have been positively correlated with the disease, as indicated by the blue color. The genes that are negatively associated with the disease are represented by the term "100% no association," which is colored yellow. ‘Both’ represents mixed findings (positive and negative) for a given gene associated with the disease (red). UD, use disorders; SI, substance-induced. E Protein interactions of DEGs in at least 3 cell types shown by String DB (combined score >0.5). The color in the gene circle represents the altered gene expression in cell types, while the colors of the lines represent the categories of interactions.

Pathway enrichment analysis showed that the DEGs in different cell types were primarily associated with 184 biological pathways of GO analysis in Neu, Ast, Mic, and Olig (FDR <0.05), while there was no altered biological pathway in OPCs (Supplementary Table S5A-D). The perturbed genes in Neu were predominantly associated with neurotransmitter regulation, such as negative regulation of catecholamine secretion, NMDA glutamate receptor clustering, and AMPA glutamate receptor clustering, including SYT4, SYT11, CHGA, NLGN1, RELN, and DLG4 genes. Notably, a total of 18 overlapping biological pathways were identified as enriched in at least two cell types (Fig. 3B). The process of nervous system development was found to be significantly disrupted in Neu, Ast, and Mic, including the DEGs MAP1B, NLGN1, ADGRV1 and ROBO2, which have been reported to play a role in axonal growth and neuronal migration in the central nervous system [51]. Recently, research on SNP-based heritability for MDD has identified nervous system development as the top biological process in GO analysis [52]. In addition, protein phosphorylation, which has been widely reported to be involved in the neurobiology of MDD, was disrupted in Ast and Mic [6].

The association between a number of our DEGs and MDD risk has been reported in previous studies, including DCC, MEF2C, and FKBP5 [53–55]. To further examine the relationship between previous findings in MDD and the identified DEGs of at least two cell types in this study, we utilized a disease-related gene reported in the global disease database DisGeNET [39] and the subdivided psychiatric disease database PsyGeNET [40]. Using DisGeNET, we found that a total of 68 DEGs were associated with depression-related diseases and phenotypes (Fig. 3C, Supplementary Table S6). Of these, FKBP5 was the most frequently reported depression-associated gene in previous studies (Supplementary Table S7). Using PsyGeNET, we found that a total of 678 associations among 259 DEGs were reported to be associated with mental illness in the literature (Fig. 3D, Supplementary Table S8). The highest number of associations (256 associations among 100 DEGs) were for “depressive disorders”, and followed by associations for “schizophrenia spectrum and other psychotic disorders” (208 associations among 151 DEGs) (Supplementary Table S8). Subsequently, a STRING DB network analysis was applied to investigate the protein-coding DEG interactions in at least three cell types [56]. Here, we found 23 DEGs that spontaneously assembled into 6 gene clusters (Fig. 3E, Supplementary Table S9; combined score >0.5). Gene cluster 1 was the largest cluster, consisting of 6 genes (FKBP5, HSP90A1, DYNC1H1, MAP1B, ZBTB16, and STMN1) mainly involved in the stress response, and gene cluster 2 was mainly involved with GABA receptors (Fig. 3E).

Neuron Subtype Deregulation in Adolescent Macaques with Depressive-Like Behaviors

To characterize the neuronal heterogeneity of the NAc, we conducted a detailed analysis of the cell subtypes of neurons present in adolescent macaques. All 36,557 neuronal nuclei were grouped into 11 subclusters (Fig. 4A), which were identified as 4 cell subtypes based on their established marker genes: D1-MSNs (n = 18,287; clusters 1, 2, 3, 6), D2-MSNs (n = 9,902; clusters 0, 10), inhibitory neurons (n = 5,024; clusters 4, 7, 9), and excitatory neurons (n = 3,344; clusters 5, 8). Firstly, inhibitory and excitatory neurons were identified using established markers including GAD1, GAD2, and SLC17A7 (Fig. 4B). Secondly, based on the canonical MSN marker Ppp1r1b (Fig. 4B), most of the neurons captured by snRNA-seq were MSNs, which is consistent with previous single-cell RNA sequencing (scRNA-seq) or snRNA-seq data in the NAc [22, 30]. Furthermore, all the MSNs were classified as discrete D1- and D2-MSNs based on established markers including Pdyn and Drd2, respectively (Fig. 4B). To further understand the heterogeneity among the neurons, we found that 11 sub-clusters could be distinguished by the enriched expression of marker genes (Fig. 4C; Supplementary Table S10). A number of MSNs were subtypes marked by unique genes, such as EBF1, TSHZ1, and HTR2C. Notably, there was also a small number of Nxph1-expressing or LHX6-expressing inhibitory neurons in the NAc. All the cell numbers of re-clusters were represented similarly in the CUMS and CON groups (Supplementary Fig. S5A). A total of 375 and 427 DEGs in D1-MSNs (Supplementary Table S11A) and D2-MSNs (Supplementary Table S11B) between CUMS and CON groups respectively (|log2FoldChange| >0.25, FDR <0.05, min.pct >0.1), and the majority of the DEGs were overlapped (Supplementary Fig. S5B). The overlapping DEGs between D1-MSNs and D2-MSNs were mainly enriched in the GABAergic synapse, including GABRA1, GABRG2, GABRG3, PRKCB, PRKCG, ADCY3, ADCY5 and GAD2 (Fig. 4D; Supplementary Table S12). Notably, QuSAGE analysis also revealed the activation of GABAergic synapses and the inhibition of D-glutamine and D-glutamate metabolism gene sets in the D1- and D2-MSNs of adolescent macaques with depressive-like behaviors (Fig. 4E). A growing body of evidence suggests that an imbalance between glutamate and GABA levels may play a role in the etiology of MDD [57].

Fig. 4.

The transcriptional features of subpopulations of neurons in the NAc in adolescent macaques with depressive-like behaviors. A Re-UMAP analysis of 36,557 neuronal nuclei, showed that 11 clusters of neurons constitute four distinct cell subtypes: D1-expressing medium spiny neurons (D1-MSNs, clusters 1, 2, 3, 6), D2-expressing medium spiny neurons (D2-MSNs, clusters 0, 10), inhibitory neurons (clusters 4, 7, 9), and excitatory neurons (clusters 5, 8). B Re-UMAP analysis showing the expression patterns of established inhibitory neuron markers (GAD1 and GAD2), the excitatory neuron marker (SLC17A7), MSN marker (Ppp1r1b), D1-MSN marker (Pdyn), and D2-MSN marker (Drd2). The color intensity of the dots represents the gene expression level of established neuron markers. C Violin plots showing the expression patterns of the 11 cluster-specific gene markers of neurons. Different neuron subtypes are color-coded as in A. The gene expression level is based on normalized data. D Schematic of the overlapped DEGs between D1-MSNs and D2-MSNs in GABAergic synapses. Up-regulated DEGs in adolescent depressed macaques are colored red: GABRA1, GABRG2, GABRG3, and GAD2; Down-regulated DEGs in adolescent depressed macaques are colored green: PRKCB, PRKCG, ADCY3, and ADCY5. E QuSAGE analysis showing the average activity level of genes enriched in GABAergic synapse gene sets (left) and D-glutamine and D-glutamate metabolism gene sets (right) of D1-MSNs and D2-MSNs. Red, activation of the gene sets; blue, inhibition of the gene sets; grey, no significant difference.

Astrocyte Subtype Deregulation in Adolescent Macaques with Depressive-Like Behaviors

In order to gain insight into the heterogeneity of astrocytes in the NAc, a comprehensive analysis of the various astrocyte cell subtypes present in adolescent macaques was conducted. We found that Ast dominated the majority (52.6%) of DEGs across all the five major cell types. Therefore, the Ast snRNA-seq data were also clustered in order to investigate the subpopulation deregulation of Ast in adolescent CUMS macaques. All 4709 Ast nuclei were classified into 8 clusters (clusters 0 to 7), which were identified as five cell subtypes based on their established marker genes, including Ast1 (clusters 0, 1, 2, 6), Ast2 (cluster 3), Ast3 (cluster 5), Ast4 (cluster 4), and Ast5 (cluster 7) (Fig. 5A). The violin plots display the expression of cell subtype-specific marker genes for each cluster (Fig. 5B). Notably, Ast4 was marked by FOS family members (FOS, FOSB, FOSL1), which have been identified as reliable markers for the brain under stress conditions [58]. The number of Ast5 cells, which express high levels of CDK6, was significantly lower in the CUMS group (Fig. 5C; P = 0.043). Furthermore, the number of Ast5 subtypes was also negatively correlated with both the frequency (r = −0.805; P = 0.005) and duration (r = −0.870; P = 0.001) of huddle posture, which is a core depressive-like behavior. A recent study has indicated that CDK6 plays a role in promoting the expansion of outer radial glia, neocortical folding, and increased motility during cell differentiation [59]. Similarly, the D-glutamine and D-glutamate metabolism gene sets were also inhibited in the five Ast cell subtypes of adolescent macaques with depressive-like behaviors, in a manner consistent with the patterns in the D1- and D2-MSNs (Fig. 5D). FKBP5 was found to be enriched in the CUMS group from Ast1 to Ast4 (Supplementary Table S13), with the majority of Ast (62.0%) that expressed FKBP5 belonging to the CUMS group (Fig. 5E). Similarly, FKBP5 was also enriched in Mic cells of the CUMS group (Supplementary Fig. S6A) in comparison to the CON group (Supplementary Fig. S6B). To gain further insight into the interactions between the Ast subtypes and D1/D2-MSNs, we applied the CellPhoneDB tool to identify cell-cell communication under CUMS and CON status. A total of 324 activated cell-cell communications were identified as activated in the context of CUMS (Supplementary Fig. S7A, Supplementary Table S14), 162 activated cell communications in CON status (Supplementary Fig. S7B, Supplementary Table S15), and 844 activated cell communications in both depression and normal status (Supplementary Table S16). In CUMS status, BMP6 demonstrated a notable degree of activation between Ast and D1-MSNs, while ERBB4 signaling was significantly activated between Ast and D1/D2-MSNs (Fig. 5F). The expression of BMP6 is significantly elevated in the brain of MDD patients [60], and BMP signaling in the hippocampus is also influenced by antidepressant treatment, affecting neural progenitor cell proliferation and behavior [61]. These findings are also consistent with previous studies suggesting that activated ERBB4 regulates depressive-like behaviors, particularly in dopamine neurons [62, 63]. In the CON status, the ephrin ligand–receptor (EFN-EPH) signaling was found to be significantly activated between Ast and D1/D2-MSNs (Fig. 5G). EFN-EPH signaling plays a role in regulating neuronal migration, guiding axons to reach their final destinations, and also mediating synapse formation and plasticity [64]. Recently, evidence has emerged suggesting a role for EFN-EPH signaling in the etiology of depression [65, 66].

Fig. 5.

The transcriptional features of subpopulations of astrocytes in NAc in adolescent macaques with depressive-like behaviors. A Re-UMAP analysis of 4709 Ast nuclei, shows that 8 clusters of astrocytes constitute 5 cell types: Ast1 (clusters 0, 1, 2, 6), Ast2 (cluster 3), Ast3 (cluster 5), Ast4 (cluster 4), and Ast5 (cluster 7). B Violin plots showing the expression patterns of the 8 clusters-specific gene markers of astrocytes. Different astrocyte subtypes are color-coded as in A. The gene expression level is based on normalized data. C Numbers of Ast5 cells is lower in the adolescent CUMS group (red) compared with the adolescent CON group (blue). D QuSAGE analysis showing the inhibition of D-glutamine and D-glutamate metabolism gene sets in the adolescent CUMS group (red, activation of the gene sets; blue, inhibition of the gene sets; grey, no significant difference). E The expression of FKBP5 in astrocytes of the NAc in the adolescent CUMS group (left) and adolescent CON group (right). The color intensity of the dots represents the gene expression level of FKBP5 in astrocytes, and a higher degree of color intensity indicates a higher expression level of FKBP5. F The cell-cell communications in CUMS group between astrocytes and D1/D2-MSNs based on the CellPhoneDB. The size of the dots represents the -log₁₀P-value, and the color intensity represents the mean value of the ligand (red) and receptor (blue). G Cell phone in the CON group between astrocytes and D1/D2-MSNs based on CellPhoneDB. The size of the dots represents the -log₁₀P-value, and the color intensity of the dots represents the mean value of the ligand (red) and receptor (blue).

Multi-Omics Data Integration of snRNA-Seq, Bulk RNA Sequencing, and Proteomics

To validate the completeness of our snRNA-seq data and to capture changes in both mRNA and protein content of the NAc in adolescent macaques with depressive-like behaviors, we generated RNA sequencing and proteomics datasets from the bulk sample from the NAc of the same macaques in the CUMS group (n = 5) and the CON group (n = 5). Furthermore, a high level of concordance was found between the bulk and single-nucleus RNA-sequencing data (Fig. 6A, R = 0.66, P = 2.2e^-16). A total of 1,186 DEGs, including 768 up-regulated and 418 down-regulated DEGs (P <0.05, Fig. 6B, Supplementary Table S17), and 847 differential-expressed proteins (DEPs), including 654 up-regulated and 193 down-regulated DEPs (P <0.05, Fig. 6C, Supplementary Table S18) were found in the CUMS group in the NAc bulk RNA sequencing and proteomics. However, only 55 genes exhibited alterations at both the mRNA and protein levels. Of these, 50 (90.9%) displayed the same direction of change in both the CUMS group with the CON group (Supplementary Table S19). Then, the correlation analysis showed that 27 of the 55 overlapped DEGs/DEPs were significantly associated with huddle posture (a core depressive-like behavior) at both the mRNA and protein levels (Supplementary Table 19). Notably, 17 out of the 55 genes were also altered in at least one cell type in snRNA-seq including FKBP5, which was up-regulated in the snRNA-seq data of Ast, Mic, OPCs, and Olig, as well as in the bulk mRNA-seq and proteomics data from the CUMS group (Fig. 6D). In GO analysis, a total of 35 and 23 biological processes were identified as altered in the NAc bulk RNA sequencing (FDR <0.05, Fig. 6E, Supplementary Table S20) and proteomics (FDR <0.05, Fig. 6F, Supplementary Table S21), respectively. Nevertheless, no overlap was found between the biological processes identified in the bulk RNA sequencing and proteomics datasets.

Fig. 6.

Integration analysis of snRNA-seq, bulk mRNA-seq, and proteomics data. A The linear correlation between the gene expression profiles of snRNA-seq and bulk mRNA-seq in the NAc. Each point represents a gene. The horizontal coordinate indicates the expression of the gene in snRNA-seq, while the vertical coordinate indicates the expression of the gene in bulk mRNA-seq. B Volcano plot of the differentially-expressed genes in bulk mRNA-seq in the NAc. Blue, down-regulated genes; red, up-regulated genes; grey, genes that are not significantly differentially expressed. C Volcano plot of the differentially expressed proteins in bulk proteomics in the NAc (blue, genes that are down-regulated in CUMS group; red, genes that are up-regulated in the CUMS group; grey, genes that are not significantly differentially expressed). D The overlapping differentially expressed genes or proteins in snRNA-seq, bulk mRNA-seq, and proteomics data (red, up-regulated in the CUMS group; blue, down-regulated in the CUMS group). ***P <0.001, **P <0.01, *P <0.05. E The biological processes of GO analysis in the differentially-expressed genes of the NAc. The sizes of the nodes indicate the number of associated genes, while the color gradient (red to green) represents the significance of enrichment, with red indicating more significant terms. F The biological processes of GO analysis in the differentially-expressed proteins of the NAc. The size of the nodes indicates the number of associated genes, while the color gradient (red to green) represents the significance of enrichment, with red nodes indicating more significant terms.

Validation of FKBP5 Expression Levels in the NAc of Adolescent CUMS Macaques and Rats

Subsequently, qRT-PCR, western blotting, and immunofluorescence were applied to validate the expression of FKBP5 in the NAc of adolescent macaques and rats exhibiting depressive-like behaviors. Following stress exposure, the CUMS group of rats (n = 8) exhibited a significantly higher incidence of depressive-like behaviors than the CON group (n = 8), accompanied by a lower sucrose preference (Fig. 7A) and higher immobility times in the forced swimming test (Fig. 7B). However, no significant difference was found between the results of the open field and elevated plus-maze tests (Supplementary Fig. S8A). Taken together, these behavioral data collectively indicate that CUMS induced depression-like behaviors in rats. The qRT-PCR results demonstrated that the mRNA level of FKBP5 in the NAc was elevated in the CUMS group of rats (Supplementary Fig. S8B). Furthermore, western blotting analysis revealed that the protein expression of FKBP5 was also elevated in the CUMS group of both macaques (Fig. 7C) and rats (Fig. 7D). Then, we applied immunofluorescence to the NAc of CUMS rats. Our findings revealed that FKBP5 in the NAc of the CUMS group exhibited higher expression levels than those in the CON group (Fig. 7E). The data indicated that the expression of FKBP5 in the NAc is elevated following chronic stress.

Fig. 7.

Validation of FKBP5 expression levels in the NAc of adolescent CUMS macaques and rats. A A lower sucrose preference in the SPT (left) in the CUMS group than in the control group in rats. B Longer immobility time in the FST (right) in the CUMS group than in the control group in rats. C Protein expression levels of FKBP5 by western blotting are elevated in the CUMS group of macaques. D Protein expression levels of FKBP5 by western blotting are elevated in the CUMS group of rats. E Protein expression levels of FKBP5 by immunofluorescence staining in the CUMS and CON groups of rats. All results are represented as the mean ± SD; ***P <0.001, **P <0.01, *P <0.05.

Discussion

A single-nucleus transcriptome from the NAc of adolescent macaques exhibiting depressive-like behaviors identified cell-type-specific gene dysregulation, which predominantly occurred in astrocytes. The expression of DGKB was significantly reduced in all neuronal and glial cells of the depressed macaques, whereas FKBP5 was significantly elevated in all of their glial cell types, but not in neurons. The subpopulation analysis of neurons and astrocytes revealed the activation of GABAergic synapse gene sets and the inhibition of D-glutamine and D-glutamate metabolism in specific neuronal and astrocytic subtypes in depressed macaques. Furthermore, the cell-cell communication between the subtypes of Ast and D1/D2-MSNs was also dysregulated in depressed status, as evidenced by the altered expression of BMP6 and ERBB4. Furthermore, the results of NAc bulk sample transcriptomics and proteomics were consistent with those of the snRNA-seq.

In light of the intricate pathogenesis of depression, which encompasses the malfunction of assorted cell types within the brain, it is imperative to investigate the underlying pathogenesis at the single-cell resolution [67]. SnRNA-seq offers novel approaches to comprehensively characterize the diversity of both the brain cell types and also the cell-cell communications [68]. Nagy et al. identified a potential association between OPCs and deep-layer excitatory neurons in the dorsolateral prefrontal cortex (dlPFC) of adult patients with depression, suggesting a regulatory role for biological pathways such as glucocorticoid receptor regulation [12]. In our previously published study, we identified the role of dlPFC microglia in neuroinflammation in adult macaques exhibiting associated depressive behaviors associated with social stress [18]. Although there are notable differences between adolescent and adult depression [8, 9], there has been a paucity of studies focusing on the pathogenesis of adolescent depression at the single-cell resolution.

A substantial body of research has elucidated the involvement of D1-MSNs and D2-MSNs in depression-related outcomes, depending on the signaling pathways of reward and aversion [69, 70]. The present study revealed that the D1-MSNs and D2-MSNs exhibited comparable transcriptional dysfunctions in the NAc of depressed adolescent macaques, primarily associated with activated GABAergic synapses and inhibited D-glutamine and D-glutamate metabolism gene sets. MSNs are a particular type of GABAergic inhibitory cell [71], and a deficit in the neurotransmitters GABA and glutamate is widely regarded as a potential etiological factor in depression [72, 73]. The evidence is mounting that the homeostasis of glutamatergic and GABAergic systems may contribute to the etiology of depression, with changes occurring in a variety of brain regions [57]. Our findings regarding the glutamatergic and GABAergic systems in the NAc using nuclear magnetic resonance spectroscopy are consistent with those of previous studies of the same region in patients with depression [74, 75]. Given the time lag and limited efficacy of current antidepressants for adolescent depression, new rapid-acting agents targeting the glutamate and GABA systems may offer superior therapeutic interventions [73]. Recent research has indicated that ketamine or scopolamine, which act on GABAergic interneurons in the NAc, may represent a promising avenue for the development of new antidepressant treatments for MDD [76–78]. All the above pieces of evidence reveal new potential antidepressant targets within the glutamate and GABA systems for the treatment of depression in children and adolescents.

Recent research has also indicated that glial cells, such as astrocytes [79], microglia [80], and oligodendrocytes [81], are significantly associated with depression. The present study underscores the significance of glial dysregulation in adolescent depression, with particular emphasis on astrocyte function, which is a primary contributor to the majority of DEGs in the NAc. Notably, astrocytes play an important role in controlling the levels of neurotransmitters in the vicinity of synapses, including GABA and glutamate [82]. Cao et al. have previously reported that astrocytic glucocorticoid receptor-mediated depressive-like behaviors in adult mice are associated with the release of endogenous ATP [79, 83]. Leng et al. reported that astrocyte-mediated neuroinflammation in the NAc is associated with depressive-like behaviors in mice [84]. Furthermore, our findings reinforce the hypothesis that astrocyte dysfunction in a depressed condition may result in impaired glutamate uptake, metabolism, and recycling at synapses [85]. Interestingly, we found dysregulation of cell-cell communication between astrocytes and MSNs in BMP6 and its receptors, which have been previously associated with psychiatric disorders such as depression and autism [61, 86]. The underlying pathogenesis of adolescent depression in the NAc may result from the dysfunction of astrocytes and MSNs.

FK506-binding protein 51, which is encoded by the FKBP5 gene, acts as a co-chaperone of heat-shock protein HSP90, which plays a pivotal role in determining the sensitivity of the glucocorticoid receptor [87]. In our snRNA-seq dataset, we found that FKBP5 was exclusively up-regulated in all the glial cells, and also in the bulk RNA sequencing and proteomics. In addition, HSP90AB1 was also up-regulated in neurons, astrocytes, oligodendrocytes, and OPCs. The association between gene polymorphism and the DNA methylation of FKBP5 with depression has been extensively documented [50]. Furthermore, the gene expression of FKBP5 has been shown to predict the efficacy of antidepressant treatments for depression [87], and selective inhibitors of FK506-binding protein 51 have the potential to become a new class of antidepressants, as they may improve neuroendocrine feedback [88]. In a recent study, Gaali and colleagues demonstrated that FKBP5 plays a role in the acute stress response in the paraventricular nucleus of the hypothalamus, particularly in Sim1⁺ neurons in mice [88]. We hypothesize that chronic stress may lead to over-activation of the HPA axis, and the key HPA axis gene FKBP5 in the NAc is upregulated, leading to depression-like behavior via the downstream pathway of autophagy, AKT signaling or NF-kB signaling [50]. The alterations in FKBP5 expression in glial cells provide insights into the disruption of the stress response in the NAc of macaques exhibiting depressive-like behaviors.

It should be noted that the current study is not without limitations. Firstly, in a manner similar to a previous snRNA-seq study of adult male MDD patients [12], only adolescent male macaques were included in this study, which has an impact on the generalizability of our results given that there are differences in brain transcriptomic changes between male and female [89]. Secondly, snRNA-seq is regarded as an alternative to scRNA-seq due to its suitability for use with frozen tissues, while a small proportion (~1%) of genes are depleted in snRNA-seq compared to scRNA-seq [90]. Thirdly, genes with low expression may be disregarded, as they are not captured by snRNA-seq [12]. Fourthly, this study does not include a spatial transcriptomics analysis, which could provide valuable insights into gene expression across thousands of cells in the context of spatial information [91].

In conclusion, this is the first study to contrast the single-nucleus transcriptomic characteristics in the NAc of adolescent macaques exhibiting depressive-like behaviors. We mapped the depression-associated genes and functions in adolescent macaques at single-cell resolution and identified altered cell-cell communications between astrocytes and D1/D2-MSNs in depressed status. Altogether, these new findings shed light on the cell-type-specific gene dysregulations in the NAc of adolescent depression and may provide new potential antidepressant targets for this age group.

Supplementary Information

Below is the link to the electronic supplementary material.

Conflict of interest

The authors declare that they have no conflicts of interest with the contents of this article.