Introduction
Bipolar disorder (BD) is a severe psychiatric condition predominantly affecting young individuals, marked by cyclical manic and depressive episodes. Beyond its profound cognitive and functional impacts, BD also elevates the risk of mortality (McIntyre et al., Reference McIntyre, Berk, Brietzke, Goldstein, Lopez-Jaramillo, Kessing and Mansur2020). While there has been considerable research into the pathophysiology of BD, its precise origins remain elusive. Identifying biomarkers rooted in the disorder’s underlying mechanisms is essential for enhancing both its diagnosis and therapeutic interventions.
MicroRNAs (miRNAs) are small, ~22 nucleotides in length, non-coding single-stranded RNAs that regulate gene expression by binding to messenger RNA (mRNA) (Bartel, Reference Bartel2004). Increasing evidence indicates that alterations in circulating miRNAs are observed in BD, suggesting a significant role in the disorder’s pathogenesis (Camkurt et al., Reference Camkurt, Karababa, Erdal, Kandemir, Fries, Bayazıt and Selek2020; Fries, Carvalho, & Quevedo, Reference Fries, Carvalho and Quevedo2018; Tabano et al., Reference Tabano, Caldiroli, Terrasi, Colapietro, Grassi, Carnevali and Buoli2020). For instance, the differentially expressed profiles of circulating miRNAs in BD patients underscored their potential involvement in neurodevelopmental and metabolic processes (Tabano et al., Reference Tabano, Caldiroli, Terrasi, Colapietro, Grassi, Carnevali and Buoli2020). Nevertheless, the concentrations of circulating miRNAs can be swayed by numerous factors and are strictly regulated by RNases, posing a challenge in accurately capturing pathological changes in the central nervous system (CNS). To overcome this, several studies have delved into analyzing miRNA expression levels in postmortem brain tissues from individuals with BD (Azevedo et al., Reference Azevedo, Carter, Meng, Turner, Dai, Schatzberg and Thompson2016; Choi et al., Reference Choi, Kao, Itriago, Zhan, Kozubek, Hoss and Delalle2017; Guella et al., Reference Guella, Sequeira, Rollins, Morgan, Torri, van Erp and Vawter2013). Yet, this line of study has been hampered by hurdles such as the difficulty in obtaining relevant brain tissue samples and the existence of only a handful of small-scale studies in the literature. Given these constraints, there is an urgent call to pinpoint peripheral biomarkers that can adeptly mirror alterations in the CNS, offering insights into the pathogenesis of the disorder.
Extracellular vesicles (EVs) are released by cells into the extracellular space and circulation (Kalluri & LeBleu, Reference Kalluri and LeBleu2020), containing a high concentration of miRNAs and specialized sorting mechanisms that can impact recipient cell processes (Garcia-Martin et al., Reference Garcia-Martin, Wang, Brandão, Zanotto, Shah, Kumar Patel and Kahn2022). While previous research has identified changes in miRNAs within EVs in the blood of individuals with mental disorders (Du et al., Reference Du, Yu, Hu, Li, Wei, Pan and Cheng2019; Fries et al., Reference Fries, Lima, Valvassori, Zunta-Soares, Soares and Quevedo2019; Wei et al., Reference Wei, Xie, Mao, Zou, Liao, Liu and Cheng2020), these EVs have a complex origin in peripheral blood, with only a small portion directly originating from brain tissue (Zhang, Yu, et al., Reference Zhang, Yu, You, Jiang, Wu, DeTure and Ikezu2023). However, recent advancements allow for the isolation of EVs derived from the brain in blood samples, utilizing specific brain cell exosome protein markers like neuronal cell adhesion molecule 1 (NCAM1). Moreover, accumulating evidence suggests that miRNAs from neural-derived EVs play key roles in various pathophysiological mechanisms associated with neuropsychiatric disorders, including Parkinson’s disease (Dutta, Hornung, Taha, & Bitan, Reference Dutta, Hornung, Taha and Bitan2023), Alzheimer’s disease (Leng et al., Reference Leng, Yuan, Pan, Su, Wang, Xue and Zhang2022), and Major Depressive Disorder (Saeedi et al., Reference Saeedi, Nagy, Ibrahim, Théroux, Wakid, Fiori and Turecki2021). Altered miRNA expression patterns in these disorders have implications for neuronal survival, neuroplasticity, and neuroinflammation (Leng et al., Reference Leng, Yuan, Pan, Su, Wang, Xue and Zhang2022; Varcianna et al., Reference Varcianna, Myszczynska, Castelli, O’Neill, Kim, Talbot and Ferraiuolo2019; Zhang, Wu, et al., Reference Zhang, Wu, Tang, Chen and Wu2023). Thus, probing the miRNAs contained within neural-derived EVs offers a promising avenue for deepening our understanding of the pathophysiological mechanisms of BD.
In this study, we isolated neural-derived EVs from the plasma of patients of both BD patients and HC. Using RNA sequencing, we profiled the miRNA expression within these neural-derived EVs. Through differential expression analyses, we discerned marked distinctions between patients with BD and HC. Notably, miR-143-3p exhibited the highest recurrence frequency in the differential expression analysis. Considering the correlation between miR-143-3p and inflammatory responses (Pan et al., Reference Pan, Li, Yu, Xie, Li, Duan and Zhou2021; Y. Wang et al., Reference Wang, Li, Shi, Wang, Xu, Li and Liu2020; Y. Wang et al., Reference Wang, Zhang, Wang, Zhang, Ye, Wang and Wang2023), as well as the widely supported role of inflammatory processes in the brain contributing to the onset and progression of BD (Jones, Vecera, Pinjari, & Machado-Vieira, Reference Jones, Vecera, Pinjari and Machado-Vieira2021; Rosenblat et al., Reference Rosenblat, Brietzke, Mansur, Maruschak, Lee and McIntyre2015; Saccaro, Crokaert, Perroud, & Piguet, Reference Saccaro, Crokaert, Perroud and Piguet2023). We manipulated miR-143-3p expression levels in the human microglia clone 3 cell line (HMC3) to further investigate the potential role of miR-143-3p in the pathogenesis of BD. To delve deeper into the functional roles of these distinct miRNAs, we carried out enrichment analyses and employed the weighted gene coexpression network analysis (WGCNA). To enhance the credibility of our findings, we validated our results by analyzing the expression of miRNA target genes in the prefrontal cortex of individuals with BD and healthy controls, using data obtained from PsychENCODE.
Methods
Subjects
From 2022 to 2023, we recruited patients diagnosed with BD from the Affiliated Mental Health Center at Zhejiang University School of Medicine. Diagnoses adhered to the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) criteria and were confirmed through evaluations conducted by two trained psychiatrists using the Structured Clinical Interview for DSM-5 (SCID-5), ensuring a standardized diagnostic process. Symptom severity was assessed using the Positive and Negative Syndrome Scale (PANSS), the 11-item Young Mania Rating Scale (YMRS), the 17-item Hamilton Depression Scale (HAMD), and the 14-item Hamilton Anxiety Scale (HAMA). The criteria for subject inclusion and exclusion are detailed in the Supplementary Methods.
Ensuring ethical adherence, our study received permission from the Institutional Ethics Committee at the Affiliated Mental Health Center of Zhejiang University School of Medicine. Moreover, before participating, all involved individuals provided their informed consent in written form, affirming the study’s alignment with the Declaration of Helsinki’s principles.
Isolation and characterization of plasma neural-derived EVs
For all participants, the peripheral blood was collected from which plasma samples were extracted and subsequently stored at −80 °C. Neural-derived EVs were then isolated from the plasma by employing a biotinylated antibody targeting human NCAM1 (Santa Cruz Biotechnology, TX, USA). To verify the authenticity of the isolated EVs, we undertook Western blot analysis. This involved using specific antibodies against EVs’ marker (CD9 and Tumor susceptibility gene 101 (TSG101), in conjunction with the neural markers NCAM1 and cell adhesion molecule L1 (L1CAM) (Supplementary Figure S1(a)). A polyclonal anti-human CD9 antibody (Proteintech, IL, USA), a polyclonal anti-human TSG101 antibody (Sino Biological, TX, USA), a polyclonal anti-human NCAM1 antibody (Proteintech, IL, USA) and a monoclonal anti-human L1CAM antibody (Thermo Fisher Scientific, MA, USA) were used according to the manufacturer’s instructions. As a comparative measure, the beads-only group was utilized as a negative control. In addition, nanoparticle tracking analysis (NTA) and transmission electron microscopy (TEM) were employed to determine the EVs’ size and morphology (Supplementary Figures S1(b,c)). For an in-depth methodology, please refer to the Supplementary Methods.
RNA extraction, library construction, and illumina sequencing
Following the isolation of EVs, total RNA was extracted using the QIAGEN exoRNeasy Midi Kit (Cat#77144, Qiagen, Hilden, Germany). To ensure the integrity of the extracted RNA, the quality of RNA was assessed via the Agilent 4200 TapeStation (Agilent Technologies, Santa Clara, CA, USA), following established protocols. For sequencing purposes, the libraries were constructed using the QIAseq miRNA Library Kit tailored for Illumina (Cat #331505, Qiagen, Hilden, Germany), strictly adhering to the manufacturer’s instructions. Subsequently, sequencing was carried out on the Illumina NovaSeq platform (Illumina, CA, USA). All sequencing analyses were conducted at Shanghai Nyuen Biotechnology Co., Ltd.
Differentially expressed and bioinformatic analyses
For differential expression, we started by tallying the read counts of EVs’ miRNAs. MiRNAs with total read counts <10 across all samples were excluded from further analysis. We utilize the DESeq2 R package (Love, Huber, & Anders, Reference Love, Huber and Anders2014) for the differential expression assessment. A miRNA was deemed as significantly differentially expressed if it displayed a false discovery rate (FDR) Q value <0.05 (corrected p-value) and a log2 fold change ˃0.263 (upregulated) or <−0.322 (downregulated). Setting thresholds at a log2 fold change >0.263 or <−0.322 when using DESeq2 is meaningful for various downstream analysis tasks and has been widely used in previous studies (Cheng et al., Reference Cheng, Ander, Jickling, Zhan, Hull, Sharp and Stamova2020; Lip et al., Reference Lip, Boekschoten, Hooiveld, van Pampus, Scherjon, Plösch and Faas2020; Love et al., Reference Love, Huber and Anders2014). Beyond contrasting the entire patient cohort with the HC, separate comparisons were conducted for the manic (BD-M) and depressive (BD-D) patient subgroups versus HC. This was done to discern if any specific miRNA changes could be identified as state markers.
The case and control samples were randomly divided into two separate groups at a 7:3 ratio, resulting in a discovery cohort of 59 cases and 27 controls, and a replication cohort of 26 cases and 12 controls. Differentially expressed analysis was performed on the discovery cohort, and any significant miRNA identified here were subsequently tested in the replication cohort. This iterative process was repeated 10 times, and consistently replicated significant miRNAs were documented.
For the identification of potential target genes of miRNAs, we utilized the MiRTarBase (Huang et al., Reference Huang, Lin, Cui, Huang, Tang, Xu and Huang2022), a comprehensive online database of miRNA-target interactions. To ensure the reliability of our findings, we only considered target genes that were backed by robust experimental validations. These validations included techniques such as reporter assays, Western blotting, and quantitative real-time polymerase chain reaction (qPCR).
The other bioinformatic analyses, including Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis, Gene Ontology (GO) enrichment analysis, and Weighted Gene Coexpression Network Analysis (WGCNA), are presented in the Supplementary Methods.
Vitro experiments
HMC3 cells were cultured and transfected with miR-143-3p mimic/inhibitor and their corresponding negative controls (mimic NC/inhibitor NC). After 24 hours, total RNA was extracted for qPCR to evaluate mRNA expression of interleukin 1β (IL-1β), IL-6, and nucleotide-binding oligomerization domain-like receptor protein 3 (NLRP3), and cell proliferation was assessed using the Cell Counting Kit-8 assay (CCK-8). Following 48 hours of transfection, protein lysates were prepared for Western blotting analysis to detect the expression levels of inducible nitric oxide synthase (iNOS) and hexokinase 2 (HK2). For further details, please refer to the Supplementary Methods.
Additional statistical analyses
For categorical data, such as gender distribution across different groups, we employed the chi-square test to ascertain any statistically significant differences between the patients and HC as well as between the various patient subgroups. For continuous data, such as age and scale scores, the Mann–Whitney U test was used to determine any significant disparities between the groups. This non-parametric test was chosen, considering that the data might not follow a normal distribution. All statistical tests were two-sided and a p-value of <0.05 was considered statistically significant. All the above statistical computations were performed using R software (Version 4.3.1).
Results
Demographic and clinical characteristics
In our study, a total of 85 patients with BD and 39 HC were enrolled. As shown in Table 1, there were no statistically significant differences in terms of sex distribution (p = 0.960) and average age (p = 0.152) between the BD patients and healthy controls. When it comes to the clinical characteristics within the BD group, as expected, 51 patients in the manic/hypomanic phase (vs. 34 patients in the depressive phase) exhibited higher scores on the YMRS scale, which indicates heightened manic symptoms. On the contrary, they had reduced scores on the HAMD and HAMA scales, denoting decreased levels of depression and anxiety, respectively. PANSS scores, which evaluate positive and negative symptoms in psychiatric disorders, did not significantly differ between the depressive and manic/hypomanic groups. At the point of enrollment, all patients were receiving medication, with the distribution as follows: lithium (49.4%), anticonvulsants (61.2%), antidepressants (17.6%), and antipsychotics (96.5%).
Clinical and demographic characteristics of subjects
Note: Data are presented as the mean ± standard deviation unless otherwise stated.
a Mann–Whitney U test.
b Chi-square test.
BD, bipolar disorder; BD-M, patients with manic-phase BD; BD-D, patients with depressive-phase BD; PANSS, Positive and Negative Syndrome Scale; YMRS, 11-item Young Mania Rating Scale; HAMD, 17-item Hamilton Depression Scale; HAMA, 14-item Hamilton Anxiety Scale; HC healthy control.
Differentially expressed miRNAs in neural-derived EVs
In the investigation of miRNAs in neural-derived EVs in BD, we compared the abundance of miRNAs in plasma neural-derived EVs from both patients and HC. Thirteen miRNAs that were upregulated (FDR <0.05 and log2 fold change ˃0.263) and one that was downregulated (FDR <0.05 and log2 fold change <−0.322) (Figure 1a and Supplementary Table S1). Notably, all but one differentially expressed miRNA (miRNA-200c-3p) demonstrated different expression between the BD-M and controls (Figure 1b). When considering the BD-D subgroup, 10 of the 14 miRNAs displayed significant differences compared with the controls. Interestingly, the expression patterns of these miRNAs remained consistent between the BD-D and BD-M groups. To gain a more comprehensive insight, we referred to the MiRTarBase for a summary of the target genes of each differentially expressed miRNA. These miRNAs underwent validation using various techniques such as reporter assay, Western blot, and qPCR (Supplementary Table S2). Furthermore, KEGG enrichment analyses for each differentially expressed miRNA unveiled several significant pathways, with the top 10 pathways detailed in Supplementary Table S2.
Differentially expressed of neural-derived EVs’ microRNAs in patients with BD. (a) Volcano plot of differentially expressed neural-derived EVs’ microRNAs in BD and control. Not Sig, not significant. Up, up-regulated. Down, down-regulated. (b) Heat maps of differentially expressed neural-derived EVs’ miRNAs in BD-M, BD-D, BD versus HC. BD, bipolar disorder; BD-M, patients with manic-phase BD; BD-D, patients with depressive-phase BD; HC, healthy controls. * p < 0.05; ** p < 0.01; *** p < 0.001.
KEGG and GO enrichment analyses of differentially expressed miRNAs
The KEGG enrichment analysis revealed that the target genes of differentially expressed miRNAs were mainly enriched in the PI3K/AKT signaling pathway, Axon guidance, and Focal adhesion (Figure 2a, Supplementary Table S3). In terms of GO analysis, significant enrichment was observed in key biological processes such as regulation of transcription by RNA polymerase II, nervous system development, and regulation of transcription, DNA-templated (Figure 2b, Supplementary Table S4). Regarding cellular composition, in addition to common components like cytoplasm, nucleoplasm, and nucleus, terms like axon, cell junction, and synapse were also notably enriched. Finally, in molecular function, there was a notable enrichment in terms related to protein binding, DNA-binding transcription factor activity, and RNA polymerase II cis-regulatory region sequence.
KEGG and GO enrichment analyses of differentially expressed microRNAs in BD. (a) KEGG enrichment analysis results. (b) GO enrichment analysis results.
MiR-143-3p was identified as a potential functional miRNA in neural-derived EVs.
MiR-143-3p stands out as the sole differentially expressed miRNAs that showed downregulation in patients. To ensure the reproducibility of miR-143-3p in the results of differential analysis, we employed a strategy that involved random splitting of subjects into discovery and replication cohorts. Notably, miR-143-3p emerged as a significant player, given its consistent recurrence in our analyses. This miRNA not only recorded the highest number of recurrences but was also successfully replicated six times (Supplementary Table S5). To uncover the potential contributions of miR-143-3p to BD pathology, we analyzed the expression of its target genes in the prefrontal cortex of patients with BD patients and healthy controls from PsychENCODE (D. Wang et al., Reference Wang, Liu, Warrell, Won, Shi, Navarro and Gerstein2018). Of the 34 target genes associated with miR-143-3p, data for 32 genes was available in the PsychENCODE. It is worth mentioning that the remaining two, namely CARMN and SCHLAP1, are not protein-coding genes (Supplementary Table S6). When we compared the gene expression profiles, we observed that 10 out of the 32 target genes exhibited significant differences between BD patients and controls, after corrections were made for multiple comparisons (the Benjamin Hochberg-corrected p value <0.05). Further details are shown in Supplementary Figure S3 and Supplementary Table S7.
Dysregulated miR-143-3p in microglia enhanced inflammatory responses and cell proliferation
To explore the impact of miR-143-3p on the pathogenesis of neuroinflammation in BD, we transfected a miR-143-3p mimic or inhibitor into HMC3 cells, a widely used model for human microglia in research (Dello Russo et al., Reference Dello Russo, Cappoli, Coletta, Mezzogori, Paciello, Pozzoli and Battaglia2018). The miR-143-3p mimic notably increased miR-143-3p expression, while the inhibitor decreased its expression in HMC3 cells (Supplementary Figure S4). Our results revealed elevated levels of IL-1β, IL-6, and NLRP3 following transfection with the miR-143-3p mimic and inhibitor after 24 hours (Figure 3a). Cell proliferation was significantly enhanced with either overexpression or underexpression of miR-143-3p, as shown by CCK-8 analysis (Figure 3b). Western blot analysis demonstrated increased expression of iNOS in HMC3 cells treated with the miR-143-3p inhibitor and mimic (Figure 3c). Additionally, miR-143-3p was found to modulate the expression of HK2.
Dysregulated expression of miR-143-3p promotes inflammation, proliferation, and activation of HMC3 cells. (a) QPCR analysis of IL6, IL-1β and NLRP3 levels in the miR-143-3p inhibitor/mimic- and NC-treated HMC3 cells. (b) The cell viability of miR-143-3p inhibitor/mimic- and NC-treated HMC3 cells. (c) Determination and quantification of iNOS and HK2 expression in miR-143-3p inhibitor/mimic- and NC-treated HMC3 cell. Data expressed as mean ± SD. * p < 0.05; ** p < 0.01; *** p < 0.001.
Perturbation of neural-derived EVs’ miRNA coexpression modules in BD
In our analysis of neural-derived EVs’ miRNA coexpression modules in BD, one BD patient and one control were identified as outliers and subsequently excluded from the WGCNA analysis. This analysis revealed four coexpression modules, excluding the grey module, which contains miRNAs not grouped into any specific module. Notably, the brown module demonstrated significant correlations with the YMRS scores (r = 0.24, p = 0.009) and the HAMA scores (r = 0.20, p = 0.02), as depicted in Figures 4a,b. Further investigation into the brown module through KEGG enrichment analysis showed a strong association with axon guidance, Ras signaling pathway, and ErbB signaling pathway, detailed in Figure 4c. Additionally, GO analysis highlighted significant enrichment in key biological processes, specifically the regulation of transcription by RNA polymerase, nervous system development, and the regulation of DNA-templated processes (Figure 4d).
Characteristics of microRNAs co-expression modules that affect clinical symptoms. (a) Pearson’s correlation coefficient (and P value in parentheses) between diagnosis, age, sex, clinical symptoms, and module eigengene. (b) Co-expression network plots for MEbrown module. Node size is proportional to node connectivity. (c) KEGG enrichment analysis of the microRNAs in MEbrown module. (d) GO enrichment analysis of the microRNAs in MEbrown module.
Examining the cellular composition, apart from the frequently observed terms like the cytoplasm, nucleus, and cytosol, other elements like cell junction, synapse, Golgi apparatus, and neuron projection emerged as prominently enriched. Additionally, the molecular function segment revealed significant associations with functionalities such as protein binding, metal ion binding, and DNA-binding transcription factor activity. More comprehensive details regarding the GO and KEGG enrichment analysis for MEbrown miRNAs are available in Supplementary Tables S8 and S9. Interestingly, no module exhibited a significant association with age and sex. This observation underscores the minimal potential confounding effects of age and sex on the plasma neural-derived EVs’ miRNA levels.
Discussion
In this study, we conducted a comparative analysis of miRNA abundance in plasma neural-derived EVs between BD patients and HC, identifying 14 differentially expressed miRNAs. Enrichment analysis revealed the potential involvement of these miRNAs in key biological pathways such as PI3K/AKT signaling, axon guidance, and focal adhesion. Then, we focused on the only significantly decreased miRNA, miR-143-3p, and found that 10 out of 32 protein-coding target genes were differentially expressed in the prefrontal cortex of BD patients compared with HC, as shown in the PsychENCODE dataset. Furthermore, miR-143-3p could promote inflammatory response and cell proliferation in microglia, regardless of its up-regulation or down-regulation. Furthermore, the WGCNA analysis pinpointed a co-expression module named MEbrown, which showed a notable correlation with YMRS scores and HAMA scores. Subsequent enrichment analysis of this module suggested a possible interplay between neural-derived EVs and neurodevelopmental processes.
Neural-derived EVs have been spotlighted in prior research due to their ability to mirror changes specific to the CNS (Saeedi, Israel, Nagy, & Turecki, Reference Saeedi, Israel, Nagy and Turecki2019; Younas, Fernandez Flores, Hopfner, Höglinger, & Zerr, Reference Younas, Fernandez Flores, Hopfner, Höglinger and Zerr2022). Consistently, our observations resonate with previous post-mortem brain studies focusing on miRNA alterations in BD. For example, the upsurge we detected in miR-29c levels within neural-derived EVs from BD patients echoes a study pinpointing a similar increase in miR-29c in EVs sourced from the prefrontal cortex (BA9) in BD patients (Banigan et al., Reference Banigan, Kao, Kozubek, Winslow, Medina, Costa and Delalle2013). The inaccessibility of post-mortem studies lends even greater significance to the exploration of EVs emanating from brain cells, offering a promising approach to deciphering the intricacies of BD. However, our findings diverge from those of previous works studying circulating EVs’ miRNAs (Ceylan, Tufekci, Keskinoglu, Genc, & Özerdem, Reference Ceylan, Tufekci, Keskinoglu, Genc and Özerdem2020). The discrepancy in miRNA changes may be attributed to the diverse origins of EVs, with neural-derived EVs representing only a small fraction of the total pool (Z. Zhang, Wu, et al., Reference Zhang, Wu, Tang, Chen and Wu2023). Compounding the complexity, prior research often hinges on smaller participant groups and tends to focus on a confined set of miRNAs. This underscores the imperative for more standardized miRNA detection protocols and a shift toward utilizing human neural-derived EVs, fostering more consistent and valuable cross-assay comparisons.
Previous studies have demonstrated the brain-specificity of miR-143-3p and its ability to modulate inflammatory responses (Guo et al., Reference Guo, Maki, Ding, Yang, Zhang and Xiong2014; Y. Wang et al., Reference Wang, Zhang, Wang, Zhang, Ye, Wang and Wang2023). Remarkably, HK2, which acts as a target gene for the differential expression of miR-143-3p in the prefrontal cortex of BD patients and HC, is selectively expressed in microglia and controls neuroinflammation through the modulation of glucose metabolism and mitochondrial functions (Hu et al., Reference Hu, Cao, Wang, Wu, Mai, Qiu and Gao2022). Our findings provide evidence that abnormal expression levels of miR-143-3p in microglia can alter the expression level of HK2 and trigger an inflammatory response, while previous studies have shown that both up-regulation and down-regulation of HK2 in microglia are involved in the activation of microglia (Hu et al., Reference Hu, Cao, Wang, Wu, Mai, Qiu and Gao2022; Leng et al., Reference Leng, Yuan, Pan, Su, Wang, Xue and Zhang2022). Considering the pronounced role inflammation plays in the pathophysiology of BD (Fiedorowicz et al., Reference Fiedorowicz, Prossin, Johnson, Christensen, Magnotta and Wemmie2015; Pereira et al., Reference Pereira, Oliveira, Silva, Madeira, Pereira and Cruz2021; Tang et al., Reference Tang, Chen, Chen, Zhong, Gong, Zhong and Huang2021), the identification of miR-143-3p in neural-derived EVs paves the way for further exploration into neuroinflammatory pathways and understanding molecular responses in clinical trials for BD.
Our study also found an increase in levels of miR-320a-3p, known for its role in regulating the AMPK/mTOR autophagy pathway (F. X. Li et al., Reference Li, Liu, Xu, Shan, Zheng, Lei and Yuan2023). Previous research has shown how any downturn in AKT–mTOR signaling can impact synaptic structural and functional plasticity in the prefrontal cortex (PFC), thereby affecting PFC-mediated cognition in patients with BD (Vanderplow et al., Reference Vanderplow, Eagle, Kermath, Bjornson, Robison and Cahill2021). Meanwhile, a study has highlighted an amplified level of miR-29c-3p in neural-derived EVs from individuals with subjective cognitive decline, when juxtaposed with controls (Y. Li et al., Reference Li, Xia, Meng, Wu, Ling, Chen and Liu2022). Given this backdrop, our discovery of elevated miR-320a-3p and miR-29c-3p in neural-derived EVs in patients with BD might hint at their potential involvement in BD-related cognitive challenges. It is worth noting that cognitive disturbances are pervasive among BD patients, persisting even during intervals devoid of manic or depressive episodes (Lima, Peckham, & Johnson, Reference Lima, Peckham and Johnson2018; Solé et al., Reference Solé, Jiménez, Torrent, Reinares, Bonnin, Torres and Vieta2017). However, the root causes of these cognitive disruptions in BD remain elusive. Our findings strongly suggest that neural-derived EVs’ miRNAs may contribute to the cognitive dilemmas faced by BD patients, a hypothesis warranting further exploration.
The analyses of GO and KEGG pathways revealed that the differentially expressed miRNAs were significantly associated with pathways such as PI3K/AKT signaling, axon guidance, and focal adhesion. Importantly, lithium, a cornerstone drug for treating BD, has been implicated in the regulation of the PI3K/AKT signaling pathway (Campbell, Campbell, & Smith, Reference Campbell, Campbell and Smith2022; Ni et al., Reference Ni, Gao, Wang, Tian, Wei, Zhao and Li2022). However, a focused analysis of the impact of lithium on EVs’ miRNAs in individuals with BD showed no significant differences in miRNA expression between those receiving lithium and those who were not (Supplementary Table S10). Furthermore, after adjusting for lithium treatment as a covariate, significant differences were observed in 11 out of the 14 miRNAs, indicating a limited effect of lithium on EVs’ miRNAs (Supplementary Table S11).
The focal adhesion’s involvement in BD is an emerging revelation, warranting deeper probes. Recent research has highlighted the direct impact of focal adhesion dysregulation on axon guidance and neuronal circuits, suggesting a foundational role in BD’s mechanistic intricacies and its potential improvement with lithium intervention (Niemsiri et al., Reference Niemsiri, Rosenthal, Nievergelt, Maihofer, Marchetto, Santos and Kelsoe2024). Further emphasizing the role of neuronal pathways, the observed aberrancies in axon guidance hint at a potential fallout from dysregulated miRNAs present in neural-derived EVs from patients with BD. Such disturbances may culminate in neuronal dysfunction, potentially correlating with the cognitive impairments frequently observed in the BD cohort (Keramatian, Torres, & Yatham, Reference Keramatian, Torres and Yatham2021; Leung et al., Reference Leung, Lau, Liang, de Dios, Suchting, Östlundh and Selvaraj2022; Shao, Golbaz, Honer, & Beasley, Reference Shao, Golbaz, Honer and Beasley2016). Likewise, while the Ras and ErbB signaling pathways, enriched in miRNA co-expression modules, were acknowledged for their pivotal roles in nervous system development and neuropsychiatric disease progression (Mei & Nave, Reference Mei and Nave2014; Sanchez-Ortiz et al., Reference Sanchez-Ortiz, Cho, Nazarenko, Mo, Chen and Parada2014), their exact role in BD remains intricate. Our insights echo the dysregulation of miRNAs in EVs might be pulling the strings behind the pathophysiology of BD, maneuvering these very biological channels. This beckons a more in-depth inquiry into the subject.
Limitations
While our study offers novel insights into the role of neural-derived EVs miRNAs in BD, several constraints necessitate a cautious interpretation of the results. First, our study lacks an independent replication of the results. To cement the credibility of our findings, ensuing research endeavors need to reproduce these outcomes. Additionally, an added layer of robustness could be incorporated by embracing cohorts representing a diverse tapestry of clinical and demographic backdrops. Entailing such varied populace groups would gauge the overarching applicability of these findings across different demographics and reduce potential biases. Furthermore, the absence of individual-level assessments to track temporal changes in miRNA expression among BD patients limits our understanding of their dynamics and response to treatment. Therefore, incorporating independent cohorts, longitudinal studies, and diverse populations is essential to strengthening the robustness and generalizability of the findings.
In addition, our study specifically focuses on a subset of differentially expressed miRNAs, selecting one for in vitro experiments and conducting preliminary investigations into its potential mechanisms. Future research should expand the focus to include a broader range of miRNAs and incorporate animal models to explore their roles more comprehensively. Although this study has identified several potential molecular pathways involved in BD, such as the Ras signaling pathway and ErbB signaling pathway, experimental validation of these pathways was not conducted. Future research must focus on validating these findings, as confirming the roles of these pathways is essential for advancing our understanding of BD’s pathological mechanisms and for identifying potential therapeutic targets.
Conclusion
In conclusion, our study first identified an aberration in the miRNA profile of plasma neural-derived EVs in patients with BD. Experimental data supports the role of miR-143-3p in regulating the immune response and proliferation of microglia. Beyond individual miRNAs, our discovery of a co-expression module associated with BD clinical manifestations further deepens our understanding. The association of these miRNAs with neurodevelopment and axonal guidance reinforces the neurogenic hypothesis of BD. With the identification of 14 differentially expressed miRNAs, we are at the cusp of potential biomarkers for BD diagnosis. Moreover, our dive into molecular pathways underscores several, previously underappreciated, pathways like focal adhesion, the Ras signaling pathway, and the ErbB signaling pathways, positioning them as potential key players in the intricate pathophysiology of BD. With these insights in hand, it becomes paramount to embark on further detailed studies. By doing so, we aim to decode the mechanisms driving the unique miRNA signatures in neural-derived EVs and their broader implications in the complex tapestry of BD.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/S0033291725000741.
Acknowledgments
We thank all of the subjects who participated in this study, and we thank all researchers who contributed to the collection and evaluation of the participants.
Author contribution
T.L. and X.Q. conceived and designed the study. H.J., B.R., and Y.Z. were responsible for drafting the manuscript and data analysis. Y.Z. and J.W. specialized in isolating plasma neural-derived EVs and contributed to data analysis. X.Y., H.Y., P.N., and Y.X. engaged in collecting clinical data and executing the psychiatric assessments. W.D., W.G., and X.H. provided essential guidance for the study protocol and are critical feedback and comments on the manuscript. All authors critically reviewed, provided feedback, and gave their approval for the final manuscript to be published.
Funding statement
This work was partly funded by National Nature Science Foundation of China (Grant No. 81920108018 to T.L., 82101598 to H.Y., 82230046 to T.L., 82371503 to X.Y., and 82371524 to X.Q.), the Key R&D Program of Zhejiang (Grant No. 2022C03096 to T.L.), Natural Science Foundation of Zhejiang Province (Grant No. LY22H090009 to X.Y.), Hangzhou Biomedical Science and Technology Support Project (Grant No. 2021WJCY195 to X.Q.), Project for Hangzhou Medical Disciplines of Excellence & KeyProject for Hangzhou Medical Disciplines (Grant No. 202004A11 to T.L.), and in part by grants from Nanhu Brain–Computer Interface Institute.
Competing interests
The authors declare no competing interests.
Ethical standard
The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008.
Open access
