Introduction

Approximately 300 million people worldwide suffer from depression, and depression may emerge as the biggest public health concern by 2030 [1]. Cognitive impairment is a common clinical symptom of depression. Major depression is associated with cognitive impairment in up to 73–94% of cases [2, 3]. Cognitive impairment associated with depression has garnered increasing attention recently [4,5,6,7]. Cognitive functioning, encompassing executive function, psychomotor speed, declarative memory, working memory, and attention, is severely impaired in both older and younger individuals with depression [4, 7, 8]. During acute depressive episodes, adults experience varying degrees of executive function (abnormalities in executive function are associated with an increased risk of suicide [9]), attention, processing speed, and memory decline [4, 5, 7]. Cognitive decline often reduces social and professional functioning [4, 10, 11]. Cognitive performance is affected in some patients even after remission [4, 10, 12]. and most individuals in the acute or remission phase experience significant difficulties concentrating [4, 13]. Additionally, cognitive impairment may increase the risk of relapse [4] and is linked to indicators of treatment response [6, 14, 15]. In a word, cognitive impairment contributes to both poor prognosis and disability in patients with depression [3, 4], resulting in increased healthcare costs and disease burden [16].

The pathogenesis of depression-related cognitive impairment involves multiple complex mechanisms. Early stress [17], childhood physical neglect [18], decreased plasma brain-derived neurotrophic factor (BDNF) levels [18], elevated cortisol levels [19], and oxidative stress [20] caused by depression contribute to cognitive impairment. Furthermore, changes in the prefrontal cortex, parietal cortex, basal ganglia, hippocampus, and amygdala are associated with depression and cognitive impairment [4, 21, 22]. Studies have reported that depressed patients exhibit negative cognitions (e.g., negativity and increased self-focus) [21, 23], and that depression-related cognitive impairment is linked to changes in the white matter of the brain [24]. Antidepressants induce neurotransmitter release in particular brain areas to elevate mood, and monoamine neurotransmitters [25, 26], including serotonin, dopamine, and norepinephrine, are linked to cognitive function [27]. Prefrontal catecholamines regulate executive functions, working memory, and attention [28, 29].

Studies have found that certain antidepressants, such as vortioxetine, escitalopram, sertraline, and ketamine [10, 21, 30,31,32], enhance cognitive function when included in an antidepressant regimen. Ketamine activates the mTORC1 signaling pathway and promotes the production of BDNF, thus increasing the levels of synaptic proteins and enhancing the number and function of synapses in the prefrontal cortex [21]. Vortioxetine improves monoaminergic transmission and glutamatergic function in the prefrontal cortex [33], while selective serotonin reuptake inhibitors influence hippocampal neurogenesis and function [31]. These changes collectively enhance neuroplasticity and ultimately contribute to improvements in cognitive function. However, a randomized longitudinal study by Shilyansky et al. found that antidepressants such as sertraline, venlafaxine, and escitalopram did not lead to significant improvements in cognitive impairment associated with depression [5]. At the same time, Keefe R.S. has pointed out that there are currently no recognized treatments for depression-related cognitive impairment and that the cognitive deficits in patients with depression should receive greater attention from relevant sectors [2, 10]. In summary, the inconsistency of existing research findings and the unclear underlying mechanisms have hindered the development of effective treatment strategies for cognitive impairment in patients with depression [10, 19]. The molecular mechanisms that influence cognitive function have not been adequately elucidated for most currently available drugs, and new methods to investigate the effects of antidepressants on cognitive function are urgently needed.

In recent years, the rise of computational biology has facilitated the exploration of complex pathological mechanisms underlying drug–disease interactions. Network pharmacology, which combines systems biology and pharmacology approaches, has been successfully used to identify key nodes in biological networks and multi-target interactions between drugs and diseases [34]. Understanding the interactions between multiple targets can enhance our understanding of the molecular mechanisms underlying complex diseases [35]. Virtual screening of large libraries of compounds can be performed as the starting point for streamlined drug discovery [36, 37] and to explore potential interactions with specific targets. [38] Molecular docking techniques are used to assess the binding modes and affinities of these compounds to the targets [39]. In this study, we employed network pharmacology, virtual screening, and molecular docking approaches to investigate the underlying molecular mechanisms for cognitive deficits associated with depression and explore the potential of existing antidepressant drugs to alleviate or prevent cognitive deficits. Figure 1 shows the workflow of this study.

Fig. 1: Study flow diagram.
figure 1

OMIM, Online Mendelian Inheritance in Man; PPI, Protein-Protein Interaction; KEGG, Kyoto Encyclopedia of Genes and Genomes; PDB, Protein Data Bank. BC, betweenness centrality; CC, closeness centrality; DC, degree centrality; AKT1, AKT serine/threonine kinase 1; AKT3, AKT serine/threonine kinase 3; CREBBP, CREB-binding protein; EGFR, epidermal growth factor receptor; EP300, histone acetyltransferase p300; INS, insulin A chain; IL-6, interleukin-6; IGF1R, insulin-like growth factor 1 receptor alpha chain; MAPK1, mitogen-activated protein kinase 1; MAPK3, mitogen-activated protein kinase 3; PIK3CA, phosphatidylinositol 4,5-bisphosphate 3-kinase catalytic subunit alpha isoform; PIK3R1, phosphatidylinositol 3-kinase regulatory subunit alpha; STAT3, signal transducer and activator of transcription 3.

Full size image

Materials and methods

Ethics approval and consent to participate

The current analysis does not require ethical approval, because our analysis only collects uploaded data information from the public database search. There are no human participants in this article and informed consent is not required.

Identifying possible targets for depression and cognitive impairment

The Online Mendelian Inheritance in Man (OMIM) database [40] (https://www.omim.org/) and the GeneCards database [41] (https://www.genecards.org/) were searched using the keywords “depressive disorder” to identify targets related to depression. The same databases were searched using the keywords “cognition disorders” to identify genes associated with cognitive impairment. Duplicates and non-protein-coding genes were eliminated from both target lists. The resulting target proteins were converted into their respective gene symbols for “Homo sapiens” using the UniProt (https://www.uniprot.org/) database [42]. The targets for depression and cognitive impairment were analyzed using the bioinformatics online tool [43] (http://www.bioinformatics.com.cn/) to generate a list of intersecting targets.

Constructing protein-protein interaction (PPI) networks and identifying hub targets

The common targets were imported into the STRING database [44] (https://cn.string-db.org/) to investigate the interactions among targets associated with depression and cognitive impairment. “Homo sapiens” was specified as the organism, the interaction confidence threshold was set at 0.9, and unconnected proteins were excluded. Cytoscape 3.10.1 [45] (https://cytoscape.org/) was used to import the filtered targets and generate the PPI network. The topological characteristics of each network node were assessed using the CytoNCA plug-in. The degree centrality, betweenness centrality, and closeness centrality were evaluated to examine the interactions between nodes and identify hub targets. Hub genes were identified after three rounds of continuous screening, using median values as cut-off thresholds. These three parameters collectively and comprehensively characterize the properties of each node within the interaction network [46]. A higher ranking indicates that the node has a more important role in the network. The targets identified through three rounds of screening were considered potential core targets for subsequent analysis [47].

Kyoto Encyclopedia of genes and genomes (KEGG) enrichment analyses

The hub targets were imported into the DAVID online platform [48] (https://david.ncifcrf.gov/) for the KEGG enrichment analysis (https://www.genome.jp/kegg/) [49]. The species was set to “Homo sapiens,” and the p-value threshold was set at < 0.05. After sorting the pathways by p-value from smallest to largest, the top 20 pathways with the highest significant enrichment were selected. This approach ensured that the highlighted pathways had the strongest statistical support and potential biological importance. Bubble plots were created to visualize the outcomes for the top 20 significant signaling pathways using the Bioinformatics online tool. Important signaling pathways were visualized using the KEGG Mapper online tool (https://www.genome.jp/kegg/mapper/).

Virtual screening and molecular docking

Based on the results of the KEGG enrichment analysis, the hub targets enriched in significant signaling pathways were analyzed to identify core targets for virtual screening and molecular docking. The crystal structures of the core targets were downloaded from the RCSB Protein Data Bank database [50] (https://www.rcsb.org/). Only protein structures from Homo sapiens were chosen to ensure biological relevance. Furthermore, high-resolution protein structures were given priority to increase the precision and reliability of the docking analysis. Water molecules and small molecules were removed, and hydrogen atoms were added to the proteins. The proteins’ active sites were identified using an online tool (https://proteins.plus/), and the docking box’s center coordinates (X, Y, Z) were saved. The antidepressant compounds were identified in the PubChem database [51] (https://pubchem.ncbi.nlm.nih.gov/), and the 3D structures were downloaded. The ligands and receptors were converted into a software-readable format using the SailVina platform (https://vina.scripps.edu/) [52] of AutoDock Vina [53] before importing the docking site data. Finally, molecular docking and virtual screening were used to determine the binding of the antidepressant compounds with the target proteins. The optimal binding conformations of the small molecules were predicted with the docking algorithm utilizing a scoring system. For each target, the top five small molecules with the highest docking scores were selected for visual analysis. The molecular docking results for candidate small molecules and the relevant targets were visualized using PyMOL (https://pymol.org/2/) and AutoDockTools 1.5.6 (https://autodocksuite.scripps.edu/adt/) software.

Results

Potential targets of depression, cognitive impairment, and common targets

After excluding non-protein-coding and median threshold truncation genes, 1386 targets associated with cognitive impairment and 1321 depression-related targets were retrieved from the GeneCards database. Following the removal of duplicate targets and entries without gene names, 613 targets associated with cognitive impairment and 571 targets related to depression were retrieved from the OMIM database. The targets from the two datasets were combined, and duplicates were eliminated, resulting in 1782 targets linked to depression and 1924 targets associated with cognitive impairment. The bioinformatics online tool was utilized to make a Venn diagram, and 1621 overlapping targets were ultimately identified (Fig. 2A).

Fig. 2: Identification of hub targets via protein-protein interaction network analysis.
figure 2

(A) A Venn diagram illustrating the overlapping targets of depression and cognitive impairment. (B–C) Construction of PPI networks of intersecting targets using the Cytoscape program and the STRING database. The interaction confidence threshold was set to 0.9 in the STRING database. (D-F) Screening of hub targets based on topological parameters. BC, CC, DC.

Full size image

Network of overlapping targets for depression and cognitive impairment and screening the hub targets

Figure 2B shows the established PPI network with 1243 nodes and 3006 edges. The information from the PPI network was imported into Cytoscape 3.10.1. Proteins without structures or links to other proteins were removed, resulting in 925 nodes and 2919 edges in the PPI network (Fig. 2C). The CytoNCA plugin was used to identify 46 hub targets based on topological parameters (degree centrality, betweenness centrality, and closeness centrality) with medians as cutoffs for three consecutive rounds (Fig. 2D–F, Supplementary Table 1).

KEGG Enrichment analyses and identification of core targets

The signaling pathways involved in impaired cognitive function with depression were identified using KEGG enrichment analysis. The top 20 enriched pathways from the 168 enriched pathways based on p-values are shown in Fig. 3A and Table 1. The hypoxia-inducible factor-1 (HIF-1) and forkhead box O (FoxO) signaling pathways were associated with cognitive impairment in depression. To identify the primary targets related to cognitive impairment induced by depression, the intersecting enriched targets of the two key signaling pathways were identified (Fig. 3B). Thirteen core targets were obtained, including cAMP response element-binding protein (CREB) binding protein (CREBBP), signal transducer and activator of transcription 3 (STAT3), phosphatidylinositol 3-kinase regulatory subunit alpha (PIK3R1), epidermal growth factor receptor (EGFR), insulin A chain (INS), insulin-like growth factor 1 receptor alpha chain (IGF1R), interleukin-6 (IL-6), phosphatidylinositol 4,5-bisphosphate 3-kinase catalytic subunit alpha isoform (PIK3CA), RAC-gamma serine/threonine-protein kinase (AKT3), RAC-alpha serine/threonine-protein kinase (AKT1), histone acetyltransferase p300 (EP300), mitogen-activated protein kinase 1 (MAPK1), and mitogen-activated protein kinase 3 (MAPK3). Detailed information about the 13 targets is shown in Table 2. Figure 3C and D show the two signaling pathways, highlighting the positions of the core targets within these pathways.

Fig. 3: KEGG pathway enrichment analysis results and the distribution of hub targets within key signaling pathways.
figure 3

(A) KEGG pathway enrichment analysis of the top 20 pathways (P < 0.05). (B) The significant KEGG signaling pathway-enriched targets were merged to identify common targets. (C, D) Maps illustrating the distribution and mechanisms of the common targets in important signaling pathways, including (C) the HIF-1 signaling pathway and (D) the FoxO signaling pathway. The common targets for both pathways are indicated in red.

Full size image
Table 1 Basic information about the KEGG enrichment results.
Full size table
Table 2 Topological information about the core targets.
Full size table

Target‑based in silico screening and molecular docking

223 antidepressant compounds were selected and downloaded in 3D format. Based on binding affinity rankings, the top five docking results were selected for visualization. The docking scores and PubChem compound CIDs are shown in Fig. 4A–B. The three compounds with the best docking results exhibited significant binding affinities to multiple targets: panuramine for INS, IGF1R, and PIK3CA; mosapramine for AKT3, PIK3R1, and IL6; and compound CID 3047803 for CREBBP, STAT3, and EP300. The first five docking scores of the 13 targets for antidepressants ranged from −7.6 to −12.9 kcal/mol. Panuramine, mosapramine, compound CID 3047803, adoprazine, and orvepitant exhibited strong binding to multiple targets (Supplementary Table 2). Notably, mosapramine showed the strongest binding in the first five dockings, interacting with seven targets, including three with the highest docking scores. The docking of mosapramine with seven targets is shown using PyMOL and Discovery Studio Visualizer software (https://www.3ds.com/) (Fig. 5, Table 3).

Fig. 4: Docking scores and PubChem Compound CIDs.
figure 4

(A) The top 5 docking scores for binding of the 13 core targets to the antidepressants. (B) The PubChem CID numbers for antidepressants that align with the docking results presented in (A). D-1 to D-5 indicate the first- to fifth-place docking scores for each target.

Full size image
Fig. 5: Molecular docking results.
figure 5

Molecular docking results for mosapramine and seven targets, including AKT3 (AKT Serine/Threonine Kinase 3), EGFR (epidermal Growth Factor Receptor), IL6 (interleukin 6), INS (insulin A chain), MAPK1(mitogen-activated protein kinase 1), MAPK3 (mitogen-activated protein kinase 3), PIK3R1 (phosphoinositide-3-kinase regulatory subunit 1).

Full size image
Table 3 Molecular docking of mosapramine with the core target proteins.
Full size table

Discussion

We identified 46 core targets using a network pharmacology approach. Based on a KEGG analysis of these targets, the FoxO and HIF-1 signaling pathways may play important roles in cognitive impairment in depression. The HIF-1 and FoxO signaling pathways regulate a variety of biological processes, including neuroimmune [54, 55], inflammatory [56,57,58], synaptic plasticit [54, 59]y, neuronal signaling [59], and oxidative stress [58, 60, 61] processes, that can be targeted to treat depression. In the brain, FoxO functions as a central regulatory factor integrating serotonergic signaling and neurotrophic pathways, exerting antidepressant and anxiolytic effects [62]. Moreover, the BDNF–FoxO1 signaling pathway in the medial prefrontal cortex was shown to regulate depression-like behaviors in postpartum mice [63]. In recent years, several studies have demonstrated that FoxO plays a critical role in improving dementia and cognitive impairment, suggesting that FoxO may represent a novel therapeutic target for ameliorating cognitive deficits [64, 65]. The mechanistic role of FoxO in cognitive impairment associated with depression warrants further investigation. Elevated HIF-1 levels enhance the mTOR signaling pathway and upregulate vascular endothelial growth factor (VEGF) expression, which is implicated in the treatment of depression [58]. One study found that HIF-1α regulated cognitive deficits in post-stroke depression model rats and may serve as a potential therapeutic target [66]. In addition, elevated HIF-1α levels have been associated with cognitive impairment in patients undergoing hemodialysis [67]. Li et al. demonstrated that FG-4592, a prolyl hydroxylase inhibitor, activates the HIF-1 and BDNF signaling pathways to promote dendritic growth and improve depression and cognitive functioning, providing strong evidence to support our findings [68]. In summary, it is hypothesized that HIF-1 and FoxO may affect cognition by influencing BDNF. It has been found that HIF-1 may play a crucial role in the BDNF and pro-BDNF activation pathway [69]. Changes in this pathway are strongly associated with both depression and cognitive impairment [69]. Furthermore, BDNF regulation of FoxO-related genes regulates neuronal survival, differentiation, and synaptic plasticity [70]. Activation of CREB/BDNF signaling induces antioxidant and antidepressant activities [1]. Recent studies have shown that quercetin promotes hippocampal neurogenesis via the CREB/BDNF signaling pathway and the forkhead box transcription factor G1, thereby enhancing both depressive symptoms and cognitive function [60]. In summary, previous studies provide indirect and direct evidence supporting the role of HIF-1 and FoxO signaling in depression and cognitive function. We utilized computational biology to validate previous experimental results, highlighting the reliability of computational biology methods.

Analysis of the enriched targets in the HIF-1 and FoxO signaling pathways revealed 13 common targets, including AKT1, AKT3, PIK3R1, PIK3CA, MAPK1, MAPK3, INS, IGF1R, STAT3, CREBBP, IL-6, EP300, and EGFR. These targets play critical roles in cell signaling, neuronal development, synaptic plasticity, learning, memory, and neuroinflammatory processes [1, 57, 71, 72]. AKT is a serine/threonine protein kinase that regulates neuronal survival, development, and signaling. Three isoforms of AKT, including AKT1, AKT2, and AKT3, are widely expressed [1]. PIK3R1 and PIK3CA are involved in phosphoinositide 3-kinase (PI3K)/protein kinase B (AKT) signaling, which plays a significant role in stress responses, neural plasticity, and cell survival [60, 71]. The expression of PI3K and AKT is markedly decreased in the brains of mice with depression, suggesting that the PI3K/AKT signaling pathway plays a significant role in depression and cognitive function [1]. The PI3K/AKT signaling pathway also regulates downstream molecular targets, including FoxO and HIF-1, processes that are associated with oxidative stress and neuroinflammatory responses [1, 54, 56, 57]. Oxidative stress and neuroinflammation are considered central mechanisms underlying both depression and neuroplasticity [55, 72].

IL-6 and MAPK-related targets may contribute to depression and cognitive impairment through the FoxO signaling pathway. IL-6 levels are elevated in the blood and cerebrospinal fluid of patients with depression [57]. Additionally, IL-6 activates the HIF-1α, Janus kinase (JAK)/signal transducer and activator of transcription (STAT), MAPK/CREB, Ras-MAPK, and PI3K signaling pathways [72]. These pathways promote neuronal development and glial cell differentiation, which influence depression and memory [72]. Other studies have shown that phosphorylated FoxO1 regulates the transcription of toll-like receptor 4 and enhances IL-6 expression [1, 57]. Baicalin alleviated neuroinflammation-induced depression-like behavior by significantly reducing IL-6 levels and promoting the phosphorylation of FoxO1 [57]. MAPK is an upstream kinase of FoxO, and its phosphorylation regulates various transcriptional functions [54]. IL-6 also activates the MAPK/CREB and Ras-MAPK signaling pathways, influencing hippocampal neurons and contributing to both depression and cognitive impairments [72]. IL-6 and MAPK-related targets synergize within the FoxO pathway, contributing to the development of both depression and cognitive impairment.

CREBBP is a direct binding protein of CREB. CREB signaling pathways play a crucial role in the nervous system and are closely associated with biological processes such as learning, memory, and neuroplasticity [54, 61, 71]. CREB regulates the transcription of genes, including BDNF and tropomyosin receptor kinase B [61]. Research demonstrates that natural products enhance memory and spatial learning via both pathways [71]. Previous animal studies have reported reduced levels of CREB phosphorylation associated with depressive behavior [1, 73]. Cordyceps sinensis was found to exert antidepressant effects through CREBBP [61]. Previous studies indicate a role of CREBBP in the pathophysiology of depression and cognitive impairment. STAT proteins are primarily activated through the JAK/STAT pathway but can also be activated via the MAPK and PI3K pathways to mediate their biological functions. The STAT pathway influences neuroplasticity and suppresses hippocampal neurogenesis [72]. This process may be synergistically involved in cognitive impairment in depression with other targets and pathways (IL-6, HIF-1α, JAK/STAT, MAPK/CREB, Ras-MAPK) [72].

During stress, 5-hydroxytryptamine regulates the FoxO and the insulin/insulin-like growth factor 1 (IGF-1)/AKT signaling pathways [54, 62]. This may contribute to neuronal atrophy and depression [54]. An additional study has shown that natural products modulate the insulin/IGF-1/AKT signaling pathway to exert neuroprotective effects [71]. Other studies have identified FoxO as a critical mediator of insulin/IGF-1 signaling, and IGF-1 plays a pivotal role in the reduction of cerebral white matter volume in individuals with depression [54]. In addition, EGFR and EP300 are important targets in treating depression [61, 74]. EGFR is expressed in the neurons of the hippocampus, cerebellum, and cerebral cortex [75]. EP300 binding to CREB proteins may play a role in the inflammatory response that contributes to depression [76]. Although there is no direct evidence to support that all of the above targets are involved in cognitive impairment in depression, the results of the present study deepen our understanding of the pathological mechanisms of cognitive impairment in depression.

To identify drugs that effectively alleviate cognitive function in individuals with depression, we conducted a virtual screening of antidepressants against the 13 targets. This screening aimed to identify antidepressants with strong binding affinity to these targets for subsequent drug optimization. Our results show that mosapramine exhibited the highest binding affinity to multiple target proteins. Mosapramine, which was developed in Japan, is an iminodibenzyl antipsychotic classified as an atypical antipsychotic, acting on dopamine 2 receptors and serotonin 2 receptors in the frontal cortex [77]. Optimal docking was observed between mosapramine and three targets: AKT3, PIK3R1, and IL-6. The corresponding docking scores were −12.9, −10.9, and −11 kcal/mol, respectively. AKT3 and PIK3R1 are involved in the PI3K/AKT signaling pathway, which plays a significant role in depression and cognitive function [1] and may regulate HIF-1 [56] and FoxO signaling pathways [54, 57] to induce neuroinflammation that affects both depression and cognition. Additionally, numerous studies have highlighted the significant role of IL-6 in cognitive impairment associated with depression [57, 72]. IL-6 may modulate the HIF-1 and FoxO signaling pathways through PI3K/AKT signaling, leading to neuroinflammation and alterations in synaptic plasticity [1, 57, 72]. In addition, another study found that mosapramine enhances FoxO protein expression in the prefrontal lobes of the brain, similar to clozapine, and may help improve the negative symptoms of schizophrenia [78], which share some commonalities with cognitive impairment [79]. Mosapramine may affect FoxO expression, thereby alleviating cognitive impairment in depression. Patients with schizophrenia and depression frequently exhibit impaired cognitive functioning. Mosapramine is structurally similar to tricyclic antidepressants and butyrophenones [77]. Previous studies have found that the antidepressants venlafaxine [80] and agomelatine [81] affect HIF-1 and FoxO signaling, ameliorate neuronal apoptosis, and exhibit antioxidant effects. Future research should focus on elucidating the intrinsic relationship between mosapramine, HIF-1, the FoxO signaling pathway, and multiple targets.

There are several limitations to this study. First, the database of disease targets has not been recently updated and does not comprehensively include all disease-related targets. Second, the database of small-molecule compounds is limited, and some newly discovered or synthesized compounds were not included. Third, construction of the PPI networks mainly depends on known and annotated interactions, which may not accurately represent dynamics taking place in the central nervous system. Fourth, this study’s network pharmacology and molecular docking methods are computational predictions that do not take into consideration pharmacokinetic characteristics (such as absorption, distribution, metabolism, and excretion). Finally, the results of this study are based on a computational approach and still need experimental validation.

Conclusions

In this study, biological and computational approaches were employed to investigate the potential molecular mechanisms underlying cognitive impairment in depression and identify candidate therapeutic drugs that may mitigate this impairment. Our findings suggest that the HIF-1 and FoxO signaling pathways are regulated by multiple targets and may represent key mechanisms contributing to cognitive dysfunction in depression. Mosapramine may impact the HIF-1 and FoxO signaling pathways through AKT3, EGFR, IL6, INS, MAPK1, MAPK3, and PIK3R1 to alleviate cognitive impairment in depression. This study provides a foundation for future research on drug repurposing and optimization. However, it is important to note that these findings require further experimental validation.