Depression is a severe affective mental disorder with typical clinical manifestations of depressed mood, reduced volitional activity, and slowed thinking and may be accompanied by somatic discomforts, such as abdominal distention, chest tightness, and pain [1]. More than 300 million people suffer from severe depression, with a lifetime prevalence rate of 17% [2]. In 2010, depression accounted for 8.2% (5.9%–10.8%) of global years lived with disability and 2.5% (1.9%–3.2%) of global disability-adjusted life years [3]. Depression causes great losses to human health and property. By the end of 2030, depression is expected to impose the world’s leading disease burden. This is because individuals with early onset depression are less likely to complete secondary education (2–3 times), and this population has a higher unemployment rate (29% at age 40) than that of the general population [4, 5]. Although the exact pathogenesis of depression has not been determined, it is generally believed that genetic factors are one of the causes of depression [6, 7]. Based on different pathogenesis hypotheses, genes related to neurotrophic factors, monoamine neurotransmitters, and neuroendocrine factors have received attention; these genes include BDNF [8], COMT [9], and NET [10]. In a large genome-wide association meta-analysis study involving 135,458 cases and 344,901 controls, 44 variants were identified as genetic risk factors for depression. However, these findings are still insufficient to provide us with a comprehensive understanding of the pathogenesis of depression and to find corresponding strategies.

Recent studies have confirmed that in different biological lineages, the protein coding sequence of the genome remains relatively stable, while the noncoding sequence shows a significant correlation with the heterogeneity and complexity of the organism [11, 12]. Long noncoding RNAs (lncRNAs), which account for 80%–90% of all noncoding RNAs (ncRNAs), are RNAs with a length of more than 200 nucleotides but no protein coding activity [13]. Compared with microRNAs (miRNAs), circular RNAs (circRNAs), and other ncRNAs, lncRNAs have long sequences and complex spatial structures and can bind to many diverse proteins. Thus, they carry more information and perform more diverse functions. LncRNAs can act at multiple levels, including the pretranscriptional, posttranscriptional, translational, and posttranslational levels. The primary mechanisms include the following: (1) acting as a “sponge” for miRNA, competitively blocking its binding to target mRNAs; (2) interfering with the expression of genes through cis or trans regulation; (3) participating in epigenetic modifications of DNA and chromatin remodeling; (4) regulating transcription factor activity and affecting their binding to gene promoter regions; (5) affecting covalent modifications of RNAs; (6) influencing alternative splicing of RNAs; (7) altering the subcellular localization of RNAs; and (8) altering the stability of RNAs [14, 15].

With the development of high-throughput technologies and various genomics approaches, we are now better equipped to comprehend the biological characteristics and disease expression profile of depression on a genome-wide scale and to construct the global expression network. To date, 21,183 lncRNAs from Homo sapiens have been included in the National Center for Biotechnology Information database (https://www.ncbi.nlm.nih.gov/). Databases for querying and analyzing lncRNAs are constantly emerging. Examples include, LNCipedia (https://lncipedia.org/), RNAcentral (https://rnacentral.org/), and LncRNADisease (http://www.cuilab.cn/lncrnadisease). An increasing number of studies have reported altered lncRNA expression profiles in both brain regions and the blood of patients with depression [16, 17, 18, 19]. In this article, we focus on a systematic review of the role of lncRNAs in depression, and aim to generate novel insights for objective diagnostic marker identification, efficacy prediction, and antidepressant therapy development [15]. Through PubMed and the Web of Science Library, we conducted an online search for all research published up to March 2023. The databases were searched by using the terms “lncRNA”, “long noncoding RNA”, and their combinations with “major depression”, “major depressive disorder”, and “depression”. The language was limited to English.

The ceRNA hypothesis was proposed in 2011 [20]. MiRNAs are negative regulators of gene expression since they can diminish the stability of target RNAs or repress their translation via complementary pairing with the 3${}^{\prime }$ untranslated region (3${}^{\prime }$UTR) of mRNAs [21, 22]. If a lncRNA and a mRNA have high sequence homology, i.e., a lncRNA and mRNA have the same miRNA response element, then the lncRNA can competitively bind the miRNA and reduce its intracellular content, thereby reducing the repressive effect of the miRNA on its target gene.

CeRNA regulatory networks, which are mediated by lncRNA‒miRNA‒mRNA, play an important role in the gene regulatory network of depression (Table 1, Ref. [23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41]) and have become a potential treatment target for depression. One of the most extensively researched neurotrophic factors, BDNF, is substantially expressed in human brain tissues, including the hippocampus, amygdala, neocortex, and cerebellum. BDNF is crucial in multiple neuropsychiatric conditions [42, 43, 44]. To date, approximately 20% of lncRNAs involved in ceRNA networks in depression have been found to be related to the regulation of BDNF expression [23, 24, 25]. For example, the regulatory role of the lncRNA GAS5 in depression implicates the lncRNA GAS5-miR-10b/BDNF axis. The lncRNAGAS5 was found to act as a miR-10b sponge in a rat model of poststroke depression (PSD). By upregulating GAS5, curcumin inhibits miR-10b, thus activating the BDNF/Trk$\beta$ pathway, promoting the expression of synapse-associated proteins, and improving PSD [24]. Huan et al. [23] documented that the lncRNA MIR155HG regulates the miR-155/BDNF axis to alleviate depressive-like behavior in mice. Table 1 lists more mechanisms by which lncRNAs control BDNF expression through their activity as ceRNAs.

LncRNAs also have multiple targets, and a single lncRNA molecule may inhibit multiple target miRNAs. In addition to the miR-10b/BDNF pathway previously mentioned, the lncRNA GAS5, for instance, may also be implicated in the regulation of gene expression in depression through the microRNA-26a/EGR1 axis [26]. MiR-26a targets EGR1, while the lncRNA GAS5 can specifically bind miR-26a. In mice with depressive-like behavior, EGR1 suppresses the activation of the PI3K/AKT pathway to induce the release of inflammatory substances and the apoptosis of hippocampal neurons. Therefore, lncRNA GAS5 reduces the inhibitory effect of miR-26a on EGR1 by acting as a ceRNA, inhibiting PI3K/AKT pathway activation and protecting hippocampal neurons from damage in mice with depressive-like behavior. This relationship between one lncRNA and multiple miRNA targets was also predicted by bioinformatics analysis [27]. For example, Chen et al. [27] identified that seven miRNAs (miR-425, miR-486, miR-28-5p, miR-29b, miR-100, miR-125b and miR-320) showed potential matching sequences with the lncRNA Gm26917.

In depression, the targets regulated by lncRNAs through their ceRNA activity are involved in several processes, including neurotrophic processes [23], immunoinflammatory processes [26], glutamatergic processes [28], and methylation modifications [29]. In the prefrontal cortex of chronic unpredictable mild stress (CUMS) rats, miR-29b-3p expression was elevated after ketamine treatment, leading to a decrease in the expression of the glutamate metabotropic receptor GRM4 [28]. Based on the inhibition of miR-29b-3p by lncRNA Gm26917, lncRNA Gm26917 is a crucial regulator of fluoxetine and ketamine and a potential target for depression treatment [30]. Ding X et al. [29] found that lncXR_351665, as a miR-152-3p sponge, upregulated the DNA methyltransferase DNMT1 to promote the development of chronic pain-induced depression. In summary, miRNAs and their target genes, as downstream targets of lncRNA regulation, establish a complex ceRNA regulatory network together with lncRNAs. LncRNAs are involved in the progression of depression through multiple targets and multiple pathways. A total of 13 lncRNAs (involving 16 ceRNA regulatory axes) are summarized in Table 1.

Single nucleotide polymorphisms (SNPs) are DNA sequence polymorphisms resulting from single nucleotide variations at the genomic level. More than 90% of all known polymorphisms are SNPs, making it the most prevalent heritable variation in humans [63]. SNP loci have been used in genome-wide association studies (GWAS) and have become the major route for the discovery of complex genetic mechanisms of disease. Depression has complex genetic variations, with numerous SNPs identified across the genome.

LncRNAs can not only interact with nucleic acid sequences by base complementary pairing to form RNA‒DNA complexes but also form relatively complex and variable spatial structures (compared to those formed by other small ncRNAs) to recruit or act as scaffolds to bind other proteins to form RNA‒protein complexes. Thus, lncRNAs have flexible and diverse mechanisms of controlling gene expression. Numerous studies have indicated this feature and regulatory mechanism of lncRNAs. Using RNA pull-down, chromatin immunoprecipitation, and RNA immunoprecipitation approaches, these studies have shown that lncRNAs bind to biomolecules such as transcription factors, proteins, mRNAs and DNA and form gene regulatory networks, revealing the biological functions and mechanisms of lncRNAs. According to the position of lncRNAs relative to the target genes, the regulation mode of lncRNAs with respect to the target genes can generally be classified into two categories: cis and trans regulation [79]. Next, we discuss the mechanism of lncRNAs in depression from two perspectives, cis and trans regulation.

5.1 Cis and trans Regulation of LncRNA

Cis regulation by lncRNAs is typically based on the observation that lncRNAs have close physical proximity to their target genes. First, lncRNAs have the ability to bind directly to the 5${}^{\prime }$-UTRs of adjacent genes and inhibit or promote the expression of the genes by blocking or recruiting transcription factors, such as RNA polII [80, 81, 82] (Fig. 2a). LncRNAs can also contribute to looping cross-talk with chromatin in adjacent genes to regulate gene expression. In some cases, lncRNAs do not directly participate in the regulation of gene expression but share the same transcription factors, including RNA polII, with adjacent genes. When transcribed, lncRNAs can compete with adjacent gene promoters to bind RNA polII or affect the local chromatin state and transcription factor binding to the promoter and enhancer regions [83, 84, 85] (Fig. 2b). There is another species of lncRNAs whose sequences contain DNA regulatory elements, such as enhancers. These DNA regulatory elements regulate the expression of the adjacent genes independent of lncRNA transcription [83]. That is, the RNA transcripts of the lncRNA genes have no effect on the adjacent genes, but mutations in the DNA sequence of the lncRNA genes or elimination of this sequence alters the expression of the adjacent target genes [83] (Fig. 2c). In summary, it is clear that cis regulation by lncRNAs has two distinct features. First, the lncRNA and the target gene are in close proximity; second, the lncRNA directly or indirectly affects the regulatory regions of the target gene, which regulates its transcription.

Fig. 2.

The cis and trans regulation modes of lncRNAs. (a–c) The cis regulation modes. (d–f) The trans regulation modes. (a) LncRNA binds directly to the 5${}^{\prime }$-UTR of adjacent genes and forms looping crosstalk to regulate gene expression. (b) LncRNA loci share RNA polII with adjacent genes. Thus, lncRNAs compete with gene promoters and inhibit their expression. (c) LncRNA loci contain the regulatory elements of adjacent genes, such as enhancers, in which the mutations alter gene expression. (d) LncRNA leaves the locus to combine with the distant gene and act as scaffolds to recruit proteins to synergistically regulate gene expression. (e) LncRNA in the nucleus is recruited into nuclear speckles to form spliceosomes and regulate the alternative splicing of RNA. (f) LncRNA combines with the 3${}^{\prime }$UTR binding proteins of the distal gene mRNA, thereby accelerating mRNA degradation. TF, transcription factor.

Via trans regulatory mechanisms, lncRNAs have a variety of potential effects on the spatial conformation of distal genes. Transcribed lncRNAs can leave the locus to bind to target genes and act as scaffolds or decoys to recruit proteins to synergistically regulate gene expression [86, 87] (Fig. 2d). The way a lncRNA functions is to some extent influenced by its cellular localization. LncRNAs can affect the structure of the nucleus. The lncRNAs in the nucleus can be recruited into nuclear speckles to form spliceosomes with numerous cleavable proteins to regulate alternative splicing of RNA and maintain the normal function of organelles in the nucleus [88, 89] (Fig. 2e). The last species of lncRNAs can also perform trans regulatory functions by influencing trans acting factors or other RNA molecules with regulatory functions (Fig. 2f). In conclusion, the above three types of lncRNA transregulatory modes share the same regulatory characteristics. The first is, the ability to affect genes that are distant from their own loci, and the second is the ability to regulate target genes at the transcriptional or posttranscriptional level.

The immune-inflammatory hypothesis is one of the explanatory hypotheses for the pathogenesis of depression. This hypothesis proposes that inflammation is crucially involved in the onset of depression or that depression is an immune system-mediated warning to the body.

Significantly elevated expression levels of different types of proinflammatory cytokines in cerebrospinal fluid and peripheral blood are common in patients with depression [95, 96]. The reason for this may be that the HPA axis, which secretes a number of hormones and neurotransmitters to regulate peripheral immunity, becomes more active in response to stress [97]. The immune system responds quickly and increases the levels of inflammatory mediators, including cytokines. Immune messages are transmitted between the peripheral and central neural systems by inflammatory cytokines, altering the central nervous system or causing corresponding behavior changes [98]. Since Smith first proposed in 1991 that cytokines could be involved in depression [99], inflammatory cytokines have been found to induce depression-like behavior [95, 100]. For example, proinflammatory cytokines, such as IL-2 and IL-16, affect the function of serotonergic neurons by inducing dioxygenases, resulting in a depressive state in the body [101]. Clinical studies have found that patients develop MDD symptoms during interferon treatment, and the risk of depression after interferon treatment is 30%–70% regardless of the treatment indication [102]. In a study by Zhou et al. [94], several lncRNAs were identified, and the overlapping or antisense genes of these lncRNAs, such as IRF2, LY6E, and HMBOX1, were shown to be associated with interferon transcription, activation, and signal transduction (Table 4). IRF2 is antisense to the lncRNA RP11-326I11.3 and encodes a transcription factor (TF) involved in interferon signal transduction. LY6E is antisense to the lncRNA RP11-273G15.2 and is an interferon-stimulated gene with immunomodulatory function. The lncRNA CTD-2647L4.4 is a sense intron of HMBOX1, which encodes a TF involved in transcriptional repression, including inhibition of interferon genes [103, 104, 105, 106, 107]. Therefore, these overlapping protein-coding genes or antisense genes of these differentially expressed lncRNAs were predicted to be potential targets of lncRNA cis regulation by Zhou et al. [94]. In short, lncRNAs may be involved in depression by cis regulating the expression of these interferon-related genes. Recent research has revealed that lncRNAs may regulate proinflammatory cytokines to lessen the symptoms of depression [108, 109]. Sun et al. [110] showed that lncRNA MSTRG.81401 significantly reduced the expression of the cytokines TNF-$\alpha$ and IL-1$\beta$ in animal models and alleviated diabetic neuropathic pain and depressive behavior in rats. Although there is no direct evidence yet that lncRNAs are involved in depression mediated by immune factors, an increasing number of studies have shown a potential regulatory relationship between lncRNAs and inflammatory factors in this disease. For example, in macrophages, the lncRNA THRIL can activate the transcription of TNF$\alpha$ by binding to its promoter [111], and TNF$\alpha$ mediates depression like behavior caused by stress induced GSK3 activation in a learned helplessness mouse model [112]. Therefore, it was suggested that there may be a potential THRIL-TNF$\alpha$ axis in depression, but this requires further experimental verification. We have summarized more potential lncRNA-cytokine regulatory axes in Supplementary Table 1.

Inflammasomes are polymeric protein complexes that resemble apoptotic bodies and have the ability to cause inflammation in the cytoplasm [113]. LncRNAs may also be involved in MDD by interfering with the formation of inflammasomes. The role of lncRNA Gm14205 in postpartum depression (PPD) was described by Zhu et al. [114]. The lncRNA Gm14205 may palliate PPD by inhibiting oxytocin receptor activation and inducing activation of NLRP3 inflammatory vesicles in astrocytes.

The gut-brain axis affects the nervous system through the interaction between the vagal nervous system and the intestinal immune system [115]. Evidence of crosstalk between the gut-brain axis and the HPA axis was demonstrated by Misiak B et al. [116]. Dyshomoeostasis of the gut flora severely interferes with central nervous system neurotransmission through the HPA axis and vagal afferent nerve fibers, accompanied by local inflammation that exacerbates anxiety or depressive behavior [117, 118]. There is no direct experimental evidence to confirm the interactions between lncRNAs and the gene regulatory network of the gut-brain axis in depression. However, recent artificial intelligence-mediated biological analysis has shown that multiple lncRNAs, including HOTAIR, MEG3, GAS5, and TUG1, may be hub factors in the lncRNA regulatory network and are influenced by improvements in the gut environment in animal models of depression [119].

The current state of depression treatment is characterized by the lack of knowledge of the pathophysiology and objective diagnostic indicators of depression, as well as the limited availability of drugs, the slow onset or even ineffectiveness of drugs, the high relapse rate, and the harmful side effects of drugs [120, 121]. It is crucial to understand the pathogenesis of depression and develop new diagnostic biomarkers and therapeutic targets. The abundant lncRNA expression profile data provide us with new tools. Considering the high heterogeneity of complex diseases such as depression, future research may need to build a good framework with multiple dimensions. Since alterations in gene expression lead to the remodeling of central nervous system structure, immunoregulation, and neurotransmission, the establishment of multidimensional and efficient diagnostic models by combining lncRNAs with other indicators, such as imaging features, protein profiles, and metabolite profiles is a promising strategy [12]. Constructing a diagnostic classifier by the nomothetic network psychiatry (NNP) approach may be a useful attempt. Maes et al. [122] integrated and ensembled multiple factors into a cause-to-outcome model by means of NNP in a study of MDD. Factors integrated in this study included genome and environmentome features (psychosocial aspects, context-centered hermeneutic data), brainome features, cognitome features, symptomatome features (different symptom domains or clinical phenotypes) and descriptive psychopathological assessments. According to the NNP, the diagnostic gold standard DSM/ICD has been challenged, and multiple false dogmas are thought to exist in mood disorder research. NNP has been attempted in studies of MDD, bipolar disorder, schizophrenia, and other disorders [122, 123, 124, 125]. However, to date, no study has provided both lncRNA expression profiles and abundant other information for modeling NNP. This is one of the future endeavors to look forward to. Furthermore, as we have reviewed before, lncRNAs may be involved in the development of depression through various mechanisms, such as ceRNA activity, epigenetic regulation, inflammation and immunity, and single nucleotide polymorphisms. The bioinformatics analysis performed in this review also suggests potential relationships between lncRNAs and methylation modifications and SNPs in depression. However, research on the mechanisms of action of lncRNAs in depression is still in its infancy. Additionally, most studies only show their correlation with the occurrence of depression by statistical analysis of differentially expressed lncRNAs and predict the regulatory mechanism by bioinformatics tools. Therefore, providing clear and powerful evidence to explain the role of lncRNAs in depression through molecular biology experimental approaches should be one of the future directions of subsequent work.

SW and LT performed the data analysis and wrote the original manuscript. NH collected and organized the literatures, and visualized the data. HW conceived and revised the manuscript. All authors contributed to editorial changes in the manuscript. All authors revised and approved the final manuscript.

Not applicable.

Not applicable.

This study was supported by the Education Department of Sichuan Province [grant numbers 18ZA0199], and the Bureau of Science and Technology and the intellectual property of Nanchong City [grant numbers 20SXQT0013].

The authors declare no conflict of interest.