How Oxidative Stress Induces Depression?

Abstract

Depression increasingly affects a wide range and a large number of people worldwide, both physically and psychologically, which makes it a social problem requiring prompt attention and management. Accumulating clinical and animal studies have provided us with substantial insights of disease pathogenesis, especially central monoamine deficiency, which considerably promotes antidepressant research and clinical treatment. The first-line antidepressants mainly target the monoamine system, whose drawbacks mainly include slow action and treatment resistant. The novel antidepressant esketamine, targeting on central glutamatergic system, rapidly and robustly alleviates depression (including treatment-resistant depression), whose efficiency is shadowed by potential addictive and psychotomimetic side effects. Thus, exploring novel depression pathogenesis is necessary, for seeking more safe and effective therapeutic methods. Emerging evidence has revealed vital involvement of oxidative stress (OS) in depression, which inspires us to pursue antioxidant pathway for depression prevention and treatment. Fully uncovering the underlying mechanisms of OS-induced depression is the first step towards the avenue, thus we summarize and expound possible downstream pathways of OS, including mitochondrial impairment and related ATP deficiency, neuroinflammation, central glutamate excitotoxicity, brain-derived neurotrophic factor/tyrosine receptor kinase B dysfunction and serotonin deficiency, the microbiota-gut-brain axis disturbance and hypothalamic-pituitary-adrenocortical axis dysregulation. We also elaborate on the intricate interactions between the multiple aspects, and molecular mechanisms mediating the interplay. Through reviewing the related research progress in the field, we hope to depict an integral overview of how OS induces depression, in order to provide fresh ideas and novel targets for the final goal of efficient treatment of the disease.

Depression

Depression is a category of mental disorder, characterized by core symptoms of depressed mood (for at least two weeks) and anhedonia, which influences a wide range and a large number of people worldwide. The global prevalence of the disease is approximately 6% and the lifetime risk is about 18%, which means about one out of six people would experience at least one episode in the lifetime (Malhi & Mann, 2018; Otte et al., 2016). According to the statistics from the World Health Organization and related report, over 280 million people have been suffering from the disease globally (Damian et al., 2021), among which is over 95 million Chinese people, including PhD students (Woolston & O’Meara, 2019; Y. Huang et al., 2019). Furthermore, the influence of global COVID-19 epidemic in recent years on the disease has started to emerge (Damian et al., 2021; Racine et al., 2021). Additionally, a considerable part of patients has not been properly diagnosed and attended, due to lack of awareness, limited medical resources, etc. For instance, the disease situation in children and adolescents remains unclarified, which receives increasing attention both locally and globally (Kong et al., 2022; Theo et al., 2020). The data indicate the disorder as an urgent social problem requiring prompt attention and management, and highlight the necessity to elucidate depression pathogenesis. The accumulating clinical and animal studies over years have provided us with in-depth understanding of the disease, and considerably promoted antidepressant research. Diversified disease hypotheses have been proposed, such as the classical hypothalamic-pituitary-adrenocortical (HPA) axis dysregulation and central serotonin (5-HT) deficiency, and the novel glutamate excitotoxicity, brain-derived neurotrophic factor/tyrosine receptor kinase B (BDNF/TrkB) dysfunction and neuroinflammation in the brain, and microbiota-gut-brain (MGB) axis disturbance (Krishnan & Nestler, 2008; Malhi & Mann, 2018).

The Vital Role of Oxidative Stress in Depression

Oxidation and reduction are the basic reactions of cell metabolism. Oxidation produces a variety of free radicals, mainly divided into reactive oxygen species (ROS) and reactive nitrogen species. The antioxidant system in the body is responsible for removing free radicals, including antioxidant enzymes and small molecules, such as superoxide dismutase (SOD), catalase (CAT), glutathione peroxidase (GPx), glutathione (GSH), etc. For example, SOD can catalyze superoxide (O₂ ⁻ ) to hydrogen peroxide, which could then be catalyzed by CAT and GPx into oxygen and water (Ali et al., 2020). However, excessive oxidation or insufficient antioxidation would cause substantial free radical accumulation, which would result in disrupted homeostasis referred as oxidative stress (OS). Take ROS, for example, which are chemical molecules that contain one oxygen atom thus are more reactive than oxygen, such as superoxide anion (O2^•−), and primarily generated by mitochondria's electron transport chain and NADPH oxidase (NOX) activity (Tauffenberger & Magistretti, 2021). Physiological levels of ROS have prominent roles in the modulation of a variety of cell signals, among which extracellular signal-related kinases 1/2 (ERK1/2), mammalian target of rapamycin, phosphatidylinositide 3-kinase (PI3K) and protein kinase B are critically involved in cell survival and growth, while nuclear factor kappa-B (NF-κB) and nuclear factor erythroid 2-related factor 2 are closely related to expressions of antioxidant enzymes and molecules, and metabolism-related transcription factors (Ballard & Towarnicki, 2020; Tauffenberger & Magistretti, 2021). However, excess ROS under OS condition would disturb normal physiological signals, induce peroxidation of nucleic acids, lipids and proteins, eventually causing cell damage and pathological reactions (Ballard & Towarnicki, 2020; Tauffenberger & Magistretti, 2021). In particular, the damage of cell membrane lipid bilayers will lead to abnormal transmembrane exchanges of a variety of molecules, ions, etc., which would disrupt physiological and biochemical homeostasis within the cell, and consequentially cause cell damage. The brain neurons, astrocytes and microglia are rich in mitochondria and NOX, thus capable of generating abundant ROS (Culmsee et al., 2018; Nayernia et al., 2014). Meanwhile, brain is a high oxygen consuming organ, with high cortical surface area, enriched with unsaturated fatty acids but lack of antioxidant enzymes, which result in low antioxidant capacity and high vulnerability to OS (Cobley et al., 2018; Patel, 2016).

Studies have revealed the pivotal role of OS in neurodegeneration diseases, such as the Alzheimer's disease and Parkinson's disease (Chen & Zhong, 2014; Subramaniam & Chesselet, 2013). Its emerging role in depression has been identified in recent years, which receives increasing attention (Bhatt et al., 2020; Wang et al., 2023). As supporting evidence, studies found significantly reduced antioxidants (e.g., SOD, GSH) and increased peroxidation biomarkers, such as carbonyl and malondialdehyde (MDA), in depressed patients and animals (Juszczyk et al., 2021; Liu et al., 2015; Visentin et al., 2020). Clinical studies found increased peroxidation biomarkers (MDA, 8-OHdG and F2-isoprostanes) and reduced antioxidants (e.g., SOD), in the plasma and serum of major depressive disorder (MDD) patients (Camkurt et al., 2016; Herken et al., 2007; Lindqvist et al., 2017; Yager et al., 2010; Table 1, oxidative stress section). Animal studies further specified stress-induced abnormalities across brain subregions, such as the reduced GSH and SOD activity, and the increased levels of ROS, MDA and carbonyl, in the prefrontal cortex (PFC) and hippocampus of depressed rats, caused by chronic unpredictable mild stress (CUMS), chronic restraint stress (CRS) and chronic social isolation, respectively (Parul et al., 2021; Zafir & Banu, 2007; Zlatković et al., 2014; Table 2, oxidative stress section). Moreover, the antioxidant effect of antidepressants, and antidepressant effect of antioxidants have been revealed in depressed patients and rodents, respectively. For example, antidepressants fluoxetine and citalopram increased the SOD activity, and reduced MDA and carbonyl levels in MDD patients (Herken et al., 2007; Khanzode et al., 2003) and CRS treated rats (Zafir & Banu, 2007), while antioxidant diallyl disulfide reversed lipopolysaccharide (LPS)-induced depressive-like behaviors in mice (Wei et al., 2021). The results indicate OS as an important depression pathogenesis, and further suggest enhancing antioxidant activity as a promising therapeutic avenue. To expound possible underlying mechanisms mediating the OS influence, the following sections describe mitochondrial impairment and related adenosine triphosphate (ATP) deficiency, neuroinflammation, glutamate excitotoxicity, BDNF/TrkB dysfunction and 5-HT deficiency in the brain, and disturbed MGB and HPA axes. We also elaborate the intricate interactions between the above aspects, and the molecular mechanisms mediating the interplay.

Table 1.

The Underlying Mechanisms of OS Induced Depression, as Revealed by Clinical Studies.

Mechanisms	Subject	Treatment/outcome	References
Oxidative stress	MDD patients	Increased 8-OHdG and F2-isoprostanes in the plasma and serum	Yager et al. (2010); Lindqvist et al. (2017)
		Increased MDA and reduced SOD in the serum	Herken et al. (2007); Camkurt et al. (2016)
		Antidepressants reduced MDA, increased SOD	Khanzode et al. (2003); Herken et al. (2007)
Mitochondrial impairment	MDD patients	Reduced ATP generation, electron transport chain and enzyme activity in brain mitochondria	Bansal and Kuhad (2016); Allen et al. (2018)
Neuro-inflammation	Hepatitis C patients	INF-α treatment induces depression	Bonaccorso et al. (2002); Keefe (2007)
	MDD patients	Elevated IL-1β, IL-6 and TNF-α in the serum	Lindqvist et al. (2017); Zou et al. (2018)
		Increased IL-6, TNF-α and TSPO in multiple brain subregions (ACC/PFC/hippocampus)	Enache et al. (2019); Gritti et al. (2021)
		Behavioral therapy reduces peripheral IL-6	Ma et al. (2022)
		Fluoxetine reduces IL-1β, IL-6 and TNF-α	García-García et al. (2022)
Glutamate excitotoxicity	MDD patients	Increased glutamate in the frontal cortex	Hashimoto et al. (2007)
		Increased glutamate in the CSF	Madeira et al. (2018)
BDNF/TrkB dysfunction	MDD patients	Reduced BDNF in the serum and plasma	Karege et al. (2005); Lee and Kim (2008)
		Reduced BDNF in the PFC and hippocampus	Autry and Monteggia (2012); Brigadsk and Leßmann (2020)
MGB disturbance	MDD patients	Reduced abundance of Bifidobacteria and Lactobacilli, increased abundance of Bacteroidea, Proteus and Actinobact eria	Sanada et al. (2020); Barandouzi et al. (2020)

Note: ACC = anterior cingulate cortex; ATP = adenosine triphosphate; BDNF = brain-derived neurotrophic factor; CSF = cerebrospinal fluid; IL-1β = interleukin-1β; IL-6 = interleukin-6; INF-α = Interferon-alpha; MDA = malondialdehyde; MDD = major depressive disorder; MGB = microbial-gut-brain; PFC = prefrontal cortex; SOD = superoxide dismutase; TrkB = tyrosine kinase B; TSPO = 18-kda Translocator Protein; 8-OHdG = 8-hidroxy-2-deoxyguanosine.

Table 2.

The Underlying Mechanisms of OS Induced Depression, as Revealed by Rodent Studies.

Mechanisms	Subject	Treatment/outcome	References
Oxidative stress	Rats	CUMS/CRS reduced GSH/SOD, increased ROS/MDA in PFC/hippocampus, induced DLB	Zafir and Banu (2007); Parul et al. (2021)
	Mice	Antioxidant DADS reversed LPS-induced DLB	Wei et al. (2021)
Mitochondrial impairment	Rats	MS affected Ca2+, increased ROS, and induced DLB	Park et al. (2019)
	Mice	Corticosterone attenuated mitochondrial acetyl-CoA and TCA cycle, and induced DLB	Xie, Shen et al. (2020)
		Astrocytic/mPFC ATP deficiency-induced DLB, which was reversed by exogenous/endogenous ATPs	Cao et al. (2013); Lin et al. (2022)
	Mice/Rats	Exercise and NMN improved mitochondrial impairment, and alleviated DLB	Park et al. (2019); Xie, Yu et al. (2020)
Neuro-inflammation	Rats	MS/CUMS increased IL-1β, IL-6, TNF-α and NF-κB in PFC/hippocampus, and induced DLB	Giridharan et al. (2019); Parul et al. (2021)
		Fluoxetine reduced CUMS increased IL-1β in PFC	Pan et al. (2014)
	Mice	Fluoxetine reduced LPS elevated hippocampal IL-1β, IL-6 and TNF-α	Duda et al. (2017)
		Anti-inflammatory drugs sinomenine and berberine alleviated stress-induced DLB	Liu et al. (2018); Qin et al. (2023)
Glutamate excitotoxicity	Mice/Rats	Injections of GLT-1 inhibitor DHK, into the LHb, amygdala and PFC, induced DLB	Cui et al. (2014); John et al. (2012, 2015)
		CUMS inhibited astrocytic GLT-1 expression in PFC/hippocampus, and induced DLB	Liu et al. (2019); Yang et al. (2020)
		Upregulated GLT-1, by Riluzole and antidepressants fluoxetine and Xiao-Yao-San, alleviated CUMS-induced DLB	Banasr et al. (2010); Ding et al. (2017); Liu et al. (2019)
		NMDA antagonists alleviated DLB	Autry et al. (2011)
BDNF/TrkB dysfunction	Mice/Rats	CUMS and CSDS reduced BDNF expression in PFC/hippocampus, and induced DLB	Mao et al. (2009); Zhang et al. (2015)
		Ketamine and its metabolite increased BDNF in mPFC, which was required for antidepressant action	Fukumoto et al. (2019; Lepack et al. (2014)
		Ketamine and fluoxetine took effects through directly activating TrkB in PFC/hippocampus	Casarotto et al. (2021)
MGB disturbance	Rats	Microbiota depletion and FMT disturbed microbiota homeostasis, and induced DLB	Hoban et al. (2016); Kelly et al. (2016)
	Mice	CSDS reduced Lactobacillus, Bacteroidales and Clostridiales, and induced DLB	Qu et al. (2017); X. Zhu et al. (2023)
		Ketamine and fluoxetine restored microbiota homeostasis and alleviated DLB	Qu et al. (2017); Lukić et al. (2019)
		Probiotic Bifidobacterium conferred resilience to CSDS, Lactococcus and FMT alleviated CUMS-induced DLB	Yang et al. (2017); Y. Zhang et al. (2019); Gao et al. (2022)

Note: ATP = adenosine triphosphate; BDNF = brain-derived neurotrophic factor; CRS = chronic restraint stress; CSDS = chronic social defeat stress; CUMS = chronic unpredictable mild stress; DADS = diallyl disulfide; DHK = Dihydroalginate; DLB = depressive-like behaviors; FMT = fecal microbiota transplantation; GLT-1 = glutamate transporter 1; GSH = glutathione; IL-1β = interleukin-1β; IL-6 = interleukin-6; LHb = lateral habenula; LPS = lipopolysaccharide; MDA = malondialdehyde; MGB = microbial-gut-brain; MS = maternal separation; mPFC = medial prefrontal cortex; NF-κB = nuclear factor kappa-B; NMDA = N-methyl-D-aspartic acid; NMN = nicotinamide mononucleotide; PFC = prefrontal cortex; ROS = reactive oxygen species; SOD = superoxide dismutase; TCA = tricarboxylic acid; TNF-α = tumor necrosis factor-α; TrkB = tyrosine kinase B.

The Underlying Mechanisms of OS-Induced Depression

Mitochondrial Impairment and ATP Deficiency in the Brain

Mitochondria are organelles that are mainly responsible for oxidative metabolism of energy substances (e.g., carbohydrates, proteins, lipids), to provide ATPs for organisms. They are one primary source of ROS generation, and in turn are the most vulnerable organelles of excess ROS. Brain neurons, astrocytes and microglia are enriched with mitochondria thus could generate abundant ATPs, which meanwhile consume a large quantity of ATPs for executing complicated brain functions, rationalizing the extensively affected brain functions caused by mitochondrial impairment and ATP deficiency. Clinical studies have found morphological and functional abnormalities of brain mitochondria in MDD patients, such as reduced ATP generation, electron transport chain and enzyme activity (Allen et al., 2018; Bansal & Kuhad, 2016; Table 1, mitochondrial impairment section). Animal studies found maternal separation (MS) and corticosterone impaired central mitochondrial metabolism, and induced depressive-like behaviors in rats and mice, which were reversed by physical exercise and nicotinamide mononucleotide administration (Park et al., 2019; Xie, Shen et al., 2020; Xie, Yu et al., 2020; Table 2, mitochondrial impairment section). Impaired mitochondrial metabolism would lead to reduced ATP generation, in a way that mitochondrial ROS (mtROS) induces cardiolipin peroxidation, which reduces the activity of complex V (ATP synthase) of mitochondrial electron transport chain (de Oliveira et al., 2021; Pointer & Klegeris, 2017; Figure 1A). Rodent studies found central ATP deficiency, astrocytes derived and medial prefrontal cortex (mPFC) localized, resulted in depressive-like behaviors in mice, which were ameliorated by both exogenous ATP administration and stimulated endogenous ATP release, indicating the pivotal role of central ATPs in depression onset and remedy, respectively (Cao et al., 2013; Lin et al., 2022). Furthermore, astrocytic epoxyeicosatrienoic acid signaling in the mPFC modulated depressive-like behaviors in mice, through regulating the astrocytic ATP release that was required for antidepressant effect (Xiong et al., 2019). The above evidence suggests critical involvement of central mitochondrial impairment and related ATP deficiency in depression.

Figure 1.

Os causes depression through inducing mitochondrial impairment (A), neuroinflammation (B), glutamate excitotoxicity (C), BDNF/TrkB dysfunction (D) and 5-HT deficiency (E) in the brain, and the MGB disturbance (F) and the HPA axis dysregulation (G). A. Mitochondrial impairment induces central ATP deficiency, through elevating mtROS and reducing cardiolipin and ATP synthase, with astrocytes derived and mPFC localized ATP deficiency closely associated with depression. OS and mitochondria have mutual influence through mtROS, neuroinflammation and excess glutamate induced Ca²⁺ influx (A-OS). B. Neuroinflammation is characterized by microglia activation, pro-inflammatory cytokine expressions, etc. OS/Mitochondria-neuroinflammation interaction (AB, B-OS): Excess ROS increases pro-inflammatory cytokines, through elevating histone acetylation and activating NLRP3/caspase-1, AP-1 and NF-κB, and aberrant astrocytic OXPHOS activity; neuroinflammation promotes IDO/TDO/KMO to generate QUIN that activates NMDA, increases glutamate, triggers Ca²⁺ influx, which causes cytochrome C release and reduces SIRT3 activity and ATP generation, or depolarizes mitochondria membrane and generates excess ROS. C. Glutamate excitotoxicity is characterized by central synaptic accumulation. Glutamate-OS interaction (C-OS): Excess ROS results in excitotoxicity through affecting Ca²⁺, GS and GLT-1 expression; excess glutamate impairs mitochondria through NMDA activation and Ca²⁺ influx. Neuroinflammation-glutamate interaction (BC): Neuroinflammation promotes excitotoxicity through increasing microglial release or blocking astrocytic reuptake, by downregulating GLT-1 or elevating GLS and QUIN; excess glutamate might increase pro-inflammatory cytokines through NMDA/AMPA receptors, but reduce the cytokines through mGluRs. D. BDNF/TrkB dysfunction is characterized by inhibited BDNF expression and TrkB activation (PFC/hippocampus). BDNF-OS interaction (D-OS): Excess ROS inhibits BDNF expression, through reducing CREB and elevating NF-κB activity; BDNF owns antioxidant property through TrkB, ERK1/2 and NF-κB/sestrin2, thus, the dysfunction promotes OS. Neuroinflammation-BDNF/TrkB interaction (BD): NF-κB mediates LPS-reduced BDNF expression through interacting with BDNF/TrkB; BDNF promotes TNF-α and IL-1β release by activating astrocytes and microglia. Glutamate-BDNF/TrkB interaction (CD): Excess glutamate might inhibit BDNF expression by activating extra-synaptic NMDA and eEF2; BDNF promotes the release or protects neurons against the excitotoxicity of glutamate, through PI3K, PLC-γ, ERK, etc., whose dysfunction weakens the influence correspondingly. E. Central 5-HT interacts with multiple aspects, whose deficiency attenuates the influence correspondingly: 5-HT presents antioxidant properties, while OS inhibits 5-HT through neuroinflammation and MGB disturbance (E-OS); Neuroinflammation inhibits 5-HT by promoting KP and modulating 5-HTT (BE); 5-HT inhibits glutamate transmission through 5-HT 1A/B receptor, which in turn regulates 5-HT function through VGLUT1 (CE); 5-HT promotes BDNF expression and TrkB activation, while BDNF affects 5-HTT function (DE). F. MGB disturbance is characterized by disturbed microbiota composition/metabolism. MGB-OS/mitochondria interaction (AF, F-OS): Microbiota affects ROS generation by modulating activities of NOX, SCFAs and antioxidants, and mitochondria metabolism; Mitochondrial genotype/mutation affects microbiota through mtROS. Neuroinflammation-MGB interaction (BF): MGB disturbance leads to gut inflammation, increases QUIN that damages BBB and results in neuroinflammation; Neuroinflammation disturbs microbiota homeostasis, through pro-inflammatory cytokines and NLRP3. BDNF-MGB interaction (DF): BDNF modulates intestinal barrier integrity; MGB disturbance reduces hippocampal BDNF expression. 5-HT-MGB interaction (EF): Microbiota disturbance inhibits 5-HT generation, by promoting intestinal/central KP, affecting 5-HT and TPH 1 gene expressions; 5-HTT genotype modulates microbiota composition. G. HPA axis interacts with multiple aspects, whose dysregulation attenuates the influence correspondingly: HPA affects ROS generation, mitochondria gene expression and metabolism through GRs, while NO mediates IL-1β and cholinergic stimulation of HPA (G-OS); HPA interacts with neuroinflammation through cytokines and GR/NF-κB signaling (BG); HPA affects glutamate release (through GR-TrkB interaction) and related gene expression (CG); GR activation reduces BDNF expression, while BDNF affects GR phosphorylation and rewrote glucocorticoid transcriptome (DG); HPA hormones increase intestinal barrier permeability and disturb microbiota composition, while MGB affects HPA through GABAergic activity (FG). AB, BC, CD, DE, EF and AF have indirect interaction through OS. AP-1 = activating protein-1; AMPA = α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic; ATP = adenosine triphosphate; BDNF = brain-derived neurotrophic factor; CREB = cAMP responsive element binding protein; EAATs = excitatory amino acid transporters; eEF2 = eukaryotic elongation factor 2; ERK = extracellular signal-related kinases; GS = glutamine synthetase; GLS = glutaminase; Glu = glutamate; GLT-1 = glutamate transporter 1; IDO = indoleamine 2,3-dioxygenase; IL-1β = interleukin-1β; IL-6 = interleukin-6; KMO = kynurenine 3-monooxygenase; KP = kynurenine pathway; MGB = microbial-gut-brain; mPFC = medial prefrontal cortex; mtROS = mitochondrial ROS; NF-κB = nuclear factor kappa-B; NLRP3 = NOD-like receptor thermal protein domain associated protein 3; NMDA = N-methyl-D-aspartic acid; NOX = NADPH oxidase; OS = oxidative stress; OXPHOS = oxidative phosphorylation; PLC-γ = phospholipase C-γ; PKC = protein kinase C; PI3K = phosphatidylinositide 3-kinase; PVN = paraventricular nucleus; QUIN = quinolinic acid; RNS = reactive nitrogen species; ROS = reactive oxygen species; SCFAs = short-chain fatty acids; SIRT3 = sirtuin 3; SRC =sparse representation-based classification; TDO =tryptophan-2,3-dioxygenase; TPH = tryptophan hydroxylase; TNF-α = tumor necrosis factor-α; TrkB = tyrosine kinase B; TRYCATs = tryptophan catabolites; VGLUT1 = vesicular glutamate transporter 1; 5-HTT = serotonin transporter.

Studies have revealed mutual interaction between OS and mitochondrial impairment, through mtROS, neuroinflammation, excess glutamate-induced Ca²⁺ influx, etc. (de Oliveira et al., 2021; Moncrieff et al., 2022; Visentin et al., 2020; Figure 1A-OS). The mutual influence between mitochondrial impairment and neuroinflammation is critically involved in depression (Hollis et al., 2022; Visentin et al., 2020; Figure 1AB). Impaired mitochondria leads to increased mtROS generation, which promotes pro-inflammatory cytokine gene expression through promoting protein-1 (AP-1) and NF-κB, and increasing histone acetylation (Adcock et al., 2005; Rahman et al., 2004), or activating caspase-1 and inflammasome NOD-like receptor thermal protein domain associated protein 3 (NLRP3) (Visentin et al., 2020; Zhou et al., 2011), or affecting astrocytic mitochondrial oxidative phosphorylation activity (Mi et al., 2023). In turn, neuroinflammation promotes mtROS generation, for instance, interleukin-1β (IL-1β) and IL-18 promote kynurenine pathway (KP) of tryptophan metabolism, through activating the enzymes of indoleamine 2,3-dioxygenase (IDO), tryptophan-2,3-dioxygenase (TDO) and kynurenine 3-monooxygenase (KMO), and generating tryptophan catabolites (TRYCATs) among which quinolinic acid (QUIN) directly activates N-methyl-D-aspartic acid (NMDA) receptors, increases synaptic glutamate, triggers Ca²⁺ influx that depolarizes mitochondrial membrane, which finally increases mtROS generation and reduces antioxidant activity (de Oliveira et al., 2021; Schwarcz et al., 2012; Visentin et al., 2020). Furthermore, increased intracellular Ca²⁺ activates neural nitric oxide synthase (NOS) and elevates nitric oxide level, decreases the activity of sirtuin 3 (SIRT3, an important deacetylase), causes mitochondrial permeability transition pores open and cytochrome C release, and finally reduces ATP generation (Visentin et al., 2020). In addition, tumor necrosis factor-α (TNF-α) phosphorylates tyrosine 304 in subunit I of cytochrome C oxidase in complex IV, to impair mitochondria and reduce ATP generation (Visentin et al., 2020).

Neuroinflammation

Neuroinflammation is the immune response of the brain to disturbed internal homeostasis, which involves microglia activation, pro-inflammatory cytokine release, ROS generation, NF-κB and NLRP3 expression and activation, etc. (Ransohoff et al., 2015). Since the early clinical findings of depressive symptoms in hepatitis C patients treated with interferon-α (Bonaccorso et al., 2002; Keefe, 2007), accumulating studies have revealed the prominent role of neuroinflammation in depression (Beurel et al., 2020; Troubat et al., 2021; Figure 1B, Tables 1–2, neuroinflammation section). Several pro-inflammatory pathways in the pathogenesis, treatment and remission of MDD have been addressed, including the caspase 1, inducible NOS (iNOS) and interferon-gamma (IFN-γ) (Inserra et al., 2019). Clinical studies found significantly elevated levels of IL-1β，IL-6 and TNF-α in the serum (Lindqvist et al., 2017; Zou et al., 2018), and increased expressions of IL-6, TNF-α and neuroinflammation biomarker 18-kda translocator protein, in the cerebrospinal fluid (CSF) and brain subregions, including the anterior cingulate cortex (ACC), PFC and hippocampus (Enache et al., 2019; Gritti et al., 2021), in MDD patients (Table 1, neuroinflammation section). Rodent studies demonstrated early life stress perturbed microglial gene expression, morphology and function in the developing rodent hippocampus (Johnson & Kaffman, 2018). MS increased microglial activation and expressions of pro-inflammatory cytokines IL-6, INF-γ and TNF-α in rat PFC (Giridharan et al., 2019). Similarly, later life social isolation stress upregulated the NF-κB activation and iNOS expressions in rat PFC (Zlatković et al., 2014; Zlatković & Filipović, 2013), while CUMS increased the expressions of TNF-α and IL-1β, NLRP3 and transcription regulator NF-κB, in the PFC and hippocampus of rats (Pan et al., 2014; Parul et al., 2021; Table 2, neuroinflammation section).

Furthermore, clinical and animal studies found anti-inflammatory effect of antidepressants (Dionisie et al., 2021; Więdłocha et al., 2018), and antidepressant effect of anti-inflammatory drugs (Simon et al., 2023). For instance, clinical studies found behavioral therapy reduced peripheral IL-6 level (Ma et al., 2022), while antidepressant fluoxetine reduced the levels of IL-1β, IL-6 and TNF-α (Garcia-Garcia et al., 2022), in MDD patients. Similarly, rodent studies found fluoxetine reduced the CUMS elevated IL-1β in rat PFC (Pan et al., 2014), and the LPS elevated IL-1β, IL-6 and TNF-α in mouse hippocampus (Duda et al., 2017). On the other hand, drugs with anti-inflammatory properties, such as the Geraniol, Sinomenine, Patchouli alcohol, Baicalin and Berberine, alleviated the the CUMS and corticosterone-induced depressive-like behaviors in mice, respectively (Bai et al., 2023; Deng et al., 2015; He et al., 2023; L. T. Guo et al., 2019; Liu et al., 2018; Qin et al., 2023). Patchouli alcohol and Berberine take effect through inhibiting the NLRP3-mediated neuroinflammation (He et al., 2023; Qin et al., 2023), while Baicalin takes effect through inhibiting the toll-like receptor 4 expression via the PI3K/AKT/FoxO1 pathway (L. T. Guo et al., 2019). A recent study further showed Saikosaponin C alleviated the chronic social defeated stress (CSDS)-induced depressive-like behaviors in mice, through inhibiting the IL-6 methylation, which indicates the prominent role of epigenetic modulation of pro-inflammatory cytokines in depression (Bai et al., 2023). The above evidence indicates neuroinflammation as an important pathogenesis of depression, and suggests anti-inflammation as a potentially effective remedy avenue.

Studies have revealed the important role of OS-neuroinflammation interaction in depression (Lindqvist et al., 2017; Rawdin et al., 2013; Figure 1B-OS). Excess ROS promotes pro-inflammatory cytokine expressions, through activating AP-1 and NF-κB, and increasing histone acetylation (Adcock et al., 2005; Rahman et al., 2004), or activating caspase-1 and NLRP3 (de Oliveira et al., 2021; Visentin et al., 2020). In turn, neuroinflammation induces mitochondrial impairment and excess ROS generation, through activating IDO/TDO/KMO to promote KP and generate QUIN, which activates NMDA, increases synaptic glutamate, triggers Ca²⁺ influx that depolarizes the mitochondrial membrane (Catena-Dell’Osso et al., 2013; de Oliveira et al., 2021; Moncrieff et al., 2022; Schwarcz et al., 2012; Visentin et al., 2020). As supporting evidence, studies showed both peripheral and central inflammation-promoted KP and elevated QUIN in the brain, through increasing the IDO/TDO/KMO activities (Connor et al., 2008; Fu et al., 2010; Lawson et al., 2013; Raison et al., 2010), which was inhibited by antidepressants (Eskelund et al., 2017). QUIN could act as NMDA receptor agonist and elevate synaptic glutamate (Tavares et al., 2002; Visentin et al., 2020), and promote OS through increasing the ROS generation and lipid peroxidation (Rios & Santamaria, 1991; Stone et al., 2000).

Neuroinflammation could induce 5-HT deficiency in the brain, through promoting KP by activating IDO/TDO, which has been recognized as a classic depression pathogenesis (Catena-Dell’Osso et al., 2013; Moncrieff et al., 2022; Figure 1BE). Tryptophan, the precursor of 5-HT, is an essential amino acid obtained from the diet, which could pass blood-brain barrier (BBB) into the brain and be metabolized to 5-HT, catalyzed by tryptophan hydroxylase (TPH) 2 and aromatic amino acid decarboxylase (AAAD), or converted into kynurenine, catalyzed by IDO (Gheorghe et al., 2019; Krautkramer et al., 2021; Schwarcz et al., 2012). Neuroinflammation-promoted KP would reduce tryptophan available for the 5-HT pathway, resulting in reduced 5-HT generation in the brain. A study indeed found inflammation and CRS enhanced IDO activity, promoted KP, and reduced 5-HT concentration in mouse hippocampus, which were reversed by the IDO inhibitor 1-methyl-tryptophan (Deng et al., 2021). Kynurenine could either be catalyzed by kynurenine aminotransferases to kynurenic acid in astrocytes, which acts as NMDA antagonist and exerts neuroprotection effect, or catalyzed by KMO to QUIN in microglia (Gheorghe et al., 2019; Krautkramer et al., 2021; Schwarcz et al., 2012). QUIN could act as NMDA receptor agonist and cause synaptic glutamate accumulation, through both stimulating the release and inhibiting the uptake, thus is neurotoxic (Tavares et al., 2002; Visentin et al., 2020). Neuroinflammation could also directly increase microglial glutamate release through increasing glutaminase (GLS), via protein kinase C and cAMP responsive element binding protein (CREB) (Chen et al., 2012; Takeuchi et al., 2006), or inhibiting astrocytic reuptake through downregulating excitatory amino acid transporters (EAATs), especially glutamate transporter 1 (GLT-1) (Haroon & Miller, 2017; Tilleux & Hermans, 2007; Figure 1BC). The above mechanisms might contribute to neuroinflammation elevated glutamate in the brain (e.g., ACC and basal ganglia) of MDD patients, assessed by magnetic resonance spectroscopy, which was closely correlation with anhedonia phenotype and increased inflammation markers (e.g., C-reactive protein) (Haroon & Miller, 2017). Alternatively, neuroinflammation might regulate glutamate signaling indirectly through OS (Miller & Raison, 2016), as studies found IL-1β inhibited glutamate reuptake and LPS stimulated glutamate release were blocked by oxide synthase and NOX inhibitors, respectively (Barger et al., 2007; Hu et al., 2000). Glutamate excitotoxicity would be achieved once excess glutamate accumulated in the synaptic space of the brain, which also serves as an important downstream pathway of OS.

Central Glutamate Excitotoxicity

Glutamate is the major excitatory neurotransmitter in the brain, produced de novo from the mitochondrial tricarboxylic acid cycle intermediate metabolite α-Ketoglutarate. The intracellular glutamate concentration is approximate 10 mmol/L, much higher than the levels in synaptic cleft (1 mmol/L), extracellular fluid, CSF and plasma (0.5–4, 10, 150 μmol/L, respectively) (Montanari et al., 2022; Pittenger et al., 2008). Glutamatergic neurons and astrocytes form glutamate-glutamine cycle to maintain synaptic homeostasis, mainly through reuptaking extracellular glutamate by astrocytic EAATs, which are the primary terminating mechanism for the actions of synaptic glutamate, as the synaptic space is short of glutamate degrading enzymes (Hertz & Zielke, 2004). Five subtypes of EAATs (referred as EAAT1-5) have been found so far, among which EAAT 1 and 2 (referred as GLAST and GLT-1 in rodents, respectively) are mainly expressed in astrocytes and primarily responsible for astrocytic reuptake of synaptic glutamate (Hertz & Zielke, 2004; Murrough et al., 2017). Studies have found GLT-1 is mainly distributed in hippocampus and cerebral cortex, while GLAST is mainly distributed in cerebellum and retina (Todd & Hardingham, 2020; Zhou et al., 2019). Glutamate cannot cross the BBB, thus excessive accumulation in the synaptic space of the brain, caused by promoted neuronal release or EAAT/GLT-1 inhibition reduced astrocytic reuptake, leads to glutamate excitotoxicity, which would damage central neurons and glial cells, and cause neurological and psychiatric disorders (Lau & Tymianski, 2010; Pajarillo et al., 2019).

Considerable studies have revealed the prominent role of glutamate excitotoxicity in depression (Murrough et al., 2017; Sanacora et al., 2008; Figure 1C, Tables 1–2, glutamate excitotoxicity section). Clinical studies found significantly increased glutamate levels in the CSF and brain subregions of MDD patients (Hashimoto et al., 2007; Madeira et al., 2018; Table 1, glutamate excitotoxicity section). In consistent, rodent studies demonstrated inducement of depressive-like behaviors in mice and rats, by injections of GLT-1 inhibitors WAY-213613 into the lateral habenula (LHb) (Lyu et al., 2021), and dihydrokainic acid into multiple brain subregions, including the LHb (Cui et al., 2014), lateral ventricle (Bechtholt-Gompf et al., 2010), infralimbic cortex (Gasull-Camós et al., 2017), central amygdala (John et al., 2015) and PFC (John et al., 2012). Moreover, CUMS exposure reduced astrocytic GLT-1 expression in the PFC and hippocampus, and induced depressive-like behaviors in rats and mice (Liu et al., 2019; Yang et al., 2020; Table 2, glutamate excitotoxicity section). Furthermore, antidepressants fluoxetine and Chinese herbal formula Xiao-Yao-San reversed depressive-like behaviors, through upregulating the CUMS inhibited GLT-1 expression in mice (Ding et al., 2017; Liu et al., 2019). In addition, the glutamatergic signal inhibitory drugs alleviated depressive-like behaviors, such as the GLT-1 promoting drug Riluzole (Banasr et al., 2010), and a variety of NMDA receptor antagonists (Autry et al., 2011). However, NMDA receptor antagonists, such as phencyclidine and dizocilpine, could impair cognitive function, which precludes its utilization as antidepressants (B. Zhang et al., 2019; Zhang et al., 2017). Meanwhile, elevated O-linked-N-acetylglucosaminylation (O-GlcNAcylation), in the mPFC of mice, had antidepressant effect (Cho et al., 2020). However, a recent study found the mRNA of O-GlcNAc transferase (OGT), a key enzyme in protein O-GlcNAcylation, was increased in MDD patients and in mPFC astrocytes of CSDS susceptible mice (Fan et al., 2023). Moreover, the OGT overexpression increased CSDS susceptibility of mice, whereas the OGT knockout reduced the astrocytic GLT-1 GlcNAcylation, and resulted in antidepressant effect (Fan et al., 2023). The results indicate the prominent role of mPFC O-GlcNAcylation in the modulation of depression, however, the detailed underlying mechanisms and the potential differential roles of neurons and astrocytes (e.g., astrocytic GLT-1) require further in-depth investigations for clarification (Fan et al., 2023; Paton & Menard, 2023). The above evidence indicates central glutamate excitotoxicity, caused by GLT-1 inhibition disrupted astrocytic reuptake of synaptic glutamate, as an important pathogenesis of depression. The results further suggest GLT-1 modulation as a potentially promising antidepressant pathway.

Studies have found mutual interaction between glutamate excitotoxicity and OS (Figure 1C-OS). Excess ROS promotes glutamate excitotoxicity, through inhibiting the astrocytic GLT-1 expression and affecting Ca²⁺ and glutamine synthetase (Dallas et al., 2007; Volterra et al., 1994). In turn, excess glutamate exacerbates OS by impairing mitochondria and promoting the ROS generation, through activating NMDA and triggering Ca²⁺ influx (Atlante et al., 2001). As supporting evidence, in vitro studies showed ROS generation was produced upon NMDA receptor activation (Lafon-Cazal et al., 1993; Reynolds & Hastings, 1995), which was dependent on mitochondria polarization, induced by glutamate caused Ca²⁺ influx that also reduced ATP generation (Castilho et al., 1999; Pereira & Oliveira, 2000; Reynolds & Hastings, 1995). Furthermore, interaction between OS and neuroinflammation might be required for the promoting effect, as oxide synthase and NOX inhibitors can block the process (Barger et al., 2007; Hu et al., 2000; Tilleux & Hermans, 2007), and hypoxia affected GLT-1 function required the present of TNF, which could cause glutamate excitotoxicity by itself (Boycott et al., 2008; Clark & Vissel, 2016). For the interaction between neuroinflammation and glutamate signaling (Figure 1BC), neuroinflammation increases microglial glutamate release as mentioned above, through affecting the expression and activity of several regulatory proteins, such as downregulating the GLT-1 (Haroon & Miller, 2017; Tilleux & Hermans, 2007), increasing the GLS (Chen et al., 2012; Takeuchi et al., 2006), or promoting the QUIN generation (de Oliveira et al., 2021; Visentin et al., 2020). In turn, glutamate might regulate the microglial release of pro-inflammatory cytokines, in a receptor subtype-dependent manner (Murugan et al., 2013). In vitro studies, performed in the primary microglial cultures, demonstrated that glutamate promoted the TNF-α and IL-1β expressions through α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic (AMPA) receptor activation (Sivakumar et al., 2010), while the NMDA inhibition reduced the hypoxia-induced TNF-α and IL-1β expression (Murugan et al., 2011), and the metabotropic glutamate receptor 5 (mGluR5) activation significantly reduced TNF-α generation (Byrnes et al., 2009). The findings might suggest glutamate upregulates pro-inflammatory cytokine expressions in activated microglia through ionotropic NMDA/AMPA receptors, whereas downregulates the expression through mGluRs.

Central BDNF/TrkB Dysfunction

Studies found glutamate quickly promoted BDNF expression in vitro, as reflected by increased levels of BDNF mRNA and protein (H. Wu et al., 2004; Jean et al., 2008). BDNF is a major type of neurotrophic factor in the brain, released by different types of cells (e.g., endothelial cells, microglia and megakaryocytes), which is critically involved in diversified functions of the brain, such as maintaining neuronal plasticity and regulating mental state (Lin & Huang, 2020). BDNF concentration is significantly higher in the serum than in the plasma and CSF, as over 90% is stored in blood platelets (Fujimura et al., 2002; Naegelin et al., 2018). Early clinical studies found significantly reduced BDNF in the serum and plasma of MDD patients (Karege et al., 2005; Lee & Kim, 2008; Table 1, BDNF/TrkB dysfunction section), probably caused by reduced platelet release (Serra-Millàs, 2016). In the brain, the gene expression and protein synthesis of BDNF and TrkB receptors have been found in multiple subregions, especially those closely related to depression (e.g., the PFC, hippocampus, hypothalamus) (Brigadski & Leßmann, 2020). BDNF evaluation in the human brain could only be acquired by analyzing postmortem brain tissues, which demonstrates subregion specific BDNF alternations in MDD patients, with the findings of reduced BDNF mRNA and protein in the PFC and hippocampus, but increased BDNF expressions in the nucleus accumbens (NAc) and amygdala (Autry & Monteggia, 2012; Brigadski & Leßmann, 2020; Zhang et al., 2016). Animal studies further showed BDNF concentrations in mouse serum and plasma were too low to be detected, which probably attributes to the lack of BDNF gene in mouse megakaryocytes, and the consequential insufficient BDNF protein synthesis (Chacón-Fernández et al., 2016). Consistent with the clinical findings, rodent studies found CUMS and CSDS impaired the BDNF signaling in the PFC and hippocampus, and induced depressive-like behaviors in rats and mice, respectively (Mao, Ip et al., 2009; Mao, Xian et al., 2010; Zhang et al., 2015; Table 2, BDNF/TrkB dysfunction section). Recent studies further found microglial BDNF depletion in the PFC increased the mouse susceptibility to CUMS-induced depressive-like behaviors (Woodburn et al., 2023), and the BDNF-TrkB signaling was critically involved in thalamic-primary auditory cortex circuit mediated resilience to CSDS in mice (Li et al., 2023). Furthermore, rodent studies showed multiple antidepressants take effect through increasing the BDNF expression, or directly binding to TrkB receptors (P. C. Casarotto et al., 2021; P. Casarotto et al., 2022). For instance, ketamine elevated the BDNF expression in the mPFC and hippocampus, which effect was suggested to be required for the behavioral actions of ketamine and its metabolite (2R, 6R)-hydroxynorketamine (Fan et al., 2021; Fukumoto et al., 2019; Lepack et al., 2014; Table 2, BDNF section). The downstream pathways of TrkB receptors mainly involve the PI3K, phospholipase C-γ (PLC-γ) and ERK signals (Kozisek et al., 2008). The above evidence indicates the critical contribution of BDNF/TrkB dysfunction to depression onset, and the requirement of BDNF expression and TrkB activation for antidepressant effect (Casarotto et al., 2022; Castren & Monteggia, 2021; Figure 1D, Tables 1–2, BDNF/TrkB dysfunction section).

Studies found mutual interaction between BDNF/TrkB signaling and OS, and neuroinflammation, respectively (Figure 1D-OS, BD). For BDNF-OS interaction, excess ROS inhibits the BDNF gene expression and protein synthesis, through reducing the CREB activity and increasing the NF-κB binding activity (A. Wu et al., 2004; Kapczinski et al., 2008; Zou & Crews, 2006). In turn, BDNF has antioxidant effect and protects neurons against OS, through TrkB, ERK 1/2 and NF-κB/sestrin2, thus BDNF deficiency might aggravate OS (S. D. Chen et al., 2017). For the interaction with neuroinflammation, BDNF promotes neuroinflammation through activating astrocytes and microglia to release TNF-α and IL-1β (Ding et al., 2020). In turn, LPS significantly reduced the BDNF protein level in mouse PFC and hippocampus, whereas increased the expression in the NAc (Guan & Fang, 2006; Zhang et al., 2014), probably through activating microglial NF-κB that interacts with the BDNF/TrkB signaling (Li et al., 2022). The prominent role of interaction between BDNF/TrkB and NF-κB in depression has been previously described (Caviedes et al., 2017). With regard to the BDNF-glutamate signaling interaction (Figure 1CD), in vitro studies found glutamate (100 μM and 100 mM) promoted the BDNF expression and secretion (H. Wu et al., 2004; Jean et al., 2008). Moreover, the higher glutamate dosage (1.0 mM) induced more BDNF release than the lower dose (0.1 mM), which suggests dose-dependent promoting effect (Taylor et al., 2003). However, the effect of toxic dosage glutamate on BDNF expression, in vitro and in vivo, remains unclear and requires further clarification. Excess glutamate might bind to the extra-synaptic NMDA receptors and activate eukaryotic elongation factor 2 (eEF2) kinase, which phosphorylates eEF2 that inhibits the BDNF expression, functioning in the opposite direction of NMDA antagonists, considering the prodepressant role of glutamate excitotoxicity, antidepressant role of NMDA blockage, and enhanced BDNF in the PFC and hippocampus (Autry et al., 2011; Bjorkholm & Monteggia, 2016; Murrough et al., 2017). In turn, BDNF protected neurons and glia cells against glutamate excitotoxicity in vitro, which effect involved Ca²⁺ homeostasis, cytokine expressions and related sparse representation-based classification (SRC) signals (Gaidin et al., 2020). The protection effect partially attributes to the BDNF upregulated astrocytic GLT-1 expression, via p42/p44 mitogen-activated protein kinase signals (ERK1/2) utilizing NF-κB (Rodriguez-Kern et al., 2003). In addition, the protection effect was blocked by inhibitors of ERK, PI3K and PLC-γ, which indicated vital involvement of the signals in the process (Almeida et al., 2005; Kume et al., 2000; Melo et al., 2013). In contrast, BDNF also rapidly and temporarily increased glutamate release in cultured neurons and astrocytes, which process was blocked by K252a and PP2 (specific inhibitors of TrkB and SRC, respectively), indicating the requirement of TrkB/SRC/PLC-γ1 pathway activation for the effect (Pascual et al., 2001; Zhang et al., 2013). Studies also demonstrated BDNF-promoted glutamate release, through activating the PLC-γ/IP3 pathway and inducing transient Ca²⁺ increase (Numakawa et al., 2001), and moreover, antidepressants imipramine and fluvoxamine potentiated the BDNF-promoted PLC-γ/IP3/Ca²⁺ signaling and glutamate release via sigma-1 receptors (Yagasaki et al., 2006). Additionally, a recent study found microglial BDNF promoted the NMDA receptor expression in the PFC of mice, further indicating the BDNF-promoted glutamatergic signaling (Woodburn et al., 2023). The above evidence might suggest BDNF modulates glutamate homeostasis in the brain, in a way that it promotes glutamate release under physiological condition (but only to limited extent), however, protects neurons and glia cells against glutamate excitotoxicity. The BDNF dysregulation would attenuate the promoted glutamate release or the excitotoxicity, correspondingly.

Despite of the well-recognized role of BDNF, the direct involvement of TrkB receptors in depression still requires further clarification. Limited research so far found TrkB inhibition caused resistant to antidepressants in TrkB.T1-overexpressing transgenic mice (Saarelainen et al., 2003), while K252a inhibited TrkB activation blocked ketamine's antidepressant effect in CUMS exposed rats (Liu et al., 2016). Moreover, TrkB activation, by administration of the agonist 7,8-Dihydroxyflavone in the PFC and hippocampus, had antidepressant effect through increasing the p-TrkB level, while the TrkB inhibitor ANA-12 administration in NAc exerted antidepressant effect by reducing the p-TrkB level, in depressed rats and mice caused by learned helplessness, CSDS and LPS administration, respectively (Shirayama et al., 2015; Zhang, Wu et al., 2014; Zhang, Yao et al., 2015). The brain subregion specific p-TrkB alternation is consistent with the regional BDNF abnormalities. More importantly, together with the findings that antidepressants can directly bind to TrkB receptors to produce antidepressant effect, results suggest the possibility of modulating depression phenotype through directly regulating TrkB expression and activation (P. C. Casarotto et al., 2021; P. Casarotto et al., 2022). The direct influence of OS, neuroinflammation and glutamate excitotoxicity on TrkB expression and activation is yet to be determined. The evidence in the present section indicates BDNF/TrkB dysfunction as a vital pathogenesis of depression, and moreover, suggests the BDNF/TrkB signaling physiological homeostasis restoration as a promising therapeutic avenue of the disease.

Central 5-HT Deficiency

The central monoamine deficiency, especially the 5-HT deficiency, is the classical depression hypothesis, based on which most first-line clinical antidepressants have been developed and widely utilized, such as tricyclic antidepressants (TCAs), monoamine oxidase inhibitors (MAOIs), and selective serotonin reuptake inhibitors (SSRIs), which take effects through upregulating the synaptic concentrations of monoamines (Malhi & Mann, 2018; Moncrieff et al., 2022). Over 90% of systematic 5-HT is synthesized by intestinal chromaffin cells, which can not cross the BBB under physiological condition, whereas its primary precursor tryptophan can (Correia & Vale, 2022; Gheorghe et al., 2019). Tryptophan could be metabolized to 5-HT, catalyzed by TPH1/AAAD in the gut and TPH2/AAAD in the brain, or alternatively, metabolized to kynurenine, catalyzed by IDO1/TDO in the gut and IDO in the brain (Gheorghe et al., 2019; Krautkramer et al., 2021). Thus, the enhanced IDO1/TDO activity would promote KP and inhibit 5-HT pathway, resulting in increased kynurenine and reduced 5-HT concentration, respectively. Gut inflammation would activate the IDO1/TDO activity and promote peripheral KP, which reduces tryptophan available to cross the BBB for central 5-HT production, while neuroinflammation enhanced IDO activity promotes central KP and directly inhibits 5-HT generation in the brain (Chen et al., 2023; Deng et al., 2021; Wang et al., 2022).

The central 5-HT could interact with OS directly, or indirectly through neuroinflammation, central glutamate and BDNF/TrkB signaling, and the MGB axis. For the interaction between 5-HT and OS (Figure 1E-OS), studies found 5-HT presented antioxidant properties in vitro, as indicted by the inhibited generations of ROS and iNOS (Vašíček et al., 2020), while N-acetyl 5-HT showed antioxidant and anti-apoptosis actions, by activating TrkB/CREB/BDNF pathway and antioxidant enzyme expression, in vitro and in vivo (Yoo et al., 2017). Recent studies demonstrated the possibility of designing fast-onset antidepressant, by dissociating 5-HTT from neuronal NOS in the DRN of mice, which indicates the critical involvement of 5-HT and OS interaction in depression pathogenesis (Guan & Pang, 2023; Kingwell, 2022; Sun et al., 2022; Ye et al., 2023). In turn, OS could cause central 5-HT deficiency through inducing neuroinflammation (Catena-Dell’Osso et al., 2013) and MGB disturbance (Stasi et al., 2019), by activating IDO/TDO enzymes and promoting KP of tryptophan metabolism, as expounded in the above sections correspondingly. For 5-HT and neuroinflammation interaction (Figure 1BE), pro-inflammatory cytokines, such as TNF-α and IL-1β, modulated the expression and function of serotonin transporter (SERT/5-HTT) (Malynn et al., 2013; Zhu et al., 2006). With regard to 5-HT and glutamate interaction (Figure 1CE), studies showed 5-HT receptors mediated the serotonergic inhibition of glutamate transmission in multiple brain subregions, through the 5-HT1B receptor inhibited presynaptic glutamate release, or the 5-HT1A receptor reduced postsynaptic glutamate receptor sensitivity (Guo & Rainnie, 2010; Nishijo et al., 2022). In turn, vesicular glutamate transporter 1 (VGLUT1)-dependent glutamate innervation of dorsal raphe nucleus (DRN) modulated the 5-HT function, as the VGLUT1^+/− mice presented decreased 5-HT neuronal activity and hippocampal 5-HT turnover (García-García et al., 2013). The glutamatergic drive of the DRN, which contains majority of 5-HT neurons in the brain, has been previously addressed (Soiza-Reilly & Commons, 2011). The involvement of 5-HT and BDNF/TrkB interaction (Figure 1DE) in depression has been previously addressed (Popova & Naumenko, 2019). 5-HT receptor activation with agonists promoted the BDNF expression and TrkB phosphorylation, in vivo and in vitro, therefore, the central 5-HT deficiency would lead to reduced BDNF expression and TrkB activation (de Foubert et al., 2007; Pascual-Brazo et al., 2012; Samarajeewa et al., 2014). In turn, BDNF dose-dependently reduced 5-HT reuptake through affecting the 5-HTT function, with the higher doses presenting more severe reuptake inhibition, in vitro (Mössner et al., 2000). However, intrahippocampal BDNF injection reduced extracellular 5-HT concentration in mice, which effect was blocked by the TrkB antagonist K252a, suggesting TrkB activation-dependent BDNF-promoted 5-HTT function (Benmansour et al., 2008). BDNF had no significant influence on the 5-HTT density in rat hippocampus (Benmansour et al., 2008), which was consistent with the finding of BDNF-dependent 5-HTT function, but not expression, in mice (Daws et al., 2007). Furthermore, loss of BDNF gene allele increased stress abnormalities and exacerbated monoamine deficiencies in the 5-HTT knockout mice (Ren-Patterson et al., 2005), and down-regulated the 5-HTT function in mouse hippocampus (Daws et al., 2007; Deltheil et al., 2008). The results suggest BDNF affected 5-HTT function, rather than the expression, at least in the hippocampus of mice. For the 5-HT and MGB interaction (Figure 1EF), microbiota disturbance could induce central 5-HT deficiency, through causing gut inflammation that activates IDO1/TDO and promotes peripheral KP, which reduces tryptophan available for central 5-HT production (Chen et al., 2023; Wang et al., 2022), or affecting the expressions of TPH1 (Reigstad et al., 2015) and 5-HT-related genes (Borrelli et al., 2016). In turn, the 5-HTT genotype could modulate the microbiota composition in rats (El Aidy et al., 2017). Moreover, the gene expression pattern of MS exposed rats resembled the pattern of rats with reduced 5-HTT gene expression, in the PFC and hippocampus, which indicates the prominent role of 5-HTT in stress response (Comasco et al., 2019). The evidence in present section indicates close relationship between the classical central 5-HT deficiency and OS, neuroinflammation, central glutamate and BDNF/TrkB dysregulation, and the MGB disturbance. Further exploring the connections between the classical and novel pathogenesis would definitely benefit the incoming studies, by providing deep insights of the intricate underlying mechanisms, and novel targets and fresh ideas for depression treatment.

The MGB Axis Disturbance

The MGB axis refers to the connection and mutual influence between the brain and gastrointestinal system, with intestinal microbiota composition (abundance and diversity) and metabolism as the key component. Microbiota affect brain function through tryptophan metabolism, vagus nerve, immune regulation, etc., while brain in turn influences microbiota through immune modulation, neuroendocrine regulation, etc. (Kundu et al., 2017). The close relationship between the MGB disturbance and depression has been extensively attended in recent years (McGuinness et al., 2022; Winter et al., 2018) (Figure 1F, Tables 1–2, MGB disturbance section). As supporting evidence, studies found disturbed microbiota composition and metabolism in depressed patients and animals, however, with both increased and reduced abundance and diversity been revealed across microbiota species/genera (mainly Firmicutes and Bacteroidetes) (Barandouzi et al., 2020; De Palma et al., 2015; Gao et al., 2022; Qu et al., 2017; Sanada et al., 2020). Clinical studies found reduced abundance of Bifidobacteria and Lactobacilli, but increased abundance of Bacteroidea, Proteus and Actinobact eria, in MDD patients (Barandouzi et al., 2020; Sanada et al., 2020; Table 1, MGB axis disturbance section). Rodent studies revealed significantly reduced abundance of Bacteroidales and Clostridiales in CSDS and social disruption treated mice (Bailey et al., 2011; Qu et al., 2017), and reduced abundance of Lactobacillus in chronic variable stress exposed rats (M. Yu et al., 2017) and CSDS susceptible mice (X. Zhu et al., 2023). In contrast, the abundance of bacteria in the genus Clostridium and Lactobacillus were increased in social disruption and MS treated mice, respectively (Bailey et al., 2011; De Palma et al., 2015) (Table 2, MGB axis disturbance section). The inconsistency could attribute to diversified microbiota species/genera and individual genetic diversity, influenced by variables of subject gender and age, stress subtype, dynamic microbiota changes across development, constant influence of daily diet, etc. On the other hand, microbiota disturbance, induced by antibiotic treatment or fecal microbiota transplantation (FMT) from CUMS exposed subjects to control recipients, resulted in depressive-live behaviors in rats and mice, respectively (Hoban et al., 2016; Kelly et al., 2016; Li et al., 2019). Furthermore, microbiota homeostasis restoration and depressive-live behavior amelioration have been achieved in mice, through antidepressant treatments (e.g., ketamine, fluoxetine and Chinese medicine formula Kai-Xin-San) (Cao et al., 2020; Lukic et al., 2019; Qu et al., 2017), probiotic administrations (Bifidobacterium, Lactobacillus rhamnosus and Lactococcus lactis) (Bravo et al., 2011; Gao et al., 2022; Yang et al., 2017), dechlorogenic sunflower seeds and tryptophan supplementation (Lu et al., 2022; Wang et al., 2022), and FMT from NLRP3 KO mice to CUMS treated mice (Y. Zhang et al., 2019). The above evidence indicates the vital role of microbiota disturbance in depression onset, and suggests microbiota homeostasis restoration as an appealing antidepressant approach.

Studies have found close interplay between microbiota and OS/mitochondria, which might play prominent roles in depression (Shandilya et al., 2022) (Figure 1AF, F-OS). Microbiota regulates ROS level through affecting the function and activities of mitochondria, short chain fatty acids (SCFAs), antioxidants and NOX, which are enriched in the intestinal tract (Aviello & Knaus, 2018; Ballard & Towarnicki, 2020; Jones & Neish, 2017). For example, lactic acid derived from Lactobacillus plantarum activated the intestinal NOX to promote ROS generation (Iatsenko et al., 2018), while molecules containing indole and SCFAs improved bacterial metabolism and reduced the ROS production, through restoring mitochondrial function and activating antioxidant activity, respectively (Ballard & Towarnicki, 2020). Furthermore, studies indicated gut microbiota as the vital mediator of OS, whose disturbance impaired mitochondria, reduced ATP production and increased ROS generation, partially through the metabolite N⁶-carboxymethyllysine in central microglia of mice (Brunt et al., 2019; Mossad et al., 2022). In addition, the probiotic Bifidobacterium infantis increased plasma tryptophan in rats, and tryptophan supplementation improved central mitochondrial metabolism and attenuated depressive-like behaviors in CUMS exposed mice, which results indicate critical involvement of tryptophan in microbiota affected mitochondrial function and depressive symptoms (Desbonnet et al., 2008; Wang et al., 2022). In turn, mitochondrial genotype and mutation significantly influenced microbiota composition through ROS regulation (Yardeni et al., 2019).

Studies have revealed mutual influence between microbiota and peripheral/gut and central/neuron inflammation, respectively. The interaction between microbiota and gut inflammation involves multiple molecular mechanisms, such as the microbiota-shaped T cell subsets and generated SCFAs, immune-related dendritic cells and immunoglobulin A, etc. (Belkaid & Hand, 2014; Hooper et al., 2012; Ratajczak et al., 2019). For microbiota-neuroinflammation interaction (Figure 1BF), microbiota disturbance leads to systemic inflammation, reduces SCFAs, and increases peripheral IL-6 and IFN-γ, which activate microglia and promote the enzymes of IDO/TDO/KMO to generate QUIN that damages BBB and elevates IL-6, IL-1β, and NLRP3 in the brain, finally resulting in neuroinflammation (Carlessi et al., 2021). On the other hand, probiotics have anti-inflammatory and antidepressant effects. Probiotic inulin reversed age-related neuroinflammation in mice, mainly including monocyte infiltration and microglial activation (Boehme et al., 2020). Bifidobacterium infantis significantly increased tryptophan concentrations, and reduced forced swim stress enhanced IL-6, IFN-γ and TNF-α levels in rat plasma (Desbonnet et al., 2008), while Bifidobacterium adolescentis and Parabacteroides reduced the CRS enhanced IDO activity and increased levels of IL-1β, TNF-α and NF-κB in mouse hippocampus (Deng et al., 2021; Y. Guo et al., 2019), which all reversed the rodent depressive-like behaviors. Additionally, studies suggested the antidepressant and anti-inflammatory effects of ketamine was mediated via its interaction with a variety of gut microbiota, such as the Lactobacillus, Actinobacteria, etc. (Getachew et al., 2018; N. Huang et al., 2019). In turn, neuroinflammation could disturb microbiota homeostasis, primarily through the NLRP3, pro-inflammatory cytokines and ROS, whose underlying mechanisms need to be further uncovered (Carlessi et al., 2021; Inserra et al., 2018). Furthermore, recent studies have revealed the crucial roles of microbiota-gut-immune-glia axis and glia-neuron communication in depression, which could potentially integrate the gut microbiota, immune signals, and central glia-neuron communication into a network, responsible for depression onset and remedy (Rudzki & Maes, 2020, 2021).

The MGB disturbance affects neurotransmitter homeostasis in the brain, especially inducing the BDNF/TrkB dysfunction and 5-HT deficiency (Figure 1DF, EF). For the MGB-BDNF interaction (Figure 1DF), studies found chronic stress (CUMS and CRS)-induced microbiota disturbance and depressive-like behaviors in mice, through reducing the hippocampal BDNF expression, which were ameliorated by the restored BDNF expression, caused by tryptophan supplementation and probiotic treatments (e.g., Parabacteroides and Bifidobacterium adolescentis) (Bercik et al., 2011; Ding et al., 2021; Wang et al., 2022; Y. Guo et al., 2019). The direct influence of MGB disturbance on TrkB expression and activation remains unclear. In turn, BDNF modulates the intestinal barrier integrity, through regulating tight junction protein expression and inhibiting intestinal epithelial cell apoptosis, whose dysfunction would impair the integrity and further disturbs the microbiota homeostasis (Y. B. Yu et al., 2017; Zhao et al., 2018). The MGB-BDNF/TrkB interaction might play more important roles in depression than we expected, however, more extensive and in-depth investigations are required to further validate their critical involvement, and reveal the detailed underlying mechanisms. With regard to the central 5-HT deficiency, considerable evidence has demonstrated its pivotal role in depression, as described in the above section (Moncrieff et al., 2022; Vahid-Ansari & Albert, 2021). Studies have found interactions between the 5-HT, gut microbiota and the MGB axis (Stasi et al., 2019) (Figure 1EF). Microbiota metabolites SCFAs increased the TPH1 mRNAs in vitro (Reigstad et al., 2015), while Bifidobacterium infantis reduced the concentration of 5-HIAA (the primary metabolite of 5-HT) in the PFC of rats (Desbonnet et al., 2008), and Lactobacillus rhamnosus increased the expression of 5-HT-related genes (e.g., TPH1a, TPH2, and slc6a4a) in the brain of zebrafish (Borrelli et al., 2016). For the involvement of the interaction in depression, studies found microbiota disturbance induces gut and central inflammation, activates the IDO/TDO enzymes and promotes peripheral and central KP, which reduces peripheral tryptophan available to cross the BBB and directly inhibits central 5-HT production, respectively. Those alternations were reversed by tryptophan supplementation and IDO/TDO inhibition, along with the depressive-like behaviors in rodents. As supporting evidence, CUMS damaged gut barrier integrity, promoted peripheral IDO activity and KP, reduced 5-HT in the serum, and induced depressive-like behaviors in mice, which were reversed by tryptophan supplementation and lentivirus-mediated IDO knockdown, respectively (Chen et al., 2023; Wang et al., 2022). In addition, CUMS and CRS-promoted neuroinflammation, reduced hippocampal 5-HT, and induced depressive-like behaviors in mice, which were reversed by the IDO inhibitor 1-methyl-tryptophan and tryptophan supplementation (Deng et al., 2021; Wang et al., 2022). The results indicate the vital role of tryptophan metabolism in the interplay between microbiota and peripheral/central inflammation. In turn, 5-HTT genotype modulated microbiota composition in rats (El Aidy et al., 2017). The evidence in the present section indicates the vital role of MGB disturbance in depression, which involves the mutual interaction with OS and mitochondrial impairment, neuroinflammation, central BDNF/TrkB dysregulation and 5-HT deficiency, respectively. The evidence also suggests MGB homeostasis restoration as a promising antidepressant avenue.

The HPA Axis Dysregulation

The HPA axis dysregulation is another classical depression hypothesis being extensively attended, and both clinical and animal studies have found depression,related HPA hyperactivity or hypoactivity, and the consequential hypercortisolism and hypocortisolism, respectively (Menke, 2019; Zhang, 2017). To simply describe the axis activity, the hypothalamus secretes corticotropin-releasing factor (CRF) that promotes the pituitary release of adrenocorticotropic hormone (ACTH), which activates the adrenal cortex release of glucocorticoids. Glucocorticoid receptors (GRs) are widely distributed in in the brain, especially depression-related subregions (e.g., PFC, hippocampus, and hypothalamus), whose activation is responsible for feedback inhibition of the CRH/ACTH release, and restoration of the HPA axis homeostasis. The dysfunction of GRs, especially the reduced expression and desensitization, would disrupt the feedback loop and result in hypercortisolism that is commonly and critically involved in depression (Menke, 2019).

The HPA axis interacts with OS directly, or indirectly through neuroinflammation, central glutamate and BDNF/TrkB signaling, and the MGB axis. For the HPA-OS interaction (Figure 1G-OS), studies found NO mediated the cholinergic stimulation of HPA axis, and the IL-1β-induced HPA response in social crowding stress exposed rats (Bugajski et al., 2006; Gadek-Michalska & Bugajski, 2005). In turn, cortisol-induced ROS generation through membrane GRs (Espinoza et al., 2017), which probably attributes to translocation of the activated GRs to mitochondria that results in enhanced calcium buffering capacity (Du et al., 2009; Psarra & Sekeris, 2009). Furthermore, mitochondrial GR activation could directly regulate mitochondrial energy metabolism and oxidative phosphorylation gene expression (Kokkinopoulou & Moutsatsou, 2021). With regard to the HPA-neuroinflammation interaction (Figure 1BG), studies have found activated GRs could translocate to nucleus, to directly inhibit pro-inflammatory cytokine gene expressions (Adcock et al., 2005; Cain & Cidlowski, 2017). The pro-inflammatory gene expression was increased in the GR depleted hippocampal microglia in mice reared under standard condition, but was reduced in the CUMS exposed mice, which suggests differential regulating effect of GRs under normal and stressful conditions (Picard et al., 2021). Another study further revealed the prominent role of interaction between GR and NF-κB in regulating hippocampal inflammatory responses, in the CUMS exposed rats (Li et al., 2021). Alternatively, GRs could indirectly take effect through OS, as GR agonist compound K regulated IL-6 and TNF-α expressions, through inhibiting the ROS generation (Cuong et al., 2009). In turn, cytokine affected GR function is closed related to glucocorticoid resistance in depression (Pace et al., 2007), and pro-inflammatory cytokines target at hypothalamus to promote glucocorticoid secretion, which in turn activates GRs that inhibit cytokine expressions, which loop links neuroinflammation to the axis activity (Cain & Cidlowski, 2017). For HPA-glutamate interaction (Figure 1CG), in vitro studies showed activated GRs interacted with TrkB to facilitate the BDNF-promoted glutamate release via PLC-γ signaling (Numakawa et al., 2009), while glucocorticoids attenuated the BDNF-dependent glutamate receptor upregulation (Kawashima et al., 2010). Moreover, chronic HPA axis disruption with corticosterone altered glutamate signaling-related gene expressions in mice, in a way that corticosterone increased mRNA expression of ionotropic glutamate receptor subunits GluR1 and GluR2 in the mPFC, reduced mRNA expression of the NMDA receptor subunits (NR1, NR2B) and mGluR5 in the hippocampus, and decreased mRNA expression the AMPA receptor subunit GluR1 in the amygdala (Kinlein et al., 2022). In turn, glutamate signaling plays a complex role in affecting the HPA responses, probably through driving acute HPA response via the ionotropic receptors postsynaptically, and inhibiting CRH neurons via group I mGluR presynaptically (Evanson & Herman, 2015b). For instance, paraventricular nucleus injection of group I mGluR agonist blunted, whereas the antagonist potentiated, the HPA response to restraint stress in rats (Evanson & Herman, 2015a). For the HPA-BDNF interaction (Figure 1DG), studies showed that GR activation down-regulated the BDNF expression in vitro (Chen et al., 2019; H. Chen et al., 2017), while BDNF polymorphism disrupted the GR phosphorylation, through which central BDNF signaling rewrote the glucocorticoid transcriptome (Arango-Lievano et al., 2019; Lambert et al., 2013). The critical involvement of HPA-MGB interaction in depression has been previously indicated (Freimer et al., 2022) (Figure 1FG). For example, stress activated HPA axis and consequentially generated hormones (e.g., glucocorticoids and catecholamines) increased the intestinal barrier permeability (Zheng et al., 2013) and disturbed microbiota composition (Moussaoui et al., 2017). In turn, microbiota homeostasis restoration with the probiotic L. farciminis attenuated the LPS-induced gut leakiness, HPA response and neuroinflammation in rats (Ait-Belgnaoui et al., 2012). Moreover, Lactobacillus rhamnosus restores central GABAergic activity to attenuate the HPA hyperactivity and exert antidepressant effect (Tette et al., 2022). The evidence in the present section indicates close association between the classical HPA axis dysregulation and novel depression pathogenesis, including OS, neuroinflammation, central glutamate and BDNF/TrkB dysregulation, and the MGB disturbance. The integration of the classical and novel pathogenesis surely would provide us with an integral overview, and supply us with more in-depth insights, of the intricate pathogenesis of the disease.

Concluding Remarks

Depression increasingly and seriously affects the mental and physical health of people worldwide. To uncover the intricate disease pathogenesis, several animal models (especially in rodents) have been established, through acute and chronic stress exposures, olfactory bulbectomy, drug administration (e.g., LPS and corticosterone), genetic manipulation, etc. (Czéh et al., 2016; Zhang, 2017). Through years of clinical and animal investigations, a variety of disease hypotheses have been proposed, such as the classical central 5-HT deficiency (Moncrieff et al., 2022) and HPA axis dysregulation (Menke, 2019), and the relative novel neuroinflammation (Troubat et al., 2021), glutamate excitotoxicity (Murrough et al., 2017), BDNF dysfunction (Castren & Monteggia, 2021) and MGB axis disturbance (Winter et al., 2018). With regard to the existing first-line antidepressants, over a third of MDD patients do not have proper response, referred as treatment-resistant depression (TRD). Classical antidepressants usually target on the monoamine system and elevate their synaptic concentration in the brain, such as the TCAs, MAOIs and SSRIs, which generally require at least two weeks of continuous treatment, and moreover, have a variety of side effects (Cipriani et al., 2018; Malhi & Mann, 2018). However, recent studies demonstrated the possibility of designing fast-onset monoamine-based antidepressants, by dissociating 5-HTT from neuronal NOS in the DRN of mice (Guan & Pang, 2023; Kingwell, 2022; Sun et al., 2022; Ye et al., 2023). Another study further showed Propofol rapidly relieved the core depressive symptom anhedonia in mice, through inhibiting dopamine transporter (but not the 5-HTT) and evoking long-lasting dopamine accumulation in the NAc (X. N. Zhu et al., 2023). There exist several other antidepressants targeting beyond the monoamine signaling, among which ketamine could rapidly and robustly alleviate depression (including TRD) with subanaesthetic doses, mainly through modulating central glutamatergic system, whose efficiency, however, has been shadowed by the addictive and psychotomimetic side effects (Berton & Nestler, 2006; Riggs & Gould, 2021). Thus, exploring novel pathogenesis of depression is necessary, for seeking more safe and efficient therapeutic treatments. In recent years, emerging evidence has revealed the vital involvement of OS in depression, however, the detailed underlying mechanisms remain unclarified. Considering the natural advantages of antioxidant treatments, such as antioxidant diet and exercise, which could avoid the side effect of drugs, it is appealing to pursue the antioxidant avenue of depression prevention and treatment. Fully uncovering how OS induces depression is the first step towards the avenue, thus we summarize and elaborate the possible downstream pathways mediating the OS influence, from the perspectives of mitochondrial impairment and related ATP deficiency, neuroinflammation, glutamate excitotoxicity, BDNF/TrkB dysfunction and 5-HT deficiency in the brain, the MGB disturbance and HPA dysregulation. We also expound the intricate interactions between the multiple aspects, and the molecular mechanisms mediating the interplay. Despite of what we have learned so far, more in-depth investigations are required to fully depict a comprehensive overview of how OS induces depression, in order to further provide fresh ideas and novel targets for the final goal of effective prevention and treatment of the disease.