The Microbiota–Gut–Brain Axis in Depression: Mechanisms, Microbiota‐Targeted Interventions, and Translational Challenges

The MGB axis allows the gut microbiota to modulate CNS function via interconnected signaling networks. In the neural pathway, the vagus nerve transmits intestinal signals to brain regions; vagotomy reduces probiotic efficacy for social deficits and anxiety [1 –5], highlighting its role. The enteric nervous system (ENS) acts as the "second brain," managing intestinal homeostasis; dysfunction causes permeability and neuroinflammation, impacting neuroplasticity [1]. In the immune pathway, two key barriers maintain homeostasis: the blood–brain barrier (BBB) and intestinal epithelial barrier (IEB). Disruption of the IEB allows harmful substances (e.g., lipopolysaccharide (LPS)) to enter the bloodstream. This increases BBB permeability and triggers proinflammatory cytokines, which contribute to depressive phenotypes [6]. Gut microbiota orchestrate immunity, with GF animals showing impairment [7, 8]. Stress or dysbiosis compromises barriers, activating TLR4/NF‐κB pathways and promoting inflammation [9]. For the endocrine pathway, probiotics enhance behavior and regulate the HPA axis [10, 11]. Chronic stress initiates cycles of intestinal barrier leakage (i.e., increased permeability of the IEB, allowing LPS and bacterial metabolites to translocate into the systemic circulation) and neuroinflammation [12], while dysbiosis induces anxiety and HPA hyperactivity [13 –15]; microbial dysregulation activates the HPA via IL‐1β and LPS [16]. Regarding metabolites, short‐chain fatty acids (SCFAs) strengthen BBB integrity and engage microglia [17]. Microbiota regulate serotonin production and synthesize GABA, reducing anxiety [18 –21] (Table 1 and Figure 1).

The human gut microbiota primarily comprises Firmicutes and Bacteroidetes [22], with others like Proteobacteria. Extrinsic factors, such as diet, environmental pollutants, medication exposure, and lifestyle habits, can profoundly disturb the composition and functional balance of gut microbiota. These disturbances often lead to immediate consequences including intestinal dysbiosis, compromised IEB function, and altered microbial metabolite synthesis—all of which contribute to the pathological processes underlying depression. Restoring eubiosis, or a stable and balanced gut microbial ecosystem, has therefore emerged as a critical target for mitigating microbiota‐associated depressive phenotypes [23]. In major depressive disorder (MDD) patients, microbiota shows abnormalities in composition, diversity, and function, linked to neuroinflammation, metabolic disorders, and behavioral abnormalities. MDD patients exhibit reduced microbiota diversity, with decreased Firmicutes abundance and fewer butyrate‐producing genera like Faecalibacterium, weakening SCFA production and anti‐inflammatory functions [24]. Conversely, Bacteroidetes abundance increases, activating microglia and promoting systemic inflammation, correlating with disease severity [25]. Probiotics like Lactobacillus and Bifidobacterium are reduced, associated with lower BDNF and impaired serotonin synthesis, while proinflammatory bacteria such as Enterobacteriaceae are elevated [26, 27]. In contrast, proinflammatory bacteria such as Enterobacteriaceae are elevated [26]. Fecal calprotectin (a marker of intestinal inflammation) is decreased in emotion‐regulating brain regions like the superior frontal gyrus—contrasting with healthy individuals [27]. Metabolic dysfunction is prevalent, with reduced SCFAs damaging the gut barrier, increasing permeability and allowing LPS entry, causing inflammation and depression‐like behavior [27]. Dysbiosis promotes gram‐negative bacteria growth, releasing LPS and triggering inflammatory storms that harm hippocampal synapses [28].

Probiotics enhance verbal episodic memory in depressed patients by increasing intestinal Lactobacillus and regulating hippocampal activation [30, 31]. A multicenter, double‐blind RCT by Nikolova et al. (n = 240) showed that adjunctive probiotics (L. rhamnosus + B. longum) improved HAM‐D scores in 52% of MDD patients, with low dropout rates (12%)—supporting clinical feasibility [30]. However, a smaller trial (n = 60) found no significant effects, likely due to unstratified enrollment (mixing high/low inflammation subgroups) and short duration (4 vs. 8 weeks) [31]. Akkermansia muciniphila, a next‐generation probiotic, reduces depressive‐like behaviors in mice by thickening the intestinal mucus layer and modulating SCFAs, with anti‐inflammatory and neuroprotective effects. Its modified outer membrane protein Amuc_1100Δ80 interacts with intestinal epithelial TLR2 to increase intestinal 5‐HT, alleviating chronic unpredictable mild stress (CUMS)–induced anxiety and depression more potently than the original protein [32]. In a mouse alcohol–LPS model, A. muciniphila improved alcohol‐related depression by strengthening the intestinal barrier, reducing serum LPS, mitigating neuroinflammation (lower TNF‐α/IL‐1β), normalizing depression‐related gene expression, and increasing hippocampal 5‐HT [33], supporting its potential to alleviate depressive symptoms. Precision probiotic interventions target specific populations: A randomized trial showed L. reuteri PBS072 and B. breve BB077 significantly improved postpartum depression (EPDS scores) by enhancing stress recovery [34].

Engineered Escherichia coli Nissle 1917 secreting BDNF reversed depression‐like behaviors in animals [35]. These strains use "smart" circuits—for example, E. coli engineered with a tryptophan‐inducible promoter produces 5‐HTP only in the gut (where tryptophan is abundant), avoiding off‐target neurochemical effects [36]. AI‐driven personalized approaches use wearable devices to monitor SCFAs and guide probiotic dose adjustments [37]. Postbiotics (inanimate microbial components/metabolites) offer new options: Heat‐killed Lactobacillus plantaris‐derived postbiotics, via metabolites like SCFAs and bile acids, protect against Salmonella‐induced depression by modulating the gut–brain axis [38].

However, probiotic efficacy varies due to baseline microbiota differences (e.g., patients with low Faecalibacterium respond better [39]). Engineered strains carry bacteremia risks in immunocompromised hosts, requiring strict preclinical safety testing (e.g., assessing bacterial translocation in immunodeficient mice [36]).

High‐fiber diets relieve neuroinflammation by boosting SCFA production via microbial fermentation. SCFAs regulate microglia, inhibit hippocampal NF‐κB, and reduce proinflammatory cytokines, improving chronic stress‐induced depression in animals [40]. Unlike medications linked to metabolic syndrome, they enhance insulin sensitivity and reduce obesity‐related depression risk via SCFAs like propionate [31]. The PREDITAP trial showed a Mediterranean diet (rich in polyphenols/Omega‐3s) reduced depression risk in adolescents and adults, with HAM‐D scores decreasing by 2.15 (4 weeks) and 2.42 (8 weeks, p < 0.05), linked to increased microbiota diversity [40]. Polyphenols (e.g., resveratrol) strengthen the intestinal barrier and inhibit systemic inflammation (20% lower C‐reactive protein (CRP)), differing from traditional calorie‐focused diets [40]. However, cross‐sectional studies report inconsistent associations, attributed to self‐reported diet recall bias [41]. Ketogenic diets (high‐fat, low‐carb) improve depression/anxiety in animal models but face clinical challenges: small, uncontrolled studies, short‐term focus, high dropout rates, and methodological heterogeneity [42]. Anti‐inflammatory diets (low processed foods, high fruits/vegetables/whole grains) reduce depression scores in patients with elevated IL‐6/TNF‐α. Curcumin and flavonoids (e.g., blueberry anthocyanins) inhibit microglial activation, making them potential options for inflamed subgroups [43].

However, personalized dietary plans consider gene‐nutrition interactions: 5‐HTTLPR s‐allele carriers are sensitive to high‐sugar diets, with low‐carb diets reducing depression risk [44]; FTO gene mutants benefit more from low‐GI diets [45]. Strict diets like ketogenic regimens have poor long‐term adherence, requiring adaptable modifications. Nonresponders need microbiome/metabolome stratification.

Fecal microbiota transplantation (FMT) regulates gut flora to treat depression. Transplanting healthy donor microbiota into depressed mice reversed behavioral abnormalities—reducing forced swimming immobility by 50% [46]. It also increased levels of the anti‐inflammatory bacterium Faecalibacterium prausnitzii. These effects are mediated by vagal signaling and HPA axis normalization [46]. Preliminary trials show FMT improves gastrointestinal symptoms and quality of life in treatment‐resistant depression (TRD), but efficacy varies: one case report noted HAM‐D reduction from 24 to 14 (8‐week follow‐up), but small sample size (n = 15) and lack of long‐term data limit generalizability [47]. FMT combined with SSRIs accelerates remission (2 weeks earlier than monotherapy) by restoring microbiota–tryptophan metabolism for serotonin synthesis [46].

However, FMT faces microbiota "reversion" (60% of patients return to baseline flora by 12 months), increasing metabolic syndrome risk (OR = 1.8) [48]. Donor selection, transplantation routes (oral vs. colonoscopy), and ethics limit clinical use [48]. FMT′s potential requires addressing donor screening standardization, long‐term safety, and individual heterogeneity. Future directions include synthetic biology‐modified microbiota and multiomics‐guided personalization.

Anti‐inflammatory antibiotics like minocycline and doxycycline show promise as adjuncts for depression, with neuroprotective and flora‐regulating properties. Some SSRIs (e.g., fluoxetine) also exhibit antibiotic activity. Notably, antibiotics do not exert uniform effects on gut microbiota; their impact varies significantly based on factors such as antibiotic class, administration dose, treatment duration, and the initial composition of the gut microbial community. While certain antibiotics may reduce the abundance of pro‐inflammatory bacteria, others can disrupt beneficial microbial populations, potentially exacerbating gut dysbiosis if not used strategically [49]. A meta‐analysis confirmed minocycline′s benefit in unipolar depression (especially psychotic subtypes) but not bipolar disorder [50]. An RCT found depressed patients with low‐grade inflammation (CRP ≥ 3 mg/L) had a 2.5‐point greater HAM‐D reduction with 4‐week minocycline versus placebo (p = 0.03) [51]. Doxycycline exerts neuroprotective, anti‐inflammatory, and antidepressant effects, restoring LPS‐induced behavioral/neuroinflammatory responses in mice by inhibiting stress‐induced NO synthesis and iNOS [52]. Selective antibiotics like rifaximin (nonsystemic, gut‐targeted) reduce inflammation/oxidative stress and improve depression by inhibiting ammonia‐producing bacteria (e.g., Klebsiella). It modulates gut flora and tryptophan metabolism to alleviate CUMS‐induced depression in rats [53].

However, broad‐spectrum antibiotics (e.g., against Gram‐negative bacteria) reduce flora diversity and risk resistance [54]. Short‐term (≤ 4 weeks) use with probiotics (e.g., Bifidobacterium) restores 90% flora diversity and reduces drug‐resistant bacteria, limiting use to infection‐associated depression subtypes with concurrent probiotic support [55]. Long‐term broad‐spectrum use causes dysbiosis, resistance, opportunistic infections (e.g., C. difficile), and metabolic disorders. Their use requires strict limitation to inflammatory subtypes, with flora monitoring.

Phages regulate gut flora homeostasis with strain‐specific targeting of pathogens, sparing beneficial bacteria. Metagenomics show MDD patients have imbalanced phage communities: reduced Clostridium bacteriophage phi8074‐B1 and Klebsiella bacteriophage vB‐KpnP‐SU552A (50%–70%), and increased Escherichia bacteriophage ECBP5 (2.3×) [56], disrupting phage–host symbiosis to promote pathogenic overgrowth, intestinal permeability, and neuroinflammation.

In chronic restraint stress‐induced depressed mice, Enterobacteriaceae, Gammaproteobacteria, and Campylobacteraceae phages are enriched, targeting proinflammatory Enterobacteriaceae [57]. Tryptophan metabolites (tryptamine and 5‐methoxytryptamine) correlate with Microviridae and Podoviridae phages, suggesting phages influence serotonin synthesis via bacterial tryptophan metabolism enzymes [57].

Phages modulate the MGB axis via dual pathways: (1) lysing proinflammatory Enterobacteriaceae to reduce LPS‐induced neuroinflammation and (2) altering bacterial tryptophan metabolism—for example, increasing Bifidobacterium abundance (via Clostridium phages) to boost 5‐HT synthesis [57]. Engineered phages could carry "therapeutic genes" (e.g., SCFA synthase) to restore microbiota function—addressing the limitation of natural phages (which only lyse pathogens, not restore beneficial metabolites).

No human trials for depression exist, but phage therapy for gut infections (e.g., C. difficile) has shown safety in Phase I trials (n = 50)—supporting translatability [56]. An upcoming pilot trial (NCT05876321) will test Enterobacteriaceae phages in 20 MDD patients with high LPS levels. Long‐term preclinical studies (6 months in mice) show no adverse effects (e.g., phage‐induced inflammation), but risks include phage resistance (pathogens developing CRISPR‐based defenses) and unintended lysis of beneficial bacteria [57].

Machine learning integrates metagenomics and clinical phenotypes to model gut microbiota–depression links. MDD patients show reduced ratios of neuroprotective butyrate‐producing bacteria to proinflammatory bacteria, a potential typing marker [58]. Deep learning links glutamate synthase (gltB/gltD) and melatonin metabolism genes to F. prausnitzii abundance, providing a basis for microbiome‐based diagnostics. Machine learning models integrating multi‐omics data (metagenomics + metabolomics + neuroimaging) outperform clinical markers alone. For example, a model using the Coprococcus/Dialister ratio predicted probiotic response with AUC = 0.82 [59]. Wearable biosensors and AI enable real‐time therapeutic adjustments: A proof‐of‐concept study used smart capsules to monitor intestinal SCFAs, with reinforcement learning adjusting probiotic doses [37]. If SCFA levels drop below 5 μmol/L (a threshold linked to symptom relapse), the AI system recommends a 20% increase in probiotic dose. The AI‐optimized group showed 17% greater HAM‐D reduction and more stable flora diversity than fixed‐dose groups [37].

However, small sample sizes (n = 30) and lack of validation in diverse populations (e.g., elderly and comorbid patients) hinder widespread use [37]. Long‐term data on algorithm accuracy is missing. Future multiomics studies should verify microbial metabolic pathway activity and cerebrospinal fluid neurotransmitter dynamics to optimize subtype identification.

Synthetic biology transforms probiotics into "living pharmaceutical factories." E. coli Nissle 1917 engineered with the tryptophan hydroxylase gene converts dietary tryptophan to 5‐HTP, increasing brain 5‐HT and reducing forced swimming immobility in CUMS mice [36]. Similarly, L. lactis engineered to secrete BDNF enhances synaptic plasticity via TrkB receptor activation—reducing immobility time by 40% in CUMS mice, vs. 25% for unmodified L. lactis [35].

Engineered strains use attenuated vectors (deletion of virulence genes) and inducible expression systems to limit off‐target effects. However, they carry risks of horizontal gene transfer (e.g., antibiotic resistance genes) in immunocompromised patients [36]. Regulatory approval as "live biotherapeutic products" (LBPs) requires long‐term safety data (e.g., 5‐year follow‐up for gene stability).

To guide clinical application and future research, we comparatively rank the above interventions based on their current evidence base and near‐term clinical potential (Table 2).