Treatment effects on sphingosine-1-phosphate and its receptors in major depressive disorder: implications for biomarkers Authors: Xiao Yang, Le Wang, Anlong Jiang, Zeyue Mu, Jiamin Chen, Manxue Hu, Feng Zhang, Xueqian Xu, Ke Zhao, Fang Liu Published: 2026 ABSTRACT BackgroundThe diagnosis of major depressive disorder (MDD) currently relies on subjective clinical assessments, highlighting a critical need for objective biological markers. The sphingosine-1-phosphate (S1P) signaling pathway, a pivotal regulator of neuro-immune interactions, has emerged as a potential contributor to MDD pathophysiology, yet its role remains incompletely understood. This study aimed to investigate plasma levels of S1P and its key receptors, S1PR1 and S1PR3, as potential diagnostic and predictive biomarkers for MDD, with a specific focus on sex differences and treatment effects.MethodsPatients with MDD (n = 56) underwent an 8-week treatment protocol were enrolled, alongside 42 healthy controls (HCs). Depression severity was evaluated using the Hamilton Depression Rating Scale (HAMD-24) and Patient Health Questionnaire (PHQ-9) at baseline and 8-week visit. The plasma levels of S1P, S1PR1 and S1PR3 were measured at baseline and week 8.ResultsAt baseline, plasma concentrations of S1P, S1PR1, and S1PR3 were significantly elevated in patients with MDD compared to HCs. All three markers significantly decreased and trended toward normal levels after 8 weeks of antidepressant treatment. Notably, a significant sex-specific difference was observed for the receptors, baseline elevations of S1PR1 and S1PR3 were more obvious in female patients than in male patients. Furthermore, baseline S1P levels significantly predicted symptom improvement—as measured by changes in both HAMD-24 and PHQ-9 scores—whereas baseline S1PR levels alone did not. In addition, a combined panel of S1P, S1PR1, and S1PR3 yielded high diagnostic accuracy, with an area under the receiver operating characteristic curve (AUC) of 0.9575.ConclusionOur data demonstrate a significant, sex-dependent dysregulation of the peripheral S1P signaling pathway in MDD, which is responsive to antidepressant treatment. The ability of baseline S1P to predict clinical outcomes and the encouraging diagnostic precision of the S1P-S1PR1-S1PR3 panel strongly support their potential as clinically applicable biomarkers. These results imply that the S1P pathway plays a crucial role in the pathophysiology of depression, offering new possibilities for diagnosis and personalized treatment strategies in psychiatry.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12888-026-08367-5. Background Major depressive disorder (MDD) is a common and serious mental disorder. In recent years, its global prevalence has continued to rise, becoming an important public health concern that significantly compromises patients’ quality of life [1, 2]. However, the diagnosis of MDD remains based on subjective clinical interviews and symptom-based criteria, as outlined in classifications such as the Diagnostic and Statistical Manual of Mental Disorders (DSM) [3]. This diagnostic strategy is limited by heterogeneity of the symptoms, high comorbidity rates, and a dependence on patient self-reporting, which can delay diagnosis and treatment [4]. The lack of objective and quantifiable biomarkers not only complicates clinical practice but also hinders the development of novel, targeted therapeutics [5]. Seeking trustworthy biomarkers in psychiatry has been challenging, primarily due to the complexity of the brain and the limited ability to sample tissue from living individuals directly [3]. Thus, the discovery of reliable biomarkers in peripheral tissues, particularly in blood, is now a main goal in biological psychiatry. Using blood-based markers is a non-invasive, cost-effective, and scalable approach to identifying systemic biological disturbances that could mirror or impact central nervous system (CNS) pathology [6, 7]. Recently, MDD has been increasingly studied by using peripheral blood from patients [8], and some molecular mediators may serve as potential biomarkers [9, 10]. Moreover, increasing evidence has shown that panels combining multiple biomarkers often provide better diagnostic and predictive accuracy compared to single analytes, as they could address the complexity of disorders like MDD [11]. For example, four metabolites (e.g., GABA, dopamine, tyramine, kynurenine) were used to create a panel specific to depression [12, 13]. Recent advances in lipid omics support that sphingolipids-including ceramides, sphingomyelins, and their metabolites-could play important roles in depression pathophysiology and biomarker development [14, 15]. Sphingosine-1-phosphate (S1P) is an essential bioactive signaling sphingolipid released by activated platelets, red blood cells, and endothelial cells in the peripheral blood [16]. S1P mediates various biological processes, including cell proliferation, apoptosis, migration, and regulation of inflammatory responses, through five G-protein-coupled S1P receptors (S1PR1–5) [17]. The S1P pathway is uniquely positioned at the crossroads of the immune, vascular, and nervous systems. In the periphery, S1P is a key regulator of lymphocyte trafficking, controlling the exit of T and B cells from secondary lymphoid organs into circulation [18]. Fingolimod and other S1P receptor modulators designed to harness this mechanism are currently under active clinical investigation for the treatment of multiple sclerosis (MS) [19]. In the brain, S1P signaling is crucial for the integrity of the blood-brain barrier (BBB), the highly selective border that protects the CNS [20]. Apolipoprotein M (ApoM)-bound S1P, acting specifically through S1PR1 on brain endothelial cells, is essential for maintaining tight junction integrity and low BBB permeability [21]. Given that endothelial dysfunction and BBB disruption have been reported in MDD [22], alterations in peripheral S1P concentrations might directly disrupt BBB function and facilitate the pathogenic crosstalk between systemic and central inflammation [20]. This provides a compelling mechanistic rationale for investigating peripheral S1P pathway components as indicators of CNS-related pathology in MDD. While S1P receptors are widely expressed in the CNS on neurons, astrocytes, microglia, and oligodendrocytes, S1PR1 and S1PR3 are of particular interest for their unique and occasionally opposing roles in neuroinflammation and neuronal function [21]. S1PR1 is the most abundant S1P receptor in the brain and is highly expressed on astrocytes and microglia [23]. Although it is crucial for developmental processes, astrocyte-specific deletion of S1PR1 has been shown to be protective in animal models of MS and neuropathic pain [24, 25], suggesting that astrocytic S1PR1 signaling contributes to neuroinflammation. S1PR3 plays a role that is more complex and dependent on the situation. In vitro studies have shown that S1PR3 activation on astrocytes can mediate pro-inflammatory signaling via the RhoA pathway, promoting the transcription of inflammatory genes [26]. However, in vivo evidence suggests a potentially protective effect on stress and depression. Research has shown that S1PR3 signaling in the medial prefrontal cortex (mPFC) promotes stress resilience by attenuating stress-induced increases in TNF-α [27]. Furthermore, S1PR3 is involved in synaptic plasticity, a crucial mechanism for learning and memory that is disrupted in patients with MDD. The SphK/S1P signaling pathway is essential for hippocampal long-term potentiation (LTP) [28], and recent findings demonstrated that selectively increasing S1PR3 expression in hippocampal neurons can improve synaptic plasticity and alleviate depressive-like behaviours in animal models [28, 29]. Although S1P signaling has become more recognized for its importance in fundamental CNS functions related to depression, clinical research on this pathway in MDD is still limited and inconsistent. Animal studies have reported both increases in S1P in specific brain regions associated with stress resilience and decreases in brain S1P associated with depressive-like behavior [30]. A key postmortem study in humans found that S1P levels in the corpus callosum of MDD brains remained unchanged, whereas a significant reduction of S1P was observed in schizophrenia brains [31]. Together, these discrepancies highlight an important need for reliable data from living MDD patients to elucidate thetrue state of the peripheral S1P pathway in clinical depression. This study aims to address the changes of S1P, S1PR1, and S1PR3 in the plasma of MDD patients. The purposes of our current study included: (i) to test the hypothesis that plasma levels of S1P and its receptors are altered in acute phase MDD patients compared to HCs and could be modulated by antidepressant treatment; (ii) to investigate potential sex-specific differences in this pathway, given the known sex disparity in MDD prevalence; (iii) to determine the value of these molecules as predictors of clinical treatment response; and (iv) to assess their potential as diagnostic biomarkers for MDD. Methods Study population The study protocol was approved by the local ethics committee of Lishui Second People’s Hospital (Approval NO. 2023-028). Written informed consent was obtained from all participants following the Declaration of Helsinki. Fifty-six patients with MDD were recruited from Lishui Second People’s Hospital, and 42 HCs from the local community between July 2024 and July 2025. All patients met the diagnostic criteria of MDD based on the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV), by a licensed psychiatrist. This study included participants aged 16–60 and with ≥ 9 years of education. Given that peripheral expressions of S1P, S1PR1, and S1PR3 are known to be altered in some other diseases [32–35], subjects with other neuropsychiatric disorders were excluded from this study. All the MDD patients completed the assessments of depressive symptoms, and no previous neuropsychiatric disorders or other medical conditions were documented in HCs. Clinical data collection Patients filled out the Patient Health Questionnaire (PHQ-9) [36], and research psychiatrists administered the HAMD-24 [37]. Depression severity was evaluated at baseline and at the 8-week follow-up period. The clinical outcomes were the changes in HAMD-24 and PHQ-9 from baseline to the follow-up at week 8. At the same time, researchers recorded the treatment modalities used for the patients, including 32 patients receiving antidepressants (Escitalopram, a Selective Serotonin Reuptake Inhibitors) and 24 patients receiving a combination of antidepressants (Escitalopram) and repetitive transcranial magnetic stimulation (rTMS). The specific doses of escitalopram were not standardized as part of this naturalistic treatment study; clinicians prescribed medications based on individual patient needs following standard clinical practice. Blood sample collection Peripheral blood samples were collected by venous puncture. Samples from MDD patients were taken at two time points, at baseline and at 8-week follow-up, whereas samples from HCs were obtained only at baseline. Samples were collected in EDTA tubes and centrifuged at 3600 rpm for 10 min at room temperature. The plasma was immediately extracted, aliquoted, and stored at − 80◦C for subsequent analyses. Plasma S1P, S1PR1 and S1PR3 concentrations detection Total S1P, S1PR1, and S1PR3 were measured using commercially available high-sensitivity enzyme-linked immunosorbent assay (ELISA) kits from Fengbin (Wuhan, China). All assays were performed by a single technician blinded to the clinical identities of the samples, following the manufacturer’s protocols. All ELISA kits were used following the manufacturer’s instructions. The reference standard was used on each ELISA plate to make a standard curve. For S1P (Fengbin, YG-901478H), six standards of different concentrations from 2400 nmol/L to 0 nmol/L (2400, 1200, 600, 300, 150, and 0 nmol/L) were prepared. For S1PR1 (Fengbin, YG-354771H), six standards of different concentrations from 8 µg/L to 0 µg/L (8, 4, 2, 1, 0.5, and 0 µg/L) were prepared. And for S1PR3 (Fengbin, YG-354773H), six standards of different concentrations from 8 µg/L to 0 µg/L (8, 4, 2, 1, 0.5, and 0 µg/L) were prepared. The detection limit for the S1P assay was 2400 nmol/L, whereas the detection limit for the S1PR1 and S1PR3 assay was 8 µg/L. In addition, sample collection and storage time did not differ between groups. The precision of all assays was validated. The intra-assay and inter-assay coefficients of variation (CV) for the S1P ELISA were 3.6% (range: 1.3–5.2%) and 7.2% (range: 4.6–9.5%), respectively. For the S1PR1 ELISA, the intra-assay and inter-assay CVs were 2.6% (range: 1.56–4.25%) and 8.1% (range: 5.6–10.2%). For the S1PR3 ELISA, the intra-assay and inter-assay CVs were 4.2% (range: 2.84–6.12%) and 6.3% (range: 3.7–8.9%). Statistical analysis Demographic categorical variables were analyzed using the chi-squared test. Continuous variables were presented as means with Standard Deviation (SD). The Shapiro-Wilk test was applied to assess the underlying distribution of all continuous variables. Normally distributed data were analyzed by an independent samples t-test or a paired-samples t-test, and data that deviated from a normal distribution were evaluated using the non-parametric Wilcoxon rank-sum test (for independent samples) or the Wilcoxon signed-rank test (for paired samples). Then, general linear models (GLM) were constructed to determine whether baseline levels of S1P, S1PR1, or S1PR3 (independent predictor variables) could predict changes in clinical depression outcomes after 8 weeks of treatment (dependent outcome variables), adjusted for age, sex and treatment as covariates. In addition, the receiver operating characteristic (ROC) curves were used to evaluate the potential diagnostic effectiveness of S1P, S1PR1, and S1PR3. An area under the receiver operating characteristic curve (AUC) value < 0.7 indicates low diagnostic accuracy, 0.7–0.9 indicates moderate accuracy, and > 0.9 indicates high accuracy. Finally, the multiple logistic regression was used to analyze the diagnostic power of these three molecules combined, controlling for the effects of gender and age. Before conducting the regression equation, the variance inflation factors (VIFs) were calculated to screen for potential multicollinearity among the predictors. In this study, all VIFs were less than 5, indicating that there were no significant multicollinearity among the independent variables S1P, S1PR1, and S1PR3. Statistical analyses were performed using the SPSS (version 27.0) and GraphPad Prism (version 10.0) software. A p-value < 0.05 was considered statistically significant. Results Sociodemographic and clinical information Table 1 displayed the demographic information and clinical characteristics for all individuals, which included 42 HCs and 56 MDD patients. All 56 patients completed the 8-week follow-up and blood sample collection. The MDD and HC cohorts were well-matched for age and sex, and there was no significant difference observed regarding age (p = 0.6714) and sex (p = 0.798). In MDD patients, HAMD-24 and PHQ-9 scores were significantly reduced at the endpoint compared to the baseline (all p < 0.001). Table 1Demographic and clinical characteristics of participantsIndexHCs (N = 42)MDD (N = 56)P-valueBaselineBaseline (W0)Week 8 (W8)HCs vs. MDDW0 vs. W8 Basic information Gender (F/M)29/1340/160.798aAge27.48 (9.376)28 (10.21)0.6714bHAMD-24NA31.79 (6.401)9.518 (4.984)< 0.001cPHQ-9NA20 (4.892)6.875 (3.908)< 0.001cValues are presented as mean (SD). HCs, healthy controls; MDD, major depressive disorder; HAMD-24, 24-Item Hamilton Depression Rating Scale; PHQ-9, Patient Health Questionnaire; S1P, sphingosine-1-phosphate; S1PR1, sphingosine-1-phosphate receptor 1; S1PR3, sphingosine-1-phosphate receptor 3; F, female; M, malea = Chi-square test; b = Wilcoxon rank-sum test; c = Wilcoxon rank-sum test Table 1Demographic and clinical characteristics of participantsIndexHCs (N = 42)MDD (N = 56)P-valueBaselineBaseline (W0)Week 8 (W8)HCs vs. MDDW0 vs. W8 Basic information Gender (F/M)29/1340/160.798aAge27.48 (9.376)28 (10.21)0.6714bHAMD-24NA31.79 (6.401)9.518 (4.984)< 0.001cPHQ-9NA20 (4.892)6.875 (3.908)< 0.001cValues are presented as mean (SD). HCs, healthy controls; MDD, major depressive disorder; HAMD-24, 24-Item Hamilton Depression Rating Scale; PHQ-9, Patient Health Questionnaire; S1P, sphingosine-1-phosphate; S1PR1, sphingosine-1-phosphate receptor 1; S1PR3, sphingosine-1-phosphate receptor 3; F, female; M, malea = Chi-square test; b = Wilcoxon rank-sum test; c = Wilcoxon rank-sum test Demographic and clinical characteristics of participants Values are presented as mean (SD). HCs, healthy controls; MDD, major depressive disorder; HAMD-24, 24-Item Hamilton Depression Rating Scale; PHQ-9, Patient Health Questionnaire; S1P, sphingosine-1-phosphate; S1PR1, sphingosine-1-phosphate receptor 1; S1PR3, sphingosine-1-phosphate receptor 3; F, female; M, male a = Chi-square test; b = Wilcoxon rank-sum test; c = Wilcoxon rank-sum test Comparison of baseline S1P, S1PR1 and S1PR3 levels between MDD patients and HCs As shown in Fig. 1A and Table 2, the S1P levels in MDD patients (2273 ± 585.1) were higher than those of HCs (1360 ± 461.8) (p < 0.001). Similarly, baseline plasma concentrations of S1PR1 (Fig. 1B) and S1PR3 (Fig. 1C) were significantly elevated in the MDD group (8.23 ± 1.811 µg/L and 8.407 ± 2.174 µg/L, respectively) relative to HCs (5.368 ± 1.895 µg /L and 5.038 ± 1.983 µg /L, respectively) (all p < 0.001). However, S1PR2, S1PR4, and S1PR5 were unchanged in MDD patients (Table S1). Then, we examined the sex differences in S1P, S1PR1, and S1PR3 levels. As shown in Fig. 1D; Table 2, the levels of S1P in female (p < 0.001) and male (p < 0.001) MDD patients were significantly different from their respective HCs. And the levels of S1PR1 (Fig. 1E, p < 0.001) and S1PR3 (Fig. 1F, p < 0.001) in female MDD patients were significantly higher than in HCs, but there was no statistical significance in males on the S1PR1 (Fig. 1E, p = 0.1642) and S1PR3 (Fig. 1F, p = 0.4189) levels. It might be due to the small sample size of male participants. Table 2Measurements of S1P, S1PR1 and S1PR3 levels at baseline and 8-week follow-upIndexHCs (N = 42)MDD (N = 56)P-valueBaselineBaseline (W0)Week 8 (W8)HCs vs. MDD (W0)W0 vs. W8 Total N = 42N = 56S1P (nmol/L)1360 (461.8)2273 (585.1)1661 (654.1)< 0.001a< 0.001cS1PR1 (µg/L)5.368 (1.895)8.23 (1.811)6.629 (2.150)< 0.001b< 0.001cS1PR3 (µg/L)5.038 (1.983)8.407 (2.174)7.761 (2.471)< 0.001b0.0316c Female N = 29N = 40S1P (nmol/L)1470 (501.8)2333 (563.5)1471 (581.1)< 0.001a< 0.001cS1PR1 (µg/L)5.090 (2.018)8.817 (1.614)6.713 (2.060)< 0.001a< 0.001dS1PR3 (µg/L)4.601 (2.085)9.121 (1.635)8.781 (1.958)< 0.001b0.3403d Male N = 13N = 16S1P (nmol/L)1115 (216.1)2123 (629.6)2138 (593.1)< 0.001b0.938cS1PR1 (µg/L)5.990 (1.470)6.762 (1.426)6.419 (2.418)0.1642b0.4637dS1PR3 (µg/L)6.013 (1.345)6.621 (2.370)5.211 (1.649)0.4189b0.0038cS1P, sphingosine-1-phosphate; S1PR1, sphingosine-1-phosphate receptor 1; S1PR3, sphingosine-1-phosphate receptor 3; a = Wilcoxon rank-sum test; b = Independent sample t-test; c = Paired-samples t-test; d = Wilcoxon rank-sum test Table 2Measurements of S1P, S1PR1 and S1PR3 levels at baseline and 8-week follow-upIndexHCs (N = 42)MDD (N = 56)P-valueBaselineBaseline (W0)Week 8 (W8)HCs vs. MDD (W0)W0 vs. W8 Total N = 42N = 56S1P (nmol/L)1360 (461.8)2273 (585.1)1661 (654.1)< 0.001a< 0.001cS1PR1 (µg/L)5.368 (1.895)8.23 (1.811)6.629 (2.150)< 0.001b< 0.001cS1PR3 (µg/L)5.038 (1.983)8.407 (2.174)7.761 (2.471)< 0.001b0.0316c Female N = 29N = 40S1P (nmol/L)1470 (501.8)2333 (563.5)1471 (581.1)< 0.001a< 0.001cS1PR1 (µg/L)5.090 (2.018)8.817 (1.614)6.713 (2.060)< 0.001a< 0.001dS1PR3 (µg/L)4.601 (2.085)9.121 (1.635)8.781 (1.958)< 0.001b0.3403d Male N = 13N = 16S1P (nmol/L)1115 (216.1)2123 (629.6)2138 (593.1)< 0.001b0.938cS1PR1 (µg/L)5.990 (1.470)6.762 (1.426)6.419 (2.418)0.1642b0.4637dS1PR3 (µg/L)6.013 (1.345)6.621 (2.370)5.211 (1.649)0.4189b0.0038cS1P, sphingosine-1-phosphate; S1PR1, sphingosine-1-phosphate receptor 1; S1PR3, sphingosine-1-phosphate receptor 3; a = Wilcoxon rank-sum test; b = Independent sample t-test; c = Paired-samples t-test; d = Wilcoxon rank-sum test Measurements of S1P, S1PR1 and S1PR3 levels at baseline and 8-week follow-up S1P, sphingosine-1-phosphate; S1PR1, sphingosine-1-phosphate receptor 1; S1PR3, sphingosine-1-phosphate receptor 3; a = Wilcoxon rank-sum test; b = Independent sample t-test; c = Paired-samples t-test; d = Wilcoxon rank-sum test Changes in S1P, S1PR1 and S1PR3 levels after 8-week treatments We also examined the effects of treatments on S1P, S1PR1, and S1PR3 levels. As shown in Fig. 2A; Table 2, MDD patients showed a significant decrease in S1P level (p < 0.001) after 8-week treatments. The levels of S1PR1 (Fig. 2B, p < 0.001) and S1PR3 (Fig. 2C, p = 0.0316) were also significantly lower after 8-week treatments. These results indicated that antidepressants therapy effectively downregulates the levels of S1P, S1PR1, and S1PR3. To further investigate, we analyzed the two treatment modalities separately: antidepressants alone (n = 32) and antidepressants combined with rTMS (n = 24). In the antidepressant-only group, S1P (Fig. S1A, p < 0.0001), S1PR1 (Fig. S1B, p = 0.0004), and S1PR3 (Fig. S1C, p = 0.0476) levels were all significantly decreased after treatment. In the combined treatment group, S1P (Fig. S1C, p = 0.0068) and S1PR1 (Fig. S1D, p = 0.018) levels were also significantly lower, while S1PR3 (Fig. S1E, p = 0.2844) did not show a significant reduction. The results suggested that no significant differences in the degree of change were found between patients receiving antidepressants alone and those receiving antidepressants combined with rTMS for S1P. Then, we analyzed whether sex has any impact on treatment outcomes. As shown in Fig. 2, the S1P (Fig. 2D, p < 0.001) and S1PR1 (Fig. 2E p < 0.001) levels in female patients were significantly decreased after 8-week treatments, but there was no statistical significance found in male patients (Fig. 2G and H, p = 0.938, p = 0.4637, respectively). However, there was an opposite result regarding the levels of S1PR3 in females (Fig. 2F, p = 0.3403) and males (Fig. 2I, p = 0.0038). Association between S1P, S1PR1, S1PR3 levels and clinical outcomes The baseline S1P levels in MDD patients predicted the changes in HAMD-24 scores (Fig. 3A, p = 0.020, F = 3.217) and PHQ9 scores (Fig. 3B, p = 0.036, F = 2.779) after 8-week treatments, with sex, age and the treatment modality controlled. In contrast, S1PR1 (Figs. S2A and B, p = 0.687, p = 0.843, respectively) and S1PR3 (Fig. S2C and D, p = 0.674, p = 0.824, respectively) failed to predict symptom changes after 8-week treatment. S1P, S1PR1 and S1PR3 as potential biomarkers for diagnosing MDD The diagnostic values of these molecules were shown on Fig. 4. The AUC value of S1P was 0.8933 (Figs. 4A and 95% CI: [0.821, 0.966]; sensitivity = 91.07%, specificity = 85.71%), the AUC value of S1PR1 was 0.8597 (Figs. 4B and 95% CI: [0.783, 0.937]; sensitivity = 80.36%, specificity = 80.95%), and the AUC value of S1PR3 was 0.8737 (Figs. 4C and 95% CI: [0.807, 0.941]; sensitivity = 85.71%, specificity = 69.05%), suggesting the good diagnostic power in diagnosing MDD. Then, we used multiple logistic regression to analyze the diagnostic power of these three molecules combined, with sex and age controlled. The results demonstrated that the combination of these three molecules could yield a better AUC value of 0.9575 (Figs. 4D and 95% CI: [0.923, 0.992]; sensitivity = 89.66%, specificity = 90.00%), and hence the model had high accuracy. Discussion This study provides a comprehensive investigation into the peripheral S1P signaling pathway in MDD. There are four key findings: First, the plasma concentrations of S1P and its receptors, S1PR1 and S1PR3, are significantly elevated in patients during an acute depressive episode, and these concentrations are reduced after 8 weeks of antidepressant therapy. Second, this dysregulation is sex-dependent, with a more pronounced increase in S1PR1 and S1PR3 in female patients. Third, baseline plasma S1P levels can predict outcomes, showing a substantial correlation with the degree of clinical improvement post-treatment. Finally, a combined panel of these three molecules demonstrated a high diagnostic accuracy in distinguishing MDD patients from healthy individuals, with an AUC of 0.9575. These findings suggest the S1P pathway in the pathophysiology of MDD and highlight its significant potential as a clinically applicable biomarker. We found a significant increase of S1P, S1PR1, and S1PR3 in the circulation of MDD patients, whereas the S1PR2, S1PR4, and S1PR5 levels were unchanged. This observation suggests a systemic dysregulation of a key lipid signaling pathway linked to inflammation and vascular function. It is important to acknowledge that plasma biomarker levels provide an indirect measure and may not directly reflect concurrent pathological changes within the central nervous system (CNS). Plasma S1P is primarily derived from erythrocytes, activated platelets, and vascular endothelial cells. Our measurement of soluble S1PR1 and S1PR3 likely reflects a combination of receptor shedding from the surface of these and other cell types, such as lymphocytes. Since MDD is commonly connected with a chronic, low-level inflammatory state and endothelial dysfunction, it is reasonable to assume that our findings represent this underlying pathology. In depression, increased platelet activation and endothelial stress could lead to more S1P being produced and released into the plasma, which may explain the higher levels observed in our patients [38]. An increase in soluble S1PR1 and S1PR3 might indicate a compensatory mechanism in response to elevated ligand levels or enhanced shedding from cell surfaces, such as activated lymphocytes or endothelial cells, during pro-inflammatory states. While our correlative data suggests an association between S1P pathway dysregulation and MDD, the upstream drivers (e.g., specific cytokine release, platelet activation) and downstream consequences (e.g., BBB permeability) require direct validation in future studies. Our results contribute important clinical evidence to a complex and conflicting body of research. Animal model studies have further revealed the mechanism of S1P in the onset of depression. In inflammatory depression models, the inhibition of the Sphk1/S1P signaling pathway can effectively alleviate inflammation and depressive symptoms, further supporting the key role of S1P in the pathology of depression [39]. A transgenic mouse model of Alzheimer’s disease, Tg2576, exhibited a significant S1P reduction in the brain and S1P elevation in plasma after stress, indicating that the dynamic changes of S1P inside and outside the brain are associated with the depressive state [40]. In addition, studies have shown that modulating the S1P and its receptor signaling pathways using a S1P receptor modulator, FTY720, can help mitigate depressive-like behaviors, neuroinflammation, and hippocampal neuron function, demonstrating the potential of the S1P signaling pathway as a therapeutic target [41]. Nonetheless, the increased peripheral S1P in living MDD patients in our study is in clear contrast to the findings of Esaki et al. (2020), who reported no change in S1P level in postmortem brain tissue from MDD patients [31]. It implies that the primary S1P dysregulation in MDD is likely a systemic event in the periphery and at the neurovascular unit, rather than a widespread change within the brain parenchyma. Increased plasma S1P likely indicates a systemic inflammatory state of MDD, which may not be reflected in postmortem brain tissue analyses. Furthermore, our findings differ from several animal models where elevated S1P in specific brain regions, such as midbrain and thalamus, was linked to stress resilience, while plasma S1P remained unchanged [30]. This suggests that localized and adaptive increases in central S1P during stress could be distinct from the systemic and potentially maladaptive elevation of peripheral S1P observed in MDD. Therefore, our study might clarify that in human MDD, the peripheral S1P signaling axis is a key site of dysregulation. S1PR1 and S1PR3 have dual roles in regulating neuroimmune responses and synaptic plasticity, pointing to the balance required for neuronal health and function, which is commonly disrupted in depression. So far, this is the first study that investigated the concentration of plasma S1PR1 and S1PR3 in MDD patients. However, human plasma ELISA is well-established for successfully quantifying these plasma-shed S1PRs, which are canonical membrane-bound GPCRs that might be released into circulation via extracellular vesicles [42], particularly in acute lung injury or stroke [43, 44]. S1PR1 mainly regulates the cell migration and survival signals of the central nervous system and participates in the regulation of neuroimmune inflammation [45]. S1PR1 interacts with multiple signaling pathways in neuroinflammation, regulating the functions of neurons and glial cells, and becoming a potential therapeutic target for neuro-psychiatric diseases [46]. S1PR3 is highly expressed in astrocytes and is upregulated under inflammatory stimulation. S1PR3 not only plays a key role in regulating inflammatory responses but also exerts an important role in regulating neuronal function and neural plasticity. In rodent models, it has been found that S1PR3 in neurons of the mPFC is a key regulatory factor for stress recovery ability [27]. In the chronic unpredictable mild stress (CUMS) model, the expression of S1PR3 in the neurons of the hippocampus is decreased, and specific overexpression of S1PR3 in hippocampal neurons promotes synaptic plasticity, alleviates depressive-like behaviors, and inhibits the upregulation of RhoA and ROCK1 in CUMS mice [28]. Another study found that the expression level of S1PR3 in the prefrontal cortex of CUMS rats was increased, and the difference between these two results may be due to the different animal species, brain regions, and cell types examined. The expression of S1PR3 in different animal species, brain regions, and cell types enables S1PR3 to participate in different inflammatory responses, regulate neuronal structure and function, and thereby affect emotional regulation and mental health [47]. One of the most novel and significant findings of this study is the sex difference in the dysregulation of S1P receptors. Although S1P levels were elevated in both male and female MDD patients, the increase in plasma S1PR1 and S1PR3 was significantly more substantial in females. This finding provides a potential molecular explanation for the epidemiological fact that MDD is nearly twice as prevalent in women as it is in men [48]. This sex-specific molecular signature is consistent with expanding research in psychoneuroimmunology that highlights significant differences in immune and inflammatory responses between males and females. Generally, adult females mount stronger innate and adaptive immune responses than males. While evolutionary advantageous for host defense, this heightened reactivity can increase susceptibility to autoimmune diseases and inflammation-related psychiatric conditions like MDD [49, 50]. Clinical research has found that depression exhibits sex-specific inflammatory responses, and elevated levels of IL-1β and IL-6 are more commonly associated with MDD in women, whereas an increase in IL-17 could be more specific to men [51]. Ovarian hormones, especially estrogen, are key modulators of immune function, inflammation, and stress responses, and their variations throughout a woman’s life are thought to increase the risk of depression [52]. Our findings identify the S1P signaling pathway as a critical point for these endocrine and immune factors. S1PR1 and S1PR3 signaling are upstream regulators of inflammatory cascades, including the production of cytokines and the activation of key transcription factors such as NF-κB [53, 54]. The increased levels of soluble S1PR1 and S1PR3 in female patients could therefore be a molecular manifestation of a more sensitized or dysregulated inflammatory response system. We hypothesize that sex hormones may directly or indirectly regulate the expression, shedding, or signaling transduction of S1P receptors on immune cells and endothelial cells, thereby enhancing the inflammatory responses and raising the risk of depression observed in women. The results strongly suggest that future biological psychiatry research should adopt a sex-stratified approach. Beyond its pathophysiological implications, our study reveals the significant translational potential of the S1P pathway in clinical psychiatry. Our findings address the two main objectives of biomarker development in MDD: (i) assisting in diagnosis and (ii) predicting or monitoring treatment response. It is clinically important that elevated baseline S1P levels are associated with a greater alleviation in depressive symptoms. This predictive biomarker could help stratify patients at diagnosis. Those with elevated baseline S1P, indicating significant inflammation, could be recognized as a subgroup prone to have a strong response to standard monoaminergic antidepressants that also possess anti-inflammatory properties. Alternatively, this marker could potentially aid in identifying patients who could benefit from supplementary therapies targeting inflammatory or sphingolipid pathways [55]. A key highlight of this study is the remarkable diagnostic performance of the combined S1P, S1PR1, and S1PR3 panel, with an AUC of 0.9575, showing high sensitivity and specificity. Research on psychiatric biomarkers has been historically challenged by the biological heterogeneity of disorders and the limited effect sizes of individual markers [3]. The enhanced effectiveness of the multi-marker panel over individual analytes supports the concept that a disease composite signature reflects multiple elements of a dysregulated pathway. If validated in larger and independent cohorts, this set of biomarkers could be a crucial tool for an objective blood-based diagnostic support for MDD, enhancing clinical assessments and potentially allowing for earlier and more accurate diagnosis. Limitations and future directions Several limitations should be acknowledged in the interpretation of our results. First, the moderate sample size of 56 MDD patients and 42 HCs limited the ability to conduct a comprehensive analysis of potential confounding factors that are associated with S1P, S1PR1, and S1PR3. The number of male participants was smaller than that of females. This suggests the necessity for replication in larger, multi-centered studies with balanced sex representation to validate our results. Secondly, the peripheral S1P was primarily derived from red blood cells and platelets, our measurements were confined to the peripheral circulation. While we have proposed mechanistic links to CNS pathology via the BBB, plasma levels still serve as an indirect indicator of central processes. Future research should adopt the advanced neuroimaging techniques, such as positron emission tomography (PET) using S1PR1-specific radioactive ligands such as [18F]TZ4877, to visualize receptor expression and occupancy in the living human brain and to directly test for central S1P pathway dysregulation in animal models of MDD or patients with MDD [56]. Third, the plasma ELISA detection did not allow for cell-specific source determination. It is important to define their origin in future studies. For example, studies employing techniques such as flow cytometry could characterize S1PR1 and S1PR3 expressions on specific immune cell subsets (e.g., T-cells, monocytes) from MDD patients to identify the cellular sources of dysregulation. Finally, this is a correlational study in nature and does not establish causality. Additional research is now required to elucidate the specific contributions of S1PR1 and S1PR3 to the pathophysiology of depression. This will involve the use of animal models and advanced in vitro systems to explore the causal effects of S1P signaling on neuroinflammation, synaptic plasticity, and depressive-like behaviors. Furthermore, exploring genetic variations in the S1PR1 and S1PR3 genes as potential risk factors for MDD could provide further evidence for the pathway’s involvement. Conclusions In conclusion, our work revealed that plasma concentrations of S1P, S1PR1, and S1PR3 are elevated in MDD patients compared to HCs. This dysregulation is state-dependent and responsive to antidepressant treatment, and exhibits distinct sex-specific differences. Simultaneously, we found that the baseline S1P levels could predict clinical outcomes, and the combination of these three molecules could provide more effective diagnostic performance. In contrast, baseline levels of S1PR1 and S1PR3 individually failed to display predictive capability. The findings strongly suggest that the S1P pathway plays a significant role in the neurobiological processes and diagnosis of depression. Fig. 1Differences in baseline S1P, S1PR1, and S1PR3 levels between HCs and MDD patients. Compared to HCs, MDD patients had significantly higher levels of S1P (A), S1PR1 (B) and S1PR3 (C). The levels of S1P (D), S1PR1 (E) and S1PR3 (F) in both female and male MDD patients were higher than HCs. Data are presented as mean ± S.D. HCs, healthy controls; MDD, major depressive disorder; F, female; M, male; ***p < 0.001 Differences in baseline S1P, S1PR1, and S1PR3 levels between HCs and MDD patients. Compared to HCs, MDD patients had significantly higher levels of S1P (A), S1PR1 (B) and S1PR3 (C). The levels of S1P (D), S1PR1 (E) and S1PR3 (F) in both female and male MDD patients were higher than HCs. Data are presented as mean ± S.D. HCs, healthy controls; MDD, major depressive disorder; F, female; M, male; ***p < 0.001 Fig. 2Changes and gender differences in S1P, S1PR1, and S1PR3 levels in MDD patients after 8-week treatments. Levels of S1P (A), S1PR1 (B), and S1PR3 (C) in the total MDD patient cohort at baseline and after 8 weeks of treatment. Statistical comparison was performed using paired-sample t-tests. Levels of S1P (D), S1PR1 (E), and S1PR3 (F) in female MDD patients at baseline and after 8 weeks of treatment. Statistical comparison was performed using paired-sample t-tests (S1P, S1PR1) or the Wilcoxon signed-rank test (S1PR3). Levels of S1P (G), S1PR1 (H), and S1PR3 (I) in male MDD patients. Statistical comparison was performed using paired-sample t-tests (S1P, S1PR3) or the Wilcoxon signed-rank test (S1PR1). ***p < 0.001; **p < 0.01; *p < 0.05 Changes and gender differences in S1P, S1PR1, and S1PR3 levels in MDD patients after 8-week treatments. Levels of S1P (A), S1PR1 (B), and S1PR3 (C) in the total MDD patient cohort at baseline and after 8 weeks of treatment. Statistical comparison was performed using paired-sample t-tests. Levels of S1P (D), S1PR1 (E), and S1PR3 (F) in female MDD patients at baseline and after 8 weeks of treatment. Statistical comparison was performed using paired-sample t-tests (S1P, S1PR1) or the Wilcoxon signed-rank test (S1PR3). Levels of S1P (G), S1PR1 (H), and S1PR3 (I) in male MDD patients. Statistical comparison was performed using paired-sample t-tests (S1P, S1PR3) or the Wilcoxon signed-rank test (S1PR1). ***p < 0.001; **p < 0.01; *p < 0.05 Fig. 3Baseline S1P levels in MDD patients predicted the changes in HAMD-24 and PHQ9 scores induced 8-week treatments. In MDD patients, baseline S1P levels in the plasma predicted the changes in HAMD-24 (A) and PHQ-9 (B) scores induced 8-week treatments Baseline S1P levels in MDD patients predicted the changes in HAMD-24 and PHQ9 scores induced 8-week treatments. In MDD patients, baseline S1P levels in the plasma predicted the changes in HAMD-24 (A) and PHQ-9 (B) scores induced 8-week treatments Fig. 4Diagnostic performances of S1P, S1PR1, and S1PR3 in diagnosing MDD. Receiver operating characteristic (ROC) curves for (A) S1P, (B) S1PR1, (C) S1PR3, and (D) the panel combining all three molecules. The area under the curve (AUC) value is indicated within each panel. AUC, area under the ROC curve Diagnostic performances of S1P, S1PR1, and S1PR3 in diagnosing MDD. Receiver operating characteristic (ROC) curves for (A) S1P, (B) S1PR1, (C) S1PR3, and (D) the panel combining all three molecules. The area under the curve (AUC) value is indicated within each panel. AUC, area under the ROC curve Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1 Supplementary Material 2 Supplementary Material 2 Supplementary Material 3 Supplementary Material 3