Depression is a prominent global cause of disability with a lifetime prevalence approaching 20% (1, 2). It can manifest at any stage of life, with the most common period for the first episode typically spanning from adolescence to middle age (1). In its most severe form, depression may lead to suicide attempts, making it one of the leading causes of death among adolescents in Europe (3). It appears to be a multifactorial disease arising from a combination of genetic or metabolic predisposition and environmental factors such as stressful life events and other stressors (4). Currently, the diagnosis of depression relies on clinical assessment, which includes evaluating depressive symptoms like low mood, reduced self-esteem, inappropriate guilt, thoughts of death and suicide, decreased concentration, loss of interest or pleasure in once enjoyable activities, and disturbances in sleep and appetite. Treatment typically involves antidepressant medication, although its efficacy is generally considered mild to moderate (5), with up to 60% of patients experience resistance to treatment (6). Despite major efforts to uncover the pathophysiological mechanisms of depression by analyzing data from genome-wide association studies (7) and RNA gene expression studies (8, 9), our understanding remains limited.

Recent technological advances, particularly within proteomic-based platforms, are providing new insights for better comprehension of depression, going beyond what conventional targeted methods can offer. With the advent of cutting-edge proteomic and transcriptomic techniques allowing simultaneous quantitative assessment of a wide spectrum of proteins or mRNA transcripts, it is now possible to explore not only the functions of individual proteins but also the roles of biochemical pathways and associated metabolites. Proteins serve as the ultimate products of RNA and DNA, often serving as the functional and modifiable units in disease mechanisms. Consequently, proteome analyses hold the potential to elucidate the pathophysiological mechanisms of depression and to identify potential biomarkers for diagnosis, treatment, and monitoring of disease progression. Studies aimed at discovering objective blood-based biomarkers in the fields of psychiatry, neurology, and neuropsychiatry have experienced exponential growth (10–12). However, due to both methodological and clinical challenges associated with psychiatric conditions, such as their inherent complexity and heterogeneous presentation, clinical efficacy remains extremely limited (13–15).

Most studies on risk factors for depression have primarily focused on the adult population. While various biological mechanisms have been explored, no biomarker has been definitively identified or validated for assessing the risk or presence of depression in adolescence and young adulthood (16–20). This is important given that approximately half of depression diagnoses in adults stem from onset in adolescence. Considering the high incidence of depression during the early decades of life and its chronicity throughout life, adolescence presents a unique opportunity to develop effective prevention strategies and alleviate the burden associated with it (21). Furthermore, more research on depression involving youth populations is essential for understanding the natural progression of the disease and identifying factors that may underlie the commonly observed differences in treatment outcomes between adults and children (22). Additionally, developing biomarkers that could predict antidepressant response in the young population would contribute to creating more personalized and effective treatments for young people, a key step in improving the prognosis of depression across the lifespan (23).

In this study, we aimed to analyze protein levels in 461 adolescents assessed for depression as part of the domestic Psychiatric Health in Adolescent Study (PSY cohort) conducted in Uppsala, Sweden, using the Proseek Multiplex Neuro Exploratory panel with Proximity Extension Assay (PEA) technology provided by Olink Bioscience. Subsequently, we conducted transcriptome analyses for the genes corresponding to the significant proteins, using publicly available cohorts, in order to identify differentially expressed transcripts between depressed individuals and controls.

In this exploratory study, the Olink Neuro Exploratory panel was used to analyze the expression levels of 92 proteins among adolescents based on their DAWBA risk in the PSY cohort, followed by transcriptome analyses in independent cohorts. As a result, we identified a total of seven proteins showing different levels between DAWBA risk groups at nominal significance, whether adjusted or non-adjusted for antidepressant use, including RBKS, CRADD, ASGR1, HMOX2, PPP3R1, CD63, PMVK. Transcriptomic analyses for these genes showed nominally significant replication of PPP3R1 in two of four cohorts, while ASGR1 and CD63 were replicated in only one cohort.

One protein that drew particular attention due to its differential expression is PPP3R1, as it showed nominal significance both before and after adjustment for antidepressant use, and it was the only protein with consistent directions of change in two transcriptome cohorts. PPP3R1 is a 19 kDa calcium-binding regulatory subunit of the calcium-dependent protein phosphatase, also known as calcineurin. Abundantly expressed in human brain neuronal cell cytosol, PPP3R1 plays a critical role in the calcineurin signaling pathway (33, 34). Studies have observed that mice with a PPP3R1 mutation, eliminating calcineurin activity in somatic cells, exhibited structural defects in anterior neural structures (35). There is increasing evidence suggesting that calcineurin signaling is implicated in the pathophysiology of depression and its treatment. The association between calcineurin activity and psychiatric disorders is based on clinical observations that the rates of anxiety and depression increase in patients treated chronically with the calcineurin inhibitor cyclosporine-A to prevent rejection after organ transplantation (36). Preclinical studies reveal depressive-like behavior induction in mice through amygdala calcineurin inhibition (37). Additionally, chronic stress in rats, a model for depression research, reduces calcineurin activity in the hippocampal CA3 region (38). Notably, calcineurin has also been found to have direct antidepressant-like effects, as the inhibition of calcineurin in the prefrontal cortex of rats induces depressive-like behavior (39). These findings taken together support the role of calcineurin in the pathophysiology of depression.

PPP3R1 was replicated in two transcriptomic cohorts, GSE53987 [prefrontal cortex (PFC)] and GSE64930 (whole blood). The differential expression of PPP3R1 in the prefrontal cortex of depressed adolescents aligns with the existing evidence that the PFC is one of the regions most consistently affected in depression. Extensive literature has strongly implicated PFC dysfunction in MDD, revealing functional, structural, and systems-level abnormalities spanning various PFC regions (40). Evidence of structural changes in the PFC associated with depression mainly comes from secondary measures. For instance, the volume of the PFC is reduced in depressed patients, and this decrease is correlated with the duration of illness (41). In depression, specific changes can be observed in the molecular composition of PFC neurons (42), which modulate their electrical properties (43). Decreases in the volume of the orbitofrontal cortex (44) and pyramidal cell density (45) have also been observed in depression. These studies suggest that PFC neurons undergo widespread physiological and morphological alterations during depression. Other lines of evidence include dysregulation of glutamatergic and GABAergic neurotransmission in the PFC. PFC glutamate metabolites are reduced in depression (46), and postmortem studies demonstrate changes in ionotropic and metabotropic glutamate receptors (47, 48). Medial PFC levels of the GABA synthetic enzyme glutamate decarboxylase-67 are also reduced in postmortem brains of depressed subjects (30), as are markers of the somatostatin/calbindin (SST) GABAergic subtype (49, 50).

Other interesting findings involve two proteins, ASGR1 and CD63, corresponding to genes that may offer new insights into the pathophysiological processes of depression. Both proteins were replicated with consistent changes, but only in one cohort, GSE98793, where CD63 remained significant after Bonferroni correction. CD63 belongs to the tetraspanin family and is characterized by its four transmembrane domains (51). It can associate with interaction partners, including integrins, receptors, kinases, and other tetraspanin proteins, on the cell surface, thus contributing to the regulation of multiple signaling pathways (51–54). CD63 also serves as a platelet dense granule and lysosomal membrane protein. Interestingly, CD63 exhibits high expression in patients with comorbid diabetes and depression compared to those with diabetes alone. This comorbidity results in enhanced platelet hyperactivation and a pro-inflammatory state, increasing susceptibility to vascular complications (55). On the other hand, ASGR1 is the major subunit of the asialoglycoprotein receptor (ASGPR), a liver-specific lectin that plays a role in glycoprotein homeostasis (56). Variants in ASGR1 are associated with lower non-high-density lipoprotein (non-HDL) cholesterol levels and a reduced risk of coronary artery disease (57, 58). Thus, ASGR1 holds clinical potential as a target for lowering blood cholesterol levels. However, there is limited information available regarding the regional or cellular specificity of ASGR1 in the human brain, and its function in the brain remains poorly understood (59).

In the search of an explanatory model for the observed associations between depression and the blood levels of PPP3R1, CD63, and ASGR1, platelets emerge as potentially pivotal players ( Figure 3 ). The presence of free CD63 in plasma strongly suggests its exosomal origin, as CD63 is widely recognized as one of the most specific exosomal markers (62). Notably, activated blood platelets are a major source of exosomes in human blood (63). This finding raises the possibility of a positive association between activated blood platelets and depression, a notion supported by multiple studies linking platelet activation to depression (60, 61). Additionally, it’s worth noting that ASGR1 plays a vital role in platelet clearance by the liver, both in humans and pigs (66, 67). Moreover, genetic variants of ASGR1 has been reported to influence platelet activation (68). Furthermore, PPP3R1, a regulatory component of the calcineurin complex responsible for phosphatase activity, has been observed to bind to platelets and prevent their aggregation (64, 65). Consequently, an increase in PPP3R1 levels may signify disruptions in platelet homeostasis.

In summary, while this model may involve some speculation, it is reasonable to suspect that the identification of three proteins related to platelet activation status in depressed adolescents could potentially reflect the neuropathological status of these adolescents. This notion is supported by a substantial body of scientific evidence suggesting that blood platelets may mirror the condition of neurons in the central nervous system (CNS) (69–74). To highlight some specific roles of platelets, studies of the serotonin uptake, storage, and secretion mechanisms in platelets have indicated major similarities to neurons (75). Studies investigating the comorbidity between cardiovascular disease and depression have proposed platelet serotonin signaling as a potential mechanism for the higher incidence of cardiac adverse events in depressed cardiovascular patients (76). Additionally, other studies have suggested that altered platelet reactivity might link stress, depression, and cardiovascular disease through changes in platelet aggregability and their content of bioactive compounds (77, 78). Platelet activation has also been implicated as a contributor to the progression of neurological conditions such as Alzheimer’s disease (79). There is potential for platelet interactions with systemic factors that change concurrently with CNS alterations, underscoring the importance of future studies elucidating the numerous systemic factors that change during depression onset, e.g. in adolescence, and how they interact with platelets.

The study’s results should be interpreted in light of several limitations. Firstly, our proteomic platform covers a rather selective portion of the proteome, potentially missing numerous proteomic pathways related to depression that our methods did not capture. Additionally, a significant number of proteins in the selected panel appear to be unexpressed in whole blood, as only 43 out of 92 were analyzable. This limitation might be attributed to the use of peripheral fluid biomarkers, as concentrations measured in the periphery may not necessarily reflect pathophysiological processes in the CNS. Furthermore, depression is a heterogeneous group of diseases often occurring concurrently with or induced by other conditions. Consequently, it is challenging to select a molecularly homogeneous group of patients or samples.

Additional limitations include the nominal significance observed for almost all proteins, which may provide weak evidence. However, it’s worth noting that nominal replication at the transcriptome level with consistent directions in independent cohorts could be considered stronger evidence. Another limitation is related to the characterization of depression across all cohorts, which is not consistent and may not represent the same phenotype. Furthermore, the populations in the cohorts are diverse and not easily comparable. Despite our efforts to account for some of the differences between the cohorts, such as age, there is still a lack of age overlap between the PSY cohort and the transcriptome cohorts. This necessitates cautious interpretation of the discordant results between the cohorts. However, on the other hand, the current findings in the PSY cohort, which reveal changes associated with depression that differ from those observed in older patients in the transcriptomic cohorts, may offer valuable insights into the etiology of depression onset during adolescence. Lastly, there were many missing values regarding medication in the PSY cohort. Since antidepressant treatment can be a significant confounder, we conducted analyses both with and without adjustment for antidepressants. Adjusting for antidepressants led to a smaller sample size due to the exclusion of missing values. We observed slightly increased effect size in the adjusted analysis that should be interpreted carefully. It could be related to biological effects of antidepressants masking protein effect in non-adjusted models or it may be of random origin as the accuracy of estimates is smaller in the adjusted analysis. Consistently, the observed shifts in the estimates of matching proteins matched between analyses (PPP3R1, RBKS, and CRADD) were less than corresponding standard errors. Lastly, even though the analysis was adjusted for antidepressants, it is possible that the real prevalence of antidepressant use may be underestimated due to underreporting from study participants. However, our estimated prevalence of antidepressant used in the PSY cohort of ~4.2% is comparable to the recent report on antidepressant use in Sweden of ~4.76 (age group 10-19 years) (80).

Our study of adolescent depression revealed protein-level and transcriptomic differences, particularly in PPP3R1, pointing to the involvement of the calcineurin pathway in depression. Our finding with regard to PPP3R1 also support the role of the prefrontal cortex in depression and reinforces the significance of investigating PFC-related mechanisms in depression. However, it’s important to consider these findings within the context of certain limitations, such as the selectivity of our proteomic platform and variations in depression characterization across cohorts. Further research is needed to validate and expand upon these findings, potentially leading to improved diagnostic and therapeutic strategies for adolescent depression.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding author.

The studies involving humans were approved by Regional Ethics Committee of Uppsala, Sweden. The studies were conducted in accordance with the local legislation and institutional requirements. Written informed consent for participation in this study was provided by the participants’ legal guardians/next of kin.

AS: Conceptualization, Data curation, Formal analysis, Methodology, Visualization, Writing – review & editing. ML: Conceptualization, Formal analysis, Methodology, Writing – original draft. DN: Methodology, Writing – review & editing. JJ: Conceptualization, Writing – review & editing. HS: Conceptualization, Funding acquisition, Supervision, Writing – review & editing.