Introduction

Schizophrenia (SCZ) is a severe psychotic disorder affecting about 20 million people worldwide1, significantly impacting patients’ quality of life. The symptoms of this illness include positive (delusion and hallucination) and negative (blunted affect, social withdrawal) symptoms, along with cognitive deficits2,3.

Compelling studies have indicated the involvement of an NMDA glutamate receptor (NMDAR) hypofunction in the pathophysiology of SCZ4. Accordingly, genetics studies have ascertained that variations in single genes associated with glutamatergic neurotransmission are related to phenotypes of SCZ5,6, and a recent Genome-Wide Association Study (GWAS) on SCZ7 has found that variants associated with this illness mainly occur in genes modulating NMDAR structure, expression and signaling.

Besides the endogenous agonists, L-glutamate (L-Glu) and L-aspartate (L-Asp), and the co-agonist glycine (Gly), two atypical amino acids in D-configuration, D-aspartate (D-Asp) and D-serine (D-Ser), modulate glutamatergic NMDAR neurotransmission. D-Ser binds to the co-agonist glycine site on GluN1, while D-Asp binds to the agonist glutamate site on GluN2 subunit of NMDARs8,9,10. In addition, D-Asp has been shown to stimulate metabotropic Glu5 receptors (mGluR5) in neonatal rat brain slices, thus extending its functional role to metabotropic receptor signaling during early postnatal life11.

Consistent with a critical role of NMDAR signaling dysfunction in SCZ, multiple studies have reported abnormally lower levels of D-Ser or D-Ser/total Ser ratio in cerebrospinal fluid (CSF) and serum12,13,14,15,16,17,18,19,20,21,22,23 of patients. Moreover, reduced D-Asp levels or a low D-Asp/total Asp ratio have been reported in the serum and post-mortem prefrontal cortex of patients with SCZ24,25,26,27. Specifically, the variation of D-Asp levels in SCZ brain has been associated with increased gene expression24 or abnormal overactivity of the D-aspartate-oxidase (DDO) enzyme, which degrades D-Asp25. Similarly, genes encoding D-amino acid oxidase (DAO), which degrades D-Ser, as well as the DAO activator (DAOA)/G72 have been associated with SCZ28,29. Finally, a variant of the gene encoding serine racemase (SR), which synthesizes D-Ser from L-Ser, has shown significant associations with SCZ in genome-wide studies30.

Consistent with these biochemical and genetic findings highlighting a prominent role of D-Ser metabolism in SCZ pathophysiology, clinical trials indicated that D-Ser add-on to antipsychotics (APs) can improve negative symptoms31,32 and deficits in auditory learning associated with SCZ33. In keeping with this, the NMDAR partial agonist at the GluN1 site, D-cycloserine, has demonstrated benefits for negative and cognitive symptoms34 of SCZ when used alongside APs35. Furthermore, AP add-on with sodium benzoate, a D-Ser catabolism inhibitor, showed beneficial effects in SCZ36,37.

Further indicating a remarkable influence of the GluN1 subunit of NMDARs in the pathophysiology of this mental illness, changes in Gly metabolism have been found in patients at their First Episode of Psychosis (FEP)38 and with full-blown SCZ39. Additionally, as reported for D-Ser, clinical investigations indicated the use of Gly as an adjunctive treatment to ameliorate SCZ cognitive deficits40.

Analogous results have been documented for asparagine (Asn), whose plasma levels were reported to be increased in FEP patients compared to healthy controls and remained high after starting AP treatment41. Similarly, Gln/Glu ratio was found to be increased in CSF of FEP patients42 and plasma of individuals with recent onset SCZ43, and reduced in FEP patients after AP treatment beginning41.

Altogether, this evidence highlights the potential impact of AAs acting at NMDAR as a putative determinant for the development of early psychosis and its progression toward SCZ. However, it remains obscure whether specific AA variation occurs as the illness progresses from at-risk premorbid to clinically manifest stages.

To clarify this issue, by using High-Performance Liquid Chromatography (HPLC), we measured the blood levels of L- and D-Asp, L- and D-Ser, L-Glu, L-Gln, Gly, L-Asn, along with the L-Gln/L-Glu, D-Ser/total Ser and D-Asp/total Asp ratios in a cohort of 73 subjects in an early stage of disease44, including 48 At-Risk Mental State (ARMS) and 25 FEP individuals, 82 patients with full-blown SCZ and 115 Healthy Donors (HD). Furthermore, we investigated the association of serum AA levels and psychosis symptom severity and performance of Working Memory (WM), a prototypical measure of cognitive impairment in psychosis45.

Materials and methods

Sample demographic and clinical characteristics

A total of 270 individuals entered the study and were classified as Healthy Donors (HD, n = 115, 43%), At-Risk Mental State (ARMS, n = 48, 18%), First Episode Psychosis (FEP, n = 25, 9%) and affected by schizophrenia (SCZ, n = 82, 30%). Demographic and clinical characteristics of participants are reported in Table 1. Subject enrollment was carried out at the Psychiatric Unit of the “Azienda Ospedaliera Universitaria Consorziale Policlinico di Bari”, Bari - Italy. Recruitment procedures were carried out in accordance with “The Code of Ethics of the World Medical Association” (Declaration of Helsinki) and Good Clinical Practice and approval was given by the local Ethics Committee (Comitato Etico Locale IRCSS Istituto Oncologico “Gabriella Serio”, Bari - Italy). All participants signed a written informed consent to participate after receiving a comprehensive explanation of the study procedures.

Table 1 Demographical and clinical variables of the groups investigated.
Full size table

Diagnosis of SCZ was made using the Structured Clinical Interview for the DSM IV-TR (SCID)46, which was administered by psychiatrists. Individuals were identified as FEP when they met criteria for DSM-IV-TR affective or non-affective psychotic episode fulfilled within the past 3 months, and when onset of psychosis happened within the past 24 months. FEP were excluded when they underwent AP medication for more than 90 cumulative days.

The ARMS state was defined by: (i) cognitive disturbances (COGDIS), as assessed by the Schizophrenia Proneness Instrument (SPI-A)47, and/or (ii) ultra-high-risk (UHR) criteria for psychosis, according to the Structured Interview for Psychosis-Risk Syndromes criteria48. ARMS were excluded when they had an intake of AP medication for more than 30 cumulative days, and when they had any intake of AP medication within the past 3 months before the study entrance44. In order to allow for inter-individual comparison, doses of AP medication assumed were standardized to olanzapine equivalents according to the methodology indicated in Leucht et al. 49,50.

Clinical symptom severity in ARMS, FEP and SCZ individuals was assessed with the Positive and Negative Syndrome Scale (PANSS)51. Healthy individuals underwent the non-patient version of the SCID52 in order to exclude the presence of any psychiatric disorder.

ARMS, FEP, SCZ and HD individuals underwent Working Memory (WM) behavioral assessment with the N-Back task as described elsewhere53.

All individuals had no history of drug or alcohol abuse within the last 6 months, head trauma with loss of consciousness, or any other clinically significant medical condition.

Human serum collection and preparation

All participants underwent a 10-ml whole blood withdrawal for subsequent serum isolation.

Blood samples, collected in anticoagulant-free tubes, were left at room temperature for about 60 min to allow for coagulation. Samples were then centrifuged at 1900 × g at 4 °C for 10 min. The upper phase containing the serum was transferred to a new tube and centrifuged again to remove any debris. Finally, serum was transferred to a new tube and stored at −80 °C until HPLC analysis.

HPLC analysis of amino acid content

Serum samples (100 µl) were mixed in a 1:10 dilution with HPLC-grade methanol (900 µl) and centrifuged at 13,000 × g for 10 min; supernatants were dried and then suspended in 100 µl of 0.2 M Trichloroacetic acid (TCA). TCA supernatants were then neutralized with 0.2 M NaOH and subjected to precolumn derivatization with o-phthaldialdehyde/N-acetyl-L-cysteine in 50% methanol. AA derivatives were resolved on a UHPLC Nexera X3 system (Shimadzu) by using a Shim-pack GIST C18 3-μm reversed-phase column (Shimadzu, 4.0 × 150 mm) under isocratic conditions (0.1 M sodium acetate buffer, pH 6.2, 1% tetrahydrofuran, and 1 ml/min flow rate). A washing step in 0.1 M sodium acetate buffer, 3% tetrahydrofuran and 47% acetonitrile, was performed after every run. Identification and quantification of AAs were based on retention times and peak areas, compared with those associated with internal standards (Fig. 1). The identity of the D-Asp peak was further evaluated by selective degradation catalyzed by a recombinant human D-aspartate oxidase (hDDO) enzyme54. hDDO enzyme (12.5 μg) was added to the samples, incubated at 30 °C for 3 h, and subsequently derivatized. The detected AA concentrations were expressed as µM, while the ratio between D-AA and total AA (D/D + L) was expressed as a percentage (%).

Fig. 1: Identification and quantification of AAs based on retention times and peak areas.
figure 1

HPLC chromatogram illustrating the D-aspartate (D-Asp), L-aspartate (L-Asp), L-asparagine (L-Asn), L-glutamate (L-Glu), D-serine (D-Ser), L-serine (L-Ser), L-glutamine (L-Gln) and glycine (Gly) peaks obtained from blood serum samples. The identity of the peaks corresponding to D-Asp and D-Ser was verified by adding internal standards to the serum sample (inset, blue line).

Full size image

Statistical analysis

Statistical analyses were performed using SPSS software v.27 (SPSS Inc., Chicago, IL, USA) and R Language v.4.3.2 (R Foundation for Statistical Computing, Vienna, Austria). Normality distribution was assessed by q–q plot and Shapiro–Wilk test. Quantitative variables were expressed by median and interquartile range (IQR), while qualitative variables by absolute or relative frequency. Associations between variables were estimated by Fisher’s Exact test. Correlation was evaluated by non-parametric Spearman’s rho. Differences between groups were studied by non-parametric Kruskal–Wallis test followed, if statistically significant, by post-hoc tests performed by Mann–Whitney test with Bonferroni’s correction. The effect of confounders was evaluated by ANCOVA on natural log-transformed variables.

Results

Clinical and demographical characteristics

Group sex and age matching revealed sex was not associated with diagnosis (p = 0.447). ARMS and SCZ groups displayed the lowest and highest age, respectively (Kruskal–Wallis, overall p < 0.001). HD were significantly older than ARMS (Bonferroni’s correction with N = 6, p < 0.01 vs all groups) and younger than patients with SCZ (Bonferroni’s correction with N = 6, p < 0.001). No difference in age was evident between HD and FEP (Table 1). Significant differences in age were also observed between ARMS vs FEP and ARMS vs SCZ groups (Table 1). Finally, FEP and SCZ groups were comparable both for age at onset (p = 0.569) and olanzapine dose equivalents (p = 0.813) (Table 1).

Distribution of amino acid levels

Levels of AAs or their D-to-total ratios are reported in Table 2 and in Figs. 2 and 3. All AAs, except L-Gln (p = 0.263) (Fig. 3A) and L-Asn (p = 0.126) (Fig. 3D), displayed significant differences across groups (Kruskal–Wallis, overall p < 0.001). In particular, L-Ser was higher in patients with SCZ than all other groups (Bonferroni’s correction with N = 6, p-value ranging from 0.017 to <0.001; +23% vs HD, +11% vs ARMS, +25% vs FEP, respectively) (Fig. 2A), while D-Ser showed significantly higher levels among ARMS than in both HD (Bonferroni’s correction with N = 6, p = 0.003; +31%) and SCZ (p < 0.001; +36%) groups (Fig. 2B). The D-Ser/total-Ser ratio (Fig. 2C) paralleled D-Ser levels, but with SCZ displaying significantly lower ratios than all other groups (−24% vs HD, −36% vs ARMS, −30% vs FEP, respectively). The use of ratio increased the statistical significance of the diagnosis effect probably due to the higher L-Ser, hence total Ser, levels among patients with SCZ compared to other groups (Fig. 2A–C).

Table 2 Concentration of amino acids in the groups investigated.
Full size table
Fig. 2: Levels of AAs and their D-to-total ratios.
figure 2

Distribution of amino acids (L-serine, A; D-serine, B; L-aspartate, D; D-aspartate, E) or their ratio (D-serine/total serine, C; D-aspartate/total aspartate, F) among the groups investigated. Overall and post-hoc comparisons were performed, respectively, by non-parametric Kruskal–Wallis test and Mann–Whitney test with Bonferroni’s correction (N = 6). Only significant p-values are reported. HD Healthy Donors, ARMS At Risk Mental State, FEP First Episode Psychosis, SCZ Schizophrenia.

Full size image
Fig. 3: Levels of AAs and their ratios.
figure 3

Distribution of amino acids (L-glutamine, A; L-glutamate, B; L-asparagine, D; Glycine, E) or their ratio (L-glutamine/L-glutamate, C) among the groups investigated. Overall and post-hoc comparisons were performed, respectively, by non-parametric Kruskal–Wallis test and Mann–Whitney test with Bonferroni’s correction (N = 6). Only significant p-values are reported. To improve the visual appearance of Gly, two extreme values (1 for HD and 1 for ARMS, close to 2000 μM) were omitted. HD Healthy Donors, ARMS At Risk Mental State, FEP First Episode Psychosis, SCZ Schizophrenia.

Full size image

A similar trend was observed for L-Asp, D-Asp and D-Asp/total-Asp ratio (Table 2 and Fig. 2D–F). In particular, similarly to L-Ser, L-Asp was significantly higher in patients with SCZ than all other groups (Bonferroni’s correction with N = 6, p-values ranging from 0.015 to <0.001; +66% vs HD, +108% vs ARMS, +58% vs FEP) (Fig. 2D), while D-Asp was significantly higher among ARMS (+61% vs HD, +45% vs SCZ) and FEP (+83% vs HD, +65% vs SCZ) than HD or SCZ (Fig. 2E). In line with results on single AAs, ARMS and FEP displayed the highest D-Asp/total-Asp ratio (Bonferroni’s correction with N = 6, p < 0.001 vs both HD and SCZ; ARMS, +98% vs HD, +212% vs SCZ, respectively; FEP, +108% vs HD, +228% vs SCZ, respectively) (Fig. 2F). While L-Gln concentrations were not significantly different among groups (Fig. 3A), level of L-Glu was significantly higher in patients with SCZ (Bonferroni’s correction with N = 6, p < 0.001; +213% vs HD, +288% vs ARMS, +189% vs FEP) (Fig. 3B). Thereby, conversely, the L-Gln/L-Glu ratio was significantly lower among SCZ patients than in all other groups (Bonferroni’s correction with N = 6, p < 0.001; −69% vs HD, −73% vs ARMS, −63% vs FEP, respectively) (Fig. 3C).

Similarly, Gly was significantly higher in patients with SCZ (Bonferroni’s correction with N = 6, p-value ranging from <0.001 to 0.006; +28% vs HD, +23% vs ARMS, +32% vs FEP, respectively) (Fig. 3E), while no difference was observed in L-Asn distribution (Fig. 3D). All differences in AA levels (except L-Asn and L-Gln) remained statistically significant after using ANCOVA on natural log-transformed variables to control for the confounding effect of age.

We then performed a comparison analysis of AA levels between male and female individuals within each clinical group. In HD patients, we found significantly higher L-Gln (p < 0.001) and L-Glu levels (p < 0.001), and a lower L-Gln/L-Glu ratio (p < 0.001) in males, compared to females (Table 3A). In ARMS individuals, L-Glu was significantly higher in males than in females (p = 0.029) (Table 3B). No sex-dependent AA level alterations were found in FEP patients (Table 3C). Finally, in patients with SCZ, L-Ser and Gly levels were significantly higher in females compared with males (borderline p = 0.051 and p = 0.014, respectively) (Table 3D).

Table 3 Association between amino acid levels, age, olanzapine equivalents and sex among healthy donors (A), ARMS (B), FEP (C) and SCZ (D) subjects.
Full size table

After, we analyzed the correlation between AA levels and age. In HD individuals, AA levels were not associated with age (Table 3A).

Within the ARMS group, age was positively correlated with L-Glu (rho=0.284, p = 0.053) and negatively correlated with D-Ser (rho = −0.429, p = 0.003), L-Gln/L-Glu ratio (rho = −0.290, p = 0.048) and Gly (rho = −0.307, p = 0.036) (Table 3B).

Among FEP subjects, age was positively correlated with levels of D-Asp (rho = 0.582, p = 0.002) and L-Glu (rho = 0.634, p = 0.001) (Table 3C). Furthermore, similarly to ARMS group, a negative correlation between age and L-Gln/L-Glu ratio (rho = −0.631, p = 0.001) was observed in the FEP group (Table 3C).

Similarly to what observed in the ARMS and FEP groups, in patients with SCZ, age was positively correlated with both D-Asp (rho = 0.256, p = 0.020) and L-Glu (rho = 0.255, p = 0.021) serum levels (Table 3D). Additionally, among patients with SCZ, age positively correlated with D-Ser (rho = 0.223, p = 0.044) (Table 3D).

Finally, in both FEP and SCZ patients, no association was detected between AA levels and olanzapine equivalents (Table 3C and D).

Association between serum amino acid levels, clinical severity and working memory performance

In ARMS and SCZ patients, correlation analyzes revealed that AA levels were not correlated with PANSS scores (Table S1). However, in the FEP group, a moderate positive correlation between PANSS Positive subscale scores and D-Asp levels (rho = 0.677, p = 0.004) was observed (Table S1).

Among HD, serum levels of L-Gln were positively correlated with accuracy as the correct response percentage at the 1-Back task (rho = 0.223, p = 0.026) and negatively correlated with reaction time at the 2-Back task (rho = −0.214, p = 0.033). Among ARMS, levels of D-Ser/total Ser ratio were negatively correlated with the 1-Back reaction time (rho = −0.383, p = 0.019), while levels of L-Glu positively correlated with 1-Back accuracy (rho = −0.334, p = 0.044). Among FEP, no association was found between AA levels and the N-Back scores (Table S2).

Among patients with SCZ, none of the AA levels correlated with the N-Back scores (Table S2).

Discussion

Different lines of research indicate that the modulation of NMDAR is implicated in the early stages of psychosis and chronic SCZ4 and that various neuroactive AAs modulating this class of ionotropic receptors, including Glu, Gly, D-Asp and D-Ser, play a role in psychosis evolution. Here, we explored the hypothesis that distinct changes in the circulating levels of NMDAR-regulating AAs track psychosis phases and impact related symptoms and cognitive deficits.

To test this hypothesis, we compared serum levels of different D- and L-AAs, as well as D-Ser/total Ser and D-Asp/total Asp ratios, among four groups of individuals diagnosed with ARMS, FEP, SCZ and HD. Moreover, we performed an exploratory analysis to investigate the relationship between AA serum profile, psychotic symptom severity and WM performance, assessed with the PANSS and the N-Back task, respectively.

Remarkably, our HPLC results highlight a significant increase in serum D-Asp and D-Ser levels in ARMS compared to HD individuals, indicating an abnormal upregulation of these atypical AAs in the pre-morbid stage of psychosis, and a specific D-Asp increase in FEP compared to both HD and SCZ subjects. Importantly, we found that both D-Ser/total Ser and D-Asp/total Asp ratios significantly increased in ARMS and FEP compared to SCZ and HD individuals, suggesting that such upregulation may be due to higher D-Ser and D-Asp blood levels in these early psychosis stages.

Our data also indicate that the blood levels of L-Ser were prominently upregulated in SCZ compared to HD, ARMS and FEP individuals, suggesting that D-Ser/total Ser ratio reduction was due to an altered Ser enantiomers conversion in SCZ patients. The present finding on L-Ser serum accumulation nicely aligns with previous studies performed on smaller-size samples of patients with SCZ12.

Given the critical importance of SR, DAO, its activator DAOA/G72, and DDO in regulating Ser and Asp L-to-D-enantiomer balance, our findings most likely involve changes in the expression or activity of these enzymes across the progression of psychosis. In line with this assumption, SR has been identified in GWAS as a risk gene for SCZ30 and the SCZ-associated allelic variant has been shown to reduce SR expression55. Accordingly, a previous study has posited the involvement of an SR activity reduction in serum L-to-D-Ser imbalance in patients with SCZ compared to healthy individuals12. Hence, the D-Ser/total Ser ratio decrease observed in SCZ may reflect a reduction of SR-mediated L-to-D-Ser conversion.

Furthermore, a very recent investigation by Camporesi and colleagues confirmed that patients with SCZ, classified as clinically limited-responders and resistant to treatments, exhibit reduced serum levels of SR, while showing a disrupted relationship between D-Ser serum levels and cognitive performance, compared to healthy individuals56. However, since SR activity is counteracted by that of DAO, we cannot rule out that increased DAO-mediated D-Ser degradation may contribute to serum L-Ser accumulation, ultimately resulting in a lower D-Ser/total Ser ratio in SCZ57.

Consistent with our biochemical observations, genetic investigation by Mossner and coll58. reported a DAOA/G72 haplotype associated with susceptibility for progression from prodromal syndromes to First Episode Psychosis. Moreover, Addington and coll59. found that polymorphisms in the 13q33.2 region of G72/G30 genes were associated with childhood-onset SCZ.

Noteworthy, we should consider that luminal-membrane transporters and DAO activity in the intestine and kidney also regulate plasma Ser enantiomer homeostasis60,61, suggesting that these physiological processes may contribute to present observations.

In line with reports implicating NMDAR-mediated glutamatergic alteration in transition from early to full-blown psychosis62,63, and considering the ability of D-Ser and D-Asp to efficiently cross the blood-brain barrier64,65 one possible explanation for our observations is that changes in peripheral metabolism of NMDAR‐related AAs reflect, to some extent, a compensatory response of the organism to an incipient glutamatergic signaling impairment in early stage of disease.

Previous publications suggest that exposure to NMDAR antagonists triggers neuroinflammation responses, resulting in psychosis-related brain structural changes66. Accordingly, increased plasma concentrations of the inflammatory mediator interleukin 6 have been reported in ARMS and AP naïve FEP individuals67, while a potential anti-inflammatory action of NMDAR-stimulating AAs was described68,69,70. This evidence suggests that the increased serum D-Ser and D-Asp levels we observed in ARMS and FEP individuals may be a biological correlate of psychosis onset via modulation of the pro-vs-anti-inflammatory balance. Consistent with our hypothesis, preclinical studies also showed that NMDAR co-agonists D-Ser- and Gly-transport inhibitors can reverse brain changes typically associated with the onset of psychosis71.

Besides D-AAs, our results also confirmed previous findings indicating SCZ-related variation of excitatory L-AA homeostasis13,14,15,17,18,19,21,27, as we found increased serum levels of L-Glu and L-Asp in full-blown SCZ patients who were undergoing treatment, compared to individuals with ARMS, FEP patients and HD individuals. As a result of increased L-Glu levels, we found a reduced L-Gln-to-L-Glu ratio in patients with full-blown SCZ, compared with both HD and ARMS/FEP individuals. In addition, our data aligns with a previous study that showed a decreased Gln/Glu ratio in patients with SCZ compared to healthy subjects43.

Given the relevance of these excitatory AAs in regulating neurotransmission at central glutamatergic synapse, future research is essential to determine whether the reported peripheral changes in the L-Gln/L-Glu ratio also reflect variations in SCZ brains, as previously observed in patients with autism and healthy volunteers72,73.

Here, we also highlighted greater serum levels of the NMDAR co-agonist Gly in patients with SCZ, compared to HD, ARMS and FEP individuals. Even though modifications in brain Gly-mediated NMDAR signaling can occur early in the psychosis spectrum39,74,75, our results indicate that Gly dysmetabolism may become peripherally relevant only after SCZ clinical full onset.

Despite the exploratory nature of our clinical investigation, we discovered that D- and L-AA serum levels were quite weakly correlated with psychosis symptom severity and WM performance.

In particular, limited to FEP individuals, serum levels of D-Asp showed a relatively moderate correlation with the PANSS Positive subscale scores. Conversely, no significant association was observed between serum neuroactive AA levels and WM measures in the same group of individuals. In the ARMS subsample, D-Ser/total Ser ratio was inversely correlated with reaction time on the 1-Back task while L-Glu levels positively correlated with this clinical parameter. Finally, in HD individuals, serum levels of L-Gln were positively correlated with accuracy on the 1-Back task and negatively correlated with reaction time on the 2-Back task.

Overall, our HPLC results revealed distinct variations in circulating AAs acting on NMDAR signaling in specific phases of psychosis that do not fully reflect more distal clinical and behavioral phenotypes, possibly due to a non-linear relationship between metabolic changes involving the body periphery and brain molecular events that eventually lead to psychosis onset. In this perspective, the positive correlation between D-Asp serum levels and PANSS Positive subscale scores in FEP but not ARMS individuals may suggest that D-Asp serum increase, while already detectable in psychosis at-risk-mental-states, becomes clinically relevant only after psychosis clinical onset.

Interestingly, such an increase is no longer observable in full-blown SCZ, where D-Asp levels return to be unrelated to PANSS scores.

Concerning the relationship between AA serum levels and WM performance, our observation of a coherent correlation between L-Gln levels and performance in both 1- and 2-Back tasks in the HD group confirms previous reports about the physiological role of Gln homeostasis in cognition76.

Nonetheless, we found that correlations between AA serum levels and N-Back task performance in pre-morbid and clinical subsamples were relatively limited or completely absent, possibly indicating that peripheral AA modifications may only moderately reflect behavioral alterations.

We are aware that a confounding factor limiting the interpretation of our results is that AA variations are most likely affected by nutrition and biochemical processes involving peripheral organ metabolism, while psychosis itself is a complex brain-routed phenomenon. However, from a system-biology perspective, considering the emerging role of continuous brain-body-microbiota crosstalk77, it is plausible that the neuroactive AA serum changes identified are related to CNS-body-periphery molecular interaction supporting the pathophysiology of psychosis. Within this perspective, future research should take into account system-level variables potentially impacting peripheral neuroactive AA metabolism, including kidney and liver function, gut microbiota composition and dietary factors. In this regard, the timing of the last meal before blood sampling on blood AA levels should be considered of critical relevance78,79,80,81.

A major strength point of our study is that we ran an inclusive HPLC analysis on main NMDAR-modulating D- and L-AAs and their precursors in over 250 individuals, therefore presenting the first large-scale serum AA profiling in psychosis staging. From this perspective, our findings offer an important contribution to the future identification of biomarkers for the early detection and staging of psychosis.