Levels of Sphingosine-1-phosphate Were Lowered in the Corpus Callosum of Patients With Schizophrenia
The study design is shown in Supplementary Figure S1. We first analyzed sphingolipids in the corpus callosum (white matter) and BA8 (cortical area) of postmortem brain samples from patients with schizophrenia and controls using the LC-ESI-MS/MS method. We detected significantly lower levels of S1P, 1 of the base-form species of sphingolipids, in the corpus callosum of subject with schizophrenia than in the controls (31% reduction, P = .009) (Figure 1B). By contrast, the levels of other base-form species of sphingolipids (sphinganine SA, SO, 1-deoxy-sphinganine, and 1-deoxymethyl-sphinganine) did not differ between the schizophrenia and control groups (Figures 1B and 1C). We also analyzed fatty-acid-acylated forms of sphingolipids (dihydroceramide, Cer, 1-deoxy-dihydroceramide, 1-deoxy-ceramide, 1-deoxymethyl-dihydroceramide, 1-deoxymethyl-ceramide [doxmeCer], lactosylceramide, hexosylceramide, and SM) (Figure 1A). In the corpus callosum, none of the acylated forms of sphingolipids differed between the schizophrenia and control groups (Table 2 and Supplementary Table S6). Although higher levels of Cer and doxmeCer, which contain fatty acids of various chain lengths, were observed in the BA8 of patients with schizophrenia than in that of the controls, the levels of the other fatty-acid-acylated species were unchanged (Table 2 and Supplementary Table S7).
Based on the results of lowered levels of S1P in the corpus callosum of patients with schizophrenia when compared with the controls, we examined S1P levels in the samples of major depressive disorder and bipolar disorder. However, we detected no significant differences of S1P levels in the corpus callosum and BA8 between patients with major depressive disorder or bipolar disorder and controls (Supplementary Figure S3), indicating that the dysregulation of S1P level is specific to the corpus callosum of patients with schizophrenia.
An elevation in the ratio of the sum of SO and Cer to S1P induces apoptosis, cell-cycle arrest, and the suppression of cell survival and cell proliferation. We found the ratio displayed a higher trend in the corpus callosum from patients with schizophrenia than in that of controls (Figure 1D).
Lowered S1P Levels in the Corpus Callosum Were not Influenced by Antipsychotic Intake
Since altered levels of S1P, Cer, and doxmeCer were observed in the brain samples of patients with schizophrenia, we further tested the correlation between each potential confounding factor (pH, age, postmortem interval, drug dose, illness duration) and S1P levels in the schizophrenia samples and the pooled samples (control + schizophrenia). We did not find any significant correlations between the confounding factors and the levels of Cer or doxmeCer in the BA8 (Supplementary Figures S4A and S4B). There was no significant correlation (P = .054) between drug dose (chlorpromazine equivalents) and S1P levels in the schizophrenia group (n = 15). According to the clinical information, 4 of the patients with schizophrenia were not on medication (Supplementary Table S1). When patients with schizophrenia (n = 11), excluding those not taking antipsychotics, were analyzed, a positive correlation between S1P level and drug dose was detected (P = .004) (Supplementary Figure S4C). However, we cannot rule out a history of antipsychotic intake during the course of disease. Collectively, antipsychotic intake is unlikely to explain the lowered levels of S1P in the patients with schizophrenia.
To further evaluate the effects of antipsychotic drugs on S1P content in the brain, we analyzed sphingolipid levels in the frontal cortex and corpus callosum of mice administered haloperidol or risperidone for 4 weeks. There were no changes in the levels of S1P in the corpus callosum and frontal cortex upon administration of haloperidol or risperidone (Figure 1E). The injection of risperidone elicited a significant reduction in SO and a trend of decreasing SA content in the corpus callosum, whereas SO and SA levels were unchanged in the frontal cortex (Figure 1E). Thus, lowered S1P levels in the corpus callosum of patients with schizophrenia are unlikely to result from drug administration, though we cannot completely exclude the possibility of antipsychotic drug effects, including those of long-term exposure, on the alteration of S1P content in patients with schizophrenia.
Expression of Genes for S1P-degrading Enzymes is Upregulated in the Corpus Callosum of Patients With Schizophrenia
To examine the underpinnings for the lowered S1P levels in the corpus callosum of patients with schizophrenia, we analyzed the expression of genes involved in the metabolism of S1P (Figure 2A) using the expanded sample set 2 (Supplementary Table S5). The outline of gene expression analysis is shown in Supplementary Figure S2. The expression levels of Sphingosine-1-Phosphate Lyase 1 (SGPL1) and Phospholipid Phosphatase 3 (PLPP3) showed a significantly higher (P = .006) and an elevated trend (P = .09), respectively, in the corpus callosum of patients with schizophrenia compared with controls (Figure 2B). We further examined the correlation of SGPL1 or PLPP3 expression level with RNA integrity number (RIN), as the RIN in the corpus callosum of patients with schizophrenia differed significantly from that of the controls. There was no significant correlation between RIN values and the expression of SGPL1 or PLPP3 (Supplementary Figure S5).
Expression analysis of sphingolipid metabolism-related genes in postmortem human corpus callosum. (A) Metabolic pathway of sphingosine-1-phosphate (S1P). S1P is synthesized from sphingosine or sphingosylphosphorylcholine by sphingosine kinase (encoded by SPHK1 and SPHK2)38 or autotaxin (ENPP2),39 respectively, and degraded to sphingosine or hexadecenal and phosphoethanolamine by S1P phosphatase (SGPP1 and SGPP2) and lipid phosphate phosphatase (PLPP1, PLPP2, and PLPP3) or S1P lyase (SGPL1), respectively.38 (B) Transcript expression levels of the genes for S1P-metabolizing enzymes in the corpus callosum of patients with schizophrenia (n = 91) and controls (n = 90). Data were normalized with the geometric mean of the 2 internal control genes (GAPDH and B2M) and are represented as the mean ± SEM. **P < .01. Differences between 2 groups were analyzed by Mann-Whitney U test. (C) Correlations between S1P levels and transcript expression levels of SGPL1 and PLPP3 in the corpus callosum of patients with schizophrenia and controls (Spearman's rank correlation coefficient). (D) Absolute quantification of expression levels of SGPL1 and PLPP3 in the corpus callosum of postmortem human brains (control, n = 6) by digital PCR. Data are represented as the mean ± SEM.
The S1P content and PLPP3 expression level displayed a trend toward negative correlation in the corpus callosum (P = .085), while S1P content and SGPL1 expression were not correlated (Figure 2C). Interestingly, the absolute transcript expression level of PLPP3 was approximately 5 times higher than that of SGPL1 in the corpus callosum (control samples) (Figure 2D). This suggests that, rather than by increased SGPL1 expression, an S1P-degrading process augmented by elevated PLPP3 expression may, at least in part, be responsible for the lowered S1P content in the corpus callosum of patients with schizophrenia.
In the BA8, the transcript expression level of SGPL1 or PLPP3 was also significantly higher in the schizophrenia group than in the controls (Supplementary Figure S6A). The expression level of SGPL1 or PLPP3 exhibited no correlation with S1P content but showed a negative correlation with the RIN (Supplementary Figures S6B and S6C); RIN values were lower in patients with schizophrenia than in controls (Supplementary Table S5). Therefore, a higher expression of SGPL1 and PLPP3 in the BA8 of patients with schizophrenia may have stemmed from a lower RIN or associated phenomena.
Expression of Genes for Sphingosine-1-phosphate Receptors Is Elevated in the Corpus Callosum of Patients With Schizophrenia
We then explored whether a compensatory mechanism is evoked following the lowered S1P levels for the upregulation of genes for S1P receptors. There are 5 S1P receptor genes (S1PR1-5), and 4 among them (S1PR1, S1PR2, S1PR3, and S1PR5) are expressed in the central nervous system (CNS). Digital PCR analysis showed that absolute transcript expression of S1PR1/S1pr1 and S1PR5/S1pr5 was abundant in the corpus callosum and frontal cortex of both humans and mice compared with that of the other receptor subtype genes (Figure 3A). Consistent with a previous literature, the expression of S1PR4/S1pr4 was not detectable in our analysis. In humans, S1PR5 expression was highest in the corpus callosum, while S1PR1 expression was highest in the BA8 (Figure 3A). In mice, S1pr1 showed the highest expression in both the corpus callosum and the frontal cortex (Figure 3A).
Transcript expression levels of the genes coding for S1P receptors. (A) Absolute quantification of the expression levels of genes for S1P receptors (S1PR/S1pr1–5) in postmortem human (control, n = 6) and mouse brains (C57/BL6J, n = 3) by digital PCR. (B) Transcript expression levels for S1P receptors in the corpus callosum of patients with schizophrenia (n = 91) and controls (n = 90). The data were normalized with the geometric mean of the 2 internal control genes (GAPDH and B2M) and are represented as the mean ± SEM. *P < .05. Difference between 2 groups was analyzed by Mann-Whitney U test. (C) Correlation between S1P content and transcript expression levels for S1PR1 and S1PR5 in the corpus callosum of patients with schizophrenia and controls (Spearman's rank correlation coefficient). CC, corpus callosum; BA8, Brodmann area 8; FC, frontal cortex.
Next, we examined the gene expression levels of 4 S1P receptors in the human corpus callosum samples (Supplementary Figure S2). Expression of both S1PR1 and S1PR5 was also significantly higher in patients with schizophrenia than in controls (P = .02 and P = .03, respectively, Figure 3B). We found a negative correlation between S1PR5 expression level and the RIN in the corpus callosum (Supplementary Figure S5), indicating that a lower RIN or associated phenomena may contribute to elevated S1PR5 expression in the corpus callosum of patients with schizophrenia. The S1PR1 expression level was positively correlated with the RIN in the corpus callosum. However, the meaning of this weak correlation is unknown, because S1PR1 expression was higher in patients with schizophrenia with lower RIN values (Supplementary Figure S5). Importantly, a negative correlation was observed between S1PR1 expression level and S1P content in the corpus callosum (Figure 3C). These results suggest that the higher expression of S1PR1 in the corpus callosum of patients with schizophrenia might have stemmed from the lowered S1P content resulting from a compensatory mechanism.
In the BA8, higher expression of S1PR1, but not S1PR5, was found in patients with schizophrenia than in controls (Supplementary Figure S6A). The transcript expression level of S1PR1 showed no correlation with S1P content and displayed a negative correlation with the RIN (Supplementary Figures S6B and S6C). Therefore, lower RIN values or associated phenomena may be involved in higher S1PR1 expression in the BA8 of patients with schizophrenia.
Schizophr Bull. 2020;46(5):1172-1181. © 2020 Oxford University Press