Over-expression of XIST, the Master Gene for X Chromosome Inactivation, in Females With Major Affective Disorders

Abstract

Background: Psychiatric disorders are common mental disorders without a pathological biomarker. Classic genetic studies found that an extra X chromosome frequently causes psychiatric symptoms in patients with either Klinefelter syndrome (XXY) or Triple X syndrome (XXX). Over-dosage of some X-linked escapee genes was suggested to cause psychiatric disorders. However, relevance of these rare genetic diseases to the pathogenesis of psychiatric disorders in the general population of psychiatric patients is unknown.

Methods: XIST and several X-linked genes were studied in 36 lymphoblastoid cell lines from healthy females and 60 lymphoblastoid cell lines from female patients with either bipolar disorder or recurrent major depression. XIST and KDM5C expression was also quantified in 48 RNA samples from postmortem human brains of healthy female controls and female psychiatric patients.

Findings: We found that the XIST gene, a master in control of X chromosome inactivation (XCI), is significantly over-expressed (p = 1 × 10(- 7), corrected after multiple comparisons) in the lymphoblastoid cells of female patients with either bipolar disorder or major depression. The X-linked escapee gene KDM5C also displays significant up-regulation (p = 5.3 × 10(- 7), corrected after multiple comparisons) in the patients' cells. Expression of XIST and KDM5C is highly correlated (Pearson's coefficient, r = 0.78, p = 1.3 × 10(- 13)). Studies on human postmortem brains supported over-expression of the XIST gene in female psychiatric patients.

Interpretations: We propose that over-expression of XIST may cause or result from subtle alteration of XCI, which up-regulates the expression of some X-linked escapee genes including KDM5C. Over-expression of X-linked genes could be a common mechanism for the development of psychiatric disorders between patients with those rare genetic diseases and the general population of female psychiatric patients with XIST over-expression. Our studies suggest that XIST and KDM5C expression could be used as a biological marker for diagnosis of psychiatric disorders in a significantly large subset of female patients.

Research in context: Due to lack of biological markers, diagnosis and treatment of psychiatric disorders are subjective. There is utmost urgency to identify biomarkers for clinics, research, and drug development. We found that XIST and KDM5C gene expression may be used as a biological marker for diagnosis of major affective disorders in a significantly large subset of female patients from the general population. Our studies show that over-expression of XIST and some X-linked escapee genes may be a common mechanism for development of psychiatric disorders between the patients with rare genetic diseases (XXY or XXX) and the general population of female psychiatric patients.

Keywords: KDM5C; Major affective disorders; X chromosome inactivation; X-linked escapee genes; XIST.

Fig. S1

Variation of protein translation activity in patients' lymphoblastoid cells. (A) Protein translation activity was measured in individual lymphoblastoid cell lines using SUnSET. Sample loading was shown by β-actin protein expression after stripping the membrane. C1–C26: healthy European Caucasian controls; B1–B28: European Caucasian bipolar patients with mania and psychosis. (B) SUnSET intensity of individual samples was normalized against β-actin. To avoid differential overall intensity between the gels, the intensity of each sample is presented as a percentage of the average intensity of all samples from the same gel. A significantly larger variation of protein translation activity was observed in the patients than in the controls (p < 0.01, F-test). Each dot represents a human subject (black = healthy controls; red = patients). CTRL: controls; BP: bipolar patients with mania and psychosis. (C) Comparison of intracellular puromycin concentrations. Two cell lines (B25 and B27) with a marked difference in SUnSET activity were selected. Equal amounts of puromycin-labeled protein were loaded on each lane. B25 and B27 cell lysates were used to block interaction between membrane-bound puromycin-labeled proteins and anti-puromycin antibodies in the Western blot blocking solution. Western blot signal was gradually decreased after adding an increasing amount of cell lysate to the blocking solution. There was no difference between the blocking activities of B25 and B27 cell lysate, indicating that there was a comparable concentration of intracellular puromycin between the two cell lines. (D) There was no correlation between protein translation activity (SUnSET) and individuals' age. Each dot represents a human subject. (E) SUnSET data were analyzed according to gender. There was no difference between male controls and male patients. A significantly larger variation of protein translation activity was observed in female patients (p < 0.01, F-test). Each dot represents a human subject (black = healthy controls; red = patients). CTRL: controls; BP: bipolar patients with mania and psychosis.

Fig. S2

Expression of X-linked genes in patients' lymphoblastoid cell lines. Each dot represents a human subject (black = healthy controls; red = patients). A trend of high expression in PGK1 (A)(t(23) = − 1.84, p < 0.1), G6PD (B)(t(23) = − 1.89, p < 0.1) in the patients before correction of multiple comparisons. There was no significant difference in HPRT1 (C)(t(23) = − 1.62, ns) and RPS4X (D)(t(23) = − 0.57, ns) expression between the controls and the patients. (E) When GAPDH was used as the reference control, a significant over-expression of XIST was also observed (t(23) = − 2.61, p = 0.016).

Fig. S3

No effect of CLZ and VPA treatment on expression of XIST, TSIX, KDM5C and KDM6A genes. Clozapine (CLZ) and valproic acid (VPA) were dissolved in methanol. Both control and patient lymphoblastoid cells were cultured in the presence of CLZ and VPA at the final concentration of 0.5 μg/ml and 50 μg/ml respectively. After 72 h incubation, CLZ and VPA were removed. Gene expression analysis was conducted in cells after 4 day recovery in normal culture medium. There was no effect of drug treatment on expression of XIST (A), TSIX (B), KDM5C (C), and KDM6A (D) genes. Error bar: SEM.

Fig. S4

No alteration of gene expression on the X chromosome in the male patients with mania and psychosis. Each dot represents a human subject (black = healthy controls; red = patients). The same set of genes was analyzed in the lymphoblastoid cells of the male patients. There was no difference in expression of KDM5C (A), KDM6A (B), G6PD (C), PGK1 (D), HPRT1 (E), and RPS4X (F) genes.

Fig. S5

No correlation between age, XIST, and protein translation activity. Each dot represents a human subject (black = healthy controls; red = patients with mania and psychosis). There was no age effect on either XIST (A) or KDM5C (B) expression. (C) There was no correlation between XIST expression and protein translation activity (SUnSET). (D) A weak positive correlation was detected between KDM5C expression and protein translation activity in the lymphoblastoid cells (Pearson's coefficient, r = 0.32, p = 0.12).

Fig. S6

X-linked gene expression and X chromosome DNA. Each dot represents a human subject (black = healthy controls; red = patients with recurrent major depression). There was no difference in expression of PGK1 (A) and G6PD (B) genes between the controls and the patients. (C) A significantly higher KDM6A (t(34) = − 3.18, p < 0.05) expression was detected in the female controls with mixed ethnic backgrounds compared with the European female Caucasian controls. (D) There was a weak correlation between KDM6A RNA expressions between different batches of cells (Pearson's coefficient, r = 0.43, p = 0.04). (E) To rule out that patients with high XIST expression may have triple XXX chromosomes, we quantified XIST genomic copies relative to β-actin genomic copies on chromosome 7 in several controls and the patients with high XIST expression. There was no difference in relative amount of XIST genomic copies between the controls and the patients.

Fig. 1

Abnormal expression of XIST and its regulator genes involved in XCI. Each dot represents a human subject. Black = healthy European Caucasian female controls (CTRL); red = European Caucasian female bipolar (BP) patients with mania and psychosis. Multiple comparisons of Student's t-tests were corrected with FDR. (A) A significantly higher level of XIST RNA expression was observed in the patients than in the controls (t(23) = − 4.23, p = 0.001). (B) Consistent with its negative role, there was a significantly lower level of TSIX expression in the patients (t(23) = 3.43, p < 0.01). (C) Consistent with its positive role, a trend of high expression of FTX was detected in the patients (t(23) = − 2.00, p < 0.1). (D) No difference was observed in JPX expression between the controls and patients (t(23) = − 0.24, ns). (E) A significantly higher level of KDM5C expression (t(23) = − 2.89, p < 0.05) was observed in the patients. (F) There was a trend of high expression in KDM6A (t(23) = − 2.00, p < 0.1) in the patients before correction of multiple comparisons. (***p < 0.001, **p < 0.01, *p < 0.05, # p < 0.1).

Fig. 2

Over-expression of XIST, KDM5C, and KDM6A in patients with recurrent major depression. Each dot represents a human subject. Black = healthy European Caucasian female controls (CTRL); red = European Caucasian female patients with recurrent major depression (MDR); gray = healthy female controls with various ethnic genetic backgrounds (mixed CTRL). Multiple comparisons of Student's t-tests were corrected with FDR. (A) A significantly larger variation was observed in XIST expression in the patients group (p = 0.001, F-test). There was also a significantly higher level of XIST expression (t(11) = − 4.16, p < 0.01, unequal variances) in the patients than in the controls. (B) Down-regulation of TSIX expression was not statistically significant (t(21) = 1.35, ns). Larger variation was observed in either KDM5C (p = 0.027, F-test) or KDM6A (p = 5.9 × 10^− 5, F-test) gene expression in the patients. There was also significantly higher expression of KDM5C (C)(t(13) = − 4.45, p < 0.001, unequal variances) and KDM6A (D)(t(10) = − 2.23, p = 0.05, unequal variances) in the patients lymphoblastoid cells. There is no difference in XIST (E) or KDM5C (F) expression between healthy European Caucasian female controls and healthy female controls with various ethnic genetic backgrounds. (**p < 0.01, *p < 0.05).

Fig. 3

XIST and KDM5C over-expression in additional groups of patients with bipolar disorder or recurrent major depression. Each dot represents a human subject. Black = healthy European Caucasian female controls (CTRL); red = patients with various ethnic backgrounds with either bipolar (mixed BP) or recurrent major depression (mixed MDR). Multiple comparisons of Student's t-tests were corrected with FDR. (A) Significantly larger variations were observed in XIST expression in either bipolar (p = 0.0038, F-test) or major depression (p = 0.0009, F-test). There was significantly higher XIST expression (t(33) = − 5.1, p < 0.00001, unequal variances) in the bipolar patients than in the controls. The mean of XIST expression of major depression group was not significantly different from the mean of the control group. (B) There is no difference in the variation of KDM5C expression between groups. Significantly higher expression of KDM5C was observed in both bipolar (t(34) = − 2.71, p < 0.01) and major depression (t(24) = − 2.84, p < 0.01) than in the controls. (C) Expression of XIST in all individuals was normalized against the mean of the European female controls that were co-cultured. After combining all data, significantly larger variation was observed in XIST expression in bipolar (p = 8 × 10^− 5, F-test), major depression (p = 1 × 10^− 10, F-test), and their combined group (p = 1 × 10^− 8, F-test). In addition to the difference in variation, there was significantly higher XIST expression in bipolar (t(52) = − 6.5, p = 1 × 10^− 7, unequal variances), major depression (t(26) = − 2.86, p = 0.004, unequal variances), and their combined group (t(83) = − 5.93, p = 1 × 10^− 7, unequal variances). (D) Significantly larger variation was found in KDM5C expression in bipolar (p = 0.002, F-test), major depression (p = 0.004, F-test), and their combined group (p = 0.001, F-test). Significantly higher KDM5C expression was observed in bipolar (t(58) = − 4.9, p = 5.4 × 10^− 6, unequal variances), major depression (t(34) = − 3.3, p = 0.001, unequal variances), and their combined group (t(94) = − 5.4, p = 5.3 × 10^− 7, unequal variances). (***p < 0.001, **p < 0.01).

Fig. 4

Over-expression of XIST and KDM5C as a potentially combinatorial biological marker for diagnosis of major psychiatric disorders. Each dot represents a human subject. Black = healthy female controls; red = female patients with either bipolar or recurrent major depression. (A) Expression of XIST and KDM5C is highly correlated in all 60 patients' lymphoblastoid cell lines (Pearson's coefficient, r = 0.78, p = 1.3 × 10^− 13). (B) Expression of XIST and KDM5C of the 36 controls' and 60 patients' lymphoblastoid cell lines are plotted together. Reference ranges of XIST and KDM5C expression were calculated from the 36 controls. M + 2SD: the dashed line corresponding to 2 standard deviations above the means of XIST (shaded with light blue) and KDM5C (shaded with light green) from the 36 controls. A significantly large number of female patients randomly collected from the general population displayed higher expression of either XIST and/or KDM5C above their reference ranges.

Fig. 5

XIST and KDM5C expression across passages and alteration of epigenetic modifications in the patients. Each dot represents a human subject. Black = healthy European Caucasian female controls (CTRL); red = European Caucasian female patients with recurrent major depression (MDR). XIST (A) and KDM5C (B) gene expression was examined in the first batch of lymphoblastoid cells (XIST-1 and KDM5C-1). After more than a month of continuous cell culture, the same set of genes was examined again in the second batch of the same cell lines with different cell passages (XIST-2 and KDM5C-2). A high correlation of XIST (Pearson's coefficient, r = 0.88, p = 2.6 × 10^− 8) and KDM5C (Pearson's coefficient, r = 0.6, p = 0.0028) expression was observed between the two batches of cells with different passages. (C) Western blot analyses of KDM5C protein expression in the lymphoblastoid cells of female patients with recurrent major depression. A single band at 180 kD, the calculated size of human KDM5C protein, was detected. C: controls; D: recurrent major depression. β-actin was used as an internal control for normalization. (D) Consistent with increased mRNA expression, significantly higher KDM5C protein expression was found in the patients (t(22) = − 2.85, p < 0.01). (E) Lymphoblastoid cell lines from 4 female controls (black) and 4 female patients (red) with either recurrent major depression or mania and psychosis were selected for chromatin immunoprecipitation experiments. Six sites, separated by ~ 1 kb (except a 2 kb between 1 and 2) around the promoter of XIST gene, were examined by ChIP-QPCR. Multiple comparisons of Student's t-tests were corrected with FDR. Significantly more H3K27me3 was observed in the patients' lymphoblastoid cells at XIST (1) (t(6) = − 3.83, p < 0.05) and XIST (3) (t(6) = − 3.2, p < 0.05). A trend of more H3K27me3 at XIST (4) (t(6) = − 2.7, p < 0.1) was also observed. (F) Two sites around the promoter of KDM5C gene were also examined by CHIP-QPCR. H3K27me3 is significantly more enriched at KDM5C (1) (t(6) = − 3.39, p < 0.05) in female patients' lymphoblastoid cells.

Fig. 6

Over-expression of XIST in postmortem brains of female patients with major psychiatric disorders. All 48 female RNA samples from postmortem human brains were kindly provided by Stanley Medical Research Institute. Expression of XIST and KDM5C was double-blindly quantified. Each dot represents a human subject. Gray = all subjects; black = healthy female controls (CTRL); red = female patients with bipolar disorder (BP), recurrent major depression (MDR), and schizophrenia (SCZ). (A) To assess cell heterogeneity of the samples, GFAP expression was quantified to evaluate relative abundance of glial cells in the samples. 2 samples displayed high levels of GFAP expression above 3 interquartile ranges (IQR) above the upper quartile using boxplot analysis. Therefore, both were determined as extreme outliers and excluded for further analyses. (B) There was one control sample with a very high level of XIST expression that is qualified as an outlier. Due to small sample size, it was retained. There was significantly higher XIST expression in bipolar disorder (t(24) = − 2.07, p = 0.025), and a trend of high XIST in major depression (t(20) = − 1.35, p = 0.096). Significantly higher XIST expression was also detected when bipolar disorder and major depression were combined or when all patients were combined. (C) There was no significant difference in KDM5C expression between the control group and any patient group. (D) There was no correlation between XIST and KDM5C expression in the postmortem brains of the patients.