Sex dimorphism controls dysbindin-related cognitive dysfunctions in mice and humans with the contribution of COMT

Abstract

Cognitive dysfunctions are core-enduring symptoms of schizophrenia, with important sex-related differences. Genetic variants of the DTBPN1 gene associated with reduced dysbindin-1 protein (Dys) expression negatively impact cognitive functions in schizophrenia through a functional epistatic interaction with Catechol-O-methyltransferase (COMT). Dys is involved in the trafficking of dopaminergic receptors, crucial for prefrontal cortex (PFC) signaling regulation. Moreover, dopamine signaling is modulated by estrogens via inhibition of COMT expression. We hypothesized a sex dimorphism in Dys-related cognitive functions dependent on COMT and estrogen levels. Our multidisciplinary approach combined behavioral-molecular findings on genetically modified mice, human postmortem Dys expression data, and in vivo fMRI during a working memory task performance. We found cognitive impairments in male mice related to genetic variants characterized by reduced Dys protein expression (p_Bonferroni = 0.0001), as well as in male humans through a COMT/Dys functional epistatic interaction involving PFC brain activity during working memory (t(23) = -3.21; p_FDR = 0.004). Dorsolateral PFC activity was associated with lower working memory performance in males only (p = 0.04). Also, male humans showed decreased Dys expression in dorsolateral PFC during adulthood (p_FDR = 0.05). Female Dys mice showed preserved cognitive performances with deficits only with a lack of estrogen tested in an ovariectomy model (p_Bonferroni = 0.0001), suggesting that genetic variants reducing Dys protein expression could probably become functional in females when the protective effect of estrogens is attenuated, i.e., during menopause. Overall, our results show the differential impact of functional variants of the DTBPN1 gene interacting with COMT on cognitive functions across sexes in mice and humans, underlying the importance of considering sex as a target for patient stratification and precision medicine in schizophrenia.

Fig. 1. Dysbindin affects the cognitive functions of mice tested in the TOR test in a sex-dependent manner.

A Cartoon depicting the behavioral task Temporal Order recognition. B Discrimination Ratio (DR). 2-Way ANOVA Gender X Genotype Interaction F(1,106) = 9.58, **p = 0.0025, Genotype effect F(1,106) = 13.20 ***p = 0.000, Gender effect F(1,106) = 0.082; p = 0.7746; Bonferroni’s multiple comparison test; p > 0.999 vs. Dys + /− female, **p = 0.0018. vs. Dys + /− male. Dys + /+ female n = 32; male: n = 12, Dys + /− female n = 35 male n = 9. C Total Exploration in seconds, Dys + /+ female n = 32; male: n = 12, Dys + /− female n = 35 male n = 9. 2-Way ANOVA did not reveal any difference between genotype (F(1,106) = 0.688) or gender (F(1, 106) = 0.7906. D Discrimination ratio (DR) analyzed according to the four phases of mice estrous cycle: Proestrus (P), Estrus (E), Metestrus (M) and Diestrus (D). Two-Way ANOVA revealed an estrous phase effect F (3,81) = 23.22, p < 0.0001 but not a genotype effect F(1,81 = 0.03564, p = 0.8507 not interaction F(3,81) = 0.5134, p = 0.6742. Bonferroni’s Multiple Comparison Test: **p < 0.01, ***p < 0.001. (Dys + /+: P n = 6, E n = 18, M n = 13, D n = 6; Dys + /−: P n = 8, E n = 15, M n = 12, D n = 11). E Total exploration in seconds, Two-way ANOVA: Estrous cycle phase F(3,81) = 1.252, p = 0.2965; Genotype F(1, 81 = 0.03240, p = 0.8576; Estrous cycle phase X Genotype F(3,81) = 0.3488, p = 0.7901; Dys + /+: P n = 6, E n = 18,M n = 13, D n = 6; Dys + /−: P n = 8, E n = 15, M n = 12, D n = 11). F Photos depicting the four phases of mice estrous cycle assessed immediately after the TOR test. G DR analyzed as “estrus” (E) and “non-estrus” (NE) phases. 2Way-ANOVA: estrous.cycle effect F(1,85) = 69.9 ***p < 0.0001, Genotype effect F(1,85) = 0.0822 p = 0.7749, Interaction E.phaseXGenotype F(1,85) = 0.0937 p = 0.7602. Bonferroni’s multiple comparison tests: ***< 0.0001 Dys + /+ E vs.NE, ***p < 0.0001 Dys + /− E vs. NE. H Total exploration time. No effect of genotype or estrous cycle phase affected the total exploration time. 2 Way-ANOVA estrous cycle effect F(1,85) = 1,571 p = 0.2135, Genotype effect F(1,85) = 0.051 p = 0.822, Interaction F(1,85) = 0.0036 p = 0.9525.

Fig. 2. Ovariectomy produces cognitive impairments in female Dys + /− mice and 17ß-estradiol rescues Dys-dependent cognitive deficits in both Dys + /− ovariectomized female and Dys + /− male mice through the COMT contribution.

A Timeline depicting the experimental procedure. B Temporal Order Recognition test. 2-way ANOVA analysis revealed a genotype effect F (1, 51) = 9.690, **p < 0.003, and a main effect of interaction surgeryXgenotype F (1,51) = 9.674, **p = 0.0031 but not a main surgery effect F (1, 51) = 3.425, p = 0.07. Bonferroni’s Multiple Comparison tests. *p = 0.011 Dys + /− Sham vs. Dys + /− OVX; ***p = 0.0001 Dys + /+ OVX vs Dys + /− OVX. C Total exploration in seconds. 2-way ANOVA, Genotype F (1,51) = 0.5846 p = 0.4480; Surgery F (1,51) = 2.06, p = 0.1573; SurgeryXGenotype F (1,51) = 0.3675, p = 0.547. Dys + /+ n = 31 (Sham n = 15, OVX n = 16), Dys + /− n = 25 (Sham n = 11, OVX n = 14). D COMT protein level alteration in the mPFC of Sham and OVX mice. 2-Way ANOVA revealed a main genotype effect F(1, 17) = 32.19 ***p < 0.0001, surgery effect F(1,17) = 14.67 **p = 0.0013 and a GenotypeXSurgery effect F(1, 17) = 7.225 *p = 0.0156. Bonferroni’s multiple comparison test ***p < 0.0001 vs Dys + /+ OVX, ^###p = 0.0007 vs Dys + /− Sham (Dys + /+: OVX n = 4, Sham n = 5; Dys + /− OVX n = 7, Sham n = 5). E Temporal order recognition test in Veh or E2-treated OVX mice. 2Way-ANOVA analysis showed a Treatment effect F(1,30) = 11.73 **p = 0.0018 and a Genotype effect F(1,30) = 12.74 **p = 0.0012 as well as the interaction GenotypeXTreatment F(1,30) = 22.99 ***p < 0.0001. Bonferroni’s multiple comparisons test: ***p < 0.0001 Dys + /− Veh vs OVX, ***p < 0.0001 Dys + /+ Veh vs Dys + /− Veh. (Dys + /+ Veh n = 8, E2 n = 9; Dys + /− Veh n = 6, E2 n = 9). F Total exploration time. 2Way ANOVA showed no effects of E2 treatment in the total exploration time. 2Way-ANOVA Treatment effect F(1,30) = 0.8988 p = 0.3507, Genotype effect F(1,30) = 3.429 p = 0.0739, Interaction Genotype X Treatment F(1,30) = 0.0339 p = 0.8551. G COMT protein level alteration OVX Veh- and E2-treated female mice. 2-Way ANOVA revealed a main Genotype X Treatment Effect F(1, 18) = 11.54, **p = 0.0032 and a main genotype effect F(1, 18) = 12.82 **p = 0.0021 but not a main treatment effect F(1,18) = 3.184, p = 0.0912. Bonferroni multiple comparison test ***p = 0.0008 vs Dys + /+ Veh, ^##p = 0.0044 vs Dys + /− E2 (Dys + /+: Veh n = 4, E2 n = 5; Dys + /− OVX n = 7, E2 n = 6). H Temporal order recognition test in Veh or E2-treated male mice. 2Way-ANOVA revealed a Genotype effect F(1,38) = 17.78 ***p = 0.0001 and a Treatment effect F(1,38) = 20.30 ***p < 0.0001 as well as an Interaction TreatmentXGenotype F(1,38) = 12.95 ***p = 0.0009. Bonferroni’s multiple comparisons test: ***p < 0.0001 Dys + /+ Veh vs. Dys + /− Veh, ***< 0.0001 Dys + /− Veh vs. Dys + /− E2 (Dys + /+ Veh n = 12, E2 n = 11; Dys + /− Veh n = 9, E2 n = 10). I Total exploration time in males. 2Way-ANOVA showed no effect of the treatment on the working memory performance: Treatment effect F(1,38) = 0.01954 p = 0.889, Genotype F(1,38) = 2.854 p = 0.0994, Interaction F(1,38) = 0.3202 p = 0.5748. All data are represented as mean ± S.e.m. L COMT protein level alteration in Veh and E2-treated male mice. 2-Way ANOVA revealed a main genotypeXtreatment effect F(1, 17) = 18.91, **p = 0.0004 and a main genotype effect F(1, 17) = 4.608, p = 0.0465 but not a main treatment effect F(1,17) = 1.129, p = 0.3029. Bonferroni’s multiple comparison test: **p < 0.001 vs Dys + /+ Veh, ^{# #}p = 0.0097 vs Dys + /− E2 (Dys + /+: Veh n = 6, E2 n = 5; Dys + /− Veh n = 5, E2 n = 5).

Fig. 3. Sex differences of DTNBP1 gene expression in samples separately for age groups, genes, and brain regions.

Box plots showing differences in terms of gene expression quantified in three brain regions, i.e., caudate, DLPFC, and hippocampus, during the perinatal period (up to the age of 6 years), juvenile period (between 12–25 years of age), younger adulthood (25-50 years of age), and older adulthood (above 50 years of age). Separate two-tailed Welch two-sample t-tests revealed significantly higher gene expression in males compared to females during the juvenile period for DTNBP1 in the hippocampus, t(15) = 3.48, *p_FDR = 0.010. In adults significantly higher gene expression was found in females compared with males; namely greater DTNBP1 expression in the DLPFC in younger, t(51) = −2.37, p_FDR = 0.065, and older adults, t(49) = −2.19, *p_FDR = 0.050, and higher DTNBP1 expression in the hippocampus in older adults, t(62) = −2.56, *p_FDR = 0.038. No other significant sex differences have been reported, all p_uncorr > 0.05. Abbreviations: F female, M male, y years of age, DLPFC dorsolateral prefrontal cortex.

Fig. 4. Interactions between Catechol-O-Methyltransferase single nucleotide polymorphism rs4680 and sex on DTNBP1 gene expression separately for age groups and brain regions.

Error bars showing the interactions resulted from separate two-way analyses of variance between COMT and sex on DTNBP1 gene expression across age groups, that is, the perinatal period (up to the age of 6 years), juvenile period (between 12 and 25 years of age), younger adulthood (25-50 years of age), and the older adulthood (above 50 years of age) and separately for caudate, dorsolateral prefrontal cortex and hippocampus. Significant interactions have been found between sex and the COMT rs4680 on DTNBP1 expression for DLPFC in young adults (F(2,87) = 3.34, *p_uncorr = 0.040), and caudate in the perinatal age group (F(1,19) = 10.22, **p_uncorr = 0.006). All other analyses of variance yielded no significant interaction effects between sex and COMT rs4680 on DTNBP1 expression, all p_uncorr > 0.05 Abbreviations: F female, M male, y. age in years, DLPFC dorsolateral prefrontal cortex.

Fig. 5. Interaction between COMT rs4633, Dysbindin Hap, and sex on brain activity during the N-back task performance.

a Brain multi-slice sections and rendering showing the significant DLPFC activity resulting from the three-way interaction between COMT genotype, Dys Hap, and sex during N-Back task located in the left DLPFC (Brodmann Area 9; MNI coordinates x = −30, y = 32, z = 38; k = 52, Z = 3.49; *p_TFCE-FWE = 0.04). The color bar indicates log₁₀ TFCE scores ranging from 1.3 to 4 as suggested by Smith and Nichols [66]. b Box plots showing the differences between the COMT genotype and Dys Hap groups in the female and male groups assessed by the two-sample t test and corrected for multiple comparisons (k = 6; p_FDR < 0.05). The BOLD estimates extracted from Brodmann Area 9 were significantly higher in COMT^Val/Val females with Dys Hap +/− compared with Dys Hap +/+ (t(27) = 3.37; **p_FDR = 0.008), while COMT^MetCar females with Dys Hap +/− presented lower BOLD estimates compared with Dys Hap +/+ (t(69) = −3.68; **p_FDR = 0.005). Furthermore, females with Dys Hap +/− and COMT^Val/Val presented higher estimates than COMT^MetCar (t(23) = 3.22; **p_FDR = 0.009), as well as females with Dys Hap +/+ and COMT^MetCar presented higher estimates than COMT^Val/Val (t(32) = 3.63; **p_FDR = 0.005). Also, the BOLD estimates extracted from BA9 were different in COMT^Val/Val males with Dys Hap +/− compared with Dys Hap +/+ (t(23) = −3.21; **p_FDR = 0.004) in the opposite direction compared with the female analysis, as well as males with Dys Hap +/− and COMT^Val/Val presented trending lower estimates than COMT^MetCar (t(27) = −2.19; p_FDR = 0.06). Only significant differences corrected for multiple comparisons have been reported in the figure (p_FDR< 0.05). c Scatterplots showing the association between brain activity estimates extracted from left Brodmann Area 9 and behavioral indices from the N-back task, i.e., the efficiency rate and the reaction time (in seconds). General linear models revealed a significant negative association between the reaction time and brain activity estimates extracted from left Brodmann Area 9 in females with the COMT^MetCar and Dys + /− (r = −0.39; p_uncorr = 0.02), while a significant positive association between the reaction time and activity estimates (r = 0.37; p_uncorr = 0.05), and a negative association between the efficiency rate and activity estimates (r = −0.38; p_uncorr = 0.04) were found in males. All regression’s r and nominal p-values are reported in the figure. BOLD Blood Oxygen Level Dependent, BA9 Brodmann Area 9.