Altered cortical synaptic lipid signaling leads to intermediate phenotypes of mental disorders

Abstract

Excitation/inhibition (E/I) balance plays important roles in mental disorders. Bioactive phospholipids like lysophosphatidic acid (LPA) are synthesized by the enzyme autotaxin (ATX) at cortical synapses and modulate glutamatergic transmission, and eventually alter E/I balance of cortical networks. Here, we analyzed functional consequences of altered E/I balance in 25 human subjects induced by genetic disruption of the synaptic lipid signaling modifier PRG-1, which were compared to 25 age and sex matched control subjects. Furthermore, we tested therapeutic options targeting ATX in a related mouse line. Using EEG combined with TMS in an instructed fear paradigm, neuropsychological analysis and an fMRI based episodic memory task, we found intermediate phenotypes of mental disorders in human carriers of a loss-of-function single nucleotide polymorphism of PRG-1 (PRG-1^R345T/WT). Prg-1^R346T/WT animals phenocopied human carriers showing increased anxiety, a depressive phenotype and lower stress resilience. Network analysis revealed that coherence and phase-amplitude coupling were altered by PRG-1 deficiency in memory related circuits in humans and mice alike. Brain oscillation phenotypes were restored by inhibtion of ATX in Prg-1 deficient mice indicating an interventional potential for mental disorders.

Fig. 1. Cortical network excitability shows significant dynamic alterations in PRG-1^R345T/WT mutation carriers.

A Experimental design for the assessment of excitability changes and E/I balance shifts in the frontal cortex of PRG-1^R345T/WT carriers using high-density EEG and TMS stimulation of the dmPFC. B SP: Global mean field potential (GMFP) over frontal cortex after single TMS-pulses (SP) over the dorsomedial prefrontal cortex (dmPFC) was significantly increased in PRG-1^R345T/WT carriers (n = 25) when compared to age and sex matched control subjects (n = 25). C SICI: After inhibitory Double TMS-pulses (SICI) over the dm PFC, GMPF over the frontal cortex in PRG-1^R345T/WT carriers (n = 25) was lower when compared to values after single pulse, however, it was significantly higher than in control subjects (n = 25). D Experimental design of the instructed fear paradigm. Unconditioned/safety Stimulus (CS^-) or a conditioned stimulus (CS⁺), which in 1/3rd of presentations was accompanied by an electric shock, were presented for 5 s following an image of a fixation cross (5–6 s). Single-pulse TMS was applied on the dmPFC 1 s after presentation of each stimulus. E Global theta power over frontal cortex at baseline was not altered in PRG-1^R345T/WT carriers (n = 25) when compared to matched control subjects (n = 25). F Theta power over the frontal cortex was not different between control subjects (n = 25) and PRG-1^R345T/WT carriers (n = 25) following CS^- presentation and single pulse TMS over dmPFC. However, following CS⁺ presentation and SP TMS, global theta power was significantly increased in control subjects, while a significant lower theta power was observed in PRG-1^R345T/WT carriers under same conditions. G Theta power ratio of GMFP changes following SP with prior CS⁺ or CS^- presentation (normalized to baseline theta power) displayed significantly reduced increase in PRG-1^R345T/WT carriers (n = 25) when compared to control subjects (n = 25) under same conditions. H Global gamma power over the frontal cortex at baseline was significantly increased in PRG-1^R345T/WT carriers (n = 25) when compared to matched controls (n = 25). I After CS^-, TMS over dmPFC resulted in significantly higher gamma power over the frontal cortex in PRG-1^R345T/WT carriers (n = 25) when compared to control subjects (n = 25). However, after presentation of a conditioned stimulus (CS⁺), gamma power was supressed and significantly lower in PRG-1^R345T/WT carriers when compared to CS^- conditions and was not different to control subjects. J Ratio of gamma power following SP with CS⁺ to SP with CS^- (both normalized to baseline gamma power) showed higher supression of gamma power in PRG-1^R345T/WT carriers (n = 25) following CS⁺ when compared to control human subjects (n = 25) under same conditions. Data in A-J are represented as violin plots covering all individual data points. Median, lower and upper quartiles are shown by dotted lines (* and ** show group differences of * >80% or **>90%, Bayesian analysis). K. Spatial configuration of the four microstate classes (A, B, C, D) according to EEG analyses. L–N EEG-analyses of the temporal microstates parameters for mean duration (shown in L), occurrence (shown in M) and time coverage (shown in N) revealed significant increases in microstate (C) and reduction in microstates (B, D) in PRG-1^R345T/WT carriers (n = 25) when compared to control subjects (n = 25). Age and gender were calculated as covariates. Analyses were performed using Bayesian one-way ANOVA. *** represent highest significance for Bayes factors >100. Data is shown as violin plots covering all individual data points. Median, lower and upper quartiles are shown by dotted line.

Fig. 2. PRG-1^R345T/WT SNP results in a phenotype consisting of mild levels of anxiety, depression and reduced stress coping in humans and mice.

A Human PRG-1^R345T/WT carriers displayed higher anxiety levels of state anxiety (STAI-S) but no difference in trait anxiety (STAI-T) when compared to control subjects as assessed by the State-Trait-Anxiety Inventory (STAI) (n = 12 control subjects and 12 PRG-1^R345T/WT carriers, Bayesian analysis). B In the positive and negative affect schedule (PANAS), PRG-1^R345T/WT carriers displayed higher levels for negative affects (PANAS-N) but no difference for positive affects (PANAS-P) when compared to control matched subjects (n = 21 control subjects and 22 PRG-1^R345T/WT carriers for PANAS-N and 22 control subjects and 22 PRG-1^R345T/WT carriers for PANAS-P, Bayesian analysis). C Prg-1^R346T/WT mice display reduced stay in the center of a novel open field (OF) arena and spent accordingly more time in its periphery when compared to WT litters (n = 11 WT mice, 14 Prg-1^R346T/WT mice, Bayesian analysis). D Fear conditioning revealed no significant difference at baseline but higher freezing 24 h after foot shock when animals were returned to the place where they previously received foot shocks (context). (n = 9 WT mice for baseline and after 24 h retrieval and n = 15 Prg-1^R346T/WT mice for baseline and 18 Prg-1^R346T/WT mice after 24 h retrieval; Bayesian analysis). E, F Social interaction and social interaction (SI) index were preserved in Prg-1^R346T/WT expressing mice. (n = 11 WT mice, 15 Prg-1^R346T/WT mice, Bayesian analysis). G Sucrose preference was significantly reduced in Prg-1^R346T/WT mice when compared to WT litters (n = 17 WT mice and n = 20 Prg-1^R346T/WT mice, Bayesian analysis). H Immobility in the tail suspension test (TST) following acute restrain stress was significantly increased in Prg-1^R346T/WT mice when compared to WT litters (n = 9 WT mice and n = 18 Prg-1^R346T/WT mice, Bayesian analysis). I, J Analysis of mice following chronic social defeat stress revealed no difference in social interaction index between Prg-1^R346T/WT mice and their WT litters in the group of resilient animals (defined by an SI index over 100). However, in the group of non-resilient animals (SI index below 100) Prg-1^R346T/WT mice displayed significantly reduced SI (n = 15 non-resilient WT mice and 26 non-resilient Prg-1^R346T/WT mice; n = 6 resilient WT mice, and 16 resilient Prg-1^R346T/WT mice; Bayesian analysis). Data are represented as box plots with whiskers covering all individual data points. Median is shown by solid line. (* and ** show group differences of * >80% or **>90%, Bayesian analysis).

Fig. 3. Analyzes on cortical network synchronization (coherence) revealed similar changes in PRG-1^R345T/WT human mutation carriers and in Prg-1^−/− mice.

A Exemplary image of EEG coherence analysis in the human entorhinal-hippocampal network. B Exemplary image of invasive mouse in-vivo measures and synchronization analysis of the murine entorhinal cortex (EC) and hippocampus (HC). C PRG-1^R345T/WT human carriers revealed significant decrease in the theta range (4–8 Hz) and an increase in the gamma range (30–100 Hz) when compared to control human subjects (n = 25 control subjects, 25 PRG-1^R345T/WT carriers, Bayesian analysis). D In-vivo analysis of WT and PRG-1^−/− mice revealed significantly decreased theta coherence (7–12 Hz) and increased gamma coherence (30–100 Hz) when compared to WT litters. (n = 15 WT mice, 13 Prg-1^−/− mice, Bayesian analysis). E, F Theta coherence (4–8 Hz) between the EC and the HC was reduced in PRG-1^R345T/WT human carriers, while gamma coherence (30–100 Hz) was significantly increased (n = 25 control subjects, 25 PRG-1^R345T/WT carriers, Bayesian analysis). G, H. Theta coherence (7–12 Hz) in the entorhinal-hippocampal network of Prg-1^−/− mice was significantly decreased, while low and high gamma coherence (30–100 Hz) was significantly increased (n = 15 WT mice, 12 Prg-1^−/− mice, Bayesian analysis). I Representative phase-amplitude coupling (PAC) plots of EC theta and HC gamma. Left PAC plot shows high PAC levels of EC 5–8 Hz theta oscillation and hippocampal gamma power in the range of 40–100 Hz in a control human subject. Right PAC plots depicts reduced modulation of HC gamma power by EC theta frequency in a PRG-1^R345T/WT human mutation carrier. J Representative PAC plot of a wild type animal (left, WT) showing strong modulation of the hippocampal gamma power by entorhinal theta frequencies. Note the phase-amplitude correlation (yellow) showing highest values between 60 and 100 Hz for hippocampal gamma power modulated by 7–12 Hz entorhinal theta frequencies. Right plot displays representative PAC of a Prg-1^−/− mouse showing reduced phase-amplitude correlation (yellow). K PAC of entorhinal theta oscillation (5 Hz, EC θ) and hippocampal gamma power (30–100 Hz, HC γ) shows reduced PAC in PRG-1^R345T/WT human mutation carriers when compared to matched control subjects (n = 25 control subjects, 25 PRG-1^R345T/WT carriers, Bayesian analysis). L EC theta power (5 Hz) was significantly decreased in PRG-1^R345T/WT human mutation carriers when compared to control subjects (n = 25 control subjects, 25 PRG-1^R345T/WT carriers, Bayesian analysis). M PAC of EC theta oscillation (10 Hz, EC θ) and hippocampal gamma power (30–100 Hz, EC γ) shows reduced coupling in Prg-1^−/− mice when compared to their wild type litters (n = 15 wild type mice and 12 Prg-1^−/− mice, Bayesian analysis). N EC theta power (10 Hz) was significantly reduced in Prg-1^−/− animals (n = 14 WT and 10 Prg-1^−/−). E–H, K–N. Data are represented as box plots with whiskers covering all individual data points. Median, lower, and upper quartiles are indicated by dotted line. (* and ** show group differences of * >80% or **>90%, Bayesian analysis).

Fig. 4. ATX inhibition restores entorhinal-hippocampal coherence and PAC to control values.

A Inhibition of the LPA-synthesizing molecule ATX by PF8380 (ATX inh.) increased EC-HC theta coherence (7–12 Hz) in Prg-1^−/− mice to WT levels (n = 15 WT mice, 7 Prg-1^−/− + ATX inh. mice, Bayesian analysis). B Low and high gamma coherence (30–100 Hz) in the entorhinal-hippocampal network in Prg-1^−/− mice was reduced to WT levels following ATX inhibition (n = 15 WT mice and 7 Prg-1^−/− + ATX inh. mice, Bayesian analysis). C Coherence spectrum of WT and Prg-1^−/− mice following ATX-inhibition shows restored coherence to WT levels (n = 15 WT mice, 7 Prg-1^−/− + ATX inh. mice, Bayesian analysis). D EC theta power (10 Hz) in Prg-1^−/− animals was increased to WT levels following ATX-inhibition by PF8380 (n = 14 WT, 7 Prg-1^−/− + PF8380, Bayesian analysis). E PAC of EC theta oscillation (10 Hz, EC θ) and hippocampal gamma power (30–100 Hz, EC γ) in Prg-1^−/− mice increased back to wild type levels following inhibition of the LPA-synthesizing enzyme ATX (ATX inh.) (n = 15 wild type mice, 7 Prg-1^−/− + ATX inh. mice, Bayesian analysis). F Representative PAC of a Prg-1^−/− animal following ATX-inhibition. Note the restored PAC between the entorhinal theta frequency (10 Hz) and the hippocampal gamma power (30–100 Hz) in the Prg-1^−/− animal after ATX inhibition. A, B, D, E Data are represented as violin plots covering all individual data points. Median, lower and upper quartiles are shown by dotted linie (* and ** show group differences of * >80% or **>90% for Bayesian analysis).

Fig. 5. Functional imaging in PRG-1^R345T/WT mutation carriers revealed higher brain activation in an episodic memory paradigm and reduced DMN connectivity.

A Experimental paradigm using fMRI. Human control subjects and PRG-1^R345T/WT carriers were subjected to an episodic memory task under fMRI conditions. B Regional brain activation during encoding in an episodic memory test shows higher bilateral activation in the anterior cingulate gyrus (ACC), in the dorsomedial prefrontal cortex (dmPFC) as well as in the insula (In) in PRG-1^R345T/WT human mutation carriers when compared to control subjects. C Regional brain activation during recall was significantly higher in PRG-1^R345T/WT mutation carriers in both hemispheres in the entorhinal region (EC) and in the parahippocampal gyrus (PHC). t-values for higher activation are represented by color bar in (B, C). D Default mode network (DMN) analysis of fMRI data during episodic memory test revealed lower activation of medial prefrontal cortex (mPFC), posterior cingulate cortex (PCC) and inferior parietal cortex (inf PC) in PRG-1^R345T/WT human mutation carriers when compared to matched control subjects. t-values for deactivation are represented by color bar. E fMRI network connectivity analyses within the DMN (within-network connectivity) showed lower connectivity of mPFC, PCC and inf PC. F EEG theta connectivity analyses within the DMN (within-network connectivity) revealed lower connectivity of mPFC, PCC, and inf PC. G, H. Resting state coherence in the prefrontal-hippocampal loop (dorsomedial prefrontal cortex, dmPFC) displayed reduction in the theta range and an increase in gamma range in PRG-1^R345T/WT human carriers when compared to matched controls All above mentioned analyses were performed in n = 25 control subjects, 25 PRG-1^R345T/WT carriers. Data are represented as violin plots covering all individual data points. Median, lower and upper quartiles are shown by dotted linie (* and ** show group differences of * >80% or **>90% for Bayesian analysis).