Prolonged epigenomic and synaptic plasticity alterations following single exposure to a psychedelic in mice

In this study, rodent models of depression, anxiety and stressor-related disorders were administered the psychedelic 2,5-Dimethoxy-4-iodoamphetamine (DOI), a lesser-known psychedelic substance of the amphetamine class. It was found that DOI produced fast-acting effects on dendritic spines in the frontal cortex and changes to chromatin organization in regions of genes involved in synaptic plasticity were observed for days after psychedelic exposure. These epigenomic changes in plasticity may sustain the long-lasting anti-depressant effects.

Abstract

“Clinical evidence suggests that rapid and sustained antidepressant action can be attained with a single exposure to psychedelics. However, the biological substrates and key mediators of psychedelics’ enduring action remain unknown. Here, we show that a single administration of the psychedelic DOI produces fast-acting effects on frontal cortex dendritic spine structure and acceleration of fear extinction via the 5-HT_2A receptor. Additionally, a single dose of DOI leads to changes in chromatin organization, particularly at enhancer regions of genes involved in synaptic assembly that stretch for days after the psychedelic exposure. These DOI-induced alterations in the neuronal epigenome overlap with genetic loci associated with schizophrenia, depression, and attention deficit hyperactivity disorder. Together, these data support that epigenomic-driven changes in synaptic plasticity sustain psychedelics’ long-lasting antidepressant action but also warn about potential substrate overlap with genetic risks for certain psychiatric conditions.”

Authors: Mario de la Fuente Revenga, Bohan Zhu, Christopher A. Guevara, Lynette B. Naler, Justin M. Saunders, Zirui Zhou, Rudy Toneatti, Salvador Sierra, Jennifer T. Wolstenhome, Patrick M. Beardsley, George W. Huntley, Chang Lu & Javier González-Maeso

Notes

This paper was also covered on The Tab.

Summary

INTRODUCTION

Psychiatric conditions such as depression, anxiety and stressor-related disorders affect millions of individuals worldwide. Standard pharmacotherapies often fail to provide clinically relevant improvements, and are accompanied by undesirable side effects.

Psychedelics, such as psilocybin, LSD and DOI, are psychoactive compounds that profoundly affect various mental domains, particularly sensory perception and thought processes. However, recent pilot clinical trials suggest that psychedelics may represent a promising long-lasting treatment for patients with depression and other psychiatric conditions.

Preclinical assays in rodent models have evaluated the effects of psychedelics as rapid-acting antidepressant medications. It has been suggested that psilocybin or LSD produces long-lasting antidepressant-like effects in rats within models of behavioral despair or passivity, as well as the extinction of cued fear memory in rats.

Mice treated with the psychedelic yet relatively selective 5-HT2A/2CR agonist DOI showed reduced immobility time in the forced swim test 24 h after administration, whereas behaviors in models of anxiety, recognition memory and sensorimotor gating remain unchanged.

Using a paradigm of contextual fear conditioning and extinction, we tested whether DOI affects 5-HT2AR-dependent signaling in fear acquisition and extinction. The results show that DOI has a modest yet statistically significant effect.

DOI had no effect on fear acquisition or fear expression, but reduced freezing time during the fear generalization and context fear extinction sessions in 5-HT2AR-/- mice. Additionally, DOI increased the rate of contextual fear extinction in 5-HT2AR+/+ mice.

Dendritic spines are central to brain plasticity. Psychedelics can promote rapid structural plasticity in pyramidal neurons, but most previous studies were conducted in vitro or relied upon manual 2D images after Golgi staining.

Using a 3D automated method for quantitative structural spine analysis, we found that 5-HT2AR-/- mice had a lower spine density in the frontal cortex than 5-HT2AR+/+ controls, and that DOI enhanced transitional stubby and dynamic thin spine density.

Based on these findings, DOI also regulates persistent synaptic functional plasticity in frontal cortex pyramidal neurons. DOI enhanced synaptic plasticity in L2/3 frontal cortex pyramidal neurons over the same time-course.

Previous studies demonstrated that a single dose of psychedelics alters the level of expression of several genes in the frontal cortex, but these studies suffered from lack of cell type specificity. Here, we used high-resolution, cell type-specific and low-input ChIP-seq and RNA-seq measurements to determine whether a single dose of DOI leads to long-lasting alterations.

We predicted enhancers by scanning the H3K27achigh regions that did not intersect with promoters, and divided all dynamic enhancers into 6 clusters with various patterns of variation after DOI injection. The effects of DOI at the epigenomic level outlast by several days the presence of the drug in native tissue.

To directly interrogate the functional relevance of these findings, we examined the GO terms and TF motifs associated with each of the clusters. These data suggest that DOI administration induces a compensatory mechanism as a result of increased glutamate release and augmentation of Egr1 and Egr2 transcription.

The long-term effects of DOI appear to be associated with epigenomic regulations of greater magnitude than changes in transcriptomic dynamics. This is supported by the finding that only 3.7% of the dynamically expressed genes present long-term variation (7 d).

To determine if there was a significant overlap of differential H3K27ac peaks with single nucleotide polymorphisms associated with depression, a null distribution was calculated using all SNPs. Five GWAS loci datasets were statistically significant.

We used weighted correlation network analysis to detect co-expression modules with highly correlated genes and performed GO analysis on the genes included in these modules to associate these modules with sample traits or experimental groups.

Module blue showed the strongest positive correlation with group 48 h and the strongest negative correlation with the control group. Among the 33 genes in this module, Inpp4a, Nfasc, and Cntnap2 are involved in axonogenesis, heart process, learning or memory, and histone modification.

Module yellow shows opposite trend of module blue and is enriched with genes related to hypoxia-inducible factor 1 (HIF-1) signaling pathway. This finding further supports the hypothesis that 5-HT2AR modulate the ventilatory response to hypoxia.

Module magenta’s eigengene showed no correlation with any experimental groups, and its genes were enriched in terms related to inflammatory response and intrinsic apoptotic signaling pathways.

Previous observations showed that psychedelics have long-lasting effects on synaptic plasticity and behavior models of depression. The present study provides direct evidence that psychedelics affect 5-HT2AR, which is responsible for these effects.

The subjective effects of psychedelics may be necessary or complementary for their post-acute clinically relevant outcomes in patients with severe psychiatric disorders such as depression. However, the recursiveness of the neocortex structure across the mammalian clade offers an unparalleled platform for characterizing non-subjective effects of psychedelics.

In this study, we found that psychedelics increase the density of immature and transitional dendritic spines in the frontal cortex, whereas the density of mature mushroom spines was unaffected. This effect was not observed in 5-HT2AR-/- mice, but may be related to changes among developmental compensatory pathways.

Our findings show that the epigenomic and transcriptomic dynamics following DOI administration are significant, and that a large fraction of epigenetic changes in enhancer regions persist for at least 7 d after DOI administration.

We studied the effects of DOI on the mouse brain epigenome and found significant overlap with GWAS-discovered variants associated with schizophrenia, depression, and ADHD. These findings may serve as a molecular warning on the risks associated to the use of psychedelics in psychotic-prone individuals.

We investigated whether the long-lasting effects of psychedelic DOI on epigenomic landscape within neuronal nuclei in the frontal cortex can be extrapolated to other brain regions relevant to substance abuse, and whether other cortical and subcortical dendritic structural and epigenetic plasticity events contribute to either clinically beneficial endpoints or unwanted side effects.

We previously reported that DOI, LSD and psilocin induce genes associated with cell morphogenesis, neuron projection and synapse structure. Our data here support a scheme whereby activation of 5-HT2AR by DOI accelerates context fear extinction likely through alterations in chromatin state at enhancer regions of genes predominantly involved in synapse organization and assembly. Recent observations have reported that a putative non-hallucinogenic psychedelic analogue promotes neural plasticity and reduces rodent behavior models relevant to depression.

Our study highlights the role of 5-HT2AR in the action of psychedelics and unveils persisting chromatin remodeling events following DOI administration. This could help understand psychopharmacological interventions whose mechanisms of action are not fully understood.

Experiments were performed on adult male mice (10-20 weeks old) randomly allocated into the different pre-treatment groups. All procedures were conducted in accordance with NIH guidelines, and animals were housed on a 12 h light/dark cycle at 23 °C with food and water ad libitum.

Locomotor activity was monitored 24h after drug administration on a computerized three-dimensional activity monitoring system. The open field was cleaned with 1% Roccal-D in between sessions.

Mice were injected with DOI or vehicle and placed in a commercial open field with equal parts light and dark areas connected through a small opening. Their preference for each area was monitored and expressed as a percentage of time spent on each.

Forced-swimming test was conducted 24h after DOI (2 mg/kg, i.p.) or vehicle administration. Mice were introduced in 2 L beakers while being held by the tail and immobility was scored during the last 4 min of the 6 min session.

Novel object recognition was conducted using tissue culture flasks filled with wood chips and opaque white light bulbs. The objects were attached to the bottom of an opaque rectangular plastic container with double-side tape. Mice that received DOI (2 mg/kg, i.p.) or vehicle 24h prior were tested for preference for a novel object as opposed to a familiar one. The preference for the novel object was calculated as percentage of total exploratory time spent exploring the novel object during the recognition stage.

Mice that had received DOI or vehicle 24h prior were presented with a startling stimulus of 119 dB (20 ms) preceded (80 ms interstimulus interval) by prepulses of 73 dB, 77 dB and 85 dB (20 ms). The startle magnitude in response to acoustic stimuli was recorded using the SR-LAB Startle Response System.

Contextual fear acquisition and extinction was tested using two commercially supplied Near Infrared Video Freeze Systems (MED-VFCNIR-M, Med Associates, Inc., St. Albans, VT), each operating four test chambers. Mice were acclimated to the behavioral testing room for at least 1h prior to experimental sessions. Two different contexts were employed: a standard polycarbonate squared cage with metallic walls and shock-delivering grid on the floor, and a modified chamber with black triangular adapter and cardamom scent on the bedding. Experimental sessions were performed on two consecutive days. The acquisition stage conditioned the mice to associate Context A with a noxious stimulus (foot-shocks), and the expression stage served as a control for conditioning and preliminary extinction training. On day 2, mice were placed in Context B for 5 min and observed for freezing behavior. After extinction training, mice were placed in Context A for 20 min.

Dendritic spine analysis was performed on neurons in the frontal cortex using an adeno-associated virus (AAV) serotype 8 expressing eYFP under the CaMKIIa promoter.

At least three weeks after surgery, apical dendritic segments 50-150 m away from the soma were randomly chosen from AAV-infected cells that express eYFP. The dendritic segments were imaged using a confocal fluorescence microscope and had to satisfy the following requirements: be completely filled, be at least 50 m from the soma, and be nonoverlapping. NeuronStudio was used to quantitatively analyze spine size and shape. Whole-cell patch-clamp recordings from layers 2/3 of somatosensory cortex were obtained from 8-10-week-old male mice as previously described (26).

Twenty-four hours prior to recording, experimental mice received a single injection of either saline (controls, n=5) or DOI (2,5-Dimethoxy-4-iodoamphetamine hydrochloride; 2 mg/kg, n= 5). Companion mice were used to match housing/social conditions across control and DOI experimental animals.

Acute coronal slices were sectioned on a VT1000S vibratome and transferred to a recovery chamber bubbled with carbogen and aCSF. The slices were perfused with 31°C oxygenated aCSF containing the GABAA receptor antagonist gabazine (10 M). L2/3 neurons with pyramidal-shaped somata were chosen for recording. EPSCs were recorded at -70 mV for 3-5 mins, followed by brief membrane depolarization to 0 mV for 10 mins, then the holding potential was returned to -70 mV for 35-40 mins.

Nuclei were isolated and sorted via FACS using a published protocol. The nuclei were placed in 3 ml of nuclei extraction buffer with 30 l of freshly added protease inhibitor cocktail.

The tissue was homogenized in a tissue grinder, and the cell suspension was centrifuged at 1000 RCF for 10 min. The nuclei pellet was resuspended in 2% normal goat serum in DPBS, and the nuclei were sorted into NeuN+ and NeuN- fractions using a BD FACSAriaTM cell sorter. 200 l of NeuN+ nuclei suspension was added into 800 l of ice-cold PBS for ChIP-seq experiment. The sample was centrifuged and resuspended in 60 l of PBS with 0.6 l of freshly added PIC and 0.6 l of PMSF.

10,000 NeuN+ nuclei were used to produce each ChIP-seq library, and two technical replicates were generated for each brain sample.

RNA-seq libraries were constructed from 5000 NeuN+ nuclei in 50 l of FACS-sorted nuclei suspension. Two technical replicates were generated for each brain sample. cDNA was prepared using the SMART-seq2 protocol (31) with minor modification. 2 ng of mRNA in 4.6 l of water was mixed with 2 l of 100 m oligo-dT primer and 2 l of 10 mM dNTP mix, and the mixture was denatured at 72 °C for 3 min. Generated cDNA was amplified by PCR and purified by SPRIselect beads. 600 pg of purified cDNA was used to produce a Nextera XT DNA Library Preparation kit.

Sequencing was performed on Illumina HiSeq 4000 with single-end 50-nt reads. The fragment size of ChIP-seq and RNA-seq libraries was measured using high sensitivity DNA analysis kit, and the concentration of each library was examined using a KAPA library quantification kit.

We predicted enhancers by identifying H3K27ac peaks that did not intersect with promoters, and then combined consensus enhancer sets for each experimental group using Diffbind. We identified differential enhancers between any two experimental groups using the bioconductor DESeq2 package.

We performed K-means clustering, GO term analysis, TF motif analysis on differential enhancers and DEGs across experimental groups, and created a list of the enhancer target genes by examining the correlation between gene expression (as reflected by RNA-seq data) and H3K27ac intensity at the candidate enhancers.

GWAS loci association was calculated by converting differential peak locations from mm10 to hg19 through UCSC’s liftOver tool, and then selecting SNPs from the 1000 Genome Project Phase 3 European population set that overlapped with the GWAS set.

RNA-seq data was analyzed using Trim Galore!, hisat2 (82) and SeqMonk v1.47.1 (Babraham Institute). Genes with over 0.3 at MAD were selected as input for the network construction.

We transformed the expression data of two genes into an adjacency matrix and then detected modules based on the topological overlap measure. The top 100 intramodular hub genes were identified and the relationship between these hub genes was visualized using Cytoscape.

Statistical analysis was performed with GraphPad Prism software version 9, and sample sizes were similar to those reported in our previous publications. Data were excluded based on previously established criterion, and statistical significance was assessed by two-way ANOVA followed by Sidak’s multiple comparison test.

Several studies have shown that 5-HT2A receptor agonists influence gene expression patterns in the mammalian brain, and a microfluidic device has been developed for low-input and multiplexed profiling of genome-wide histone modifications. Smart-seq2 for sensitive full-length transcriptome profiling in single cells, A. S. Nord, A. E. West, Neurog2 Deficiency Uncovers a Critical Period of Cell Fate Plasticity and Vulnerability among Neural-Crest-Derived Somatosensory Progenitor Cells, WGCNA: an R package for weighted correlation network analysis, 39.

Improved mental-health outcomes linked to psychedelic microdosing, chromatin regulation in complex brain disorders, dysregulation of the dopamine system in the pathophysiology of schizophrenia and depression, and white matter abnormalities in major depression biotypes identified by diffusion tensor imaging. The histone methyltransferase G9a plays an essential role in cocaine-induced plasticity, according to a study published in Science. Several studies have shown that 5-HT2A receptors are increased in frontal cortex pyramidal neurons in a genetic rat model of schizophrenia-relevant features, and that these receptors may be the site of action of hallucinogenic and antipsychotic drugs in pyramidal cell apical dendrites.

A practical and powerful approach to multiple testing is presented by Y. Benjamini, Y. Hochberg, P. J. Rousseeuw, D. U. Gorkin, I. Barozzi, Y. Zhao, Y. Zhang, H. Huang, A. Y. Lee, B. Li, J. Chiou, A. Wildberg, B. Ding, B. Zhang, M. Wang, J. S. Several tools are available to analyze the biological themes among gene clusters, such as ChIPseeker, clusterProfiler, HISAT, B. Langmead, S. L. Salzberg, featureCounts, and B. Zhang, S. Horvath.

The authors thank Jennifer Jimenez and Yashu Sampathkumar for technical experiments, analyzed data and wrote the manuscript. C.L. and J.G.-M. supervised the research and obtained funding.

DOI reduced passivity and fear extinction in mice 24 h after a single injection (i.p.) of DOI (2 mg/kg) or vehicle. DOI had no effect on exploratory behavior, dark-light choice test, novel-object recognition test, or PPI of startle. DOI induced fear extinction in 5-HT2AR+/+ and 5-HT2AR-/- mice. The effects of DOI on fear acquisition were analyzed using two-way repeated measures ANOVA, two-way ANOVA with Sidak’s multiple comparison test, or Student’s t test.

DOI affected synaptic structural elements in the frontal cortex of 5-HT2AR+/+ and 5-HT2AR-/- mice. Total, stubby, thin, mushroom, and AAV-injected cortical dendritic segments were affected. A single dose of DOI (2 mg/kg) significantly enhanced cortical LTP in comparison with saline-injected mice assayed 24-h post-injection. LTP was induced by pairing extracellular presynaptic stimulation of L4 neurons with brief (10 min) postsynaptic depolarization of overlying L2/3 neurons to 0 mV.

DOI (a thiol) induced epigenomic and transcriptomic variations in neuronal nuclei of the mouse frontal cortex. The differential enhancers and transcripts are clustered based on normalized H3K27ac signal, and the overlap between differential enhancers and NHGRI-EBI GWAS SNP sets is significant.

A heatmap of normalized gene expression profiles in the co-expression module and a visualization of the intramodular connections among the top 100 hub genes in each module.