Analysing RNA-seq from the rat prefrontal cortex after chronic LSD, the authors show that LSD rewires gene co‑expression networks to become less centralised but more complex, producing an overall increase in signalling entropy consistent with heightened molecular plasticity. This molecular increase in entropy parallels human neuroimaging reports of greater brain entropy and, via network topology, the study nominates candidate transcriptional regulators and implicates specific cell types in psychedelic action.
Psychedelic drugs are gaining attention from the scientific community as potential new compounds for the treatment of psychiatric diseases such as mood and substance use disorders. The 5-HT 2A receptor has been identified as the main molecular target, and early studies pointed to an effect on the expression of neuroplasticity genes. Analysing RNA-seq data from the prefrontal cortex of rats chronically treated with lysergic acid diethylamide (LSD), we describe the psychedelic-induced rewiring of gene co-expression networks, which become less centralized but more complex, with an overall increase in signalling entropy, typical of highly plastic systems. Intriguingly, signalling entropy mirrors, at the molecular level, the increased brain entropy reported through neuroimaging studies in human, suggesting the underlying mechanisms of higher-order phenomena. Moreover, from the analysis of network topology we identify potential transcriptional regulators and imply different cell types in psychedelics’ activity.
Psychedelic drugs such as LSD, psilocybin and DOI act primarily at the 5-HT2A receptor and produce profound acute alterations in perception, behaviour and mood. Earlier work has shown that these compounds can mimic psychotic-like states and have been used in animal models of schizophrenia, but more recent clinical and preclinical studies have also reported rapid and sometimes long-lasting beneficial effects in mood and substance use disorders. Neuroimaging studies in humans indicate that psychedelics increase the variability or information content of brain signals (often described as increased brain entropy), and rodent studies report induction of neuroplasticity-related and immediate-early genes in brain areas with dense 5-HT2A expression such as the prefrontal cortex (PFC). At the molecular level, entropy has been used to describe both plastic, stem-like states and pathological or age-related loss of organisation, depending on the metric and context. This study analyses high-depth RNA-seq data from the medial PFC of rats chronically treated with LSD to characterise how psychedelics reorganise gene co-expression networks and transcriptional entropy. The researchers aim to determine whether chronic LSD leads to sustained changes in network topology and signalling (transcriptional) entropy, to identify potential transcriptional regulators underlying those changes, and to explore whether different cell compartments may be implicated. They additionally compare chronic-treatment signatures with available datasets from acute single-dose experiments to infer temporal dynamics of the transcriptional response.
Papers cited by this study that are also in Blossom
Barbanoj, M. J., Riba, J., Clos, S. et al. · Psychopharmacology (2007)
Carhart-Harris, R. L. · Neuropharmacology (2018)
Carhart-Harris, R. L., Bolstridge, M., Rucker, J. et al. · Lancet Psychiatry (2016)
Carhart-Harris, R. L., Leech, R., Shanahan, M. et al. · Frontiers in Human Neuroscience (2014)
The researchers re-analysed previously published transcriptomic datasets. The primary dataset comprises RNA-seq from medial PFC (mPFC) of rats chronically treated with LSD; the extracted text gives inconsistent timing for sample collection after treatment cessation (one passage reports samples taken 3 weeks after discontinuation, another refers to three months), and the extracted text does not clearly report the total number of animals or biological replicates for the main RNA-seq dataset. Two additional datasets were used for temporal comparisons: a microarray dataset from rat PFC 90 minutes after acute LSD administration (noted to consist of three pooled independent replicates of two animals each) and an RNA-seq dataset of mouse cortical neurons collected 24h, 48h and 72h after DOI administration. Pre-processing for RNA-seq included adapter trimming and quality control with TrimGalore, alignment to the Rattus norvegicus genome (rn6) with Hisat2, read counting with HTSeq-count, RPM normalisation, and a filter retaining genes with at least 5 reads in at least 10 samples. Microarray data were normalised with RMA. Splicing junction usage was quantified with SGSeq. Transposable elements were quantified with TEtranscripts after aligning reads with permissive multi-mapping settings. Differential expression analysis used DESeq2. Gene Ontology enrichment was performed with clusterProfiler. Co-expression networks were constructed using WGCNA (signed network, power beta=12), yielding modules whose activity was summarised by module eigengenes (MEs). Module differential activity was tested by comparing MEs between treatment groups with Wilcoxon tests and FDR correction. Potential transcriptional regulators were explored by selecting centrally connected transcription factors within modules and testing enrichment of their known targets using the ChEA database. Signalling entropy was calculated with the SCENT R package using either a protein–protein interaction (PPI) network or the study-specific co-expression network; between-sample entropy was defined as 1 minus the Pearson correlation between samples. Subsampling (random selection of 10 million reads per sample) was used to check that results were not driven by sequencing depth.
Differential expression analysis of the chronic-LSD mPFC RNA-seq identified up-regulation of neuroplasticity and neurotransmission genes, including previously reported bdnf up-regulation and enrichment for dendrite development. Circadian rhythm genes were enriched among up-regulated genes. Notably, genes involved in covalent chromatin modification and histone modification were also up-regulated, with Tet1 (involved in DNA demethylation) among those increased, suggesting effects on the epigenetic machinery. Down-regulated genes were enriched for oxidative phosphorylation and included genes with GTP-binding, hydrolase activity and ribosomal structural functions. WGCNA produced eighteen co-expression modules; six modules showed differential activity between LSD and control conditions. Modules with higher activity in LSD-treated rats were enriched for chromatin organisation, vesicle-mediated synaptic transport, and cell–cell adhesion. Using network topology and an external transcription factor target database, the researchers nominated Tcf4 as a potential regulator of the vesicle transport module (FDR < 3×10^-4 across available gene sets) and Arnt as a potential regulator of the cell–cell adhesion module (FDR = 0.0009), with no other TFs meeting the stringent criteria. Measuring within-sample signalling entropy (using both a PPI-based method and the constructed co-expression network) revealed a significant increase in transcriptional complexity in the PFC of chronically LSD-treated rats relative to saline controls, persisting after the withdrawal period reported in the dataset. Complementary measures—number of splicing junctions used per sample and the fraction of reads mapping to transposable elements—were also increased in the LSD group. These results were robust to subsampling to equal read depth across samples. By contrast, between-sample entropy (group-level variability) decreased in the LSD-treated group, indicating reduced divergence between individual samples in the treated group at the gene expression and splicing levels. Ranking genes by change in per-gene entropy singled out RNA processing and splicing, chromosome organisation, DNA repair, cell cycle and extracellular matrix among pathways with the largest entropy increases. Genes with altered splicing patterns were enriched for targets of the Nova splicing factor. Overall co-expression network connectivity decreased in LSD-treated samples, consistent with a decentralisation of the network; however, three modules displayed increased intramodular connectivity and centralisation. Those modules were enriched for mesenchyme development, immune system regulation and extracellular matrix terms, interpreted as indicative of microenvironmental or non-neuronal cell involvement. Within most modules, hubs lost connections and peripheral nodes gained connections, with entropy increases concentrated in peripheral nodes. Temporal comparisons with other datasets suggested that modules upregulated by chronic LSD treatment are also over-expressed after a single DOI administration in mice at 24h and return to baseline by 72h. Tcf4 showed a similar acute induction pattern with DOI. Down-regulated modules in the chronic LSD dataset did not show a consistent pattern after single doses. In the acute LSD microarray dataset no significant changes were detected, potentially due to small sample size (N=3 pooled replicates), platform differences or limited gene coverage. Signalling entropy showed a non-significant trend in the acute LSD dataset but was significantly increased in DOI-treated neurons at 24h and returned to baseline by 72h. The small sample sizes in the auxiliary datasets limited reliable measurement of between-sample entropy in those experiments.
The authors interpret their findings as evidence that chronic LSD treatment produces long-lasting transcriptional reorganisation in rat mPFC that includes up-regulation of neuroplasticity, neurotransmission and epigenetic regulator genes, and an increase in transcriptional or signalling entropy. They note that the up-regulation of Tet1 and enrichment for histone modification terms point to engagement of the epigenetic machinery as a candidate mechanism for persistent transcriptional changes. The overall increase in signalling entropy is presented as the molecular parallel of increased brain entropy observed in human neuroimaging studies under psychedelics; the authors argue that the pattern of changes (including increased alternative splicing and transposable element activity) more closely resembles a plastic, stem-like transcriptional state than an age-related disorder state. From network topology the researchers propose Tcf4 as a plausible regulator of a vesicle-transport/synaptic module; Tcf4 has known roles in neuritogenesis and spine density. The authors also highlight a dichotomy in module reorganisation: most modules lose centralisation and connectivity (peripheral nodes gain relative importance), whereas a subset of modules related to mesenchyme, immune response and extracellular matrix gain connectivity and centralisation, suggesting involvement of non-neuronal cell populations or microenvironmental reorganisation in long-term LSD effects. They propose a temporal model in which a single psychedelic dose triggers a transient reorganisation of gene networks and increased signalling entropy, and repeated administrations are required to produce sustained transcriptional rewiring and lasting increases in entropy that could underlie persistent behavioural effects. Conversely, prolonged decentralisation and elevated entropy after repeated dosing might also relate to psychotic-like phenotypes observed in chronically treated animals. The authors acknowledge several limitations: inconsistencies and limited reporting in the primary dataset regarding exact post-treatment timing, small sample sizes (notably the acute LSD microarray with pooled replicates), cross-platform and cross-species comparisons (rat vs mouse, RNA-seq vs microarray), and the inability of bulk RNA-seq to distinguish whether increased signalling entropy arises from increased cell-type diversity versus increased potential within individual cells. They state that single-cell RNA-seq and longer time-course experiments will be necessary to disentangle these alternatives and to better characterise the cellular sources and temporal persistence of the transcriptional changes. The authors conclude that their analyses suggest molecular mechanisms that may underlie psychedelics-induced increases in brain entropy, implicate epigenetic alteration, alternative splicing and transposable element activity, and point to possible involvement of non-neuronal compartments such as mesenchymal and immune-related elements in the long-term effects of psychedelics.
2.1. Chronic LSD exposure has a long-term effect on the epigenetic machinery We made use of the previously described system of rats chronically treated with LSD, which show persistent altered locomotor activity and social interactions long after drug discontinuationand have hence been proposed as a rat model system for the study of schizophrenia. RNA-seq data on the mPFC of these rats 3 weeks after cessation of drug showed persisting transcriptional effects, and an enrichment for differential regulation of schizophreniarelated genes. We further pursued the investigation of these high depth RNA-seq data to study the rearrangements of gene co-expression networks in response to prolonged repeated LSD administration. First, from the analysis of differentially expressed genes (Suppl. Table), we observed increased expression of genes related to neuroplasticity and neurotransmission, confirming previous reports of bdnf up-regulation and of increased dendritogenesis upon psychedelics treatment. Indeed, "dendrite development" is amongst the top significantly enriched Gene Ontology (GO) categories for up-regulated genes (Suppl. Table). Also, we found a significant enrichment for circadian rhythm genes, possibly underlying the alteration of sleep cycles. Interestingly, "covalent chromatin modification" and "histone modification" showed strong enrichment for up-regulated genes, among which we found Tet1, involved in the erasure of DNA methylation. This indicates that repeated psychedelics' administration affects the epigenetic machinery, thus suggesting a mechanism for their long-term effects. Downregulated genes were found to be mostly involved in oxidative phosphorylation (Suppl. Table). We then built gene co-expression networks, to investigate potential regulatory relationships between genes and their changes upon treatment. We applied the widely used WGCNA algorithm (Weighted Gene Co-expression Network Analysis)and identified eighteen clusters of genes (modules, Suppl. Table), six of which show differential activity between LSD and Ctrl conditions (Figure), as quantified through the module eigengene (ME), an ideal meta-gene representing the whole module's expression. Modules higher in LSD-treated rats are enriched for regulation of chromatin organization, vesicle-mediated transport in synapse, and cell-cell adhesion (Suppl. Table), indicating that our observations on differentially expressed GO categories are robust and independent of the analysis method. Although modules down-regulated upon treatment do not show a significant enrichment for GO biological processes, they include genes with molecular functions such as GTP-binding, hydrolase activity, and structural constituents of ribosomes (Suppl. Table). It is important to note, however, that these are not the only enriched categories for each module (Suppl. Tables), and labels assigned based on the most significantly enriched GO categories should not be interpreted strictly to exclude other categories. Using the network's topological features, we looked for potential regulators of the LSD modules. Indeed, genes centrally located within the network and displaying many connections (hubs) have been reported to be crucial for the system's maintenance). Hence, we selected genes amongst the first 10 central transcription factors (TFs) in each module and tested whether there is independent evidence of their regulatory activity on the same module's genes using the ChEA database. With these criteria, we identified Tcf4 as a potential regulator of the "Vesicle transport" module (false discovery rate for each of the three available gene sets <3*10 -4 ), and Arnt as a potential regulator of the "Cell-cell adhesion" module (false discovery rate = 0.0009), but no other genes passed these stringent filters.
We investigated the signalling variability of gene networks by measuring transcriptional entropy. As mentioned in the introduction, transcriptional entropy has been paradoxically associated with both plasticity and aging. Nevertheless, different measures have been used between the two contexts: in the first, entropy is quantified from the number of signalling paths in a protein-protein interaction network or from the number of expressed genes in each sample, whereas in the second, differences between samples have been used as a metric to define entropy. To distinguish these scenarios, we named these two kinds of entropies as "signalling" and "between-sample" entropy, schematically described in Figure.
obtained and then summarized as the overall subject specific signalling entropy. B) Between-sample entropy is defined as the variability between different subjects in a group. Hence, homogenous groups have low entropy, while diversified groups have high entropy. To ensure robustness of our observations, we measured signalling entropy based on both proteinprotein interaction (PPI) networks, as in the originally published methodor on our co-expression network. Moreover, as additional measures of within-sample variability or information content, we quantified the number of splicing junctions used in each sample and the proportion of reads mapped to transposable elements. With all metrics we found a significant increase in the transcriptional complexity of the PFC of rats chronically treated with LSD with respect to saline controls, even after a withdrawal period of 3 weeks (Figure). Of note, none of these measures are influenced by sequencing depth, since differences are retained also by randomly sampling the data to have the same number of reads in each sample (Suppl. Fig.). On the contrary, we observed a decrease in between-sample entropy, revealed by the lower samples' divergence in the LSD condition at the gene expression and splicing usage levels (Figure). Taken together, these observations suggest that LSD treatment induces increased plasticitylike and reduced aging-like entropy.
We next studied which nodes most strongly contribute to the overall signalling entropy increase, ranking the genes based on their average entropy change between the LSD and saline conditions. Performing a GSEA analysis on the ranked gene list, RNA processing and splicing, chromosome organization, DNA repair, cell cycle, extracellular matrix, and chromosome organization resulted as significantly enriched amongst the genes with the highest increase in entropy. Of note, increased entropy in genes regulating RNA processing and splicing could explain the larger set of splicing junctions detected in the LSD treatment group. Interestingly, the genes with altered splicing patterns are enriched for genes regulated by the splicing factor Nova. In line with an increase in entropy, the overall connectivity of the co-expression network decreases (Figure), as previously shown for cancer networks. Interestingly, not all modules rearrange their connections in the same way: despite most modules decreasing their co-expression in the LSD treated group, indicating a de-centralization of the network, three display increased intramodular connectivity, indicating a tighter regulation of corresponding functions (Figure). For the same three modules, the nodes that increase coexpression connections tend to be the most central, while for the remaining modules we observe the opposite trend (Figure). Accordingly, in most modules entropy increases more strongly for peripheral nodes, and only a few modules make an exception (Figure). Of these, the "mesenchyme development" module shows coherent correlation sign between centrality and change in connectivity/entropy. This indicates that for most modules the nodes increasing signalling connections, reflecting on both entropy and intra-modular weighted degree, are the most peripheral in the module, while hubs tend to lose connections (Figure). Intriguingly, the modules showing strong opposite trends, and therefore increasing the compactness and centralization of the network, are enriched for mesenchyme development, regulation of immune system and extracellular matrix, categories that could reflect microenvironmental reorganization, and suggesting that different cell types might be affected by prolonged LSD treatment in different ways. To explore the temporal dynamics of the transcriptional changes that we observe in chronically treated rats, we took advantage of two additional transcriptomic datasets, analysing PFCs from rat brains harvested 90 min after an acute administration of LSD (C. D., or neurons isolated from the PFC of mice after 24h, 48h or 72h of DOI administration. We envisioned at least two possible temporal dynamics: in the first model, psychedelics induce the well-established transcriptional changes in neuroplasticity genes without immediately affecting the organization of co-expression modules nor transcriptional complexity, exerting this effect only after repeated treatments; in a second model, the network changes that we detect are induced acutely, and potentially show different post-treatment dynamics in the case of single or multiple treatments. Therefore, we tested whether our findings could be replicated in the two aforementioned singledose datasets. Co-expression modules increasing in expression with chronic LSD treatment are strongly over-expressed also after a single treatment with DOI, and return to the baseline at 72h (Suppl. Fig.). Additionally, the potential transcriptional regulator of the "vesicle transport" module, Tcf4, shows the same trend upon DOI treatment (Suppl. Fig.). Modules downregulated upon chronic LSD treatment do not show a clear pattern, with only a slight trend toward down-regulation at 24h and 48h of the GTP-binding module (Suppl. Fig.). This suggests that, while the first three groups of genes are induced by acute treatment and require multiple administrations for sustained expression, the last three modules are repressed only upon chronic treatment. No significant differences could be detected in the microarray of LSD's acute effects, possibly related to the small sample size (N=3), to differences in the technologies, or to the smaller number of genes analysed in the array with respect to genome-wide RNA-seq (Suppl. Fig.). Nevertheless, we cannot exclude the lack of differential expression being due to differences in the biological system. Similarly, we tested the change in signalling entropy: in the LSD dataset, we observed a trend towards increased signalling entropy in the treated group, which nevertheless did not reach significance, likely due to the small sample size, while entropy is significantly increased by DOI at 24h after treatment, returning to baseline at 72h. Unfortunately, the small number of samples in the two additional datasets did not allow to have a reliable measure of between-sample entropy.
Psychedelic drugs are gaining attention from the scientific community as potential new compounds for the treatment of psychiatric diseases such as depression and PTSD, but also as highways to explore the neurobiology of human consciousness. Also, the psychedelics' induced alterations in perception and thought processes mimic symptoms of mental disorders such as schizophrenia, and hence they have been employed in animal models to study this disease. Neuroimaging studies have shown their effects on brain connectivity networks in humans, where they induce the reorganization of interacting areas leading to a diminished activity of the DMN and to an increased variability of the electromagnetic signal, quantified as an increase in the entropy. The main molecular target has been identified as the serotonin 5-HT2A receptor. Early studies pointed to an effect on neuroplasticity-related genes, but an extensive investigation on psychedelics' induced gene expression changes is still lacking. Nevertheless, gene expression patterns and interactions strongly characterize the functioning of biological systems, implying that behavioural changes are often mediated by alterations in gene expression. The transcriptome, studied through microarray or RNA-sequencing, is the most accessible layer of gene expression, and RNA-seq is increasingly employed to explore the molecular mechanisms implicated in physiological, pathological and drug-induced processes. Analysing RNA-seq data from prefrontal cortices of rats chronically treated LSD three months after drug discontinuation (e.g. no drug or recent drug on-board), we observe long-lasting changes in gene expression, particularly in neuroplasticity and neurotransmission genes, in line with previous reports, but also in circadian rhythm and, importantly, in epigenetic modifiers. Indeed, Tet1, involved in the erasure of DNA methylation, increases its expression upon LSD, and the "histone modification" GO category showed strong enrichment for up-regulated genes. This indicates that repeated psychedelics' administration affects the epigenetic machinery, thus proposing a mechanism for their long-term effects. Interestingly, Tet1 is recruited by Egr1, one of the classic immediate-early genes which expression is induced by psychedelics. Of note, it has been recently shown that a single dose of DOI is able to alter the PFC epigenetic status in mice up to 7 days after administration, suggesting that a single exposure is sufficient to induce epigenetic reorganization. Applying gene co-expression network analysis, we confirm these observations identifying modules of tightly connected LSD-induced genes enriched for chromatin modification, vesicle mediated transport in synapse and cell-cell adhesion. On the other hand, down-regulated modules are enriched for GTP-binding, hydrolase activity and ribosome. From the topology of the co-expression network, we propose Tcf4 as a potential regulator of the up-regulated "Vesicle transport" module. Notably, Tcf4 regulates neuritogenesis and neuronal migration, and its loss decreases spine density in the cortex and in the hippocampus. Importantly, we describe an overall increase in transcriptional/signalling entropy, potentially reflecting an overall increase in available transcriptional states. Intriguingly, transcriptional entropy mirrors, at the molecular level, the increased brain entropy reported through neuroimaging studies in human. High transcriptional entropy is typical of stem cells, and decreases with differentiation, paralleling cells' developmental potential. Nevertheless, transcriptional entropy has also been associated with aging, reflecting the disruption of a well-organized and functional system. Using multiple metrics, we show that LSD-induced transcriptional entropy is more reflective of a plastic stem-like state than an aged state, suggesting the induction of a potentiality-expansion process with organized and reproducible features. Moreover, we imply additional players in transcriptional diversification, since LSD-treated rats display re-activation of transposable elements and increased alternative splicing sites' usage. In particular, the most reliably alternatively spliced genes are enriched for targets of the Nova splicing factor, known to control the splicing of synaptic proteins. Transposable elements (TEs) are a class of repeated DNA sequences with the ability to mobilize and change locations in the genome. Despite being mostly inactive in somatic cells, efficient transposition was detected in neural progenitor cells and mature neurons. They have been reported to be both beneficial and pathological to the organism, and associated with both neurodegenerative and psychiatric disease and plasticity. For example, in first episode schizophrenia, hypomethylation of HERV-K locus was reported, and L1 insertions were found significantly elevated in post-mortem dorsolateral prefrontal cortex of patients with schizophrenia. Nevertheless, they have been proposed to have a fundamental role in promoting evolution and also increasing cells' variability through the generation of somatic mosaicism, creating a greater potential for the adaptation of genetic networks. Disentangling individual modules' topological reorganization, we distinguish two opposite processes: 1) most modules decrease their connectivity paralleling the increased entropy, and redistribute connections towards module's periphery, hence reducing its centralization (Figure); 2) a few modules increase their overall connectivity and centralization. Interestingly, the modules increasing connectivity are enriched for GO categories indicative of the involvement of microenvironment: mesenchyme development, regulation of immune response and extracellular matrix. This suggests a potential involvement of non-neuronal cells in long-term LSD effects. In particular, Mesenchymal stem cells (MSCs) are stem cells found in many adult tissues, including brain,and can differentiate into neuronsand glial cellsto replace damaged tissues, and thus promoting neuroprotection, regeneration and repair. They have immunomodulatory properties, mitigating the inflammation related to stroke or neurological diseases. Interestingly, both mesenchyme-and immune-related modules display similar topological restructuring. In line with our observations suggesting non-neuronal populations' involvement, in particular of the immune system, psychedelics have been shown to exert anti-inflammatory effects, which led to their proposal as treatments for neurodegenerative diseases such as Alzheimer disease. Finally, we explored the temporal dynamics of co-expression modules and signalling entropy by comparing our results with other available datasets. We acknowledge that none of these conditions is ideal for our aim, since the LSD dataset has been generated with a different platform (Affymetrix U34A), possibly generating technical batch effects, and comprises only three pooled independent replicates of two animals each, limiting the statistical power, while the second utilizes a different psychedelic drug, DOI, with non-perfectly overlapping effects with LSD. Another limitation is that the LSD set used rats, and the DOI set mice. Nevertheless, both LSD-induced modules and signalling entropy display similar trends in PFC neurons upon DOI treatment, increasing at 24h and returning to background after 72h. Hence, we propose a model of the transcriptional response to psychedelics where a single dose triggers a transient reorganization of the gene networks that is sustained at long-term only after several administrations (Figure). We hypothesise that the short-term reorganization and entropy increase induced by LSD allows for the formation of new synaptic connections and hence novel neuronal networks that can be maintained after the treatment, resulting in the long-lasting beneficial effects that psychedelics have proven to exert in clinical settings. Frequent and repeated administrations, however, may result in a prolonged increase in the entropy and decentralization of co-expression networks that could be reflective of the psychotic-like state observed in the chronically treated rats. Signalling entropy increase could be interpreted as increased cell diversity within the same subject, but could also reflect higher individual cells' potential. With the available data, it is impossible distinguishing between these two mechanisms, for which single cell RNA-seq experiments would be necessary. The indirect activation by LSD of several neuronal cell types has previously been shown, but how each cell population responds to this class of compounds and how cell heterogeneity is affected remains largely unknown. Additionally, comparing experiments with different administration schedules showed differences in the persistence of transcriptional rewiring, and additional time-course experiments with longer time frames will be needed to elucidate the dynamics of gene expression response. In conclusion, analysing transcriptomic data of LSD treated rats, we suggest some of the possible molecular mechanisms potentially underlying psychedelics-induced increase in brain entropy observed at higher psychedelic levels, identify epigenetic alterations that could explain long-term effects, and imply alternative splicing, transposable elements' activity and the involvement of additional cell components such as the mesenchyme in the mechanism of action of psychedelics.
RNA-seq data of rats chronically treated with LSD: Data collection is described in. Reads were trimmed and quality checked with TrimGalore (), aligned to the Rattus norvegicus genome (rn6) with Hisat2, and counts obtained with HTSeqcount. Resulting processed data were RPM normalized and only genes with at least 5 reads in at least 10 samples were retained for further analyses. Microarray data of rats after 90min of LSD treatment: Experimental setting and data collection are described in C. D.. Raw data were normalized with the rma function from the Affy package. Probes were mapped to gene symbols with the GPL85 annotation from Gene Expression Omnibus. RNA-seq of mice's cortical neurons upon treatment with DOI: Counts were downloaded from Gene Expression Omnibus (GSE161626). Technical replicates were averaged and then normalized and log transformed similarly to rats' PFC data. Splicing junctions were quantified with SGSeq. Transposable elements were quantified with TEtranscripts, after aligning the reads with Hisat2 allowing reporting up to 1000 mapping sites per read.
Subsampling was performed by randomly selecting 10 million reads from the original fastq files, and repeating the pre-processing, as described above.
Differential gene expression was performed with DESeq2on count data. Gene Ontology enrichment was calculated with the enrichGO function from the clusterProfiler package, using "Biological Process" or "Molecular Function" GO categories and default parameters. For modules' enrichment, all genes belonging to a module different from the "grey" (unconnected) module were used as background. Transcription Factors (TFs) were defined based on the Gene Ontology category GO0003700, downloaded from Biomart (). Targets of each TF were obtained from the ChEA database, and downloaded from(ChEA_2016). An exploratory enrichment analysis of differentially spliced genes has been performed with Enrichr (). The Wikipathway gene lists has been downloaded from(Wikipathway_2021_Human) and used for testing the enrichment of significantly differentially spliced genes. The GSEA analysis of genes with the highest entropy change was performed with the fgseaand msigdbr packages (), using the category "C5" (comprising GO categories).
The co-expression network was obtained with the blockwiseModules function from the WGCNA package, setting the parameters networkType and TOMtype to "signed", and beta=12. The module eigengene (ME) was calculated with the function moduleEigengenes, and weighted degree within each module (kWithin) was calculated with the function intramodularConnectivity.fromExpr, setting networkType="signed" and power=12. Differential activity of modules was tested comparing ME between treatment groups with the Wilcoxon test, and then correcting for multiple testing and obtaining false discovery rates for each module. Connectivity represents genes' weighted degree, with weights based on the adjacency matrix, either across the whole network or considering only a module's genes. The connectivity in a specific condition was obtained by calculating the adjacency using only samples belonging to that group Module eigengene in different datasets were obtained as the projection of samples on the first PC of a PCA (function prcomp) performed on rats' PFC data using only the genes of the chosen module.
Signalling entropy was calculated with the SCENT R package, either using the provided PPI network or the co-expression network built in this work. Input data (RNAseq) were log transformed with an offset of 1.1.
Between-sample entropy was obtained as 1-correlation, with the correlation obtained from pairwise Pearson's correlation between samples using gene expression profiles or splicing junction usage profiles.
ID mapping and orthologs identification was performed with the biomaRt R package. In case of multiple IDs mapping to the same symbol, the ID with the highest average expression across all dataset's values was chosen.
All statistical analyses were performed with R 4.0.4 (R Core Team, 2018). Packages used for plotting are R base graphics, ggplot2and ggsignif (), and pheatmap (project.org/package=pheatmap). Groups' means were compared with the Wilcoxon rank sum test, while correlations' statistical significance was calculated with the cor.test function. Only tests passing the threshold FDR<0.05 (false discovery rate) are shown. inhibited by siRNA on differention of rat marrow mesenchymal stem cells into neurons. Zhongguo ying yong sheng li xue za zhi = Zhongguo yingyong shenglixue zazhi = Chinese journal of applied physiology, 31(3), 254-258.
Create a free account to open full-text PDFs.
Carhart-Harris, R. L., Roseman, L., Bolstridge, M. et al. · Scientific Reports (2017)
Davis, A. K., Barrett, F. S., May, D. G. et al. · JAMA Psychiatry (2021)
Erritzoe, D., Roseman, L., Nour, M. R. et al. · Acta Psychiatrica Scandinavica (2018)
Family, N., Maillet, E. L., Williams, L. T. J. et al. · Psychopharmacology (2019)
Flanagan, T. W., Nichols, C. D. · International Review of Psychiatry (2018)
Gonza ´lez-Maeso, J., Weisstaub, N. V., Zhou, M. et al. · Neuron (2007)
Griffiths, R. R., Johnson, M. W. · Journal of Psychopharmacology (2016)
Lebedev, A. V., Kaelen, M., L€ Ovd En, M. et al. · Human Brain Mapping (2016)
Ly, C., Greb, A. C., Cameron, L. P. et al. · Cell Reports (2018)
Maclean, K. A., Johnson, M. W., Griffiths, R. R. · Journal of Psychopharmacology (2011)
Madsen, M. K., Fisher, P. M., Burmester, D. et al. · Neuropsychopharmacology (2019)
Martinotti, G., Santacroce, R., Pettorruso, M. et al. · Brain Sciences (2018)
Nichols, D. E. · Pharmacology and Therapeutics (2004)
Preller, K. H., Burt, J. B., Adkinson, B. et al. · eLife (2018)
Preller, K. H., Burt, J. B., Adkinson, B. et al. · Biological Psychiatry (2020)
Preller, K. H., Razi, A., Zeidman, P. et al. · PNAS (2019)
Dos Santos, R. G., Osório, F. L., Crippa, J. A. et al. · Therapeutic Advances in Psychopharmacology (2016)
Schartner, M., Carhart-Harris, R. L., Barrett, A. B. et al. · Scientific Reports (2017)
Schmid, Y., Enzler, F., Gasser, P. et al. · Biological Psychiatry (2015)
Studerus, E., Kometer, M., Hasler, F. et al. · Journal of Psychopharmacology (2010)
Vollenweider, F. X., Vollenweider-Scherpenhuyzen, M. F. I., Bäbler, A. et al. · NeuroReport (1998)
Vollenweider, F. X., Preller, K. H. · Nature Reviews Neuroscience (2020)