Genetic Networks of Drug Resistance in Epilepsy: Outputs of a Computational and Translational Disease Framework

Abstract

Identification of Gene Regulatory Networks Affected Across Drug-Resistant Epilepsies

François L, Romagnolo A, Luinenburg MJ, Anink JJ, Godard P, Rajman M, van Eyll J, Mühlebner A, Skelton A, Mills JD, Dedeurwaerdere S, Aronica E. Nat Commun. 2024. 15(1):2180. PMID: 38467626. doi: 10.1038/s41467-024-46592-2

Epilepsy is a chronic and heterogeneous disease characterized by recurrent unprovoked seizures, that are commonly resistant to antiseizure medications. This study applies a transcriptome network-based approach across epilepsies aiming to improve understanding of molecular disease pathobiology, recognize affected biological mechanisms and apply causal reasoning to identify therapeutic hypotheses. This study included the most common drug-resistant epilepsies (DREs), such as temporal lobe epilepsy with hippocampal sclerosis (TLE-HS), and mTOR pathway-related malformations of cortical development (mTORopathies). This systematic comparison characterized the global molecular signature of epilepsies, elucidating the key underlying mechanisms of disease pathology including neurotransmission and synaptic plasticity, brain extracellular matrix and energy metabolism. In addition, specific dysregulations in neuroinflammation and oligodendrocyte function were observed in TLE-HS and mTORopathies, respectively. The aforementioned mechanisms are proposed as molecular hallmarks of DRE with the identified upstream regulators offering opportunities for drug-target discovery and development.

Commentary

Identification of mechanisms underlying drug resistance is a clinical and research priority in epilepsy, in which a third of patients fail to control their seizures with antiseizure medications. It accelerates identification of novel pharmacological targets and preclinical research, ultimately bringing rational and meaningful new treatment options.

With the advent of multiomics and the plethora of information that comes with it, gene functional annotation, that is, the process of describing a gene's biological identity is crucial to begin linking genetic findings to health and disease. This first step consists of determining the gene's biological functions, pathways, localization, molecular function, (sub)cellular location, and expression domain. Subsequently, functional enrichment analysis allows identification of overrepresented genes that might have a relationship with disease phenotypes. Determining how these overrepresented genes and increased transcription can result in molecular changes and phenotypes associated with disease, and more so severity of disease symptoms, remain a challenge.

“Causal Reasoning Analytical Framework for Target discovery” (CRAFT) is a computational framework for drug target discovery that combines gene regulatory information with causal reasoning, proposed and validated by Srivastava et al, first in animal models of epilepsy and correlated with human brain tissue. ¹ CRAFT applies a systems genetics approach starting from gene expression data from target tissue, providing a predictive framework for identifying cell membrane receptors with a direction-specified influence over disease-related gene expression profiles. ¹ It identifies potential upstream regulators by predicting the interaction between cell membrane receptor proteins, transcription factors, and downstream target genes. ²

The study by Francois et al applies the CRAFT platform to 162 brain tissue specimens from operated epilepsy patients, to identify molecular underpinnings of drug resistance. ² Their study adds dimensionality and further translational value to comparisons between disease and health, as similarities and discrepancies in pathways extracted from CRAFT in animal models and human disease can be further validated and explored, correlated to neuropathological diagnoses.

The studied cohorts comprised typical cases of drug resistance, that is, temporal lobe epilepsy with hippocampal sclerosis (TLE-HS, n = 64) and epilepsies with malformation of cortical development (MCD) including focal cortical dysplasia (FCD) (n = 17 FCD IIa, n = 33 FCD IIb) and tuberous sclerosis complex cortical tubers (n = 21). Epilepsy tissue samples were further compared to control tissue from autopsy (cortex and hippocampi). The authors worked under the hypothesis that data driven gene coexpression modules can build a global model of epilepsy pathobiology based on the assumption that some biological pathways may be differentially regulated in the disease state due to perturbations of gene expression. ²

They performed RNA sequencing from brain tissue extracts and a total of 28 366 genes underwent downstream analysis. The gene coexpression module analysis identified modules related to similar biological functions across the different epilepsy patient cohorts, but also some cohort-specific modules.

To further understand these modules, the authors studied how they clustered in “regulomes,” based on their transcriptional regulation. Four categories of regulomes, based on their relationship with the pathological state of the tissue, were characterized as follows: constitutive (no change between controls and epilepsy), enhanced (significant increased activity in epilepsy), activated (only present and active in epilepsy), and pathology-specific (differentially coexpressed in a specific epilepsy cohort) regulomes.

A total of 29 regulomes (with two to 10 gene modules) were identified, 14 of which with a predicted effect in epilepsy. Enhanced regulomes were identified in neuronal function, neuroinflammation, immune response, neurotransmission and synaptic plasticity showing enrichment for nicotine signaling, transmission across chemical synapse, and chemical synaptic transmission. Activated regulomes were identified for brain extracellular matrix and energy metabolism (oxidative phosphorylation/respiratory electron transport). Pathology-specific regulomes were found for neuroinflammation and immune response in TLE-HS and neuronal support and myelination in mTORopathies.

With this landscape, identifying transcriptional regulators for these regulomes opened the possibility to understand how their up or downregulation would impair or enhance downstream biological processes, likewise what was pursued by Srivastava et al in epilepsy animal models when proposing CRAFT as framework. ¹

The study then geared toward further understanding the activated regulomes linked to energy metabolism. The authors first identified KDM1A, which has been reported to modulate oxidative phosphorylation in metabolic tissues by genome-wide binding and transcriptome analyses, as a common transcriptional regulator. Although no cell-specific enrichment was identified, they then confirmed through immunohistochemistry that the epilepsy tissue showed a consistent expression of KDM1A in astrocytes. They aimed at determining the effect of KDM1A downregulation and subsequently performed in vitro validation of the role of KDM1A in fetal astrocytes derived from abortion at 12–16 weeks gestation, treated at 3 and 6 h with PMA/ionomycin stimulation (a process that activates T-cells and induces the production of cytokines). This led to impairment of cell metabolism including mitochondria electron transport chain, response to oxidative stress, oxidoreductase complex signaling, ATPase activity, and cellular respiration. In addition, there was impairment of inflammatory response pathways including IL-1 mediated signaling pathways, NF-kB signaling, T- and B-cells receptor signaling pathways.

Through a computational framework for clustering of genes modules that are altered in brain tissue resected from patients with drug-resistant seizures, which had been previously validated in animal models of disease, this study steps forward in the construction of a translational disease framework. Studying different pathologies allows for a larger transcriptomics dataset and makes comparison across various underlying pathologies possible, although in this cohort there were no patients under the category of MRI-negative or nonspecific tissue pathology. MRI-negative patients represent a significant proportion of surgical candidates, with an even greater potential to benefit from pharmacological developments that could arise from common unraveled mechanisms, in particular if found to share common dysregulations with a “lesional” subgroup.

Another important aspect of this cohort of mixed pathologies is that of determining relevance of common versus divergent findings (dysregulations). It might be easier to associate divergent findings in different pathologies sharing a common endpoint of drug resistance in a causal relationship. Common dysregulations however, might be the result of the establishment of drug resistance or a co-occurring process that unfolds as pathophysiology of drug resistance develops. CRAFT innovative framework predicts that these steps can be experimentally tested and back tracked by identification and modulation of upstream regulators and quantification of downstream targets, such as demonstrated in this study and this remains promising.

Whereas common modules and regulomes associated with drug resistance could be identified in the epilepsy pathologies studied here, it remains to be confirmed if this could be extrapolated to a larger but more homogeneous subgroups. In the MCD cohort studied, not all affected modules identified could be biologically annotated in the present study. Although FCDs more frequently occur in frontal and temporal cortices, they can still be found in various cortical topographies. Differently from MTLE-HS, in neocortical epilepsy, analysis of cortex tissue from different anatomical neocortical regions poses a challenge when amassing enough control samples for each patient-specific resected area is not possible. This might be important when interpreting results in this subgroup of patients as compared to controls. Future work could include comparisons of more homogeneous cohorts of FCD tissue (germline and somatic mutations profile as well as anatomical topography of the lesion) and statistically adequate number of corresponding control samples.