Abstract
Rare and Common Epilepsies Converge on a Shared Gene Regulatory Network Providing Opportunities for Novel Antiepileptic Drug Discovery
Delahaye-Duriez A, Srivastava P, Shkura K, Langley SR, Laaniste L, Moreno-Moral A, Danis B, Mazzuferi M, Foerch P, Gazina EV, Richards K, Petrou S, Kaminski RM, Petretto E, Johnson MR. Genome Biol 2016;17:245.
BACKGROUND: The relationship between monogenic and polygenic forms of epilepsy is poorly understood and the extent to which the genetic and acquired epilepsies share common pathways is unclear. Here, we use an integrated systems-level analysis of brain gene expression data to identify molecular networks disrupted in epilepsy. RESULTS: We identified a co-expression network of 320 genes (M30), which is significantly enriched for non-synonymous de novo mutations ascertained from patients with monogenic epilepsy and for common variants associated with polygenic epilepsy. The genes in the M30 network are expressed widely in the human brain under tight developmental control and encode physically interacting proteins involved in synaptic processes. The most highly connected proteins within the M30 network were preferentially disrupted by deleterious de novo mutations for monogenic epilepsy, in line with the centrality-lethality hypothesis. Analysis of M30 expression revealed consistent downregulation in the epileptic brain in heterogeneous forms of epilepsy including human temporal lobe epilepsy, a mouse model of acquired temporal lobe epilepsy, and a mouse model of monogenic Dravet (SCN1A) disease. These results suggest functional disruption of M30 via gene mutation or altered expression as a convergent mechanism regulating susceptibility to epilepsy broadly. Using the large collection of drug-induced gene expression data from Connectivity Map, several drugs were predicted to preferentially restore the downregulation of M30 in epilepsy toward health, most notably valproic acid, whose effect on M30 expression was replicated in neurons. CONCLUSIONS: Taken together, our results suggest targeting the expression of M30 as a potential new therapeutic strategy in epilepsy.
Commentary
Systems biology takes a “big-data” integration approach and focuses on interaction networks to understand complex systems (1). A basic tenet of this type of approach is that the whole is greater than the sum of its parts. Given that the brain is a complex system constantly engaged in a dynamic balancing act, a systems approach may have potential to uncover novel therapeutic strategies for epilepsy that are aimed at rebalancing the nervous system.
In the study by Delahaye-Duriez and colleagues, the authors integrated data from a number of sources to identify gene co-regulatory networks that contribute to epilepsy. The first step was to determine co-regulated genes from normal human brain tissue in the absence of disease. For this, they relied on a publicly available resource, the UK Brain Expression Consortium (UKBEC), which includes genome-wide gene expression across nine different brain regions (2). Using expression data from 88 post-mortem brains, they inferred 34 “consensus” co-expression modules that were common across brain regions and 13 “differential” modules where co-expression varied across brain regions. These modules were subsequently investigated in two additional, publicly available human gene expression datasets: Brainspan that covers fetal to postnatal human brain development (3), and GTEx that includes genotype and tissue expression data across multiple brain regions and nonbrain tissues (4). They found that approximately 60% of modules were preserved in the Brainspan and GTEx datasets. A final step was to compare across species and look for preserved modules in healthy mouse hippocampus RNA-seq datasets (5), which revealed approximately 58% of modules were preserved between human and mouse. From this comparative analysis, they identified 18 consensus and 80 differential modules that reproduced across varied brain expression datasets. These were hypothesized to be important gene networks critical for brain function.
Taking these networks forward, they next asked whether any of these modules were enriched for genes in which de novo mutations cause monogenic epileptic encephalopathy, autism spectrum disorder, schizophrenia, or developmental disorders. Only a single module—consensus module M30—was significantly enriched for epileptic encephalopathy genes, while no modules were enriched for the other neurodevelopmental phenotypes. On the one hand, this is remarkable as one might have expected to find enrichment in module(s) for other neurodevelopmental phenotypes. However, this may be due to selection bias in the way the modules were initially defined. The criteria required preservation across all ages and across species, which may have omitted age-sensitive modules or those not conserved between human and mouse. Seizures and epilepsy occur across a wide range of ages and species and, thus, it is not surprising that epilepsy genes reside in a more general module. Consistent with this, genes implicated in more common forms of epilepsy by a recent GWAS meta-analysis (6) are also enriched in module M30. Gene expression data from human temporal lobe epilepsy resections or mouse epileptic hippocampus (pilocarpine post-SE or Dravet syndrome) also showed enrichment for module M30 genes. The convergence of many lines of evidence on module M30 suggests that it is important for influencing susceptibility to seizures or epilepsy.
An interesting question becomes what comprises module M30? It contains many monogenic epilepsy genes as this was one feature used to define the module, but other M30 members have not yet been directly implicated in epilepsy. M30 contains a total of 320 genes and is enriched for genes involved in neural processes, such as “transmission of nerve impulse,” “synaptic transmission,” “synaptic vesicle transport,” and “GABA signaling.” The genes of module M30 begin expression in early to mid-fetal development, peak around birth, and persist through adulthood. Many M30 genes are targets of NRSF/REST regulation, which may underlie the observed co-expression. Downregulation of REST activity is associated with neuronal maturation and dysregulation of REST is observed in epilepsy, as well as other brain diseases (7). There is a high degree of protein–protein interactions among the M30 gene products (Figure 1), with the genes associated with monogenic epilepsy being among the most highly connected or, in other words, the monogenic epilepsy genes are often nodes.
Module M30 protein–protein interaction network. Proteins encoded by M30 genes have a high degree of protein–protein interaction. Image generated by STRING v.10.0 (string-db.org).
A final step in this study was to evaluate whether targeting of the M30 network might be a viable therapeutic strategy. A priori, one could look at the M30 gene list and see that many single molecule classes targeted by anticonvulsants are present, as might be expected. However, they used a purely data-driven approach leveraging Connectivity Map (CMap), a database that reports transcriptional signatures of 1,300 compounds (8). They examined a subset of 152 CMap genes that showed differential expression above a threshold cutoff in at least one cell line (all non-neuronal). The top-hit with the strongest overlap of drug-induced differential expression of M30 genes was valproic acid (VPA), an anticonvulsant that also acts as an HDAC inhibitor to modulate gene expression. Other hits included trichostatin A, another HDAC inhibitor, and Whithania somnifera (ashwagandha), which has been shown to have anticonvulsant properties (9). They replicated the VPA findings in an experiment in which they exposed neurons in culture to VPA and observed upregulation of ~50% of genes in the M30 module. This supports the possibility of in vitro screening for M30 upregulation as a potential drug discovery strategy. It would be interesting to determine whether other more targeted anticonvulsants influence M30 gene expression with chronic exposure, as well as evaluating whether compounds identified as REST modulators show anticonvulsant potential (10).
Overall, this study provides proof-of-principle that a systems biology approach may provide a novel therapeutic strategy. Additionally, the M30 module could potentially be leveraged as a resource for biomarker development. However, it is important to appreciate that the datasets used to feed the system strongly influence the output. A caveat of this particular study is that defining a network using monogenic epilepsy genes heavily weighted the outcome. In spite of this limitation, the valuable take-away message may be the gestalt that emphasizes a network approach to therapy over a focused single-target paradigm. Focusing on rebalancing brain networks toward the non-disease state may enable future development of anti-epileptogenic rather than anti-seizure drugs.