Search tips
Search criteria 


Logo of epicurrLink to Publisher's site
Epilepsy Curr. 2017 Nov-Dec; 17(6): 374–376.
PMCID: PMC5706363

A Look Behind the Curtain: Epilepsy Microarray Consortium


Transcriptional Profile of Hippocampal Dentate Granule Cells in Four Rat Epilepsy Models.

Dingledine R, Coulter DA, Fritsch B, Gorter JA, Lelutiu N, McNamara J, Nadler JV, Pitkänen A, Rogawski MA, Skene P, Sloviter RS, Wang Y, Wadman WJ, Wasterlain C, Roopra A.Sci Data 2017;4:170061. [PubMed]

Global expression profiling of neurologic or psychiatric disorders has been confounded by variability among laboratories, animal models, tissues sampled, and experimental platforms, with the result being that few genes demonstrate consistent expression changes. We attempted to minimize these confounds by pooling dentate granule cell transcriptional profiles from 164 rats in seven laboratories, using three status epilepticus (SE) epilepsy models (pilocarpine, kainate, self-sustained SE), plus amygdala kindling. In each epilepsy model, RNA was harvested from laser-captured dentate granule cells from six rats at four time points early in the process of developing epilepsy, and data were collected from two independent laboratories in each rodent model except SSSE. Hierarchical clustering of differentially-expressed transcripts in the three SE models revealed complete separation between controls and SE rats isolated 1 day after SE. However, concordance of gene expression changes in the SE models was only 26–38% between laboratories, and 4.5% among models, validating the consortium approach. Transcripts with unusually highly variable control expression across laboratories provide a 'red herring' list for low-powered studies.

The recent publication by Dingledine and colleagues is a unique report focused primarily on experimental design lessons learned from a large, multicenter gene expression profiling study in rat epilepsy models. Although the raw data was originally deposited in the NCBI Gene Expression Omnibus (GEO) database in 2013 (1), the current report gives an in-depth description of the study design and dataset, as well as providing guidance for future studies. This report is timely given the current attention on rigor and reproducibility in the scientific community.

Understanding the changes underlying epileptogenesis has been a long-standing goal of epilepsy research. The development of microarray technology in the 1990s enabled the interrogation of expression of tens of thousands of genes (2) and held promise for profiling gene expression changes during epileptogenesis. Numerous microarray profiling studies attempted to characterize changes in gene transcription that might provide insight into epileptogenesis (3–6). However, inconsistencies in results failed to provide a consensus list of transcriptional changes that underlie the epileptogenic process. Comparing results among studies was complicated by differences in study design, including model selection, tissue selection, assay platform, and analytic techniques. To address these shortcomings, a multicenter epilepsy microarray consortium was assembled with the goal of generating a robust transcriptional profiling dataset through the joint effort of seven sites in the United States and Europe.

The guiding principle was that robust gene expression changes observed across multiple sites and models are key drivers of epileptogenesis. Therefore, the study was rigorously designed to enable cross-lab and cross-model validation of data from a uniform cell type processed and analyzed using standardized methods on a single platform. The careful attention to study design provided the opportunity to empirically assess reproducibility across labs and models.

The study used four rat induced epilepsy models, including two chemoconvulsant models (pilocarpine, kainate) and two electroconvulsant models (self-sustained status epilepticus [SSSE], amygdala kindling). Three of these models (pilocarpine, kainate, SSSE) employ induction of status epilepticus (SE), while amygdala kindling does not involve SE. All groups used male Sprague Dawley rats obtained almost exclusively from a single vendor (Charles River, Wilmington, MA), with one exception. This is important as there are documented differences in epilepsy susceptibility following SE in Sprague Dawley rats obtained from different vendors (8–9). Of the two kainate model sites, one obtained Sprague Dawley rats from Charles River, while the other site obtained their Sprague Dawley rats from Harlan. There are reported differences in propensity for spontaneous recurrent seizure development and neuro-degeneration following SE in Sprague Dawley rats obtained from Charles River versus Harlan (8), making this an important caveat for cross-lab validation of the kainate data.

Previous microarray studies used a variety of brain tissue sources, contributing to the study-to-study variability in gene expression profiles. The epilepsy microarray consortium decided to focus on a single, defined cell type and used laser-capture microdissection (LCMD) to isolate dentate granule cells. This avoided complications of mixed cell types and cell-death that is prominent in other hippocampal regions like CA3 and CA1. Additionally, prior work suggested that dentate granule cells had a robust transcriptional response to seizures (6), and the dentate gyrus is an important gatekeeper for seizure generation in the hippocampal circuit. Tissue samples from each site were collected at several time points (1, 3, 10 days) following SE-induction or at specific stages of kindling (stage 2; 1st or 10th stage 4/5). Frozen samples were then sent to a single site for LCMD to avoid potential differences in technique or cell isolation across multiple sites.

Microarray technology relies on hybridization of sample input to a solid surface that is tiled with probes corresponding to known transcripts in the genome. Thus, it interrogates the sample of interest using a finite set of probes. This study used the RAE230A microarray that covers over 10,000 unique rat transcripts. Microarray has a dynamic range of 10^3–10^4, limited by background hybridization at the low-end and probe saturation effects at the high-end. Although the dynamic range is smaller than newer RNA-seq based approaches, microarray studies had a cost-savings advantage when this study was conducted. Microarray hybridization was performed at the NINDS NIMH Translational Genomics Institute (TGEN) in Phoenix, Arizona. This high-volume facility used standard sample handling and processing to generate gene expression data. Image files and hybridization data was deposited it the NCBI GEO database in accordance with minimum information about a microarray experiment (MIAME) standards for microarray studies (10).

Data processing followed standard techniques, with additional cross-laboratory and cross-model comparisons focused on the SE-based models. Examination of control rat samples from all sites showed an overall coefficient of variation (CV) of expression of 14%, which is on par with the expected within-group variation in a typical microarray experiment. These could be subdivided into a stable expression group with CV <3% and a highly variable group with CV >3 standard deviations above the mean. The authors suggest that the highly variable group of genes may be potential “red herrings” for smaller scale studies. Cluster analysis of the three SE models showed that the most consistent effects across sites are seen 1 day after SE, versus 3- and 10-day data that did not cluster consistently. Cross-lab and cross-model comparisons were largely incongruent, with only 4.5% of differentially expressed genes (DEGs) shared across models and labs. However, this list of 73 shared DEGs may represent a robust set of core epileptogenesis genes. Cross-lab within-model comparisons fared better, with 38% of shared DEGs for kainate and 25% of shared DEGs for pilocarpine. Final SSSE data were only obtained from a single site, precluding cross-lab comparison, while analysis of the amygdala dataset was not presented in this report.

Although there were many DEGs that were not shared between sites, this perhaps is not surprising as subtle differences in SE models between labs have long been acknowledged. A difference between labs that may be particularly important is the percentage of animals that go on to develop spontaneous recurrent seizures following SE. A limitation of an RNA-based study of early-stage epileptogenesis is that it is impossible to know which rats would have ultimately developed spontaneous recurrent seizures. Another potential source of variability is the degree of neurodegeneration. The study design did include conserving a portion of tissue for histopathology, so it is possible that this issue will be addressed in future reports from the epilepsy microarray consortium.

This report highlights the challenges of cross-model and cross-lab reproducibility. There are numerous models of induced epilepsy, and it is not clear which most closely approximates human post-SE epilepsy. The most possible answer is that each model approximates some aspects of the human condition, while no single model fully captures all of them. In terms of cross-lab validation, subtle variations in SE induction protocols between labs may contribute, along with potentially overlooked differences in animal care environment. This preliminary report supports multi-center design and cross-lab validation as a powerful approach for highlighting robust differences and filtering out lab-specific noise.

Supplementary Material


1. Dingledine R, Coulter D, Nadler JV, Wadman W, Wasterlain C, McNamara JO, Rogawski M, Sloviter R. Transcriptional profiling of dentate granule cells in 4 rat epilepsy models. 2013; NCBI Gene Expression Omnibus:GSE47752.
2. Schena M, Shalon D, Davis RW, Brown PO. Quantitative monitoring of gene expression patterns with a complementary DNA microarray. Science 1995; 270: 467– 470. [PubMed]
3. Newton SS, Collier EF, Hunsberger J, Adams D, Terwilliger R, Selvanayagam E, Duman RS. Gene profile of electroconvulsive seizures: induction of neurotrophic and angiogenic factors. J Neurosci 2003; 23: 10841– 10851. [PubMed]
4. Hunsberger JG, Bennett AH, Selvanayagam E, Duman RS, Newton SS. Gene profiling the response to kainic acid induced seizures. Brain Res Mol Brain Res 2005; 141: 95– 112. [PubMed]
5. Gorter JA, van Vliet EA, Aronica E, Breit T, Rauwerda H, Lopes da Silva FH, Wadman WJ. Potential new antiepileptogenic targets indicated by microarray analysis in a rat model for temporal lobe epilepsy. J Neurosci 2006; 26: 11083– 110110. [PubMed]
6. Borges K, Shaw R, Dingledine R. Gene expression changes after seizure preconditioning in the three major hippocampal cell layers. Neurobiol Dis 2007; 26: 66– 77. [PMC free article] [PubMed]
7. Cacheaux LP, Ivens S, David Y, Lakhter AJ, Bar-Klein G, Shapira M, Heinemann U, Friedman A, Kaufer D. Transcriptome profiling reveals TGF-beta signaling involvement in epileptogenesis. J Neurosci 2009; 29: 8927– 8935. [PMC free article] [PubMed]
8. Langer M, Brandt C, Löscher W. Marked strain and substrain differences in induction of status epilepticus and subsequent development of neurodegeneration, epilepsy, and behavioral alterations in rats. Epilepsy Res 2011; 96: 207– 224. [PubMed]
9. Brandt C, Bankstahl M, Töllner K, Klee R, Löscher W. The pilocarpine model of temporal lobe epilepsy: marked intrastrain differences in female Sprague–Dawley rats and the effect of estrous cycle. Epilepsy Behav 2016; 61: 141– 152. [PubMed]
10. Brazma A, Hingamp P, Quackenbush J, Sherlock G, Spellman P, Stoeckert C, Aach J, Ansorge W, Ball CA, Causton HC, Gaasterland T, Glenisson P, Holstege FC, Kim IF, Markowitz V, Matese JC, Parkinson H, Robinson A, Sarkans U, Schulze-Kremer S, Stewart J, Taylor R, Vilo J, Vingron M. Minimum information about a microarray experiment (MIAME) –toward standards for microarray data. Nat Genet 2001; 29: 365– 371. [PubMed]

Articles from Epilepsy Currents are provided here courtesy of American Epilepsy Society