Genomics of Ischemia
The quest to understand the neurobiology of ischemic brain injury has relied on the investigational techniques available over the years. Historically, morbid anatomy was the approach used through the midportion of the last century.1 The establishment of the Journal of Neurochemisty in 1956 marks the assent of tools to understanding brain injury beyond the whole organ level. Cellular and molecular approaches supervened especially after the discovery of apoptosis or programmed cell death by Kerr and colleagues in 1972,2 which lead to a focus on cellular and molecular cell death mechanisms. We now know that ischemic cell death is accompanied by transcriptional and translational events that may modulate ischemic injury. The induction of death-inducing gene products, such as the caspases, and the regulation of prosurvival gene products, such as antiapoptotic bcl-2 family genes, are now well known.3 More broad-based approaches to gene modulation in disease, such as subtractive hybridization libraries, have been used to discover unanticipated changes in gene expression during ischemia. With the cloning of the mouse genome, screening of complete gene expression patterns became a reality. Affymetrix chips now make it possible to use a single array to survey expression from ≈40 000 gene transcripts simultaneously. Gene array experiments can now capture “snapshots” of dynamic cellular processes at particular times and under particular conditions. These expression patterns can then be reconstructed into a dynamic model of a cell or an entire organ. Although genome-wide expression profiling offers unparalleled data acquisition, it also presents new and daunting challenges in data acquisition, storage, analysis and mining of these very large data sets. Statistical methodology for gene expression applications is a rapidly “moving target”. Clustering has become a relatively popular method of exploring patterns of expression in microarray data. In the common scenario of functional interpretation, clustering results can be used to infer function when an unannotated gene clusters with genes of known function. However, it is also well known that each clustering method is biased toward searching for a particular type of pattern in the data. Some new approaches have been developed for these data sets. The identification of potential pathways from transcription factor analysis is an automated and scalable bioinformatics approach for the identification and analysis of candidate regulatory interactions in a specific experimental setting by detecting potential promoter sequences for genes of interest and then finding and identifying transcription factor-binding sites on the sequences. One can then analyze the transcription factor-binding site occurrences for over/under-representation compared with a reference (Figure 1).
Another approach to microarray data sets is visualization. A visual analysis system designed for integrating and concurrently analyzing the contents of large, complex, multimedia information collections incorporate pattern recognition algorithms and a variety of effective visualization tools. Figure 2 shows such an image in which information about input RNA, labeled cRNA target, and GeneChip assay performance for 49 different samples is displayed. The assays selected are those that produced the highest % Present Calls, a measure of array sensitivity and overall assay performance. The yellow lines linking values between rows allow one to see how individual samples or groups of samples perform in each measurement category. Not surprisingly, the most differentiated pattern in this image is that showing that high % Present is associated with low 3′/5′ ratios for the control genes, GAPDH and actin. Low 3′/5′ ratios for GAPDH and actin control probes sets on Affymetrix arrays are known to be associated with high quality, nondegraded input RNA. We are using Starlight visualization to explore large collections of data to facilitate our efforts to identify those criteria that best predict array performance.
The Genomics session is offered to show where the field is and what the potential might be. This session includes presentations on ischemic tolerance/preconditioning looked at by microarray technology and the new insights that this approach can bring. There is a presentation discussing an old area—the blood-brain barrier—that genomics is now addressing. There are 2 presentations on a remarkable work in progress—the genomic signature in blood induced by cerebral ischemia and a host of other disorders.
- Received July 14, 2006.
- Accepted August 6, 2006.