Microarray data affords a broad survey of gene expression and mutations and is, therefore, a particularly valuable technology. Unfortunately, large amounts of tissue are needed for this purpose and the analyses are expensive. The HBTRC has found that while multiple laboratories request tissue for such studies, we have concluded that it is not efficient or cost effective to distribute precious material to all such laboratories. Also, such studies have the potential to result in patentable findings, which may result in delays in publications for proprietary reasons. The inclusion of such microarray data into our National Brain Databank will be a valuable resource for the neuroscience community (see below); however, the use of different technologies by other laboratories and delays in the inclusion of data into the Databank because of proprietary considerations would compromise the success of this initiative. With consultation from our External Scientific Advisory Board, we have concluded that these problems can be overcome by having the HBTRC perform gene expression profiling and SNP analyses using the most broadly employed technology available and immediately placing the data into the National Brain Databank. The Affymetrix Microarray Analysis system (Fig. 1) has become the "gold standard" for this type of work and it is this that we have deployed at our facility.

	Fig. 1:   Affymetrix Gene array at low power (left) and with the zoom function, the probe sets are visible.

Fresh frozen post-mortem tissue is available and used for all cases included in the National Brain Databank. Approximately 50mg of tissue is needed to generate an RNA extract with sufficient RNA to generate a reliable and repeatable signal using the Affymetrix Microarray System. Extraction of total RNA is performed using the Totally RNA(TM) Kit from Ambion, the RNA Company and extracts are further "cleaned-up" using a Qiagen RNeasy(R) kit. The total RNA aliquots are stored at -70 C until further batch processing. The quality of the RNA is assessed using an Agilent 2100 Bioanalyzer to obtain an electropherogram of the 18S and 28S peaks (Fig.2). If the RNA extract does not show an acceptable profile and ratio, the extraction procedure is repeated. Tissue pH is computed for each case, however the 18S/28S ratio has been found to be a much more reliable index of the RNA integrity

Biotinylated target RNA is prepared from the Total RNA extracts of each patient according to recommended Affymetrix Protocols (see flowchart at http://www.affymetrix.com/support/technical/datasheets/t7_datasheet.pdf). Eight ug total RNA is used for cDNA synthesis with a SuperScript II double-stranded cDNA synthesis kit (Invitrogen, Carlsbad, CA). The SuperScript II kit employs a standard two-step reverse transcription reaction to generate cDNA complimentary to target mRNA. In vitro transcription of biotinylated RNA (bRNA) is performed with an Enzo-IVT kit (Affymetrix, Santa Clara, CA), which utilizes T7 RNA polymerase to produce

Fig. 2:  rRNA electropherogram showing 18S and 28S peaks from an HBTRC case.

a linear amplification of target cDNA generated during the previous reverse transcription step. Target bRNA from each case is then fragmented and hybridized to an Affymetrix Genechip (both human U-95A and U-133 chips are included in the databank) Genechip arrays are stained in the Affymetrix Genechip Fluidics Station 400 and scanned twice in an Agilent Genearray Scanner 2500. The microarray scans are visually inspected for evidence of artifacts, including "edge" effects or scratches, which may lead to aberrant readings for genes located on those portions of the chips. If important genes of interest are not located in these regions, the defects are edited out. In practice, however, fewer than 5% of our chips show any such artifacts. The Affymetrix MicroArray Suite 5.1 software is used to control the scanner, average the two scanned images, define the probe cells, and compute an intensity value for each cell.

Assessing the quality of the scanned chip images is a critical quality control step for the National Brain Databank. Affymetrix's MAS software allows us to examine the image and numeric data for each individual gene or "probe set" (i.e. consisting of the series Perfect Match (PM) and Mismatch (MM) probe cells) on each chip. Values for % genes present, gene 3'/5' ratios (an indication of mRNA quality), and chip scale factor are all returned by the MAS software and used together with Agilent rRNA 18S/28S ratios to make a decision on whether to include or exclude a chip (and thus case) in the National Brain Databank. Generally speaking, an acceptable level of "present" calls is approximately 35-45%. In practice, the average %"present" calls obtained at the HBTRC has been in the range of 41-52%, suggesting that our RNA extracts are of reasonably high quality. When the "present" calls fall below 35%, the sample extraction and assay are repeated; however, if there is no improvement, the case may be rejected for use in gene expression profiling studies. The DNA Chip Analyzer (dChip) software package recently developed by Li and Wong (2001) at the Harvard University School of Public Health is being used to further evaluate the quality of the chip data from a given run, by assessing the standard errors for the expression indices for each probe array. This is a statistical method for reducing errors due to non-specific cross-hybridizing probes and image contamination. Since the probe array results are being placed in the National Brain Databank as raw data, no further analyses will be performed by the HBTRC; however, investigators who are granted access to the database will be able to adjust the stringency of their analyses as desired.

After investigators have analyzed the gene expression profiling data present on the National Brain Databank and developed specific working hypotheses, the HBTRC will provide frozen sections postmortem human brain tissue to perform post hoc studies of genes-of-interest defined by the microarray studies. In the dorsolateral prefrontal cortex, the tissue is particularly abundant, while in others, such as the substantia nigra or amygdala, the quantity is much more limited, making it more challenging to satisfy investigator needs. We will begin by providing at least a few investigators with frozen sections cut under highly controlled conditions and placed either in vials or mounted on slides. In the past, the HBTRC would typically send blocks, but this severely limits tissue availability, particularly in brain disorders where the key region-of-interest is extremely small.