Microarray approach for cloned library quality assessment and the comprehensive identification of positives from library screens

coding regions represented in the library. We chose to label the library DNA using random-primed synthesis that incorporated biotinylated nucleotides directly. Alternatively, we could have chosen PCR using biotinylated primers or nucleotides, but this approach might have selected against some classes of inserts (e.g., long inserts) and has other complications (see below). The library DNA (400 ng) was labeled in a 100-μL reaction using the BioPrime® DNA Labeling System (Invitrogen, Carlsbad, CA, USA) following the manufacturer’s protocol, except that we labeled for 2 h and used only 4 μL of stop buffer at the end of the reaction. We ran 5 μL of the material on a 4% polyacrylamide gel to monitor the DNA labeling and confirm the manufacturer’s estimate of 2–5 μg of synthesized product. After ethidium bromide staining, the newly synthesized DNA appeared as a strongly stained smear between 40 and 140 bp. An added benefit of using the random-primed labeling protocol is that these fragments are appropriately sized to provide a good population of targets for the microarray hybridization (3). This could be a significant issue if intact (i.e., much larger) labeled PCR products are used directly. For the chip hybridization, 80 μL of the labeling reaction were brought to a total volume of 305 μL in the standard hybridization solution recommended by Affymetrix. (In the preparation of the hybridization cocktail, we considered the 80 μL BioPrime reaction as if it were contributing only water to the mixture). The U133A GeneChip was pretreated, hybridized (45°C for 16 h), and washed using the manufacturer’s procedure typically used for RNA profiling analysis (3). The chip was scanned, and the data was processed by the MAS 5 software (Affymetrix) using the default settings for the analysis. For the spleen library hybridization shown in Figure 1, the software called one-third (33.7%) of the genes represented on the chip as “P” (present). While this is a respectable representation, it is somewhat lower than the percentages obtained from typical tissue-specific RNA profiling experiments (4). However, the overall signals we obtained in our hybridizations are also somewhat lower, indicating that our protocol using a few micrograms of double-stranded DNA for Microarray approach for cloned library quality assessment and the comprehensive identification of positives from library screens