An evaluation of the discriminating probe sets selected from the U133A and U133B microarrays reveled that 20% of the identified class discriminators correspond to genes that were represented on the U95Av2 microarray, but had not been selected as class discriminators using the latter platform. A possible explanation for this observation is that the improved oligonucleotide design of the U133 microarrays results in an increase in both sensitivity/specificity, and thus allows the identification of class discriminators that had previously on the U95Av2 microarray fallen below statistical significance. To confirm this possibility, real-time RT-PCR (Taqman; Perkin-Elmer/Applied Biosystems, Foster City, CA) assays were developed for eleven genes that met these criteria. The probes selected were against discriminating genes for BCR-ABL (Neuropillin1, 212298_at; and ATP10C, 214255_at), E2A-PBX1 (FLJ12280, 213909_at; and FAM3C, 201889_at), Hyperdiploid > 50 chromosomes (EST, 212419_at), T-ALL (RAB32, 204214_s_at; and TRIM, 217147_s_at), and TEL-AML1 (Desmocollin 3, 206033_s_at; Epsin 2, 203464_s_at; NOVA1, 205794_s_at; and HAP1, 211222_s_at). Taqman primers were developed for each of these genes using the target sequences against which the microarray oligonucleotide probes were designed. Amplified product was quantitated using SYBR green and the resulting values normalized using the commercially available Taqman GAPDH primer and probe sequences.
RNA from representative diagnostic bone marrow samples of each class (four per class) were diluted to 35 ng/ml then treated for 15 minutes with 1.0 unit of DNase I (Invitrogen, Carlsbad, California) using the Invitrogen protocol to remove any contaminating DNA. Samples were reverse transcribed following the Taqman RT protocol using 1X RT mix (Perkin-Elmer/Applied Biosystems) and MultiScribe (Applied Biosystems). Samples were incubated at 25oC for 10 minutes, 48oC for 30 minutes, 95oC for 5 minutes, then placed on ice for 2 minutes. Real-time PCR was performed on a PE Applied Biosystems 7700 prism using oligonucleotide primers designed using Primer Express. The following primers were used: (1) Desmocollin 3, 5'- GCA ACC AGT ATC ACT TCC CTG TT -3' (forward primer), and 5'- GAA TTA CAA ATT CGG GCA TAC ATG -3' (reverse primer); (2) Epsin 2, 5'- GGA GGT GCA ATG GGA TGG -3' (forward), and 5'- GCC CTG CAC CGT CTG AAG -3' (reverse); (3) Nova 1, 5'- CAT CCC AGC TGC TCC TTT CT -3' (forward), and 5'- AGG CTG GAC GAA ATT CAG ACA -3' (reverse); (4) HAP1, 5'- AGC GCC TTC CCA ACA ATG T -3' (forward), and 5'- ACA AGA ACC AGG GTG GCT ACC -3' (reverse); (5) Neuropilin 1, 5'- AAA GAC ATT TGT TGG GAG TCA CAT T -3' (forward), and 5'- CAT GGT GAT CAA TAT TTT CCT GGA A -3' (reverse); (6) ATP10C, 5'- TTG AGT GCA TCC CAG CAT TC -3' (forward), and 5'- CCA GGG ATG TAA TAC CTA CAA GGC -3' (reverse); (7) FLJ12280, 5'- TTG CCT TGC TTA GAG AAT TAC TGC -3' (forward), and 5'- ACC CTG AGG GCC TAG AAA TCT G -3'; (8) FAM3C, 5'- TGT GTA TGA AGT ATC TCA AAC TGG AAC AT -3' (forward), and 5'- ATG CAT CTA AAC TTT AGG TTC GAA ATT -3' (reverse); (9) FAB32, 5'- CCC TGC TGT CCT CTT GGC T -3' (forward), and 5'- AAT TGG TCC ACC TGG GAA GG -3' (reverse); (10) TRIM, 5'- TTA CAA AGT ATT TTT CCC AAA GAT AGC TT -3' (forward), and 5'- TTT CTC CAA GTG ACT ATC TCT GGC TAG -3' (reverse); and (11) 212419: EST, 5'- GAG AAG GCT CCG ACG TCT CC -3', and 5'- CCA GGA ATA AGG CAC ACA ACG -3' (reverse).
The PCR reactions and amplication protocols were set up as previously described.3 The reverse transcribed cDNA was amplified in 1X SYBR green master mix containing a Taq Polymerase with 100 nM each of a forward and reverse primer. The GAPDH control for each cDNA was set up according to the Applied Biosystems protocol. Amplification proceeded as outlined in the “Quickstart Guide: ABI PrismTM Dissociation Curve Software” instructions. Data collection was performed by linking a dissociation curve profile on the end of the PCR run. Dilution controls were set up for each gene of interest as well as for GAPDH, and standard curves were generated at the end of the run. The expression levels were quantitated for each gene, and then normalized to the quantitated GAPDH. The normalized real-time result for the four samples of each class were average together and the mean for that class was compared to the mean intensity value for the same four samples from the microchip array. This comparison is shown in Table S18 below.
Although the selected probe sets were from the lower end of the ranked list of discriminating genes for each class, their expression was significantly higher by Taqman measurements in the identical class that was identified to have high expression by microarray analysis. The high degree of correlation between the microarray data and that obtained using Taqman methodology suggests that the U133 microarray platform was correctly identifying these genes as class discriminators.