Unambiguous Identification of the Expressed MAGE-A Genes on a DNA Microarray

To the Editor:

We have described a post-PCR detection method for the 12 MAGE-A sequences on a DNA microarray (1) and compared the results with a method using pairs of primers unique for each sequence (2). The microarray assay did not differentiate between MAGE-A3 and MAGE-A6 amplicons, which differed by only 1 nucleotide, and could not distinguish between PCR products amplified from mRNA and those amplified from genomic DNA. In addition, the assay could not identify false-negative results related to RNA degradation or to enzyme inhibition during reverse transcription-PCR.

Here we describe an assay that is designed to overcome these drawbacks and appears to be more sensitive. We used 3 primer pairs: 1 for amplification of the 12 MAGE-A sequences (1); 1 for specific amplification of MAGE-A3; and 1 for amplification of an endogenous, ubiquitously expressed control gene (MAGE-D2) (3). The low-density microarray includes new capture probes for MAGE-A3 and -D2. The assay involved DNase treatment of total RNA, reverse transcription of mRNA with oligo(dT) primer, triplex PCR amplification in the presence of biotin-dATP/biotin-dCTP, and hybridization of the resulting amplicons on a DNA microarray (see the Experimental Protocol in the Data Supplement that accompanies the online version of this Letter at http://www.clinchem.org/content/vol51/issue12).

The new capture probe for MAGE-A3 corresponded to a sequence downstream from the location of the reverse primer DPASCONB4, and a primer pair (DPSA3 and DPASA3) defining a MAGE-A3 amplicon encompassing the new probe sequence was added to the PCR mixture. With a sample expressing MAGE-A3, this method generated 2 MAGE-A3 amplicons: one by extension of the consensus MAGE-A primers and the other by extension of the MAGE-A3 primers; only the latter, however, was detected on the microarray carrying the new MAGE-A3 probe. We verified the absence of cross-hybridization of MAGE-A6 sequences with the new MAGE-A3 probe by analyzing 8 tumor samples known to express MAGE-A6 but not MAGE-A3 (as determined by the comparison method). MAGE-A6 was detected in all samples, whereas no signal was detected with the MAGE-A3 probe in any sample.

To determine the detection limit of the assay, we tested various amounts (1.0, 0.1, 0.3, and 0.01 μg) of DNase-treated total RNA from MZ2-MEL.3.0 cells. This melanoma cell line expresses MAGE-A1, -A2, -A3, -A6, and -A10 and weakly expresses MAGE-A5 and -A12 (2)(4). With 1 μg of RNA, all of the expected MAGE-A genes were detected (see Table 1 in the online Data Supplement). With <1 μg RNA, only MAGE-A5 and -A12 were not detected. With 0.01 μg of RNA, MAGE-A1, -A2, -A3, -A6, and -A10 were still detected. Compared with the previous microarray assay (1), the present results indicate an improvement in sensitivity by a factor of 10 for the detection of MAGE-A2, -A3, and -A10; by a factor of 33 for MAGE-A12; and by a factor of 100 for MAGE-A1 and -A6. This improvement probably results from the labeling of the amplicons with 2 nucleotides (biotin-11-dATP and biotin-11-dCTP) rather than biotin-16-dUTP alone (1).

We analyzed 52 cDNA samples from tumor tissues and cell lines. Two independent PCR amplifications were performed on each cDNA sample, and the resulting products were hybridized on separate microarrays. For genes MAGE-A1, -A2, -A3, -A4, -A6, -A10, or -A12, we observed a 100% correlation between the results of the 2 methods when the gene expression determined with the comparison method was >10% of that found in the reference cell line and a 98.3% correlation when no expression was detected with the comparison method (Table 1⇓ ). Discrepancies were observed only for MAGE-A12 in 3 samples. A possible explanation for the negative results by the comparison PCR assay is that MAGE-A12 is not detected optimally because the sequence of the sense primer is located close to the 5′ end of the first exon of MAGEA12. The sensitivity of this assay could be improved by selecting sense and antisense primers in the last exon of MAGE-A12 (F. Brasseur, personal communication). In the MAGECHIP® assay, the MAGE-A12 amplicon derives from a sequence located entirely in the last exon.

Tumor samples with MAGE-A expression ≥3% may express enough MAGE antigen to be recognized by cytolytic T lymphocytes (5). When the expression of MAGE-A1, -A2, -A3, -A4, -A6, -A10, or -A12 determined by the comparison method was ≥3%, we observed a correlation of 98.8% with the MAGECHIP results (Table 1⇑ ).

The inclusion of a capture probe for MAGE-A7 (pseudogene) allows detection of false positives resulting from the presence of contaminating genomic DNA in the RNA. Because the PCR primers are from a single exon (for MAGE-A) or from exons flanking a 190-bp intron (for MAGE-D2), PCR products can be generated from DNA and can contribute to the signal. After DNase treatment, we found no signal for the MAGE-A7 probe in any sample. The MAGE-D2 control was clearly positive in all samples.

In conclusion, the assay described here should facilitate tumor diagnosis related to therapeutic vaccinations involving MAGE-A gene products.