CHARACTERISTICS OF RNA THAT AFFECT
cDNA GENERATION AND ARRAY HYBRIDIZATION
Christine L. Miller and Robert H.
Yolken
Different outcomes can be expected from
array hybridizations that utilize different methodologies. At least some
of the disparity may be attributable to the relative ability to access distinct
pools of RNA. We have compared two different methods of RNA extraction,
and two methods of oligonucleotide priming, in order to optimize the yield and
range of cDNA templates that can be generated from post-mortem brain
tissue. The cDNA product is influenced by the overall quality of the RNA
starting material, as well as structural characteristics of specific RNA species
that influence the ability of reverse transcriptase to carry out its
function. From the same pool of homogenized brain tissue, the Trizol
method of extraction resulted in a lower quality of RNA than did the Qiagen
method, as visualized on an agarose gel and reflected in quantification of
RT-PCR products. In addition, the RNA extracted by Trizol was
markedly less stable at 37°C. Although both methods of extraction yielded
RNA that primed better with random hexamers (RH) than with oligo-dT (dT), the
difference between priming with dT and priming with RH was much greater for the
Trizol product than the Qiagen product. Data will be presented to show
that such differences in priming likely result from structural features specific
to certain RNAs, structures which appear to be disrupted by the Qiagen column
method.
There remain, however, RNA species in
the Qiagen extract that do not prime with dT but prime very well with RH.
This finding would predict that array methodologies that rely on RH priming
would detect a greater percentage of targets than those that rely on dT
priming. Other points to be considered include the practice of pooling
samples, either at the RNA or cDNA stage, which could theoretically lead to
strong interactions between sense template present in one sample and antisense
template present in another sample. The methods of RNA extraction and probe
generation may be important variables in the performance and interpretation of
microarray expression analyses.
SLIDES: