Glass cDNA microarrays can be used to profile the expression of thousands o
f gene targets in a single experiment. However, the potential for hybridiza
tion cross-reactivity needs to be considered when interpreting the results.
Here, we describe hybridization experiments with a model array representin
g four distinct functional classes (families): chemokines, cytochrome P-450
isozymes. G proteins, and proteases. The cDNA clones selected for this arr
ay exhibited pairwise sequence identities ranging from 55% to 100%, as dete
rmined by a homology scoring algorithm (LALIGN). Targets for microarraying
were amplified by PCR and spotted in 4-fold replication for signal averagin
g. One designated target from each family was further amplified by PCR to i
ncorporate a T7 promoter sequence for the production of synthetic RNA trans
cripts. These transcripts were used to generate fluorescent hybridization p
robes by reverse transcription at varying input concentrations. As expected
, hybridization signals were highest at the matching target elements. Targe
ts containing less than 80% sequence identity relative to the hybridization
probe sequences showed cross-reactivities ranging from 0.6% to 12%. Target
s containing greater than 80% identity showed higher cross-reactivities (26
%-57%). These cross-reactive signals were analyzed for statistical correlat
ion with the length of sequence overlap, percent sequence identity, and hom
ology score determined by LALIGN. Overall, percent sequence identity was th
e best predictor of hybridization cross-reactivity. These results provide u
seful guidelines for interpreting glass cDNA microarray data.