Maize Genetics Cooperation Newsletter vol 81 2007
Microarrays have become a popular method to monitor gene expression levels on a genomic scale. We have been using the array produced by the Maize Oligonucleotide Array Project at University of Arizona. We have generally followed the protocols provided on the project website (www.maizearray.org), and we have also tried modifications. Based on our experience with microarray experiments, we have the following recommendations for people who may be interested in starting a microarray experiment. Please feel free to contact us if you have any questions or would like more information.
1. Successful modifications to the project protocols. We used the protocols for cRNA targets provided on the website with the following alterations.
a. During RNA purification step we adjusted the total elution volume to 100 �l (65 1st, 45 2nd) instead of the recommended 60 �l. Our yields ranged from 20-40 �g aRNA.
b. We experienced up to 50% reduction in yield during the Cy Dye coupling step due to cRNA adherence to the column. To help alleviate this we used 50�C DEPC water for the elution steps and also heated the entire column during the elution for ~ 5 minutes in a 50�C hybridization oven. This increased the yield of labeled aRNA to about 80%.
c. In fear of washing the oligos off the microarray slides, we opted to skip the rehydration steps as recommended in the protocol under DNA Probe Immobilization and simply cross-linked and washed the slides as described.
2. Use of aliens as a control feature for cRNA targets. Aliens are control RNAs that can be added to the total RNA as a standard for data normalization and scanning. Stragene alien sequences 1-10 are printed on the maize array. To take advantage of this control, we used mRNA spikes from the Stratagene SpotReport� Alien� cDNA Array Validation System in our amplification and hybridizations. In our hands, the aliens created problems during scanning as they drastically reduced the signals from other features. We also found that the aliens could not be used to manually adjust the scanner for equal red and green intensities. Our core facility has a Perkin Elmer ScanArray 4000 and Genepix software for microarray scanning and analysis. Using this scanner and software the auto PMT setting was found to be optimal for adjusting the signal intensities for all scans (the saturation levels were adjusted from the default settings of .05% to .005 % when using the auto PMT setting).
3. Use of Dyesaver for fluor preservation. Dyesaver, by Genesphere, is a toluene-based material coating which is applied to the slides after hybridization and washing. It is recommended to help preserve the fluorochromes from degradation, especially the Cy5 which is more easily degraded than the Cy3 dye. Our experience is that Dyesaver is expensive and may not be necessary for repeated scanning. As an experiment, we used the �practice� slides supplied to us by the Maize Array Project for two identical hybridizations, one with Dyesaver and the other without. The data from both slides produced similar results. The slide without the Dyesaver was scanned at least 4 times with only a minor loss in fluor intensity, using the auto PMT setting, with laser power settings between 70-90%. Slides that were coated in Dyesaver did maintain their integrity for several months, unlike untreated slides which expire rapidly. The major disadvantages of Dyesaver were the toxic toluene fumes which made it unpleasant to work with, high evaporation rate of the dye during storage drastically reducing the number of slides on which one can actually use the dye, and the overall green hue it gives to the slides.
4.
Data analysis using Bioconductor
freeware (bioconductor.org), which uses the R computing environment (www.r-project.org),
requires writing customized Perl scripts.
The main advantage to using Bioconductor is that it is one of the
most powerful software packages to use for microarray statistical
analysis. The main disadvantage,
however, is that it uses a command driven user interface and therefore is not
user friendly for most scientists.
We needed to create customized scripts effectively filter and normalize
our data. In Bioconductor we used
the Limma package (also available at www.r-project.org), which uses the
empirical Bayesian method to create a linear model to evaluate genes with
significant differential expression.
We would be happy to share our scripts for filtering, normalization and
linear model analysis.
Please Note: Notes
submitted to the Maize Genetics Cooperation Newsletter may be cited only with
consent of authors.