Total RNA was extracted from abomasal tissue using TRIzol reagent (Invitrogen Inc., Carlsbad, CA) according to manufacturer’s protocols. Quantity of RNA was assessed using a UV spectrophotometer, and gel electrophoresis was used to ensure that RNA was of high quality with minimal degradation.
Label
Cy3, Cy5
Label protocol
Total RNA from abomasum (12 µg) was used as a template for reverse transcription (Superscript III, Invitrogen). Resulting amino-allyl modified cDNA was then labeled with either Cy3 or Cy5 dye (Amersham Pharmacia Biotech, Piscataway, NJ). Unincorporated dye was removed, and labeled cDNA was concentrated to a volume of less than 10 µL using Microcon 30 spin columns (Millipore Corp.). Concentrated cDNAs were mixed with 100 µL of SlideHyb #3 (Ambion, Inc., Austin, TX) and incubated at 70°C for 5 minutes before hybridization to glass microarrays.
Total RNA was extracted from abomasal tissue using TRIzol reagent (Invitrogen Inc., Carlsbad, CA) according to manufacturer’s protocols. Quantity of RNA was assessed using a UV spectrophotometer, and gel electrophoresis was used to ensure that RNA was of high quality with minimal degradation.
Label
Cy3, Cy5
Label protocol
Total RNA from abomasum (12 µg) was used as a template for reverse transcription (Superscript III, Invitrogen). Resulting amino-allyl modified cDNA was then labeled with either Cy3 or Cy5 dye (Amersham Pharmacia Biotech, Piscataway, NJ). Unincorporated dye was removed, and labeled cDNA was concentrated to a volume of less than 10 µL using Microcon 30 spin columns (Millipore Corp.). Concentrated cDNAs were mixed with 100 µL of SlideHyb #3 (Ambion, Inc., Austin, TX) and incubated at 70°C for 5 minutes before hybridization to glass microarrays.
Hybridization protocol
Abomasal samples were hybridized to Michigan State University’s National Bovine Functional Genome Consortium (NBFGC) cDNA microarray according to procedures outlined by Suchyta et al., 2003-physiological Genomics. Hybridized arrays were washed at room temperature in 0.2X SSC, rinsed in double-distilled water, and dried by centrifugation.
Scan protocol
Images were scanned using the GeneTAC LS IV (Genomics Solutions) scanner and accompanying software (version 3.01).
Description
Spot alignment was performed using MolecularWare DigitalGENOME Pro 2.5 (MolecularWare, Cambridge, MA), and spots were visualized for abnormalities. Spots that were abnormal in shape, intensity saturation, or array imperfections were removed from subsequent analyses. Total intensity values were obtained for each dye channel and normalized as described below to determine differences in gene expression.
Data processing
Scanned florescent intensities for each individual array were log2 transformed to correct for right-tailed skewness. Transformed data were inspected for Cy3 versus Cy5 dye bias using M vs A scatter plots for each array, where log intensity ratios M = log2(Cy3/Cy5) were plotted against mean log intensities A = [log2 (Cy3) + log2 (Cy5)]/2 for each spot on the array (Yang et al., 2002). Adjustments were made for differential dye binding using locally weighted regression procedures (PROC LOESS) of SAS (SAS Institute, Cary, NC) (Madsen et al., 2004)). Boxplots of normalized data from each array were compared, and median intensity and variance were found to differ. To directly compare arrays, median intensity value was subtracted from all values for each array to obtain a median intensity of zero. Deviations from the geometric mean of each array were calculated, and the mean of the absolute value of the deviation was used to adjust the variance of each array. This adjustment equalized variance across arrays within day and infection status. Raw images and LOESS-transformed normalized spot intensities are available at http://www.ebi.ac.uk/microarray-as/ae/. A two-step mixed model analysis using the mixed-model procedure of SAS (SAS Institute, Cary, NC) was then performed (Wolfinger et al., 2001). The first step adjusted for array-specific variation using the following model:
yijk = µ + αi + βj + εijk
where yijk is the normalized intensity value, µ is the overall mean, αi is the random effect for array i, βj is the fixed effect for dye j, and εijk is the residual. Residuals were then carried over to the second step, which tested for gene-specific effects of breed composites using the model:
εijkl = µl + αil + βjl + γkl + βγjkl + φijkl
where all effects have the same definition as before except effects are now gene-specific (indicated by subscript l), γkl is the fixed effect for breed type k, and φijkl is the residual. Differences in gene expression of hair and wool sheep were assessed using two-tailed t-tests of least square means for each experimental group.
