BioProject

The objective of the present study was to use our own fabricated bovine liver cDNA microarray to profile genome-wide expression patterns of genes in the liver of cow throughout pregnancy. More...

The objective of the present study was to use our own fabricated bovine liver cDNA microarray to profile genome-wide expression patterns of genes in the liver of cow throughout pregnancy. Ten-microgram of pooled liver total RNA obtained from nonpregnant cycling cows was reverse transcribed in the presence of cyanine-3-fluorescent dye (Cy3)-conjugated dUTP (Amersham Pharmacia Biotech) by using Superscript II reverse transcriptase (Life Technologies, Rockville, MD) to yield Cy3 fluorescence-labeled control cDNA probe. Similarly, liver total RNA obtained from cows on days 19, 27, 49, 58, 150 and 245 of pregnancy was separately reverse transcribed in the presence of Cy5-conjugated dUTP by using Superscript II reverse transcriptase to yield Cy5 fluorescence-labeled test cDNA probe. After a 2-h reverse transcription reaction at 42oC, RNA was degraded by addition of 1.5 µl of an alkaline solution (1N NaOH / 20 mM EDTA), and neutralized by addition of 1.5 µl 1N HCl. The labeled control cDNA probe and test cDNA probe were then mixed and washed by centrifugation with TE (Tris-EDTA, pH 7.0) first and then with TE, 20 µg Salmon sperm DNA, 20 µg polyA and 20 µg yeast RNA in a Microcon YM-30 column (Millipore, Bedford, MA). The concentrated probe mix was adjusted with TE to a volume of 24.8 µl, denatured at 100oC for 2 min, spun for a few min and made to a final hybridization volume of 31 µl with 5.27 µl 20XSSC and 0.93 µl 10% SDS and kept at room temperature until used. Before hybridization, microarray glass slides were immersed in 0.2% SDS for 2 min followed by two washes each 2 min in autoclaved Milli Q water, and subjected to blocking in 180 ml of Methyl Pyrrolidinone (Aldrich) containing 2.7 g Succinic Anhydride (WAKO, Japan) and 15.3 ml 1M Boric acid (pH 8.0) solution for 20 min. Slides were then washed with autoclaved Milli Q water, and the spotted cDNA targets (microarray) were denatured in boiling water for 2 min. The slides were dehydrated in 100% ethanol for 2 min and air dried by spinning at a low speed. The mixture of labeled probes, for example, the control (Cy3) and test (Cy5), were then applied to the microarray, placed a cover slip carefully and incubated at 65oC for 14 – 16 h in a custom slide chamber with humidity maintained by a small reservoir of 3 x SSC. In addition, 3 - 6 µl of 3 x SSC was spotted at each corner of the slide, as far away from the array as possible. After incubation for 14 – 16 h at 65oC, the slides were washed in 2 x SSC / 0.03% SDS for 2 min, followed by washing in 2 x SSC, 1 x SSC and 0.2 x SSC, each for 2 min with gentle agitation, and air dried by spinning at a low speed. All hybridizations were performed in duplicate except those of days 19 and 27, where no reverse labeling was done. In addition, day 245 of pregnancy had three samples hybridized with two of them being labeled using the same dye combination and the other is reverse labeled. In the reverse labeling procedure, for example, the control liver cDNA and test liver cDNA, which had been first labeled with fluorescent dyes Cy3 and Cy5, respectively, were then labeled with Cy5 and Cy3, respectively. The microarray hybridization images were immediately scanned by GenePix 4000B (Axon Instrument, Union City, CA). The images acquired by scan were analyzed by GenePix Pro 4.0 software. The absolute feature (microarray spot) and background intensity of Cy3 and Cy5 for each feature on the array was obtained for the hybridized images. The local background intensity of each spot was smoothed by the local weight regression smoother (lowess smoother) and subtracted from feature intensity data. The subtracted intensity data were subjected to non-parametric regression and local variance normalization. The non-parametric regression can reduce intensity-depended biases. Compared with the linear regression, the accuracy is improved, if the points in scatter plot of Cy3 vs. Cy5 were not distributed around the straight line. The local variance normalization controls the noise (background) in the raw data to a greater extent where the normalized data, in many cases, did not show significant fold-differences in comparison with background-subtracted raw intensity ratios, which frequently displayed higher fold-differences. Thus, the variance method that we employed to bovine liver array data has produced much reliable normalized ratios. To group liver genes with similar expression patterns throughout bovine pregnancy, two clustering algorithm were employed. The hierarchical clustering algorithm was performed using the software, Cluster 3.0, created and made freely available by Dr. Michiel de Hoon (http://bonsai.ims.u-tokyo.ac.jp/~mdehoon/software/cluster/software.htm) based on the original version written by Dr. Michael Eisen (http://rana.lbl.gov/), and the software TreeView (http://rana.lbl.gov/). For hierarchical clustering, genes with either 2 or more than 2-fold induction or 2 or more than 2-fold repression at least at one point of the sampling period were selected. All ratio values were log transformed (base 2) to treat inductions or repressions of identical magnitude as numerically equal but with opposite sign, and subjected to centroid linkage clustering. Self-Organizing Maps (SOMs) algorithm was performed using the software GeneCluster 2 made freely available by Whitehead Institute Center for Genome Research (http://www-genome.wi.mit.edu/cancer/software/genecluster2/gc2.html). The SOMs were performed on the present data to better visualization of clusters of genes with their expression patterns plotted on an x-y axis. For SOMs, genes with either 1.5-fold (and above) induction or 1.5-fold (and above) repression at least at one point of the sampling period were selected. The minimum fold difference i.e., 1.5-fold, chosen for the analyses in the present study was based on the calculation that the technical variation contributing to the expression level of each individual gene due either to different labeling efficiencies of cyanine dyes or labeling procedure or both, was found to be within the range of 1.41-fold induction or repression. Therefore, either the induction or the repression of a particular gene by 1.5-fold or above was considered significant. Keywords = differential liver gene expression Keywords = bovine pregnancy Keywords = hierarchical clustering Keywords = self-organizing maps Keywords: time-course Less...