An oligonucleotide microarray containing 50-mer oligonucleotides representing 9277 unique Atlantic halibut genes has been designed, printed and is currently being used for the study of gene expression in developing halibut. The oligonucleotides are based on all of the Atlantic halibut data available at the time of printing; these included ESTs and complete cDNAs derived from the Pleurogene sequencing project as well as sequences deposited in GenBank by other groups as of September 2006. Of the Pleurogene ESTs, 5040 are functionally annotated; the remainder are unknown (1016) or are similar to unannotated sequences in GenBank (1626). In addition to Atlantic halibut features, several control features have been incorporated, including an oligonucleotide representing a heterologous plant gene (92 spots) and empty spots containing buffer only (1344). The array contains 48 subgrids, each comprised of 32 columns and 26 rows. Every feature is printed at least four times as side-by-side quadruplicates, resulting in a microarray containing 39936 features. This microarray has been utilized to identify genes differentially expressed in larval Atlantic halibut during the developmental period from post-hatch to post-metamorphosis.
Keywords: Development stage comparison
Overall design: Two colour design. Oligonucleotide DNA probes were designed from approximately 13000 available Atlantic halibut ESTs. The majority of the ESTs were derived by sequencing from the 5’ end of clones from 13 different normalized libraries from eight different tissues and five different developmental stages (Douglas et al., 2007). All EST data have been deposited in GenBank (accession numbers EB029285-EB041700 & EB080851-EB080975), and preliminary annotations are available on the Pleurogene website (www.pleurogene.ca). Additional unique ESTs from other research groups that were present in GenBank were also used. Oligonucleotide probes wFere designed using ArrayDesigner software (Premierbiosoft, Palo Alto, CA, USA). Probe sequences were selected to have minimal secondary structure, a GC ratio between 40-60%, a melt temperature of 75 ± 5oC and to be 45-55-mer in length. Uniqueness of the selected probes was ensured by BLAST comparison against all of the ESTs and the zebrafish genome.
Halibut DNA probes were synthesized at the Atlantic Microarray Facility (Moncton, NB, Canada) by the phosphoramidite method at a 40 nmole scale on a Gene Machines PolyPlex 2 oligonucleotide synthesizer (Genomic Soutions, Ann Arbor, MI, USA). Isolated DNA was analyzed by Electrospray ionisation-Mass Spectrometry using an Agilent VL1100MSD mass spectrometer (Agilent Technologies, Inc., Mississauga, ON, Canada). DNA of low quality was re-synthesized. In addition to the halibut probes, 1344 control spots containing buffer only, and 92 control spots containing an oligonucleotide specific to the chlorophyll synthetase G4 from Arabidopsis thaliana (GenBank Acc. U19382) were included. Solutions (20 uM) in sodium phosphate Schott-Nexterion Spot buffer (Schott, Jena, Germany) were prepared in 384-well plates (Genetix X7020) and spotted in quadruplicate on epoxide microarray slides (Schott) to give 39936-feature microarrays. An OmniGrid 100 microarrayer (Genomic Solutions) equipped with SMT-S50 silicon print pins (Parallel Synthesis Technology, Santa Clara, CA, USA) was used.
Spotted microarrays were analyzed by SpotQC (IDT Inc., Coralville, IA, USA) for any defects in spotting that may have occurred. Microarrays were stored at room temperature in vacuum-sealed microarray pouches (Corning Inc., Corning, NY, USA).
Larval Sampling
Larvae were sampled at five different stages: hatching (1 dph; 10 mm), mouth opening (21 dph;15 mm), midway to metamorphosis (64 dph; 20 mm), premetamorphosis (91 dph; 25 mm), and post-metamorphosis (104 dph; 30-35 mm). Three biological replicates were sampled at each stage. Larval samples were pooled as follows: ~20 larvae each for post-hatch stage and 5 larvae each for all of the later developmental stages.
RNA Isolation
RNA was extracted from three biological replicates of pooled larvae at each of the five stages using the RNeasy Plus Mini kit (Qiagen, Mississauga, ON, Canada) according to the manufacturer’s protocol. RNA quality was assessed using a Bioanalyser 2100 (Agilent Technologies, Inc.) and RNA was only used if the RIN value was at least 8.0. A universal RNA reference sample was prepared that consisted of equal amounts of RNA from all 15 RNA preparations.
cDNA Preparation and Labeling
RNA (7.5 ug) from universal RNA reference and experimental samples was reverse transcribed and labeled using the SuperscriptTM Plus Direct cDNA labeling system (Invitrogen, Carlsbad, CA, USA) with Alexa 555 (experimental) and Alexa 647 (reference) fluorophores. An 8 ng spike of Arabidopsis chlorophyll synthetase G4 transcript was also included in the reverse transcription reaction as in internal control. Labeled targets were hydrolyzed and purified using the QIAquick PCR purification kit (Qiagen). Purified labeled targets were quantitated using a NanoDrop® ND-1000 spectrophotometer (NanoDrop® Technologies Inc., Wilmington, DE, USA) and used if cDNA yield was greater than 20 ng/uL and fluorophore incorporation was at least 1.5 pmol/uL. Labeled targets (20 ul each of control and experimental samples) were mixed with 20 uL DIG EasyHyb buffer (Roche Applied Science, Laval, PQ, Canada), denatured at 95 oC for 3 min, centrifuged briefly, and hybridized to the microarrays.
Hybridization to Microarray
Slides were blocked in 100 mL of Nexterion Block E (Schott) at 50 oC for 1 h. They were rinsed in 100 ml dH2O for 1 min at room temperature and dried by centrifuging at 200 x g for 5 min. Lifter Slip cover slips (Erie Scientific, Portsmouth, NH, USA) were washed in dH2O and 95% ethanol, wiped dry, and gently placed over the microarray. Hybridzsation buffer containing the labeled targets (60 uL) was carefully applied to one corner of the Lifter Slip and immediately placed in the hybridization chamber (Die Tech, San Jose, CA, USA) to prevent drying. A small volume (60 uL) of hybridization buffer was added to wells in the bottom of the hybridization chamber to maintain moisture content, the chamber was closed, and incubated overnight at 30 oC in a water bath. Microarrays were washed once for 10 min at room temperature in 2X SSC with 0.2% SDS, twice for 10 min each at room temperature in 0.1X SSC, and dried by centrifugation at 200 x g for 5 min.
Scanning and Data Analysis
Microarrays were scanned at 543 and 633 nm using a ScanArray® 5000XL Microarray Acquisition scanner (Packard Bioscience, Billerica, MA, USA) at a resolution of 10 um. Laser power was set at levels between 80 and 100% and photomultiplier tube settings were set at values ranging from 70 to 95% to adjust Alexa555 and Alexa647 channels on individual slides. Spot intensities were measured using ScanArray® Express (version 3.0.1.0001) microarray analysis software (Perkin Elmer LAS, Waltham, MA, USA), and the CSV files loaded into the ArrayPipe (Hokamp et al., 2004) server at the Institute for Marine Biosciences. Appropriate labels for common reference and treatment were set as follows: Channel1 for all slides was annotated as ‘C’ (common reference), and Channel2 was annotated as T1 through T5, with T1 corresponding to samples from post-hatch and T5 corresponding to samples from post-metamorphosis. Processing steps in Arraypipe consisted of ‘Flagging Markers’ (control features labeled as ‘buffer only’ and ‘Arabidopsis’) for exclusion from further analysis; background correction using the ‘limma normexp’ module with an offset of 50; subgrid (print-tip) normalization via the ‘limma loess’ (Smyth and Speed, 2003) module with span = 0.3, bin = 0.01, and iterations = 4. Replicate features (4 replicates for each feature) were merged by computing the median value of all unflagged replicates. In all operations, features flagged by ScanArray® Express were excluded.
Spot ratios between pairs of successive developmental stages were compared individually using significance analysis of microarrays (SAM) (Tusher et al., 2001). To enable comparison of ratios between successive developmental stages, the output files for each slide were downloaded from Arraypipe, and the normalized, merged log2 ratios for each stage were input to the SAM Excel module. Genes used in SAM were unflagged in all replicates for both developmental stages being compared. In each comparison, the first developmental stage was labeled ‘1’, and the second ‘2’. The SAM mode was ‘two class unpaired’, using a t-test with 100 permutations.
For principal components analysis (PCA), the background-corrected, normalized, merged log2 ratios from ArrayPipe were imported into Matlab®. The data set was truncated to contain only those ratios that were unflagged across all 15 slides, resulting in a data matrix of 588 rows (features) by 15 columns (slides). Each column of the data matrix was mean centered, and analyzed by PCA via singular value decomposition (SVD) (Anderson et al., 1999). In this work, only the scores of the first two principal components (PCs) were examined.
Less...