Genome-wide analysis of antisense transcription with Affymetrix exon array.(Methodology article)

Authors: Xijin Ge (corresponding author) [1]; Wendy S Rubinstein [2,5]; Yong-chul Jung [3]; Qingfa Wu [4]

Background

Genome-wide analyses of mRNA and cDNA sequences have revealed large-scale antisense transcription in various animal and plant species. Based on analysis of mouse full length cDNA generated by FANTOM2 consortium, Okazaki

et al . identified 2,481 pairs of overlapping sense/antisense transcripts [1]. An updated analysis using FANTOM3 cDNAs and mouse cDNA at GenBank showed that there are 4,520 transcription units forming sense and antisense pairs on exons [2]. For human, Yelin et al . [3] and Chen et al . [4] reported 2,667 and 2,940 pairs, respectively. The prevalence of natural antisense transcripts (NATs) is also supported by data derived by sequence tag-based technologies such as serial analysis of gene expression (SAGE) [5, 6, 7] and massively parallel signature sequencing (MPSS) [8].

Microarray-based studies also confirmed the prevalence of antisense transcripts. Using strand-specific oligonucleotide probes Yelin

et al . studied the expression of both strands of 264 genomic loci in 19 human cell lines and detected antisense transcription in 112 (42.5%) of these loci [3]. Kiyosawa et al . studied the expression of 1,947 mouse NAT pairs in five types of cells and tissues using custom-made oligonucleotide arrays that distinguish the expression of sense versus antisense transcripts [9]. A research group at Affymetrix [10, 11] used a novel direct RNA end-labeling method to detect the orientation of transcripts with a tiling array integrating chromosome 21 and 22 at a resolution of every probe per 35 base pair (bp). The analysis was later extended to 10 chromosomes at 5 bp resolution [12]. The data strongly support the observation that transcripts encoded on both DNA strands often results in complementary mRNAs. By focusing on 1% of human genome, the ENCODE project generated tiling array data suggesting that the majority of the bases of human genome is transcribed [13].

To investigate potential functions of the large number of NATs clearly requires cost-efficient technology for genome-wide expression analysis in a strand-specific manner. Because of the lack of technology for routine analysis, an interesting study took advantage of incorrectly orientated probes in commercial arrays [14]. About 25% of the probes on the first version of Affymetrix mouse U74A and U74B arrays were designed and manufactured in the wrong orientation [15]. Werner

et al . used these faulty arrays to detect the expression of antisense transcripts in mouse brain and kidney [14]. Their result show that the antisense transcription is tissue specific and that the sensitivity of commercial arrays is sufficient to assess NATs in total RNAs.

In this proof-of-concept study, we demonstrate the applicability of Affymetrix Exon array to detect antisense transcripts at the whole-genome level. By modifying the standard cDNA synthesis and labeling process, we could labels single-strand cDNA in the reverse direction as compared to the standard protocol. Thus we can use Affymetrix Exon array to detect transcripts from antisense strands at over 1 million exonic loci across the human genome. Unlike previous expression arrays that target the 3' end of annotated genes, the Affymetrix Human Genome Exon array includes probes for all known and predicted exons. Most exons are represented by a probe-set consisting of 3 to 4 probes of 25 bp. For convenience, a 25 bp probe will be referred to as a "feature" and a probe-set will be called "probe" in the rest of the paper. We present a protocol that allows the independent labeling of sense and antisense strand of an RNA sample in combination with exon arrays will bring routine examinations of the antisense transcriptome within reach.

Results

The Affymetrix Exon array is an inexpensive high-density oligonucleotide microarray that has two unique features: (1) it includes probes for all known and predicted exons, and (2) its signals are strand-specific because of the generation and labeling of single-stranded DNA targets. Exon arrays are currently available for human, mouse, and rat. In the standard protocol [16], an initial cycle of reverse transcription (RT) converts RNA into cDNA using random primers linked with the T7 promoter. This is followed by second-strand cDNA synthesis. The double-stranded cDNA is then used as a template for

in vitro transcription (IVT) with T7 RNA Polymerase, which produces many copies of cRNAs that are reverse complementary to original RNA molecules. In the second cycle of cDNA synthesis, random primers are used to reverse transcribe the cRNA to obtain single-stranded DNA. The DNA is then fragmented and labeled in preparation for hybridization. After two RT cycles, the final single-strand DNA product is in the same orientation as original RNA. Based on annotation databases, the probes have been pre-manufactured to be reverse complementary to RNA sequences, so that the labeled product could hybridize with them.

We have tested a …