Global view of gene expression analysis

Since the inception of microarray-based expression profiling, microarrays have used a 3' (three prime) biased labeling system. To date, this approach has made significant contributions towards the better understanding of the mechanisms governing many diseases.

However, it is now understood that more than 60 per cent of genes are alternatively spliced, the process in which the exons in pre-mRNAs are spliced out, changing the transcript's sequence and contributing to the hundreds of thousands of transcript isoforms that help to define the biology of the system being studied.

Of specific importance to human disease is the fact that up to 50 per cent of disease causing mutations may result in transcript splicing alternations and 20 per cent of cancer-causing mutations can result in exon-skipping events.

Unfortunately, the traditional 3' approach to expression profiling is limited by assumptions that the 3' end of each gene is clearly defined and that each transcript has an intact Poly-A tail. As such, 3' expression profiling cannot detect alternative spliced transcripts with the same 3' end, transcripts from the same gene locus with a different 3' tail, non-polyadenylated transcripts, genomic deletions and other genomic alterations (see figure 1).

Microarray pioneer Affymetrix has now developed a new Whole Transcript Assay (WT Assay) and exon-based microarrays that are designed to reveal a level of transcriptional complexity not previously detectable with conventional 3' microarray expression approaches.

The main purpose of the WT Assay is to provide a more complete and accurate picture of overall gene expression. The assay uses a random priming method for generating labeled sense targets throughout the entire length of RNA transcripts, thereby negating the limitations of 3' expression profiling.

"There are two types of arrays that are compatible with the new assay - the GeneChip Exon 1.0 ST and Gene 1.0 ST arrays, which are available for human, mouse and rat," says Dr Robert Henke, chief scientific officer of Millennium Science, the Australian distributor for Affymetrix.

"The Exon 1.0 ST Arrays provide three levels of expression information from a single array - differential gene-level expression, differential exon-level expression, and discovery of alternative splice events."

Henke says some of the alternative splicing events detectable by exon arrays include exon skipping, differential isoform expression, alternative 5' transcriptional start sites, truncated transcripts, genomic deletions and translocations, and detection of transcripts with undefined or non-poly adenylated 3'ends.

"Exon arrays target over one million exons with about four probes per exon, providing an average of 40 probes spread across the entire length of each targeted gene," Henke says. "Through the use of this exon-centric probe content, researchers have the ability to analyse both alternative splicing and differential expression of individual exons within each gene.

"For gene-level expression analysis, the full set of probes spanning the entire transcript are used to generate a single data value that represents the expression level of all transcripts generated by the gene."

Splice variations and glioblastomas

Techniques such as whole transcript assays and exon arrays are unveiling alterations in exon usage that may play a critical role in diseases such as cancer.

For example, researchers at Millennium Pharmaceuticals in the US have used Affymetrix's Exon 1.0 ST Array to discover specific exon cassettes within the CD44 gene that are expressed in primary colon cancer cell lines, but are absent from metastatic colon cancer and Hela cell lines. These results suggest that CD44 splice variants might serve as diagnostic or prognostic markers for colon cancer.

Exon arrays are also being used to distinguish glioblastomas from oligodendrogliomas in order to help clinicians better diagnose brain cancer. Dr Pim French and Dr Justine Peeters from the Erasmus Medical Centre in the Netherlands have used exon arrays to compare different tumour samples and identify novel exon-skipping events and associated genes.

They have performed an analysis of differentially regulated splice variants and novel exons in glial brain tumours.

"Splice variants have a key role in biology," Peeters says. "Both tissue and developmental stage-specific alternative splicing contributes to significant protein diversity."

This team's primary aim it its research project is to explore how disease-related deregulation of splicing might be critical in pathogenesis and contribute to disease diversity and complexity.

"Compared to other primary tumour types, there is not so much known about the underlying molecular causes that contribute to the onset, progression and treatment of gliomas," Peeters says. "By identifying splice variants that are differentially regulated between histological subgroups we can uncover more of the biology involved in these aggressive tumours."

---PB---

Greater coverage

The Erasmus Medical Centre was a major test site for Affymetrix Exon 1.0 ST Array and has previously published a study in Cancer Research on glial tumours using the Affymetrix U133 Plus 2.0 expression arrays (a 3' focused microarray).

According to Peeters, the Exon 1.0 ST array provides greater genome coverage, as well as the possibility of detecting regulatory mechanisms such as exon skipping, intron retention and alternative promoter usage. The Exon array also demonstrated the ability to identify and characterize glial tumour subgroups based on different analyses methods.

"We were able to identify and molecularly separate these subgroups based on both the expression of the exons, as well as the associated transcript expression," she says.

"We were able to detect differentially regulated splice variants, novel exons and possible translocated transcripts and we have also been able to detect exon skipping mutations."

For Pim French, the key goal wasn't to find more markers for specific subtypes of glial brain tumours but to utlise the potential of such arrays to allow researchers to find causal genetic changes, like the pathological splice variant of EGFR.

According to French, a large proportion of glial brain tumours have a genetic deletion within the epidermal growth factor receptor (EGFR) locus.

"This deletion results in the expression of a pathological splice variant that is constitutively active," he says. "This splice variant has been demonstrated to play a role in tumour formation and is associated with response to EGFR inhibitors."

Using the Exon array, "I was most excited to identify pathological splice variants like the one in EGFR. In fact, we found a few in that gene we were not aware of.

"Such pathological splice variants will not be detected with other expression profiling platforms."

"The information contained on the exon array is greater than the older 3' arrays and such genome coverage give more possibility of answering more diverse biological questions," Peeters says. "Having run disease-related samples on the arrays gives endless possibilities of re-mining the data."