High-throughput sequencing challenges microarray turf

The precisely choreographed interplay of cellular gene activity is controlled by a vast cast of DNA-binding proteins - transcription factors and enzymes mostly. Recently, two independent research terms have been able to map the locations across the genome where a specific DNA-binding protein latches onto the DNA.

Using Illumina/Solexa's Genome Analyzer, the groups are using a new method called ChIP-sequencing (ChIPSeq) - a combination of chromatin immunoprecipitation and next-generation, or parallel, sequencing.

The feat was performed "with a speed and precision that goes beyond what has been achieved with previous technologies," University of Washington geneticist Stanley Fields wrote in an accompanying essay to one of the papers in Science.

ChIP is a well-established lab technique to identify those specific sites where proteins latch onto the DNA. Cells are treated with a chemical to fossilise the links between DNA and protein, the chromatin is then isolated, the DNA broken up, and the attached proteins immunoprecipitated.

Finally, the DNA stuck to the protein can be released and analysed. Until now, the most high-throughput application of this technique involved using microarrays containing thousands of gene spots able to identify binding sites for transcription factors and the like.

Writing in Science, David Johnson and colleagues at Stanford University used ChIPSeq to identify the binding sites for the transcription factor NRSF (neuron-restrictive silencer factor), which turns off neuronal genes in non-neuronal cells.

The DNA motif that NRSF recognises consists of a 21-base pair core fragment. As high-read numbers contribute to high sensitivity and comprehensiveness in large genomes, Johnson and his team performed a ChIP experiment in a T-cell line. They sequenced the released DNA fragments - some two to five million per sample - of which about half were successfully mapped back to the reference genome sequence.

The Stanford group discovered a total of 1946 NRSF-binding locations in the human genome, including DNA motifs controlling more than 100 other transcription factors and 22 micro-RNAs. The most common binding target was identified more than 6700 times in the experiment. Most of the sites were identified as expected, but so too were some previously unrecognised binding motifs that did not fit the previously known rules for NRSF binding.

The authors conclude that ChIPSeq is a cost-effective alternative to microarray methods, with a significant upside. "Other ultrahigh-throughput sequencing platforms, such as the one from 454 Life Sciences, could also be used to assay ChIP products, but whatever sequencing platform is used, our results indicate that read number capacity and input ChIP DNA size are key parameters," the team write.

Meanwhile, Gordon Robertson, Steven Jones and colleagues at the British Columbia Cancer Agency Genome Sciences Centre in Vancouver performed a similar analysis, again using the Illumina Genome Analyzer because of its high throughput. Reporting online in Nature Methods, this team looked at binding of a transcription factor called STAT1. The Vancouver group generated a total of more than 28 million fragments (in two types of cells), identifying more than 42,000 putative STAT1-binding regions.

The group suggests that ChIPSeq might be an order of magnitude cheaper than microarray alternatives, with the eight flow cell lanes in the Genome Analyzer offering excellent design flexibility. Fewer materials are required, and the method can be applied to any organism - it is not restricted to available gene arrays.

Fields writes that the advantages of ChIPSeq over ChIP-chip include the ability to interrogate the entire genome rather than just the genes represented on a microarray. (For example, Johnson points out that a similar experiment using conventional microarrays would require roughly 1 billion features per array.)

There is also the benefit of sidestepping known hybridisation complications with microarray platforms. "Perhaps most usefully," writes Fields, "ChIPSeq can immediately be applied to any of those [available] genomes, rather than only those for which microarrays are available."

Fields anticipates that similar experiments will quickly identify the binding locales of numerous other transcription factors, structural chromatin components, histone proteins and various enzymes. The addition of ChIPSeq to the next-generation sequencing repertoire, as well as the ability to quantify captured gene sequences in a single sample, illustrate the growing breadth of next-generation sequencing applications, he says.

Fields concludes his essay with a provocative thought: "The technology that is most threatened by the widespread adoption of ultrahigh-throughput sequencing? The DNA microarray."

Kevin Davies is editor-in-chief of Bio-IT World.