Name: John Edwards
Title: Associate research scientist, since 2006
Columbia University Genome Center
Experience and Education:
- Postdoctoral researcher, Chemical Engineering Department and Columbia Genome Center, 2003-2006
- PhD, Chemical Engineering, Columbia University, 2003
- BS, Chemical Engineering, Johns Hopkins University, 1998
Columbia University’s John Edwards has been collaborating with Applied Biosystems to use the company’s new SOLiD platform to profile DNA methylation on a genome-wide basis.
Next week, he is scheduled to talk about the project in a workshop organized by ABI during the American Society for Human Genetics annual meeting in San Diego.
In Sequence spoke to Edwards last week about his experience with the SOLiD technology.
How did your collaboration with Applied Biosystems come about, and how long has it been going on for?
It started almost a year ago now when they started mentioning that they had their SOLiD technology. They were looking to bring on collaborators who had a project that could benefit from SOLiD and who knew how to handle the data.
We had previously developed a methylation fractionation technique, where we can isolate methylated and unmethylated compartments of the genome. In our previous publication [Genome Research, 16: 157-163, 2006], we used Sanger sequencing and sequenced the ends to characterize the compartments. We always knew that if we had a way to sequence much more for cheaper, we could turn this method into a high-throughput screen that would profile the methylation state of the whole genome.
Have you had experience with high-throughput sequence data?
Yes, high-throughput before these new machines came out. I was involved with an EST project with Jingyue Ju here at the Genome Center, where we sequenced, in collaboration with colleagues here at Columbia and in Florida, nearly 200,000 ESTs from Aplysia californica by Sanger sequencing. A few years ago, that was quite a lot. Now, that’s changing very quickly. Because of this EST project, we had already developed all the techniques to handle the data and could easily scale things up.
Can you tell me about your methylation fractionation approach?
We use a battery of methylation-sensitive, and methylation-dependent, enzymes. The methylation-dependent enzyme will cut essentially at all methylated CpG sites in the genome. So we digest away the methylated compartment, leaving only the unmethylated compartment of the genome. Then we use methylation-sensitive enzymes that digest only at unmethylated CpGs, and through exhaustive digest leave only the methylated compartment.
Once we have these two compartments separated out, we can do end-sequencing, sequence short tags off the ends of each clone, and map them back to human genome sequence. Then you know where the whole clone came from, allowing you to assign methylation state.
When ABI said, ‘We can do paired ends that are going to be 25 bases in length on the SOLiD system,’ I did some analysis and showed that even with 20- to 25-base-pair tags you can still do this mapping back to the genome.
What did you do in your project with ABI?
In collaboration with Tim Bestor and Victoria Haghighi, here at Columbia, we are looking at methylation changes in breast cancer. There are two well-known features about methylation in breast cancer: One is hypermethylation of tumor suppressor genes. This is thought to turn them off and to play a role in carcinogenesis. The second feature is global hypomethylation of the genome, which tends to occur at repetitive elements.
You get groups that look at either one or the other, but it has been difficult, because of a lack of adequate technology, to do whole-genome methylation profiling to look at both these features at once.
By using our technique, we can assess the methylation status of the whole genome in one assay, including the methylation status of repetitive elements and of the promoter regions of all genes.
Our initial study, combining the fractionation technology we developed previously with SOLiD sequencing, uses tumor samples, normal samples, and a cell line to fully characterize the methylation changes in breast cancer.
Using SOLiD, we were able to profile the methylation status of the whole genome for the same cost it took to profile 1 percent of the genome using Sanger sequencing in our previous work.
Have you completed the project yet?
The project is still ongoing. The first sample we ran looked at a breast cancer cell line, analyzing both methylated and unmethylated compartments in one run. We approached whole-genome coverage in our first run, and that was on an early SOLiD instrument a few months ago. We hope that within the next couple of runs we can optimize the method to get whole-genome coverage of the methylation status for a couple of different samples per run.
As soon as we get all loose ends tied up, hopefully in a few months, we plan to publish our results and then we will make all the data publicly available in an easy-to-use format.
What has been your experience with the data?
The SOLiD data has been a lot better than what I had thought it would be, actually. I was pretty skeptical about these things when they first came out. For tag mapping, we don’t have the [same] needs for low error rates that you might have if you are looking for SNPs or some other studies, so I cannot speak for the quality of data for these studies. But for this type of mapping strategy, it has worked wonderfully well. We got a lot more data per run than we thought they would provide when we first started talking about a collaboration.
Did you also consider Illumina’s system for this project?
Not for this project. We have used Illumina for another project, a microRNA project, which is something I know a lot of people are doing on their system. One main characteristic of our approach is that we need to be able to sequence from both ends. It turns out that repetitive elements play a big part in the methylation landscape of the genome. So by using a paired-end technique, even if one of the tags falls, let’s say, into an Alu or L1 repeat, most of the time, the other tag will fall into a unique part of the genome, so you are still able to map that fragment uniquely.
We are developing this in collaboration with ABI using SOLiD at this point because their system meets our requirements, but on the other hand, I think the fractionation technique could be used on any system as long as you can sequence tags from both ends. Since Illumina doesn’t have paired ends yet, and since 454 doesn’t even come close to the throughput, SOLiD is the way to go for it now. As SOLiD instruments are becoming more available, we are planning to examine methylation profiles in much more than just breast cancer.
For us, it’s just an exciting time. Whole-genome methylation profiling is something that has really been lacking, there haven’t been ways to do this. I think our approach is one solution, but I think these machines are really going to open up what you can do as far as epigenetic profiling goes, and really start to change and revolutionize the field.