By Julia Karow
This article was originally published March 19.
Startup Halcyon Molecular has developed a method for arraying long, stretched-out single-stranded DNA molecules on a substrate and is working on reading the sequence of labeled DNA bases by transmission electron microscopy.
Two weeks ago at the National Human Genome Research Institute's Advanced Sequencing Technology Development meeting in Chapel Hill, NC, J. Provine, the company's director of nanofabrication, talked about the progress the company has made over the last several years.
The firm was founded by Michael and William Andregg in 2003 and moved to the San Francisco Bay area in 2008. Last summer, it received an undisclosed amount of venture capital funding from the Founders Fund, and last fall, through a collaboration with George Church at Harvard Medical School, it was awarded $1.29 million in funding from the NHGRI (see In Sequence 10/13/2009).
To sequence single-stranded DNA, Halcyon wants to label the molecule with base-specific heavy metal atoms, stretch it out on a TEM substrate, image it using a scanning transmission electron microscope, and determine its base sequence with single-base resolution.
According to the NIH grant abstract, the technology promises long reads of at least 150 kilobases and up to 4 megabases. Unlike most existing sequencing technologies, repeat-rich DNA regions of the genome are not expected to be any more difficult to sequence than other areas. Based on models, the researchers estimate that TEM sequencing may be able to sequence a human genome to more than 99.9999-percent consensus accuracy and completeness in less than 10 minutes, at a cost of less than $100.
Provine said that the company is using "well-known chemistries" that have been developed by others over the last few decades to label the DNA bases specifically with heavy metal atoms. For example, the researchers use osmium to label thymine, and platinum to label guanine. Experimental data he presented showed that the specificity of the label is not always perfect — for example, a G-specific label sometimes also attaches to A — but he said there are methods available to prevent off-target labeling.
In order to stretch out the labeled DNA strands and deposit them on a TEM substrate, the company has developed a threading technology called "individual molecule placement rapid nanotransfer," or IMPRNT, which constitutes its "core technology," according to Provine.
IMPRNT uses a needle that grabs a single DNA strand from a solution and stretches it out in free space before depositing it onto a substrate. At the moment, company researchers are able to place DNA strands 10 nanometers apart and grab single DNA strands — rather than several at once — up to 10 percent of the time.
The aim under the NHGRI grant is to narrow the distance between the strands on the substrate to 3 nanometers, and to increase the yield of single molecules to about 50 percent, Provine said, which can be achieved by better position controllers, optics, software, and needles. The firm has already built microfabricated needle arrays that allow it to pull out up to a million DNA strands per hour.
Halcyon has also been working on better TEM substrates to hold the DNA during imaging. These substrates need to be as thin and as flat as possible, provide a large enough imaging area, be robust to temperature changes, and unreactive with DNA, Provine said. Company researchers have so far used carbon films, which they have transferred onto nanofabricated support frames, but are currently developing batch-fabricated films to replace them.
Finally, the researchers have started to image DNA strands labeled with platinum, using a STEM at Oak Ridge National Laboratory. In an image of a DNA molecule consisting of GATC repeats, most guanines appeared to be labeled, each by a single platinum atom.
The goal is to tag several types of bases with different labels simultaneously, but Provine said recording differences between osmium and platinum, for example, "is challenging." To that end, the researchers are exploring labeling the bases with clusters of atoms rather than single atoms, he said. They are also looking for new methods to speed up the scanning process. It currently takes several hours to get up to 50 images, which cover more than a micrometer of DNA, equivalent to more than 2,000 bases.