Researchers at the Stanford Genome Technology Center have developed a new method for amplifying genomic regions in parallel that could be used to prepare DNA for high-throughput sequencing.
In a study published in PNAS last week, the scientists used their approach, which is based on long padlock probes, to successfully amplify more than 90 percent of almost 600 target sequences from 30 human kinase genes.
The uniformity of the products is better than that achieved by other multiplex amplification methods, they say, and with further improvements and additional funding, the method could be scaled up to the entire human exome.
The new method is similar to a multiplex amplification strategy published by George Church’s group at Harvard University last year, whichuses modified molecular inversion probes (see In Sequence 10/16/2007).
A key improvement by the Stanford researchers is the use of long padlock probes that are more than 300 bases instead of 70 bases in length, allowing them to capture and amplify up to 500 bases of DNA, instead of up to 200 bases, with greater uniformity and success than the previously published approach.
The long probes are “critical for the success of the amplification,” said Michael Mindrinos, associate director of the Stanford Genome Technology Center and one of the inventors of the new method, dubbed “Spacer Multiplex Amplification ReacTion,” or SMART technology.
For their new method, Mindrinos and his colleagues first devised a protocol for constructing the long, single-stranded probes. “There is no conventional methodology to make those long probes,” he said.
The protocol uses double-stranded bacteriophage lambda DNA as a template that is common to all probes. Target-specific ends unique to each probe are added to this by PCR amplification. A single-stranded probe is then generated from each double-stranded product, which can be amplified indefinitely. “You make it once, and you have it forever,” Mindrinos said.
In a proof-of-principle study, the Stanford scientists constructed long padlock probes to target 575 DNA sequences, representing exons from 30 human kinase genes. More than 90 percent of the targets were successfully amplified.
The researchers then pooled the 10 percent of the targets that failed, probably because of their high GC content, and were able to amplify about half of them.
For 70 percent of the targets the product abundance was within a 10-fold range, and for all targets it was within a 100-fold range. That kind of uniformity, Mindrinos said, was “much better than [that of] previously published studies.”
By comparison, the Church group, in their published study, targeted approximately 50,000 exons and amplified about 20 percent of them with poorer uniformity.
“Even though we are much better than anybody else, we still think we need to do further improvements if we want to use sequencing technologies from ABI or Illumina effectively and efficiently.”
However, Jay Shendure, an author of that study and an assistant professor at the University of Washington, told In Sequence that “some very simple optimizations” have allowed him and his colleagues to increase the success rate of their method from 20 percent to more than 90 percent, meaning more than 50,000 of 55,000 targets were amplified.
“So I don’t think that the long spacers are necessary to achieve that,” he said.
However, he agreed that the long padlock probes seem to improve uniformity and the ability to target longer stretches of sequence.
Compared to array-based DNA enrichment methods, like those developed by Roche NimbleGen and Febit, Mindrinos said his method has several advantages.
The success rate, compared to published studies using array-based capture methods, is better, he said, and those methods are unable to easily address problems arising from secondary structure and GC-rich areas.
In addition, array-based approaches, he said, require more target DNA to start with. “It is well known that in solid-phase hybridization, like any array technology, you have to use 100 times more material to get a signal,” he said. His team’s technology requires 250 nanograms of DNA per reaction, he added.
Carsten Russ, a research scientist in the genome sequencing and analysis program at the Broad Institute, who has been developing a hybridization-based capture method (see In Sequence 4/8/2008), told In Sequence that padlock probe-based methods “in general have the advantage that they do not deliver ‘overhang,’ or ‘near-target,’ bases outside the precise target for sequencing.”
However, padlock-based methods are “very sensitive to SNPs falling under the targeting sequences, which can cause the probes to fail systematically in samples with SNPs in those regions,” he said.
Hybridization-based methods, on the other hand, “tend not to lose a target because of a SNP in the probe region,” he said, and data published to date indicate “they have more even yields of target than padlocks.”
Mo’ Better Probes
Mindrinos said that he and his colleagues are currently working on scaling the method up, initially to between 5,000 and 6,000 exons targeted, and expect to have results within the next two months.
At the same time, they are working to further improve the uniformity of the products, an important step for coupling the method with second-generation sequencing.
“Even though we are much better than anybody else, we still think we need to do further improvements if we want to use sequencing technologies from ABI or Illumina effectively and efficiently,” he said.
The goal is to achieve a tenfold or smaller range of abundance for all products, and initial results are promising, he added.
“This is our major research focus right now, because if we can achieve that, we have positioned ourselves in an extremely good place to scale up [and] to do the whole genome very soon.”
So far, the researchers have performed pilot sequencing studies using 454’s technology, which the Stanford Genome Technology Center has in house, and plan to sequence more amplification products by 454 sequencing once they have improved their uniformity.
The main obstacle to scaling up the technology is a lack of funding, according to Mindrinos. He estimated that the cost of generating probes to target all exons of the human genome today is approximately $3 million to $5 million, a one-time investment.
The cost is significant because the oligonucleotides needed for the probes must be created in individual reactions rather than as a pool. “What we need is a cheap way to make individual oligos,” he said. However, the cost of oligo synthesis is expected to decline significantly over the next few years, he added.
Mindrinos said his team is also starting to look for academic collaborations and is exploring possible partnerships with companies that could help commercialize the technology, either as a tool or by using it to generate content that could be sold.
“We have not made a final decision, but we are looking very carefully for partnerships, either in the academic world or in the business world,” he said.
Stanford University owns the method, which was invented by Mindrinos; Sujatha Krishnakumar, a research associate; and Ron Davis, director of the Stanford Genome Technology Center.