Skip to main content
Premium Trial:

Request an Annual Quote

Scientists from Febit Look to Commercialize New Gene Synthesis Method Using Next-Gen Sequencing


By Monica Heger

This story was originally published Jan. 5.

Next-gen sequencing could find a new application in synthetic biology — as a sample-prep step to make DNA for gene synthesis. Scientists from the German company Febit are looking to spin out a new company that would commercialize a new method of DNA synthesis based on the approach, which they say could reduce the price of synthetic DNA by several fold.

Currently, methods of preparing DNA for gene synthesis, which include techniques such as microarray-derived or column-derived nucleotides, cost around 30 cents per nucleotide and are prone to high error rates. Moreover, high-throughput methods of creating desired DNA sequences are lacking.

Along with Harvard scientists, the Febit team recently described a method in Nature Biotechnology that uses sequencing on the Roche 454 GS FLX as a way to reduce the error rate of creating DNA sequences by about a factor of 500 in a high-throughput manner, and at about a tenth of the cost of conventional methods.

Mark Matzas, head of synthetic biology at Febit, told In Sequence that he is looking to create a startup company to commercialize the method and offer it as a service to customers looking to design synthetic genes for a range of applications including the biofuels industry and pharmaceuticals. He said he is currently seeking investors and partners in order to set up a facility for the service, which he estimated the team could begin offering within the next one to two years at between 6 to 9 cents per base pair.

However, the exact timeline has not yet been worked out because some investors in Febit currently hold patents in the synthetic biology domain, and whether the startup could use that technology is a question that has not yet been resolved, Matzas said.

"We have the technology we've published, and we have the idea to commercialize that. But the exact way is not yet clear," he said. "It depends on what will happen with those patents."

Matzas said it is unlikely that Febit itself would want to commercialize the new technology, because the company has decided to focus on blood-based miRNA biomarker discovery, as it disclosed last summer. As part of that refocus, it cut its entire genomic tool and services business, resulting in a 60 percent reduction in headcount (IS 6/29/2010).

Additionally, the company had launched a synthetic biology subsidiary in 2007 called Synbio to commercialize a service that used its Geniom One array synthesizer to design, fabricate, and sell genes for use in various applications. Those R&D activities were also cut as part of the restructuring in June.

Tackling Cost and Accuracy

Currently, gene synthesis is plagued by two problems: high cost and high error rate. The cheapest way to create DNA is to start with microarray-derived oligonucleotides. But, because those oligos are prone to errors, they must be individually verified with Sanger sequencing, which is expensive and time consuming. Another method of preparing oligonucleotides, known as column-derived nucleotides, is more accurate, but also more costly. Typically, synthetic DNA costs around 30 cents per base pair, prohibitively expensive for creating synthetic genes, said Matzas.

The Febit and Harvard team were able to address both cost and accuracy issues by using only the microarray-derived nucleotides and replacing the Sanger sequencing step at the end with sequencing on the 454 GS FLX.

Microarray-derived oligonucleotides have an advantage over column-derived oligonucleotides in that researchers can purchase them cheaply and in bulk, Peer Stähler, an author of the paper and former chief scientific officer at Febit, told In Sequence. But, "they have a relatively high error rate. So there are good ones and bad ones, and they come in a big, complex mix."

[ pagebreak ]

The 454 sequencing acts as a "filtering step," said Stähler, sorting through the different sequences. The correct sequences will then be located at the 454 bead — with different sequences at each bead — and they can then be identified with a microscope and picked out of the sequencer with a micropipette. The 454 is particularly well-suited for the method because its beads are easily accessible and also large enough to be picked up using a micropipette, he said.

To test the method, the team started with a microarray-derived pool of oligonucleotides. They selected 319 beads, enriching the pool of oligos for those sequences and then sequencing them on the 454. To test the accuracy, they used Illumina sequencing to sequence the oligo pool both before and after 454 sequencing. In the "untreated" pool of oligos, only 3.1 percent of the reads matched to the set of 319 sequences. After 454 sequencing, however, 84.3 percent of the reads matched, indicating that the desired targets were correctly selected and enriched.

They then tested the ability to assemble gene fragments from the resulting oligos. They created two gene fragments, each 220 base pairs, by combining nine to ten of the sequenced amplicons in a PCR-based gene assembly reaction. The assemblies were then cloned and checked using Sanger sequencing. Seven out of the eight clones matched the target. The clone that did not match contained insertions and deletions all within a 23-base pair-wide region, which the authors attributed to being caused by "misassembly, rather than errors in the building blocks."

Next, they tested their ability to create a functional gene and were able to create a fully functional gene longer than 7,000 base pairs from DNA fragments obtained from 29 beads of the 454.

Finally, the authors wanted to calculate the error rate for the entire process. Estimating that errors would come either from wrong sequencing calls or from amplification errors, they predicted that they would see one error for every 21 kilobase pairs, an improvement over current methods that would translate into a 5- to 10-fold reduction in cost.

Chuck Merryman, an assistant professor at the J. Craig Venter Institute who does research on de novo construction of whole genomes, said the method could have far-reaching applications. "Essentially, every drug we have, every [biological] sensor, every biofuel — all of these are DNA-encoded elements. So, being able to change the DNA sequence means new drugs and new biofuels."

One limitation of the method, he said, is that it will only be cost effective for large-scale projects, because one run of the 454 costs around $15,000. "You need to be planning on making a lot of synthetic DNA," he said. "For a project looking at 15 to 20 proteins, it won't be cost effective. You need to be making thousands of proteins or looking at an entire genome."

That's one reason why it makes sense to offer the gene synthesis as a service, said Matzas, because one sequencing run could produce oligos for multiple customers. In the current paper, Matzas said they were able to produce DNA for about 3 cents per base pair. At that price, it would make sense to sell the synthetically designed DNA at between 6 to 9 cents per base pair, already well below the current price, he said.

Additionally, the method would ideally be fully automated so that the beads will be machine-picked, rather than individually hand-picked, a step that Matzas said the team is now working on.

He said that initially, the team would use the 454 platform, although he also thought the SOLiD machine would be amenable because, like the 454 platform, it has an open architecture, so the beads are accessible. The SOLiD beads are only about a micron in diameter, however, compared to the 454 beads, which are about 40 microns. So while switching to the SOLiD would help reduce the price further because it would produce even more reads, the platform would require full automation, since the beads are too small to be hand-picked.

Matzas said the method could have a range of applications, including producing synthetic genes for biofuels and pharmaceuticals, and that the first application would likely depend on the investors.

Further down the road, the method could be used to build entire genetic circuits to test how different gene combinations work together and for designing optimal proteins and antibodies for drug development, said Stähler. "It will allow you to screen a lot of variants quickly," he said. This will "really open up the field of in silico computer-designed proteins."

Have topics you'd like to see covered in In Sequence? Contact the editor at mheger [at] genomeweb [.] com.