Microarray synthesis of DNA oligonucleotides costs a fraction of a penny per base. But such methods can’t make pieces longer than about 200 bases. And stitching those oligonucleotide fragments into gene-length sequences typically must be done one strand at a time, losing the multiplex advantage of the array synthesis.
A team led by Calin Plesa, Angus M. Sidore, and Sri Kosuri of UCLA now reports a method called DropSynth that synthesizes multiple genes in parallel from a pool of microarray-generated oligonucleotides (Science 2018, DOI: 10.1126/science.aao5167).
The researchers design the oligonucleotides so that each piece needed for a particular gene has the same “barcode,” or identifying stretch of gene sequence. They then use microbeads tagged with a sequence complementary to that barcode to pull those oligonucleotides out of an oligonucleotide pool.
“Each microbead probably has millions of copies of the barcode,” Kosuri says. That means that each microbead will have many copies of each fragment in the gene.
The UCLA team processes the microbeads in a water-in-oil emulsion, with one bead in each emulsion droplet. “The way we generate the emulsion is actually pretty simple,” Kosuri says. “You just take an oil and your oligo mixture and vortex it for 30 seconds. You don’t need any specialized equipment.”
Once in the emulsion, enzymes clip the oligonucleotides from the microbeads. The pieces are assembled into the gene by a method called polymerase cycling assembly, and the resulting genes are recovered from the emulsion. The genes are placed on plasmids containing an assembly barcode. The researchers used next-gen sequencing methods to sequence the assembled genes, using the assembly barcodes to flag genes that were stitched together properly.
The main benefit of the multiplexing approach, Kosuri says, is that a gene pool is now accessible for approximately the same price as an oligonucleotide pool. “We’re not adding much to the cost by doing an emulsion assembly,” he says.
The method’s drawback is that the resulting gene library is “messy,” Kosuri says. The efficiency of making any particular gene product is only about 5%, which is limited by the coupling efficiency of making the oligonucleotides and by the fidelity of the enzymes used in the assembly process.
“I don’t think this is going to be the new way that a gene synthesis company might do synthesis, because most users want an exact sequence,” Kosuri says. But he does expect DropSynth will allow individual labs to make large libraries of genes. “Say you want to design 1,000 proteins by gene design; you couldn’t really synthesize that without lots of money.”
“As a field, cost-effective methods for synthesizing DNA—particularly gene-sized DNA fragments—have really lagged behind methods for sequencing DNA. This has been a bottleneck for many areas, but in particular for the field of protein design,” says Jay Shendure of the University of Washington. “If this method proves robust, it represents a major advance and will facilitate protein design and many other goals.”