by Jeffrey M. Perkel
The genome, so the saying goes, is the “blueprint” for life. And until recently, it was a relatively invariant one.
There were exceptions, of course, both natural and manmade. Systematic genomic modifications in B and T cells fuel the diversity of the immune system. Less controlled genomic tinkering in somatic cells produces cancer. In the lab, researchers have for a long time had the ability to insert random bits of DNA into the genomes of bacteria, plants and even mice, whether to destroy or insert a gene, to mark it with a genetic tag or to amp up or control its expression. That’s how knockout and knock-in mice are made.
Yet such tools have to some extent been cumbersome to use, painfully slow and certainly unsuitable for high-throughput, genomics-scale work. In any event, the technology of knockout mice doesn’t really translate to human (or other somatic mammalian) cells, as knockout technologies rely on the vigorous recombination pathways that are active in the embryonic stem cells those methods use.
One alternative, of course, is simply to write the desired genome from scratch, building it up piece by piece from synthetic oligonucleotides—a trick J. Craig Venter and colleagues pulled off in 2010.  But that process is both financially and technically beyond the reach of most researchers for now; even if it weren’t, scientists wouldn’t really know what to write. Yet researchers have devised other tools that enable them to tinker with the genome with a precision not previously possible.
If there’s one thing all genome-editing techniques have in common, it’s that they don’t try to reinvent the wheel, molecularly speaking. Rather than devising some new strategy to get cells to implement the desired change, these methods let the cells themselves do the heavy lifting. They leverage natural DNA repair mechanisms.
One such mechanism is homologous recombination, in which the cell uses one homologous piece of DNA as the template to “repair” another. The process, says Chris Torrance, chief scientific officer at Horizon Discovery, is a “seamless,” high-fidelity mechanism to repair naturally occurring strand breaks during DNA replication. It also provides a way to promote genetic diversity by enabling crossover events during meiosis.
Yet homologous recombination also can be exploited experimentally. The researcher simply needs to feed the process an exogenous template containing a desired sequence change. That’s how transgenic mice are created, and it’s the theory behind the “recombineering” technology commercialized by Gene Bridges. Recombineering, explains Stefanie Hager, a senior scientist at Gene Bridges, “is basically a more versatile and powerful alternative to copy-and-paste cloning” that’s used to create templates for generating knockout and knock-in mouse strains (among other applications).
The goal is to build a vector in which the desired genetic change is flanked by 5- to 6- kb “homology arms,” which drive the mammalian cell homologous-recombination machinery. The process occurs in two stages. In the first, the user creates (via homologous recombination in bacteria) a template DNA containing some desired modification—say, conditional deletion of exon 5 of a particular gene. The researcher does this by introducing into a recombination-competent strain of E. coli a cloned DNA (such as a bacterial artificial chromosome) harboring the genomic region of interest.
The process is iterative. In the first round, the genomic region is transferred into a high copy number plasmid in bacterial cells via a recombination process that relies on bacteriophage recombinases. The resulting template plasmid is then modified in later recombination rounds, for instance to flank exon 5 with loxP sites (which enables excision of the exon sequence in animals expressing the Cre recombinase) and introduce a selectable marker for selection of targeted embryonic stem cells.
According to Hager, this completed template plasmid, which is about 15 to 20 kb in size, is then shipped to the user who, in the process’ second stage, introduces it into mammalian cells for site-specific genome editing. This time the editing is driven by the mammalian homologous-recombination machinery.
“We use the recombineering technology to generate targeting vectors in E. coli in vivo,” Hager says. “But you have to rely on the cells’ own homologous-recombination mechanism to actually integrate that targeting construct [into the final mammalian cell target].” As a result, Gene Bridges’ vectors mostly are used to generate knockout and knock-in mice, as mouse embryonic stem cells are highly efficient at homologous recombination; human somatic cells (such as immortal cell lines) are not.
“Almost any human cell line in culture that is somatic in origin, like cancer cells, actually shut[s] off homologous recombination in favor of other DNA repair mechanisms that are more error-prone, such as non-homologous end-joining” says Torrance.
According to Hager, users can request or produce one of two formats using the Gene Bridges process. Targeting vectors are used to introduce specific genomic modifications at defined loci into mouse embryonic stem cells, (for instance, to create conditional knock-out or knock-in strains). The second format is a modified bacterial artificial chromosome, which is injected into fertilized mouse eggs to produce random insertion events—a standard transgenic mouse.
Users can create these vectors themselves, using a Gene Bridges kit, or opt to have the company build the targeting vectors for them. According to Hager, the company’s Heidelberg-based service lab “has run several hundred DNA engineering projects” since its founding seven years ago. Most projects, she says, can be completed in eight to 10 weeks for between 8,000 and 10,000 euro.
Horizon Discovery’s Genesis platform also takes advantage of homologous recombination. In addition, it leverages the serendipitous finding that a natural, nonpathogenic DNA virus can be exploited to do this with greater efficiency than ever before.
Specifically, Genesis removes all the viral genes present in the wild-type adeno-associated virus (rAAV) to deliver a stretch of DNA with homology to the target locus to be altered. The key is that rAAV is a single-stranded DNA (ssDNA ) virus; as it turns out, Torrance says, human somatic cells can carry out homologous recombination very effectively using ssDNA templates.
“We are finding that homologous recombination may not be a single mechanism or pathway,” Torrance explains. “There appears to be a pathway for double-stranded species and a separate pathway for single-stranded species, and the single-stranded pathway is not shut off in human somatic cells.”
According to Torrance, Genesis can handle vectors of up to about 4.7 kb, including 2 kb of flanking DNA and 2.5 kb or so for any desired changes. That makes the technology amenable to point mutations and insertion of small genetic tags such as green fluorescent proteins but not for the direct insertion of very large transgenes. The process also is highly precise, Torrance adds. “We can take out a whole gene or put in a point mutation in a kinase to destroy the active site.”
To date, Horizon Discovery has produced more than 350 Genesis-modified human cell lines, says Torrance. The company also will undertake custom work, but the cost is such that it may not be practical for academics. Consequently, the company has begun implementing a “Centers of Excellence” program to “teach academic labs to do this themselves,” he says, adding, “It costs nothing but consumables and time.”
Sigma-Aldrich and Cellectis promote a genome-editing strategy that can leverage homologous recombination as well as a second DNA-repair mechanism: non-homologous end-joining (NHEJ).
In NHEJ, a double-stranded DNA break is repaired by stitching the two ends of a DNA molecule together. Generally, the process repairs the break flawlessly; sometimes, though, the system makes mistakes, which would be deleterious unless the goal was to make a knock-out mutation.
“Every so often, NHEJ is an imperfect process,” says Keith Hansen, product manager for the Emerging Technology division at Sigma-Aldrich. “So at times you will have nucleotides deleted or inserted at the cut site. That causes a frameshift, and the cell no longer expresses functional protein.”
To introduce the double-stranded break that kicks off NHEJ, both Sigma-Aldrich and Cellectis have devised strategies to target nucleases to specific sequences of interest. Sigma-Aldrich’s nucleases are called CompoZr® ZFNs, or zinc-finger nucleases. Cellectis’ are called TAL effector nucleases (TALENs). In both cases, a DNA-binding protein is coupled to a DNA nuclease (for example, FokI), which cleaves the DNA.
Researchers can target a CompoZr ZFN to a specific genomic region by assembling three to five zinc-finger binding domains, each of which targets a specific three-nucleotide stretch. Five such domains, then, can target 15 base pairs. But, because FokI works as a dimer, ZFNs must as well, and the complete ZFN complex actually targets a 30 bp sequence. The majority of CompoZr ZFNs have at least a 30 bp target sequence to increase nuclease sensitivity, Hansen says.
TALENs work similarly. TALENs are comprised of a set of repeats, each 34 or so amino acids long. Each repeat targets a single nucleotide, based on a pair of amino acids in the middle of the repeat. By matching up these repeats with the targeted sequence, a researcher can, in theory, target any sequence in the genome.
According to Luc Selig, vice president for marketing and sales at Cellectis Bioresearch, this ability to target TALENs nucleotide by nucleotide enhances the tool’s flexibility and there is little context dependence when assembling the various repeat modules. “There is no restriction in what you can do at the molecular level,” he says. A researcher could, for instance, design a TALEN to knock out the start codon for any gene.
Both companies deliver to the user a plasmid or mRNA encoding the nuclease, which is introduced into the cell to cut the genomic DNA. In the absence of a “repair” template—that is, a new sequence to insert, located on a DNA donor plasmid containing the appropriate homology arms—the cell repairs this DNA damage with NHEJ, producing a knockout. Alternatively, researchers can supply an exogenous template, which is incorporated via homologous recombination.
Sigma-Aldrich offers its CompoZr® ZFN technology in three flavors. Users can order Custom ZFNs through the company’s service arm. Or they can order Knockout ZFNs for any gene in the human, mouse or rat genomes through the company’s Knockout ZFN program. Users who simply wish to insert a transgene into a genome can use a Targeted Integration Kit, which introduces the user’s DNA into a designated “safe-harbor locus” in the genome of humans or mice. In the human genome, that locus is on chromosome 19.
Researchers at the Wellcome Trust Sanger Center in the United Kingdom and colleagues recently used this approach to repair a mutant gene in induced pluripotent stem cells (iPSCs). Using ZFNs in combination with a mammalian transposon called ‘piggyBac,’ the team repaired both copies of the alpha-1-antitrypsin gene in iPSCs derived from an individual in whom that gene was mutated—without leaving any potentially damaging genetic “scars.” In so doing, the researchers restored normal gene function to those cells, which functioned after differentiation into hepatocytes both in vitro and when injected into mice. 
The results, the authors wrote, “provide the first proof of principle, to our knowledge, for the potential of combining human iPSCs with genetic correction to generate clinically relevant cells for autologous cell-based therapies.”
In August, Sigma-Aldrich cut the price of its CompoZr products in half, down to $12,000 for Custom ZFNs, $6,000 for Knockout ZFNs and $2,500 for Targeted Integration Kits. Cellectis offers custom TALENs for $5,000 and up, says Selig.
MAGE & CAGE
Each of the approaches described above can be used to generate defined genomic changes. Yet they do so singly, one at a time. Harvard University geneticist George Church has developed a pair of approaches that enable a broader genomic redesign, at least in bacteria.
In 2009, his team described multiplex automated genome engineering (MAGE), an approach in which bacterial cells are transformed with up to 10 degenerate oligonucleotides (each targeting a specific sequence of the bacterial genome) and iteratively tested and retransformed to promote rapid genomic variation. 
MAGE, says Church, allows researchers to develop combinatorial changes and rapidly test them for fitness. “We can make billions of changes in a day and see if they are viable,” he says.
Indeed, each cycle takes only two to two-and-a-half hours. In just 35 cycles, or three days, his team was able to optimize lycopene production by a specially engineered E. coli strain by up to five-fold.
More recently, Church’s team combined MAGE with a second technique, hierarchical conjugative assembly genome engineering (CAGE), to replace all 314 TAG stop codons in E. coli with TAA.  The team used MAGE to make 32 pools, each containing 10 TAG-to-TAA changes. They then genetically combined those pools via bacterial conjugation to produce a series of eight strains, each containing one-eighth of a TAG-to-TAA-corrected genome.
Current work to merge those eight into the final, completely edited strain is ongoing, Church says, and could be used both to foil viruses and to promote protein engineering. That’s because the complete lack of a single stop codon in the organism means it’s possible to either remove the TAG-processing translational components (termination factors and tRNAs) altogether or reprogram them to, for instance, insert non-natural amino acids whenever they encounter a TAG codon.
As impressive as that feat of “genomineering” is, it’s only a start, says Church. “MAGE and CAGE could be even faster in E. coli, and we are beginning to adapt MAGE to Bacillus, Lactococcus, yeast and human cells,” he says. And researchers in Church's lab have just begun to test tools to make more extreme genome-wide changes (such as remapping 61 of the genetic code’s 64 codons, rather than just the one his team initially did) without “breaking” the system altogether.
“Our methods treat the chromosome as both an editable and an evolvable template, permitting the exploration of vast genetic landscapes,” Church and colleagues wrote in their CAGE publication. And those methods are available and affordable for other research groups now. Says Church, “It looks to me that you should be able to do a fairly radical redesign of a 5 million-base genome in a few months for thousands of dollars, not decades and millions of dollars.”
 Gibson, DG, et al., "Creation of a Bacterial Cell Controlled by a Chemically Synthesized Genome," Science, 329:52-56, 2010.
 Yusa, K, et al., “Targeted gene correction of á1-antitrypsin deficiency in induced pluripotent stem cells,” Nature, published online Oct. 12, 2011, doi:10.1038/nature10424.
 Wang, HH, et al., “Programming cells by multiplex genome engineering and accelerated evolution,” Nature, 460:894-8, 2009.
 Isaacs, FJ, et al., “Precise manipulation of chromosomes in vivo enables genome-wide codon replacement,” Science, 333:348-53, 2011.
The image at the top of the page is from Sigma-Aldrich's video "CompoZr ZFN Technology for Targeted Genome Editing".