Main

Researchers at the J. Craig Venter Institute (JCVI) in Rockville, Maryland, USA, recently described the construction of a bacterium controlled by a synthetic genome1. Whether they have made the first synthetic organism or merely the first organism to be completely under the control of a synthetic bacterial genome is debated. The more sober commentators have pointed out that the advance falls short of creating an entirely synthetic organism and that only the first and arguably most trivial step has been achieved: that of generating a synthetic genome and using it to 'reboot' a pre-existing cell. This can be thought of as the equivalent to ditching Windows and installing Linux in its place. Gibson et al.1 would perhaps reject this metaphor, pointing out that “the DNA software builds its own hardware”. They argue that after several generations of existence cells of Mycoplasma mycoides JCVI-syn1.0 were rebuilt based on a genome that started off inside a computer.

Irrespective of any disagreement about how to define the resulting organism, this work is an important milestone in the journey towards the creation of uncontested artificial life. It was not a simple task and was necessarily preceded by other breakthroughs. JCVI researchers were responsible for two of the key steps that lead directly to their recent paper. The first of these was the synthesis and assembly of the Mycoplasma genitalium genome (M. genitalium JCVI-1.0) based on sequence data alone2. The de novo generation of an M. genitalium genome was an order-of-magnitude increase on previous DNA synthesis experiments, which were limited to complete viral genomes and partial bacterial genomes. The 583 kb genome was initially synthesized as 101 cassettes of 5–7 kb overlapping by ∼80 bp. These cassettes were assembled into four pieces of ∼144 kb each in three stages using in vitro recombination. To produce a single contiguous DNA molecule representing the entire genome, the researchers used in vivo recombination in yeast.

The second breakthrough was to successfully replace one bacterial genome with another. Lartigue et al.3 cloned the genome of M. mycoides into a yeast centromeric plasmid that was then transplanted into a Mycoplasma caricolum cell. A key stumbling block was the restriction system, which protects the bacterial cell by digesting unmethylated foreign DNA at particular sites. As the donor M. mycoides genome was not methylated when it was in yeast, it became a target for the M. capricolum restriction system. To circumvent this defence system, the donor genome could be treated with purified methyltransferases.

The techniques developed in these studies were then combined in the most recent work to generate a synthetic M. mycoides genome and transfer it to an M. capricolum recipient cell1. The genome of M. mycoides is substantially larger than that of M. genitalium (0.6 Mb compared with 1.1 Mb), so the researchers had to synthesize 1078 cassettes of 1080 bp each, again with 80 bp overlaps. Groups of these cassettes were recombined in yeast, producing 109 assemblies of 10 kb each. This process was repeated with the 10 kb assemblies to produce 11 assemblies of 100 kb and then, eventually, a single molecule. The assembled synthetic genome was then transplanted from yeast into the recipient M. capricolum cell. Despite extensive checking and correction of errors, several unintended mutations were introduced during the procedure, including several single-nucleotide polymorphisms, as well as a transposon insertion from E. coli and an 85 bp duplication caused by non-homologous end joining that, together, disrupted two non-essential genes. To distinguish their genome from that of natural isolates, the researchers also incorporated several short nucleotide 'watermark' sequences placed so as to cause minimal disruption to functional parts of the genome. The resulting cells contained only the synthetic genome and were able to replicate. On the basis of growth assays and proteomic analysis, the cells seemed to be essentially indistinguishable from M. mycoides.

This work shows that current genome-sequencing techniques produce data of sufficient quality to reconstitute a functional bacterial genome from an in silico copy. Furthermore, it provides proof of principle for producing genome sequences with any number of alterations without the use of extensive, laborious genetic manipulation. Applications may be far off, however, as it has proved difficult to design novel functional modules with more than a couple of genes. Nonetheless, the JCVI researchers already have plans. Perhaps the most ambitious and potentially beneficial of these is to produce a photosynthetic bacterium that can use water to generate hydrogen gas for use as a multipurpose fuel.