Synthia and the shotgun: what became of the synthetic cell

In December 2007 the most talked-about organism in synthetic biology did not yet exist. It already had a nickname, though. Critics of J. Craig Venter had started calling his planned minimal bacterium "Synthia", a label coined that summer by the watchdog ETC Group, and the name stuck to everything that followed.

What alarmed them was paperwork. Patent applications from the J. Craig Venter Institute, published over the course of 2007, sketched a stripped-down bacterial genome, a way of stitching DNA fragments together in a test tube, and a vision of running thousands or even millions of genome variants through robotic tests. Read together, they looked like an attempt to own the operating system of a new kind of biotechnology before anyone had booted it.

Seventeen years on, the forecast can be checked against what was actually built. Some of it arrived roughly on schedule. Some of it never arrived. And the patents ended up somewhere few people were watching in 2007.

What was actually on file

Two documents mattered most.

The first, titled simply "Minimal bacterial genome", was filed in October 2006 and published in May 2007. It claimed a set of a few hundred essential protein-coding genes from Mycoplasma genitalium, a tiny bacterium that lives in the human urogenital tract. It also claimed a free-living organism built from that set, versions engineered to make hydrogen or ethanol, and ways of using them. This was the "Mycoplasma laboratorium" of the press coverage.

The second, "Method for in vitro recombination", came from Daniel Gibson and Hamilton Smith. It described joining overlapping pieces of double-stranded DNA in a tube. An enzyme chews back the ends to leave single-stranded tails, complementary tails find each other, and a polymerase and a ligase fill the gaps and seal the joins. No restriction enzymes, no shuttling through bacteria at every step, and many fragments joined in a fixed order in one reaction.

Put the two together and you get what critics called "shotgun synthesis", a nod to the shotgun sequencing Venter had used at Celera: break a genome into cassettes, swap or add a few, reassemble the lot in one go. In principle that replaces the old routine of cutting a single gene into a plasmid and coaxing it into a cell. Venter had talked for years about "combinatorial genomics", running huge numbers of such variants in parallel the way drug companies screen chemical libraries.

One more piece was on the table that year. In mid-2007 the institute reported that it had moved the entire genome of one mycoplasma species into the cell of another, and that the recipient took on the donor's identity. A genome could be installed like software. The open question was whether a synthetic one would run.

2008 to 2010: the genome boots

In January 2008 the team reported a complete chemically synthesised copy of the M. genitalium genome, close to 600,000 base pairs. It was built from 101 overlapping cassettes, and the largest pieces were finally joined inside yeast, which turned out to be far better at stitching big DNA molecules than anything in a tube. A year later Gibson published the one-step, constant-temperature version of the in vitro method that now carries his name.

The synthetic M. genitalium genome was never booted. The bacterium grows painfully slowly, and the transplantation trick had worked with a faster species. So the group switched to Mycoplasma mycoides. In May 2010 it reported JCVI-syn1.0: a synthetic copy of the M. mycoides genome, about 1.08 million base pairs, assembled in yeast and transplanted into cells of Mycoplasma capricolum. The cells grew and divided under the control of the synthetic DNA.

The details were part engineering, part theatre. The genome carried "watermark" sequences encoding contributors' names, a web address and several quotations, one of them a slightly mangled version of Richard Feynman's line about understanding only what you can create. A single missing base in an essential gene had held the project up for months, which is a fair summary of what writing genomes was like at the time.

It pays to be precise about what syn1.0 was. The genome was a near-copy of a natural one, with the watermarks and a few minor changes. The cell it went into was a natural cell, with natural membranes, ribosomes and proteins. "Synthetic genome in a borrowed cell" is accurate; "synthetic life" was a headline. In December 2010 a US presidential bioethics commission reviewed the work and recommended "prudent vigilance" rather than a moratorium, which is more or less where official policy has stayed.

2016: the minimal cell, and what it did not know

The minimal genome finally arrived in March 2016 as JCVI-syn3.0, described in Science (doi:10.1126/science.aad6253). It had 473 genes on a genome of about 531,000 base pairs, fewer genes than any free-living organism found in nature. It was derived not from M. genitalium but from M. mycoides, and it got there through repeated rounds of design, build and test. Early designs based on what the literature said was essential simply did not work. Genes that looked dispensable one at a time turned out to be needed in combination.

The most quoted figure was this: 149 of the 473 genes, nearly a third, had no known function. A cell built to be as simple as possible still contained a large unknown.

Syn3.0 was not a robust organism. It doubled roughly every three hours and divided into irregular shapes. A later strain, syn3A, put back a handful of genes and behaved better, and a 2021 study traced normal cell division to seven of those added genes. In 2023 researchers reported in Nature that a minimal-cell line evolved for about 2,000 generations had regained much of the fitness lost when its genome was pared down. Minimal, it turns out, is a starting point for evolution as much as an endpoint for design.

The yeast that is half synthetic

The other big project went after a eukaryote. The Synthetic Yeast Genome Project, Sc2.0, led by Jef Boeke and run by an international consortium, set out to rewrite all 16 chromosomes of baker's yeast. This was not copying. The designers removed transposons and most introns, changed every TAG stop codon to TAA, moved the transfer RNA genes onto an entirely new chromosome, and scattered short recombination sites through the genome.

Those sites are the clever part. When a recombinase enzyme is switched on, the synthetic chromosomes shuffle: genes are deleted, duplicated or rearranged at random, and the resulting strains can be screened for useful traits. The system is called SCRaMbLE. Functionally, it is the combinatorial genomics Venter described in 2007, except that the diversity is generated inside the living cell rather than by robots assembling genomes one by one.

The first fully synthetic yeast chromosome appeared in 2014, and five more in 2017. In November 2023 the consortium published a set of papers reporting that all 16 chromosomes had been synthesised and debugged, and that a strain with more than half of its genome synthetic grew much like ordinary yeast. Merging everything into one cell was still to come.

Who owns synthetic life now

This is where the fears of 2007 met the facts of 2024. The "Minimal bacterial genome" application, the one that seemed to claim a living organism and its uses, was eventually abandoned without being granted. The minimal cell that did appear came from a different species and a different gene list.

The method application did better. It was granted in the United States in August 2010 as patent 7,776,532, later passed to the company now called Telesis Bio, and runs until 2026. Gibson assembly became one of the standard ways to build DNA in academic and industrial labs, sold as kits under licence. If you wanted to bet on who would profit from synthetic genomics, the tube was a better bet than the bug.

US law has, if anything, made designed genomes easier to defend. When the Supreme Court ruled in the Myriad case in 2013 that isolated natural DNA sequences cannot be patented, it left synthetic sequences that do not occur in nature on the patentable side of the line. A recoded or minimal genome is about as clearly man-made as DNA gets.

The bigger shift is that value moved to the infrastructure: DNA synthesis, assembly, design software and automated "biofoundries" that build and test strains at scale. That is where a handful of firms can control a bottleneck, the pattern traced in our piece on concentration ratios in agribusiness, and it is where arguments about governing synthetic biology, the subject of Lab life 3.0, have increasingly gone.

Goodbye, genetic engineering?

The 2007 question was whether shotgun-style genome assembly would make conventional genetic engineering obsolete. It did not, and the reason has a name: CRISPR. Since 2012 the cheap, precise way to change an organism has been to edit the genome it already has rather than write a new one. Whole-genome synthesis remains slow and expensive, and it is used mostly for questions editing cannot answer. What is the least a cell needs? What happens if the genetic code itself is rewritten? A Cambridge group's recoded E. coli, reported in 2019 and running on a reduced set of codons, belongs in that category.

Editing, meanwhile, became the thing regulators had to deal with, as the European Union's current fight over new genomic techniques shows.

So Synthia exists, more or less, under a less catchy name. She is small, slow and partly a mystery to her makers. The shotgun works, though mostly inside yeast. And the patent that survived was the one on the tool.