Richard Fox might be the world’s ultimate power user in computational biology, and when he heard about Inscripta’s new technology, he couldn’t wait to have it. So he joined the company.“We’ve been tinkering up to now,” he told me, limited by the two kinds of gene editing tools now available: tools that can edit thousands of genes with very limited edit types like knockouts, and tools that can edit a single, specific region with base-pair precision. “Both are great, but we want to open up a much richer variety of edit types across the entire genome.”Inscripta’s platform -- which will be unveiled at SynBioBeta 2019 on October 1-- promises that researchers will not have to choose between depth and breadth of genome editing, with the ability to insert, swap, or delete elements like promoters, terminators, transcription factor binding sites, and other large motifs anywhere and everywhere. “It's the combination of the number of targets across the genome coupled with a high variety of edit types -- a quadrant that heretofore has been accepted as inaccessible by researchers.”It’s an ability that Fox and his fellow protein engineers had dreamed about for years. As Inscripta’s Executive Director for Data Science, he brings his passion for applying technology to big, important problems. “It's really about taking the technology and applying it to interesting problems to stimulate the field and fuel their imagination,” Fox says.Having a wide variety of edits throughout the genome isn’t enough, though. You also need very high editing efficiency -- as much as 20%-50% of cells need to be edited, compared to one or a few percent we’re achieving with today’s technologies.“On a 96-well plate, if only one thing is edited, that's just an extreme waste of your screening throughput,” says Fox. But if, say, 40% are edited, then your odds of success increase dramatically. He explains that when there’s a mistake that leads to non-cutting -- either in the design or manufacturing of the reagents -- those cells grow fat, dumb and happy, while the cells of interest that have undergone the very insulting editing process get delayed by one, two, or more logs. “They get outcompeted big time in a pooled setting,” says Fox.Once you have high edit rates, you also need traceability. Inscripta’s platform does this by adding a barcode to the editing cassettes. “Once you carry out your editing,” Fox explains, “you can do selections where you're looking at all the edited cells competing with each other and read back with next-gen sequencing the frequency of the barcodes. So you can ask, which edits are being enriched or depleted in the population?”
Fox waited almost two years to share the results from Inscripta’s platform technology, which will enable desktop digital genome engineering at a depth and breadth previously unheard-of. The device will make its much-anticipated debut at SynBioBeta 2019 on October 1 in San Francisco. Watch Fox’s remarks at SEED 2019. Photo courtesy Inscripta.For commercial applications, freedom-to-operate is another crucial piece of the puzzle. Inscripta’s editing is based on the royalty-free MAD7 nuclease, avoiding the potentially onerous licensing terms of Cas9. “People fully own the products of their research,” says Fox. Unless you’re incorporating the enzyme into a product for sale, there are no reach-throughs or royalties.“Research has been stymied because of the onerous licensing terms that are out there for other nucleases,” Fox explains. “The MAD7 enzyme is already opening up many more applications that have been forbidden because people don't want to get into the big IP minefield.”For two decades, Fox directed computational biology groups at companies like Codexis, Dupont Pioneer, and most recently Intrexon. At Inscripta, he uses that experience in applying Inscripta’s platform to the hardest applications in the industry, testing its limits and providing feedback on product development.“One of our team’s primary missions is to break stuff and learn from these failures to improve the system,” he says. “It’s still following a passion to apply the technology, more so than actually building it. But because we are a small company and everybody wears many hats, I've played a role in product development as well.”While Fox is on the applications side, many of Inscripta’s leaders are world-class tool builders. CEO Kevin Ness previously launched QuantaLife and 10x Genomics, two companies widely recognized for breakthroughs in the genome reading space. Board Chairman John Stuelpnagel co-founded Illumina and was instrumental in making it one of the most successful life science companies. And Inscripta’s investors include Venrock, the technology investment firm that helped start little companies like Apple, Intel, and Illumina.So while the platform might have been kept proprietary to do genome-scale editing for a very successful company, it was not in the company’s DNA, so to speak. Instead, Inscripta is trying to enable true forward engineering for everyone. I ask: how will this change the way we engineer biology?“At the risk of being a little melodramatic,” Fox says, “a quote from Shakespeare’s Hamlet comes to mind: ‘There are more things in heaven and Earth, Horatio, than are dreamt of in your philosophy.’ The idea is that there's so much we don't know about biology, and despite the best efforts, it's very hard to engineer these biological systems from anything like first principles. So for the foreseeable future, we're going to have to let nature be our guide. And so empirical design-create-test-learn processes will, I think, continue to play a central role in engineering biological systems.”“In terms of what a synthetic biologist gets from this,” he continues, “is the ability to test many more ideas than we currently even dream of. To unlock all kinds of new sources of diversity for both forward and reverse engineering. There is a great deal of dark matter in genome sequence space that we still don’t understand”Here, Fox points out that the Inscripta platform’s ability to go both wide and deep is equally suited for both industrial and academic kinds of users, pursuing either forward engineering goals like optimizing a production pathway, or generating and interrogating more fundamental scientific hypotheses.“Protein engineering has long been characterized as a competition between rational versus irrational strategies,” Fox cites by way of example. “While somewhat of a false dichotomy, I think irrational methods have dominated and done well compared to purely rational methods. However with pathways and genomes the space is so big that there's a huge opportunity for the literature, databases, and modeling techniques to feed into a large pipeline of ideas. But you still have to test so many things, because our models are not sophisticated enough to really be highly predictive.”Regardless of what kind of user, Fox anticipates that the platform will be used to generate large amounts of genotype and phenotype data “to really understand biology at a deeper level, as well as understand the grammar or patterns around a lot of processes that we don't really have any handle on at all at this point.” For example, he believes it could lead to a better understanding about how regulation works, the specifics of promoters and terminators, and what really drives expression including the elucidation of complex interactions between elements.One example of the power of this platform is an application to overproduce lysine, an amino acid used in animal feed and other applications, with a global market of around $7 billion. Inscripta used the lysine biosynthesis pathway in E. coli to demonstrate the power of its new platform. In a single experiment, the company not only validated decades of published results on lysine biosynthesis, but it also found numerous edits, including amino acid changes, knockouts, and promoter insertions across the genome that gave rise to greater productivity, and many of those discoveries have not been previously reported in the literature.With the lysine example, “We're starting to talk to collaborators who can help us with more of this phenotyping, to try to understand more of the why and the how,” Fox says, “as opposed to getting a single answer.” For example, the lysine study found a number of genome-wide knockouts and promoters whose roles are not immediately obvious looking at the existing literature and databases. To understand the biology, one has to understand the relationship of the genotype to the phenotype, the metabolome, the transcriptome, and so on.And this gets to a challenge the Inscripta platform presents: there is a lot of assay power needed to capture and analyze what’s going on in these unprecedented libraries.“We think that the scale at which we can create things will need to be matched with a test capability that fully leverages what you're creating,” says Fox. Looking at existing capabilities, he says, the bottleneck has been at the “build” or “create” stage, so much so that people's imaginations may have been stunted in pushing the limits of the “test” phase.“I think what's going to happen is we're going to fundamentally invert the impedance mismatch between build and test... I think there will be a new golden age for the Agilents and others out there who are developing high-throughput test capabilities to catch the output of our platform.”As our conversation concluded, I asked: What excites you most about being able to make 200,000 edits across genomes? The answer is philosophical.“I think our technology is ideally suited to generate the kind of data that, in the age of machine learning and other forms of artificial intelligence... that we’ll start to get to the point where we actually do have predictive models of where to intervene. However, it might not be the kind of knowledge that the synthetic biology thought it was going to obtain when it set out conquer biology a decade or so ago. And that might lead to some interesting philosophical discussions about what we mean by knowledge or predictability. If it works, and you can predict, but it is difficult to understand or interpret, is that okay? From a forward engineering standpoint, it's perfectly okay. But it might lead to a different definition of what it means to know something.”The supreme ability to predict outcomes and prescribe ways to intervene that improve system performance, even in the absence of interpretive ability, is the holy grail of genome engineering. At least for gene editing, that is a grand vision and something to be excited about.