NEW YORK – RootPath, a startup with ties to the US and China, is racking up partnerships around the world, based on its ability to create large numbers of synthetic DNA constructs.
"We're first and foremost a synthetic biology company — we have a broadly enabling platform and we will pursue different applications," said Cofounder and CEO Xi Chen. The core technology for the firm is a method called PathFinder DNA assembly to create hundreds of synthetic genes, thousands of bases long, using DNA origami principles and synthetic oligo pools. The company has also developed technologies to pick out particular constructs from its reactions.
"Making genes is a lot like building brick walls," Chen said. "We've programmed the bricks to be robots that can recognize each other. So, if you have a dump truck of bricks, they will self-assemble; the bricks themselves are doing all the work."
But not all of the genes are identical. "We're building a thousand genes at the same time," he said. This offers benefits for applications where variation is valuable, such as in T-cell receptor (TCR) and antibody profiling.
The technology has led the firm into external collaborations with researchers at Barcelona's Vall d'Hebron Institute of Oncology and the Parker Institute for Cancer Immunotherapy. Chen said the firm has "a handful" more, but he declined to name them due to confidentiality agreements.
"The technology commercially available only allows you to test a few TCRs at an affordable price," said Alena Gros, a tumor immunology researcher at Vall d'Hebron and a member of the firm's scientific advisory board. Testing 10 TCRs costs approximately €1,000 ($1,066.) "That's not enough TCRs for the projects we were working on," she said. Now, with RootPath, she can test 1,000 TCRs at a time, getting the synthetic DNA for free as part of her collaboration.
RootPath will pursue some of the potential applications itself, beginning with cell therapy. The firm has a T-cell receptor pipeline and is already collaborating with a Chinese partner institution to conduct "investigator-initiated, first-in-human trials," Chen said, using a "a regulatory channel specially designed for cell therapy products."
Chen founded the company in 2017 along with Ely Porter, the firm's chief technology officer; Yinqing Li of Tsinghua University; Jun Wang, a pathologist and cancer immunology researcher at New York University Langone Health; and Cheryl Cui, partner at Nest.Bio Ventures and CEO of Bota Biosciences.
Though Chen completed his doctorate in biochemistry at the University of Texas at Austin with a focus on DNA nanotechnology, he said he came back to the field after swearing to leave it behind. After graduating, he joined Nest.Bio Ventures, which runs a biotech incubator in the biotech hotbed of Kendall Square in Cambridge, Massachusetts.
"When I joined Nest.Bio, I said I was going to give up all my pre-existing knowledge," Chen said. But he identified T cells as a promising field within therapeutics. "The biggest problem is cheap, high-throughput synthesis of TCR genes," he said. "We can get sequences more and more easily, but we don't know which ones are functional until we can test them. The bottleneck is the ability to synthesize them."
Combining DNA origami with molecular programming — the ability to quantitatively engineer biomolecules in complex manners — "can really solve this problem," he said. "The core of the secret sauce is in the sequence design. You have to engineer 1,000 orthogonal reactions, at the right speed, and you can't have cross-talk."
RootPath has incorporated concepts such as logic gates and feed forward loops in their reactions, along with simulation tools, to engineer tens of thousands of short, pooled oligos to self-assemble into thousands of genes. "In conventional DNA assembly, like Gibson assembly, you're executing three or four instructions in a 10-microliter reaction. We can execute 3,000 instructions in that same 10 microliter reaction."
Since its founding, the company has grown to approximately 90 employees and two locations in China. A "small but fierce" core team operates out of the Boston area, while the company has a GMP facility in Hangzhou, China, near Shanghai and a new facility in Guangzhou. It has done so with more than $60 million from investors including Sequoia Capital China, Matrix Partners China, CDH Investments, and Sky9 Capital.
PathFinder technology starts with oligo pools and produces genes several Kb long. "In theory, as long as you can PCR-amplify the gene, we can make that length," Chen said, suggesting it could easily reach 7 kb, even 10 kb. Some gene pools contain DNA longer than 5 kb, he said; however, TCRs and antibody genes are usually 1 kb to 2 kb.
Chen wouldn't say where the company gets its oligo pools, although they are available from several DNA synthesis companies including Twist Bioscience, Danaher's IDT, and GenScript. The pools are "very, very cheap," he said. "So low, we consider them free."
Chen claimed that "you can synthesize any arbitrary sequence the computer spits out." Moreover, the technology can produce different sequences in the same reaction. "A lot of people try to assemble multiple genes in one test tube," he said. "They can make five, maybe 10 genes, if the sequences are very different from each other. If you're building 10 variants of the same genes, or of the same class, that have similar amino acid or DNA sequences, then things go to hell. Our method is not plagued by that."
Each reaction can produce between 500 and 1,000 different genes. The assembly pool does have errors, but the firm has also developed so-called "nano toothpick" technology to pick out "perfect" sequences, with a per-base error rate of less than one in 30,000 bases. Each assembled molecule gets a unique barcode, similar to a unique molecular identifier used in next-generation sequencing. Erroneous molecules are discarded, and the barcodes are used to isolate the final product.
TCR profiling is the most immediate application for RootPath's synthesis method. Gros' lab is interested in experiments that consider 1,000, patient-derived TCRs at a time. They take T cells from peripheral blood of cancer patients, sequence them at a single-cell level to obtain the TCR genes, and synthesize them to screen for tumor recognition. Moreover, each TCR in the library is coded with three different nucleotide sequences featuring different codons, in hopes of boosting robustness.
With synthetic TCRs, "you're not at the mercy of how frequent the TCR is in the patient sample," she said. Tumor-specific TCRs are not frequent in blood samples, which can lead to a skewed library. "With RootPath, you choose the TCRs, so you don't have this bias," she said.
Gros said the first library she got took two months to synthesize. "Regular gene synthesis for one TCR could take four weeks," she noted.
RootPath has "always been very responsive" to questions, she noted, and has even helped her lab with single-cell sequencing and is working on getting cell surface receptor protein data. "These analyses are all part of the collaboration and are not offered to customers," Chen noted.
As an adviser to the company, Gros receives compensation, which she and RootPath declined to disclose. Gros noted that she has been collaborating with the company for nearly two years and was not on the advisory board "until recently."
Personalized T-cell therapy is "a match made in heaven" for the firm's technical capabilities, Chen said, especially tumor-infiltrating lymphocytes. Conventional approaches for this cell therapy grow up T cells from the patient tumor sample ex vivo and reinfuse them. It can work well in highly immunogenic tumors, Chem said, but it doesn't always work so well in others, including lung and colorectal cancers. "The reason is that just because a T cell is in the tumor does not mean it reacts to the tumor," Chen said. And some that are, have been exhausted. Chen suggested that finding tumor-reactive T cells using the firm's technology is "quite straightforward."
Single-cell sequencing can provide the TCR sequences. "The hardest part of assaying them is making 1,000 genes," Chen said. With PathFinder, the cost for doing so goes down.
Getting into the T-cell therapy space has led to other innovations, Chen said. "No CRO can assay 1,000 TCRs in two days. We had to build that," he said, as well as the "not trivial" task of delivering individual genes to T cells and integrating pooled screens and arrays. Data from a pilot study have been "very, very promising," Chen said, and the firm's collaboration will be moving into its next phase in the second half of the year.
Other potential applications include antibody discovery, mining bacterial genomes for enzymes in industrial applications, and even gene therapy. "For TCR genes, we only work with collaborators," Chen noted. "For other types of genes, we are working with many beta testers now. If all goes well, we will commercially offer the service to customers."
The company is one of several recent entrants to the market promising longer, purer synthetic DNA.
RootPath continues to work on its per-base error rate and turnaround time. Whereas Gros' first libraries took two months, it now takes three or four weeks. "In the near future, after automation is online, we can shrink it to two weeks," Chen said.
More broadly, Chen hopes his company can bridge the gap between genome sequences and functional readouts. "Key resources and values in biotech in the next 10 years will be created by going back and forth between the two," he said. "What we're going to do is remove the speed bumps in between."