NEW YORK – Stellaromics, a startup founded by Karl Deisseroth at Stanford University and Xiao Wang at the Broad Institute, has launched an early-access program for its Pyxa 3D spatial transcriptomics platform. While most of the interest so far has come from neuroscientists, at least one lab is planning to use it to study cancer.
Nigel Jamieson, a physician-scientist at the University of Glasgow in Scotland who runs a lab focused on multiple spatial biology methods, is planning to use Pyxa to study the tumor microenvironment and immuno-oncology. "We can see how the cancer invades," he said. "In colon cancer, the tumor cells bud off. That's quite difficult to capture in 2D."
With the ability to look at tissue sections 5 µm to 20 µm or 20 µm to 100 µm thick by confocal imaging, Pyxa offers much more depth than existing high-resolution spatial transcriptomics methods, which analyze standard tissue sections around 5 µm thick. "You're working in a box, as opposed to on a sheet of paper," Jamieson said.
Presently, Pyxa can only analyze mouse brain samples, so Jamieson's first samples have been of mouse brain cancer tissues. However, his lab is working with Stellaromics on optimizing a gene panel for use in human tissue, which he plans to use with colon cancer liver metastases. "We will very likely have additional sample type protocols in early access, but they have yet to be determined," Stellaromics CEO Todd Dickinson said in an email.
Ultimately, Pyxa data should help with cell phenotyping, Jamieson said.
Pyxa is launching with the ability to detect panels of about 250 RNA targets. Earlier this month, Stellaromics announced an $80 million Series B financing round to help support development and commercialization of the platform. The early-access program has five sites, according to the firm, and commercial shipments are scheduled to begin in December. It is also running a services program to allow researchers to test the technology and generate data for grant applications.
The Pyxa platform runs 12 samples at a time, each in its own well with a diameter of 16 mm, or a detection area of approximately 200 mm2. Runs take two days or more, depending on the number of genes being analyzed. Sample prep, including clearing tissues of certain proteins and lipids, happens off-instrument and takes three to five days.
The panel size makes Pyxa competitive with 10x Genomics' Xenium in situ spatial biology platform, Jamieson said, while Bruker's NanoString CosMx platform can essentially assay the whole transcriptome. He operates both of these platforms — which can also detect proteins — in his lab, among others.
In an email, Stellaromics said its standard 250-gene panel detects between 100 and 150 transcripts per cell in 100 µm tissues. "The customer could add custom genes to get more transcripts," according to the firm. "We see some cells with more than 300 transcripts."
Jamieson has not yet received his Pyxa instrument but expects it by early summer. At first, he will have to run fresh-frozen tissue samples. "Ultimately we want to also run this on [formalin-fixed] paraffin-embedded (FFPE) tissue," Jamieson said. "That's where a lot of the technologies are moving. I'll be happier when we have that ironed out." Having flexibility in panel design will also be important, he added.
Stellaromics is also working on the ability to detect proteins and ribosome-bound mRNA. The firm declined to discuss pricing for both the Pyxa instrument and per-sample rate for consumables at this time.
Of course, other researchers will be using Pyxa as Wang and Deisseroth have: to study the brain. Xin Jin, a researcher at the Scripps Research Institute, began using STARMap — the academic method being commercialized by Stellaromics — through a collaboration and plans to purchase Pyxa to continue that work. Her lab has long used single-cell sequencing assays to study "how different cell types talk to each other and assemble into circuits and ultimately give rise to thoughts and behaviors," she said. "Adding spatial information will be crucial."
Pyxa's ability to capture information on the connections between cells could be "extremely valuable," she said. "Inherently, the brain operates at long range." Neurons can connect all the way to the spinal cord, for instance. "That's an extreme example. But there's so much information flow that's not happening in a thin [tissue] section," she said.
A challenge facing all spatial transcriptomics technologies is how to phenotype a given cell. The current standard of analyzing 5 µm sections means many human cells, which can be around 10 µm in diameter, aren't getting captured in whole. "You're not getting all the information from that sphere," Jaimeson said. "At a fundamental level, I'm interested to see how cellular phenotyping can be improved if you see the whole cell and the cells next to it," he said, adding that algorithms for phenotyping a cell are often based on the cells next to it.