NEW YORK – Single-cell transcriptomics have changed biology, providing an unbiased view of gene expression. Now, researchers are gearing up to look selectively at gene expression in single cells, which could provide the same resolution for genes of interest at up to a 90 percent discount.
In a paper published March 30 in Nature Biotechnology, researchers led by the Jonathan Weissman lab at the University of California, San Francisco showed how they were able to use hybrid capture to get a representative sample of the transcriptome: the L1000 "landmark" gene set.
In their paper, the researchers said that hybrid capture increased the percentage of mRNA molecules aligning to target genes to 87 percent, up from 6 percent with no enrichment. They chose hybrid capture because PCR-, bead-, and linear amplification-based methods can be limited by the number of targets. Enriched gene expression profiles were also highly correlated with unenriched profiles at the global, single-cell, and single-gene levels.
Joseph Replogle, a graduate student at UCSF and the first author of the paper, suggested that the method could reduce sequencing costs by up to 90 percent for a given experiment, a statistic echoed by 10x.
"Rather than sequence at a depth of 10,000 reads per cell, you can do 1,000 reads per cell and, for your genes of interest, still have as much coverage," Replogle said. "At one-tenth the sequencing depth, I see more molecules per gene of interest. This has allowed me to functionally cluster my genetic perturbations. Those are the metrics that matter to me."
The UCSF method was part of a larger effort to scale up single-cell CRISPR screens to explore how genetic perturbations changed gene expression. "You can focus your sequencing and sequence more deeply on the things you really care about," explained Replogle. Often, that includes transcription factors, kinases, or cell surface receptors and not the more highly expressed and so-called housekeeping genes. "Most reads are going to those. In genomics we're all limited by money all the time. For most single-cell experiments, about half the cost is … single-cell reagents and half is sequencing. This change that equation."
In parallel, 10x Genomics has been developing a targeted single-cell gene expression product, also based on hybrid capture technology. Announced in February, the firm is taking orders now for several panels to be used with the Chromium single-cell platform and, soon, with the Visium spatial genomics platform. 10x had already worked closely with the Weissman lab on other aspects of their CRISPR screens and applied the company's own method to the lab's gene expression libraries during beta testing.
The data are good enough that Replogle said he would probably use 10x's product going forward rather than his lab's homebrewed method. "There are slight differences," he said. "Their platform is comparable to what we were already doing, if not even a little bit better."
Targeted gene expression has long been part of 10x's plans, according to cofounder and CSO Ben Hindson. "Once you've done some discovery in an unbiased manner, a natural thing to do is start to zero in a little more in a targeted fashion," he said. Customers, especially those in biopharma, have been asking for such a product. "It's always been in our minds, but in … the last 12 months, we put a significant effort into evaluating what the solutions are."
For now, 10x's single-cell targeted gene expression assay will run on its Chromium controller. The workflow is standard up to and including the PCR amplification. From there, the assay performs an optimized hybrid capture to pull out genes of interest.
"That gives you a lot of flexibility," Hindson said. "You can come back to the well, if you will, and keep sampling it."
10x declined to disclose where it gets its hybrid capture probes. The UCSF researchers chose Twist Bioscience as their supplier over IDT and Agilent. "The three companies have slightly different pricing models," Replogle said in an email. "I think IDT is a better deal for smaller bait sets, while Twist is better for larger bait sets/more reactions."
As announced, panels will include a pan-cancer panel of 1,253 genes spanning 33 cancer types; an immunology panel of 1,056 genes; and a human gene signature product for signaling pathways and drug targets, about 1,142 genes. 10x also plans to soon release a neuroscience panel, a custom content and custom panel designer, and extend targeted gene expression to the Visium platform.
10x has begun enrolling customers for an early-access program that it expects to run over the next several months; however, the company declined to disclose who it is in talks with. Hindson said the firm is targeting two or three customers with a mix of research and biopharma organizations.
To complement the assay, 10x has developed computational pipelines and visualization software. Once 10x is able to provide custom content, including entire panels, it will also launch a panel design software.
Hindson suggested that 10x would be open to making more targeted panels, for example, an infectious disease panel.