BALTIMORE – Chinese sequencing giant BGI has developed a genome-wide spatial transcriptomics technology that promises single-cell resolution, high sensitivity, and a large field of view that it plans to commercialize.
In four separate studies published in Cell and Developmental Cell this week, BGI-Research scientists and their collaborators showcased the technology's strength by constructing spatiotemporal transcriptomic maps of developing mice, Drosophila, zebrafish, and Arabidopsis.
The company also unveiled the SpatioTemporal Omics Consortium (STOC), a global research initiative aiming to promote multiomics tools to propel the understanding of spatiotemporal biology.
Dubbed spatial enhanced resolution omics-sequencing, or Stereo-seq, BGI's spatial RNA profiling method marries the company's proprietary DNA nanoball (DNB) array sequencing technology with in situ RNA capture, said Ao Chen, BGI-Research's chief scientist of spatiotemporal omics who co-invented the technology. According to Chen, Stereo-seq was originally conceived in 2019 in an effort to develop the company's existing sequencing technology for single-cell and spatial omics applications.
There are existing methods for sequencing RNA from tissues, Chen said, but these often come with limitations. For instance, bulk RNA sequencing, which profiles RNA across the tissue, cannot deliver spatial information or achieve single-cell resolution of the transcriptome. Meanwhile, single-cell RNA sequencing, as its name suggests, can drill down to the cellular level but lacks the spatial component.
RNA profiling technologies that are capable of providing spatial details—including fluorescence in situ hybridization (FISH)-based methods, laser capture microdissection (LCM), or other existing in situ RNA capture approaches—tend to suffer from other limitations, Chen said, such as having limited RNA targeting capability, being unable to achieve single-cell resolution, or covering a small field of view at the millimeter level.
In contrast, he said, BGI's Stereo-seq can not only achieve subcellular resolution of 500 nanometers but also deliver a panoramic centimeter-level view.
Chen said the Stereo-seq workflow typically starts with cryo-sectioning the target tissue into approximately 10-micron layers, which are about a single cell thick. Each layer is then placed onto the Stereo-seq chip, which is permeated with hundreds of millions of randomly barcoded DNA nanoballs and pre-sequenced to decode the spatial coordinates for in situ RNA capture. Afterward, cDNAs are synthesized through reverse transcription and harvested for further amplification and library construction. Eventually, the cDNA library is sequenced and analyzed to reveal the spatially resolved transcriptome.
"With Stereo-seq, now it is possible to really profile the entire embryo," said Longqi Liu, BGI's chief scientist for single-cell omics, who co-invented Stereo-seq. In perhaps the most prominent paper out of the four studies that the company and its collaborators published this week, Liu's team used Stereo-seq to spatially profile the transcriptome of 53 sagittal sections of mouse embryos from 9.5 to 16.5 days, a period when embryonic development is occurring at the fastest rate.
To demonstrate Stereo-seq's high sensitivity, the team also produced a series of high-resolution maps showing the location of 281,377 segmented cells while identifying an average of 1,107 unique molecular identifier (UMI) and 529 genes for each cell.
With the data they generated, the BGI scientists were able to piece together a spatiotemporal transcriptomic atlas of mouse organogenesis, offering a window into the transcriptional landscape down to the single-cell level to help understand the development of vital organs, according to Liu.
To further explore the utility of Stereo-seq for disease research, Liu's team harnessed the mouse organogenesis atlas to illuminate the genesis of certain mammalian genetic diseases. The group focused on Robinow syndrome, a congenital defect that can lead to a cleft lip and palate as well as limb shortness. Although scientists know the disorder is caused by mutations in WNT5A, a cytokine regulating the Wnt signaling pathway, the exact mechanism remains somewhat unclear. Liu and his collaborators tracked the gene's expression in the mouse embryos and observed high expression in lips and toes, unveiling the spatiotemporal window of disease development.
"I think one of the strengths that [the BGI team] demonstrated is this really large-scale data collection," said Fei Chen, a core institute member at the Broad Institute of MIT and Harvard. Having led the efforts to invent Slide-seq, a barcoded DNA bead-based technology that also aims to profile genome-wide RNA expression at high spatial resolution, Fei Chen said the underlying concept of BGI's Stereo-seq is similar to that of many other spatial RNA capture technologies in that it also deploys some kind of barcoded pixels to spatially capture the nucleic acids from the tissue.
Still, Fei Chen said what makes the BGI approach stand out to him, judging by the Cell study, is that the technology has an "extremely large" field of view achieved by the "quite impressive" array size. According to the paper, Stereo-seq chips so far have sizes ranging from 50 mm2 to 174.24 cm2. In addition, he praised the method's ability to improve capture yield without compromising resolution, which he said is one of the pitfalls for the field.
However, Fei Chen noted that the enhanced tissue area will almost always come with higher sequencing costs, a potential drawback for Stereo-seq.
In addition, BGI's DNB sequencing technology is not available in some parts of the world, including the US. To that aspect, Fei Chen said it is currently unclear how accessible the Stereo-seq technology will be to researchers worldwide once it is commercialized.
BGI's instrumentation subsidiary, MGI Tech, said in March that it will be able to commercialize products based on its CoolMPS technology starting in August of this year.
Lastly, Fei Chen said the amount of data generated by the technology will necessitate significant informatics resources, as well as new analysis tools.
BGI's chief scientist of bioinformatics, Yuxiang Li, agreed that the Stereo-seq technology pushes spatial transcriptomics to having to deal with "a real big data problem, which is unseen before."
To cope, Li said, the company has developed two bioinformatics pipelines, Stereo-seq Analysis Workflow (SAW) and Stereopy, that both catered to the computational needs of Stereo-seq. While SAW focuses on the generation of gene expression profiles from the raw data files, Stereopy mainly handles the downstream analysis, Li said, adding that the company is developing more tools for cell binning, data denoising, cell clustering, annotation, and 3D reconstruction.
In terms of accessibility, although the technology is not yet available on the market, Ao Chen said BGI has already made Stereo-seq available to hundreds of researchers around the world under academic collaborations, especially through the STOC initiative.
Launched by BGI as an open scientific collaboration to enhance the understanding of cellular interactions, the consortium has already attracted scientists from institutions such as Harvard University, the University of Oxford, MIT, the University of Cambridge, the Karolinska Institute, and the Genome Institute of Singapore.
According to Ao Chen, BGI plans to foster worldwide collaborations around Stereo-seq in three ways. These include establishing demo labs around the world—especially in the US, Europe, Australia, and Singapore—to which the collaborators can send their samples for processing. In addition, he said the company is willing to help collaborators build their own Stereo-seq labs, where BGI can provide reagent and software support. Lastly, Ao Chen said the company can share Stereo-seq data with researchers around the world, enabling scientists to use the data to decipher the biology behind the sequences.
Similarly, Ao Chen anticipates that once Stereo-seq is commercialized, it will be offered both as a service and an instrument, just like the company's sequencing technology. He noted that BGI is currently developing an automated platform for Stereo-seq, although he did not disclose a specific commercialization timeline.
Although BGI scientists estimated in the Cell paper that the current DNB capture chip costs about 220 RMB, or roughly $35, per square millimeter, Ao Chen acknowledged that the cost for the technology right now is "still too high," adding that for Stereo-seq to be widely adopted, a price reduction of at least 50 percent would be desirable.
As for turnaround time, Ao Chen said a typical Stereo-seq workflow normally takes less than a week from sample prep to analysis. He said BGI has also devised a high-throughput Stereo-seq model, where the downstream cDNA libraries from different samples can be pooled and sequenced together, allowing the company to process hundreds of samples per month.
Moving forward, Ao Chen said the company plans to continue optimizing the Stereo-seq technology, in addition to improving its cost. Specifically, he said BGI is working on further enhancing the resolution to allow researchers to define the cell boundaries more accurately. Moreover, he said, the company aims to achieve multiomics capabilities for Stereo-seq by integrating genomics, epigenomics, and proteomics profiling on one chip.
In addition to the model organisms investigated in the studies published this week, Liu said BGI scientists and their collaborators have also tested Stereo-Seq on the monkey brain, which requires a larger chip size. Most importantly, he said, the company is gearing up to apply the technology to human samples this year, such as cancer tissues, with the ultimate goal to unveil cell-cell interactions as well as the cellular spatial landscapes in the context of human disease, development, and evolution.
"Stereo-Seq opens a new era in life science," Liu said. "In the future, there are a lot of possibilities."