This story has been updated to include additional information from Bruker.
NEW YORK – The last time the Advances in Genome Biology & Technology annual meeting was held in person — February of 2020, for those who have been counting — spatial transcriptomics technologies had barely broken out of academic labs.
10x Genomics had launched its Visium Spatial Gene Expression solution just a few months earlier, and targeted, highly resolved methods, such as Bio-Techne's ACD RNAScope could handle just a few different genes at a time.
At this year's AGBT meeting, held last week in Orlando, Florida, at least nine companies showed just how far the market has developed. Germany's Resolve Biosciences, a spatial genomics startup with ties to Qiagen, held the conference's top sponsorship, using the opportunity to tout its Molecular Cartography platform. NanoString Technologies sponsored a pre-meeting workshop where it dramatically unveiled its CosMx spatial molecular imager, offering higher resolution than its GeoMx digital spatial profile. And 10x Genomics provided the first in-depth look at its Xenium in situ analyzer, built from parts acquired from Cartana and ReadCoor in 2020.
Akoya Biosciences threw its hat into the ring, announcing a new RNA chemistry for its PhenoCycler-Fusion platform. Vizgen introduced a method for its Merscope platform to make it work with formalin-fixed, paraffin-embedded samples. Bruker touted its Canopy CellScape platform for spatial proteomics and Acuity platform for 3D genome organization. Also in the market are platforms from Rebus Biosystems, Veranome Biosystems, and Spatial Genomics.
The myriad options for acquiring spatial data are "fairly overwhelming," said Tom Wilson, director of the University of Michigan's Advanced Genomics core. He said his lab and others on campus run several spatial platforms already, including Visium and NanoString's GeoMx. "I've lost track of how many vendors are competing to get your attention on this," he said.
"It's a lot to take in," echoed Shawn Levy, senior VP of applications at sequencing startup Element Biosciences, who previously ran a genomics lab at the HudsonAlpha Institute for Biotechnology.
Resolve Biosciences
Resolve announced last week that it has begun installing its workflow at customer sites, whereas previously samples were run at its Germany- or San Jose, California-based service labs. The workflow is fully automated, Resolve CEO Jason Gammack said, and consists of a sample processing unit, imaging system, and high-performance computer for image processing and analysis. The firm has also added the ability to do FFPE samples, in addition to cell cultures and fresh frozen tissue.
Resolve remains somewhat opaque about its transcript detection chemistry. The process targets a panel of 100 genes, with gene-specific probes for each molecule.
"We label each target molecule with gene-specific probes. These probes are colorized and imaged," Gammack said, though exactly how this step occurs is still unclear. Gammack further noted that the probes are small and diffuse easily to ensure they find their target. "If it's in there, we can tag it," he said.
The probe detection part of the process uses multiple rounds of imaging to create a combinatory color barcode unique to each transcript. Fluorescent readout is added and removed in each cycle. The "barcode" is ultimately the sequence in which different readout channels are expected to light up, said Jeroen Aerts, a customer technology advisor at Resolve, said during a sponsored talk.
The method offers 90 percent sensitivity compared to single-molecule FISH and a false positive rate of less than 1 percent, Gammack said. "We can treat every dataset as a single-cell dataset," he said, by segmenting the image based on the transcripts it expresses. Resolve has now completed 70 projects in 37 different tissue types. The method works well in high-fat tissues, which have higher background levels of autofluorescence and might confound spatial methods, he added.
In a statement provided to Resolve for its AGBT presentation, Martin Guilliams, an immunology researcher at Belgium's VIB, said Molecular Cartography "came out as the very best of the class" in a benchmarking study of multiple platforms. "It worked right the first time on challenging clinical samples. It will be the gold standard for years to come," he said.
In a sponsored talk, Michael Snyder a genomics researcher at Stanford Medicine, described the use of Molecular Cartography as part of his work at the Stanford Tissue Mapping Center, where his team used it on small and large intestine samples.
They ran their 100-gene panel on samples from two healthy donors. Out of 60 tissue section, 44 passed QC, he said, and all but two probes worked. They used the Baysor cell segmentation algorithm, he noted.
This yielded 7,500 cells per section, on average. Their data showed averages of eight genes per cell and 71 molecules per cell.
This data helped show that two cell types with similar differential gene expression patterns were mapped to different parts of the tissue. Despite their UMAP plot similarity, "spatially, they're quite different," he said. "There's clearly different behaviors going on in these two regions."
The first instruments have been installed in Snyder's lab, VIB, the Novo Nordisk Foundation Center at the University of Copenhagen, and the German Cancer Research Center (DKFZ). Gammack said the firm is also developing its second-generation chemistry, targeting launch in early 2024.
"We will be raising funds," Gammack said. "The money is out there," however, he noted that the slide of US stock markets since highs of several months ago have created challenges. "Our comps are down 50 to 80 percent," he said, making it hard to reconcile his perceived total valuation with what investors may be willing to value the company at based on a potential IPO (or on current revenues).
NanoString
With a dedicated pre-conference workshop featuring presentations from early-access customers including Weill Cornell Medicine's Christopher Mason, and more than 15 oral and poster presentations throughout the AGBT conference, NanoString was definitely not bashful in promoting its newly announced single-cell spatial multiomics offerings.
In the spatial multiomics symposium leading to the conference, filled with zesty pop music and kaleidoscopic stage lights, NanoString rolled out — literally — its CosMx Spatial Molecular Imager on stage, amidst a cinematic smoke effect.
Touting the company's first spatial multiomics single-cell platform "the best in class," NanoString's technical sales director Jennifer Hart told attendees that CosMx, currently with its 1,000-plex RNA panel, can detect more genes than any other in situ imaging platform out there today.
Additionally, she said for each gene, up to five RNA-detection oligonucleotide probes, or tiles, of ISH probes were designed to independently detect different regions of the RNA target. Finally, she said each reporter construct contains a controlled number of 15 to 60 dyes, depending on desired sensitivity, assembled with fluorophore-conjugated oligos with photocleavable (PC)-linkers. All reporters are single-color, containing one of four fluorophores: Alexa Fluor-488, ATTO 532, Dyomics-605, or Alexa Fluor-647.
In terms of multiplexing, the company said that, with the tissue scan area of up to 300 mm2, CosMx currently can handle 1,000 targets for RNAs and 100 for proteins. As for throughput and turnaround time, the official boasted that the "tunable workflow" of CosMx can yield two to 20 slides per week, roughly equivalent to 4 million single cells, though she said the exact throughput and turnaround time for CosMx SMI depends upon the plex and tissue area analyzed. She also said the company is planning to significantly improve CosMx throughput and turnaround time in the next year, as well as enabling simultaneous RNA and protein detection in late 2023.
While NanoString did not disclose CosMx pricing on stage, the company later told GenomeWeb its sticker price is $295,000. As for reagents, the company said the 1,000-plex RNA panel costs $3,300 per slide, while the 100-plex RNA Panel costs $1,850 per slide and the custom RNA panel has a list price of $2 per target per slide. However, NanoString said it has yet to finalize the price for the 100-plex protein panel. The company also said all pre-validated panels include four protein markers for morphology and single-cell segmentation as well as all consumables from sample preparation to data generation.
NanoString also unveiled AtoMx SIP, a new cloud-based informatics portal for use with GeoMx and CosMx data that will be launched later this year. According to the company, AtoMx promises "an integrated ecosystem with streamlined workflows" to manage, analyze, and share data from its GeoMx and CosMx platforms. Its scalable solution can reduce data computing time from days to hours while enabling secure data sharing and collaboration across laboratories and institutions.
During the symposium, NanoString's Hart said the company has already garnered over 30 customers in the early-access program and presold over 35 systems, which will be shipped to customers starting later this year.
Grant Kolar, associate research professor at Saint Louis University School of Medicine and an early CosMx customer, said his lab used the platform to generate a spatially resolved, single-cell atlas of human pancreatic islet cells.
At NanoString's symposium he showed a human pancreas atlas mapped from nine FFPE samples — including three from patients with Type 2 diabetes mellitus (T2D), three from metabolically healthy but obese (MHO) patients, and three normal controls — using the CosMx 1000-plex RNA panel spiked with more than 30 custom genes of interest.
"We identified 463 islets out of the dataset for us to compare," Kolar said. "That's huge for us." The data helped his team make insights into the relative abundance and distribution of these cell subpopulations in healthy controls and their redistribution in the context of MHO and T2D. In addition, Kolar's team investigated the abundance and distribution of immune cell populations that were present in the tissue, providing insight into the ongoing immune response in the context of tissue damage.
Another NanoString customer, Miranda Orr, assistant professor of gerontology and geriatric medicine at Wake Forest University School of Medicine, used CosMx in a study of Alzheimer's disease in a mouse model.
Her team looked at 1,003 RNA targets covering a range of neuroscience applications in wild-type and tau-transgenic mouse brain tissues. The CosMx data revealed cell-type differences between the two sample types, including fewer neurons in the tau-transgenic sample compared with the wild type, and an increase in oligodendrocytes, a phenotype that Orr said her team hadn't fully appreciated before. Additionally, she said CosMx enabled her team to associate the subcellular localization of different gene transcript with different aspects of cellular senescence, data that she suggested can help reveal the underlying biology of the disease.
With CosMx, "I think we can answer and will be answering some of the same questions we had for a decade or longer, but now with such higher resolution and such better accuracy," Orr told the audience. "It's just so phenomenal … all I can say is I can't wait to get one."
Akoya Biosciences
Akoya, a company known for its spatial protein analysis offerings, also stepped into the spatial genomics market by unveiling a new RNA chemistry for its PhenoCycler-Fusion platform to a packed sponsor's suite at the meeting.
The RNA-detection workflow largely resembles that of the firm's protein-detection assay, according to Akoya Senior R&D Director Julia Kennedy-Darling, and uses a targeted, amplification-based chemistry to detect markers through cycles of hybridization, imaging, and dehybridization. The RNA workflow requires an overnight sample prep step, resulting in prep time of about a day, while the protein assay typically takes less than that.
Oligo-labeled antibodies detect the proteins within the tissue; both the RNA and protein chemistries use dye-labeled oligos to read out the signal for subsets of targets across multiple cycles.
Akoya's RNA chemistry can process 1 million cells in as little as 10 minutes, Kennedy-Darling claimed, and will be able to do back-to-back preps to analyze RNA and protein markers from the same sample, a feature that is not currently available but is slated for release next summer. Moreover, it could enable researchers to analyze around 100 biomarkers per sample for both RNA and protein with single-cell and subcellular resolution. Ramachandran added that the company has plans to offer 1,000 plex assays down the road. "If you can get to 100, you can get to 1,000," he said.
PhenoCycler-Fusion was able to produce "strong expression across a broad range of targets" in various human FFPE tissues, Kennedy-Darling said, adding that the performance of the protein assay is not diminished by the presence of the RNA detection.
With the integrated multiomics workflow, we are "opening up our business to a market segment we haven't touched before, which is those genomic customers," Akoya CEO Brian McKelligon said in a post-launch interview. He hopes the firm can place multiple thousands of instruments in that market over the next several years. The firm also believes this platform can compete with auto-stainers in the clinical chemistry market, where he said there are approximately 13,000 to 15,000 placements.
Akoya currently has an installed base of roughly 750 instruments, mostly concentrated at oncology labs, Ramachandran noted. To allow these existing customers to run the RNA chemistry on their platforms, Akoya plans to roll out instrument upgrades to most customers by the end of this year, he said.
Given that the new chemistry is early in product development, Ramachandran said the company has yet to settle on a price. However, he said that for protein analysis, the current cost is around $1,000 per sample for a 50-plex assay.
Though having fewer presentations than NanoString, Akoya featured in several posters at the conference highlighting data from early-access customers including the Wellcome Sanger Institute and the Broad Institute. According to Akoya's Ramachandran, these institutions are the first customers to get access to the company's RNA chemistry, though in both collaborations, Akoya ran the assay for the researchers.
In one poster presented by Nadav Yayon, a researcher at the Wellcome Sanger Institute, Akoya's Hi-plex RNA detection assay was used to analyze thymus tissues, a lymphoid organ beneath the breastbone. The researchers reported that the assay can detect many RNA targets simultaneously within the thymic tissue without perturbing its integrity, enabling the localization and identification of key cell types involved in biology. Specifically, the researchers showcased that the technology was able to distinguish unconventional T cells, activated dendritic cells, and the spatial correlation of immune cells with stroma cells, illuminating the unanswered questions related to the trajectory of T cell differentiation.
In another poster presented by Alex Tsankov, a researcher at Icahn School of Medicine at Mount Sinai who was a postdoc at the Broad, the researchers performed spatial phenotyping of immune cells in non-small-cell lung carcinoma patients. The results showed that the RNA assay can detect individual RNA molecules at single-cell resolution. Additionally, the researchers used the method to annotate certain cell types while accessing the immune, tumor, and stromal regions of the tissue to map the neighborhoods present within the tumor microenvironment.
During Akoya's launch event, a researcher from the audience asked a question that resonated with many: How does Akoya's technology differentiate from others?
"The number one reason to choose [Akoya] is speed," Akoya CBO Ramachandran answered, adding that the company's platform is "probably 10 times faster than any of the spatial biology systems out there." What's more, he said the company is aiming to increase the speed of the system every six months over the next year and a half, and hopefully by next summer, the system will be capable of running 10 to 30 samples per week with 100 RNAs per sample.
Moving forward, Ramachandran said one of the goals for Akoya is to figure out a way to scale RNA detection up 1,000 RNAs per sample. In addition, he said the company aims to determine the best applications for its RNA-plus-protein workflow while defining content for its 100-plex RNA panels.
To that, in addition to the current collaborations with Sanger and Broad, the company also plans to forge more early-access collaborations in Q4 this year to generate more datasets for the new technology.
10x Genomics
10x recently accelerated its development timeline for Xenium, announcing it would be ready by the end of the year.
Company officials expect the launch to mirror that of its Chromium single-cell platform. "When we brought the first version out, we knew we had lots of room to grow," 10x Cofounder and Chief Technology Officer Michael Schnall-Levin said of Chromium. "I think with Xenium it's going to be the same thing."
10x will provide "a broad menu of panels" for Xenium that can target many FFPE sample types, such as brain, lung, and gut, 10x Staff Scientist Francesca Meschi said at an AGBT sponsored lunch workshop. "We also want to hear from you what are the specific tissues or applications that really are going to empower you to use this technology," she said.
In her talk, she described how in Xenium, DNA probes hybridize to multiple tissue sections placed on each slide, while multiple copies of gene-specific barcodes are generated by ligation. The chemistry could be considered a redesigned Cartana chemistry, she suggested, with higher sensitivity from proprietary enzymes. "We have made massive improvements to Cartana from the probe design, chemistry, optics, and software and analysis to make this a high sensitivity, extremely specific in situ platform," a 10x spokesperson added in an email.
Fluorescent oligos then bind the gene-specific barcodes. "This is hands off," she said. "Each cycle, a gene is turned off or on in a specific color" creating an optical signature specific to each gene.
The platform can perform onboard data analysis, handling up to 1 terabyte of data. It can take each image and convert the optical signature into a gene identification, which happens in parallel to the run. It can also assign transcripts to cells and transfer data to the firm's Xenium Explorer software suit to visualize data off-instrument.
Throughput is up to about 1,700 mm2 per week. Each run can analyze multiple slides — enough to cover about 12 mouse brain sections — with room to increase that in the future, she said.
She showed data for a 200-plex gene panel in mouse brain tissue, which detected all transcripts, including "exemplary cell type-specific markers," and showed good correlation with Visium, she added. While the first version of the product will be able to analyze up to 400 genes, "the biochemistry is to build in a way that has the flexibility to expand for higher capacity," she said.
An experiment with human breast cancer samples enabled 10x researchers to achieve differential expression analysis at single-cell resolution, revealing biomarkers that are indicative of tumor progressions, such as the loss of myoepithelial cells. The platform can also reveal structural morphological information about the tumor, providing further biological insights, she said.
For now, Meschi said Xenium can only detect RNA, but will be able to detect RNA and protein from the same sample sometime in 2023.
"We're building a strong foundation," she said. "We're building this in a flexible way so that we can keep adding on." The platform remains on track to be available by the end of the year, she said.
Schnall-Levin suggested that Xenium will appeal to people already using Chromium and Visium who want to work on tissues, although he hopes to reach researchers who don't already do a lot of sequencing. He said he was most curious to get feedback from researchers who have never done single-cell sequencing.
"Ultimately, this has huge clinical potential" with translational and clinical trial researchers, he said.
But wait, there's more!
One of the academic presentations suggested that spatial analysis may be able to incorporate epigenetic information. Chongyuan Luo, a researcher at the University of California, Los Angeles, presented on his lab's development of so-called photonic-indexing sequencing. At a high level, the chemistry works by controlling in situ ligation reactions with light. "It can be adapted to any kind of reaction," Luo said, including RNA-seq, single-cell ATAC-seq, or single-cell methylation sequencing.
And a methods paper published last month in Star Protocols demonstrated how to detect both protein and mRNA markers in human FFPE samples on the Bruker Canopy ZellScannerOne. The research was led by Dirk Busch of the Technical University of Munich.
There are plenty of differentiators between the various spatial platforms on the market. Important considerations include the number of analytes the platform can analyze, the ability to do multiple analyses on the same tissue, multiplex capability, spatial resolution, data processing, and vendor reliability, Wilson, the Michigan core lab director, said. And the ability to process FFPE samples is "a paramount requirement," he said.
In that vein, performance and reliability in a variety of conditions could be the difference-maker.
"How robust these technologies are in other people's labs and hands, and how well they can process not-so-well-preserved post-mortem samples are very important," said Joana Petrescu, a researcher at the New York Genome Center who uses spatial omics technologies to help investigate neurodegenerative disorders.
Some labs may be able to try a lot of different systems, but Wilson said his lab won't be able to afford to.
"We're in a waiting game right now," he said. "I need to do a little homework."