NEW YORK – Don't junk your old Illumina HiSeq 2500s just yet. A small but growing community of researchers is working on turning that sequencer into a spatial biology platform.
Researchers at the New York Genome Center and Columbia University have repurposed the imaging and microfluidic systems onboard the sequencer to implement an automated version of the antibody-based iterative indirect immunofluorescence imaging (4i) method, which offers the ability to look at multiple proteins in thousands of cells at a time with single-cell resolution.
Their rig, paired with the 4i chemistry, can detect four different proteins at a time at subcellular resolution and in three dimensions, and offers on-board temperature control. As such, it provides a possible alternative to commercial platforms such as the Akoya PhenoCycler.
"It is not quite as shiny and user friendly [as commercial platforms], and it's perhaps a little slower," said Silas Maniatis of the NYGC Technology Innovation lab, who oversaw the development of the platform. "But when you're talking a multi hundred-thousand-dollar instrument and hundreds of dollars in reagents with a defined panel versus having everything completely open for $10,000 maximum, it's pretty appealing."
Last August, the team published a preprint describing PySeq2500 — the suite of tools and a method for creating new flow cells to unlock the system’s capabilities — on BioRxiv. Already, their method is catching on.
Researchers in Loyal Goff's lab at Johns Hopkins University have built a PySeq rig, and researchers led by Robin Coope of Canada's Michael Smith Genome Sciences Centre at BC Cancer are attempting to convert an old HiSeq to do spatial transcriptomics with MERFISH (multiplex error-robust fluorescence in situ hybridization), an RNA detection method being commercialized by Harvard University spinout Vizgen. The NYGC team has also used it for immune-SABER (signal amplification by exchange reaction), among other chemistries.
PySeq is just one of many spatial omics methods to emerge in the last few years, but it could be more than a curiosity. It could help researchers get into spatial biology without tying themselves to a commercial instrument. Observing the plethora of spatial omics platforms on display at last month's Advances in Genome Biology and Technology conference, Kunal Pandit, a senior research engineer at NYGC who spearheaded the technical development of PySeq, was apprehensive. "I was thinking 'There are a lot of companies. Not all of them are going to survive,'" he said. For a lab to spend hundreds of thousands of dollars on a platform from a company that might not be there a few years down the line could be costly. "We can use [PySeq] for our spatial technologies and see who comes out on top," he suggested.
PySeq was born out of necessity, namely a desire to avoid the wrath of other scientists vying to access "our only really viable microscope," said Joana Petrescu, an MD-PhD candidate at Columbia University and a codeveloper of the PySeq platform. She wanted to use 4i in her work on neurodegeneration in amyotrophic lateral sclerosis (ALS) and dementia. The method offers more multiplexing than a single round of observation and can use unmodified antibodies, but each round takes about eight hours, and the whole process can take multiple days. Doing it on NYGC's shared Zeiss confocal microscope and immunofluorescence scanner was not an option.
She connected with Pandit, who had heard about hacking HiSeqs through an internet forum. NYGC had "a bunch of old HiSeqs that were decommissioned and just sitting there," Pandit said, so he was able to grab two. "The first one we completely took apart, to figure out how each component worked," he said. The second one was converted into the actual automated rig.
"I think the HiSeq 2500 will go down in history as one of the more impactful mass-produced scientific instruments," said Coope, the BC Cancer researcher. "They were a big step forward in capacity, a lot of them [were] run for a comparatively long time with good reliability, and important biological insights were gained. The instruments themselves are full of very high-quality optical and fluidic systems which have lots of life left, so it's great that a way has been found to give these platforms potential new uses."
"We're honoring the legacy by finding another use for it," he said. How many HiSeq 2500s are out there to potentially be converted is unclear. Illumina did not respond to a request for the number of active or recently decommissioned instruments before deadline but Pandit estimated that Illumina sold a couple thousand instruments. Illumina has discontinued the platform and will stop selling sequencing reagents for the HiSeq on February 28, 2023.
A key engineering challenge for the NYGC team was creating a custom flow cell design that would work with the system but also be inexpensive. They used regular glass slides and coverslips but needed to develop the right spacers to allow any sample to be analyzed. In addition to hand drilling holes on the slides to line up with the reagent ports on the HiSeq, the key solution was to buy a Cricut cutting device, the kind one can find at a craft store for a few hundred dollars, to cut special tape to make the flow cell spacer.
Anyone looking to implement PySeq will still need to buy a Cricut, but the flow cell slides — pre-drilled — are available through Potomac Photonics.
The other challenge was writing software to control the machinery. "The really amazing thing about the system and [the] code [Pandit] wrote is that it is compatible with the imaging as is," Petrescu said.
"You can do any protocol," Maniatis said, noting that the software controls all the important timing aspects of reagent delivery and imaging, such as autofocus.
While Coope's lab is still in the early stages of exploration, he's hopeful that his collaboration with Sam Aparicio, a fellow BC Cancer researcher who has lab-built MERFISH instrumentation, will help lead to success. Key hardware components for MERFISH setups are fluidics and fluorescence microscopy. "On paper, there's sufficient fluidics [on the HiSeq 2500] to do the equivalent of assays you could do with a lab-made MERFISH hardware," he said, noting that some labs have developed whole-transcriptome assays using that technology. "There are challenges using the 2500 for this application, however, particularly around resolution and fluorophore-filter compatibility, so we are still in the assessment stage," he added.
As a spatial proteomics platform, a PySeq rig may not offer best-in-class features compared to commercial systems, such as the Akoya PhenoCycler, which may have higher multiplexing or a wider field of view. But with temperature control between 4° C and 65° C, one can get more stringent hybridization for oligo-conjugated antibodies, for example.
The NYGC team touted accessibility. "I'm not someone who has a coding background," Petrescu said, "so the fact I can use [Pandit's] code and run experiments with minimal learning is helpful." She added that the flexibility is attractive. "We're not locked into any particular chemistry or sample types," she said. "That, we're very pleased with."
One important aspect that PySeq doesn't offer is customer support. The team has been in contact with Illumina representatives about the project, "and it's all good," Maniatis said. But Pandit said the company is not interested in actively supporting it as a platform for spatial biology applications. All the software is open source, Pandit said, so researchers are unlikely to run into Illumina software license issues.
Despite Illumina's professed disinterest, Pandit wondered if the project might inspire someone at the company to take a closer look.
"It begs the question, 'What if Illumina got in the spatial game?'" he said.