Spatial Transcriptomics: A window into disease

Newsletter Signup - Under Article / In Page

"*" indicates required fields

The emerging spatial transcriptomics field is letting researchers map out gene expression in tumor and brain samples. How could this change the way we treat disease?

DNA and RNA sequencing technology has undergone a revolution in the last decade, with companies working to provide ever cheaper and faster ways to explore our genetics.

This stride forward in sequencing has, in turn, led to the development of personalized medicines to treat a range of diseases such as cancer. For example, physicians can plan the most appropriate immunotherapy for a patient based on DNA and messenger (m)RNA extracted and sequenced from a tumor sample.

However, this sequencing approach has its limits.

“You want to look at not just what immune cell types are there, but where the immune cell types are, what immune cells are present, and are they actually invading the tumor or not,” said Richard Terry, founder of tech company ReadCoor, which was acquired by the California-based firm 10X Genomics in 2020.

“This rich complexity is impossible to get at once you start destroying that rich contextual information.”

There are already methods for visualizing mRNA sequences on tissue sections, such as RNAscope, an in situ hybridization technology developed by the U.S. company Advanced Cell Diagnostics. However, the maximum number of different genes you can analyze at a time with this technique is limited.

“People like to use it and it’s very quantitative and sensitive, but you can only look at three or four genes,” explained Malte Kühnemund, director of R&D at 10X Genomics. “What you really need is a method that can look at 50-200 genes.”

How did the technology come about?

In a 2013 Nature study, a research group led by Mats Nilsson in Stockholm, Sweden, reported a way to fuse the spatial information of in situ hybridization with the detail of transcriptomics in a technique called spatial transcriptomics, or in situ sequencing. Other groups in Stockholm developed their own versions of the technology, and the city became a European hub for spatial transcriptomics.

One spatial transcriptomics method — published in Science in 2016 — involves rows of spots on a microscope slide that are made of millions of DNA primers and unique DNA barcodes. A tissue sample is then placed on the slide and each primer sticks to a target mRNA sequence nearby in the tissue section. The captured mRNA sequences are then converted into DNA, and each sequence also has the location barcode included. The researchers can then use the barcode to map the mRNA onto the part of the tissue where it was captured.

This technique can be used to track a lot more genes than previous spatial techniques. Along with other spatial sequencing technologies such as spatial genomics and proteomics, it has advanced the Human Cell Atlas, an international collaboration aiming to functionally map the cells of the human body.

Spatial transcriptomics: a home in Europe and beyond

Spatial transcriptomics techniques have quickly gained a strong footing in the scientific community and have even been picked up by the industry in the relatively short time frame they have been around, explained Maike Priester, director of Product Management and Marketing at the German oncology company Indivumed.

“Backed by the fast-paced research to improve data resolution in the field, biotech companies like 10x Genomics have developed techniques that provide spatial visualization of genes down to a cellular level. Considering that this resolution was still at a regional level within tissues less than seven years ago, this is a remarkable development,” said Nicole Kerstedt, director of the Genomics Department at Indivumed.

Moreover, spatial transcriptomics can be used in conjunction with other scientific advances in the field of multi-omics, which provides insights using datasets from different biological analyses focused on the genome, proteome, microbiome, and more. Combining different techniques in such a manner can add a high dimensionality to clinical research.

Swedish companies have drawn a lot of attention in the spatial transcriptomics space. For example, the Stockholm-based firm Spatial Transcriptomics was purchased by 10X Genomics in late 2018. 10X Genomics also snapped up another Stockholm startup Cartana, which was co-founded by Kühnemund.

“If you analyze a sample with us, you get 100 genes,” Kühnemund said regarding Cartana’s technology. “It’s a lot cheaper than doing 30 different in situ hybridizations.”

There is a growing wave of European startups working on their own forms of spatial sequencing. The Swedish company Single Technologies closed a $10 million seed round in 2022 to bankroll a high-throughput method. Also in 2022, Resolve Biosciences bagged $70 million in a series B round to fund the development of spatial sequencing technology.

Outside of Europe, other players in the spatial sequencing sector include NanoString and Bruker Spatial Genomics as well as up-and-comers like Curio Bioscience in the U.S.

How is spatial transcriptomics changing the immuno-oncology field?

Most spatial transcriptomics technology is marketed for research purposes, and a number of developers are also targeting it for diagnostics in the future. One of the biggest applications is immuno-oncology.

Currently, spatial transcriptomics companies offer immunotherapy developers the ability to map out gene expression in tumor samples and discover new drug targets or biomarkers. However, this is only the start of what the field could achieve in the next ten years in immuno-oncology.

Another intriguing application could be as a diagnostic tool in personalized medicine. This is because each tumor has a unique microenvironment that can suppress immune cells, and block the effects of particular immunotherapies such as checkpoint inhibitors. Knowing more about the microenvironment, such as the checkpoint inhibitor targets on immune cells, could let physicians choose more appropriate treatments for patients.

Fundamentally, the immune system acts through interactions between cells and immuno-oncology generally seeks to modulate immune interactions, explained Neil Kennedy, chief commercial officer of Curio Bioscience.

“Spatial transcriptomics has been a boon to researchers in the immuno-oncology field, enabling them to analyze gene expression across the entire transcriptome within the context of tissue architecture, offering a powerful tool to explore the complexities of tumor-immune interactions within the microenvironment. This comprehensive data helps identify key pathways and interactions that might be missed with targeted ‘omics’ approaches, leading to enhanced insights and potential advancements in I-O testing and treatments,” said Kennedy.

Spatial transcriptomics: an emerging tool in neurology research

Another hot area for spatial transcriptomics is neurology research, a historically difficult field due to the complexity of the human brain.

“If you look at a group like Roche, they’re very focused on immuno-oncology now, but their roadmap for the next few decades is to get deeply into improving brain health and brain disease through therapeutics,” noted Terry. “But to do that, you need a much deeper understanding of how the human brain works today.”

Neurological conditions currently being studied by spatial transcriptomics companies include neurodegenerative diseases, where NanoString, for example, hunts for marker genes for these conditions. Furthermore, spatial transcriptomics can also help pharmaceutical clients to study drug targets in the brain.

“You can actually map out where the target is in the brain, for example, and not only the approximate location inside the organ, but also in what type of cells,” Kühnemund said. “You actually now know what type of cells you are targeting. That’s completely new information for them and it might help them to understand side effects and so on.”

Another potential application of the technology is viral infection. NanoString’s technology has been used to map out lung damage caused by COVID-19, which could lead to new therapies. Additionally, spatial sequencing could shed light on how HIV is able to attach to immune cells and enter a latent stage.

“That’s not well understood for most viruses – what environmental factors actually trigger that or how do they come back out of that,” Terry said.

The technique of spatial transcriptomics in some ways resembles the aims of single-cell sequencing, a type of sequencing that can provide much more detail about cells than bulk sequencing. While Terry questioned the hype of single-cell sequencing, Michael Schnall-Levin, founding scientist at 10X Genomics, believes that the two techniques are actually complementary.

“Single-cell sequencing is critical for defining the ‘what’ in a sample — what the different cell types and states are, how they are defined, and how they differ,” he explained. “Spatial transcriptomics is critical for defining the ‘where’ — where cells are located, how cells interact, and how cell states differ across a tissue.”

Spatial transcriptomics: not without its challenges

This is not to say that spatial transcriptomics is free from its share of hurdles today. The high dimensionality brings with it the challenge of tackling large datasets and the requirement of powerful visualization programs compatible with big data.

The high demand due to the versatile applicability of spatial transcriptomics also means researchers face constant pressure to improve cellular data resolution.

Spatial transcriptomics techniques are also yet to be optimized for high-throughput research methods, a hurdle that is essential to overcome before they can be effectively applied in large-scale drug discovery and development, Kerstedt pointed out.

Similar thoughts were echoed by Evan Macosko, academic co-founder of California-based spatial mapping company Curio Bioscience. “While most spatial transcriptomics methods – sequencing-based or imaging-based – have achieved theoretical resolutions lower than the size of a cell, cellular segmentation prevents the assembly of pure, single-cell transcriptomes.”

Macosko thinks that either computational strategies for segmentation must improve or moving in the direction of Curio’s Trekker technology could tackle this problem. This single-cell mapping kit platform utilizes the intrinsic segmentation offered by tissue dissociation to ensure single-cell resolution.

This also points to scalability being a concern. Scale is necessary to make complete maps of large (human) organs, and to perform properly powered case-control studies, explained Macosko.

“Most approaches are either too expensive or time-consuming to be routinely deployable in large tissue sections or in many tissue sections,” said Macosko.

Moreover, as many approaches are probe-based, which requires knowledge of the RNA sequence in advance of the experiment, “discovering unusual splice-forms, studying non-standard biological species, and combining with DNA barcoding approaches are particularly reliant upon untargeted, sequencing-based strategies,” Macosko commented.

The importance of the ‘right’ biospecimen in spatial transcriptomics

Despite such hurdles, spatial transcriptomics techniques in use today, along with the kits and workflows that enable them, are very comprehensive, asserted Priester. Key to this has been the use of high-quality biospecimens.

“The outcome of your research is only as good as the biospecimen you put in. Good quality samples are required to generate reliable results that reflect molecular reality,” elaborated Priester.

“Developing drugs or companion diagnostics on biased information collected in a non-optimal manner can lead to massive mis-investments for pharma firms while doing the scientific community and all of their hard work a huge disservice.”

Biospecimens that have been outside of the body for too long or otherwise modified can lose vital information. This is even more important in oncology, where tumor cells extracted from the body can still undergo morphological or physiological changes due to continued contact with oxygen. Such changes can mean that experimental results are altered: wrong pathways or incorrect biomarkers are picked up.

To control this, many academic and commercial biobanks today have standard operating procedures for sample fixation, processing, and storage. However, since many firms outsource biospecimen collection, sample variability may be introduced at the very first step of sample collection, irrespective of the protocol in the steps that follow.

The future of spatial transcriptomics and multi-omics biospecimen

With developments in digital technology and machine learning, spatial transcriptomics could serve to propel the shift toward personalized medicine further. In the near future, the information from these tools could potentially enhance diagnostic accuracy and enable optimal treatment decision-making suited to each individual patient.

“With each individual tumor being different, it becomes crucial to look in-depth with support from the latest techniques,” said Priester.

Nevertheless, the field is moving fast, and could one day be combined with artificial intelligence (AI) to improve diagnostics.

“I think it’s going to be tricky for pathologists to look at that data,” Kühnemund said. “I think this will go hand in hand with the digital pathology revolution where computers are doing the analysis and they spit out an answer. That’s a lot more precise than what any doctor could possibly do.”

As spatial transcriptomics branches out into more fields of medicine, the market is expanding fast. This could, in turn, make the technique more mainstream and better integrated into future healthcare.

“We believe there will be incredible technological leaps in the next ten years that will enable new biological understanding that’s completely impossible today,” Schnall-Levin said.

Meanwhile, Macosko has high hopes for the future of this technology.

“Spatial transcriptomics will instead be spatial genomics – a means of quantifying diverse biomolecules, within cells in tissues, including genomes, transcriptomes, and proteomes. The technology will become sufficiently easy, affordable, and scalable that it can be a standard means of tissue exploration. Instead of a histological stain, pathologists and biologists will have access to high-dimensional data from specimens for purposes of diagnosis and hypothesis generation.”

This article was originally published in May 2021 and has since been worked on by Sudha Sundaram and Roohi Mariam Peter.