Combining Spatial Biology and CRISPR to Fight Cancer

A new spatial functional genomics approach for studying tumors unites CRISPR-based genetic screens with multiplex imaging and spatial transcriptomics. The method, Perturb-map, offers scientists an unprecedented opportunity to test how specific genes control the architecture of a tumor, and how cancer cells regulate the tumor microenvironment (TME) to protect themselves from immune cell attack. This article explains how Perturb-map works and the promise it holds for advancing cancer research.

Perturb-map and the TME

In recent years, cancer researchers have gained an appreciation of the importance of the TME, comprising a heterogeneous mixture of cancer cells, stroma, and immune cells, as well as vasculature and extracellular factors such as collagens. Notably, multiple types of immune and stroma cells reside in the TME, with varied functions—some act to protect cancer cells, while others aim to eliminate them. The spatial architecture of cells within the TME affects many processes critical to cancer growth, metastasis, and response to therapies. Yet when tumor cells are studied individually, the spatially coded information of each cell type within the TME is lost upon tissue dissociation.

A research group led by Brian Brown, Director of the Icahn Genomics Institute, and Associate Director of the Marc and Jennifer Lipschultz Precision Immunology Institute at the Mount Sinai School of Medicine, has developed a method called Perturb-map to study how different genetic perturbations affect the TME. Perturb-map is an in vivo spatial functional genomics approach that combines the genetic editing power of CRISPR with multiplex imaging and spatial transcriptomics, which preserves the spatial architecture of tumors. With Perturb-map, researchers can study how extracellular influences on cells in the TME (e.g., secreted factors such as chemokines or interleukins) manifest in spatially coded information. With this new spatial functional genomics approach, we can learn more about the genes that cancer cells use to control the TME—evading attacks from immune cells and resisting cancer immunotherapies.

A protocol of many techniques

Brown characterizes the Perturb-map protocol as generally straightforward, particularly for those with previous multiplex imaging experience. The group used CRISPR gene editing to generate multiple knockouts (KOs) in parallel, transducing cancer cells with lentiviral vectors that express both a CRISPR guide RNA and a protein barcode (Pro-Code). They then injected the cells into mice to generate tumors in various tissues to model human cancers. Importantly, because the CRISPR gRNA is linked to a unique Pro-Code, they can create dozens or even 100s of different specific gene knockouts.

Brown’s group knocked out 35 different genes in parallel and then assessed tumors by a variety of parameters, including growth, histopathology, immune cell composition, and molecular state. To detect Pro-Code-tagged cells, they used multiplex immunohistochemistry consecutive staining on a single slide (MICSSS). According to Brown, the trickiest part is analyzing the imaging data, because tools in this nascent area are still developing. “This is not specific to Perturb-map—it’s a general challenge of the field right now, but part of what makes spatial biology so exciting,” he says. “We are in a new frontier of biology, and because everything is so novel, we can learn more than we did with older techniques.”

An innovative approach

Thanks to the Pro-Code tags, Brown’s group was able to examine the gene expression of gene-edited cells, in an innovative pairing of CRISPR and spatial transcriptomics. Using Perturb-map together with spatial transcriptomics enabled them to investigate how dozens of knockouts affect gene expression in tumors. “Perturb-map greatly amplifies the discovery power of this technology,” says Brown.

To accomplish this, Brown’s group turned to the Visium spatial transcriptomics platform from 10x Genomics. According to Jacob Stern, Director of Product Management at 10x Genomics, their Chromium and Visium spatial biology platforms often support innovative research like Perturb-map. “Our products are designed to streamline workflows at scale for high-quality data, making them good substrates for pairing with other innovations,” says Stern. “Our spatial technologies have been cited in over 300 publications and are behind breakthroughs in oncology, immunology, neuroscience, and more, fueling powerful discoveries that are transforming the world’s understanding of health and disease.”

By pairing Perturb-map with spatial transcriptomics, Brown’s group found that knocking out the SOCS1 and Tgfbr2 genes affected tumor growth in distinct ways, based on their modes of action in the TME. The SOCS1 gene codes for a negative regulator of interferon-gamma, while the Tgfbr2 gene codes for TGF-beta receptor 2. Knocking out each gene resulted in a tumor growth advantage.

Even more compelling for clinical applications is the effects of knockouts on the spatial architecture of the TME—and what this could mean for therapeutic efficacy. SOCS1-KO tumors were highly infiltrated by T cells, while Tgfbr2-KO tumors excluded T cells, forming a fibro-mucinous state within the TME. Clinically, this is significant because patients with tumors containing fewer T cells tend to show a poor response to immunotherapy drugs.

Some of the current immunotherapy drugs work by “turning on” T cells’ capacity to kill cancer cells. However, if T cells are largely absent within some tumors, so-called “pockets of resistance” to medication can persist. Brown’s group suggests a potential solution: to administer immunotherapies to “turn on” T cells concomitant with a SOCS1 inhibitor to enhance T cell infiltration. Indeed, the group showed that when animals were treated with the therapeutic antibody anti-PD-1, SOCS1-KO tumors were more responsive to treatment compared to controls.

Looking ahead

Brown’s group is already expanding on their published Perturb-map work, which used markers for 4–5 different immune cell types. “We are now working to be able to measure 30 different cell markers by using cyclic immunofluorescence (CyCIF),” Brown says. “This covers all major immune and stroma cell types, plus other types of measurements like different cell state markers.”

The Brown lab is applying Perturb-map to a number of different cancers besides lung cancer, including ovarian, pancreatic, and breast cancer. “Our collaborators in Cambridge, U.K. have been putting the Pro-Codes into T cells to track them by flow cytometry, and we are planning to try tracking T cells spatially by Perturb-map,” says Brown. Going forward, Perturb-map offers a powerful tool to identify gene controlling tumor architecture. “Perturb-map can provide a lot of information already, but we are working to super charge it with CyCIF,” he says. A deeper understanding of how cancer cells alter the TME in order to thrive and evade immunotherapies may hasten cancer’s end.