Spatial Transcriptomics Is Transforming Immuno-Oncology Research

Cancer has claimed more than nine million lives each year since 2015. The World Health Organization estimates that one in every four to five people worldwide is expected to develop cancer over their lifetime. While cancer mortality rate has decreased over time due to technological advances, cancer incidence is rising, highlighting a need to develop better anti-cancer therapy.

Cell-based immunotherapy such as chimeric antigen receptor T (CAR-T) cell therapy is a promising way to treat cancer. Immune cells, with cytotoxic CD8+ T cells being most popular so far, are genetically engineered to express chimeric antigen receptor proteins targeting cancer-associated antigens to eliminate cancer cells. Thus far, there are at least 500 registered clinical trials investigating the use of CAR-T cells for treating various cancer including melanoma and glioblastoma. 

Although CAR-T cell therapy has worked well for liquid cancer such as leukemia, it has limited efficacy against solid tumors. This is because solid tumors have a complex microenvironment with diverse composition of cells engaging in intercellular crosstalk. These cellular interactions can confer fitness to tumor cells via mechanisms like immuno-suppression to reduce the infiltration and metabolic activities of CD8+ T cells.

Transforming cancer research with spatial transcriptomics

The cancer microenvironment contains many novel cell types that are interacting over time and spatially enriched in different parts of a tumor. By characterizing these cell types and their crosstalk, researchers are hoping to identify new therapeutic targets for CAR-T cell therapy. 

Single cell RNA sequencing has accelerated the characterization of novel cell types, but the tissue dissociation step that occurs prior to sequencing leads to the loss of spatial information. On the other hand, in situ hybridization (ISH) is limited to a subset of the transcriptome and may not provide sufficient information to accurately classify cells. 

Spatial transcriptomics such as technology from 10x Genomics makes use of spatially barcoded oligonucleotide microarrays for unbiased mapping of transcripts over tissue sections. This technique is increasingly popular to decipher the complexity in a tumor microenvironment as it overcomes the limitation of low throughput in ISH and complements single cell RNA sequencing to map rare cell types within tumor tissues. In 2020, the spatial transcriptomics method was also chosen by Nature Methods for the “Method of the Year” award, joining other powerful techniques like optogenetics, light sheet microscopy, organoids, and single cell transcriptomics.

Spatial transcriptomics experiments can be performed in a few steps: (1) tissue preparation where cryopreserved tissues are sliced into 10-20 µm thick before being mounted onto the gene expression slide; (2) tissue permeabilization to release mRNA from cells that would bind to spatially barcoded oligonucleotides on the slide. Reverse transcription converts mRNA to cDNA by polymerase chain reaction. Researchers can also perform immuno-labeling of proteins to visualize co-localization of proteins and its corresponding mRNA; (3) barcoded cDNA is sequenced before being analyzed using tailored software and cross-referenced with datasets from various tissue atlases.

Understanding immune-cancer crosstalk in tumors

Spatial transcriptomics is a valuable tool for immuno-oncology research. Pancreatic cancer is one of the deadliest cancers, with a five-year survival rate of six percent. After surgery, most patients still suffer from cancer recurrence, with five-year survival rate only up to 25 percent. Unfortunately, many patients are unsuitable for post-surgery chemo and radiotherapy (CRT) due to high risk of morbidity. A team led by Aviv Regev, professor at the Massachusetts Institute of Technology, reported the use of single nucleus and spatial transcriptomics to understand the effects of neoadjuvant (before surgery) CRT to improve molecular subtyping of pancreatic cancer.

The team found that compared to untreated tumors, tumors treated with neoadjuvant CRT became more basal-like and had immune compartments that were distinctly different. For instance, treated tumors had lower proportion of B cells and regulatory T cells and higher proportion of CD4+ T cells and macrophages. The distinct immune infiltrates for basal-like and classical-like pancreatic cancer tissues may represent a way for more targeted therapeutic interventions. 

It was also found that post-CRT tumors contained only conventional type 1 dendritic cells that are known to activate cytotoxic lymphocytes for antitumor immunity. These data are consistent with pre-clinical and clinical reports that CRT can induce immunogenic cell death by improving tumor antigen availability and cross presentation of antigen by dendritic cells. 

In a recent study published in Cell, Andrew L. Ji and colleagues integrated single cell RNA sequencing with spatial transcriptomics and multiplexed ion beam imaging to characterize the immune landscape of cutaneous squamous cell carcinoma (cSCC). The team found that migrating dendritic cells upregulated IDO1, which is known to inhibit T cell cytotoxic activity and promote differentiation into regulatory T cells. Spatial transcriptomics also revealed that these cell types were predominantly enriched at the inflammatory leading edge of the tumor.

Improving cellular resolution

Despite the utility of spatial transcriptomics techniques, it is important to know that they can only sample RNA from tens of cells, and not single cells due to spatial resolution of about 100 µm (3-30 cells), and RNA capture efficiency onto the spatially resolved barcodes remains low (~5-10 percent). This means that rare cells might be missed or misclassified, which can complicate accurate molecular subtyping of cells.

To improve the spatial resolution, Vickovic et al. developed high-definition spatial transcriptomics by depositing barcoded poly(d)T oligonucleotides into 2 µm wells using a randomly ordered bead array-based fabrication process. They next demonstrated the clinical potential of their newly-developed technique using a tumor section from a histological grade 3 breast human epidermal growth factor receptor 2 (HER2+) cancer patient. The team showed that immune sub-populations could be identified within the tumor slice, and invasive cancer-specific areas were high in keratin 19 (KRT19) and Erb-B2 receptor tyrosine kinase 2 (ERBB2) genes that are implicated in cancer.

“The spatial transcriptomics technologies based on barcoded surfaces (beads or glass slides), i.e. with NGS sequencing as read out, strives to increase resolution and improve the sensitivity. The high-definition spatial transcriptomics paper shows that one can achieve sub-cellular resolution (2 micron) with barcoded beads. The field is still striving to increase that further and there is a recent bioRxiv publication from BGI that shows that 0.2 µm DNA balls can be used for spatial analysis,” says Joakim Lundeberg, professor at the KTH Royal Institute of Technology.

The same team continued with a study integrating the use of spatial transcriptomics with the latent Dirichlet approach to better characterize the molecular properties of HER2+ breast tumor from 10 patients, focusing on the types and infiltration of immune cells. They applied spatial cell scoring to reveal areas with immune cells with the goal to enhance cancer immunotherapy. For instance, they found that two deceased patients belonged to an immune score group with lack of immune cell infiltration in the invasive cancer region, and it is interesting to further investigate how immune-tumor crosstalk affect prognosis and overall survival with more patient samples to provide personalized cancer medicine.

“Spatial transcriptomics techniques such as Visium and HDST have already proven to provide new information in terms of cancer heterogeneity; for example, providing a higher granularity in the analysis of multifocal cancers (prostate cancer), immune infiltration (breast cancer), and description of the heterogeneity of the leading edge of the tumor (squamous cell cancer) that would not be evident if single cell RNA sequencing data wasn't coupled with spatial methods,” adds Lundeberg.

In 2019, Rodriques et al. developed the Slide-seq method that offers transcriptome-wide detection of RNA with 10 µm spatial resolution. To further improve RNA capture efficiency for more accurate analysis, the same team introduced Slide-seqV2, a work published in late 2020, which provides 50 percent RNA capture efficiency that is roughly tenfold better than Slide-seq, and nearly as sensitive as single cell RNA sequencing. The team accomplished this by improving library generation, bead synthesis, and array indexing. For instance, they added a second-strand synthesis step after reverse transcription to increase the numbers of cDNAs that could be amplified by polymerase chain reaction.

All in all, innovations in material fabrication, sequencing, and bioinformatics are rapidly improving spatial resolution and reducing the costs of spatial transcriptomics to transform immuno-oncology research. By providing an unprecedented way to characterize spatial locations of individual cells in a tumor and their cross-talks, spatial transcriptomics can help improve molecular subtyping of cancer and enhance cancer immunotherapy.