Integrating Single-Cell and Spatial Transcriptomics: An Exciting New Era

How the first complete cell census and atlas of a mammalian brain was built

Illustration of a brain map AI generated by DALL-E

In a huge collaborative effort, researchers from the BRAIN Initiative Cell Census Network (BICCN) have combined single-cell and spatial transcriptomics to catalog the type and location of cells present across all parts of the mouse brain in unprecedented detail. As reported in a series of articles published in Nature on 13th December 2023, the resulting high-resolution atlas reveals the hidden depths of cellular and regional diversity and spatial cell-cell interactions between distinct brain regions alongside novel cell-type marker combinations (1–3). However, these discoveries were only possible thanks to integrating data from different cutting-edge single-cell and spatial transcriptomic technologies, where each method elegantly complimented the other to harness the maximum amount of spatial and transcriptomic information possible. So, in the third installment of our spatial transcriptomic series, we explore how the integration of multimodal data types in this ambitious project was fundamental to the whole mouse brain atlas and how it creates opportunities for the future.

Single-cell transcriptomics provides the “what,” spatial transcriptomics the “where”

Previous efforts to generate brain cell atlases primarily relied on single-cell RNA sequencing (scRNA-seq) to define cell types (the “what”) thanks to the technology’s profiling depth and scalability (4, 5).

However, due to tissue homogenization necessary before scRNA-seq, studies often lacked any spatial information (the “where”) to place cell types in their native tissue context crucial to understanding cell microenvironments and cellular tissue architecture.

Where studies did integrate scRNA-seq and spatial transcriptomics, these often focused on particular brain areas, making accurate comparisons across brain regions difficult (6, 7).

A high-resolution spatial transcriptomic atlas of the mouse brain

Over six years and US$375 million in the making, researchers from the National Institute of Health-funded BICCN have now applied single-cell and spatial transcriptomics to the entire adult mouse brain to discover a staggering 5,322 different cell types and their precise locations (1–3).

The atlas reveals deep connections between a cell’s transcriptomic identity and location while offering clues into the evolution of the intricate interactions of different brain regions.

The mouse brain atlas also paves the way for future studies into the 200 billion cells contained in the human brain , a feat now deemed possible thanks to technological and analytic advances in integrating complex high-content data types.

Integrating spatial transcriptomics and single-cell RNA-seq

All studies contributing to the atlas integrated single-cell and spatial transcriptomics to some degree to overcome the technological limitations of each method and make groundbreaking discoveries (1–3).

For instance, Yao and colleagues integrated scRNA-seq with the imaging-based spatial transcriptomic method multiplexed error-robust fluorescence in situ hybridization (MERFISH) (1). They found striking diversity in the cell types present in different brain regions combined with a transcription factor ‘code’ that defines cell types across all parts of the brain with the same classification power as complete transcriptomes.

But, while MERFISH achieves spatial single-cell resolution, it can only detect the expression of hundreds to around a thousand pre-selected gene panels, potentially leading to the spatial misassignment of cell types or the absence of key marker genes. For biomarker detection or drug discovery studies, annotating cell types in incorrect locations or missing gene expression data could be crucial.

Spatial transcriptomic cell type inference with single-cell RNA-seq

So, to more accurately annotate the location of cell types and to infer broader gene expression profiles for MERFISH data, the researchers first performed scRNA-seq on microdissected brain regions with 10X Genomics Chromium technology.

Using data from approximately four million cells post-quality control, 8,460 cell type marker genes were established and the researchers used the 500 genes most predictive of cell type as probes in the MERFISH analysis. This panel was then used to profile 4.3 million cells with MERFISH at single-cell resolution across the entire mouse brain.

Next, bespoke machine learning algorithms, based on the k-nearest neighbor method, imputed the expression of the 8,460 marker genes using the 500 genes directly detected in each MERFISH-imaged cell. This imputation allowed the researchers to accurately detect and annotate over 5,000 different cell types while pinpointing their precise locations within the complex geography of the whole brain.

In a related study, Zhang and colleagues used the scRNA-seq data from Yao et al, to develop a different MERFISH panel of over 1,000 genes but increased the number of cells imaged to approximately 10 million (2). Combined with similar data analysis techniques, this increased spatial coverage drove the discovery of over 100 molecularly defined brain regions, including spatial expression gradients and cell-type-specific interactions.

Spatial transcriptomic atlases of the future

These studies elegantly demonstrate the power of integrating complementary data sets to make biological discoveries greater than the sum of their parts. However, large-scale or bespoke studies where vast multimodal data sets must be integrated to harness the power of the technologies and overcome their weaknesses will require advanced or novel data analytic approaches.

Scaling up to a 200 billion cell atlas of the human brain will be no mean feat, but powerful data analysis methods are increasingly poised for the task.

As experts in both single-cell and spatial transcriptomic data analyses, Nexco Analytics is dynamically positioned to perform comprehensive analyses or offer bespoke advice on any large-scale multimodal transcriptomic or multi-omic data integration projects, however complex.

Please contact us to find out how we can help you conquer your next data analysis challenge.

References

(1) Yao Z, van Velthoven CT, Kunst M, Zhang M, McMillen D, Lee C, Jung W, Goldy J, Abdelhak A, Aitken M, Baker K. A high-resolution transcriptomic and spatial atlas of cell types in the whole mouse brain. Nature. 2023 Dec 14;624(7991):317–32. Available at: https://doi.org/10.1038/s41586-021-03950-0.

(2) Zhang, M., Pan, X., Jung, W. et al. Molecularly defined and spatially resolved cell atlas of the whole mouse brain. Nature. 2023 Dec 14;624(7991):343–354. Available at: https://doi.org/10.1038/s41586-023-06808-9.

(3) Langlieb J, Sachdev NS, Balderrama KS, Nadaf NM, Raj M, Murray E, Webber JT, Vanderburg C, Gazestani V, Tward D, Mezias C. The molecular cytoarchitecture of the adult mouse brain. Nature. 2023 Dec 14;624(7991):333–42. Available at: https://doi.org/10.1038/s41586-023-06818-7.

(4) Kozareva V, Martin C, Osorno T, Rudolph S, Guo C, Vanderburg C, Nadaf N, Regev A, Regehr WG, Macosko E. A transcriptomic atlas of mouse cerebellar cortex comprehensively defines cell types. Nature. 2021 Oct 7;598(7879):214–9. Available at: https://doi.org/10.1038/s41586-021-03220-z.

(5) Yao Z, van Velthoven CT, Nguyen TN, Goldy J, Sedeno-Cortes AE, Baftizadeh F, Bertagnolli D, Casper T, Chiang M, Crichton K, Ding SL. A taxonomy of transcriptomic cell types across the isocortex and hippocampal formation. Cell. 2021 Jun 10;184(12):3222–41. Available at: https://doi.org/10.1016/j.cell.2021.04.021.

(6) Zeisel A, Hochgerner H, Lönnerberg P, Johnsson A, Memic F, Van Der Zwan J, Häring M, Braun E, Borm LE, La Manno G, Codeluppi S. Molecular architecture of the mouse nervous system. Cell. 2018 Aug 9;174(4):999–1014. Available at: https://doi.org/10.1016/j.cell.2018.06.021.

(7) BRAIN Initiative Cell Census Network (BICCN). A multimodal cell census and atlas of the mammalian primary motor cortex. Nature. 2021 Oct 7;598(7879):86–102. Available at: https://doi.org/10.1038/s41586-021-03950-0.

  • lundi 15 janv. 2024, 12:58
  • single-cell-sequencing, spatial-omics, bioinformatics, spatial-transcriptomics, transcriptomics
  • Share this post
Contactez-nous

Nos locaux:

Nexco Analytics Bâtiment Alanine, Startlab Route de la Corniche 5A 1066 Epalinges, Switzerland

Appelez-nous

+41 76 509 73 73     

Laissez-nous un message

contact@nexco.ch

N'hésitez pas à nous laisser un message

Nous nous ferons un plaisir de vous répondre et de trouver des solutions optimales à vos besoins.