Human cerebral cortex is tremendously complex with distinct functional areas. These functionally discrete areas lie on a large cortical sheet that has diverse cell types, cytoarchitectures, and transcriptomic variations. Previously, lots of efforts were spentto deciphering the areal cellular organizations and their molecular signatures. Recent studies that use single-cell/nucleus RNA-sequencing to profile the whole transcriptomes of human cortical areas extensively categorize the distinct transcriptomic cell types in these areas but the spatial information was lost during cell/nucleus isolation. Thus, a spatial transcriptomic tool is required to further uncover the cytoarchitectures and their spatial transcriptomes.
Recently, many spatial transcriptomic tools have emerged but have different aspects of disadvantages such as low sensitivity, low spatial resolution, and low throughput. These limitations restrict their direct application to postmortem human braintissue that have large size, general low tissue quality, and high autofluorescence signals. Here, we developed Decoded Amplified taRgeted Transcripts with Fluorescent in situ Hybridization or DART-FISH, a versatile padlock probe-based technology capable of profiling hundreds of genes in centimeter-sized human tissue sections at cellular resolution. As proof of principle, we measured 121 genes in human primary motor cortex (M1C), benchmarked, and validated its reproducibility, sensitivity, and specificity.
In addition, we applied DART-FISH to measure spatial expression of 484 genes in 9 human cortical areas along the rostrocaudal axis. Meanwhile, SNARE-seq2 was used to profile the whole transcriptomes of isolated nuclei from these areas. With the alignment of DART-FISH cells with pre-annotated SNARE-seq2 nuclei, spatially resolved transcriptomic cell atlases of human cerebral cortex were generated. In line with previous studies in M1C, these cortical areas are all composed of 24 cell subclasses of excitatory neurons, inhibitory neurons, and non-neuronal cells. Also, the found the cytoarchitectures of these cortical areas are similar except the rostral most anterior cingulate cortex (ACC) and the caudal most primary visual cortex (V1C). V1C showed the most drastic specialization, with tremendous expansion of layer 4 and increase of L4 IT neurons. To sum up, our findings suggest that these cortical areas have subtle cytoarchitectural and transcriptomic variations, except those areas at the poles of the rostrocaudal axis.