NEW YORK – A large international team came up with single-cell and single-nucleus transcriptomic resources to characterize the diverse cell types found in the human body, along with related gene expression and splicing features that offer clues to processes at play within and across tissues and organs.
In a series of studies published in Science on Thursday, thousands of researchers from the Human Cell Atlas (HCA) initiative outlined findings from their cross-organ, single-cell transcriptome analyses, which resulted in a spatially informed cell atlas that spans some two dozen tissue or organ types.
"People often think of the genome as the blueprint of the organism, but that's not really correct. The genome is more of a parts list because every cell uses a different set of parts," Stanford University bioengineering researcher Stephen Quake, co-president of the Chan Zuckerberg Biohub Network, explained during an online press briefing on Tuesday.
"What we've been able to do … is help understand and interpret how different parts of the genome are used to define different cell types and create molecular definitions of all the different cell types in the tissues that we studied collectively in these papers," he explained.
For the first of the papers, Quake and colleagues with the Tabula Sapiens Consortium arm of HCA described efforts to sequence RNA in almost half a million individual cells spanning 24 tissue types using fluorescence-activated cell sorting and smart-seq2 or 10x Genomics microfluidic droplet capture and amplification approaches.
The collection included cells originating in 17 tissue samples from the same donor individual, the team explained, and in 14 tissue types from another individual, the researchers explained. They also assessed up to five tissues apiece from 13 more donor individuals with a range of ethnic backgrounds, medical histories, and ages.
"Tissue experts used a defined cell ontology terminology to annotate cell types consistently across the different tissues, leading to a total of 475 distinct cell types with reference transcriptome profiles," the authors noted. The full dataset, they added, can be explored online with the cellxgene tool through the Tabula Sapiens data portal.
Of 483,152 cells that passed quality control steps, the investigators noted that 264,824 were immune cells, while more than 104,100 epithelial cells, nearly 82,500 stromal cells, and almost 31,700 endothelial cells rounded out the collection. Within the T-cell sets, for example, they flagged clonal populations across tissues and saw B-cell mutation rates that varied depending on the tissue or organ considered.
The team also highlighted the gene expression and RNA splice variant similarities and differences within and between cells and tissue types, uncovering new splice variants along the way. The group also characterized distinct gut microbial community features found in different parts of the intestine, Quake explained, by sequencing individual microbes in situ at different spatial sites in the gut.
"The Human Cell Atlas is an open, scientist-led consortium that's mapping all the cell types in the healthy human body, providing an unprecedented resource for studying health and disease," Sarah Teichmann, co-chair of the HCA Organizing Committee and head of cellular genetics at the Welcome Sanger Institute, told reporters during Tuesday's briefing.
Teichmann, who led two of the new immune system-focused HCA studies, called the latest work "a real milestone towards creating a full draft human cell atlas" in terms of the tissues analyzed and the number of cell types identified.
"To understand the immune system, we need to look not only at immune cells in the blood, which is the most practical tissue to acquire for humans and has been the most commonly studied tissue," she explained, "but we also need to understand the tissues that create immune cells — the professional immune organs, the embryonic liver, bone marrow, and thymus — and the tissues where immune cells migrate to, such as barrier tissues of skin, gut, and lung, and where immune cells mature, such as lymph nodes and spleen."
In one of the papers, for example, Teichmann and her colleagues dug into data for developing immune cells sets, unearthing unexpected immune stem and progenitor cells in embryonic gut and skin barrier tissues and predicted signaling between T cells in the thymus that was validated with organoid models.
The team also considered tissue-specific and tissue-agnostic features for adult immune cells, along with related migration characteristics, using a new machine learning-based cell annotation tool called CellTypist to assess scRNA-seq and immune-specific recombination sequencing on some 360,000 cells plucked from 16 adult tissues originating in 12 donors.
"With CellTypist, we found that some families of immune cells such as macrophages have common signatures across tissues … whereas others such as memory T cells have different flavors depending on which tissues they reside in," Teichmann said, noting that specific findings from the immune studies "have implications for therapies that enhance or suppress an immune response to fight disease and for designing vaccines."
In another Science study, other members of the team focused on tissue-specific effects of rare or common variants implicated in monogenic neuromuscular, metabolic, or immune conditions or in complex diseases linked to common variants — analyses that relied on new analytical algorithms informed by single-nucleus RNA-seq on 209,126 nuclei isolated from frozen tissue samples across eight tissue types in 16 participants from the Genotype-Tissue Expression effort.
"There are many applications, but one of the main applications of this kind of map is to identify the cell types and the programs where our disease genes act," explained that study's senior author and HCA co-Chair Aviv Regev, a professor on leave at the Broad Institute and the Massachusetts Institute of Technology and head of Genentech Research and Early Development at Genentech/Roche.
Although cells across the body share the same DNA, the effects of disease-related variants are typically limited to specific tissue or organ types, she explained. Consequently, investigators hope to tap into the new single-cell or single-nucleus transcriptome sets and cell maps to better understand where and how diseases manifest in the body to improve related diagnostic and treatment options.
Among other findings, the team saw signs that muscular dystrophy involves genetic alterations in genes expressed in non-muscular cell types that impact broader muscle tissue function, for example; while genes implicated in a heart condition called atrial fibrillation also appeared to contribute to functions in skeletal muscle, esophageal tissue, and the prostate.
"We believe that this kind of work really lays a foundation for building and exploring more comprehensive atlases, by expanding to more tissue types and covering the whole body," Regev explained, "as well as giving a map to those who want to understand the genetic basis of disease, to move from discovering the gene to knowing where the gene acts, which is the first step that we need both for diagnostics and therapeutic development."
"Collectively, these pan-tissue studies bring us closer to building a comprehensive human single-cell atlas," Peking University researchers Zedao Liu and Zemin Zhang wrote in an accompanying perspectives article in Science. "Future analyses may include larger and diverse cohorts. It is also crucial to apply this approach in the setting of disease, such as cancer."