A decade of molecular cell atlases

Abstract

The development of molecular cell atlases has been crucially dependent on technological advances, which can be traced back three decades.Important technologies include the development of single-cell cDNA libraries, whole-transcriptome analysis such as next-generation sequencing, and microfluidic tools that enable high-throughput cell processing.A vibrant international community has made important scientific contributions with the use of these tools to discover and characterize cell types over an extended period, starting around 2011–2012.Whole-organism cell atlases are now reaching fruition for a number of important species, including human, and they serve as phenotypic companions to the genome for these animals. In recent years there has been tremendous progress towards deep molecular characterization of cell types using single-cell transcriptome sequencing, creating so-called ‘cell atlases’. These atlases provide a basic understanding of how different cell types of the same organism – which all share the genome – make distinct use of subsets of genes from the genome to create a variety of distinct cell types across tissues with specialized functions. In this opinion article I discuss some of the history and technological innovations that led to the development of whole-organism atlases. In recent years there has been tremendous progress towards deep molecular characterization of cell types using single-cell transcriptome sequencing, creating so-called ‘cell atlases’. These atlases provide a basic understanding of how different cell types of the same organism – which all share the genome – make distinct use of subsets of genes from the genome to create a variety of distinct cell types across tissues with specialized functions. In this opinion article I discuss some of the history and technological innovations that led to the development of whole-organism atlases. Biology has a long history of creating cell atlases based on imaging morphology, both by optical microscopy [1.Ramón y Cajal, S. Histologie du système nerveux de l’homme & des vertébrés. Maloine, 1909Google Scholar] and by electron microscopy [2.Fawcett D.W. The cell, its organelles and inclusions. An atlas of fine structure. W.B. Saunders, 1967Google Scholar]. These approaches famously enabled the creation of a lineage map of all of the cells of the nematode Caenorhabditis elegans [3.Sulston J.E. Horvitz H.R. Post-embryonic cell lineages of the nematode, Caenorhabditis elegans.Dev. Biol. 1977; 56: 110-156Crossref PubMed Scopus (2442) Google Scholar], and in his Nobel Prize lecture Sidney Brenner proposed extending such an approach by defining cell type by gene expression and creating a ‘CellMap’ to understand all of the cell types in an organism [4.Brenner S. Sydney Brenner – Nobel Lecture: Nature’s gift to science. Nobel Prize Outreach, 2002Google Scholar]. In recent years, technological advances have enabled scientists to create deep molecular characterizations of cell types across many organs, tissues, and organisms. The generation of such molecular cell atlases has been intimately connected to the development of single-cell transcriptome technology, which itself dates back 30 years. The full history of these developments should be reviewed elsewhere, but in summary it began with visionary experiments to create amplified single-cell cDNA libraries, which were probed for the expression of individual genes with low-throughput methods. Soon thereafter, gene expression microarrays were used to obtain the first transcriptome-wide insights into single cell biology, and ultimately sequencers became the read-out method of choice for single-cell cDNA libraries (Box 1).Box 1The historical development of single-cell transcriptomics technologiesPioneering work done independently by Eberwine [44.Van Gelder R.N. et al.Amplified RNA synthesized from limited quantities of heterogeneous cDNA.Proc. Natl. Acad. Sci. U. S. A. 1990; 87: 1663-1667Crossref PubMed Scopus (1023) Google Scholar] and Iscove [45.Brady G. et al.Representative in vitro cDNA amplification from individual hematopoietic cells and colonies.Methods Mol. Cell. Biol. 1990; 2: 17-25Google Scholar] in 1990 showed how to make and amplify cDNA libraries from single cells, which were analyzed by probing small numbers of genes via Southern hybridization. This approach was adopted by others and used for a variety of discoveries, and in 2003 single-cell cDNA libraries were for the first time analyzed at scale by microarray hybridization, using both homemade arrays with hundreds [46.Chiang M.K. Melton D.A. Single-cell transcript analysis of pancreas development.Dev. Cell. 2003; 4: 383-393Abstract Full Text Full Text PDF PubMed Scopus (183) Google Scholar] and thousands [47.Kamme F. et al.Single-cell microarray analysis in hippocampus CA1: demonstration and validation of cellular heterogeneity.J. Neurosci. 2003; 23: 3607-3615Crossref PubMed Google Scholar] of genes and commercial whole-transcriptome arrays [48.Tietjen I. et al.Single-cell transcriptional analysis of neuronal progenitors.Neuron. 2003; 38: 161-175Abstract Full Text Full Text PDF PubMed Scopus (204) Google Scholar,49.Seshi B. et al.Multilineage gene expression in human bone marrow stromal cells as evidenced by single-cell microarray analysis.Blood Cells Mol. Dis. 2003; 31: 268-285Crossref PubMed Scopus (70) Google Scholar]. Eventually, single-cell libraries were analyzed by direct high-throughput sequencing, initially in 2006 by Sanger sequencing [50.Moroz L.L. et al.Neuronal transcriptome of Aplysia: neuronal compartments and circuitry.Cell. 2006; 127: 1453-1467Abstract Full Text Full Text PDF PubMed Scopus (255) Google Scholar] and then in 2009 by next-generation sequencing [51.Tang F. et al.mRNA-seq whole-transcriptome analysis of a single cell.Nat. Methods. 2009; 6: 377-382Crossref PubMed Scopus (1705) Google Scholar], with several advances in quick succession over the next few years [5.Wu A.R. et al.Quantitative assessment of single-cell RNA-sequencing methods.Nat. Methods. 2013; 11: 41-46Crossref PubMed Scopus (493) Google Scholar,16.Islam S. et al.Characterization of the single-cell transcriptional landscape by highly multiplex RNA-seq.Genome Res. 2011; 21: 1160-1167Crossref PubMed Scopus (563) Google Scholar,52.Ramsköld D. et al.Full-length mRNA-seq from single-cell levels of RNA and individual circulating tumor cells.Nat. Biotechnol. 2012; 30: 777-782Crossref PubMed Scopus (965) Google Scholar,53.Hashimshony T. et al.CEL-seq: single-cell RNA-seq by multiplexed linear amplification.Cell Rep. 2012; 2: 666-673Abstract Full Text Full Text PDF PubMed Scopus (706) Google Scholar]. There is also an important part of the history of single-cell transcriptome analysis revolving around the development of single-cell PCR, which is not discussed here due to space constraints. Similarly, the development of new algorithms and computational approaches was also a powerful enabler of the field as it now exists. Pioneering work done independently by Eberwine [44.Van Gelder R.N. et al.Amplified RNA synthesized from limited quantities of heterogeneous cDNA.Proc. Natl. Acad. Sci. U. S. A. 1990; 87: 1663-1667Crossref PubMed Scopus (1023) Google Scholar] and Iscove [45.Brady G. et al.Representative in vitro cDNA amplification from individual hematopoietic cells and colonies.Methods Mol. Cell. Biol. 1990; 2: 17-25Google Scholar] in 1990 showed how to make and amplify cDNA libraries from single cells, which were analyzed by probing small numbers of genes via Southern hybridization. This approach was adopted by others and used for a variety of discoveries, and in 2003 single-cell cDNA libraries were for the first time analyzed at scale by microarray hybridization, using both homemade arrays with hundreds [46.Chiang M.K. Melton D.A. Single-cell transcript analysis of pancreas development.Dev. Cell. 2003; 4: 383-393Abstract Full Text Full Text PDF PubMed Scopus (183) Google Scholar] and thousands [47.Kamme F. et al.Single-cell microarray analysis in hippocampus CA1: demonstration and validation of cellular heterogeneity.J. Neurosci. 2003; 23: 3607-3615Crossref PubMed Google Scholar] of genes and commercial whole-transcriptome arrays [48.Tietjen I. et al.Single-cell transcriptional analysis of neuronal progenitors.Neuron. 2003; 38: 161-175Abstract Full Text Full Text PDF PubMed Scopus (204) Google Scholar,49.Seshi B. et al.Multilineage gene expression in human bone marrow stromal cells as evidenced by single-cell microarray analysis.Blood Cells Mol. Dis. 2003; 31: 268-285Crossref PubMed Scopus (70) Google Scholar]. Eventually, single-cell libraries were analyzed by direct high-throughput sequencing, initially in 2006 by Sanger sequencing [50.Moroz L.L. et al.Neuronal transcriptome of Aplysia: neuronal compartments and circuitry.Cell. 2006; 127: 1453-1467Abstract Full Text Full Text PDF PubMed Scopus (255) Google Scholar] and then in 2009 by next-generation sequencing [51.Tang F. et al.mRNA-seq whole-transcriptome analysis of a single cell.Nat. Methods. 2009; 6: 377-382Crossref PubMed Scopus (1705) Google Scholar], with several advances in quick succession over the next few years [5.Wu A.R. et al.Quantitative assessment of single-cell RNA-sequencing methods.Nat. Methods. 2013; 11: 41-46Crossref PubMed Scopus (493) Google Scholar,16.Islam S. et al.Characterization of the single-cell transcriptional landscape by highly multiplex RNA-seq.Genome Res. 2011; 21: 1160-1167Crossref PubMed Scopus (563) Google Scholar,52.Ramsköld D. et al.Full-length mRNA-seq from single-cell levels of RNA and individual circulating tumor cells.Nat. Biotechnol. 2012; 30: 777-782Crossref PubMed Scopus (965) Google Scholar,53.Hashimshony T. et al.CEL-seq: single-cell RNA-seq by multiplexed linear amplification.Cell Rep. 2012; 2: 666-673Abstract Full Text Full Text PDF PubMed Scopus (706) Google Scholar]. There is also an important part of the history of single-cell transcriptome analysis revolving around the development of single-cell PCR, which is not discussed here due to space constraints. Similarly, the development of new algorithms and computational approaches was also a powerful enabler of the field as it now exists. As the first single-cell transcriptome proof-of-principle experiments emerged, the challenge became how to scale the techniques to large numbers of cells; the early work was manual and extremely labor intensive, and it became clear that new technology would be needed to increase throughput. I was convinced that microfluidic automation would offer the answer, and my group worked for several years on chips that would automate single cell cDNA library construction. These were eventually commercialized by a company I founded called Fluidigm, which created the first commercial system for single-cell transcriptomic automation at scale [5.Wu A.R. et al.Quantitative assessment of single-cell RNA-sequencing methods.Nat. Methods. 2013; 11: 41-46Crossref PubMed Scopus (493) Google Scholar]. Named the C1, it was used by hundreds of laboratories around the world and led to many beautiful discoveries, of which only a few are cited here [6.Treutlein B. et al.Reconstructing lineage hierarchies of the distal lung epithelium using single-cell RNA-seq.Nature. 2014; 509: 371-375Crossref PubMed Scopus (851) Google Scholar, 7.Shalek A.K. et al.Single cell RNA seq reveals dynamic paracrine control of cellular variation.Nature. 2014; 510: 363Crossref PubMed Scopus (565) Google Scholar, 8.Zeisel A. et al.Brain structure. Cell types in the mouse cortex and hippocampus revealed by single-cell RNA-seq.Science. 2015; 347: 1138-1142Crossref PubMed Scopus (1625) Google Scholar, 9.Pollen A.A. et al.Low-coverage single-cell mRNA sequencing reveals cellular heterogeneity and activated signaling pathways in developing cerebral cortex.Nat. Biotechnol. 2014; 32: 1053-1058Crossref PubMed Scopus (555) Google Scholar, 10.Pollen A.A. et al.Molecular identity of human outer radial glia during cortical development.Cell. 2015; 163: 55-67Abstract Full Text Full Text PDF PubMed Scopus (439) Google Scholar]. The C1 broke important new ground by making it easy to perform single-cell transcriptomics on 96 cells at a time, but people’s thirst for higher throughput was becoming unquenchable. Dave Weitz at Harvard led the way to the next generation of single-cell automation, via microfluidic droplets. A few years earlier, my student Todd Thorsen and I had discovered that we could produce microfluidic emulsions on chips and use them to capture and assay cells [11.Thorsen T. et al.Dynamic pattern formation in a vesicle-generating microfluidic device.Phys. Rev. Lett. 2001; 86: 4163-4166Crossref PubMed Scopus (1658) Google Scholar]. Dave took that technology and matured it, demonstrating a variety of interesting screening applications. He then teamed up in separate collaborations co-led by Steve McCarroll and Aviv Regev [12.Macosko E.Z. et al.Highly parallel genome-wide expression profiling of individual cells using nanoliter droplets.Cell. 2015; 161: 1202-1214Abstract Full Text Full Text PDF PubMed Scopus (3327) Google Scholar] and Marc Kirschner [13.Klein A.M. et al.Droplet barcoding for single-cell transcriptomics applied to embryonic stem cells.Cell. 2015; 161: 1187-1201Abstract Full Text Full Text PDF PubMed Scopus (1717) Google Scholar] to show that it was possible to trap cells and oligo-coated beads in the same droplet along with lysis buffer, so the cells would break open and their mRNA would be captured on the beads. Each bead had unique barcodes on the oligos that could be used to track the cellular identity through the rest of the process. This enabled nearly two orders of magnitude more cells to be analyzed per experiment, albeit at a significantly lower sensitivity. This general approach was successfully commercialized by the company 10× Genomics, whose products are now dominant in the field [14.Kulkarni A. et al.Beyond bulk: a review of single cell transcriptomics methodologies and applications.Curr. Opin. Biotechnol. 2019; 58: 129-136Crossref PubMed Scopus (117) Google Scholar]. Several other technologies are looking promising to increase throughput even further, such as cell capture in nanoliter wells, in situ cell barcoding, and various spatial approaches. This opinion article is entitled ‘A decade of cell atlases’ because 2011–2012 were seminal years in conceptualizing the notion of a transcriptomic cell atlas. In 2011, Mike Clarke and I published a paper using highly parallel single-cell PCR to create molecular definitions of the diverse cell types as well as tumors of the human colon [15.Dalerba P. et al.Single-cell dissection of transcriptional heterogeneity in human colon tumors.Nat. Biotechnol. 2011; 29: 1120-1127Crossref PubMed Scopus (517) Google Scholar]. In doing so, we discovered new progenitor cell types in the normal human crypts, and we characterized how the cell-type distribution changed in tumors and tumor xenografts; these were some of the earliest novel cell types to be discovered by single-cell transcriptomics. That year Sten Linnarsson also published one of the early papers using single-cell RNA-seq to characterize cell types, in which he concluded by suggesting ‘us[ing]…very large-scale single-cell transcriptional profiling to build a detailed map of naturally occurring cell types’ [16.Islam S. et al.Characterization of the single-cell transcriptional landscape by highly multiplex RNA-seq.Genome Res. 2011; 21: 1160-1167Crossref PubMed Scopus (563) Google Scholar]. In 2012, I wrote a grant application – in collaboration with Mike Clarke, Irv Weissman, and Tom Sudhof – entitled ‘A cell type atlas for mice and humans’ in which we proposed to fuse the single-cell transcriptomic technology developed as part of my research with the first-rate cell biology of the other team members to create a systematic atlas of important tissues and organs from mouse and human. The application was not funded, but we continued with the collaborative research that had inspired the proposal and continued to use single-cell approaches to characterize individual tissues from human and mouse. Others were clearly thinking along the same lines. Sarah Teichman discussed the notion of a human body map in with Mike at the Sanger that year and Aviv Regev the Cell at the to characterize in a large variety of cell In the Sanger their Cell and Sten Linnarsson and published an in which they using single-cell transcriptomics to characterize cell types across et al.Single-cell technologies whole-organism Rev. 2013; PubMed Scopus Google Scholar]. single-cell RNA-seq had been used to discover new cell types in the D.A. et parallel single-cell RNA-seq for of tissues into cell 2014; PubMed Scopus Google Scholar] by and also discovered novel cell types, and in the lung [6.Treutlein B. et al.Reconstructing lineage hierarchies of the distal lung epithelium using single-cell RNA-seq.Nature. 2014; 509: 371-375Crossref PubMed Scopus (851) Google Scholar] by and we discovered a new progenitor cell that year also more detailed the creation of a human cell atlas In Aviv Regev and Sarah Teichman a in with a to discuss the of a and various should be into an international and I the and Sten Linnarsson with from from a large and from the as well as several other including and The were all and we to a in a few which by a at in which more The group a detailed paper that the of is now called the A. et Cell Scopus Google Scholar], a group of from around the world have a approach with a to ultimately the separate into a single atlas. the there is scientific but several and have up to a variety of on separate organs, tissues, and whose all to the of the scientific community (Box by many has the cell atlas have to the development of cell atlases. this is not an it a of the of from a variety of that have to the was from to by a variety of including by the and by a first to microfluidic tools for single-cell transcriptomics and then to use them to characterize a variety of of my that enabled a of at a small scale the formation of the in and other members of the The a in single-cell analysis that from to which many The began creating a transcriptional atlas of cell types in by the in The the the and the were early of the starting in and The also began to in by the in The made a from 2003 to the to the creation of a human which in many important cell atlas and The the from to which led to early work in single-cell other are now to the cell atlas including the the and have to the development of cell atlases. this is not an it a of the of from a variety of that have to the was from to by a variety of including by the and by a first to microfluidic tools for single-cell transcriptomics and then to use them to characterize a variety of of my that enabled a of at a small scale the formation of the in and other members of the The a in single-cell analysis that from to which many The began creating a transcriptional atlas of cell types in by the in The the the and the were early of the starting in and The also began to in by the in The made a from 2003 to the to the creation of a human which in many important cell atlas and The the from to which led to early work in single-cell other are now to the cell atlas including the the and In parallel with I continued to for a large and by I had to from and that would enable the of the at In of that we the a new research with of over years. and I were of the with a both to enable by at and and to large that to research in and in my a cell atlas. That year also the of a led by which became an important of the and other various the we made two which in were The first was that we would an of the of a cell atlas – one that transcriptome that we as on the and he in three group and was in transcriptomics and on approaches to understanding the cell, which led to the et the of human cellular Scholar], on developing new of high-throughput optical which enable one to the cellular development of the organism B. et large imaging and single Scholar]. also developed collaborations with two that were cell atlases. was with the whose work making a atlas led them to more in cell The other was with the led by had a full decade creating all human had been on understanding are in cells and tissues et map of the human 2015; PubMed Scopus Google et map of the human PubMed Scopus Google Scholar], but they that they were in a to cell-type identity across a large number of human of their years on at the and which to the were also to and Sten Linnarsson for a the and the in The important we made at the was to whole-organism cell atlases individual organs This by as one we would be in using transcriptomics to characterize across tissues in I that this would be a not only to make a single-cell atlas of the mouse but also to use it to only a few mice were to work we developed a approach that enabled to tissues and organs from mouse. This as the were for and so The into a large collaborative many at and because we to in to with both the and the of the cell as a mouse was and with to their back to their laboratories for the et al.Single-cell transcriptomics of mouse organs a PubMed Scopus (851) Google Scholar] and et single-cell transcriptomic atlas tissues in the PubMed Scopus Google Scholar] were was lung for this and he also to create a cell atlas of the mouse from the same created the a transcriptomic cell atlas of mouse an Scholar]. the of these to the we to the to the same we had used in both mouse and which is to we to obtain more tissues from the same individual were in several that a from the to my to the that we were to build a with the that in and that the was to to The year the of and characterization of organs and tissues from mouse at single cell a Scholar], the first whole-organism cell also published a cell atlas of the elegans that year et single-cell transcriptional profiling of a PubMed Scopus Google Scholar]. The for whole-organism atlases the year in with of cell atlases of the et type atlas and lineage of a by single-cell PubMed Scopus Google et type transcriptome atlas for the PubMed Scopus Google Scholar], the A. et cell type and revealed by whole-organism single-cell Full Text Full Text PDF PubMed Scopus Google Scholar], and the et of gene expression in at single-cell Scopus Google Scholar, et al.Single-cell of gene expression and lineage in the PubMed Scopus Google Scholar, et al.Single-cell of during PubMed Scopus Google Scholar] as well as work on other mouse cell atlases by both D.A. et single-cell atlas of in Full Text Full Text PDF PubMed Scopus Google Scholar] and et the Cell by Full Text Full Text PDF PubMed Scopus Google Scholar]. there is not space to review them also published important work on cell a distinct of cell atlases one at a time et Cell of gene Scopus (117) Google Scholar, et of a human cell landscape at single-cell PubMed Scopus Google Scholar, S. et al.Single-cell transcriptome profiling of an Cell of Biol. 21: PubMed Scopus Google this is the that was by the et a Cell from the Full Text Full Text PDF PubMed Scopus Google Scholar]. In the its and across and tissues and The a single cell transcriptomic atlas of organs from individual human Scholar]. based on separate work from the G. et molecular to gene Scholar] and Sanger et cell analysis reveals and across the human Scholar] were and the published an important at the same time et type transcriptomics map of human Scopus Google Scholar]. these molecular atlases are to the genome of an As by Sydney Brenner more two the genome a but not provide on how various are used in different cell the various cell atlases now in to the he and provide molecular characterizations of which genes are used in the various cell types, a molecular of cell I the many to the work here and that due to space I could not that to the development of the I Sten Aviv and Sarah Teichman for both their of the development of the field and their on the are

Topics & Concepts