A Single-Cell Roadmap of Lineage Bifurcation in Human ESC Models of Embryonic Brain Development - PubMed (original) (raw)

. 2017 Jan 5;20(1):120-134.

doi: 10.1016/j.stem.2016.09.011. Epub 2016 Oct 27.

John K Mich 1, Sherman Ku 1, Vilas Menon 1, Anne-Rachel Krostag 1, Refugio A Martinez 1, Leon Furchtgott 2, Heather Mulholland 1, Susan Bort 1, Margaret A Fuqua 1, Ben W Gregor 1, Rebecca D Hodge 1, Anu Jayabalu 1, Ryan C May 1, Samuel Melton 3, Angelique M Nelson 1, N Kiet Ngo 1, Nadiya V Shapovalova 1, Soraya I Shehata 1, Michael W Smith 1, Leah J Tait 1, Carol L Thompson 1, Elliot R Thomsen 1, Chaoyang Ye 1, Ian A Glass 4, Ajamete Kaykas 1, Shuyuan Yao 1, John W Phillips 1, Joshua S Grimley 5, Boaz P Levi 6, Yanling Wang 7, Sharad Ramanathan 8

Affiliations

A Single-Cell Roadmap of Lineage Bifurcation in Human ESC Models of Embryonic Brain Development

Zizhen Yao et al. Cell Stem Cell. 2017.

Abstract

During human brain development, multiple signaling pathways generate diverse cell types with varied regional identities. Here, we integrate single-cell RNA sequencing and clonal analyses to reveal lineage trees and molecular signals underlying early forebrain and mid/hindbrain cell differentiation from human embryonic stem cells (hESCs). Clustering single-cell transcriptomic data identified 41 distinct populations of progenitor, neuronal, and non-neural cells across our differentiation time course. Comparisons with primary mouse and human gene expression data demonstrated rostral and caudal progenitor and neuronal identities from early brain development. Bayesian analyses inferred a unified cell-type lineage tree that bifurcates between cortical and mid/hindbrain cell types. Two methods of clonal analyses confirmed these findings and further revealed the importance of Wnt/β-catenin signaling in controlling this lineage decision. Together, these findings provide a rich transcriptome-based lineage map for studying human brain development and modeling developmental disorders.

Keywords: human embryonic stem cells; lineage; neurogenesis; single cell RNA-seq.

Copyright © 2017 Elsevier Inc. All rights reserved.

PubMed Disclaimer

Figures

Figure 1

Figure 1. In vitro neural differentiation generates cortical and non-cortical cells

(A) Schematic representation of in vitro neural differentiation of hESCs. (B) Representative images of immunostaining on D12, D26 and D54 of H1 differentiated cells, with DAPI in blue. Scale bar: 100 μm. (C) Representative traces of calcium activity as imaged with FURA2-AM (top traces) and after blockade by TTX (bottom traces). Data quantified from three representative experiments (n = 1148 cells at D54, from 3 biological replicates) (bottom); RFU = relative fluorescent units. (D) Principal component analysis of population RNA-Seq data demonstrates the reproducibility of differentiation methods across multiple experiments from both H1 and H9 stem cell lines. (E) Population RNA-Seq expression of genes that mark indicated brain regions. (F) Schematic of targeted loci of the SOX2Cit/+ and DCXCit/Y reporter cell lines. (G) Quantitation of percent citrine positive cells during differentiation by flow cytometry. Mean ± SD is shown from 3 (SOX2Cit/+) and 6 (DCXCit/Y) biological replicates. See also Figures S1-2.

Figure 2

Figure 2. Identification of cell types through single-cell transcriptomics

(A) Single cell-profiling strategy and (B) methodology of cell type identification from CelSeq single cell RNA-Seq data. (C) Principal component analysis of all single cells used for analysis based on high variance genes (Table S1). (D) Representative heatmap showing most distinguishing genes at D54 used to identify cell types (top, Data S1), and bootstrapping analysis of D54 cells. (E) Violin plots for most distinguishing and commonly used marker genes. Max Log2(UMI + 1) values are to the right, and the number of cell per cell type is listed above the cell type name. (F) Constellation diagram of cell types identified at D12, D19, D26, D40, and D54. Cell type name, relative number of cells per time point sampled are shown by circle size, strength of intra- and inter-cell type clustering from bootstrapping analysis are indicated by circle boarder with and edge width, and POU3F2 and LHX2 expression status are shown by circle boarder color. Neuronal cell types (yellow) were defined as cell types with strong DCX expression, Progenitors with strong SOX2 expression (gray), Transitional types express both SOX2 and DCX (gray-yellow transition), and Other cell types (pink) express genes indicative of non-neuronal lineages. See also Figures S3-4, Table S1, and Data S1-2.

Figure 3

Figure 3. Validation of single-cell RNA-Seq data and creation of preplate cortical neurons

(A-D) Differentiating cultures were immunostained with the indicated cell type markers and types of progenitors (SOX2+ cells) and neurons (TUJ1+ cells) were quantified. n = 4-6 independent experiments per immunostaining condition, scale 50 μm. The quantification by immunostaining (left) is directly compared to single-cell gene expression data generated by CelSeq (middle) and SmartSeq2 (right). Cells were scored positive for CelSeq if their UMI was >4 and for SmartSeq2 if TPM>0. (E) Chromogenic in situ hybridization data show early co-expression of Tbr1, Reln, Eomes, and Lhx2 in E11.5-E13.5 mouse cortex, but no overlap of Tbr1 and Eomes after E15.5. Data are from Allen Developing Mouse Brain Atlas, scale is 100 μm. See also Figure S5, Table S5.

Figure 4

Figure 4. Stem cell-derived cell types resemble forebrain and mid/hindbrain cell types

(A) Spearman correlation of D54 neuronal cell types to E13.5 Allen Brain Atlas of the Developing Mouse Brain based on genes differentially expressed between cell types and tissue regions (Table S1). Mouse regional gene expression levels are derived from in situ hybridization staining intensity. (B) Correlation of single D54 neurons with regions of the human brain from the Brainspan Atlas of the Developing Human Brain. Spearman correlations are based on genes differentially expressed between cell types and tissue regions (Table S1). (C) Fluorescence micrographs of 122 dpc cortex and 132 dpc hindbrain. LHX2 marks human cortical but not hindbrain progenitors, while SOX2 marks progenitors in both regions. Scale is 100 μm. (D) Immunohistochemistry of cortical and hindbrain cell type markers. Top: Nissl stain and representation of tissue architecture are shown; below: tissue representation based on DAPI staining. VZ ventricular zone, OFL outer fiber layer, oSVZ outer subventricular zone, IZ intermediate zone, CP cortical plate, nuc medullary nuclei. The entire tissue section was scored and each dot represents a positive cell. Scale is 1 mm. Inset: fluorescence micrograph showing a representative image (location indicated by red box); scale is 25 μm. See also Table S1.

Figure 5

Figure 5. Comparison of single stem cell-derived forebrain cells to primary human single cells

(A) Flow cytometry plot showing primary cell populations that were sorted (n = 4) and profiled using FRISCR (n = 2). Mean ± SD of population percentage derived from four brains. (B) Expression of conserved and divergent gene modules between primary human cortical single cells and _in vitro_-differentiated progenitors and neurons at D26 and D54 (Data S1). For each block, progenitors are to the left, intermediate progenitors in the middle, and neurons to the right. (C) Results of gene ontology analysis of conserved and divergent gene expression modules. The top five most significant biological processes with a Bonferroni correction value <10-1 are shown. (D, E) Heatmaps show genes differentially expressed between D26 and D54 progenitors (D) and neurons (E) (Data S1). Column color bars are as in (B), row color bars show if expression is elevated at D26 (black) or D54 (gray). Black line shows genes with expression enriched in primary progenitors relative to primary neurons (D), or primary neurons relative to primary progenitors (E). See also Data S1.

Figure 6

Figure 6. A lineage tree from single cell transcriptomics that is modulated by canonical Wnt/beta-catenin signaling

(A) Example of one triplet of transcriptomic cell types showing strong evidence for an intermediate state (in blue). The non-intermediate states (red and green) express genes only along one of the two horizontal axes, whereas the intermediate state expresses both sets of genes and also expresses a set of marker genes (vertical axis) that is not highly expressed in either of the other two states. Axis values represent means of normalized gene expression over all the genes on a given axis. Transcriptomic types are named as in Figure 2. (B) In silico lineage tree assembled from triplets showing strong evidence of an intermediate state. Circles around groups of cell types indicate that they are not distinguishable in terms of lineage or progression using the tree-building algorithm. Transcriptomic types are named and colored as in Figure 2. Arrows indicate proposed lineage/progression links, and key asymmetrically regulated genes indicated by numbers next to the lineage arrow are listed in (C). (D) Differentiating cultures were treated with 2 μM XAV-939 or 1 μM CHIR-99021 during CI phase, then fixed and immunostained at D26 for LHX2 (green), POU3F2 (red), and SOX2 (not shown) to identify progenitors of each lineage branch. Scale 50 μm. (E) The proportion of SOX2+ progenitors costaining for LHX2 and/or POU3F2 were quantified. n = 3 independent experiments, >1000 cells counted per condition, * P < .05, ** P < .01, *** P < .001 by unpaired t-test relative to control. (F) Canonical Wnt/beta-catenin target genes AXIN2 and TNFRSF19 (but not pan-progenitor genes SOX3 or NCAM1) are disproportionally expressed by the minor progenitor cluster D12_P_S100A11 relative to the major cluster D12_P in the SmartSeq2 dataset. ** P < .01, ns not significant, by Fisher's exact test. See also Figure S5.

Figure 7

Figure 7. Clonal analysis confirms distinct POU3F2 and LHX2 branches of the human brain lineage tree

(A) Schematic of viral barcoding experiment, indicating time of infection and collection. (B) Representative fluorescence micrographs captured from live differentiating clones, exhibiting non-neuronal (Clone A) and neuronal morphologies (Clone B). Scale bar is 250 μm (Clone A) and 500 μm (Clone B). (C) All multicellular clones are shown as individual rows and each D54 cell type as columns. All single cell clones are combined in the bottom row (“singlets”). The size of each dot indicates the number of cells per cell type from a clone. Dots are colored as cell types in Figure 2E. (D) Clonal analysis of cell-autonomous lineage potential was performed by re-plating D26 progenitors at clonal density on feeder mouse astrocytes, then analyzing outgrown colonies for cell composition at D54 by immunostaining. (E and G) Example colonies are stained with antibodies for HNA + TuJ1 (blue), LHX2 (green), and either BCL11B (E, red) or TFAP2B (G, red). Colonies were grouped into categories of LHX2-containing and non-LHX2-containing. Scale bar, 50 μm. (F) Colonies that contain LHX2+ cells are more likely to contain BCL11B+ cells as compared to colonies lacking LHX2+ cells. (H) Colonies that contain LHX2+ cells are less likely to contain TFAP2B+ cells as compared to colonies lacking LHX2+ cells. In F and H, five independent differentiations were analyzed, and 25-38 colonies per immunostaining cocktail per experiment were inspected for the presence of cell types. *** P < .001, ** P < .01 by unpaired t-test. See also Figures S6-7, Table S2.

Similar articles

Cited by

References

    1. Bakken TE, Miller JA, Ding SL, Sunkin SM, Smith KA, Ng L, Szafer A, Dalley RA, Royall JJ, Lemon T, et al. A comprehensive transcriptional map of primate brain development. Nature. 2016;535:367–375. - PMC - PubMed
    1. Bertacchi M, Pandolfini L, Murenu E, Viegi A, Capsoni S, Cellerino A, Messina A, Casarosa S, Cremisi F. The positional identity of mouse ES cell-generated neurons is affected by BMP signaling. Cellular and molecular life sciences : CMLS. 2013;70:1095–1111. - PMC - PubMed
    1. Bulfone A, Martinez S, Marigo V, Campanella M, Basile A, Quaderi N, Gattuso C, Rubenstein JLR, Ballabio A. Expression pattern of the Tbr2 (Eomesodermin) gene during mouse and chick brain development. Mechanisms of Development. 1999;84:133–138. - PubMed
    1. Bystron I, Rakic P, Molnar Z, Blakemore C. The first neurons of the human cerebral cortex. Nature neuroscience. 2006;9:880–886. - PubMed
    1. Camp JG, Badsha F, Florio M, Kanton S, Gerber T, Wilsch-Brauninger M, Lewitus E, Sykes A, Hevers W, Lancaster M, et al. Human cerebral organoids recapitulate gene expression programs of fetal neocortex development. Proceedings of the National Academy of Sciences of the United States of America. 2015;112:15672–15677. - PMC - PubMed

Publication types

MeSH terms

Substances

LinkOut - more resources