Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2013 Apr;15(4):363-72.
doi: 10.1038/ncb2709. Epub 2013 Mar 24.

Characterization of transcriptional networks in blood stem and progenitor cells using high-throughput single-cell gene expression analysis

Victoria Moignard  1 , Iain C Macaulay, Gemma Swiers, Florian Buettner, Judith Schütte, Fernando J Calero-Nieto, Sarah Kinston, Anagha Joshi, Rebecca Hannah, Fabian J Theis, Sten Eirik Jacobsen, Marella F de Bruijn, Berthold Göttgens
Affiliations

Characterization of transcriptional networks in blood stem and progenitor cells using high-throughput single-cell gene expression analysis

Victoria Moignard et al. Nat Cell Biol. 2013 Apr.

Erratum in

  • Nat Cell Biol. 2013 May;15(5):544

Abstract

Cellular decision-making is mediated by a complex interplay of external stimuli with the intracellular environment, in particular transcription factor regulatory networks. Here we have determined the expression of a network of 18 key haematopoietic transcription factors in 597 single primary blood stem and progenitor cells isolated from mouse bone marrow. We demonstrate that different stem/progenitor populations are characterized by distinctive transcription factor expression states, and through comprehensive bioinformatic analysis reveal positively and negatively correlated transcription factor pairings, including previously unrecognized relationships between Gata2, Gfi1 and Gfi1b. Validation using transcriptional and transgenic assays confirmed direct regulatory interactions consistent with a regulatory triad in immature blood stem cells, where Gata2 may function to modulate cross-inhibition between Gfi1 and Gfi1b. Single-cell expression profiling therefore identifies network states and allows reconstruction of network hierarchies involved in controlling stem cell fate choices, and provides a blueprint for studying both normal development and human disease.

PubMed Disclaimer

Figures

Figure 1
Figure 1. Single cell gene expression analysis of a core haematopoietic transcriptional regulatory network
(a) Schematic of the haematopoietic hierarchy, with the megakaryocyte-erythroid lineage in red, the myeloid lineages in orange and the lymphoid lineage in blue. Cell types investigated in this study are outlined in the colours used to represent these populations in subsequent figures, and encompass both early multipotent stem and progenitors and committed progenitors for each of the major haematopoietic lineages. Cell surface phenotypes were LSK CD150+CD48− HSC (also gated as CD34loFlt3−), LSK Flt3hi LMPP, Lin−IL7Rα+KitloSca-1lo CLP, CD41loCD16/32hi GMP (also gated Lin−c-Kit+CD150−), CD16/32loCD41−CD150+CD105lo PreMegE (also gated Lin−c-Kit+). LT-HSC, long-term haematopoietic stem cell; MPP, multi-potent progenitor; LMPP, lymphoid-primed multi-potent progenitor; CMP, common myeloid progenitor; CLP, common lymphoid progenitor; GMP, granulocyte-monocyte progenitor; PreMegE, pre megakaryocyte erythroid progenitor; NK cell, natural killer cell. (b) Network diagram of data curated from the literature and protein interaction databases (STRING and FunctionalNet) illustrating the complex interactions between 18 core haematopoietic transcription factors. Green lines indicate functional relationships and red lines indicate direct protein-protein interactions. Activating and inhibitory connections are not distinguished.
Figure 2
Figure 2. Haematopoietic transcription factors show heterogeneous expression in haematopoietic stem and progenitor cells
Density plots for 18 transcription factors, the stem cell factor receptor c-Kit, and the housekeeping gene Ubc, in five haematopoietic stem and progenitor populations. The density indicates the fraction of cells at each expression level, allowing direct comparison of the expression level of each gene in all five populations. Green, HSC; Blue, LMPP; Purple, CLP; Red, GMP; Orange, PreMegE.
Figure 3
Figure 3. Single cell gene expression analysis reveals cell type-specific regulatory codes
(a) Hierarchical clustering of 597 haematopoietic stem and progenitor cells according to the expression of the 18 TFs. Coloured bar indicates cell type of origin: Green, HSC; Blue, LMPP; Purple, CLP; Red, GMP; Orange, PreMegE. (b) Principal component projections of the 597 haematopoietic stem/progenitor cells, in the first and second components (top), from the expression of all 18 TFs. Principal component loadings (bottom) indicate the extent to which each gene contributes to the separation of cells along each component. (c) Gaussian process latent variable model (GPLVM) illustrating variations of 18 dimensional gene expression patterns between and within cell types in 2D. GPLVMs are a non-linear generalisation of PCA that allow for the analysis of more complex gene expression patterns than PCA and can thus potentially better represent variations between and within populations of different cell types. The uncertainty of the mapping from 2D to the 18 dimensional TF space is encoded in grey (white low uncertainty, grey high uncertainty). (d) Relevance map showing the most important genes across the GPLVM map. The colours correspond to the distance of the respective gene from the origin in a standard loadings plot (red far away/important, blue close to origin). (e) Expression maps for Gata2 (left) and Gfi1 (right), with high expression in red and low or absent expression in blue. Gata2 is expressed primarily in the HSC and PreMegE clusters, and Gfi1 in LMPPs, GMPs and some CLPs.
Figure 4
Figure 4. Single cell expression analysis of haematopoietic TFs identifies previously unrecognised putative regulatory interactions between key TFs
(a) Hierarchical clustering of Spearman Rank correlations between pairs of TFs for all 597 cells together and for the different cell types individually as indicated. Genes in all heatmaps are ordered according to the clustering performed for all data. Positive correlations (red) may result from the coordinate expression or lack of expression of pairs of factors in individual cells, while negative correlations (blue) can result either from the expression of one factor in the absence of the other, or from high expression of one factor and reciprocal low expression of the other in the same cell. (b) Network diagrams showing putative activating relationships between TFs suggested by significant positive correlations (top, red lines) and antagonistic relationships suggested by significant negative correlations (bottom, blue lines) in the whole data set. Known relationships are highlighted with bold lines. This highlights a putative transcription factor triad in which Gfi1 is negatively correlated with Gata2 and Gfi1b, but Gata2 and Gfi1b are positively correlated.
Figure 5
Figure 5. Direct repression of Gata2 by GFI1 through a distal enhancer element provides a mechanism for negatively correlated expression
(a) ChIP-seq analysis of Gfi1 in primary mast cells indicates that GFI1 binds to the Gata2 locus at the −83 kb regulatory element. (b) Representative embryos demonstrating LacZ staining for the Gata2 −83kb, −3kb and combined −3/−83kb regulatory element reporter constructs. The −83kb region alone (SV/LacZ/−83) showed consistent staining of the midbrain, hindbrain and spinal cord, but no haematopoietic staining. The −3kb enhancer (−3/SV/LacZ) had only hindbrain staining. The −83/−3kb combined element (−3/SV/LacZ/−83) showed the neural activities of both individual enhancers, but also staining in the dorsal aorta (right-hand top panel) and foetal liver haematopoietic cells (right-hand bottom panel). Images of sections taken at 40× magnification. (c) A luciferase reporter construct carrying both regulatory elements (−3/SV/luc/−83) was transfected into the HPC7 haematopoietic progenitor cell line, which expresses high levels of Gata2 but low levels of Gfi1. Co-transfection with a Gfi1 expression construct caused a 40% reduction in reporter activity. Luciferase activity is shown relative to −3/SV/luc/−83 and bars are the mean and standard deviation of three biological replicates.
Figure 6
Figure 6. Direct activation of Gfi1b by GATA2 through distal enhancer elements
(a) ChIP-seq analysis of GATA2 in primary mast cells indicates that GATA2 binds to the Gfi1b locus at multiple locations, including the promoter, a region in the first intron and the +13, +16 and +17 kb regulatory elements. (b) Representative embryos demonstrating LacZ staining for the Gfi1b promoter as well as the +13, +16 and +17 kb regulatory elements. The promoter shows no staining. The +13 kb region shows weak expression in a subset of circulating blood cells. The +16 kb region has strong staining in haematopoietic clusters in the dorsal aorta, and in a subset of foetal liver cells. The +17 kb region shows staining in a small subset of foetal liver haematopoietic cells. Images of sections taken at 40× magnification. (c) Luciferase reporter constructs carrying the wild type +16kb (SV/luc/+16 WT) and +17 kb (SV/luc/+17 WT) kb regulatory regions transfected in 416B cells displayed high levels of luciferase activity, particularly for the +16 kb region. Mutation of the two conserved GATA sites in the +16 kb region (SV/luc/+16 GATA m12) reduced luciferase activity by >95%. Mutation of the two conserved and one partially conserved GATA sites in the +17 kb region (SV/luc/+17 GATA m123) also reduced luciferase activity by >95%. m12 and m123 indicate that GATA sites 1, 2 and 3 were mutated. Luciferase activity is shown relative to SV/luc. Experiments were performed in biological duplicate or triplicate on two separate occasions. Shown is one representative experiment displaying the mean and standard deviation for three biological replicate transfections. (d) A putative regulatory triad including GATA2, GFI1 and GFI1B suggested by the data. In this regulatory triad, GFI1 and GFI1B are mutually inhibitory, while GATA2 can activate expression of Gfi1b and GFI1 can repress expression of Gata2.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

See all "Cited by" articles

References

    1. Orkin SH, Zon LI. Hematopoiesis: an evolving paradigm for stem cell biology. Cell. 2008;132:631–644. - PMC - PubMed
    1. Ottersbach K, Smith A, Wood A, Gottgens B. Ontogeny of haematopoiesis: recent advances and open questions. Br J Haematol. 2010;148:343–355. - PubMed
    1. Takahashi K, Yamanaka S. Induction of pluripotent stem cells from mouse embryonic and adult fibroblast cultures by defined factors. Cell. 2006;126:663–676. - PubMed
    1. Davis RL, Weintraub H, Lassar AB. Expression of a single transfected cDNA converts fibroblasts to myoblasts. Cell. 1987;51:987–1000. - PubMed
    1. Gering M, Yamada Y, Rabbitts TH, Patient RK. Lmo2 and Scl/Tal1 convert non-axial mesoderm into haemangioblasts which differentiate into endothelial cells in the absence of Gata1. Development. 2003;130:6187–6199. - PubMed

Publication types

MeSH terms