Robust hematopoietic specification requires the ubiquitous Sp1 and Sp3 transcription factors

Abstract

Background: Both tissue-specific and ubiquitously expressed transcription factors, such as Sp-family members, are required for correct development. However, the molecular details of how ubiquitous factors are involved in programming tissue-specific chromatin and thus participate in developmental processes are still unclear. We previously showed that embryonic stem cells lacking Sp1 DNA-binding activity (Sp1^ΔDBD/ΔDBD cells) are able to differentiate into early blood progenitors despite the inability of Sp1 to bind chromatin without its DNA-binding domain. However, gene expression during differentiation becomes progressively deregulated, and terminal differentiation is severely compromised.

Results: Here, we studied the cooperation of Sp1 with its closest paralogue Sp3 in hematopoietic development and demonstrate that Sp1 and Sp3 binding sites largely overlap. The complete absence of either Sp1 or Sp3 or the presence of the Sp1 DNA-binding mutant has only a minor effect on the pattern of distal accessible chromatin sites and their transcription factor binding motif content, suggesting that these mutations do not affect tissue-specific chromatin programming. Sp3 cooperates with Sp1^ΔDBD/ΔDBD to enable hematopoiesis, but is unable to do so in the complete absence of Sp1. Using single-cell gene expression analysis, we show that the lack of Sp1 DNA binding leads to a distortion of cell fate decision timing, indicating that stable chromatin binding of Sp1 is required to maintain robust differentiation trajectories.

Conclusions: Our findings highlight the essential contribution of ubiquitous factors such as Sp1 to blood cell development. In contrast to tissue-specific transcription factors which are required to direct specific cell fates, loss of Sp1 leads to a widespread deregulation in timing and coordination of differentiation trajectories during hematopoietic specification.

Keywords: ATAC-seq; Blood cell development; Differentiation trajectories; Embryonic development; Single-cell RNA-seq; Transcription factor cooperation.

Fig. 1

The complete absence or the truncation  of Sp1 do not cause widespread changes in chromatin accessibility. a Schematic representing in vitro differentiation of ESC. b FACS analysis of Flk1 expression in E14 WT and E14 Sp1^ΔDBD/ΔDBD cells derived from embryoid bodies (EB) at day 3.25 of in vitro differentiation. Left panel: representative FACS profiles for Flk1-PE staining, right panel: graph showing Flk1 expression and isotype controls for E14 WT and Sp1^ΔDBD/ΔDBD cells (n = 4, **indicates p < 0.001). c Graph showing the percentage of EB releasing macrophages in a macrophage release assay (n = 3, * indicates p < 0.05). d Rescue of Flk1 expression levels by re-expression of wild-type Sp1 in Sp1^−/− and Sp1^ΔDBD/ΔDBD cells (n = 3 for Sp1^−/− and n = 4 for Sp1^ΔDBD/ΔDBD, p values are indicated on the graph). e Pearson correlation plot of accessible chromatin regions in ESC as determined by ATAC-seq, in WT cells and Sp1 mutant ESC clones. f Heat maps showing the fold difference in accessible chromatin sites, as determined by ATAC, between WT and Sp1^ΔDBD/ΔDBD E14 ESC (left panel) and WT and Sp1^−/− A17Lox ESC (right panel). The red box indicates WT-specific ATAC sites, while the blue box indicates ATAC sites specific to either Sp1^ΔDBD/ΔDBD or Sp1^−/− cell lines

Fig. 2

Sp3 partially compensates for the absence of Sp1 in cells with disrupted Sp1 binding. a Average profiles of Sp1 ChIP enrichment in ESC (left panel) and Flk1+ cells for WT and CRISPR clones. b Heat maps showing a ranking of normalised tag counts from Sp1 ChIP-seq in A17Lox WT ES cells. Ranked alongside are tag counts from the Sp3 ChIP-Seq in the same cells and motifs for CTCF, NFY, YY1, ESRRB, NANOG, OCT4 and SOX2. Heat maps (public datasets, see Additional file 1: Table 1) showing ChIP-seq enrichment for CTCF, NFY, YY1, ESRRB, NANOG, OCT4 and SOX2. The histone modifications H3K9Ac, H3K27me3, H3K4me3 and H3K27Ac are shown alongside. c Heat maps showing fold difference in ChIP-seq enrichment between Sp1 and Sp3 ChIP in E14 WT ES cells. Ranked along the same coordinates are Sp1 and Sp3 ChIP tag counts in Sp1^ΔDBD/ΔDBD and Sp1^−/−  cells. Peaks were classed as specific if they showed equal to or more than twofold change in tag count difference between the Sp1 and Sp3 ChIPs. The specific and shared groups are indicated by coloured bars alongside (red indicates Sp1 specific peaks and blue indicates Sp3 specific peaks), and the number of peaks within each group is shown. d Heat maps showing fold difference in ChIP-seq tag count enrichment between Sp1 and Sp3 ChIP peaks in Flk1+ cells. Ranked along the same coordinates are Sp1 and Sp3 ChIP in Sp1^ΔDBD/ΔDBD  cells. Peaks were classed as specific if they showed more than twofold change in tag count difference between the Sp1 and Sp3 ChIPs. The specific and shared groups are indicated by coloured bars alongside (red indicates Sp1 specific peaks and blue indicates Sp3 specific peaks), and the number of peaks within each group is shown. e Zoomed-in heat map of the 1975 Sp1 specific sites in ESC shown in Fig. 2c. Right-hand panels show the Sp3 ChIP-seq enrichment ranked according to the changes in Sp3 occupancy at these sites in Sp1^ΔDBD/ΔDBD and Sp1^−/−  cells. The blue bar indicates peaks which are increased in the mutant cell lines compared to WT, and the green bar indicates peaks which are reduced at least twofold in the mutant cell lines compared to WT. f Zoomed-in heat map of 873 Sp1 specific sites in Flk1+ cells shown in Fig. 2d. Right-hand panel shows the Sp3 ChIP-seq enrichment ranked according to the changes in Sp3 occupancy at these sites. The blue bar indicates peaks which are increased in the mutant cell lines compared to WT, and the green bar indicates peaks which are reduced at least twofold in the mutant cell lines compared to WT. g Bar graphs indicating numbers of motifs within the 1975 Sp1 specific peaks in ESC. Separate graphs show all peaks, distal peaks and promoter peaks. The number of peaks is indicated above each bar. h Bar graphs indicating numbers of motifs within the 873 Sp1 specific peaks in Flk1+ cells. Separate graphs show all peaks, distal peaks and promoter peaks. The number of peaks is indicated above each bar

Fig. 3

Sp3^−/− ES cells exhibit a defect in hematopoietic differentiation. a Genome browser screenshots depicting examples of shared and unique binding sites. Top panel: Sp1 is an example of an Sp1/Sp3 shared promoter binding site; middle panel: Zswim6 is an example of an intragenic Sp1 specific peak (highlighted by the red box); bottom panel: Yod1 is an example of an Sp3 specific promoter peak (highlighted by the blue box). b Flk1 FACS staining on WT and Sp3^−/− cells at day 3.25 of EB differentiation (n = 5, **indicates p < 0.01). c Kit FACS staining of WT and Sp3^−/− cells at day 2 of blast culture (n = 4, *indicates p < 0.05). d Percentage cell populations from WT and Sp3^−/− cells at day 2 of blast culture (n = 4, n.s. indicates not statistically significant). e Percentage of macrophage-releasing EB in a macrophage release assay comparing WT and Sp3^−/− cells (n = 4, *indicates p < 0.05). f Number of primitive erythroid colonies in an EryP colony formation assay comparing WT and Sp3^−/− cells (n = 4, *indicates p < 0.05)

Fig. 4

Deregulated gene expression in Sp1^ΔDBD/ΔDBD and Sp3^−/− cell populations. a Hierarchical clustering of differentially regulated genes between WT, Sp1^ΔDBD/ΔDBD and Sp3^−/− cell populations from each stage of the differentiation series according to the row Z score. Data were separated into 13 clusters, and the number of the cluster is shown to the left of the heat map. Representative genes from each cluster are shown to the right of the heat map. b Box plots representing the Z score of gene expression of each individual cluster from (a). c Heat map representing the grouping analysis of differentially regulated genes for Sp1^ΔDBD/ΔDBD relative to WT and Sp3^−/− relative to WT in ESC. The coloured sidebar indicates the 8 clusters assigned according to the changes in gene expression. GO terms for selected groups are shown to the left of the heat map. The table beneath shows the number of deregulated genes in each group and the number of Sp1 and Sp3 target genes within each cluster. d Heat map representing the grouping analysis of differentially regulated genes for Sp1^ΔDBD/ΔDBD relative to WT in Flk1+ cells, and Sp3^−/− relative to WT in Flk1+ cells. The coloured sidebar indicates the 8 clusters assigned according to the changes in gene expression. GO terms for selected groups are shown to the left of the heat map. The table beneath shows the number of deregulated genes in each group and the number of Sp1 and Sp3 target genes within each cluster

Fig. 5

Identification of different cell populations in differentiating WT and Sp1^ΔDBD/ΔDBD cells. a t-SNE visualisation of the indicated cell populations within Sp1^ΔDBD/ΔDBD Kit+ sorted cells from day 2 of blast culture. The left panel shows WT cells; right panel shows Sp1^ΔDBD/ΔDBD cells. Each dot represents the transcriptome of a single cell, colour coded according to the assigned cellular identity as shown to the right of the panels. b Expression of selected marker genes superimposed onto the different clusters for WT and Sp1^ΔDBD/ΔDBD cells

Fig. 6

Sp1^ΔDBD/ΔDBD cells contain an additional megakaryocyte cluster as compared to WT cells. a t-SNE visualisation of combined WT and Sp1^ΔDBD/ΔDBD populations. Each dot represents a single cell. In the left panel, dots are coloured to represent the genotype as indicated. In the right panel, dots are coloured according to the cell identity as indicated. b The expression level of selected differentially expressed genes was superimposed onto the t-SNE plots shown in Fig. 6a. Gene names are indicated to the right of the panel. c Venn diagrams representing the overlap between Sp1 ChIP-seq targets in ESC (light green circles) or Flk1+ cells (dark green circles) with up-regulated genes (red circles) or down-regulated genes (blue circles) determined from single-cell differential gene expression analysis. The number of genes for each group is indicated

Fig. 7

Absence of functional Sp1 results in highly disordered differentiation trajectories. a Pseudotime trajectory plots of WT and Sp1^ΔDBD/ΔDBD Kit + sorted cells from day 2 of blast culture. Each dot represents the transcriptome of a single cell, colour coded according to the assigned cellular identity as shown to the right of the panels. b Expression patterns of selected TFs projected on the trajectory plots for WT and Sp1^ΔDBD/ΔDBD from day 2 of blast culture