An in vitro model of human hematopoiesis identifies a regulatory role for the aryl hydrocarbon receptor

Abstract

In vitro models to study simultaneous development of different human immune cells and hematopoietic lineages are lacking. We identified and characterized, using single-cell methods, an in vitro stromal cell-free culture system of human hematopoietic stem and progenitor cell (HSPC) differentiation that allows concurrent development of multiple immune cell lineages. The aryl hydrocarbon receptor (AHR) is a ligand-activated transcription factor influencing many biological processes in diverse cell types. Using this in vitro model, we found that AHR activation by the highly specific AHR ligand, 2,3,7,8-tetrachlorodibenzo-p-dioxin, drives differentiation of human umbilical cord blood-derived CD34+ HSPCs toward monocytes and granulocytes with a significant decrease in lymphoid and megakaryocyte lineage specification that may lead to reduced immune competence. To our knowledge, we also discovered for the first time, using single-cell modalities, that AHR activation decreased the expression of BCL11A and IRF8 in progenitor cells, which are critical genes involved in hematopoietic lineage specification processes at both transcriptomic and protein levels. Our in vitro model of hematopoiesis, coupled with single-cell tools, therefore allows for a better understanding of the role played by AHR in modulating hematopoietic differentiation.

Graphical abstract

Figure 1.

Single-cell transcriptomic analysis reveals development of multiple hematopoietic cell types in an in vitro model of human hematopoietic differentiation. (A) Experimental design: in vitro culture system of hematopoiesis from human cord blood HSPCs and schema of sample processing and data analysis. (B) Different cell types were identified based on expression of a suite of lineage-specific genes (refer to “Methods”). (C) Representation using UMAP of all cells from control group collected over the 28-day period. (D) Development of cell clusters over time is shown. (E) Proportion of cells of each cluster across time. (F) Percent of cells belonging to major cell types that develop from untreated HSPCs were identified using flow cytometry. Error bars show mean ± standard error of the mean from 2 independent experiments. HSC, hematopoietic stem cells; MPP, multipotent progenitors; cDC2, type 2 classical dendritic cells; pDC, plasmacytoid dendritic cells; Meg-Ery, megakaryocyte-erythroid progenitors.

Figure 2.

AHR activation by TCDD alters the development of several hematopoietic lineages. (A) Changes in cell clusters from vehicle and TCDD groups across the 28-day developmental period. Some of these changes are highlighted with red circles. (B) Percent of cells belonging to each scRNA-seq data associated cell cluster for Vehicle and TCDD-treated groups across the 28-day developmental period. (C) Representative UMAP plots of cells captured by flow cytometry on day 21, with major cell populations identified through expression of different lineage markers. Comparison of UMAP plots of an equal number of cells (30,000) from both vehicle and TCDD-treated groups; (D) Percent of cells belonging to major hematopoietic lineages captured using flow cytometry for vehicle and TCDD-treated groups across the 28-day developmental period. Data presented are composite of 6 independent experiments. Statistical significance of differences in percentage of cells between treatments at any time point was calculated using a 2-tailed paired t test. ∗P < .05; ∗∗P < .01; ∗∗∗∗P < .0001. Error bars show mean ± standard error of the mean. Refer to supplemental Figures 5 and 6 for gating strategy. cDC2, type 2 classical dendritic cells; CLP, common lymphoid progenitor.

Figure 2.

AHR activation by TCDD alters the development of several hematopoietic lineages. (A) Changes in cell clusters from vehicle and TCDD groups across the 28-day developmental period. Some of these changes are highlighted with red circles. (B) Percent of cells belonging to each scRNA-seq data associated cell cluster for Vehicle and TCDD-treated groups across the 28-day developmental period. (C) Representative UMAP plots of cells captured by flow cytometry on day 21, with major cell populations identified through expression of different lineage markers. Comparison of UMAP plots of an equal number of cells (30,000) from both vehicle and TCDD-treated groups; (D) Percent of cells belonging to major hematopoietic lineages captured using flow cytometry for vehicle and TCDD-treated groups across the 28-day developmental period. Data presented are composite of 6 independent experiments. Statistical significance of differences in percentage of cells between treatments at any time point was calculated using a 2-tailed paired t test. ∗P < .05; ∗∗P < .01; ∗∗∗∗P < .0001. Error bars show mean ± standard error of the mean. Refer to supplemental Figures 5 and 6 for gating strategy. cDC2, type 2 classical dendritic cells; CLP, common lymphoid progenitor.

Figure 3.

AHR activation suppresses critical genes and key transcription factors involved in B cell and dendritic cell development. (A) Dot plot of the average gene expression of TCDD-treatment induced differentially expressed genes in MPP (multipotent progenitor) cluster on days 7 and 14. All DEGs are associated with a Bonferroni adjusted P value <.05. (B) Density of cells along MPP to lymphoid cells trajectory and expression of the top 4 highly variable genes that are differentially expressed by TCDD treatment and are involved in the development of lymphoid cells. (C) Percent BCL11A protein–expressing cells in overall population from 5 independent experiments. (D) Percent BCL11A protein–expressing cells in CD10⁺ CD19^– cells of vehicle and TCDD-treated groups across days from 5 independent experiments. (E) Transcription factor activity of IRF8 (analyzed by SCENIC) in pDC (plasmacytoid dendritic cell) cluster. Statistical significance of differences in TF activity between treatments at any time point was calculated using a Wilcoxon rank sum test. ∗P < .05. (F) Percent IRF8 protein–expressing cells in overall population from 5 independent experiments. (C-D,F) Protein expression was measured using flow cytometry. Error bars show mean ± standard error of the mean. Statistical significance of differences in percentage of cells between treatments at any time point was calculated using a 2-tailed paired t test. ∗ P < .05, ∗∗ P < .01, ∗∗∗ P < .001.

Figure 4.

Single-cell transcriptomic analysis shows that the macrophage cluster cells that develop in presence of TCDD have a reduced M2 macrophage signature. (A) Expression of select TCDD-treatment induced differentially expressed genes in macrophages on days 21 and 28; ∗Bonferroni adjusted P value <.05. (B) Macrophage cluster was selected, and an M2 macrophage score was calculated based on expression of several genes (refer to “supplemental Methods”). M2 macrophage score within the macrophage cluster, and across time for vehicle and TCDD-treated groups is shown. Statistical significance of differences in M2 macrophage score between treatments at any time point was calculated using a Wilcoxon rank sum test. ∗P < .05. (C) TF activity (calculated with SCENIC) in the macrophage cluster for selected TFs across time is shown. Statistical significance of differences in TF activity between treatments at any time point was calculated using a Wilcoxon rank sum test. ∗P < .05. (D) GSEA using the Hallmark pathways database shows significantly enriched pathways in cells of the macrophage cluster owing to TCDD treatment at day 28. All enriched pathways have a Benjamini-Hochberg adjusted P value <.05. DOWN, downregulated pathways; NES, normalized enrichment score; UP, upregulated pathways.