Interferon-α signaling promotes embryonic HSC maturation

Abstract

In the developing mouse embryo, the first hematopoietic stem cells (HSCs) arise in the aorta-gonad-mesonephros (AGM) and mature as they transit through the fetal liver (FL). Compared with FL and adult HSCs, AGM HSCs have reduced repopulation potential in irradiated adult transplant recipients but mechanisms underlying this deficiency in AGM HSCs are poorly understood. By co-expression gene network analysis, we deduced that AGM HSCs show lower levels of interferon-α (IFN-α)/Jak-Stat1-associated gene expression than FL HSCs. Treatment of AGM HSCs with IFN-α enhanced long-term hematopoietic engraftment and donor chimerism. Conversely, IFN-α receptor-deficient AGMs (Ifnαr1(-/-)), had significantly reduced donor chimerism. We identify adenine-thymine-rich interactive domain-3a (Arid3a), a factor essential for FL and B lymphopoiesis, as a key transcriptional co-regulator of IFN-α/Stat1 signaling. Arid3a occupies the genomic loci of Stat1 as well as several IFN-α effector genes, acting to regulate their expression. Accordingly, Arid3a(-/-) AGM HSCs had significantly reduced transplant potential, which was rescued by IFN-α treatment. Our results implicate the inflammatory IFN-α/Jak-Stat pathway in the developmental maturation of embryonic HSCs, whose manipulation may lead to increased potency of reprogrammed HSCs for transplantation.

Figure 1

Screen for pathways corresponding to developmental HSC maturation reveals Jak-Stat1. (A) WGCNA. The horizontal bar represents all genes in the sample. Identified co-expressed genes (modules) are assigned individual colors in the first row. In subsequent rows, red indicates upregulation and blue indicates downregulation in each sample. The module of interest, which shows low expression in the AGM and higher expression in the FL and BM is indicated by an arrow. (B) Expression levels of genes in this module identified in (A) during embryonic development. (C) Top 5 gene ontology biological process terms identified from genes in this module suggesting inflammatory pathway activation. (D) Top 5 KEGG pathway terms identified from genes in this module. Jak-Stat pathways are shown in red. (E) GSEA of samples displayed in columns compared against AGM HSCs. Red indicates that the gene set is more highly enriched in the AGM compared with tissues indicated, whereas blue indicates that the gene set is less enriched. The color scale is determined by the absolute enrichment score from the GSEA. (F) Quantitative RT-PCR for IFN-α and IFN-γ transcripts in whole E11.5 AGM, and E11.5 and E13.5 FL; n = 4. IFNs are detected in the AGM at Ct values <30, confirming their presence. (G) Percent of E11.5 AGM (VE-cadherin⁺CD45⁺) and E13.5 FL HSCs (Lineage⁻Sca1⁺c-Kit⁺), which are positive for intracellular staining of phospho-Stat1 (pacific orange). Absolute MFI for phospho-Stat1 is quantified. Relative MFI was determined by dividing MFI of sample by that of the IgG isotype control; n = 5-7. Statistical significance: *P < .05; **P < .01; ***P < .001. Ct, cycle threshold; IgA, immunoglobulin A; mRNA, messenger RNA; YS, yolk sac.

Figure 2

IFN-α treatment promotes long-term engraftment of AGM HSCs. (A) Flow cytometry for IFN receptors Ifnγr1 (top) and Ifnαr1 (bottom), which are present in CD45⁺ cells of the AGM; n = 3. (B) Example of phospho-Stat1 activation in CD45⁺ E11.5 AGM cells treated with IFN-α (bottom) with isotype control (top). (C) Phospho-Stat1⁺ cells in the different cell compartments in the E11.5 AGM in response to IFN-α. The column indicates the absolute number of positive cells per embryo; n = 3. (D) Phospho-Stat3⁺ response in the different cell compartments in the E11.5 AGM in response to IFN-α. The column indicates the absolute number of positive cells per embryo; n = 3. (E) Phospho-Stat5⁺ response in the different cell compartments in the E11.5 AGM in response to IFN-α. The column indicates the absolute number of positive cells per embryo; n = 3. (F) Immunoblot for phospho-Stat1 in response to dose-titrations of IFN-α in adult splenocytes. (G) Example of donor chimerism analysis for multilineage engraftment. (H) Boxplots showing the effect of IFN-α on long-term hematopoietic engraftment of E11.5 AGM HSCs in the peripheral blood; 2 e.e. were transplanted with 2 × 10⁵ helper splenocytes. Two-way analysis of variance (ANOVA) was performed. (I) Boxplots showing the effect of IFN-α on long-term hematopoietic engraftment of AGM HSCs in the BM at 21 weeks posttransplantation. Wilcoxon rank-sum test was performed; n = 4-6. (J) Quantification of lineage contributions of B, T, and myeloid cells at 21 weeks posttransplantation in the peripheral blood; n = 4-6. (K) Boxplots showing engraftment from secondary transplantation of 2 × 10⁶ BM cells from IFN-α–treated and control AGMs in (H) with 3 × 10⁵ competitor BM cells. Two-way ANOVA was performed. Statistical significance: *P < .05; **P < .01; ***P < .001. n.s., not significant.

Figure 3

IFN-α enhances competitive transplant by enhancing HSC quiescence without affecting stem cell frequency or homing. (A) Limiting dilution assays with IFN-γ–treated, IFN-α–treated, untreated wild-type (WT), or Ifnαr1^−/− E11.5 AGMs; n = 13 for a total of 89 mice. (B) Homing assay for detection of CD45⁺GFP⁺Lin⁺ and CD45⁺GFP⁺Lin⁻ in the BM 18 hours posttransplantation; n = 5 (right). Representative flow cytometry plot showing gating for CD45 and GFP (left); n = 5. (C) Boxplots showing donor chimerism attributable to control or IFN-α treatment during AGM transplantation. (D) Boxplots showing engraftment from competitive transplant with IFN-α–treated, untreated, or Ifnαr1^−/− AGMs against 3 × 10⁵ competitor BM cells. Two-way ANOVA was performed. (E) MFI of MHC class I molecules detected on donor-derived hematopoietic cells 6 hours after IFN-α treatment (top) or in the periphery at 6 weeks posttransplantation (bottom); n = 3-6. (F) Cell-cycle analysis of the donor-derived (CD45.2⁺) LSK HSCs in the BM of transplantations involving IFN-α–treated or untreated AGMs. Samples were analyzed at 4 weeks posttransplant in the primary recipients (n = 3; middle) or 36 weeks posttransplant in the secondary recipients (n = 4-5; bottom) by flow cytometry with Pyronin Y and Hoescht 33342 (top); n = 4-5. (G) Cell-cycle analysis of the donor-derived (CD45.2⁺) LSK HSCs in the BM of transplantations involving IFN-α–treated or untreated AGMs. Samples were analyzed at 4 weeks posttransplant in the primary recipients (n = 3) or 36 weeks posttransplant in the secondary recipients (n = 4-5) by flow cytometry with Pyronin Y and Hoescht 33342. (H) Boxplots showing engraftment from competitive transplant with 5000 Sca1⁺ cells from E14.0 FL from WT, Ifnαr1^−/−, or Ifnγr1^−/− against 2 × 10⁵ competitor BM cells. Two-way ANOVA was performed. Statistical significance: *P < .05; ***P < .001. ns, not significant; SLAM, signaling lymphocyte activation molecules.

Figure 4

Microarray analysis shows partial maturation of IFN-α–treated AGM HSCs. (A) MDS shows 3 distinct groups of embryonic HSCs maturation, which correspond to AGM, “transition,” and FL signatures. IFN-α–treated AGM HSCs and subset of E12.5 FL HSCs are grouped as having this “transition” signature. Data from this study is depicted by open circles, whereas data from McKinney-Freeman et al (GSE37000) is depicted by closed circles. (B) Intersection of differentially expressed genes between E13.5-14.5 FL HSCs and AGM, and between AGM and IFN-α–treated AGM (nominal P value cutoff .05). (C) Top 5 KEGG pathway terms identified from FL upregulated and IFN-α–regulated gene set from (B). (D) Top 5 KEGG pathway terms identified from FL downregulated and IFN-α–regulated gene set from (B). MDS, multidimensional scaling.

Figure 5

Hematopoietic defect in Arid3a KO embryos is rescued by IFN-α. (A) Microarray analysis of transcription factors upregulated by IFN-α in vivo in adult HSCs reveals Arid3a (GSE14361). (B) Immunoblot of Arid3a in the AGM, YS, and PLA. (C) Percentage of Arid3a⁺ cells in the various VE-cadherin/CD45 compartments showing abundance of Arid3a in hematopoietic cells. Sorted cells were cytospun and stained for Arid3a; n = 4-5. (D) Microarray analysis showing the presence of Arid3a expression in AGM HSCs (VE-cadherin⁺CD45⁺) (GSE37000) via presence/absence call. Dotted line indicates the presence/absence cutoff. (E) Boxplot quantification of nuclear Runx1⁺ surrounding the dorsal aorta in the E10.5 AGM of Arid3a ^+/+ and ^−/− sections; n = 13-19. (F) Boxplot quantification of nuclear Runx1⁺ surrounding the dorsal aorta in the E11.5 AGM of Arid3a ^+/+, ^+/−, and ^−/− sections; n = 6-9. (G) CFU assays from E11.5 Arid3a WT and KO AGMs; n = 3. Error bars indicate standard error of the mean (SEM). (H) Boxplots of donor chimerisms of Arid3a ^+/+, ^+/−, and ^−/− E11.5 to E12.5 AGMs transplanted at 1 e.e. and analyzed at 6, 10, and 14 weeks posttransplantation; 5 × 10⁵ splenic helper cells were used. (I) Boxplots of donor chimerisms of E11.5 Arid3a ^+/+, ^+/−, and ^−/− AGMs transplanted at 0.5 e.e. and analyzed at 6, 12, and 16 weeks posttransplantation. Two-way ANOVA was performed. Statistical significance: *P < .05; **P < .01; ***P < .001. PLA, placenta; YS, yolk sac.

Figure 6

Genomic binding of ARID3A and STAT1. (A) Quantitative RT-PCR of Arid3a and IFN-related genes in Arid3a ^+/+, ^+/−, and ^−/− E11.5 AGM; n = 3-8. Error bars indicate SEM. (B) Immunostaining of E11.5 Arid3a WT and KO AGMs for phospho-Stat1. Scale bar = 100 μm. (C) ChIP-seq tracks ARID3A and STAT1 at the genomic loci of IFN-αR1 in K562 cells. (D) ChIP-seq tracks ARID3A and STAT1 at the genomic loci of IRF1 in K562 cells. (E) ChIP-seq tracks ARID3A and STAT1 at the genomic loci of STAT1 in K562 cells. (F) ChIP-seq data showing global overlap of binding sites between ARID3A and STAT1 but not the hormone receptor TR4. (G) Immunoprecipitation of ARID3A (left). PI IgG was used as a control. Co-immunoprecipitation of STAT1 with ARID3A (right). K562 cells were exposed for 90 to 120 minutes of IFN-α (0.5 ng/mL). (H) Confirmation of ChIP-seq via quantitative ChIP-PCR in K562 cells normalized by input control; n = 4. Error bars indicate SEM. (I) Confirmation of ChIP-seq via quantitative ChIP-PCR in CB CD34⁺ cells normalized by input control; n = 2. Statistical significance: *P < .05; **P < .01. ao, dorsal aorta; CB, cord blood; cv, cardinal vein; DAPI, 4′,6-diamidino-2-phenylindole; het, heterozygous; IB, immunoblot; IgG, immunoglobulin G; nc, notochord; ns, not signicant; PI, pre-immune; ur, urogenital ridge.

Figure 7

IFN-α response in Arid3a KO cells. (A) Immunoblot for phospho-Stat1 and Stat1 in Arid3a ^+/− and ^−/− AGMs treated with IFN-α. (B) Immunoblot confirmation of ARID3A KD in K562 cells via shRNAs. (C) Immunoblot for phospho-Stat1 and Stat1 in ARID3A KD cells showing response to IFN-α. (D) Representative model indicating the role of IFN-α during embryonic hematopoiesis. In contrast to IFN-γ, which promotes HSC emergence, IFN-α promotes partial maturation of AGM HSCs. Arid3a is a transcription co-regulator of IFN effector genes. When Arid3a is absent, inflammatory signaling via IFNs is dampened. Saturating the system with Stat1 via exogenous IFN-α treatment is able to overcome this defect. KD, knockdown; sh, short hairpin.