The mouse Runx1 +23 hematopoietic stem cell enhancer confers hematopoietic specificity to both Runx1 promoters

The transcription factor RUNX1 is a critical regulator of definitive hematopoiesis, and genomic aberrations of the gene encoding RUNX1 are frequently found in human acute leukemia.¹  In the mouse, Runx1 null mutations result in the absence of functional hematopoietic stem cells (HSCs) and definitive progenitors, leading to embryonic lethality.^2-6  During development, Runx1 is first expressed in the emerging hematopoietic system, including definitive HSCs.^7,8  Its highly regulated spatiotemporal expression pattern and pivotal role in HSC emergence prompted us to study its transcriptional regulation, to obtain insight into the molecular mechanisms underlying de novo HSC generation. We recently identified the Runx1 +23 hematopoietic enhancer, located 23.5 kb downstream of the ATG in exon 1.⁹  We showed that this +23 enhancer targets reporter gene expression, from a heterologous hsp68 core promoter, to the emerging HSCs and putative HSC-fated cells in the mouse embryo, and acts directly downstream of Gata2, SCL, and Ets transcription factors. Whether the +23 enhancer is equally active with the endogenous Runx1 promoters has not been assessed.

Runx1 is transcribed from 2 alternative promoters (Figure 1A), a distal P1 and proximal P2, with the P1 being specific to vertebrates.^10-13  Both the P1 and P2 promoters were reported to be transcriptionally active in the emerging hematopoietic system of the mouse embryo, at the stages of yolk sac (YS), aorta-gonad-mesonephros (AGM), and fetal liver (FL) hematopoiesis, with P1-derived transcripts particularly prevalent among enriched FL HSCs.^11,14,15  The P2 promoter was shown to be active in HSC-fated cells¹⁶  and to be critically required for FL hematopoiesis.¹⁷  In vitro transfection assays suggested that neither P1 nor P2 RUNX1 promoter elements harbored tissue-specific cis-regulatory elements.¹⁰  However, in vivo mouse promoter assays have not been reported, and it is therefore not clear to what extent cis-elements elsewhere in the locus are required to mediate the activity of the 2 Runx1 promoters.

Genomic fragments spanning the mouse Runx1 P1 and P2 promoters were generated by polymerase chain reaction (PCR) using the primers listed (Table S1, available on the Blood website; see the Supplemental Materials link at the top of the online article). Fragments were cloned upstream of the luciferase gene in pGL3 vectors (Promega, Southampton, United Kingdom) and upstream of the LacZ reporter, with or without the Runx1 +23 enhancer.⁹  Cloning strategy and maps are available on request. Mutagenesis of conserved Runx motifs in the P1 promoter was performed as described,⁹  using the primers listed in Table S1.

Luciferase assays were performed in the 416B myeloid progenitor cell line.⁹  Mouse F0 transgenic embryos were generated and analyzed as described.⁹  All mouse work was in accordance with United Kingdom Home Office regulations.

Tissues were harvested and chromatin immunoprecipitation (ChIP) performed as described.⁹ 

To establish whether mouse Runx1 promoter sequences have any cis-regulatory activity of their own, mouse P1 (526 bp) and P2 (205 bp) promoter fragments (Figure 1B) were cloned upstream of the LacZ reporter gene for analysis in F0 transgenic embryos. These elements spanned the areas of multispecies conservation around the mouse homologs of core promoter sequences defined in the human RUNX1 promoters¹⁰  and were confirmed to be active in vitro (Figure 1C). Analysis of P1 LacZ and P2 LacZ promoter-reporter constructs in F0 transgenic embryos (n = 10 and n = 20, respectively; Figure 1D) established that neither transgene had any consistent activity in specific hematopoietic, or nonhematopoietic,^7,18  sites of Runx1 expression.

That the mouse Runx1 promoter fragments lack tissue specificity in vivo confirms and extends the in vitro analysis of human P1 and P2 promoter fragments¹⁰  and corresponds to preliminary in vivo data using larger mouse promoter constructs (Ditsa Levanon and Yoram Groner, oral communication, 2006). It differs, however, from a recent report on zebrafish runx1 promoter fragments, showing that those do contain hematopoietic cis-regulatory activity in vivo, with the P2 marking the zebrafish equivalents of mouse AGM and FL.¹⁹  Interestingly, the zebrafish P2 core promoter contains multiple E-box and Gata motifs, whereas no E-box and only one Gata motif is present in the mouse P2 promoter (Figure S1). Because SCL and Gata2 are known to control Runx1 +23 hematopoietic enhancer activity through binding to E-box and/or Gata motifs,^9,20  the difference in zebrafish and mouse P2 promoter specificity may be explained by their differential binding of SCL and/or Gata2, with the zebrafish P2 showing sequence characteristics reminiscent of the +23 enhancer. Indeed, such a model is further supported by our observation that substantial SCL and Gata2 binding, although present at the Runx1 +23 enhancer in primary FL or 416B cells, was not found at the endogenous mouse P2 promoter (Figure 1E).

Next, we examined whether the Runx1 +23 hematopoietic enhancer maintains its specific cis-regulatory activity when acting with either of the Runx1 promoters (instead of the heterologous hsp68 promoter⁹ ), or whether there is any differential spatiotemporal specificity. For this, mouse P1 and P2 promoter fragments were cloned upstream of the LacZ +23 reporter-enhancer fragment, and the activity of the resulting constructs analyzed in F0 transgenic mouse embryos. In extensive analysis of both P1 LacZm +23 (n = 58) and P2 LacZ +23 (n = 44) F0 transgenic embryos, Xgal staining was reproducibly observed at sites of developmental hematopoiesis (Figure 2A,B), in proportions and a pattern similar to that seen in hsp68LacZ +23 F0 transgenics, ie, in the blood islands of the E8 YS, in hematopoietic clusters and a few endothelial and/or mesenchymal cells of the E10 dorsal aorta and vitelline/umbilical arteries, and in E12 FL cells.⁹  Thus, the Runx1 +23 enhancer confers hematopoietic specificity to both the Runx1 P1 and P2 promoters, and neither promoter contributes to, or significantly alters, the +23-specific spatiotemporal activity. Given that both promoters are active in mouse developmental hematopoiesis^11,14-17  and human stem and progenitor cells²¹  (Figure S2), we predict that P1/P2 promoters also interact with the +23 enhancer in the endogenous Runx1 locus during HSC emergence.

To begin to address the transcriptional regulation of the Runx1 promoters, we examined the relevance of 2 deeply conserved Runx motifs for P1 activity (Figure 2C). These Runx motifs are conserved down to frog²²  and are present in the P1 promoters of all 3 Runx genes,¹³  suggesting a deeply conserved role in positive or negative auto- and/or cross-regulation.²³  Indeed, the Runx1 P1 was demonstrated to be subject to a positive auto-regulatory and a negative cross-regulatory loop in myeloid and B-cell lines, respectively.^22,24  In mouse AGM hematopoiesis and during ES cell differentiation, the Runx1 P1 was suggested to be subject to auto-regulation.^7,14  Thus, we mutated the pair of deeply conserved Runx motifs and analyzed the activity of the resulting P1_mutLacZ +23 construct in the developing hematopoietic system of F0 transgenic embryos. Surprisingly, mutation of these motifs did not perceivably alter in vivo transgene activity, compared with the P1LacZ +23 construct, in the E10 YS, in the hematopoietic clusters of the dorsal aorta and vitelline/umbilical arteries, or in the E12 FL (Figure 2D,E). We concluded that the pair of conserved Runx motifs is not required for P1 promoter activity during mouse developmental hematopoiesis. Whether they play a role in other organ systems or in interactions with other Runx1 cis-elements remains to be established.

In conclusion, we firmly established that the Runx1 +23 HSC enhancer confers its hematopoietic specificity to both Runx1 P1 and P2 promoter fragments in vivo. Furthermore, we demonstrated that there is no role for the pair of conserved P1 Runx motifs in a long-proposed Runx1 auto-regulatory loop in developmental hematopoiesis. Finally, the observation that the P1 and P2 promoter fragments on their own lacked any reproducible Runx1-specific activity in vivo indicates that both the specific hematopoietic and nonhematopoietic expression of Runx1 critically relies on enhancer elements elsewhere in the locus. Future identification of cell- and/or stage-specific Runx1 hematopoietic enhancers will provide the basis for building gene-regulatory networks underlying HSC specification and differentiation.

The authors thank Bill Wood for valuable comments on the manuscript; Marieke von Lindern, Yoram Groner, and Ditsa Levanon for advice and fruitful discussions; past and present members of the de Bruijn group for help and discussions; Jackie Sharpe and Bill Wood for efficiently managing the mouse transgenics unit; and Catherine Porcher for the SCL antibody.

This work was supported by the Medical Research Council and Leukemia Research Fund, United Kingdom.

Contribution: T.B. performed the research, analyzed data, and drafted the paper; E.L.K.A., S.R.B.B., and A.J. performed research and analyzed data; P.-S.L. and J.S.-S. generated transgenics; B.G. performed the genomics analyses and critically commented on the paper; and M.F.T.R.d.B. designed the study, analyzed data, and wrote the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Marella F. T. R. de Bruijn, Medical Research Council Molecular Haematology Unit, Weatherall Institute of Molecular Medicine, John Radcliffe Hospital, Oxford, OX3 9DS, United Kingdom; e-mail: marella.debruijn@imm.ox.ac.uk.