CDX2 regulates ACE expression in blood development and leukemia cells

During vertebrate, definitive hematopoietic stem cells (HSCs) are first generated in the aorta-gonad-mesonephros (AGM) region of the posterior embryo.¹  We previously showed that the angiotensin-converting enzyme (ACE) is a cell-surface marker of human HSCs in both adult bone marrow (BM)²  and all hematopoietic niches developmentduring human ontogeny.³  In the embryo, ACE identifies HSCs associated with the ventral side of the aortic endothelium (in the AGM region) and of vitelline and umbilical arteries, both in humans³  and mice.⁴  ACE is a key component of the renin-angiotensin system (RAS), catalyzing the production of angiotensin II (AngII),⁵  which we have recently shown to be involved in hematopoietic emergence during ontogeny.⁴  ACE overexpression, leading to AngII increase, has also been reported in the BM of patients with acute myeloid leukemia (AML), influencing malignancy^6,7 ; however, its mechanisms of regulation are still unknown.⁸  The caudal-related homeobox gene 2 (CDX2) encodes an important transcription factor involved in tissue expansion and patterning of the posterior embryo.^9,10  At midgestation, it becomes limited to the presumptive mid-/hindgut endoderm and maintains selectively in the adult intestinal epithelium¹¹  where it has tumor-suppressor activity.^12-14  Conversely, although CDX2 is not expressed in adult HSCs, ectopic expression occurs in 80% of acute leukemia, which is considered a driver of leukemogenesis.^15-17  Here, we investigated the ACE and CDX2 expression patterns during human and mouse hematopoietic development, as well as in AML, and we conducted functional assays supporting that CDX2 participates in ACE regulation.

Human embryos were obtained from voluntary abortions and staged using anatomic criteria and the Carnegie classification as described.¹⁸  Umbilical cord blood was collected from healthy newborns. BM of AML patients (n = 29) and healthy subjects (n = 6) was collected after obtaining written consent according to the Declaration of Helsinki and the guidelines of the French National Ethics Committee. C57BL/6 mice were provided by Charles River and housed in a certified animal facility (number H-67-482-21). Murine embryos were staged by somite counting. Para-aortic splanchnopleura (P-Sp) or AGM regions were dissected as described.¹⁹ 

Additional methods are described in supplemental Material and methods.

Although CDX2 is crucial for the development of the posterior embryonic structures and ACE expression also emerges in the posterior region during hematopoietic ontogenesis, the first aim of this study was to perform a comparative study of CDX2 and ACE expression patterns in both human and mouse embryos. At early stages of murine development (embryonic day 9 [E9]), both proteins were synthetized in the extraembryonic YS endoderm and in the intraembryonic P-Sp (including the gut endoderm): the region endowed with hematopoietic ability^20,21  (supplemental Figure 1). Because our previous observations indicated ACE as a marker of HSCs in the AGM region,^3,4  we focused on CDX2 at this site. By reverse transcription–polymerase chain reaction (RT-PCR), CDX2 transcripts were detected in the AGM region that also expressed ACE (supplemental Figure 2). Indeed, CDX2 protein was found in intra-aortic hematopoietic clusters (IAHCs), also positive for ACE, in 33- to 36-day human embryos and in E10.5 mouse embryos (Figure 1A). This observation was strengthened by RNAscope in situ hybridization, which localized CDX2 transcripts in IAHCs, in associated endothelial cells (hemogenic endothelium), as well as in scattered mesodermal cells (Figure 1B). To further document the spatial evolution of ACE and CDX2 messenger RNA (mRNA) patterns in this blood-forming site at early stages of mouse development (E10.5-E11.5), mesenchymal cells (MCs), endothelial cells (ECs), and IAHC cells were sorted from the AGM region and processed for quantitative RT-PCR (RT-qPCR) analyses (Figure 1C-D). At E10.5, CDX2 and ACE transcripts were expressed in all 3 cell populations. Compared with the bulk of AGM cells, the proportion of CDX2 transcripts in ECs and HSCs was lower than that of ACE, but it was higher in MCs, consistent with the wider function of CDX2 in posterior morphogenesis at this stage.^9,10  Nevertheless, this pattern was rapidly evolving because, at E11.5, both transcripts were still detected in ECs and MCs, but were turned off in IAHCs. The loss of CDX2 expression observed in murine IAHCs at midgestation is in line with the absence of CDX2 in fetal liver– and BM-derived HSCs and is also confirmed in humans because, despite its expression in IAHCs, CDX2 was absent in CD34⁺ progenitors isolated from human fetal liver and cord blood (supplemental Table 1). Furthermore, ACE⁺ MCs isolated from early human embryos also expressed CDX2, which by contrast is absent in ACE⁻ vascular endothelial cells derived from the same embryos (supplemental Figure 3).

Based on the similarities between ACE and CDX2 patterns during embryonic hematopoiesis, we investigated whether CDX2 could participate in the gene regulation of ACE and/or of other RAS components. For this purpose, 2 leukemia cell lines, Nalm6 and K562, were transfected with plasmid pFlag-CDX2 encoding the human CDX2 homeoprotein.²²  CDX2 increased the endogenous expression of ACE mRNA (Figure 2A), whereas no effect was observed on any other RAS element (supplemental Figure 4). To get further insight into the underlying molecular mechanism, we constructed the reporter luciferase plasmid pACE-Luc containing the 461-bp promoter and the first 37 bp of exon 1 of the human ACE gene up to the translation start site. Luciferase activity in Nalm6 and K562 cells transfected with pACE-Luc was stimulated by cotransfection with pFlag-CDX2 (Figure 2B). The ACE promoter contains a putative consensus CDX-type binding site²³  overlapping the TATA box and conserved in mice in the configuration: TTTATAA (CDX-type site in bold; TATA box underlined). Changing the TTTATA sequence into TTTGTA in pACEm-Luc strongly compromised the CDX2-stimulatory effect (Figure 2B). This was not due to destroying the TATA box because cotransfection with plasmid pTBPspm3, which encodes a mutant TBP able to bind the TGTA sequence instead of the classical TATA box,²⁴  did restore the basal activity of the promoter but not the stimulatory effect of CDX2 (Figure 2B). In addition, chromatin immunoprecipitation with anti-Flag or anti-CDX2 antibodies revealed the CDX2 protein linked to chromatin containing the CDX-type binding site of the ACE promoter in nuclear extracts of pFlag-CDX2–transfected cells (Figure 2C).

Next, we analyzed the expression of ACE and CDX2 in human BM mononuclear cells (BMMCs) derived from healthy individuals and AML patients (Figure 2D). RT-qPCR revealed a basal expression of ACE transcripts in the BMMCs of healthy individuals (cycle threshold [Ct] = 33 cycles), and a significant increase in AML patients. Unlike ACE, CDX2 was virtually absent in BMMCs of healthy individuals (no RT-qPCR signal was detected below Ct = 50 cycles), but aberrant expression was detected at various levels in AML patients. Statistical analysis revealed a significant positive correlation (r = 0.718; P = .0005) between ACE and CDX2 levels in AML patients (Figure 2E). Moreover, a coexpression pattern between ACE and CDX2 was observed in leukemic blasts sorted from 7 AML patients (supplemental Figure 5).

Altogether, these data suggest that homeotic transcription factor CDX2, a major determinant of growth and patterning of the posterior embryo, is a part of the regulator network involved in ACE regulation during hematopoietic emergence. Strikingly, although absent in normal BM-derived HSCs, CDX2 becomes abnormally expressed in AML cells, at levels that correlate with the increase of ACE expression. This opens new perspectives for understanding the deleterious effect exerted by CDX2 upon abnormal reexpression in leukemia through the proliferative effect exerted by AngII on hematopoietic cell emergence and growth.^4,25 

The authors thank T. Ayoub for technical support.

This work was supported by INSERM and by grants from the University of Strasbourg, the Ligue Contre le Cancer Bourgogne France Comté, the Fondation ARC pour la Recherche sur le Cancer (PJA 20131200190), the Région Grand-Est (program OMAGE), and ANR (ANR-14-CE11-0008). E.J. was awarded a fellowship from INSERM-Alsace and from the Fondation ARC pour la Recherche sur le Cancer. K.B. was awarded a fellowship from INSERM-Alsace and from the Société Française d'Hématologie. R.E.O. was supported by a grant from ANR.

Contribution: R.E.O. and E.J. performed experiments and analysis, and contributed to interpretation of results and the writing of the manuscript; K.B. designed and conducted experiments, and analyzed and interpreted the data; I.G., C.D.-D., and B.G. conducted experiments and analyzed the data; C.G. and F.L. contributed to data interpretation and final discussion; B.L. and L.V. provided materials and contributed to data interpretation; M.N. designed and performed plasmid constructions and mutagenesis; J.-N.F. contributed to experiment design, interpretation of results, and the writing of the manuscript; and M.T. supervised the overall project, designed experiments, analyzed the data, and wrote the manuscript.

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

The current affiliation for R.E.O. is Ingénierie Moléculaire et Physiopathologie Articulaire (IMoPA), UMR7365 CNRS–University of Lorraine, Vandœuvre Les Nancy, France.

Correspondence: Manuela Tavian, University of Strasbourg, INSERM, IRFAC/UMR-S1113, Strasbourg, France; e-mail: manuela.tavian@inserm.fr.