Intra-Aortic Clusters Undergo Endothelial to Hematopoietic Phenotypic Transition during Early Embryogenesis

Intra-aortic clusters (IACs) attach to floor of large arteries and are considered to have recently acquired hematopoietic stem cell (HSC)-potential in vertebrate early mid-gestation embryos. The formation and function of IACs is poorly understood. To address this issue, IACs were characterized by immunohistochemistry and flow cytometry in mouse embryos. Immunohistochemical analysis revealed that IACs simultaneously express the surface antigens CD31, CD34 and c-Kit. As embryos developed from 9.5 to 10.5 dpc, IACs up-regulate the hematopoietic markers CD41 and CD45 while down-regulating the endothelial surface antigen VE-cadherin/CD144, suggesting that IACs lose endothelial phenotype after 9.5 dpc. Analysis of the hematopoietic potential of IACs revealed a significant change in macrophage CFC activity from 9.5 to 10.5 dpc. To further characterize IACs, we isolated IACs based on CD45 expression. Correspondingly, the expression of hematopoietic transcription factors in the CD45(neg) fraction of IACs was significantly up-regulated. These results suggest that the transition from endothelial to hematopoietic phenotype of IACs occurs after 9.5 dpc.


Introduction
During mouse embryogenesis, hematopoiesis begins at the extra-embryonic yolk sac (YS) at 7.5 days post-coitum (dpc) and shifts to fetal liver after mid-gestation, then to spleen and finally to bone marrow shortly before birth. There are two distinct waves of hematopoietic emergence: a transient wave, primarily restricted to erythropoiesis in YS blood islands prior to the connection of the circulation from the YS to the embryo; and a definitive wave originating in both the YS and embryo proper. The embryonic site has been identified in the aortic region, in the para-aortic splanchnopleura (p-Sp)/aorta-gonad-mesonephros (AGM) region [1][2][3][4][5][6]. Functional hematopoietic stem cells (HSCs) that can reconstitute adult recipients are first identified in the AGM region at 10.5 dpc after ex vivo organ culture [7]. The cells at 10.5 dpc that were not cultured ex vivo rarely reconstitute adult recipients, whereas those at 11.5 dpc can regardless [7][8][9]. Therefore, the cells that acquire HSC activity after culture step, have been termed ''pre-HSC''s. Although several reports characterize the surface marker expression on both pre-HSCs at 10.5 dpc and HSCs at 11.5 dpc, the developmental process of HSC generation still remains unclear [8][9][10][11]. Cell populations capable of reconstituting neonatal recipients are detected in the p-Sp/AGM region at 9.5 dpc [12][13]. These observations suggest that ancestor cells of HSC from the p-Sp/AGM region at 9.5 dpc require special microenvironments to acquire HSC activity and that HSCs undergo phenotypic changes from 9.5 to 10.5 dpc. In the AGM region, intra-aortic/arterial clusters (IACs) are observed attached to floors of large arteries in several species including chicken, mouse and humans [3]. Mouse IACs have been characterized morphologically and are primarily located in three large arteries, namely, the dorsal aorta (DA), the omphalomesenteric (vitelline) artery (OMA; VA) and the umbilical artery (UA) [3,[14][15]. IACs express both hematopoietic (CD41 and CD45) and endothelial (CD31, CD34 and VE-cadherin) surface markers [3,[15][16] suggesting that IACs are likely equivalent to ancestor cells of HSC and/or pre-HSCs and are derived from endothelial cells (ECs) at aortic/arterial regions. Although recent genetic approaches and novel tracing methods demonstrate that IACs are derived from ECs in zebrafish and mice, it is unclear how IACs form and acquire HSC activity [17][18][19][20][21][22][23][24][25].
To address how IACs form and function in HSC generation, we first visualized IACs by immunohistochemistry and confocal imaging and were found to simultaneously express CD31, CD34 and c-Kit. This approach enabled us to investigate the phenotypic characterization of IACs by flow cytometry and hematopoiesis assays. Here, we demonstrate a significant transition from endothelial to hematopoietic cell phenotype of IAC cells after 9.5 dpc.

Visualization of IACs in mouse embryos
Previous studies identified intra-aortic/arterial clusters (IACs) primarily by immunocytochemistry and microscopy [3,[14][15]. Recently, we successfully visualized hematopoietic cell clusters in mouse placenta using thick (20 mm) cryo-sections and antibodies recognizing the embryonic HSC markers c-Kit, CD31 and CD34 and applied this method to quantifying IACs [26]. Cell aggregates consisting of more than three c-Kit-positive cells were defined as an IAC. Here, we used confocal microscopy to expand upon our previous study and characterize the cell types found within IACs according to c-Kit, CD31 and CD34 expression ( Figure 1). The first IACs were observed as spherical structures in the omphalomesentric artery (OMA) at 9.0 dpc (12-14 somite pairs [SP]) ( Figure 1A, left). Between 9.5 dpc (18-22 SP) to 10.5 dpc (30)(31)(32)(33)(34), large arteries such as the dorsal aorta (DA), OMA and umbilical artery (UA) form [14]. IACs were observed in DA, OMA and UA at 10.5 dpc, and the size of IACs in the OMA and UA was significantly larger than those seen in the DA ( Figure 1A, right). Localization of IACs in DA was not restricted to the ventral wall of DA, but rather some IACs were observed at dorsal and lateral sides of the wall (data not shown). All IACs in the DA, OMA and UA at 10.5 dpc simultaneously expressed c-Kit, CD31 and CD34 ( Figure 1B-D). IACs expressing c-Kit in the different arteries analyzed were also positive for Ki-67, a marker of cell proliferation, regardless of location, suggesting that cells within IACs are highly proliferative ( Figure 1E).
IAC formation in the developing human embryo is poorly defined. Having defined the developmental progression of IAC in the mouse above, we next examined IAC morphology and phenotype in a 32 day-old human embryo. Immunohistochemistry of embryonic human cryosections was performed using antihuman CD34 and CD45 antibodies. As shown in Figure 4E, IACs can be detected in ventral wall of the dorsal aorta. CD34 was expressed by a wide range of vascular endothelial cells throughout the embryo. CD45 was restricted to round and in many cases clearly circulating cells. However, within the IAC observable on the ventral wall of the dorsal aorta, cells expressing both CD34 and CD45 can be seen. This reflects the expression pattern we have identified in embryonic mouse IACs.

Transcription factor hierarchy in IAC development
We next observed IAC formation by immunohistochemistry and flow cytometry in mouse embryos harboring mutations associated with aberrant embryonic hematopoiesis [27][28][29][30][31][32]. Immunohistochemical analysis of Runx1 -/embryos lacked IACs in the DA, OMA and UA. Flow cytometric analyses confirmed the absence of c-Kit + /CD31 + /CD34 + cells in Runx1 -/embryos compared to wild type embryos ( Figure 5A-B). Evi-1 -/embryos also lacked IACs in the DA, OMA and UA by immunohistochemistry. However, a small frequency of c-Kit + /CD31 + /CD34 + cells could be detected by flow cytometry ( Figure 5C). In c-Myb -/embryos, IACs were observed at the DA, OMA and UA, and c-Kit + /CD31 + /CD34 + cells were also observed by flow cytometry ( Figure 5D). Collectively, these results demonstrate that Runx1 is essential for IAC formation while Evi-1 appears to be playing a function downstream of Runx1 in this process.

Discussion
During embryogenesis, a unique cell biological shift takes places in which endothelial cells with adherens junctions detach from each other, alter gene expression and become hematopoietic cells. This process is limited both anatomically and temporally. We here demonstrated that the transition from endothelial to hematopoietic phenotype of IACs occurs from 9.5 dpc in the mouse embryo, earlier than previously described. Furthermore, we show that IACs are identifiable in the human embryo based on CD45 expression, implying that this process in mice is applicable to human.
Previously, we reported an immunohistochemistry visualization technique revealing hematopoietic cell clusters in placenta using thick (20 mm) cryo-sections and antibodies recognizing embryonic HSC markers [26]. Here, we applied this technique to obtain high quality confocal images of intra-aortic/arterial clusters (IACs) in the AGM region. We defined IACs as c-Kit + /CD31 + /CD34 + cells. Recently, c-Kit + /CD31 + /SSEA-1cells were also identified in the AGM region [11]. As CD31 is expressed on both IACs and primordial germ cells (PGCs), it was necessary to exclude PGCs according to SSEA-1 expression. As shown in Figure 2 and 5, we Figure S2. (B-G) Expression of hematopoietic and endothelial cell markers was analyzed on CD31 + /CD34 + /c-Kit + cells at 9.5 dpc (left) and 10.5 dpc (right) with the following antibodies: (B) VE-cadherin/CD144 (an endothelial cell marker), (C) CD41 (the earliest hematopoietic cell marker), (D) CD45 (a pan-leukocyte marker), (E) Sca-1 (a late fetal and adult HSC marker), (F) CD150 and (G) EPCR (adult HSC markers). At least 1,000 cells were assessed for each surface antigen. Representative profiles are shown. (H) Percentage of expression was summarized. At least 3 independent experiments were performed. Mean 6 2SD was calculated and shown at the top of bars. (I) One thousand sorted CD45-negative or CD45-positive CD31 + /CD34 + /c-Kit + cells were cultured in semisolid medium containing the hematopoietic cytokines, SCF (Stem Cell Factor), IL (Interleukin)-3, IL-6 and EPO (Erythropoietin). Left and right panels show each fraction and the total number of colonies, respectively. GEMM (colony-forming units of granulocyte erythrocyte monocyte macrophages); GM (of granulocyte macrophages); M (of macrophages); G (of granulocytes); BFU (burst forming units of erythroid cells). (J) 50-100 sorted CD31 + /CD34 + /c-Kit + cells at 9.5 dpc, as well as CD45-negative and CD45-positive CD31 + /CD34 + /c-Kit + cells were transplanted into busulfan-treated Ly5.1 mouse neonates. Approximately one year after transplantation, blood samples were collected and analyzed for CD45.2 expression by flow cytometry. Representative profile of flow cytometric analysis and its negative and positive controls are shown in Figure  S3 and S6, respectively. doi:10.1371/journal.pone.0035763.g002 could observe a small number of CD31 + /CD34cells, which are likely to be PGCs. Since PGCs do not express CD34 at this stage, we could positively select the IAC fraction based on our definition by flow cytometry [33]. Our observation of IACs is compatible with the result showing large IACs were primarily observed in omphalomesentric artery (OMA) and umbilical artery (UA) at 10.5 dpc [11]. In the mouse, IACs protruding into the lumen of arteries were previously reported at 9.5 dpc in studies using microscopy and Tie-2 immunohistochemistry [14,34]. Prior to 9.5 dpc, we identified the first IACs, which formed a spherical structure, in the OMA at 9.0 dpc ( Figure 1A). The OMA appears at 8.0 dpc and directly connects with the dorsal aorta (DA). The OMA anastomoses with the DA after 9.5 dpc and loses its connection with the UA by 10.5 dpc [14,35]. Our data ( Figure 1E) indicate that IACs are proliferative, based on Ki-67 staining. Taken together, it is likely that the first IACs in the OMA proliferate and are distributed into large arteries, such as the DA and UA, as the arterial system develops. Although several reports provide direct evidence that endothelial cells (ECs) generate IACs, we cannot rule out the possibility that either mesodermal cells, the ancestors of hematopoietic cells, or so-called hemangioblasts, which give rise both to ECs and hematopoietic cells, generate IACs by another pathway [17][18][19][20][21][22][23][24][25]. When VE-cadherin + /CD45cells were sorted out from AGM regions at 10.5 dpc, and co-aggregated with OP9 stromal cells, these cells acquired HSC activity [8]. As embryos develop, VE-cadherin + /CD45 + cells from AGM regions at 11.5 dpc can reconstitute adult recipients without culture step, whereas both VEcadherin + /CD45 +/cells can after aggregation culture with OP9 stromal cells. It suggests that the transition from endothelial to hematopoietic phenotype in pre-HSCs occurs between 10.5 and 11.5 dpc. According to our flow cytometric analysis of IACs, the transition from endothelial to hematopoietic phenotype occurs after 9.5 dpc (Figure 2). Although we found that 33% of c-Kit + /CD31 + / CD34 + cells at 9.5 dpc express VE-cadherin, most IACs defined as c-Kit + /CD31 + /CD34 + cells by flow cytometry did not contribute to blood vessel structure. VE-cadherin is expressed in IACs as well as in ECs [16]. It is likely that sorted VE-cadherin + /CD45cells from AGM regions at 10.5 dpc contained ECs with HSC potential in addition to some IACs. Further studies are necessary to determine how ECs contribute to IAC generation. CD150 belongs to the SLAM family and its expression is developmentally regulated on the surface of HSCs. At 11.5 dpc, CD150cells can reconstitute adult recipients, but CD150 + cells not [10]. In this study, CD150 expression was examined on c-Kit + /CD31 + /CD34 + cells by flow cytometry and the percentage of CD150 expression was not changed ( Figure 2F, H). It will be interesting to compare the CD150 expression between 10.5 and 11.5 dpc.
The pan-leukocyte marker CD45 is a transmembrane glycoprotein that functions as a protein phosphotyrosine phosphatase. Although loss of the CD45 gene results in T and B lymphocyte anomalies in adult, there appears to be no significant abnormality in HSC development during embryogenesis [36][37][38]. We observed that CD45 protein expression was up-regulated in c-Kit + /CD31 + / CD34 + cells between 9.5 and 10.5 dpc ( Figure 2D). Our results are compatible with the report showing that CD45 is expressed on the surface of IACs at 10.5 dpc, but not on the IACs at 9.5 dpc [11]. In agreement with previous reports, we observed no significant differences in HSC activity based on neonatal transplantation, whereas myeloid potential differs based on colony formation assay between CD45-negative and CD45-positive c-Kit + /CD31 + / CD34 + cells, suggesting that CD45 expression is not required for hematopoietic cell identity ( Figure 2I, J) [39][40]. However, pre-HSCs that can reconstitute both adult and neonatal recipients appear at 10.5 dpc, whereas presumptive ancestor cells of HSC that can reconstitute only neonatal but not adult recipients appear at 9.5 dpc [7,[12][13]. In accordance with flow cytometric data, some IACs expressed CD45 while others did not in both 10.5 dpc mouse embryos and 32 day-old human embryos ( Figure 4B-E). Taken together, although CD45 does not function in HSC development, its expression on the cell surface might serve as a marker of pre-HSC maturation from ancestor cells of HSC. With regard to myeloid potential, only macrophage development differs (Figure 2I). At 10.5 dpc, macrophages are reportedly c-Kit -/ CD31 -/CD45 + cells, and we could observe some c-Kit -/CD45 + cells in the AGM regions (Figure 4) [11]. CD45 expression on c-Kit + /CD31 + /CD34 + cells might be the diverging point of myeloid potential. Furthermore, we identified CD45 gene expression in CD45-negative c-Kit + /CD31 + /CD34 + cells, suggesting that these cells are primed to differentiate into CD45-positive c-Kit + / CD31 + /CD34 + cells. Expression levels of Runx1, c-Myb, SCL and Gata2 were highest in CD45-low positive c-Kit + /CD31 + /CD34 + cells, implying that the transition from endothelial to hematopoietic phenotype of IACs occurs in CD45-low positive c-Kit + / CD31 + /CD34 + cells, as these transcription factors are reportedly important for the switch to hematopoietic cells [22]. Evi-1 is involved in vasculo-angiogenesis in addition to HSC development [31]. Therefore, high expression level of Evi-1 gene in CD45negative c-Kit + /CD31 + /CD34 + cells implies that this population still preserves some endothelial identity.
Several evidences reveal that HSCs are generated from ECs [17][18][19][20][21]. Taken together, our results corroborate HSC-generation from ECs and imply that IACs gradually acquire hematopoietic phenotype after 9.5 dpc. Understanding how IACs are generated could lead to an understanding of how to manipulate HSC generation from ES/iPS cells and thus be applicable to future clinical applications.

Mice
Ly5.1 (Sankyo Labo Service, Tokyo, Japan) mice, Ly5.2 adult C57/BL6 mice (Kyudo, Tosu, Japan), ICR mice (SLC, Hamamatsu, Japan), Runx1 +/mice (provided by Dr. Speck at University of Pennsylvania), Evi-1 +/mice (JAX mice and Services, Bar Harbor, ME) and c-Myb +/mice (JAX mice and Services) were used in these studies. To analyze cells, pregnant mice were sacrificed at 9.0-10.5 dpc and somite pair number was counted. Embryos at 9.0 dpc with 12-14 somite pairs (SP), 9.5 dpc with 18-22 SP and 10.5 dpc with 30-34 SP were dissected out, respectively. Animals were handled according to the Guidelines for the Care and Use of Laboratory Animals of Kyushu University. This study was approved by Animal Care and Use Committee, Kyushu University (Approval ID: A21-068-0).

Mouse immunohistochemistry
Embryos were dissected out and fixed in 2% paraformaldehyde in PBS, followed by equilibration in 30% sucrose in PBS. Embryos were embedded in OCT compound (SAKURA, Tokyo, Japan) and frozen in liquid nitrogen. Tissues were sliced at 20 mm on a Leica CM1900 UV cryostat, transferred to glass slides (Matsunami, Osaka, Japan) and dried thoroughly. Sections were blocked in 1% BSA in PBS and incubated in PBS containing 1% BSA with appropriate dilutions of the following primary antibodies: goat anti-mouse c-Kit (R&D Systems, Minneapolis, MN), rat antimouse CD31 (BD Biosciences, San Diego, CA), rat anti-mouse CD34 (BD Biosciences), rat anti-mouse CD45 (Biolegend) and rat anti-mouse Ki-67 antigen (Dako Corporation, Carpinteria, CA) at 4C overnight. After washing in PBS three times, sections were incubated with appropriate dilutions of the following secondary antibodies: Alexa Fluor 488 donkey anti-rat IgG (Invitrogen, Carlsbad, CA), Alexa Fluor 488 donkey anti-goat IgG (Invitrogen), Alexa Fluor 546 donkey anti-goat IgG (Invitrogen) and Alexa Fluor 568 donkey anti-goat IgG (Invitrogen), as well as TOTO-3 (Invitrogen) to stain nuclei, at room temperature for 30 minutes. Samples were mounted on coverslips using fluorescent mounting medium (Dako Corporation) and assessed using a FluoView 1000 confocal microscope (Olympus, Tokyo, Japan).

Human tissues
Human embryos were obtained from voluntary abortions performed according to guidelines and with the approval of the French National Ethics Committee. In all cases, written consent allowing use of the embryo for research was obtained from the patient. Developmental age was estimated based on anatomical criteria and the Carnegie classification as previously described [41][42].

Human immunohistochemistry
Embryos were fixed overnight at 4uC in PBS plus 4% paraformaldehyde (Sigma-Aldrich), rinsed twice in PBS, then in PBS/15% sucrose (Sigma-Aldrich) for at least 24 hours. Tissues were then embedded in PBS with 15% sucrose and 7.5% gelatin (Sigma-Aldrich), frozen and stored at -80uC. Frozen sections (5 mm) were stored at -20uC until use, and then thawed and hydrated in PBS [37]. For double-staining, the TSA Plus Fluorescence amplification system was used, according to the manufacturer's instructions (NEN-Perkin Elmer). Endogenous peroxidases were inhibited for 20 minutes in PBS containing 0.2% hydrogen peroxide (Sigma-Aldrich). Sections were washed in PBS and non-specific binding sites were blocked with PBS/5% goat serum (Vector Laboratories) for 1 hour. Sections were then incubated with uncoupled antibody to CD45 (overnight at room temperature). After rinsing, sections were incubated with biotinylated goat anti-mouse IgG antibody (Immunotech) for 1 hour and then with peroxidase-labeled streptavidin (Immunotech) for 1 hour. Staining was revealed using fluorescent tyramide (TMR, Tetramethylrhodamine). Residual peroxidase activity was inhibited in PBS/0.2% hydrogen peroxide for 10 min at RT. After 3 washings in PBS, slides were treated with an Avidin/Biotin blocking kit according to the manufacturer's instructions (Vector Laboratories). Sections were washed and incubated with anti-CD34 antibody at room temperature for 2 hours, then with biotinylated goat anti-mouse IgG antibody (Immunotech) for 1 hour at RT, and with Alexa 488-labeled streptavidin for 1 hour. Slides were mounted in Vectashield medium (Vector Laboratories). Monoclonal antibodies to CD34 (IgG1, clone Qbend-10) and CD45 (IgG1, clone Hle-1) were purchased from Immunotech and Becton-Dickinson Biosciences, respectively.

Cell preparation
The caudal portion of embryos containing the p-Sp/AGM region was used to obtain a single cell suspension. Tissues were incubated with 1 mg/ml collagenase in medium supplemented with 10% fetal bovine serum for 30 minutes at 37C and filtered through 40-mm nylon cell strainers (BD Biosciences).

RNA extraction and real-time PCR analysis
Total RNA was isolated using the RNAqueous 4PCR kit (Ambion Inc., Austin, Texas). mRNA was reverse transcribed using a High-Capacity RNA-to-cDNA kit (Life Technologies, Carlsbad, CA). The quality of cDNA synthesis was evaluated by amplifying mouse ß-actin using PCR. Thirty thermal cycles were used as follows: denaturation at 95uC for 10 sec, annealing at 60uC for 20 sec, followed by extension at 72uC for 20 seconds. Gene expression levels were measured by real time PCR with TaqManH Gene Expression Master Mix and StepOnePlus TM real time PCR (Life Technologies). All probes were from TaqManH Gene Expression Assays (Life Technologies). All analyses were performed in triplicate wells; mRNA levels were normalized to ßactin and the relative quantity (RQ) of expression was compared with a reference sample.

Colony formation assay
Sorted cells were suspended in 3 ml of MethoCultH GF M3434 (Stemcell Technologies) distributed into three 35 mm dishes and then incubated in 5% CO 2 at 37uC. Colonies were counted up 14 days later using an inverted phase contrast microscope CKX41 (Olympus, Tokyo, Japan).

Transplantation assay
To examine neonatal repopulating HSCs, sorted cells were transplanted into busulfan-treated Ly5.1 mouse neonates as described previously [9,15]. Briefly, time-pregnant mice were injected on days 17 and 18 after conception with 15 mg of busulfan/gram body weight of the mother (Sigma-Aldrich, St.Louis MO). Isolated cells were suspended in 25 ml PBS and transplanted into neonates at the time of delivery using a 100 ml Hamilton syringe (Hamilton, Reno, NV). Approximately one year after transplantation, blood samples were collected, lysed in BD Pharm Lyse (BD Biosciences) and analyzed for CD45.2 expression by flow cytometry.