Hematopoietic stem cells (HSCs) emerge from and expand in the mouse placenta at mid-gestation. To determine their compartment of origin and define extrinsic signals governing their commitment to this lineage, we identified hematopoietic cell (HC) clusters in mouse placenta, defined as cells expressing the embryonic HSC markers CD31, CD34 and Kit, by immunohistochemistry. HC clusters were first observed in the placenta at 9.5 days post coitum (dpc). To determine their origin, we tagged the allantoic region with CM-DiI at 8.25 dpc, prior to placenta formation, and cultured embryos in a whole embryo culture (WEC) system. CM-DiI-positive HC clusters were observed 42 hours later. To determine how clusters are extrinsically regulated, we isolated niche cells using laser capture micro-dissection and assayed them for expression of genes encoding hematopoietic cytokines. Among a panel of candidates assayed, only stem cell factor (SCF) was expressed in niche cells. To define niche cells, endothelial and mesenchymal cells were sorted by flow cytometry from dissociated placenta and hematopoietic cytokine gene expression was investigated. The endothelial cell compartment predominantly expressed SCF mRNA and protein. To determine whether SCF/Kit signaling regulates placental HC cluster proliferation, we injected anti-Kit neutralizing antibody into 10.25 dpc embryos and assayed cultured embryos for expression of hematopoietic transcription factors. Runx1, Myb and Gata2 were downregulated in the placental HC cluster fraction relative to controls. These observations demonstrate that placental HC clusters originate from the allantois and are regulated by endothelial niche cells through SCF/Kit signaling.
INTRODUCTION
During mouse embryogenesis, hematopoiesis begins in the yolk sac (YS), producing mainly primitive erythroid cells at 7.5 days post coitum (dpc). Shortly thereafter, definitive myelo-erythroid progenitor cells appear in the YS, which seed the fetal liver (Cumano et al., 1996; Ferkowicz and Yoder, 2005; Li et al., 2003; McGrath and Palis, 2005; Palis et al., 1999). This process, termed primitive hematopoiesis, diminishes at 12.5 dpc, when definitive hematopoiesis, which sustains the adult blood system through hematopoietic stem cells (HSCs), begins in fetal liver (Sugiyama and Tsuji, 2006). Although there is controversy over where HSCs are generated – in the extra-embryonic YS or intra-embryonic para-aortic-splanchnopleural mesoderm (P-Sp)/aorta-gonad-mesonephros (AGM) region – recent studies suggest that both the YS and P-Sp/AGM region contain HSCs capable of reconstituting adult bone marrow hematopoiesis (Cumano et al., 1996; Matsuoka et al., 2001; Medvinsky and Dzierzak, 1996; Samokhvalov et al., 2007; Yoder et al., 1997a; Yoder et al., 1997b). Thereafter, these AGM HSCs are thought to circulate and colonize fetal liver, where HSC expansion occurs (Cudennec et al., 1981; Ema and Nakauchi, 2000; Houssaint, 1981; Johnson and Moore, 1975; Sugiyama et al., 2005). In addition to these sites, several reports suggest that the placenta functions not only in gas exchange and fetal nutrition but also in hematopoiesis at approximately mid-gestation (Dancis et al., 1968; Dancis et al., 1977; Melchers, 1979). It is also reported that a significant proportion of hematopoietic progenitor cells (HPCs), including highly proliferative potential colony forming cells (HPP-CFCs), are located in the mouse placenta (Alvarez-Silva et al., 2003). HSCs are detected at this site by 11.5 dpc and the number of long-term reconstituting (LTR)-HSCs dramatically increases from 11.5 dpc to 12.5 dpc, resulting in a 15-fold increase in HSC activity compared with that of the AGM region (Gekas et al., 2005; Ottersbach and Dzierzak, 2005). Taken together, these findings indicate that mouse placenta is likely to be a site for HSC generation and expansion at mid-gestation.
HSCs are regulated by intrinsic programming and by extrinsic signaling from so-called niche cells. However, it is unclear how HSC generation and expansion is regulated in the placenta. To address this issue, we identified hematopoietic cell (HC) clusters, defined as cells expressing embryonic HSC markers such as CD31 (Pecam1 – Mouse Genome Informatics), CD34 and Kit, by immunohistochemistry, enabling us to follow the origin of HC clusters and identify surrounding niche cells. We then used vital dye labeling to demonstrate that HC clusters in placenta originate from the allantois, an embryonic compartment of the placenta. Furthermore, we showed that HC clusters are regulated by vascular niche cells through SCF (Kitl – Mouse Genome Informatics)/Kit signaling.
MATERIALS AND METHODS
Immunohistochemistry
Mouse embryos were dissected out and fixed in 2% paraformaldehyde in PBS, followed by equilibration in 30% sucrose in PBS. Placentas were embedded in OCT compound (SAKURA, Tokyo, Japan) and frozen in liquid nitrogen. Tissues were sliced at 20 μm with 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 Kit (1:500; R&D Systems, Minneapolis, MN), rat anti mouse-CD31 (1:500; BD Biosciences, San Diego, CA), rat anti mouse-CD34 (1:500; BD Biosciences), rat anti-mouse CD41 (1:300; BioLegend, San Diego, CA), rat anti-mouse CD45 (1:300; BioLegend) and rat anti-mouse F4/80 (1:300; BioLegend) at 4°C 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 (1:300; Invitrogen, Carlsbad, CA) and Alexa Fluor 568 donkey anti-goat IgG (1:300; Invitrogen), as well as TOTO-3 (1:1500; Invitrogen) to stain nuclei, at room temperature for 30 minutes. Samples were mounted on coverslips using fluorescent mounting medium (Dako Corporation, Carpinteria, CA) and were assessed using a FluoView 1000 confocal microscope (Olympus, Tokyo, Japan). Cell aggregates consisting of more than four Kit/CD31 or Kit/CD34 double-positive cells were defined as a hematopoietic cluster.
Cell preparation
Placentas without deciduas and umbilical vessels were used to obtain a single cell suspension. Tissues were passed through 21-gauge needles, incubated with 1 mg/ml collagenase in medium supplemented with 10% fetal bovine serum for 30 minutes at 37°C and filtered through 40 μm nylon cell strainers (BD Biosciences). In analysis and sorting of HSCs, density gradient centrifugation using lymphocyte cell separation medium (Cedarlane Laboratories, Eugene, OR) was performed to harvest mononuclear cells. After centrifugation, the cell pellet was used as the placental cell population.
Flow cytometry and cell sorting
As macrophages are found in HSC preparations (CD31, CD34, Kit), anti-mouse F4/80 antibody was added to identify and to exclude them from analysis. Antibodies used for analysis of the HSC population were: FITC-conjugated anti-mouse CD41 (eBioscience, San Diego, CA), FITC-conjugated anti-mouse Sca-1 (eBioscience), FITC-conjugated anti-mouse EPCR (endothelial protein C receptor) (known as CD201) (Stem Cell Technologies, Vancouver, BC), PE-conjugated anti-mouse CD31 (BD Biosciences), APC-conjugated anti-mouse F4/80 (BioLegend), PE-Cy7-cojugated anti-mouse CD45 (BioLegend), APC-Cy7-conjugated anti-mouse Kit (eBioscience) and Pacific Blue-conjugated anti-mouse CD34 (eBioscience). For endothelial and mesenchymal cell populations, FITC-conjugated anti-mouse Ter-119 (eBioscience), PE-conjugated anti-mouse CD31 (BD Biosciences), APC-conjugated anti-mouse Kit (BD Biosciences), PE-Cy7-cojugated anti-mouse CD45 (BioLegend) and Pacific Blue-conjugated anti-mouse CD34 (eBioscience) were used. Flow cytometric analysis and cell sorting were carried out using a FACS Aria cell sorter (BDIS, San Jose, CA). The data files were analyzed using FlowJo software (Tree Star, San Carlos, CA).
RNA extraction and real-time PCR analysis
Total RNA was isolated using the RNAqueous 4PCR kit (Ambion, Austin, TX). 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 β-actin using PCR. Thirty thermal cycles were used as follows: denaturation at 95°C for 10 seconds, annealing at 60°C for 20 seconds, followed by extension at 72°C for 20 seconds. Gene expression levels were measured by real time-PCR with TaqMan Gene Expression Master Mix and StepOnePlus real-time PCR (Life Technologies). All probes were from TaqMan Gene Expression Assays (Life Technologies). All analyses were performed in triplicate wells; mRNA levels were normalized to β-actin and the relative quantity (RQ) of expression was compared with a reference sample.
Enzyme-linked immunosorbent assay
Proteins were extracted from a flow-sorted endothelial and mesenchymal cell population using a Q Proteome Mammalian Protein Preparation kit (Qiagen, Valencia, CA). SCF in both populations was assayed using an (enzyme-linked immunosorbent assay) ELISA kit (Mouse SCF Immunoassay, R&D Systems), according to the manufacturer's instructions. The optical density was measured in a Thermo Multiskan EX plate reader (Thermo Fisher Scientific, Waltham, MA).
Laser capture micro-dissection
For this procedure, embryos were not fixed and equilibration in sucrose in PBS was not undertaken. Tissues sliced at 20 μm on a cryostat were transferred to glass slides (Matsunami), placed on ice and immediately stored at –80°C until use. After thawing, frozen cryosections were washed in PBS three times and incubated with 1:500 goat anti Kit (R&D Systems) for 60 minutes. After washing in PBS three times, sections were incubated with 1:300 Alexa Fluor 488-conjugated donkey anti-goat IgG (Invitrogen) for 60 minutes to detect Kit-positive cells. In this analysis, fluorescent Kit-positive cell aggregates were considered to be HC clusters, and surrounding cells were marked and cut by laser. Those cells were captured as niche cells and transferred to microcentrifuge tubes containing 10 μl extraction buffer using a Pico Pure RNA Isolation Kit (Molecular Devices, Silicon Valley, CA). Laser capture micro-dissection (LCM) was carried out using an ArcturusXT Laser Capture Microdissection System (Molecular Devices). Immunohistochemistry for LCM was carried out at 4°C, and all solutions were treated with diethylpyrocarbonate (Wako, Osaka, Japan).
CM-DiI labeling of the allantois
CM-DiI (Invitrogen), which binds to the cellular membrane, is non-toxic and remains fluorescent for at least for 4 days, was injected into the basal part of allantois of ICR mouse embryos at 8.25 dpc, prior to chorio-allantoic fusion (Downs and Harmann, 1997). Injected embryos were subjected to whole embryo culture (WEC). Dissection and manipulation of embryos was completed within an hour of starting WEC (Sugiyama et al., 2007).
Intra-cardiac injection
For this procedure, 0.2-0.5 μl of anti-Kit neutralizing antibody (ACK2) (eBioscience) and purified rat IgG2b (isotype control; eBioscience) were administered to ICR mouse embryos at 10.25 dpc, a stage at which we can cut the avascular area of yolk sac with minimal bleeding and inject materials into the heart, by intra-cardiac injection, as previously described (Kulkeaw et al., 2009; Sugiyama et al., 2003). Briefly, embryos in PBS were visualized under a stereomicroscope (Leica Microsystems MZ6, Wetzlar, Germany). Both uterus and deciduo capsularis were removed, and the yolk sac was cut along the yolk sac arteries with care to avoid excessive hemorrhage. The amnion was opened to allow needle access to the heart. The needle was made from glass capillary tubes (Narishige GC-10, Japan) using a micropipette puller (Narishige, Tokyo, Japan). Injected embryos were placed in the WEC system within 1 hour of isolation.
Whole mouse embryo culture (WEC)
Injected embryos were transferred to culture bottles containing rat serum supplemented with 2 mg/ml glucose in a WEC system (Ikemoto Scientific Technology, Tokyo, Japan) and cultured for 6 hours in the case of intra-cardiac injection or 42 hours in the case of CM-DiI labeling at 37°C in the dark with a continuous supply of gas (60% O2 and 5% CO2 balanced with N2) (Kulkeaw et al., 2009; Osumi-Yamashita et al., 1997; Sugiyama et al., 2003; Sugiyama et al., 2005; Sugiyama et al., 2007). After WEC, embryos exhibiting no conspicuous bleeding or anomalies were analyzed.
Mutant strains
Runx1+/– mouse strains were kindly provided by Dr N. Speck. Evi1–/– and Myb–/– mouse strains were obtained from the Jackson Laboratory (Bar Harbor, ME, USA).
RESULTS
Visualization and characterization of HC clusters in the mouse placenta
Previous studies have identified placental HC clusters primarily by microscopic inspection (Ottersbach and Dzierzak, 2005; Rhodes et al., 2008). To extend these studies, quantify clusters, understand their relationship with other placental components and identify niche cells, we carried out immunohistochemistry using thick (20 μm) cryosections and antibodies recognizing embryonic HSC markers – namely, CD31, CD34 and Kit (Baumann et al., 2004; North et al., 2002; Yoder et al., 1997a). As shown in Fig. 1A,B, using confocal microscopy we defined HC clusters as aggregates of more than four Kit+/CD31+/CD34+ cells. Clusters were attached to the endothelial wall of capillary vessels, the so-called vascular labyrinth region, from 10.5 dpc to 12.5 dpc and were morphologically similar to those seen in the AGM region at 10.5 dpc (Fig. 1C). To further characterize HC clusters, placental tissue was dissociated and analyzed by flow cytometry after first removing macrophages expressing F4/80 (Emr1 – Mouse Genome Informatics). Other HSC markers [such as CD41 (Itga2b – Mouse Genome Informatics), EPCR (Procr – Mouse Genome Informatics; CD201), Sca-1 (Ly6a – Mouse Genome Informatics) and the pan-leukocyte marker CD45 (Ptprc – Mouse Genome Informatics)] were expressed on Kit+/CD31+/CD34+/F4/80– cells in the placenta (Fig. 2). Among Kit+/CD31+/CD34+/F4/80– cells, expression of CD41, EPCR (CD201) and Sca-1 decreased from 10.5 dpc to 11.5 dpc, whereas CD45 expression increased over this period. At 12.5 dpc, the embryonic HSC marker CD31 was expressed on 90.2 % of CD34+/Sca-1+/Kit+ cells, which were previously defined as LTR-HSCs (see Fig. S1 in the supplementary material) (Gekas et al., 2005; Ottersbach and Dzierzak, 2005). We next observed HC cluster formation in embryos harboring various mutations associated with aberrant embryonic hematopoiesis (Goyama et al., 2008; Mucenski et al., 1991; North et al., 1999; Okuda et al., 1996; Wang et al., 1996; Yuasa et al., 2005). Specifically, in Runx1–/– embryos, no HC clusters were observed inside capillary vessels in the placenta (Fig. 1D, upper). In addition, in Evi1–/– (Mecom–/– – Mouse Genome Informatics) embryos, no HC clusters were observed inside capillary vessels at the time of abnormal vessel formation (Fig. 1D, middle). Finally, in Myb–/– embryos, HC cluster size overall was larger than that seen in wild-type embryos (Fig. 1D, lower).
Confocal images of HC clusters expressing CD31/CD34/Kit in the placenta and aortic region. Sections both of placenta and AGM region were made from ICR mouse embryos at 10.5 dpc, stained with antibodies and observed under confocal microscopy. (A,B) HC clusters in 10.5 dpc placenta. (A) CD31 (red), Kit (green) and TOTO-3 (blue). (B) CD34 (red), Kit (green) and TOTO-3 (blue). (C) HC cluster in the aorta at 10.5 dpc. CD34 (red), Kit (green), and TOTO-3 (blue). (D) Altered phenotype of HC clusters in the Runx1–/– (upper), Evi1–/– (middle) and Myb–/– (lower) placentas at 10.5 dpc.
Confocal images of HC clusters expressing CD31/CD34/Kit in the placenta and aortic region. Sections both of placenta and AGM region were made from ICR mouse embryos at 10.5 dpc, stained with antibodies and observed under confocal microscopy. (A,B) HC clusters in 10.5 dpc placenta. (A) CD31 (red), Kit (green) and TOTO-3 (blue). (B) CD34 (red), Kit (green) and TOTO-3 (blue). (C) HC cluster in the aorta at 10.5 dpc. CD34 (red), Kit (green), and TOTO-3 (blue). (D) Altered phenotype of HC clusters in the Runx1–/– (upper), Evi1–/– (middle) and Myb–/– (lower) placentas at 10.5 dpc.
Flow cytometric analysis of CD31+/CD34+/Kit+/F4/80– placental cells using surface expression HSC markers. Single cell suspensions of placentas at 10.5, 11.5 and 12.5 dpc, were prepared and analyzed by flow cytometry. The cells that express CD31, CD34 and Kit (markers of HC clusters), but not F4/80 (a macrophage marker) were first gated. Expression of HSC markers, such as CD41, EPCR (CD201), Sca-1 and CD45 was analyzed on CD31+/CD34+/Kit+/F4/80– cells at 10.5 dpc (upper), 11.5 dpc (middle) and 12.5 dpc (lower).
Flow cytometric analysis of CD31+/CD34+/Kit+/F4/80– placental cells using surface expression HSC markers. Single cell suspensions of placentas at 10.5, 11.5 and 12.5 dpc, were prepared and analyzed by flow cytometry. The cells that express CD31, CD34 and Kit (markers of HC clusters), but not F4/80 (a macrophage marker) were first gated. Expression of HSC markers, such as CD41, EPCR (CD201), Sca-1 and CD45 was analyzed on CD31+/CD34+/Kit+/F4/80– cells at 10.5 dpc (upper), 11.5 dpc (middle) and 12.5 dpc (lower).
To further characterize HC clusters, which we define immunohistochemically as cells expressing CD31, CD34 and Kit, we double-stained tissues with CD41, CD45 and F4/80. CD41 marks embryonic HSCs (Mikkola et al., 2003a; Rhodes et al., 2008). HC clusters at 10.5 dpc expressed Kit but not CD41, whereas those at 11.5 dpc expressed both Kit and CD41 (Fig. 3A-C). The intensity of CD41 expression in HC clusters expressing Kit was relatively weak compared with single CD41+ cells. The pan-leukocyte marker CD45 was also weakly expressed HC clusters at the AGM region (Godin and Cumano, 2002). HC clusters at 10.5 dpc expressed Kit but not CD45, whereas those at 11.5 dpc expressed both Kit and CD45 (Fig. 3D,E). Like CD41 expression, the intensity of CD45 expression in HC clusters expressing Kit was relatively weak compared with single CD45+ cells. HC clusters at 10.5 dpc did not express the macrophage marker F4/80 (Fig. 3F). However, circulating Kit+ cells inside blood vessels weakly expressed F4/80 (Fig. 3G,H). It has been reported that F4/80+ macrophages populate the placental mesenchyme (Rhodes et al., 2010). In agreement, we observed some single F4/80+ cells in the mesenchyme, in addition to Kit+/F4/80+ cells in circulation in the placenta (data not shown). Taken together, our data suggest that by immunohistochemical analysis, combined CD31/CD34/Kit positivity is sufficient to identify HC clusters.
Origin of HC clusters in placenta
To identify the origin of HC clusters in the placenta, we examined them at stages earlier than 10.5 dpc. The allantois, which originates from the embryo, and the extra-embryonic chorion fuse at 8.5 dpc and primary villi begin to develop at 9.0 dpc (Watson and Cross, 2005). HC clusters expressing CD31/CD34/Kit were observed at the allantois and chorionic plate at 9.5 dpc (Fig. 4A). However, HC clusters were not observed in either the allantois or chorion at 8.5 dpc, prior to placenta development (see Fig. S2 in the supplementary material). To follow the fate of allantoic cells, the basal part of the allantois was tagged with CM-DiI at 8.25 dpc and tagged embryos were cultured in a WEC system (Fig. 4B) (Khakoo et al., 2006; Krishnamurthy et al., 2008; Kulkeaw et al., 2009; Osumi-Yamashita et al., 1997; Silva et al., 2006; Sugiyama et al., 2003). After 42 hours in culture, all embryos developed normally (data not shown). The allantois fused to the chorion, forming both the umbilical cord and placenta, in which CM-DiI fluorescence could be detected (n=4) (Fig. 4C). Sections of embryos tagged with CM-DiI were stained for Kit by immunohistochemistry. CM-DiI/Kit-positive HC clusters were observed in the developing placenta, strongly suggesting that these clusters are derived from the allantois (Fig. 4D,E; see Fig. S3 in the supplementary material).
Proliferative status of HC clusters in placenta
Kit+/CD31+/CD34+/F4/80– cells sorted from placenta at 12.5 dpc exhibited immature morphology, appearing as blast cells when stained with May-Grunwald Giemsa (Fig. 5A). To understand the kinetics of HC cluster formation in placenta from 10.5 dpc to 12.5 dpc, we calculated the number of Kit+/CD31+/CD34+/F4/80– cells in the placenta at various developmental stages. We observed that the number of Kit+/CD31+/CD34+/F4/80– cells per placenta increased as the embryo developed (427, 1540 and 3227 cells at 10.5, 11.5 and 12.5 dpc, respectively), suggesting that they are proliferating (Fig. 5B). We next investigated cell cycle status of these cells. Kit+/CD31+/CD34+/F4/80– cells were flow sorted and stained with an antibody against Ki-67, a marker of cell proliferation (Fig. 5C) (Scholzen and Gerdes, 2000). The proportion of Ki-67+ cells in sorted Kit+/CD31+/CD34+/F4/80– cells was 80.4%, 77.2% and 48.2% at 10.5, 11.5 and 12.5 dpc, respectively (Fig. 5D), indicating that HC cluster cells in the placenta at 10.5 and 11.5 dpc are more proliferative than cluster cells at 12.5 dpc.
Confocal images of CD41, CD45 and F4/80 placental expression. Placenta sections were made from ICR mouse embryos both at 10.5 and 11.5 dpc, stained with antibodies and observed under confocal microscopy. (A,D) Images of 10.5 dpc placenta. (B,C,E-H) Images of 11.5 dpc placenta. (A-C) Sections were stained with anti-Kit antibody (red), anti-CD41 antibody (green) and TOTO-3 (blue). (A) HC cluster at 10.5 dpc expressed Kit, but not CD41. Arrow indicates HC cluster. (B) HC cluster at 11.5 dpc expressed Kit but not CD41. Some single cells strongly expressed CD41. Arrow indicates HC cluster. (C) HC cluster at 11.5 dpc expressing both Kit and CD41. The intensity of CD41 expression was relatively weak. Arrow indicates HC cluster. (D,E) Sections were stained with anti-Kit antibody (red), anti-CD45 antibody (green) and TOTO-3 (blue). (D) HC cluster at 10.5 dpc expressing Kit, but not CD45. Arrow indicates HC cluster. (E) HC cluster at 11.5 dpc expressed both Kit and CD45. The intensity of CD45 expression was relatively weak. Arrow indicates HC cluster. (F-H) Sections were stained with anti-F4/80 antibody (red), anti-Kit antibody (green) and TOTO-3 (blue). (F) HC cluster at 10.5 dpc expressing Kit, but not F4/80. Arrow indicates a Kit–/F4/80+ cell. (G) Blood vessel at 11.5 dpc. Some single cells expressed F4/80. (H) High magnification view of boxed area in G. Arrow indicates c-Kit+/F4/80+/+ cell circulating inside of blood vessel. For all images, original magnification is ×40.
Confocal images of CD41, CD45 and F4/80 placental expression. Placenta sections were made from ICR mouse embryos both at 10.5 and 11.5 dpc, stained with antibodies and observed under confocal microscopy. (A,D) Images of 10.5 dpc placenta. (B,C,E-H) Images of 11.5 dpc placenta. (A-C) Sections were stained with anti-Kit antibody (red), anti-CD41 antibody (green) and TOTO-3 (blue). (A) HC cluster at 10.5 dpc expressed Kit, but not CD41. Arrow indicates HC cluster. (B) HC cluster at 11.5 dpc expressed Kit but not CD41. Some single cells strongly expressed CD41. Arrow indicates HC cluster. (C) HC cluster at 11.5 dpc expressing both Kit and CD41. The intensity of CD41 expression was relatively weak. Arrow indicates HC cluster. (D,E) Sections were stained with anti-Kit antibody (red), anti-CD45 antibody (green) and TOTO-3 (blue). (D) HC cluster at 10.5 dpc expressing Kit, but not CD45. Arrow indicates HC cluster. (E) HC cluster at 11.5 dpc expressed both Kit and CD45. The intensity of CD45 expression was relatively weak. Arrow indicates HC cluster. (F-H) Sections were stained with anti-F4/80 antibody (red), anti-Kit antibody (green) and TOTO-3 (blue). (F) HC cluster at 10.5 dpc expressing Kit, but not F4/80. Arrow indicates a Kit–/F4/80+ cell. (G) Blood vessel at 11.5 dpc. Some single cells expressed F4/80. (H) High magnification view of boxed area in G. Arrow indicates c-Kit+/F4/80+/+ cell circulating inside of blood vessel. For all images, original magnification is ×40.
Regulation of HC clusters by niche cells
To investigate extrinsic factors that regulate HC cluster proliferation, we used an LCM system to collect niche cells surrounding HC clusters expressing Kit at 11.5 dpc. This technique enables us to isolate precisely specific cell compartments in tissue sections (Gomez and Harrison, 2009). The experimental strategy is shown in Fig. S4 in the supplementary material. We obtained niche cells comprising both endothelial and mesenchymal cells. To collect total RNAs by LCM in numbers sufficient to perform further analysis meant that we had to shorten the immunostaining period. We also could not use confocal microscopy to visualize HC clusters due to hardware limitations. We found that among Kit, CD31, CD34 and CD41 antibodies, the Kit antibody was most sensitive and specific to stain sections quickly and identify small HC clusters. Expression of hematopoietic cytokine genes such as SCF, Tpo, Flt3l, Il3, Il6, Il11, GM-CSF (Csf2 – Mouse Genome Informatics), G-CSF (Csf3 – Mouse Genome Informatics), Epo and Osm (see Fig. 6A legend for abbreviations) was evaluated by real-time PCR in isolated niche cells (Fig. 6A). SCF, Tpo, Flt3l, Il6, GM-CSF and Osm expression was detected in placental tissue containing various cell types. When we compared isolated niche cells with placental cells, SCF expression was four times higher in niche compared with placental cells, suggesting that SCF is a potential extrinsic factor regulating HC cluster proliferation. To further characterize placental niche cells, we used flow cytometry to sort endothelial cells and mesenchymal cells from both placenta at 11.5 dpc and the AGM region at 10.5 dpc, as HC clusters are prominent at 10.5 dpc and disappear by 11.5 dpc in the AGM region, whereas HC clusters are apparent at 11.5 dpc in the placenta (Godin and Cumano, 2002). The endothelial cell population was defined as CD31+/CD34+/Kit–/Ter119–/CD45– and the mesenchymal as CD31–/CD34–/Kit–/Ter119–/CD45– (Fig. 6B). When we analyzed expression of SCF, Tpo, Flt3l and Il6 by real-time PCR, SCF expression was detected primarily in both endothelial and mesenchymal cells of the placenta and AGM region (Fig. 6C). SCF expression levels in endothelial cells were 2.5-fold and 8-fold higher than in mesenchymal cells in the placenta and AGM region, respectively. To confirm SCF protein expression, we undertook ELISA analysis and detected SCF only in placental endothelial cells (0.49 ng/103 cells) and in the AGM region (0.95 ng/103 cells) (Fig. 6D). To determine which endothelial cells expressed SCF, we also evaluated co-expression of SCF with CD31 or CD34 at 11.5 dpc. Expression of SCF protein was observed associated with capillary vessels expressing CD31 or CD34, but not in all endothelial cells (Fig. 6E,F). In particular, endothelial cells attached to HC clusters expressed SCF, suggesting that endothelial cells surrounding HC clusters function as niche cells through SCF expression. To investigate whether SCF/Kit signaling regulates HC clusters in the placenta, we administered an anti-Kit neutralizing antibody (ACK2) by intra-cardiac injection to embryos at 10.25 dpc (Czechowicz et al., 2007; Ogawa et al., 1993). Injected embryos were then cultured in a WEC system for 6 hours (Sugiyama et al., 2003; Kulkeaw et al., 2009). Following culture, injected embryos were harvested and their placentas dissociated for flow cytometric analysis. Cells expressing CD31+/CD34+/Kit+ (equivalent to HC clusters) were flow sorted and expression of hematopoietic transcription factors SCL (Tal1 – Mouse Genome Informatics), Runx1, Myb and Gata2 was examined by real-time PCR (Fig. 7). When compared with a control sample from embryos injected with isotype control IgG, expression of Runx1, Myb and Gata2 was significantly downregulated, whereas that of SCL was unchanged. These data suggest that SCF/Kit signaling regulates HC clusters through Runx1, Myb and Gata2.
Origin of placental HC clusters. (A) Sections of placenta were made from ICR mouse embryos at 9.5 dpc. Confocal image of HC clusters expressing CD31/CD34/Kit in the allantois and chorionic plate at 9.5 dpc. (B) CM-DiI was injected into the basal part of allantois of ICR mouse embryos at 8.25 dpc before chorio-allantoic fusion. Injected embryos are shown under a standard stereomicroscope (left) and under a fluorescence stereomicroscope (right). (C) CM-DiI tagged embryos subjected to whole embryo culture for 42 hours are shown under a stereomicroscope (left) and a fluorescence stereomicroscope (right). Upper panels show CM-DiI tagged embryo; lower panels show a non-tagged control. (D) The placenta of a CM-DiI tagged embryo after 42 hours culture was immunostained with anti-Kit antibody. Expression of Kit (green), CM-DiI (red) and TOTO-3 (blue) was observed by confocal microscopy. (E) High magnification view of an HC cluster (boxed area in D).
Origin of placental HC clusters. (A) Sections of placenta were made from ICR mouse embryos at 9.5 dpc. Confocal image of HC clusters expressing CD31/CD34/Kit in the allantois and chorionic plate at 9.5 dpc. (B) CM-DiI was injected into the basal part of allantois of ICR mouse embryos at 8.25 dpc before chorio-allantoic fusion. Injected embryos are shown under a standard stereomicroscope (left) and under a fluorescence stereomicroscope (right). (C) CM-DiI tagged embryos subjected to whole embryo culture for 42 hours are shown under a stereomicroscope (left) and a fluorescence stereomicroscope (right). Upper panels show CM-DiI tagged embryo; lower panels show a non-tagged control. (D) The placenta of a CM-DiI tagged embryo after 42 hours culture was immunostained with anti-Kit antibody. Expression of Kit (green), CM-DiI (red) and TOTO-3 (blue) was observed by confocal microscopy. (E) High magnification view of an HC cluster (boxed area in D).
Analysis of proliferation in placental HC clusters. (A) Morphology of CD31+/CD34+/Kit+/F4/80– cells at 12.5 dpc. (B) The number of CD31+/CD34+/Kit+/F4/80– cells per placenta at 10.5 dpc, 11.5 dpc and 12.5 dpc. (C) Confocal image demonstrating Ki-67 expression in CD31+/CD34+/Kit+/F4/80– cells at 10.5 dpc, 11.5 dpc and 12.5 dpc. (D) The proportion of Ki-67+ cells (Ki-67+ cells/TOTO-3+ cells) at 10.5 dpc, 11.5 dpc and 12.5 dpc.
Analysis of proliferation in placental HC clusters. (A) Morphology of CD31+/CD34+/Kit+/F4/80– cells at 12.5 dpc. (B) The number of CD31+/CD34+/Kit+/F4/80– cells per placenta at 10.5 dpc, 11.5 dpc and 12.5 dpc. (C) Confocal image demonstrating Ki-67 expression in CD31+/CD34+/Kit+/F4/80– cells at 10.5 dpc, 11.5 dpc and 12.5 dpc. (D) The proportion of Ki-67+ cells (Ki-67+ cells/TOTO-3+ cells) at 10.5 dpc, 11.5 dpc and 12.5 dpc.
DISCUSSION
Localization of HC clusters in placenta
To examine mechanisms governing niche cell regulation of HSCs in the mouse placenta, it was necessary to gain insights into their cellular interactions through observation of their morphology and evaluating cells based on marker expression. Previous studies have characterized placental HSCs primarily by flow cytometry, cell culture and transplantation (Alvarez-Silva et al., 2003; Gekas et al., 2005), while immunohistochemical analysis of HC clusters has not been extensively undertaken. We successfully identified Kit+/CD31+/CD34+ HC clusters in the mouse placenta and AGM region. Clusters in the placenta were attached to endothelial cells, as has been observed in the AGM region, a site of HSC generation, suggesting that the placenta might be a site for HSC generation. We determined whether HSC surface markers, such as CD41 (Corbel and Salaun, 2002; Corbel et al., 2005; Ferkowicz et al., 2003; Matsubara et al., 2005; Mikkola et al., 2003a; Mitjavila-Garcia et al., 2002), CD45 (Matsubara et al., 2005; North et al., 2002), EPCR (CD201) (Balazs et al., 2006) and Sca-1 (de Bruijn et al., 2002) were expressed in placental HC clusters using flow cytometry. Although 59.8% of CD31+/CD34+/Kit+/F4/80– cells expressed CD41 when analyzed by flow cytometry (Fig. 2), no strong CD41 signal on HC clusters was detected by immunohistochemstry at 10.5 dpc (Fig. 3). Another group has used an enzyme/antibody technique to stain HC clusters, whereas we employed fluorescent antibodies (Rhodes et al., 2008). The discrepancy and difference in results might be due to the difference of staining method or to the nature of the antibody, i.e. either appropriate for flow cytometry or for immunohistochemistry. We found we could identify HC clusters using a combination of CD31, CD34 and Kit antibodies, regardless of whether they expressed CD41 or CD45. We also investigated HC clusters in the placenta of Runx1–/–, Evi1–/– or Myb–/– mouse embryos. Runx1 is essential for definitive hematopoiesis, and its expression marks the site of de novo generation of hematopoietic progenitors (North et al., 1999; Okuda et al., 1996; Wang et al., 1996). We observed an absence of HC clusters in Runx1–/– placentas. Evi1 is important for HSC generation and expansion, and HSC development in the para-aortic-splanchnopleural mesoderm (P-Sp) region is severely impaired in Evi1–/– embryos (Goyama et al., 2008; Yuasa et al., 2005). Similar to our observations of Runx1–/– embryos, we detected no HC clusters in the Evi1–/– placenta. Myb is essential for HSC maturation and proliferation, and Myb–/– embryos die at 15.5 dpc from impaired definitive hematopoiesis in the fetal liver, although primitive hematopoiesis appears normal (Mucenski et al., 1991). In contrast to Runx1–/– or Evi1–/– placenta, we observed HC clusters in the Myb–/– placenta, and these clusters were larger than those seen in wild-type animals. Although Myb–/– embryos can form endothelial sheet, they fail to generate hematopoietic cells in vitro (Mukoyama et al., 1999). Taken together with our result, they might lack the potential of differentiation from HC clusters to relatively mature hematopoietic cells. Overall, abnormal HC cluster formation seen in knockout mouse embryos indicates that the placenta plays an important role in that process.
Cytokine expression in niche cells. (A) Relative expression of cytokine genes in niche cells isolated by LCM indicated as `Niche' (black bar), compared with placental tissue indicated as `Bulk' (white bar). SCF, stem cell factor; TPO, thrombopoietin; FLT3-L, Flt3-ligand; IL3, interleukin 3) IL6, interleukin 6; IL11, interleukin 11; GM-CSF, granulocyte-macrophage colony stimulating factor; G-CSF, granulocyte colony stimulating factor; EPO, erythropoietin; OSM, oncostatin M. (B) Sorting by flow cytometry of endothelial (CD31+/CD34+/Kit–/Ter119–/CD45–) and mesenchymal cells (CD31–/CD34–/Kit–/Ter119–/CD45–) from the placenta at 11.5 dpc and the AGM region at 10.5 dpc. (C) Relative expression of SCF, TPO, Flt3l and Il6 genes in placental endothelial and mesenchymal cells at 11.5 dpc placenta and the AGM region at 10.5 dpc. (D) Expression of SCF protein (ng/103 cells) measured by ELISA in placental endothelial and mesenchymal cells at 11.5 dpc and the AGM region at 10.5 dpc. (E-H) Placenta sections were made from ICR mouse embryos at 11.5 dpc, stained with antibodies and observed under confocal microscopy. The antibody combination is as follows. (E,F) CD31 (red), SCF (green) and TOTO-3 (blue). (G,H) CD34 (red), SCF (green) and TOTO-3 (blue). Confocal images demonstrate placental localization of SCF at 11.5 dpc. (F,H) Higher magnification view of boxed areas in E and G, respectively. (E) SCF is expressed in endothelial cells surrounding HC clusters that express CD31. (G) SCF is expressed in HC clusters in addition to endothelial cells that express CD34.
Cytokine expression in niche cells. (A) Relative expression of cytokine genes in niche cells isolated by LCM indicated as `Niche' (black bar), compared with placental tissue indicated as `Bulk' (white bar). SCF, stem cell factor; TPO, thrombopoietin; FLT3-L, Flt3-ligand; IL3, interleukin 3) IL6, interleukin 6; IL11, interleukin 11; GM-CSF, granulocyte-macrophage colony stimulating factor; G-CSF, granulocyte colony stimulating factor; EPO, erythropoietin; OSM, oncostatin M. (B) Sorting by flow cytometry of endothelial (CD31+/CD34+/Kit–/Ter119–/CD45–) and mesenchymal cells (CD31–/CD34–/Kit–/Ter119–/CD45–) from the placenta at 11.5 dpc and the AGM region at 10.5 dpc. (C) Relative expression of SCF, TPO, Flt3l and Il6 genes in placental endothelial and mesenchymal cells at 11.5 dpc placenta and the AGM region at 10.5 dpc. (D) Expression of SCF protein (ng/103 cells) measured by ELISA in placental endothelial and mesenchymal cells at 11.5 dpc and the AGM region at 10.5 dpc. (E-H) Placenta sections were made from ICR mouse embryos at 11.5 dpc, stained with antibodies and observed under confocal microscopy. The antibody combination is as follows. (E,F) CD31 (red), SCF (green) and TOTO-3 (blue). (G,H) CD34 (red), SCF (green) and TOTO-3 (blue). Confocal images demonstrate placental localization of SCF at 11.5 dpc. (F,H) Higher magnification view of boxed areas in E and G, respectively. (E) SCF is expressed in endothelial cells surrounding HC clusters that express CD31. (G) SCF is expressed in HC clusters in addition to endothelial cells that express CD34.
Altered Runx1, Myb and Gata2 expression in CD31+/CD34+/Kit+ cells following inhibition of placental SCF/Kit signaling. Intra-cardiac injection of ACK2, a neutralizing antibody, was used to block Kit receptor function in 10.25 dpc mouse embryos followed by whole-embryo culture for 6 hours. CD31+/CD34+/Kit+ cells were flow sorted from placenta of injected embryos and analyzed by real-time PCR (*P<0.05).
Altered Runx1, Myb and Gata2 expression in CD31+/CD34+/Kit+ cells following inhibition of placental SCF/Kit signaling. Intra-cardiac injection of ACK2, a neutralizing antibody, was used to block Kit receptor function in 10.25 dpc mouse embryos followed by whole-embryo culture for 6 hours. CD31+/CD34+/Kit+ cells were flow sorted from placenta of injected embryos and analyzed by real-time PCR (*P<0.05).
In the placental vasculature, the presence of HC clusters composed of up to ten cells has been reported previously (Ottersbach and Dzierzak, 2005; Rhodes et al., 2008). However, we identified larger HC clusters, comprising more than 30 cells, as well as surrounding cells by using thick placental cryosections (20 μm) and confocal microscopy. This methodology might also be useful to investigate interactions between HSCs and surrounding cells in other hematopoietic organs.
Flow cytometric analysis of placental cells
Although we visualized HC clusters by immunohistochemistry, HC clusters expressing CD31, CD34 and Kit may contain hematopoietic progenitors. CD34+/Kit+ cells in the placenta at E12.5 reportedly contain HSCs and hematopoietic progenitors (Gekas et al., 2005). This finding suggests that a combination of CD31, CD34 and Kit antibodies is sufficient to identify HC clusters by morphology, but not specific for HSCs. Therefore, our clusters probably contain both HSCs and progenitors. Although flow cytometry could be employed to purify these two cell populations, those preparations could be contaminated by circulating hematopoietic cells or other cell types. It has been reported that F4/80+ macrophages populate the placental mesenchyme (Rhodes et al., 2010). In agreement, we observed some single F4/80+ cells in the mesenchyme, in addition to c-Kit+/F4/80+/+ cells in circulation in the placenta (Fig. 3). Therefore, after removing F4/80+ cells from the CD31+/CD34+/Kit+ population, we considered that the remaining clusters contained primarily HSCs and hematopoietic progenitors. Our data suggest that a combination of CD31, CD34 and Kit antibodies can be employed to identify HC clusters, regardless of contamination by mature macrophages. By combining CD41 with CD31, CD34 and Kit to sort HC clusters by flow cytometry, we may be able to purify HSCs from this population.
Origin of placental HSCs
Although the YS, AGM region and FL are well recognized organs for hematopoiesis, the small number of HSCs generated in the YS and AGM region cannot completely account for the number of HSCs in FL prior to HSC expansion. In addition, there is a 2-day time lag between HSC generation in the AGM and initiation of HSC expansion in FL. These observations suggest the presence of another hematopoietic site for HSC generation to fill a time gap between AGM region and FL (Kumaravelu et al., 2002). In avian embryos, quail-chick grafting experiments have demonstrated that the allantois (which is equivalent to mammalian placenta) generates definitive hematopoietic cells de novo (Caprioli et al., 1998; Caprioli et al., 2001). It has also been reported that mouse placenta contains HSCs and hematopoietic progenitors (Gekas et al., 2005; Ottersbach and Dzierzak, 2005). The hematopoietic potential of the allantois and chorion isolated prior to establishment of circulation and their fusion has been studied in the mouse placenta (Corbel et al., 2007; Zeigler et al., 2006). These studies showed that both the allantois and chorion exhibit myelo-erythroid potential, implying that definitive hematopoiesis occurs in the placenta. The presence of myelo-erythroid and B- and T-cell progenitors in the placenta of Ncx1–/– mouse embryos, which lack a heartbeat and therefore input from circulating cells to the placenta, supports the notion that HSCs are autonomously generated (Rhodes et al., 2008). However, in vivo experiments to examine the origin of HSCs in mouse placenta has not been performed owing to the difficulty in manipulating embryos within a thick uterine membrane. To overcome this problem, we used a WEC system, enabling us to follow events outside the uterus in vivo (Kulkeaw et al., 2009). When four embryos injected with CM-DiI at 8.25 dpc were cultured in WEC for 42 hours, all developed an umbilical cord and placenta. Here, we determined the origin of HC clusters in the placenta by tagging the allantois at 8.25 dpc with CM-DiI and culturing injected mouse embryos. All Kit+ cell aggregates were CM-DiI+, and no Kit+ aggregate was DiI-negative, although some single CM-DiI–/Kit+ cells were observed (data not shown). It is possible that not all allantoic cells were CM-DiI-tagged and that non-tagged cells gave rise to single Kit+ cells, given the technical difficulty of tagging all cells. We injected CM-DiI into the basal part of allantois, implying that HC clusters are originated from this part. It is also possible that chorionic cells per se may give rise to Kit+ cells: chorion reportedly has a potential to generate myeloid and definitive erythroid cells (Corbel et al., 2007; Zeigler et al., 2006). Thus, although some HC clusters may have been derived from chorion, it is more likely that the mouse placenta does autonomously generate HSCs and that the allantois is at least a major source of placental HSCs. As shown in Fig. 4A, HC clusters first form cell aggregates. Although several reports suggest that HC clusters in the AGM region are derived from endothelial cells expressing VE-cadherin (Dzierzak and Speck, 2008), the HC clusters in the placenta probably did not originate from endothelial cells. Interestingly, Fraser et al. demonstrated that VE-cadherin is also expressed in HC clusters in the AGM region, indicating that VE-cadherin is not a specific marker of endothelial cells (Fraser et al., 2003). It may be further necessary to evaluate the origin of HC clusters both in the AGM region and the placenta in the future.
Niche regulation of placental HSCs
HSCs are regulated by niche cells surrounding HSCs. However, it remains unclear how embryonic HSCs are regulated by niche cells. In the bone marrow, expression of niche cell markers such as N-cadherin and CXCL12 enables their isolation by flow cytometry and has contributed greatly to an understanding of niche regulation (Arai and Suda, 2007; Sugiyama et al., 2006). Conversely, investigation of the placental niche has been impeded by a lack of markers for placental niche cells. To address this issue, we isolated niche cells surrounding HC clusters in placenta by LCM. Using this system, we obtained niche cells despite the lack of markers. HC clusters were found inside blood vessels, suggesting that niche cells are mostly composed of endothelial cells. In addition, we sorted out both endothelial and mesenchymal cells, and performed real-time PCR with SCF gene (Fig. 6). Our gene expression analysis revealed that SCF is predominantly expressed in niche cells, and protein expression analysis suggested that SCF is predominantly expressed in niche endothelial cells. In agreement, we found that SCF is predominantly expressed in endothelial cells, in particular cells surrounding HC clusters by immunostaining. Interestingly, SCF was expressed in clusters as well as in endothelial cells, implying an autocrine mechanism. It would be of interest to investigate whether SCF plays a role in specification as well as niche development. To understand the role of the SCF/Kit signal in regulating placental HSCs, we performed a loss-of-function experiment in vivo to inhibit SCF/Kit signaling in the mouse placenta using a WEC system with 10.25 dpc embryos – a stage suitable for manipulation. SCL is not required for HSC development once commitment to hematopoietic lineages has occurred (D'Souza et al., 2005; Mikkola et al., 2003b; Robb et al., 1995; Shivdasani et al., 1995). However, Gata2 is crucial for definitive hematopoiesis and functions in the generation and expansion of HSCs in the AGM region (Ling et al., 2004; Lugus et al., 2007; Tsai et al., 1994). Our study confirmed that expression of Runx1, Myb and Gata2 was significantly downregulated compared with control samples in Kit loss-of-function analyses but SCL expression was not altered. Kit receptor activation plays a major role in regulating survival, proliferation and self-renewal of HSC phenotypes (Kent et al., 2008), but how SCF/Kit signal regulates Runx1, Myb and Gata2 remains unclear. In addition to SCF/Kit signaling, other signals may regulate HC clusters. SCF secreted by niche cells may modulate proliferation of CD31+/CD34+/Kit+ cells between 10.5 and 12.5 dpc, as shown in Fig. 5, although this proliferation might be due to an accumulation of the hematopoietic cells in the placental vasculature as this organ increases in size. Decrease of Ki-67 positive cells might be due to the downregulation of SCF by niche cells.
IL3 reportedly increases the number of HSCs in the AGM region (Robin et al., 2006). However, these authors demonstrated that IL3 has no effect on HSC activity in the placenta at 10.5 dpc, an observation compatible with our data showing that IL3 is not expressed in placental niche cells (Fig. 6A). Hedgehog, BMP4, bFGF and VEGF signals from the surrounding micro-environment are required for mesodermal cells to commit to hematopoietic cells (Dzierzak and Speck, 2008). In the AGM region, location plays a role in regulating HSC generation: ventral tissues induce AGM HSCs, whereas dorsal tissues suppress them (Peeters et al., 2009). Hedgehog protein(s) have been identified as positive effectors that increase the number of AGM HSCs (Peeters et al., 2009). Moreover, there is greater expansion of placental HSCs from 11.5 dpc to 12.5 dpc than of hematopoietic progenitors at this site, suggesting that other signals in the placental niche probably inhibit HSC differentiation (Gekas et al., 2005).
This is the first report to identify and examine the function of cytokine signals regulating HSCs in the mouse placenta. Our study is also evidence that LCM is a useful tool with which to study molecular mechanisms in specific cell aggregates. Recently, it was demonstrated that human placenta contains HSCs and that stromal cells (derived from human placenta) could support hematopoiesis (Robin et al., 2009). Clarifying how the niche regulates HSCs in the placenta could lead to an understanding of how to manipulate HSC generation from ES/iPS cells and, thus, be applicable to future clinical applications.
Acknowledgements
This research was partially supported by a Grant-in-Aid for Exploratory Research; by the Project for Realization of Regenerative Medicine; by Special Coordination Funds for Promoting Science and Technology of the Ministry of Education, Science, Sports and Culture; and by a Bilateral Program of the Japan Society for the Promotion of Science. We thank the Research Support Center, Graduate School of Medical Sciences, Kyushu University for technical support; Drs M. Ogawa, C. Meno and S. Oki for helpful discussion; Dr R. Jones for critical reading of our manuscript; and Drs K. Kulkeaw and T. Inoue for technical support in our laboratory.
The authors declare no competing financial interests.
Supplementary material
References
Competing interests statement