Embryonic Hematopoietic Progenitor Cells Reside in Muscle before Bone Marrow Hematopoiesis

In mice, hematopoietic cells home to bone marrow from fetal liver prenatally. To elucidate mechanisms underlying homing, we performed immunohistochemistry with the hematopoietic cell marker c-Kit, and observed c-Kit(+) cells localized inside muscle surrounding bone after 14.5 days post coitum. Flow cytometric analysis showed that CD45(+) c-Kit(+) hematopoietic cells were more abundant in muscle than in bone marrow between 14.5 and 17.5 days post coitum, peaking at 16.5 days post coitum. CD45(+) c-Kit(+) cells in muscle at 16.5 days post coitum exhibited higher expression of Gata2, among several hematopoietic genes, than did fetal liver or bone marrow cells. Colony formation assays revealed that muscle hematopoietic cells possess hematopoietic progenitor activity. Furthermore, exo utero transplantation revealed that fetal liver hematopoietic progenitor cells home to muscle and then to BM. Our findings demonstrate that hematopoietic progenitor cell homing occurs earlier than previously reported and that hematopoietic progenitor cells reside in muscle tissue before bone marrow hematopoiesis occurs during mouse embryogenesis.


Introduction
In mice, the site of embryonic hematopoiesis changes over an approximately 20-day gestation period. Primitive hematopoiesis begins in the yolk sac (YS), producing mainly primitive erythroid cells at 7.5 days post coitum (dpc). This process is transient and decreases by 12.5 dpc [1]. Adult-type hematopoiesis, termed definitive hematopoiesis, is characterized by the appearance of cells with definitive erythroid, lymphoid and hematopoietic stem cell (HSC) potentials. Definitive myelo-erythroid progenitor cells appear in the YS around 8.25 dpc and are then seeded to fetal liver (FL) [2]. HSCs are likely generated in the YS, intra-embryonic para-aorticsplanchnopleural mesoderm/Aorta-Gonad-Mesonephros (AGM) region, and placenta [3][4][5][6][7]. Previously, we reported that circulating c-Kit-positive hematopoietic cells (HCs) home to FL [8]. Both morphological observation and in vitro experiments indicated that FL itself does not produce hematopoietic stem/progenitor cells (HSPCs) but is rather colonized by HCs originating elsewhere after 9.5 dpc [9][10][11][12]. Taken together, HSPCs likely circulate and home to FL, where their number dramatically increases as definitive erythropoiesis occurs extensively at mid-gestation [11][12][13]. After HSC expansion in FL, HSCs home to the fetal spleen, where they differentiate from 13.5 to 14.5 dpc [14]. As HSCs with reconstitution ability are first detected in bone marrow (BM) at 17.5 dpc, they likely home to this site to start life-long hematopoiesis [15].
It remains unclear why hematopoietic sites dramatically shift during embryogenesis. Previously, we demonstrated that Dlk-1-positive hepatoblasts function as niche cells to regulate HSC homing and differentiation by secretion of extra-cellular matrix (ECM) proteins and cytokines, such as erythropoietin (Epo) and stem cell factor (SCF) [16,17]. ECMs, which typically function in cell adhesion, cell-to-cell communication and differentiation, often partner with integrins in these processes [18][19][20]. In FL of beta-1 integrin (fibronectin receptor beta, CD29) knockout chimeric embryos, beta-1 integrin-positive HCs homed to the FL, while those lacking beta-1 integrin did not [19,21]. We also demonstrated that HSPCs and erythroid cells in FL express beta-1 integrin, while circulating erythroid cells do not, suggesting that beta-1 integrin regulates FL homing [21,22]. The ECM protein fibronectin is a beta-1 integrin ligand and reportedly promotes homing ability of HCs in vitro [22]. Given that fibronectin is highly expressed in FL, it likely regulates homing of HCs expressing beta-1 integrin.
Although mechanisms underlying HC homing to FL from the circulation have been investigated, how cells home from the FL to embryonic BM is not fully understood. Fetal BM forms by 15.5 dpc [15], but HSC activity is not detected there until 17.5 dpc, suggesting that HSCs remain in the FL or other tissues. Here, to investigate mechanisms underlying fetal BM homing, we performed immunohistochemistry of embryonic bones and surrounding tissues. We observed c-Kit-positive HCs residing in muscle tissue surrounding bones late in gestation. In addition, muscle HCs showed HPC ability, as determined by colony formation assays. These findings suggest that HPCs reside in muscle tissue before homing to the fetal BM.

Materials and Methods
Mice C57BL/6 mice (Nihon SLC, Hamamatsu, Japan and Kyudo, Tosu, Japan) and enhanced green fluorescence (EGFP) Tg mice (Research Institute for Microbial Diseases, Osaka University, Osaka, Japan) were used in this study. Animals were handled according to Guidelines for the Care and Use of Laboratory Animals of Kyushu University. This study was approved by the Animal Care and Use Committee, Kyushu University (Approval ID: A25-119-1).

Cell preparation
Left and right femurs and muscle tissues surrounding those structures of 14.5 to 19.5 dpc C57BL/6 mouse embryos were used to obtain single cell suspensions. Tissues were trimmed from femurs and incubated with 3 mg/ml collagenase in medium containing 10% fetal bovine serum (FBS) for 20 minutes at 37°C. Cells were then filtered through 70 μm nylon cell strainers (BD Biosciences, San Jose, CA). BM cells were flushed out with PBS containing 2% FBS using 29 to 32G needles with syringes (TERUMO, Tokyo, Japan) and filtered through 40 μm nylon cell strainers (BD Biosciences). FLs from 14.5 dpc and 16.5 dpc embryos were dissected out, and single cell suspensions were prepared by digesting tissues with 3 mg/ml collagenase in medium containing 10% FBS for 20 minutes at 37°C. Cells were then filtered through 40 μm nylon cell strainers (BD Biosciences). BM cells from femurs and tibias of 3-month-old adult C57BL/6 mice were dissected out and then flushed out with PBS containing 2% FBS from 27G needles and syringes (TERUMO). Cells were then filtered through 40 μm nylon cell strainers (BD Biosciences). Fetal blood was obtained from 16.5 dpc embryos. Blood cells were then washed 3 times with PBS containing 2% FBS.

Immunohistochemistry
Femurs and surrounding muscle tissues of 14.5 to 19.5 dpc embryos were dissected and fixed in 2% paraformaldehyde (PFA) in PBS at 4°C for overnight. Fixed tissues were then washed 3 times using PBS, equilibrated with 30% sucrose in PBS, embedded in OCT compound (SAKURA, Tokyo, Japan) and frozen in liquid nitrogen. Tissues were cut into 20 μm slices using a Leica CM1900 UV cryostat (Leica Microsystems, Tokyo, Japan), transferred to glass slides (Matsunami glass, Osaka, Japan) and dried thoroughly. Colony formation and high proliferative potential colony forming cell (HPP-CFC) assays OP9/OP9 Delta1 co-culture OP9/OP9 Delta1 cell lines were maintained in α-MEM medium (Wako Pure Chemical Industries, Osaka, Japan) supplemented with 20% FBS and 0.1% penicillin/streptomycin at 37°C in 5% CO 2 . Cells were passaged every 3-4 days. For co-culture experiments, muscle tissue CD45 (+) c-Kit(+) cells obtained from 16.5 dpc C57BL/6 mouse embryos were suspended in α-MEM medium supplemented with 20% FBS, 5 ng/ml Flt3-L (PEPROTECH, NJ, USA), 1 ng/ml IL-7 (PEPROTECH) and 0.1% penicillin/streptomycin and cultured with either OP9 or OP9 Delta1 cell lines. Cells were passaged every 3-4 days. On day 16, surface expression of CD19 and B220 (B lymphoid markers) was analyzed in the case of cells cultured with the OP9 line, while surface expression of the T lymphoid markers CD4 and CD8 was analyzed for cells cultured with OP9 Delta1 line by flow cytometry.

Organ culture
Organ culture of fetal muscle tissue was performed as reported previously with minor modifications [5]. Muscle tissues surrounding the femur at 16.5 dpc were placed onto sterilized 0.65 μm filter membranes (Millipore, Bedford, UK) and cultured in StemSpan (STEMCELL Technologies) supplemented with 0.1% penicillin/streptomycin at 37°C in 5% CO 2 . After 72 hours, tissues were collected and single cell suspensions were prepared by incubation in 3 mg/ml collagenase in medium containing 10% FBS for 15 minutes at 37°C. Colony formation assays were then performed as described above.
Exo utero surgical transplantation C57BL/6 pregnant mice at 14.5 dpc and 16.5 dpc were anaesthetized using 1.0% isoflurane and an animal anesthetizer device (MK-AT210D, Muromachi Kikai Co., LTD, Tokyo, Japan). Mice were then placed on the plate warmed to 37°C, their abdominal area was shaved, and the skin and abdominal wall were incised. The uterine wall was then cut on the side opposite the placenta. Glass needles were prepared from glass capillary tubes (Narishige, Tokyo, Japan) using a micropipette puller (PN-30, Narishige). Under a microscope, 3 μl of a cell suspension containing 1.

Statistical analysis
Results are expressed as the mean ± standard deviation (SD) in all analyses. Paired samples were compared using Student's t-test.

Characterization of hematopoietic progenitor cells in muscle tissue surrounding embryonic femur
To evaluate migration properties of HPCs, we investigated their localization in late-gestation mice. We carried out immunohistochemistry using cryosections of fetal femur and surrounding tissues at 14.5 to 19.5 dpc. Hematoxylin-eosin staining indicated normal structure of the femur BM cavity ( To further investigate surface marker expression of c-Kit(+) cells, we performed flow cytometric analysis of single cell suspensions of muscle tissues prepared from 14.5 to 19.5 dpc embryos using the pan-leukocyte marker CD45 (also expressed on fetal HPCs) [24], the HPC marker c-Kit, and the HSC marker Sca-1 [25]. CD45(+) c-Kit(+) cells were fractionated into 2 subpopulations based on Sca-1 expression, an analysis enabling us to distinguish HSCs from HPCs (Figs 2A and S3) and BM (S4 Fig) [26,27]. To investigate the kinetics of CD45(+) c-Kit (+) Sca-1(+) cells compared to CD45(+) c-Kit(+) Sca-1(-) cells in embryonic muscle tissue and BM, we determined the total number of cells in these subpopulations per two femurs and per two sets of femur-surrounding muscle tissue at mid-to late-gestation ( Fig 2B). From 14.5 to 19.5 dpc, the number of CD45(+) c-Kit(+) Sca-1(-) cells decreased 17.5-fold in muscle tissue (14.5 dpc: 1241±624 cells; 19.5 dpc: 71.0±17.5 cells), while it increased 2.9-fold in BM (14.5 dpc: 109±100 cells; 19.5 dpc: 312±166 cells) (Fig 2B). During this period, CD45(+) c-Kit(+) Sca-1(-) cells were more abundant in muscle tissue than in BM, while they became more abundant in BM by 19.5 dpc. The number of CD45(+) c-Kit(+) Sca-1(+) cells in muscle tissue peaked at 15.5-16.5 dpc and declined gradually by 19.5 dpc (Fig 2C). Likewise, the number of CD45(+) c-Kit(+) Sca-1(+) cells in BM gradually increased and peaked at 19.5 dpc, suggesting that BM hematopoiesis had been initiated at this stage. This trend was similar to one reported by Christensen et al, showing that long-term repopulating (LTR)-HSC activity is first detected in BM at 17.5 dpc [15]. To further characterize muscle HCs at 16.5 dpc, we analyzed surface marker expression by flow cytometry. We observed that muscle HCs express other hematopoietic surface markers, such as CD34 (4.22±2.75%), CD150 (2.85±1.34%) and EPCR (6. To examine cell morphology, CD45(+) c-Kit(+) Sca-1(-) cells and CD45(+) c-Kit(+) Sca-1 (+) cells were sorted from 16.5 dpc muscle tissues, stained with May-Grünwald Giemsa solution, and observed by microscopy. Both populations of cells appeared morphologically immature, as defined by a high nucleus/cytoplasm ratio and blue-colored cytoplasm ( Fig 2D). Overall, these findings indicate that CD45(+) c-Kit(+) HCs localize in muscle surrounding bone and are more abundant in muscle than in BM between 14.5 and 17.5 dpc.

Muscle CD45(+) c-Kit(+) cells possess colony forming capacity
To evaluate their hematopoietic potential, we sorted CD45(+) c-Kit(+) muscle cells at 16.5 dpc and performed colony formation assays. To do so, we seeded 1,000 sorted cells in a 35-mm dish and determined the number of colonies on day 14. As shown in Fig 4A,   Next, we evaluated the influence of the circulation on embryonic muscle tissue using organ culture [5]. Using the experimental design shown in Fig 4C, we performed colony formation assays after 72 hours of organ culture, and 50,000 cells were seeded per dish. After 72 hours we assessed the surface phenotype of cells and both CD45(+) c-Kit(+) Sca-1(+) and CD45(+) c-Kit(+) Sca-1(-) cells were observed (S8 Fig). At the end of the culture period, we found that 60.5% of cells incorporated propidium iodide (a marker of dead cells), and 172 cells among 50,000 cells of muscle after culture expressed both CD45 and c-Kit. On day 14, we observed formation of hematopoietic colonies (27.7±2.08 colonies) consisting of CFU-G (9.70±2.52 colonies), CFU-M (5.00±1.73 colonies), CFU-GM (13.0±1.00) and CFU-GEMM (0.70±0.58 colonies) (Fig 4D). Overall, colony formation analysis revealed that muscle HCs possesses HPC activity.

Fetal CD45(+) c-Kit(+) cells migrate from liver to muscle to BM
To investigate migration of fetal CD45(+) c-Kit(+) cells, we sorted EGFP(+) CD45(+) c-Kit(+) cells from FL of EGFP transgenic (Tg) mouse embryos at 14.5 dpc or from muscle tissue of Tg embryos at 16.5 dpc and transplanted them into corresponding tissues of C57BL/6 recipient embryos using exo utero surgical techniques [30]. After 24 hours, we undertook flow cytometry and/or immunohistochemistry to monitor EGFP(+) donor cells in muscle tissue or BM, based on the procedure shown in Fig 5A. EGFP(+) CD45(+) c-Kit(+) cells were sorted from FL for transplantation (Fig 5B, upper). Flow cytometric analysis showed that EGFP(+) CD45(+) c-Kit (+) cells were present in muscle tissue of 15.5 dpc recipients (Fig 5B, middle). Among EGFP(+) CD45(+) c-Kit(+) cells, 17.6±1.13% were Sca-1(+) and 82.4±1.13% were Sca-1(-) (S9A Fig). We also assessed expression of hematopoietic transcription factors in EGFP(+) donor cells 24 hours after transplantation. EGFP(+) donor cells that had migrated expressed Gata2, Tal1, Myb and Runx1 but not Mecom (S9B Fig). Transplanted cells were also observed in other areas of muscle tissue that surround the tibia and humerus (S9C Fig). Immunostaining of muscle tissue surrounding femurs also revealed the presence of EGFP(+) donor cells (Fig 5B, lower). We then sorted EGFP(+) CD45(+) c-Kit(+) cells from muscle for transplantation into recipient muscle (Fig 5C, upper). Immunostaining of femur showed that EGFP(+) donor cells were present in BM (Fig 5C, lower). EGFP(+) donor cells also remained in muscle tissue in all recipients (data not shown). We further investigated migration of muscle CD45(+) c-Kit(+) cells from FL to fetal BM by tracking cells for up to 72 hours. As shown in S9D and S9E Fig, Qdot585-labeled donor cells transplanted into FL were detected in BM 72 hours later, indicating that they had migrated into fetal BM. Table 1 summarizes data relevant to exo utero surgery transplantation. Taken together, these observations indicate that during embryogenesis CD45(+) c-Kit(+) cells migrate from FL to muscle tissue and thereafter move from muscle tissue to BM.
We then sorted EGFP(+) CD45(+) c-Kit(+) cells of EGFP transgenic mouse embryos from FL at 14.5 dpc and from muscle tissue at 16.5 dpc and transplanted cells into corresponding tissues of C57BL/6 mouse recipients utilizing exo utero transplantation. In some experiments, CD45(+) c-Kit(+) cells were sorted from FL of C57BL/6 mouse embryos at 14.5 dpc and labeled with Qdot585 dye. After 24 hours, the presence of EGFP(+) or Qdot585(+) cells was evaluated using flow cytometry and/or immunohistochemistry. Transplantation results are summarized.

Discussion
Here, we demonstrate that HPCs reside in embryonic muscle tissue before BM hematopoiesis. These muscle HPCs migrate from FL and move to the BM, suggesting that HPC homing occurs earlier than previously reported and that these HPCs reside in muscle for a period as they make their migration.
The fetal blood circulation system becomes functional by 14.5 dpc and is a potential source of HCs [15]. To exclude the possibility that blood contamination is the source of embryonic muscle HPCs observed here, we performed CD45, c-Kit and CD31 immunostaining of serial sections of muscle tissues at 14.5 dpc and 16.5 dpc (Fig 1C) and found that most CD45(+) c-Kit(+) cells were localized in muscle tissue and their absolute number was greater outside of blood vessels in muscle tissue than in inside vessels. (Fig 1D). We also performed colony formation assays with muscle tissue cells after 72 hours of organ culture to exclude the influence of circulation. That analysis confirmed the existence of HPCs in muscle tissues at 16.5 dpc   [15]. Likewise, our flow cytometry analysis indicates that fetal blood contains CD45(+) c-Kit(+) cells but at a low frequency (7.41x10 -4 ±5.94x10 -4 % of living cells) at 16.5 dpc (Figs 1C and 1D and S10). We conclude that muscle HPCs observed here are not likely due to contamination by blood. Gene expression analysis indicated that HPCs in muscle express hematopoietic transcription factor genes such as Gata2, Tal1, Myb and Runx1, but (with the exception of Gata2) at levels lower than in HPCs in other tissues. This finding implies that CD45(+) c-Kit(+) cells in embryonic muscle possess decreased hematopoietic potential, although they show similar surface marker expression and are potentially in a transition stage between cells leaving the FL and those in the BM. To assess their hematopoietic potential, we performed colony formation assays. Previously, we reported that CD31(+) CD34(+) c-Kit(+) cells derived from intra-aortic clusters at 10.5 dpc generated 113.4 colonies per 1,000 cells; here, we observed that 1,000 CD45 (+) c-Kit(+) cells from 16.5 dpc muscle gave rise to 66.2 colonies, indicative of a relative decrease in hematopoietic potential [34]. Interestingly, muscle HPCs lack expression of Mecom, an oncogenic transcription factor regulating embryonic HPC activity [35]. Mecom knockout mice exhibit impaired hematopoietic and vascular development and die at 10.5 dpc [36]. The availability of Mecom-IRES-GFP reporter mice allows fractionation by flow cytometry of BM cells negative for lineage markers and positive for Sca-1 and c-Kit (LSK cells) as GFP (+) (showing high Mecom expression) and GFP(-) (showing low or no Mecom expression) cells [35]. Colony formation assays showed that sorted GFP(+) and GFP(-) LSK cells generated hematopoietic colonies, but that a greater number was generated from GFP(+) cells [35]. This observation suggests that Mecom expression levels define hematopoietic potential and supports the idea that CD45(+) c-Kit(+) cells in muscle exhibit decreased hematopoietic potential. Low Myc expression in muscle HPCs reported here also implies decreased hematopoietic potential and a more quiescent status, as Myc stimulates cell proliferation and controls the balance of self-renewal activity, quiescence and differentiation [37][38][39]. An interesting exception is our observation that muscle HPCs express Gata2 at higher levels than do HPCs from other tissues. When we quantified Gata2 expression using the comparative Ct method, Gata2 was the most highly expressed hematopoietic transcription factor in muscle HPCs (S6 Fig). Gata2 knockout embryos die by 11.5 dpc due to severe anemia, and cells from their YS and AGM regions exhibit relatively low hematopoietic potential, as assessed by colony formation [40]. Gata2 deletion in VE-cadherin expressing cells decreases the number of HPCs in the YS, AGM region and FL and impairs intra-aortic cluster formation. On the other hand, Gata2 overexpression in mouse BM HSCs decreases the number of both CFU-C and CFU-S and impairs HSC activity [41]. Myb knockout embryos also die around 15.5 dpc due to severe anemia [42], and Myb overexpression in mouse ES cells impairs hematopoietic differentiation in vitro [43]. Thus, decreased hematopoietic potential observed in muscle HPCs might be due to the relative expression of Gata2 (high) to Myb (low).
It remains unclear why migrating HPCs transiently reside in muscle. Nonetheless, exposure to an embryonic muscle environment may be required to modulate cells' hematopoietic potential, possibly through signals that regulate appropriate levels of hematopoietic transcription factors. Between 10 and 15 dpc, FL actively develops and is colonized by HCs to become a major hematopoietic organ [44]. Then, as embryos develop and hepatogenesis becomes more active between 15 dpc and post-natal stages, the FL hematopoietic compartment likely becomes smaller. As the FL environment becomes rich in cytokines and ECM proteins that accelerate hematopoietic differentiation, HPCs may exit the FL to avoid stimulation by hematopoietic differentiation factors. Since the vascular structure of fetal BM is not yet well-developed at midgestation [45], BM may not constitute an environment that can sustain hematopoietic cells at 16.5 dpc, implying that HPCs require other environments to pause. Given that HPCs survive during an organ culture step (Fig 4D), an alternate hypothesis is that muscle tissues secrete factors required to sustain HPCs at 16.5 dpc.
Homing mechanisms used by HCs to move from FL to fetal BM are not fully understood. Here, we used an exo utero technique to show that HPCs found in muscle tissue surrounding bones migrate from FL to muscle and then to BM (Fig 5). Future analysis should address developmental implications of this migration to expand our knowledge of how HCs are regulated.