Avian pluripotent haemopoietic progenitor cells: detection and enrichment from the para-aortic region of the early embryo

The development of the haemopoietic system involves the successive colonization of the haemopoietic rudiments by extrinsic stem cells (Metcalf and Moore, 1971; Le Douarin and Jotereau, 1973; Le Douarin et al., 1975). In mammals, it is still assumed that the embryonic origin of the haemopoietic stem cells (HSC) is the yolk sac (Metcalf and Moore, 1971). In birds, it was clearly demonstrated by means of chimaeras composed of an embryo associated with a xenogenic or allogenic yolk sac that HSC derive from the embryo proper and not from the yolk sac (Dieterlen-Lièvre, 1975; Martin et al., 1978; Lassila et al., 1979, 1982). In the avian embryo, haemopoiesis also occurs in the para-aortic region; between day 3 and day 9 of incubation (E3 and E9) in the chick embryo (Romanov, 1960). Haemopoietic foci develop from progenitors that are known to segregate in situ from the endothelial or/and mesenchymal components at E3-4 (Dieterlen-Lièvre and Martin, 1981).

The differentiation potentialities of E4 para-aortic cells were previously investigated by assaying their colony-form-ing capacities in an in vitro clonal culture system. In agar cultures with chicken serum, monocytic colony-forming cells (M-CFC) were exclusively detected from the paraaortic cell population and not from embryos deprived of their aorta (Cormier et al., 1986). In serum-free plasma clot cultures, we could detect progenitors with various potentialities, unipotent progenitors of the monocytic, granulocytic and erythrocytic lineages and bipotent granulo-monocytic progenitors, M-CFC, G-CFC, BFU-E and GM-CFC, respectively (Cormier and Dieterlen-Lièvre, 1988). The present paper reports the demonstration of progenitors with the dual potentiality to give rise to cells of the myeloid and erythroid lineages. Avian normal pluripotent haemopoietic progenitors have not been described before.

Para-aortic regions from 60 to 120 E4 outbred White Leghorn chick embryos were retrieved as described previously (Cormier et al., 1986). They were dissociated by a 1 h treatment in 0.1% collagenase (Sigma) solution in Ca-Mg-PBS at 37°C, followed by repeated pipetting. Para-aortic fragments from E4 chick embryos are comprised of the thin endothelial layer of the aorta surrounded by loose mesenchyme. Thus, complete dissociation is easily obtained. Nevertheless, the cell suspension was filtered over a nylon tissue to obtain a single cell suspension. Then, the cell suspension was centrifuged for 15 min at 1000 r.p.m. (150 g). The cell pellet was resuspended in alpha-medium (Gibco) and the cells were counted using the trypan blue exclusion test.

For sorting experiments, para-aortic cell suspensions were prepared from 120 embryos to obtain sufficient cell number (2-3×106 cells).

Para-aortic cells were incubated for 30 min at 4°C with the monoclonal antibodies (mAbs). After washing in Hanks’ balanced salt solution containing 3% fetal calf serum (HBSS-FCS), cells were incubated for 30 min at 4°C with goat anti-mouse immunoglobulin (Ig) conjugated with fluorescein isothiocyanate (FITC), or biotinylated and washed twice in HBSS-FCS. The cell suspension was diluted in PBS to a maximum concentration of 106 cells/ml and filtered over a nylon tissue before analysis and sorting on a FACS 440 (Beckton-Dickinson). Dead cells and doublets were eliminated by excluding the ‘high forward light scattering window’ and the ‘low and high orthogonal light scattering windows’. The purity of the sorted populations was about 95%. After sorting, cells were collected in sterile tubes containing 100 μl BSA and centrifuged for 15 min at 1000 r.p.m.. They were resuspended in alpha-medium and counted using the trypan blue exclusion test before plating in plasma clot cultures.

The VI-A2 mAb, which recognizes all the chicken haemopoietic cells except mature B lymphocytes and erythrocytes (Yassine et al., 1989) and the anti-BEN mAb, which recognizes a cell surface molecule of the Ig superfamily shared by selective sets of neurons, bursal epithelial cells, activated T cells and haemopoietic progenitor cells (Pourquié et al., 1990; Corbel et al., 1992a,b), were used.

Para-aortic cells were seeded in plasma clot cultures performed in serum-free medium supplemented with fibroblast-conditioned medium as described previously (Cormier and Dieterlen-Lièvre, 1988). Cultures were performed in duplicate and incubated for 3 days at 40°C in a humidified atmosphere. Cultures were harvested and stained as described previously (Cormier and Dieterlen-Lièvre, 1988). In each experiment, increasing cell numbers were plated (from 500 to 10,000 cells per plasma clot). The frequency of the various progenitors reported in this study was determined from cultures in which colonies were well separated and which corresponded to the limit dilution condition for the development of the myelo-erythroid colonies.

We previously reported that the para-aortic cells cultured in serum-free medium in the presence of fibroblast-conditioned medium (FCM) give rise to three types of colonies, monocytic-, granulocyticand granulo-monocytic colonies (Cormier and Dieterlen-Lièvre, 1988). In this paper, other batches of FCM were used, which induced the development of three other colony types. The first one consisted of colonies composed of immature myeloblasts surrounded by maturing monocytes and/or granulocytes (Fig. 1A). Erythroid colonies also developed; at day 3, they were composed of erythroblasts which degenerated thereafter without reaching haemoglobinized state. The frequency of these colonies, about 70 from 12,500 cells, was similar to the frequency of the BFU-E-derived colonies described previously (Cormier and Dieterlen-Lièvre, 1988). Another colony type was composed of erythroblastic cells and maturing cells of the myelomonocytic lineage. All the combinations were obtained: colonies with erythroblasts and monocytes, colonies with erythroblasts and maturing granulocytes (Fig. 1B) and colonies with erythroblasts, monocytes and granulocytes (Fig. 1C). One colony with erythroblasts, monocytes, granulocytes and thrombocytes was also obtained and is illustrated in Fig. 1D. This last type is extremely rare, since only a single colony developed in all the cultures performed.

Table 1 reports the results of five independent cultures. Each value is the number of colonies which developed in two plasma clots seeded with 12,500 para-aortic cells. Therefore, the frequency of immature progenitors belonging to the myelomonocytic lineage is about 4 for 10,000 cells and the frequency of myelo-erythroid progenitors is about 1 for 10,000 cells.

FACS analysis demonstrated that about 8% of the paraaortic cells were stained by the VI-A2 mAb (Fig. 2), which is specific for chicken haemopoietic cells (Yassine et al., 1989). In three independent cell sorting experiments, the colony-forming capacity of unsorted, VI-A2⁻ and VI-A2⁺ cells was analyzed. As reported in Table 2, all the CFC types were enriched in the VI-A2⁺ population. Unipotent progenitors of the myelomonocytic lineage were enriched by a factor of 4, as compared to the unsorted population. Similarly, immature myeloid progenitors (designated by My-CFC since they give rise to myeloblast colonies) and myelo-erythroid progenitors (designated ME-CFC) were enriched 4-fold in the VI-A2⁺ population. Erythroid progenitors (Eb-CFC, since they give rise to erythroblast colonies) were enriched by a factor of 8, twice as high as the myeloid progenitor enrichment factor.

The anti-BEN mAb recognizes a cell surface molecule of the Ig superfamily (Pourquié et al., 1992) which is expressed by CFC and maturing haemopoietic precursor cells from bone marrow, as we recently demonstrated (Corbel et al., 1992b). Double staining of para-aortic cells in suspension showed that 40% of them are BEN⁺ and these were not stained by the VI-A2 mAb (Fig. 2). Among the 8% of the E4 para-aortic cells that were VI-A2⁺, less than 1% weakly expressed BEN. The BEN⁺ para-aortic cells at this developmental stage are mainly primary sympathetic ganglionic cells, as shown by Pourquié et al. (1990).

These results suggested that most of the haemopoietic cells from the para-aortic region do not express the BEN cell surface molecule. To confirm this, we analysed the haemopoietic differentiation potentialities of both BEN⁺ and BEN⁻ populations in plasma clot cultures with FCM. Table 3 reports the distribution of the different para-aortic CFC in the BEN⁺ or BEN⁻ cell populations sorted in five independent experiments. In each of these experiments, unipotent myeloid CFC were distributed in both BEN⁺ and BEN⁻ cell populations. About 2/3 of them were detected among the BEN⁻ cells, and 1/3 among the BEN⁺ cells. No enrichment of M/G-CFC was obtained in one of the two sorted populations. On the contrary, the My-CFC, Eb-CFC and ME-CFC were selectively detected in the BEN-negative cell population. My-CFC were enriched in two of the five experiments, by factors of 2 and 5, respectively. Erythroid progenitors were enriched about two times. It should be particularly noted for the purpose of this paper that ME-CFC were highly enriched (around 10-fold) in the BEN⁻ cell population, since their frequency reached 2 to 5 for 10,000 cells.

This paper reports the first demonstration of avian normal haemopoietic progenitors with both myelomonocytic and erythroid potentialities. This demonstration is based on the in vitro development of myelo-erythroid colonies from embryonic para-aortic cells. These mixed colonies were identified and counted in cultures in which individual colonies were well separated; only the colonies in which erythroid and myelomonocytic cells were highly mixed were identified as mixed colonies. The para-aortic cell suspensions were filtered over nylon tissue (two times for sorting experiments) and were single cell suspensions, as it could be checked during counting of the cells. The frequency of the mixed colonies in unfractionated cell suspensions was determined in cultures seeded with a low cell number (12,500 in Table 1 and 5,000 in Tables 2 and 3), which corresponded to the limit dilution condition. After cell sorting, the frequency of the mixed colonies could be increased 4 and 10 times in two cell fractions (VI-A2 ⁺ and BEN⁻ cells, respectively) and was determined in cultures seeded with a still lower cell number (2,000 cells), which corresponded to the limit dilution condition for the sorted populations concerned. Thus it is highly unlikely that the mixed colonies reported in this study derive from agregates or juxtaposition of unipotent progenitors.

The pluripotent progenitors detected in this paper are more primitive than those that could be detected in the clonal assays so far established in chicken. Late and primitive erythroid progenitors (CFU-E and BFU-E, respectively) develop provided that the culture medium contains anaemic chicken serum (Samarut and Bouabdelli, 1980). Agar cultures containing chicken serum and fibroblast-conditioned medium were initially developed to assay the clonogenic potential of normal and transformed monocytic progenitors (Dodge and Moscovici, 1973). By substituting serum for defined nutrient elements, we recently improved the assay with FCM, since it permitted us to detect granulocytic and granulo-monocytic progenitors (Cormier and Dieterlen-Lièvre, 1988).

In mammals, intensive studies on haemopoietic growth factors have permitted the in vitro identification of pluripotent progenitors, the CFU-GEMM, which give rise to colonies composed of granulocytes, erythrocytes, mono-cytes and megakaryocytes, as early as 1977 in mouse (Johnson and Metcalf) and 1979 in human (Fauser and Messner, 1979). The development of various cell sorting methods combined with various assays (in vivo and in vitro) led to the identification of subsets of progenitors with various selfrenewal and differentiation potentialities (for review, see Watt and Visser, 1992), and recently, Huang and Terstappen (1992) have identified a common precursor for haematopoietic and stromal cells. The chicken pluripotent progenitors detected in this study differ from the mammalian CFU-GEMM in two points. The myelo-erythroid colonies described in this paper contained fewer cells (up to one hundred-fold fewer) than those derived from mammalian CFU-GEMM, which could comprise 10⁵ cells. Furthermore, after only 3 days in culture, the mixed colonies are composed of differentiated cells and do not contain immature cells. This indicates that they do not contain clonogenic cells, contrary to the mixed colonies developed from the mammalian CFU-GEMM, which still contain clonogenic cells at days 7 to 14 (Metcalf et al., 1979). This also indicates that chicken ME-CFC undergo fewer divisions before giving rise to differentiated cells, or that our culture conditions are not yet optimal for their growth. Recently, we have observed that, in the presence of TGFα, para-aortic cells give rise to very large erythroblastic colonies (data not shown), similar to the colonies developed from bone marrow cells in the presence of this growth factor (Pain et al., 1991).

The pluripotent-CFC, immature myeloid-CFC and most of the unipotent myeloid-CFC from the E4 para-aortic region do not express the BEN Ig superfamily molecule, in contrast to the CFC from bone marrow (Corbel et al., 1992b). Molecular cloning of the BEN molecule (Pourquié et al., 1992) demonstrated that BEN is identical to the SC1 homophilic adhesion molecule (Tanaka et al., 1991). Therefore, BEN is likely to represent an adhesion molecule. The absence of the BEN molecule at the surface of most of the para-aortic CFC at E4 is in agreement with a homophilic adhesion property of BEN. Indeed, at E4, haemopoietic cells are widely dispersed and isolated in the mesenchyme network. BEN expression is observed at a later embryonic date when haemopoietic cells are aggregated close together in large foci (Corbel et al., 1992b). Immunocytological staining of plasma clot cultures also demonstrated that haemopoietic progenitors from the para-aortic region acquired BEN expression during the colony formation (data not shown).

In the present study, myelo-erythroid progenitors were detected from an embryonic population located in the region of the aorta. During ontogeny, various haemopoietic organs function successively. The first one, the yolk sac, functions with its own stem cells until E7. Later, haemopoiesis is supported by HSC that emigrated from the embryo to the yolk sac (Martin et al., 1978). By contrast to the yolk sac, the rudiments of the intra-embryonic haemopoietic organs are colonized by extrinsic HSC that are known to originate from the embryo itself, rather than from the yolk sac (Dieterlen-Lièvre, 1975). Before the spleen and the bone marrow become functional, haemopoietic cells develop in the ventral wall of the aorta towards E2-3 and give rise to diffuse haemopoietic foci in the dorsal mesentery between E6 and E9 (Romanov, 1960; Dieterlen-Lièvre and Martin, 1981). This early intra-embryonic haemopoietic population is diffuse in the mesenchyme around the aorta rather than located in a definite organ and could contain HSC that migrate to the haemopoietic organs. The in vitro clonal analysis that we performed give data demonstrating the high haemopoietic potential of the paraaortic population. Our initial studies demonstrated that myeloid and erythroid progenitors are two to eight times more frequent than in the bone marrow (Cormier and Dieterlen-Lièvre, 1988). Myelo-erythroid progenitors such as those reported from the para-aortic region in the present study could never be detected from bone marrow cells from hatching chickens, although the same batches of FCM were used. This indicates that they might be much less frequent in bone marrow than in para-aortic region. Myelo-erythroid progenitors represent 4o/ooo of the para-aortic haemopoietic cells, as defined by the VI-A2 mAb. Moreover, it is not unlikely that the enrichment value of ME-CFC in the VI-A2⁺ population could be biased by the elimination of eventual positive interactions between CFC and non-haemopoietic cells. Thus, this study gives further data demonstrating the richness of the para-aortic region in haemopoietic stem cells.

In birds, assays for the more primitive progenitors are poorly developed. In an in vivo assay using the principle of the mouse CFU-S assay (Till and McCulloch, 1961), Samarut and Nigon (1976) identified chicken primitive erythroid progenitors which, after injection in irradiated chicken, give rise to large erythrocytic colonies in the bone marrow. Recently, we described chicken long-term bone marrow cultures in which the production of M-CFC, G-CFC, GM-CFC and BFU-E and the differentiation of myelo-monocytic cells during 7 months indicated the presence of progenitors with a particularly high self-renewal capacity (Cormier and Dieterlen-Lièvre, 1990). The growth conditions used in the present study permitted us to obtain the development of myelo-erythroid progenitors with high reproducibility (thirteen experiments were performed, four different preparations of FCM were used and mixed colonies developed in each of the experiments). Therefore, they provide an in vitro assay for avian pluripotent progenitors that could be used for characterizing the potentialities of avian haemopoietic cells.

I am very grateful to Dr C. Corbel who collaborated for the analysis of the BEN expression. I thank Drs C. Corbel and F. Dieterlen-Lièvre for critical reading of the manuscript; Drs B. Bruner and O. Pourquié for providing monoclonal antibodies; M. Klaine for her skilful technical assistance; P. Vaigot for performing flow cytometry and S. Gournet and Y. Rantier for the illustrations.

This work was supported by the Centre National de la Recherche Scientifique and INSERM grant no. 881008.