Microenvironment created by stromal cells is essential for a rapid expansion of erythroid cells in mouse fetal liver

Hematopoietic stem cells and their progenitor cells can proliferate and differentiate in the microenvironment of bone marrow, spleen and fetal liver (Metcalf and Moore, 1971; Till and McCulloch, 1961). Understanding of the regulation of hematopoiesis requires a knowledge of the required microenvironment in vitro. By long-term bone marrow cultures (Dexter et al. 1977; Gordon et al. 1987; Zipori et al. 1985), it was demonstrated that bone marrow stromal cells support the proliferation and differentiation of hematopoietic stem cells in vitro. Several stromal cell lines have been successfully established from bone marrow cultures (Tsai et al. 1986; Kodama et al. 1986; Hunt et al. 1987; Whitlock et al. 1987; Itoh et al. 1989) and used in the study of the cellular interactions that are important in the regulation of hematopoietic stem cell proliferation and differentiation.

Spleens in adult mice and livers in fetal mice are other possible microenvironments for hematopoietic stem cells. No detailed analysis has been reported on the microenvironment of these organs for hematopoiesis, however. Recently we established mouse stromal cell lines (MSS lines) from the spleens of newborn mice, which support the proliferation and differentiation of the erythroid progenitor cells in the presence of erythropoietin (epo) (Yanai et al. 1989). Large erythroid colonies were developed from the fetal liver erythroid progenitor cells on the MSS cell layers after 6 days of incubation. These layers also supported the maturation of the erythroid cells. Thus, the established spleen stromal cells may create the in vitro micro-environment adequate for erythropoiesis in mouse spleen. This was the first clear demonstration of the in vitro reconstruction of erythropoietic microenvironment.

Fetal liver is a major erythropoietic organ in the mouse. The microenvironment produced by fetal liver stromal cells may be essential for the growth and differentiation of erythropoiesis. Thus, in this work, we established stromal cell lines (FLS lines) from fetal livers and used them for in vitro reconstruction of the erythropoietic microenvironment of fetal liver.

Establishment of fetal liver stromal (FLS) cell lines

Stromal cell lines (9 FLS lines) were established from primary cultures of fetal liver cells from 13-day embryos of C57BL/6J mice. The livers of 13-day embryos were removed aseptically and prepared for culture by flushing with RTTC 80-7 medium (Yamane et al. 1981; Kyokuto Pharmaceutical Co. Ltd., Tokyo, Japan) using 26 1/2-gauge needle. The cells were washed with same medium, then cultured on 35 mm plastic dishes (Falcon 3001, Becton Dickinson, Oxnard, CA) with RITC 80-7 containing 2 % fetal bovine serum (FBS, Gibco New Zealand Ltd), 10μgm1^-1 transferrin (Sigma, St. Louis, MO), 1μgml^-1 insulin (Shimizu Seiyaku Co. Ltd., Tokyo), 10 μg ml^-1 epidermal growth factor (recombinant EGF, Ajinomoto Co. Ltd,, Tokyo), and 2.5 ng ml^-1 ethanolamine. The medium was changed on the 1st and 5th days. After 8 days in cultivation, primary cultures were subcultured by 25/xgml^-1 pronase (Behringer Mannheim, W. Germany) at room temperature for 5 min. Only adhered cells were subcultured, but contaminating macrophages did not detach from the substrate during such mild pronase treatment. After 8 weeks of successive culture and 8 passages, the cells were cloned by the colony formation during 3 to 4 weeks.

After the initial screening of 9 clones for the ability to support erythropoiesis, 3 clones (FLS3, FLS5, and FLS11) were used for the further experiments. Mouse fibroblast cell line, BALB3T3 A31 (Kakunaga and Crow, 1980), and mouse spleen stromal cell line, MSS 62 (Yanai et al. 1989), were used for comparisons. All cultivations were done at 37°C in a humidified atmosphere of 5 % CO2 in air.

Preparation of hemopoietic progenitor cells and cocultivation of hemopoietic cells on stromal cell layers were performed as described previously (Yanai et al. 1989). Hemopoietic progenitor cells were prepared from the liver of 13-day embryos of C57BL/6J mice (Mishina et al. 1986). Monolayers of FLS cell lines, MSS62 and BALB 3T3 were made in multiwell plates (Falcon 3047, Becton Dickinson, Oxnard, CA). Hemopoietic progenitor cells were cultured on these layers in 0.8 % methylcellulose semisolid culture media containing Iscove’s modified Dulbecco medium (IMDM; Gibco), 30% heat-inactivated FBS, 1% bovine serum albumin, 100mM 2-mercaptoethanol and various concentration of erythropoietin (epo) (recombinant human epo, generously supplied by Kirin-Amgen Co. Ltd). After cocultivation, erythroid colonies were scored by direct staining with benzidine.

The growth of a single erythroid progenitor cell to form the large erythroid colony on the stromal cell layers was monitored under the microscope with a video camera and the number of cells were scored every 12 h for 4 days of culture.

Erythroid progenitor cells were fractionated by the method described previously (Mishina et al. 1986). Fetal liver cells were suspended in a solution containing 155 mM NH4CI, 10 mM KHCO3, and 1 mM EDTA, and incubated to selectively destroy the differentiated erythroid cells before separation, thus enriching CFU-E (colony-forming unit-erythroid). Then, the cell suspension was layered on Percoll (Pharmacia) density gradient which was preformed by ultracentrifugation at 25000g for 15 min with 60% Percoll and separated by centrifugation at 1200g for 10 min. Fractions were taken from the gradient with an 18-gauge needle. The density of each fraction was monitored by density marker beads (Pharmacia).

FLS cell lines were established from primary cultures of livers of 13-day gestation fetuses of C57BL/6J mice (Fig. 1). These established FLS cell lines were maintained with RITC 80-7 supplemented with 2% FBS, because they do not adapt to culture conditions with higher FBS concentrations.

The ability of the FLS cell lines to support erythropoiesis was examined by culturing erythroid progenitor cells from mouse fetal liver on monolayers of FLS cell lines in a semisolid medium in the presence or absence of epo. The differentiated colonies were counted after benzidine-staining of hemoglobin-producing (benzidine-positive) cells. After 2 days of cultivation of fetal liver cells in the presence of 0.1 Uml^-1 of epo in the semisolid medium, dark-blue benzidine-positive colonies derived from colony forming unit-erythroid (CFU-E) were formed, but they subsequently disappeared. Then, other types of colonies continued to proliferate on the stromal cell layers and developed to large randomly distributed ones after several days cultivation in the presence of 0.1 U ml^-1 epo as shown in Fig. 2.

In three independent experiments, FLS3, 5 and 11 Unes stimulated the formation of large erythroid colonies (Table 1). Formation of large colonies is dependent on the FLS cell layers and epo, since large colonies were not formed from fetal liver cells even in the presence of IL-3+epo or GM-CSF+epo without the cell layers and the large colony formation on the FLS cell layers was not stimulated with IL-3 or GM-CSF in the presence of epo (data not shown). On the FLS cell layers, benzidine-negative cells appeared after 4 days culture, but those colonies was not large, consisting of fewer than 50 cells. Mouse fibroblast cell fine BALB 3T3, which we used for comparison, did not support erythropoiesis, although it has been shown to support granulocyte proliferation (Roberts et al. 1987). FLS cell lines could support the formation of as large erythroid colonies from the progenitor cells of adult bone marrow as do those from fetal liver cells.

The properties of the erythroid progenitor cells that form the large colonies on the stromal layers were examined by comparison of their epo sensitivity and cell density with those of the CFU-E-derived colonies. The dose-dependency of epo on the formation of both types of colonies from the erythroid progenitor cells of fetal fiver is shown in Fig. 3. The large colonies on the FLS cell layers were formed at the same concentration of epo required for the formation of CFU-E-derived colonies. We obtained the same epo-dependency on the formation of both types of colonies from the progenitor cells of adult bone marrow (data not shown). Thus, large erythroid colonies on FLS cell layers were produced at epo concentration as low as 0.05Uml^-1, enough for CFU-E, but not for BFU-E. To examine whether both types of colonies are generated from different progenitors, the erythroid progenitor cells of fetal livers were fractionated using Percoll gradient centrifugation. The CFU-E-derived and the large erythroid colonies on the stromal layers were formed from the progenitors that migrated in the same density (Fig. 4). These results show that both types of colonies are generated from erythroid progenitor cells with the same epo sensitivity and cell density. From the examination of the morphological character of the erythroid cells of the colonies by May-Gruenwald-Giemsa staining, the majority were identified as erythroblasts, but some were enucleated mature erythrocytes. Many cytoplasmic buddings were observed among these erythroblasts in the process of enucleation. Thus, the enucleation process of erythroid cells may be facilitated on FLS cell layers.

The erythroid cells grew on top of the FLS cell layers with some cells dispersed or detached from the layer. The growth was so rapid that colonies consisted of more than 500 and sometimes over 1000 cells after 4 days of culture. Their orange color revealed an accumulation of hemoglobin and most cells in the colonies were stained with benzidine. To show the rapid proliferation and differentiation of single erythroid progenitor cells, we followed the formation of colonies from several independent single cells on FLS cell layers by observing the colonies every 12 h using a microscope attached to a video TV set. Cell numbers of the individual continuously growing colonies were counted as shown in Fig. 5. The result clearly indicated that individual colonies were derived from single cells and 500 to 1000 hemoglobin-positive cells were generated from a single erythroid progenitor cell on the FLS cell layer after 4 to 5 days. The generation time of the colony-forming erythroid cells was from 6 to 10 h (average=9.6 h). The colonies were stained everyday, and most of cells in a single colony were benzidine-positive, starting from the small colonies consisting of 20-30 cells to the large colonies of over 1000 cells (Fig. 2). Thus, the erythroid cells producing hemoglobins could divide more than 10 times on the FLS cell layers in the presence of epo.

The stromal cells of hematopoietic organs are thought to create a microenvironment adequate for the proliferation and differentiation of hematopoietic stem cells and their progenitor cells. The survival, proliferation and differentiation of the hematopoietic stem cells and their progenitor cells in an in vitro culture were earlier demonstrated on layers of stromal cells established from bone marrow (Tsai et al. 1986; Kodama et al. 1986; Hunt et al. 1987; Whitlock et al. 1987; Itoh etal. 1989). It has been proposed that stromal regulation of hematopoiesis occurs by cell-cell contact (Kodama et al. 1986; Itoh et al. 1989), by extracellular matrices (Zipori et al. 1985; Bentley and Foidart, 1980; Zuckerman and Wicha, 1983; Campbell et al. 1987), or through the secretion of hematopoietic growth factors from the stromas (Tsai et al. 1986; Hunt et al. 1987; Lipton and Nathan, 1983).

However, there has been no report of established stromal cells that support erythropoiesis. In the previous work, we demonstrated clearly that the established mouse spleen stromal (MSS) cell lines selectively support formation of the large erythroid colonies only in the presence of epo. In this work, we have tried to reconstruct the erythropoietic micro-environment in vitro with the established mouse fetal liver stromal cell (FLS) lines to explain the phase of rapid expansion of the erythropoietic cell population observed in mouse fetal liver during 12 to 16 days of gestation (Wolf and Trentin, 1967).

On the FLS cell layers, we detected the large colonies composed of 500 to 1000 benzidine-positive erythroid cells. They were produced only on the FLS cell layers from the erythroid progenitor cells from fetal livers and from adult bone marrows in the presence of epo. The erythroid cells producing hemoglobins could divide more than 10 times on the FLS cell layers in the presence of epo. A large expansion of a single erythroid progenitor cell observed in our stromal cell Unes is striking since such large colonies were not observed in the in vitro erythroid colony formation in the usual semisolid assay. Although expansion of late erythroid progenitor cells occurs at the stage of BFU-E (burst-forming unit-erythroid) in the semisolid medium without feeder cells after 7 days of culture (Gregory and Eaves, 1977), the colonies are smaller and their cell population is not homogeneous. The large colony-forming cells from fetal liver and adult bone marrow observed in the present study require the same optimum epo concentration (as low as 0.05Uml^-1), enough for CFU-E, but not for BFU-E. They are not separated from CFU-E by fractionation of the erythroid progenitor cells of fetal livers. BFU-E can be separated from the CFU-E by density gradient experiments (Heath et al. 1976; Mishina et al. 1986), requires a higher epo-concentration than CFU-E and cannot be detected in the fetal liver erythroid cells in the semisolid culture (Gregory and Eaves, 1977; Iscove etal. 1974). In fact, large colonies were not formed from fetal liver cells in the IL-3+epo or GM-CSF+epo without FLS cell layers. Thus, the progenitors that form the large erythroid colonies appear not be BFU-E, but may be at a stage close to or identical with the CFU-E.

The in vitro microenvironment created by the fetal Ever stromal cell layers is essential for the large expansion of the colonies from erythroid progenitor cells. An in vitro microenvironment that supports more than 10 cycles of cell division of the erythroid progenitor cells can explain the phase of rapid expansion of the erythropoietic cell population observed in the fetal liver (Wolf and Trentin, 1967). The in vitro microenvironment seems to be created by direct contact and/or short-range communication between the erythroid progenitor cells and the stromal cell layers since large erythroid colonies were not formed by separation of the progenitor cells from the stromal cells by a diffusion chamber or a nucleopore filter and the large erythroid colonies were formed only in the presence of semisolid support (data not shown). Supplementing the conditioned medium of stromal cells did not show any effects on the colony formation (data not shown). Thus, IL-3, GM-CSF or other diffusive factors that stimulate the erythroid colony formation in the presence of epo may not be produced from the FLS cells. In this microenvironment, many mature erythrocytes and erythroblasts showing cytoplasmic budding were produced as also observed on the MSS cell layers (Yanai et al. 1989). Thus, FLS cells may support the maturation of enucleation of the erythroblasts in vitro. Morphological and biological studies on the MSS cell lines suggest these cell fines may be derived from endothelial cells (Yanai et al. in preparation) and most of FLS cell lines examined in this study seemed to be epithelial-like cells in their morphology. Although their morphological features are different, they show similar selective ability to support erythropoiesis in vitro. It will be interesting to examine at the cellular and molecular levels how the clonally established stromal cell creates the microenvironment sufficient for the proliferation and differentiation of erythroid cells.

This work was partly supported by Special Coordination Funds from the Science and Technology Agency of the Japanese Government and Grant-in-Aid for Cancer Research from the Ministry of Education, Science and Culture of Japan.