Isolation, growth and identification of colony-forming cells with erythroid, myeloid, dendritic cell and NK-cell potential from human fetal liver.

The study of hematopoietic stem cells (HSCs) and the process by which they differentiate into committed progenitors has been hampered by the lack of in vitro clonal assays that can support erythroid, myeloid and lymphoid differentiation. We describe a method for the isolation from human fetal liver of highly purified candidate HSCs and progenitors based on the phenotypes CD38(-)CD34(++) and CD38(+)CD34(++), respectively. We also describe a method for the growth of colony-forming cells (CFCs) from these cell populations, under defined culture conditions, that supports the differentiation of erythroid, CD14/CD15(+) myeloid, CD1a(+) dendritic cell and CD56(+) NK cell lineages. Flow cytometric analyses of individual colonies demonstrate that CFCs with erythroid, myeloid and lymphoid potential are distributed among both the CD38(-) and CD38(+) populations of CD34(++) progenitors.

Introduction

Maintenance of blood cell homeostasis involves the daily production of a tremendous number of blood cells from a relatively small pool of stem cells. Regulation of hematopoiesis is complex involving both hormonal and local regulatory mechanisms that can promote or inhibit the growth and differentiation of stem cells into progenitors and, in turn, mature blood cells. In adult mammals, hematopoiesis occurs primarily in the bone marrow. However, hematopoiesis can occur in other locations as well such as in the thymus, which is specialized in the generation of T cells. In addition, small numbers of natural killer (NK) cells, B cells, dendritic cells (DCs) and myeloid cells are also generated in the thymus (1, 2). The spleen offers another potential site for hematopoiesis, especially in small animals. Hematopoietic progenitors have also been found in the adult liver (3).

Hematopoietic progenitors and stem cells are distributed continually to favorable sites of hematopoiesis. This is evidenced by the presence of a small number of these cells that can be found circulating in the peripheral blood (4). The numbers of circulating progenitors and stem cells can increase dramatically when hematopoietic activity is increased. Treatment with various hematopoietic growth factors or recovery from chemotherapeutic injury can result in the mobilization of hematopoietic progenitors and stem cells (5). This is reminiscent of mammalian development, a period of rapid growth in hematopoietic tissues marked by high levels of circulating progenitors and stem cells and a remarkable series of transitions between hematopoietic tissues (6, 7). Hematopoiesis begins extraembryonically in the blood islands of the yolk sac (6). Intraembryonic hematopoiesis is first observed in the paraaortic splanchnopleura or aorta-gonad-mesonephros region and moves to the liver, the primary site of hematopoiesis during much of fetal life (7). In mice, hematopoiesis transitions from the liver to the spleen in the older fetus, and then to the bone marrow around the time of birth. In humans, the relocation of hematopoiesis from the liver to the bone marrow begins early in the second trimester and the spleen is never a significant location for hematopoiesis (8).

A large number of culture systems have been developed for the study of hematopoiesis in vitro. One such system that has been prominent in much of the research in this field is the colony-forming cell (CFC) or colony-forming unit-culture (CFU-C) assay (9, 10). In its basic form, the CFC assay is a clonal assay used for the enumeration of hematopoietic progenitors. Progenitors are suspended in a combination of culture medium, growth factors and a semi-solid matrix such as methylcellulose or agarose. Over time the suspended progenitors proliferate and differentiate forming a colony of blood cells that can be readily counted using an inverted or dissecting microscope. The CFC assay has evolved since its beginnings in the mid 1960's. Layers of feeder cells, a source of growth factors, were replaced by conditioned media and eventually recombinant cytokines, discovered in part through the use of the CFC assay. There is also no longer a need for serum in the culture medium as various formulations of defined serum components have been devised that can substitute serum (11, 12).

Stem cells are defined by their extensive proliferative capacity, including a limited ability for self-renewal, and their pluripotentiality. Although no CFC assay has ever been developed that can definitively detect stem cells, CFC assays have been developed to measure the properties of extensive proliferative capacity and pluripotentiality. High-proliferative potential CFCs (HPP-CFCs) are a primitive subpopulation of progenitors, likely including stem cells, that are characterized by their ability to form large colonies containing 1 x 104 cells (13). HPP-CFCs were first detected when two sources of conditioned media were combined to provide a synergistic combination of cytokines that supported the growth of primitive progenitors, a characteristic requirement of primitive progenitors. The blast cell colony assay, similar to the HPP-CFC assay, is another variant of the CFC assay that detects cells with an extensive proliferative capacity based, in part, on their potential to generate secondary colonies when replated (14). The conditions that support the generation of myeloid and erythroid cells from CFCs were developed early in the evolution of the CFC assay. However, growth of lymphoid progenitors and multipotent progenitors with lymphoid potential hasn't been possible until more recently. With the discovery of the cytokines interleukin (IL)-7 and IL-15, the clonal growth of murine B cell progenitors (15-17) and NK-cell progenitors (18) was achieved. Indeed, murine multipotent CFCs can form colonies with myeloid, erythroid and lymphoid cells (19, 20). Similar successes with human lymphoid and multipotent CFCs have trailed the murine results. Initially, human NK-cell colonies were grown from committed lymphoid CFCs isolated from thymic tissue (21). Recently, our laboratory reported the growth of CFCs with erythroid, myeloid, DC and NK cell potential (22). These CFCs were isolated from human fetal liver tissues, grown and detected by the methods detailed in this report.

Materials and Methods

Human fetal liver progenitors

Livers were harvested from midgestation fetuses obtained with maternal consent from elective abortions. Research with fetal tissue was performed with approval of the Committee of Human Research at our institute. Two subpopulations of progenitors were isolated by the method detailed in the Protocol section: CD38-CD34++lineage- (lineage or Lin = CD3, CD14, CD19, CD20, CD56 and CD235a) and CD38+CD34++Lin- fetal liver cells (23). For simplicity, these two populations of cells will be referred to as CD38- and CD38+ progenitors. The CD38- population is enriched in hematopoietic stem cells whereas the CD38+ population is an intermediate population of progenitors that still expresses high levels of CD34 and is believed to be derived from the CD38- stem cells (23, 24).

Colony-forming cell assay

A detailed description of the culture conditions used to grow multipotent CFC is given in the Protocol section. Briefly, 50 to 100 cells were plated/culture dish and 3 to 8 dishes were analyzed for each experiment. Progenitors were grown in semi-solid medium under serum-deprived conditions supplemented with six recombinant human cytokines: kit ligand (KL), flk-2/flt-3 ligand (FL), GM-CSF, thrombopoietin (TPO), erythropoietin (EPO) and interleukin-15 (IL-15). KL, FL, TPO and IL-15 were purchased from R&D Systems, Inc. (Minneapolis, MN, http://www.rndsystems.com). EPO was purchased from Amgen (Thousand Oaks, CA, http://www.amgen.com and GM-CSF (Leukine) from Immunex Corporation (Seattle, WA, http://www.immunex.com). The serum-deprived medium used has been shown to be superior to serum-replete medium in supporting the growth of myeloid and erythroid cells (25, 26). The cytokines used were chosen based on previous experiments that indicated that the combination KL+FL+GM-CSF+IL-15 supports the generation NK cells from CD38- progenitors (27). Furthermore, GM-CSF, FL, KL and IL-15 have also been shown to support the growth of DCs (28-30). EPO and ML were further added to the cultures to support erythroid and megakaryocytic development, which did not appear to interfere with the development of the myeloid, dendritic and NK cells (22). CFCs were cultured for 3 weeks and analyzed for lineage composition.

Analyses of colony lineage composition

Colonies containing erythrocytes were identified visually by the presence of hemoglobinized cells. The presence of myeloid (CD14+ and/or CD15+ cells) and NK cells (CD56+ cells) was determined by flow cytometric analyses of live cells isolated from individually plucked colonies. Anti-CD14-PE (clone TüK4) (Caltag Laboratories, Burlingame, CA, http://www.caltag.com), anti-CD15-PE (clone VIMC6) (Caltag Laboratories) and anti-CD56-FITC (clone C5.9) (Exalpha Corporation, Boston, MA, http://www.exalpha.com) were the monoclonal antibodies (mAbs) used to stain the colonies. In some experiments, anti-CD1a-PE (clone T6) (Beckman Coulter, Inc., Miami, FL, http://www.beckmancoulter.com) was used to detect DCs combined with anti-CD14-APC (clone M5E2) (BD Biosciences, http://www.bdbiosciences.com), anti-CD15-APC (clone HI98) (BD Biosciences) and anti-CD56-FITC. A detailed protocol for the analysis of colony lineage composition is presented in the Protocol section. A FACScan or FACSCalibur flow cytometer (BD Biosciences) was used to analyze the stained colonies.

Results and Discussion

The early stages of lymphoid development are only partially characterized. Most schema depicting hematopoietic differentiation place the separation of lymphoid progenitors from myeloid-erythroid progenitors as the first major branch point in hematopoiesis. This early division of lymphopoiesis from myeloerythropoiesis does have some basis in experimental observation. Galy et al. isolated a CD10+CD38+ committed lymphoid progenitor from among early hematopoietic progenitors that express high levels of CD34 (31). These lymphoid progenitors were capable of generating T cells, B cells, NK cells and DCs. Since there is abundant evidence that stem cells reside among CD38-CD34++ cells (24), a model in which pluripotent CD38- stem cells differentiate into multipotent CD38+ lymphoid and myeloerythroid progenitors can be envisioned.

To further research the development of lymphoid progenitors, we set out to define in vitro conditions for the clonal growth of progenitors with lymphoid, myeloid and erythroid potential. Two cell populations expressing high levels of CD34 were studied, CD38- candidate stem cells and CD38+ progenitors. These cells were isolated from human fetal liver according to the criteria depicted in Fig. 1. Cells were isolated based on their lack of staining with PI (Fig. 1A), low side-light scatter (Fig. 1B) and lack of expression of a panel of Lin markers (Fig. 1C). Cells were sorted using regions similar to the ones shown in Fig. 1D. These regions are drawn conservatively since analyses of the sorted cells (Fig. 1E and F) always indicate that the sorted cells have spread beyond the original regions used to sort the cells. Nonetheless, highly purified populations of progenitors and candidate stem cells can be obtained. These cells were cultured under serum-deprived conditions using a methylcellulose-based colony assay system. The cytokine cocktail KL+FL+TPO+GM-CSF+EPO+IL-15 was used, which supported the growth of erythroid, myeloid, NK cells and DCs. Currently, conditions that support the development of human B- and T-cells in colony assays remain elusive, although B cells have been generated under defined liquid culture conditions (27). We speculate that the efficiency of B-cell formation remains below the threshold for detection of these cells within individual colonies (22).

Fig. 1

Isolation of primitive hematopoietic progenitors from human fetal liver.

Hematopoietic progenitors were isolated by FACS following enrichment by two rounds of immunomagnetic bead depletion of Lin+ cells and isolation of light-density cells. Dead cells and debris are removed using a region (A, R1) drawn to include PI- events with a moderate forward scatter. Cells with a low to medium side scatter are then selected using a second region (B, R2). Any remaining Lin-FITC+ cells are then removed with a third region (R3) as shown in C. Expression of CD34 and CD38 are then viewed on events gated to fall within regions 1-3 (D). CD38+CD34++ and CD38-CD34++ events are selected using regions 4 and 5, respectively, as indicated. These regions are drawn conservatively as analysis of the sorted cells (E and F) usually indicates some slight spread of the populations outside the sort regions. Ideally, the isolated CD38+ and CD38- progenitors will be nearly contiguous, but non-overlapping, populations.

Colonies grown from the two cell populations were scored using a dissecting microscope. To better insure clonality we have settled on plating only 50 sorted cells/plate and analyzing up to 8 dishes for each cell population. At this starting density we obtained an average 9.5 and 11 colonies/dish for CD38- and CD38+ progenitors, respectively (22). Even at these low densities colonies were sometimes closely spaced or overlapping. However, we have had little difficulty in harvesting these colonies by omitting to harvest portions of the colony that may be contaminated with another colony's cells. Before harvest, the estimated size and the presence or absence of hemoglobinized erythrocytes was recorded as each colony was visualized. Examples of myeloid colonies are shown in Figs. 2A and 2B and representative myeloerythroid colonies are shown in Figs. 2C-F. After recording the attributes of a colony, the colony was drawn into a 200μpipette tip (Figs. 2D-F) and dispersed in a solution of blocking buffer and mAbs.

Colony formation by CD38- progenitors.

Examples of myeloid colonies are shown in A and B. Mixed lineage colonies, derived from HPP-CFCs, containing myeloid and erythroid cells are shown in C and D. Harvest, using a micropipette, of one of the colonies shown in D is shown in E. The small number of cells remaining after the harvest of the colony is shown in F. All colonies were grown in KL+FL+ML+GM-CSF+EPO+IL-15. Colonies were viewed on a 2mm grid.

After staining, individual colonies were analyzed by flow cytometry (Fig. 3). Live and dead cells were distinguished using PI. Considerable numbers of dead cells and debris were removed from analysis using a gate similar to the one shown in Fig. 1A. As shown in Fig. 3, cell populations were observed with different light scatter properties (22). Erythrocytes, NK cells and a subpopulation of the CD14+ and CD15+ cells had a low side scatter profile (Fig. 3A, blue events). Most cells with a high side scatter had a high autofluorescence and many also expressed CD14 (Fig. 3. red events). These high side scatter cells represent primarily DCs as discussed below.

Fig. 3

Determination of lineage content of individual colonies by flow cytometry.

Results of three colonies differing in their lineage composition are shown. Low side scatter cells are shown in blue and high side scatter cells are shown in red as indicated in A. A colony containing myelocytes and erythrocytes, but no NK cells, is shown in B. Note the two populations of cells CD14/CD15+ cells have distinctive side scatter profiles. A multilineage colony containing erythrocytes, myelocytes, NK cells and most likely CD14+ DCs is shown in C. A colony containing mostly CD14+ DCs and NK cells is shown in D. All colonies were derived from CD38- progenitors (22).

We analyzed 313 colonies grown from CD38- progenitors (n = 5 experiments) and 110 colonies grown from CD38+ progenitors (n = 4 experiments) (22). Progenitors capable of giving rise to myeloid, erythroid and NK cells were enriched among the CD38- fraction, representing 13.7% of all CFC compared to 2.7% among all CD38+ CFCs. These results are consistent with the belief that stem cells reside among the CD38- progenitors, in that multipotent progenitors were enriched among the CD38- population. Nonetheless, the majority (51.8%) of CD38- CFCs formed colonies containing only myeloid cells. Another 21.7% had both myeloid and erythroid cells, but no identifiable NK cells. Similarly, pure myeloid (40.0%) and myeloerythroid (39.1%) colonies represented the bulk of CD38+ CFCs. A small (5.5%) subpopulation of CD38+ CFCs generated erythroid colonies with no identifiable myeloid cells. Such erythroid colonies represented <1% of CD38- CFCs.

Except for a single colony derived from CD38+ CFCs, no pure NK-cell colonies were observed. It should be noted that even this one NK-cell colony contained a number of cells that failed to stain with either the CD14/CD15 or the CD56 mAbs. Not all cells mature in each colony at the same rate. Consequently, our detection methods may fail to ascertain the complete lineage composition of all colonies, especially in small colonies. Analysis of the colonies after 3 weeks of growth represents a compromise between earlier time points at which fewer cells express CD14 or begin to be hemoglobinized and later time points when cells begin to die and the cultures decline (26, 32). Presently, we have not had success in expanding the mature NK cells generated in vitro, although the NK cells are functionally mature and capable of killing appropriate target cells (27). In liquid cultures the number of NK cells relative to the numbers of myeloid and erythroid cells is low (generally <5%). Likewise, a mean of only 24.9 CD56+ cells were detected/colony (22). This low number of NK cells required the threshold for detection to set a low 3 events above background staining. Nonspecific binding of the FITC-conjugated IgG2b antibody was very low to non-existent enabling detection of low numbers of CD56-FITC+ cells, whereas PE-conjugated mAbs have slightly higher background staining. NK cells were defined as not only CD56+ events but also as cells with a low side scatter (Fig. 3), which further reduces the chances of false positives. As culture conditions for the growth of lymphocytes improve we anticipate greater numbers of NK-cell colonies to be detected and the detection of CFCs with B-cell potential to become feasible.

Since a high proliferative capacity is another property of stem cells, we also analyzed the fraction of colonies that grew from HPP-CFCs. HPP-CFCs represented 16.4% of the CD38- CFCs and 22.0% of the CD38+ CFCs (22). Nearly all of these colonies contained both myeloid and erythroid cells. It should be noted that support of the erythroid lineage in our cultures increased the numbers of HPP-CFCs observed over the number observed in the absence of EPO. This is because the presence of erythrocytes in a colony greatly increases the size of the colony. For instance, HPP-CFCs are restricted to the CD34++ fraction of fetal liver cells when only myelopoiesis is supported (33), whereas further support of erythropoiesis results in colonies with over 1 x 104 cells arising from even the more mature CD34+Lin- population of fetal liver cells (26). When analyzed by flow cytometry, 59.4% of CD38- HPP-CFC generated colonies containing erythroid, myeloid and NK cells, whereas 46.3% of CD38+ HPP-CFC generated such colonies (22). Therefore, this analysis further suggested an enrichment of stem cells among the CD38- fraction, but the high frequency of multipotent HPP-CFC among CD38+ progenitors was unexpected. We initially held out the possibility that the presence of multilineage HPP-CFCs among the CD38+ fraction of progenitors suggests that rare stem cells may also reside among this fraction of fetal liver cells (22). However, recent analysis of the bone marrow reconstituting ability in NOD/SCID mice of CD38-CD34++/+ and CD38+CD34++/+ progenitors, isolated from umbilical cord blood, showed that long-term reconstituting ability was limited to the CD38- fraction (34). We have also subsequently tested the reconstituting potential of the two fractions of fetal liver progenitors and found that only the CD38- fraction of progenitors could generate active hematopoiesis in the bone marrow of NOD/SCID mice 8 weeks after transplantation (authors unpublished observations). Thus, it appears that the CD38+ CFCs with myeloid, erythroid and lymphoid potential represent a population of multipotent progenitors rather than stem cells.

Further analysis of the two subpopulations of CD14/CD15+ cells present in the hematopoietic colonies (Fig. 3) indicated that the high side-scatter cells were CD14+ DCs (22). These cells were identified as myeloid DCs (DC1) based on their expression of CD2, CD4, CD11a, CD11c, CD40, CD80, CD83, CD86, HLA-DR and HLA-DQ (35-37). Not all high side scatter cells expressed all of these markers, but the general pattern suggested that many of the high side scatter cells represented DCs at varying stages of differentiation. In addition, the marker CD1a was found expressed on most of the DCs generated under our culture conditions, further indicating a myeloid origin for these cells (37).

CD1a+ cells were found in approximately three-quarters of all colonies grown from CD38- and CD38+ progenitors. Examples of CD1a expression on cells isolated from representative colonies are shown in Fig. 4. These colonies were stained with anti-CD56-FITC, anti-CD1a-PE, anti-CD14-APC, anti-CD15-APC and PI, enabling us to distinguish non-DC myeloid cells from DCs. The stained cells were analyzed on a two-laser FACSCalibur flow cytometer. Colonies containing CD14/CD15+ myeloid cells, but no DCs, were observed such as shown in Fig. 4B. Varying levels of myeloid and DCs were also observed as exampled by Fig. 4C-E. Note that the expression of CD14 on CD1a+ cells was variable in these colonies (CD15 is not expressed by the high side-scatter cells under these culture conditions). Fig. 4F is an example of a colony containing primarily DCs and no myeloid cells. Thus, using the techniques described the clonal analysis of CFCs capable of erythroid, myeloid, DC and NK cell differentiation is feasible.

Fig. 4

Flow cytometric analysis of myelocyte and DC content in individual colonies.

Results of five colonies differing in their lineage composition are shown. Low side scatter cells are shown in blue and high side scatter cells are shown in red as indicated in A. A colony with myeloid cells but no CD1a+ DCs is shown in B. C-E show colonies with varying frequencies of CD1a+ and CD14/CD15+ cells. F is an example of a colony primarily composed of CD1a+ DCs. All colonies shown were derived from CD38- progenitors.

We anticipate that such analyses of the lineage potential of CFCs will be useful in research aimed at understanding the early events of hematopoiesis. Analyses of myeloerythroid progenitors have indicated that hematopoietic differentiation is, in large part, a stochastic process (38). Our analysis of lineage potential of CD38- and CD38+ CFCs would agree with this assessment. Differentiation, mostly a loss of erythroid potential, was found to occur at the CD38- stage of differentiation whereas CD38+ progenitors still contained many multipotent progenitors. In regards to NK-lymphopoiesis in particular, our data suggests that multiple pathways of differentiation may exist. Recently, Hao et al. demonstrated the existence of a lymphoid committed progenitor among the CD38- fraction (39). Thus, based on our work and that of Galy et al. (31), the data suggest that segregation of the lymphoid from the myeloid and erythroid pathways can occur either very early in hematopoietic differentiation or later at the CD38+ stage of differentiation. The possibility also exists for multiple pathways of NK cell differentiation. A lymphoid pathway arising from a common T-, B-, NK- and lymphoid dendritic-cell may define one pathway for NK cell differentiation, whereas a myeloid pathway for NK-cell differentiation may also exist (22, 37). Such a hypothesis would be consistent with previous observations regarding the stochastic nature of hematopoietic differentiation. Further analysis of the multilineage potential among subsets of early hematopoietic progenitors will help to understand the process of hematopoietic differentiation and the environmental signals that control it.

1Abbreviations: Fluorescein isothiocyanate (FITC), phycoerythrin (PE), allophycocyanin (APC), American Type Culture Collection (ATCC).

Specificity-Label	Clone	Product #	Isotype	Source
CD3-FITC1	SK7	349201	IgG1	BD Biosciences, http://www.bdbiosciences.com
CD14-FITC1	TüK4	MHCD1401	IgG2a	Caltag Laboratories, http://www.caltag.com
CD19-FITC	SJ25C1	340409	IgG1	BD Biosciences
CD20-FITC	HI47	MHCD2001	IgG3	Caltag Laboratories
CD34-FITC	581	IM 1870	IgG1	Beckman-Coulter, http://www.beckmancoulter.com
CD34-APC1	581	IM2472	IgG1	Beckman-Coulter
CD38-PE	HB7	347687	IgG1	BD Biosciences
CD56-FITC	C5.9	0562	IgG2b	Exalpha Inc., http://www.exalpha.com
CD235a	10F7MN	HB-8162	IgG1	ATCC1, http://www.atcc.org
CD235a-FITC	11E4B-7-6	IM2212	IgG1	Beckman-Coulter Inc
None/FITC	polyclonal	MG101	IgG1	Caltag Laboratories
None/PE	polyclonal	MG104	IgG1	Caltag Laboratories
None/APC	polyclonal	IM 2475	IgG1	Beckman-Coulter Inc.