Progenitor cells divide symmetrically to generate new colony-forming cells and clonal heterogeneity


  • Stephen B. Marley,

    1. Leukaemia Research Fund Centre for Adult Leukaemia, Department of Haematology, Imperial College Faculty of Medicine, London, UK
    Search for more papers by this author
  • John L. Lewis,

    1. Leukaemia Research Fund Centre for Adult Leukaemia, Department of Haematology, Imperial College Faculty of Medicine, London, UK
    Search for more papers by this author
    • *

      Present address: YCR Cancer Research Unit, Department of Biology, University of York, York, UK.

  • Myrtle Y. Gordon

    1. Leukaemia Research Fund Centre for Adult Leukaemia, Department of Haematology, Imperial College Faculty of Medicine, London, UK
    Search for more papers by this author

Professor M. Y. Gordon, LRF Centre for Adult Leukaemia, Department of Haematology, Imperial College Faculty of Medicine, Du Cane Road, London W12 ONN, UK. E-mail:


Summary. Self-renewal is the most fundamental property of haemopoietic stem and progenitor cells. However, because of the need to produce differentiated cells, not all cell divisions involve self-renewal. We have used a colony replating assay to follow the fates of individual haemopoietic progenitor cell clones. For this, human myeloid colony-forming cells (CFCs) were cultured by standard methodology. Onset of proliferation and growth rates were established by a video recording method. Individual colonies were replated several times to document the rate of clonal extinction, and the numbers of secondary, tertiary and quaternary CFCs. The clonogenic population exhibited similar kinetics in terms of onset of proliferation and growth rate. Clonal extinction was progressive so that only 30 ± 7% (mean ± standard error of the mean; n = 4) of the original primary colonies formed quaternary colonies after the third replating step. However, individual primary CFCs that produced colonies throughout the experiment generated, on average, 40 ± 8 secondary and tertiary CFCs overall. The values obtained in standard culture conditions were modified when granulocyte colony-stimulating factor (G-CSF) or G-CSF plus interleukin 3 were used to stimulate colony growth, showing that the kinetics of colony formation respond to extrinsic regulation. Examination of the replating potential of individual secondary colonies in the clones demonstrated that they generated different numbers of tertiary colonies. The data best fit a stochastic model of haemopoietic cell development where event probabilities can be modified by extracellular factors.

The ability of the haemopoietic system to maintain blood cell production throughout life depends on the capacity of haemopoietic stem cells to self-renew and also to produce daughter cells, which differentiate and ultimately replenish all lineages of mature blood cells. This means either that a single cell must divide to produce two daughter cells which adopt different fates, or that some stem cells produce two new stem cells while others produce two differentiated cells. However, the characteristics of self-renewing stem cell divisions are not well defined. Theoretically, balanced self-renewal and differentiation could be achieved in one of the following ways. First, the outcome of cell division may be determined before mitosis, as a result of unequal distribution of factors such as transcription factors in the parent cell (Horvitz & Herskowitz, 1992), so that stem cells divide to form one new stem cell and one differentiated cell. However, this model does not allow for stem cell regeneration or expansion (Gordon & Blackett, 1994, 1998; Morrison et al, 1997). Second, daughter cells may be similar to the parent but be differentially exposed to extracellular differentiation signals, such as cytokines. The disadvantage of this model is that it will result in an increase in stem cell number, albeit probably minor and temporary. Third, stem cells may divide symmetrically with some forming two new stem cells while others form two differentiating cells. This would allow stem cell recovery and expansion by altering the relative proportions of the stem cells which self-renew versus those which differentiate (Gordon & Blackett, 1994, 1998; Morrison et al, 1997) and would also explain why it has not been easy to purify stem cells to functional homogeneity.

Strategies for investigating self-renewal by stem cells are limited by their rarity in haemopoietic tissue and the difficulties experienced in culturing them in vitro. Candidate stem cells have been isolated phenotypically and cultured as single cells to examine the pattern of cell division (Huang et al, 1999; Punzel & Ho, 2001; Punzel et al, 2002) or conservation of cell phenotype (Goff et al, 1998). Brummendorf et al (1998) sorted single CD34+CD38 fetal liver cells and cultured them in individual microtitre wells and found some clones produced CD34+CD38 cells that could be recloned and cultured for several passages. Ema et al (2000) tested separated stem cell daughters for their capacity to engraft mice in a competitive repopulation assay and obtained evidence that stem cell division was asymmetric in functional terms.

It seems unlikely that self-renewal by stem cells alone is sufficient to account for the amplification in cell number that occurs between the stem cell population and mature circulating blood cells, and more likely that self-renewal persists for several cell generations. It has been suggested that there is a transit amplifying cell population, which shares the property of self-renewal with stem cells (Potten & Loeffler, 1990; Loeffler & Potten, 1997; Gordon & Blackett, 1998). Indeed, models of haemopoietic cell development generally include reducing self-renewal as a component (Slayton et al, 2001; Quesenberry et al, 2002). Moreover, it has been difficult to demonstrate stem cell expansion in vitro, but remarkably easy to expand progenitor cell numbers (Peters et al, 1995, 1996). By replating granulocyte–macrophage colony-forming unit (CFU-GM) colonies, we have found that substantial numbers of new colony-forming cells (CFCs) were formed during colony development. Consequently, this assay provides a clonal model system for investigating aspects of self-renewal and differentiation. It also provides the opportunity to investigate whether division is symmetrical or asymmetrical in terms of self-renewal and differentiation, if probabilities of self-renewal and differentiation can be modified by extrinsic factors such as cytokines, and whether the self-renewal properties of daughter cells reflect those of their parents.

Materials and methods

Cells for study.  All samples were obtained from bone marrow donors for transplantation into allogeneic recipients, with informed consent and research ethics committee approval. The mononuclear cell fraction was separated using Lymphoprep (Nycomed, Oslo, Norway), washed three times and resuspended at 106 cells per ml in alpha medium (Gibco, Paisley, UK) supplemented with 15% fetal calf serum. Ten millilitres of cell suspension were placed in a 25-cm2 tissue culture flask and incubated for 2 h at 37°C in humidified 5% CO2 in air to remove plastic-adherent cells.

Culture of primary CFCs.  Plastic non-adherent cells were plated at 105 cells per ml in methylcellulose-containing serum (Methocult H4230; StemCell Laboratories, France) and supplemented with recombinant human cytokines [1 ng/ml granulocyte–macrophage colony stimulating factor (GM-CSF), 100 ng/ml granulocyte colony stimulating factor (G-CSF), 5 ng/ml interleukin 3 (IL-3), 20 ng/ml stem cell factor (SCF); all from First Link, West Midlands, UK] in 35 mm Petri dishes. For some experiments, colonies were grown in G-CSF (100 ng/ml) only or a combination of G-CSF and IL-3 (25 ng/ml). The cultures were incubated at 37°C in humidified 5% CO2 in air for 7 d.

Replating experiments.  Replating primary colonies and observing secondary colony formation provides the means to investigate the kinetics of colony formation. A primary colony that contains secondary CFCs is considered to be derived from a cell whose progeny perform one or more self-renewal divisions before terminal differentiation. The probability of clonal extinction (pE) is, therefore, provided by the proportion of primary colonies that contain no secondary CFCs, and the probability of clonal survival (pS) is 1-pE.

To examine clonal survival and changes in the numbers of CFCs per clone during sequential replating, primary colonies were grown as described above. After 7 d incubation, individual colonies consisting of more than 50 cells were plucked from the Petri dish using sterile Eppendorf pipettes, with a new tip for each colony. Each colony was then transferred to a separate well of a 96-well microtitre plate containing 100 µl of methylcellulose plus serum and cytokines. The colonies were dispersed to a single cell suspension and thoroughly mixed with the methylcellulose. The microtitre plates were incubated for a further 7 d after which the presence and number of secondary colonies of more than 50 cells in each well was scored.

The entire contents of each well containing one or more secondary colonies was transferred to a new well of a 96-well microtitre plate containing 100 µl methylcellulose plus serum and cytokines (total volume 200 µl). The contents were dispersed and mixed with the fresh methylcellulose, the plates incubated for another 7 d, and the presence and numbers of tertiary colonies per well scored. This procedure was repeated using 48-well microtitre plates and a total volume of 500 µl to obtain data for quaternary colony formation.

To study the replating ability of individual CFCs within a clone, individual primary colonies were replated into 500 µl of methylcellulose plus serum and cytokines in 48-well plates. After a further 7 d incubation, the numbers of secondary colonies per well were scored. Then each secondary colony in a well was plucked separately and transferred to a well of a 96-well microtitre plate containing 100 µl methylcellulose plus serum plus cytokines. The well of origin in the 48-well plate (i.e. the clone) was recorded for each secondary colony. Seven days later, the numbers of tertiary colonies in each well were scored.

Video recording.  Colony cultures were set up as described above. At daily intervals, the plates were removed from the incubator for a short time (< 15 min) and placed on a grid on a microscope stage (Lewis et al, 1994). The grid was designed so that the plate could be replaced in exactly the same position at each time interval and the positions of the developing colonies could be accurately recorded. Whole culture plates were filmed using a close circuit TV camera mounted on the microscope and attached to a video recorder. The recordings were reviewed retrospectively, and the positions and size of the colonies at each time point were transferred to a paper replica of the grid.

Statistical analysis. The results were analysed using Student's t-test.


Colony growth in primary culture

In order to determine the onset of proliferation and subsequent growth kinetics, the growth curves of 11 colonies were plotted. Figure 1 shows that all CFCs first divided 1–3 d after plating. Once cell division had commenced, the growth rates of the colonies were similar, with a calculated cell doubling time of 26 ± 0·8 h (mean ± standard error of the mean, SEM). Morphologically, the colonies were predominantly granulocytic (compact and composed of small cells) and the differential cell count on May–Grunwald-stained slide preparations of picked colonies was 94% granulocytic, 1% monocytic and 5% blasts. In three experiments, colonies (n = > 50 colonies per experiment) were picked on d 7 to determine colony cellularity. They consisted of 62 ± 3·5, 59 ± 2·6 and 58 ± 3·4 cells (mean ± SEM).

Figure 1.

Growth curves of 11 individual colonies reaching the 50-cell threshold by d 7 of incubation.

Clonal extinction during sequential replating

A total of 350 primary colonies in four separate experiments were replated into secondary cultures and 58 ± 13% (mean ± SEM) of them formed secondary colony cells (Fig 2) so that the probability of clonal extinction (pE) was 42%. All of the secondary colonies in each well were dispersed and replated. This step revealed that 46 ± 11% of the original 350 colonies produced tertiary CFCs. The procedure was repeated and 30 ± 7% of the primary colonies produced quaternary CFCs. Thus, the development of clones of haemopoietic progenitor cells was associated with progressive exponential clonal extinction, as has been predicted by computer simulation (Vogel et al, 1969; Gordon & Blackett, 1995), and which was consistent with symmetrical division of progenitor cells in terms of their retention and loss of clonogenic capacity.

Figure 2.

Rate of clonal extinction during three replating steps, performed at weekly intervals.

Numbers of CFCs in clones surviving replating

The 58% of primary colonies which produced secondary colonies contained, on average, 11·9 ± 2·4 (mean ± SEM) CFCs (Fig 3). The numbers of CFCs per colony fell to 10·9 ± 2·4 and 6·2 ± 1·3 after the second and third replates respectively. The 30% of clones which formed colonies after the first, second and third replating steps produced, on average, a total (secondary + tertiary + quaternary) of 40 ± 8 (mean ± SEM) CFCs, ranging in number from five to 141 per clone. These results showed that the clonogenic myeloid progenitors had a considerable capacity for amplification that cannot be explained by asymmetric progenitor cell division.

Figure 3.

Average number of CFCs (CFU-GM) per surviving clone at each replating step.

Influence of cytokines on the pE and numbers of CFCs in extant clones

To determine whether cytokine stimulation influenced the kinetics of myeloid colony formation, cultures were set up using G-CSF alone or G-CSF + IL-3. Table I shows that the cytokines could significantly modify both colony survival and the multiplicity of clonogenic cells within the extant clones.

Table I.  Influence of cytokine conditions on colony extinction (pE) and numbers of CFCs within colonies.
  • *

    Number of secondary CFCs per primary replated colony.

  • pE, probability of extinction.

pE0·76 ± 0·020·49 ± 0·060·004
CFC/colony*4·50 ± 0·599·41 ± 0·980·03

Intraclonal variation

To determine whether individual members of a clone behave in the same way or differ from one another, primary colonies were first replated into secondary cultures. Then each individual secondary colony was replated separately and analysed for tertiary colony formation. The results (Table II) showed that individual secondary colonies within a clone also contained variable numbers of CFCs. For example, in clone number 6 the primary CFC produced four secondary CFCs. When these were individually replated, one extinguished, and the other three produced 1, 2 and 5 tertiary CFCs respectively. However, there was a significant relationship between the numbers of secondary CFCs and the numbers of tertiary CFCs within individual clones (Fig 4).

Table II.  Numbers of tertiary CFCs produced by individual secondary colonies belonging to 19 individual clones.
Secondary colonies/clone1212343338174932182131375
Secondary colonies replated121234333183832182027375
Tertiary CFC/0030000000000000000
secondary colony 1 0411100000000101
     42330103 000201
      5  010  000 11
         020  000 13
         020  000 1 
         080  000 2 
         0100  000   
         0 0  000   
         0 0  000   
         0 0  001   
         0 0  001   
         0 0  001   
         0 0  001   
         0 0  101   
         0 0  303   
         0 0  303   
         1 0  613   
         1 0   15   
         1 0   25   
         1 0    5   
         2 0    9   
         2 0    11   
         2 0    12   
         2 0    12   
         2 0    13   
         3 1        
         3 1        
         3 1        
         5 1        
         7 1        
Figure 4.

Relationship between the numbers of secondary CFCs and the numbers of tertiary CFCs in individual clones.


How a single cell can divide to produce two daughter cells that adopt different fates is one of the central questions in developmental biology, and in haemopoietic cell proliferation and differentiation in particular. According to the hierarchical model of haemopoiesis, stem cells have the capacity to self-renew and to differentiate to form cells belonging to the multiple lineages found in the blood. However, it is clear that not all of these fates can be adopted by a single stem cell (Solberg, 2001). Also, the idea that there is a sudden transition from stem cell to progenitor cell rather than a continuum is becoming less strongly held (Slayton et al, 2001; Quesenberry et al, 2002). This view is consistent with the idea that stem cells and progenitor cells may differ, in terms of their self-renewal capacity, by degree but not by kind. It is also consistent with the idea that progenitor cells may function as an amplifying transit population in haemopoiesis (Gordon & Blackett, 1994; Loeffler & Potten, 1997). We used a colony replating assay to address questions about clonal extinction, clonal expansion and the generation of progenitor cell heterogeneity through proliferation. This system had the advantage that it enabled investigation of the clonal production of progenitor cells with the same colony-forming capacity as the parent cell.

When we followed the proportion of colonies that extinguished (i.e. contained no new CFCs) at each replating step, we found a continuous exponential decline in colony survival. Similarly, Brummendorf et al (1998) found that some clones extinguished after 16 d while others proliferated for 60 d or more; Cho and Muller-Sieberg (2000) found that 50% of stem cell-derived clones lost haemopoietic repopulating ability, and culture of marrow cells in vitro leads to impairment of engraftment ability (Peters et al, 1995, 1996). A similar pattern of clonal extinction has been predicted for haemopoietic stem cells, even under steady-state conditions (Vogel et al, 1969; Gordon & Blackett, 1995). However, if the majority of stem cells remain quiescent, the expression of stem cell clonal deletion may be delayed considerably. Therefore, it is possible that both haemopoietic stem cells and progenitor cells have capacities for self-renewal, albeit of differing degrees, and clonal extinction, but that extinction is accelerated in the progenitor cell population because of its greater propensity for being in active cell cycle.

To determine whether expansion of progenitor cell number could be detected during colony formation, we measured the numbers of CFCs in extant clones surviving replating. The results showed that the 58% of primary CFCs that contained secondary CFCs were capable of a ∼10-fold expansion of progenitor cell numbers. By following individual clones through the three replating steps, we were able to demonstrate that a single primary CFC could produce up to 140 new CFCs over the course of the experiment. This result cannot be explained by asymmetric progenitor cell division (i.e. one progenitor forming one clonogenic cell and one non-clonogenic cell) because, theoretically, the maximum number of stem cells per clone would always be one. It could be explained by either of two models whereby the decision to differentiate is made before or after mitosis.

The signals determining which cells retain or lose their clonogenic capacity on division are not well understood. However, combinations of cytokines have formed the basis of most in vitro stem/progenitor cell expansion protocols. Nonetheless, opinions about the importance of cytokines for progenitor cell fate differ (Enver et al, 1998; Metcalf, 1998). Here we have shown that the probability of clonal extinction and the numbers of clonogenic cells per colony (a reflection of self-replication divisions) can be modified by extracellular signals, supporting a cell-extrinsic mechanism as part of the regulation of haemopoietic progenitor cell kinetics.

We found no evidence of uniform proliferative behaviour among individual CFCs within a single clone. This agrees with earlier results from in vitro studies using clonogenic assays (Brummendorf et al, 1998) but not with recent results from in vivo experiments using serial transplantation (Muller-Sieberg et al, 2002). It is noteworthy that both we, and Brummendorf et al (1998), recloned progenitor cells at each replating step; in contrast, serial transplantation was done by transferring a fixed number of cells containing unknown numbers of stem cells. It is relevant, therefore, that there was a relationship between the numbers of clonogenic cells in successive generations within a clone because the results of serial transplantation may reflect the fact that large clones are more ‘successful’ than small clones as they contain more clonogenic progeny and can persist for longer (Gordon & Blackett, 1995).

In summary, haemopoietic colony formation and microcosmic haemopoietic cell development are characterized by clonal extinction and a balance between cell proliferation and cell differentiation within developing clones. The probabilities of proliferation and differentiation can be modified by extrinsic factors such as cytokines but there was no evidence of uniform proliferative behaviour among individual members of a clone. These data are in accordance with computer simulations of stem cell behaviour (Vogel et al, 1969; Gordon & Blackett, 1995) and seem to fit a stochastic model of haemopoietic progenitor cell development where event probabilities can be modified by extracellular conditions.


This work was supported by a Specialist Programme Grant from the Leukaemia Research Fund of Great Britain.