A cautionary tale: how to delete mouse haemopoietic stem cells with busulphan.

In this study treating mice with the 'correct' dose of busulphan did not necessarily destroy all haematopoietic stem cells. In certain circumstances host stem cells survived undetected and subsequently resumed haemopoiesis. This may apply to the use of busulphan clinically. We found that the following conditions determined the deletion of mouse stem cells using busulphan: (1) graft size--grafting more than 106 marrow cells ( approximately 0.3% of the animal's stem cells) concealed the survival of stem cells; (2) dose of busulphan--insufficient busulphan did not kill all host stem cells; (3) old or improperly stored busulphan failed to delete all host stem cells; furthermore (4) the survival of host stem cells should be assessed by typing many kinds of circulating cells; and (5) tests should be carried out to determine if busulphan has killed all host stem cells by typing circulating blood cells at appropriate intervals.

Gap-junction communication pathways in germinal center reactions.

Intercellular channels called gap junctions enable multicellular organisms to exchange information rapidly between cells. Though gap junctions are held to be ubiquitous in solid tissues, we have only recently found them in the lymphoid organs. Functional direct cell-cell communication has now been confirmed by us and other groups in bone marrow, thymus, and in secondary lymphoid tissues. What functions do they serve in the lymphoreticular system where, so far, only cytokines/growth factors and adhesion molecules have been considered as regulators? Here we show evidence for and refer to published work about functional direct cell-cell communication through gap junctions in germinal center reactions and make proposals for their role in the immune response. We found a large amount of the connexin43 (Cx43) gap junctions in the germinal centers of secondary lymphoid follicles. Ultrastructurally and immunohistologically, most of the junctions were detected on the processes of follicular dendritic cells (FDC) enveloping nondividing centrocytes in the light zone of germinal centers where B-cell selection is thought to take place. Further support for this finding came by revealing the Cx43 mRNA in situ at the same location as the protein. On antigen challenge, gap junctions appeared on the FDC as they formed meshworks in germinal centers. In order to find out which germinal center cells communicate directly, we separated FDC-rich, low-density, B-cell fractions from human tonsil. In culture, we injected single FDC with the low-molecular-weight fluorescent dye, Lucifer Yellow (M(r) 457 Da), which passed between adjacent FDC and sometimes from FDC to B cells. Based on these findings and their assigned functions in other tissues, gap junctions may contribute to germinal center reactions in the following ways: (1) they may regulate follicle pattern formation by controlling FDC growth, (2) they may be involved in FDC-B-cell signaling contributing to the final rescue of selected B cells from apoptosis, and (3) they may enable FDC to work as a functional syncytium providing a cellular internet for integrating germinal center events. Data supporting these interpretations are briefly discussed.

Connexin43 gap junctions in normal, regenerating, and cultured mouse bone marrow and in human leukemias: their possible involvement in blood formation.

Communicating channels called gap junctions are thought to play a ubiquitous part in cell growth and development. Based on earlier work, we have recently found functional evidence of their presence in human and mouse bone marrow. In this study we studied the cell-type association of the gap junction channel-forming protein, connexin, in mouse and human bone marrow under different physiological and pathological conditions and tested the pathway of communication in bone marrow cultures. For high-resolution antigen demonstration we took advantage of semi-thin resin sections, antigen retrieval methods, immunofluorescence, and confocal laser scanning microscopy. Connexin43 (Cx43) and its mRNA were consistently expressed in human and rodent marrow. Cx37 was found only in the arteriolar endothelium, but neither Cx32 nor -26 were expressed. In tissue sections, the immunostained junctions appeared as dots, which were digitally measured and counted. Their average size was 0.40 mm in human and 0.49 mm in mice marrow. There were at least twice as many gap junctions in the femoral midshaft of 6-week-old mice (1.75 x 10(5)/mm3) as in those older than 12 weeks (0.89 x 10(5)/mm3). Most Cx43 was associated with collagen III+ endosteal and adventitial stromal cells and with megakaryocytes. Elsewhere, they were few and randomly distributed between all kinds of hematopoietic cells. In the femoral epiphysis of juvenile mice, stromal cell processes full of Cx43 enmeshed three to six layers of hematopoietic cells near the endosteum. The same pattern was seen in the midshaft of regenerating mouse marrow 3 to 5 days after cytotoxic treatment with 5-fluorouracil. Functional tests in cultures showed the transfer of small fluorescent dyes, Lucifer Yellow and 2',7'-bis-(2-carboxyethyl)-5, 6-carboxyfluorescein, between stromal cells and in rare cases between stromal and hematopoietic cells too. The stromal cells were densely packed with Cx43 and we found aggregates of connexon particles in their membrane replicas. In normocellular human bone marrow, gap junctions were as rare as in adult mouse and similarly distributed, except that they were also on adipocytic membranes. In a few leukemic samples, characterized by an increased stromal/hematopoietic cell ratio, there were two- to fourfold more Cx43 (2.8 x 10(5) to 3.9 x 10(5)/mm3) than in the normal (1.0 x 10(5) to 1.2 x 10(5)/mm3). The cases included a hypoplastic acute lymphoblastic leukemia, an acute myeloid leukemia (French-American-British classification M4-5), a case of myelodysplastic syndrome with elevated number of megakaryocytes, and a CD34+ acute hemoblastosis, probably acute myeloid leukemia (French-American-British classification M7). Taken together, our results indicate that direct cell-cell communication may be involved in hematopoiesis, ie, in developmentally active epiphyseal bone marrow and when there is a demand for progenitors in regeneration. However, gap junctions may not play as important a role in resting adult hematopoiesis and in leukemias.

Does transmembrane communication through gap junctions enable stem cells to overcome stromal inhibition?

When long-term bone marrow cultures are treated with Amphotericin B (AB) their haemopoietic stem cells (HSC) cease growing. This is not a toxic effect of the drug because once that is removed, HSC resume clonal growth and, given sufficient time, form as many cells as HSC in untreated cultures. Amphotericin B-evoked inhibition of blood formation is probably mediated by transmembrane communication between HSC and stroma for the following reasons: (1) AB does not stop HSC forming colony-forming units in culture (CFU-c) when HSC are separated from stroma by culturing them on Transwell inserts above the stroma. (2) Conditioned media (CM) from AB-containing or normal long-term cultures (LTC) does not inhibit normal marrow cells forming colonies in semi-solid cultures without stromal underlays. (3) AB itself does not stop bone marrow cells forming colonies in semi-solid cultures nor does it stop stromal cells growing or prejudice their long-term maintenance. (4) Furthermore, growing stromal cells with AB does not alter the number of transcripts they form for cytokines and chemokines to any large extent, including TGF-beta1. We have extensive, though circumstantial, evidence that gap junctions are involved in this communication. AB only stopped the growth of HSC when we blocked intercellular communication via gap junctions (GJIC) (tested by micro-injection of lucifer yellow). Lipophilic compounds that do not affect GJIC had no effect on the growth of HSC. Looking at a series of stromal cell lines from foetal liver and neonatal bone marrow we found that extensive GJIC correlated with stromal support of the late-appearing clones formed by primitive HSC (week 3-5 cobblestone-area forming cells, CAFC). We propose that the proliferation of HSC is regulated via transmembrane communication between stromal and HSC. Our findings support the proposal that gap junctions play a part in this stromal-dependent regulation.

Direct cell/cell communication in the lymphoid germinal center: connexin43 gap junctions functionally couple follicular dendritic cells to each other and to B lymphocytes.

Direct cell/cell communication occurs through gap junctions (GJ). We mapped GJ expression in secondary lymphoid organs and found, for the first time, a high density of connexin43 (Cx43) GJ in follicular dendritic cells (FDC) in close association with lymphocytes (Krenacs T. and Rosendaal M., J. Histochem. Cytochem. 1995. 43: 1125-1137). In this work, we used a combination of ultrastructural, immunocytochemical, molecular methods, and functional dye transfer experiments to study which germinal center cells are involved in direct cell/ cell communication and how GJ expression is regulated during antigen responses. One week after injecting the footpad of mice with 50 micrograms lysozyme, Cx43 GJ were detected on elongated cells in the paracortex of their popliteal lymph nodes. Repeated challenge led to the formation of secondary follicles with enlarged FDC meshwork full of Cx43 GJ. This positive correlation may reflect an importance for GJ in the pattern formation of FDC and lymphoid follicles. In human tonsil, the density of GJ and FDC was highest in the light zone of germinal centers where the fate of B cells is thought to be decided. Cx43 colocalized with CD21 and CD35 antigens in the vicinity of desmosomal junctions on FDC embracing lymphocytes. Freeze-fracture hallmarks of GJ of 200-400 nm were also found on FDC in the vicinity of desmosomal plaques. Furthermore, Northern blot analysis showed the consistent presence of Cx43 mRNA in human tonsil and spleen. Most Cx43 message was localized in situ to cells with FDC morphology and some to a few germinal center lymphocytes. To investigate functional cell coupling, we set up FDC/B cell cultures from the low density cell fractions of human tonsils. Cx43 plaques associated with lymphocytes were detected both on elongated FDC processes in early cultures (up to 4 h) and in established FDC/B cell clusters (between 4 and 24 h). In early cultures, we injected FDC with Lucifer Yellow, a fluorescent dye which passes through GJ: the dye spread into adjacent FDC and occasionally from FDC into CD19+ B cells. Based on these results, we propose that direct cell/cell communication through Cx43 GJ is involved in FDC/FDC and in FDC/B cell interactions. The functionally coupled FDC meshwork may serve as a communication channel synchronizing germinal center events. FDC may also deliver crucial direct signals through GJ involved in the rescue of high-affinity B cell clones from apoptotic cell death.

Immunohistological detection of gap junctions in human lymphoid tissue: connexin43 in follicular dendritic and lymphoendothelial cells.

We investigated the expression of gap junction connexins26, -32, and -43 in normal, reactive, and diseased human lymphoid tissue with single and double immunolabeling and confocal laser scanning microscopy. In all tissues, connexin43 positivity was detected in follicular dendritic cells positive for CD21 and CD35 antigens, around lymphoendothelial cells moderately positive for Factor VIII, CD31 and cathepsin-D antigens; and somewhat in vascular endothelia including high endothelial venules strongly positive for Factor VIII and CD31 antigens. The ultrastructural hallmark of gap junctions, pentalaminar structures with appropriate spacing, was found in follicular dendritic cell processes. Connexin43 was also detected between smooth muscle and stromal cells of the gut, in capsular fibroblasts, and in tonsil epithelium. Neither connexin32 nor -26 was revealed, except for connexin26 in the tonsil epithelium. In follicular dendritic cells, connexin43 co-localized closely with the desmosomal proteins desmoplakin and desmoglein, suggesting that cell adherence has a role in gap junction formation. Most connexin43 was observed in sinus lining cells of lymph nodes involved in malignancies and in follicular dendritic cells in the light zone of germinal centers where maturing but still proliferating lymphocytes are situated. In the light of their distribution, gap junctions may play a part in regulating the growth of germinal centers and in integrating activating or controlling signals in follicular dendritic and sinus lining cell networks. Because connexin43 is the connexin of stromal cells, finding it in follicular dendritic cells in consistent with the proposal that these cells originate from resident stromal cells.

Gap junctions in blood forming tissues.

More than ten research groups have now reported the presence of gap junctions in blood-forming tissue or cultured cells. It is time to accept that these cell-coupling structures are present in this tissue. To find out what they are doing here we need to develop appropriate experimental techniques. This review covers the particular problems of investigating direct cell-cell communication by gap or other junctions in undisturbed haemopoietic tissue. It then describes and assesses the published reports of haemopoietic gap junctions. Recently, in the author's laboratory, three means of increasing the number of gap junctions 50- to 100-fold in mouse marrow have been described, as well as techniques for doing so in culture. There is a complete report of this work here. At present it is quite unclear what function gap junctions serve in blood-formation, perhaps it is some consolation that 30 years after their ultramicroscopic discovery it is also true for all other unexcitable tissues. Possibly the ability to up-regulate their expression in haemopoietic tissue will help us find out what their role is in blood formation.

Up-regulation of the connexin43+ gap junction network in haemopoietic tissue before the growth of stem cells.

The early developmental stages of haemopoiesis are thought to be regulated by paracrine growth factors and by the haemopoietic environment. Are gap junctions involved here? Gap junctions are structures in cell membranes allowing the direct transfer of ions and small molecules between adjacent cells and are known to be involved in development. We have found that although connexin43 gap junctions are rare (0.00016 +/- 0.0002/microns2 tissue) in normal adult mouse marrow their expression is 80-fold higher (0.0292 +/- 0.0147/microns2) in neonatal marrow. One difference between neonatal and adult haemopoietic tissue is that in the latter more haemopoietic cells are dividing. To test if more gap junctions were due to increased division we altered adult blood-formation by mobilizing or destroying end cells--granulocytes and red cells--or by forcing stem cells to divide by making them regenerate an ablated blood-forming system. Mobilizing end cells had no effect on the number or distribution of gap junctions in marrow but forced stem cell division caused a 100-fold increase in gap junction expression and did so before any recognizable haemopoietic cells formed. There were greater than normal numbers of gap junctions in radio-protected adult mouse marrow. The cells coupled by gap junctions are TE-7+ mesodermally derived fibroblasts, STRO-1+ stromal cells, and CD45+ and CD34+ haemopoietic cells. We propose that there is a latent network of Cx43+ gap junctions in normal quiescent marrow. In response to events that call for active division of stem cells this network is amplified and coupled to haemopoietic stem cells, perhaps enabling them to divide.

Direct cell-cell communication in the blood-forming system.

In mammals, bone marrow is the principal tissue where blood is formed during adult life. Paracrine factors are generally considered to control this process but there is considerable evidence that gap junctions are present in haemopoietic tissues. Gap junctions have been implicated in developmental and patterning roles, and we set out to characterize the cells which are coupled, and to provide evidence for their role(s) in blood cell formation. Direct cell-cell communication, shown by dye-transfer, occurs between haemopoietic cells and certain stromal cells. In culture these stromal cells form a mat in which they retain their dye-coupling properties. Freeze-fracture electron microscopy confirms that this coupling is via gap junctions. When haemopoietic cells are cultured on top of these mats dye spreads upwards from the stromal cells into the haemopoietic cells above. Experiments in which haemopoietic cells were cultured alone, with stromal cell conditioned medium, or in direct contact with stromal cell underlays, were therefore carried out. The results of these experiments provide evidence that gap junctional communication may be playing a vital role in maintaining populations of precursor cells which would otherwise differentiate into end cells, leading to the ultimate demise of the system.

The effect of stem cell proliferation regulators demonstrated with an in vitro assay.

Spleen colony formation after transplantation of bone marrow cells into irradiated mice has been used as an assay for hematopoietic stem cells (CFU-S), but has serious limitations intrinsic to an in vivo assay. In this report we describe experiments using an in vitro clonogenic assay that is especially suitable for studies of stem cell regulation as defined growth factors and normal untreated bone marrow can be used. We have demonstrated that the colony-forming cells have proliferative properties in common with CFU-S and respond to specific proliferation regulators previously detected using the spleen colony assay.

Patterns of haemopoietic recovery after stress--II. Treatment with fluorouracil.

The mouse haemopoietic system is not permanently damaged by repeated injections of cytotoxic fluorouracil. It contains approximately normal numbers of nucleated femoral and spleen colony-forming cells (CFUs) after seven monthly injections of the drug and normal numbers of high proliferation potential colony-forming cells (HPP-CFC) after five serial injections. Furthermore, the mouse is fully fertile after seven injections of fluorouracil. The mouse recovers quickly after treatment because it regenerates from cells which were not killed by the drug. Within 14 days of treatment with fluorouracil there are almost twice the normal number of femoral macrophage and high proliferation potential colony-forming cells (M-CFC and HPP-CFC). These numbers then fall but are returning to normal 6 weeks after the drug was administered. In this quick recovery the response of the haemopoietic system differs from its response to the loss of blood cells caused by sub-lethal irradiation, or lethal irradiation, or treatment with busulphan. When mice are treated twice with fluorouracil, the second injection 14 days after the first, the number of femoral M-CFC two days after the second injection, is 16-fold the number in controls, but the number of femoral HPP-CFC is only twice the number in controls. When the interval between the two injections is 21 days, the number of femoral M-CFC is almost 8% of that of mice treated once, but the number of HPP-CFC is 67%. The characteristic response of each type of cell to repeated treatment with fluorouracil is probably due to the number of its precursors which are killed by the drug and to the interval between successive injections. A second injection of fluorouracil, 28 days after the first, speeds the growth rate of HPP-CFC. Their doubling time is 6 h shorter than that of mice treated once. Haemopoietic tissue from mice treated repeatedly with fluorouracil can only outgrow normal marrow under certain conditions. The nature of these conditions and the mechanisms involved are discussed in the light of contradictory findings.

Patterns of recovery of high proliferation potential colony-forming cells after stressing the haemopoietic system--I.

The rates at which the number of high proliferation potential colony-forming cells and other haemopoietic cells recovered after different first stresses and a standard second stress were studied. The following first stresses were compared; different doses of sublethal irradiation (2.57, 4.84 and 5.5 Gy) followed by endogenous repopulation; lethal irradiation followed by exogenous repopulation; lethal irradiation of radio-protected mice followed by endogenous repopulation; and treatment with busulphan. Six to 16 weeks after these first stresses a standard second stress was applied. This was i.v. injection of fluorouracil. Two, four and six days later the number of high proliferation potential colony-forming cells in femora was determined and recovery curves for these cells were calculated. Their number increased exponentially in this period in all mice studied except radio-protected, lethally irradiated ones. In these, the exponential increase occurred between four and eight days after fluorouracil. The rate of increase was faster than normal in sublethally irradiated and radio-protected, lethally irradiated animals whose haemopoietic systems repopulated endogenously; in lethally irradiated, exogenously reconstituted animals it was the same as normal and in busulphan-treated animals it was slower than normal. The marrow from these differently stressed mice was also cultured with seven doses of two synergistic factors to contrast the growth of high proliferation potential colony-forming cells in the mice whose first stresses had differed. The cultures were assessed automatically by the CLIP 4 image processor. The high proliferation potential colony-forming cells of sublethally irradiated and busulphan-treated mice required more synergistic factor than normals to form a given number of cells/femur.

Haemopoietic progenitors in different parts of one femur perform different functions during regeneration.

Femoral haemopoietic tissue was divided into cells released by flushing and cells released by grinding and washing flushed femora. The flushed femur contained 5 times more nucleated cells than the ground femur, 40 times more macrophage colony-forming cells and 6 times more developmentally late, day 8, and developmentally early, day 13, spleen colony-forming cells. However, the ground femur contained 2 times more developmentally early high proliferation potential colony-forming cells and 3 times more late ones. Haemopoietic regeneration of mice treated with fluorouracil was compared in samples obtained by flushing alone and grinding flushed femora. The number of nucleated cells recovered by flushing fell thirteen-fold by the sixth day after administration of the drug and the number recovered by grinding fell six-fold by the eighth day. Developmentally early high-proliferation-potential colony-forming cells which were recovered by grinding doubled their number in half the time taken by similar cells recovered by flushing. These observations are consistent with haemopoietic cells in different parts of the same bone performing different functions during regeneration. Large numbers of high-proliferation-potential colony-forming cells were not found in the circulation until 8 days after treatment with fluorouracil. Five days after mice had been treated with fluorouracil, when their blood forming systems were regenerating, early high-proliferation-potential colony-forming cells in one sample of marrow were derived from different founder cells than were late cells in the same sample. At the same time, early high-proliferation-potential colony-forming cells in the ground sample of a femur were derived from different founder cells than were cells at the same stage of development in the flushed sample of the femur. These observations are consistent with the view that haemopoietic regeneration after treatment with fluorouracil is due to the growth of few founder cells whose progeny have migrated little within 5 days of drug treatment.

Correspondence between the development of hemopoietic tissue and the time of colony formation by colony-forming cells.

The development of a haemopoietic tissue and the time when colony-forming cells in it formed detectable colonies were studied with in vivo spleen colony-forming units (CFUs) and in vitro high-proliferation-potential colony-forming cells (HPP CFC). Cells that form colonies first are developmentally more mature than those doing so later. Marrow containing mature spleen colony-forming cells formed fewer cells in the femora of recipients than that which contained early colony-forming cells. The growth curve of developmentally early high-proliferation potential-colony-forming cells was steeper than that of later cells. The time period before colony-formation occurs is a property of the colony-forming cell and is not due to regulatory mechanisms in the animal or to regulatory cells in the haemopoietic stroma.

Regulation of very primitive, multipotent, hemopoietic cells by hemopoietin-1.

Hemopoietin-1 (H-1) is known to act synergistically with CSF-1, a mononuclear phagocyte growth factor, to induce the development of primitive hemopoietic cells. To determine whether purified H-1 also acts on multipotent hemopoietic cells, its ability to act synergistically with interleukin-3 (IL-3) and erythropoietin (Epo) was tested in methyl cellulose cultures of murine bone marrow cells. In the presence of IL-3, H-1 increased the number of colonies formed by primitive, multipotent cells by approximately 30-fold. H-1 alone or H-1 plus Epo produced no colonies. Forty percent of the colonies induced by H-1 plus IL-3 contained cells that could be subcultured at least twice, whereas cells from colonies induced by IL-3 alone could not be similarly subcultured. Thus H-1 permits CSF-1 or IL-3 to act on cells more primitive than those acted on by either growth factor alone. The results indicate that H-1 acts on the most primitive hemopoietic cells yet shown to proliferate and differentiate in culture.

Assessing cultured colonies automatically.

The number of colonies formed by macrophage colony-forming cells and high proliferation potential colony-forming cells was assessed by an image processor. The processor counted and sized colonies accurately, reproducibly, rapidly (2 s/dish) and objectively. The processor also measured the amount of light (in grey levels) the colonies transmitted. The optical density of a colony (the sum of its grey levels) was related to its cellularity. Thus the image processor compared both the number of colonies in samples and their cellularity. Samples of marrow containing high proliferation potential colony-forming cells of different proliferative capacity were prepared by injecting fluorouracil into mice and collecting their marrow 2-10 days later (marrow samples called FU2-FU10). These samples were cultured with one of three sources of synergistic factor titrated over seven dilutions. Colonies contained approx. 5 X 10(4) cells after 11 days culture but the way that FU2-FU10 marrow grew depended on the interval between treating donors with fluorouracil and collecting their marrow. Samples collected 2-4 days after fluorouracil formed more colonies containing more cells with small increases of synergistic factor whereas samples collected after 8-10 days did neither. It was important to culture samples of marrow with the appropriate synergistic factor for the interval after fluorouracil. Factor(s) derived from the 5637 cell line acted optimally on high proliferation potential colony-forming cells in samples collected 2-8 days after fluorouracil, and factor(s) derived from Wehi 3B cells on high proliferation potential colony-forming cells in samples collected 6-10 days after fluorouracil. Factor(s) derived from placental conditioned medium acted well on samples collected between 2 and 10 days. The proliferative capacity of samples of marrow could also be compared by estimating growth curves for high proliferation potential colony-forming cells in samples collected at successive intervals after fluorouracil.

Haemopoiesis by clonal succession?

Alloenzymes of phosphoglycerate kinase (Pgk-1A and 1B), an X-linked enzyme of the glycolytic pathway that is expressed on all somatic cells, were used to study properties of developmentally early haemopoietic cells of mice (high proliferation potential colony-forming cells, HPP CFC) This study showed that: The colonies formed by HPP CFC are clones; There are distinct populations of HPP CFC that can respond to different haemopoietic growth factors; There are substantial differences in the proportions of Pgk alloenzymes of similar sets of HPP CFC found in similar and neighbouring bones; and In the whole animal, there is no evidence of any single population of HPP CFC that is permanently derived from a limited set of precursors.

Early haemopoietic responses to Salmonella typhimurium infection in resistant and susceptible mice.

The response of colony forming cells in the bone marrow and spleen of resistant (CBA) and susceptible (C57BL) mice to Salmonella typhimurium infection was studied for 4 days after infection. The number and size of the colonies were assessed. The resistant strain exhibited an immediate response to challenge, sharply increasing the number of colonies to 2.5 times normal over 2-3 days after infection. In contrast the susceptible strain gave a slowly increased response to the same challenge, which never exceeded 1.2 times normal and fell to 0.8 times the normal. When mouse strains were immunized there was a clear distinction between the splenic and bone marrow cellularity. Immunization appeared to enhance the splenic cellularity in resistant mice but failed to in susceptible mice. In the bone marrow of susceptible mice, however, there was some evidence of an elevated response.

Physical and kinetic properties of haemopoietic progenitor cell populations from mouse marrow detected in five different assay systems.

Properties of haemopoietic progenitor cells detected in several different assays have been compared in order to position them within the haemopoietic developmental lineage. The spleen colony-forming cell (CFUs), the high proliferation potential colony-forming cell (HPP-CFC) and two granulocyte-macrophage colony-forming cells (GM-CFC-1 and GM-CFC-2) have been studied. Two experimental techniques were used: separation of cells on the basis of their buoyant density and comparison of the survival of haemopoietic cells after donor mice had been injected with the cytotoxic drug 5-fluorouracil (5-FU). On linear BSA gradients the modal buoyant densities of CFUs, HPP-CFC and GM-CFC-1 were the same, 1.070 g cm-3; the density of GM-CFC-2 was higher, 1.075 g cm-3. GM-CFC-2 colonies were much smaller and contained far fewer cells than HPP-CFC or GM-CFC-1 colonies, even after prolonged culture, and this suggests that dense haemopoietic progenitors have less proliferation potential. This was confirmed by comparison of the size of colony formed, under identical culture conditions, by progenitors of different densities. Mean colony diameter was inversely related to the density of the progenitor cell. With the exception of GM-CFC-1, low density progenitors were more resistant to the cytotoxic effects of 5-FU than high density precursor cells (GM-CFC-2). Consequently, the GM-CFC-1 could be distinguished from GM-CFC-2 on the basis of buoyant density and from the other low density populations on the basis of post-FU kinetics. The reasons why the GM-CFC-1 should be more sensitive to 5-FU than other low density progenitors are discussed and the relation of these low density precursors to one another in terms of their position within the haemopoietic developmental lineage is elucidated.