1- The need for a truly serum-free medium for the growth of adult human circulating early erythroid progenitors. 

1.1- The objective of creating a truly serum-free medium

The objective of obtaining successful serum-free (SF) conditions for tissue culture is to be able to examine defined growth requirements reliably, as well as to study the effect of specific, non-nutritional, peptide growth factors (GFs) and any other molecules, in the absence of the myriad of uncontrolled stimulatory and inhibitory factors present in serum. As we shall see, the same problems posed by serum-containing (SC) cultures arise when apparently serum-free (i.e. "serum-free") media are prepared with serum-derived, purified components that are contaminated with unidentified factors. The objective of creating a truly SF medium is to be able to completely define the culture conditions, using recently available recombinant growth factors and the highest purity for all ingredients. These are conditions that depend upon the refinement of our current technologies and are perforce subject to constant improvement.  

1.2- Serum-deprived and serum-free media for erythropoiesis in vitro  and the problem of cleaning albumin

The development of serum-deprived (SD, or "serum-free", "SF") media capable of supporting erythroid growth and differentiation originated in the 1980 paper of Iscove, Guilbert and Weyman (1). These workers described how the replacement of serum by charcoal-delipidated and deionized bovine serum albumin (BSA), iron-saturated transferrin and BSA-adsorbed, sonicated lipids (oleic and linoleic acids, lecithin and cholesterol), provided an assay system for the growth of murine late erythroid progenitors from bone marrow (colony-forming units, erythroid, or CFU-E which form erythrocytic colonies) (1). But in order to support the growth of murine early erythroid progenitors (burst-forming units, erythroid, or BFU-E), this culture system required the presence of serum (fetal bovine serum, FBS), albeit at a much reduced concentration compared to that used for plasma or methylcellulose SC assays (2). (For a diagram of the developmental and conceptual relationships involved in erythropoiesis, the reader is referred to Fig.1 and to Appendix A.)

Subsequently, Stewart et al. developed a SD medium capable of supporting the growth of murine BFU-E, by altering the method of preparation of the lipid mixture, which was directly solubilized in ethanol with unfiltered cholesterol and physically adsorbed to BSA (3), instead of being simply sonicated with the BSA (2). The number of murine erythropoietic bursts that developed under these SD conditions was directly dependent upon the concentration of albumin in the medium (3). However, these workers also found that charcoal-delipidated BSA failed to support burst-formation, and thus concluded that either the delipidation process irreversibly damages the BSA and/or that some substance associated with albumin and required for burst-formation is removed by the treatment (3). Hence, burst-formation could only be seen to occur in the presence of undefatted, contaminated, deionized BSA.

This role of the albumin is central to the problem of developing a truly SF medium for burst-formation. Albumin has two main functions in such a medium: it buffers pH changes and it serves as a non-specific carrier of ions, lipids, small nutrient molecules and peptide growth factors, which it delivers to the cells by virtue of its adhesive properties (1-2). Amongst the substances it transports, albumin also conveys a molecule or molecules necessary for burst-formation (3-4).

1.3- The problem of undefined sources of   burst-promoting activity

Application of this methodology to the growth of adult human bone-marrow BFU-E led Dainiak et al. to the findings that burst-formation also required non-delipidated albumin (delipidation only occurred if the albumin was precoated with dextran before being adsorbed with charcoal) as well as purified lipoproteins (4). Moreover, this SD medium only performed satisfactorily if cell concentrations >105 were plated (4). Clearly, some unknown factor, a burst-promoting activity (BPA), normally present in serum and present equally in purified albumin and lipoprotein preparations, was necessary for promoting the proliferation of human and murine BFU-E (1,3,4). The negative result that followed delipidation of albumin could be partially counterbalanced by the addition of leukocyte conditioned medium (LCM), but it was also demonstrated that accessory cells present in the marrow suspension could contribute an endogenous BPA (4).

In fact, such BPAs were found to be present in media conditioned by human T-lymphocytes (5-8), monocytes (6-8), macrophages (9), radio-resistant marrow (10) and peripheral blood mononuclear cells (11). Other investigators were able to employ a delipidated BSA and to avoid using undefined lipoproteins, by adding undefined, serum-deprived, T-cell conditioned medium together with hemin and purified bovine pancreatic insulin, to the culture system (12). Under these circumstances, it was possible to obtain burst-formation in the absence of erythropoietin (Epo) (12). But it was also reported that, notwithstanding the delipidation of BSA (using the dextran step), the purified Epo preparations were also contaminated with BPA (13); furthermore, use of recombinant human (rHu) Epo made it necessary to provide a BPA by adding purified lipoproteins (14-15).

The ubiquitous presence of BPA(s) in serum, in non-defatted albumin, in purified ingredients (lipoproteins, Epo and others), as well as in conditioned medium from accessory cells underlines the importance of utilizing maximally defined components (e.g. recombinant growth factors) in order to overcome the insufficiencies of SD media systems. This is especially so when purified components (albumin, lipoproteins, Epo, insulin, etc) are prepared from serum and serve therefore as a source of undefined contaminants that can be critical for initiating and maintaining burst-formation. 

1.4- The need for a defined BPA in a serum-free medium

The problem of providing a defined BPA has been the major stumbling block in the development of a truly defined SF medium for human adult burst-formation. Most culture protocols are, in fact, solely SD assay systems that utilize undefatted albumins (typically deionized Cohn Fraction V) and lipoprotein fractions (16-20), even when they employ rHu Epo together with recombinant human Interleukin-3 (rHu IL-3) or granulocyte/macrophage colony-stimulating factor (rHu GM-CSF) (18-20). Despite incipient claims to SF culture systems (12-14,16-17,19), there is only one published protocol, which utilizes a fatty-acid-free and globulin-free BSA (FAF/GF BSA), that actually qualifies as an SF medium, even if it is not completely defined chemically (21-22). This SF medium supports the multi-lineage growth of hemopoietic progenitor cells (purified as null cells) from human bone marrow (BM) and, unlike most of the SD protocols which required the presence of crude, non-enriched BM cell populations at high concentrations in order to work (4,12,16- 17), the medium of Sonoda et al. performed well with low cell concentrations (21-22). In this assay system, erythroid burst-formation from human adult BM BFU-E was observed in vitro to require a defined BPA; in this respect, rHu IL-3 was found to be significantly more effective than was rHu GM-CSF (22). However, the same workers also found evidence that, aside from IL-3 and GM- CSF, other BPA-like factors present in serum were equally relevant for burst-formation (22).

1.5- The need for a serum-free medium for the growth of human adult circulating BFU-E

At present there is no culture system for the growth of human adult BFU-E from peripheral blood (PB) comparable to the SF system that Sonoda et al. designed for the growth of human BM BFU-E (21-22). Moreover, human adult BM is a richer source of BFU-E than is human adult PB, and a large proportion of BM BFU-E are actively undergoing DNA synthesis, whereas most, if not all, normal BFU-E circulating in the blood are not cycling (23-25). In a parallel fashion, it has also been shown that, like BM BFU-E, human fetal liver BFU-E are also cycling (24). In this sense, other workers have also reported that human embryonic BFU-E, but not adult PB BFU-E, can be grown with rHu Epo alone (i.e. without IL-3 or GM-CSF) in a SD medium, when seeded at very low cell concentrations (26); the medium employed, however, also contained undefatted BSA and purified lipoproteins, and it was therefore typically contaminated with undefined BPA. So far, published research on burst-formation from PB has only been performed with SD media, using BPA-contaminated BSA (13,16-17,19-20, 26-28), BPA-contaminated Epo (13), purified lipoproteins (14-15, 20, 26, 27), and often large quantities of purified insulin (12,17,18, 26, 28).

In light of this, if one is to ascertain the cellular and molecular regulation of erythropoiesis in vitro, using normal human adult PB BFU-E, it is necessary to create a SF medium based upon a BPA-free BSA (such as FAF/GF BSA), which will make use of recombinant growth factors only, and will contain a defined, exogenous BPA. These criteria also demand that the medium be capable of supporting low seeding densities of mononuclear cells, and that they be purified of accessory cells. Only under such ideal conditions can we ascertain the exact nature of the chemical and signal-transduction interactions that mediate control of erythroid cell growth. Moreover, the need for such a SF medium is all the more pressing if we are to understand the altered cellular responses to hormonal stimulation that underlie hyperplastic and neoplastic myeloproliferative disorders, such as Polycythemia vera. 

2- BPA-like factors in serum-containing, serum-deprived and serum-free systems. 

2.1- The general concept of a BPA 

In general, BPA is any factor which stimulates the production in vitro of erythroid bursts. It may be required for the initiation of burst-formation and/or for its maintenance. However, this activity is deemed to be distinct from that of Epo, in that the latter, although mitogenic for late erythroid progenitors, acts fundamentally as an erythroid-specific differentiation factor; BPA-like factors, in contrast, target early erythroid progenitors and are thought to be solely agents of cellular proliferation, which may or may not similarly affect other hemopoietic lineages (29). Since the early model of erythropoiesis derived from murine studies postulated that the progenitor of each erythrocytic colony was the CFU-E, then, if an erythroid burst was thought to be composed of erythrocytic colonies, it would follow that the (early) progenitor of the burst must give rise to a cluster of CFU-E (or late progenitors) (30). Accordingly, erythropoiesis would be dual-regulated: by BPA in the early proliferative stages, and by Epo in the terminal stages of differentiation (31-32). Nevertheless, examination of CFU-E and BFU-E progenitor kinetics in murine BM were at variance with this model of erythropoiesis (33-34). Moreover, the assumption that bursts are composed of cells that reach maturation (terminal hemoglobinization) at different times in culture and are therefore derived from progenitors that were initially at different levels of differentiation, was confirmed by the findings of an erythroid progenitor that bridged the generation gap between the early BFU-E and the CFU-E; this was the late BFU-E present in murine (35) and human (36) BM, which was reported to be intermediate in its Epo dose-response curve, cell size, derived colony size and maturation time, when compared to the early BFU-E and the  CFU-E, and their derived colonies (see Fig.1). Nonetheless, the effect of BPA would be to increase the number of cell divisions separating the early BFU-E from the final product of growth, the red blood cell (RBC). As such, the net effect of BPA would be to increase burst size and the number of burst-component colonies (32,37). What was not readily apparent was whether BPA was involved in recruiting BFU-E into cycling, eliciting their departure from the quiescent Go state, or simply ensured G1->S progression and survival of cycling BFU-E (29); in other words, whether BPA acted as a competence or a progression signal for BFU-E proliferation, or both; and whether the observed BPA phenomena involved one or more distinct molecules.

 2.2- Early isolation of BPAs; GM-CSF and IL-3 

We referred above to different cell-types that produce BPAs present in their conditioned medium (CM) (4-11). Other typical examples of BPA-containing CM, besides those already mentioned, are: Pokeweed-stimulated spleen-cell conditioned medium (PW SCM), used in the classical study that first demonstrated how a BPA could enhance the survival of murine BM BFU-E during a preincubation period of 3-6 days, in the absence of Epo, as well as reduce the Epo requirement of BFU-E (31); and CMs from established T-cell and monocytic lines (38-40). Using such diverse sources, several groups of investigators have attempted to purify different human BPAs: from BM CM (41), from PW SCM or other lectin-stimulated SCMs (42-43), from urinary preparations (44), from a T-lymphoblast cell line (45-46) and more recently, from plasma membranes of B-lymphocytes (47). However, two of these BPA candidates are of doubtful BPA-like activity, inasmuch as they can only be shown to enhance the growth of erythrocytic colonies derived from late erythroid progenitors (CFU-E) and have not been shown to be capable of maintaining the proliferation of early BFU-E (45-47). All of the other purified preparations investigated for their BPA-like properties, with one exception, have remained undefined, and none has been tried in a SF medium. The exception to both generalizations is GM-CSF (22), first found in lectin-stimulated SCM (42-43). Nevertheless, it appears that the action of rHu GM-CSF targets populations of intermediate BFU-E progenitors, and not the primitive population of quiescent BFU-E, as found with BM BFU-E in a SF medium (22). Other findings, obtained in a SD medium, suggested that rHu GM-CSF is only capable of maintaining BFU-E proliferation, but is not capable of recruiting or increasing the number of PB or BM BFU-E (18,19).

The most promising BPA found so far, though not originally investigated for its BPA-like effect, is IL-3, which was first purified from the CM of the murine cell line WEHI-3 (48-50), and has been shown to stimulate the growth of several hemopoietic lineages (51). Early findings in a SD medium that employed purified Epo and IL-3 preparations to ascertain burst-formation from murine spleen cells, showed that the IL-3-induced multilineage colonies, normally devoid of erythrocytic components, would contain hemoglobinized cells only if Epo were added (52). This neatly fitted the classical model of erythropoiesis and the role of a BPA in this model; IL-3 was acting as just such a BPA. Once the human GF was cloned and became commercially available (rHu IL-3), this effect could also be shown for human adult burst-formation from PB or BM in SD culture (18-19). More recent SF studies have confirmed the action of rHu IL-3 as a primary BPA for BM BFU-E and its capacity to increase burst size in a dose-dependent manner (21-22). Even though there is no current demonstration that primitive BFU-E have IL-3 receptors, work performed with murine erythroblasts and the B6SUtA cell line indicates that, whereas Epo reduces the number of IL-3 receptors (as erythropoietically committed cells become progressively more differentiated), in turn IL-3 down-modulates the number of high-affinity Epo receptors needed for differentiation (53-54). Accessory cells would be capable of promoting burst-formation by providing IL-3, inasmuch as the IL-3 gene has been found to be transcribed only in activated T-cells and macrophages, which are typically present in BM but absent in PB (55-56).

2.3- Other peptide BPAs: IL-4, IL-9, SCF 

More recently, besides IL-3 and GM-CSF, other cytokines have been examined for BPA-like properties, though originally, like IL-3, they were not known or investigated for this quality. Currently, there are three under consideration: rHu IL-4, rHu IL-9, and mast-cell GF, also known as stem cell factor (MGF = SCF).

Murine IL-4 (previously known as a B-cell stimulating factor or as eosinophil differentiation factor), which was purified from the supernatant of murine T-cell clones stimulated with irradiated T-cell-depleted CBA spleen cells (57), was observed to act like a BPA for murine BFU-E (58-59). However, in SF medium, rHu IL-4 has been shown to be incapable of supporting either the proliferation or the survival of human BM BFU-E (60). 

 IL-9 was originally described as a murine T-cell GF which was detected in the supernatant of the P40 T-cell line (61); rHu IL-9 was subsequently made available (62) and shown to be less than half as potent as IL-3 in terms of the plating efficiency of PB BFU-E, as performed in a high-serum assay (63). Very recently, other workers found it to be even less potent under similar assay conditions (147). This has been recently confirmed with a SD assay for highly purified BM CD34+CD33- BFU-E (64-65). Moreover, this could well fit in with the SF findings of Sonoda et al. (22) that the BPA present in serum is neither IL-3 nor GM-CSF.

Lastly, and very recently, a novel hematopoietic GF (encoded by the mouse Steel locus on chromosome 10), which is the ligand for the c-kit proto-oncogene receptor (which, in turn, is the gene product of the W locus on chromosome 5), has been purified from the CM of a buffalo rat liver cell line (66), as well as sequenced in its soluble rat (67), mouse (68-69) and human (67) forms, and alternative cell-bound mouse form (70). As it turns out, the Steel ligand is responsible for an activity that promotes the proliferation of mast cells (MGF) (69), and is identical to the stem cell stimulatory activity (SCF) found in the BM of 5-fluorouracil-treated animals (66-68,70). Its effects on the proliferation of myeloid and lymphoid lineages are not remarkable, though it affects all lineages, as would be expected from a stem-cell factor (71). However, it synergizes with other cytokines (e.g.. GM-CSF, IL-3 and Epo (67,72-73)) to produce multi-lineage colonies (thought to be derived from the colony-forming unit, granulocytic/erythroid/mast and monocytic type, CFU-GEMM) from human BM (67,72-75). These colonies require Epo in order to express their erythroid potentiality (67,72-75). Nevertheless, all these studies were performed either in the presence of high quantities of serum (67,73-75) or in SD media (72). There are at present no definitive data showing that the response of target cells to SCF is actually proliferative (71) and it may be that, in what specifically concerns its potential BPA-like properties, it functions as a permissive rather than as an instructive factor or vice versa. Studies in SF media are necessary in order to determine the exact nature of the function of SCF in early erythropoiesis: whether it is capable of recruiting quiescent BFU-E, or is simply a support factor involved in the maintenance of BFU-E proliferation, as GM-CSF and fibroblast growth factor (FGF) appear to be (16,17,20,22). 

2.4- Evidence for the BPA role of hemin 

Aside from the investigations of the effect of insulin-like  GF-I (IGF-I) on erythropoiesis (which we shall discuss later and which have suggested that IGF-I is a general mitogen devoid of BPA-like activity), the only other molecule studied for its BPA-like qualities is not a cytokine but the small biosynthetic nonpolypeptide that serves as the prosthetic group of hemoglobin and of some cytochromes: iron protoporphyrin IX, or its ferric chloride form (hemin). Hemin, which is released during the breakdown of RBCs, has been shown to act as a normal maturation factor of erythropoiesis capable of inducing strong hemoglobinization of erythrocytic colonies (76-79). But hemin has also been shown to have distinct proliferative effects in hemopoiesis. In SC and SD cultures of murine BM, hemin can be shown to enhance the proliferation of pluripotential CFU-GEMM (80-82) and primitive BFU-E (12, 83-84). Similar results have been reported for rat liver cells (78). However, there are conflicting results as to the effect of hemin on human BM and PB BFU-E, results in SC culture suggesting that it works as a BPA (79,85), whereas SD culture assays have not shown any significant effect (12), or have suggested its effect is inhibitory (86).

Although these studies suggest that the stimulatory effect of hemin is due to its iron delivery role, they did not rule out a specifically proliferative action of the porphyrin molecule. But a recent comparison of the iron intake promoted by ferric salicylaldehyde isonicotinoyl hydrazone (Fe-SIH, a more efficient vehicle for iron delivery than either transferrin or hemin), in murine BM CFU-GEMM indicated that a significant part of hemin's effect upon the growth of these stem cells, and possibly upon BFU-E, was independent of its role in supplying iron (87).

 Hemin has also been recently reported to synergize with IL-3 in SD cultures of murine BM CFU-GEMM (88). While either molecule can support the growth of CFU-GEMM-derived colonies in SC cultures, neither hemin nor IL-3 was capable of doing so under SD conditions; but they could effectively support multilineage colony-formation, in the same conditions, when added together (88). Hemin, which binds specifically to albumin on a 1:1 molar ratio (89), is a prime candidate for the BPA, or one of the BPAs, that are known to be present in albumin (3,4,16). Hemin, which is known for its capacity to regulate the transcription of murine globin genes, has also been shown to increase g -globin synthesis in BM BFU-E (whose growth it increased up to 7-fold) from normal and sickle-cell (HbSS) human donors (90). Part of this effect is apparently mediated by a plastic-adherent cell (90). 

3 - The human myeloproliferative disorder Polycythemia vera: a controversial erythropoietic mechanism

 3.1- A definition of the disorder and its symptomatic complications

Polycythemia rubra vera (PV) is a chronic myeloproliferative disorder of unknown cause which was first identified by Vaquez in 1892, as a cyanotic erythrocytosis (91), and subsequently described as a new clinical entity, a "true erythroid polycythemia", by Osler in 1903 (92). Only afterwards did the disease become recognized as a generalized hemopoietic disorder characterized by hyperplasia of all three myeloid cell lineages (erythrocytosis, leukocytosis, thrombocytosis) (93-94) and by the re-appearance of myeloid tissue in extramedullary sites (95). Nonetheless, this disease has a special emphasis on the erythroid lineage, exhibiting increased red-cell mass (an increase in the total volume of circulating RBCs) due to an overproduction of RBCs (91,96-97). This often occurs in the presence of depressed serum and urinary levels of Epo (the hormone responsible for the normal production of RBCs) (96-97). Accordingly, the major clinical criteria of the PV Study Group (PVSG) for the diagnosis of PV are an increased total red cell mass and splenomegaly in the presence of normal O2 saturation (96). However, splenomegaly is only detectable in 20-30% of PV patients, the remaining being classified under PV if they also exhibit two of the four minor criteria: an increase in white-cell count, an increase in platelet count, an increase in neutrophil alkaline phosphatase or an increase in serum vitamin B12 or its transport protein (96).

Because of these general characteristics, PV has long been thought to represent a reversion of hematopoiesis to a mechanism and to sites that are normally only involved in hematopoiesis during fetal and embryonic development. The mean age at diagnosis is 60 and, if unchecked,  the disease is fatal, as it evolves towards anemia, myelosclerosis and acute myelogenous leukemia, with an increased chance of thrombotic and hemorrhagic episodes (96).

 3.2- Experimental analysis and mechanistic models of the erythrocytosis associated with PV:

3.2.1- Earliest findings and the Epo-hypersensitivity model

The earliest results of research into PV erythropoiesis in vitro indicated that BM cultures of PV progenitors showed only a small increase in Hb expression and in DNA synthesis in response to (exogenous) Epo added to the cultures, when compared to cultures of normal BM cells (98-99). Sera from PV patients were also observed to increase the Epo-responsive cell compartment in hyper-transfused mice (100). At the time, it was speculated that this serum factor acted either to expand the number of Epo-responsive cells or to increase the Epo-sensitivity of these cells (100).

The creation of a plasma clot SC culture system in the early 70's, in our laboratory (101), led to the demonstration that the production of day 7 (late) erythroid colonies (shown via isoenzyme analysis to be clonally derived from CFU-E (102)) by normal human BM cells, is strictly dependent upon the concentration of Epo added to the medium; if no Epo was added, no or very few CFU-E-derived colonies were detected (103). It was then shown by Prchal and Axelrad, using the same SC culture system, that BM cells of PV patients generated erythroid colonies in the absence of added Epo (103-104). This constituted the original finding of spontaneous erythroid colony-formation in these patients, and these colonies were termed EEC, ie. endogenous erythroid colonies. EEC were observed to constitute between 10 and 40% of all erythroid colonies obtained with PV BM cells (103, 96).

To explain this finding, Prchal and Axelrad put forth three hypotheses (104). EEC could be due to:

I) the acquisition of an increased sensitivity to Epo by PV erythroid progenitor cells (the Epo-hypersensitivity hypothesis);

II) the presence of a subpopulation of erythroid progenitor cells that were sensitized in vivo  by Epo, prior to their assay;

III) a neoplastic transformation that made these cells independent from the physiological control of Epo.

Subsequently, it was found that PV is a clonal disorder of a multipotential stem cell, a finding which demonstrated the common etiology of the multilineage hyperplasia. Studies done with female PV patients that were heterozygous for the G-6-PD gene (present in the X-chromosome) showed that their PB cells did not present the expected mosaic aspect characteristic of normal PB cells (with either of the two G-6-PD alleles, A and B, being expressed in different cells (102)), and almost always presented instead the expression of only one allele (105-106). Moreover, SC methylcellulose cultures of BM cells from these heterozygous PV females showed that all their erythroid colonies, in the absence of Epo, expressed only the same allele A which was present in PB, whereas the addition of Epo provoked the emergence of erythoid colonies expressing both alleles, A and B (105-106). These findings indicated that, even though PV patients have normal Epo-dependent progenitor cells in their bone-marrow, these normal progenitors were somehow suppressed, in light of the fact that only the progeny of the PV clone could be detected in vivo (106-107). Nevertheless, it was also established that not all erythroid progenitors belonging to the abnormal clone were necessarily capable of giving rise to EEC (107).  In other words, the polycythemic clone also contained "normal Epo-responsive" cells (107).

Early evidence for the first of the hypotheses, the Epo- hypersensitivity model, put forth by Prchal and Axelrad to explain the EEC phenomenon characteristic of PV, came from the work of Zanjani's group, who found that adding an anti-Epo antiserum resulted in a substantial decrease of EEC grown in the absence of exogenous Epo (108). These and other researchers also found that the addition of Epo to the plasma clot system resulted in an increased number of erythrocytic colonies (108-109). These investigators thus argued that EEC occurred because of an abnormal sensitivity to Epo (108), likely to be present in fetal calf-serum (FCS) or in the citrated plasma utilized in these assay systems (101). 

Concurrently with this, it was discovered that larger, composite colonies of erythroid (sub)colonies, analogous to murine BFU-E-derived bursts (30), could be obtained in SC plasma-clot or methylcellulose cultures of normal human PB or BM, grown for 9 to 18 days (110-111). It was also noted that there were no, or very few, CFU-E-derived erythrocytic-like colonies that could be grown in vitro from normal human PB (110,112-113). As found previously for their murine counterparts, the human erythroid bursts appeared to require higher doses of Epo than that needed to grow erythrocytic colonies, and this suggested that the CFU-E entities were later progenitors derived from the earlier BFU-E cell (111-112). A third entity, previously shown to exist in murine erythropoiesis, was also postulated on the basis of its intermediate response to Epo (111). This was the day 9 to 12 human BFU-E (designated by Eaves and Eaves as the mature BFU-E), which would also bridge the clonal derivation between the other two progenitors - the primitive day 12 to18 BFU-E present in PB or BM, and the late day 7 BM CFU-E (111). The same workers also found that the growth of primitive BFU-E required, in addition to Epo, either the presence of an adherent cell layer or the addition of LCM (111). However, Aye found opposite results using the same SC methylcellulose culture assay-system: human nonadherent BM cells were observed to produce erythroid day 12 to 14 colonies in the absence of LCM (or adherent cells), even though LCM was capable of greatly enhancing their growth (114). This led Aye to suggest that the erythroid PV phenotype is Epo-independent (114), whereas the data obtained with PV BM cells by Eaves and Eaves was taken to suggest that the hypersensitivity to Epo, which would be characteristic of PV, was associated with the mature BFU-E progenitors, if not also with the primitive BFU-E (111). Moreover, the findings of Eaves and Eaves also suggested that Epo-hypersensitivity was responsible for the observation of 'spontaneous' burst-formation (endogenous erythroid bursts, EEB) obtained, in the absence of Epo, in cultures of PV BM progenitor cells (111-113,115-116). Hence, Eaves and Eaves concluded that normal primitive BFU-E were already regulated by Epo and that, in PV, this normal population coexisted with an abnormal clonal population that was more sensitive to Epo and which therefore had a growth advantage under the conditions of low Epo levels present in the serum and marrow of PV patients (111). Other researchers confirmed the presence of endogenous bursts in plasma-clot cultures of PV BM cells, but claimed that the abnormal sensitivity to Epo was expressed only at the CFU-E level; otherwise, they argued, if normal BFU-E were to be dependent upon Epo for their differentiation into CFU-E, the abnormal population of erythroid progenitors in PV would be seen to increase in vivo from the BFU-E to the CFU-E levels, and this was not observed (117). They concluded therefore, that 'spontaneous' bursts or EEB are not derived from BFU-E that are hypersensitive to Epo (117).

Nevertheless, the model of an Epo-hypersensitivity on the part of PV erythroid progenitor cells appeared to be further confirmed by  findings similar to those of Eaves and Eaves: in the absence of exogenous Epo and in a SC methylcellulose assay, no spontaneous colonies or bursts could be observed with normal human adult PB cells, whereas EEC and EEB were present under the same conditions when normal newborn cord blood or PV PB cells were plated (118). Moreover, another study, performed under SD culture conditions in the absence of exogenous Epo, found that no EEC or EEB could be detected with PV BM cells (27). This SD study, which seemed to definitively invalidate the third hypothesis of Prchal and Axelrad (i.e. of an Epo-independence on the part of the abnormal PV clone), also established that the PV clone constituted an heterogeneous population exhibiting varying degrees of hypersensitivity to Epo, up to 10x the sensitivity of the normal populations (27).  

3.2.2- Evidence against a prior Epo-sensitization in vivo model of PV

Meanwhile, Eaves and Eaves' group also showed that, by replating day 9 colonies to obtain secondary EEC, whatever was responsible for the production of EEC could be transmitted from one cell generation to the next (119). Thereby, they ruled out the second hypothesis of a prior sensitization in vivo of PV progenitor cells by Epo (119). They also revised their previous contention of an Epo-hypersensitive response on the part of PV BFU-E, by showing that the expression of the proposed Epo-hypersensitive phenotype occurred after the primitive BFU-E stage, in light of the fact that the primitive BFU-E, which yielded the apparent Epo-hypersensitive progenitors, also produced significant numbers of normal Epo-responsive cells (119).

3.2.3- The fetal reversion hypothesis: evidence for and against it,  and evidence against the Epo-hypersensitivity model

In this context, it is relevant to examine how erythropoiesis in PV shares some characteristics with fetal erythropoiesis. Furthermore, because the Epo-hypersensitivity phenomenon appeared to be closely associated with stem cell neoplasia, Eaves' group and other workers suggested that PV could be caused by the reactivation of gene networks operative in extramedullary fetal hemopoiesis, which are normally turned off after birth, rather than being the result of a 'new acquisition' (eg. caused by infection with a retrovirus that alters the normal cellular expression of growth factor genes) (118-119). Consonant with this hypothesis, it was reported that normal human umbilical cord blood progenitors (118) and fetal and newborn sheep progenitors (120), unlike their normal adult counterparts, can produce EEC like those in PV. Moreover, fetal erythropoiesis is characterized by low circulating Epo levels and is not regulated by O2 deficiency (121).

Consonant with this fetal-reversion hypothesis, it was found that adult PV EEC and EEB (grown without exogenous Epo) contain from 27% to 50% fetal hemoglobin, Hb-F (122), a form which is low in normal adult bursts and virtually absent in normal adult CFU-E-derived colonies (122-123). In normal erythropoiesis, Hb-F content decreases in the progenitor cells with the advancing stage of their maturation, from the CFU-GEMM to the CFU-E stage (124-125). However, this finding of a high Hb-F content in PV PB erythroid colonies was obtained under serum-containing conditions, and it has since been shown by the same workers that FCS contains activities that promote the expression of Hb-F in adult PB or BM BFU-E cultures (126). Moreover, PV circulating erythroblasts have normal or near normal Hb-F levels in vivo (122), even though there are reported cases of observed increases in Hb-F (127). It has also been shown that neither Epo, nor IL-3 or GM-CSF could increase Hb-F in adult human bursts grown in SC or SD media, whereas the addition of fetal bovine serum could (128-129). However, these results have been directly contradicted by others who found, also under SD conditions, that both IL-3 and GM-CSF induce a dose-dependent rise of Hb-F production in normal adult PB bursts (130).

3.2.4- Evidence for an Epo-independent model

There are even more severe problems with the Epo-hypersensitivity explanation of the erythroid hyperplasia observed in PV. First, Zanjani's anti-Epo antiserum (108) was polyclonal, ie. directed against multiple antigens, and made against purified rather than recombinant Epo. This left open the possibility that some of the antibody specificities were directed against entities other than Epo, specially given that certain Epo preparations are known to be contaminated with other GFs. Secondly, not only is there evidence for the presence of small amounts of Epo and Epo-like activities in serum and in plasma (which are components of the SC clot or SC methylcellulose culture media), but, what is still more important, undefined BPAs are now known to be present in the BSA, in the lipoprotein fractions and in the purified Epo and insulin preparations employed by the same SC or SD assay systems which were utilized to operationally define the criteria for EEC and EEB. Hence, with such undefined systems there is little that can be reliably learned with respect to cellular responses to growth factors, and we cannot therefore rule out the possibility that the abnormal PV clone might be hypersensitive to other growth factors besides, or instead of, Epo. Thirdly, a whole other series of data on this subject started to turn around the entire field of inquiry, a change which we shall now examine.

In fact, Zanjani's results were directly contradicted by Najman et al. who found no diminution of the number of EEC in culture when a polyclonal antibody obtained against purified Epo was added (131); as this finding was obtained in SD culture conditions, it therefore more strongly suggested that the abnormal progenitors in PV give rise to truly Epo-independent colonies (131). More recently, another group confirmed this result for EEB, but in SC culture (132). Small numbers of EEC were also reported to have been found in PB cultures of PV patients, indicating the existence of circulating Epo-independent CFU-E in PV (133).  Coupled to the original findings of Aye concerning the effect of LCM on burst-formation (114), which we referred to above, these results suggested a validation of the third hypothesis, i.e. Epo-independence.

Furthermore, these results seemed to implicate the activity of BM accessory cells in the induction and selection of the abnormal phenotype in PV (113-114). In agreement with this possibility, another group of researchers reported that lectin-induced CM of PB adherent cells could triple the number of EEB in the absence of Epo (134); and that, after a 24-hour adherence treatment of PV PB mononuclear cells (MNC), the growth of EEB was completely abrogated (134) (though we were told nothing about the viability of the cells after such long incubation nor about its conditions). Nonetheless, the same workers reported that EEB could be restored totally, if adherent cells were subsequently added, or partially, if their conditioned medium was added (134). This would constitute evidence that, even in a SC medium, some BPA provided by adherent cells was necessary for EEB-formation by PV progenitor cells. More recently, another SC study showed that PB nonadherent MNC, capable of producing EEB in the absence of added Epo, produced much higher numbers of these endogenous bursts when IL-3 or GM-CSF was added to the cultures (132). As these IL-3- or GM-CSF-promoted EEB present in PV were not abrogated by an anti-Epo antiserum (132) that precluded burst-formation by normal progenitors, the Epo-hypersensitivity hypothesis, and the hypothesis of a reversion of Epo-sensitivity to a fetal-like stage (118,120), appeared to lack a valid foundation.

 To make matters worse, EEB from normal adult BM nonadherent MNC were obtained in SD conditions: Eridani's group reported that some endogenous bursts were still observed both with PV marrow and, to a lesser degree, with the marrow of normal subjects, even when FCS, Epo, LCM and adherent cells were removed (28). In the absence of any adherent cells, this clearly raised three possibilities (see Fig. 2) as to the source and identity of the growth factor(s) responsible for this EEB-inducing activity(-ies). It should be noted, that all of these possibilities are intimately connected with the SF problem of defining biologically just what a BPA is (is a BPA just instructive or are permissive factors also BPAs? And, just how many BPAs are required for optimal growth?):

1.- It is possible that lipids could be involved in acting as a source of BPA, given that defatting all albumins (BSA (3) and human serum albumin (HSA) (4)) prevents normal erythroid burst-formation unless IL-3 is added, as found under SF conditions for BM MNC null cells (22). It is equally well established that lipoprotein fractions contain some BPA (14,114) and that burst-formation requires lecithin, cholesterol, and oleic or linoleic acids (1-3). Moreover, as Eridani's findings were performed with a BSA previously shown to contain a BPA-like activity for both murine and human burst-formation (3,14), the normal and PV EEB observed could have resulted from an activity associated with lipids or with lipoproteins present in the albumin.

2.- The BPA-like activity which is associated with BSA may  be a substance other than the lipids or small lipoproteins, given that the delipidation method would just as effectively remove, besides lipids and lipoproteins, any other factors, small peptides and other molecules, such as hemin or retinoids, also present in BSA or in HSA. But a BPA-like activity has only been identified for one such molecule, hemin, and only for murine erythropoiesis (79-87). Furthermore, this hypothesized non-lipid BPA present in albumins and possibly involved in eliciting the PV phenotype, may also be present in other purified preparations (like the BPA found in urinary Epo preparations (13)), or in serum (2). At any rate, it is likely that the factor or factors involved in promoting EEB in Eridani et al. 's SD study, might be linked to the low purity of the BSA employed, given that other investigators have detected no normal human bursts in the presence of defatted albumins, when recombinant Epo alone was used, in the absence of lipoproteins and insulin (4,13).

3.- The BPA-activity could be an effect of insulin or of an insulin-like activity. It has been found by Kurtz et al. that, in SD conditions, supraphysiological concentrations (>10-8 M) of insulin can elicit Epo-independent erythroid colony formation (normal EEC) by mouse fetal liver and adult BM CFU-E (135), as is observed with the endogenous colonies of PV progenitors. Furthermore, the observed combined effect of Epo and insulin was smaller than the sum of their single effects, suggesting that, in mice, we might be dealing with two overlapping classes of CFU-E which are partially distinct as to their respective sensitivities to two comparable factors- Epo and insulin (135). Further work by the same group showed that the Epo-like activity present in FCS and responsible for the EEC formation in the absence of exogenous Epo is insulin-like growth factor I (IGF-I), in light of the finding that it could be abolished by a polyclonal antibody directed against pure human IGF-I (121). In previous investigations of fetal and adult murine erythropoiesis, the same workers had shown that, under SD conditions, IGF-I promotes normal EEC in the absence of added Epo (136). Other researchers have also detected the presence of an Epo-like activity, or activities in FCS needed, for example, for the erythroid differentiation of K562 cells (137), and capable of inducing Hb-F production by normal adult PB BFU-E (126). It has also been reported that NIH 3T3 fibroblasts, transfected with wild-type and chimeric insulin and type I IGF-I receptor cDNAs, exhibit a maximal growth-response to either insulin or IGF-I alone, which is equivalent to that obtained with 10% FCS (138). Others showed that insulin or IGF-I were required for the optimal growth of highly purified populations of replated CFU-E in SD medium (an increase of 100%), but that adherent cells were not necessary (139). However, in the absence of added Epo and adherent cells, no EEC could be detected with insulin alone (139). Under these SD conditions, but in the presence of rHu Epo, a 100-fold greater concentration of insulin was needed to equal the maximal IGF-I effect upon CFU-E growth (139).

Further support for our hypothesis of the endogenous burst-formation characteristic of PV being the result of a (known or unknown) insulin-type activity present in the culture systems may be inferred from the finding that IGF-I has an enhancing effect on the growth of human and murine BM BFU-E in SD media (16). The same authors also showed that increasing the concentration of Fr. V BSA from 1 to 3% was almost comparable to the maximal effect that IGF-I had in murine burst-formation, though they did not draw any conclusion from this, despite their speculation that IGF-I acts directly on the BFU-E (16). Studies of children (PB and BM) (140) and adult (BM) (141) BFU-E grown under SC conditions have also demonstrated that IGF-I enhances burst-formation 1.5-fold; and that, in the case of children's BFU-E, no Epo is necessary to observe burst-formation (140). The enhancing effect of human Growth Hormone (hGH) upon adult BM burst-formation was shown to act via a paracrine mechanism with IGF-I acting as mediator, inasmuch as an anti-IGF-I antibody was capable of abrogating the hGH effect (141). The same investigators also found that neither PB MNC nor adherence-depleted BM MNC responded to hGH with increased colony production (141). They therefore suggested that, as monocyte-depleted marrow is equivalent to PB and incapable of producing IGF-I in response to hGH, then the monocytes present in PB probably do not normally do so either (141). This, however, was performed both in the presence of serum and of a saturating quantity of LCM, both of which could just as well contain significant quantities of IGF-I, and could thus possibly be precluding the detection of any further IGF-I effects. But in this sense of IGF-I being a mediator of accessory cell modulation of erythropoiesis, it has been reported that human alveolar macrophages only synthesize IGF-I upon antigenic activation (142). Dainiak et al. have also studied the effect of insulin and insulin-like growth factor II (IGF-II) using a SD system containing LCM and purified HSA, but, unlike the findings of Kurtz et al., they reported a synergistic effect of insulin and IGF-II with both Epo and LCM (143). They concluded that insulin and IGF-II were potentiators of early erythropoiesis (143). Finally, there is only one SF study of the action of IGF-I (22) which has shown that adult human BM BFU-E require IL-3 for proliferation, even when Epo and IGF-I are present at optimal concentrations. This strongly suggests that IGF-I does not possess a BPA-like activity for BM BFU-E, in the sense that it cannot recruit BFU-E into proliferation, and that its stimulatory or mitogenic activity upon the BFU-E and its descendents is only that of a potentiator of the Epo and BPA effects. Thus, IGF-I does not function as a competence signal capable of inducing BFU-E into proliferation, even if other researchers have found that both IGFs I and II are capable of promoting DNA synthesis in adherence-depleted murine fetal liver cells (FLC) in an SF liquid culture (144). But, as FLC can also produce bursts in response to Epo alone in an unicellular SD assay (26), and have also been shown to be undergoing DNA synthesis (24), the mitogenic activity of IGF-I cannot be taken to mean that IGF-I functions as a competence factor, or an instructive signal for erythropoiesis in primitive BFU-E.

Taken together, these results, which can be read to favor the third possibility, could well explain the normal EEB found by Eridani's group- for they used insulin in their "SF" culture along with a non-defatted BPA-containing BSA, even when no apparent source of BPA (LCM, lipoproteins, adherent cells) and no Epo were added.  An interpretation such as this could also explain a possible cause for the discrepancy that exists between the lines of evidence for an Epo-hypersensitivity model and those for a model of Epo-independence with respect to PV erythropoiesis in vitro: the experiments that suggested Epo-hypersensitivity did not use insulin in their culture media, although contaminating BPA was present (in FCS, in BSA, in LCM, in purified Epo, etc); whereas those that suggested instead Epo-independence contained supraphysiological levels of insulin, along with contaminating BPA. Moreover, peptides of the IGF family are also common contaminants of albumin (145-146). This preliminary interpretation suggested to us that the abnormal erythroid proliferation in PV may be connected to the effects of insulin and/or IGFs.

4- Statement of the problem

The question of whether PV erythroid progenitors are really Epo-hypersensitive or instead hypersensitive to a factor or factors other than Epo (be it to a BPA-like molecule, like IL-3, or to a potentiator molecule, like IGF-I) could not be resolved satisfactorily as long as the media used for the cultivation of erythroid progenitor cells were contaminated by undefined or unidentified GFs. To solve this question, we needed a serum-free medium, as fully defined as is presently possible, based upon a defatted albumin. Moreover, for logistic reasons, this SF medium had to support burst-formation by circulating erythroid progenitors. In fact, when we first set out to work on erythroid progenitors in vitro  from PV patients, we planned to use BM cells. However, BM samples from patients with PV could not be obtained in sufficient numbers because PV can be effectively managed without resorting to bone marrow aspiration. We therefore turned to peripheral blood as a source of erythroid progenitors, but no technique existed for cultivating these cells under SF conditions.

To achieve this task, we endeavored to create a medium which was based upon a BPA-free BSA, and a BPA-free, recombinant Epo preparation, in order to test the effects of hemin, IL-3 and IGF-I on the production of erythroid bursts and erythrocytic colonies by adult, normal and PV PB cells. If we were successful it would be the first SF medium capable of supporting the growth of BFU-E from peripheral blood. 

5- Our point of departure

We first confirmed Stewart et al. 's findings (3) using a non-defatted BSA (Fr. V), ie. that murine BM BFU-E can be grown with purified Epo alone, under these conditions, and their progeny can be detected histochemically with benzidine staining (results not shown). We then applied this methodology to adult human BM but using 4x crystallized, deionized and charcoal-delipidated ICN BSA. From a titration of the concentration of BM MNC, we found we could grow erythrocytic colonies with a higher (>2x) plating efficiency than in the presence of serum (cf open and closed circles in Fig. 3), but we were unable to grow any bursts in SF conditions, even when >105 MNC were plated and IGF-I was added (cf open and closed squares in Fig. 3). In all cases, 3 U/mL of BPA-free human purified Epo was used. But if we were to replace this delipidated BSA by Fr. V BSA, we could grow BM BFU-E from normal donors using the same BPA-free Epo preparation (see Fig. 4). However, we needed to use either a benzidine stain or a Soret band-pass filter to score these bursts and detect their hemoglobinization. Using these methods, we were able to observe endogenous burst-formation in the absence of Epo (bar 1, Fig. 4), and the number of these EEB increased when 30 nM IGF-I was added (bar 3, Fig. 4). Under these circumstances, the activities of IGF-I and Epo appeared to be interchangeable and non-additive (cf bars 2 and 3 with bar 4, Fig. 4).

These results indicated that the BPA present in nondefatted BSA is sufficient to support some burst-formation, even when BPA-free Epo is employed. It also confirmed that a BPA-free albumin was necessary for the definition of a SF medium, and that, in such conditions, a defined BPA would have to be provided. This was only emphasized by the fact that, under the SD conditions described in Fig. 4, practically no bursts could be grown from PB MNC, even in the presence of the same concentrations of IGF-I and Epo together (only one donor out of six examined was positive when Epo and IGF-I were added together: 37+/-3 bursts/2 x 105 PB MNC). But these preliminary findings also suggested that normal, endogenous burst-formation was probably related to an insulin-like activity, possibly equivalent to one of the contaminants in nondefatted albumin. To build a SF medium for the growth of PB BFU-E we thus decided to utilize the same fatty-acid- and globulin-free albumin (FAF/GF BSA) employed by Sonoda et al. in the only SF medium currently existing for BM BFU-E (22). Under these conditions, we hoped to define the minimal requirements for optimal burst-formation by circulating erythroid progenitors.

7-REFERENCES

1. Iscove NN, Guilbert LJ, Weyman C (1980) Complete replacement of serum in primary cultures of erythropoietin-dependent red cell precursors (CFU-E) by albumin, transferrin, iron, unsaturated fatty acid, lecithin and cholesterol. Exp Cell Res 126:121.

         2. Guilbert LJ, Iscove NN (1976) Partial replacement of serum by selenite, transferrin, albumin, and lecithin in haemopoietic cell cultures. Nature 263:594. See also: Iscove NN (1984) Culture of lymphocytes and hemopoietic cells in serum-free medium. In Liss AR (ed) Methods for serum-free culture of neuronal and lymphoid cells, New York, NY, p. 169.

3. Stewart S, Zhu B-d, Axelrad, A (1984) A "serum-free" medium for the production of erythropoietic bursts by murine bone marrow cells. Exp Hematol 12:309.

4. Dainiak N, Kreczko S, Cohen A, Pannell R, Lawler J (1985) Primary human marrow cultures for erythroid bursts in a serum-substituted system. Exp Hematol 13:1073.

5. Nathan DG, Chess L, Hillman DG, Clarke BJ, Breard J, Merler E, Housman D (1978) Human erythroid burst forming unit: T-cell requirement for proliferation in vitro. J Exp Med 147:324.

6. Zuckerman KS (1980) Stimulation of human BFU-E by products of human monocytes and lymphocytes. Exp Hematol 8:924.

7. Reid CD, Baptista LC, Chanarin I (1981) Erythroid colony growth in vitro from human peripheral blood null cells: evidence for regulation by T lymphocytes and monocytes. Br J Haematol 58:171.

8. Linch DC, Lipton JM, Nathan DG: (1985) Identification of three accessory cell populations in human bone marrow with erythroid burst-promoting properties. J Clin Invest 75:1278.

9. Gordon LI, Miller WJ, Branda RF, Zanjani ED, Jacob HS (1980) Regulation of erythroid colony formation by bone marrow macrophages. Blood 55:1047.

10. Wagemaker G, Peters MF, Bol SJ (1979) Induction of erythropoietin responsiveness in vitro by a distinct population of bone marrow cells. Cell Tissue Kinet 12:521.

11. Meytes D, Ma A, Ortega JA, Shore NA, Dukes PP (1979) Human erythroid burst promoting activity produced by phytohemagglutinin-stimulated radio-resistant peripheral blood mononuclear cells. Blood 54:1050.

12. Eliason JF, Odartchenko N (1985) Colony formation by primitive hemopoietic progenitor cells in serum-free medium. Proc Natl Acad Sci (USA) 82:775.

13. Konwalinka G, Geissler D, Peschel C, Breier C, Grunewald K, Odavic R, Braunsteiner H (1986) Human erythropoiesis in vitro and the source of burst-promoting activity in a serum-free system. Exp Hematol 14:899.

14. Konwalinka G, Breier C, Peschel C, Geissler D, Lisch HJ, Braunsteiner H (1986) Effect of human plasmalipoproteins on erythropoietic progenitor cells in serum-free cultures. Blut 52:191.

15. Konwalinka G, Breier C, Geissler D, Peschel C, Wiedermann CJ, Patsch J, Braunsteiner H (1988) Proliferation and differentiation of human erythropoiesis in vitro : effect of different human lipoprotein species. Exp Hematol 16:125.

16. Akahane K, Tojo A, Urabe A, Takaku F (1987) Pure erythropoietic colony and burst formations in serum-free culture and their enhancement by insulin-like Growth Factor I. Exp Hematol 15:797.

17. Migliaccio G, Migliaccio AR (1987) Cloning of human erythroid progenitors (BFU-E) in the absence of fetal bovine serum. Br J Haematol 67:129.

18. Migliaccio G, Migliaccio AR, Adamson JW (1988) In vitro differentiation of human granulocyte/macrophage and erythroid progenitors: comparative analysis of the influence of recombinant human erythropoietin, G-CSF, GM-CSF, and IL-3 in serum-supplemented and serum-deprived cultures. Blood 72:248.

19. Misago M, Chiba S, Kikuchi M, Tsukada J, Sato T, Oda S, Eto S (1989) Effect of recombinant human interleukin 3, granulocyte-macrophage colony-stimulating factor and granulocyte colony-stimulating factor on human BFU-E in serum-free cultures. Int J Cell Cloning 7:37.

20. Gabbianelli M, Sargiacomo M, Pelosi E, Testa U, Isacchi G, Peschel C (1990) "Pure" human hemotopoietic progenitors: permissive action of basic fibroblast growth factor. Science 249:1561.

21. Sonoda Y, Ogawa M (1988) Serum-free culture of human hemopoietic progenitors in attenuated culture media. Am J Hematol 28: 227.

22. Sonoda Y, Yang Y, Wong GG, Clark SC, Ogawa M (1988) Erythroid burst-promoting activity of purified recombinant human GM-CSF and Interleukin-3: studies with anti-GM-CSF and anti-IL-3 sera and studies in serum-free cultures. Blood 72:1381.

23. Eaves CJ, Humphries RK, Eaves AC (1979) In vitro characterization of erythroid precursor cells and the erythropoietic differentiation process. In Stamatoyannopoulos G, Nienhuis AW (eds) Cellular and molecular regulation of hemoglobin switching. New York, Grune and Stratton, p 251.

24. Peschle C, Migliaccio AR, Migliaccio G, Ciccariello R, Lettieri F, Quattrin S et al. (1981) Identification and characterization of three classes of erythroid progenitors in human fetal liver. Blood 58: 565.

25. Shekhter-Levin S, Amato D, Axelrad AA (1984) The proliferative state of early erythropoietic progenitor cells (BFU-E) in human umbilical cord blood: low probability of finding BFU-E in DNA synthesis. Exp Hematol 12: 650.

26. Valtieri M, Gabbianelli M, Pelosi E, Bassano E, Petti S, Russo G et al. (1989) Erythropoietin alone induces erythroid burst formation by human embryonic but not adult BFU-E in unicellular serum-free culture. Blood 74: 460.

27. Casadevall B, Vainchenker W, Lacombe C, Vinci G, Chapman J, Breton-Gorius J and Varet B (1982) Erythroid progenitors in Polycythemia vera: demonstration of their hypersensitivity to erythropoietin using serum free cultures. Blood 59:447.

28. Eridani S, Dudley JM, Sawyer BM, Pearson TC (1987) Erythropoietic colonies in a serum-free system: results in primary proliferative polycythaemia and thrombocythaemia. Br J Haematol 67:387.

29. Porter PN, Ogawa M (1983) Erythroid burst-promoting activity (BPA). In Dunn CDR (ed) Current concepts in erythropoiesis. New York, John Wiley & Sons, p. 81.

30. Axelrad AA, McLeod DL, Shreeve MM, Heath DS (1973) Properties of cells that produce erythrocytic colonies in vitro. In Robinson WA (ed) Proc Second International Workshop on Hemopoiesis in Culture, Airlie House, VA, p. 226.

31. Axelrad AA, McLeod DL, Suzuki S, Shreeve MM (1978) Regulation of the population size of erythropoietic progenitor cells. In Clarkson B, Marks PA, Till JE (eds) Differentiation of normal and neoplastic hematopoietic cells. New York, Cold Spring Harbor, p. 155.

32. Axelrad AA, Burgess A, Iscove N, Wagemaker G (1981) Culture of Hematopoietic cells: workshop report. In Liss AR (ed) Control of cellular divisions and development, New York, NY. Part B, p. 421.

33.  Hara H, Ogawa M (1976) Erythropoietic precursors in mice with phenylhydrazine-induced anemia. Am J Hematol 1:453.

34. Iscove NN (1977) The role of erythropoietin in regulation of population size and cell cycling of early and late erythroid precursors in mouse bone marrow. Cell Tissue Kinet 10:323.

35. Gregory CJ, Eaves AC (1978) Three stages of erythropoietic progenitor cell differentiation distinguished by a number of physical and biologic properties. Blood 51:527.

36. Gregory CJ, Eaves AC (1977) Human marrow cells capable of erythropoietic differentiation in vitro: definition of three erythroid colony responses. Blood 49:855.

37. Ogawa M, Porter PN, Terasawa T, Brockbank KG (1980) Effects of burst-promoting activity (BPA) on hemoglobin synthesis in culture. Exp Hematol 8:90.

38. Iscove NN, Schreier M (1979) Involvement of T-cells and macrophages in generation of burst-promoting activity (BPA). Exp Hematol 7: 4 (abs).

39. Ascensao JL, Key NE, Earenfight-Engler T, Koren HS, Zanjani ES (1981) Production of erythroid potentiating factor(s) by a human monocyte cell line. Blood 57:170.

40. Linch DC, Donahue RE (1985) Production of human active BPA from TCGF independent T cell lines that do not excrete HTLV: proof of direct action of MLA-144 derived BPA using purified BFU-E. Br J Haematol 61:71.

41. Porter PN, Ogawa M (1982) Characterization of human erythroid burst-promoting activity derived from bone marrow conditioned media. Blood 59:1207.

42. Burgess AW, Metcalf D, Russel SH, Nicola NA (1980) Granulocyte/macrophage-megakaryocyte-eosinophil and erythroid-colony stimulating factors produced by mouse spleen cells. Biochem J 185:301.

43. Metcalf D, Johnson GR, Burgess AW (1980) Direct stimulation by purified GM-CSF of the proliferation of multipotential and erythroid precursor cells. Blood 55:138.

44. Dukes PP, Ma A, Clemons GK, Meytes D (1985) Measurement of human erythroid burst-promoting activity by a specific cell culture assay. Exp Hematol 13:59.

45. Golde DW, Quan SG, Cline MJ (1978) Human T-lymphocyte cell line producing colony stimulating activity. Blood 52:1068.

46. Golde DW, Bersch N, Quan SD, Lusis AJ (1980) Production of erythroid potentiating activity by a human T-lymphoblast cell line. Proc Nat Acad Sci (USA) 77:593.

47. Feldman L, Sytkowski AJ (1990) Human B-lymphoctyte derived burst promoting activity (B-BPA) exerts discrete effects on the development of both BFU-E and CFU-E during erythropoiesis. Clin Res 38:299a.

48. Iscove NN, Roitsch CA, Williams N, Guilbert LJ (1982) Molecules stimulating early red cell, granulocyte, macrophage and megakaryocyte precursors in culture: similarity in size, hydrophobicity and charge. J Cell Physiol Suppl 1:65.

49. Bazill GW, Haynes M, Garland J, Dexter TM (1983) Characterization and partial purification of a haemopoietic cell growth factor in WEHI-3 cell conditioned medium. Biochem J 210:747.

50. Ihle JN, Keller J, Henderson L, Klein F, Palaszynski E (1982) Procedures for the purification of interleukin 3 to homogeneity. J Immunol 129:2431.

51. Ihle JN, Keller J, Oroszlan S, Henderson LE, Copeland TD, Fitch F et al.  (1983) Biological properties of homogenous interleukin 3. I. Demonstration of WEHI-3 growth factor activity, mast cell growth factor activity, P cell-stimulating factor activity, colony-stimulating factor activity and histamine-producing cell-stimulating factor activity. J Immunol 131:282.

52. Suda J, Suda T, Kubota K, Ihle JN, Saito M, Miura Y (1986) Purified Interleukin-3 and erythropoietin support the terminal differentiation of hemopoietic progenitors in serum-free culture. Blood 67:1002.

53. Fraser JK, Nicholls J, Coffey C, Lin FK, Berridge MV (1988) Down-modulation of high-affinity receptors for erythropoietin on murine erythroblasts by Interleukin 3. Exp Hematol 16:769.

54. Broudy VC, Lin N, Nakomoto B, Mui AL, Krystal G, Papayannopoulou (1989) Erythropoietin (Epo)-induced erythroid differentiation significantly reduces IL-3 receptor expression in B6SUtA cells. Blood 75:74a (abs).

55. Wolin MJ, Nimer SD (1989) Expression of human interleukin-3. Blood 74:333a (abs).

56. Ernst TJ, Ritchie AR, Stopak KS, Griffin JD  (1989) Human monocytes produce IL-3 in response to stimulation with the calcium ionophore, A23187. Blood 74:116a (abs).

57. Sanderson CJ, O'Garra A, Warren DJ, Klaus GG (1986) Eosinophil differentiation factor also has B-cell growth factor activity: proposed name interleukin 4. Proc Natl Acad Sci (USA) 83:437.

58. Rennick D, Yang G, Muller-Sieburg C, Smith C, Arai N, Takabe Y, Gemmell L, (1987) Interleukin 4 (B-cell stimulating factor 1) can enhance or antagonize the factor-dependent growth of hemopoietic progenitor cells. Proc Natl Acad Sci (USA) 84:6889.

59. Rennick D, Jackson J, Yang G, Wideman J, Lee F, Hudak S (1989) Interleukin-6 interacts with interleukin-4 and other hematopoietic growth factors to selectively enhance the growth of megakaryocytic, erythroid, myeloid and multipotential progenitor cells. Blood 73:1828.

60. Sonoda Y, Okuda T, Yokota S, Maekawa T, Shizumi Y, Nishigaki H et al. (1990) Actions of human interleukin-4/B-cell stimulatory factor-1 on proliferation and differentiation of enriched hematopoietic progenitor cells in culture. Blood 75: 1615.

61. Van Snick J, Goethals A, Renauld J-C, Van Roost E, Uyttenhove C, Rubira MR et al. (1989) Cloning and characterization of a cDNA for a new mouse T cell growth factor (P40). J Exp Med 169:363.

62. Yang YC, Ricciardi S, Ciarletta A, Calvetti J, Kelleher K, Clark SC (1989) Expression cloning of a cDNA encoding a novel human hematopoietic growth factor: Human homologue of murine T-cell growth factor P40. Blood 74:1880.

63. Donahue RE, Yang Y-C, Clark SC (1990) Human P40 T-cell growth factor (interleukin-9) supports erythroid colony formation. Blood 75:2271.

64. Lu L, Yang YC (1990) Effect of human interleukin (IL)-9 on enriched erythroid progenitors in normal human bone marrow under serum-depleted culture conditions. Exp Hematol 18:608 (abs).  

65. Lu L, Leemhuis T, Srour EF, Yang YC (1990) Enhancing effect of recombinant human interleukin (IL)-9 on enriched sorted CD34+ DR+ CD33- cells from normal human bone marrow. Blood 76:104a (abs).

66. Zsebo KM, Wypych J, McNiece IK, Lu HS, Smith KA, Karkare SB et al.  (1990) Identification, purification, and biological characterization of hematopoietic stem cell factor from buffalo rat liver-conditioned medium. Cell 63:195.

67. Martin FH, Suggs SV, Langley KE, Lu HS, Ting J, Okino KH et al.  (1990) Primary structure and functional expression of rat and human stem cell factor DNAs. Cell 63:203.

68. Zsebo KM, Williams DA, Geissler EN, Broudy VC, Martin FH, Atkins HL et al. (1990) Stem cell factor is encoded at the SI locus of the mouse and is the ligand for the c-kit tyrosine kinase receptor. Cell 63:213.

69. Anderson DM, Lyman SD, Baird A, Wignall JM, Eisenman J, Rauch C et al. (1990) Molecular cloning of mast cell growth factor, a hematopoietin that is active in both membrane bound and soluble forms. Cell 63:235.

70. Huang E, Nocka K, Beier DR, Chu TY, Buck J, Lahm HW et al. (1990) The hematopoietic growth factor KL is encoded by the SI locus and is the ligand of the c-kit receptor, the gene product of the W locus. Cell 63:225.

71. Witte ON (1990) Steel locus defines new multipotent growth factor. Cell 63:5.

72. Migliaccio G, Migliaccio AR, Valinsky K, Langely K, Zsebo K, Visser JW, Adamson JW (1990) Recombinant rat stem cell factor (rrSCF) induces proliferation and differentiation of primitive hematopoietic progenitor cells (HPC) in serum-deprived cultures. Blood 76:156a (abs).

73. McNiece I, Langley K, Zsebo KM (1991) Recombinant human stem cell factor synergises with GM-CSF, G-CSF, IL-3 and Epo to stimulate human progenitor cells of the myeloid and erythroid lineages.  Exp Hematol 19:226.

74. Broudy VC, Smith FO, Lin N, Zsebo K, Egrie J, Bernstein ID (1990) Recombinant human stem cell factor (rHuSCF) stimulates the growth of hematopoietic colonies from patients with acute nonlymphocytic leukemia (ANLL) by binding to a specific receptor. Blood  76:134a (abs).

75. Broxmeyer HE, Cooper S, Lu L, Hangoc G, Anderson D, Cosman D et al. (1991) Effect of murine mast cell growth factor (c-kit proto-oncogene ligand) on colony formation by human marrow hematopoietic progenitor cells. Blood 77: 2142.

76. Porter PN, Meints RH, Mesner K (1979) Enhancement of erythroid colony growth in culture by hemin. Exp Hematol 7:11.

77. Ibrahim NG, Lutton JD, Levere RD (1982) The role of haem biosynthetic and degradative enzymes in erythroid colony development: the effect of haemin. Br J Haematol 50:17.

78. Mayeux P, Felix JM, Billat C, Jacquot R (1986) Induction by hemin of proliferation and of differentiation of progenitor erythroid cells responsible for erythropoietin. Exp Hematol 14:801.

79. Kaye FJ, Weinberg RW, Scholfield JM, Alter BP (1986) The effect of hemin in vitro and in vivo on human erythroid progenitor cells. Int J Cell Cloning 4:432.

80. Monette FC, Holden SA, Sheehy MJ, Matzinger EA (1984) Specificity of hemin action in vivo at early stages of hematopoietic cell differentiation. Exp Hematol 12:782.

81. Monette FC, Sigounas G (1986) The sensitivity of murine multipotential stem cell colony (CFU-GEMM) growth to interleukin-3, erythropoietin and hemin. Exp Hematol 14:519.

82. Monette FC, Sigounas G (1988) Growth of murine multipotent stem cells in a simple "serum-free" culture system: role of interleukin-3, erythropoietin and hemin. Exp Hematol 16:250.

83. Monette FC, Holden SA (1982) Hemin enhances the in vitro growth of primitive erythroid progenitor cells. Blood 60:527.

84. Holden SA, Steinberg HN, Matzinger EA, Monette FC (1983) Further characterization of the hemin-induced enhancement of primitive erythroid progenitor cell growth in vitro . Exp Hematol 11:953.

85. Lu L, Broxmeyer HE (1983) The selective influence of hemin and products of human erythrocytes on colony formation by human multipotential (CFU-GEMM) and erythroid (BFU-E) progenitor cells in vitro . Exp Hematol 11:721.

86. Migliaccio G, Migliaccio AR, Adamson JW (1990) The biology of hematopoietic growth factors: studies in vitro under serum-deprived conditions. Exp Hematol 18:1049.

87. Monette FC, Wu D, Ponka P (1990) Mechanism of action of hemin at early stages of multipotential cell growth in vitro : role of hemin as a supplier of iron in serum-deprived marrow cultures. Exp Hematol 18:554 (abs).

88. Monette FC, Sigounas G, Wu D (1988) Synergism between hemotopoietic growth factors: modulation of interleukin-3- dependent multipotential stem cell colony growth by hemin in "serum-free" marrow cultures. J Cell Biol 107:700a (abs).

89. Kragh-Hansen U (1981) Molecular aspects of ligand binding to serum albumin. Pharmacol Rev 33:17.

90. Weinberg RS, He L, Alter BP (1989) Hemin stimulation of globin synthesis in normal (Hb AA) and sickle cell (Hb SS) BFU-E derived colonies is mediated by adherent accessory cells. Blood  74:322a (abs).

91. Vaquez HM (1892) Sur une forme spéciale de cyanose s'accompagnant d'hyperglobulie excessive et persistante. CR Soc Biol (Paris) 44:384.

92. Osler W (1903) Chronic cyanosis with polycythemia and enlarged spleen: a new clinical entity. Am J Med Sci 126:187.

93. Türk W (1904) Beitrage zur kenntnis des symptomenbildes polycythamie mit milztumor und zyanose. Wien Klin Wochenschr 17:153.

94. Hutchinson R, Miller CH (1906) A case of splenomegalic polycythaemia, with report of post-mortem examination. Lancet 1:744.

95. Hirsch R (1935) Generalized osteosclerosis with chronic Polycythemia vera.  Arch Pathol 19:91.

96. Lemoine F, Najman A, Baillou C, Stachowiak J, Boffa G, Aegerter P et al. (1986) A prospective study of the value of bone marrow erythroid progenitor cultures in polycythemia. Blood 68:996.

97. de Klerk G, Rosengarten PC, Vet RJ, Goudsmit R (1981) Serum erythropoietin (ESF) titers in polycythemia. Blood 58:1171.

98. Krantz SB (1968) Response of Polycythemia vera marrow to erythropoietin in vitro . J Lab Clin Med 71:999.

99. Moriyama Y (1969) Studies on regulation of erythropoiesis in Polycythemia vera. Jpn J Clin Invest 10:701.

100. Ward HP, Vautrin R, Kurnick J (1974) Presence of a myeloproliferative factor in patients with Polycythemia vera and agnogenic myeloid metaplasia. Proc Soc Exp Biol Med 147:305.

101. McLeod DL, Shreeve MM, Axelrad AA (1974) Improved plasma culture system for production of erythrocytic colonies in vitro : quantitative assay method for CFU-E. Blood 44:517.

102. Prchal JF, Adamson JW, Steinmann L, Fialkow PJ (1976) Human erythroid colony formation in vitro : Evidence for clonal origin. J Cell Physiol 89:489

103. Prchal JF, Axelrad AA (1974) Bone marrow responses in Polycythemia vera.  New Engl J Med 290:1382.

104. Prchal JF, Axelrad AA (1974) Abnormal requirements for erythrocytic colony formation in vitro in Polycythemia vera. Delivered at the International Society of Hematology Meeting, Jerusalem, Israel.

105. Adamson JW, Fialkow PJ, Murphy S, Prchal JF, Steinmann L (1976) Polycythemia vera : stem-cell and probable clonal origin of the disease. New Engl J Med 295:913.

106. Prchal JF, Adamson JW, Murphy S, Steinmann L, Fialkow PJ (1978) The in vitro response of normal and abnormal stem cell lines to erythropoietin. J Clin Invest 61:1044.

107. Adamson JW, Singer JW Catalano P, Murphy S, Lin N, Steinmann L et al. (1980) Polycythemia vera: Further in vitro studies of hematopoietic regulation. J Clin Invest 66:1363.

108. Zanjani ED, Lutton JD, Hoffman R, Wasserman LR (1977) Erythroid colony formation by Polycythemia vera bone marrow in vitro : dependence on erythropoietin. J Clin Invest 59:841.

109. Golde DW, Bersch N, Cline MJ (1977) Polycythemia vera : hormonal modulation of erythropoiesis in vitro . Blood 49:399.

110. Clarke BJ, Housman D (1977) Characterization of an erythroid precursor cell of high proliferative capacity in normal human peripheral blood. Proc Natl Acad Sci (USA) 74:1105.

111. Eaves CJ, Eaves AC (1978) Erythropoietin (Ep) dose-response curves for three classes of erythroid progenitors in normal human marrow and in patients with Polycythemia vera. Blood 52:1196.

112. Eaves AC, Eaves CJ (1984) Erythropoiesis in culture. In McCulloch EA (ed) Clinics in Haematology, Toronto, WB Saunders Company, p.371.

113. Eaves AC, Eaves CJ (1983) In vitro studies of erythropoiesis in Polycythemia vera. In CDR Dunn (ed) Current concepts in erythropoiesis, John Wiley & Sons, p. 167.

114. Aye MT, Sequin JA, McBurney JP (1979) Erythroid and granulocytic colony growth in cultures supplemented with human serum lipoproteins. J Cell Physiol 99:233.

115. Prchal JF (1978) Erythroid progenitors in peripheral blood of normal and polycythemic subjects. ICN-UCLA Symposium on hematopoietic cell differentiation, p. 421 (abs.).

116. Lacombe C, Casadevall B, Varet B (1978) Mise en évidence dans le sang au cours de la maladie de Vasquez d'une différenciation érythroblastique in vitro en 14 jours sans adjonction d'érythropoietine. C R Heb Acad Sci (Paris) 286:551.

117. Lacombe C, Casadevall N, Varet B (1980) Polycythemia vera: in vitro studies of circulating erythroid progenitors. Br J Haematol 44:189.

118. Tchernia G, Mielot F, Coulombel L, Mohandas N (1981) Characterization of circulating erythroid progenitor cells in human newborn blood. J Lab Clin Med 97:322.

119. Cashman JD, Eaves CJ, Eaves AC (1988) Unregulated proliferation of primitive neoplastic progenitor cells in long-term Polycythemia vera marrow cultures. J Clin Invest 81:87.

120. Zanjani ED, Weinberg RS, Nomdedeu B, Kaplan ME, Wasserman LR (1978) In vitro assessment of similarities between erythroid precursors of fetal sheep and patients with Polycythemia vera. In Murphy MJ Jr. (ed) In vitro aspects of erythropoiesis, Berlin, Springer Verlag, p. 118.

121. Kurtz A, Hartl W, Jelkmann W, Zapf J, Bauer C (1985) Activity in fetal bovine serum that stimulates erythroid colony formation in fetal mouse livers is insulinlike growth factor I. J Clin Invest 76:1643.

122. Papayannopoulou Th, Buckley J, Nakamoto B, Kurachi S, Nute PE, Stamatoyannopoulos G (1979) Hb F production in endogenous colonies of Polycythemia vera .  Blood 53:446.

123. Kidoguchi K, Ogawa M, Karam JD, Martin AG (1978) Preferential synthesis of fetal hemoglobin (Hb F) in culture by human erythropoietic precursors in the marrow and peripheral blood. Clin Res 26:350A.

124. Fauser AA, Messner HA (1979) Fetal hemoglobin in mixed hemopoietic colonies (CFU-GEMM), erythroid bursts (BFU-E) and erythroid colonies (CFU-E): assessment by radioimmune assay and immunofluorescence.  Blood 54:1384.

125. Papayannopoulou Th, Nakamoto B, Kurachi S, Stamatoyannopoulos G (1981) Globin synthesis in erythroid bursts that mature sequentially in culture. I. Studies in cultures of adult peripheral blood BFU-Es. Blood 58:969.

126. Constantoulakis P, Nakamoto B, Papayannopoulou Th, Stamatoyannopoulos G (1990) Fetal calf serum contains activities that induce fetal hemoglobin in adult erythroid cell cultures. Blood 75:1862.

127. Hoffman R (1979) Fetal hemoglobin in Polycythemia vera : cellular distribution in 50 unselected patients. Blood 53:1148

128. Umemura T, Constantoulakis P, Papayannopoulou Th, Stamatoyannopoulos G (1990) Differentiation kinetics and globin gene expression by circulating human BFU-E in suspension cultures. Exp Hematol 18:1116.

129. Migliaccio AR, Migliaccio G, Brice M, Constantoulakis P, Stamatoyannopoulos G, Papayannopoulou Th (1990) Influence of recombinant hematopoietins and of fetal bovine serum on the globin synthetic pattern of human BFU-E. Blood 76:1150.

130. Gabbianelli M, Pelosi E, Labbaye C, Valtieri M, Testa U, Peschle C (1990) Reactivation of Hb F synthesis in normal adult erythroid bursts by IL-3.  Br J Haematol 74:114.

131. Najman A, Baillou C, Drouet X, Krystal G, Isnard F, Gorin NC (1987) Late erythroid progenitors in Polycythemia vera are independent of erythropoietin. Blood 70:264a (abs ).

132.  de Wolf JT, Beentjes JA, Esselink MT, Smit JW, Halie RM, Clark SC, Vellenga E (1989) In Polycythemia vera human interleukin 3 and granulocyte-macrophage colony-stimulating factor enhance erythroid colony growth in the absence of erythropoietin. Exp Hematol 17: 981.

133. Prchal JF, Axelrad AA, Crookston JH (1974) Erythroid colony formation in plasma culture from cells of peripheral bloood in myeloproliferative disorders. Blood 44:912 (abs).    

134. Nissen C, Hasler E, Moor K, Moser Y, Speck B (1986) Polycythemia vera : spontaneous growth of hemoglobinized colonies is mediated by adherent cells.  Exp Hematol 14:549 (abs).

135. Kurtz A, Jelkmann W, Bauer C (1983) Insulin stimulates erythroid colony formation independently of erythropoietin. Br J Hemaetol 53:311.

136.  Kurtz A, Jelkmann W, Bauer C (1982) A new candidate for the regulation of erythropoiesis: insulin-like growth factor I. FEBS Lett 149:105.

137. Hicks DG, Ohlsson-Wilhelm BM, Farley BA, Kosciolek BA, Rowley PT (1985) K562 cell erythroid differentiation: requirement for a factor in fetal bovine serum. Exp Hematol 13:273.

138. Lammers R, Gray A, Schlessinger J, Ullrich A (1989) Differential signaling potential of insulin- and IGF-I-receptor cytoplasmic domains. EMBO J 8:1369.

139. Sawada K, Krantz SB, Dessypris EN, Koury ST, Sawyer ST (1989) Human colony-forming units-erythroid do not require accessory cells but do require direct interaction with insulin-like growth factor I and/or insulin for erythroid development. J Clin Invest 83:1701.

140. Claustres M, Chatelain P, Sultan C (1987) Insulin-like growth factor I stimulates human erythroid colony formation in vitro.  J Clin Endocrinol Metab 65:78.

 141. Merchav S, Tatarsky I, Hochberg Z (1988) Enhancement of erythropoiesis in vitro by human growth hormone is mediated by insulin-like growth factor I. Br J Haematol 70:267.

142. Trapnell BC, Nagaoka I, Nukiwa T, Rom W, Borok Z, Crystal RG (1988) Modulation of insulin-like growth factor-I gene expression in human mononuclear phagocytes. J Cell Biol 107:702a (abs).

143. Dainiak N, Kreczko S (1985) Interaction of insulin, insulinlike growth factor II, and platelet-derived growth factor in erythropoietic culture. J Clin Invest 76:1237.

144. Werther GA, Haynes K, Johnson GR (1990) Insulin-like growth factors promote DNA synthesis and support cell viability in fetal hemopoietic tissue by paracrine mechanisms. Growth Factors 3:171.

145. Congote LF (1984) High-performance liquid chromatographic separation of serum erythrotropin and erythropoietin. J Chromat 310:396.

146. Congote LF (1987) Extraction of an erythrotropin-like factor from bovine serum albumin (Cohn Fraction V). In Vitro Cell Develop Biol 23:361.

147. Holbrook ST, Ohls RK, Schibler KR, Yang YC, Christensen RD (1991) Effect of interleukin-9 on clonogenic maturation and cell-cycle status of fetal and adult hematopoietic progenitors. Blood 77:2129.