Effects of recombinant human thrombopoietin on megakaryocyte colony formation and megakaryocyte ploidy by human CD34+ cells in a serum-free system.

In serum-free cultures of human CD34 cells, recombinant human thrombopoietin (TPO) induced megakaryocyte colony formation a dose-dependent fashion that was further enhanced by the presence of interleukin-3 (IL-3) and stem cell factor (SCF), but not by IL-6, IL-11 or erythropoietin. TPO gave rise to much smaller colonies and at an earlier time than IL_3, indicating that TPO affects predominantly more mature megakaryocytic progenitors. In liquid cultures. TPO increased the percentage and the absolute number of > or = 8N megakaryocytes, but it did not shift their modal ploidy from 2N. TPO-induced endomitosis was totally inhibited by the presence of or previous exposure of cells to, IL-3 and /or SCF. The mechanism by which TPO overcomes in vivo the negative effects of IL-3 and SCF on megakaryocyte ploidy remains unknown.

Regulation and specificity of MNDA expression in monocytes, macrophages, and leukemia/B lymphoma cell lines.

The expression of the human myeloid cell nuclear differentiation antigen (MNDA) was observed specifically in cells of the granulocyte-macrophage lineage in our earlier reports. The specificity of MNDA expression for cells in the granulocyte-macrophage lineage was reexamined in cell lines established from patients with Philadelphia chromosome-positive chronic myeloid leukemia. Cell lines that expressed MNDA exhibited myeloid cell features and granulocyte or monocyte differentiation could be induced in vitro, while cell lines exhibiting properties of very early stage cells or multipotential cells did not express MNDA. Cells originating from cases of Burkitt's lymphoma were negative. By contrast, three lymphoblastoid cell lines (immortalized in vitro with Epstein-Barr virus) were weakly positive and MNDA was up-regulated by interferon-alpha (IFN-alpha) treatment. As we reported previously, MNDA mRNA level in adherent monocytes is elevated by IFN-alpha; in this study, we further assessed MNDA expression in in vitro monocyte-derived macrophages. Three additional agents (endotoxin, phytohemagglutinin, and phorbol ester) and other conditions that affect function, cytokine production, differentiation, and/or growth of monocytes were examined for their ability to alter MNDA expression. The results varied with the agent, cell type, and stage of differentiation. Changes in MNDA expression occurred slowly (hours to days), suggesting that MNDA could mediate changes realized over a long period. The results also reveal a discordance in certain MNDA positive cells between steady-state levels or changes in levels of protein and mRNA indicating that the regulation of MNDA expression occurs at more than one point. Changes in MNDA expression are consistent with a role in opposing macrophage differentiation and activation of monocytes/macrophages.

Replication and endoreplication in developing megakaryocytes in vitro.

To study the relation in time between replication and endoreplication and the relation between appearance of platelet-specific proteins and endoreplication in maturing megakaryocytes, peripheral blood mononuclear cells highly enriched in hematopoietic progenitors were cultured in liquid cultures and plasma clots in the presence of either interleukin-3 (IL-3) and stem cell factor (SCF) or medium conditioned by blood mononuclear cells stimulated by phytohemagglutinin (PHA). In plasma clots, megakaryocytic (MK) colonies appeared first on day 5 and reached a maximum by day 8, whereas the number of cells per colony increased until day 10, indicating that there was a single wave of MK colony formation. In liquid cultures, the first immunologically recognizable megakaryocytes appeared on day 5 and expressed GPIIb/IIIa and thrombospondin only, but all other platelet-specific protein markers appeared within 24 hours. Replating cells from liquid medium into plasma clots showed that 92 +/- 8% of day 6 GPIIb/IIIa-positive cells are capable of replicating. Their replicative potential decreased with age, however, so that between days 6 and 11, a linear correlation was noted between the logarithm of the percentage of megakaryocytes with replicative capacity and their age in culture. Replication ceased completely after day 10. In the presence of IL-3, polyploid megakaryocytes appeared at the same time that GPIIb/IIIa was expressed, and the megakaryocyte distribution into ploidy classes remained unchanged until day 20. In the presence of PHA-leukocyte conditioned medium (PHA-LCM), ploidy of megakaryocytes was shifted toward higher classes after day 6, and the process of endoreplication was completed by day 10. No changes in ploidy distribution were noted between days 10 and 20. These findings indicate that in the cohort of megakaryocytes derived from colony-forming units-megakaryocyte (CFU-MK), endoreplication can occur at an early stage of development, proceeds synchronously with replication, and is completed before the megakaryocytes exhaust their replicative potential.

The human myeloid cell nuclear differentiation antigen gene is one of at least two related interferon-inducible genes located on chromosome 1q that are expressed specifically in hematopoietic cells.

We have previously shown that the human myeloid cell nuclear differentiation antigen (MNDA) is expressed at both the antigen and mRNA levels specifically in human monocytes and granulocytes and earlier stage cells in the myeloid lineage. A 200 amino acid region of the MNDA is strikingly similar to a region in the proteins encoded by a family of interferon-inducible mouse genes, designated Ifi-201, Ifi-202, Ifi-203, etc, that are not regulated in a cell- or tissue-specific fashion. However, a new member of the Ifi-200 gene family, D3, is induced in mouse mononuclear phagocytes but not in fibroblasts by interferon. The same 200 amino acid region, duplicated in the mouse Ifi-200 gene family, is also repeated in the recently characterized human IFI 16 gene that is constitutively expressed specifically in lymphoid cells and is induced in myeloid cells by interferon gamma. The 1.8-kb MNDA mRNA, which contains an interferon-stimulated response element in the 5' untranslated region, was significantly upregulated in human monocytes exposed to interferon alpha. Characterization of the MNDA gene showed that it is a single-copy gene and localized to human chromosome 1q 21-22 within the large linkage group conserved between mouse and human that contains the Ifi-200 gene family. The IFI 16 gene is also located on human chromosome 1q. Our observations are consistent with the proposal that the MNDA is a member of a cluster of related human interferon-regulated genes, similar to the mouse Ifi-200 gene family. In addition, one mouse gene in the Ifi-200 gene family and the human MNDA and IFI 16 genes show expression and/or regulation restricted to cells of the hematopoietic system, suggesting that these genes participate in blood cell-specific responses to interferons.

Aplastic anemia and pure red cell aplasia.

The role of known hematopoietic growth factors in the pathogenesis of aplastic anemia and congenital hypoplastic anemia has been extensively studied and no evidence has been obtained that deficiency of these factors contributes to the hypoproliferative state in these disorders. Clonal hematopoiesis seems to be present at least in a small percentage of cases of aplastic anemia, a finding that needs further investigation. Androgens were shown to be beneficial only for women with aplastic anemia treated with antilymphocyte globulin. Unrelated-donor bone marrow transplantation is becoming a realistic approach for children and very young adults with aplastic anemia, but in older groups the survival is very poor. New observations on abnormalities of lymphokines and cytokines in Fanconi's anemia have been described, but their pathogenetic significance remains unknown. A large number of studies have excluded the possibility that abnormalities of c-kit/SCF genes and their expression are responsible for the erythroid aplasia in Diamond-Blackfan syndrome. Cyclosporine was found to be an effective treatment for pure red cell aplasia associated with chronic lymphocytic leukemia. The cell membrane receptor for B19 parvovirus has been identified as the P antigen. Long-term studies showed that in 20% of patients with homozygous sickle cell disease, infection by B19 does not cause erythroid aplasia.

Interferon alpha selectively affects expression of the human myeloid cell nuclear differentiation antigen in late stage cells in the monocytic but not the granulocytic lineage.

The human myeloid cell nuclear differentiation antigen (MNDA) is expressed constitutively in cells of the myeloid lineage, appearing in myeloblast cells in some cases of acute myeloid leukemia and consistently being detected in promyelocyte stage cells as well as in all later stage cells including peripheral blood monocytes and granulocytes. The human myeloid leukemia cell lines, HL-60, U937, and THP-1, express similar levels of immunochemically detectable MNDA. Although, the level of MNDA mRNA in primary monocytes is very low it was up-regulated at 6 h following the addition of interferon alpha. The effect of interferon alpha on the MNDA mRNA is also observed in the cell lines HL-60, U937, and THP-1. The MNDA mRNA level in primary granulocytes was unaffected by addition of interferon alpha and other agents including interferon gamma, endotoxin, poly (I).poly (C), and FMLP. The MNDA mRNA level in the myeloid cell lines was also unaffected by the latter four agents. Induction of differentiation in the myeloid cell lines with phorbol ester induces monocyte differentiation which was accompanied by a decrease in MNDA mRNA level. This reduced level of mRNA could then be elevated with subsequent interferon alpha treatment. The effects of phorbol ester on MNDA mRNA appeared to be associated with induced differentiation since inhibiting cell proliferation did not alter the level of MNDA mRNA and cell cycle variation in MNDA mRNA levels were not observed. The ability of interferon alpha to up-regulate MNDA mRNA in phorbol ester treated myeloid cell lines is consistent with the observations made in primary monocytes.(ABSTRACT TRUNCATED AT 250 WORDS)

Chronic administration of transforming growth factor-beta suppresses erythropoietin-dependent erythropoiesis and induces tumour necrosis factor in vivo.

Transforming growth factor beta is a known inhibitor of the proliferation and differentiation of early haematopoietic progenitors but has no effect on mature erythroid cells in vitro. Mice injected with rhTGF beta 1 exhibited severe and progressive suppression of erythropoiesis manifested by a decline of reticulocyte count, marrow erythroblasts and marrow and spleen CFU-E, which could be prevented by administration of erythropoietin. This suppression of erythropoiesis was associated with the appearance of tumour necrosis factor in the blood, development of pronounced cachexia and depression of serum erythropoietin levels. TGF beta induces TNF in vivo that leads to cachexia, decrease of serum erythropoietin levels and suppression of erythropoietin dependent erythropoiesis.

Granulocyte/macrophage colony-stimulating factor stimulates human polymorphonuclear leukocytes to produce interleukin-8 in vitro.

Interleukin-8 (IL-8) is a potent chemotactic factor for polymorphonuclear leukocytes (PMN). Here we examine whether PMN synthesize and release IL-8 in response to stimulation by selected inflammatory cytokines. PMN isolated from normal heparinized peripheral human blood were incubated in RPMI culture medium at 37 degrees C in 5% CO2, with and without granulocyte/macrophage colony-stimulating factor (GM-CSF). The culture supernatants were tested for chemotactic activity using a modified Boyden chamber. Immunoreactive IL-8 protein was measured by ELISA with a monoclonal antibody specific for IL-8. GM-CSF (0.01 to 50 ng/ml) stimulated PMN to produce chemotactic activity in a dose- and time-dependent manner. The amount of chemotactic activity reached maximal levels after 3 h of incubation with GM-CSF. Treatment of culture media supernatants with rabbit antiserum against IL-8 blocked the GM-CSF-induced chemotactic activity. IL-8 protein concentrations detected by ELISA closely paralleled the chemotactic bioactivity in both the dose-response and kinetic studies. Northern blot analysis of total RNA from PMN using a 30 mer oligonucleotide complementary to mRNA for IL-8 yielded a single 1.6-kb band. Its intensity increased 4-fold 2 h after treatment of PMN with GM-CSF. These data suggest that peripheral blood PMN can be stimulated by GM-CSF to synthesize and secrete bioactive IL-8. Since both IL-8 and GM-CSF accumulate in sites of acute inflammation, PMN may induce IL-8 gene expression in response to GM-CSF and thereby amplify the acute inflammatory response by recruiting additional PMN into inflammatory sites.

Case report: granulocyte colony-stimulating factor overcomes severe neutropenia of large granular lymphocytosis.

Large granular lymphocytosis (LGL) is characterized by enhanced proliferation of T lymphocytes that have antibody-dependent cell-mediated cytotoxicity or natural killer cell activity and that often produce severe cytopenias, including neutropenia. When a 68-year-old man with seropositive rheumatoid arthritis and severe neutropenia was examined, he was found to have LGL with a T cell gene rearrangement, indicating the presence of a clonal population of T lymphocytes. The patient was admitted with a fever of 102 degrees F and a nonhealing ulcer over the right tibia. When the infection did not respond to intravenous antibiotics, granulocyte colony-stimulating factor (GCSF) therapy was started at 5 micrograms/kg subcutaneously each day. The neutrophil count promptly increased and the patient subsequently defervesced and was able to have a skin graft placed, which healed without difficulty. GCSF, which is known to be an effective therapeutic agent for neutropenia associated with chemotherapy and bone marrow transplantation, also was a very valuable treatment for the life-threatening neutropenia of LGL.

Polycythemia vera. II. Hypersensitivity of bone marrow erythroid, granulocyte-macrophage, and megakaryocyte progenitor cells to interleukin-3 and granulocyte-macrophage colony-stimulating factor.

Polycythemia vera (PV) is a clonal disease of the hematopoietic stem cell characterized by a hyperplasia of marrow erythropoiesis, granulocytopoiesis, and megakaryocytopoiesis. We previously reported that highly purified PV blood burst-forming units-erythroid (BFU-E) are hypersensitive to recombinant human interleukin-3 (rIL-3). Because these cells may be only a subset, and not representative of marrow progenitors, we have now studied partially purified marrow hematopoietic progenitor cells. Dose-response experiments with PV marrow BFU-E showed a 38-fold increase in sensitivity to rIL-3 and a 4.3-fold increase in sensitivity to recombinant human erythropoietin (rEpo) compared with normal marrow BFU-E. In addition, PV marrow colony-forming units-granulocyte-macrophage (CFU-GM) and CFU-megakaryocyte (CFU-MK) also showed a marked hypersensitivity to rIL-3 and to human recombinant granulocyte-macrophage colony-stimulating factor (rGM-CSF). Dose-response curves with rGM-CSF and blood BFU-E showed a 48-fold increase in sensitivity. No effect of rIL-4, rIL-6, human recombinant granulocyte-CSF (rG-CSF), or macrophage-CSF (rM-CSF) was evident, nor was there any effect of PV cell-conditioned medium on normal BFU-E, when compared with normal cell-conditioned medium. Autoradiography with 125I-rEpo showed an increase in Epo receptors after maturation of PV BFU-E to CFU-E similar to that shown with normal BFU-E, but no increase of specific binding of 125I-rIL-3 by PV CD34+ cells was seen compared with normal CD34+ cells. These studies show that PV marrow hematopoietic progenitor cells are hypersensitive to rIL-3 and rGM-CSF, similar to PV blood BFU-E. While the mechanism does not appear to be due to enhanced binding of rIL-3, the hypersensitivity of PV progenitor cells to IL-3 and GM-CSF may be a key factor in the pathogenesis of PV.

Ineffective hematopoiesis in folate-deficient mice.

A folate-free amino acid-based diet provided an opportunity to characterize the effects of folate depletion on growth, tissue folate levels, and hematopoiesis of mice under well-standardized conditions. Weanling mice were fed a folate-free, amino acid-based diet supplemented with either 0 or 2 mg folic acid/kg diet for 35 to 48 days. Folate concentrations were decreased in liver, kidney, serum, and erythrocytes in mice fed the folate-free diet. The folate-deficient mice had anemia, reticulocytopenia, thrombocytopenia, and leukopenia, all of which reverted to normal after folic acid was reintroduced to the diet. Hematopoietic organs of folate-deficient mice had alterations that were similar to those seen in folate-deficient humans except that in mice, the hyperplasia of hematopoietic tissue occurred in the spleen rather than in the marrow. Ferrokinetic studies showed a normal 59Fe-transferrin half-life, but the percentage of 59Fe-incorporation into red blood cells at 48 hours was markedly subnormal. The number of committed hematopoietic progenitors at the stages of erythroid colony-forming units (CFUs), megakaryocyte CFUs, and granulocyte-macrophage CFUs were all increased in folate-deficient mice. However, the progeny of these progenitors was markedly decreased in folate-deficient mice. Thus, the folate-deficient mice had "ineffective hematopoiesis" leading to pancytopenia, and they therefore provide a murine model of megaloblastic anemia.

Inhibition of human erythroid colony-forming units by interleukin-1 is mediated by gamma interferon.

IL-1 inhibits erythropoiesis in vivo and in vitro. This inhibition was studied by comparing the effect of recombinant human IL-1 (rhIL-1) on highly purified CFU-erythroid (E) generated from peripheral blood burst-forming units-erythroid (BFU-E) (mean purity 44.4%) with its effect on unpurified marrow CFU-E (mean purity 0.36%). Colony formation by marrow CFU-E was significantly inhibited by rhIL-1, while colony formation by highly purified CFU-E was not inhibited. However, purified CFU-E colonies were inhibited by rhIL-1 in the presence of autologous T-lymphocytes, and also by cell-free conditioned medium prepared from T-lymphocytes stimulated by rhIL-1. This inhibitory effect was ablated by neutralizing antibodies to gamma interferon (IFN), but not by antibodies to human IL-1, tumor necrosis factor, or beta IFN. Colony formation by highly purified CFU-E was also inhibited by recombinant human gamma IFN (rh gamma IFN). IL-1 and gamma IFN play significant roles in the pathogenesis of the anemia of chronic disease. These studies indicate that rhIL-1 inhibits CFU-E colony formation by an indirect mechanism involving T-lymphocytes and requiring gamma IFN and that gamma IFN itself is most probably the direct mediator of this effect.

Treatment of refractory pure red cell aplasia with cyclosporine A: disappearance of IgG inhibitor associated with clinical response.

Remissions were obtained in 6/9 evaluable patients with pure red cell asplasia (PRCA) refractory to other immunosuppressive agents who were treated with cyclosporine A (CsA). Four of these patients have remained in continuous remission off all treatment for 4-19 months. Another patient who stopped CsA abruptly relapsed, but responded to reinstitution of therapy. The sixth patient died of a cerebrovascular accident while in remission on a low dose of CsA. Acute side effects were minimal and were responsive to dose reduction. One patient developed a lymphoma while in an unmaintained remission, and one patient who did not respond to CsA was found to have a lymphoma approximately a year after stopping treatment. In vitro studies of autologous erythroid progenitors in a patient with an IgG inhibitor of erythropoiesis showed a reduction of autoantibody associated with the response to CsA. The antigen to which this inhibitor is directed was expressed only during the marrow erythroid burst-forming unit (BFU-E) period of erythroid differentiation. CsA can induce sustained remissions in cases of PRCA refractory to other multiple agents, and these remissions may be associated with a reduction in autoantibody to erythroid progenitor cells. Further studies of patients with PRCA who respond to CsA may lead to an improved understanding of this disorder.

Inhibition of human colony-forming-unit erythroid by tumor necrosis factor requires accessory cells.

Recombinant tumor necrosis factor (rTNF) inhibits erythropoiesis in vivo and in vitro. To study the mechanism of this inhibition, the effect of rTNF on highly purified human CFU-erythroid (E) (mean purity 63.5%), which were generated from peripheral blood burst-forming units-erythroid (BFU-E), was compared to its effect on unpurified human marrow CFU-E (mean purity 0.21%). Although growth of colonies from marrow CFU-E was inhibited by rTNF, no significant effect on purified BFU-E-derived CFU-E colony growth was found. Removal of accessory marrow cells by soy bean agglutinin (SBA) ablated the inhibition of marrow CFU-E colonies by rTNF. Inhibition of colony growth was then restored by adding back SBA+ cells, but not by adding T lymphocytes or adherent cells. Conditioned medium prepared from bone marrow mononuclear cells stimulated by rTNF inhibited the growth of colonies from highly purified BFU-E derived CFU-E resistant to direct inhibition by rTNF. These findings indicate that rTNF does not directly inhibit CFU-E, but requires accessory cells to decrease erythropoiesis. These accessory cells reside in the SBA+ cell fraction, but are neither T cells nor adherent cells. Therefore, in order to produce anemia, TNF must induce release or production of a factor that directly inhibits erythroid colony growth.

Thrombopoiesis-stimulating factor: its effects on megakaryocyte colony formation in vitro and its relation to human granulocyte-macrophage colony-stimulating factor.

The effects of thrombopoiesis-stimulating factor (TSF) on human marrow megakaryocyte colony formation in vitro were studied by the plasma clot method. TSF was found to stimulate megakaryocyte as well as granulocyte-macrophage colony formation in vitro at optimal concentrations of 200-300 pg/ml of medium containing 2.5% horse serum. This colony-stimulating effect of TSF was not affected by polyclonal antibodies to human (h) interleukin 3 (IL-3) or to granulocyte colony-stimulating factor (G-CSF) but was neutralized by monoclonal or polyclonal antibodies to human granulocyte-macrophage colony-stimulating factor (hGM-CSF). In order to differentiate among cross-reactivity between TSF and hGM-CSF, induction of colony growth via release of GM-CSF, and presence of hGM-CSF in TSF preparations, TSF was tested on murine marrow cells, which are not responsive to hGM-CSF. TSF induced growth of murine megakaryocyte colony-forming units (CFU-MK) and granulocyte-macrophage colony-forming units (CFU-GM) in vitro with a dose response similar to that observed on human marrow cells; however, this effect could not be neutralized by antibodies to either human or murine GM-CSF. Using a double-antibody enzyme-linked immunosorbent assay, TSF preparations were found to contain 36 +/- 4 U of hGM-CSF per picogram of TSF protein. These findings indicate that hGM-CSF is responsible for the megakaryocyte colony-promoting effects of TSF on human marrow cells in vitro.

Reversible hepatomegaly and diabetes mellitus in an adult with disseminated histiocytosis X.

Histiocytosis X rarely disseminates in an adult. The authors describe an unusual patients who presented with multiple areas of cutaneous and bone involvement. During the course of his disease he developed massive hepatomegaly. Aggregates of vacuolated histiocytes were found on liver biopsy. He subsequently developed diabetes mellitus complicated by ketoacidosis. Both his hepatomegaly and diabetes resolved spontaneously. No pancreatic nor pituitary abnormalities were identified. The combination of histiocytosis X, hepatomegaly, and diabetes mellitus has not been previously reported. The medical literature is reviewed with an emphasis on disseminated histiocytosis X in adults and the mechanism of glucose intolerance in liver disease.

The pathogenesis of anemia in rheumatoid arthritis: a clinical and laboratory analysis.

Principal concepts concerning the anemia of RA are summarized in Tables 7 and 8. These concepts have been validated by our analysis of 93 anemic RA patients and by our review of the literature. The fact that anemia in RA may have one or more etiologies, occasionally in the same patient, mandates a reasoned approach to the analysis of anemia in every RA patient in whom it may occur. In particular, iron deficiency is common and determination of bone marrow iron content via an aspirate may be required for a definitive diagnosis. In those RA patients with anemia of chronic disease, the best therapy remains control of the underlying disease, most commonly with second line drugs and/or corticosteroids. The place for recombinant erythropoietin in the therapy of this anemia has not been defined; one specific role for erythropoietin may be in the preparation of RA patients for elective surgery, particularly hip arthroplasty, where correction of the anemia may either obviate the need for transfusion or may allow for donation of blood for purposes of autologous transfusion perioperatively. The pathogenesis of the anemia of chronic disease, as seen in RA anemia, is not completely understood. Inflammatory mediators, particularly the cytokines, appear to be important factors in the impairment of erythropoiesis. The mechanism by which these cytokines impair erythroid progenitor growth and hemoglobin production in developing erythrocytes is an important area for future study.

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.

The presence of heterogeneous erythroid progenitor cells, contaminant cells, or serum may alter erythroid colony development in vitro. To obtain highly purified colony-forming units-erythroid (CFU-E), we cultured partially purified human blood burst-forming units-erythroid (BFU-E) in methylcellulose with recombinant human erythropoietin (rHuEPO) for 7 d and generated cells that consisted of 30-60% CFU-E, but no BFU-E. A serum-free medium was used that allowed development of the same number of erythroid colonies as serum containing medium, but with a greater percentage of larger colonies. This medium consisted of delipidated crystalline bovine serum albumin, iron saturated transferrin, lipid suspension, fibrinogen, thrombin, Iscove's modified Dulbecco's medium/F-12[HAM], and insulin plus rHuEPO. When CFU-E were cultured in a limiting dilution assay and the percentage of nonresponder wells was plotted against cell concentration, both serum-free cultures and serum-containing cultures yielded overlapping straight lines through the origin indicating that CFU-E development did not depend on accessory cells and that insulin acted directly on the CFU-E. Human recombinant interleukin 3 (IL-3) and/or granulocyte-macrophage colony-stimulating factor had no effect on CFU-E growth, while they markedly enhanced BFU-E growth. Physiological concentrations of recombinant human insulin-like growth factor I (IGF-I) enhanced CFU-E growth in the absence of insulin and, together with rHuEPO in serum-free medium, provided a plating efficiency equal to that of serum-containing medium. Limiting dilution analysis in serum-free medium with IGF-I showed a straight line through the origin indicating that IGF-I also acted directly on the CFU-E and not through an effect on accessory cells. These data demonstrate that CFU-E do not require accessory cells, but do require IGF-I and/or insulin which act directly on the CFU-E.