Studies on hematopoietic stem cells: XI. Lack of erythroid burst-forming units (BFU-E) in patients with aplastic anemia.

The marrow concentration of erythropoietic precursors was examined in normal donors and patients with idiopathic aplastic anemia using a plasma clot culture system. On time course observations the heterogeneity of human erythroid precursors assayable in culture was demonstrated. To evaluate human erythropoiesis in vitro, the benzidine-positive colonies were divided into three groups: small colony, containing 8-50 cells; medium-sized colony, containing 50-500 cells; and large colony, containing more than 500 cells. The majority of the large colonies assumed the morphology of erythropietic bursts (BFU-E) consisted of several subcolonies. The small colonies were counted as CFU-E1, the medium-sized as CFU-E2, and the large as BFU-E to evaluate the erythroid precursor cell compartment in aplastic anemia. The marrow concentration of CFU-E1 and CFU-E2 was shown to be quantitatively diminished in aplastic anemia. In addition, there was no ability of the marrow cells from aplastic patieints to grow BFU-E in vitro even in the presence of a large dose of erythropoietin. This lack of BFU-E colony growth may play an important role in the mechanism of the erythropoietic deficiency in aplastic anemia.     

Erythroid and granulocyte-macrophage colony formation in myelodysplastic syndromes

Colony formation by haematopoietic progenitors from the bone marrow was studied in 44 patients with a myelodysplastic syndrome. Erythroid progenitors BFU-E and CFU-E were cultured in methyl cellulose, and granulocyte-macrophage precursors CFU-GM in agar. 3 of 32 patients showed normal numbers of BFU-E colonies; in all the other cases the number of these colonies was below the normal range. CFU-E colony formation was subnormal in all cases. 23 of 44 patients grew normal numbers of colonies and clusters in CFU-GM cultures. These patients had refractory anaemia with ring sideroblasts (FAB-classification) or 5q-karyotype anomaly in the marrow. Patients lacking both of these findings exhibited reduced colony formation or excessive growth of colonies and/or clusters, with few exceptions. In conclusion, we found that erythroid colony formation was defective in all cases. Normal granulocyte-macrophage colony formation was associated with refractory anaemia with ring sideroblasts or the presence of 5q- karyotype anomaly.     

In vitro steroid sensitivity testing: A possible means to predict response to therapy in primary hypoproliferative anemia

We investigated the effects of various steroids on erythroid colony formation by normal human bone marrow and peripheral blood, and by marrow and peripheral blood from 18 patients with primary hypoproliferative anemia. These agents were variously found to enhance both CFU-E and BFU-E derived colony growth by normal human cells. Fluoxymesterone and dexamethasone were the most active inducers of CFU-E proliferation, and etiocholanolone and dexamethasone were the most potent burst augmenters. Androsterone did not significantly influence BFU-E proliferation in 66% of the marrow cultures from hematologically normal donors. Colony formation by erythroid progenitor cells of the patients with hypoproliferative anemia was reduce (20 ± 10 CFU-E derived colonies/6 × 10⁴ marrow cells; 12 ± 5 BFU-E derived colonies/1 × 10⁵ blood cells) when compared to growth by normal cells (65 ± 14 CFU-E derived colonies/6 × 10⁴ marrow cells; 21 ± 9 BFU-E derived colonies/1 × 10⁵ blood cells). Colony formation by marrow or peripheral blood cells of eight patients with steroid-responsive anemia was only moderately reduced (26 ± 11 CFU-E derived colonies/6 ± 10⁴ marrow cells; 17 ± 3 BFU-E derived colonies/1 × 10⁵ blood cells) when compared to growth by marrow cells of three steroid-unresponsive patients (3 ± 1.5 CFU-E derived colonies/6 × 10⁴ cells). Whereas the addition of steroids of the same class to marrow and peripheral blood cultures of the steroid-responsive patients enhanced colony growth by 60–300%, their addition to marrow cultures of the steroid-unresponsive patients increased colony growth by less than 60%. It appears that further investigations using in vitro culture techniques as predictors of response to steroid therapy in patients with hypoproliferative anemia may be warranted.     

Transient erythroblastopenia of childhood: Varied pathogenesis

The mechanism of anemia in four patients with transient red cell aplasia of childhood (“erythroblastopenia”) was studied at the time of diagnosis by assessing the colony growth of marrow erythroid progenitors in methylcellulose tissue cultures. Marrow from Patient 1 yielded high normal numbers of BFU-E colonies that were completely abolished on addition of autologous serum or IgG. Patient 2 had normal BFU-E growth that markedly declined when autologous serum or IgM was added to the cultures, but growth remained unchanged with added autologous IgG or peripheral blood mononuclear cells (PBMC). Marrow from Patient 3 yielded low CFU-E and BFU-E numbers with standard plating techniques, but colonies strikingly increased when marrow fractions from an albumin density gradient were cultured. PBMC from Patient 3 suppressed control marrow CFU-E and BFU-E, but serum had no effect. Patient 4 had normal CFU-E and BFU-E that increased with autologous serum and remained unchanged with autologous PBMC. We conclude that the red cell aplasia in Patients 1, 2, and 3 was due to suppressed erythropoiesis via IgG, IgM, and cell-mediated inhibition, respectively. In contrast, in Patient 4 no immune mechanism was demonstrated. Whereas transient red cell aplasia has a uniform clinical presentation, there are at least four pathogenetic mechanisms that can be detected in vitro.     

Erythroleukemia: in vitro studies of erythropoiesis.

We studied the growth of erythroid burst-forming units (BFU-E) and erythroid colony forming units (CFU-E) from bone marrow and blood in six patients with erythroleukemia. Five patients grew CFU-E, while BFU-E were found in the marrow of two and in the peripheral blood of only one patient. In all cases with colony growth, the numbers of colonies were markedly decreased with respect to normal controls. Patient BFU-E were composed of fewer clusters than those of controls. BFU-E and CFU-E growth was dependent on the addition of erythropoietin to the medium, and no growth was observed in absence of erythropoietin. At present it is not known if the growth obtained is derived from residual normal erythropoietic stem cells or from abnormal erythroid precursors of the leukemic cells.     

Congenital dyserythropoietic anaemia type II (CDA-II): chromosomal banding studies and adherent cell effects on erythroid colony (CFU-E) and burst (BFU-E) formation

Bone marrow CFU-E and BFU-E from a patient with CDA-II formed erythroid colonies and bursts which contained multinucleated erythroblasts in vitro . Adherent cell depletion of the patient's marrow increased CFU-E derived colonies six-fold (98 ± 17 v. 640 ± 15 per 10⁵ marrow cells plated) and co-culture of CDA-II marrow adherent cells with CDA-II adherent cell depleted marrow significantly suppressed erythroid colony formation. Similar adherent cell suppression of the patient's BFU-E also occurred. Adherent cell depletion of normal marrow did not increase CFU-E derived colony formation (488 ± 63 v. 495 ± 108) and decreased BFU-E derived burst formation. Addition of normal adherent cells to normal marrow increased erythroid colony and burst formation. Karyotype and chromosomal banding studies of the patient's multinucleated cells did not show chromosomal inversions, deletions or translocations.     

Demonstration of three distinct immunological disorders on erythropoiesis in a patient with pure red cell aplasia and autoimmune haemolytic anaemia associated with thymoma

A patient with pure red cell aplasia (PRCA) and autoimmune haemolytic anaemia (AIHA), associated with a thymoma which had already been removed, was studied in order to investigate the pathogenesis of PRCA and AIHA. The autoantibody eluted from the surface of the patient's red blood cells (RBC) reacted with the large E antigen of the Rh complex. Immunoglobulin-G (IgG) purified from the patient's serum suppressed CFU-E and BFU-E but not CFU-GM colony formation in the presence of complement. This antibody was not adsorbed with large E antigen. T-lymphocytes in the bone marrow suppressing autologous CFU-E and BFU-E colonies were demonstrated. Thus, three distinct immunological disorders on erythropoiesis were present in this patient with PRCA and AIHA associated with thymoma in a thymectomized state.     

Pure red cell aplasia: lymphocyte inhibition of erythropoiesis

The pathogenesis of pure red cell aplasia (PRCA) was studied in a patient who had no evidence of malignancy. In marrow culture, no erythroid colonies (from late erythroid progenitors [CFU-E]) but normal numbers of well-haemoglobinized erythroid bursts (from early erythroid progenitors [BFU-E]) were found, indicating that BFU-E existed in the patient but that their subsequent in vivo differentiation was inhibited. Autologous coculture studies suggested that inhibition was mediated by the patient's ER+ lymphocytes. After remission was induced with cyclophosphamide, autologous ER + cells no longer suppressed in vitro erythropoiesis. However, cryopreserved ER + cells, obtained with anaemia, suppressed BFU-E growth from remission marrow. An expanded population of large granular lymphocytes (LGL) with ER +, F_cγ+. T3 +, T8 +, HNK-1 +, Ia —, M1 — phenotype and no functional natural killer (NK) cell activity was noted during PRCA that reverted to normal with remission. For this patient, both in vivo and in vitro evidence demonstrates a cellular inhibition of erythropoiesis at the level of differentiation between BFU-E and CFU-E.     

Human interleukin (IL)-9 specifically stimulates proliferation of CD34+++DR+CD33- erythroid progenitors in normal human bone marrow in the absence of serum.

The cDNA encoding human interleukin (IL)-9 has recently been cloned and the recombinant molecule found to enhance erythroid colony formation in vitro by bone marrow, peripheral blood, and cord blood cells. In our present report, recombinant human (rhu) IL-9 was evaluated, alone and in combination with other cytokines, for its effect on colony formation by erythroid progenitor (erythroid burst-forming units, BFU-E) and precursor (erythroid colony-forming units, CFU-E) cells in low density (LD), nonadherent LD density T-lymphocyte-depleted (NALT-), and immunofluorescence-sorted CD34+++DR+ and CD34+++DR+CD33- cells from normal human bone marrow. When highly enriched CD34+++DR+ and CD34+++DR+CD33- cells were plated at 200 and 100 cells/ml in the presence of 5% (vol/vol) 5637-cell-conditioned medium and erythropoietin (Epo) under serum-containing conditions, 46 and 51 day-14 BFU-E were observed, respectively. The enhancing effect of rhuIL-9 was similar to that of 5637 CM on colony formation by Epo-dependent BFU-E and CFU-E in these enriched sorted CD34+++DR+ and CD34+++DR+CD33- cells under serum-containing and serum-depleted culture conditions. No significant synergistic or additive effect of rhuIL-9 was noted when used in conjunction with rhu interleukin 3 (rhuIL-3), rhu interleukin 6 (rhuIL-6), and/or rhu granulocyte-macrophage colony-stimulating factor (rhuGM-CSF) under the same culture conditions. The cloning enhancing effect elicited by human IL-9 is Epo dependent, although IL-9 alone sustains the survival of erythroid progenitor cells in vitro, as assessed by delayed additions of Epo to the cultures. The ability of human IL-9 to stimulate BFU-E and CFU-E colony formation using low numbers of highly enriched progenitor cells in serum-depleted conditions demonstrates the direct effect of IL-9 on erythroid progenitors and implicates its potential role in the enhancement of erythropoiesis.     

Heme metabolism and in vitro erythropoiesis in anemia associated with hypochromic microcytosis

Heme metabolism and in vitro erythropoietic growth (CFU-E, BFU-E) were examined in bone marrow cells taken from two siblings with apparent familial hypochromic microcytic anemia. Bone marrow cells from both patients grew adequate numbers of CFU-E and BFU-E colonies in culture in the presence of erythropoietin. In addition, small numbers of endogenous CFU-E were seen in 7-day cultures. Assays on bone marrow cells taken from both patients revealed that baseline delta-aminolevulinic synthase activity was considerably reduced, but increased six to seven fold (to normal levels) when patients' cells were exposed to pyridoxal phosphate (PLP). In both cases, ferrochelatase and delta-aminolevulinic acid dehydratase activities were normal. Bone marrow heme oxygenase showed no significant differences in activities between normals and patients values in the absence or presence of PLP. In contrast, heme synthesis by patients' bone marrow was less than that of normals. This study demonstrates that bone marrow cells from patients with this rare disorder have some disturbances in heme metabolism, whereas erythropoiesis appeared to be normal when cultured with adequate nutrients in vitro.     

Humoral suppression of erythropoiesis in systemic lupus erythematosus (SLE) and rheumatoid arthritis

Anemia due to inadequate red cell production often accompanies systemic lupus erythematosus and rheumatoid arthritis. We investigated its pathogenesis in 17 patients with these disorders, using a plasma clot culture system. In serum from normal donors and nonanemic patients CFU-E derived colony formation was not significantly altered by normal marrow cells (mean 74±12 colonies/6 × 10⁴ cells), whereas colony formation was inhibited (mean 36 ± 6 colonies/6 × 10⁴ cells) in serum from 10 anemic patients. In serum from anemic patients proliferation of the more primitive BFU-E was also reduced in three cases. In two patients with a humoral inhibitor, colony growth was suppressed by autologous marrow cells. In another patient without an inhibitor, colony formation was not suppressed by autologous bone marrow.The physical properties of this inhibitor are compatible with those of an immunoglobulin. Moreover, its presence is related to disease activity and it can be removed by successful therapy with either corticosteroids or plasma exchange. Circulating inhibitors of erythropoiesis may play an important role in causing severe anemia in patients with these rheumatic diseases.     

Suppression of normal human erythropoiesis by human recombinant DNA-produced alpha-2-interferon in vitro

Interferons (IFN) have been shown to suppress the proliferation of human erythroid progenitors (BFU-E, CFU-E) in vitro. We have previously demonstrated that the inhibition of erythroid colony formation by gamma-IFN in vitro is mediated, in part, through the activation of monocytes and T-lymphocytes. In order to examine the mechanism(s) underlying the inhibitory action of one type of recombinant alpha-IFN (alpha-2-IFN) on erythropoiesis, the effect of different doses (80-10,000 U) of alpha-2-IFN on erythroid colony formation by normal human bone marrow cells in the presence or absence of monocytes and/or T cells was studied. The addition of alpha-2-IFN to whole marrow caused the suppression of BFU-E (10%-68%) and CFU-E (5%-75%) in a dose-dependent fashion. This inhibition occurred with the direct addition of alpha-2-IFN to culture plates but not with brief preincubation of marrow cells with alpha-2-IFN followed by washing of the cells. By contrast, brief exposure of marrow cells to gamma-IFN resulted in significant suppression of erythroid colony formation. The inhibitory action of alpha-2-IFN was not influenced by erythropoietin. Removal of monocytes and/or T cells prior to the addition of alpha-2-IFN failed to significantly reduce the suppressive effects of this molecule (BFU-E: 21%-66%; CFU-E: 20%-83%). Coculture of purified monocytes or T-lymphocytes preexposed to alpha-2-IFN with autologous bone marrow cells did not cause suppression of erythropoiesis; monocytes or T cells similarly treated with gamma-IFN, however, inhibited autologous BFU-E and CFU-E in vitro. These results demonstrate that, unlike gamma-IFN, the inhibitory effect of alpha-2-IFN on erythroid colony formation in vitro is not mediated to any significant degree through monocytes and T-lymphocytes. The suppressive effect of alpha-2-IFN occurs either directly at the erythroid progenitor(s) level and/or through accessory cell(s) other than monocytes and T cells.     

Tumor necrosis factor-alpha and hematopoietic progenitors: effects of tumor necrosis factor on the growth of erythroid progenitors CFU-E and BFU-E and the hematopoietic cell lines K562, HL60, and HEL cells

Macrophages can modulate the growth of hematopoietic progenitors. We have examined the effects of tumor necrosis factor-alpha, a product of activated macrophages, on human erythroid progenitors (CFU-E, BFU-E) and the hematopoietic cell lines K562, HL60, and HEL cells. Tumor necrosis factor (TNF) significantly inhibited CFU-E and BFU-E growth at concentrations as low as 10(-11)-10(-12) M (0.2 U/ml), although erythroid colony and burst formation were not totally ablated. Preincubation of marrow samples with TNF for 15 min was sufficient to suppress erythroid colony and burst formation. Addition of TNF after the start of culture inhibited CFU-E- and BFU-E-derived colony formation if TNF was added within the first 48 h of culture. Additionally, TNF inhibited the growth of highly purified erythroid progenitors harvested from day 5 BFU-E. The colonies which formed in cultures treated with TNF were significantly smaller than those formed in control cultures. TNF (10(-8)-10(-10) M) also suppressed the growth of the hematopoietic cell lines K562, HL60, and HEL cells, with 40%-60% of the cells being sensitive to TNF. Preincubation of HL60 cells with TNF for 15 min significantly inhibited their growth. K562, HL60, and HEL cells expressed high-affinity receptors for TNF in low numbers (6000-10,000 receptors per cell). Fluorescence-activated cell sorter analysis of TNF binding to HEL cells demonstrated that the majority of these cells expressed TNF receptors. These data suggest that: (1) TNF is a rapid irreversible and extremely potent inhibitor of CFU-E, BFU-E, and hematopoietic cell lines K562, HL60, and HEL cells; (2) TNF appears to be acting on a subpopulation of erythroid cells, predominantly CFU-E, BFU-E, and possibly proerythroblasts; (3) TNF appears not to require accessory cells such as lymphocytes or macrophages to inhibit erythroid progenitors; and (4) the presence of TNF receptors on hematopoietic cells is not sufficient to confer sensitivity to TNF since the majority (80%-95%) of HEL cells express TNF receptors while only 40%-60% are inhibited by TNF.     

Chronic exposure to tumor necrosis factor in vivo preferentially inhibits erythropoiesis in nude mice

The anemia of chronic disease (ACD) is associated with conditions in which macrophage activation occurs. Activated marrow macrophages suppress erythropoiesis in vitro and produce tumor necrosis factor (TNF). Therefore, we tested the effects of chronic in vivo exposure to TNF to determine if it was a candidate for a mediator of ACD. Nude mice were inoculated with Chinese hamster ovary (CHO) cells expressing the human TNF gene or with control cells containing the transfection vector alone. The TNF mice promptly became reticulocytopenic, and after 3 weeks their corrected reticulocytes were 2.6% +/- 0.7% as compared with 7.3% +/- 4% in control mice. The hematocrit at 3 weeks was 28.4% +/- 1.7% in TNF mice as compared with 46% +/- 0.8% in control mice. This anemia was also associated with low serum iron and normal iron stores and increased erythropoietin (Epo) levels. The TNF mice showed an absolute monocytosis with twice the number of circulating monocytes as control mice and had M-colony-stimulating factor (CSF) activity in their serum. The TNF mice also became mildly thrombocytopenic. Marrow CFU-E and BFU-E were profoundly decreased (1.2 +/- 0.2 x 10(3) v 8.6 +/- 0.2 x 10(4) CFU-E per femur, and 6.5 +/- 1 x 10(2) v 8.5 +/- 0.2 x 10(4) BFU-E per femur). Splenic CFU-E and BFU-E were similarly depressed. In contrast, marrow CFU-GM and CFU-GEMM were not affected. The residual BFU-E in TNF mice were relatively resistant to TNF as compared with control mice. These data demonstrate that TNF preferentially inhibits erythropoiesis in vivo and may be important in the pathogenesis of ACD.     

Appearance of adherent cells suppressive to erythropoiesis during an early stage of Plasmodium berghei infection in mice.

We examined the effect of adherent cells from bone marrow or spleen of mice infected with Plasmodium berghei on dyserythropoiesis. Significant reduction in number of erythroid progenitors (erythroid colony-forming units: CFU-E and erythroid burst-forming units: BFU-E) was observed in bone marrow as early as 1 day after P. berghei infection. When adherent cells were removed from bone marrow or spleen cells of infected mice, the number of CFU-E and BFU-E was clearly increased. Furthermore, addition of adherent cells from infected mice to nonadherent cells from normal mice inhibited erythroid colony formation significantly in a dose-dependent manner. These results suggest that the adherent cells obtained from bone marrow or spleen of mice in the early stage of P. berghei-infection have a suppressive effect on erythropoiesis.     

Suppression of mouse erythroid colony formation by L1210 leukemia cells.

The effect of L1210 transplantable leukemic cells on in vitro formation of erythroid colonies from CD2F1 mouse bone marrow progenitor cells (CFU-E) was investigated. Clonal cell culture was carried out by a methylcellulose technique. Human urinary erythropoietin served as the stimulator. After 44 hours of incubation aggregates of eight or more erythroid cells were scored as colonies. The number of CFU-E which could be demonstrated in marrow cells from mice that had been injected intravenously 6 days before with 5 x 10(4) L1210 cells was far below that obtained from normal marrow cells. When 1.3 x 10(5) marrow cells from leukemic mice or L1210 ascites cells were cultured with an equal amount of normal cells, the number of CFU-E expressed was reduced by 51% and by 86%, respectively, relative to controls with normal cells only. Neither lethally irradiated L1210 cells (4500 rad) nor L1210 cell conditioned media suppressed erythroid colony formation. It is suggested that in L1210 leukemia erythropoiesis is decreased because of a cell-to-cell inhibitory action of the leukemia cells on CFU-E.     

Erythroid progenitors in Rauscher leukemia virus variant-A-induced erythropoietic dysplasia in mice

In vitro erythropoiesis by bone marrow and spleen cells was assessed in normal mice and during progression of Rauscher leukemia virus, variant-A (RLV-A) disease in mice. As RLV-A disease progressed from early through terminal stages, there was a marked increase in the numbers of in vitro splenic CFU-E and BFU-E. Conversely, bone marrow CFU-E and BFU-E demonstrated a concomitant decrease in numbers with disease progression. At no time were erythropoietin-independent (endogenous) erythroid colonies generated. The results suggest that compartmental alterations in erythroid precursors occur during progression of RLV-A.     

The regulatory role of interleukin 2-responsive T lymphocytes on early and mature erythroid progenitor proliferation

To analyze the role of T lymphocytes in human erythropoiesis, we evaluated the effect of recombinant interleukin 2 (IL 2) on marrow CFU- E and BFU-E colony formation in vitro. IL 2 resulted in an increase in CFU-E and BFU-E colony numbers in a dose-dependent manner. This increase could be prevented by anti-Tac, a monoclonal antibody to the IL 2 receptor. Moreover, anti-Tac on its own resulted in an overall decrease in colony numbers. Depletion of marrow adherent cells did not alter the effect of either IL 2 or anti-Tac on colony growth. Following the removal of marrow T lymphocytes, CFU-E and BFU-E colony formation proceeded normally; however, the effects of IL 2 and anti-Tac were markedly diminished. Readdition of T lymphocytes to the cultures restored the IL 2 effect. Although T lymphocytes were not themselves essential for in vitro erythropoiesis, our studies suggest that IL 2 and IL 2-responsive T cells can regulate both early and mature stages of erythroid differentiation.     

Clonal analysis of peripheral blood and haemopoietic colonies in patients with aplastic anaemia and refractory anaemia using the polymorphic short tandem repeat on the human androgen-receptor (HUMARA) gene

The clonalities in white blood cells (WBC) of blood and nucleated bone marrow cells from patients with refractory anaemia and aplastic anaemia were examined by polymerase chain reaction (PCR) methods using the polymorphic short tandem repeat (STR) on the human androgen-receptor gene (HUMARA). Peripheral blood samples were obtained from 12 female patients, six with aplastic anaemia (AA) and six with refractory anaemia (RA). Peripheral blood was fractionated into granulocytes, lymphocytes, T lymphocytes and B lymphocytes. DNA was extracted from each fraction. Bone marrow samples were obtained from seven female patients (three with AA and four with RA). Sorted CD34 positive cells were cultured in a semisolid culture system. DNA was extracted from a 14-day haemopoietic colony. The clonal pattern was assessed using HUMARA gene STR polymorphism and the differential methylation pattern of nearby cytosine residues by PCR methods. Four of six (67%) AA and two of six (33%'RA patients had a monoclonal proliferating pattern in their granulocytes. The ratio of the numbers of minority colonies per majority colonies (m/M ratio) was examined for seven patients (three AA and four RA). In patients who had a clonal haemopoietic pattern in peripheral WBC the ratio was under 0.4 but not zero. In contrast, patients exhibiting a polyclonal pattern had an m/M ratio above 0.8. We concluded that some normal or heterogenous haemopoietic clones, not only MDS but also AA, may remain in the bone marrow, although almost all colonies were derived from a single pathogenic clone when the clonality pattern exhibited monoclonality in peripheral blood analysis.     

Expression of latent hematopoietic progenitor cells in cultures of newborn and adult baboon liver

The anatomic site of hematopoiesis changes during fetal development from the yolk sac to the liver and finally to the marrow. Factors controlling this switch in the site of hematopoiesis are unknown. We assayed erythroid colony (CFU-E) and erythroid burst (BFU-E) formation in fetal, newborn, and adult baboon liver and marrow to determine the growth requirements of primate hematopoietic progenitor cells from different anatomic sites and developmental stages. We cocultured fetal, newborn, and adult liver and marrow nonadherent cells with adherent cells from these organs to assess the role adherent cells may play in determining the site of hematopoiesis. Fetal liver, fetal marrow, newborn marrow, and adult marrow cultures formed CFU-E and BFU-E colonies in vitro. In contrast, newborn and adult liver cell cultures very rarely formed colonies. However, when newborn or adult liver nonadherent cells were cocultured with marrow adherent cells, CFU-E and BFU-E colonies were detected. The colonies that formed in the newborn and adult liver cultures were derived from the liver and not from the marrow cells or peripheral blood trapped in the liver. These data suggest that in contrast to fetal liver, newborn and adult liver may not be hematopoietic organs in normal primates in vivo because of changes in the growth requirements of hematopoietic progenitor cells present in these organs.