Main

Hematopoietic growth factors influence the proliferation, differentiation, and survival of hematopoietic progenitor cells(1, 2). G-CSF, Epo, thrombopoietin, and several of the ILs, are examples(1–3). Each of the hematopoietic growth factors acts by binding to specific cell-surface receptors(4). Chen et al.(5), Nicola et al.(6–8), and Walkeret al.(9) showed that the binding of one variety of hematopoietic growth factor to its specific receptor can have an effect on subsequent expression of receptors for other factors, leading to a feature that Nicola termed “trans-down modulation”(7, 8). Bradley(10), Hellman and Grate(11), Mayani et al.(12), Ulich et al.(13), de Haan et al.(14), and our group(15, 16) have all shown that, if erythropoiesis is accelerated sufficiently, neutrophil production can diminish. However, in clinical medicine, marked erythropoietic stimulation usually does not result in neutropenia. For instance, administration of rEpo to adults who have anemia of chronic renal disease accelerates erythropoiesis but does not result in neutropenia(17). When the number of pluripotent progenitors is severely limited, however, as occurs in culture dishes, splenectomized mice, neonatal animals, or preterm human infants, very high concentrations of Epo can result in diminished neutrophil production(14–16, 18–20).

rG-CSF, a neutrophil-specific growth factor, has proven useful in treating neutropenia, and its administration to adults and older children has not caused a reduction in erythrocyte production(21, 22). The potential use of rG-CSF in neonates is currently under examination(23–25). It is not clear, however, whether administration of rG-CSF in this patient population will have a deleterious effect on erythropoiesis. Indeed, de Haan et al.(14) showed that rG-CSF administration to splenectomized mice reduced erythroid progenitor cells (BFU-E and CFU-E) and diminished the hematocrit, and Stallmach et al.(26) reported that chorioamnionitis, during the second trimester of pregnancy, resulted in accelerated fetal granulocytopoiesis but diminished erythropoiesis. In view of these reports, we believed it important to examine the effect of rG-CSF on erythrocyte production from human fetal and neonatal hematopoietic progenitors, before engaging further in trials to test the potential usefulness of rG-CSF in this population.

METHODS

Subjects. The liver and long bones (femurs, tibias, humeri, and radii) were obtained from seven human abortuses at 13-24 wk of gestation. Only fetuses that were normal by ultrasound examination and from elective pregnancy termination were studied. Pregnancy terminations were carried out by suction curettage (13-16 wk of gestation) or by cervical dilatation and extraction curettage (17-24 wk of gestation). Umbilical cord blood samples from five term and five preterm placentas were obtained by direct aspiration from an umbilical vein immediately after delivery of the placenta. The studies were approved by the University of Florida Institutional Review Board.

Sample preparation and cell enumeration. Samples were processed within 1 h of the pregnancy termination. After sterilely weighing each liver, cell suspensions were prepared by placing a weighed piece of liver in sterileα-MEM (HyClone Laboratories, Logan, UT) triturating the tissue, and very gently passing the pieces through serially smaller gauge needles beginning with a no. 16 and ending with a no. 23. Suspensions of bone marrow cells were prepared by flushing the long bones with sterile α-MEM. Nucleated cell counts were performed electronically (Coulter Electronics, Hialeah, FL), and total volumes of the cell suspensions were documented.

Quantification of hematopoietic progenitors. Cells from fetal liver and bone marrow as well as cord blood light-density cells obtained by layering over Ficoll (Pharmacia Biotech Inc., Uppsala, Sweden), were cultured in quadruplicate 35-mm culture dishes in α-MEM containing 1.1% methylcellulose (Sigma Chemical Co., St. Louis, MO), 30% FCS (HyClone), 10% BSA (Sigma Chemical Co.), and 5 × 10-5 mol/Lβ-mercaptoethanol (Eastman Organic Chemicals, Rochester, NY)(27). Sets of plates for CFU-E assays contained 75× 103 light density cells/mL, and 1-2 U/mL rEpo (Amgen, Thousand Oaks, CA). Cultures to evaluate BFU-E colonies contained 5-10 × 103 cells/mL, 20 U/mL Epo, and 10 ng/mL stem cell factor (Amgen). G-CSF(Amgen) was added to the CFU-E and BFU-E cultures in concentrations of 0, 0.1, or 10 ng/mL, to observe for an effect on erythroid colony generation. The culture plates were incubated at 37 °C, 5% CO2, and high humidity. Plates were examined with the aid of an inverted microscope. CFU-E were identified as small, isolated clusters or pairs of clusters of distinctly red-orange erythroblasts at 6-8 d(28–30), whereas erythroid bursts (BFU-E) were recognized as red-orange bursts composed of three or more clusters after 12-14 d(28–30).

Quantification of normoblasts per erythroid progenitor. Afterin situ scoring, the culture dishes containing BFU-E colonies were washed extensively, as we have previously detailed, to recover the cells from the dishes(31). The number of cells per culture dish was determined with the aid of a hemocytometer, and 500-1000 differential cell counts were performed on Wright-stained cells. In this way, the absolute number of normoblasts, neutrophils, macrophages, and so forth, in each culture dish could be determined, and thus the average number of normoblasts per developing erythroid clone could be assessed.

Sample size and data analysis. No pilot data were available with which to calculate a power analysis to be used in the design of these experiments; thus, for each comparison, sample sizes were set at n≥ 20. Statistical comparisons were made by the Mann Whitney U test using MINITAB Statistical Software (MINITAB Inc., State College, PA; PC version, release I, 1991). Comparison p values <0.05 were considered significant; all others were reported as nonsignificant.

RESULTS

The addition of rG-CSF, in concentrations of 0.1 and 10.0 ng/mL, to cultures of fetal bone marrow cells and fetal liver cells had no effect on the number of CFU-E colonies subsequently generated (Table 1). Similarly, the addition of rG-CSF to fetal bone marrow cells, to fetal liver cells, and to light-density cells from the umbilical cord blood of preterm and term infants had no effect on the number of BFU-E colonies generated (Table 2). Addition of rG-CSF also had no effect on the number of normoblasts generated in the erythroid colonies from fetal bone marrow, fetal liver, or umbilical cord blood(Table 3).

Table 1 Effect of rG-CSF on clonal maturation of fetal CFU-E
Full size table
Table 2 Effect of rG-CSF on clonal maturation of fetal BFU-E
Full size table
Table 3 Effect of rG-CSF on the number of normoblasts which developed per erythroid colony
Full size table

Colonies of neutrophil origin were present [CFU-granulocyte/macrophage and CFU-Mix (multipotent progenitor)] in all of the clonogenic assays examined for BFU-E. Inasmuch as the focus of this study was not neutrophil development and the culture conditions were not set to optimize their growth, we have not reported any neutrophil colony numbers.

DISCUSSION

Competitive interaction of hematopoietic growth factors is minimal. For instance, the various hematopoietic growth factors do not compete with each other as ligands for binding to cell surface receptors; rather, each of the recognized factors interacts specifically with its own unique variety of receptor(3). In this way the acceleration of erythropoiesis, after stimulation with Epo, does not occur at the expense of granulocytopoiesis or thrombocytopoiesis. Under certain unique experimental circumstances, however, certain competitive interactions are observed. For instance, when the number of multipotent progenitors is limited, as in a culture plate that contains a limited population of nonrenewing cells, the addition of high concentrations of rEpo results in a reduction in neutrophil production in that plate(14, 15, 20, 31). Also, when fetal or neonatal animals, who have a small number of multipotent progenitors per g of body weight, are treated with extraordinarily high concentrations of rEpo, reduced neutrophil production can result(16). The transient hyporegenerative neutropenias observed in neonates who have very high endogenous concentrations of Epo might be the result of this mechanism. Examples include the hyporegenerative neutropenia observed in severely anemic fetuses with Rh hemolytic disease(32) and the hyporegenerative neutropenia observed in severely anemic donors in the fetal twin-twin transfusion syndrome(33).

In experiments where rG-CSF was administered in large doses to mice, a reduction in erythropoiesis was observed(14, 34, 35), but studies in humans have not shown this effect. In fact, several reports indicate that, when rG-CSF is administered to adult patients, erythropoiesis either remains unchanged or increases(21, 22). For instance, an increase in the blood concentration of BFU-E has been observed after rG-CSF administration to cancer patients, and an increase in BFU-E and an acceleration in erythropoiesis have been reported after rG-CSF administration to patients with human immunodeficiency virus infection. The mechanism explaining the increase in erythropoiesis after rG-CSF treatment of adult patients is not clear, but might be related to its capacity to increase cell cycling of multipotent progenitors, observed by Ikebuchi et al.(36).

Whether the administration of rG-CSF to human neonates will significantly affect (increase or decrease) erythropoiesis is not known, but such information is crucial to proceeding with clinical trials. To test this hypothesis in an in vitro system, we evaluated various concentrations of rG-CSF that included peak concentrations measured in the blood on neonates receiving rG-CSF treatment(24, 25, 37). Recombinant G-CSF was added to cultures of hematopoietic progenitor cells obtained from the three anatomic sites of hematopoiesis in the human fetus: the liver, bone marrow, and blood. In the fetus and neonate, mature erythroid progenitors, CFU-E, are predominantly found in the liver and bone marrow, and are relatively rare in blood(38). Therefore we examined the effect of rG-CSF on CFU-E from liver and marrow only (Table 1). Primitive erythroid progenitors (BFU-E) and multipotent hematopoietic progenitors are found in fetal (and umbilical cord) blood, as well as in the bone marrow(38). Therefore we examined the effect of rG-CSF on these progenitors using umbilical cord blood, fetal liver, and fetal bone marrow. We reasoned that, if rG-CSF caused a “down-modulation” of erythropoiesis from fetal hematopoietic progenitors, this could be observed either by a reduction in the number of erythroid colonies generated(Tables 1 and 2), or by a reduction in the number of normoblasts that developed from individual erythroid colonies(Table 3), or by both mechanisms simultaneously. In fact, we observed no effect of rG-CSF on either aspect of erythropoiesis.

Ultimately, clinical trials using rG-CSF will be needed to clearly determine its effect on erythroid progenitors in vivo. Anticipated difficulties that may occur in assessment of the effects of rG-CSF administration in vivo include the existence of anemia before rG-CSF administration, variations in hematocrit with gestational age, as well as the interaction of multiple growth factors, cytokines, and other regulatory cofactors that may be present in vivo. Any conclusions that are made on the effect of cytokines in vitro may need to be considered in the context of the complex issues that may occur in vivo.

Obviously, it is not known whether the administration of multiple doses of rG-CSF to neonates will be analogous to the exposure we tested in vitro. Nevertheless, hematopoietic progenitors in culture seem somewhat more susceptible to downmodulation than do progenitors in intact animals(16). Because we observed no evidence of reduced erythropoiesis in vitro when we added high concentration of rG-CSF to progenitors of human fetuses and neonates, it seems reasonable to predict that rG-CSF administration to neonates may not impede their erythropoiesis. However, this observation may be subject to a β-type error in which the relatively limited number of experiments performed may have precluded the ability to detect a difference between groups.