Main

Diminishing the number of transfusions administered to preterm neonates is widely maintained to be a desirable goal, as the risk of adverse reactions, or infection accompanying multiple transfusions, although small, is real. Multiple large clinical trials have shown that rEpo treatment can reduce both the number and total volume of transfusions required in this patient population(1–5). The cost of treatment with rEpo may be significantly less than the cost of transfusions, although this analysis is dependent on the costs of rEpo, and whether single or multiple doses are obtained per vial of drug(6–8). No significant adverse effects of rEpo administration to preterm infants have been identified, although an unexplained, possibly coincidental association with sudden infant death syndrome has been noted(9–11). The experience with rEpo in human infants to date is insufficient to guarantee the absence of significant toxicity, particularly for complications which occur rarely. Additional long-term studies are needed to assess the neurodevelopmental effects of Epo administration to this population.

Epo, a 34-kD glycoprotein, was originally thought to act only on erythroid progenitor cells, stimulating their proliferation and differentiation by binding to a specific 66-kD membrane receptor(12–17). There is now growing evidence in animal models that Epo-R are also present in some nonhematopoietic tissues such as endothelial cells(18, 19) and fetal cells of neural origin(20), although the physiologic role of Epo-R at these sites is unclear. These receptors do, however, appear to be functional as assessed by in vitro and in vivo studies. We will concentrate primarily on work done in tissues of neural origin, as there is accumulating evidence that Epo may be active within the developing CNS.

Masuda et al.(20) first demonstrated the presence of Epo-R on two rodent cell lines of neural origin (PC12 and SN6 cells) which responded to rEpo by increasing the intracellular concentration of monoamines, and by rapidly increasing cytosolic concentrations of free calcium. The possibility that this receptor is developmentally regulated was suggested by Liu et al.(21) in work observing substantial expression of Epo-R within the early fetal murine brain falling to undetectable levels by gestational d 16.

A prerequisite for assigning a physiologic role to Epo-R within the fetal CNS is the presence of the ligand, Epo, within this system, either by crossing the blood-brain barrier, or by synthesis within the CNS. Yasuda et al.(22) provided such support for a role of Epo in neurogenesis in a study of postimplantation mouse embryos, where both Epo and Epo-R messages were identified in the primitive streak. Our own work in human infants has documented the presence of Epo within the spinal fluid of premature and term neonates, which appears to be developmentally modulated(23).

Studies directed at determining the function of Epo within the CNS have demonstrated oxygen-regulated Epo synthesis by primary cultures of astrocytes from rat cerebrum(24). The Epo produced within the brain had a different extent of sialylation, and was more active in vitro than was rEpo. Tan et al.(25), studying in vivo exposure of rats to hypoxic conditions (7.5% oxygen) showed an induction of Epo mRNA in brain tissue. The possibility that Epo might function as a neurotropic factor is raised by a study showing that rEpo increased survival of damaged rat cholinergic septal neurons produced by fimbria-fornix transections(26, 27).

Thus, an endogenous source of Epo within the developing CNS and the presence of functional Epo-R have been established in murine models, although their role, if any, is not clear. Our laboratory has demonstrated the presence of Epo within the spinal fluid of neonates, and we now show that Epo-R are present in the CNS of developing humans. We speculate that if functional Epo-R are expressed within the CNS of the midtrimester fetus, as they are in developing rodents, treatment of premature infants with rEpo might have unanticipated neurodevelopmental consequences, and we echo the caveats voiced by others that, before treatment with rEpo becomes the standard of care for the treatment of anemia of prematurity, further neurodevelopmental studies to determine the physiologic relevance of this finding should be carried out.

METHODS

Human fetal specimens. Spinal columns were obtained from five human abortuses at 13 to 17 wk of gestation. Only fetuses that were normal by ultrasound examination and underwent elective pregnancy termination were studied. Pregnancy terminations were carried out by suction curettage. No contact with the mother or attempt to influence her decision was made. The studies were approved by the University of Florida Institutional Review Board. The spinal cords were obtained after severing the dorsal lamina from the spinal column, blotting the spinal cord several times on sterile gauze, and extensively washing the cords in sterile minimal essential medium, α modification (HyClone Laboratories, Logan, UT). The meninges were removed from some of the spinal cord samples under a dissecting microscope.

Preparation of total RNA. Total cellular RNA was extracted from the washed spinal cords using the guanidinium thiocyanate method(28). The spinal cord tissue and OCIM1 cells (an erythroleukemia cell line used as a positive control for Epo-R) were homogenized in 4 M guanidinium thiocyanate, and the RNA was separated from DNA and protein by phenol/chloroform extraction, followed by isopropanol precipitation. The purity and concentration of the RNA were determined spectrophotometrically.

RT of RNA and amplification of cDNA. RT of RNA and amplification of cDNA were performed by using the GeneAmp RNA PCR kit(Perkin-Elmer Corp., Norwalk, CT) in a DNA thermal cycler (Perkin-Elmer Corp.). One microgram of total RNA was incubated with 50 U of reverse transcriptase for 1 h at 37 °C. The reverse transcriptase was inactivated at the end of the reaction by heating to 95 °C for 5 min. Amplification of first strand cDNA was carried out under the following conditions: 10 mM Tris, pH 8.3, 50 mM KCl, 2.5 mM MgCl2, 2.0 mM dNTP, 0.2 μM upstream and downstream primers, 10% RT reaction mix, 5 U/μL Ampli Taq 94 °C for 1 min, 57 °C for 1 min, 72 °C for 2 min for 35 cycles, followed by 10-min elongation at 72 °C. The following oligonucleotide primers were used to amplify a 197-bp fragment from the reverse transcribed Epo-R mRNA: sense primer 5′-GCA-CCG-AGT-GTG-TGC-TGA-GCA-A, and antisense primer 5′-GGT-CAG-CAG-CAC-CAG-GAT-GAC. These primers amplified the region from nucleotide 632 to 829 of the Epo-R sequence(29). The primers used to identify α-globin mRNA amplified the region from nucleotide 37 to 236 of the α-globin sequence resulting in a 199-bp fragment(30) and the primers used to identifyβ-actin mRNA amplified the region between nucleotides 1038 and 1876 of the β-actin sequence resulting in a 661-bp fragment(31).

Specific Epo-R mRNA identification by digestion products and sequence analysis. To further confirm whether the 197-bp RT-PCR product was indeed Epo-R, the product was digested with the restriction endonuclease Ava II, according to the protocol of Anagnostou et al.(19). This digestion should produce two fragments, one of 57 and the other 140 bp. A positive control was obtained by digesting the RT-PCR product of the OCIM1 cell line with AvaII. Specificity of the RT-PCR product was additionally confirmed by direct sequencing of an amplified PCR product using the Taq DyeDeoxy Terminator protocol developed by Applied Biosystems (Perkin-Elmer Corp., Foster City, CA). The labeled extension products were analyzed on an Applied Biosystems Model 373 DNA sequencer.

Immunofluorescent staining. Representative sections from mid-trimester human spinal cords were fixed overnight at 4 °C in 94% absolute ethanol/3% glacial acetic acid/3% distilled water. The tissue was processed manually and embedded in paraffin. Sections of 7 μm were cut, mounted on drops of water, and heated to 37 °C for 1 h. They were then deparaffinized in Clear-Rite-3 (Richard Allen, Richland, MI) and rehydrated in graded ethanol solutions to PBS. Nonspecific staining was blocked using 1% normal goat serum. Monoclonal anti-human Epo-R antibody (mh2er/16.5.1, mouse IgGl, affinity purified on protein A, Genetics Institute, Cambridge, MA) was added at a dilution of 1:50 and incubated for 3 h. After rinsing thoroughly in PBS, secondary antibody (goat anti-mouse IgG conjugated to fluorescein (Sigma Chemical Co., St. Louis, MO) was applied at a 1:50 dilution and incubated for 1 h. Sections were mounted in glycerol with anti-quenching agent and viewed under epifluorescence with a Nikon FXA photomicroscope. In the negative control, nonimmune mouse serum was substituted for the primary anti-Epo-R antibody.

RESULTS

The upper panel of Figure 1 shows the RT-PCR products obtained from the five mid-trimester fetal spinal cords. The single, anticipated, 197-bp band was amplified from RNA isolated from all five spinal cords. To determine whether circulating erythroid cells, which might be contaminating the spinal cord preparations were responsible for the Epo-R RNA, RT-PCR was performed with primers specific to human α-globin(Fig. 1, middle panel). Only one spinal cord sample and OCIM1 cells showed the presence of α-globin mRNA. RT-PCR was also performed using primers for β-actin, and in each case a single band of appropriate size was obtained (Fig. 1, lower panel).

Figure 1
figure 1

Lanes 1 and 8 contain 100-bp DNA markers. Lanes 2 through 6 contain RT-PCR products from 15-, 13-, 14-, 15-, and 16-wk gestation fetal spinal cords, respectively. Lane 7 contains the RT-PCR product from OCIM1 cells, as a positive control. The upper panel shows the RT-PCR products obtained using primers specific to human Epo-R (197-bp product expected). The middle panel shows the RT-PCR products resulting from primers specific toα-globin (199-bp product expected), and the lower panel shows the RT-PCR products resulting from primers specific to β-actin (661-bp product expected).

Full size image

To further confirm that the 197-bp product, resulting from RT-PCR with primers specific to Epo-R, was indeed Epo-R mRNA, we digested the product with the restriction endonuclease AvaII, which resulted in the expected restriction fragments of 57 and 140 bp(29) (Fig. 2). As the cytokine family to which Epo-R belongs contains a number of sequence homologies, we obtained direct nucleotide sequence information from an amplified gene product. The results confirmed amplification between nucleotides 632 and 829 of the published Epo-R sequence(29).

Figure 2
figure 2

Specific digestion of the 197-bp Epo-R RT-PCR product with the restriction endonuclease AvaII, showing restriction fragments of 57 and 140 bp. Lanes 1 and 6 contain 100-bp DNA markers. Lane 2 contains the undigested RT-PCR product from human fetal spinal cord, lane 3 contains the RT-PCR product digested with AvaII, yielding the two smaller bands of 140 and 57 bp. Lanes 4 and 5 contain the uncut and cut products, respectively, from the positive control OCIM1 cells.

Full size image

Figure 3 shows the Epo-R immunostain of the fetal spinal cords. Ependymal cells enclosing the central canal express Epo-R, as demonstrated in panel A, by the red staining of these cells. The regions proximal to the anterior and posterior median sulci also are intensely immunopositive. Panel B shows the negative control.

Figure 3
figure 3

Immunofluorescent cytochemical stains of a spinal cord from a 17-wk gestation fetus showing (panel A) the cells to which anti-Epo-R antibody is bound. Panel B is a negative control.

Full size image

DISCUSSION

Approximately 8% of live births in the United States are delivered prematurely(32). These infants are at significant risk for receiving erythrocyte transfusions during their hospital stay, due initially to iatrogenic blood loss, and later to the anemia of prematurity(4, 33). Diminishing the number of transfusions to which premature infants are exposed is considered to be a worthwhile goal(33). Toward that aim, multiple, randomized, placebo-controlled, clinical trials have been conducted, and they indicate that rEpo treatment can decrease the number of transfusion needed(1–3, 5, 34). In fact, this use of rEpo has been viewed as physiologic, because preterm infants appear to have a very limited capacity to increase their serum concentrations of Epo during anemia or as a compensation for blood loss(3, 34).

The action of Epo occurs by way of its binding to specific cell-surface receptors. The Epo-R is a member of the super-family of hematopoietic cytokine receptors, which includes receptors for: IL-2, IL-3, IL-4, IL-5, IL-6, IL-7, granulocyte/macrophage and granulocyte colony-stimulating factors, prolactin, and growth hormone. The extracellular domains of these molecules share homology in the region required for receptor binding and activation. The binding of Epo to its receptor is followed by receptor-mediated endocytosis of Epo, and involves tyrosine phosphorylation(35).

The conventional understanding has been that the action of Epo was restricted to the proliferation and differentiation of erythroid progenitor cells, because these were the cells known to express Epo-R(12–17). Contrary to this belief, functional Epo-R have now been demonstrated on several nonerythroid tissues. Such receptors in endothelial cells obtained from multiple sources(human umbilical vein, bovine pulmonary artery, and bovine adrenal capillary) not only show mitogenic activity when stimulated by rEpo, but also stimulate the release of angiotensin-1(36). Of particular concern are the reports of developmentally regulated, functional Epo-R on cells of neural origin(21). Indeed, previous and present studies suggest that the effects of Epo in the fetus might not be limited to erythropoiesis, as both the receptor and its ligand are present within the developing CNS during fetal development. The existence of alternatively spliced isoforms of the Epo-R is controversial, and further study will be required to know whether the receptors we identified represent alternatively spliced isoforms, or whether they are the same as the receptors found on erythroid cell lines. Although the physiologic significance of this ligand-receptor pair in the human fetal CNS is unclear, we maintain that further studies are indicated to delineate potential neurodevelopmental effects of rEpo administration for anemia of prematurity. For example, it would be important to know whether rEpo, used at doses appropriate to treat anemia of prematurity, crosses the blood brain barrier in premature human infants. Localization of Epo-R within the human brain, brainstem, and spinal cord would be interesting, and might help to direct further neurodevelopmental studies. For example, if there was a preponderance of such receptors in the reticular activating system, one might pay particular attention to the effects of rEpo on symptoms of apnea. Such a finding might also shed light on the association (real or coincidental) of rEpo treatment with sudden infant death syndrome. It is also possible that rEpo might have beneficial neurodevelopmental effects, however, all these potential effects, good or bad, are at this point speculation, and a combination of further laboratory elucidation of the possible function of Epo within the CNS and long-term clinical studies following treated infants would answer these questions.