Main

Infants born at the limits of viability are susceptible to specific morbidities. One such morbidity involves the abnormal growth and maturation of the vascular retina, termed retinopathy of prematurity (ROP). Early studies in preterm infants evaluating preventive strategies for ROP involved limitations in light exposure and the addition of a variety of nutrients to the preterm infant's diet; however, no strategy other than prevention of premature birth has been shown to be effective in preventing ROP (14). The severity of the disease can be attenuated by preventing hyperoxia in the first weeks of life, as well as preventing episodes of hyperoxia and hypoxia (57). Despite significant focus on decreasing the incidence of ROP in extremely low birth weight infants, the disease remains prevalent (14).

Watanabe and colleagues reported elevations in vitreal erythropoietin (Epo) concentrations in patients with proliferative diabetic retinopathy (8), a disease with possibly similar mechanisms of vascular injury and neovascularization as seen in ROP. Those authors suggested that Epo is a potent ischemia-induced retinal angiogenic factor. In that study, blockage of Epo in a mouse model of hypoxia-induced retinopathy was shown to inhibit retinal neovascularization and endothelial cell proliferation in vivo.

The role of Epo in the developing eye and its involvement in the development of retinal vascular disease in preterm infants has not been examined. If Epo concentrations are negligible in the developing eye, then Epo administration after preterm delivery might abnormally influence growth after delivery. We hypothesized that Epo mRNA and protein are present in the mid-gestation human fetal eye, and that Epo mRNA and protein concentrations increase with increasing gestation. Our objectives were to quantify Epo mRNA and protein in the serum and vitreous during fetal development between 12 and 24 wk gestation. In addition to Epo mRNA concentrations, we sought to quantify hypoxia inducible factor 1α (HIF) mRNA concentrations to determine whether, similar to other fetal tissues, retinal concentrations remain relatively unchanged throughout the gestations tested and could be used to normalize target mRNA data. In fetal liver and kidney, HIF can be used as an endogenous control. HIF activity is posttranscriptionally controlled through an oxygen dependent ubiquitin proteosome degradation pathway (9,10); therefore, concentrations remain stable and do not fluctuate under hypoxic conditions (11).

METHODS

A total of 52 samples were processed for measurement of Epo mRNA and protein. Not all samples had obtainable protein or tissue. Gender could not be determined on most samples (12) and that information was not collected. Samples were collected between 12 wk and 24 wk gestation, for a total of 36 matched vitreous and serum samples (8 samples from 12 to 14 wk, 10 samples from 15 to 18 wk, 10 samples from 19 to 21 wk, and 8 samples from 22 to 24 wk). The aqueous and vitreous were collected from both fetal eyes and combined for each sample. Serum was obtained from all samples when available by collecting blood still present in the umbilical cord arteries and vein. Serum and vitreous were stored at −20°C until analyzed by enzyme-linked immunosorbent serologic assay (R&D Systems, Minneapolis,1 MN).

The retina was removed from both eyes, and total RNA was extracted from 45 samples using TriZol (Invitrogen, Carlsbad, CA). Total RNA was quantified by spectrophotometer. The RNA was reverse transcribed using the cDNA Archive kit according to manufacture's instruction (Applied Biosystems Inc. [ABI] CA). Primers and probe specific for Epo (ABI) were used to determine mRNA concentrations in all RNA samples using real time PCR (PRISM 7500 Fast Thermocycler, ABI). cDNA reverse-transcribed from total RNA was amplified over 40 cycles.

In addition to Epo mRNA concentrations, HIF mRNA concentrations were measured in 23 retinal samples. In previous studies, the use of common “housekeeping” RNA, such as β-actin or GAPDH, did not remain constant under hypoxic conditions (13) or over the gestations evaluated in the second trimester. We tested HIF to determine whether, similar to other fetal tissues (11), retinal concentrations remain relatively unchanged throughout the gestations tested and could be used to normalize target mRNA data.

For quantitative PCR measurements, a standard curve was created by reverse transcribing serial dilutions of total RNA isolated from Heb3B cells, an immortalized cell line that produces Epo mRNA and protein constitutively. Undiluted Hep3B total RNA was assigned a value of “100” for both Epo and HIF assays (100 ng total RNA per 20 μL RT reaction).

Differences in mRNA concentrations and in Epo concentrations at each gestational age group (plotted as mean plus or minus SE) were analyzed using an unpaired t test and ANOVA analysis. The study was deemed not to constitute human subject research by the Institutional Review Board at the University of New Mexico, as no identifiable patient data were collected.

RESULTS

Epo mRNA concentrations were measurable in all fetal retinal samples tested. Epo mRNA concentrations increased with increasing gestational age (Fig. 1; R = 0.65, p < 0.05). Similar to previous fetal tissues tested, HIF mRNA concentrations were unchanged over the range of gestational ages tested (Fig. 2; R = 0.23, p = NS). The quantity of HIF mRNA per total RNA measured in fetal retina was similar to that seen in RNA isolated from Hep3B cells (total RNA for all samples was 100 ng/20 μL reverse transcription reaction volume).

Figure 1
figure 1

Retinal Epo mRNA concentrations in the developing human eye. Epo mRNA concentrations (45 samples measured) increase with increasing gestational age (r = 0.65, p < 0.05). Values are presented as retinal Epo mRNA concentration/Hep3B Epo mRNA concentrations per total RNA loaded.

Figure 2
figure 2

Retinal HIF-1α mRNA concentrations in the developing human eye. HIF mRNA concentrations (23 samples measured) were similar over increasing gestational ages (r = 0.23, p = NS). Values are presented as retinal HIF mRNA concentration/Hep3B HIF mRNA concentrations per total RNA loaded.

Both serum and vitreous Epo concentrations increased with increasing gestation (Fig. 3; p < 0.05, 12–14 wk vs. 18–20 wk and 21–24 wk). Vitreous Epo concentrations ranged from 1 to 12.5 mU/mL at 12–14 wk (median 7.5 mU/mL), 5–16.6 mU/mL at 15–17 wk (10.1 mU/mL), 1–29 mU/mL at 18–20 wk (14.1 mU/mL), and 15.7–34.1 mU/mL at 21–24 wk (26.8 mU/mL). Vitreous Epo concentrations were greater than serum Epo concentrations at each gestational age group tested (p < 0.05, 12–14 wk and 21–24 wk; p < 0.01, 15–17 wk and 18–20 wk).

Figure 3
figure 3

Serum and vitreous Epo concentrations. A total of 36 samples were measured for serum and vitreous Epo concentrations: 8 samples from 12–14 wk, 10 samples from 15–18 wk, 10 samples from 19–21 wk, and 8 samples from 22–24 wk. Values shown are mean ± SEM. Both serum (clear bars) and vitreous (solid bars) Epo concentrations increased with increasing gestation (p < 0.05, 12–14 wk vs. 18–20 wk and 21–24 wk). Vitreous Epo concentrations were greater than serum Epo concentrations at each gestational age group tested (p < 0.05, 12–14 wk and 21–24 wk; p < 0.01, 15–17 wk and 18–20 wk).

DISCUSSION

The role of Epo in the developing eye has not been evaluated. Moreover, the role of Epo in the development of retinal vascular disease in preterm infants has not been determined. We found increasing concentrations of both Epo mRNA and protein with increasing gestation in the human fetal eye. Similar to other fetal tissues, we measured significant but unchanging HIF mRNA concentrations with increasing gestation. The finding of increasing Epo protein concentrations with increasing gestation, and importantly, Epo concentrations that exceeded those found in serum at each gestation tested suggests a role for Epo in human retinal development. In animal models, exposure to low levels of oxygen, deemed “hypoxic conditioning,” results in up-regulation of Epo and its receptor (14). The increase in Epo concentration is thought to play a role in protecting against retinal damage, primarily by inhibiting gene expression for enzymes involved in apoptosis. In addition, neuroprotection of retinal photoreceptors can be induced by hypoxic preconditioning or by increasing systemic Epo concentrations (15).

Watanabe and colleagues reported significant elevations in vascular endothelial growth factor (VEGF) and Epo in the vitreous of patients with proliferative diabetic retinopathy, although regression analysis showed that Epo was more strongly associated with active disease than VEGF (8). Those authors suggested that Epo is a potent ischemia-induced retinal angiogenic factor, because blockage of Epo is shown to inhibit retinal neovascularization and endothelial cell proliferation in vivo. Other authors have described Epo as an endogenous retinal survival factor (1517).

Research has demonstrated that in addition to hematopoietic properties, Epo plays a developmental role in angiogenesis (18,19). Epo receptors are found on endothelial cells, neuronal cells, and accessory cells throughout the developing fetus. In response to Epo administration, angiogenic events such as endothelial cell proliferation, organization, and chemotaxis are demonstrated in vitro and in vivo. Epo is also shown to promote neurogenesis and has neuroprotective properties as well. Epo is found in the cerebral spinal fluid of both animals and humans, both neonatal and adult (20). Epo receptor expression has been demonstrated in most cerebral cell types, including endothelial cells, neurons, and astrocytes (2022). Epo and Epo receptor expression is elevated during early development, but decreases up to 100-fold after birth (22,23). Epo receptor deficient mice have an increased rate of cerebral apoptosis, as well as areas of cerebral hypoplasia (24).

Similar to adult renal and fetal hepatic cells, neuronal cells have increased Epo gene expression in response to hypoxia (25). This pathway is mediated primarily by hypoxically regulated vascular growth factors (26). A variety of insults, including cerebral ischemia, seizures, and head injury result in increased growth factors. In response, the Epo/Epo receptor system is up-regulated, ultimately promoting cell viability (2527). This neuroprotection is a result of repressing apoptosis and blocking nitric oxide-meditated toxicity (28,29). Animal studies have demonstrated dose-dependent effects, including reduced infarct volume, increased blood flow, and improvement in long-term outcome/relearning (14,30,31).

Although numerous randomized placebo-controlled studies have not reported significant differences in the incidence of ROP between Epo treated and placebo infants, retrospective reports have postulated that use of Epo may be associated with increased rates of ROP (3235). Brown and colleagues reported an association between cumulative doses of Epo and risk for ROP (33); however, the authors were unable to separate the relationship among extreme prematurity, Epo doses, and ROP. A recent Cochrane review (36) speculated about a relationship between Epo and ROP, primarily due to the inclusion of a study, summarized in a letter to the editor, of early Epo and iron administered to preterm infants (32). Moreover, although numerous studies have reported an increased risk of ROP in relation to blood transfusions (37,38), no studies have reported Epo administration as a factor increasing the risk of ROP (24).

The relationship between exogenous Epo and ROP remains speculative, but should be monitored closely in future Epo studies. In addition, the relationship between endogenous Epo and ROP remains to be determined. It is possible that elevated vitreal Epo concentrations in utero, followed by decreased Epo concentrations after extremely preterm delivery create an abnormal cycle of vascular stimulus and suppression that eventually leads to ROP. We speculate that treatment of preterm infants with exogenous Epo may influence vitreal Epo concentrations; however, the timing of Epo treatment remains to be determined. We conclude that Epo is involved in human retinal vascular development and continue to evaluate the role of Epo in the developing human eye.