Mammalian milk contains the hematopoietic growth factor erythropoietin (Epo), the primary hormone regulating erythropoiesis. For some time, indirect evidence has supported the presence of milkborne Epo (1–4). More recently, Epo has been assayed in human and rat milk (5–9). Many milkborne growth factors are protected from proteolytic degradation and exert biological functions in the neonate (10). Simulated in vitro digestion experiments showed that significant amounts of milkborne Epo resisted degradation (5,7,9).
The animal literature is conflicted as to whether erythropoiesis is stimulated after feeding Epo. Although it amounts to indirect evidence, rats suckling anemic dams had higher hematocrit and hemoglobin (Hb) levels and reticulocytosis compared with control (11). In contrast, injecting Epo into dams resulted in similar plasma Epo levels and hematocrit in the Epo offspring compared with the control offspring (8). Additionally, Epo in artificial formula did not increase plasma Epo levels or hematocrit percentages compared with controls (8). Although no erythropoietic response was seen in suckling rats, enteral Epo stimulated local intestinal growth (8), leading the authors to conclude that insufficient Epo remained intact to stimulate a systemic erythropoietic response.
Human studies also conflict. One small controlled trial fed Epo to premature infants and found higher plasma Epo levels, higher peak reticulocyte counts, and lower plasma ferritin levels than controls (12), whereas a slightly larger study found plasma Epo levels, reticulocytes, and hematocrit percentages similar to controls (13).
Because iron supplementation is deemed necessary with parenteral Epo administration in prematurity (14), we investigated whether iron plays a role in the erythropoietic efficacy of enteral Epo in suckling rats. The potential advantage of treating the anemia of prematurity with enteral Epo would be that the enteral route is less invasive compared with parenteral treatment. However, another potentially more significant benefit under investigation is the nonerythropoietic local effects of enteral Epo (7,8,13,15). Because of the recent work examining local intestinal effects, delineation of all potential biological effects of enteral Epo is scientifically and clinically meritorious. We hypothesized that enteral Epo stimulates erythropoiesis in suckling rats if accompanied by sufficient iron supplementation. A secondary hypothesis was that milkborne Epo accompanied by enteral iron also would stimulate small intestinal growth.
MATERIALS AND METHODS
Recombinant human Epo (Amgen, Thousand Oaks, CA) was purchased and diluted to appropriate dosing concentrations in 2.5% bovine serum albumin in phosphate buffered saline (PBS). Ferrous sulfate was administered as Fer-In-Sol (Mead Johnson, Evansville, IN) and diluted with 0.1% bovine serum albumin in PBS.
Institutional Animal Care and Use Committees at both the University of Arizona and the University of Wisconsin, Madison approved these studies. CD IGS Sprague-Dawley rats were purchased from Charles River Laboratories (Wilmington, MA). After breeding and delivery of pups, litter size was reduced to 10 pups per litter on postnatal day 1 to standardize litter size and growth. Each dam was bred no more than 3 times.
Postnatal day 4 rats were removed from their dams, and weight and sex determined. Rats were fasted in a cage overlying a 33°C heating pad. Before surgery, urination was provoked by gentle stimulation of the anogenital area. Rats artificially fed the cow milk–based rat milk substitute (RMS) were anesthetized with halothane or isoflurane, and a percutaneous gastrostomy tube placed (16). Briefly, a 16-gauge Cathlon angiocatheter (Johnson & Johnson, Arlington, TX) was inserted through the greater curvature of the stomach. PE-20 tubing with flared end was then inserted through the sleeve until lodged in the stomach. A plastic spacer was secured against the abdominal wall to prevent catheter dislodgement. A loop of exposed tubing was fixed under nuchal skin for further security. The rats were then placed in pint-size polypropylene cups with crushed corncob bedding and floated in a water bath (Precision model 270, Chicago, IL) kept at 40°C. The PE-20 tubing was attached to PE-150 tubing leading to a refrigerated syringe delivering Pedialyte (Ross Laboratories, Columbus, OH) at 150 μL · g−1 · day−1 to 200 μL · g−1 · day−1 for 12 hours for recovery. RMS was then infused. Along with multivitamins and minerals, RMS contained 24 mg/L ferrous sulfate/7 H2O, and 4.8 mg/L elemental iron (Sigma, St Louis, MO). Rats received 1.8 mg · kg−1 · day−1 of elemental iron in RMS, based on mean group weight each day. For all RMS experiments, formula was infused at 370 μL · g−1 · day−1, based on mean daily group weight (16). Volume of artificial feeding was selected to match weight gain of dam-fed litters with 10 rats (16). Twice daily, rats were examined and the perianal region stroked to promote micturition and defecation.
The Epo dose was based on the observation that subcutaneous human Epo at 425 U · kg−1 · day−1 stimulated erythropoiesis in suckling rats (17). Although equal amounts of Epo were recovered from marrow after equivalent bolus dosing of enteral in rat milk or subcutaneous Epo (9), Epo is less protected from proteolytic degradation by artificial formula (30% intact) than rat milk (60% intact) (7). Animals were randomized to 6 artificial feeding groups: control, control+Fe, Epo 425 U · kg−1 · day−1, Epo 425+Fe, Epo 1700 U · kg−1 · day−1, and Epo 1700+Fe, with the 1700 dose to account for potential degradation in formula. For each experiment, no more than 20 rats could be studied simultaneously, requiring individual rats from 2 litters to be randomized to 2 or 3 different artificial-feeding groups. A control or control + Fe group always was included, depending on whether the treatment included iron. Epo 425 or Epo 1700 was infused in RMS. Because oral ferrous sulfate 6 mg · kg−1 · day−1 sustained erythropoiesis and normalized iron indices in dam-fed rats injected with Epo 425 (17), rats were randomized to receive enteral ferrous sulfate 6 mg · kg−1 · day−1 or placebo in a daily bolus.
Blood, Marrow, and Tissue Collection
At postnatal day 12, all of the animals were given lethal anesthesia with inhaled halothane or isoflurane. Cardiac puncture was performed and whole blood was placed into EDTA Microtainers (Becton Dickinson, Franklin Lakes, NJ). Complete blood counts (Hb, red blood cell, and erythroid indices) were determined with an MD16 counter (Coulter Diagnostics, Hialeah, FL). The erythroid index, or mean cell volume, falls with iron-deficient microcytosis, and red cell distribution width (RDW) rises with increased variability of red blood cell size (18). Reticulocyte percent was determined microscopically by a blinded reviewer, after staining with brilliant cresyl blue (Sigma). Absolute reticulocyte count was determined as the product of red blood cell count and the percentage of reticulocytes.
Plasma was obtained by centrifuging whole blood at 3000 rpm for 6 minutes and stored at −70°C for later evaluation. Plasma Epo levels were determined via double antibody radioimmunoassay (Diagnostic Systems Laboratory, Webster, TX), with a minimal detection level of human Epo of 1.0 U/L, using a human standard curve. Erythrocyte zinc protoporphyrin/heme (ZnPP/H) was obtained by a hematofluorometer, (Aviv Biomedical, Lakewood, NJ) after rinsing to remove interfering pigments (19). ZnPP/H rises during iron-deficient erythropoiesis and is more sensitive compared with standard markers of iron stores in humans and rats (17,20).
Fibular bone marrow samples were flushed with PBS containing 5% albumin and 60 U/mL heparin using 23-gauge needles. Marrow was rinsed, triturated, resuspended in PBS, and centrifuged on ethanol-washed slides (Cytospin 3 Centrifuge, Shandon, Runcorn, UK), then dried, stained with Dif-Quik (Dade Behring, Newark, DE), and counted for myeloid:erythroid (M:E) ratios that normally fall with stimulation of erythropoiesis.
To confirm the previously reported local trophic intestinal effects of Epo (8), duodenum was harvested from the control+Fe and Epo 1700+Fe groups. After the rats were killed, a midline incision was made and the duodenum was harvested, cut open, and placed on a 0.45-μm Metricel membrane filter (Pall Life Sciences, Ann Arbor, MI) before fixing in formalin. Paraffin-blocked 10-μm sections were placed on slides and stained with hematoxylin and eosin. Slides were evaluated for villous length. Five digitized images of duodenum at ×10 were evaluated from each animal using Metamorph software (Molecular Devices, Sunnyvale, CA). An investigator blinded to treatment measured villous height for each digitized image. Images were digitized with a minimum of 4 measurable intact villi in the field. The tallest villi were measured to minimize underestimation due to the angle and position of section planes (8,21).
Sample size was estimated at 12/group, with 80% power, α level <0.05, anticipating a 12-g/L rise in Hb concentration. Data were analyzed by 1-way analysis of variance for multiple comparisons, with Fisher t test, Kruskal-Wallis test for nonparametric data, or unpaired t tests. Data are expressed as mean ± standard deviation (SD). ZnPP/H and plasma Epo underwent log conversion for comparison purposes. ZnPP/H is shown in log10 scale. Population normal ranges for dam-fed rats, mean ± 2 SD (17), are shown for comparison on Fig. 1.
Rat weight from control (26.2 ± 4.0 g), control+Fe (27.5 ± 4.0 g), Epo 425 (27.4 ± 3.3 g), Epo 425+Fe (28.6 ± 3.5 g), Epo 1700 (27.5 ± 3.2 g), and Epo 1700+Fe (26.1 ± 4.1 g) were similar to each other and to dam-fed population normal values (27.1 ± 1.0 g). Figure 1A shows that Hb did not rise with Epo 425, but Hb levels in control +Fe, Epo 425+Fe, and Epo 1700 were higher than control, P < 0.001. Mean Hb level was highest in Epo 1700+Fe, P < 0.003. Figure 1B shows that reticulocyte numbers were higher in Epo 1700 and Epo 1700+Fe than other groups (P < 0.04). Figure 1C shows that mean cell volume was higher in the groups treated with iron, P < 0.01, but was highest in Epo 1700+Fe, P < 0.0002. Figure 1D shows that ZnPP/H fell in response to iron treatment, P < 0.02, and was intermediate in Epo 1700+Fe groups, P < 0.002.
Other erythropoietic indicators were examined. Compared with control (0.73 ± 0.19), marrow M:E ratios were lower in all groups, P < 0.01. M:E ratios were similar in control+Fe (0.58 ± 0.15), Epo 425 (0.54 ± 0.15), Epo 425+Fe (0.55 ± 0.15), Epo 1700 (0.55 ± 0.04), and Epo 1700+Fe (0.57 ± 0.07). Plasma Epo levels (antihuman Epo) were similar to the levels reported in Juul et al (8). Plasma Epo levels in control (7.1 ± 2.8 U/L) were similar to all groups, including Epo 1700 (11.6 ± 13.4 U/L), but trended lower in the Epo 1700+Fe group (1.9 ± 1.7 U/L), P = 0.11. Several plasma Epo levels in the Epo 1700+Fe group were below detection limits of the assay.
We compared RDW to further interrogate iron status, and found that RDW was higher in control (21.2 ± 2.5%) than in the other groups, P < 0.001. RDW in the Epo 425 group was intermediate (19.0 ± 2.3%), P < 0.001, with control+Fe (16.9 ± 1.7%), Epo 425+Fe (17.0 ± 1.9%), Epo 1700 (17.5 ± 2.4), and Epo 1700+Fe (18.3 ± 3.0%) groups similar to each other.
Because Juul et al (8) observed trophic effects of Epo on local intestinal tissue, we compared the weight of small intestine in Epo 1700+Fe to control+Fe. Small intestine trended heavier in Epo 1700+Fe (53.8 ± 6.4 mg/g/rat) compared with control+Fe (48.6 ± 8.1 mg/g/rat), P = 0.07. Microscopic duodenal villous height (425 ± 50 μm) was greater in the Epo 1700+Fe treatment group (n = 7) compared with control+Fe (352 ± 34 μm; n = 7), P < 0.01, Figure 2.
This study is the first to show that the combination of enteral Epo and ferrous sulfate stimulated a systemic erythropoietic effect in newborn rats. The literature supports the bioactivity of milkborne growth factors because neonates exhibit higher gastric pH, immaturity of gastrointestinal enzymes, and protective milkborne protease inhibitors (10,22,23). In suckling rats or humans, mother's milk afforded the greatest protection (60% intact), with less protection from formula (30% intact) or saline (10% intact) (5,7,9). We previously showed that enteral Epo remained intact in both local and distal tissues in newborn rats (9). Two hours after Epo bolus feed, approximately 10% of Epo remained intact in the intestine, 5% remained intact in plasma, and nearly 3% remained intact in marrow, similar to percentages seen with subcutaneous administration of the same dose (9).
Inferential evidence in the literature supports that milkborne Epo exerts erythropoietic effects in offspring of rat dams made hypoxic or anemic, compared with offspring of control dams (1,11,24). However, Juul et al (8) injected lactating dams with Epo in an effort to increase milkborne Epo levels, but found similar hematocrit percentages in Epo and control offspring. Because milk from anemic or hypoxic dams may contain higher levels of other erythropoietic growth factors known to independently increase Hb levels in artificially fed rats, such as insulin-like growth factor I (25), lack of other factors may explain the failure of Hb to rise in the Juul study.
Enterally fed Epo in adult animals had no effect (26). The actual daily dose of Epo in rat milk is unclear, but human studies support an average physiological intake of 40 U · kg−1 · day−1 (5). As a comparison with normal human physiological intake, when pharmacological-dose Epo was gavage-fed for 4 days to rats otherwise suckling their dam, investigators observed increased reticulocyte percentage, but no change in hematocrit (4). In the recent study by Juul et al (8), artificially reared pups were fed Epo 1000 U · kg−1 · day−1 for 2 weeks. Juul observed significant local trophic intestinal effects, but no rise in Epo level nor hematocrit, suggesting that intact and functional Epo did not reach distal erythropoietic tissue (8). Because we previously found intact enteral Epo in the marrow of suckling rats (9), our alternative explanation could be that insufficient marrow erythroid iron delivery limited the erythropoietic response in these previous studies. Because the present study combined Epo with iron supplementation and observed both local intestinal effects and distal erythropoietic effects in the iron-treated groups (elevated Hb, reticulocytes, and decreased marrow M:E ratio), this supports that iron plays a critical role in the erythropoietic response. Higher mean cell volume, lower ZnPP/H, and lower RDW values also support that increased erythrocyte iron delivery improved the erythropoietic response.
We observed that Epo and iron feeding decreased marrow M:E ratios, but circulating reticulocytes rose only with Epo 1700 ± iron. The apparent disruption of immature erythroid numbers from the marrow reaching the circulation could be explained by the finding that Epo and iron both stimulate erythropoiesis, but that Epo also serves as a marrow-releasing factor, through a process previously described but incompletely understood (27,28).
Consistent with the published rat data (8), data from 1 uncontrolled human (29), and 2 controlled human studies (12,13), we observed low systemic plasma Epo levels after pharmacological enteral Epo intake, compared with plasma levels seen after parenteral therapy. Potential explanations of this finding are as follows: First, given the relatively lesser protection in formula, compared with rat milk, the Epo 1700 approximates the subcutaneous 425 “intact” dosing previously shown to raise Hb (7). Second, because Epo was ingested in a near-continuous fashion, a minimal peak plasma level could be anticipated. Epo exerts effects in a paracrine—as well as an endocrine—fashion, such that plasma Epo levels may not correlate to erythropoietic response after enteral feeding. Therapeutic plasma Epo levels are the sum of the intact enteral dose (one third of the administered dose in formula or two thirds of the dose in rat or human milk), receptor utilization, and a small amount that is cleared in urine. Epo elimination largely is due to pharmacological transduction and degradation through the receptor (30). Plasma Epo levels trended highest in the Epo 1700 group and lowest in the high Hb group (Epo 1700+Fe), supporting that erythroid receptor–mediated degradation is a major determinant of plasma Epo levels. Pharmacokinetic data after enteral administration are unknown, but glycosylation sites are responsible for maintaining plasma Epo levels (31). However, as is seen with other growth factors (10), we speculate that intestinal juices cleave a portion of Epo glycosylation sites as Epo moves through the gastrointestinal tract. Removal of sialic acid of Epo does not impede erythropoietic activity, but partially sialated Epo is cleared more rapidly from the plasma, resulting in higher levels in erythropoietic tissue than fully glycosylated Epo (31). In support of this speculation, plasma Epo levels were relatively low, compared with parenteral dosing (8). Although the Epo 1700+Fe daily intake was higher than the 1000 U · kg−1 · day−1 intake in Juul et al (8), the plasma Epo levels in both studies were comparable. No matter the cause(s) of low circulating Epo levels, we observed a stimulation of erythropoiesis when iron accompanied the Epo 1700 dose.
In the studies feeding Epo and iron to premature humans, one found increased reticulocytes and decreased plasma ferritin levels (12), but a second observed similar reticulocytes and hematocrit percentages (13). It is possible that enteral Epo dosing in humans could stimulate erythropoiesis in the anemia of prematurity, given that initial studies of parental Epo in premature infants also were underwhelming because of unclear dosing and confounding influences on study endpoints, such as transfusion number or hematocrit (14).
Consistent with previous work, the addition of enteral iron did not alter the local nonerythropoietic effects of Epo on rat intestine. Our data support previous in vitro and in vivo data showing a key role for Epo in stimulating growth and healing of gastrointestinal epithelia (8,30,32).
Our study is limited by the possibility that erythrocyte iron delivery is relatively more important in extremely rapidly growing rats, compared with humans. In human infants, iron supplementation alone generally does not increase Hb levels as in this study. Because rats responded to ferrous sulfate, the iron content and/or iron bioavailability in artificial formula may be inadequate, also a possibility in the work of Juul (13). Although findings may not directly translate to humans, the possibility of increasing enteral Epo dose and/or optimizing iron may be applicable.
In conclusion, we confirmed that enteral iron did not alter previous findings that Epo fed to suckling rats exerted local trophic effects and that a poor erythropoietic response is observed without sufficient iron. However, we are the first to observe increased Hb levels with sufficient daily enteral Epo intake and enteral ferrous sulfate supplementation. Based on these findings, work should continue to examine the impact of iron supplementation on both the erythropoietic and nonerythropoietic effects of enteral Epo.
The authors acknowledge the technical support of Michael W. Rausch, JD, Robert Johnson, PhD, Suann S. Woodward, Aisha K. David, Lisa Ngo, and Kristin Repyak.
1. Grant W. Influence of anoxia of a lactating rat on the blood of normal baby rats [abstract]. Am J Physiol 1952; 171:728–729.
2. Hyzy J. Search for erythropoietin in the milk of sheep. Acta Physiol Polonica 1969; 20:11–22.
3. Bielecki M, Przala F, Lazewska M. Level of erythropoietin in the woman milk. Acta Physiol Polonica 1972; 23:435–439.
4. Carmichael RD, Gordon AS, LoBue J. The effects of maternal phlebotomy and orally-administered erythropoietin (Ep) on erythropoiesis in the suckling rat. Biol Neonate 1978; 33:119–131.
5. Kling PJ, Sullivan TM, Roberts RA, et al
. Human milk as a potential enteral source of erythropoietin. Pediatr Res 1998; 43:216–221.
6. Krafft A, Breymann C, Huttner R, et al
. Erythropoietic quality of maternal milk. Lancet 1999; 354:778.
7. Juul SE, Zhao Y, Dame JB, et al
. Origin and fate of erythropoietin in human milk. Pediatr Res 2000; 48:660–667.
8. Juul SE, Ledbetter D, Joyce AE, et al
. Erythropoietin acts as a trophic factor in neonatal rat intestine. Gut 2001; 49:182–189.
9. Miller-Gilbert AL, Dubuque SH, Dvorak B, et al
. Enteral absorption of erythropoietin in the suckling rat. Pediatr Res 2001; 50:261–267.
10. Hamosh M. Bioactive factors in human milk. Pediatr Clin North Am 2001; 48:69–86.
11. Carmichael RD, LoBue J, Gordon AS. Neonatal erythropoiesis. I. Peripheral blood erythropoietic parameters: data suggest erythropoietin transfer via maternal milk. Endocr Regul 1992; 26:83–88.
12. Ballin A, Bilker-Reich A, Arbel E, et al
. Erythropoietin, given enterally, stimulates erythropoiesis in premature infants. Lancet 1999; 353:1849.
13. Juul SE. Enterally dosed recombinant human erythropoietin does not stimulate erythropoiesis in neonates. J Pediatr 2003; 143:321–326.
14. Ohlsson A, Aher S. Early erythropoietin for preventing red blood cell transfusion in preterm and/or low birth weight infants. Cochrane Database Syst Rev 2006; (3):CD004863.
15. Barney CK, Lambert DK, Alder SC, et al
. Treating feeding intolerance with an enteral solution patterned after human amniotic fluid: a randomized, controlled, masked trial. J Perinatol 2007; 27:28–31.
16. Dvorak B, McWilliam DL, Williams CS, et al
. Artificial formula induces precocious maturation of the small intestine of artificially reared suckling rats. J Pediatr Gastroenterol Nutr 2000; 31:162–169.
17. Dubuque SH, Dvorak B, Woodward SS, et al
. Iron deficient erythropoiesis in neonatal rats. Biol Neonate 2002; 81:51–57.
18. Monzon CM, Beaver BD, Dillon TD. Evaluation of erythrocyte disorders with mean corpuscular volume (MCV) and red cell distribution width (RDW). Clinical Pediatrics 1987; 26:632–638.
19. Hastka J, Lasserre JJ, Schwarzbeck A, et al
. Washing erythrocytes to remove interferents in measurements of zinc protoporphyrin by front-face hematofluorometry. Clin Chem 1992; 38:2184–2189.
20. Rettmer RL, Carlson TH, Origenes ML, et al
. Zinc protoporphyrin/heme ratio for diagnosis of preanemic iron deficiency. Pediatrics 1999; 104:e37.
21. Mayhew TM, Carson FL. Mechanisms of adaption in rat small intestine: regional differences in quantitative morphology during normal growth and experimental hypertrophy. J Anat 1989; 164:189–200.
22. Koldovsky O. The potential physiological significance of milk-borne hormonally active substances for the neonate. J Mamm Gland Biol Neoplasia 1996; 1:317–322.
23. Britton JR, Koldovsky O. Development of luminal protein digestion: implications for biologically active dietary polypeptides. J Pediatr Gastroenterol Nutr 1989; 9:144–162.
24. Carmichael RD, Gordon AS, LoBue J. Erythropoietic response of hypertransfused neonatal rats suckled by anemic mothers. Anat Rec 1978; 191:227–238.
25. Kling PJ, Taing KM, Dvorak B, et al
. Insulin-like growth factor-I stimulates erythropoiesis when administered enterally. Growth Factors 2006; 24:218–223.
26. Feleppa AE. Effectiveness of oral versus intraperitoneal administration of erythropoietin in stimulating erythropoiesis. Scand J Haematol 1973; 10:186–188.
27. McCuskey RS, Meineke HA. Studies of the hemopoietic microenvironment. V. Erythropoietin-induced release of vasoactive substance(s) from erythropoietin-responsive stem cells. Proc Soc Exper Biol Med 1977; 156:181–185.
28. McCuskey RS. Erythropoietin: effect on the living fetal hepatic microvascular system in situ. Life Sci 1967; 6:2129–2133.
29. Britton JR, Christensen RD. Enteral administration of recombinant erythropoietin to preterm infants. J Perinatol 1995; 15:281–283.
30. Widness JA, Veng-Pedersen P, Schmidt RL, et al
. In vivo 125I-erythropoietic pharmacokinetics are unchanged after anesthesia, nephrectomy, and hepatectomy in sheep. J Pharmacol Exp Ther 1996; 279:1205–1210.
31. Spivak JL, Hogans BB. The in vivo metabolism of recombinant human erythropoietin in the rat. Blood 1989; 73:90–99.
32. Juul SE, Joyce AE, Zhao Y, et al
. Why is erythropoietin present in human milk? Studies of erythropoietin receptors on enterocytes of human and rat neonates. Pediatr Res 1999; 46:263–268.