Premature infants (PIs), whose gestation is <37 weeks, are vulnerable to diseases linked to oxidative stress, which is essentially an imbalance between intracellular pro- and antioxidants. In this context, oxidative stress can be accrued from several sources (1–5). For example, supplemental oxygen therapy is usually administered to these infants because of their immature respiratory systems, resulting in exposure to higher oxygen partial pressures (6). Such a transition from the relatively hypoxic uterine environment to a hyperoxic external environment results in the excessive production of damaging molecules such as superoxide and hydroxyl radicals, as well as reactive oxygen species (ROS) (7). Coupled with the immature antioxidant defense systems of PIs, free radical and ROS overproduction can induce redox imbalances, which have been linked to a variety of morbidities associated with preterm birth, including necrotizing enterocolitis, chronic lung disease, periventricular leukomalacia, bronchopulmonary dysplasia, and retinopathy of prematurity (1,3–5).
Breast milk (BM) is widely regarded as the best food for preterm infants. In addition to providing nutritional, gastrointestinal, immunological, and developmental benefits, there is good evidence that BM has antioxidant properties (8). Antioxidant enzymes, including superoxide dismutase, glutathione peroxidase, and catalase (CAT), in BM are believed to assist the destruction of H2O2 and other ROS (9,10). Furthermore, scavengers of free radicals, including α-tocopherol, cysteine, and ascorbic acid are also present in human BM, and at concentrations considerably higher than in cow's milk (11).
To date, several investigators have tested antioxidant properties of BM. Friel et al (12) compared the antioxidant protection of BM with that conferred by infant formulae. In their study, physiologic oxidative stress was induced in BM and infant formulas by hypoxanthine/xanthine oxidase, and ascorbate free-radical electron spin resonance intensity, a marker for the degree of ongoing free-radical oxidative stress, was monitored. Their results showed that BM was more effective than infant formula in reducing ascorbate radical production.
In a study conducted by Shoji et al (9), cultured IEC-6 cells were preincubated with milk with 100-fold dilutions of defatted BM, bovine milk, or 3 artificial formulae for 24 hours, followed by a 30-minute 0.5-mmol/L hydrogen peroxide treatment to induce oxidative stress. Results showed that BM treatment maintained the highest cell survival rate (50%) compared with no pretreatment (27%), bovine milk treatment (6%), or artificial formula treatment (13%–16%) of cells. These researchers concluded that BM acts as an antioxidant in the gastrointestinal tract of infants.
The basis for the present study is 2-fold: first, the luminal surface of the intestinal mucosa is considerably more complex than previously modeled. A mucin layer covers the apical surface of enterocytes and forms a microenvironment that modulates nutrient, fluid, and electrolyte balance. Second, BM that comes into contact with cells of the intestinal tract is considerably altered from its native state, having been subjected to acid hydrolysis and enzymatic digestion. Whether this highly chemically modified matrix retains antioxidant properties has not been established. Therefore, to test the antioxidant properties of BM on enterocytes under conditions more reflective of the in vivo state, we developed a combined in vitro digestion and enterocyte coculture model, the former was developed specifically to mimic parameters in the preterm infant gastrointestinal system. Based on direct and indirect endpoints of oxidative stress, our data indicate that digested BM confers antioxidant protection in enterocytes. Furthermore, we suggest that this model may be more suitable for assessing other nutrient/mucosal interactions.
MATERIALS AND METHODS
BM samples were collected from 21 mothers of PIs (born between 29 and 37 weeks of gestation) during the second week of lactation. Consent and ethical approval according to the University of Manitoba Human Research Ethics Board were obtained from women who participated in the study. BM samples were collected from mothers by using a breast pump or hand expression, transported on ice to the laboratory, frozen immediately, and stored at −80°C until analysis. To minimize individual variance, BM was pooled before any analysis.
In Vitro Digestion Protocol
An in vitro digestion model mimicking gastric and intestinal phases of digestion in the preterm infant was modified from that used by Etcheverry et al (13). Briefly, in the gastric phase, 1.73 mL of 140 mmol/L NaCl and 5 mmol/L KCl solution was added to 3.27 mL of BM to mimic the in vivo dilution of ingested BM by digestive fluids (13). After adding 0.25 mL of pepsin (Sigma, St Louis, MO) (0.2 g pepsin dissolved in 5 mL of 0.1 N HCl), BM samples were adjusted to pH 4.0 by addition of HCl and placed in a 37°C shaking incubator for 30 minutes. BM samples were then adjusted to pH 6.0 by addition of NaHCO3 and incubated for an additional 30 minutes.
In the intestinal phase, 1.25 mL of pancreatin/bile (Sigma, St Louis, MO) solution (0.05 g of pancreatin and 0.3 g of bile extract dissolved in 25 mL of 0.1 mol/L NaHCO3) and 17.2 mU of lactase (Sigma, St Louis, MO) were added to BM samples. BM samples were next adjusted to pH 7.0 by 1 mol/L NaHCO3 and to a final volume of 10 mL by 140 mmol/L NaCl and 5 mmol/L KCl solution. After gentle agitation in a 37°C incubator for 2 hours, intestinal phase end products were stored in a −20°C freezer until further analysis.
Preparation of Mucin
Porcine gastric mucin, 250 mg, (Sigma, St Louis, MO) was dissolved in 10 mL of a saline solution (pH 7, 140 mmol/L NaCl and 5 mmol/L KCl). To chelate extra iron, 0.5 g Chelex 100 (Fisher Biosciences, Ottawa, Canada) was added to the mucin solution and stirred until all of the components were dissolved. The mucin/Chelex admixture was filtered twice through glass fiber in Pasteur pipettes, then aliquoted and stored at −20°C. In all of the experiments involving mucin treatments, cocultures were first pretreated with mucin.
Coculture Cell Model
To model the human intestinal mucosa, Caco-2BBE (ATCC, Rockville, MD) cells and HT29-MTX (clone E12) cells (a gift from Drs Thomas Kissel, University of Marburg, Marburg, Germany and Per Artursson, University of Lund, Lund, Sweden) were cocultured at a ratio of 7:3 in plates precoated with 1% Matrigel Matrix (BD Biosciences, Bedford, MA). This ratio was selected because it approximates the relative proportion of enterocytes to goblet cells in the small intestine (14). Each well was plated with a total of approximately 5 × 106 cells/mL; this relatively high density precluded changes in the original ratio of Caco-2BBE:HT-29MTX cells because the latter have a much faster doubling time. To confirm this assertion, the numbers of periodic acid-Schiff–positive (PAS+) and PAS-negative (PAS−) cells were counted in 5 representative-phase contrast micrographs taken of cocultures 21 days postseeding; the average (± standard deviation) number of PAS−:PAS+ cells in these fields was 113.2 ± 5.8:51.8 ± 4.3 The rationale behind using a coculture rather than a monoculture is that in vivo, the intestinal epithelium comprises mucin-secreting cells that are functionally similar to HT-29MTX cells, as well as absorptive/secretory cells morphologically and functionally similar to Caco-2 cells. Although both cell lines are in a secretory state, HT-29MTX cells constitutively express several mucin-gene products, and our initial expectation was that this source of endogenously secreted mucin would be sufficient to protect the enterocyte monolayer from bile salts and digestive enzymes. Our rationale for using a 7:3 ratio of Caco-2BBE:HT-29MTX is that this is within a physiologically realistic range used by other researchers (see Discussion), and it provided a continuous mucin layer as assessed by whole-mount alcian blue staining. We further reasoned that a 7:3 ratio would likely confer greater protection from digestive enzymes than lower seeding ratios. Furthermore, we compared the transepithelial electrical resistance (TEER) of cultures seeded at 9:1, 7:3, and 1:1 Caco-2BE:HT-29 MTX ratios at 21 days postseeding, without mucin these values were 317 ± 3.6, 285.7 ± 6.3, and 243.94 Ω/cm2, respectively. These values are significantly different from each other (n = 3; P < 0.05). Furthermore, initial tests with cocultures seeded at a 9:1 ratio and exposed to digested BM yielded extremely low adhesion rates. Taken together, we determined that the 7:3 seeding ratio would give the best combination of mucin production, barrier function, and adhesion.
Culture medium was prepared by supplementing 500 mL Dulbecco Modified Eagle Medium (Gibco, St Burlington, Canada) with 5 mL of 100 U/mL penicillin, 100 μg/mL of streptomycin solution, 10 mL of 2 mmol/L L-glutamine solution, 5 mL of 1 mmol/L sodium pyruvate solution, 5 mL of 10 μg/mL human transferring solution, and 60 mL of 10% fetal bovine serum. Fresh culture medium was routinely changed every 2 days.
Coculture Responses to Digested BM
To verify that cocultures expressed expected characteristics similar to the in vitro state, cell morphology and mucin secretion were examined. To assess the stability of the coculture model during in vitro tests, BM cell viability, cell adhesion, and barrier integrity were measured before and after exposure to BM, with and without pretreatment with exogenous porcine mucin.
To examine the physical effects of digested BM on cell morphology either with or without exogenous mucin, cocultures on Transwell polycarbonate filters (4-μm pore size) (Corning, Wilkes Barre, PA) were fixed in Karnovsky fixative (3% glutaraldehyde, 2% paraformaldehyde) and imbedded in epon-araldite (30% epon [v/v] 0.8% araldite [v/v]) (EM Sciences, Ithaca, NY); 2% sodium dodecyl succinic anhydride (Fisher Scientific, Ottawa, Canada); N,N,N′,N′-tetramethylethylenediamine (Fisher Scientific) and sectioned on a microtome (LKB, Kars, Canada). Epon-araldite sections were stained with 1% toluidine blue and mounted under oil for microscopic examination. To determine whether cocultures expressed mucins, the cocultures were grown on Transwell membranes, fixed in 2% formaldehyde, imbedded in JB-4 methacrylate, sectioned (10-μm thick), and stained in PAS reagent.
For cell adhesion measurement, cocultures were grown in 96 black-well plates until after confluence (day 21); to determine the effectiveness of exogenous mucin in promoting cell adhesion, specific cocultures were pretreated with mucin at the concentrations described above. After incubation with digested BM, cocultures were incubated with 10 μmol/L calcein AM (Invitrogen, Eugene, OR) for 30 minutes. Calcein AM is a nonfluorescent dye that easily permeates intact live cells and is hydrolyzed into calcein by intracellular esterases (15). De-esterified calcein is cell impermeant and emits strong fluorescence after being excited at 485 nmol/L. The intensity of calcein was measured on a Fluoroskan Ascent fluorometer using appropriate excitation and emission filters (Thermo Labsystems, Helsinki, Finland).
Cell viability was measured by the live/dead assay method in which calcein AM is an indicator of live cells and ethidium bromide (EtBr) (Fisher Scientific) intercalates with DNA in dead cells (16). Cell viability was expressed as the ratio of emission intensity of calcein to that of EtBr.
Cell barrier integrity was measured by TEER using an REMS autosampler (WPI, Sarasota, Fl). Cocultures were grown on Transwell inserts (4-μm pore size) in 24-well plates until confluence. After 30-minute BM incubation, 0.5 mL of fresh culture medium was added to each apical chamber, and 1.5 mL of fresh culture medium added to each basolateral chamber, taking care to avoid disrupting cell monolayers. TEER values were then measured and compared with untreated controls incubated in cell culture medium.
Antioxidant Properties of BM
ROS-quenching properties of BM were measured by 2 relatively nonselective ROS probes, 5-(and-6)-chloromethyl-2,7-dichlorodihydrofluorescein diacetate acetyl ester (CM-H2DCFDA) (Invitrogen) and dihydrorhodamine 123 (DHR) (Invitrogen). CM- H2DCFDA is a cell-permeant fluorochrome that is hydrolyzed by intracellular esterases to form CM-H2DCF, which is retained intracellularly and is oxidized by peroxyl and hydroxyl radicals. Fluorescence intensity is proportional to the level of oxidation, permitting quantification of intracellular ROS (17). DHR is similarly a fluorochrome that is oxidized by superoxide, hydroxyl radicals, and hydrogen peroxide and is thus a complementary indicator of intracellular oxidative status (18,19).
For both fluorescence assays, cocultures were grown in 96 black-well (clear bottom) plates until confluence (approximately 14 days). Except for the control group, cells in each microplate well were loaded with 10 μmol/L of either CM-H2DCFDA or DHR for 30 minutes. Probes were then aspirated and cocultures were incubated with different BM treatments (see Results) and oxidatively challenged by addition of 1 mmol/L fresh hydrogen peroxide. After aspiration of treatment solutions and wash with phosphate-buffered saline, fluorescence intensities were measured with a Fluoroskan Ascent fluorometer (485 nM excitation/527 nM emission) after 30 minutes at 37°C.
We also measured oxidative DNA damage using the comet assay. Briefly, after BM treatments, live cells were imbedded in low-melting-point agar (38°C), which was then plated on specially coated glass microscope slides (Trevigen Inc, Gaithersburg, MD). After agarose discs solidified, cells were disrupted in lysis buffer (73.05 NaCl, 14.61 g EDTA, 0.6 g Tris, and 5 mL Triton dissolved in 500 mL of distilled water) and DNA denatured in alkaline solution (6 g NaOH pellets and 29.2 g EDTA dissolved in 500 mL of distilled water) (20). Under an electric field, heterogeneously fragmented DNA (as induced by oxidative damage) migrates electrophoretically toward the cathode, forming elongated smears of smaller DNA fragments, or a “comet-like tail.” DNA was visualized by SYBR Gold (Trevigen Inc) and observed with a Leitz transmitted fluorescence microscope equipped with a 16× neofluar lens and the appropriate filters (Zeiss, Germany). The degree of DNA damage was determined by tail moment, which is a function of comet tail length and fluorescence intensity of DNA in the comet tail (21). Images were captured with a Retiga Q-imaging digital camera and analyzed by CASP software (courtesy of Andrzej Wojcik, University of Wroclaw, Institute of Theoretical Physics, Poland). Culture medium treatment was used as the control group and 20 cells were randomly selected for tail moment measurement in each treatment.
In the present study, loss of cell viability, cell adhesion, and barrier integrity were observed after a 30-minute incubation in digested BM. It seems reasonable to suggest that this was due to either the activity of digestive enzymes or bile-salt toxicity. Rather than further dilute or fractionate digested BM, or use BM dialysate, as has been done in previous studies, we chose to mitigate the effects of digested BM by addition of additional exogenous mucin to the apical surface of cocultures. Given the complex 3-dimensional architecture of the intestinal mucosal in vivo and the protection afforded by villar trapping of mucin, this approach was judged to be the most appropriate for the purpose of assaying native BM. Mucin was prepared according to the methods used by Jin et al. Briefly, 250 mg of porcine mucin (Sigma) was dissolved in 10 mL pH 7.4 phosphate-buffered saline. After addition of 0.5 g Chelex 100 chelating resin (Bio-Rad, Hercules, CA), the mucin solution was filtered twice through pipettes clogged by glass fiber to remove ions existing in mucin. In experimental conditions involving the use of exogenous mucin, cultures were pretreated with mucin (25 mg/mL) before treatments.
For each assay, 1-way analysis of variance (ANOVA) with the Tukey post-hoc test to describe the relation between means was used, and P < 0.05 was regarded as statistically significant.
Light microscopy revealed confluent cells closely apposed to each other and attached to the polycarbonate filter (Fig. 1A). Cells with distinctively differentiated morphologies were evident: Some cells clearly expressed brush borders; others had numerous inclusions resembling mucin granules. Results of PAS staining showed that mucus secretion, beginning at day 12 postseeding, reached a maximum by day 21 (Fig. 1B).
Both cell viability and cell adhesion assays suggested that digested BM treatment led to significant cell death and cell detachment compared with nondigested BM treatment. Data from the TEER assay indicated that the monolayer barrier integrity was compromised by incubation for 30 minutes in digested BM (data not shown). We reasoned that cell integrity was likely disrupted by still-active digestive enzymes added during the in vitro digestion process.
To overcome the adverse effects of in vitro digested BM, exogenous mucin was added to cocultures before experiments. Digestive enzymes added directly to cell cultures resulted in a significant loss in cell viability, as did digested BM without earlier treatment with mucin; however, there were no significant differences in live:dead ratios between cultures pretreated with mucin and in vitro digested BM treated and the culture medium–treated group after 30-minute exposure (Fig. 2). Similarly, compared with baseline values, exogenous mucin prevented any significant changes in cell adhesion after 30-minute exposure to digested BM (Fig. 3). As demonstrated in Figure 4, a 30-minute incubation in digested BM reduced cell barrier integrity despite mucin addition, and TEER values were recovered within 3 hours postexposure. In contrast, cell barrier integrity did not recover in cultures without exogenous mucin.
To determine the effects of treatments on cell morphology, cells were fixed and processed as described above. Longitudinal thick (2 μmol/L) sections of cells revealed that after 3 hours recovery, cell integrity in cocultures not supplemented with exogenous mucin was visibly altered. Microvillar loss, cellular vacuolisation, and complete disruption of patches of monolayer were evident (Fig. 5A). In contrast, microscopic cell structure and microvilli were well preserved after a 30-minute incubation with in vitro digested BM treatment if cells were pretreated with exogenous mucin (Fig. 5B). Taken together, these data suggest that exogenous mucin conferred a protection that enabled cocultures to remain structurally intact after a 30-minute exposure to digested BM.
Effects of BM samples on reducing oxidative stress in the coculture model were measured by CM-H2DCFDA assay, DHR assay, and comet assay. BM samples were diluted by a factor of approximately 3 during the in vitro digestion process. To account for this dilution, nondigested BM samples were diluted 1:3 times for comparison of antioxidant properties with in vitro digested BM.
In the CM-H2DCFDA assay, cocultures were incubated with different treatments before 30-minute exposure to oxidative stress induced by 1 mmol/L hydrogen peroxide. Figure 6 illustrates the effects of these treatments on intracellular ROS. Nondigested BM, 1:3-diluted nondigested BM, and 10 U/mL CAT (Sigma) exhibited equal potential in reducing intracellular oxidative stress compared with the culture medium–treated group. Similar results were obtained using the DHR assay (Fig. 7).
Oxidative DNA damage induced by 500 μmol/L hydrogen peroxide yielded higher numbers of “cometized” cells and elevated comet tail moments. Representative fluorescence micrographs of intact and damaged DNA after whole-cell electrophoresis are presented in Figure 8A. In the absence of frank DNA damage, no significant electrophoretic migration is detectable. DNA damage as measured by comet tail moment is depicted in Figure 8B. Nondigested BM and CAT significantly minimized tail moment compared with the positive control (H2O2 treatment). No statistically significant differences were seen between 1:3 diluted nondigested BM and digested BM with respect to mitigation of DNA damage. All of the BM treatments resulted in DNA tail moments that were significantly lower than the medium + peroxide treatment.
To better understand the protective effects of human BM on the intestinal epithelium, we combined both in vitro digestion and a Caco-2BBE/HT29-MTX coculture in a novel model. Caco-2BBE include cells that express a brush border, tight junctions, and an apical/basolateral polarization. HT29-MTX cells secrete several mucin products (22,23), and our intention was to develop a heterogeneous culture able to form a barrier with complex absorptive and secretory characteristics. In addition, mucin secreted by HT29-MTX cells serve to create an apical microenvironment similar to that found in vivo (14). Finally, the Matrigel substrate contains collagen, laminins, and fibronectin normally found in basement membranes, facilitating cell attachment, multiplication, and differentiation (24).
To develop an accurate representation of digestive processes in the gastrointestinal tract of PI, a thorough literature review was used to inform optimal luminal pH and enzyme levels, as well as transit time in preterm infants. Normally, gastric digestion occurs for approximately 1 hour and intestinal incubation for 2 hours, based on the study by Bode et al (25), and gastric pH was adjusted to 4.0 in the first 30-minute incubation and raised to 6.0 in the second 30-minute incubation to simulate the gradual gastric pH change of PIs during BM feeding (26). Furthermore, in the intestinal phase, exogenous lactose was added to compensate the insufficient lactose secretion by the cell culture model.
In vitro digestion was used by Jovani et al (27) and Etcheverry et al (13) to hydrolyze infant formulae or BM. These researchers used specific digestive enzyme concentrations, incubation times, and variations in intestinal pH, although the basis for some of these is unclear. We believe that our in vitro digestion protocol differs from that of the other investigators in that it is specifically based on empiric measurements of all of the above parameters. One key difference of note is that pH values used during the gastric digestion phase in our protocol are considerably higher than those used in other studies, which usually conform to adult pH values.
Our results indicated that mucin addition enabled cocultures to maintain their characteristics after incubation for 30 minutes with digested BM. Compared with previous studies, in which digestive enzymes were heated or dialyzed, addition of exogenous mucin in the present study is more consistent with physiological conditions, providing a similar microenvironment to that found in vivo (28,29). Although we had initially hoped that HT-29MTX cells alone would provide adequate mucin protection from digestive enzymes, we speculate that compared with the more complex, folded arrangement of mucosal villi in the intestine likely helps to better retain secreted mucin. In our flat, planar coculture model, the exogenous mucin proved necessary to maintain barrier integrity. On one hand, the mechanism(s) underlying the toxicity of digested milk can be ascribed to still-active proteolytic enzymes. On the other hand, it is equally plausible that the bile salts added during the in vitro digestion process are toxic, as has been noted by Lowes and Simmons (30), who determined that the human enterocyte cell lines Caco-2 and T-84 are sensitive to unconjugated bile acids such as cholic acid. Whatever the source(s) of the insult, further improvements to this model, such as the use of 3-dimensional, folded growth substrates, may eliminate the need for additional mucin. The mechanism(s) by which mucin confers protection against digestive enzymes is unclear; what is well known is the importance of mucin in maintaining an adluminal microenvironment requisite for mucosal protection, solute and fluid transport, and pH balance.
It is important to note that Mahler et al (31) have recently used Caco-2/HT-29 MTX cocultures with exogenous porcine mucin to study the effects of in vitro digested foods on iron transport, using both 9:1 and 7:3 seeding ratios, and also report that exogenous mucin has a protective effect against digestive enzymes. Our model differs in that we have developed a process specifically designed to reflect the in vivo digestive processes in the preterm infant, and we have also used a different Caco-2 subclone as well as an artificial basement membrane matrix.
Aycicek et al (10) probed total antioxidant capacity and total peroxides of plasma. Their results showed that breast-feeding enhanced plasma antioxidant capacity and lowered plasma total peroxide compared with formula feeding. In another study by Shoji et al (32), infants were classified into different groups based on types of feeding. Urinary 8-hydroxy-2′-deoxyguanosine (8-OHdG), 1 of the biomarkers of oxidative DNA damage, was measured and data suggested urinary 8-OHdG excretion of the breast-fed group was significantly lower in comparison to that of the formula-fed group. The researchers concluded that BM acts as an antioxidant during infancy.
Data from the present study indicated that in vitro digested milk reduced oxidative stress in the coculture cell model. This finding is consistent with results of the previous studies, confirming the notion that BM reduces oxidative stress in a physiological system. The combination of the coculture, exogenous mucin, and digested BM can be argued to be a good approximation of both the physiological microenvironment of intestinal mucosa and the substrate to which it is exposed in vivo. However, based on the data presented herein, we cannot discern the actual mechanisms underlying the protective effects of BM; nor is it possible to determine whether the antioxidant activity was conferred in the extracellular matrix or in the intracellular environment. Work is in progress in our laboratory to determine the specific component(s) in digested BM that protect(s) against oxidative stress, and whether this protection is a consequence of biochemical scavenging of free radicals and/or upregulation of antioxidant enzymes via genomic effects. We anticipate that this model will constitute a useful tool in both studies of the effects of enteral nutrition in vitro as well as for pharmacological and bioavailability studies.
We thank Mr Allistair Carrothers for technical assistance.
1. Moison RM, de Beaufort AJ, Haasnoot AA, et al. Uric acid and ascorbic acid redox ratios in plasma and tracheal aspirate of preterm babies with acute and chronic lung disease. Free Radic Biol Med 1997; 23:226–234.
2. Anderson D. Handbook of Pediatric Nutrition. Sudbury, MA: Jones & Bartlett; 2004.
3. Cervantes-Munguia R, Espinosa-Lopez L, Gomez-Contreras P, et al. Retinopathy of prematurity and oxidative stress. An Pediatr (Barc) 2006; 64:126–131.
4. Okur H, Kucukaydin M, Kose K, et al. Hypoxia-induced necrotizing enterocolitis in the immature rat: the role of lipid peroxidation and management by vitamin E. J Pediatr Surg 1995; 30:1416–1419.
5. Welty SE. Is there a role for antioxidant therapy in bronchopulmonary dysplasia? J Nutr 2001; 131:947S–950S.
6. Saugstad OD. Oxygen for newborns: how much is too much? J Perinatol 2005; 25(Suppl 2):S45–S50.
7. Saugstad OD. Oxygen toxicity in the neonatal period. Acta Paediatr Scand 1990; 79:881–892.
8. Callen J, Pinelli J. A review of the literature examining the benefits and challenges, incidence and duration, and barriers to breastfeeding in preterm infants. Adv Neonatal Care 2005; 5:72–92.
9. Shoji H, Oguchi S, Fujinaga S, et al. Effects of human milk and spermine on hydrogen peroxide-induced oxidative damage in IEC-6 cells. J Pediatr Gastroenterol Nutr 2005; 41:460–465.
10. Aycicek A, Erel O, Kocyigit A, et al. Breast milk provides better antioxidant power than does formula. Nutrition 2006; 22:616–619.
11. Goldman AS, Thorpe LW, Goldblum RM, et al. Anti-inflammatory properties of human milk. Acta Paediatr Scand 1986; 75:689–695.
12. Friel JK, Martin SM, Langdon M, et al. Milk from mothers of both premature and full-term infants provides better antioxidant protection than does infant formula. Pediatr Res 2002; 51:612–618.
13. Etcheverry P, Wallingford JC, Miller DD, et al. Calcium, zinc, and iron bioavailabilities from a commercial human milk fortifier: a comparison study. J Dairy Sci 2004; 87:3629–3637.
14. Hilgendorf C, Spahn-Langguth H, Regardh CG, et al. Caco-2 versus Caco-2/HT29-MTX co-cultured cell lines: permeabilities via diffusion, inside- and outside-directed carrier-mediated transport. J Pharm Sci 2000; 89:63–75.
15. Neri S, Mariani E, Meneghetti A, et al. Calcein-acetyoxymethyl cytotoxicity assay: standardization of a method allowing additional analyses on recovered effector cells and supernatants. Clin Diagn Lab Immunol 2001; 8:1131–1135.
16. Gray DW, Morris PJ. The use of fluorescein diacetate and ethidium bromide as a viability stain for isolated islets of Langerhans. Stain Technol 1987; 62:373–381.
17. Shanker G, Aschner JL, Syversen T, et al. Free radical formation in cerebral cortical astrocytes in culture induced by methylmercury. Brain Res Mol Brain Res 2004; 128:48–57.
18. Biziukin AV, Korkina LG, Velichkovskii BT. Comparative use of 2,7-dichlorofluorescein diacetate, dihydrorhodamine 123, and hydroethidine for studying oxidative metabolism of phagocytosing cells. Biull Eksp Biol Med 1995; 119:361–365.
19. Yi J, Gao F, Shi G, et al. Apoptosis susceptibility of tumor cells to arsenic trioxide and the inherent cellular level of reactive oxygen species. Chin Med J (Engl) 2002; 115:603–606.
20. Tice RR, Agurell E, Anderson D, et al. Single cell gel/comet assay: guidelines for in vitro and in vivo genetic toxicology testing. Environ Mol Mutagen 2000; 35:206–221.
21. de Lima Leite A, Ferreira Santiago J Jr, Mauad Levy F, et al. Absence of DNA damage in multiple organs (blood, liver, kidney, thyroid gland and urinary bladder) after acute fluoride exposure in rats. Hum Exp Toxicol 2007; 26:435–440.
22. Sambuy Y, De Angelis I, Ranaldi G, et al. The Caco-2 cell line as a model of the intestinal barrier: influence of cell and culture-related factors on Caco-2 cell functional characteristics. Cell Biol Toxicol 2005; 21:1–26.
23. Lesuffleur T, Barbat A, Dussaulx E, et al. Growth adaptation to methotrexate of HT-29 human colon carcinoma cells is associated with their ability to differentiate into columnar absorptive and mucus-secreting cells. Cancer Res 1990; 50:6334–6343.
24. Frisk T, Rydholm S, Andersson H, et al. A concept for miniaturized 3-D cell culture using an extracellular matrix gel. Electrophoresis 2005; 26:4751–4758.
25. Bode S, Dreyer M, Greisen G. Gastric emptying and small intestinal transit time in preterm infants: a scintigraphic method. J Pediatr Gastroenterol Nutr 2004; 39:378–382.
26. Omari TI, Davidson GP. Multipoint measurement of intragastric pH in healthy preterm infants. Arch Dis Child Fetal Neonatal Ed 2003; 88:F517–F520.
27. Jovani M, Barbera R, Farre R, et al. Calcium, iron, and zinc uptake from digests of infant formulas by Caco-2 cells. J Agric Food Chem 2001; 49:3480–3485.
28. Ait-Oukhatar N, Bouhallab S, Bureau F, et al. In vitro digestion of caseinophosphopeptide-iron complex. J Dairy Res 2000; 67:125–129.
29. Bosscher D, Lu Z, Janssens G, et al. In vitro availability of zinc from infant foods with increasing phytic acid contents. Br J Nutr 2001; 86:241–247.
30. Lowes S, Simnons NL. Human intestinal cell monolayers are preferentially sensitive to disruption of barrier function from basolateral exposure to cholic acid: correlation with membrane transport and transepithelial secretion. Pflugers Archiv 2001; 443:265–273.
31. Mahler GJ, Shuler ML, Glahn RP. Characterization of Caco-2 and HT29-MTX cocultures in an in vitro digestion/cell culture model used to predict iron bioavailability. J Nutr Biochem 2009;20:494–502.
32. Shoji H, Oguchi S, Shimizu T, et al. Effect of human breast milk on urinary 8-hydroxy-2′-deoxyguanosine excretion in infants. Pediatr Res 2003; 53:850–852.
© 2010 Lippincott Williams & Wilkins, Inc.